Acoustic Waves

ACOUSTIC WAVES
FROM MICRODEVICES TO
HELIOSEISMOLOGY

Edited by Marco G. Beghi

Acoustic Waves From Microdevices to Helioseismology
Edited by Marco G. Beghi

Published by InTech
Janeza Trdine 9, 51000 Rijeka, Croatia

Copyright 2011 InTech
All chapters are Open Access distributed under the Creative Commons Attribution 3.0
license, which permits to copy, distribute, transmit, and adapt the work in any medium,
so long as the original work is properly cited. After this work has been published by
InTech, authors have the right to republish it, in whole or part, in any publication of
which they are the author, and to make other personal use of the work. Any republication,
referencing or personal use of the work must explicitly identify the original source.

As for readers, this license allows users to download, copy and build upon published
chapters even for commercial purposes, as long as the author and publisher are properly
credited, which ensures maximum dissemination and a wider impact of our publications.

Notice
Statements and opinions expressed in the chapters are these of the individual contributors
and not necessarily those of the editors or publisher. No responsibility is accepted for the
accuracy of information contained in the published chapters. The publisher assumes no
responsibility for any damage or injury to persons or property arising out of the use of any
materials, instructions, methods or ideas contained in the book.

Publishing Process Manager Sandra Bakic
Technical Editor Teodora Smiljanic
Cover Designer Roko Kerovec
Image Copyright Bocos Benedict, 2011. Used under license from Shutterstock.com

First published October, 2011
Printed in Croatia

A free online edition of this book is available at www.intechopen.com
Additional hard copies can be obtained from [email protected]

Acoustic Waves From Microdevices to Helioseismology, Edited by Marco G. Beghi
p. cm.
ISBN 978-953-307-572-3

Contents

Preface IX
Part 1 Theoretical and Numerical Investigations
of Acoustic Waves 1
Chapter 1 Analysis of Acoustic Wave in Homogeneous and
Inhomogeneous Media Using Finite Element Method 3
Zi-Gui Huang
Chapter 2 Topological Singularities in Acoustic Fields due to Absorption
of a Crystal 21
V. I. Alshits, V. N. Lyubimov and A. Radowicz
Chapter 3 An Operational Approach to the
Acoustic Analogy Equations 49
Dorel Homentcovschi and Ronald Miles
Chapter 4 Exact Solutions Expressible in Hyperbolic and Jacobi
Elliptic Functions of Some Important Equations of
Ion-Acoustic Waves 67
A. H. Khater and M. M. Hassan
Chapter 5 Acoustic Wave 79
P. K. Karmakar
Part 2 Acoustic Waves as Investigative Tools 123
Chapter 6 Acoustic Waves: A Probe for the Elastic Properties
of Films 125
Marco G. Beghi
Chapter 7 Evaluation Method for Anisotropic Drilling Characteristics of
the Formation by Using Acoustic Wave Information 147
Deli Gao and Qifeng Pan
VI Contents

Chapter 8 Machinery Faults Detection Using
Acoustic Emission Signal 171
Dong Sik Gu and Byeong Keun Choi
Chapter 9 Compensation of Ultrasound Attenuation in Photoacoustic
Imaging 191
P. Burgholzer, H. Roitner, J. Bauer-Marschallinger,
H. Grn, T. Berer and G. Paltauf
Chapter 10 Low Frequency Acoustic Devices for Viscoelastic Complex
Media Characterization 213
Georges Nassar
Chapter 11 Modeling of Biological Interfacial Processes Using
ThicknessShear Mode Sensors 239
Ertan Ergezen, Johann Desa, Matias Hochman, Robert Weisbein
Hart, Qiliang Zhang, Sun Kwoun, Piyush Shah and Ryszard Lec
Chapter 12 Analysis of Biological Acoustic Waves by Means of the
PhaseSensitivity Technique 259
Wojciech Michalski, Wojciech Dziewiszek and Marek Bochnia
Chapter 13 Photoacoustic Technique Applied to Skin Research:
Characterization of Tissue, Topically Applied Products and
Transdermal Drug Delivery 287
Jociely P. Mota, Jorge L.C. Carvalho,
Srgio S. Carvalho and Paulo R. Barja
Chapter 14 AcousticGravity Waves in the Ionosphere During Solar
Eclipse Events 303
Petra Kouck Knov and Zbyek Mona
Part 3 Acoustic Waves as Manipulative Tools 321
Chapter 15 Use of Acoustic Waves for Pulsating
Water Jet Generation 323
Josef Foldyna
Chapter 16 Molecular Desorption by LaserDriven Acoustic Waves:
Analytical Applications and Physical Mechanisms 343
Alexander Zinovev, Igor Veryovkin and Michael Pellin
Chapter 17 Excitation of Periodical Shock Waves in SolidState Optical
Media (Yb:YAG, Glass) at SBS of Focused LowCoherent
Pump Radiation: Structure Changes, Features of Lasing 369
N.E. Bykovsky and Yu.V. Senatsky
Contents VII

Chapter 18 An Optimal Distribution of Actuatorsin Active Beam
Vibration Some Aspects, Theoretical Considerations 397
Adam Braski
Part 4 Acoustic Wave Based Microdevices 419
Chapter 19 Multilayered Structure as a Novel Material for Surface
Acoustic Wave Devices: Physical Insight 421
Natalya Naumenko
Chapter 20 SAW Parameters Analysis and Equivalent Circuit
of SAW Device 443
Trang Hoang
Chapter 21 Sources of ThirdOrder Intermodulation Distortion in Bulk
Acoustic Wave Devices: A Phenomenological Approach 483
Eduard Rocas and Carlos Collado
Chapter 22 Shear Mode Piezoelectric Thin Film Resonators 501
Takahiko Yanagitani
Chapter 23 Polymer Coated Rayleigh SAW and STW Resonators for Gas
Sensor Applications 521
Ivan D. Avramov
Chapter 24 Ultrananocrystalline Diamond as Material for Surface
Acoustic Wave Devices 547
Nicolas Woehrl and Volker Buck
Chapter 25 Aluminum Nitride (AlN) Film Based Acoustic Devices:
Material Synthesis and Device Fabrication 563
Jyoti Prakash Kar and Gouranga Bose
Chapter 26 Surface Acoustic Wave Devices for Harsh Environment 579
Cinzia Caliendo
Chapter 27 Applications of InFiber AcoustoOptic Devices 595
C. Cuadrado-Laborde, A. Dez, M. V. Andrs,
J. L. Cruz, M. Bello-Jimenez, I. L. Villegas,
A. Martnez-Gmez and Y. O. Barmenkov
Chapter 28 Surface Acoustic Waves and NanoElectromechanical
Systems 637
Dustin J. Kreft and Robert H. Blick

Preface

The subject of acoustic waves might easily be considered a mature one, quite
specialized, with narrow and circumscribed fields of interest and of application. The
present book is an evidence of the opposite: it witnesses how the concept of acoustic
wave, a collective displacement of matter which perturbs an equilibrium
configuration, is a pervasive concept, which emerges in very different fields. This type
of phenomena can be analyzed from different points of view, it can be exploited in
different ways, and is the object of active investigations. The present book, far from
pretending to give an exhaustive overview of the subject, offers instead a sampling of
various points of view, of applications, and of research objectives which are actively
pursued.
It must first be remembered that acoustic waves are supported by all the forms of
matter: solids, liquids, gases and plasmas. And if similarities among the different
phenomena are deep enough for them to deserve the same name, nevertheless the
peculiarities connected to the various media are significant. Although the range of
involved length and time scales is huge, going from sub-micrometric layers exploited
in microdevices to seismic waves propagating in the Suns interior, the more profound
peculiarities of the various cases concern the very heart of the phenomena, namely the
type of forces which, in different types of media, tend to restore the equilibrium
configuration.
These phenomena can be approached under different points of view. A first type of
approach aims at a better comprehension of phenomena. Many aspects of acoustic
waves are nowadays well understood, but the investigation is obviously never ending.
A line of research aims at the theoretical exploration, also by relatively sophisticated
mathematical analyses, of various aspects of phenomena whose basic laws are well
established. Concerning acoustic waves in elastic solids, Huang recalls the characters
of such waves in homogeneous isotropic media. Then he exploits recent computational
tools to analyze the modifications occurring in media which are periodically
inhomogeneous, like composite materials. Alshits, Lyubimov & Radowicz investigate
instead the elastic waves in solids which are homogeneous but anisotropic, like single
crystals. They show that the addition of a dissipative term to the elasto-dynamic
equations has consequences which go far beyond the intuitive introduction of a
damping. This term can modify the same topology of the slowness surface, inducing a
X Preface

splitting of acoustic axes. Homentcovschi & Miles review and reformulate in an
operational way the acoustic analogy theory which describes how noise is generated
in the interaction of gas flow with stationary or mobile bodies; the application of this
approach to a range of technologies (jets, propellers, aircrafts) is easily imagined.
Khater & Hassan consider various nonlinear evolution equations which are well
established in plasma physics and fluid dynamics, and which admit wave solutions,
either periodic waves or solitary waves. They seek exact solutions, which helps to
understand phenomena more than purely numerical solutions. Karmakar considers
various aspects involved in perturbations of plasmas, from ion acoustic excitation to
turbulence, and focuses on the effects of the inertia of electrons, which is much smaller
than that of ions but is not always completely negligible. He then combines various
arguments to give a picture of solar wind plasma, which needs the description of the
solar surface boundary.
A second type of approach exploits acoustic waves as probes to gain information
about the properties or the behavior of a system. Beghi revises various methods based
on acoustic waves which aim at the elastic characterization of materials, namely of thin
films. Gao & Pan consider a specific problem of significant technical relevance for the
oil and gas industry: the drillability of rocks, and in particular its anisotropy. They
shows how the outcome of laboratory acoustic tests correlates with the drilling
properties of rocks. Gu & Choi consider instead the acoustic emission from rotating
machinery, and show how it can be exploited for the early detection of faults.
Burgholzer and coworkers focus on the photoacoustic imaging technique, and in
particular on the image reconstruction to achieve the tomographic capability: they
analyze methods to compensate for ultrasound attenuation in the media being
observed.
Since acoustic waves are relatively a non invasive probe, they can be exploited also on
delicate materials and on biological systems. Nassar presents various applications to
delicate systems in the agro-industry, like cheese undergoing ripening, for which
dedicated low frequency sensors had to be developed. Erzegen and co-workers
characterize the performance of the multi-resonant thickness shear mode sensor,
exploited with a genetic algorithm for data processing: this type of sensor is devoted to
the characterization of biological interfaces. Finally, two chapters present
measurements performed in vivo. Michalski, Dziewiszek & Bochnia discuss the
performance of phase sensitive techniques to characterize non linear systems, and
show how these techniques can be applied to cochlear microphonics to study ear
behavior. Mota, Carvalho & Barja present photoacoustic measurements performed on
human skin, to characterize the skin itself, and the transdermal drug delivery.
A completely different system is found in the ionosphere, where acoustic-gravity
waves are found. Koucka & Mosna show how the ionogram technique can be
exploited to investigate the ionosphere, in particular exploiting the waves excited by
the shadow of an eclipse.
Preface XI

A third type of approach exploits acoustic waves to perform some kind of
manipulation. Foldyna shows how acoustic transducers and waveguides can be
exploited to generate and control pulsating water jets, which can be used as machining
tools. Zinovev, Veryovkin & Pellin discuss the Laser Induced Acoustic Desorption
technique to vaporize solid material to be analyzed by mass spectrometry. This
technique is less prone to induce modifications of the analyte than the more
widespread MALDI technique, although these authors show by some experiments that
the operational mechanism is still not well understood. At the other extreme, that of
high intensity laser pulses, Bykovsky & Senatsky demonstrate how stimulated
Brillouin scattering can generate shock waves, able to induce permanent modifications
of the materials, like phase changes and cracks. Finally, Branski considers the problem
of active vibration control of beams, and investigates the optimal distribution of
actuators to perform such a control.
A fourth type of approach exploits the properties of acoustic waves to design various
types of devices, mainly micro devices. The most widely exploited type of device has a
simple basic structure: a substrate, at least one layer of piezoelectric material, an
interdigitated transducer (IDT) operating as an emitter, and another one operating as
receiver. These type of devices were originally introduced as delay lines and filters,
and were then developed also for other purposes.
Before discussing this type of device, it must be remembered that other types of
devices also exist. Kreft & Blick discuss applications of surface acoustic waves to
quantum electronics, made possible by devices like quantum dots and by the
interaction of surface acoustic waves with the electron gas. This type of device is
nanomechanical, and also exploits IDT, with acoustic waveguides to match their
acoustic impedance to that of nanomechanical devices. The chapter by Cuadrado-
Laborde and co-workers considers instead the in-fiber photonic devices, and the
acousto-optic modulator which obtained exciting traveling or standing acoustic waves
by a piezoelectric actuator. This way, a dynamic and controllable modulation of the
fiber properties is obtained by the acousto-optic effect. They review a wide variety of
configurations, showing how different devices can be obtained, including Q-switched
lasers and mode locking lasers.
Returning to the most widespread type of microdevice, its interest is witnessed by the
numerous chapters devoted to it. Naumenko reviews the most common design
configurations, and presents detailed analyses of their behavior. Hoang presents the
most adopted method of analysis of such devices, based on equivalent circuits and the
so called Mason model. The method is adopted also in other chapters, and Hoang
gives a detailed introduction of the method itself, also presenting the applications to
basic configurations. Rocas & Collado analyze, for these devices, the various sources
which can introduce a non perfectly linear behavior, leading to 3rd order
intermodulation distortion. Most of the devices of this type exploit longitudinal
acoustic waves, or surface waves polarized in the plane normal to the surface. Waves
XII Preface

transversally polarized in the plane of the device surface are less considered.
Yanagitani compares the performances of the two types of operation, showing the
possible advantages of transversal waves. Avramov performs a similar comparison,
between surface waves of the Rayleigh type and of the transverse type, for devices
which are polymer coated to act as gas sensors. Some chapters focus instead on the
production and the characterization of various materials which are of interest for the
production of this type of devices. Buck considers ultranano crystalline diamond:
diamond is the acoustically fastest material, which allows operation at the highest
frequencies. Both Kar & Bose and Caliendo consider AlN layers, a piezoelectric
material whose properties are interesting under several respects. Caliendo also
considers multilayers, including platinum and sapphire layers.
As mentioned above, the various approaches documented in this book represent a
sampling of the wide spectrum of methods and techniques involving acoustic waves.
This book is offered to the scientific community in the hope of promoting a cross
fertilization of ideas and of approaches.

Marco G. Beghi
Politecnico di Milano, Energy Department and NEMAS Center,
Milano,
Italy

Part 1
Theoretical and Numerical Investigations
of Acoustic Waves

1
Analysis of Acoustic Wave in
Homogeneous and Inhomogeneous Media
Using Finite Element Method
Zi-Gui Huang
Department of Mechanical Design Engineering, National Formosa University
Taiwan
1. Introduction
Even though the propagation of elastic/acoustic waves in inhomogeneous and layered
media has been an active research topic for many decades already, new problems and
challenges continue to be posed even up to now. In fact, during the last few years, renewed
interests have been witnessed by researchers in the various fields of acoustics, such as
acoustic mirrors, filters, resonators, waveguides, and other kinds of acoustic devices, in
relation to wave propagation in periodic elastic media. In acoustics and applied mechanics,
these developments have been triggered by the need for new acoustic devices in order to
obtain quality control of elastic/acoustic waves.
What sort of material can allow us to have complete control over the elastic/acoustic waves
propagation? We would like to discuss and answer this question in this chapter. It is well
known that the successful applications of photonic band-gap materials have hastened the
related researches on phononic band-gap materials. Analysis of Acoustic Wave in Homogeneous
and Inhomogeneous Media Using Finite Element Method explores the theoretical road leading to
the possible applications of phononic band gaps. It should quickly bring the elastic/acoustic
professionals and engineers up to speed in this field of study where elastic/acoustic waves
and solid-state physics meet. It will also provide an excellent overview to any course in
elastic/acoustic media.
Previous research on photonic crystals (Johnson & Joannopoulos, 2001, 2003; Joannopoulos
et al., 1995; Leung & Liu, 1990) has sparked rapidly growing interest in the analogous
acoustic effects of phononic crystals and periodic elastic structures. The various techniques
for band structure calculations were introduced (Hussein, 2009). There are many well-
known methods of calculating the band structures of photonic and phononic crystals in
addition to the reduced Bloch mode expansion method: the plane-wave expansion (PWE)
method (Huang & Wu, 2005; Kushwaha et al., 1993; Laude et al., 2005; Tanaka & Tamura,
1998; Wu et al., 2004 ; Wu & Huang, 2004), the multiple-scattering theory (MST) (Leung &
Qiu, 1993; Kafesaki & Economou, 1999; Psarobas & Stefanou, 2000; Wang et al., 1993), the
finite-difference (FD) method (Garica-Pabloset et al., 2000; Sun & Wu, 2005; Yang, 1996), the
transfer matrix method (Pendry & MacKinnon, 1992), the meshless method (Jun et al., 2003),
the multiple multipole method (Moreno et al., 2002), the wavelet method (Checoury &
Lourtioz, 2006; Yan & Wang, 2006), the pseudospectral method (Chiang et al., 2007), the
finite element method (FEM) (Axmann & Kuchment, 1999; Dobson, 1999; Huang & Chen,


4
2011; Wu et al., 2008), the mass-in-mass lattice model (Huang & Sun, 2010), and the
micropolar continuous modeling (Salehian & Inman, 2010).
Many studies on phononic band structures from the past decade use the PWE, MST, and FD
methods to analyze the frequency band gaps of bulk acoustic waves (BAW) in composite
materials or phononic band structures. Studies adopting the PWE method investigate the
dispersion relations and the frequency band-gap feathers of the BAW and surface acoustic
wave (SAW) modes. Other studies use the layered MST to study the frequency band gaps of
bulk acoustic waves in three-dimensional periodic acoustic composites and the band
structures of phononic crystals consisting of complex and frequency-dependent Lame
coefficients. Other researchers applied the finite-difference time-domain method to predict
the precise transmission properties of slabs of phononic crystals and analyze the mode
coupling in joined parallel phononic crystal waveguides.
The techniques for tuning frequency band gaps of elastic/acoustic waves in phononic
crystals are very important, and remain exciting research topics in the physics community.
The filling fraction, rotation of noncircular rods, different cuts of anisotropic materials, and
the temperature effect all produce large frequency band gaps in the BAW and SAW modes
of periodic structures. A previous review paper (Burger et al., 2004) discusses the technique
used to optimize the unit cell material distribution, achieving the largest possible band gap
in photonic crystals for a given cell symmetry. Studies over the past decade focus on the
theoretical and numerical analysis of phononic structures based on circular or square
cylinders embedded in background materials. In this case, the PWE method can easily
calculate the dispersion relations by constructing the structural functions with Bessel or Sinc
functions. However, research on the more complicated problem of waves in the reticular
and other special periodic band structures has not started until recently.
This chapter uses the 2D and 3D finite element methods to discuss the wave velocities of
isotropic and anisotropic materials in homogeneous media. It also considers the tunable
band gaps of acoustic waves in two-dimensional phononic crystals with reticular geometric
structures (Huang & Chen, 2011). The concept of adopting a reticular geometric structure
comes from the variations of similar geometry in bio-structural reticular formation and
fibers. The PWE method used to calculate the structural functions of densities and elastic
constants cannot numerically analyze the Gibbs phenomenon. Therefore, this chapter adopts
the FEM to discuss this special periodic band structure. Changing the filling fraction, scale
parameters, and rotating angles of reticular geometric structures can tune the frequency
band gaps of mixed polarization modes. This technique is suitable for analyzing the
phenomenon of frequency band gaps in special band structures.
2. Theory
In this chapter, based on the theorems of solid-state physics and the finite element method
with Bloch calculations, equation of motion of the acoustic modes in two-dimensional
inhomogeneous media, phononic band structures, are derived and discussed in detail. In the
beginning, the concepts of the real space and k space are introduced while the Brillouin
zone is also addressed in the text. Generalized techniques of Bloch calculations in finite
element method are used to analyze the acoustic modes in two-dimensional homogeneous
and inhomogeneous media, phononic band structures, consisting of materials with general
anisotropy. The mixed and transverse polarization modes and quasi-polarization modes are
investigated in the text.
Analysis of Acoustic Wave
in Homogeneous and Inhomogeneous Media Using Finite Element Method

5
2.1 Real space and k space
It is well-known that the analysis of wave motion in infinite periodic structures is difficult in
real space. For dealing with the periodic structures, the Fourier series and Blochs theorem
are used to expand the periodic parameters such as the density, material constants,
displacement fields, or potential. Regarding to the transformation of the real space and k
space, the reciprocal lattice vectors (RLVs) are adopted from the solid-state physics. In
general, we consider a three-dimensional phononic crystal with primitive lattice vectors
1
a ,
2
a , and
3
a . The complete set of lattice vectors is written as { }
1 2 3
| l l l = + +
1 2 3
R R a a a , where
l
1
, l
2
, and l
3
are integers. The associated primitive reciprocal lattice vectors
1
b ,
2
b , and
3
b
are determined by (Kittel, 1996)
2 ,
( )
ijk j k
i

=

1 2 3
a a
b
a a a

(1)
where
ijk
is the three-dimensional Levi-Civita completely antisymmetric symbol. The
complete set of reciprocal lattice vectors is written as { }
1 2 3
| N N N = + +
1 2 3
G G b b b , where
N
1
, N
2
, and N
3
are integers. Figure 1 shows the primitive unit cell in two-dimensional real
space while the Fig. 2 shows the relationship between the real space and k space. A property
between the primitive lattice vectors and associated primitive reciprocal lattice vectors is
2
i j ij
= b a , where
ij
is the kronecker symbol. Note that the associated primitive
reciprocal lattice vectors are constructed as k space from the concept of crystal diffraction.

2
a
1
a
1
a
1
a
2
a
2
a
I
II
III

Fig. 1. Primitive unit cell in real space

k space
Real space
2
b
1
b
2
a
1
a

Fig. 2. Relationship between the real space and k space
We will find that, in following sections, the discrete translational symmetry of a phononic
crystal allows us to classify the elastic/acoustic waves with a wave vector k. The


6
propagating modes can be written in Bloch form, consisting of a plane wave modulated
by a function that shares the periodicity of the lattice (Joannopoulos et al., 1995):
.
i i
e e

= = +
k r k r
k k k
P (r) u (r) u (r R)

(2)
The important feature of the Bloch states is that different values of k do not necessarily lead
to different modes. It is clear that a mode with wave vector k and a mode with wave vector
k+G are the same mode, where G is a reciprocal lattice vector. The wave vector k serves to
specify the phase relationship between the various cells that are described by u. If k is
increased by G, then the phase between cells is increased by GR, which we know is 2n (n=
l
1
N
1
+l
2
N
2
+ l
3
N
3
is an integer) and not really a phase difference at all. So incrementing k by G
results in the same physical mode. This means that we can restrict our attention to a finite
zone in reciprocal space in which we cannot get from one part of the volume to another by
adding any G. All values of k that lie outside of this zone, by definition, can be reached from
within the zone by adding G, and are therefore redundant labels shown in Fig. 3. This zone
is the so-called Brillouin zone.

k
y
k
x
k'
K
G
a
-

a
-

a
G

Fig. 3. All values of k that lie outside of this zone, by definition, can be reached from within
the zone by adding G

1
0
8 7 6
5 4
3 2
2 /a
1 Bril louin zone
st
3 Brilloui n zone
rd
2 Bril louin zone
nd

Fig. 4. Brillouin zones in a square lattice

7
By the periodicity of the reciprocal lattice, any reciprocal lattice point which represents a
wave vector k outside the first Brillouin zone can be found a corresponding point in the first
Brillouin zone. Therefore, the wave vectors k can always be confined in the first Brillouin
zone. In the square lattice, only the wave vectors k in the region of the first Brillouin zone
between a to a (the lattice constant is a) need to be considered. The Fig. 4 shows the
first, second, and third Brillouin zones. For more details, it is best to consult the first few
chapters of a solid-state physics text, such as Kittel, 1996, or consult the appendix of popular
photonic text like Joannopoulos et al. 1995 and Johnson & Joannopoulos, 2001, 2003.
2.2 Equation of motion
This section provides a brief introduction of the theory of analyzing acoustic wave
propagation in inhomogeneous media like as phononic band structures. The theory in this
chapter can also be used to discuss acoustic wave propagation in homogeneous media
because a homogeneous medium is symmetric with respect to any periodicity.
In an inhomogeneous linear elastic medium with no body force, the equation of motion of
the displacement vector ( , ) t u r can be written as
( ) ( , ) [ ( ) ( , )],
i j ijmn n m
u t C u t = r r r r

(3)
where ( , ) ( , , ) z x y z = = r x is the position vector, t is the time variable, and ( ) r and ( )
ijmn
C r
are the position-dependent mass density and elastic stiffness tensor, respectively. The
following discussion considers a periodic structure consisting of a two-dimensional periodic
array (x-y plane) of material A embedded in a background material B shown in Fig. 5. It is
noted that when the properties of materials A and B tend to coincide, the homogeneous case
is recovered.

x
y
A
B
B A
z
x
y
0
Half space
r
0
a

Fig. 5. Periodic structures with square lattice. When the properties of materials A and B tend
to coincide, the homogeneous case is recovered
To calculate the dispersion diagrams of periodic structures, this study uses COMSOL
Multiphysics software to apply the Bloch boundary condition to the unit cell domain in the
FEM method. Based on the periodicity of phononic crystals, the displacement and stress
components in the periodic structure are expressed as follows:
( , ) ( , ),
i
i i
u t e U t
=
k x
x x

(4)


8
( , ) ( , ),
i
ij ij
t e T t

=
k x
x x (5)
where
1 2
( , ) k k = k is the Bloch wave vector, and 1 i = ; ( , )
i
U t x and ( , )
ij
T t x are periodic
functions that satisfy the following relation (Tanaka et al., 2000):
( , ) ( , ),
i i
U t U t + = x R x

(6)
( , ) ( , ),
ij ij
T t T t + = x R x (7)
where R is a lattice translation vector with components of
1
R and
2
R in the x and y
directions. The relationships between the original variables ( , )
i
u t x , ( , )
ij
t x , ( , )
i
u t + x R ,
and ( , )
ij
t + x R about the Bloch boundary conditions are characterized as:

( )
( , ) ( , ) ( , ) ( , ),
i i i i
i i i i
u t e U t e e U t e u t
+
+ = + = =
k x R k R k x k R
x R x R x x

(8)

( )
( , ) ( , ) ( , ) ( , ).
i i i i
ij ij ij ij
t e T t e e T t e t
+
+ = + = =
k x R k R k x k R
x R x R x x (9)
The Bloch calculations in this study record the variation of the displacements, stress fields,
and eigen-frequencies as the wave vector increases. By using the FEM, the unit cell is
meshed and divided into finite elements which connect by nodes, and is used to obtain the
eigen-solutions and mechanical displacements. The types of finite elements used in this
chapter are the default element types, Lagrange-quadratic, in COMSOL Multiphysics. In
order to simulate the dispersion diagrams, the wave vectors are condensed inside the first
Brillouin zone in the square lattice. According to the above theories, the results of dispersion
relations in a band structure along the are characterized and presented in the
following sections.

Fig. 6. Brillouin regions of the square and rectangular lattices
This chapter considers a periodic homogeneous medium with square lattice and phononic
structures with square and rectangular lattices. These lattices consist of periodic structures
that form two-dimensional lattices with lattice spacing R (square lattice) and lattice spacing
aR (rectangular lattice). The term a is a scale from 0.1 to 2.0 in this chapter. The periodic
structures are parallel to the z-axis. Figures 6(a) and 6(b) illustrate the Brillouin regions of
the square lattice and rectangular lattice, respectively. In the square lattice, Fig. 6(a) shows

9
the irreducible part of the Brillouin zone, which is a triangle with vertexes , , and .
Similarly, Fig. 6(b) shows the irreducible part of the Brillouin zone of a rectangular lattice
due to the geometric anisotropy, which is a rectangle with vertexes , , , and Y , and
the same as discussing the material anisotropy (Wu et al., 2004).
The finite element method divides a unit cell with a three-dimensional model into finite
elements connected by nodes. The FEM obtains the eigen-solutions and contours of a mode
shape. To simulate the dispersion diagrams, the wave vectors are condensed inside the first
Brillouin zone in the square and rectangular lattices. Using the theories above, the following
section presents the results of dispersion relations in a band structure for the
square lattice or isotropic materials, and Y rectangular lattice or anisotropic
materials. Note that the 2D FEM model calculates the dispersion relations of mixed
polarization modes, while the 3D FEM model describes the dispersion relations of mixed
and transverse polarization modes.
3. Acoustic wave in homogeneous media
It can be noted that a homogeneous medium is symmetric with respect to any periodicity,
and it can be shown that the results for an infinite homogeneous medium can be cast in the
form appropriate for a periodic medium. In this section, we introduce the mixed
polarization modes and transverse polarization modes in a homogeneous medium.
Displacement fields (polarizations) are also investigated and used to distinguish the
different modes in the dispersion relations. The aluminum and quartz are adopted for
examples and discussed in the section. The wave velocities of different propogating modes
are also observed and discussed.
3.1 Isotropic medium
In Fig. 5, when the properties of materials A and B tend to coincide, the homogeneous case
is recovered. Consider a periodic structure consisting of aluminum (Al) circular cylinders
embedded in a background material of Al forming a two-dimensional square lattice with
lattice spacing R. It means this is a homogeneous medium in a 3D FEM model. Figure 7
shows the dispersion relations along the boundaries of the irreducible part of the Brillouin
zone . The vertical axis is the frequency (Hz) and the horizontal axis is the
reduced wave vector
*
/ k kR = . Here, k is the wave vector along the Brillouin zone. The
Youngs modulus E, Poissons ratio , and density of the material Al utilized in this
example are E=70 GPa, =0.33, and =2700 kg/m
3
.
As the elastic waves propagate along the x axis, the nonvanishing displacement fields of the
shear horizontal mode (SH), shear vertical mode (SV), and longitudinal mode (L) are u
y
, u
z
,
and u
x
respectively. It is noted that wave velocity
, ,
/ 2 *
S L S L
c d dk R m = = , so the slopes of
dispersion curves in the section of Fig. 7 are exactly the straight lines and can be
explained as the wave velocities of shear (S) and longitudinal (L) modes. Here, m
S,L
are the
slopes of shear and longitudinal modes in Fig. 7. It is noted that the wave velocities of shear
horizontal mode and shear vertical mode are the same in an isotropic material. From the
results in Fig. 7, the wave velocities of shear and longitudinal modes are 3119 and 6174 m/s.
As we know, the wave velocities of shear and longitudinal modes in an isotropic material
can be obtain from


10

1
3122 / ,
2(1 )
S
E
c m s

= =
+

(10)

(1 )
6031 / .
(1 )(1 2 )
L
E
c m s
= =
+
(11)
Note that the FEM method can easily describe the mode characteristics. Figure 8 shows the
vibration mode shapes of unit cell for shear and longuitudinal modes in X point. In this
example, Fig. 8(a) is a shear horizontal mode with mode vibrating displacement along the y
direction when the wave propagates along the x direction ( direction). Also, Fig. 8(b) is
a shear vertical mode with mode vibrating displacement along z direction, and Fig. 8(c) is a
longitudinal mode with mode vibrating displacement along x direction. The arrows shown
in Fig. 8 are the polarizations.

Fig. 7. The dispersion relations of homogeneous and isotropic material Al along the
boundaries of the irreducible part of the Brillouin zone

(a) (b) (c)
Fig. 8. (a) shear horizontal mode (b) shear vertical mode, (c) longitudinal mode in the Al

11
3.2 Anisotropic medium
Similarly, the method in this chapter is used to discuss the wave velocities of acoustic modes
in an anisotropic material. Consider a periodic structure consisting of quartz circular
cylinders embedded in a background material of quartz forming a two-dimensional square
lattice with lattice spacing R. This is also a homogeneous medium. The quartz is a
piezoelectric and anisotropic material. The density =2651 kg/m
3
. The elastic constants,
piezoelectric constants, and relative permittivity of quartz utilized in this example are
shown in Tables 1-3. The piezoelectric material, quartz, is a complete structural-electrical
material, and thus all piezoelectric material properties were defined and entered into the
FEM model. Figure 9 shows the dispersion relations along the boundaries of the irreducible
part of the Brillouin zone Y due to the material anisotropy. In the
calculations, the x-y plane is parallel to the (001) plane and the x axis is along the [100]
direction of quartz. The vertical axis is the frequency in Hz unit and the horizontal axis is the
reduced wave vector.

86.7362 6.98527 11.9104 17.9081 0 0
6.98527 86.7362 11.9104 -17.9081 0 0
11.9104 11.9104 107.194 0 0 0
17.9081 -17.9081 0 57.9428 0 0
0 0 0 0 57.9492 17.9224
0 0 0 0 17.9224 39.9073
Table 1. The elastic constants of quartz in GPa unit

-0.19543 0.19543 0 -0.1212 0 0
0 0 0 0 0.12127 0.19558
0 0 0 0 0 0
Table 2. The piezoelectric constants of quartz in C/m
2
unit

4.4093 0 0
0 4.4092 0
0 0 4.68
Table 3. The relative permittivity of quartz
Shown in section of Fig. 9, the cross symbols represent the quasi shear horizontal
(quasi-SH) mode. The square symbols represent the quasi shear vertical (quasi-SV) mode
and the open circle symbols represent the quasi longitudinal (quasi-L) mode. The wave
velocities of quasi-SH, quasi-SV, and quasi-L modes along x axis are 3306, 5116, and 5741
m/s. Similarly, The wave velocities of quasi-SH, quasi-SV, and quasi-L modes along y axis
( Y section) are 3922, 4311, and 6009 m/s respectively.
Figure 10 also shows the vibration mode shapes of unit cell for quasi-SH, quasi-SV, and
quasi-L modes in X point. The arrows shown in Fig. 10 are the polarizations. In this
example, the quasi-longitudinal and quasi-transverse waves are almost indistinguishable
from the truly longitudinal and truly transverse waves of Fig. 8.


12

Fig. 9. The dispersion relations of homogeneous material quartz along the boundaries of the
irreducible part of the Brillouin zone Y

(a) (b) (c)
Fig. 10. (a) quasi shear horizontal mode (b) quasi shear vertical mode, (c) quasi longitudinal
mode in the quartz
From the discussion, it shows that the method adopted in this chapter can be used to discuss
the wave propagations in isotropic and anisotropic media.
4. Acoustic wave in inhomogeneous media
Previous studies on photonic crystals raise the exciting topic of phononic crystals. This
section presents the results of acoustic waves in inhomogeneous media, Al/Ni periodic
structures and phononic crystals with reticular geometric structures. It also discusses the
tunable band gaps in the acoustic waves of two-dimensional phononic crystals with
reticular geometric structures using the 2D and 3D finite element methods. This section

13
calculates and discusses the band gap variations of the bulk modes due to different sizes of
reticular geometric structures. Results show that adjusting the orientation of the reticular
geometric structures can increase or decrease the total elastic band gaps for mixed
polarization modes.
4.1 Periodic structure with two media
It is necessary and worthy to provide evidence supporting the FEM methods (COMSOL
Multiphysics) ability to perform Bloch calculations with two media. This chapter compares
the dispersion relations of Al/Ni band structure using the PWE method with the results of
using the FEM method. Consider a phononic structure consisting of Al circular cylinders
embedded in a background material of Ni to form a two-dimensional square lattice with
lattice spacing R. Figure 11 shows the dispersion relations along the boundaries of the
irreducible part of the Brillouin zone in Fig. 6(a) with filling ratio 0.6. The vertical axis
represents the normalized frequency
*
/
t
R C = and the horizontal axis represents the
reduced wave number
*
/ k kR = . Here, C
t
and k are the shear velocity of Ni and the wave
vector along the Brillouin zone, respectively. The Youngs modulus E, Poissons ratio ,
and density of the material Ni utilized in this example are E=214 GPa, =0.336, and
=8905 kg/m
3
.
The diamond symbols represent the dispersion relations of the transverse polarization
modes (shear vertical modes), and the cross symbols represent the mixed polarization
modes (shear horizontal mode coupled with longitudinal mode) in the PWE method. The
open circles represent the dispersion relations of all modes in the FEM method with a 3D
model. The results of the FEM method match well with those of the PWE method. In the
similar cases, when the differences of mass densities and elastic constants between the two
periodic materials are larger, the convergence of the PWE method is slower and costs more
CPU time.

Fig. 11. Comparison of Bloch calculations between the PWE and FEM methods


14
As the elastic waves propagate along the x axis, the nonvanishing displacement fields of the
shear horizontal mode, shear vertical mode, and longitudinal mode are u
y
, u
z
, and u
x

respectively. For the sequence modes appear, the modes are always the same. When
representing the whole wave vector space by the first Brillouin zone alone, they appear as
further branches from higher Brillouin zones. In this example, the phase velocities of the SV
0

mode (diamond symbols) are larger than those of the SH
0
mode. The boundary of the
Brillouin zone X-M of Fig. 11 represents the dispersion of the bulk waves with propagating
direction varied 0 deg~ 45 deg counterclockwise away from the x direction.
4.2 Periodic structure with single medium
Figure 12(a) depicts a two-dimensional phononic crystal with the reticular geometric
structures of square lattice. These reticular structures are parallel to the z-axis. In a perfect
two-dimensional phononic crystal, the periodic structure is constant in the z direction and
the size of the structure is infinite in the x and y directions. To analyze the dispersion
relations of all bulk acoustic modes in this band structure, the FEM should consider the 3D
model in Fig. 12(c). The dimensions of the unit cell in Fig. 12(a) are c=d=0.8R and R=h=1 in
the calculations.

Fig. 12. (a) square lattice with lattice spacing R and (b) rectangular lattice with lattice spacing
aR along x-axis and R along y-axis, (c) a unit cell with reticular structures in a 3D FEM
model
The material of the reticular structures in the unit cell in this chapter is aluminum. Figure
12(c) shows a diagram of the unit square lattice in a 3D FEM model. The periodicity of
phononic crystals along the z direction is used to calculate the dispersion relations of the
mixed and transverse polarization modes. The types of finite elements used for the 2D and
3D cases are the default element types, Lagrange-Quadratic, in COMSOL Multiphysics.
Figure 13 shows the dispersion relations of the mixed and transverse polarization modes
along the boundaries of the irreducible part of the Brillouin zone in Fig. 6(b) with the scales
R=h=1, c=0.8, and a=1.2. The horizontal axis represents the reduced wave number along
Y and the vertical axis represents the frequency (Hz). Note that this band
structure shows no full band gap of the mixed and transverse polarization modes. Adopting
the 2D FEM model to discuss the mixed polarization modes in this kind of band structure
shows that there is only one full frequency band gap in Fig. 13, located at 3311 ~ 3400 Hz.
Figure 13 compares the 3D and 2D FEM models. Open circles represent the dispersion

15
relations of mixed polarization modes in the 2D FEM model, while solid circles represent
the results of all bulk modes in the 3D FEM model. Figure 14 shows the eigenmode shapes
with 44 supercell of total displacements for M
1
and M
2
modes indicated in Fig. 13. These
figures clearly show the phenomena of wave localizations in this reticular geometric
structure. Note that the FEM method can easily describe the mode characteristics. In this
chapter, M
1
is a shear horizontal mode with mode vibrating displacement along the y
direction when the wave propagates along the x direction ( direction). Also, M
2
is a
shear vertical mode with mode vibrating displacement along z direction, and it does not
couple with the mixed polarization modes.

Fig. 13. The dispersion relations of the mixed and transverse polarization modes along the
boundaries of the irreducible part of the Brillouin zone with the scales R=h=1, c=0.8, and
a=1.2

M
1
M
2

Fig. 14. The eigenmode shapes with 44 supercell of total displacements for M
1
and M
2

modes indicated in Fig. 13


16
The following discussion addresses several parameters of the reticular geometric in this
chapter. First, the effect of filling fraction is discussed when the parameters c=d varied from
0.1 to 0.9 in Fig. 12(a). Figure 15 shows the distribution of the total band gaps of mixed
polarization modes, in which only one total band gap appears at approximately 3560 ~ 3736
Hz in c=d=0.8. The horizontal axis represents the parameter c, and the vertical axis
represents frequency (Hz). Figure 15 also shows the 2D diagrams of the reticular geometric
structures with c=d=0.1, 0.5, and 0.8.

Fig. 15. The band gap width with parameters c=d varying from 0.1 to 0.9 when the vertical
range is selected from 3500 to 4500 Hz
On the other hand, the scale a in Fig. 12(b) varies from 0.1 to 2.0 along the x direction and
the width of the unit cell along y direction remains 1.0 in the Bloch calculations. Changing
the scale a from 0.1 to 2.0 can tune the full frequency band gaps of mixed polarization
modes. Using detailed calculations of dispersion relations of reticular geometric structures
with scale a=0.1 to 2.0, Fig. 16 shows the band gap widths with the scale a from 0.1 to 2.0
when the vertical range ranges from 2400 to 5200 Hz. The horizontal axis ranges from 0 to
2.0, and the vertical axis represents frequency (Hz). No full frequency band gap exists when
the scale a are 0.1, 0.2, 0.3, 0.4, 0.6, 0.7, 1.5, 1.6, and 1.7. These results clearly show that
changing the scale a can increase or decrease the full frequency band gap.
It is noted that the unit cells with a=0.5 and 2.0 are the same in the Bloch calculations.
However, the dispersion phenomena is similar except for the scalar of the eigenmode
frequencies in the vertical axis of dispersion relations. In both cases, there is only one total
band gap of the mixed polarization modes. The location of the band gap ranges from
approximately 5009 to 5017.4 Hz with a=0.5, while that for a=2.0 ranges from approximately
2504.5 to 2508.7 Hz.

17

Fig. 16. The band gap widths with the scale a from 0.1 to 2.0
Finally, the rotating angles of reticular geometric structures were changed to analyze the
distribution of total band gaps. Figure 17 shows the 2D diagrams of unit rectangular
lattices in different rotating angles D=30 deg, 45 deg, 75 deg, and 90 deg. In these cases,
the widths of aluminum remain constant, 0.14R, in the reticular geometric structures with
different rotating angles in the calculations. Figure 18 shows the band gap widths of
rectangular lattices with different rotating angles of reticular geometric structures. Based
on the symmetry of the geometry, the different angles in the Bloch calculations were
adopted from 15 deg ~ 90 deg. In the calculated results, no band gap is detected from D=5
deg to 65 deg.

Fig. 17. 2D diagrams of unit rectangular lattices in different rotating angles D=30 deg, 45
deg, 75 deg, and 90 deg


18

Fig. 18. The band gap widths of the rectangular lattices with different rotating angles of
reticular geometric structures
5. Conclusion
This chapter examines and discusses the acoustic waves in homogeneous medium and
inhomogeneous medium, periodic structures with two media and one medium with
geometrical periodicity. The wave velocities of shear and longitudinal modes in an isotropic
material and those of quasi-SV, quasi-SH, and quasi-L modes in an anisotropic material are
obtained using the finite element method. This method also discusses the tunable frequency
band gaps of bulk acoustic waves in two-dimensional phononic crystals with reticular
geometric structures using the 2D and 3D finite element methods. This study adopts the
finite element method to calculate dispersion relations, avoiding the numerical errors, Gibbs
phenomenon, from the PWE method. Results show that changing the filling fraction, scale a,
and the rotating angles of unit lattices in the reticular geometric structures can increase or
decrease the elastic/acoustic band gaps. The effect discussed in this chapter can be utilized
to enlarge the phononic band gap frequency and may enable the study of the frequency
band gaps of elastic/acoustic modes in special phononic band structures.
6. Acknowledgment
The authors thank the National Science Council (NSC 97-2218-E-150-006, 98-2221-E-150-026,
and 99-2628-E-150-001) of Taiwan for financial support.
7. References
Axmann, W. & Kuchment, P. (1999). An efficient finite element method for computing
spectra of photonic and acoustic band-gap materials. J. Comput. Phys. Vol. 150, pp.
468, ISSN 0021-9991

19
Burger, M. S., Osher, J., & Yablonovitch, E. (2004). Inverse Problem Techniques for the
Design of Photonic Crystals. IEICE Trans. Electron, E87C, 258-265.
Checoury, X. & Lourtioz, J. M. (2006). Wavelet method for computing band diagrams of 2D
photonic crystals. Optics Communications Vol. 59, pp. 360, ISSN 0030-4018
Chiang, P. J., Yu, C. P., & Chang, H. C. (2007). Analysis of two-dimensional photonic crystals
using a multidomain pseudospectral method. Phys. Rev. E Vol. 75, pp. 026703,
ISSN 1539-3755
Dobson, D. C. (1999). An efficient method for band structure calculations in 2D photonic
crystals. J. Comput. Phys. Vol. 149, pp. 363, ISSN 0021-9991
Garica-Pablos, D., Sigalas, M., Montero de Espinosa, F. R., Torres, M., Kafesaki, M., and
Garcia, N. (2000). Theory and Experiments on Elastic Band gaps. Phys. Rev. Lett.
Vol. 84, pp. 4349, ISSN 0031-9007
Huang, G. L. & Sun, C. T. (2010). Band Gaps in a Multiresonator Acoustic Metamaterial.
ASME J. Vib. Acoust. Vol. 132, pp. 031003. ISSN 1048-9002
Huang, Z. G. & Chen, Z. Y. (2011). Acoustic Waves in Two-dimensional Phononic Crystals
with Reticular Geometric Structures. ASME J. Vib. Acoust. Vol. 133(3), pp.031011,
ISSN 1048-9002
Huang, Z. G. & Wu, T.-T. (2005). Temperature effects on bandgaps of surface and bulk
acoustic waves in two-dimensional phononic crystals. IEEE Trans. Ultrason.
Ferroelectr. Freq. Control Vol. 52, pp. 365, ISSN 0885-3010
Hussein, M. I. (2009). Reduced Bloch mode expansion for periodic media band structure
calculations. Proceedings of the Royal Society A Vol. 465, pp. 2825-2848, ISSN 1364
5021
Joannopoulos, J. D., Meade, R. D. & Winn, J. N. (1995) . Photonic Crystals: Molding the flow
of light, ISBN: 978-0691124568, Princeton University Press, Princeton, NJ.
Johnson, S. G. & Joannopoulos, J. D. (2001). Block-iterative frequency-domain methods for
Maxwells equations in a planewave basis. Optics Express Vol. 8, pp. 173, ISSN
1094-4087
Johnson, S. G. & Joannopoulos, J. D. (2003). PHOTONIC CRYSTALS: The road from theory
to practice, ISBN 978-0792376095, Kluwer academic publishers, Boston.
Jun, S., Cho, Y. S., & Im, S. (2003). Moving least-square method for the band-structure
calculation of 2D photonic crystals. Optics Express Vol. 11, pp. 541, ISSN 1094-4087
Kafesaki, M. & Economou, E. N. (1999). Multiple-scattering theory for three-dimensional
periodic acoustic composites. Phys. Rev. B Vol. 60, pp. 11993, ISSN 1098-0121
Kittel, C. (1996). Introduction to Solid State Physics, ISBN 978-0-471-41526-8, 7th ed., John
Wiley & Sons. Inc.
Kushwaha, M. S., Halevi, P., Dobrzynski, L. & Djafari-Rouhani, B. (1993). Acoustic Band
Structure of Periodic Elastic Composites. Phys. Rev. Lett. Vol. 71, pp. 2022, ISSN
0031-9007
Laude, V., Wilm, M., Benchabane, S., Khelif, A. (2005). Full band gap for surface acoustic
waves in a piezoelectric phononic crystal. Phys. Rev. E Vol. 71, pp. 036607, ISSN
1539-3755
Leung, K. M. & Liu, Y. F. (1990). Full vector wave calculation of photonic band structures in
face-centered-cubic dielectric media. Phys. Rev. Lett. Vol. 65, pp. 2646, ISSN 0031-
9007


20
Leung, K. M. & Qiu, Y. (1993). Multiple-scattering calculation of the two-dimensional
photonic band structure. Phys. Rev. B Vol. 48, pp. 7767, ISSN 1098-0121
Moreno, E., Erni, D., & Hafner, C. (2002). Band structure computations of metallic photonic
crystals with the multiple multipole method. Phys. Rev. B Vol. 65, pp. 155120, ISSN
1098-0121
Pendry, J. B. & MacKinnon, A. (1992). Calculation of photon dispersion relations. Phys. Rev.
Lett. Vol. 69, pp. 2772, ISSN 0031-9007
Psarobas, I. E. & Stefanou, N. (2000). Scattering of elastic waves by periodic arrays of
spherical bodies. Phys. Rev. B Vol. 62, pp. 278, ISSN 1098-0121
Salehian, A. & Inman, D. J. (2010). Micropolar Continuous Modeling and Frequency
Response Validation of a Lattice Structure. ASME J. Vib. Acoust. Vol. 132,
pp.011010, ISSN 1048-9002
Sun, J. H. & Wu, T.-T. (2005). Analyses of mode coupling in joined parallel phononic crystal
waveguides. Phys. Rev. B Vol. 71, pp. 174303, ISSN 1098-0121
Tanaka, Y. & Tamura, S. (1998). Surface acoustic waves in two-dimensional periodic elastic
structures. Phys. Rev. B Vol. 58, pp. 7958, ISSN 1098-0121
Tanaka, Y., Tomoyasu, Y., & Tamura, S. I. (2000). Band structure of acoustic waves in
phononic lattices: Two-dimensional composites with large acoustic mismatch.
Phys. Rev. B Vol.62, no. 11, 73877392.
Wang, X., Zhang, X. G., Yu, Q. & Harmon, B. N. (1993). Multiple-scattering theory for
electromagnetic waves. Phys. Rev. B Vol. 47, pp. 4161, ISSN 1098-0121
Wu, T.-T. & Huang, Z. G. (2004). Level repulsion of bulk acoustic waves in composite
materials. Phys. Rev. B Vol. 70, pp. 214304, ISSN 1098-0121
Wu, T.-T., Huang, Z. G., & Lin, S. (2004). Surface and bulk acoustic waves in two-
dimensional phononic crystals consisting of materials with general anisotropy.
Phys. Rev. B Vol. 69, pp. 094301, ISSN 1098-0121
Wu, T.-T., Huang, Z. G., Tsai, T. C. & Wu, T. C. (2008). Evidence of complete band gap and
resonances in a plate with periodic stubbed surface. Applied Physics Letters Vol.
93, pp. 111902, ISSN 0003-6951
Yan, Z. Z. & Wang, Y. S. (2006). Wavelet-based method for calculating elastic band gaps of
two-dimensional phononic crystals. Phys. Rev. B Vol. 74, pp. 224303, ISSN 1098-
0121
Yang, H. Y. D. (1996). Finite difference analysis of 2-D photonic crystals. IEEE Trans.
Microwave Theory Tech. Vol. 44, pp. 2688, ISSN 0018-9480
2
Topological Singularities in Acoustic Fields
due to Absorption of a Crystal
V. I. Alshits
1,2
, V. N. Lyubimov
1
and A. Radowicz
3

1
A.V. Shubnikov Institute of Crystallography, Russian Academy of Sciences, Moscow,
2
Polish-Japanese Institute of Information Technology, Warsaw,
3
Kielce University of Technology, Kielce
1
Russia
2,3
Poland
1. Introduction
The influence of energy dissipation on the properties of bulk elastic waves in crystals is not
at all reduced to trivial decrease in their amplitudes along propagation. In anisotropic media
the situation is much more complicated than it looks like at first glance, at least for such
specific directions of propagation as acoustic axes. The latter are defined as directions
0
m
along which a degeneracy of the phase speeds of two isonormal waves occurs (Fedorov,
1968; Khatkevich, 1962a, 1964). The corresponding points of the contact of the degenerate
sheets of the phase velocity surface P may be tangent or conical (Alshits & Lothe, 1979;
Alshits, Sarychev & Shuvalov, 1985) (Fig.1).

1
v
2
v
3
v
c
0
m
t
0
m

Fig. 1. Schematic plot of the section fragment of the three sheets of the phase velocity surface
( )
v m ( = 1, 2, 3) containing one tangent and two conical points of degeneracy

Taking into account that formally the wave attenuation may be described as an imaginary
perturbation of the phase speed, one could expect due to the damping either a shift or a split
of the acoustic axis, of course if it is not created by a symmetry. As we shall see below, for an
acoustic axis of general position it is just splitting what is realized, and with quite a radical
transformation of the local geometry of the phase velocity surface. The other possible reason
for sensitivity of the wave properties to a small attenuation is related to a polarization
aspect. Indeed, it is known (Alshits & Lothe, 1979; Alshits, Sarychev & Shuvalov, 1985) that
the acoustic axes indicate on the unit sphere of propagation directions
2
1 = m the singular
points in the vector fields of polarizations which are characterized by the definite vector


22
rotation around these points on 2 or , i.e. by the Poincar indices n= 1 or 1/2 (Fig.2).
It is clear that a split of such singular points must be quite catastrophic for the
corresponding polarization distribution. And that really occurs.

n = 1 n = -1 n = 1/2
n = -1/2

Fig. 2. Singular polarization distributions around the two types of tangent degeneracy
points n=1 and the two types of conical degeneracies n=1/2
The above peculiarities are associated with space distribution of wave characteristic rather
than with individual properties of bulk elastic waves. Meanwhile, as we shall see, in
absorptive crystals the individual wave properties close to degeneracy directions also
manifest quite unusual features, such as an almost circular polarization, in contrast to a
quasi-linear one in the non-degenerate regions.
The theory of acoustic axes in non-absorptive anisotropic media is quite complete. For a
review we address readers to the paper by Shuvalov (1998). The theory gives the general
criteria of the degeneracy occurrence and describes all possible types of acoustic axes
classifying them with respect to a local geometry of the degenerate velocity sheets and to
specific features of the vector polarization fields around the degeneracy directions. This
classification (Alshits & Lothe, 1979; Alshits, Sarychev & Shuvalov, 1985) includes more
types than we presented in Figs. 1 and 2. However, apart from a line degeneracy known in
hexagonal crystals, the rest additional types relate to the model media with accidentally
coinciding or vanishing material constants (or some their combinations). Such media are
beyond our interest in this paper. Note in addition that conical acoustic axes may exist in
real crystals even in quite non-symmetric directions and always exist along the symmetry
axis 3. In contrast, tangent degeneracies are realized in practice only due to a high symmetry
of the crystals and are known only along symmetry axes and 4. As was shown in (Alshits
& Lothe, 1979; Alshits, Sarychev & Shuvalov, 1985) , all the model acoustic axes together
with any tangent or line degeneracies are unstable and must disappear, split or be
transformed into other types under any small triclinic perturbation of the elastic moduli
tensor
c . The only stable type of acoustic axes is the conical type. Under any real
perturbation
c a conical degeneracy never split or disappear, but can only shift.

The wave attenuation can be interpreted as a perturbation of the tensor
c , however not real

but imaginary. As was mentioned in (Alshits, Sarychev & Shuvalov, 1985), under such a
non-hermitian perturbation even a conical degeneracy may lose its stability. Later Shuvalov
& Chadwick (1997) rigorously investigated the stability of different acoustic axes with
respect to a weak thermoelastic coupling. Their conclusion was: all types of degeneracies are
unstable including a conical acoustic axis which splits into a pair. The same problem for
viscoelastic and thermo-viscoelastic media has been studied by Shuvalov & Scott (1999,
2000) with similar conclusions.

Topological Singularities in Acoustic Fields due to Absorption of a Crystal

23
It is evident that the considered physical mechanisms of the damping definitely do not
disturb symmetry of the crystal and therefore cannot shift or split degeneracies along
symmetry axes , 4 and 3. It means that any really existing tangent degeneracies and
conical degeneracies along symmetry axes 3 must be stable under the damping
perturbation. This statement was proved by Alshits & Lyubimov (1998) for viscoelastic
media.
In this chapter we shall consider the attenuation in terms of viscoelasticity following to the
approach of the papers (Alshits & Lyubimov, 1998, 2011). We shall analyse in detail the
mentioned above geometrical peculiarities and polarization singularities related to a pair of
the so-called singular acoustic axes representing a new type of stable degeneracy and arising
as a result of the considered split of a conical acoustic axis. On this basis we shall develop an
extension of the classical theory of internal conical refraction (Barry & Musgrave, 1979; De
Klerk & Musgrave, 1955; Fedorov, 1968; Khatkevich, 1962b; Musgrave, 1957) for an
absorptive crystal. As will be shown, the damping provides very radical and non-trivial
modifications of fundamental features of the phenomenon.
2. Statement of the problem and general relations
Let us consider the viscoelastic medium characterized by the density and the tensors of
elastic moduli
c and viscosity
. The dynamic displacement field ( , ) t u r in such medium is

described by the known equation (Landau & Lifshitz, 1986)

i ijkl l,kj ijkl l,kj
u c u u = , (1)
where the vectors u and u are the velocity and acceleration fields and the usual notation
..., ...
k k
/ x is accepted. For the bulk wave
exp[ ( )] ( ,t) C ik - vt = u r A m r (2)
propagating along the wave vector k = k mwith the amplitude C, the frequency kv = , the
phase speed v and the polarization A, eqn. (1) is transformed into Christoffel`s equation

2

( ) Q iQ v = A A (3)
where
Q and
Q are the real symmetric matrices

i l jk ijkl
Q mc m = ,
i jk ijkl l
Q m m = . (4)
Note, that the imaginary addition Q i -
to the usual acoustic tensor
Q , in contrast to the
latter, is dependent on the frequency. Eqn. (3) determines the three complex eigenvectors
A and the three corresponding complex eigenvalues

2
v ( = 1, 2, 3), i.e. the three phase

speeds
v as functions of the direction m:

i = + A A A ,

v v iv = . (5)
Below the frequency will be supposed to be real. Hence, by (5), the value
k should contain
an imaginary addition determining the decay of the wave along its propagation:


24
1

v
k k ik i
v iv v v
| |
+ = +
|

\ .
. (6)
The complex phase speeds of eigenwaves are found from the equation

2

(Q iQ )
v

=
A A
A A
. (7)
The polarization vectors
A as eigenvectors of the symmetric matrix Q i - Q

for non-
degenerate directions m of propagation must be mutually orthogonal

= A A , . (8)
As regards to their normalization, we cannot use the customary condition
2
1
= A , bearing
in mind the possibility of a circular polarization for which
2
0
= A . Instead, the normalizing

factor will be chosen so that

2 2 2
| | 1

= + = A A A . (9)
For a further development let us divide the basic eqn. (3) on the real and imaginary parts

2 2
2 2

( ) ,

( ) .

Q Q v v 2v v
Q Q v v 2v v
+ = +
=
A A A A
A A A A
(10)
Multiplying these equations by
,
A or
,
A ( ) and combining the results one obtains

2v v Q Q , = + A A A A (11)

2 2

( ) ,

v v Q Q = + A A A A (12)

2 2 2 2

( ) ( )

v v v v Q Q 2 v v v v . + = + + A A A A A A A A (13)
Eqns. (11)-(13) are exact. The first two of them show that the imaginary part of the phase
speed
v being linear in small viscosity
is small compared to
v independently of the
direction m. In accordance with Eqn. (13), one can also conclude that

A A << however not
for any m, but only far enough from acoustic axes, when the difference

v v is not small.
In this case the value | |

A = A is also linear in
and therefore small. Let us decompose the

vector
A on the two components:

= + A A A
||
, where

A A and

A A
||
|| . Thus,
the ellipticity /

A A

= of the wave polarization due to the damping is also small almost
everywhere beyond small domains around acoustic axes. Let us estimate this ellipticity to
the first order in
.
Being perpendicular to
A , the vector
A may be expressed in leading approximation as a

superposition of two isonormal vectors
A and
A which are almost orthogonal to
A ,
( ) ( ) ( ) ( )

+ + A A A A A A A A A A A A A . (14)


25
In view of eqn. (13) this gives far from degeneracies

2 2 2 2

( ) ( )

Q Q
v v v v

= +

A A A A
A A A . (15)
In fact, for considered non-singular directions the component
A
||
is physically
unimportant. Indeed, the vector amplitude of the wave (2) in the accepted linear
approximation is equal
( ) (1 ) ( )
1

i
C C i i C i C i
i

| |
= + + = + + +
|
+
\ .
A
A A A A A A A
|| ||
||
, (16)

where the notations = /

A A
|| ||
and (1 ) C C i = +
||
are introduced.
With (15) and (9), the wave ellipticity is readily estimated as | |

A

= A . For similar
non-singular directions the speeds
v and
v determined by eqns. (11) and (12) may be

expressed in the same leading approximation as

Q
v
2v

=
A A
,
2

v Q = A A . (17)
On the other hand, eqn. (15) demonstrates the tendency to increasing ellipticity when the
wave normal m approaches the degeneracy direction (

v v = or

v v = ) and one of the
denominators in (15) decreases becoming singular. Of course, in the vicinity of the
degeneracy it is necessary to replace eqns. (15) and (17) by some other relations.
3. General formalism for the neighbourhood of an acoustic axis
In fact, eqns. (15) and (17) quite hold for the description of the non-degenerate wave branch
even along the direction where two other branches are degenerate. In the further
development we shall choose for the non-degenerate wave characteristics the number 3 = .
In this notation, by eqn. (15), the vector
3
A must be small addition to
3
A . In view of the
orthogonality condition (8) this allows us in the leading approximation to replace the
complex polarization vectors
1,2
A by their projections on the plane orthogonal to the vector
3
A . This must work even close to acoustic axes where the imaginary components of
1,2
A
might be comparable in the length with their real counterparts. We are following here the
ideology developed in the theory of acoustic axes for the case of zero damping (Alshits &
Lothe, 1979; Alshits, Sarychev & Shuvalov, 1985).
Thus, let
0
m is the direction of the acoustic axis in the crystal with the switched off
attenuation. By definition, along
0
m there must be
1 2 0
v v v = and, apart from the non-
degenerate wave with the speed
3
v and the polarization
03
A , any polarization in the
degeneracy plane
03
D A is permissible (Fig.3).
Let us choose in the D plane an arbitrary pair of unit orthogonal vectors
01
A and
02
A
forming with
03
A the orthonormal right-handed basis
01 02 03
{ , , } A A A (Fig. 3).
Now switch on the damping and consider eqn. (3) close to
0
m at
0
= + m m m:


26

2
0 0 0

( )( )( ) ( )

c i v v + + = + m m m m A A . (18)

03
A
02
A
01
A
D
0
m

Fig. 3. Allowed polarizations along the acoustic axis
0
m at switched off attenuation
In the linear approximation eqn. (18) is transformed to

2
0 0 0

( ) ( )

Q Q v 2v v + = + A A , (19)
where

0 0

( ) Q Q = m , = +
0 0 0

Q c c iQ m m m m ,
0 0

( ) Q Q = m . (20)
The complex polarization vectors
A may be decomposed in the basis

01 02 03
{ , , } A A A as

0
a = A A , (21)
where , = 1, 2, 3 and the summation over is assumed. Substituting the linear
superpositions (21) at = 1 and = 2 into eqn. (19) one obtains

2 2
1 0 1 1 0 03 13 03
2 2
2 0 2 2 0 03 23 03
2 ( ) ,
2 ( ) .
Q v v v v a
Q v v v v a
= +
= +
A A A
A A A
(22)

Eqns. (22) show that the coefficients
13
a and
23
a must be linearly small. So, indeed in the
leading approximation one can replace the polarization vectors
1
A and
2
A by their
projections on the D-plane

1 11 01 12 02
a a + A A A ,
2 21 01 22 02
a a + A A A . (23)
Multiplying eqns. (22) by
1
A or
2
A we obtain the two linear systems determining the
coefficients
a in (23):

11 0 1 11 12 12
12 11 22 0 1 12
( ) 0,
( ) 0;
Q 2v v a Q a
Q a Q 2v v a
+ =
+ =

11 0 2 21 12 22
12 21 22 0 2 22
( ) 0,
( ) 0;
Q 2v v a Q a
Q a Q 2v v a
+ =
+ =
(24)
where

0 0
1, 2
ij i j
Q Q , i, j = = A A . (25)


27
The conditions for the existence of nontrivial solutions of the systems (24) give the common
quadratic equation determining both
1
v and
2
v

2
11 0 22 0 12
( 2 )( ) 0 Q v v Q 2v v Q = (26)
with the roots determining the unknown additions
1,2
v to the degenerate speed
0
v :

1,2 0
v is R = s m . (27)
Here the notations are introduced

2 2
( ) ( ) R ip iq = + p m q m ; (28)

0
01 01 02 02 0 01 02 02 01 0
0 0
1 1

( ) , ( )
2 2
c c c c
v v
= = +
`
)
s
A A A A m q A A A A m
p
; (29)

11 22 12
0 0
4 2
s
Q Q Q
, q ;
v v
p

= =
`
)
(30)

0 0 0
ij i j
Q Q = A A . (31)
The introduced vectors
0
s , p and q have the following projections on
0
m :

0 0 0 0 0
, 0 v = = = s m p m q m . (32)
Note, that the vectors
0
s , p and q were first introduced by Fedorov (1968) in his theory of
internal conical refraction. Then the same vectors were used in the theory of acoustic axes
(Alshits, Sarychev & Shuvalov, 1985). With (27), systems (24) are easily solved which allows
us to find the polarization vectors
1,2
A (23) (not normalized at this stage):
= +
1,2 01 02
( ) ( ) iq ip R A q m A p m A . (33)
It is easily checked that
1 2
0 = A A , i.e. the orthogonality property, eqn. (8), is fulfilled.
Actually, eqns. (27) and (33) contain all necessary information for our further analysis.
However, in the next section we shall have to make preliminary step aside.
4. On the acoustic axes along directions of high symmetry
Note, that the above formalism linear in small parameters does not work for the case of
tangent acoustic axes along which p = q = 0 (Alshits, Sarychev & Shuvalov, 1985) and one
should keep the higher order terms in all expansions. The above criterion for a tangent
degeneracy can be satisfied either because of an accidental vanishing of some combinations
of material parameters (i.e. in model crystals) or due to a high symmetry of the direction
0
m . That is why tangent degeneracies are known in real crystals only along 4- and -fold
symmetry axes. In the first case the both Poincar indices n = 1 (Fig. 2) are possible, in the


28
latter case only the index n =+1 can occur (Alshits, Sarychev & Shuvalov, 1985). We already
mentioned that model media are beyond our interest in this paper. As to symmetrical
tangent acoustic axes, their reaction to switching on the damping is predictable without
any calculations. The answer is rather natural: existing due to a symmetry which is not
disturbed by the attenuation, they keep their directions and linear polarizations of the elastic
waves propagating along them also retain, though the phase speeds
v of these waves
certainly take small imaginary components.
Indeed, the tensors
c and
have completely the same symmetrical structure. It is well

known, that along the direction
0
m of the symmetry axis or 4 the tensor
0 0 0
Q c = m m has
eigenvectors
01 02
, A A and
03
A coinciding with the basis vectors of the crystallographic
coordinate system. Clearly, the tensor
0 0 0
Q = m m (20) must have the same eigenvectors.

Hence, the combined complex Christoffel tensor
0 0

Q - iQ Q - iQ = along the direction
0
m
admits purely real polarizations of three isonormal eigenwaves: one longitudinal
(
03
A parallel to z) and two transverse (
01
A parallel to x and
02
A parallel to y). It is easy to
check that the degeneracy along
0
m also retains. In accordance with eqn. (7)

2
1 1331 1331 55 55
2
2 2332 2332 44 44
2
3 3333 3333 33 33
,
,
.
v c i c i
v c i c i
v c i c i

= =
= =
= =
(34)
But the symmetry axes or 4 present only in tetragonal, cubic and hexagonal crystals where

44 55
c c = ,
44 55
= (35)
and therefore
1 2
v v = . Accordingly, the degenerate tensor
0 0

Q - iQ has the spectral
representation

0 0 44 33 44 33 33 33 03 03

[ ( )] ( ) Q iQ c c i I c i = + A A (36)
where
I denotes the unit tensor and the symbol means a dyadic product. So, one can see,
that any linear combination
01 02
+ A A is also an eigenvector of
0 0

Q - iQ and any
transverse wave may propagate along
0
m (Fig.3). Note in addition, that in the considered
situation the nominator
01 02
Q A A of the singular term in eqn. (15) vanishes together with

45
. This explains why our qualitative expectations of increasing imaginary components
1,2
A close to
0
m have not been realized.
The same arguments are equally applicable for the directions of 3-fold symmetry axes along
which the conical acoustic axes necessarily occur (Alshits, Sarychev & Shuvalov, 1985) being
characterized by the polarization singularity with the Poincar index n = 1/2 (Fig. 2).
However in this case p q 0 and the formalism developed in the previous section does
work and may be used for the demonstration of the validity of the above considerations. For
example, in the case of the 3-fold symmetry axis in trigonal crystal one has:
1331 2332 55 44
0 p = = and
1332 54
0 q = = but
1331 2332 44
2 0 s + = . As a
result, the phase speed perturbations
1,2
v , eqn. (27), have the equal imaginary components
and the difference between
1
v and
2
v remains purely real and vanishes at 0 m . Thus,


29
again there is neither a split nor a shift of the degeneracy. And in accordance with eqn. (33)
the degenerate polarization fields
1,2
A in the neighbourhood of the acoustic axis in the
leading approximation remain linear, i.e. their imaginary components are small and vanish
at 0 m .
But such a trivial situation occurs only along the symmetry axes , 4 and 3. As we shall see,
any other point of degeneracy, even in a symmetry plane, manifests instability with respect
to attenuation and singular behaviour of basic wave parameters close to new acoustic axes.
Let us consider the other symmetric case known in real crystals: the line of degeneracy
which occurs in some transversely isotropic media. According to (Alshits, Sarychev &
Shuvalov, 1985), along such line: 0 = p q i.e. the vectors p and q must be parallel or one of
them should vanish (let = q psay). Note that at 0 = p q the point degeneracy is also
possible (in model crystals, of course). In this case for its description one should keep in
expansions the terms of the higher order. But for a line degeneracy the leading
approximation used above is completely sufficient and eqns. (27) and (33) may be applied
for an analysis. The condition of the degeneracy
1 2
v v = is equivalent to the requirement
of the vanishing square root in (27) which brings us to the following system

2 2 2 2
( ) ( ) 0 p q + = p m q m , (37)
( ) ( ) 0 p q q + = p m m . (38)
At = q p this system becomes clearly contradictive,
2 2 2
( ) /(1 ) p q = + + p m ,
0 = p m ,
and has no solutions unless the both parameters p and q simultaneously vanish which
does not occur in hexagonal crystals. Thus the line degeneracy 0 = p m under the
damping perturbation must completely disappear which coincide with the corresponding
conclusion in (Shuvalov & Chadwick, 1997). But looking at eqns. (27) and (33) one can say
more. It follows from (27) that at 0 = p m the perturbations
1,2
v are purely imaginary.
This means that real components of the phase speeds
01
v and
02
v coincide as before on the
same line 0 = p m . However, the imaginary components
1,2
v are different on this line
which eliminates the degeneracy. As regards to polarizations, the only peculiarity of the
polarization field on the line 0 = p m is the lack of even a symbolic ellipticity: by (33) it is
purely linear.
5. Split of acoustic axes of general positions
At the switched off attenuation eqn. (27) transforms to the known equation (Alshits,
Sarychev & Shuvalov, 1985) describing local geometry of sheets of the phase velocity surface
P:
1,2 0 1,2 0
( ) ( ) v v v = + + m m m in the vicinity of the degeneracy point
1 0 2 0 0
( ) ( ) v v v = = m m :

2 2
1,2 0
v = + s m (p m) (q m) . (39)


30
If the vectors p and q are neither vanishing nor parallel to each other, 0 p q , then
eqn.(39) describes a conical contact of the sheets
1,2
( ) v m and simultaneously of the sheets
1,2
1 / ( ) v m of the slowness surface S. This is a conical degeneracy of a general type not
related to symmetry of a crystal.
As we know, a switching on the attenuation causes a small imaginary addition to a phase
speed of the wave: v v iv = . As a result, apart from the wave surfaces P and S the new
surface of attenuation, ( ) v m , arises. And the real components of the phase speeds
1,2
( ) v m also manifest important changes providing a topological transformation of the wave
surfaces P and S.
Let us return to eqns. (37), (38). At 0 p q this system describes the split of conical
acoustic axis due to a damping. The new positions of the degeneracies
1
m and
2
m are
given by the intersection of the ellipse

2 2
1

r r
| | | |
+ =
| |
\ . \ .
m p m q
,
2 2
r p q = + , (40)
with the straight line 0 = m M passing through the end of the vector
0
m perpendicularly to
the vector (Fig. 4)
( sin cos ) p q r = + = + M p q p q , (41)
where the angle is introduced by the expressions
sin p / r = , cos q / r = . (41a)

m
+
m
M
p
q
/q | p |
/p | p |
(b)
m
+
m
M
p
q
(a)
t

Fig. 4. Schematic plot of the ellipse, eqn. (40), in the general case (a), and for the case of the
conical acoustic axis
0
m splitting from the symmetry plane of a crystal (b)
In accordance with eqn. (32) the both vectors p and q are orthogonal to
0
m . Therefore the
ellipse (40) (Fig.4) belongs to the plane tangent to the unit sphere 1 = m m at the point
0
= m m which indicates the center of the ellipse. Thus, switching on the damping causes
the split of the conical axis
0
m into the two singular axes directed along the wave normals
0

= + m m m where


31

0
0
( )
( )
p q

+
=

m p q
m
m p q
. (42)
Note that the projections of

m (42) on p and q vectors look rather simple
, q p

= = m p m q . (43)
Let us consider the example of splitting of a conical axis belonging to the symmetry plane S
of the crystal. It is evident that in this case the polarization vector
03
A also belongs to the
plane S. The other vectors of our basis may be chosen so that, say, the vector
01
A is directed
along the normal to the plane S, and the vector
02
A belongs to the same plane S together
with the vectors
0
m and
03
A (Fig. 5a). It is easily checked that in the given case due to a
crystal symmetry, which is not less than monoclinic, there must be
=
01 01 0
0, , q q A p A m || || (44)
(Fig. 5b). By eqns. (43), (44), the split from the symmetry plane is determined by the vectors

01
|p |
= m A . (45)

02
A
0
m
S
03
A
01
A (a)
(b)
02
A
+
m
S
01
A q||
+
m
m
0
m
03
A
p

Fig. 5. Acoustic axis
0
m in symmetry plane S (a), and its splitting due to the damping (b)
For the found mutual orthogonality of the vectors p and q ellipse (40) looks very symmetric
(Fig. 4b). Thus, in the considered particular case the split of the acoustic axis occurs in the
plane orthogonal to S and the angle of splitting is proportional to the damping (Fig. 5b)
2 2 |m | |p |/q
= . (46)
6. Local geometry of the velocity surfaces in the vicinity of split axes
Let us now return to eqns. (27), (28). We shall not divide eqn. (27) on the real and imaginary
parts. It is more convenient to analyse this equation in its combined form. First of all, let us
note that the expression under square root in eqn. (28) along the line 0 = m M is purely
real, being negative between the degeneracy points (i.e. inside the ellipse, eqn. (40) and
Fig.4) and positive beyond them (i.e. outside the ellipse). But this means that on the part of
this line which is inside of the ellipse, the square root is purely imaginary. Accordingly, on


32
this part of the line the real components of phase speed
1
( ) v m and
2
( ) v m must coincide
which creates the lines of self-intersection of the wave surfaces
1,2
( ) v m and
1,2
1 / ( ) v m .
Quite similarly, we come to the conclusion that the corresponding sheets of the attenuation
surface
1,2
( ) v m must intersect each other over the line 0 = m M outside the ellipse (40).
Fig. 6 gives a schematic illustration of such self-intersection of the slowness surface.

m +
m

Fig. 6. Self-intersection of the slowness surface
1,2
1 / ( ) v m and split acoustic axes
Now let us study the above surfaces close to one of the new degeneracies, say,
0

+ +
= + m m m . We are going to find the phase speeds of isonormal waves at the contour
(Fig. 4a): ( )

+
= + m m m . The contour lies in the plane orthogonal to
0
m and its radius is
supposed to be small:
| | | |
+
= << m m . Denote
( ) ( )
= m t , (47)
where t is the unit vector making the angle with the vector p:

0
( ) [ cos ( )sin ] / p = + t p m p . (48)
Thus, by changing from 0 to 2, the vector
m (47) path-traces the contour around the

degeneracy point
+
m (Fig. 4a).
With (47), (48), eqn. (27) gives in the leading approximation over at the contour :

+
= =
1,2 1,2 0
( , ) Re[ ( )] v v v f , (49)
where
( ) 2 ( ) f i = t N M , q p = + N p q . (50)
As is seen from (49), the dependence
1,2
( ) v at 0 is characterized by an infinite
derivative over in any section
0
, where the angle
0
relates to a transition of the
vector
m through the self-intersection line,

0
Re[ ( )] 0 f = . This singularity of the function


33
1,2
( ) v at the end of the wedge of self-intersection corresponds to a sharpening tip of the
slowness surface
1,2
1 / ( ) v m and to a plane fan of the normals to this surface at the contour
when 0 (Fig. 7).

) ( ) (
1 2
m m v v =

Fig. 7. The fragment of the internal degenerate sheet ) ( 1/
2
m v of the slowness surface close
to the singular point at the end of the wedge of the self-intersection and the plane fan of
normals to the surface at this point
7. Polarization field singularities around the split acoustic axes
At the same contour (47) polarization vectors (33) up to normalizing factors are equal

1,2 01 02
( )
1
f
i
q ip
| |
= + |
|
+
\ .
A A A . (51)
For the further analysis the function ( ) f here should be concretize to the form

2
( ) cos sin cos f A g ip = + p q , (52)
where
2 A (p iq ) / p = ,
0
( ) g = m p q . (53)
It is easily checked, that at the rotation of unit vector t, (48), over the whole circuit, i.e. at
varying from 0 to 2, the function ( ) f (52), changes its sign. Indeed, the phase of the
complex function
( ) ( )exp[ ( )] f / A R i (54)
must be twice less than the phase of its square

2 2
exp(2 ) cos sin cos R i g ip = + p q . (55)
On the other hand, one can find from (55) the following relations

2
1
( ) Arctg
2 tg
p
g
| |
=
|
+
\ .
p q
,
2 2
2
cos 2
( cos sin )
gp
g
=
+ p q
. (56)


34
= 2
= 0
= 3/2
= /2
Im[f()/A]
Re[f()/A]
= 2
= 0
= /2
= 3/2
Im[f()/A]
Re[f()/A]
(a)
(b)

Fig. 8. The function ( ) f / A in the complex plane at g > 0 (a) and g < 0 (b)
This gives (see also Fig. 8)
(2 ) (0) sgn g = , (57)
(2 ) (0) f f = . (58)
Thus, after the whole turn over the contour around the degeneracy point at
+
m (Fig. 4a)
one has the identical transformation of the polarization field (51) in itself in the form

1 2
(2 ) (0) = A A ,
2 1
(2 ) (0) = A A . (59)
In other words, each of two orthogonal polarization ellipses rotates exactly on /2 being
transformed into the polarization of the isonormal wave (Fig. 9). And simultaneously the
complex velocities
1,2 1,2 1,2
v v iv = also are interchanging with their counterparts (Fig. 10).

Fig. 9. The rotation of the polarization ellipses
1,2
A in the degeneracy plane D when the
wave normal m is scanning the contour . The case g > 0 is shown when n = .
The found singularity of the polarization field at the degeneracy point
+
m (Fig. 9) may be
characterized by the Poincar index defined as the value of the total polarization rotation (in
the 2 units) at a complete path-tracing over the contour around this point. The found turn
of the polarization ellipses is equal /2, and the direction of the rotation, by eqn. (57), is
determined by the sign of the parameter g (53). Hence, one has (Alshits & Lyubimov, 1998)


35

0
1
sgn[ ( )]
4
n = m p q (60)
It is easily verified that the same relation is valid for the second degeneracy point
m .

) ( 1/
1
m v
) ( 1/
2
m v
2 1
v v =
2 1
v v =
) (
1
m v
) (
2
m v

Fig. 10. The surfaces
1,2
1 / ( ) v m and
1,2
( ) v m in the vicinity of the acoustic axis
+
m . The
transition between the sheets of the surfaces when m is scanning the contour around
+
m
Thus the physical equivalence of two pictures at = 0 and = 2 is realized not by a
coincidence of the wave characteristic inside each of the branches, as it occurs at zero
damping, but by the identity of their superpositions. This becomes topologically possible
due to such a new feature of the slowness surfaces as their self-intersections (Fig. 10). In the
absence of damping, when the degenerate wave sheets locally have the only contact point,
one of the branches along any direction is always faster than the other. And the related
polarization cross, contained of isonormal linear non-directed vectors, has non-equivalent
differently coloured crosspieces. Hence for a coincidence of such cross with itself it is
required its turn on the minimum angle , instead of /2, as in the above case (Figs. 9, 10).
The turn on /2 is sufficient only when the change of colours of crosspieces occurs during
the turn.

Fig. 11. The field of elliptic polarizations of degenerate branches in the vicinity of split axes
of an absorptive crystal for g < 0. The Poincar indices at small contours are n = -1/4, and
the combined index at the external contour is n = -1/2


36
That is why (Alshits, Sarychev & Shuvalov, 1985) in the absence of the damping a conical
axis along
0
m is characterized by the Poincar index (1 /2)sgn n g = . This is the minimal
index for a real polarization field. Its splitting into the two singularities (60) due to
switching on attenuation satisfies the index conservation law. On the other hand, the
same combined index 1/2 arises at the path-tracing of the both points
m (Fig. 11).
8. Conical refraction in absorptive crystals
Internal conical refraction of elastic waves in crystals is a good example of a non-trivial role
of anisotropy, which may create new phenomena principally impossible in isotropic media.
The energy flux P of the wave in crystal is, as a rule, non-parallel to its direction m of
propagation. For any wave normal m the direction of the Poynting vector P is determined
by the orientation of the normal n to the slowness surface. At the choice of the wave normal
along a conical acoustic axis each polarization vector in the degeneracy plane D (Fig. 3)
relates to the definite Poynting vector, i.e. to the definite normal to a cone. Rotation of the
polarization in the plane D (e.g. in a circularly polarized wave) should create a precession of
the energy flux P.
This phenomenon called the internal conical refraction was theoretically predicted and
experimentally discovered by De Klerk & Musgrave (1955). They found a circular cone of
refraction along the 3-fold symmetry axis in the cubic crystal Ni. Later on the more general
cases of the refraction cones of elliptic section were theoretically studied (Barry & Musgrave,
1979; Khatkecich, 1962b; Musgrave, 1957) and experimentally found (Aleksandrov &
Ryzhova, 1964). The complete theory of this phenomenon is presented in the monographs
(Fedorov, 1968; Sirotin & Shaskolskaya, 1983). Below we shall develop an extension of this
theory for absorptive crystals following to the recent paper (Alshits & Lyubimov, 2011).
8.1 Conical refraction in the absence of attenuation
As we have seen, in a crystal without damping along the acoustic axis
0
m , apart from the
non-degenerate wave with the polarization vector
03
A , an infinite number of elastic waves
may propagate with arbitrary polarization in the degeneracy plane D (Fig. 3). Thus in the
basis
01 02
{A , A } belonging to the same plane, for any angle the vector

01 02
( ) cos sin = + A A A (61)
determines polarization of the eigenwave propagating along
0
m with the phase speed
0
v .
Certainly, the wave with a circular polarization
01 02
i = + A A A can also propagate along the
same direction.
Consider a monochromatic plane wave propagating along the acoustic axis
0
m with the
polarization A and the phase speed
0
v :

0
( , ) C exp( ) t i = u r A ,
0 0 0
( ) k v t = m r . (62)
The Poynting vector of such wave is equal (Fedorov, 1968)

0 0
(Re ) (Re ) / c v = P u u m . (63)

For linear and circular polarizations one has, respectively,


37

0 lin
Re ( )cos C = u A , (64)

cir 01 0 02 0
Re [ cos sin ] C = u A A . (65)
In these two cases the Poynting vectors are given by different expressions:

2 2 2
0 0 lin
( cos2 sin2 )sin C = + + P s p q , (66)

2 2
cir 0 0 0
( cos2 sin2 ) C = + P s p q . (67)
Quite similarly one finds the elastic energy densities
2
(Re ) W = u :

2 2 2
0 lin
sin W C = ,
2 2
cir
W C = . (68)
With eqns. (66)-(68), the ray velocities of the considered waves are equal
W sin2 cos2 /
0 lin lin lin
q p s P s + + = = , (69)

0 0 0 cir cir cir
sin2 cos2 / q p s P s + = = W . (70)
During the period of the circularly polarized wave at a complete turn of the polarization
vector in the degeneracy plane D, the ray velocity vector
cir
s (70) twice circumscribes a cone
(Fig. 12). At that the end of the vector
cir
s twice path-traces the ellipse

0 0 0
cos2 sin2 = = + s s s p q . (71)

0 0
m v
01
A
A()
02
A

03
A
s
0
m

s
D
0
s

Q

q
p

Fig. 12. The cone of the internal conical refraction
In view of (32), the plane Q of the ellipse is orthogonal to
0
m , and the directions of path-
tracing of the vectors s and Reu
cir
are the same when g > 0 and opposite when g < 0.
For a linearly polarized wave the same refraction cone is described by the vector
lin
s (69)
when the angle changes within the interval 0 2 (Fig. 12). This particular scheme was
realized in the first experiments of De Klerk & Musgrave (1955).


38
8.2 The polarization ellipses at the ridge of the wedge of self-intersection
Consider now the wave characteristics of an absorptive crystal at the ridge of the wedge of
self-intersection of the slowness surface. For a description of the set of wave normals related
to the ridge between the two degeneracy points at the slowness surface let us introduce the
vector sin

+
= m m . At changing from -/2 to +/2 the vector
m moves through all

the ridge from one degeneracy (
-
m ) to another one ( )
+
m . Substitution
= m m into (33)
gives the polarization vectors at any point of the line of self-intersection. Making use of
relations (43) one obtains

01 02
1,2
(sin sin cos ) [cos sin (sin cos )]
2(1 sin cos )
i i

+ +
=
A A
A

, (72)
where the normalizing (9) is fulfilled and notations (41a) are used.
We remind that at the ridge of the wedge the real components of the phase speeds
1,2
v
coincide:
1 2
v v v = = . The imaginary components
1,2
v coincide only at the end points of the
ridge, = /2. In view of (6), the real components of the displacement vectors
1,2
u take the
form

1,2 1,2 1,2 1,2 1,2 1,2 1,2
Re ( , ) exp( )Re[ exp( )] exp( )

t C k i C k = u r m r A m r U . (73)
We introduced here the wave normal
m and the real phase
at the ridge,

0
= + m m m ,

k t = m r , (74)
and the dimensionless displacement vectors

1,2 1,2
Re[ exp( )]
i = U A . (75)
It is essential that in eqn. (73) a trivial damping of the wave
1,2
exp( )
k m r is separated
from the vectors
1,2
U describing much more important for us effects of attenuation.
In the considered stationary problem a choice of the time origin is certainly unessential and
may be different for isonormal waves, independent from each other. Hence, the vectors
1,2
U
as well as the polarization vectors
1,2
A are defined to the sign. Below this sign will be
chosen so that our expression would be more compact.
Note, that at scanning the ridge by the wave normal
m , the elliptic polarization

determined by eqns. (72), (75) is sharply changing. It is easily checked that this ellipticity
provides rotations of the vectors
1,2
U along the same directions corresponding to the right-
hand screw along the propagation, until sin 0 > , and to the left-hand screw, when
sin 0 < . At the ridge ends ( = /2) where the degeneracies occur, the isonormal waves,
naturally, coincide:
1 2 0
= U U U . In both cases the polarization is circular however with
different rotation signs:

0 01 02
/2
0 01 02
/2
1
( cos sin ),
2
1
( cos sin ).
2

=
=
=
= +
U A A
U A A
(76)


39
Here the angle is excluded from the arguments by the choice of the time origin.
In any other points of the ridge the polarization ellipses of isonormal waves are different. In
the middle point = 0 the isonormal waves have linear polarizations orthogonal to each other:

1 01 02 0
0
2 01 02 0
0
cos sin sin ,
4 2 4 2
sin cos sin .
4 2 4 2

=
=
| | | |
= +
`
| |
\ . \ . )
| | | |
=
`
| |
\ . \ . )
U A A
U A A
(77)
One can show that linear polarization retain on a whole line passing through the middle of
the ridge ( 0 = ) perpendicular to it (at the unit sphere 1
2
= m this line passes through
point
0
m with local orientation along vector M).
Expressions for the polarization ellipses of isonormal waves at the ridge are remarkably
simplified in the considered above particular case related to the unperturbed acoustic axis
0
m situated in the symmetry plane of the crystal. In this case 0 q = . Supposing for
definiteness that 0 p > , one can put = /2. Then, instead of (72), the polarization vectors
of the isonormal waves are equal

1 01 02
2 01 02
cos( /2) sin( /2),
sin( /2) cos( /2).
i
i
= +
= +
A A A
A A A
(78)
And the rotation of the displacement vectors
1,2
U (75) over the ellipses is now described by

1 01 02
2 01 02
cos( /2)cos sin( /2)sin ,
sin( /2)cos cos( /2)sin .

=
=
U A A
U A A
(79)
These expressions represent ellipses in a parametric form. The lengths of the horizontal and
vertical semi-axes of the first ellipse are equal |cos(/2)| and |sin(/2)|, respectively. For
the second ellipse the same length relate to the vertical and horizontal semi-axes. At the
ridge ends = /2 the above lengths of the semi-axes are equal to each other, and the
polarization becomes circular. With a displacement of the observation point
m from the
ridge ends to its middle the large semi-axes increase and the small semi-axes decrease to
zero at = 0.
Thus, both general expressions (76), (77) and the particular example (79) lead to the same
picture of polarization distribution at the ridge of wedge of self-intersection. At passing
along this line the isonormal waves, starting from a circular polarization of definite sign,
monotonously decrease their ellipticity to zero in the middle of the ridge, where ellipses are
transformed into non-directed vectors. At the second half of the ridge the ellipticity changes
its sign and monotonously increases becoming circular at the other degeneracy point. Fig. 13
illustrates this behavior of polarization at the line of self-intersection of the slowness surface.
Consider now the kinematics of the motion of the displacement vectors of isonormal waves
along the polarization ellipses. Express the radius-vectors ( 1, 2)
= U at the ellipse in polar

coordinates ( ) ,

U :

01 02
( cos sin )

U = + U A A . (80)


40
Certainly, the lengths
U of these radius-vectors at the ellipse depend on the azimuth
.
Comparing eqns. (72), (75) and (80), one has

2
1,2
(sin cos )cos2 sin cos sin2
1
1 cos
2 1 sin cos

U

+ | |
=
|
|
\ .
, (81)

1,2
cos sin (sin cos )tg
tg
sin sin cos tg

=
+
. (82)
Differentiating the latter expression with respect to time, it is easy to find the angular
velocities
1,2
of the radius-vectors
1,2
U at their ellipses:

1,2 2
1,2
sin
2

U
= , (83)
where we put
. As is seen from (83), the angular velocities differently behave in

time at different observation points at the ridge. Along the acoustic axes ( = /2), when
the isonormal ellipses coincide into one circle (76), the denominator in (83) is equal to 1, and
the circular motion has a constant angular velocity:
1 2
= = . Here the upper and lower
signs relate to different directions of the rotation at = /2 (Fig. 13).

2
=
4
=
2
=
0 =
4
=

Fig. 13. Polarization distribution for isonormal waves at the line of self-intersection of the
slowness surface
With decreasing || a non-uniformity of the motion increases and at ||<< 1 acquires a
singular character, when during the most part of the period the velocities
1,2
are very
small, and the azimuth angles
1,2
related to them are almost fixed. In this regime, the
vectors
1,2
U pass the most part of the ellipse in a short time with very high velocity. This is
clearly seen from the analytical formulae related to the discussed above particular case of
the acoustic axis splitting from the symmetry plane (Fig. 14):

=
=
1
2
tg tg( /2)tg ,
tg ctg( /2)tg ;
(84)
=
1,2
sin
1 cos cos2
. (85)


41
Eqns. (81)-(85) and Fig. 14 show that the functions
1,2
( )
and
1,2
( )
have the period

twice less than the period of the wave. This means that the half-turn of the displacement
vector over the polarization ellipse exhausts all its physically different orientations.

(a)
(b)
1
2
3
4
5

-/2 /2 0
1
2
3
/g v /
sec
2 1

1

1
2
3
4
5
/2
-/2
-/2
0
/2
1
2
3
4
5
2

-/2
/2

0
/2
-/2
1
2
3
4
5 1
2
3
-/2 0 /2
/g v /
sec
1 2

Fig. 14. Numerical plot of the azimuths
1,2
(84) (a) and the coinciding normalized speeds
sec
1,2 2,1
/ v / g (85), (106) (b) versus the phase
for the series of observation points

at the line of self-intersection of the slowness surface for a particular case of the acoustic axis
splitting from the symmetry plane. 1 - 5 =

, 2 - 15
, 3 - 45
, 4 - 75
, 5 - 90

The other kinematic characteristics of the considered motion could be the so-called sectorial
velocities defined as area circumscribed by a rotating vectors
1,2
U per unit time:

sec 2
1,2 1,2 1,2
1 1
sin
2 4
v U = = . (86)


42
The found expression is valid for unrestricted anisotropy. It is identical for the both
isonormal waves being independent of the time. However velocity (86) strongly depends on
the position of the observation point at the line of self-intersection. In particular, it
vanishes in the center of the ridge ( = 0), where the polarization becomes linear.
8.3 Universal refraction cone at the line of self-intersection and kinematics of ray
velocity precession on this cone
Let us find the ray velocities of isonormal waves (73) at the ridge of self-intersection of the
slowness surface
1,2
1/ ( ) v m . Substituting (73) into (63) one obtains the energy flux

2 2 2 2 2 2
1,2 1,2 1,2 0
| exp( )| [ ( ) ( ) 2 ]
C k F G G F FG = + + P m r s p q

, (87)
where

sin sin sin cos cos F = (88)
cos sin sin (sin cos )cos

G = +
. (89)
The energy density in the isonormal waves in the same terms is equal

2 2 2 2 2
1,2 1,2 1,2 1,2
(Re ) | exp( )| ( )
W C k F G = = + u m r

. (90)
Accordingly, the ray velocities of these waves are given by

1,2 1,2 1,2 0 1,2 1,2
/ cos2 sin2 W = = + s P s p q , (91)
where

1,2 1,2 1,2
( , , ), tg /
F G = =

. (92)
Eqn. (91) is transformed from classic expression (70) for crystals without damping after the
substitution in the latter
0 1,2
. This means that in an absorptive crystal at any point of
the ridge of self-intersection the ends of the ray velocity vectors
1,2
s move along the same
trajectories, described by the universal ellipse

1,2 1,2 1,2
cos2 sin2 = + s p q . (93)
The form of this ellipse is completely determined by the vectors p and q, and is independent
of the parameters
1,2
. In other words, it is insensitive neither to the phase
of the wave,
nor to the angles and , related to parameters of damping and to a position of the
observation point. The principal semi-axes of universal ellipse (93), coinciding with ellipse
(71) for a non-attenuating medium, are equal

( )
2 2 2 2 2 2
1,2
1
( ) 4( )
2
= + + p q p q p q . (94)
Though the vectors p and q (29) do depend on a choice of the basis
01 02
{ } A , A , one can
easily check that their combinations
2 2
+ p q and p q determining semi-axes (94) are
invariant with respect to orientation of this basis in the degeneracy plane D (Fig. 3).


43
With identical trajectories of the ray velocities precession at the whole ridge, the kinematics
of their motion is very sensitive to the position of the observation point. It may be
shown that at the ridge ends = /2 the values
0
( 2) / and
0
(- 2) / differ by only
signs:

0
( /2)
= , (95)
which gives, by (91),

= +
=
0
0
( /2) cos2 sin2 ,
(- /2) cos2 sin2 .

s s p q
s s p q
(96)
This shows that the precession of the ray velocity vector along one of split acoustic axes is
identical with the analogous process for a circularly polarized wave (70) propagating along
the unsplit acoustic axis in the crystal without damping. Directions of rotation of the ray
velocities s(/2) (96) have different signs. It is easily checked that at g > 0 they coincide with
corresponding directions of circular polarization (76), and at g < 0 are opposite to them.
In spite of the found identity of cones (70) and (96), there is an important difference between
the related to them pictures of conical refraction. In the crystal without damping the ray
velocities forming the refraction cone are directed along the appropriate normals to the
slowness surface at the conical point of degeneracy. And the normals to the analogous
surface in the vicinity of one of the split axes, as we have seen (Fig. 7), form a plane fan,
which has nothing to do with a cone of ray velocities (96) (Fig. 12).
With passing of the observation point from the end of the ridge to its center, the motion of
the ray velocity around universal cone (96) becomes less and less uniform. At the center
point = 0 the motion deceases at all: the isonormal vectors
1,2
s are frozen at definite
positions. Indeed, at = 0 eqns. (88), (89) and (92) give the values
1,2
independent of time:

1,2
2 4

= . (97)
The corresponding fixed vectors of ray velocity are equal

1,2 0 0
(0) ( sin cos ) / r = + = s s p q s M . (98)
As it would be expected, this result relates to expression (77) for a linear polarization of
isonormal waves in the same way, as eqn. (69) to expression (64) from the refraction theory
for crystals without damping. One can show that in this point ( = 0) of the ridge the two
normals to the slowness surface are parallel not to vectors (98), but to their components
belonging to the plane
0
{ , } m M orthogonal to the ridge.
One should note that the fixed in time positions of the ray velocities (98) on ellipse (93)
depend on the attenuation (the angle , eqn. (41a)), whereas the universal ellipse does not
know about . This means that points (98), generally speaking, do not coincide with the
ends of the principal semi-axes of the ellipse. Of course, in more symmetric situations the
coincidence may occur, as it happens in the case of the splitting of the acoustic axis from a
symmetry plane. What is more important that the vectors
1,2 1,2 0
(0) - = s s s , as it follows


44
from eqns. (98), (41) and Fig. 4, remain universally orthogonal to the ridge of the wedge at
any changes of the angle .
It is evident that at any small deviation of from zero the fixed vectors (98) acquire some
increments dependent on the phase
. This will renew a motion of the ray velocities

1,2
s
over the cone. However, if not to pass far from the middle point = 0, the most part of the
period the vectors
1,2
s will retain orientations close to directions (98). And the time-
averaged vectors
1,2
s in these points will be close to directions (98). This means that in the
middle domain of the self-intersection line of the slowness surface, the refraction will have
rather a wedge than a cone character.
In the considered above particular case of the acoustic axis splitting from the symmetry
plane, one can put /2 = which remarkably simplifies expressions (88) and (89), and
together with them also the formulae for angle parameters
1,2
(92):

1
tg ctg( /2)tg
= , =
2
tg tg( /2)tg
. (99)
The discussed problem of kinematics of the precession of ray velocities at the line of self-
intersection of slowness surface may be quantitatively described. Introduce the polar
coordinates (
1,2 1,2
, S ) of the positions of the ends of the radius-vectors
1,2
s at ellipses (93):

0
1,2 1,2 1,2 1,2
cos sin S
p p

| |
= +
|
\ .
m p p
s . (100)
Comparing (100) with (93) one obtains

2 2 2 2
2 2
1,2 1,2 1,2 2 2
2 G
( sin cos )
q F p G
S
F G
+
= =
+
p q
q p

, (101)

2
1,2 1,2
ctg ( - ctg ) / p g = p q . (102)
Differentiating the latter equation gives the angular velocities

1,2 1,2 2
1,2

g
S
=

. (103)
Here the derivatives
1,2
are found from (92):

1,2 1,2 2 2 2
1,2
sin
2 ( /2)
( /2)

FG FG
F G U

= = = +
+ +

, (104)
where
2
1,2
U and
1,2
are given by functions (81) and (83) shifted in phase: /2

+ .
The sectorial velocities
1,2
sec
v of the motion of the vectors
1,2
s over universal ellipse (93) are
found in analogy with eqn. (86):

2
1,2 1,2 1,2 1,2 1,2
1 1
( /2)
2 2
sec
v S g g = = = +

. (105)


45
Thus, this velocity differs from the angular velocity of the polarization, eqn. (83), only by the
dimensional factor g and by the retardation /2. Substituting into (105) the angular velocity
1,2
(85) for the considered above symmetric example, one obtains the more compact form
for the sectorial velocity:

1,2 2,1
sin
( ) ( )
1 cos cos2
sec

g
v g
. (106)
Here it is bearing in mind that the phase shift of the velocity
1,2
in simplified variant (85) is
equivalent to the transition at the counterpart branch:
1,2 2,1
( /2) ( )

+ = . The found
relation (106) allowed us to use in Fig. 14 the same curves for a characterization of both
angular velocities of polarization and sectorial velocities of ray speeds. The shown
dependencies adequately reflect the discussed above properties of the ray velocity
precession at the line of self-intersection of the slowness surface. Angle velocities (103)
behave in a similar way, especially in the region of small . With closing to acoustic axes =
/2, variations of angular velocities in time are smoothing, but retain finite until p q, in
contrast to the velocities
1,2
sec
v , which tend to constant at these limits.
9. Conclusions
Thus, we have found that specific features of the influence of attenuation on the basic wave
properties are associated with two main qualities of the damping: i) it does not disturb the
symmetry of a crystal, and ii) formally, it provides an imaginary, i.e. non-hermitian,
perturbation of the acoustic tensor. Due to the first quality there is almost no influence of the
damping on the acoustic axes which exist due to symmetry of the crystal (tangent
degeneracies along and 4-fold symmetry axes and conical degeneracies along 3-fold axes).
On the other hand, the conical acoustic axes of any other orientations manifest instability
with respect to an imaginary perturbation of the acoustic tensor. They split into pairs of
degeneracies of new type (the so-called singular acoustic axes), which never occurs without
damping. In the neighbourhood of split acoustic axes the polarization of elastic waves
proves to be strongly elliptical becoming almost circular close to the degeneracy points. A
rotation of the polarization ellipses around those points is described by the Poincar index n
= 1/4. The slowness surface acquires lines of self-intersection connecting the split singular
acoustic axes. Only the end points of these lines correspond to true degeneracies where the
imaginary components of phase speeds of isonormal waves also coincide. The latter
coincidence also occurs on the whole equi-damping lines at the attenuation surface. These
self-intersection lines at the two different surfaces (Fig. 10) after their projection on the unit
sphere
2
1 = m of propagation directions continue each other at the degeneracy points.
Topological transformations of wave surfaces and polarization fields create new features of the
phenomenon of internal conical refraction. Still an extension of the theory may be done in
terms of the same classic refraction cone bounded by the universal ellipse. As we have seen, in
crystals without damping the classic picture of conical refraction automatically arises for a
circularly polarized wave propagating along conical acoustic axis. In an absorptive crystal the
same cone and universal ellipse as a trajectory of precession of the ray velocity vectors retain at
the whole self-intersection line of the slowness surface between split degeneracy points.


46
Along singular axes the refraction does not differ from the classical picture: the isonormal
waves degenerate into one circularly polarized wave with the ray precession of constant
sectorial velocity
sec
v g = at the ellipse. A screen illumination related to such precession
would look as a completely drawn ellipse (Fig. 15a). Some increase of intensity in the
vicinity of large semi-axes
max
( ) S is explained by a slower motion of the vector
0
s in this
region (its linear speed at the ellipse is equal
sec
2v / S ). When the observation point
passes along the ridge of the wedge to its middle, both the precession of the vectors
1,2
s and
the illumination pattern become less and less uniform (Fig. 15b,c). And in the center ( =
0) only two points (Fig. 15d) will turn to be illuminated. They relate to the isonormal
waves with linear polarization: the refraction becomes purely wedge-like. Thus, with
scanning the ridge by the wave normal m the refraction continuously transforms from
purely conic to purely wedge type.

p
q
() (b) (c) (d)

Fig. 15. Schematic plot of gradual transition from purely conical refraction (a) along a
singular acoustic axis to purely wedge refraction in the middle of the ridge (d) for a
particular case of the acoustic axis splitting from the symmetry plane
In conclusion, let us discuss the observability of the above beautiful and nontrivial physical
effects. In principal, there is no threshold level of damping for the split of acoustic axes. Just
the less damping, the less is the solid angle inside of which all the peculiarities manifest
themselves. If this angle is less than the angle of the acoustic beam divergence, then we shall
not observe neither splitting of acoustic axes, nor any accompanied effects. Thus, for the
observability of our predictions the split angle (46) must exceed the divergence of the
beam. The best experimentally realizable collimation of sound beams is limited by the
diffraction divergence, which is estimated as ~/d, where is the wave length and d is the
diameter of the beam. So, with increasing frequency the angle increases and the
beam divergence, on the contrary, decreases. Thus, we deal here with a frequency
threshold from below. The order of the splitting angle is determined by the estimate
2
0
m ~ / = (46), where is the shear modulus and is the viscousity. Substituting
this estimate to the inequality / d > , one obtains the following lower threshold for
the frequency = /2

s
th
2
c
~
d
> , (107)


47
where
s
c is the sound speed. There are known a series of physical mechanisms of the sound
attenuation . Often phonons play in it an important role. The phonon viscosity at room (or
high) temperature T may be estimated as

3
B ph ph
(3 ) ~ k T / a . (108)
Here
ph
is the phonon relaxation time,
B
k is the Boltzmann constant, and a is the lattice
parameter. Substituting into eqns. (107), (108)
s
c ~
5
3 10 cm/s, ~
11
10
2
dyn /cm , d ~ 0.5
cm, T 300 K, a ~
-8
3 10 cm,
ph
~
-9
10 s, we come to the estimate
th
~100 MHz. Thus, at
rather high-frequencies, which however belong to experimentally available ultrasound
range, the properties and effects described in this chapter appear quite observable.
10. Acknowledgment
Authors are grateful to A.L. Shuvalov for helpful discussions and to W. Gerulski for the help
in computations related to the illustrations. The support of the Polish Foundation MNiSW
(grant No N N 501252334) is gratefully acknowledged. V.I.A. is also grateful to the Kielce
University of Technology for a hospitality and support.
11. References
Aleksandrov, K.S. & Ryzhova, T.V. (1964). Internal conical refraction of elastic waves in
ammonium dihydrogen phosphate. Kristallografiya, Vol. 9, No. 3 (June 1964) 373-
376, ISSN 0023-4761 [Sov. Phys. Crystallography, Vol. 9, No. 3 (1964) 298-300, ISSN
1063-7745]
Alshits, V.I. & Lothe, J. (1979). Elastic waves in triclinic crystals. Kristallografiya, Vol. 24, No.
4, 6 (Aug., Dec. 1979) 972-993, 1122-1130, ISSN 0023-4761 [Sov. Phys. Crystallography,
Vol. 24, No. 4, 6 (1979) 387-398, 644-648, ISSN 1063-7745]
Alshits, V.I. & Lyubimov, V.N. (1998). Elastic waves in absorptive media: peculiarities of
wave surfaces and singularities in the polarization fields. In: Dissipation in Physical
Systems, A. Radowicz (Ed.), pp. 15-43, Politechnika Swietokrzyska, ISSN 0239-4979,
Kielce. Proceedings of 2
nd
Workshop on Dissipation in Physical Systems, Borkow,
Poland, September 1-3, 1997
Alshits, V.I. & Lyubimov, V.N. (2011). Conical refraction of elastic waves in absorptive
crystals. Zh. Eksp. Teor. Fiz., Vol. 140, No. 2(8) (Aug. 2011) [JETP, Vol. 113, No. 2
(2011), ISSN 1063-7761]
Alshits, V.I.; Sarychev, A.V. & Shuvalov, A.L. (1985). Classification of degeneracies and
analysis of their stability in the theory of elastic waves in crystals. Zh. Eksp. Teor.
Fiz., Vol. 89, No. 3(9) (Sept. 1985) 922-938, ISSN 0044-4510 [Sov. Phys. JETP, Vol. 62,
No. 3 (1985) 531-539, ISSN 1063-7761]
Barry P.A. & Musgrave, M.J.P. (1979). On elliptical conical refraction of elastic waves in
tetragonal crystals. Q. J. Mech. & Appl. Math., Vol. 32, No. 3 (March 1979) 205-214,
ISSN 0033-5614
De Klerk, J. & Musgrave, M.J.P. (1955). Internal conical refraction of transverse elastic waves
in a cubic crystal. Proc. Phys. Soc. Lond. B, Vol. 68, No. 2 (Feb. 1955) 81-88, ISSN
1088-0370


48
Fedorov, F.I. (1968). Theory of Elastic Waves in Crystals, Plenum Press, ISBN, New York
Khatkievich, A.G. (1962a). The acoustic axes in crystals. Kristallografiya, Vol. 7, No. 5 (Oct.
1962) 742-747, ISSN 0023-4761 [Sov. Phys. Crystallography, Vol. 7, No. 5 (1963) 601-
604, ISSN 1063-7745]
Khatkievich, A.G. (1962b). Internal conical refraction of elastic waves. Kristallografiya, Vol. 7,
No. 6 (Dec. 1962) 916-920, ISSN 0023-4761 [Sov. Phys. Crystallography, Vol. 7, No. 6
(1963) 742-745, ISSN 1063-7745]
Khatkievich, A.G. (1964). Special directions for elastic waves in crystals. Kristallografiya, Vol.
9, No.5 (Oct. 1964) 690-695, ISSN 0023-4761 [Sov. Phys. Crystallography, Vol. 9, No. 5
(1964) 579-582, ISSN 1063-7745]
Landau L.D. & Lifshitz, E.M. (1986). Theory of elasticity. Pergamon Press, ISBN, London
Musgrave, M.J.P. (1957). On an elliptic cone of internal refraction for quasi-transverse waves
in tetragonal crystals. Acta Crystallogr., Vol. 10, No. 4 (Apr. 1957) 316-318, ISSN
Shuvalov, A.L. (1998). Topological features of polarization fields of plane acoustic waves in
anisotropic media. Proc. R. Soc. Lond. A, Vol. 454, (Nov. 1998) 2911-2947, ISSN 1471-
2946
Shuvalov, A.L. & Chadwick, P. (1997). Degeneracies in the theory of plane harmonic wave
propagation in anisotropic heat-conducting elastic media. Phil. Trans. R. Soc. Lond.
A, Vol. 355 (Jan. 1977) 156-188, ISSN 1471-2962
Shuvalov, A.L. & Scott, N.H. (1999). On the properties of homogeneous viscoelastic waves.
Q. J. Mech. Appl. Math., Vol. 52 (Sept. 1999) 405-417, ISSN 0033-5614
Shuvalov, A.L. & Scott, N.H. (2000). On singular features of acoustic wave propagation in
weakly dissipative anisotropic thermoviscoelasticity. Acta Mechanica, Vol. 140, No
1-2 (March 2000) 1-15, ISSN
Sirotin Yu.I. & Shaskolskaya, M.P. (1979). Fundamentals of Crystal Physics (in Russian),
Nauka, Moscow [(1982) translation into English, Mir, ISBN, Moscow]
0
An Operational Approach to the Acoustic
Analogy Equations
Dorel Homentcovschi and Ronald Miles
Department of Mechanical Engineering
State University of New York at Binghamton
USA
1. Introduction
Great progress has been made in the last sixty years in the study of the important problem of
noise generated by the interaction of ow with stationary or mobile bodies such as occurs
in jets, rotating blade propulsion machinery (propellers, turbofans helicopter rotors) and
last but not least in aircraft at all ranges of ight and speed. An important part of this
progress was based on a rigorous theory known as the Acoustic Analogy initiated by Sir
James Lighthill in (Lighthill, 1952) and (Lighthill, 1954). Lighthill considered a free ow, as
for example with a jet engine, and the nonstationary uctuations of the stream represented
by a distribution of quadrupole sources in the same volume. The ow parameters such
as the surface pressure and the Lighthill tensor T
ij
are assumed known from solving the
aerodynamic problem in the region of sound generation or furnished by measurements.
For the rst time, this revealed a clear distinction between Aerodynamic Theory, meant to
determine mainly the aerodynamic parameters as the lift and damping on the moving object
(and also supplying the data for the noise determination) and the Aeroacoustic Theory needed
for studying the noise produced, generally at large distances, by the ying (or moving)
objects. A primary aim of the following is to show that by using an operational calculus based
on the multidimensional Fourier Transform all the theory involved in obtaining the Ffowcs
Williams-Hawkings formula (Ffowcs Williams and Hawkings, 1969) can be performed using
only classical mathematical analysis.
Curles contribution (Curle, 1955) is a formal solution of the Acoustic Analogy which takes
stationary hard surfaces into consideration. The theory developed by Ffowcs Williams and
Hawkings (FW-H) (Ffowcs Williams and Hawkings, 1969) is valid for aeroacoustic sources in
relative motion with respect to a hard surface, as is the case in many technical applications for
example in the automotive industry or in air travel. The calculation involves quadrupole,
dipole and monopole terms. An important point is that FW-H theory, developed in
(Ffowcs Williams and Hawkings, 1969) assumed that the boundary surface coincides with the
physical body surface and is impenetrable. In both Aerodynamic and Aeroacoustic theories
the domain was the same: the innite air domain external to the moving body.
When the Aeroacoustic Theory was developed by Lighthill, Curle, Ffowcs Williams and
Hawkings there were not a lot of experimental or theoretical data to be used as input to
their aeroacoustic theoretical work. For this reason, they derived mainly qualitative results
3
2 Will-be-set-by-IN-TECH
which were quite useful in guiding many signicant acoustic experiments and in designing
low noise propulsion machinery.
The situation has changed dramatically in the last 20 years. The rapid growth in high
speed digital computer technology, the availability of turbulent ow simulation codes as
well as high quality measured uid dynamic data and advances in the theory of partial
differential equations, resulted in obtaining the needed data for many important problems of
aeroacoustics. However, the development of reliable Computational Fluid Dynamics (CFD)
methods made them also useful in the evaluation of the near-eld aerodynamic parameters.
Unfortunately, a fully CFD-based computational aeroacoustic methodology is so far too
inefcient and beyond the capability of supercomputers of today.
To avoid these computational difculties, the philosophy of approaching the Aeroacoustic
problem has changed by introducing a surface S as a "permeable" control surface. The
surface S is assumed to include inside, in the volume V, all the nonlinear ow effects and
noise sources. This splitting of the problem into a linear problem for an innite domain and
a nonlinear setting for a bounded region allows the use of the most appropriate numerical
methodology for each of them. In the bounded domain V the CFD methods or advanced
measurement techniques will be used for obtaining the aerodynamic near-eld and providing
the data on the surface S needed for the external, innite domain modelling. The analysis
of the ow information inside V is, in general, expensive either using experiments or CFD.
Therefore, it is advantageous to make the volume V as small as possible.
The FH-W equation involving a permeable surface is the proper model for determining
the far-eld pressure in the innite domain. The case of permeable surfaces was analyzed
by Ffowcs Williams in (Dowling and Ffowcs Williams, 1982) and (Crighton, et al.), by
Francescantonio in (Francescantonio, 1997) who called it the KFWH (Kirchhoff FW-H)
formula, by Pilon and Lyrintzis in (Pilon and Lyrintzis, 1997) calling it an improved Kirchhoff
method and by Brentner and Farassat in (Brentner & Farassat, 1998). Besides the accessibility
to the surface data the advantage of the methods using a permeable control surface is that
the surface integrals and the rst derivative needed can be evaluated more easily than the
volume integrals and the second derivatives necessary for the calculation of the quadrupole
terms when the traditional Acoustic Analogy is used.
The Acoustic Analogy approach and especially the theory based on the FW-H equation is
the most widely used tool for deterministic noise sources. The beauty (and the power) of this
model is that all the manipulations are completely rigorous without any ad hoc reasoning.
Besides the approach based on generalized functions, used in most papers approaching the
FW-H formula, we note the work by Goldstein (Goldstein, 1976) where all the formulas
are obtained starting with a generalized Greens formula. A similar approach was used in
(Wu and Akay, 1992) . However, the algebra involved in their construction is substantial.
In the second of the two reports by Farassat (Farassat, 1994),(Farassat, 1996), which covers
the details of the mathematics used for the wave equation with sources on a moving
surface, the author correctly claims that the Ffowcs Williams-Hawkings famous paper
published in 1969 used a level of mathematical sophistication including multidimensional
generalized functions (distributions) and differential geometry unfamiliar to most researchers
and designers working in the eld. Many people use Diracs (generalized) function starting
with its integral denition. On the other hand, to learn about more complicated operations
such as the derivative of a generalized function (distribution) we need a change of paradigm
in the way we look at ordinary functions. For some people involved in practical applications
this is not a simple task. The power of the theory of generalized functions stems from its
50 Acoustic Waves From Microdevices to Helioseismology
An Operational Approach to the Acoustic Analogy Equations 3
operational properties. Thus, for example, discontinuous solutions of linear equations using
the Greens function are easily obtained by posing the problemin generalized function spaces.
In the following we show that the theory connected with the FW-H formula can be made
much simpler, without manipulating multidimensional generalized functions, by using an
operational calculus based on the multidimensional Fourier Transform. The approach
based on using the Fourier Transform preserves all the good operational properties of
the generalized functions without the need to introduce a new sophisticated mathematical
tool. The method was used previously by us (Homentcovschi and Singler, 1999) for a direct
introduction to the Boundary Element Method.
In the case of permeable surfaces an alternate method for solving the Aeroacoustic problem
for the innite external domain is based on the Kirchhoff formula for the wave equation. Due
to its use in Aeroacoustic theory we included also in Section 6 the Kirchhoff formulation for
the solution of the wave equation in the case of mobile surfaces. The proof is based again on
3-D Fourier Transform of discontinuous functions and the nal formula includes the volume
sources and the surface sources as well. As a comparison of the two approaches (that based
on FW-K equation and that using Kirchhoffs equation) we notice that Kirchhoffs method
requires less memory because fewer quantities on the control surface needed to be stored.
On the other hand, Brentner and Farassat in (Brentner & Farassat, 1998) have shown that the
FW-H equation is superior to the Kirchhoff formula for aeroacoustic problems because it is
based on conservation laws of uid mechanics rather than on the wave equation. Thus, the
FW-Hequation is valid even if the integration surface is in the nonlinear regionbeing therefore
more robust with the choice of control surface. Another advantage of the FW-Hmethod is that
it does not require computation of the normal derivatives on the permeable surface.
A comprehensive review of the use of Kirchhoffs method in computational aeroacoustics
was given by Lyrintzis in (Lyrintzis, 1994). The same author reviewed the advances in
the use of surface integral methods in aeroacoustics, including Kirchhoffs method and
permeable Ffowcs-Williams Hawkings methods in (Lyrintzis, 2003). Morino in (Morino,
2003) addresses commonalities and differences between aeroacoustics and aerodynamics. A
discussion about the acoustic analogy and alternative theories for jet noise prediction is the
subject of (Morris and Farassat, 2002). Finally we note some interesting work about this
subject included in the book edited by (Raman, 2009).
It is our hope that the elementary derivations included in this chapter will make the
application of the FW-H equation more clear, avoiding in the future comments such as
those generated by (Zinoviev and Bies, 2004) (See (Farassat, 2005), (Zinoviev and Bies, 2005),
(Farassat and Myers, 2006), (Zinoviev and Bies, 2006)).
2. The equations of the acoustic analogy and their operational form
In order to apply the 3-D Fourier Transform to a certain physical variable this has to be
dened in the whole space. Thus, to utilize the Fourier Transform to examine the sound
eld in a nite physical domain, it is necessary to imbed this domain within an innite space.
For example, in the case of studying the air motion around a nite body, the interior of the
body is considered as an air-lled domain separated from the exterior, innite domain by
an impermeable surface, S
b
enclosing the body. This introduces a rst class of discontinuity
surfaces of the motion between two air-lled regions. The second class contains the natural
discontinuity surfaces S
s
inside the ow domain as shock fronts, wakes, etc. Finally, to aid
computations, it is often helpful to also dene a virtual permeable surface, S
p
, enclosing body
along with the portion of the air domain where viscous effects and the nonlinear terms in
51 An Operational Approach to the Acoustic Analogy Equations
the Navier-Stokes equations are important. It is convenient to evaluate the aerodynamic
eld in this region numerically using CFD, due to the difculties imposed by viscosity and
nonlinear effects. These calculations furnish data describing the hydrodynamic state on the
virtual permeable surface, S
p
. In the domain exterior to the permeable surface the acoustical
analogy is applied to predict the sound eld. In other words, the eld inside the virtual
permeable surface is assumed to be strongly inuenced by hydrodynamic effects, while the
eld in the external, innite domain can be modeled according to the usual assumptions of
linear acoustics.
2.1 The equations of the acoustic analogy
In the case where the whole space R
3
is lled by a compressible viscous uid, containing
several discontinuity surfaces, the ow inside each domain is governed by the following
equations:
the continuity equation (mass conservation)
t
+

x
j
_
u
j
_
= 0, (1)
and the conservation of momentum in the form written by Lighthill in Refs. (Lighthill, 1952)
and (Lighthill, 1954)
(u
i
)
t
+ c
2

x
i
=
T
ij
x
j
. (2)
Here is the density, c is the velocity of sound in the uniform medium, u
j
is the component of
uid velocity in the direction x
j
(j = 1, 2, 3), and a repeated index implies a summation over
these values, and,
T
ij
= P
ij
+ u
i
u
j
c
2

ij
,
is Lighthills stress tensor. Also, we denoted by
P
ij
= p
ij
+
_
u
j
x
j
u
j
x
i
+
2
3
_
u
ij
_
,
the compressive stress tensor,
ij
is the Kronicker delta function, p is the pressure and is the
viscosity.
Remark 1. The acoustical analogy equations (1) and (2) are in fact the exact uid ow equations in
the form written by Lighthill. This means that if one solves these equations correctly for a problem
satisfying the Lighthill assumptions, then one will get the correct answer to the aerodynamic and
aeroacoustic problems simultaneously.
In the case where inside the ow domain there are discontinuity surfaces of type S
s
(as
shock fronts) the conservation laws for mass and momentum yield the Rankine-Hugoniot
type junction conditions
[ (u
n
v
n
)] = 0 (3)
_
u
(u
n
v
n
) + P
j
n
j
_
= 0. (4)
We denote by square brackets the jump of its content across the discontinuity surface (see also
formula (54) in Appendix A). u
n
is the velocity projection on the normal to the surface, n, and
v
n
is the normal velocity of the surface.
Finally, on solid (deformable) impermeable surfaces S
b
inside the ow the obvious
non-penetration condition proves true
u
n
v
n
= 0. (5)
2.2 The operational form of the acoustic analogy equations
Since the density
0
and pressure p
0
at large distances (in the unperturbed uid) are different
fromzero we introduce the perturbation of the density
=
0
and the perturbation of the
stress tensor as P
j
= P
j
p
0
j
(where
j
is the Kronecker delta) and will write equations
(1) and (2) as
t
+

x
j
_
u
j
_
= 0 (6)
_
u
j
_
t
+ c
2

x
j
=
T
j
x
, (7)
where
T
j
= P
j
+ u
j
u
c
2
j
According to the previous discussion, these equations are valid within the owdomains in the
whole space. Assuming, as in Appendix A, that there is a discontinuity surface S, separating
the inner domain D
(i)
and the external domain D
(e)
, by applying the Fourier transform to
equations (6) and (7) and using the formulas (53) and (68) there results the operational equations
d

dt
+ ik
j
u
j
=
_
S
[ (u
n
v
n
)] e
iky
dS (8)
d u
j
dt
+ ik
j
c
2
= ik
j
+
_
S
_
u
j
(u
n
v
n
) + P
j
n
_
e
iky
dS (9)
Here the overhead tilde denotes a Fourier Transform (see Appendix A), k(k
1
, k
2
, k
3
) is the
wave vector and y(y
1
, y
2
, y
3
) is the position vector of the integration (source) point. The
operational form of the conservation equations contains in the left-hand sides the Fourier
transform of corresponding terms in (1) and (2) and in the right-hand sides the integrals
accounting for the inuence of the discontinuity surfaces. As was noted in Remark 1 in
Appendix A, in the general case, in the right-hand side of equations (8) and (9) the sum of
the contributions of all discontinuity surfaces will appear.
Remark 2. The square brackets in equations (8) and (9) can also be written as [
0
u
n
+
(u
n
v
n
)]
and
_
u
(u
n
v
n
) + P
j
n
j
_
respectively.
2.3 Some particular cases
2.3.1 Shock-type discontinuity surfaces (S
s
)
In the case of a discontinuity surface S
s
(of shock-type) inside the uid domain the junction
conditions (3) and (4) will cancel out the integral terms in equations (8) and (9). Consequently
the shock-type discontinuity surfaces are not introducing any supplementary terms in the
operational equations.
2.3.2 Impermeable solid deformable surfaces (S
b
)
In the case of a solid with a deformable and impermeable boundary surface the condition (5)
will cancel out the integral term in the operational form of the continuity equation (8) and a
part of the integral in the momentum equation (9). In this case, the Fourier transform of the
traction P=(P
1
, P
2
, P
3
) of the surface on the uid enters in the integral term of the momentum
equations as P
j
= P
j
n
. The operational momentum equation now contains in the right-hand

side the action of the solid surface S
b
on the uid ow.
3. Solution of the operational form of the acoustical analogy equations.
3.1 The case of a permeable surface
_
S
p
_
Let S
p
be a discontinuity surface of the ow variables inside the external uid ow such that
in the domain D
(i)
we have p
(i)
= p
0
,
(i)
=
0
, v
(i)
= 0.
We write the system of equations (8), (9) as
d w
dt
=

A w+
f (10)
where
w
T
= (
, u
1
, u
2
, u
3
), (11)
A=
0 i k
1
i k
2
i k
3
c
2
i k
1
0 0 0
c
2
i k
2
0 0 0
c
2
i k
3
0 0 0
, (12)
f
1
=
_
S
[
0
u
n
+
(u
n
v
n
)] e
iky
dS
f
j+1
= ik
T
j
+
_
S
_
u
j
(u
n
v
n
) + P
j
_
e
iky
dS, j = 1, 2, 3.
(13)
The solution of equation (10) can be obtained by using the exponential

H(t) of the matrix

A
as
w=
_
t
H
_
t t
f
_
t
_
dt
(14)
The exponential of the matrix

Acan be written as
H=
_
h
j
_
(15)
h
1,1
= cos (ckt) ,

h
1,+1
=
i k
sin(ckt)
ck
, = 1, 2, 3
h
j+1,1
=
i k
j
c sin(ckt)
k
, j = 1, 2, 3 (16)
h
j+1,+1
=
j
+
k
j
k
(cos (ckt) 1)
k
2
, j, = 1, 2, 3
where k
2
= k
2
1
+ k
2
2
+ k
2
3
= |k|
2
. By introducing the function
F(k, t) =
d
dt
_
S
_
0
u
n
+
(u
n
v
n
)
_
e
iky
dS (17)
ik
j
_
S
_
u
j
(u
n
v
n
) + P
j
_
e
iky
dS + ik
j
ik
j
the components of the vector w can be written as
(k, t) =
_
t
F(k, )
sin(ck (t ))
ck
d (18)
u
j
(k, t) =
_
t
ik
T
j
(k, )d +
_
t
d
_
S()
_
u
j
(u
n
v
n
) + P
j
_
e
iky
dS (19)
+ik
j
_
t
F(k, )
cos (ck (t )) 1
k
2
d
The method given in this section has the advantage of furnishing the integral representations
for the operational density and operational velocity as well. A simpler deduction of the
representation of the operational density is given in the next section.
3.2 The equation for the operational density
By eliminating the Fourier transform of the momentum u
j
between equations (8) and (9) the
operational equation satised by the density perturbation becomes
d
2

dt
2
+ c
2
k
2
=

F(k, t). (20)
The relationship (20) is the operational form of the nonhomogeneous wave equation. The
general solution of the homogeneous wave equation in operational form can be written as
h
= A(k) cos(c k t) + B(k) sin(c k t))
and by using Lagranges method of variation of parameters there results the same
representation formula (18) for the operational density as the solution of equation (20).
3.3 The case of an impermeable surface (S
b
)
In the case where the surface S
b
is impermeable, the condition (5) cancels out some terms in
formula (17). In this case the nonhomogeneous term is
F
b
(k, t) =
d
dt
_
S
0
u
n
e
iky
dS ik
j
_
S
P
j
e
iky
dS + ik
j
ik
j
(21)
which coincides with the nonhomogeneous term in the operational form of the FW-H
equation.
By introducing the new variables suggested by Francescantonio (Francescantonio, 1997), in
the form modied in (Brentner & Farassat, 1998)
U
j
=
_
1

0
_
v
j
+
u
j
0
(22)
L
j
= P
j
+ u
j
(u
n
v
n
) (23)
the nonhomogeneous termof the operational wave equation in the case of a permeable surface
coincides with that corresponding to an impermeable case
F(k, t) =
d
dt
_
S
0
U
n
e
iky
dS ik
j
_
S
L
j
e
iky
dS + ik
j
ik
T
j
(24)
The terms U
j
and L
j
can be interpreted respectively as a modied velocity and a modied
traction, which take into account the ow across S.
4. Determination of velocity
4.1 The lift component of velocity
The lift component of velocity is given by inverse Fourier transform of the term
_
u
j
_
L
= ik
j
ik
r
_
t
d
_
S
(y, ) n
r
(y, ) e
iky
cos (ck (t )) 1
k
2
dS
Therefore, by using formula (73) we obtain
_
u
j
_
L
=

2
x
j
x
r
_
t
d
_
S
r
H (t r/c)
4r
dS
where
r = |x y|
_
(x
1
y
1
)
2
+ (x
2
y
2
)
2
+ (x
3
y
3
)
2
.
x(x
1
, x
2
, x
3
) being the position vector of the observation point.
5. Determination of density
Since the general permeable case can be studied by means of an equivalent impermeable case
we shall determine the density in the case where the nonhomogeneous term is (21).
5.1 The case of sources on a surface
Consider the simpler case when the noise sources are on a rigidsurface having only translation
and rotation motions. Then,
F
S
(k, t) =
_
S
Q(y, t) e
iky
dS (25)
where Q(y, t) is the surface intensity. In this case we have

S
(k, t) =
_
t
d
_
S
Q(y, ) e
iky
sin(ck (t ))
ck
dS (26)
Hence,
S
(x, t) =
1
(2)
3
_
e
ikx
dk
_
t
d
_
S
Q(y, ) e
iky
sin (ck (t ))
ck
dS
=
_
t
d
_
S
Q(y, ) dS
_
sin(ck (t ))
(2)
3
ck
e
ik(xy)
dk (27)
=
_
t
d
_
S
Q(y, )
( (t r/c))
4c
2
r
dS
Here the relationship (71) and the property (t) = (t) have been used.
The only contribution in the last integral in formula (27) comes from the time
e
which is the
solution of the equation
g(, y, t, x) =0 (28)
where
g(, y, t, x) = t +
|x y|
c
(29)
In other worlds, the value of the density at the observation point x at the moment t is
determined by the noise sources at the emission (radiating) time
e
on the emission (radiating)
surface S
e
S
e
.
It is now necessay to consider the coordinate systems. Let a xed point P
0
reside on a material
surface such as an airplane wing or a blade etc. We consider two coordinate frames:
a) Aframe xed relative to the moving material surface. This is the frame used by a designer
to describe the structure geometry for purposes of fabrication. The variable will be called the
Lagrangian variable of the point P
0
and in the case supposed here (nondeformable material
surface) is independent of time.
b) A coordinate frame xed with respect to the undisturbed medium having the origin O (see
Fig.1). The position of the observation point is given by the position vector x. The position of
the point P
0
is described by the position vector y (, ). The position vectors x and y will give
the Eulerian coordinates of the observation and source terms, respectively. The formula
Fig. 1. Coordinate frame

y = +
_

cM
_
,
_
d
(30)
gives the connection between the Lagrangian and Eulerian coordinates of the point P
0
. For
xed the equation y = y (, ) gives the trajectory of the point P
0
and cM =dy/d is the
velocity of the source point with respect to the undisturbed medium (source convection
velocity). Formula (30) can be viewed as a transformation between the Lagrangian and
Eulerian coordinates of the point P
0
. The inverse transformation will be denoted by =
(y, ). In the case where the transformation involves only translations and rotations we have
det (y/) = det (/y) = 1. Fig.1 shows the observers position x at time t, the emission
(radiation) position of the material surface (S
e
S
e
), the position of the same surface at the
observation time (S
t
), the position vector y
e
at the emission time and at the observation time
y (, t), the trajectory of the point P
0
, the convection velocity of the source at the emission
moment cM
e
, the radiation vector r
e
= x y
e
, and the emission distance r
e
= |r
e
| .
To change the integration variable in the last integral in formula (27) from to g, we calculate
d g
d
= 1
1
c
r
y
= 1
r
r

v
c
= 1 M
r
(31)
where M
r
is the Mach number at the point in the radiation direction at the time .
The density perturbation can therefore be written as
S
(x, t) =
_
S
e
_
Q(y, )
4c
2
r |1 M
r
|
_
ret
dS

_
S
(
e
)
Q
(,
e
)
4c
2
r
e
|1 M
r
e
|
dS
(32)
Here, |1 M
r
e
| is the Doppler factor. The square brackets []
ret
imply that the contents are to
be evaluated at the retarded (emission or radiating time)
e
given implicitly by g () = 0.
The emission position is y
e
= y (,
e
) , the emission distance r
e
of the source point to the
observer position x is r
e
= |x y (
e
,
e
)| , and Q
(,
e
) = Q(y
e
,
e
) .
5.1.1 The thickness noise
The thickness noise is given by the term
F
thickness
(k, t) =
d
dt
_
S
0
u
n
e
iky
dS
An analysis similar with that of the section 5.1 yields
thickness
(x, t) =

t
_
S
_

0
u
n
4c
2
r |1 M
r
|
_
ret
dS
(33)
5.1.2 The loading noise
The last term in relationship (21) describes the loading noise
F
loading
(k, t) = ik
j
_
S
P
j
e
iky
dS
Its contribution to the perturbed density
is
loading
(x, t) =

x
j
_
S
_
P
j
4c
2
r |1 M
r
|
_
ret
dS
(34)
We have given here a very short presentation of formulas for thickness noise and loading
noise. A more complete presentation about these formulas and their implementation can be
found in (Farassat, 2007).
5.2 The quadrupole noise term
The last term in formula (21) corresponds to a quadrupole noise source:
q
(x, t) = F
1
_
ik
j
ik
T
j
_
=
_
t
d
_
ik
j
ik
T
j
sin(ck (t ))
(2)
3
ck
e
ikx
dk (35)
=

2
x
j
x
_
t
d
_
D
(e)
T
j
(y, ) dy
_
sin(ck (t ))
(2)
3
ck
e
ik(xy)
dk,
where D
(e)
denotes the 3-D domain occupied by volume sources at the moment . The last
integral in formula (35) was calculated in Appendix B. Introducing its expression given by
formula (71) there results
q
(x, t) =

2
x
j
x
_
t
d
_
D
(e)
T
j
(y, )
( (t r/c))
4c
2
r
dy, (36)
the relationship (36) becomes
q
(x, t) =

2
x
j
x
_
D
(e)
()
_
T
j
(y, )
4c
2
r |1 M
r
|
_
ret
d. (37)
where the effect of source convection is revealed by the Doppler factor; convection effectively
increases the source strength by |1 M
r
|
1
. Further transformations of the formula (37) useful
for its numerical implementation were made by Farassat and Brentner (Farassat and Brentner,
1988) and by Brentner in (Brentner, 1997). In the case where the discontinuity surface is
permeable (of type S
p
) this term is missing, the surface being usually chosen outside the
space containing the quadruple sources. In the general case the sum of the solutions
q
(x, t) ,
thickness
and
loading
completely specify the density eld.
6. The Kirchhoff method in Aeroacoustics
Besides the Acoustic Analogy approach for the solution of the Aeroacoustic noise, another
widely used method is based on Kirchhoffs solution of the wave equation. We start with the
nonhomogeneous wave equation for the pressure perturbation
_

1
c
2
2
t
2
_
p
(x, t) = g (x, t) (38)

where g (x, t) represents the density of pressure sources. By applying the Fourier transform
with respect to the spatial variables and using formulas (58) and (70) we obtain the operational
form of the wave equation (38)
d
2
d t
2
+ c
2
k
2
= c
2
g (k, t) +

G (k, t) (39)
where
G (k, t) =
_
S
t
_
c
2
_
p
n
_
+ v
n
_
p
t
__
e
iky
dS (40)
_
S
t
c
2
i k n
_
p
e
iky
dS
d
d t
_
S
t
v
n
_
p
e
iky
dS
Equation (39) is similar to equation (20). Consequently, its solution can be written as
(k, t) =
_
t
_
c
2
g (k, ) +

G(k, )
_
sin(c k (t ))
c k
d (41)
Hence the contribution of the nonhomogeneous term in equation (38) can be written as
p
g
(x, t) = c
2
_
t
d
_
g (k, )
sin(ck (t ))
(2)
3
c k
e
ikx
dk
=
_
t
d
_
g (y, )
(t |x y| /c)
4 |x y|
dy
=
_
g (y, t |x y| /c)
dy
4 |x y|
(42)
Finally, the contribution of the terms corresponding to the boundary conditions on the mobile
surface S can be written as
p
G
(x, t) =

x
i
_
S
t
_
p
n
i
4r |1 M
r
|
_
ret
dS

t
_
S
t
_
p
M
n
4r |1 M
r
|
_
ret
dS (43)
_
S
t
__
p
n
+ M
n
p

_
1
4r |1 M
r
|
_
ret
dS
which coincides with the relationship (5.3) given in (Ffowcs Williams and Hawkings, 1969).
7. Concluding remarks
Acoustic Analogy is one of the greatest contributions to the eld of acoustics of the previous
century. It is a major extension of acoustics made by Sir M. J. Lighthill (and other contributors)
who formulated for the rst time the science of how sound is created by uid motion. This
theory completes the previous work by famous researchers in the eld of acoustics who had
discovered how sound propagates through various media and across surrounding surfaces.
In this chapter we have attempted to simplify the application of the Acoustic Analogy by
showing how to apply it using only classical mathematical analysis tools.
8. Acknowledgment
This work has been supported by the National Institute on Deafness and Other
Communication Disorders grant R01 DC009429 to RNM. The content is solely the
responsibility of the authors and does not necessarily represent the ofcial views of the
National Institute on Deafness and Other Communication Disorders or the National Institutes
of Health.
9. Appendix A: Fourier Transform of piecewise differentiable functions
Let D
(i)
(t) be a bounded mobile domain in R
3
having a smooth boundary surface S
t
. Denote
by D
(e)
(t) the domain external to the surface S
t
(See Fig.2): D
(i)
S D
(e)
= R
3
, D
(i)
D
(e)
=
. Let also
(i)
(x, t) be a continuous differentiable function dened in the closed domain
D
(i)
R and
(e)
(x, t) a continuous differentiable function dened in D
(e)
R. We dene
also
(x, t) =
_

(i)
(x, t) , in D
(i)
(e)
(x, t) , in D
(e)
(44)
Fig. 2. Boundary Surface
Generally, the function (x, t) is discontinuous across the surface S. We call it a
piecewise differentiable function (pdf). Assuming that the function
(e)
(x, t) is decreasing
sufciently rapidly at innity (for more precise conditions about the function
(e)
(x, t) see
(Homentcovschi and Singler, 1999)) we can take the Fourier Transform of the function (x, t)
with respect to space variables
(k, t) F { (x, t)} =
_
(x, t) e
ikx
dx (45)
Here x = (x
1
,x
2
,x
3
) , k = (k
1
,k
2
,k
3
) , k x = k
1
x
1
+ k
2
x
2
+ k
3
x
3
is the inner product of the two
vectors, dx = dx
1
dx
2
dx
3
and the integral is extended over the whole R
3
space. The inversion
formula can be written as
(x, t) F
1
{ (k, t)} =
1
(2)
3
_
(k, t) e
ixk
dk (46)
where dk =dk
1
dk
2
dk
3
. Accounting for relationship (44) we can write
F { (x, t)} =
_
D
(i)
(i)
(x, t) e
ikx
dx +
_
D
(e)
(e)
(x, t) e
ikx
dx (47)
9.1 Fourier transform of the derivative with respect to a spatial variable of a piecewise
differentiable function
9.1.1 The rst basic formula
We write
F
_

x
1
_
=
_
D
(i)
(i)
(x, t)
x
1
e
ikx
dx +
_
D
(e)
(e)
(x, t)
x
1
e
ikx
dx (48)
But
_
D
(i)
(i)
x
1
e
ikx
dx =
_
D
(i)
x
1
_
(i)
e
ikx
_
dx + ik
1
_
D
(i)
(i)
(x, t) e
ikx
dx (49)
The rst integral in the right-hand side of relationship (49) can be replaced, by using the
divergence theorem by an integral over the boundary surface S
t
.
_
D
(i)
x
1
_
(i)
e
ikx
_
dx =
_
S
t
n
1
(i)
(y, t) e
iky
dS (50)
where n = (n
1
, n
2
, n
3
) is the external unit normal to S
t
. Therefore,
_
D
(i)
(i)
x
1
e
ikx
dx = ik
1
_
D
(i)
(i)
(x, t) e
ikx
dx +
_
S
t
n
1
(i)
(y, t) e
iky
dS (51)
Similarly,
_
D
(e)
(e)
x
1
e
ikx
dx = ik
1
_
D
(e)
(e)
(x, t) e
ikx
dx
_
S
t
n
1
(e)
(y, t) e
iky
dS (52)
Finally, the relationships (49), (51), and (52) give the rst basic formula:
F
_

x
1
_
= ik
1
(k, t)
_
S
n
1
[ (y, t)] e
iky
dS (53)
Here we denoted by square bracket the jump of the function (x, t) across the surface S
t
[ (y, t)] = lim
x
(e)
y
(e)
_
x
(e)
, t
_
lim
x
(i)
y
(i)
_
x
(i)
, t
_
(54)
for y S
t
, x
(i)
D
(i)
and x
(e)
D
(e)
. Similar relationships to (53) can be proved for the
derivatives in the directions x
2
and x
3
.
Remark 3. It is clear that the obtained relationships can be extended immediately to the case where
there are more discontinuity surfaces of the given function. The resulting formulas will contain sums
of integrals corresponding to each discontinuity surface.
9.1.2 Other formulas
The relationship (53) gives also
F {}

= ik (k, t)
_
S
t
n [ (y, t)] e
iky
dS (55)
In the case where we write V(x, t) = (V
1
(x, t) , V
2
(x, t) , V
3
(x, t)) where V
j
(x, t) is a piecewise
differentiable function dened by a relationship similar to (44) we can write also the formulas
F { V}
V = ik
V(k, t)
_
S
t
n [V(y, t)] e
iky
dS (56)
F {V}
V = ik
V(k, t)
_
S
t
n[V(y, t)] e
iky
dS (57)
The formulas (55), (56) and (57) permit the calculation of the Fourier Transforms of a gradient
of a scalar eld of a divergence and a curl of a vector eld in the case of piecewise differentiable
scalar and vector elds.
Moreover, we can write
F { } F { } = ik
(k, t)
_
S
t
n [ (y, t) ] e
iky
dS (58)
= ik
_
ik (k, t)
_
S
t
n [ (y, t)] e
iky
dS
_
_
S
t
_

n
(y, t)
_
e
iky
dS
Finally,
= k
2
(k, t)
_
S
t
ik n [ (y, t)] e
iky
dS
_
S
t
_

n
(y, t)
_
e
iky
dS (59)
where k
2
= k
2
1
+ k
2
2
+ k
2
3
= |k|
2
.
9.2 Fourier transform of the time derivative of a piecewise differentiable function
9.2.1 The displacement velocity of a mobile surface
Let S(y
1
, y
2
, y
3
, t) = 0 be the equation of the mobile surface S
t
. Then, for S(y
01
, y
02
, y
03
, t
0
) on
S
t
0
we can write
0 = S(y
1
, y
2
, y
3
, t) S(y
01
, y
02
, y
03
, t
0
) = (60)
_
S
y
1
dy
1
dt
+
S
y
2
dy
2
dt
+
S
y
3
dy
3
dt
+
S
t
_
(t t
0
) ,
the partial derivatives of the function S being calculated at a certain point x
lying between
the points x (t
0
) and x (t) . But,
n
1
=
S/y
1
|grad S|
, n
2
=
S/y
2
|grad S|
, n
3
=
S/y
3
|grad S|
(61)
are the components of the external normal unit vector n and
dy
1
dt
= v
1
,
dy
2
dt
= v
2
,
dy
3
dt
= v
3
, (62)
are the projections on the velocity vector of a point on the surface S
t
on the three axes. The
relation (60) yields
v
n
=
S/t
|grad S|
(63)
which is the displacement velocity of the geometric surface S
t
. We mention that the
displacement velocity of a surface has the direction of the normal vector to this surface.
9.2.2 Reynolds transport theorem
For calculating the Fourier Transform of a time derivative of a certain function we write
_
D
(i)
(t)
(i)
(x, t) e
ikx
dx
_
D
(i)
(t
0
)
(i)
(x, t
0
) e
ikx
dx = (64)
_
D
(i)
(t)
(i)
(x, t) e
ikx
dx
_
D
(i)
(t)
(i)
(x, t
0
) e
ikx
dx+
_
D
(i)
(t)
(i)
(x, t
0
) e
ikx
dx
_
D
(i)
(t
0
)
(i)
(x, t
0
) e
ikx
dx =
_
D
(i)
(t)
_
(i)
(x, t)
(i)
(x, t
0
)
_
e
ikx
dx+
_
D
(i)
(t)D
(i)
(t
0
)
(i)
(x, t
0
) e
ikx
dx
Now, dividing by (t t
0
) and taking the limit for t t
0
the rst term gives the Fourier
transform of the time derivative of the function
(i)
(x, t) while in the second integral we can
write dx =v
n
(t t
0
) dS (see (Jacob, 1959), (Currie, 2003)). Finally, we obtain the following
form of the Reynolds transport theorem
d
dt
_
D
(i)
(t)
(i)
(x, t) e
ikx
dx =
_
D
(i)
(t)
(i)
(x, t)
t
e
ikx
dx (65)
+
_
S
t
v
n
(y, t)
(i)
(y, t) e
iky
dS
v
n
being the displacement velocity of the surface S (t) .
9.2.3 The second basic formula
We calculate now
F
_
(x, t)
t
_
=
_
D
(i)
(t)
(i)
(x, t)
t
e
ikx
dx +
_
D
(e)
(t)
(e)
(x, t)
t
e
ikx
dx
By using formula (65) we get
_
D
(i)
(t)
(i)
(x, t)
t
e
ikx
dx =
d
dt
_
D
(i)
(t)
(i)
e
ikx
dx
_
S
t
v
n

(i)
(y, t) e
iky
dS (66)
_
D
(e)
(t)
(e)
(x, t)
t
e
ikx
dx =
d
dt
_
D
(e)
(t)
(e)
e
ikx
dx +
_
S
t
v
n
(e)
(y, t) e
iky
dS (67)
The sum of formulas (66) and (67) gives the second basic formula
F
_
(x, t)
t
_
=
d (k, t)
dt
+
_
S
t
v
n
(y, t) [ (y, t)] e
iky
dS. (68)
Formula (68) permits us to calculate the Fourier transformof the time derivative of a piecewise
differentiable function. For the second time derivative we can write
F
_
2
(x, t)
t
2
_
=
d
dt
d
dt
+
_
S
t
v
n
(y, t)
_
(y, t)
t
_
e
iky
dS (69)
By using again formula (68) we obtain nally,
t
2
=
d
2

dt
2
+
d
dt
_
S
t
v
n
[] e
iky
dS +
_
S
t
v
n
_
t
_
e
iky
dS (70)
such that in the Fourier transform of second time derivative of the piecewise differentiable
function enters the jump of the function across the discontinuity and the jump of the rst
time derivative of as well.
10. Appendix B: Determination of the Greens function for the wave equation
By using formula 11, given at page 364 in (Gelfand and Shilov, 1964) we can write
F
1
_
sin (ckt)
k
_
=
(t r/c)
4c r
(71)
where r = |x| .
By taking the derivative with respect to parameter t in the both sides of relationship (71) there
results
F
1
{cos (ckt)} =

(t r/c)
4c
2
r
(72)
Similarly, integrating the formula (71) over the interval (0, t) we obtain
F
1
_
1 cos (ckt)
k
2
_
=
H (t r/c)
4r
(73)
H (x) being the Heavisides function.
11. References
Brentner, K. S., An efcient and robust method for predicting helicopter high-speed impulsive
noise, Journal of Sound and Vibration, 1997, 203(1), 87-100.
Brentner, K. S. &Farassat, F. 1998 Analytical comparison of the acoustic analogy and Kirchhoff
formulation for moving surfaces. AIAA J. 36(8), 1998, 13791386.
Crighton, D. G., Dowling, A. P., Ffowcs Williams, J. E., Heckl, M., and Leppington, F. G.,
Modern Methods in Analytical Acoustics: Lecture Notes, SpringerVerlag, London,
1992. Chap.11, Sec. 10.
Curle, N., The inuence of solid boundaries upon aerodynamic sound, Proceedings of the
Royal Society A 231 (1955) 505514.
Currie, I. G., Fundamental Mechanics of Fluids, 3rd edition, Marcel Dekker, 2003, pg.12.
Dowling, A. P., and Ffowcs Williams, J.E., Sound and Sources of Sound, Wiley &Sons, New
York, 1982. Chap. 9, Sec. 2.
Farassat, F., Introduction to generalized functions with applications in aerodynamics and
aeroacoustics, Corrected Copy (April 1996), NASA Technical Paper 3428, 1994,
Farassat, F., The Kirchhoff formulas for moving surfaces in aeroacousticsthe subsonic and
supersonic cases, NASA Technical Memorandum 110285, 1996.
Farassat, F., Comments on the paper by Zinoviev and Bies On acoustic radiation by a rigid
object in a uid ow, Journal of Sound and Vibration 281 (2005) 12171223.
Farassat, F., Derivation of Formulations 1 and 1A of Farassat, NASA Technical Memorandum
214853, 2007.
Farassat, F., and Brentner, K. S., The uses and abuses of the acoustic analogy in helicopter rotor
noise prediction, Journal of the American Helicopter Society, 1988, 33, 29-36
Farassat, F., Myers, M.K., Further comments on the paper by Zinoviev and Bies, On acoustic
radiation by a rigid object in a uid ow, Journal of Sound and Vibration 290 (2006),
538-547.
Ffowcs Williams, J.E., and Hawkings, D.L., Sound generation by turbulence and surfaces
in arbitrary motion, Philosophical Transactions of the Royal Society A 264 (1969)
321342.
Francescantonio, P. Di,A new boundary integral formulation for the prediction of sound
radiation, J. Sound Vibr. 202 (1997) (4), pp. 491509.
Gelfand, I. M., and Shilov, G. E., Generalized Functions, Volume 1, Academic Press, NewYork
and London, 1964.
Goldstein, M. E., Aeroacoustics, McGraw-Hill Book Co., 1976.
Homentcovschi, D., and Singler, T., An introduction to BEM by integral transforms,
Engineering Analysis with Boundary Elements, 23 (1999) 603-609.
Jacob, C., Introduction Mathematique a la Mechanique des Fluides, Gauthier-Villlars, Paris,
1959.
Lighthill, M.J., On sound generated aerodynamically, Proceedings of the Royal Society of
London, A 211 (1952) 564586.
Lighthill, M.J., On sound generated aerodynamically: Turbulence as a source of sound,
Proceedings of the Royal Society of London, A 222 (1954), 1-32.
Lyrintzis, A. (1994). Review: The use of Kirchhoffs method in computational aerodynamics,
Transactions of the ASME, Journal of Fluid Engineering 116(4): 665676.
Lyrintzis, A.S., Surface integral methods in computational aeroacoustics.-Fromthe (CFD) near
eld to the (Acoustic) far-eld, Aeroacoustics, 2 (2003) pp. 95-128.
Morino, L. (2003). Is there a difference between aeroacoustics and aerodynamics? An
aeroelasticians viewpoint, AIAA JOURNAL 41(7): 12091223.
Morris, P.J. and Farassat, F., Acoustic analogy and alternative theories for jet noise prediction,
AIAA Journal, 40 (2002), pp. 671-680.
Pilon, A. R., and Lyrintzis, A. S., Integral Methods for Computational Aeroacoustics, AIAA
paper No. 97-0020, presented at the 35th Aerospace Science Meeting, Reno, NV, Jan.
1997.
Raman, G. (editor), Computational Aeroacoustics, Multiscience Publishing Co. Ltd., 2009,
507pp.
Wu, X-F, and Akay A., Sound radiation from vibrating bodies in motion, J. Acoust. Soc. Am.,
91 (1992) pp.2544-2555.
Zinoviev, A., Bies, D.A., On acoustic radiation by a rigid object in a uid ow, Journal of
Sound and Vibration 269 (2004) 535548.
Zinoviev, A., Bies, D.A., Authors Reply to: F. Farassat, Comments on the paper by Zinoviev
and Bies On acoustic radiation by arigid object in a uid ow, Journal of Sound
and Vibration 281 (2005) 12241237.
Zinoviev, A., Bies, D.A., Authors Reply, Journal of Sound and Vibration 290 (2006) 548-554.
0
Exact Solutions Expressible in Hyperbolic and
Jacobi Elliptic Functions of Some Important
Equations of Ion-Acoustic Waves
A. H. Khater
1
and M. M. Hassan
2
1
Mathematics Department, Faculty of Science, Beni-Suef University, Beni-Suef
2
Mathematics Department, Faculty of Science, Minia University, El-Minia
Egypt
1. Introduction
Many phenomena in physics and other elds are often described by nonlinear partial
differential equations (NLPDEs). The investigation of exact and numerical solutions, in
particular, traveling wave solutions, for NLPDEs plays an important role in the study of
nonlinear physical phenomena. These exact solutions when they exist can help one to well
understand the mechanism of the complicated physical phenomena and dynamical processes
modeled by these nonlinear evolution equations (NLEEs). The ion-acoustic solitary wave
is one of the fundamental nonlinear wave phenomena appearing in uid dynamics [1] and
plasma physics [2, 3]. It has recently became more interesting to obtain exact analytical
solutions to NLPDEs by using appropriate techniques and symbolical computer programs
such as Maple or Mathematica. The capability and power of these software have increased
dramatically over the past decade. Hence, direct search for exact solutions is now much more
viable. Several important direct methods have been developed for obtaining traveling wave
solutions to NLEEs such as the inverse scattering method [3], the tanh-function method [4],
the extended tanh-function method [5] and the homogeneous balance method [6]. We assume
that the exact solution is expressed by a simple expansion u(x, t) = U() =
N
i=0
A
i
F
i
()
where A
i
are constants to be determined and the function F() is dened by the solution
of an auxiliary ordinary differential equation (ODE). The tanh-function method is the well
known method as a direct selection of the function F() = tanh( ). Recently, many exact
solutions expressed by various Jacobi elliptic functions (JEFs) of many NLEEs have been
obtained by Jacobi elliptic function expansion method [7-10], mapping method [11, 12],
F-expansion method [13], extended F-expansion method [14], the generalized Jacobi elliptic
function method [15] and other methods [16-20]. Various exact solutions were obtained by
using these methods, including the solitary wave solutions, shock wave solutions and periodic
wave solutions.
The main steps of the F-expansion method [13] are outlined as follows:
Step 1. Use the transformation u(x, t) = u(); = k(x t) +
0
,
0
is an arbitrary constant,
and reduce a given NLPDE, say in two independent variables,
F(u, u
t
, u
x
, u
tt
, u
xx
, ...) = 0, (1.1)
4
to the (ODE)
G(u, u
, u
, ...) = 0, u
=
du
d
. (1.2)
In general, the left hand side of Eq. (1.1) is a polynomial in u and its various derivatives.
Step 2. The F-expansion method gives the solution of (1.1) in the form
u(x, t) = u() =
N
i=0
a
i
F
i
(), a
N
= 0, (1.3)
where a
i
(i = 0, 1, 2, ..., N) are constants to be determined and F() satises the rst order
nonlinear ODE in the form
(F
())
2
= q
0
+ q
2
F
2
() + q
4
F
4
(), (1.4)
where q
0
, q
2
and q
4
are constants and N in Eq. (1.3) is a positive integer that can be determined
by balancing the nonlinear term(s) and the highest order derivatives in Eq. (1.1).
Step 3. Substituting the F-expansion (1.3) into (1.2) and using (1.4); setting each coefcient of
the polynomial to zero yields a system of algebraic equations involving a
0
, a
1
, ...a
N
, k and .
Step 4. Solving these equations, probably with the aid of Mathematica or Maple, then
a
0
, a
1
, ...a
N
, k and can be expressed by q
0
, q
2
, q
4
.
Step 5. Substituting these results into F-expansion (1.3), then a general form of traveling
wave solution of the NLPDE (1.1) can be obtained. Many solutions of equation (1.4) have
been reported in [13, 14]. Substituting the values of q
0
, q
2
, q
4
and the corresponding JEF
solution F() into the general form of solution, we may get several classes of exact solutions
of equations (1.1) involving JEFs.
Also, we give a brief description of the mapping method to seek the traveling wave solutions
of (1.1) in the form u(x, t) = u(), = kx t +
0
,
0
is an arbitrary constant. Thus, Eq.
(1.1) reduces to Eq. (1.2), whose solution can be express in the form
u() =
n
i=0
A
i
f
i
(), (1.5)
where n is a balancing number, A
i
are constants to be determined and f () satises the
nonlinear ODE
f
2
() = 2 p f () + q f
2
() +
2
3
r f
3
(). (1.6)
Here p, q and r are constants. After substituting Eq. (1.5) into the ODE (1.2) and using
Eq. (1.6), the constants A
i
, k and may be determined. By using the solutions of auxiliary
nonlinear equation (1.6), many JEF solutions of NLEEs have been obtained [19, 20].
The JEFs sn() = sn(, m), cn() = cn( , m) and dn() = dn(, m) are double periodic and
have the following properties:
sn
2
() +cn
2
() = 1, dn
2
() + m
2
sn
2
() = 1.
In the limit m 1, the JEFs degenerate to the hyperbolic functions, i.e.,
sn(, 1) tanh(), cn(, 1) sech(), dn(, 1) sech().
Detailed explanations about JEFs can be found in [21].
Some of the nonlinear models in uids, plasma and dust plasma are described by canonical
models and include the Korteweg-de Vries (KdV) and the modied KdV equations [22-25].
Exact Solutions Expressible in Hyperbolic and Jacobi Elliptic Functions of Some Important Equations of Ion-Acoustic Waves 3
The evolution of small but nite-amplitude solitary waves, studied by means of the
Korteweg-de Vries (KdV) equation, is of considerable interest in plasma dynamics. In the
study of multidimensional version two type of nonlinear waves are well known, the so called
Kadomtsev-Petviashvilli (KP) equation and Zakharov - Kuzentsov (ZK) equation. Employing
the reductive perturbation technique on the system of equations for hydrodynamics and the
dynamics of plasma waves to derive such equation.
We construct several classes of exact JEF solutions of some nonlinear evolution equations of
plasma physics by using the mapping method and the F-expansion method. The rest of this
chapter is organized as follows: in section 2, we present the JEF solutions to the KdVequation,
combined KdV - modied KdV equation. In section 3, we apply the F-expansion method to
the Schamel- KdV equation. Moreover, using the ansatz solution (1.5) and the solutions of
nonlinear ODE (1.6), many exact solutions of Schamel equation, ZK equation and modied
fth order KdV equation are given in sections 4, 5, 6.
2. The KdV and modied KdV equations
The Korteweg de-Vries (KdV) equation
u
t
+ uu
x
+ u
xxx
= 0,
models a variety of nonlinear phenomena, including ion acoustic waves in plasmas, dust
acoustic solitary structures in magnetized dusty plasmas, and shallow water waves. On the
other hand, the modied KdV equation (mKdV)
u
t
+ bu
2
u
x
+ u
xxx
= 0,
models the dust-ion acoustic waves, electromagnetic waves in size-quantized lms, ion
acoustic solitons, trafc ow problems, and in other applications. The KdV equation and
the modied KdV equation are completely integrable equations that have multiple-soliton
solutions and possess innite conservation quantities. The KdVequation is the earliest soliton
equation that was rstly derived by Korteweg and de Vries to model the evolution of shallow
water wave in 1895. In the study of the KdV equation, traveling wave solution leads to
periodic solution which is called cnoidal wave solution [22, 23]. Exact solutions of KdV
equation have been studied extensively since they were rst found. Solitary wave solutions
and periodic wave solutions were obtained for the KdV and modied KdV equations [3, 7,
22]. The JEF solutions to two kinds of KdV equations with variables coefcients have been
constructed by using the method of the auxiliary equation [19]. The reductive perturbation
method [24] has been employed to derive the KdV equation for small but nite amplitude
electrostatic ion-acoustic waves [23, 25, 26]. The basic equations describing the system in
dimensionless variables is studied by El-Labany [26] and the KdV equation for the rst-order
perturbed potential has been obtained using the reductive perturbation method.
We consider the combined KdV and mKdV equation [22, 27, 28]
u
t
+ uu
x
+ u
2
u
x
+ u
xxx
= 0, = 0. (2.1)
where , and are constants. Equation (2.1) is widely used in various elds such as
quantum eld theory, dust-acoustic waves, ion acoustic waves in plasmas with a negative
ion, solid-state physics and uid dynamics.
Let u = u(), equation (2.1) transformed to the reduced equation
u
+ uu
+ u
2
u
+ k
2
u
= 0. (2.2)
69
Exact Solutions Expressible in Hyperbolic
and Jacobi Elliptic Functions of Some Important Equations of Ion-Acoustic Waves
Balancing u
with u
2
u
yields N = 1, so the F-expansion method gives

u(x, t) = a
0
+ a
1
F(). (2.3)
Substituting (2.3) into (2.2) and equating the coefcients of like powers of F() to zero, we
obtain a set of algebraic equations. Solving these algebraic equations, we obtain the exact
solutions of (2.1) as follows:
When q
0
= 1, q
2
= 1 m
2
, q
4
= m
2
, solutions of Eq. (1.4) is F() = sn , we have
u =

2
k
6m
2
sn
_
k(x + (
2
+4k
2
(m
2
+1)
4
)t) +
0
_
, (2.4)
If q
0
= m
2
1, q
2
= 2 m
2
, q
4
= 1, the solution of Eq (1.4) is F() = dn. Thus, we obtain
the periodic wave solutions of Eq. (2.1)
u =

2
k
dn(k(x +

2
4k
2
(2 m
2
)
4
t) +
0
), (2.5)
Selecting the values of the q
0
, q
2
and q
4
of equation (1.4) and the corresponding function
F, we can construct various JEF solutions of (2.1). Other JEF solutions are omitted here for
simplicity. If we put = 0 in (2.4), we get the periodic solution of the modied KdV equation
which coincides with that given by Liu et al. [7]. Moreover, the solutions (2.5) to equation (2.1)
given in [28] are recovered. With m 1 in (2.4) , (2.5), the solitary wave solutions to (2.1)
given in [7, 27, 28] are also recovered.
We notice that the solutions of the KdV equation cannot obtain from (2.4) and (2.5) as = 0.
In this case, the general form of cnoidal wave solutions of the KdV equation are given by
u(x, t) =
3q
4
q
2
F
2
(), =
_

4q
2
(x t) +
0
. (2.6)
Thus we can obtain abundant cnoidal wave solutions of the KdV equation in terms of JEFs.
Some periodic wave solutions of the KdV equation and modied KdV equation have been
studied in [7,23, 28]. As m 1, these solutions will degenerate into the corresponding
solitary wave solutions.
3. The JEF solutions of Schamel- KdV equation
We consider the Schamel- KdV equation [29, 30]
u
t
+ (u
1/2
+ u)u
x
+ u
xxx
= 0, = 0 (3.1)
where , and are constants and u is the wave potential.
In order to nd the periodic wave solution of (3.1), we use the transformations u = v
2
,
v(x, t) = V(); = k(x t) +
0
, then (2.7) becomes
V V
+ (V
2
+ V
3
)V
+ k
2
[VV
+3V
] = 0. (3.2)
The balancing procedure implies that N = 1. Therefore, the F-expansion method gives the
solution
V(x, t) = V() = a
0
+ a
1
F(), (3.3)
where a
0
and a
1
are constants to be determined and F() is a solution of Eq. (1.4). Substituting
Eq. (3.3) into Eq. (3.2) and equating the coefcients of the like powers of F to zero, yields a set
of algebraic equations for a
0
, a
1
, k and :
[a
2
1
+12k
2
q
4
]a
1
= 0,
[a
2
1
+3a
0
a
2
1
+6k
2
a
0
q
4
]a
1
= 0,
[ +2a
0
+3a
2
0
+4k
2
q
2
]a
1
= 0,
[ + a
0
+ a
2
0
+ k
2
q
2
]a
0
= 0.
(3.4)
Solving these algebraic equations, we gave a general form of traveling wave solutions of Eq.
(3.1)
u =
4
2
25
2
_
1
2q
4
q
2
F()
_
2
. (3.5)
Therefore, we obtained in [30] the JEF solutions of Eq. (3.1) as follows:
When q
0
= 1, q
2
= 1 m
2
, q
4
= m
2
, solutions of Eq. (1.4) is F() = sn , we have
u
1
=
4
2
25
2
_
1
_
2m
2
m
2
+1
sn
_
2
5
6(m
2
+1)
(x +
16
2
75
t) +
0
__
2
, < 0, (3.6)
If q
0
= 1 m
2
, q
2
= 2m
2
1, q
4
= m
2
, F() = cn, thus yields the exact solutions of Eq.
(3.1)
u
2
=
4
2
25
2
_
1
_
2m
2
2m
2
1
cn
_
2
5
6(2m
2
1)
(x +
16
2
75
t) +
0
__
2
, > 0,
(3.7)
If q
0
= m
2
1, q
2
= 2 m
2
, q
4
= 1, the solution of Eq (1.4) is F() = dn. So, we obtained
the exact solutions of Eq. (3.1) in the form
u
3
=
4
2
25
2
_
1
_
2
2m
2
dn
_
2
5
6(2m
2
)
(x +
16
2
75
t) +
0
__
2
, > 0,
(3.8)
Many types of JEF solutions of Eq. (3.1) are given [30]. As m 1, Eqs. (3.6)-(3.8) degenerate
to
u
4
=
4
2
25
2
_
1 tanh
_

5
3
(x +
16
2
75
t) +
0
__
2
, < 0,
u
5
=
4
2
25
2
_
1
2 sech
_
2
5
6
(x +
16
2
75
t) +
0
__
2
, > 0,
(3.9)
The solitary wave solutions (3.9) in terms of tanh are equivalent to the solutions given in [31].
The JEF solutions of (3.1) may be describe various features of waves and may be helpful in
understanding the problems in ion acoustic waves.
71
4. Schamel equation and modied KP equation
The equation describing ion-acoustic waves in a cold-ion plasma where electrons do not
behave isothermally during their passage of the wave is
u
t
+ u
1/2
u
x
+ u
xxx
= 0. (4.1)
Schamel [29] derived this equation and a simple solitary wave solution having a sech
4
prole
was obtained. Therefore the Schamel equation (4.1) containing a square root nonlinearity is
very attractive model for the study of ion-acoustic waves in plasmas and dusty plasmas.
In order to nd the periodic wave solution of (4.1), we use the transformations
u = v
2
, v(x, t) = V(); = kx t +
0
,, then (4.1) becomes
V V
+ k V
2
V
+ k
3
[VV
+3V
] = 0. (4.2)
According to the mapping method, we assume that Eq. (4.2) has the following solution:
V() = A
0
+ A
1
f (), (4.3)
where A
0
and A
1
are constants to be determined and f () satises Eq. (1.6).
Substitution of Eq. (4.3) into Eq. (4.2) and selecting the values of p, q and r, we have the
solutions of Eq. (4.1) which was given in [20] as follows:
Case 1. p = 2, q = 4 (1 + m
2
), r = 6 m
2
. In this case, we have f () = sn
2
. Thus the
periodic wave solutions of Eq. (4.1) are
u
1
(x, t) = 100
2
k
4
_
1 + m
2
1 m
2
+ m
4
3 m
2
sn
2
_
2
,
= kx 16 k
3
1 m
2
+ m
4
t +
0
.
(4.4)
Case 2. p =
(1m
2
)
2
2
, q = 2(1 + m
2
), r =
3
2
. The solutions of Eq. (1.6) are f () =
(mcn dn)
2
. Thus the exact solutions of Eq. (4.1) are
u
2
(x, t) =
25
2
k
4
4
_
2(1 + m
2
)
1 +14m
2
+ m
4
+3 (mcn dn)
2
_
2
,
= kx 4 k
3
1 +14m
2
+ m
4
t +
0
.
(4.5)
Case 3. p =
m
2
2
, q = 2(m
2
2), r =
3 m
2
2
. The solutions of Eq. (1.6) are
f () =
_
msn
1dn
_
2
. So, we obtained the exact solutions of Eq. (4.1)in the form
u
3
(x, t) =
25
2
k
4
4
_
2(2 m
2
)
16 16m
2
+ m
4
3 m
4
_
sn
1dn
_
2
_
2
,
= kx 4 k
3
16 16m
2
+ m
4
t +
0
.
(4.6)
There are several exact solutions for the Eq. (4.1) which are omitted here for simplicity. As
m 1, these solutions reduce to the solitary wave solutions
u
4
(x, t) = 900
2
k
4
sech
4
(kx 16 k
3
t +
0
),
u
5
(x, t) = 100
2
k
4
[2 3 sech
2
(kx +16 k
3
t +
0
)]
2
.
(4.7)
u
6
(x, t) =
225
2
k
4
4
_
1
_
tanh(kx4 k
3
t+
0
)
1+sech(kx4 k
3
t+
0
)
_
2
_
2
.
(4.8)
The KdV equation in two dimensions, known as Kadomtsev Petviashivili (KP) equation [32],
was derived for ion-acoustic waves in a non magnetized plasma by Kako and Rowlands [33].
Therefore the modied KP equation containing a square root nonlinearity is very attractive
model for the study of ion-acoustic waves in plasma and dusty plasma [34- 36]. Extensive
work has been devoted to the study of nonlinear waves associated with the dust ion-acoustic
waves, particularly the dust ion-acoustic solitary and shock waves in dusty plasmas in which
dust particles are stationary and provide only the neutrality [37]. The KP equation is derived
[38] for the propagation of nonlinear waves in warm dusty plasmas with variable dust charge,
two-temperature ions and nonthermal electrons by using the reductive perturbation theory.
Consider the modied KP equation
(u
t
+ u
1/2
u
x
+ u
xxx
)
x
+ u
yy
= 0, (4.9)
where and are constants. The modied KP equation (4.9) for ion-acoustic waves in a multi
species plasma consisting of non-isothermal electrons have been derived by Chakraborty and
Das [34]. We applied the mapping method with the ansatz solution (4.3) and the solutions of
auxiliary equation (1.6) to nd the solutions of equation (4.9) (see [39]).
5. The ZK equation and modied ZK equation
The equation
u
t
+ u
2
u
x
+ u
xxx
+ u
yyx
= 0, (5.1)
is the modied ZK in (2+1) dimensions which is a model for acoustic plasma waves [40, 41].
The ZK equation was rst derived for describing weakly nonlinear ion- acoustic waves in a
strongly magnetized lossless plasma in two dimension [41]. The ZK equation and modied
ZK equation possess traveling wave structures [28, 42]. Peng [42] studied the exact solutions
of ZK equation by using extended mapping method. Various types of solutions of Schamel-
KdV equation and modied ZK equation arising in plasma and dust plasma are presented in
[43].
We apply the F-expansion method to the modied ZK equation. Thus, Eq. (5.1) has a solution
in the form
u() = a
0
+ a
1
F(), = k(x + ly t) +
0
.
Substituting this equation into Eq. (5.1), we obtain the following classes of exact solutions of
the modied ZK equation:
u = m
_
6
(m
2
+1)
sn(
_

(m
2
+1)(1+l
2
)
(x + ly t +
0
)),
u = m
_
6
(2m
2
)
dn(
_

(2m
2
)(1+l
2
)
(x + ly t +
0
)).
(5.2)
In the following we apply the mapping method to the ZK equation
u
t
+ u u
x
+ u
xxx
+ u
yyx
= 0. (5.3)
In this case, we have n = 1. Thus Eq. (5.3) has a solution in the form
u() = A
0
+ A
1
f (), = kx + ly t +
0
.
73
Substituting this equation into Eq. (5.3) to determine A
0
, A
1
, k, and using the solutions
of auxiliary equation (1.6), we obtained the following classes of exact solutions of the ZK
equation [39]:
u
1
(x, y, t) =

k
+
4(1+m
2
)(l
2
+k
2
)

12m
2
(l
2
+k
2
)
sn
2
( kx + ly t +
0
),
u
2
(x, y, t) =

k
+
4(m
2
2)(l
2
+k
2
)
+
12(l
2
+k
2
)
dn
2
( kx + ly t +
0
),
(5.4)
u
3
=

k

2(1+m
2
)(l
2
+k
2
)
+
3(l
2
+k
2
)
[mcn( ) dn( )]
2
,
(5.5)
u
4
(x, y, t) =

k

2(1+m
2
)(l
2
+k
2
)

3(1m
2
)(l
2
+k
2
)
_
cn( kx+lyt+
0
)
1sn( kx+lyt+
0
)
_
2
.
(5.6)
When m 1, some of these solutions degenerate as solitary wave solutions of ZK equation.
The solutions (5.3) are coincide with the solutions given in [44].
Recently, some properties of the quantum ion-acoustic waves were also investigated in
dense quantum plasmas by studying the quantum hydrodynamical equations in different
conditions, which includes the quantum Zakharov Kuznetsov equation, the extended
quantum Zakharov Kuznetsov equation, and the quantum Zakharov system [45]. The
three-dimensional extended quantum Zakharov Kuznetsov (QZK) equation [46] was
investigated in dense quantum plasmas which arises from the dimensionless hydrodynamics
equations describing the nonlinear propagation of the quantum ion-acoustic waves. The
three-dimensional extended QZK equation was given in [46]
t
+ (A+ B
2
)
x
+ C
zzz
+ D(
xxz
+
yyz
) = 0, (5.7)
where A, B, C and D are constants. This equation has the following JEF solutions (see [45, 46]):
1
=
A
2B
+ mk
_
6E
B
sn( k(x + ly + z t +
0
)), =
4BEk
2
(1+m
2
)+A
2
4 B
, BE < 0,
2
=
A
2B
+ mk
_
6E
B
cn( k(x + ly + z t +
0
)), =
4BEk
2
(12m
2
)+A
2
4 B
, BE > 0,
(5.8)
with E = C
2
+D(1 +l
2
). Moreover, many types of analytical solutions of the extended QZK
equation are constructed in terms of some powerful ansatze, which include doubly periodic
wave solutions, solitary wave solutions, kink-shaped wave solutions, rational wave solutions
and singular solutions [46].
6. The modied fth order KdV equation
Higher order KdV equations have many applications in different elds of mathematical
physics. For example the fth-order KdV equations can be derived in uid dynamics and
in magneto-acoustic waves in plasma and its exact solutions was given in [47-51]. The
higher-order KdV equation can be derived for magnetized plasmas by using the reductive
perturbation technique. Traveling wave solutions of Kawahara equation and modied
Kawahara equation have been studied [9, 48, 49]. Moreover, the solitary wave solutions of
nonlinear equations with arbitrary odd-order derivatives were studied by many authors [47,
51].
Consider the modied fth order KdV equation
u
t
+ u
2
u
x
+ c
3
u
xxx
+ c
5
u
xxxxx
= 0, (6.1)
where , c
3
and c
5
are constants. Here, we review the exact traveling wave solutions of
equation (6.1) using exact solutions of the auxiliary equation (1.5) and applied the mapping
method. Thus, Eq. (6.1) has the solutions in the form
u() = A
0
+ A
1
f (), = kx t +
0
, (6.2)
Substituting equation (6.2) into (6.1) and equating the coefcients of like powers of f to zero,
yields a system of algebraic equations for A
0
, A
1
, k and and then solve it. Therefore, the
solutions of the modied fth order KdV equation (6.1) was given in [39] as follows:
u
1
=
(10 k
2
c
5
(1+m
2
)+c
3
)
10c
5
k
2
_
90 c
5
4
_
mcn dn
_
2
,
= k
_
x +
(15c
2
5
k
4
(m
4
+14m
2
+1)+c
2
3
)
10c
5
t
_
+
0
.
(6.3)
If we choose A
0
= 0, equation (6.3) takes the form
u
2
=
3c
3
2(1+m
2
)
10 c
5
_
mcn dn
_
2
,
=
_
c
3
10c
5
(1+m
2
)
_
x +
(23m
4
+82m
2
+23) c
2
3
200c
5
(1+m
2
)
2
t
_
+
0
.
(6.4)
Moreover, we have obtained the exact solutions
u
3
=
3m
2
c
3
2(m
2
2)
10 c
5
_
msn
1dn
_
2
,
=
_
c
3
10c
5
(2m
2
)
_
x +
(23m
4
128m
2
+128) c
2
3
200c
5
(m
2
2)
2
t
_
+
0
,
(6.5)
u
4
=
3c
3
2(12m
2
)
10 c
5
_
sn
1cn
_
2
,
=
_
c
3
10c
5
(12m
2
)
_
x +
(128m
4
128m
2
+23) c
2
3
200c
5
(12m
2
)
2
t
_
+
0
.
(6.6)
There are several other JEFs of Eq. (6.1) which are omitted here for simplicity. When m 1,
then (6.4)-(6.6) become the solitary wave solutions
u
5
=
3c
3
_
10c
5
sech
2
(
1
2
_
c
3
5c
5
(x +
4 c
2
3
25c
5
t) +
0
), (6.7)
u
6
=
3c
3
2
_
10 c
5
_
tanh
1 sech
_
2
, =
_
c
3
10c
5
_
x +
23 c
2
3
200c
5
t
_
+
0
. (6.8)
We notice that Eq. (6.7) is the solution given by Example 2 in Ref. [47].
Finally, we can construct various types of exact and explicit solutions of the generalized ZK
equation
u
t
+ ( + u
p
) u
p
u
x
+ u
xxx
+ u
yyx
= 0, (6.9)
by using suitable method and using an appropriate transformation. Also, we can study the
exact solution of the generalized KdV equation ( = 0) which studied by many authors
[22, 23, 31]. The generalized ZK equation was rst derived for describing weakly nonlinear
ion-acoustic waves in strongly magnetized lossless plasma in two dimensions and governs
the behavior of weakly nonlinear ion-acoustic waves in plasma comprising cold ions and
75
hot isothermal electrons in the presence of a uniform magnetic eld. Eq. (6.9) includes
considerable interesting equations, such as KdV equation, mKdV equation, ZK equation
and mZK equation. Exact traveling wave solutions for the generalized ZK equation with
higher-order nonlinear terms have obtained in [52-54]. Moreover, we can use the symbolic
computations and apply the mapping method with the ansatz solution (1.5) to nd the several
classes of traveling wave solutions of the fth order KdV equation
u
t
+ c
1
u u
x
+ c
2
u
xxx
+ u
xxxxx
= 0.
This equation appears in the theory of shallow water waves with surface tension and the
theory of magneto-acoustic waves in plasmas [9]. Wazwaz [55] studied soliton solutions of
fth-order KdV equation. We can use a suitable method to construct the exact solutions of
some special types of nonlinear evolution equations aries in plasma physics such as Liouville,
sine-Gordon and sinh-Poisson equations.
7. References
[1] G. Whitham, Linear and Nonlinear Waves, New York, Wiley (1974).
[2] R. Davidson, Methods in Nonlinear Plasma Theory, New York, Academic Press (1972).
[3] M. J. Ablowitz and P. A. Clarkson, Solitons, Nonlinear Evolution Equations and Inverse
Scattering Transform, Cambridge, Cambridge University Press (1991).
[4] W. Maliet, Solitary wave solutions of nonlinear wave equations, Am. J. Phys. 60 (1992)
650-654;
W. Maliet, The tanh method: a tool for solving certain classes of nonlinear evolution
and wave equations, J. Comput. Appl. Math. 164-156 (2004) 529-541.
[5] E.G. Fan, Extended tanh-function method and its applications to nonlinear equations,
Phys. Lett. A 277 (2000) 212-218;
E.G. Fan and Y.C. Hong, Generalized tanh method to special types of nonlinear
equations, Z. Naturforsch. A 57 (2002) 692-700.
[6] M. Wang, Exact solutions for a compound KdV - Burgers equation, Phys. Lett. A 213
(1996) 279-287.
[7] S. K. Liu, Z. T. Fu, S. D. Liu, and Q. Zhao, Jacobi elliptic function expansion method and
periodic wave solutions of nonlinear wave equations, Phys. Lett. A 289 (2001) 69-74;
[8] Z.T.Fu, S. K. Liu, S. D. Liu, and Q. Zhao, New Jacobi elliptic function expansion method
and new periodic solutions of nonlinear wave equations, Phys. Lett. A 290 (2001) 72-76.
[9] E.J. Parkes, B.R. Duffy and P.C. Abbott, The Jacobi elliptic-function method for nding
periodic wave solutions to nonlinear evolution equations, Phys. Lett. A 295 (2002)
280-286.
[10] H. T. Chen and H. Q. Zhang, Improved Jacobin elliptic method and its applications,
Chaos, Solitons and Fractals 15 (2003) 585-591.
[11] Y. Peng, Exact periodic wave solutions to a new Hamiltonian amplitude equation, J.
Phys. Soc. Japan 72 (2003) 1356-1359;
Y. Peng, New exact solutions to a new Hamiltonian amplitude equation II, J. Phys. Soc.
Japan 73 (2004) 1156-1158.
[12] Y. Peng, Exact periodic wave solutions to the Melnikov equation, Z. Naturforsch A 60
(2005) 321-327.
[13] Y. B. Zhao, M. L. Wang and Y. M. Wang, Periodic wave solutions to a coupled KdV
equations with variable coefcients, Phys. Lett. A 308 (2003) 31-36.
[14] J. Liu and K. Yang, The extended F-expansion method and exact solutions of nonlinear
PDEs, Chaos, Solitons and Fractals 22 (2004) 111-121.
[15] H. T. Chen and H. Q. Zhang, New double periodic and multiple soliton solutions of
the generalized (2+1)-dimensional Boussinesq equation, Chaos, Solitons and Fractals 20
(2004) 765-769.
[16] S. A. Elwakil, S. K. El-labany, M. A. Zahran and R. Sabry, Modied extended
tanh-function method for solving nonlinear partial differential equations, Phys. Lett. A.
299 (2002) 179-188.
[17] M. A. Abdou and S. Zhang, New periodic wave solutions via extended mapping
method, Commun. Nonlinear Sci. Numer. Simul. 14 (2009) 2-11.
[18] D. Baldwin,

U. G oktas, W. Hereman et al., Symbolic computions of exact solutions
expressible in hyperbolic and elliptic functions for nonlinear PDEs, J. Symb. Comput.
37 (2004) 669-705.
[19] Taogetusang and Sirendaoerji, The Jacobi elliptic function-like exact solutions to two
kinds of KdV equations with variable coefcients and KdV equation with forcible term,
Chinese Phys. 15 (2006) 2809-2818.
[20] A. H. Khater, M. M. Hassan, E. V. Krishnan and Y.Z. Peng, Applications of elliptic
functions to ion-acoustic plasma waves, Eur. Phys. J. D 50 (2008) 177-184.
[21] M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions, Dover, New York
(1965).
[22] P. G. Drazin and R. S. Johnson, Solitons,: An Introduction, Cambridge University press,
Cambridge (1989).
[23] A. Jeffrey and T. Kakutani, Weak nonlinear dispersive waves: A discussion centered
around the Korteweg de Vries equation, SIAM Rev. 14 (1972) 582-643.
[24] H. Washimi and T. Taniuti, Propagation of ion acoustic solitary waves of small
amplitude, J Phys Rev Lett. 17 (1966) 996 -998.
[25] E.K. El-Shewy, H.G. Abdelwahed and H.M. Abd-El-Hamid, Computational solutions
for the Korteweg
UdeVries equation in warm Plasma, Computat. Methods in Sci.

Technology 16 (2010) 13-18.
[26] S. K. El-Labany, Contribution of higher-order nonlinearity to nonlinear ion acoustic
waves in a weakly relativistic warm plasma. Part1. isothermal case, J. Plasma Phys. 50
(1993) 495 .
[27] J. F. Zhang, New solitary wave solution of the combined KdV and m KdV equation, Int.
J. Theoret. Phys. 37 (1998) 1541-1546.
[28] A. H. Khater and M. M. Hassan, Travelling and periodic wave solutions of some
nonlinear wave equations, Z. Naturforsch. 59 a (2004) 389-396.
[29] H. Schamel, A modied Korteweg de Vries equation for ion acoustic waves due to
resonant electrons, J. Plasma Phys. 9 (1973) 377-387.
[30] A. H. Khater, M. M. Hassan and R. S. Temsah, Exact solutions with Jacobi elliptic
functions of two nonlinear models for ion-acoustic plasma waves, J. Phys. Soc. Japan
74 (2005) 1431-1435.
[31] M. M. Hassan, Exact solitary wave solutions for a generalized KdV - Burgers equation,
Chaos, Solitons and Fractals 19 (2004) 1201-1206.
[32] B. B. Kadomtsev and V. I. Petviashvili, On the stability of solitary in weakly dispersive
media, Soviet Phys. Dokl. 15 (1970) 539-541.
[33] M. Kako and G. Rowlands, Two-dimensional stability of ion acoustic solitons, Plasma
Phys. 18 (1976) 165-170.
[34] D. Chakraborty and K. P. Das, Stability of ion acoustic solitons in a multispecies plasma
consisting of non-isothermal electrons, J. Plasma Phys. 60 (1998) 151-158.
[35] A. H. Khater, A. A. Abdallah, O. H. El-Kalaaway and D. K. Callebaut, Backlund
transformations, a simple transformation and exact solutions for dust-acoustic solitary
77
waves in dusty plasma consisting of cold dust particles and two-temperature isothermal
ions, Phys. Plasmas 6 (1999) 4542-4547.
[36] A. H. Khater and M. M. Hassan, Exact Jacobi elliptic function solutions for some special
types of nonlinear evolution equations, Il Nuovo Cimento 121 B (2006) 613- 622.
[37] A. A. Mamun and P.K. Shukla, Cylindrical and spherical dust ion-acoustic solitary
waves, Phys. Plasmas 9 (2002) 1468.
[38] H.R. Pakzad, Soliton energy of the Kadomtsev Petviashvili equation in warm dusty
plasma with variable dust charge, two-temperature ions, and nonthermal electrons,
Astrophys Space Sci. 326 (2010) 69 -75.
[39] A. H. Khater, M. M. Hassan and D. K. Callebaut, Travelling wave solutions to some
important equations of mathematical physics, Reports on Math. Phys. 66 (2010) 1-19.
[40] K. P. Das and F. Verheest, Ion acoustic solitons in magnetized multi-component plasmas
including negative ions, J. Plasma. Phys. 41 (1989) 139-155.
[41] V. E. Zakharov and E. A. Kuznetsov, On three-dimensional solitons, Sov. Phys. JETP 39
(1974) 285-286.
[42] Y.Z. Peng, Exact travelling wave solutions of the Zakharov Kuznetsov equation, Appl.
Math. Comput. 199 (2008) 397- 405.
[43] M. M. Hassan, New exact solutions of two nonlinear physical models, Commun. Theor.
Phys. 53 (2010) 596-604.
[44] M. M. Hassan, Exact solutions for some models of nonlinear evolution equations, J.
Egypt Math. Soc. 12 (2004) 31- 43.
[45] R. Sabry, W.M. Moslem, F. Haas, S. Ali, P.K. Shukla, Nonlinear structures: Explosive,
soliton and shock in a quantum electron- positron-ion magentoplasma, Phys. Plasmas
15 (2008) 122308.
[46] Z. Yan, Periodic, Solitary and rational wave solutions of the 3D extended quantum
Zakharov Kuznetsov equation in dense quantum plasmas, Phys. Let. A 373 (2009)
2432-2437.
[47] E.J. Parkes, Z. Zhu, B.R. Duffy and H. C. Hang, Sech-polynomail traveling solitary-wave
solutions of odd-order generalized KdV equations, Phys. Lett. A 248 (1998) 219-224.
[48] Z.J. Yang, Exact solitary wave solutions to a class of generalized odd-order KdV
equations, Int. J. Theor. Phys. 34 (1995) 641-647.
[49] D. Zhang, Doubly periodic solution of modied Kawahara equation, Chaos, Solitons and
Fractals 25 (2005) 1155-1160.
[50] A. H. Khater, M. M. Hassan and R. S. Temsah, Cnoidal wave solutions for a class of
fth-order KdV equations, Math. Comput. Simulation 70 (2005) 221-226.
[51] J. Sarma, Solitary wave solution of higher-order Korteweg de Vries equation, Chaos,
Solitons and Fractals 39 (2009) 277 - 281.
[52] L.- H. Zhang, Travelling wave solutions for the generalized Zakharov-Kuznetsov
equation with higher-order nonlinear terms, Appl. Math. Comput. 208 (2009) 144-155.
[53] C. Deng, New exact solutions to the Zakharov -Kuznetsov equation and its generalized
form, Commun. Nonlinear Sci. Numer. Simulat. 15 (2010) 857 - 868.
[54] SUN Yu-Huai, MA Zhi-Min and LI Yan, Explicit solutions for generalized
(2+1)-dimensional nonlinear Zakharov- Kuznetsov equation, Commun. Theor. Phys. 54
(2010) 397-400.
[55] A.-M., Wazwaz, Soliton solutions for the fth-order KdV equation and the Kawahara
equation with time-dependent coefcients, Phys. Scr. 82 (2010) 035009.
5
Acoustic Wave
P. K. Karmakar
Department of Physics, Tezpur University, Napaam, Tezpur, Assam
India
1. Introduction
An acoustic wave basically is a mechanical oscillation of pressure that travels through a
medium like solid, liquid, gas, or plasma in a periodic wave pattern transmitting energy
from one point to another in the medium [1-2]. It transmits sound by vibrating organs in the
ear that produce the sensation of hearing and hence, it is also called acoustic signal. This is
well-known that air is a fluid. Mechanical waves in air can only be longitudinal in nature;
and therefore, all sound waves traveling through air must be longitudinal waves originating
in the transmission form of compression and rarefaction from vibrating matter in the medium.
The propagation of sound in absence of any material medium is always impossible. Therefore,
sound does not travel through the vacuum of outer space, since there is nothing to carry the
vibrations from a source to a receiver. The nature of the molecules making up a substance
determines how well or how rapidly the substance will carry sound waves. The two
characteristic variables affecting the propagation of acoustic waves are (1) the inertia of the
constituent molecules and (2) the strength of molecular interaction. Thus, hydrogen gas,
with the least massive molecules, will carry a sound wave at 1,284.00 ms
-1
when the gas
temperature is 0
0
C [1]. More massive helium gas molecules have more inertia and carry a
sound wave at only 965.00 ms
-1
at the same temperature. A solid, however, has molecules
that are strongly attached, so acoustic vibrations are passed rapidly from molecule to
molecule. Steel, for an instant example, is highly elastic, and sound will move rapidly
through a steel rail at 5,940.00 ms
-1
at the same temperature. The temperature of a medium
influences the phase speed of sound through it. The gas molecules in warmer air thus have a
greater kinetic energy than those of cooler air. The molecules of warmer air therefore
transmit an acoustic impulse from molecule to molecule more rapidly. More precisely, the
speed of a sound wave increases by 0.60 ms
-1
for each Celcius degree rise in temperature
above 0
0
C.
Acoustic waves, or sound waves, are defined generally and specified mainly by three
characteristics: wavelength, frequency, and amplitude. The wavelength is the distance from
the top of one waves crest to the next (or, from the top of one trough to the next). The
frequency of a sound wave is the number of waves that pass a point each second [1]. Sound
waves with higher frequencies have higher pitches than sound waves with lower
frequencies and vice versa. Amplitude is the measure of energy in a sound wave and affects
volume. The greater the amplitude of an acoustic wave, the louder the sound and vice versa.
An acoustic wave is what makes humans and other animals able to hear. A persons ear
perceives the vibrations of an acoustic wave and interprets it as sound [1]. The outer ear, the
visible part, is shaped like a funnel that collects sound waves and sends them into the ear


80
canal where they hit the ear drum, which is a tightly stretched piece of skin that vibrates in
time with the wave. The ear drum starts a chain reaction and sends the vibration through
three little bones in the middle ear that amplify sound. Those bones are called the hammer,
the anvil, and the stirrup.
Furthermore, acoustic waves from a purely hydrodynamic point of view are small-
amplitude disturbances that propagate in a compressible medium (like a fluid) through the
interplay between fluid inertia, and the restoring force of fluid pressure. The propagation of
small-amplitude disturbances in homogeneous medium is observed as acoustic waves such
as water waves, and in self-gravitationally stratified medium like stellar atmosphere [36-37,
41-44], acoustic-gravity waves such as p-modes, g-modes, f-modes, etc., as found by helio-
and astero-seismological studies. Acoustic waves propagating through a dispersive medium
may get dynamically converted into solitons or shocks depending on the physical
mechanisms responsible for their saturation. When fluid nonlinearity (convective effect) is
balanced by dispersion (geometrical effect), solitons usually result [4]. Conversely, shocks
are formed if fluid nonlinearity is balanced by dissipation (damping effect). The nonlinear
hydrodynamic equations of various forms (like KdV equation, Burger equation, NLS
equation, BO equation, etc.) in the context of the generation, structure, propagation, self-
organization and dissipation of solitons or shocks have long been developed applying the
hydrodynamic views of the usual conservation laws of flux, momentum and energy [4].
Similar outlook is needed to understand the formation of other nonlinear localized
structures of low frequency acoustic waves like double layers, vortices, etc. They are
important in a wide variety of space, astrophysical and laboratory problems for the
investigation of dynamical stability against perturbation [3-4]. In addition, these equations
have wide applications to study a nonlinear, radial, energetic, and steady-flow problem that
provides a first rough approximation to the physics of stellar winds and associated acoustic
wave kinetics, which are responsible for stellar mass-loss phenomena via supersonic flow
into interstellar space [2].
Acoustic mode in plasmas of all types [2-44], similarly, is actually a pressure driven
longitudinal wave like the ordinary sound mode in neutral gas. In normal two-component
plasmas, the electron thermal pressure drives the collective ion oscillations to propagate as
the ion sound (acoustic) wave. Here the electron thermal pressure provides the restoring
force to allow the collective ion dynamics in the form of ionic compression and rarefaction
to propagate in the plasma background and ionic mass provides the corresponding inertial
force. Thermal plasma species (like electrons) are free to carry out thermal screening of the
electrostatic potential. In absence of any dissipative mechanism, the ion sound wave moves
with constant amplitude. For mathematical description of the ion sound kinetics, the plasma
electrons are normally treated as inertialess species and the plasma ions, with full inertial
dynamics. However, recent finding of ion sound wave excitation in transonic plasma
condition of hydrodynamic equilibrium offers a new physical scope of acoustic turbulence
due to weak but finite electron inertial delay effect [5-12]. Qualitative and quantitative
modifications are introduced into its nonlinear counterpart as well, under the same
transonic plasma equilibrium configuration [12]. The transonic transition of the plasma flow
motion quite naturally occurs in the neighborhood of boundary wall surface of laboratory
plasmas, self-similar expansion of plasmas into vacuum, in solar wind plasmas and different
astrophysical plasmas, etc. The self-similar plasma expansion model predicts supersonic
motion of plasma flow into vacuum. This model is widely used to describe the motion of
intense ion plasma jets produced by short time pulse laser interaction with solid target [17-

Acoustic Wave

81
23]. Recently, the self-similar plasma expansion into vacuum is modeled by an appropriate
consideration of space charge separation effect on the expanding front [13].
According to the recently proposed inertia-induced ion acoustic excitation theory [5-8], the
large-scale plasma flow motion feeds the energy to the short scale fluctuations near the pre-
sheath termination at sonic point. This is a kind of energy transfer process from large-scale
flow energy to wave energy through short scale instability of cascading type. In order to
maintain the turbulence type of hydrodynamic equilibrium, there must be some source to
feed large-scale flow and sink to arrest the infinite growth of the excited short waves. The
growing wave energy could be used to re-modify the global transonic equilibrium such that
the transonic transition becomes a natural equilibrium with smooth change in flow motion
from subsonic to supersonic regime. Of course, this is a quite involved problem to handle
the self-consistent turbulence theory of transonic plasma in terms of anomalous transport
[5]. Now one may ask how to produce such boundary layer with sufficient size of the
transonic plasma layer for laboratory experimentations?
This, in fact, is an experimental challenge to design and set up such experiments to produce
extended length of the transonic zone to sufficient extent to resolve the desired unstable
wave spectral components. Creation of a thick boundary layer of transonic flow dynamics
is, no doubt, an important task. This zone lies between subsonic and supersonic domains,
and is naturally bounded by low supersonic and high subsonic speeds. It should be
mentioned here that the sonic velocity corresponds to the phase velocity of the bulk
plasma mode of the dispersionless ion acoustic wave. In case of sheath edge boundary,
transonic layer could be probed by high-resolving diagnosis of the Debye length order.
The desired experiments of spectral analysis of the unstable ion acoustic waves in
transonic plasma condition may be quite useful to resolve the mystery of sheath edge
singularity. Using de-Lavel nozzle mechanism of hydrodynamic flow motion, experiments
could be designed to produce transonic transition layer of desired length and
characteristics [6-8].
Study of the ambient acoustic spectrum associated with plasma flow motion can be termed
as the acoustic spectroscopy of equilibrium homogeneous plasma flows [6, 26]. This may be
useful for expanding background plasmas [13], solar wind plasmas and also in space
plasmas through which the space vehicles motion and aerodynamic motion occur [3, 25,
28]. Basic principles of the acoustic spectroscopy have concern to the linear and non-linear
ion acoustic wave turbulence theory and properties of the transonic plasma equilibrium [5-
12, 26]. These properties may be used to develop the required diagnostic tools to study and
describe the hydrodynamic equilibrium states of plasma flows by suitable observations and
analysis of the waves and instabilities they exhibit. In fact, the ambient turbulence-driven
plasma flow is quite natural to occur in toroidal and poloidal directions of the magnetic
confinement of tokamak device. Similar physical mechanism is supposed to be operative in
the transonic transition behavior of equilibrium plasma flow motion [5-12]. Thorough
investigations of acoustic wave turbulence theory in transonic plasma condition will be
needed to explore transonic flow dynamics on a concrete footing.
Recently, there has been an outburst of interest in plasma states where the assumption of
static equilibrium practically is violated [28-30]. Great deals of research activities are now
going on in transonic and supersonic magnetohydrodynamic (MHD) flows in laboratory
and astrophysical plasmas. Similar activities are also important for understanding the
designing of supersonic aerodynamics having relevance in spacecraft-based laboratory


82
experimentations of space plasma research as well [8, 30]. This is also argued that future
tokamak reactors need the consideration of rotation of fusion plasma with high speeds that do
not permit the assumption of static equilibrium to hold good. This may be brought about due
to neutral beam heating and pumped divertor action for the extraction of heat and exhaust.
In astrophysics [3, 28-32, 35-44], the primary importance of plasma flows is revealed in such
diverse situations as coronal flux tubes, stellar winds, rotating accretion disks, torsional
modes, and jets emitted from radio galaxies. This is to argue that the basic understanding of
the acoustic wave dynamics in transonic plasma system constitutes an important subject of
future interdisciplinary research [5-12, 26-30]. This may be useful for development of the
appropriate diagnostics for acoustic spectroscopy to measure and characterize the
hydrodynamic equilibrium of flowing transonic plasmas [8-10]. Such concepts of acoustic
wave dynamics in a wider horizon may also applied to understand some helio- and astero-
seismic observations in astrophysical contexts.
Most of the plasma devices of industrial applications like dense plasma focus machine,
plasma torches, etc. depend on the plasma flows that violate the static equilibrium [26-30].
In fusion plasmas of future generation too, the static approximation of the equilibrium
plasma description may not be suitable to describe the acoustic wave behavior. In future
course of fusion research, rotational motions of fusion plasmas in poloidal and toroidal
directions may decide the equilibrium. This is important to state that in toroidal plasmas,
the geodesic acoustic mode becomes of fundamental importance in comparison to the
ordinary sound modes [30]. This may be more important when these rotational motions are
in the defined range of the transonic limit. Simplicity is correlated to the local mode
approximation of the acoustic wave description in transonic limit of uniform and
unidirectional plasma flow motion without magnetic field.
The lowest order nonlinear wave theory of the ion acoustic wave dynamics predicts that the
usual KdV equation is not suitable to describe the kinetics of the nonlinear traveling ion
acoustic waves in transonic plasma condition [8-9, 12, 26]. A self-consistent linear source
driven KdV equation, termed as d-KdV equation, is prescribed as a more suitable nonlinear
differential equation to describe the nonlinear traveling ion acoustic wave dynamics in
transonic plasma condition. By mathematical structure of the derived d-KdV equation, it
looks analytically non-integrable and physically non-conservative dynamical system [8-9].
Due to linear source term, an additional class of nonlinear traveling wave solution of
oscillatory shock-like nature is obtained. This is more prominent in the shorter scale domain
of the unstable ion acoustic wave spectrum, but within the validity limit of weak
nonlinearity and weak dispersion.
If there is multispecies ionic composition in a plasma system, varieties of plasma sound
waves are likely to exist depending on, in principle, the number of inertial ionic species. In
plasmas containing two varieties of dust or fine suspended particles, two distinct kinds of
natural plasma sound modes are possible [15-16]. Such plasmas, termed as the colloidal
plasmas [16], have become the subject of intensive study in various fields of physics and
engineering such as in space, astrophysics, plasma physics, plasma-aided manufacturing
technique, and lastly, fusion technology [14-23]. The dust grains or the solid fine particles
suspended in low temperature gaseous plasmas are usually negatively charged. It is also
observed that plasmas including micro scale-sized and nano scale-sized suspended particles
exist in many natural conditions of technological values. Such plasmas have been generated
in laboratories with a view to investigate the dust grain charging physics, plasma wave
physics as well as some acoustic instability phenomena.

Acoustic Wave

83
The two distinct sound modes, however, in bi-ion colloidal plasma are well-separated in
space and time scales due to wide range variations of mass scaling of the normal ions and
the charged dust grains and free electrons populations. The charged dust grains are termed
as the Dust Grain Like Impurity Ions (DGLIIs) [15] to distinguish from the normal impurity
ions. The present contributory chapter, additionally, applies the inertia-induced acoustic
excitation theory to nonlinear description of plasma sound modes in colloidal plasma [15-
16] under different configurations. Two separate cases of ion flow motion and dust grain
motions are considered. It is indeed found that the modified Ion Acoustic Wave (m-IAW) or
Dust Ion Acoustic (DIA) wave and the so-called (Ion) Acoustic Wave (s-IAW) or Dust Acoustic
(DA) wave both become nonlinearly unstable due to an active role of weak but finite inertial
correction of the respective plasma thermal species [15, 26]. Proper mass domain scaling of
the dust grains for acoustic instability to occur is estimated to be equal to that of the
asymptotic mass ratio of plasma electron to ion as the lowest order inertial correction of
background plasma thermal species. This contributory chapter is thus a review organized to
aim at some illustrative examples of linear and nonlinear acoustic wave propagation
dynamics through transonic plasma fluid, particularly, under the light of current scenario.
Some important reported findings on nonlinear acoustic modes found in space and
astrophysical situation [31-44], like in solar plasma system [10-11, 31-44], will also be
presented in concise to understand space phenomena. Incipient future scopes of the
presented contribution on transonic flow dynamics in different astrophysical situations will
also be briefly pointed out.
2. Physical model description
A simple two-component non-isothermal, field-free and collisionless plasma system under
fluid limit approximation is assumed. The plasma ions are supposed to be drifting with
uniform velocity at around the sonic phase speed under field-free approximation. Global
plasma equilibrium flow motion over transonic plasma scale length at hydrodynamic
equilibrium is assumed to satisfy the global quasi-neutrality. Such situations are realizable
in the transonic region of the plasma sheath system as well as in solar and other stellar wind
plasmas [3, 10-11, 35-41]. Its importance has previously been discussed [1, 5-12], where the
ion-beam driven wave phenomena are supposed to be involved in Q-machine or in
unipolar/ bipolar ion rich sheath formed around an electrode wall or grid in Double Plasma
Device (DPD) experiments of plasma sheath driven low frequency instabilities of relaxation
type [5]. The unstable situation is equally likely to occur on both the sides of the sheath
structures with plasma ion streamers [12].
3. Linear normal acoustic mode analyses
3.1 Basic governing equations
The basic set of governing dynamical evolution equations for the linear normal mode
behavior of fluid acoustic wave consist of electron continuity equation, electron momentum
equation, ion continuity equation, and ion momentum equation [5-6]. The set is closed by
coupling the plasma thermal electron dynamics with that of plasma inertial ion dynamics
through a single Poissons equation for electrostatic potential distribution due to localized
ambipolar effects. Applying Fouriers wave analysis for linear normal mode behavior of ion
acoustic wave over the basic set of governing dynamical evolution equations [5-6], the linear
dispersion relation is derived as follows


84

( )
( )
( )
te De
a
k v k
k v
2 2 2 2
2
2 2
0 2
1
. .
+
+ =
(1)
All the notations in the equation (1) are usual and conventional. Here is the Doppler-
shifted frequency of the ion acoustic wave,
a
is the ion acoustic wave frequency in
laboratory frame of reference, k is the angular wave number of the ion acoustic wave such
that k
De
is a measure of the acoustic wave dispersion scaling and v
te
is the electron thermal
velocity. Now the kinematics of any mode can be analyzed in two different ways: one in lab-
frame and the other, in Doppler-shifted frame of reference. This is to note that the obtained
dispersion relation differs from those of the other known normal modes of low frequency
relaxation type of instability, ion plasma oscillations and waves. This is due to the weak but
finite electron inertial delay effect in the dispersion relation of the wave fluctuations. This is
mathematically incorporated by a weak inertial perturbation over electron inertial dynamics
over the leading order solution obtained by virtue of electron fluid equations neglecting
electron inertial term.
It is thus obvious from the mathematical construct of equation (1) that the LHS is a non-
resonant term whereas RHS is a resonant term. The RHS gets artificially transformed into a
resonant term if and only if
i
k v
0
. 0 < . Now, it can be inferred that equation (1) represents a
resonantly unstable situation at Doppler shifted resonance frequency of
a
k v
0
. , if
and only if k v
0
. <0. This means that only the mode counter moving with respect to the
plasma beam mode gets resonantly unstable. The resonance growth rate for this resonant
instability [5-6] is found to be of the following form

( ) ( )
i
a De
e
m
k k v
m
1
2 2 2
0
2 1 . . = + (2)
This is important to add that the resonance condition required by equation (1) dictates the
propagation direction of the unstable ion acoustic wave (counter moving with respect to
plasma ion streams) at reduced frequencies. It is clear from equation (2) that there is the
physical appearance of two distinct classes of eigen mode frequencies of the resonantly
coupled mode-mode system of linearly growing ion acoustic oscillations in lab-frame: near-
zero frequency (standing mode pattern) and non-zero frequency (propagating mode
pattern). These two distinct eigen modes are generated by the process of repeated Doppler-
shifting of the ion acoustic wave frequency under the unique mathematical compulsion of
the hydrodynamic tailoring of the electron fluid density perturbation over ion acoustic time
scale. The unstable condition decides the resonant acoustic excitation threshold value for the
onset of the instability in terms of normalized value of the eigen mode frequency of the
acoustic fluctuations.
3.2 Graphical analysis
It is well-known that the graphical method is a more informative, simple and quick tool for
analyzing the stability behavior of a plasma-beam system even without solving dispersion
relation. To depict the clear-cut picture of the poles, relation (1) is rewritten as,
( )
( ) ( ) ( )
a
te
De
F k k v
k k v k v
2
2 2
2 2 2 2 2
0 0
1 1
, .
1 . .
(
= = (
+ + + (

(3)

Acoustic Wave

85
It is clear from the equation (3) that two poles are possible to exist in -space at 0 = and
k v
0
. = for k v
0
. 0 < . According to graphical method, the beam-plasma system will exhibit
instability only when the curve of ( ) F k , versus has multiple singular values in -
space having finite minima in between the two successive singularities, which do not
intersect with the line ( ) ( )
De
F k k
2 2
, 1 1 = + . The required condition for minimization of
( ) F k , in -space can be obtained by equating dF d 0 = . Now this condition, when
applied to equation (3), results into the following equality to derive the value of where
dispersion function is supposed to be minimum
( ) ( )
a a
k v
2 2 2
0
. 0. + + = (4)
In principle, equation (4) is to be solved to determine the value of . This is obvious to note
that this equality is satisfied at resonance value of
a
k v
0
~ . ~ for k v
0
. 0 < . Now to
indemnify the complex nature of , the functional value of ( ) ( )
De
F k k
2 2
, 1 1 > + . This can,
however, be further simplified to yield the following inequality to determine the threshold
value for the onset of the inertia-induced instability

( ) ( ) ( )
te a De
k v k v k
2
2 2 2 2 2 2
0
. 1 . > + (5)
The threshold condition for the instability is satisfied for equality sign at resonance
frequency
a
k v
0
~ . ~ that characterizes the case of a marginal instability. A few typical
plots of the function ( ) F k , in -space for shorter and longer acoustic wavelengths
(perturbation scale lengths) are represented in Fig. 1.
3.3 Numerical analysis
Numerical techniques for solving polynomials over years have developed to a vast extent
for solving polynomials even with complex coefficients and complex variables. For the
present case, the Laguerre's algebraic root-finding method [6] to solve the normalized form
of polynomial equation has been used. The polynomial
( )
P
'
in the normalized form of the
dispersion relation (1) in ion-beam frame is given below

( )
P a a a a a
' ' ' 2 ' 3 ' 4
0 1 2 3 4
0. = + + + + = (6)
Here all the normalized notations used are usual, generic and defined by
pi a a pi De te te s i e
k k v v c m m
' ' '
, , ' = = = = = and
s
M v c
0
= . The expressions for
the various coefficients in the polynomial
( )
P
'
are defined as follows
( )( )
te a
a k k v
'2 '2 ' 2 '2
0
1 = + ,
a
1
0 = ,
( ) ( )( )
te
a k M k k v
2
2 2 2
2
'. 1 ' ' ' = + + ,
a k M
3
2 '. = , and


86
a
4
1 = .
It is found that out of four possible roots of
( )
P
'
, only two roots are complex and these are
the complex conjugates as a pair. For all the complex conjugated roots, only the complex
root with positive imaginary part is useful, since this determines the growth rate of the
instability. Real and imaginary parts of the corresponding complex roots are then plotted as
shown in Figs. 2 and 3, respectively. Numerical characterization of the unstable mode of the
instability clearly depicts the resonant character of the electron inertia-induced resonant
acoustic instability [5].
3.4 Evaluation of wave energy
This is important to evaluate the wave energy in order to have a more complete picture of
the basic source mechanism of the discussed instability. In presence of the beam, it is
expected that one of the modes involved, has positive energy and the other has negative
energy. The dispersion relation (1) can be put in the laboratory frame for a more clear
identification and characterization of the positive and negative energy modes in the form of
dispersion function ( ) k , as follows
( )
( )
pe pi
De te
k
k k v
k v
2 2
2 2 2 2 2
0
1
, 1 1 0.
.

| |
= + + = |
|

\ .
(7)
The average electric field energy stored in a propagating electrostatic (created by ambipolar
effect) wave in a medium is given by the following relation [6]
( ) ( ) ( ) W k E k k
2
0
1
, , ,
2
= (

. (8)

Here
0
is the dielectric constant of free space, ( ) E k , is the electric field amplitude of the
ion acoustic fluctuations and ( )
E
W E k
2
1 2 , = is the corresponding counterpart of electric
energy of the acoustic fluctuations through free space. Applying the equations (7) and (8),
the following can explicitly be derived

( )
( )
pi
E De te
k W
W k k v
k v
2
2
3 2 2 2 2
0
2 , 2
.
.
= = + (
(

(9)
Now, clearly, it is evident that the second term of equation (9) contributes negative energy
value to the defined wave-plasma system. This occurs as because the sign of this term
becomes negative for the values of k v
0
. < , which is the case for the reported instability.
From a few typical plots in Fig.4, one can notice that the total wave energy suffers a sharp
transition from negative to positive values at resonance frequency point of zero energy
value. The resonance point lies in the domain of near-zero and non-zero frequencies in lab-
frame. According to conventional definition and understanding, the wave energy
expression in equation (9) classifies the near-zero frequency mode as the negative energy
mode. Then immediately the non-zero frequency mode may be classified as the positive
energy mode.

Acoustic Wave

87
This is important to clarify that the theoretical concept of near-zero frequency mode is an
outcome of the mathematical construct of weak but finite electron inertial response to the
ion acoustic wave fluctuations. The blowing up character, as shown in Fig. 4, of the total
wave energy in opposite directions suggests referring the discussed instability to as an
'explosive instability' in accordance with the law of conservation of energy. It signifies the
transonic plasma condition with the resonant mode-mode coupling of the positive and
negative energy modes. The time average of the hydrodynamic and wave potential energies
of the considered wave-plasma system over the growth time scale is conserved during the
energy exchange process between the unstable resonant eigen modes and the main source of
ion flow dynamics. These two modes are clearly identified from equation (9) as the natural
resonant modes of the defined plasma system that undergo linear resonant mode-mode
coupling to produce the defined wave instability.
3.5 Estimation of quenching time
Under the cold ion approximation, even the small electrostatic potential will be able to
distort the ion particle motion and associated trajectories, affecting the driving source flow
velocity of the resonant instability under consideration. In wave frame, the streaming ion
energy ( )
i
E can be expressed by the following relation

i i
E m v
k
2
0
1
2
| |
=
|
\ .
. (10)
For ( )
o s
v k c v
0
~ >> , which is a valid case for the considered instability [5], the
condition for ion orbit distortion becomes of the following form,

w i
W mn v
2
0 0
1
2
. (11)
From this condition, the quenching time is estimated under the assumption that the wave
amplitude grows sufficiently from thermal noise level to physically measurable level such that
( )
t
E i
W t We
= . (12)
Here
i
W is the initial energy of the acoustic wave amplitude, which is of the order of the
thermal fluctuations, i.e.,
i e De
W T
3
~ and is the unnormalized linear growth rate. Using the
resonance values of
s
k c v
0
= and
s
k v kc
0
. ~ as derived in [5] for long wavelength case
of resonant mode, equation (12) for the quenching time with the help of (9) can be
rewritten as follows

( ) ( )( )
e
De De
i De
m M
M n k
m k M
2
1 2
3 2 2
0
1 1
1 ln
2 4 1

(
=
(
(

. (13)
For some typical plasma parameters in hydrogen plasma,
( )
De
n
3
0
ln ~15 30 . For
De
k ~0.3, 0.1, 0.05 near resonant M as in Figs. 2 and 3, equation (13) gives 1 > , i.e.,
q pi
>
. This physically means that the resonant growth time scale is greater than that of the plasma
ion oscillation time scale. Thus the resonant nature of the instability is observable in the
present analysis.


88
3.6 Physical consequences
Wave energy analyses are carried out to depict the graphical appearance of poles (Fig. 1),
nature of real parts of the roots (Fig. 2), nature of imaginary parts of roots (Fig. 3) and
positive-negative energy modes (Fig. 4).

Fig. 1. Graphical appearance of resonance poles as a variation of the dispersion function
( ) F k , with normalized Doppler-shifted frequency for dispersion scaling (a)
De
k 0.3 = , (b)
De
k 0.1 = , and (c)
De
k 0.01 =

Fig. 2. Variation of the real part of the normalized Doppler shifted eigen mode frequency
( )
with respect to Mach number ( ) M for different values of 0.30, 0.10, 0.05
De
k =
It is found that the instability arises out of linear resonance mode-mode intermixed coupling
between the negative and positive energy modes. The total energy of the coupled mode-
mode system comprising of hydrodynamical potential energy and wave kinetic energy,
however, is in accordance with the law of conservation of energy in the observation time
scale on the order of ion acoustic wave time scale. Identification and characterization of the
resonance nature of the said instability through transonic plasma is presented in order to
explore the acoustic richness in terms of collective waves, oscillations and fluctuations. This
is an important point to be mentioned here that the same type of instability features are

Acoustic Wave

89
expected to happen in plasma-wall interaction process and sheath-induced instability
phenomena in other similar situations as well.
There are different sorts of analytical and numerical tools for studying the linear instabilities
in a given plasma system. Energy method, based on energy minimization principle and the
normal mode analysis, based on equilibrium perturbations are the two basic mathematical
tools for analyzing the stability behavior of the given plasma systems. However, the latter is
most popular and simple for common use in analyzing the threshold conditions of the
instabilities and their growth rates. In the normal mode analysis, a linear dispersion relation
is derived which can be put in the form of a polynomial with real or imaginary coefficients.
The limitation of the analytical method depends upon the degree of the polynomial.
Computational technique broadly takes into account two ways of investigating instability.
First, an unstable mode can be deduced by the derived dispersion relation. The obtained
polynomial is then solved to delineate the complex roots having concern to the desired
instability. Second, a more comprehensive computational method involves solving for the
time dependent solution. Simulation technique used to solve the basic set of equations is
supposed to give more complete picture of the space and time evolution of the wave
phenomena. However, there is another very informative and simple method for analyzing
the derived dispersion relation to predict for the unstable behavior of the plasma system
under consideration. This is the graphical method in which the dispersion relation is
graphically represented for different values of resonance characterization parameters.
Source perturbation scale length ( )
De
k and deviation from sonic point ( ) M 1 are the
characterization parameters for the defined acoustic resonance.

Fig. 3. Variation of the normalized growth rate of the electron inertia-induced resonant
acoustic instability with Mach number for (a) 0.3
De
k = , (b) 0.1
De
k = and (c) 0.05
De
k =
showing that transonic plasma is rich in wide range acoustic spectral components and
hence, an unstable zone
This is quite natural and interesting to argue that the transonic plasma condition offers a
unique example where the physical situation of localized hydrodynamic equilibrium of
quasi-neutral plasma flow dynamics exists. Previous publication reports that the transonic
plasma layer, assumed to have finite extension, can be considered as a good physical
situation to study the acoustic instability, wave and turbulence driven by electron inertia-
induced ion acoustic excitation physics.


90

Fig. 4. Explosive nature of the electron inertia-induced ion acoustic wave instability as an
outcome of an interplay for the linear resonant mode-mode coupling of positive and
negative energy eigen modes. It shows how the normalized wave energy varies with
normalized frequency under a set of fixed values of Mand
De
k as (a) 0.85, 0.30
De
M k = = ;
(b) 0.842, 0.100
De
M k = = ; and (c) 1.79, 0.01
De
M k = =
In the present sections of the chapter, many features about the electron inertia-induced ion
acoustic wave instability are observed. For example, we physically identify and demonstrate
the following features of the instability obtained by theoretical and numerical means of
analysis of the desired dispersion relation:
1. The transonic plasma layer is an unstable zone of hydrodynamic equilibrium of
quasineutral plasma gas flow motion,
2. The instability is an outcome of the linear resonant mode-mode coupling of positive
and negative energy modes,
3. The quenching time of the instability is estimated for some typical values of plasma and
wave parameters as mentioned in the previous section. It is found to moderately exceed
the ion plasma oscillation time scale, and
4. Lastly, this indicates that in lab frame observation the unstable mode of ion acoustic
wave fluctuations at reduced frequencies may look like a purely growing mode. This is
very likely to occur for almost entire unstable frequency domain of the frequency
transformed ion acoustic waves.

Acoustic Wave

91
In fact, the electron inertial responses naturally appear only at electron oscillation frequency.
However, the transonic plasma condition creates a natural physical situation for the same to
occur even at the ion acoustic wave frequency of the transformed reduced values. The linear
process of resonant mode-mode coupling produces this and makes the coupled system of
wave modes unstable.
We have identified and demonstrated the following features of the instability obtained by
theoretical and numerical analysis of the dispersion relation: (i) The transonic plasma layer
is indeed an unstable zone of hydrodynamic equilibrium of quasi-neutral plasma gas flow
motion. (ii) The instability is an outcome of linear resonant mode-mode coupling of positive
and negative energy modes. (iii) The normalized values of Doppler-shifted resonant
frequencies of the unstable ion acoustic wave fluctuations in ion beam frame come out to be
almost equal to 0.5. (iv) The estimated quenching time of the instability exceeds the ion
plasma oscillation time scale moderately and hence, (v) In the lab-frame, the unstable modes
of ion acoustic wave fluctuations at reduced frequencies may look physically like a purely
growing mode.
This is further argued that the physical insights as listed above can be useful as theoretical,
graphical and numerical recipes to (1) formulate and solve the problems of saturation
mechanisms of the unstable ion acoustic wave fluctuations, (2) formulate and solve the
problems of the ion acoustic wave turbulence, and (3) design and setup experiments to
study the basic physics of linear and nonlinear ion acoustic wave activities in unique
transonic plasma system. These investigations may be useful to improve the existing
conceptual framework of physical and mathematical methods of two-scale theory of plasma
sheath research to resolve the long-term mystery of the sheath edge singularity. These, in
brief, are added to judge the didactic vis--vis the scientific qualities of the current research
work too much specialized in the subject of ion acoustic wave physics.
3.7 Comments
The main conclusive comment here is that the graphical method successfully explains the
unstable behavior of the fluid acoustic mode of the ion acoustic wave fluctuations in drifting
plasmas with cold ions and hot electrons. A more vivid picture of linear resonant mode-
mode coupling of positive and negative energy waves is obtained. This is important to note
that simple formulae for wave energy and quenching time calculations [6] are used. This
calculation further confirms the earlier results of stability analysis of drifting plasmas
against the acoustic wave perturbations [5]. It is, therefore, reasonable to think of logical
hypothesis of wave turbulence model approach to solve the sheath edge singularity problem
[1, 4]. Actually, the local normal mode theory of the discussed instability implies that the
entire transonic plasma zone should be rich in wide frequency range spectrum of the ion
acoustic wave fluctuations. This leads to develop the conceptual framework of situational
definition of the Debye sheath edge to behave as a turbulent zone with finite extension [12].
This hypothetical scenario of the transonic plasma condition can be examined by
appropriate experiments of measuring wide range spectral components of the ion acoustic
wave fluctuations.
This is a nontrivial problem to explicitly characterize the turbulent properties of the
transonic region. The more realistic problem of wave turbulence analysis demands the self-
consistent consideration of flow induced quasi-neutral plasma with inhomogeneity in
equilibrium plasma background. Similar situations are likely to occur in stellar wind
plasmas, where, the transonic behavior is brought about by deLaval nozzle mechanism [6-10]


92
of gas flow through a tube of varying cross section. Recent experimental observation [12] in
double plasma device (DPD) reports an instability even in a condition of symmetric
bipotential ion-rich sheath case. Its frequency falls within zero frequency range and its
source is believed to lie in presheath.
Finally, in a nutshell, it is concluded that the graphical method of analyzing the dispersion
relation of the inertia-induced instability offers a simple and more informative method of
practical importance in transonic plasma equilibrium. Moreover, the plasma environment of
Debye sheath edge locality offers a realistic situation for self-excitation of the ion acoustic
wave turbulence through resonant ion acoustic wave instability. This is induced by
hydrodynamic tailoring of the ion acoustic wave-induced electron density fluctuations. Of
course, no experimental observation of instability in transonic plasma has yet been reported
to directly compare with the theoretical results. However it cannot be undermined in
understanding wave turbulence phenomenon of flowing plasmas. This is informative to add
that the frequency and amplitude transformation of the normal ion acoustic wave into
unstable ion plasma wave at higher frequency is reported in high intense laserplasma
interaction processes [6-7] through the nonlinear ponderomotive action. This leads to the
formation of soliton, double layers, etc. through the saturation mechanism of strong laser-
plasma interaction processes due to non-zero average value of the spatially varying electric
field associated with laser pulse.
4. Nonlinear normal acoustic mode analyses
A large amount of literature of theoretical and experimental investigations has been
produced on the solitary wave propagation in plasmas since the theoretical discovery of ion
acoustic soliton [4, 11-12, their references]. Varieties of physical situations of drifting ions of
high energy with [5-12] and without [13-33] electron inertial correction have been
considered in the ion acoustic wave dynamics. It is shown that the electron inertial motion
becomes more important than the ion relativistic effect. Such situations exist in Earth's
magnetosphere, stellar atmosphere and in Van Allen radiation belts [3]. Similar studies have
been carried out in plasmas with additional ion beam fluid with full electron inertial
response in motion [12 and references].
A number of experiments were performed in the unstable condition of beam plasma system
in laboratory in order to observe soliton amplification [12]. There are many theoretical
calculations and experiments on linear [7-8] and nonlinear [9-11] wave propagation
properties of acoustic waves to see their behavior near the transonic point. For an assumed
transonic region, it has been theoretically shown that the small amplitude acoustic wave
fluctuations exhibit linear resonant growth of relaxation type under the consideration of
weak but finite electron inertial delay effect [12-13]. In contrast to earlier claim [3] that the
complex nature of coefficients in KdV equation prevents the soliton formation, we argue
that their interpretation seems to be physically inappropriate. Instead, by global phase
modification technique [12], we show that the usual soliton solution exists (even under the
unstable condition), but only for infinitely long wavelength source perturbations.
Otherwise, oscillatory shock-like solutions are more likely to exist.
Under fluid approximation, the self-consistently closed set of basic dynamical equations
for transonic plasma system with all usual notations in normalized form is given as
follows

Acoustic Wave

93
Electron continuity equation:

e
e
v
v
t x x
. 0

+ + =

, and (14)
Electron momentum equation:

e e e e
e
i e
m v v n
v
m t x x n x
1
.
| |
+ =
|

\ .
. (15)
This is to remind the readers that equation (15) is obtained by substituting zero-order
solution of Boltzmann electron density distribution into the normal electron continuity
equation. In fact, in the asymptotic limit of
e i
m m 0 , electron continuity equation as such
is redundant as because the left hand side (electron inertial effect) of (15) is ignorable.
Equation (14) basically offers a scope to introduce the weak but finite role of electron to ion
inertial mass ratio on the normal mode behavior of acoustic wave.
Ion continuity equation:
( )
i
i i
n
n v
t x
0,

+ =

(16)
Ion momentum equation:

i i
i
v v
v
t x x

+ =

, and (17)
Poisson equation:

e i
n n
x
2
2

=
. (18)
Following form of the derived d-KdV equation obtained from the above equations by the
standard methodology of reductive perturbation [12] describes the nonlinear ion acoustic
wave dynamics under transient limit (~soliton transit time scale) in a new space defined by
the stretched coordinates ( ) , . This is to mention that ( ) ( ) x t e , ,

= and 0
under the transient time action of the propagating ion acoustic soliton through transonic
plasma
K M K
t x x
3
0 0 0 3
1
2

+ + =

. (19)
Here the notations K
0
and M
0
termed as complex response coefficients [11-12, 26] and the linear
resonant growth rate () of the ion acoustic wave with complex Doppler-shifted Mach
number
D Dr Di
M M iM = + and lab-frame Mach number
r i
M M iM = + in transonic
equilibrium appearing in equation (19) are as follows,
K A B
1 2
2 2
0
( = +

where,


94
( )
r Dr Dr i
m
Dr i
M M M M
A
M M
3 2
3
2 2
3
| |
|
= +
|
+
\ .
, and
( )
i i Dr i
m
Dr i
M M M M
B
M M
3 2
3
2 2
3
| |
|
= +
|
+
\ .
,
M C D
1 2
2 2
0
( = +

where,
( )
{ }
( )
( )
Dr i Dr i
r i r i
m
Dr i
M M M M
M M M M
C
M M
2
2 2 2 2 2
2 2 2 2
4 2
2 2
3 4
4
1
1
2
(

(
=
(
+
(

, and
( )
( )
( )
Dr i Dr i r i r i
m
Dr i
M M M M M M M M
D
M M
2 2 2 2
4 2
2 2
12 4
1
2

= +
`
+
)
,
( )
i
De i
e
m
k v
m
1 2
0
2 1
| |
=
|
\ .
.
The notations are usual and generic as discussed earlier [12]. In the system, plasma ions are
self consistently drifting or streaming through a negative neutralizing background of hot
electrons having relatively zero inertia. The time response of the electron fluid here is
normally ignored. As a result, the unique role of weak but finite electron inertia to
destabilize the plasma ion sound wave in transonic plasma equilibrium even within fluid
model approach of normal mode description is masked.
Now equation (19) after being transformed into an equivalent stationary ODE form by the
Galilean transformations is numerically solved as an initial value problem. Some very small
simultaneous values of , and
2 2
are required for the numerical programme
to proceed. A few numerical plots for the desired nonlinear evolutions are shown in Figs. 5-
6. This is to note that the calculated amplitudes (as shown in Figs. 5a-6a) are the solutions of
the present d-KdV equation (19) with bounded and unbounded phase potraits (as shown in
Figs. 5b-6b). Now, the actual amplitudes of the resulting solutions can be deduced by
multiplying the numerically obtained values with ( )
DE
k
2
2
~ 10

[12]. In principle, the
parameter is an arbitrary smallness parameter proportional to the dispersion strength or
the amplitude of the weakly dispersive and weakly nonlinear plasma wave.
The unique motivation here is to characterize the possible nonlinear normal mode structure
of ion acoustic fluctuations under unstable condition of the ion drifts [8-9,12]. By this very
specific example, we show that the complex nature of the coefficients of the derived KdV
equation in the unstable zone of transonic plasma doesn't prevent the existence of localized
nonlinear solutions including usual soliton solution, too. The concept of global phase

Acoustic Wave

95
modification technique (DPMT) [11-12, 26] results into a d-KdV equation [8-9, 12] with
variable nonlinear and dispersion coefficients.
Two distinct classes of solutions are obtained: soliton and oscillatory shock-like structures.
Amplification and damping of the driven KdV soliton over the usual KdV soliton is noted
for extremely large wavelength (dc) acoustic driving in source term as shown in Figs. 5-6.
The amplification near resonance is associated with considerable reduction in nonlinear
coefficient than unity as confirmed by numerical calculation. In other cases of shorter
acoustic driving in source term as shown in Figs. 5-6, nonlinear solutions of oscillatory
shock-like nature are obtained depending on the small deviation from resonant values. It is
clearly seen that the peaks of oscillatory shock-like solutions are of either sinusoidal or non-
sinusoidal nature with continuous elevation of the initial values of the successive peaks
beyond the main nonlinear acoustic peak.
Most of the experimental results in Double Plasma Device (DPD) are reported to show that
the obtained theoretical results may have practical relevance to understand the basic physics
of ion acoustic wave activities in the transonic region [12] as in Fig. 7. The experiment is
performed in a DPD of 90 cm in length and 50 cm in diameter equipped with multi-dipole
magnets for surface plasma confinement [12]. The chamber is divided into source and target
by a mesh grid of 85% transparency kept electrically floating. It is evacuated down to a
pressure of ( )
5
5 6 10
Pa with a turbomolecular pump backed by a rotary pump. Ar-gas

is bled into the system at a pressure ( )
2
3 5 10
Pa under continuous pumping condition.

The source and target plasmas are produced by dc discharge between the tungsten filament
of 0.1 mm diameter and magnetic cages.

(a) (b)
Fig. 5. Profile of (a) ion acoustic potential ( ) with normalized space variable ( ) , and (b)
phase space geometry of ion acoustic potential in a phase space described by and ( )

with
8
2.5 10
De
k

= (fixed) for Case (1):
7
1.0 10 ,

= Case (2):
7
2.5 10 ,

= Case (3):
7
5.0 10 ,

= and Case (4):
7
7.5 10

=
The plasma parameters are measured with the help of a plane Langmuir probe of 5 mm
diameter and Retarding Potential Analyzer (RPA) of 2.2 cm in diameter. The probe and the
analyzer are movable axially by a motor driving system so as to take data at any desired
-15 -10 -5 0 5 10 15
0
0.5
1
1.5
2
2.5
3
Ion acoustic potential profile in coordination space
Normalized position
I
o
n

a
c
o
u
s
t
i
c

p
o
t
e
n
t
i
a
l
Case (1)
Case (2)
Case (3)
Case (4)
0 0.5 1 1.5 2 2.5 3
-1.5
-1
-0.5
0
0.5
1
1.5
Ion acoustic potential profile in phase space
Ion acoustic potential
I
o
n

a
c
o
u
s
t
i
c

p
o
t
e
n
t
i
a
l

g
r
a
d
i
e
n
t
Case (1)
Case (2)
Case (3)
Case (4)


96
position. The plasma parameters are: electron density n cm
8 9 3
0
10 10 = , electron
temperature
e
T eV 1.0 1.5 = and ion temperature
i
T eV 0.1 = . An ion-acoustic wave is
excited with a positive ramp voltage of which the rise time is controllable and is applied to
the source anode of the system. Propagating signals are detected by an axially movable
Langmuir probe which is biased to V 4 + with respect to the plasma potential in order to
detect the perturbation in the electron current saturation region. The current is then
converted into voltage by a resistance of 100 and the resultant signals are fed to the
oscilloscope. The probe surface is repeatedly cleaned with ion bombardment by applying
V 100 to it for a short time scale.

(a) (b)
Fig. 6. Same as Fig. 5 but with
1
1.0 10
De
k

= (fixed) for Case (1):
5
1.0 10 ,

= Case (2):
5
2.0 10 ,

= Case (3):
5
3.0 10 ,

= and Case (4):
5
5.0 10

=
.

Fig. 7. Experimental profiles of variation of plasma density perturbation ( ) n against time
( ) t at different position of the probe from the grid is shown. Along the x-axis, each division
represents 10 s and along the y-axis, the density perturbation scale is given as 0.09
e
n n =

-15 -10 -5 0 5 10 15
0
0.5
1
1.5
2
2.5
Ion acoustic potential profile in coordination space
Normalized position
I
o
n

a
c
o
u
s
t
i
c

p
o
t
e
n
t
i
a
l
Case (1)
Case (2)
Case (3)
Case (4)
0 0.5 1 1.5 2 2.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Ion acoustic potential profile in phase space
Ion acoustic potential
I
o
n

a
c
o
u
s
t
i
c

p
o
t
e
n
t
i
a
l

g
r
a
d
i
e
n
t Case (1)
Case (2)
Case (3)
Case (4)

Acoustic Wave

97
It is also seen that the value of
Dr
M 1 = where resonance occurs remains invariant to
spectral variation in source term even by orders of
1 10
1.0 10 1.0 10

. The nonlinear and
dispersive coefficients exhibit very sensitive role on even slight variation of from its
resonance. It is noted that as the value of 0 = , the usual KdV soliton is recovered,
irrespective of any wave number value in the source term. The source term plays an
effective role only when finite
De
k and -values are assigned simultaneously
4.3 Comments
As per experimental observations, oscillatory shocks are reported to emerge from the
transonic zone in the target plasma as shown in Fig. 7. One can qualitatively argue that as
soon as the solitary wave passes through the unstable transonic zone, it may experience the
transient phase modifications leading thereby to the formation of oscillatory shock. The
observed damping of the oscillatory shock may be correlated to the non-resonant type of
dissipation through phase incoherence among ion acoustic spectral components of the usual
solitary wave. It seems to be more plausible to argue that the input energy to the usual
soliton due to transonic plasma equilibrium may be shared among different spectral
components through adiabatic energy exchange processes. This is concluded here that the
complex coefficients of the KdV equation should, in principle, not become the criterion for
the non-existence of localized nonlinear solutions including usual solilton, too. But the usual
soliton solution exists only for infinitely long wavelength source perturbation. This
conclusion is derived subject to the validity condition of our arguments of global temporal
phase modification of usual soliton amplitude under unstable condition of the plasma
medium. The unstable condition of the medium may cause structural deformation of the
non-driven KdV solution. Such deformations may result into sinusoidal (linear) or non-
sinusoidal (nonlinear) peaks of oscillatory shock-like solution depending on the wavelength
of the source perturbations [8-9, 12].
Applying the wave packet model for a moving soliton leaving behind an acoustic tail of
dispersive waves known as precursor or acoustic wind (in soliton frame), the asymmetry
can be associated with elevation of the bottom potential by a finite dc value superposed with
periodic repetition of linear or nonlinear peaks. The amplification or suppression of a single
soliton can be possible only for infinitely long wavelength (dc) source. For shorter
wavelength source driving, the transition from usual soliton solution to oscillatory shock-
like solutions is more likely to occur. It is, in brief, concluded that the present mathematical
study of d-KdV equation offers a significant contribution of analytical supports to our
numerical prediction of structural transformation of the traveling nonlinear ion acoustic waves
in transonic plasma equilibrium of desired quality. It clearly shows that the actual solution of
d-KdV equation is a resultant of linear mixing (superposition) of soliton and shock both.
Dominating features of the individual nonlinear modes is decided by an appropriate choice
of the specific values of unstable wave number (or wavelength) for a given value of the ion
flow Mach number. It is obvious to note that in zero growth limit of d-KdV equation, the
shock-term disappears and only soliton remains. This limit is correlated with dc range of the
chosen unstable wave number of quite weaker dispersion strength. As the dispersion
strength becomes significant to influence the original soliton strength of weak nonlineariy
and weak dispersion in the defined transonic plasma of finite extension, structural
modification of the usual KdV soliton profile occurs.
We further argue that the linear and nonlinear normal mode behaviors of the ion acoustic
waves in transonic plasma condition differ qualitatively from those derived for static and


98
dynamic equilibriums without electron inertial correction. The finite but weak hydrodynamic
tailoring of the electron fluid motion on ion acoustic space and time scales brings about this
difference. It is then argued that the plasma flows in transonic equilibrium should exhibit rich
spectrum of linear and nonlinear ion acoustic waves and oscillations. Of course, under
Vlasov model the hot electrons with streaming velocity comparable to the phase speed of
the ion sound wave may destabilize the ion sound mode through wave-particle resonance
effect [8 and references] too. However, our excitation mechanism of ion sound wave differs
from the other known mechanisms [8]

to excite the same ion sound wave on many grounds
[8]. This kind of theoretical scenario of transonic plasmas offers a unique scope of acoustic
spectroscopy to describe the internal state of transonic equilibrium of plasma flows [28].
These calculations have potential applications [26] extensively to understand plasma
acoustic dynamics in colloidal plasmas too, but under transonic equilibrium configuration.
A generalized statement thereby is reported that all possible sound modes in multi-species
colloidal plasmas with drift motions (of inertial ionic species) could be destabilized by the
inertial delay effect of the corresponding plasma thermal species that carry out thermal
screening of acoustic potential developed due to respective inertial ionic species. Of course,
threshold values may differ depending on the choice of the plasma sound mode under
consideration. In technological application point of view, one may argue that the proposed
theoretical model for inertia-induced acoustic instability mechanism may be utilized to
make a plasma-based micro device for acoustic amplifier [26]. The amplified acoustic signals
(developed due to respective inertial ionic species) from the amplifier could be detected,
received and analyzed for the diagnosis and characterization of hydrodynamic flow of
plasmas with embedded inertial dust contaminations. These analyses may have potential
applications in different ion acoustic wave turbulence-related situations like aerodynamics,
solar wind and space plasmas, fusion plasmas, industrial plasmas and plasma flows in
astrophysical context, etc.
5. Astrophysical normal acoustics
A plasma-based Gravito-Electorstatic Sheath (GES) model is proposed to discuss the
fundamental issues of the solar interior plasma (SIP) and solar wind plasma (SWP). Basic
concepts of plasma-wall interaction physics are invoked. Here the wall is defined by a
continuous variation of gravity associated with the SIP mass. The neutral gas approximation
of the inertially confined SIP is relaxed, and as such the scope of quasi-neutral plasma
sheath formation is allowed to arise near the self-consistently defined solar surface
boundary (SSB). Analytical and numerical results are obtained to define the SSB and discuss
the physics of the surface properties of the Sun, and hence, those of the SWP.
5.1 Physical model description
The SIP system can be idealized as a self-gravitationally bounded quasi-neutral plasma with
a spherically symmetric surface boundary of nonrigid and nonphysical nature. The self-
gravitational potential barrier of the solar plasma mass distribution acts as an enclosure to
confine this quasi-neutral plasma. An estimated typical value ~10
-20
of the ratio of the solar
plasma Debye length and Jeans length of the total solar mass justifies the quasi-neutral
behavior of the solar plasma on both the bounded and unbounded scales. Here the zeroth-
order boundary surface can be defined by the exact hydrostatic condition of gravito-
electrostatic force balancing of the enclosed plasma mass at some arbitrary radial position

Acoustic Wave

99
from the center of the mean solar gravitational mass. With this much background in mind,
let us now formulate the problem of the physical and mathematical descriptions GES
formation around the SSB. For simplicity, we consider spherical symmetry of the inertially
confined SIP mass, which helps to reduce the three dimensional problem of describing the
GES into a simplified one dimensional problem in the radial direction. Thus, only a single
radial degree of freedom is required for description of the dynamical behavior of the SWP
under the assumed spherically symmetric self-gravitating solar plasma mass distribution.
The idea of the GES formation can be appreciated with quantitative estimates of the gravito-
thermal coupling constants for the SIP electrons and hydrogen ions. Henceforth, ions and
hydrogen ions will be used in the sense of the same ionic species. These parameters [10]
can be defined and estimated as follows: The gravitothermal coupling constant for electrons
can be estimated as
e B e e
k T m g R 10

= , for a mean electron temperature of
e
T
5
~ 10 K
and as
e
800 for mean
e
T
6
10 = K. The notation
B
k (=1.380610
-23
JK
-1
) denotes the
Boltzmann constant. Similarly, the gravito-thermal coupling constant for ions can be
estimated as ( )
i i e e i e
Tm T m 1 = << for mean
e
T
5
~ 10 K, and
i
1 for mean
e
T K
6
~ 10 .
Here g GM R
2

= denotes the value of the solar surface gravity. The values of the other
constant quantities are taken to be G
8
6.6726 10
= dyn cm
2
g
-2
, M
33
1.90 10
= g, and
R
10
6.97 10
= cm.
These estimates are based on the condition of an isothermal SIP, where T
e
and T
i
respectively

denote the electron and ion temperatures. It is now easy to see that the electrons can very
well overcome the gravitational potential barrier at the SSB in the standard solar model,
whereas the ions cannot. This is the reason why a surface polarization-induced space charge
(electrostatic) field is likely to appear, due to thermal leakage of the electrons from the SSB
in the radially outward direction. Moreover, the neutral gas approximation for the SIP may
not be a good one for describing the properties of the SSB. Similar realizations have already
occurred to previous authors [5, 9, 11, 14] for the SWP as well. We take the SIP to be an ideal
nonisothermal plasma gas with relatively cold ions. The mean electron temperature
e
T
6
10 >
K for the SIP emerges as a more suitable choice for our theoretical consideration.
According to our GES-model analyses, the GES divides into two scales: one bounded, and
the other unbounded. The former includes the steady state equilibrium description of the
SIP dynamics bounded by the solar self-gravity. This extends from the solar center to the
self-consistently defined and specified SSB. The unbounded scale encompasses the SWP
dynamics extending from the SSB to infinity. The SIP electrons can easily escape from the
defined SSB. On the other hand, the SIP ions cannot cross the gravitational potential barrier
of the solar mass on their thermal energy alone. However, surface leakage of the SIP
electrons is bound to produce an electrostatic field by virtue of surface charge polarization.
This, in turn, provides an additional source to act on the SIP ions to further energize and
encourage them cross over the solar self-gravitational potential barrier.
In order to describe the plasma-based GES physics of our model system, we adopt a
collisionless unmagnetized plasma fluid for simplicity in mathematical development to
obtain some physical insight into the solar wind physics. The role of magnetic field is also
ignored (just for mathematical simplicity) in discussing the collisionless SIP and SWP


100
dynamics. Applying the spherical capacitor charging model [3], the coulomb charge on the
SSB comes out to be
SSB
Q C ~120 . The mean rotational frequency of the SSB about the centre
of the SIP system is is determined to be
SSB
f Hz
12
~ 1.59 10
[42]. Applying the electrical

model [42] of the Sun, the mean value of the strength of the solar magnetic field at the SSB in
our model analysis is estimated as
SSB SSB SSB
B Q f T
2 11
4 ~7.53 10

= , which is negligibly
small for producing any significant effects on the dynamics of the solar plasma particles.
Thus the effects of the magnetic field are not realized by the solar plasma particles due to the
weak Lorentz force, which is now estimated to be ( )
L
F e v B N
33
3.61 10
= corresponding
to a subsonic flow speed v cm s
1
~ 3.00

with the input data available [2, 42] with us and
hence, neglected. Therefore our unmagnetized plasma approximation is well justified in our
model configuration. In addition, the effects of solar rotation, viscosity, non-thermal energy.
For further simplification, the electrons are assumed to obey a Maxwellian velocity
distribution. Although these approximations may not be realistic, but they may be
considered working hypotheses to begin with an ideal situation. Deviations indeed exist
from a Maxwellian velocity distribution. We however use it as a working hypothesis for our
model considerations. As a result, the usual form of the Boltzmann density distribution for
plasma thermal electrons with all usual notations is given as

e
N e .
= (20)
Here
e e
N n n
0
= denotes the normalized electron density. The generic notation
e
e T =
denotes the normalized value of the plasma potential associated with the GES on the
bounded scale and with the SWP on unbounded scale. The general notation
e
n stands for
the nonnormalized electron density and
i
n m
0

= defines the average bulk density of the

equilibrium SIP. The notation 1.43
= g cm
-3
stands for the average but constant solar
plasma mass density and
i
m
24
1.67 10
= g for the ionic (protonic) mass. Again e

represents the electronic charge unit and , the nonnormalized plasma potential associated
with both the GES and SWP.
The hydrogen ions are described by their full inertial response dynamics. This includes the
ion momentum equation as well as the ion continuity equation. The first describes the
change in ion momentum under the action of central gravito-electrostatic fields of potential
gradient and forces induced by thermal gas pressure gradients. The latter equation is
considered a gas dynamic analog of plasma flowing through a spherical chamber of radially
varying surface area. In normalized forms, the ion momentum equation is

i
T
i
dM d dN d
M
d d N d d
1

= . (21)
Here the minus sign in the gravitational potential term indicates the radially inward
direction of the solar self-gravity. The deviation from the conventional neutral gas treatment
of the SIP is introduced through the electric space charge-induced force (first term on right-
hand side) effect. The normalized expression for conservation of ion flux density is

i
i
dN dM
N d M d
1 1 2
0.

+ + = (22)

Acoustic Wave

101
The normalizations are defined as follows:
e
e
T
,
=
s
C
2
,
=
e
e
n
N
n
0
, =
i
i
n
N
n
0
, =
i
s
v
M
c
, =
J
r
,
=
s
J
J
c
,
=
e
s
i
T
c
m
1 2
,
| |
=
|
\ .

( )
J
G
1 2
4 ,
=
i
T
e
T
T
. =
The notations and respectively stand for the dimensional (unnormalized) values of the
plasma electrostatic potential and the self-solar gravitational potential as variables
associated with the GES. The dimensional values of the electron and ion population density
variables are respectively denoted by
e
n and
i
n . Likewise, the dimensional ion fluid
velocity variable is represented by the symbol
i
v . The notation stands for the normalized
variable of the self-solar gravitation potential. The notation
i
N denotes the normalized
value of the ion particle population density variable. Notation M stands for the ion flow
Mach number.
The notations r and stand for the nonnormalized and normalized radial distance
respectively from the heliocenter in spherical co-ordinates. The other notations
J
,
s
c and
J
defined as above stand for the Jeans length, sound speed and Jeans frequency
respectively. Finally, the notation
T
as defined above stands for the ratio of ion to electron
temperature. The ion flux density conservation (eq. 22) contains a term that includes the
effect of geometry on the ion flow dynamics of the SIP mass, self-gravitationally confined in
a spherical region, whose size is to be determined from our own model calculations.
Equations (21) and (22) can be combined to yield a single expression representing the well-
known steady state hydrodynamic flow,

( )
T T
dM d d
M
M d d d
2
1 2
.

= + (23)
There is an obvious difference in the above equation from the corresponding momentum
equation under the neutral gas approximation for the SIP. The difference appears, as
discussed above, in the form of a space charge effect originating from the Coulomb force on
a collective scale (first term on the right-hand side of eq. (21)).
The gravito-electrostatic Poisson equations complement the steady dynamical equation (23)
for a complete description of the gravito-electrostatic sheath structure, which is formed
inside the non-rigid SSB. This is important to emphasize that in the case of a real physical
boundary, the plasma sheath is always formed both inside and outside the boundary
surface in its close vicinity [12]. The normalized forms of the gravitational and electrostatic
Poisson equations for the SWP description are respectively given by

i
d d
N
d d
2
2
2

+ = , and (24)


102

De
e i
J
d d
N N
d d
2
2
2
2

| |
(
+ = |
(
|

\ .
. (25)
Here
( )
De e
T n e
1 2
2
0
4 = denotes the plasma electron Debye length of the defined SIP
system. The other quantities are as defined above as usual. Equations (21)(25) constitute a
completely closed set of basic governing equations with which to discuss the basic physics
of the GES-potential distribution on the bounded scale. Of course, the discussion also
includes the associated ambipolar radial flow variation of the SIP towards an unknown SSB
which we have to determine self-consistently in this problem with GES-based theory. For a
typical value
e
T
6
10 = K, one can estimate that
De J
20
10

which implies that the Debye
length is quite a bit smaller than the Jeans scale length of the solar plasma mass. Thus, on
the typical gravitational scale length of the inertially bounded plasma, the limit
De J
0
represents a realistic (physical) approximation. By virtue of this limiting condition, the
entire SIP extending up to the solar boundary and beyond obeys the plasma approximation.
Thus, the quasi-neutrality condition as given below holds good

e i
N N N e .
= = = (26)
This is to note that equation (26) does not mean that the plasma ions are Boltzmannian in
thermal character, but inertial species. Equation (26) can be differentiated once in space and
further rewritten as,

dN d
N d d
1
.

= (27)
By virtue of the plasma approximation, one can justify that the GES of the SIP origin should
behave as a quasi-neutral space charge sheath on the Jeans scale size order. The formation
mechanism of the defined GES, however, is the same as in the case of plasma-wall
interaction process in laboratory confined plasmas. From equations (26)-(27), it is clear that
for the electrostatic potential and its gradient being negative, causes the exponential
decrease of the plasma density. Finally, the reduced form of the basic set of autonomous
closed system of coupled nonlinear dynamical evolution equations under quasi-neutral
plasma approximation is enlisted as follows

( )
T T
dM d d
M
M d d d
2
1 2
,

= + (28)

d dM
d M d
1 2
0,

+ + = and (29)

d d
e
d d
2
2
2
.

+ = (30)
This set of differential evolution equations constitutes a closed dynamical system of
governing hydrodynamic equations that will be used to determine the existence of a
bounded GES structure on the order of the Jeans scale length
( )
J
in our GES-model of the

Acoustic Wave

103
subsonic origin of the SWP of current interest. Thus the solar parameters ( ) M , ( )
s
g and
( ) representing the equilibrium Mach number, solar self-gravity and electrostatic
potential, respectively, will characterize the gravito-electrostatic acoustics in our approach.
5.3 Theoretical analysis of solar surface boundary
5.3.1 Analytical calculations
We first wish to specify the overall condition for the existence of the SSB. Such existence
demands the possibility of a self-consistent bounded solution for the solar self-gravity. The
boundary will correspond to a maximum value of the solar self-gravity at some radial
distance from the heliocenter. This defines a self-consistent location of the SSB. Before we
proceed further, let us argue that the radially outward pulling bulk force effect of the GES-
associated potential term in equation (28) demands a negative electrostatic potential
gradient, that is, d d 0 < . This makes some physical sense because the ion fluid has to
overcome the gravitational barrier to create a global-scale flow of the SIP in a quasi-
hydrostatic way.
Now, if we invoke the concept of exact hydrostatic formation under gravito-electrostatic
force balancing( ) d d d d , the surface potential can be solved to get

.
Here the unknown boundary values of
= ,
= and
SSB
M M = are to be self-
consistently specified numerically. The notation ( )
SSB
M stands for the Mach value
associated with the SIP flow at the SSB. Now, by the exact hydrostatic equilibrium condition
in the set of equations (28)-(30), one can get the following set of equations for the SSB
description:

( )
T T
dM
M
M d
2
1 2

= , (31)

d dM
d M d
1 2
0

+ + = , and (32)

d d
e
d d
2
2
2

+ = . (33)
For purpose of the GES analysis, we define the solar self-gravitational acceleration as
s
g d d = . Equation (34) thus reads

s
s
dg
g e
d
2

+ = . (34)
Finally, the SIP and hence, the SSB are described and specified in terms of the relevant solar
plasma parameters ( ) M , ( )
s
g and ( ) representing respectively the equilibrium
Mach number, solar self-gravity and electrostatic potential as a coupled dynamical system
of the closed set of equations recast as the following
Solar self-gravity equation:

s
s
dg
g e
d
2
,

+ = (34a)


104
Ion continuity equation:

d dM
d M d
1 2
0,

+ + = and (34b)
Ion momentum equation:

( )
s
dM
M g
M d
2
1 2
,

= (34c)
where ( )
T i e
T T 1 1 = + = + ,
e
T is the electron temperature and
i
T is the inertial ion
temperature for the bounded solar plasma on the SIP-scale as already mentioned.
Let us now denote the maximum value ( ) g
of solar gravity at some radial position
=
where
= and apply the necessary condition for the maximization of

s
g at a spatial
coordinate
= . This condition ( )
s
dg d 0

=
= when used in equation (34) yields
g e 2

= . However, it is not sufficient to justify the occurrence of the maximum value of
s
g until and unless the second derivative of
s
g is shown to have negative value. To derive
the sufficient condition for the maximum value of
s
g at
= , let us once spatially

differentiate equation (34) to yield

s s
s
d g dg d
g e
d d d
2
2 2
2 2
.

+ = (35)
Now the condition for the maximization of
s
g at the location
= can be discussed by
considering d d 0 < in equation (35) under the exact hydrostatic equilibrium
approximation ( )
d d d d g

= near the solar surface to yield the following
inequality

s
d g
g e g g e
d
2
2 2 2
2 2
0.

=
| |
= = <
|
\ .
(36)
From these analytical arguments one can infer that the maximization of g
s
indeed occurs
at some arbitrary radial position that satisfies the inequality: e
2
2 ( 2.33)
> = for
~ 1
(Figs. 8b, 9b, and 10b). Numerically the location of the SSB is found to lie at
~3.5
that matches with g e 2

= for 1.07
= and g 0.6
= . It satisfies the
analytically derived inequality (36) too. Now the other two equations (32)-(33) can be
simultaneously satisfied in the SSB only for a subsonic solar plasma ion flow speed if
Mach number gradient acquires some appropriate negative minimum near zero
( )
( )
M dM d
6
~ ~ 10

.
It is indeed seen numerically that near the maximum solar self-gravity of the SIP mass, the
first and third terms in equation (32) are almost equal and hence the Mach number gradient
term, which is negative in the close vicinity of the SSB, should be smaller than the other two
terms so as to satisfy equations (31) and (32), simultaneously. Actually, the three equations
(31)-(32) and (34) are solved numerically to describe the SSB of the maximum self-
gravitational potential barrier properly where
s
g associated with the self SIP mass is
maximized.

Acoustic Wave

105
5.3.2 Numerical calculations
Determination of the autonomous set of the initial values of the defined physical variables is
a prerequisite to solve the nonlinear dynamical evolution equations (34a)-(34c) in general as
an initial value problem. The initial values of the physical variables like ( ) ( )
s
M g , and
( ) are defined inside the solar interior and are determined on the basis of extreme
condition of the nonlinear stability analysis [4]. The self-consistent choice of the initial values
is obtained by putting
i
i
dM d e
2
= ,
i
s
dg d 0
= and
i
d d 0
= in these three
equations (34a)-(34c). But the realistic SWP model demands that
i
d d 0.
Finally, we
determine the expressions for a physically valid set of the initial values of the given physical
variables as follows,

i
i i
M e
2
1
2

= (37)

i
si i
g e
1
2

= (38)
This is to note that the initial values of
i
and
i
are chosen arbitrarily. As discussed later,
we find that the SSB acquires a negative potential bias (
s
~ 1 ) of about -1 kV. It also
acquires the maximum value of solar interior gravity ( g ~0.6
) and minimum value of non-

zero SIP flow speed (
SSB
M
7
~ 10
) at the SSB. The value of the electrostatic potential gradient

at the surface comes out to be ~ 0.6. This means that the strength of the GES-associated
solar surface gravity and electrostatic potential gradient is almost equal. As a result the SSB
is defined by some constant values of the physical variables (
s
g M , , ). The SSB values of
these parameters are determined through spatial evolution of the coupled system of
equations (34a)-(34c) from the given initial values (37)-(38) inside the SIP zone.
We have used the well-known fourth-order Runge-Kutta method (RK 4) for numerical
solutions of equations (34a)-(34c). By numerical analysis (Figs 8-11), we find that the solar
radius is equal to thrice of the Jeans length
( )
J
for mean solar mass density
( )
g cm
3
1.41 .

= . From this observation one can easily estimate that

j
R /3.5

= . Now
comparing our own theoretical value of the solar mass self-gravity with that of the standard
value, we arrive at the following relationship between the solar plasma sound speed ( )
s
c
and the Jeans length
( )
J

( )
s J
c
2 4
0.6 2.74 10 = cm/sec
2
. (39)
By substituting the value of the Jeans length expressed in terms of the solar radius, one can
determine and specify the mean value of the electron temperature, which is
e
T ~ 10
7
K. The
sound speed in the SIP under the cold ion model approximation is thus obtained as
s
c
7
~ 3 10 cm s
-1
. Note that the SWP speed at 1 AU is fixed by the sound speed of the SWP
medium, which is determined and specified by the requirement that a transonic transition
solution occurs on the unbounded scale of the SWP dynamics description.
The gravito-acoustic coupling coefficient could be estimated as
g a i e
mg R T ~2.0

= . In
absence of the gravitational force, the bulk SIP ion fluid will acquire the flow speed
corresponding to M~ 1.41 for a negative potential drop of the order of ( )
e
T e over a


106
distance equal to that of the solar radius. If one estimates the value of gravito-acoustic
coupling coefficient at this velocity defined by M~ 1.41 , it comes out to be unity. Thus the
quasi-hydrostatic type of equilibrium gravitational surface confinement of the SIP is
ensured. The GES-potential induced outward flow of the SIP is also ensured. Due to
comparable strength of the solar surface gravitational effect of deceleration, the net SIP ion
fluid flow is highly suppressed to some minimum value (~1.0-3.0 cm/sec) corresponding to
M
7
~10
at the SSB.
An interesting point to note here is that near the defined SSB, the electrostatic potential
gradient terminates into an almost linear type of profile. The value of its gradient value will
provide an estimate of the second order derivatives contribution into the electrostatic
potential which measures the level of local charge imbalance near the solar surface. From
our computational plots (Figs. 8b, 9b and 10b), this local charge imbalance comes out to be
of the order 0.17, which is equivalent to 17% ion excess charge distributed over a region of
size on the order of the plasma Debye sheath scale length. However, the same level of the
electrostatic local charge imbalance on the Jeans scale length does not require the inclusion
of the role of the Poisson term for the evolution of the electrostatic potentials profile under
the GES-model. Hence, in this sense the GES is practically equivalent to a quasi-neutral
plasma sheath with its potential profile tailored and shaped by the potential barrier of the
self-gravity of the SIP mass distribution.
5.3.3 Properties of solar surface boundary
Table I lists the defined initial values of the physical variables (
s
g , , M) as already
discussed and their corresponding boundary values numerically obtained for the
description of the desired SSB. The initial values of
s
g , , and M are associated with the
normalized mean SIP mass density, enclosed within a tiny spherical globule having
normalized radius equal to an arbitrarily chosen value of
i
.

Parameter At the Initial
Radial Point ( )
i

At the Solar Surface
Boundary ( )

Initial Values
Potential d
d
0
=
d
d
~ 0.62
, ~ 1.00

i
, arbitrarily chosen
Gravity
s
g
s
dg
d
0
=
s
dg
d
0
= , g ~0.60

i
si i
g e
1
2

= , derived

Mach
number M
i
dM
e
d
2
=
dM
d
0
= ,
SWP
M
7
~ 10

i
i i
M e
2
1
2

= , derived
Table 1. Initial and Boundary Values of Relevant Solar Parameters
From the numerical plots shown in Figs. 8-10, we find that the minimum Mach number
( )
SSB
M at the specifically defined SSB comes out to be of order 10
-7
. For this value of Mach
number, equation (31) can be simplified to show that near the boundary,
dM d M
8
3 10 ~ 0

= . This corresponds to a quasi-hydrostatic type of the SSB

equilibrium. It arises from gravito-electrostatic balancing with an outward flow of the SIP
having a minimum speed of about 1-3 cm s
-1
. With these inferences one can argue that the

Acoustic Wave

107
SWP originates by virtue of the interaction of the SIP with the SSB. Hence an interconnection
between the Sun and the SWP can be observed by applying the GES model. Here the
boundary is not sharp but distributed over the entire region of the solar interior volume.
The basic principles of the GES coupling govern the solar surface emission process of the
subsonic SWP.
As depicted in table I, the time-independent solar
s
g -profile associated with the SIP mass
distribution terminates into a diffuse surface boundary. This is characterized and defined by
the quasi-hydrostatic equilibrium
s
g g d d ~
= , which occurs at about ~3.5
= (see
Figs. 8-10). As such, the basic physics of the subsonic origin of the SWP from the SSB is
correlated with the bulk SIP dynamics. We note that the precise definition of the SSB
influences the SWP velocity at 1 AU. Other models report similar observations too [3, 31-41].
The dependence on the ion to electron temperature ratio is quite visible in Fig. 11a. Let us
now discuss the numerical results in the figures individually.
Figure 8 depicts the time-independent profiles of (
s
g , , M) and their variations with the
ion-to-electron temperature ratio
t
for fixed values of the initial point (
i
0.01 = ) and
plasma sheath potential (
i
0.001 = ). As shown in Fig. 8a, the location of the SSB remains
the same but its maximum value changes, and a most suitable choice of
t
= 0.4 is identified
for which the quasi-hydrostatic condition is fulfilled. The E-field profile is invariant for all
chosen values
t
=0-0.5. Again, as shown in Fig. 8b, the electrostatic potential corresponding
to
t
~ 0.4 comes out to be ~ 1
= (i.e. ~1 kV). Similarly, Fig. 8c depicts the minimum

Mach value of
SSB
M ~10
-7
for
t
~0.4 varying by a factor of 2 for other values of
t
.
Figure 9 depicts the time-independent profiles of
s
g , , and Mand their variations with
initial position for fixed values of
t
=0-0.4 and
i
0.001 = . It can be seen that the most
suitable choice of the initial position for our fixed values
i
and
T
comes out to be
i
~0.01 , which is consistent with the earlier value shown in Fig. 8a. Moreover, the
minimum value of M~10
-7
(Fig. 9c) is also consistent with the earlier value shown in Fig. 8c.
Figure 10 depicts the time-independent profiles of
s
g , , and M and their variations with
electrostatic potential for fixed values of
i
0.01 = and
T
0.4 = . It is observed fascinatingly
that the most suitable choice of the initial value of the normalized electrostatic potential for
our fixed values of
i
and
T
comes out to be
i
0.001 = .
It is notable that high initial drop of M-profile occurs as shown in Fig.8c, Fig. 9c and Fig. 10c.
This indicates the over dominance of the solar interior gravity up to about ~ 1.5 , and
thereafter, the E-field becomes comparable, balancing at about ~ 3.5 . Thus the normalized
width of the gravito-electrostatic sheath could be estimated and denoted by
G E
~2

. This is a
quasi-neutral space charge region with positive charge (ion) excess near the defined SSB
wall. Thus a self-consistent bounded solution of nonlinearly coupled gravito-electrostatic
potential profiles exists. It forms a quasi-hydrostatic equilibrium at the SSB for the choice of the
appropriate set of the initial parameter values
i i
0.001 & 0.01 = = for
T
0.4 = . This is not a
rigid boundary at all. As a result the SSB is capable to exhibit many kinds of global oscillation
dynamics governed by the nonlinear coupling of the gravito-electrostatic sheath potentials.
For a laboratory hydrogen plasma, the normalized floating potential can be estimated as
f
3.76 = under the flat surface approximation. Now, if we consider the numerically
calculated minimum value of M for our solar surface characterization, the estimated value of


108
f
is about 20 using the flat surface approximation. This is a crude estimate because the
solar surface potential drop occurs over a distance of the order of the Jeans scale where the
effect of curvature should not be ignored. The numerically obtained solar surface potential
is quite a bit smaller than the floating potential. Simply put, this means that the defined SSB
of the GES-model draws a finite amount of electron-dominated electric current that flows
toward the heliocenter.
Let us invoke a generalized concept of the plasma sheath, which is traditionally associated
with a localized electrostatic potential only in the plasma physics community. We argue that
any localized nonneutral space charge layer (in our case, on the order of the Jeans length) is
the result of a self-consistent nonlinear coupling of gravito-electrostatic force field
variations. This is what we mean by the GES, which of course, obeys the global quasi-
neutrality condition because of the smallness of the ratio
De J
for the SIP parameters.
5.4 Acceleration of solar interior plasma
5.4.1 Basic equations for SWP descriptions
We have already argued that the subsonic origin of the SWP from the SSB is an outcome of
the condition of quasi-hydrostatic equilibrium at the boundary. This is a result of the
comparable, but competiting strengths of the gravitational deceleration and the electrostatic
acceleration of the SIP near the SSB. Now we will try to look at the problem of solar wind
acceleration from subsonic to supersonic speed. This is referred to as the transonic transition
behavior of the outward-moving SIP in the form of the SWP. Let us now argue that the
radial variation of Mach number and electrostatic potential beyond the defined SSB should
be described by the following autonomous set of coupled nonlinear differential equations

( )
T T
s J
dM d GM
M
M d d C
2
2 2
1 2 1

= + , and (40)

d dM
d M d
1 2
0

+ + = . (41)
Let us note that the constant SIP mass acts as an external object to offer a source of gravity
for tailoring and monitoring the outgoing SIP flow with the initially subsonic speed
specified at the defined SSB. The Poisson equation for gravity is now redundant. It is
important to comment that the electrostatic force field is not imposed from outside to
control the solar winds motion. In fact, the required electric field for the SWP acceleration is
of internal origin. Equations (40) and (41) can be combined to yield a single coupled form as
given below

( ) ( )
T T
s J
dM GM
M
M d C
2
2 2
1 2 1
1 1 .

( + = +

(42)
The quantity
s J
a GM c
2
0

= (normalization coefficient) is treated as a free parameter, which

eventually provides a way to estimate the SWP electron temperature. The value of this
parameter is determined by the condition that a transonic solution for the SWP exists for a
given set of initial values of the required physical variables. The above equation can now be
rewritten as

Acoustic Wave

109

( ) ( )
T T
dM a
M
M d
2 0
2
1 2
1 1 .

( + = +

(43)
5.4.2 Numerical results
Equations (41) and (43) can be solved by numerical methods (by Runge-Kutta IV method) to
determine the time independent M and profiles associated with the SWP for some
arbitrary values of a
0
. However, we choose the minimum value of a
0
that yields transonic
transition solutions. It is obvious from equation (43) that the critical distance will exist at
c T
a
0
2(1 ) = = + . This critical distance corresponds to R ~14

from the defined solar
surface. As shown in Fig. 10, the critical point for transonic transition, indeed, exists for a
0
=
95 for narrow range variation of
T
0.0 0.1 = , for the already derived solar surface values of
SSB
M
7
10 & 1.0
= = as a set of initial values. The M-values at a distance of 1AU from the

defined SSB, i.e., at ~750 are about 3.3 and 3.5 for
T
0.0& 0.1 = , respectively, as shown
in Fig. 11a. The corresponding values of the electrostatic potential at the same distance are
found to be 31& 30 = for
T
0.0&0.1 = , respectively as shown in Fig. 11b. This is to
note that for higher values of
T
, solar breeze solutions are obtained.
Substitution of
J
R 3.5

= in the defined expression of a
0
= 95, we can estimate c
s
~100
km/sec and T
e
~100 eV for the SWP. The critical distance for transition behavior to occur for
M
SSB
~10
-7
(Fig. 11a) as an initial value for Mach number exists at about R 14

distance apart
from the defined solar surface. This is to note that if we consider M
SSB
~10
-6
as an initial
value for the numerical solution of equations (41) and (43), the transonic transition occurs
for a
0
= 71 that yields almost the same values of c
s
~100 km/sec and T
e
~100 eV for the SWP.
But the critical transition location point exists at about R 10

distance apart from the defined
solar surface. This implies the initial value of M
SSB
plays an important role in the proper
fixation of a
0
that determines the exact location of transonic transition point and the SWP-
property. Accordingly, the speed of the SWP at 1 AU comes out to be 330-350 km s
-1
, as
shown in Fig. 11a (dotted vertical line).
Let us now look at Fig. 11b which the electrostatic potentials profile for a predetermined set
of initial values of
SSB
M
7
10
= , and ~ 1
at the SSB, as in the case of Fig. 11a. It can be

seen that the normalized value of the SWP-associated electrostatic potential at 1 AU is about
-30 to -31 for
T
0.0 0.1 = . With some simple calculations, as illustrated in the next
subsection, we can argue that beyond the transonic transition, the SWP seems to obey the
zero-electric current approximation, but not before. This is inferred from the floating surface
condition, which is defined by the equalization of escaping flux of the SWP particles in
accordance with the law of conservation of particle flux.
5.4.3 Theoretical estimation of floating potential
In absence of any particle source and/or sink of a stellar origin under spherical geometry
approximation, we get an expression for the steady state mass conservation of the SWP
flow

iT i SSB SSB
r n v r n v
2 2
0
. = (44)
In normalized form the above expression (44) for N
i
=n
i
/n
0
=1, can be written as


110

( )
SSB SSB
M M
2 2
. = (45)
Now using normal practice for floating potential estimation under net zero-electric current
approximation, i.e. J
e
=J
i
,, one gets

( ) ( )
f
i e SSB SSB
e m m M M
2 2
.
= = (46)
Now, from equation (47) the normalized floating potential at any normalized radial position
from the SSB can be expressed as

e SSB
f SSB
i
m
M
m
2
log .
(
| |
= (
|
( \ .

(47)
By simple calculations, one can generate the following comparative data of theoretical
estimation of the SWP floating potential (using above expression (47)) at different distances
from the obtained SSB as follows.

f

3.50 (at
) -19.57
47.50 (at
c
) -24.78
100 -26.27
200 -27.66
300 -28.47
400 -29.04
500 -29.49
600 -29.86
700 -30.16
750 (1 AU) -30.30
Table 2. Values of the Floating Potential
It looks as if the SSB was in non-floating condition as because it does not acquire floating
potential during evolution of the GES-potential distribution of the SIP. However, beyond
the critical distance and up to a distance of 1 AU, the calculated values of the floating
potential almost match with those of the SWP obtained numerically (Fig. 11b). This
implies that a finite divergence-free electric current exists at the SSB up to the transonic
transition region! Beyond the transonic point, zero electric current approximation seems
to hold good.
It is commented that the zero-electric current approximation at the SSB assumed in previous
model calculations [3, 11, 31-41] for the qualitative description of the SWP properties seems
to be physically unjustified. Furthermore, our model calculation does not require outside
imposition of the electric field to ensure the validity of the zero-electric current
approximation at the SSB. Probably the imposition of the zero-electric current
approximation is not suitable for proper description of the SWP properties. Now the natural
question may arise, What happens to the SWP current after the transonic transition? It

Acoustic Wave

111
seems the current dissipates mainly through a channel of inertial resistance of the plasma
ions due to solar gravity.
5.5.1 Description of numerical results
The proposed GES-model predicts that the GES formation (of the SIP origin) drives the
subsonic SWP at the solar surface. The quasi-hydrostatic equilibrium defines the solar
boundary and ensures the GES formation. Numerically ~ 1
, M
SSB
~10
-7
, and
g d d ~ 0.60
prescribe the defined solar boundary (Table I). It requires specific initial
values
i i
0.001& 0.01 = = in the solar interior for
T
0.4 = .

(a) (b)

(c)
Fig. 8. Variation of normalized values of (a) solar interior gravity d d (upper group of
curves) and electric field d d (lower curve), (b) electrostatic potential , and (c) speed M
associated with solar interior plasma flow dynamics with normalized position ( ) from the
heliocenter ( ) 0 = . The values of initial position 0.01
i
= and initial electrostatic potential
0.001
i
= are held fixed. The lines correspond to the cases
T
= 0.0 (graph 1), 0.1 (graph 2),
0.2 (graph 3), 0.3 (graph 4), 0.4 (graph 5), and 0.5 (graph 6) respectively. The defined solar
surface boundary lies at a radial position ~ 3.5
(implying ~ 3.5
J
R
) with circled points

corresponding to the solar surface values
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Solar interior gravity (A) and solar interior E-field (B) profiles
Normalized position
N
o
r
m
a
l
i
z
e
d

s
o
l
a
r

i
n
t
e
r
i
o
r

g
r
a
v
i
t
y

(
A
)

a
n
d

E
-
f
i
e
l
d

(
B
)
(1)
(2)
(3)
(4)
(5)
(6)
O
O
A
B
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-2.5
-2.25
-2
-1.75
-1.5
-1.25
-1
-0.75
-0.5
-0.25
0
Normalized solar interior electrostatic potential profile
Normalized position
N
o
r
m
a
l
i
z
e
d

e
l
e
c
t
r
o
s
t
a
t
i
c

p
o
t
e
n
t
i
a
l
(1) (2) (3) (4) (5) (6)
O
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
x 10
-7
Mach No. profile with normalized position
Normalized position
M
a
c
h

N
o
.
(1) (2) (3) (4) (5) (6)
O


112
Item Parkers model GES Model
1. Deals with an unbounded solution of
steady state hydrodynamic equilibrium
of the solar wind (SW)

Deals with bounded (SIP)
and unbounded (SWP)
solutions of a continuum
steady state hydrodynamic
equilibrium
2. Considers a single neutral fluid (gas)
model approximation for the SW gas
flow dynamics
Considers a two-fluid ideal
plasma (gas) model for the
SIP and SWP gas flow
3. Predicts an unbounded solution of
supersonic expansion of the SW
provided that a sub-sonic flow pre-
exists at the SSB
Predicts a bounded solution
of the SIP mass distribution
with its subsonic outflow at
the SSB
4. Genesis of the subsonic solar surface
origin of the SW is not precisely known:
discusses the acceleration of the SW by
analogy with the de Laval nozzle

Discusses the genesis of the
subsonic SSB origin of the SIP
in terms of the basic
principles of the GES
acceleration of ions: the
transonic acceleration
mechanism of the SWP is the
same as Parkers
5. Does not specify precisely the SW-base
definition and prescription for the self-
consistent SSB
Offers a precise definition of
and prescription for a self-
consistent SSB
6. Standard solar surface is electrically
uncharged and unbiased
SSB acquires a negative
electrostatic potential (~1 kV)
at the cost of thermal loss of
the electrons
7. Does not consider plasma-boundary
wall interaction, plasma sheath
formation and spontaneous thermal
leakage through squeezing mechanism
Considers it
8. Concept of floating surface (at which no
net electric current) is not involved
It is involved
9. Considers one-scale (SW) theoretical
description
Considers two-scale (SIP and
SWP) theoretical description
10. Extensive research has already been
done on the SW acceleration and
heating

Opens a new chapter of the
GES-based theory for interior
(bounded) and exterior
(unbounded) solar plasma
flow dynamics
Table 3. Parker versus GES Model

Acoustic Wave

113
The GES-formation occurs due to solar surface leakage of the thermal electrons of solar
interior plasma outgoing radially outwards. It causes an appreciable space charge
polarization effect near the boundary. The depth of the electrostatic potential well for the
plasma ions, so formed, is such as to allow the incoming ions from the solar interior bulk
plasma to acquire the kinetic energy of ion motion to overcome the maximum gravitational
potential barrier height near the boundary. The SIP ions come out of the solar gravitational
barrier with a minimum speed M
SSB
~10
-7
. From the floating potential calculation with no
net current flow, we infer that the solar surface boundary drives out some finite electric
current in the outward flow. That is, it seems a finite electric current loss of the SIP occurs
through its leakage process near the SSB! It can be shown that the total surface charge in the
boundary, however, comes out to be about 10
20
times the electronic charge. Table III gives a
glimpse of distinction between Parkers model and GES-model of the subsonic SWP origin
and its acceleration from subsonic to supersonic flow speed.

(a) (b)

(c)
Fig. 9. Same as Fig. 8 but with the ion-to-electron temperature ratio 0.40
T
= and
electrostatic potential 0.001
i
= held fixed. The lines correspond to the cases with initial
positions
i
= 10
-4
(graph 1), 10
-3
(graph 2), 10
-2
(graph 3), 10
-1
(graph 4), 0.2 (graph 5), 0.5
(graph 6), and 1.0 (graph 7), respectively. The circled points indicate the most suitable choice
of the solar surface values
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Normalized position
N
o
r
m
a
l
i
z
e
d

s
o
l
a
r

i
n
t
e
r
i
o
r

g
r
a
v
i
t
y

(
A
)

a
n
d

E
-
f
i
e
l
d

(
B
)
(7) (6) (5) (4).....
O
O
A
B
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-2.5
-2
-1.5
-1
-0.5
0
Normalized position
N
o
r
m
a
l
i
z
e
d

e
l
e
c
t
r
o
s
t
a
t
i
c

p
o
t
e
n
t
i
a
l
(7)
(6)
(5)

O
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
0
0.25
0.5
0.75
1
1.25
1.5
x 10
-6
Mach No. profile with normalized position
Normalized position
M
a
c
h

N
o
.
(1)
(2)
(3)
O


114
This is to clarify that the GES-model is a quite simplified one in the sense that it does not
include any role of magnetic forces, interplanetary medium or any other complications like
rotations, viscosity, etc. It opens a new chapter for further study on the coupled system of
the solar interior and exterior plasma flow dynamics.
According to GES-model, the normal acoustic modes of the global solar surface oscillations
can be analyzed in terms of the local and global gravito-electrostatic plasma sheath
oscillations governed by the basic principles of linear and nonlinear nonlocal theory of the
Jeans collapse model [24-25] of charged dust clouds in plasma environment.
The magnified view of the Mach number variation in transonic transition zone of the SWP
(Fig. 11c) indicates the existence of an extended region having almost uniform sonic flow
speed. It can be deduced from Fig. 11c that the transonic point does not always coincide
with the critical point. We define the critical point as a radial point (away from that defined
solar surface) where the net force on the SWP ions, due to GES-induced E-field and external
gravity due to total solar interior plasma mass, becomes almost zero.

(a) (b)

(c)
Fig. 10. Same as Fig. 8 but with the initial position 0.01
i
= and ion-to-electron temperatue
ratio 0.40
T
= held fixed. The lines correspond to the cases of
i
= 0.0 (graph 1), -0.001
(graph 2), -0.01 (graph 3), -0.1 (graph 4), -0.5 (graph 5), -0.6 (graph 6), and 1.0 (graph 7),
respectively. The circled points indicate the most suitable choice of the solar surface values
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Normalized position
N
o
r
m
a
l
i
z
e
d

s
o
l
a
r

i
n
t
e
r
i
o
r

g
r
a
v
i
t
y

(
A
)

a
n
d

E
-
f
i
e
l
d

(
B
)
O
O
(7) (6) (5) (4) (3) ...
B
A
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-2.5
-2
-1.5
-1
-0.5
0
Normalized position
N
o
r
m
a
l
i
z
e
d

i
n
t
e
r
i
o
r

e
l
e
c
t
r
o
s
t
a
t
i
c

p
o
t
e
n
t
i
a
l
O
(7) (6) (5) (4) (3) ....
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x 10
-6
Normalized position
M
a
c
h

N
o
.
Mach No. profile with mormalized position
(1) (2) (3)
(4)
(5)
(6)
(7)
O

Acoustic Wave

115
Numerical solution in the GES-model reproduces the Parker model values of the SWP speed
at 1 AU (Fig. 11a) for the numerically predetermined set of initial values of M
SSB
~10
-7
and
SWP ion-to-electron temperature ratio
T
0.0 0.1 = . The estimated critical point for the
transonic transition to occur ( )
c
r R ~ 14

differs from that ( )
c
r R ~ 10

in Parkers model. We
find that the latter can be obtained with a choice of M
SSB
~10
-6
(or larger than this by orders
of magnitude) as an initial Mach value at the solar surface.

(a) (b)

(c)
Fig. 11. Variation of normalized values of (a) speed M, (b) electrostatic potential , and (c)
speed Min the transonic transition zone associated with SWP flow dynamics with respect to
normalized position ( ) from the solar surface boundary ( ) 3.5
= in magnified form. The

predetermined solar surface boundary parameter values of
7
~ 10
SSB
M

, ~ 1.0
and
2
0
95
s J
a GM c
= = are considered as the set of initial values. The lines correspond to the
cases of
T
= 0.0 (graph 1), 0.1 (graph 2), 0.2 (graph 3), 0.3 (graph 4), and 0.4 (graph 5),
respectively. The critical distance lies at 47.5
c
, which corresponds to a radial position of
~ 14 r R
from the solar surface boundary

This is to clarify that the GES-model is a quite simplified one in the sense that it does not
include any role of magnetic forces, interplanetary medium or any other complications like
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800
-35
-32.5
-30
-27.5
-25
-22.5
-20
-17.5
-15
-12.5
-10
-7.5
-5
-2.5
0
Electrostatic potential profile associated with SWP
Normalized position
N
o
r
m
a
l
i
z
e
d

e
l
e
c
t
r
o
s
t
a
t
i
c

p
o
t
e
n
t
i
a
l
Graph 1
Graph 2
Graph 3
Graph 4
Graph 5
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800
0
0.25
0.5
0.75
1
1.25
1.5
1.75
2
2.25
2.5
2.75
3
3.25
3.5
3.75
4
Mach profile associated with SWP
Normalized position
M
a
c
h

N
o
.
Graph 1
Graph 2
Graph 3
Graph 4
Graph 5
0 5 10 15 20 25 30 35 40 45 50
0
0.15
0.3
0.45
0.6
0.75
0.9
1.05
1.2
1.35
1.5
Mach profile associated with SWP in transonic transition zone
Normalized position
M
a
c
h

N
o
.
Graph 1
Graph 2
Graph 3
Graph 4
Graph 5
Sonic line(M=1.0488 for Ti/Te=0.1)


116
rotations, viscosity, etc. It opens a new chapter for further study on the coupled system of
the solar interior and exterior plasma flow dynamics.
According to GES-model, the normal acoustic modes of the global solar surface oscillations
can be analyzed in terms of the local and global gravito-electrostatic plasma sheath
oscillations governed by the basic principles of linear and nonlinear nonlocal theory of the
Jeans collapse model [24-25] of charged dust clouds in plasma environment.
The magnified view of the Mach number variation in transonic transition zone of the SWP
(Fig. 11c) indicates the existence of an extended region having almost uniform sonic flow
speed. It can be deduced from Fig. 11c that the transonic point does not always coincide
with the critical point. We define the critical point as a radial point (away from that defined
solar surface) where the net force on the SWP ions, due to GES-induced E-field and external
gravity due to the total SIP mass, becomes almost zero.

Fig. 12. Variation of net normalized GES force per unit mass associated with solar interior
plasma flow dynamics with respect to normalized position ( ) from the heliocentre ( ) 0 = .
The initial position 0.01
i
= and initial electrostatic potential 0.001
i
= are held fixed. The
lines correspond to the cases of
T
= 0.5 (graph 1), 0.4 (graph 2), 0.3 (graph 3), 0.2 (graph 4),
0.1 (graph 5), and 0.0 (graph 6), respectively. The defined solar surface boundary is found to
lie at a radial position ~ 3.5
for a more suitable choice of

T
= 0.4-0.5 variations
Now, one can see that the sonic point for
T
= 0.0 (graph 1 in Fig. 11c) falls around 10
whereas the transonic point for the same value of
T
= 0.0 falls around the critical point
c
47.5 . Similarly, one can see that the sonic point for
T
= 0.1 (graph 2 in Fig. 11c) falls
around 15 whereas the transonic point for the same value of
T
= 0.1 falls around the
non-critical point 50 . From these numerical observations of the transonic transition
region, one can clearly notice that an extended zone of about 40
J
exists having almost a
uniform sonic flow speed of the SWP between sonic and transonic points.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
-1
-0.75
-0.5
-0.25
0
0.25
0.5
0.75
1
1.25
1.5
1.75
2
Normalized position
N
e
t

G
E
S
-
f
o
r
c
e

p
e
r

u
n
i
t

m
a
s
s
Net normalized GES-force profile
Graph 1
Graph 2
Graph 3
Graph 4
Graph 5
Graph 6
O

Acoustic Wave

117
In this region, the inertia-induced acoustic excitation theory [5-12] may have potential
applications provided it is improved with a proper inclusion of the solar gravity under
nonlocal normal-mode analysis. Similar situations are likely to arise in laboratory plasmas of
gravitationally sensitive multi-ion colloidal plasma systems [14-15, 24-25].
This is to point out that the intense acoustic fluctuations appearing in the Mach-profile (Fig.
11c) are merely the results of a numerical instability arising due to the mathematically
indeterminate situations localized mainly near the sonic speed. These fluctuations, however,
are found to disappear beyond the critical distance. Physically, however, the physical
appearance of such indeterminate structures in the graphical plots is because of some
chaotic interference and intermittency of acoustic background fluctuations in the emitted
SWP. It may produce some dissipative effects in course of the electrodynamical process of
the electrodynamical interaction of the SWP particles with background particles of ambient
stellar atmosphere.
5.6 Comparison with exospheric model
Exospheric model [38-39] is a simple kinetic model for the solar coronal plasma expansion.
This model assumes that beyond a given altitude termed as the exobase from the SSB,
binary collisions between the SWP-particles are negligible. The coronal plasma expansion is
believed to occur due to thermal evaporation of the hot protons that have velocity exceeding
the escape velocity so as to cross over the barrier of the external solar gravitational field. The
generalized model [19-23] considers the non-Maxwellian velocity distribution function for
the coronal plasma electrons.
Of course, this model has succeeded to explain the observation of the high speed SWP. This
model indeed explains the high speed SWP without requiring any additional source to heat
the coronal plasma electrons. In this model, the exobase is defined by the condition that the
density scale length equals the mean free path of the SWP particles. Due to the complexities
of coronal based physics and multiplicities of plasma species different exobases are likely to
coexist. Moreover, the appropriate electrostatic potential is determined by applying the
approximations of both local quasi-neutrality and zero current. Of course, this model has
succeeded in explaining the observation of the high-speed SWP. But according to our model
calculations, the zero-current approximation of the exospheric model seems to be valid only
on large scales but not near and above the SSB.
By our GES-model analyses, a finite electron-dominated [10-11] current with a positive finite
divergence exists on the solar interior scale for d d 0 < . Immediately after the SSB, i.e., on
the unbounded scale of the SWP, a divergence-free current exists. This seems to exhibit a
discontinuous behavior. How to resolve this? In reality electron temperature has variable
profiles on both the bounded and unbounded scales. Probably a self-consistent profile of
two distinct electron temperatures on two regions of bounded and unbounded scales
separately may resolve the interfacial transition problem of the proposed two-scale theory of
the GES-associated solar plasma current system.
According to our GES theory and model calculations, the zero-electric current
approximation of exospheric model calculations [38-39] requires further review. The
appropriate electrostatic potential estimate from numerical analysis emphasizes that the
zero electric current approximation is an outcome of the GES model on the large scale of the
SWP. Now the question may naturally arise, What happens to the SWP current after the
transonic transition? It seems that the electron-dominated electric current dissipates mainly
through a channel of inertial resistance of the plasma ions due to solar gravity as a barrier.


118
The other dissipation channels of the electric current may be through the SWP heating
generation of fluctuations in thermal noise level, etc. The uniform flow region of the SWP is,
in addition, found to have a large number of conservation rules [11] under the lowest order
inertial correction of the thermal electrons in the solar plasma system approximately from
applied mathematical point of view. The details of the associated physical mechanism and
fluctuations will be communicated to somewhere else.
5.7 Comments
Before we conclude with any physical comment, we must admit that the neglect of
collisional dissipation and deviation from a Maxwellian velocity distribution of the plasma
particles is not quite realistic. But our GES model under these simple approximations may
provide quite interesting results. For example, it provides deep physical insight into the
interconnection between the Sun and the SWP. The violation of the zero-current
approximation is indeed noted in the neighborhood of the SSB and above. Of course, the
zero current approximation seems to be satisfied beyond the transonic region. This
conclusion is based on the well-known condition of the floating surface boundary in basic
plasma sheath physics.
An estimated value
De J
20
~10

of the ratio of the solar plasma Debye length and the Jeans
length of the total solar mass justifies the quasi-neutral behavior of the solar plasma on both
the bounded and unbounded scales.
Applying the spherical capacitor charging model, the coulomb charge on the SWP at a
distance of ~ 1 AU comes out to be . For rotation frequency of the solar plasma system
corresponding to the mean angular frequency about the centre of the system (Gunn 1931),
the mean value of the strength of the solar magnetic field associated with the SWP in our
model analysis is estimated as
SWP SWP SWP
B Q f T
2 11
4 ~3.01 10 = . This is obviously
considerably higher for producing any significant effects on the dynamics of the SWP
particles. Thus the effects of the magnetic field are not ignorable for the SWP particles due to
the significantly strong Lorentz force, which is now estimated to be
( )
L SWP
F e v B N
2
0
1.64 10
= corresponding to a supersonic flow speed v kms

1
0
~ 340.00

.
Thus the Lorentz force may have some remarkable effect on the SWP particles and hence,
may not justifiably neglected for the unbounded scale description. It justifies the convective
and circulation dynamics to be considered in that context. Therefore our unmagnetized
plasma approximation may not prove well justified in our GES model configuration for the
SWP flow dynamics description. Although collision processes are dominant in the realistic
solar interior [2, 39-44], collisionless models [2, 39] are also equally useful for the solar
plasma description. Thus our collisionless model approximation for mathematical simplicity
may be justified here. In our GES model, the calculated values of the mean free paths for the
solar plasma electrons,
e
m
198
~ 1.50 10 and for ions,
i
m
132
~ 3.05 10 justify the
collisionless model approximation. This approximation holds good justifiably under the
fulfillment of the validity condition
e i J
, . >>
One can note that the SIP electron temperature, specified by
e
T
1
, differs from (exceeds) the
SWP electron temperature specified
e
T
2
by one order of magnitude. This is discussed
already discussed above. It simply means the SWP has been relatively cooled. It is quite
natural for expanding plasma gas to be cool. This is to further comment that these two
different electron temperatures are considered constant over their respective scales.

Acoustic Wave

119
Actually, a discontinuity exists at the interface of the bounded and unbounded scales. This
is an open problem to resolve.
Let us clarify once again that equalizing the maximum value of the numerically determined
solar self-gravity with the standard value specifies the SIP electron temperature. Similarly,
the appropriate choice of the defined constant a
0
specifies the SWP electron temperature,
which ensures that a transonic solution of the SWP dynamics exists. Now, with this
simplified treatment our theoretical model calculations yield the following conclusions.
1. Contrary to the general belief that the SWP emerges from the SSB, our theory provokes
us to argue that the genesis of the subsonic origin of the SWP at the SSB in fact lies in
the SIP dynamics. It is governed by the basic principles of the GES formation near the
SSB and beyond. The surface boundary is located at a radial distance defined by ~ 3.5
(Figs. 810) from the heliocentric origin. This specific location in the plots (Figs. 8a, 9a,
and 10a) is marked by a vertical line with small circles.
2. Thereafter, the outward moving SIP forms the SWP with a highly subsonic speed at the
SSB. Initially the outward acceleration of the SWP is quite rapid allowing a transonic
transition solution to exist for a specific choice of
T
0.0 0.1 = (Fig. 11a). This occurs as
a consequence of the predominantly self-consistent electric field associated with the
SWP (Fig. 11b). It produces a transonic transition region of sufficient length scale with the
critical point lying at about R 14

(Fig. 11c) from the SSB.
3. It is noted that initially the gravitational potential barrier decelerates the SIP dynamics
rapidly. As soon as the E-field of the SIP origin gathers sufficient strength, an outward
flow occurs with a reduced minimum speed of M
SSB
~10
-7
(Figs. 8c, 9c, 10c) at the SSB
defined by the quasi-hydrostatic equilibrium condition at a point of the maximum solar
gravity, as clearly depicted in Fig. 12. This figure clearly shows the strong solar self-
gravity up to the solar boundary and relatively weaker strength of the solar external
gravity beyond the boundary.
4. According to our model calculations, the SSB behaves as a negatively biased grid with a
bias potential of about 1 kV. The surface draws a finite current dominated by the
thermal electrons and flows towards the surface. As a result, the solar surface
oscillations may naturally be attributed to the resulting consequences of the GES
oscillations. Under the neutral ideal gas approximation of the SIP, this property cannot
be deduced.
5. We therefore conclude that our GES-based model may be useful to study the properties
of the SSB and the properties of the slow speed SWP. Of course, the properties of the
high speed SWP description under our model will require a kinetic treatment as already
reported by previous workers in the case of the exospheric model.
A few more reminders are in order:
1. The exact location of the SSB and that of 1 AU distance as specified in Figs. 8-12 on the
normalized scale are estimated for the normalization factor, which is, decided by the
SIP parameters.
2. In the absence of magnetic field in our model approach, the Lorenz force term is absent,
but it will be needed for further improvements under the fluid and/or kinetic regime to
see the realistic dynamics of the solar plasma system. However, the estimated mean
value of the solar magnetic field
SIP
B T
11
~7.53 10
in the SIP justifies and supports

our unmagnetized plasma approximation in the present context.


120
3. The GES-model is a useful theoretical construct with which to study SWP dynamics in
terms of solar interior dynamical behavior (generator of the SWP) through active
dynamical coupling processes of solar exterior regions in the light of localized electric
space charge effects.
Finally, it is important to comment that the further improvements and modifications to the
model will be needed to make it more realistic for actual SWP conditions. These form the
basic problems of future research on the GES model. The genesis of the SWP is now found
to be associated with the coupling of the SIP potential and self-gravitational potential of the
SIP mass. We finally argue that the lines of communications should be kept open between
theorists and observers and solar and stellar physicists, and more importantly also between
the solar and plasma physics communities, in order to further the study of stellar wind
plasmas. Ours is a first step, albeit very simplified and external-field free and ideal, in this
particular direction. We have tried to provide an integrated theoretical outlook on the SIP
dynamics on the bounded scale, and SWP on the unbounded scale. This model could further
be useful to study the properties of the helio-seismic dynamics of the Sun and other stars
[36-37] too.
5.8 Overall summary
The presented chapter reviews the latest findings of normal acoustic mode analyses through
different types of transonic plasma equilibrium models [5-12] under the lowest order inertial
correction of plasma thermal species. Different types of acoustic resonances are observed in
transonic plasma equilibria depending on different plasma inertial ions. The linear analyses
show the graphical nature of the associated resonance poles. This implies that transonic
plasma is an unstable zone, which is rich in wide range spectra of acoustic wave
fluctuations. The acoustic wave kinetics in the nonlinear normal mode analyses in different
types of plasmas [8-9, 12, 26] is describable by a linear source driven KdV (d-KdV) equation.
After integration, it shows two distinct classes of soluations, i. e., solitons and oscillatory
shocks. The fundamental condition to observe inertia-induced (ion) acoustic wave resonant
excitation is that the ion flow speed must be uniform. Accordingly, the same applies to the
solar wind dynamics [10-11, 35-41] in self-gravitating plasma systems as well. A large number
of conservation laws of applied mathematical significance associated with the d-KdV flow
dynamics are also pointed out [9] in transonic plasma domain in different situations including
solar plasma. Of course, convective and circulation dynamics which are the primary sources of
magnetic field [41], are neglected throughout for simplicity. Similar observations of acoustic
kinetics of the formation of soliton-type structures are also found in self-gravitating dust
molecular clouds in presence of partially ionized dust grains through the active mechanism of
gravito-acoustic coupling processes [27]. Some future scopes including realistic sources of
acoustic perturbation of the presented analyses are also pointed out in brief.
Very similar to Geoseismology dealing with the Earths interior through the various seismic
(acoustic) waves produced during the earthquakes, Helioseismology is the study of the
various linear and nonlinear surface waves and oscillations of solar origin (like p-mode, and
f-mode) to measure the internal structure and dynamics of the Sun [36-44]. The acoustic
dynamics in the Sun (or Star) is understandable by considering it to have a resonant cavity
like an organ pipe in which acoustic waves are trapped (by reflections or refractions) [41].
One of the earliest studies of solar oscillations and fluctuations established that the power
spectrum of the Suns full disk contained a multitude of Doppler shift peaks between 2.5
mHz - 4.5 mHz [36-37 and references therein]. The Global Oscillation Network Group

Acoustic Wave

121
(GONG), Stellar Observations Network Group (SONG), Helio- and Asteroseismology
(HELAS) Network, and Birmingham Solar Oscillations Network (BiSON) are examples of
recent studies being undertaken to measure these surface oscillations through space and
ground based remote-sensing observations [36-37, 44]. Michelson Doppler Imager (MDI)
onboard Solar and Heliospheric Observatory (SOHO) and recently, Helioseismic Magnetic
Imager (HMI) onboard Solar Dynamics Observatory (SDO) also measure these oscillations
from space [36-37, 41, 44]. Significant power has been observed at frequencies ranging from
1.4 mHz to 5.6 mHz, corresponding to periods of 3 to 12 minutes. They are called 5 minute
oscillations due to their dominant mean period [44]. Besides, the behavior of the solar
intermediate-degree modes (during extended minimum) is also investigated to explore the
time-varying solar interior dynamics with the help of contemporaneous helioseismic GONG
and MDI data [44]. The basic physics behind these helioseismic and helioacoustic observations
(in situ) reported in the literature, however, needs to be more clearly understood in a broader
horizon. Moreover, there are many more experimental observations [3, 35-44] on seismic
activities that will require self-gravitating plasma wave theory for further development of our
stability analyses and seismic diagnostics. In conclusion, we strongly believe that the
presented mathematical strategies and techniques of linear and nonlinear acoustic mode
analyses amidst more realistic plasma-boundary interaction processes may have some
potential applications in such future helio- and astero-seismic directions.
6. Acknowledgements
I am thankful to Ms. Sandra Bakic, Publishing Process Manager, InTech - Open Access
Publisher, for the invitation, chapter proposal and continuous cooperation. I am also
thankful to each and everybody of the InTech family whoever involved for extending
cooperation. Lastly, I gratefully recognize the Intech Editorial Board for giving me this rare
opportunity to publish this chapter in the book Acoustic Waves From Microdevices to
Helioseismology without any article processing charge.
7. References
[1] B. W. Tillery, Physical Sciences (8
th
Edition, McGraw-Hill 2009), Ch. 5, pp. 115-134.
[2] F. F. Chen, Introduction to Plasma Physics and Controlled Fusion (Plenum Press, New York
and London, 1988), Ch.1, pp. 1-17.
[3] S. R. Cranmer, Space. Sci. Rev. 101, 229 (2002).
[4] M. Lakshamanan and S. Rajasekhar, Nonlinear Dynamics: Integrability, Chaos, and Patterns
(Springer-Verlag Heidelberg 2003), Ch. 11-14, pp. 341-454.
[5] C. B. Dwivedi and R. Prakash, J. Appl. Phys. 90, 3200 (2001).
[6] P. K. Karmakar, U. Deka and C. B. Dwivedi, Phys. Plasmas 12, 032105 (2005).
[7] P. K. Karmakar, U. Deka and C. B. Dwivedi, Phys. Plasmas 13, 104702 (1) (2006).
[8] P. K. Karmakar and C.B. Dwivedi, J. Math. Phys. 47, 032901(1) (2006).
[9] P. K. Karmakar, J. Phys.: Conf. Ser. 208, 012059(1) (2010).
[10] C. B. Dwivedi, P.K. Karmakar and S.C. Tripathy, Astrophys. J. 663 (2), 1340 (2007).
[11] P. K. Karmakar, J. Phys.: Conf. Ser. 208, 012072(1) (2010).
[12] U. Deka, A. Sarma, R. Prakash, P. K. Karmakar and C. B. Dwivedi, Phys. Scr. 69, 303
(2004).
[13] P. Mora, Phys. Rev. Lett., 90, 185002-1 (2003).


122
[14] C. B. Dwivedi, Pramana-J. Phys. 55, 843 (2000).
[15] C. B. Dwivedi, Phys. Plasmas 6, 31 (1999).
[16] M. S. Sodha and S. Guha, Advances in Plasma Physics, edited by A. Simon and W. B.
Thompson (Wiley, New York, 1971) vol. 4.
[17] C. K. Goertz, Rev. Geophys. 27, 271 (1989).
[18] F. Verheest, Space Sci. Rev. 77, 267 (1996).
[19] H. Thomas, G. E. Morfill, V. Demmel, J. Goerce, B. Feuerbacher and D. Mohlmann, Phys.
Rev. Lett. 73, 652 (1994).
[20] G. S. Selwyn, J. E. Heidenreich and K. L. Haller, Appl. Phys. Lett. 57, 1876 (1990).
[21] L. Boufendi, A. Bouchoule, P. K. Porteous, J. Ph. Blondeau, A. Plain and C. Laure, J.
Appl. Phys. 73, 2160 (1993).
[22] A. Gondhalekar, P. C. Stangeby and J. D. Elder, Nucl. Fusion 34, 247 (1994).
[23] B. N. Kolbasov, A. B. Kukushkin, V. A. Rantsev Kartinov, and P. V. Romanov, Phys.
Lett. A 269, 363 (2000).
[24] N. N. Rao, P. K. Shukla and M. Y. Yu, Planet. Space Sci. 38, 345 (1990).
[25] P. K. Shukla, Waves in dusty, solar wind and space plasmas, (AIP Conference proceedings,
Leuven, Belgium) 537, 3 (2000).
[26] P. K. Karmakar, Pramana- J. Phys. 68, 631 (2007).
[27] P. K. Karmakar, Pramana- J. Phys. 76 (6), 945 (2011).
[28] J. P. Goedbloed, R. Keppens and S. Poedts, Space Sci. Rev. 107, 63 (2003).
[29] S. P. Kuo and D. Bivolaru, Phys. Plasmas 8 (7), 3258 (2001).
[30] K. Itoh, K. Hallatschek and S-I Itoh, Plasma Phys. Control. Fusion 47, 451 (2005).
[31] I. Ballai, PADEU 15, 73 (2005).
[32] I. Ballai, E. Forgacs and A. Marcu, Astron. Nachr. 328 (8), 734 (2007).
[33] P. K. Shukla and A. A. Mamun, New J. Phys. 5, 17.1 (2003).
[34] M. Khan, S. Ghosh, S. Sarkar and M. R. Gupta, Phys. Scr. T116, 53 (2005).
[35] M. S. Ruderman, Phi. Trans. R. Soc. A, 364, 485 (2006).
[36] W. J. Chaplin, Astron. Nachr. 331 (9), 1090 (2010).
[37] J. C. Dalsgaard, Astron. Nachr. 331 (9), 866 (2010).
[38] J. Lemaire and V. Pierrard, Astrophys. Space Sci. 277, 169 (2001).
[39] E. Marsch, Liv. Rev. Solar Phys. 3, 1 (2006) (http: //livingreviews.org/lrsp-2006-1).
[40] U. Narain and P. Ulmschneider, Space Sci. Rev. 75, 453 (1996).
[41] A. Nordlund, R. F. Stein and M. Asplund, Living Rev. Solar Phys. 6, 2 (2009) (http:
//livingreviews.org/lrsp-2009-2).
[42] R. Gunn, Phys. Rev. 37, 983 (1931).
[43] V. M. Nakariakov and E. Verwichte, Living Rev. Solar Phys. 2, 3 (2005) (http:
//livingreviews.org/lrsp-2005-3).
[44] S. C. Tripathy, F. Hill, K. Jain and J. W. Leibacher, Astrophys. J. 711, L84 (2010).
Part 2
Acoustic Waves as Investigative Tools

6
Acoustic Waves:
A Probe for the Elastic Properties of Films
Marco G. Beghi
Politecnico di Milano, Energy Department and NEMAS Center, Milano
Italy
1. Introduction
Films and thin films are exploited by an ever increasing number of technologies. The
properties of films can be different from those of the same material in bulk form, and can
depend on the preparation process, and on thickness. Specific techniques are needed for
their measurement. Whenever films or thin layers have structural functions, as in micro
electro-mechanical systems (MEMS), a precise characterization of their stiffness is crucial for
the design of devices. The same can be said for the design of devices which exploit thin
layers to support surface acoustic waves (surface acoustic wave filters). More generally,
knowledge of the elastic properties is interesting because such properties depend on the
structural properties.
The most widespread technique for the mechanical characterization of films, instrumented
indentation, induces both elastic and inelastic strains. It also characterizes irreversible
deformation, but the extraction of the information concerning the elastic behaviour is non
trivial. To overcome this difficulty, several measurement methods have been developed,
which exploit vibrations as a probe of the material behaviour. These methods intrinsically
involve only elastic strains, and are non destructive. This is true at any length scale, and is
peculiarly useful at micrometric and sub-micrometric scales, where the exploitation of other
types of probes can become critical.
Both propagating waves and standing waves can be exploited, with excitation which can be
either monochromatic (e.g. resonance techniques) or impulsive, therefore broadband,
requiring an analysis of the response either in the time domain (the so called picoseconds
ultrasonics) or in the frequency domain (the so called laser ultrasonics). The propagation
velocities of vibrational modes are obtained, from which the stiffness is derived if an
independent value of the mass density (the inertia) is available.
Older resonance techniques have been developed to be operated with thin slabs, also
exploiting optical measurements of displacement. In the measurement of films and small
structures, the advantages of light, a contact-less and inertia-less probe, are substantial, and
are increasingly exploited. The laser ultrasonics technique, commercially available since
some years ago, measures waves travelling along the film surface. The so called
picosecond ultrasonics technique measures waves travelling across the film thickness; it is
a relatively sophisticated optical technique, which exploits femtosecond laser pulses in a
pump-and-probe measurement scheme. For best performance it needs the deposition of


126
an interaction layer, possibly combined with microlithography techniques to obtain a
patterned layer.
Techniques for optical detection of acoustic vibrations include inelastic scattering of light:
Brillouin scattering. Brillouin spectroscopy does not excite vibrations, but relies on thermal
excitation, which has the broadest band but small amplitude, and measures the spectrum of
inelastically scattered laser light. Brillouin Spectroscopy and Surface Brillouin Spectroscopy
are relatively simple optical techniques, which do not require a specific specimen
preparation. They operate at sub-micrometric acoustic wavelengths, which are selected by
the scattering geometry, and have been exploited to characterize the elastic properties of
bulk materials and of films.
An overview of this variety of techniques is presented here, underlining analogies and
differences. The increasing demand of precise characterization raised the point of precision
and accuracy achievable by vibration based techniques, and specifically by the optical
techniques. In the overview, attention is drawn to the steps or the parameters which are the
limiting factor for the achievable precision, and to the way of characterizing them. The
effects of inaccuracies of the mass density are common to all the techniques based on
vibrations, while other sources of uncertainty are more specific to each technique.
2. Acoustic modes in elastic solids
In the continuum description the instantaneous configuration of a solid undergoing
deformation can be represented by the displacement vector field ( , ) t u r , where
( )
1 2 3
, , u u u = u , ( )
1 2 3
, , x x x = r and t is time. The local state of the solid being represented by
the strain and stress tensors, the linear elastic behaviour is characterized by the tensor of the
elastic constants
ijmn
C , which can be conveniently represented by the matrix of the elastic
constants C
ij
. When other phenomena, like e.g. viscoelasticity or elasticity of higher orders,
can be neglected, the tensor of the elastic constants fully characterizes the stiffness. Inertia is
characterized by the mass density . Within a homogeneous linear elastic solid, in the
absence of body forces, the equations of motion for the displacement vector field are
homogeneous, and read (Auld, 1990; Every, 2001; Kundu, 2004)

2 2
2
, ,
, 1, 2, 3
i m
ijmn
j m n j n
u u
C i
t x x

= =

.

(1)
These equations describe vibrational elastic excitations, which are typically called acoustic
also in the ultrasonic frequency range. The basic solutions of Eqs.(1), and the most important
ones when boundary effects are irrelevant, are the plane acoustic waves, or modes (Auld,
1990; Kundu, 2004), of the form
( ) { }
exp A i t = (

u e k r
,

(2)
where k is the wavevector, = 2 f the circular frequency, f the frequency, A
an arbitrary
complex amplitude, and e the polarization vector, which is normalized. The continuum
description, underlying Eq. (1), is appropriate until the wavelength 2 / = k is much
larger than the interatomic distances. The three translational degrees of freedom of each
infinitesimal volume element correspond, for each wavevector k, to three independent

Acoustic Waves: A Probe for the Elastic Properties of Films

127
modes, having different polarization vectors and frequencies. In general the phase velocity
/ v f = = k depends on both the direction of k and the polarization vector e. In an
infinite homogeneous medium, travelling waves of the type given by Eq. (2) exist for any
frequency f compatible with the mentioned lower limit for wavelength.
In the simplest case, the isotropic solid, the matrix of the elastic constants is fully determined
by only two independent quantities; the only non null matrix elements are C
11
= C
22
= C
33
,
C
44
= C
55
= C
66
, C
12
= C
13
= C
23
= C
11
- 2C
44
. In this case the shear modulus G coincides with
C
44
, while Young modulus E, Poissons ratio and bulk modulus B are respectively given
by (Every, 2001; Kundu, 2004)

( ) ( )
44 12 44 44 11 44
12 44 11 44
3 2 3 4 C C C C C C
E
C C C C
+
= =
+
, (3)

( )
12 11 44
11 12 11 44
2
1
2 2
C C C E
C C C C G

= = =
+
, (4)

11 12
11 44
2 4
3 3
C C
B C C
+
= = , (5)
In the isotropic case the phase velocities are independent from the direction of k, only
depending on the relative orientation of e with respect to k; one of the three modes is
longitudinal ( e k ) and has velocity
l
v , the other two are transversal ( e k ), are
independent (the two polarization vectors are orthogonal) and degenerate: they have the
same velocity
t
v (Auld, 1990; Kundu, 2004). The two velocities are

11
C /
l
v = and
44
C /
t
v = .

(6)
In the non isotropic case more than two independent quantities are needed to determine the
matrix of the elastic constants, and the phase velocities, beside depending on the direction
of k, have a more complex dependence on the C
ij
values.
In a finite geometry the search for standing waves having the harmonic time dependence of
the type
i t
e

transforms Eq. (1) into an eigenfunction / eigenvalue equation of the
Helmholtz type (Auld, 1990); an appropriate set of basis functions allows to transform this
equation into a matrix eigenvalue problem (Nakamura et al., 2004). The eigenvalues are
proportional to
2
, the square of the frequencies of the acoustic modes of the structure, or
natural frequencies of the structure. In other words, the finiteness of the geometry converts
the continuum spectrum of frequencies of the modes of the infinite medium, given by Eq.
(2), into the discrete spectrum of the natural frequencies. These frequencies depend on the
(C
ij
/) values and on the geometry.
In a schematic way: also in non isotropic media the acoustic velocities depend on stiffness
and inertia as in Eqs. (6):
2
/ v C = , indicating generically by C the relevant combination of
elastic constants and, possibly, direction cosines of k. In the simplest case, the one
dimensional geometry of length L, the standing waves are identified by the constructive self
interference condition /2 ( /2) / L n n v f = = (n is an integer number), such that
( /2) / / f n C L = . Therefore, a measurement of the frequencies of the acoustic modes


128
allows to derive ( ) ( )
2 2
/ /2 C fL n = . Also in more complex geometries, the dependence is
of the same type
( )
2
C fL N =

(7)
where L is now a characteristic length of the structure (for a slender rod, essentially one
dimensional, the length), and N is a dimensionless numerical factor which, beside the mode
order n, can depend on dimensionless quantities like geometrical aspect ratios or Poissons
ratio. The factor N also depends on the character of the mode whose frequency f is being
measured, and therefore on the specific modulus C which is involved.
Structures can be finite in one or two dimensions and practically infinite in others, as it
happens e.g. in a slab or a long cylinder. The free surface of an otherwise homogeneous
solid is a case of semi-infiniteness along a single dimension. The translational symmetry is
broken in the direction perpendicular to the surface, and new phenomena, absent in the
infinite medium, are found: the reflection of bulk waves, and the existence of surface
acoustic waves. Namely, at a stress free surface Eqs. (1) admit a further solution: the
Rayleigh wave, the paradigm of the surface acoustic waves (SAWs). Such waves have
peculiar characters (Farnell & Adler, 1972): a displacement field confined in the
neighborhood of the surface, with the amplitude which declines with depth, a wavevector
parallel to the surface, and a velocity lower than that of any bulk wave, such that the surface
wave cannot couple to bulk waves, and does not lose its energy irradiating it towards the
bulk. Pseudo surface acoustic waves can also exist, which violate this last condition. The
velocity
R
v of the Rayleigh wave cannot be given in closed form; in the isotropic case a
good approximation is (Farnell & Adler, 1972)

0.862 1.14
1
R t
v v

+
.

(8)
The continuum model of a homogeneous solid does not contain any intrinsic length scale.
Accordingly, all the solutions for this model are non dispersive, meaning that the velocities
(Eqs. (6) and (8)) are independent from wavelength (or frequency).
More complex modes occur in non homogeneous media. Layered media are a particularly
relevant case, in which new types of acoustic modes can occur; namely, modes confined
around the interfaces and modes which are essentially guided by one layer or another, like
the Sezawa waves. In this case, also in the continuum model the physical system has an
intrinsic length scale, identified by the layer thicknesses. For wavelengths much smaller
than the thicknesses wave propagation occurs within each layer as if it was infinite, with
reflections and refractions at the surfaces. Instead, for wavelengths comparable to, or larger
than, the thicknesses, the acoustic modes extend over several layers, and are modes of the
whole structure. Such modes are dispersive: their velocities depend on wavelength, or, more
precisely, on the wavelength to thickness ratio(s). Also the simplest surface wave, the
Rayleigh wave of a bare homogeneous substrate, is modified by a layer deposited on it, and
becomes dispersive: the propagation velocity depends on wavelength, therefore on
frequency. The velocities of the acoustic modes in layered structures can be numerically
computed, as non trivial functions of the properties of the substrate and the layer(s), and of
the wavelength to thickness ratio. The dispersion relations (k) or ( ) v f are thus obtained.


129
3. Stiffness measurements
3.1 Vibration based methods
It has always been recognized that since the phase velocities of acoustic waves and the
natural frequencies of the acoustic modes depend on stiffness and inertia, their
measurement gives access, by Eq.(6) or Eq. (7), to the elastic constants C
ij
, if the mass density
, and possibly the geometry, are known. Many experimental methods have been devised,
which exploit vibrations to measure the elastic properties of solids. These methods measure
the dynamic, or adiabatic, elastic moduli; these moduli do not coincide with the isothermal
moduli which are measured in monotonic tests (if strain rate is not too high), but in elastic
solids the difference between adiabatic and isothermal moduli seldom exceeds 1% (Every,
2001). Furthermore, when the elastic constants are needed to design a device which operates
dynamically, like most microdevices, the dynamic moduli are exactly the ones which are
needed in the design process.
Some methods measure the wave propagation velocity by measuring the transit time over a
finite, macroscopic distance, other methods measure the frequency of standing modes
defined by the sample geometry, or the frequency of propagating waves of well defined
wavelength. The excitation can be either monochromatic, at a frequency which typically
should be adjustable until resonance conditions are achieved, or broadband. The latter is
typically obtained by an impulsive excitation, which can be provided by a mechanical
percussion or by a laser pulse. Generally, the response to a broadband excitation is
spectrally analyzed. The availability of ultrafast lasers (femtosecond laser pulses) also allows
an analysis in the time domain, by pump-and-probe techniques.
In homogeneous specimens each propagation velocity, or the frequency of each standing
wave, has a single value, from which the corresponding elastic modulus can be derived. In
non homogeneous specimens, typically in supported films, each propagation velocity can
depend on wavelength or frequency. Dispersion relations (k) or ( ) v can be measured over
a finite interval of frequency or wavelength, and the film properties can be obtained fitting
the computed dispersion relations to the measured ones.
3.2 Vibration excitation and detection techniques
The various experimental methods operate in different frequency ranges. The range is
determined by both the excitation and the detection techniques, and is strictly correlated to
the spatial resolution. It is worth remembering that the acoustic velocities in typical elastic
solids like metals and ceramics are of the order of a few km/s = mmMHz, and that a phase
velocity v links the frequency f to a characteristic length L, which can be a characteristic
dimension of a structure supporting standing waves, or the wavelength of a travelling wave.
Characteristic lengths of centimetres imply frequencies in the tens of kHz range, which are
easily excited by a mechanical percussion and measured by a microphone; Nieves et al.
(2000) estimate at around 0.1 MHz the upper limit of the frequencies excited by the
mechanical percussion, with a steel ball of a very few millimetres. Piezoelectric actuators,
and sensors are also available.
Characteristic lengths of several micrometers correspond to frequencies in the tens of MHz
range. Structures of this size can be built by micromachining techniques, and their vibration
can be excited and detected by capacitive actuators and sensors. Measurement techniques of
this type are essentially a miniaturization of the vibrating reed technique (Kubisztal, 2008).
Czaplewski et al. (2005) built flexural and torsional resonators of tetrahedral amorphous


130
carbon (also known as amorphous diamond), by standard techniques for the production of
micro-electromechanical systems. They exploited piezo-electric actuation, and an
interferometric technique to measure the oscillation. They were able to perform
measurements at variable temperatures, determining the elastic moduli of this material as
function of temperature. They analyzed the uncertainty sources, finding that the leading
contribution to the uncertainty comes, for the flexural oscillator, from the value of the mass
density of this material, while for the torsional oscillator it comes from the exact dimensions
of the thin member undergoing torsion.
In larger structures, waves at frequencies in the tens of MHz range can be excited and
detected by piezoelectric elements, possibly operating simultaneously as actuators and
sensors. Excitation can also be performed by a laser pulse; if the pulse is short enough, the
upper limit of the measurable frequency range can be set by the piezoelectric sensor. Optical
detection techniques are also available. Specific devices like interdigitated transducers
(IDTs) can be built by lithographic techniques on, or within, an appropriate layer stack, which
must include a piezoelectric layer. Such devices emit and receive waves at the wavelength
which resonates with the periodicity of the transducer, typically at micrometric scale. This
configuration was exploited to measure the material properties (Bi et al. 2002; Kim et al., 2000),
but it is seldom adopted, because it requires the production of a dedicated micro device.
Micrometric and sub-micrometric wavelengths correspond to frequencies in the GHz to tens
of GHz range. Detection of such frequencies requires optical techniques; excitation of such
frequencies can be obtained by laser pulses of short enough duration.
The variety of vibration based methods to measure the stiffness of solids the can be
classified according to various criteria. In this chapter methods are reviewed grouping them
by the main vibration excitation techniques: mechanical excitation, either periodic or by
percussion, laser pulse excitation, and inelastic light scattering (Brillouin spectroscopy).
Similarly to Raman spectroscopy, Brillouin spectroscopy does not excite vibrations at all,
but relies on the naturally occurring thermal motion. This gives access to the broadest band,
but with small vibration amplitudes, which require time consuming measurements.
3.3 Precision and accuracy
In all the techniques based on vibrations the elastic constants themselves are not the direct
outcome of the measurement, but are derived from direct measurements of a primary
quantity like frequency or velocity, and auxiliary quantities like thickness, or mass density.
The uncertainty to be associated to the resulting value of each elastic constant must be
evaluated considering the uncertainties associated to each of the raw measurements. For a
quantity q which is derived from directly measured quantities a, b and c, the uncertainty
q

depends on the primary uncertainties
a
,
b
and
c
. For a functional dependence of the
type f Aa b c

= , where A is a numerical constant, the usual error propagation formula can
be written in terms of the relative uncertainties ( ) /
a
a , ( ) /
b
b , ( ) /
c
c as

2
2 2 2
2 2 2 q
a b c
q a b c

| |
| | | | | |
= + +
|
| | |
\ . \ . \ .
\ .

(8)
However, the various uncertainties can have different meanings and consequences. The
frequency is typically measured either identifying the frequency of a periodic signal which
achieves resonance, or by the spectral analysis of the response to a broadband excitation. In


131
both cases each frequency reading is associated to a finite degree of uncertainty, mainly due
to random errors. In a set of repeated measurements such errors are uncorrelated, and tend
to be cancelled by an averaging process; the error of each measurement affects the
dispersion of results around the average, but not the average itself. In other words, these
errors affect precision, but not accuracy. The accuracy of results is at most affected by the
finite accuracy in the calibration of the frequency meter, of whichever nature it be. When
then deriving the elastic moduli, the frequency reading can be exploited as such (see e.g.
Eq.(7)), or via the determination of a propagation velocity. In both cases, the obtained
moduli also depend on further auxiliary parameters. In a very simple case, from Eq. (6) we
have
2
11
C
l
v = , and the resulting value of C
11
is also affected by the finite uncertainty of
the best available value of , exploited in the derivation. However, in a set of repeated
measurements the uncertainty of does not contribute to the dispersion of results around
the average, but it affects the average itself. This means that it affects accuracy, but not
precision. The same can be said for the sample geometry (see Eq. (7)).
4. Mechanical excitation
Mechanical excitation can be either impulsive and broadband, as obtained by a percussion,
or narrow band, as provided by a periodic excitation. Most methods exploiting mechanical
excitation rely on the identification of the natural frequencies, or resonances, of a structure.
With a periodic excitation, such frequencies are identified scanning the excitation frequency
until resonance conditions (maximum oscillation amplitude for given excitation force) are
detected. With a broadband excitation the response (measured amplitude) is frequency
analyzed to identify the resonant frequencies.
Among the methods adopting harmonic excitation, acoustic microscopy (Zinin, 2001)
exploits a piezoelectric actuator, typically in the form of an acoustic lens, and often
operating also as a transducer. The acoustic lens is mechanically coupled to the sample by a
liquid drop. Acoustic microscopy can be operated with imaging purposes; in the
quantitative acoustic microscopy version (Zinin, 2001) it aims at measuring the acoustic
properties of the sample.
Beside acoustic microscopy, two main types of methods have been developed. The first one
measures the bulk properties. It adopts macroscopic homogeneous samples, and can exploit
either broadband or narrow band excitation. These methods have also been ruled by norms
(ASTM, 2008, 2009). A second group of methods, collectively called Resonance Ultrasound
Spectroscopy, aims at measuring the properties of thin supported films. It almost invariably
exploits a periodic excitation, whose frequency is swept in order to achieve resonance
conditions.
4.1 Measurement of bulk properties
Macroscopic homogeneous samples are self supporting. They can be tested as free standing
samples, provided the disturbances to free oscillations are minimized. Such a minimization
includes sample suspension by thin threads, or specimen support by adequate material
(cork, rubber), the supports having contact of minimum size, and positions at the nodes of
the fundamental vibrational mode of interest, either flexural or torsional. Also the sensor
contact, if oscillation is detected by a contact device, must be devised aiming at the
minimization of the disturbance induced by the contact. Non contact detection techniques


132
are available, including all optical techniques, and acoustic techniques: in the proper
frequency range, oscillations can be detected through the air, by a microphone, and even
excited, with harmonic excitation, by a similar technique, exploiting an audio oscillator and
an audio amplifier. The optical techniques, intrinsically contact-less and inertia-less, have
the broadest band, only limited by the light detection and analysis apparatus.
Since the full characterization of the elasticity of an isotropic medium requires two
independent parameters, it typically requires excitation of at least two modes of different
nature. For a specific simple geometry, the slender rod of length L, test methods have been
regulated by norms (American Society for Testing and Materials [ASTM], 2008, 2009, and
other ASTM norms cited by these two). Mainly flexural and torsional modes are considered,
for a slender rod of mass m, and either rectangular section of width b and thickness t, or
circular section of diameter D. Mass density being measured as ( ) m tbL / = , Eq.(7) takes in
this case, for the rectangular section, the forms (ASTM, 2008)
( ) ( ) ( ) ( )
f E
E m tbL f L L t T L t
2
2
/ 0.9465 / , / = (

(9a)
( ) ( ) ( )
t G
G m tbL f L T b t b L
2
/ 4 / , / = (

(9b)
where
f
f and
t
f are the frequencies of respectively the fundamental flexural and torsional
modes, and
E
T and
G
T are numerical factors, functions of the indicated dimensionless
quantities. Similar formulas hold for rods of circular section. The fundamental frequencies
can be identified by either sweeping the frequency of a periodic excitation (ASTM, 2008), or
by the broadband excitation by a mechanical percussion (ASTM, 2009), followed by the
frequency analysis of the response. In both cases the displacement can be sensed by a
contact transducer or by a microphone. The estimates for E and G being coupled by the
value of Poissons ratio (Eqs. 4 and 9a), an iterative procedure is indicated, to obtain
consistent estimates. The algorithm of Eq.(8) applied to Eq. (9a) gives

|
|
.
|
\
|
+ |
.
|
\
|
+ |
.
|
\
|
+ |
.
|
\
|
+ |
.
|
\
|
= |
.
|
\
|
f L b t m E
f
L b t m E

2
2
2
2 2
2
2 2
2 3 3
(10)
similar expressions being obtained for ( ) G
G
/ and for rods of circular section; such
equations allow to derive the uncertainties to be associated to the obtained moduli, from the
intrinsic uncertainties of the primary quantities. It was estimated (ASTM, 2008) that the
major sources of uncertainty come from the fundamental frequency f and from the smallest
specimen dimension (thickness or diameter). Uncertainties of the moduli in the 1% range are
achievable.
Other free standing sample geometries were considered, and analyzed by detailed finite
elements computations, to identify the appropriate values of the numerical factor N (see Eq.
(7)). Nieves et al. (2000) consider a cylinder with length equal to diameter (L = D). They
excite vibrations by a longitudinal percussion, and detect the displacement by an optical
technique; several modes are typically observed. Alfano & Pagnotta (2006) consider instead
a thin square plate, and, analyzing the displacement distributions of the first modes, they
identify the best positions for the plate supports and for perpendicular percussion. They
measure the response by a microphone. DEvelyn & Taniguchi (1999) similarly exploited


133
thin disks, exciting different modes by different impact points of a hollow zirconia bead,
and measuring the response by a microphone. They exploit computations of the resonant
frequencies of thin disks performed by others, and they estimate the accuracies of these
computations to 1 % or better.
In all these cases, from the numerical computations the mode frequencies are tabulated or
interpolated; Nieves et al. exploit the scaling parameter ( ) G L / / . The ratios of the
mode frequencies depend essentially on Poissons ratio: the ratios of the observed
frequencies allow therefore to identify the modes and to evaluate Poissons ratio. The
frequency values allow then, by the scaling parameter, to derive the shear modulus.
Both Nieves et al. and Alfano & Pagnotta perform detailed analyses of the measurement
uncertainties. They both find that, also with their experimental set-up, the frequency
measurement has a crucial role in the precision of the obtained moduli. Since Alfano &
Pagnotta consider thin plates, they find that the precision of the thickness t is also crucial,
simply because, thickness t being much smaller than the plate size a, a small value of the
relative uncertainty t
t
/ is more difficultly achievable than for a
a
/ .
For comparison purposes, Nieves et al. (2000), beside considering the axial modes, also
excite, by tangential percussion, the torsional modes, whose frequencies can be computed in
closed form. Once the torsional modes are discriminated from the bending modes, the
results agree to better than 1%, indicating a precision of this order. They also perform
measurements by the pulse-echo method, finding instead discrepancies of 2% or more; they
suggest that this method, which measures the propagation velocity of travelling waves, might
become intrinsically less accurate when performed in a confined geometry of small size.
4.2 Resonance ultrasound spectroscopy
The Resonance Ultrasound Spectroscopy (RUS) methods have been developed (Migliori et
al., 1993; Ohno, 1976; Schwarz et al., 2005; So et al., 2003) aiming in particular at the
measurement of the properties of supported thin films. A recent implementation (Nakamura
et al., 2004, 2010) exploits, as film support, a thin plate which, to be measured, is located on a
tripod. One of the three legs is rigid, and contains a thermocouple which monitors the
sample temperature; the second leg is a piezoelectric actuator, feeding a harmonic excitation
whose frequency is swept, the third one is a piezoelectric sensor, which detects the
oscillation amplitude. The resonance spectrum, i.e. the oscillation amplitude as function of
the frequency, is measured sweeping the excitation frequency. Several peaks are found,
sometimes with partial overlaps; their amplitudes, but not their frequencies, depend on the
position of the piezoelectric sensor. The measurement is precise enough to clearly detect the
difference between measurements in air and in vacuum, and the reproducibility of the
resonance frequencies is at the 0.1 % level (Nakamura et al., 2010). The elastic constants are
found fitting the computed frequencies to the measured ones; to this end, mode
identification is crucial. Identification is performed keeping the excitation frequency at
resonance, and scanning the specimen surface by a laser-Doppler interferometer. The map
of the out-of-plane displacement of the vibrating specimen is thus obtained, which allows an
unambiguous mode identification.
The elastic constants of the substrate are previously found by performing the same type of
measurement on a bare substrate, and the elastic constants of the film are derived from the
measured modifications of the resonance spectrum of the substrate. Since the vibration
amplitude of the standing waves in a plate is maximum at the surface, the sensitivity of the


134
method is better than the ratio of film thickness to the support thickness. In particular, the
sensitivity of each resonance frequency to each elastic constant is assessed evaluating
numerically the derivatives
ij
f C / , and from these values the uncertainties
ij
C are
derived from the estimated uncertainties f as

ij
ij
C f
f C
1
/
=

. (11)
Deposited thin films often have a significant texture, with one crystalline direction
preferentially oriented perpendicularly to the substrate surface, and random in-plane
orientations, resulting, at a scale larger than that of the single crystallite, in in-plane
isotropy, with different out-of-plane properties. This type of symmetry corresponds to the
hexagonal symmetry, in which the tensor of the elastic constants is determined by five
independent quantities. Among these, the resonance frequencies turn out to be almost
insensitive to C
44
, which therefore remains not determined, while the highest sensitivity is to
C
11
(Nakamura et al., 2010).
5. Laser pulse excitation
A laser pulse which is absorbed induces a sudden local heating which, by thermal
expansion, produces an impulsive mechanical loading. Such a mechanical impulse excites
waves in a broad frequency range, which are then detected. The accessible frequency band
can be limited by either the excitation bandwidth or the detection device. According to the
nature of the material being investigated, the deposition of an interaction layer can be
needed. In particular, a short absorption depth is required, in order to excite a pulse which
has small spatial and temporal duration, and therefore a broad band. The power density
threshold for ablation must also be considered: measurements are typically conducted with
high repetition rate pulses, and if ablation occurs the specimen undergoes a continuous
modification during the measurement.
Two main configurations have been developed up to maturity; they differ for the
propagation geometry and for the technique to detect vibrations. They are respectively
called laser ultrasonics and picoseconds ultrasonics.
5.1 Laser ultrasonics
The so called laser ultrasonics technique is mainly exploited to characterize thin supported
films. Oscillations are excited by a focused laser pulse, and propagation along the surface is
detected, measuring the surface displacement at a distance from excitation pulse ranging
from millimetres to centimetres. It is mainly the Rayleigh wave, modified by the presence of
the film, which is excited and detected. The laser pulse, typically of nanosecond duration, is
focused by a cylindrical lens on a line. The sudden expansion of this line source launches
surface waves of limited divergence, propagating along the surface, perpendicularly to the
focusing line. The component of the surface displacement normal to the surface itself can be
measured, at various distances from the line source, by optical interferometry (Neubrand &
Hess, 1992; Withfield et al., 2000), or, in a simpler and more robust way, by a piezoelectric
sensor (Lehmann et al., 2002; Schneider et al., 1997, 1998, 2000).
The recorded displacement is frequency analyzed, yielding the dispersion relation ) ( f v for
a frequency interval that can extend over a full frequency decade (e.g. 20 to 200 MHz).


135
Fitting the computed dispersion relation to the measured one allows to derive the film
properties. The width of the measured frequency interval can allow to to derive the Young
modulus and also the film thickness (Schneider et al., 1997, 2000). The uncertainties of the
results is evaluated numerically by the fitting procedure. Since the observed propagation
distance is of the order of millimetres, the obtained properties are representative of an
average over the propagation distance. The performance of the method could be pushed to
the measurement of the properties of diamond-like carbon films having thickness down to 5
nm (Schneider et al., 2000). By stretching the observed propagation path to 20 mm it was
possible to reduce the uncertainty of the measured propagation velocity of the Rayleigh
wave to below 0.25 m/s. It was thus possible to detect the small variation of the Rayleigh
velocity (5081 m/s for the bare (001) silicon substrate) induced by the presence of the film.
In a different configuration, the laser pulse is focused on a point instead of a line, with
consequent excitation of waves which expand in all the radial directions, and a common
path interferometer is adopted, whose light collection point scans the specimen surface
(Sugawara et al., 2002). It is thus possible to visualize the wavefronts, circular for isotropic
samples and non circular for anisotropic ones, with time resolution of the order of
picoseconds and spatial resolution of a few micrometres.
5.2 Picosecond ultrasonics
The so called picosecond ultrasonics technique owes its name to the picosecond laser pulses
which were available at the time it was introduced (Thomsen et al., 1984, 1986). It is
nowadays implemented by femtosecond laser pulses, and is intrinsically suited to
characterize thin supported films and multilayers. It follows the optical pump-and-probe
scheme (Belliard et al., 2009; Bienville et al., 2006; Bryner et al. 2006; Vollmann et al. 2002).
The pump beam, a femtosecond laser pulse, is focused, by a spherical lens, at the specimen
surface and, at least partially, absorbed. The focusing spot, a few to tens of micrometers
wide, is orders of magnitude larger than the characteristic lengths of the involved
phenomena: mainly the optical absorption length, and also the thermal diffusion length and
the acoustic wave propagation length, both for a femtosecond time scale. Bryner et al. (2006)
estimate that with an aluminium surface, a near infrared laser (800 nm), and a pulse width
of 70 fs, the absorption depth and therefore the dominant acoustic wavelength are of the
order of 10 nm. Quite consistently, for a similar laser pulse but for Pt and Fe ultra thin films,
Ogi et al. (2007) estimate at several THz the upper bound of the frequencies excited by the
laser pulse.
The thermal and mechanical fields are thus (almost perfectly) laterally uniform, and
essentially one dimensional: the absorbed pulse has a depth of the order of nanometres, and
propagates like an acoustic wave with a plane wavefront which travels perpendicular to the
surface, towards the specimen depth. At each interface this wave is partly transmitted and
partly reflected, according to the acoustical impedances of the layers, and gives rise to
echoes, which return to the surface, where they again are reflected back.
The surface is then probed by the probe beam, much weaker than the pump beam, which
reaches the surface with a variable delay, controlled by a delay line. The probe beam,
similarly to the acoustic wave, is partly transmitted and partly reflected at each interface,
leading, in each layer, to a forward and a backward electromagnetic field, whose complex
amplitudes can be calculated using a transfer-matrix formalism. The external reflectivity of
the surface is the ratio of the backward to the forward complex amplitudes in the outer


136
space. This reflectivity is modified by the acoustic strain, by two mechanisms. Firstly, each
interface is displaced by the acoustic wave, and, secondly, the refractive index in each layer
is modified by the elastic strain, by the acousto-optic (also called photoelastic) coupling. In
particular, it must be remembered that the traveling acoustic pulse extends over a depth of a
few nanometers, inducing a localized perturbation of the refractive index, which partially
scatters the optical beam. Interference can occur between the beam reflected at the outer
surface and that reflected by the traveling acoustic pulse.
Interferometric techniques allow to measure the variation of both amplitude and phase of
the reflected beam, thus measuring the variation of the complex reflection coefficient. With
probe pulses in the femtosecond range, and varying the probe pulse delay, the time
evolution of the surface reflectivity can be monitored with high temporal resolution. This
time evolution typically shows several features. The diffusion, towards the sample depth, of
the heat deposited by the laser pulse gives a slowly varying reflectivity background. The
echoes of the acoustic pulse which, after partial reflection at the film/substrate interface, are
again reflected at the outer surface are generally visible. The so called Brillouin oscillations,
due to the interference between the beam reflected at the outer surface and that reflected by
the traveling acoustic pulse, can then be found.
The analysis of the various features allows to characterize the waves which cross the layers
travelling perpendicularly to the surface. In the derivation of the film properties, the
knowledge of film thickness, typically obtained by X-ray reflectivity, has a crucial role; the
uncertainty about thickness is one of the leading terms in the uncertainty of the final results.
The achievable resolution depends on the excited wavelength. In copper the absorption
depth is larger than the value cited above for aluminium. Since the smallness of the absorption
depth determines the localization of the acoustic pulse and the achievable resolution, the
deposition of an aluminium interaction layer, which guarantees a very small absorption depth,
is a common practice. The interaction layer, typically a few tens of nanometres thick, then
participates to the vibrational behaviour of the structure being investigated. Accurate
measurements of stiffness therefore require consideration of the effects of the interaction
layer, e.g. by measurements with layers of different thicknesses, followed by an
extrapolation to null thickness (Mante et al., 2008). Obviously this deconvolution of the
effects of the interaction layer contributes to the uncertainty of the final results.
Near infrared lasers are a common choice, because at shorter wavelength more complex
phenomena can occur, which were attributed to electronic interband transitions (Devos &
Cote, 2004); obviously, if one is interested in elastic properties, electronic transitions are a
spurious effect to be avoided.
By picoseconds ultrasonics it was possible to characterize a layer stack, including a buried
layer of about 20 nm thickness (Bryner et al., 2006). The uncertainty for the elastic constants
of this layer is estimated at 20-25%, which is however remarkable for a layer of this type.
The lowest limit for layer detection is also estimated at about 10 nm thickness. In a different
configuration, namely a single Pt or Fe layer on a silicon substrate or a borosilicate glass
substrate, Ogi et al. (2007) were able to characterize metallic films of thickness down to 5
nm. They found, at nanometric thicknesses, a dependence of the elastic moduli on thickness.
This was explained by the impossibility of plastic flow at such low thicknesses: the elastic
strains can thus reach levels which are non reachable in thicker samples, such that higher
order elastic constants are no longer negligible.
Periodic Mo/Si multilayers (superlattices) were investigated by Belliard et al. (2009),
exploring various periodicities in the nanometric range. They detect bulk waves crossing


137
back and forth the multilayer; such waves show pulses of the order of 10 ps, which
correspond to propagation lengths much larger than the superlattice period. These waves
therefore see the whole multilayer as an effective medium. The theoretical prediction for the
properties of the effective medium and the reflection coefficient at the superlattice / Si
substrate interface are confirmed by the experimental findings. They also detect higher
frequency oscillations, which correspond to localized waves. The periodic multilayer acts as
a Bragg reflector, and opens forbidden gaps in the spectrum. It can confine a mode in the
neighbourhood of the outer surface (acoustic-phonon surface modes), but only if the outer
layer is the lower acoustic impedance layer (in this case Si), which therefore acts as a perfect
reflector. The properties of such modes could be correctly predicted only taking into account
the nanometric top silicon oxide layer, which spontaneously forms at he silicon surface. The
presence of this additional layer is also consistent with the X-ray reflectivity measurements.
The behaviour of a Mo cavity sandwiched between Mo/Si mirrors was also analyzed.
The picoseconds ultrasonics technique was also exploited to investigate non laterally
homogeneous specimens (Bienville et al., 2006; Mante et al., 2008). One limitation of this
technique in the measurement of the elastic constants is that it involves only plane waves
travelling perpendicular to the surface, thus allowing only the out-of-plane elastic
characterization of the film. To overcome this limitation, Mante et al. and Robillard et al.
(2008) proposed a technique by which the film to be characterized, and the aluminium
interaction layer deposited on it, are cut by lithographic techniques to obtain a periodic
square lattice of square (200 nm 200 nm) pillars. As confirmed by the measurements
performed on square lattices of different lattice constants, the pump pulse also excites, in
this nanostructured film, acoustic collective modes of the pillars, which propagate along the
surface in various directions. Various branches are measured, from which also the inplane
properties of the film can be measured, achieving a complete elastic characterization.
6. Brillouin spectroscopy
Aggregates of atoms, from molecules to clusters, to nanoparticles and nanocrystals, up to
mesoscopic and macroscopic aggregates, can interact with electromagnetic waves either
elastically or inelastically. Inelastic interactions include emission/absorption phenomena,
and inelastic scattering. We consider here inelastic scattering by vibrational excitations. At
the molecular scale the atomic structure of matter has a crucial role, and quantum
phenomena are relevant. At this level, vibrational excitations are the vibrations of molecules,
or, in a crystal, the vibrations of the internal degrees of freedom of each unit cell, which form
the so called optical branches of the dispersion relation, or optical phonons. Broadly
speaking, inelastic scattering by these excitations is called Raman scattering.
Aggregates above the nanometric scale also support collective vibrational excitations which
begin to resemble to acoustic waves, and can be described by a continuum model. In a
crystal, the vibrations of the degrees of freedom of the centre of mass of each unit cell form
the so called acoustic branches of the dispersion relation, or acoustic phonons. In the long
wavelength limit they are the are acoustic waves, accurately described by the continuum
model (Eq. (1)). Broadly speaking, inelastic scattering by these excitations is called Brillouin
scattering.
Visible light has sub micrometric wavelength. In media, either crystalline or amorphous,
which are homogeneous, and therefore translationally invariant, over at least a few
micrometres, vibrational excitations of sub-micrometric wavelength have a well defined


138
wavevector (see Eq. (2)). Due to translational invariance, the kinematics of scattering selects
the excitations whose wavevector is close to the Brillouin zone center (k = 0). The acoustic
and optical phonon branches have very different behaviours close to the zone center (k
0). The acoustic branch frequency
a
goes to zero, with phase velocity
a
/k and group velocity
/
a
k which tend to coincide with the sound velocity (which depends on polarization and,
in an anisotropic crystal, on the wavevector direction), while the optical branch frequency
o

typically goes to a maximum, with group velocity /
o
k which goes to zero.
Correspondingly, with visible light and with typical properties of solids, the two types of
branches produce inelastic scattering with frequency shifts ranging from a fraction of cm
-1

to a few cm
-1
(i.e from a few GHz to tens of GHz) for Brillouin scattering, and from
hundreds to thousands of cm
-1
(i.e from THz to tens of THz) for Raman scattering. The
spectral analysis of so widely different frequency ranges requires different types of
spectrometer. However, for both types of scattering the experiments are performed without
exciting the vibrations, but relying on the naturally occurring thermal motion.
Brillouin spectrometry thus offers a fully optical, and therefore contact-less, method to
measure the dispersion relations of bulk and surface acoustic waves, whose wavelength is
determined by the scattering geometry and the optical wavelength, and is typically sub-
micrometric. The frequency results from the medium properties, and typically falls in the
GHz to tens of GHz range. Measurements are performed illuminating the sample by a
focused laser beam, and analyzing the spectrum of scattered light, which is dominated by
the elastically scattered light, but also contains weak Stokes/anti-Stokes doublets due to
inelastic scattering by thermally excited vibrations (Beghi et al., 2004; Comins, 2001; Every,
2002; Grimsditch, 2001; Sandercock, 1982).
In sufficiently transparent materials scattering can occur in the bulk, by bulk acoustic waves.
The coupling mechanism is the elasto-optic (or acousto-optic) effect: the periodic
modulation of the refractive index by the periodic strain of the acoustic wave. In both
transparent and opaque materials scattering can also occur by surface acoustic waves, by the
ripple mechanism: the periodic corrugation of the surface due to the surface wave.
In more detail: the incident beam, of angular frequency
i
and wavelength
0
, impinges on a
sufficiently transparent sample with wavevector
i
q and is refracted into the wavevector
i
q .
Scattered light, of wavevector
s
q , emerges with wavevector
s
q . The probed wavevector,
( ) ' '
s i
= k q q , is determined by
0
, the directions of
i
q and
s
q , and the refractive index n.
Light inelastically scattered by a vibrational excitation of angular frequency (k) gives a
Stokes/anti-Stokes doublet at frequencies
s
=
i
. Detection, in the spectrum of
scattered light, of such a doublet allows to measure = |
s
-
i
| and to derive the
excitation velocity / v = k . In both transparent and opaque samples scattering occurring
by surface waves only depends on the components of wavevectors parallel to the surface:
the probed wavevector is ( )
s i
' ' = k q q

, and the surface wave velocity is v / = k
. In
other words, the spontaneous thermal motion can be viewed as spatially Fourier
transformed into an incoherent superposition of harmonic waves having all the possible
wavevectors; the scattering geometry (the directions of wavevectors
i
q and
s
q ) selects a
specific wavevector k or k
which is probed by the inelastic light scattering event.

Although also other geometries have been exploited (Beghi et al. 2011), in Brillouin
spectroscopy the most frequently adopted scattering geometry is backscattering:
s i
= q q .
For bulk scattering it corresponds to 2 '
i
= k q , such that


139

0
2
2 n
= k , (12)
which depends on the refractive index, but depends on geometry only when the sample is
anisotropic, while for surface scattering it means
i
2 = k q

, i.e.

0
2
2 sin
= k
, (13)
where is the incidence angle (the angle between the incident beam and the surface
normal). In this case the probed wavevector depends the incidence angle, but not on the
refractive index, because Snells law implies that upon refraction the optical parallel
components
||
q remain unchanged.
The data analysis for Brillouin spectroscopy results is common to all the methods, like laser
ultrasonics, which measure the velocity of travelling waves. In the simplest cases the
velocity is a function of the elastic constants which can be given in closed form. For instance,
if scattering by the longitudinal bulk wave, of velocity
11
C /
l
v = (Eq. (6)) is detected, C
11

is directly obtained as
2 2 2
11
C /
l
v k = = , and its uncertainty
11
C
is evaluated by Eq. (8).
In other cases, and namely in the case of supported films, the mode velocity can be
computed as function of the elastic constants only numerically. In that case it must be
remembered that the stiffness of an elastic solid is determined by as many independent
parameters as are needed to completely identify the tensor of the elastic constants. In other
words the stiffness is identified by a point in a multidimensional space, the dimensionality
being 2 in the simplest case of the isotropic medium, and being higher for lower symmetry
media.
Focusing here on the isotropic case, the stiffness can be represented, among other possible
choices, by the (E,) couple (see Eqs. (3) and (4)). In Brillouin spectroscopy, for each mode
the velocity is measured for various wavevectors k (in laser ultrasonics it is measured for
various frequencies) as ( )
m
v k with uncertainty
( ) v k
; since it can also be computed as
( ) , ,
c
v E k , the stiffness, represented by the (E,) couple, can be determined by a standard
least squares minimization procedure. The sum of squares is computed as
( )
( )
2
2
( )
, , ( )
,
c m
k v k
v E k v k
E

| |
= |
|
\ .
, (14)
where, for each wavevector k, the sum is further extended to all the detected acoustic
modes. Following standard estimation theory, the minimum of ( )
2
, E identifies the most
probable value
( )
, E of the ( ) , E couple, and the isolevel curves of the normalized
estimator ( ) ( ) ( ) ( )
2 2 2
, , / , E E E identify the confidence region at any
predetermined confidence level (Beghi et al., 2001, 2004, 2011; Lefeuvre et al., 1999). In some
cases a well defined minimum of ( )
2
, E is found, allowing a good identification of the
parameters (Beghi et al., 2001; Comins et al., 2000; Zhang et al., 2001a), while in other cases a
broad, valley-shaped minimum is found. In such cases a good identification of the
parameters is not possible (Beghi et al., 2002; Zhang et al., 1998), although sometimes some
combination of the parameters can be identified with better precision than individual


140
parameters (Comins et al., 2000; Zhang et al., 1998, 2001a). When several acoustic modes are
measured, the larger amount of available information allows a precise and complete elastic
characterization, as obtained e.g. for SiC films of micrometric thickness (Djemia et al, 2004).
In Eq. (14) each value ( )
m
v k , being obtained as / k, has an uncertainty
( ) v k
which depends
in turn on the uncertainty of the frequency of each spectral doublet, and on the precision of
the incidence angle (see Eq. (13)) or of the refractive index (see Eq. (12)). It can be noted that
the uncertainties of frequency and angle are of the random type, which affects precision but
not accuracy, while the uncertainty of the refractive index affects accuracy but not precision
(see Section 3.3). As with other techniques, also the uncertainties concerning the mass
density and the layer thickness(es) affect accuracy but not precision. These uncertainties
were the object of detailed investigations (Beghi et al., 2011; Stoddart et al., 1998). The effects
of the uncertainties of the quantities which are directly measured (primary uncertainties)
on the values of the elastic constants which are finally obtained were evaluated. It was
found that with appropriate sets of measurement uncertainties at the 1% level are reachable
(Beghi et al., 2011).
As already noted, Brillouin spectroscopy measures the acoustic modes at frequencies of the
order of GHz to tens of GHz, therefore at wavelengths much shorter than those
corresponding to frequencies of tens to hundreds of MHz, typically observed with
piezoelectric excitation and/or detection. This gives Brillouin spectroscopy an intrinsically
higher sensitivity to the properties of films, or to the perturbation induced by the presence
of very thin films. Brillouin spectroscopy was exploited to characterize tetrahedral
amorphous carbon films of thicknesses of hundreds of nanometres (Chirita et al., 1999), tens
of nanometres (Ferrari et al., 1999), down to a few nanometres (Beghi et al., 2002). It was also
shown that inelastic light scattering can be sensitive to nanometric thickness differences
(Lou et al., 2010). By Brillouin spectroscopy it was also possible to characterize buried layers
in silicon-on-insulator structures (Ghislotti & Bottani, 1994).
On the other hand, the techniques which excite vibrations operate with oscillation
amplitudes significantly larger than those measured by Brillouin spectroscopy; this allows
more precise measurements of frequencies, which at least partially compensates for the
lower intrinsic sensitivity due to the larger wavelengths. Combinations of techniques were
also exploited: thicker tetrahedral amorphous carbon films films (3 micron) were
characterized combining Brillouin spectroscopy and laser ultrasonics. A wide range of
frequencies was thus accessible, allowing a detailed characterization of the elastic properties
of the film (Berezina et al., 2004). A combination of Brillouin spectroscopy and picosecond
ultrasonics was instead exploited to characterize superlattices formed by periodic
multilayers of permalloy/alumina, with various periodicities at the nanometric scale
(Rossignol et al., 2004). Picosecond ultrasonics characterizes the out-of-plane properties by
waves travelling normal to the surface, while Brillouin spectroscopy characterizes the in-
plane properties by waves travelling along the surface. The combination of techniques
elucidated the effects of the interfaces.
Another algorithm for data analysis, different from that outlined above, has also been
recently proposed (Every et al., 2010). Both algorithms refer to the types of waves most
frequently measured in Brillouin spectroscopy of films or layered structures: surface
acoustic wave, or pseudo surface acoustic waves, which essentially travel parallel to the
surface, or however have a significant wavevector component parallel to the surface. It can
also be mentioned that it was also possible, by Brillouin spectroscopy, to detect standing


141
acoustic waves trapped within a film, which are reflected back and forth, crossing the film
perpendicularly to its surface (Zhang, 2001b).
Brillouin spectroscopy lends itself to the characterization of structures other than films or
layers. In particular, single-walled carbon nanotubes were characterized, measuring
Brillouin scattering by a free-standing film of pure, partially aligned, single-walled
nanotubes, and analyzing the results in terms of continuum models (Bottani et al., 2003).
The dependence of the measured spectra on the angle between the exchanged wavevector
and the preferential direction of the tubes shows that the tube-tube interactions are weak:
the tubes are vibrationally almost independent. The tubes are modelled as continuous
membranes, at two different levels: at the first one the membrane is infinitely flexible, only
able to transmit in-plane forces, while at the second level of approximation the tube wall is
treated as also able to transmit shear forces and torques not belonging to the shell surface.
In both cases scattering was essentially due to longitudinal waves travelling along the tubes.
Taking into account that AFM images suggest that the tube segments contributing to
scattering are not in the infinite tube length approximation, it was possible to derive the 2D
Young modulus for the tube wall, achieving the first dynamic estimation of the stiffness of
the tube wall. Scattering from carbon nanotubes was observed also in a different geometry,
with an ordered array of tubes, clamped at one end (Polomska et al., 2007).
Due to its intrinsic contact-less nature, Brillouin spectroscopy is the natural choice for the
measurement of elastic properties in conditions, like high temperature and/or high
pressure, in which physical contact with the specimen is difficult, if possible at all. Brillouin
spectroscopy only requires optical access, which can be obtained by an appropriate window,
and even in the extreme conditions achievable in a diamond anvil cell, optical access is
guaranteed by the transparency of the same diamond anvils.
Measurements were performed at high temperatures (Pang, 1997; Stoddart, 1995; Zhang et
al., 2001a), as well as at low temperatures, which were crucial to single out a particular
mechanism of hypersound propagation in alkali-borate glasses (Carini et al., 2008). After
pioneering experiments at high pressures (Crowhurst et al., 1999; Whitfield et al., 1976,), in
recent years dedicated Brillouin spectrometers were built at synchrotron facilities, allowing
simultaneous performance of high resolution x-ray diffraction and Brillouin spectroscopy
(and possibly other optical investigations like Raman spectroscopy, fluorescence,
absorption) on specimens subjected to extreme pressures in a diamond anvil cell, with
possible heating (Murakami et al., 2009; Prakapenka et al., 2010; Sinogeikin et al., 2006). This
set-up, of particular interest for geophysicists since it allows to characterize the behaviour of
minerals in the conditions which are found in the Earths interior, was exploited to perform
measurements on SiO
2
glass (Murakami & Bass, 2010) and other minerals, but also on
polymers (Stevens et al., 2007) and liquid methane (Li et al., 2010).
7. Conclusion
The stiffness of films, characterized by the elastic constants, depends on the film
microstructure, and its precise characterization is crucial when thin layers have structural
functions. The interest in the measurement of the elastic constants is witnessed by the
number of new techniques, or of improvements of existing techniques, being proposed.
The techniques which exploit either propagating acoustic waves or standing oscillations
involve exclusively elastic strains: they therefore offer the most direct and clean access to the
elastic properties, and potentially the most accurate measurements. Among the methods


142
based on vibrations, those which exploit, for excitation and/or detection, the contact-less
and inertia-less nature of light, have an important role.
An overview of the variety of existing methods was presented here, trying to present a
unified picture, and underlining the peculiarities of each of them, in particular for what
concerns the experimental uncertainties. It turns out that, under appropriate conditions and
experimental procedures, several techniques can achieve significant precision and accuracy.
8. References
Alfano, M. & Pagnotta, L., (2006). Measurement of the dynamic elastic properties of a thin
coating, Review of Scientific Instruments, Vol.77, Paper No. 056107
ASTM E1875-08 (2008). Standard Test Method for Dynamic Youngs Modulus,Shear Modulus, and
Poissons Ratio by Sonic Resonance, ASTM International, West Conshohocken, PA.
ASTM E1876-09 (2009). Standard Test Method for Dynamic Youngs Modulus, Shear Modulus,
and Poissons Ratio by Impulse Excitation of Vibration, ASTM International, West
Conshohocken, PA.
Auld, B. A. (1990). Acoustic fields and Waves in Solids, Robert E. Krieger Publishing Company,
Malabar, Florida
Beghi, M. G., Bottani, C. E. & Pastorelli, R. (2001). High accuracy measurement of elastic
constants of thin films by surface Brillouin scattering. In Mechanical properties of
structural films, C. Muhlstein, C. & Brown, S.B. (Eds.), pp. 109-126. ASTM STP 1413,
American Society for Testing and Materials, Conshohoken, PA
Beghi, M.G., Ferrari, A. C., Teo, K.B.K., Robertson, J., Bottani, C.E., Libassi, A. & Tanner, B.K.
(2002). Bonding and mechanical properties of ultrathin diamond-like carbon films,
Applied Physics Letters, Vol.81, pp. 3804-3806
Beghi, M.G., Every, A.G. & Zinin, P.V. (2004). Brillouin scattering measurement of SAW
velocities for determining near-surface elastic properties, in Ultrasonic
nondestructive evaluation, T. Kundu (Ed.), pp. 581-651, CRC Press, Boca Raton, FL.
Revised edition in press (2011)
Beghi, M. G., Di Fonzo, F., Pietralunga, S., Ubaldi, C. & Bottani, C. E. (2011). Precision and
accuracy in film stiffness measurement by Brillouin spectroscopy, Review of
Scientific Instruments, Vol.102, Paper No. 053107
Belliard, L., Huynh, A., Perrin, B., Michel, A., Abadias, G. & Jaouen, C. (2009). Elastic
properties and phonon generation in Mo/Si superlattices, Physical Review B Vol.80,
Paper No. 155424
Berezina, S., Zinin, P. V., Schneider, D., Fei, D. & Rebinsky, D. A. (2004). Combining
Brillouin spectroscopy and laser-SAW technique for elastic property
characterization of thick DLC films, Ultrasonics, Vol.43, pp. 87 - 93
Bi, B., Huang, W.-S., Asmussen, J. & Bolding, B. (2002). Surface acoustic waves on
nanocrystalline diamond, Diamond and Related Materials, Vol.11, pp. 677--680.
Bienville, T., Robillard, J.F., Belliard, L., Roch_Jeune, I, Devos, A. & Perrin, B. (2006).
Individual and collective vibrational modes of nanostructures studied by
picosecond ultrasonics, Ultrasonics, Vol.44, pp. e1289-e1294
Bottani, C.E., Li Bassi, A., Beghi, M.G., Podesta, A., Milani, P., Zakhidov, A., Baughman, R.,
Walters, D.A. & Smalley, R.E. (2003). Dynamic light scattering from acoustic modes
in single-walled carbon nanotubes, Physical. Review. B , Vol.67, paper n. 155407


143
Bryner, J., Profunser, D.M., Vollmann, J., Mueller, E. & Dual, J. (2006). Characterization of
Ta and TaN diffusion barriers beneath Cu layers using picosecond ultrasonics,
Ultrasonics, Vol.44, pp. e1269-e1275
Carini, G., Tripodo, G. & Borjesson, L. (2008). Thermally activated relaxations and
vibrational anharmonicity in alkali-borate glasses: Brillouin scattering study.
Physical Review B, Vol.78, paper n. 024104
Chirita, M., Sooryakumar, R., Xia, H., Monteiro, O. R. & Brown, I. G. (1999). Observation of
guided longitudinal acoustic modes in hard supported layers. Physical Review B,
Vol.60, pp. 5153-5156.
Comins, J. D., Every, A. G., Stoddart, P. R., Zhang, X., Crowhurst, J. C. & Hearne, G. R.
(2000). Surface Brillouin scattering of opaque solids and thin supported films.
Ultrasonics, Vol.38, pp. 450-458.
Comins J.D. (2001). Surface Brillouin scattering, in Handbook of Elastic Properties of Solids,
Liquids, and Gases, M. Levy, H. E. Bass and R. R. Stern & V. Keppens (Eds.), Volume
I: Dynamic Methods for Measuring the Elastic Properties of Solids, pp. 349-378,
Academic Press, New York
Crowhurst, J.C., Hearne, G.R., Comins, J.D., Every, A.G. & Stoddart, P.R.. (1999). Surface
Brillouin scattering at high pressure: Application to a thin supported gold film.
Physical Review B, Vol.60, pp. R14990-R14993
Czaplewski, D.A., Sullivan, J.P., Friedmann, T.A. & Wendt, J.R. (2005). Temperature
dependence of the mechanical properties of tetrahedrally coordinated amorphous
carbon thin films, Applied Physics Letters, Vol.87, paper n. 161915
DEvelyn M.P. & Taniguchi, T. (1999). Elastic properties of translucent polycrystalline cubic
boron nitride as characterized by the dynamic resonance method, Diamond and
Related Materials, Vol.8, pp. 1522-1526
Devos, A. & Cte, R. (2004). Strong oscillations detected by picoseconds ultrasonics in
silicon: evidence for an electronic structure effect, Physical Review B, Vol.70, paper n.
125208
Djemia, P., Roussign, Y., Dirras, G.,F. & Jackson, K.M. (2004). Elastic properties of SiC films
by Brillouin light scattering, Journal of Applied Physics, Vol. 95, pp. 2324-2330
Every, A.G. (2001). The Elastic Properties of Solids: Static and Dynamic Principles, In:
Handbook of Elastic Properties of Solids, Liquids, and Gases, M. Levy, H. Bass, R. Stern
& V. Keppens (Eds.), Volume I: Dynamic Methods for Measuring the Elastic Properties
of Solids, pp. 3-36, Academic Press, New York
Every, A.G. (2002). Measurement of the near surface elastic properties of solids and thin
supported films, Measurement Science and Technology, Vol.13, pp. R21-R39
Every, A.G., Kotane, L.M. & Comins, J. D. (2010). Characteristic wave speeds in the surface
Brillouin scattering measurement of elastic constants of crystals. Physical Review B,
Vol.81, paper n. 224303
Farnell, G.W. & Adler, E.L. (1972). Elastic wave propagation in thin layers. In: Physical
Acoustics, W.P. Mason & R.N. Thurston (Eds.), Vol. 9, pp. 35-127, Academic, New
York
Ferrari, A. C., Robertson, J., Beghi, M. G., Bottani, C. E., Ferulano, R. & Pastorelli, R. (1999).
Elastic constants of tetrahedral amorphous carbon films by surface Brillouin
scattering, Applied Physics Letters, Vol.75, pp. 1893-1895.


144
Ghislotti, G. & Bottani, C. E. (1994). Brillouin scattering from shear horizontal surface
phonons in silicon-on-insulator structures - Theory and experiment, Physical Review
B, Vol.50, pp. 12131-12137.
Grimsditch, M., (2001). Brillouin scattering, in Handbook of Elastic Properties of Solids, Liquids,
and Gases, M. Levy, H. E. Bass and R. R. Stern & V. Keppens (Eds.), Volume I:
Dynamic Methods for Measuring the Elastic Properties of Solids, pp. 331-347, Academic
Press, New York
Kim, J. Y., Chung, H. J., Kim, H. J., Cho, H. M., Yang, H. K. & Park, J. C. (2000). Surface
acoustic wave propagation properties of nitrogenated diamond-like carbon films,
Journal of Vacuum Science and Technology A, Vol.18, pp. 19931997
Kubisztal, M., Kubisztal, J., Chrobak, A., Haneczok, B., Budniok, A. & Rasek, J. (2008).
Elastic properties of Ni and Ni + Mo coatings electrodeposited on stainless steel
substrate, Surface and Coatings Technology, Vol.202, pp. 22922296
Kundu, T. (2004). Mechanics of elastic waves and ultrasonics non-destructive evaluation, in
Ultrasonic nondestructive evaluation, T. Kundu (Ed.), pp. 1-142, CRC Press, Boca
Raton, FL. Revised edition in press (2011)
Lefeuvre, O., Pang, W., Zinin, P., Comins, J.D., Every, A.G., Briggs, G.A.D., Zeller, B.D. &
Thompson, G.E. (1999). Determination of the elastic properties of a barrier film on
aluminium by Brillouin spectroscopy. Thin Solid Films, Vol.350, pp. 53-58.
Lehmann, G., Hess, P., Weissmantel, S., Reisse, G., Scheible, P. & Lunk, A. (2002). Youngs
modulus and density of nanocrystalline cubic boron nitride films determined by
dispersion of surface acoustic waves, Applied Physics A, Vol.74, pp. 4145
Li, M., Li, F.F., Gao, W., Ma, C.L., Huang, L.Y., Zhou, Q.A. & Cui, Q.L. (2010). Brillouin
scattering study of liquid methane under high pressures and high temperatures.
Journal of Chemical Physics, Vol.133, paper n. 044503
Lou, N., Groenen, J., Benassayag, G. & Zwick, A. (2010). Acoustics at nanoscale: Raman-
Brillouin scattering from thin silicon-on-insulator layers, Applied Physics Letters,
Vol.97, article n. 141908
Mante, P.A., Robillard, J.F. & Devos, A. (2008). Complete thin film mechanical
characterization using picoseconds ultrasonics and nanostructured transducers:
experimental demonstration on SiO
2
, Applied Physics Letters, Vol.93, paper n. 071909
Migliori, A., Sarrao, J.L., Visschera, W.M., Bella, T.M., Leia, M., Fisk, Z. & Leisure, R.G.
(1993). Resonant ultrasound spectroscopic techniques for measurement of the
elastic moduli of solids, Physica B, Vol.183, pp. 124.
Murakami, M., Asahara, Y., Ohishi, Y., Hirao N. & Hirose, K. (2009). Development of in situ
Brillouin spectroscopy at high pressure and high temperature with synchrotron
radiation and infrared laser heating system: Application to the Earth's deep
interior. Physics of the Earth and Planetary Interiors, Vol.174, pp. 282-291
Murakami, M. & Bass, J.D. (2010). Spectroscopic Evidence for Ultrahigh-Pressure
Polymorphism in SiO
2
Glass. Physical Review Letters, Vol.104, pp. 025504.
Nakamura, N., Ogi, H. & Hirao, M. (2004). Resonance ultrasound spectroscopy with laser-
Doppler interferometry for studying elastic properties of thin films, Ultrasonics,
Vol.42, pp. 491494
Nakamura, N., Nakashima, T., Oura, S., Ogi, H. & Hirao, M. (2010). Resonant-ultrasound
spectroscopy for studying annealing effect on elastic constant of thin film,
Ultrasonics, Vol.50, pp. 150154


145
Neubrand, A. & Hess, P. (1992). Laser generation and detection of surface acoustic waves:
Elastic properties of surface layers, Journal of applied physics, Vol.71, pp. 227-238
Nieves, F.J., Gascn, F. & Bayn, A. (2000). Precise and direct determination of the elastic
constants of a cylinder with a length equal to its diameter, Review of Scientific
Instruments, Vol.71, pp. 2433-2439
Ogi, H., Fujii, M., Nakamura, N., Shagawa, T. & M. Hirao, M. (2007). Resonance acoustic-
phonon spectroscopy for studying elasticity of ultrathin films, Applied Physics
Letters, Vol. 90, paper n. 191906.
Ohno, I. (1976). Free vibration of a rectangular parallelepiped crystal and its application to
determination of elastic constants of orthorhombic crystal, Journal of Physics of the
Earth, Vol.24, pp. 355379.
Pang, W., Stoddart, P.R., Comins, J.D., Every, A.G., Pietersen, D. & Marais P.J. (1997). Elastic
properties of TiN hard films at room and high temperatures using Brillouin
scattering, International Journal of Refractory Metals and Hard Materials, Vol.15, pp.
179-185.
Polomska, A.M., Young, C.K., Andrews, G.T., Clouter, M.J., Yin, A. & Xu, J. M. (2007).
Inelastic laser light scattering study of an ordered array of carbon nanotubes.
Applied Physics Letters, Vol.90, paper n. 201918
Prakapenka, V. (2010). On-line Brillouin Spectroscopy at GSECARS: Basic Principles and
Application for High Pressure Research, Synchrotron Radiation News Vol.23, pp. 14-
15.
Robillard, J.-F., Devos, A., Roch-Jeune, I. & Mante, P. A. (2008). Collective acoustic modes in
various two-dimensional crystals by ultrafast acoustics: Theory and experiment,
Physical Review B, Vol.78, paper n. 064302
Rossignol, C., Perrin, B., Bonello, B., Djemia, P., Moch, P., & Hurdequint, H. (2004). Elastic
properties of ultrathin permalloy/alumina multilayer films using picosecond
ultrasonics and Brillouin light scattering, Physical Review.B, Vol.70, paper n. 094102
Sandercock J.R. (1982). Trends in Brillouin scattering Studies of opaque materials,
supported films, and central modes, in Light Scattering in solids III, M. Cardona and
G. Gntherodt (Eds.), pp. 173-206, Springer, Berlin
Schneider, D., Schwarz, T., Scheibe, H.-J. & Panzner, M. (1997) Non destructive evaluation
of diamond-like carbon films by laser induced surface acoustic waves, Thin Solid
Films, Vol.295, pp. 107116
Schneider, D., Schultrich, B., Scheibe, H.-J., Ziegele, H. & Griepentrog, M. (1998). A laser
acoustic method for testing and classifying hard surface layers, Thin Solid Films,
Vol. 332, pp. 157163.
Schneider, D., Witke, T.H., Schwarz, T.H., Schoneich, B. & Schultrich, B. (2000). Testing
ultra-thin films by laser-acoustics, Surface and Coatings Technology, Vol.126,
pp.136-141
Schwarz, R.B., Hooks, D.E., Dick, J.J. & Archuleta, J.I. (2005). Resonant ultrasound
spectroscopy measurement of the elastic constants of cyclotrimethylene
trinitramine, Journal of Applied Physics, Vol.98, paper N. 056106.
Sinogeikin, S., Bass, J., Prakapenka, V., Lakshtanov, D., Shen, G.Y., Sanchez-Valle, C &
Rivers, M. (2006). Brillouin spectrometer interfaced with synchrotron radiation for
simultaneous X-ray density and acoustic velocity measurements. Review of
Scientific Instruments, Vol.77, article n. 103905


146
So, J.H., Gladden, J.R., Hu, Y.F., Maynard, J.D. & Qi Li (2003). Measurements of elastic
constants in thin films of colossal magnetoresistance material, Physical Review
Letters, Vol.90, paper N. 036103.
Stevens, L.L., Orler, E.B., Dattelbaum, D.M., Ahart, M. & Hemley, R. J. (2007). Brillouin
scattering determination of the acoustic properties and their pressure dependence
for three polymeric elastomers. Journal of Chemical Physics, Vol.127, article n. 104905
Stoddart, P.R., Comins, J.D. & Every, A.G. (1995). Brillouin-scattering measurements of
surface-acoustic-wave velocities in silicon at high-temperatures. Physical Review B,
Vol.51, pp. 17574-17578
Stoddart, P.R., Crowhurst, J.C., Every, A.G. & Comins, J. D. (1998). Measurement precision
in surface Brillouin scattering, Journal of the Optical Society of America B, Vol.15, pp.
2481-2489.
Sugawara, Y., Wright, O.B., Matsuda, O. & Gusev, V.E. (2002). Spatiotemporal mapping of
surface acoustic waves in isotropic and anisotropic materials, Ultrasonics, Vol.40,
pp. 55-59
Thomsen, C., Strait, J., Vardeny, Z., Maris, H.J., Tauc, J. & Hauser, J.J. (1984). Coherent
phonon generation and detection by picoseconds light pulses, Physical Review
Letters, Vol. 53, pp. 989-992
Thomsen, C., Grahn, H.T., Maris, H.J. & Tauc, J. (1986). Surface generation and detection of
phonons by picoseconds light pulses, Physical Review B, Vol. 34, pp. 4129-4138
Vollmann, J., Profunser, D.M. & Dual, J., (2002). Sensitivity improvement of a pump-probe
set-up for thin film and microstructure metrology, Ultrasonics, Vol.40, pp. 757763
Whitfield, C.H., Brody, E.M. & Bassett, W.A. (1976). Elastic moduli of NaCl by Brillouin
scattering at high pressure in a diamond anvil cell. Review of Scientific Instruments,
Vol.47, pp. 942-947
Whitfield, M. D., Audic, B., Flannery, C. M., Kehoe, L. P., Crean, G. M. & Jackman, R. B.
(2000). Charactrization of acoustic Lamb wave propagation in polycrystalline
diamond film by laser ultrasonics, Journal of Applied Physics, Vol.88, pp. 2984
2993.
Zhang, X., Comins, J.D., Every, A.G. & Stoddart, P.R. (1998). Surface Brillouin scattering
studies on vanadium carbide, International Journal of Refractory Metals and Hard
Materials, Vol.16, pp. 303-308
Zhang, X., Stoddart, P.R., Comins, J.D. & Every, A. G. (2001a). High-temperature elastic
properties of a nickel-based superalloy studied by surface Brillouin scattering,
Journal of Physics: Condensed. Matter, Vol.13, pp. 2281-2294
Zhang, X., Sooryakumar, R., Every, A.G. & Manghnani, M.H. (2001b). Observation of
organpipe acoustic excitations in supported thin films. Physical Review B, Vol.64,
paper n. 081402.
Zinin, P.V (2001). Quantitative acoustic microscopy of solids, in Handbook of Elastic Properties
of Solids, Liquids, and Gases, M. Levy, H. E. Bass and R. R. Stern & V. Keppens
(Eds.), Volume I: Dynamic Methods for Measuring the Elastic Properties of Solids, pp.
187-226, Academic Press, New York
7
Evaluation Method for Anisotropic Drilling
Characteristics of the Formation by Using
Acoustic Wave Information
Deli Gao and Qifeng Pan
China University of Petroleum at Beijing, Beijing
China
1. Introduction
In drilling engineering, we must have solid understanding of the underground geological
environment which is not only complicated and diversified but also someway concealed.
Thus, in order to find an effective method of predicting it, a long-term research and practice
must be required. In drilling engineering for oil & gas, problems including borehole
deviation & its control, wellbore instability & its control, influence directly drilling quality &
efficiency of a deep or complicated well for exploration and production of oil & gas fields.
For instance, because of the complicated surface and underground conditions as well as the
depth (over 5000 m) of oil & gas reservoirs in western China, such unstable factors such as
hole deviation and instability often encountered with each other in deep drilling
engineering. Because we did not have access to the geological parameters of the formations
to be drilled including the rock drillability anisotropy and so on, huge economic loss had
been caused and the steps to explore and produce oil & gas in western China had been
seriously restricted. Hence, there are many researches and development programs to do for
the right cognition and scientific evaluation of the geological environments, and for the
further study of mechanism of the drilling process instability, and so on. The solution of
these problems is the key to improve the performance of drilling & HSSE (health, safety,
security, environment) and lower well construction cost.
The factors influencing the instability can be sorted into subjective category and objective
category. In the objective category, there are the types of geological structure and in-situ
stress, rock anisotropy, porosity, permeability, lithology, pressures, and mineral
components, as well as rock strength and weak layer of the formation to be drilled, and so
on. In the subjective category, there are the performance of down hole drilling system, the
drilling parameters (weight and torque on bit, etc.), the drilling fluid performances (water
loss, viscosity, rheological property and density) and its hydration on shale, the direction
and open time of wellbore, the erosion and surge pressure of drilling fluid on the hole
wall, the interaction between drillstring and hole wall. Thereby, in researches on the
instability, the factors from the both categories should be taken into comprehensive
consideration.
Whether in vertical drilling or in directional drilling, it is always a complicated academic
and technological problem how to control the well trajectory exactly along the designed


148
track to reach the underground targets. In rotary drilling, the forming of wellbore & its
trajectory is the result of the rock-bit interaction. In this interaction, the drill bit anisotropy
and its mechanical behavior (i.e. the drill bit force and tilt angle) are important factors that
can directly affect the well trajectory. The mechanical behavior depends on by the bottom
hole assembly (BHA) analysis. Accordingly, principal factors influencing the well trajectory
generally contain BHA, drill bit, operating parameters in drilling, drilled wellbore
configuration and the formations to be drilled. Of which the BHA, drill bit and operating
parameters in drilling are the factors that can be artificially controlled, and the formation
property (such as rock drillability and its anisotropy) is the objective factor which can not be
changed by us. The trajectory can be predicted before drilling and also can be determined
after drilling through surveys and calculations. Besides, the drilled wellbore will not only
generate a strong reaction on the drill bit force and the drillstring deflection, but also will
exert an influence on the anisotropic drilling characteristics of the formation. Due to the
above-complicated factors, the hole deviation is always inevitable, which may seriously
influence the wellbore quality and the drilling performance.
The well trajectory control is the process which forces drill bit to break through
formations along the designed track forward by applying reasonable techniques. The
anisotropic drilling characteristics of the drill bit & the formation and their interaction
effects are the factors which will cause a direct influence on the well trajectory control.
Thereby, it is a complicated scientific and technological problem for us how to make the
cognition, evaluation and utilization of anisotropic drilling characteristics of the
formation, as well as the prediction & control of mechanical action of the drill bit on the
formations.
Rock drillability anisotropy of the formation to be drilled has significant effects on the well
trajectory control so that it is very important to evaluate it. Definitions of rock drillability
anisotropy and acoustic wave anisotropy of the formation to be drilled are presented in this
chapter. The acoustic velocities and the drillability parameters of some rock samples from
Chinese Continental Scientific Drilling (CCSD) are respectively measured with the testing
device of rock drillability and the ultrasonic testing system in laboratory. Thus, their
drillability anisotropy and acoustic wave anisotropy are respectively calculated and
discussed in detail by using the experimental data. Based on the experiments and
calculations, the correlations between drillability anisotropy and acoustic wave anisotropy
of the rock samples are illustrated through regression analysis. Whats more, the correlation
of rock drillability in directions perpendicular to and parallel to the bedding plane of core
samples is studied by means of mathematical statistics. Thus, a mathematic model is
established for predicting rock drillability in direction parallel to the formation bedding
plane by using rock drillability in direction perpendicular to the formation bedding plane
with the well logging or seismic data. The inversion method for rock anisotropy parameters
(,) is presented by using well logging information and the acoustic wave velocity in
direction perpendicular to the bedding plane of the formation is calculated by using acoustic
wave velocity in any direction of the bedding plane. Then, rock drillability in direction
perpendicular to the bedding plane of the formation can be calculated by using acoustic
wave velocity in the same direction. Thus, rock drillability anisotropy and anisotropic
drilling characteristics of the formation can be evaluated by using the acoustic wave
information based on well logging data. The evaluation method has been examined by case
study based on oilfield data in west China.
Evaluation Method for Anisotropic Drilling Characteristics
of the Formation by Using Acoustic Wave Information

149
2. Anisotropic drilling characteristics of the formation
Although many theories have been proposed to explain the hole deviation since the 1950s
(Gao et al, 1994), it is only the rock drillability anisotropy theory (Lubinski & Woods, 1953)
that was recognized by petroleum engineers and widely applied to petroleum engineering
because it can be used to quantify the anisotropic drilling characteristics of the formation
and to explain properly the actual cases of hole deviation encountered in drilling
engineering. The theory suggested that since values of rock drillability are not always the
same in the directions perpendicular and parallel to the bedding plane of the formation, the
formation will bring the bit a considerable force, which may likely cause changes on the
original drilling direction and hole deviation.
The orthotropic or the transversely isotropic formations are the typical formations
encountered frequently in drilling engineering. The anisotropic effects of the formations
(rock drillability) on the well trajectory must be considered in hole deviation control and
directional drilling. Based on the rock-bit interaction model, the formation force is defined
and modeled in this section to describe quantitatively anisotropic drilling characteristics of
the formations to be drilled.
2.1 Definition of rock drillability anisotropy
Because of rock drillability anisotropy, the real drilling direction does not coincide with the
resultant force direction of the drill bit (supposed that it is isotropic) on bottom hole. Besides
calculating the drill bit force by BHA (bottom hole assembly) analysis, rock drillability
anisotropy of the formation must be considered in hole deviation control.
The formation studied here is typical orthotropic one, and the transversely isotropic
formation discussed previously is regarded as its particular case. Let
d
e
,
u
e
and
s
e
represent unit vectors in the directions of inner normal, up-dip and strike of the
formation respectively, as shown in Fig.1.There are different physical properties along
different directions of them. in Fig.1 represents dip angle of the formation to be drilled.
Rock drillability anisotropy of the formation can be expressed by rock drillability anisotropy
index. If the components of penetration rate of the drill bit (isotropic) along inner normal,
up-dip and strike of the orthotropic formation are noted as
dip
R ,
str
R and
n
R respectively,
correspondingly the net applied forces are
dip
F ,
str
F and
n
F respectively, the rock
drillability can be defined as:

n
n
n
R
D
F
= ,
dip
dip
dip
R
D
F
= ,
str
str
str
R
D
F
= (1)
Rock drillability anisotropy of the orthotropic formation may be represented by two indexes
(
r1
I and
r2
I

) which are defined as:

dip
r 1
n
D
I
D
= ,
str
r 2
n
D
I
D
= (2)
Dip angle and strike of the formation can be obtained from the analysis of well logging and
geological structure survey. The values of
r1
I and
r2
I for the orthotropic formation can be
evaluated by the experimental analysis or using the acoustic wave information.


150

Fig. 1. Descartes coordinates for the formation geometry
2.2 The formation force
Assumed that the drill bit is isotropic for eliminating the effects of its tilt angle on hole
deviation, the effects of the orthotropic formation on hole deviation can be presented by the
formation force analysis. The two parameter equations related to the formation forces can be
derived from the rock-bit interaction model (Gao & Liu, 1989):

22 13 12 23
11 22 12 21
11 23 21 13
11 22 12 21
t t t t
G
t t t t
t t t t
G
t t t t
(3)
Where G
and G
are called as the building angle parameter (positive for building up the
inclination of well trajectory) and the drifting azimuth parameter (positive for left walking
of well trajectory) of the formation respectively , and the
ij
t (i, j=1,2,3) can be expressed as
(Gao & Liu, 1990):

( ) 1 1 2 1
1 ( )
0,
1,
ij r ij r ij r r ij
ij
t I I a I I c
i j
i j
= + +
(4)
where
ij ji
a a = ,
ij ji
c c = (i, j=1, 2, 3) can be calculated by the following equations:
dip angle
horizontal plane

151

( )
( )
( )( )
( )
( )
2
11
12
13
21
12 21
2
22
23
sin cos cos sin cos
cos sin cos sin cos sin sin
cos sin cos sin cos sin sin cos cos cos
cos sin cos sin cos sin sin
sin sin
sin sin sin sin cos cos cos
a
a
a
a
a a
a
a

=
=
= +
=
=
=
= +
( )
( )
31 13
32 23
2
33
sin sin cos cos cos
a a
a a
a
= +

)
(5)

( )
( )
2
11
12
2
13
21 12
2
22
23
31 13
32 23
2
33
sin cos
cos sin cos
(sin ) sin cos
(cos )
cos sin sin
sin sin
c
c
c
c c
c
c
c c
c c
c

=
`
=

)
(6)
Where = ; and are respectively azimuth and inclination of well trajectory on the
bottom hole; and are respectively dip angle and up dip azimuth of the formation to be
drilled.
It is obviously that the values of G
and G
are not only controlled by rock drillability

anisotropy of the formation, but also affected by the formation geometry and the well
trajectory. Therefore, G
and G

can be used to describe the anisotropic drilling
characteristics of the formation to be drilled. Thus, the formation force can be
mathematically defined as:

ob
ob
GF G W
GF G W

=
(7)
Where GF
and GF
are called as the inclination force (positive for building up the

inclination) and the azimuth force (positive for decreasing the azimuth) of the formation
respectively, and
ob
W is weight on bit. It should be pointed out that both GF
and GF
are
only an equivalent expression of anisotropic drilling characteristics of the formation and
they are completely different from the mechanical action forces of the drill bit on the
formation. Rock drillability anisotropy of the formation is the internal cause of the
generations of GF
and GF
, while weight on bit is the its external cause.


152
2.3 G
and G
of the transversely isotropic formation

By using equations (5) and (6) and making
r1 r2 r
I I I = = , equation (3) can be simplified as the
following expressions of G
and G
for the transversely isotropic formation:

( )( )( )
( ) ( ) ( )
r
2 2
r r
1 cos sin cos sin cos cos cos sin sin cos
1 sin sin cos sin cos sin cos
I
G
I I

+
=
(
+ +

(8)

( ) ( )
( ) ( ) ( )
r
2 2
r r
1 sin sin cos cos sin sin cos
1 sin sin cos sin cos sin cos
I
G
I I

+
=
(
+ +

(9)
Where all the symbols here express the same meanings as the previous ones.
3. Experiments on rock anisotropy
Evaluation of rock drillability anisotropy is necessary for hole deviation control in drilling
engineering. Many efforts have been made to evaluate rock drillability of the formation
through the core testing, the inverse calculation and the acoustic wave. Proposed in this
section is an alternative solution by using the acoustic wave to evaluate rock drillability
anisotropy of the formation. First, a correlation between the P-wave velocity anisotropy
coefficient and the rock drillability anisotropy index of the formation which are calculated
according to the core testing data in laboratory, is established by means of mathematical
statistics. Then, a mathematical model is obtained for predicting the rock drillability
anisotropy index by using the P-wave velocity anisotropy coefficient. Thus, rock drillability
anisotropy of the formation can be evaluated conveniently by using the well logging or
seismic data (Gao & Pan, 2006).
3.1 Rock drillability anisotropy
3.1.1 Definition
The transversely isotropic formation is a typical anisotropic formation, whose anisotropy
can be expressed by a rock drillability anisotropy index:

h
r
v
D
I
D
= (10)
where
v v v
D V F = and
h h h
D V F = are respectively rock drillability parameters in the
directions perpendicular and parallel to the bedding plane of the transversely isotropic
formation;
v
V &
v
F and
h
V &
h
F are the corresponding components of the penetration rate
& the net applied force of the isotropic bit to the formation.
When the rock drillability is tested in laboratory using the core samples, the weight on the
bit and the rotary speed are constant so that rock drillability anisotropy index of the
transversely isotropic formation can also be expressed as:

r
h
T
I
T
=
v
(11)

153
where T
v
and
h
T are two parameters representing the drilling time (seconds) in directions
perpendicular and parallel to bedding plane of the core samples respectively. The
standard definition of rock drillability can be expressed by the following equation (Yin,
1989):

d 2
log K T = (12)
where
d
K is the rock drillability and T the drilling time. Taking two sides of equation (11)
into logarithm to the base 2, we can obtain the following equations:

2 r 2 v 2 h dv dh d
log log log I T T K K K = = = (13)

d
r
2
K
I

= (14)
3.1.2 Rock samples
Fourteen core samples used in laboratory came from the measured depth interval of
48m1027 m of the well KZ-1 for scientific drilling in China, which were supplied by the
Engineering Center for Chinese Continental Scientific Drilling (CCSD). In the directions
perpendicular and parallel to the bedding plane, these core samples were cut into shapes of
cube or cuboid and their surfaces of both ends were polished and kept parallel to each other,
with an error of less than 0.2 mm. Then, the machined samples were put into an oven with a
temperature of 105-110C and roasted for 24 h. Finally, all of the samples can be used for the
testing of rock drillability after cooling down to room temperature.
3.1.3 Testing method
The rock drillability can be measured with a device for testing the rock drillability (shown
in Fig.2). During the measurement, some weight is applied on the micro-bit by the
function of a hydraulic pressure tank with the fixed poises, so that the weight on the
micro-bit is kept at a constant value. The measured depth to be drilled to is set with the
standard indicator, and the drilling time is logged with a stopwatch. Both the roller bit
(bit of this kind has three rotating cones and each cone will rotate on its own axis during
drilling) drillability and the PDC (the acronym of Polycrystalline Diamond Compact) bit
drillability can be tested with the above-mentioned instrument, which is of the following
standard data.
The diameter of the micro-bit is 31.75 mm.
Weight is 9020 N on the roller bit and 50020 N on the PDC bit.
The rotary speed is 551 r/min.
The total depth to be drilled to is 2.6 mm for the roller bit with a pre-drilled depth of 0.2 mm
and 4 mm for the PDC bit with a pre-drilled depth of 1.0 mm.
During testing the rock drillability, the micro-bit is often checked so that each of the worn
micro-bits should be replaced in time to ensure the testing accuracy. The testing points of
drilling time for each tested side of a rock sample should be gained as many as possible and
their average value is taken as the test value of the side. The grade value of each side
drillability of the rock sample can be calculated by equation (16) with the test data of drilling
time for each side of the rock sample.


154
1
2
3
4
5
6
7
8
9
10

Fig. 2. Testing device for rock drillability(Note: 1. Rock sample; 2. micro-bit; 3. cutting tray;
4. turbine rod; 5. lever; 6. weight; 7. meter for measuring depth; 8. bar with thread for
adjusting lever; 9. worktable; 10. compaction bar with thread)
3.1.4 Experimental result and analysis
Some testing results of rock drillability for the 14 core samples from CCSD are obtained in
laboratory and shown in Table 1 and Table 2.

No. of the
cores from
CCSD
Measured
depth, m
Rock drillability with the roller bit (K
dRB
) Rock
drillability
anisotropy
index
Perpendicular to the
bedding plane
Parallel to the
bedding plane
9 48 6.03 6.12 0.94
38 145 9.18 9.86 0.62
57 197 10.29 10.79 0.71
104 305 11.11 11.39 0.82
143 400 8.21 8.17 1.03
179 504 8.70 8.99 0.82
218 607 9.25 10.29 0.49
252 698 10.64 10.78 0.91
281 775 8.78 9.21 0.74
288 795 10.17 8.22 3.86
304 834 7.92 8.57 0.64
340 925 8.89 9.08 0.88
363 998 9.15 10.20 0.48
373 1027 8.12 10.21 0.23
Table 1. Experimental results of rock drillability with the roller bit

155
No. of the
cores from
CCSD
Measured
depth, m
Rock drillability with the PDC bit (K
dPDC
) Rock
drillability
anisotropy
index
Perpendicular to the
bedding plane
Parallel to the
bedding plane
9 48 4.55 4.34 1.16
38 145 8.57 10.65 0.24
57 197 9.89 10.06 0.89
104 305 10.78 10.90 0.92
143 400 7.82 7.40 1.34
179 504 8.48 8.47 1.01
218 607 8.61 9.14 0.69
252 698 9.52 9.86 0.79
281 775 8.22 9.03 0.57
288 795 9.55 7.31 4.72
304 834 6.06 7.71 0.32
340 925 8.14 8.95 0.57
363 998 8.18 8.64 0.73
373 1027 7.93 8.77 0.56
Table 2. Experimental results of rock drillability with the PDC bit
It is observed clearly from Table 1 and Table 2 that the rock samples from CCSD have the
anisotropic characteristics in the rock drillability. The rock drillability perpendicular to the
bedding plan is different from that parallel to the bedding plane, whether it is for the roller
bit or for the PDC bit. For the roller bit, indices of drillability anisotropy of the rock samples
are ranged from 0.23 to 0.94, except the anisotropy indices of rock samples of 143# and 288#,
which are 1.03 and 3.86 respectively. The case is similar to the PDC bit; indices of drillability
anisotropy of the rock samples are between 0.24 and 0.92, except the anisotropy indices of
rock samples of 9#, 143#, 179# and 288#, corresponding to 1.16, 1.34, 1.01 and 4.72,
respectively. Generally, the rock drillability perpendicular to the bedding plan is less than
that parallel to the bedding plane, so that the formation can be penetrated more easily in the
direction perpendicular to the bedding plane.
3.2 Acoustic anisotropy of rock sample
3.2.1 Definition
It is supposed that the formation is the transversely isotropic, and thus the acoustic
anisotropy of the formation rock can be expressed by an acoustic anisotropy index (
v
I ):

v av ah
/ I V V = (15)
where
av
V and
ah
V are the acoustic velocities in rock along the directions perpendicular and
parallel to the bedding plane of the formation respectively.
3.2.2 Testing method
With the method of making the ultrasonic pulse penetrating through a rock sample, the
acoustic velocities
av
V and
ah
V can be measured in laboratory. The ultrasonic testing


156
system used in laboratory is shown in Fig. 3, in which the ultrasonic transducers can
provide a frequency of 0.5 MHz and the butter and honey can be used as its coupling media.
The pulse generator can generate electric pulses with a strength range of 1-300 V. The width
and iteration frequency of the electric pulse can be adjusted and controlled. During testing,
the signal generator makes an electric pulse signal which will touch off the emission end of
the energy exchanger to generate ultrasonic pulses. The ultrasonic pulses (acoustic waves)
propagating through the rock sample are incepted by the reception end of the energy
exchanger. Finally, the propagation time and the signal strength of the ultrasonic pulses
(acoustic waves) through the rock sample are logged by a digital memory oscillograph.
In order to reduce the errors from the artificial operations, the emission end of the energy
exchanger is aimed at its reception end as accurately as possible during testing. Before each
test, the ultrasonic testing system should be calibrated using the aluminum rod to ensure the
accuracy of the test results. Testing for each point of a rock sample is conducted for three
times in the actual testing. The average value of the test data of three times for each point is
taken as a final test result for the point of a rock sample. With the test data, the acoustic
velocity may be calculated by the following equation:

0
l
V
t t
=
(16)
where V is the acoustic velocity; l is length of the rock sample, mm; t is propagation time of
the acoustic wave, s; and
0
t is delayed time of the testing system, s.

Fig. 3. The ultrasonic testing system
3.2.3 Experimental result and analysis
Some ultrasonic test results of the 14 core samples from CCSD are logged with the above
test method and with the ultrasonic testing system in laboratory, and the rock acoustic
velocities shown in Table 3 can be calculated by equation (16).
It can be obviously observed from Table 3 that the rock samples from CCSD are of the rock
acoustic anisotropy. The rock acoustic velocity perpendicular to the bedding plan is
different from that parallel to the bedding plane. Based on the acoustic velocity data in Table
3, the acoustic anisotropy of the rock samples can be calculated by equation (15). The
Ultrasonic generator
Rock sample
Printer
Computer

157
acoustic anisotropy indices of the rock samples are between 0.85 and 0.98, except the 363#
and 373#, which are 0.77 and 0.76 respectively. For the test of the rock samples from CCSD,
the rock acoustic velocity perpendicular to the bedding plan is less than that parallel to the
bedding plane, as shown in Table 3. The main reason for this difference is that there are
many fractures with different scales in the rock sample. When the acoustic wave penetrates
through the fractures, the fractures cause a loss of the pulse energy so as to make the
acoustic velocity reduce more quickly, on the other hand, the pulse energy is dissipated in
the process of propagation. According to some progress in geophysics (Patrick & Richard,
1984), the fractures can play a role in guiding the wave when the elastic wave has
propagated in the direction parallel to the bedding plane of the rock sample, and play a role
in obstructing the wave when the elastic wave has propagated in the direction
perpendicular to the bedding plane. Therefore, the propagation of the acoustic wave
penetrating through the rock sample is probably controlled by such a kind of geophysical
mechanism.

No. of the
core from
CCSD
Measured
depth, m
P-wave velocities of the rock samples, m/s
Acoustic
anisotropy index of
the rock sample
Perpendicular to bedding
plane (V
av
)
Parallel to bedding
plane (V
ah
)
9 48 4387 4457 0.98
38 145 4442 5120 0.87
57 197 6365 6826 0.93
104 305 5431 5714 0.95
143 400 4410 4928 0.89
179 504 4568 4744 0.96
218 607 4177 4671 0.89
252 698 5805 5990 0.97
281 775 4776 5516 0.87
288 795 5142 5453 0.94
304 834 4392 5129 0.86
340 925 4325 5096 0.85
363 998 3816 4928 0.77
373 1027 3761 4930 0.76
Table 3. Experimental results of acoustic velocities of the rock samples
3.3 Correlations between I
r
and I
v

With the experimental data in table 1 to table 3 and the corresponding calculations, it can be
found that the rock drillability anisotropy is inherently related to the acoustic anisotropy of
the rock samples. Therefore, exponential function, logarithmic function, polynomial
function, and linear function, are used to make a regression analysis of the data obtained by
experiments. With the matching & extrapolating effects of these regression functions
comprehensively considered, exponential function is finally selected as the regression model
of correlation between
r
I and
v
I . The results of regression calculations for the correlations
are listed in Table 4.
In Table 4,
rRB
I is denoted as the drillability anisotropy index of the rock sample with a
roller bit,
vp
I as the acoustic anisotropy index of P-wave through the rock sample,
dRB
K as


158
the rock drillability difference between both directions perpendicular and parallel to the
bedding plane of the rock sample with a roller bit, calculated by equation (13), and
rPDC
I as
the drillability anisotropy index of the rock sample with a PDC bit.

Regression functions
vp ( )
rRB
e
a bI
I
+
=
vp ( )
dRB
e
a bI
K
+
=
vp ( )
rPDC
e
a bI
I
+
=
a
Value 3.418 8.129 4.286
Standard error 0.840 1.893 1.491
t-ratio 4.0679 4.295 2.875
Prob(t) 0.00226 0.00157 0.02068
b
Value 3.401 -9.993 4.366
Standard error 0.914 2.361 1.661
t-ratio 3.721 4.232 2.628
Prob.(t) 0.00397 0.00174 0.03026
Correlation coefficient R 0.793 0.832 0.710
F-ratio 16.92 22.41 8.09
Prob.(F) 0.0021 0.0008 0.0217
Table 4. Results of the regression calculations
4. Evaluation method based on acoustic wave information
Many studies have been made to evaluate the rock drillability anisotropy with the core
testing method (Gao & Pan, 2006) and the inversion method (Gao et al, 1994). However, as
for the core testing method, its result may not reflect the actual rock drillability anisotropy
since the experimental conditions are different from the downhole conditions. Moreover, the
profile of rock drillability anisotropy along the hole depth can not be established because of
the limitation of the core samples. The inversion method needs to work with a bottom hole
assembly (BHA) analysis program and some parameters in the inversion model are not easy
to obtain so that its applications are limited to some extent. Thus, the evaluation method will
be presented in this section so as to predict rock drillability anisotropy of the formation by
using the acoustic wave information (Gao et al, 2008).
4.1 Acoustic wave velocity of the formation
The formation studied here is the transversely isotropic formation which is frequently
encountered in drilling for oil & gas. Experimental investigation shows that layered rock has
the transversely isotropic characteristics.
4.1.1 Phase velocity in the transversely isotropic formation
For the transversely isotropic media, Hookes law can be written as

159

11 11 66 13
11 66 11 13
13 13 33
44
44
66
2 0 0 0
2 0 0 0
0 0 0
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
xx xx
yy yy
zz zz
yz yz
xz xz
xy xy
C C C C
C C C C
C C C
C
C
C

| | | || |
| | |
| | |
| | |
| | |
= | | |
| | |
| | |
| | |
| | |
| | |
\ . \ .\ .
(17)
Elastodynamic equation of the elastic media can be obtained from the textbook and
expressed as

2
2
2
2
2
2
0
0
0
xx yx zx
yy zy xy
zz xz yz
u
X
x y z t
v
Y
y z x t
w
Z
z x y t

+ + + =

+ + + =
+ + + =
(18)
where X , Y , and Z are respectively the body force in directions of x, y and z (Xu, 2011). u,
v and w are the corresponding displacements. is the density of the elastic media,
3
g /cm .
Substituting equation (17) into equation (18) and solving with geometric equations without
considering body force, we can get the following wave equation :

2 2 2 2 2 2
11 66 44 11 66 13 44
2 2 2 2
2 2 2 2 2 2
66 22 44 11 66 13 44
2 2 2 2
2 2 2 2 2
44 44 33 13 44 1
2 2 2 2
( ) ( )
( ) ( )
( ) (
u u u u v w
C C C C C C C
t x y z x y x z
v v v v u w
C C C C C C C
t x y z x y y z
w w w w u
C C C C C C
t x y z x z

= + + + + +

= + + + + +

= + + + + +

2
3 44
)
v
C
y z
(19)
Because of the symmetry of the stress and strain in the direction normal to z direction, the
wave equation can be simplified to two dimensions without any loss of generality. In the
plane of y=0(that is the xz plane), the wave equation (19) can be written as

2 2 2 2
11 44 13 44
2 2 2
2 2 2
66 44
2 2 2
2 2 2 2
44 33 13 44
2 2 2
( )
( )
u u u w
C C C C
t x z x z
v v v
C C
t x z
w w w u
C C C C
t x z x z

= + + +

= +

= + + +

(20)
The solutions of equation (20) are


160

2 2
33 44 11 33
1
( ) [ ( )sin ( )]
2
Pa
v C C C C D = + + + (21)

2 2
33 44 11 33
1
( ) [ ( )sin ( )]
2
SVa
v C C C C D = + + (22)

2 2 2
66 44
( ) sin cos
SHa
v C C = + (23)
Where
2 2 2
33 44 13 44 33 44 11 33 44
2 2 4 1/2
11 33 44 13 44
( ) {( ) 2[2( ) ( )( 2 )]sin
[( 2 ) 4( ) ]sin }
D C C C C C C C C C
C C C C C

= + + + +
+ + +
,
where
Pa
v is phase velocity of the P-wave;
SVa
v is phase velocity of P-SV wave;
SHa
v is
phase velocity of SH-wave; is phase angle which is the angle between the wave front
normal and the unique (vertical) axis as shown in Fig.4.

O
A
g
0
1
a
v
g
v
ray
(source)
wavefront

Fig. 4. Phase angle and group angle
It is defined that

11 33
33
2
C C
C

= (24)

66 44
44
2
C C
C

= (25)

* 2
13 44 33 44 11 33 44
2
33
1
[2( ) ( )( 2 )]
2
C C C C C C C
C
= + + (26)
and

0 33
/ C =
0 44
/ C = (27)

161
where
0
is the vertical P-wave velocity;
0
is the vertical SV-wave velocity; is rock
density. , and
*
are rock anisotropy parameters of the formation.
Substituting equation (24), (25), (26) and (27) into equation (21), (22) and (23), we can get

2 2 2 *
0
( ) [1 sin ( )]
Pa
v D = + + , (28)

2 2
0 0
2 2 2 *
0
2 2
0 0
( ) [1 sin ( )]
SVa
v D

= + (29)

2 2 2
0
( ) [1 2 sin ]
SHa
v = + (30)
where

2 * 2 2
0 0 0 * 2 2 4 1/2
2 2 2 2 2 2 2
0 0 0 0 0
1 4 4(1 / )
( ) (1 ){[1 sin cos sin ] 1}
2 (1 / ) (1 / )
D

+
= + +

(31)
Letting
2
= and substituting it into equations (28), (29), (30) and (31), we can get

( )
( )
0.5
,90 0
,90 0
0.5
,90 0
1 2
1 2
Pa
SVa
SHa
v
v
v

= +
= +
(32)
For the case of weak rock anisotropy (i.e. the quantity of , , and
*
is small),
expanding equation (31) in the Taylor series at fixed and neglecting the second order of
small quantity can be simplified as

*
* 2 2 4
2 2
0 0
sin cos sin
(1 / )
D

+
(33)
Using equations (33) in equation (28) and (29), expanding
Pa
v ,
SVa
v and
SHa
v in a Taylor
series at the fixed and neglecting the second order of small quantity can be expressed as

2 2 4
0
( ) (1 sin cos sin )
Pa
v = + + (34)

2
0
2 2
0
2
0
( ) [1 ( )sin cos ]
SVa
v

= + (35)

2
0
( ) (1 sin )
SHa
v = + (36)
Where
*
2 2
0 0
1
2 (1 / )

(
= +
(

.


162
4.1.2 Phase velocity and group velocity
The phase velocity is the velocity in the direction of the phase propagation vector, normal to
the surface of constant phase, which is also called the wave front velocity since it is the
propagation velocity of the wave front along the phase vector. The phase angle is formed
between the direction of phase vector and the vertical axis. In contrast, the ray vector points
always from the source to the considered point on the wave front. The energy propagates
along the ray vector with the group velocity, while the group angle is formed between the
propagation direction and the vertical axis. The difference between the phase angle and the
ray angle is illustrated in Fig.4.
The relationship between the phase angle and the group angle can be expressed by the
following equation:
tan( )
a
g
a
dv
d
v
= (37)
Where
g
is the group angle;
a
v is the phase velocity; is the phase angle. Expanding
equation (37) leads to

1 tan
tan( ( )) (tan ) /(1 )
a a
g
a a
dv dv
v d v d

= + (38)
The group velocity (
g
v ) is related to the phase velocity (
a
v ) as shown by the following
formula
( )
2
2 2
( ) ( )
a
g g a
dv
v v
d

| |
= +
|
\ .
(39)
Where
g
v is the group velocity.
The following section makes the solution for the relationship between the group velocity
and the phase velocity, the group angle and the phase angle of the P-wave at any angle.
Another rock anisotropy parameter is introduced and expressed as

* 2 2
13 44 33 44
2 2
0 0 33 33 44
1 ( ) ( )
2 (1 / ) 2 ( )
C C C C
C C C

(
+
= + =
(

(40)
Letting
2 2
0 0
1 / t = , equation (38) can be rescaled as

2
2
2
0
( )
1 sin ( )
Pa
v
D
= + + (41)
Where
2 4 2 2
1 1
( ) 4( 2 2 )sin 4 (2 )sin
2 2
D t t t t t = + + + .
Making derivation to both sides of equation (41) , we can get

163

2
0 2 2
( ) sin cos
2( 2 2 )sin 2 ( )
( ) ( )
Pa
Pa
dv
t t t t R
d v R

( = + + +

(42)
Where
( )
1
2 4 2 2
2
( ) 4( 2 2 )sin 4 (2 )sin 2 ( ) R t t t t D t = + + + = + .
Substituting equations (41) and (42) into equation (38), we can get

{ }
2
3 2 1 4 3
2
3 2 4
2 ( ) ( ) 2 sin ( ) 2 ( ) tan
tan
2 ( ) ( ) sin ( )
g
M M M M M
M M M

(

=
(

(43)
Where
2
1
2 2 M t t = + ;
2
( ) 4 ( ) 2 M t R t = + ;
3
( ) 2 ( ) M t R t = + ;
2
4
( ) ( ) 2 ( ) M t tR R = + .
Substituting equation (41) and (42) into equation (39), we can get the following equation:
( )
{ }
1
2
2
4 2 4 2 2 2
0 1 3
1
( ) ( ) ( ) sin cos 2 sin ( )
( ) ( )
Pg g Pa
Pa
v v R M M
v R

( = + +

(44)
Where
Pg
v is the group velocity of the P-wave;
Pa
v is the phase velocity of the P-wave.
In the case of 0 =

and
90 =

,
( )
0
Pa
dv
d

=
, the phase velocity is equal to the group
velocity.Thus, when the group angle of the P-wave at the considered point of the formation
is given, its phase angle can be calculated by using equation (43). Phase velocity and group
velocity of the point can also be made out by using equation (41) and (44).
4.1.3 Methodology for determining rock anisotropy parameters
From the above discussion, we know that if rock anisotropy parameters , and , are
known, the wave velocity at any direction can be calculated by using acoustic wave velocity
perpendicular to the bedding of the formation. In other words, the acoustic wave velocity
perpendicular to the bedding of the formation can be make out if the wave velocity at any
direction and rock anisotropy parameters, , and , are known.
It is assumed that the formation to be drilled is transversely isotropic with symmetry axis
perpendicular to the bedding of the formation and the formation properties do not change
significantly from one well to another. Acoustic wave logging provides a way to measure
the velocity of P-wave or S-wave in the formation (or slowness time). The schematic figure
for measuring the velocity of P-wave or S-wave is illustrated in Fig.5.


164
In the figure 5, S
1
and S
2
are monopole sonic transducers. R
1
, R
2
, R
3
and R
4
are sonic
receivers. When the sonic is transmitted from S
1
, time difference between R
2
and R
4
is
recorded. In the same way, when the sonic is transmitted from S
2
, time difference between
R
1
and R
3
is recorded. The average of time difference between R
2
and R
4
and time difference
between R
1
and R
3
is the velocity in the formation measured.

Fig. 5. The principle of acoustic wave logging
Since available S-wave velocity is limited in logging data, we restrict ourselves to take
consideration of the P-wave only. The frequency of the acoustic wave logging is about
20kHz~25kHz which has long wave length. Since a monopole sonic transducer has a
mini-bulk, a monopole borehole sonic tool may be approximated by a point source in line
with an array of point receivers. The group velocity surface is the response from a point
source and so the monopole sonic tool response is approximated as a point source coupled
with a series of point receivers in an infinite media (neglecting borehole effects). Therefore,
we measure group velocity with borehole sonic tools.
Three parameters, the vertical P-wave velocity (
0
) and the anisotropy parameters and
can be recovered using borehole sonic measurement at different angles relative to the axis of
symmetry by following objective function:

2
1
1
( ) ( )
n
P Pmi Pci
i
v v v
n

=
= (

(45)
where ( )
Pmi
v is the measured P-wave velocity; ( )
Pci
v is the P-wave velocity calculated by
equation (32); n is the total number of the measured signals. The goal of the inversion is to
find the optimization value of C
11
, C
13
, C
33
and C
44
, to minimize value of
P
v , by which
0
,
and can be calculated, as shown in Fig.6 (Gao et al, 2008).

165

Fig. 6. Flow chart for the inversion calculations of rock anisotropy parameters


166
4.2 Prediction model of rock drillability anisotropy
Based on the previous section 3, a calculation model has been established to predict rock
drillability anisotropy of the formation:

1 dv 2
r
2
C K C
I
+
= (46)

dv 3 4
ln( ) K C C t = + (47)
Where K
dv
is the rock drillability perpendicular to the bedding plane of the formation; t is
the time interval of acoustic wave in the same direction, us/m;
j
C (j=1,2,3,4) are the
regression coefficients based on the experimental data and the survey data in drilling
engineering. For example, by the regression analysis based on some oilfield data in west
China, we can get such coefficients as C
1
=0.05246, C
2
=-0.76732, C
3
=32.977, C
4
=-4.950.
4.3 Evaluation method of rock drillability anisotropy based on acoustic wave
From equations (46) and (47), it is shown that the key point for the evaluation of rock
drillability anisotropy is how to obtain the rock drillability perpendicular to the bedding
plane of the formation which depends on the time interval of acoustic wave in the same
direction. Thus, the evaluation of rock drillability anisotropy comes down to determine the
time interval of acoustic wave perpendicular to the bedding plane of the formation.
Provided that the formation is of the transversely isotropy and has the symmetry axis
perpendicular to the bedding plane of the formation, the angle between hole axis and the
formation normal can be calculated by the following formula which is derived from
transformation of the formation coordinates to the bottom hole coordinate.
arccos cos cos sin sin cos( )
f
= (

(48)
where is the angle between hole axis and normal of the formation; is hole inclination,
degree or radian; is azimuth, degree or radian; is stratigraphic dip, degree or radian;
f
is azimuth of the formation tendency, degree or radian.
When rock anisotropy parameters of a hole section is known, its acoustic wave velocity
perpendicular to the bedding plane of the formation can be calculated by the following
procedures:
1. Calculating group angle according to stratigraphic dip angle & up dip direction, and
inclination & azimuth of hole.
2. Making an initial guess for the acoustic wave velocity
,0 P
v .
3. Reading shear wave velocity from shear wave logging or calculating it by equation (46).
4. Calculating phase angle by equation (43).
5. Calculating phase velocity and group velocity of the P-wave by equation (41) and
equation (44), respectively.
6. Comparing the P-wave group velocity with the measured velocity.
7. If group velocity of the P-wave matches the measured velocity,
,0 P
v is what we find.
Otherwise, we should repeat step 2 to step 7 until they are matched.
The flow chart for inversion of the acoustic wave velocity perpendicular to the bedding
plane of the formation is shown in figure 7.
The rock drillability can be calculated by equation (47) after obtaining the time interval of
the acoustic wave perpendicular to the bedding plane of the formation. Thus, the profile of
rock drillability anisotropy index can be established by using equation (46).

167

Fig. 7. Inversion method of the acoustic wave velocity perpendicular to the bedding plane
5. Case study
Based on some well logging data and drilling information from Qinghai oilfield in west
China, the case study is presented in this section to verify the evaluation method for the
anisotropic drilling characteristics of the formation to a certain extent.
Based on these data in table 5, rock drillability anisotropy of the fromation and its
anisotropic drilling characteristics can be calculated by using the evaluation method
described above. The inversion result of shale anisotropy parameters is shown in table 6.

Well
Number
Well logging information
Well 5
gamma-ray, compensated acoustic wave and compensated density,
inclinometer data and geologic stratification data
Well 6
inclinometer data and geologic stratification data
Well 7
inclinometer data and geologic stratification data, and some other records
Table 5. Well drilling & logging information from some completed wells at the Honggouzi
conformation in Qinghai oilfield


168
v
P,0
(m/s)
root-mean-
square (m/s)
4029.1 1.6833 1.6098 98.6153
Table 6. The inversion result of shale anisotropy parameters for the Honggouzi
conformation
From the data in the table 6, we can see that the shale is of strong rock anisotropy. The
acoustic wave front of the shale section is shown in figure 8.

Fig. 8. Acoustic wave front of the shale section of the Honggouzi conformation
Based on the wave front of the shale shown in Fig.8, the rock anisotropy parameters of the
formation at any measured depth, and , can be approximately calculated by the
following equations:

cur cur
sh
min max min
(2 1) (2 1)
( ) ( )
sh c
sh c
GR G G
V
V
V
GR GR GR GR GR

=
=

`
=

=
)
(49)
where and are rock anisotropy parameters of the formation at any measured depth;
c

and
c
are rock anisotropy parameters of the shale section;
sh
V is the shale content, %;
GR is gamma ray value;
max
GR is the maximum value of gamma ray;
min
GR is the minimum
value of gamma ray;
cur
G is the Hilchie index whose value is 3.7 for the Neogene
Stratigraphy and 2 for old strata.
After obtaining the rock anisotropy parameters( and ), we can calculate the acoustic
wave perpendicular to the bedding plane of the formation by using the inversion method

169
shown in Fig.7 and the well logging data of acoustic wave. Thus, the rock drillability
anisotropy of the formation can be calculated by equations (46) & (47), and the
corresponding anisotropic drilling characteristics can be evaluated by equations (8) & (9), as
shown in figure 9.

Fig. 9. The evaluation results of anisotropic drilling characteristics of the formation at the
Honggouzi conformation in Qinghai oilfield
6. Conclusion
The orthotropic formation and the transversely isotropic formation are the typical
formations encountered frequently in drilling engineering. Based on rock-bit interaction
model, the two parameter equations have been derived for us to calculate the anisotropic
drilling characteristics of them as soon as rock drillability anisotropy of the formations is
evaluated quantificationally by using the oilfield data.
The correlation between rock drillability anisotropy and acoustic wave anisotropy of the
formation can be matched to each other by an exponential function which is of the best
extrapolative performance and relativity. Coefficients in the model are various for different
formations to be drilled.
To a certain extent, the research results presented here have shown a new way for us to
evaluate conveniently rock drillability anisotropy of the formation by using well logging or
seismic data. Case study shows that this evaluation method is better for applications of rock
drillability anisotropy of the formation in drilling engineering.


170
7. Acknowledgment
The authors are grateful for the financial support from the national research project (Grant
No. 2010CB226703) and the supply of many core samples by CCSD.
8. References
Gao, D.; Dai, D. & Pan, Q. (2008). Evaluation of deflecting characteristics of anisotropy
formation by using well-log information. ACTA PETROLEI SINICA, Vol.29, No.6,
(December 2008), pp. 927-932, ISSN 0253-2697
Gao, D. & Pan, Q.(2006). Experimental study of rock drill-ability anisotropy by acoustic
velocity. Petroleum Science, Vol. 3, No.1, ( March 2006), pp.50-55, ISSN 1672-5107
Gao, D. (1995). Predicting and scanning of wellbore trajectory in horizontal well using
advanced model. Proceedings of the Fifth International Conference on Petroleum
Engineering Held in Beijing, China, SPE 29982, pp.297-308, 14-17 Nov. 1995
Gao, D.; Liu, X. & Xu B.(Dec. 1994). Prediction and Control of Well Trajectory, Petroleum
University Press, ISBN 7-5636-0584-3/TE95, Dongying, China
Gao, D. & Liu, X.(1990). Anisotropic drilling characteristics of the typical formations. Journal
of the University of petroleum, China, Vol.14, No.5, (Oct. 1990), pp.1-8, ISSN 1000-7393
Gao, D. & Liu, X.(1989). A new model of rock-bit interaction. Oil Drilling and Production
Technology, Vol.11, No.5, (Oct. 1989), pp.23-28, ISSN 1000-7393
Lubinski, A. & Woods,H.(1953). Factors affecting the angle of inclination and dog-legging in
rotary bore holes. DPP, 1953: 222-242
Patrick, J. & Richard L.(1984). An experimental test of P-wave anisotropy in stratified media.
Geophysics, Vol.49,No.4, (1984),pp. 374-378
Xu, B.(Jan. 2011). Concise Elasticity and Plasticity, Higher Education Press, ISBN 978-7-04-
030725-2, Beijing, China
Yin H.(1989). Study of formation anisotropy-rock drillability. Oil Drilling and Production
Technology, Vol.11, No.1, (Feb. 1989),pp.15-22, ISSN 1000-7393
8
Machinery Faults Detection
Using Acoustic Emission Signal
Dong Sik Gu and Byeong Keun Choi
Gyeongsang National University
Republic of Korea
1. Introduction
Application of the high-frequency acoustic emission (AE) technique in condition monitoring
of rotating machinery has been growing over recent years. This is particularly true for
bearing defect diagnosis and seal rubbing (Mba et al., 1999, 2003, 2005; Kim et al., 2007;
Siores & Negro, 1997). The main drawback with the application of the AE technique is the
attenuation of the signal and as such the AE sensor has to be close to its source. However, it
is often practical to place the AE sensor on the non-rotating member of the machine, such as
the bearing or gear casing. Therefore, the AE signal originating from the defective
component will suffer severe attenuation before reaching the sensor. Typical frequencies
associated with AE activity range from 20 kHz to 1 MHz.
While vibration analysis on gear fault diagnosis is well established, the application of AE to
this field is still in its infancy. In addition, there are limited publications on application of
AE to gear fault diagnosis. Siores explored several AE analysis techniques in an attempt to
correlate all possible failure modes of a gearbox during its useful life. Failures such as
excessive backlash, shaft misalignment, tooth breakage, scuffing, and a worn tooth were
seeded during tests. Siores correlated the various seeded failure modes of the gearbox with
the AE amplitude, root mean square, standard deviation and duration. It was concluded
that the AE results could be correlated to various defect conditions (Siores et al., 1997).
Sentoku correlated tooth surface damage such as pitting to AE activity. An AE sensor was
mounted on the gear wheel and the AE signature was transmitted from the sensor to data
acquisition card across a mercury slip ring. It was concluded that AE amplitude and energy
increased with increased pitting (Sentoku, 1998). In a separated study, Singh studied the
feasibility of AE for gear fault diagnosis. In one test, a simulated pit was introduced on the
pitch line of a gear tooth using an electrical discharge machining (EDM) process. An AE
sensor and an accelerometer for comparative purposes were employed in both test cases. It
was important to note that both the accelerometer and AE sensor were placed on the
gearbox casing, it was observed that the AE amplitude increased with increased rotational
speed and increased AE activity was observed with increased pitting. In a second test,
periodically occurring peaks were observed when natural pitting started to appear after half
an hour of operation. These AE activities increased as the pitting spread over more teeth.
Singh concluded that AE could provide earlier detection over vibration monitoring for
pitting of gears, but noted it could not be applicable to extremely high speeds or for


172
unloaded gear conditions. (Singh et al., 1996) Tan offered that AE RMS (Root Mean Square)
levels from the pinion were linearly correlated to pitting rates; AE showed better sensitivity
than vibration at higher toque level (220Nm) due to fatigue gear testing using spur gears.
He made sure that the linear relationship between AE, gearbox running time and pit
progression implied that the AE technique offers good potential in prognostic capabilities
for monitoring the health of rotating machines. (Tan et al., 2005, 2007)
On the other hand, the signal processing method for AE signal was studied using bearing
and gearbox. In the results of the research (Sheen, 2008, 2010; Yang et al., 2007), the envelope
analysis was found to be useful to detect fault in rolling element bearing. The fault detection
frequency of bearing can be presented in the power spectrum. Wavelet transform was used
for the signal processing method for the gearboxes (Wu et al., 2006, 2009), but wavelet
transforms can give the different results with the envelope analysis. It can be shown the
defect frequency, but the efficiency is lower than that of envelope analysis. Thus, the signal
processing method for AE signal has not been completed until now, and it must be
developed in the future.
Therefore, in this paper, a signal processing method for AE signal by envelope analysis with
discrete wavelet transforms is proposed. For the detection of faults generated by gear
systems and a cracked rotor using the suggested signal processing, these were installed in
each test rig system. In gearbox, misalignment was created by a twisted case caused by arc-
welding to fix the base and bearing inner race fault was generated by severe misalignment.
Through the 15 days test using AE sensor, misalignment was observed and bearing faults
were also detected in the early fault stage. To identify the sensing ability of the AE, vibration
signal was acquired through an accelerometer and compared with the AE signal. Also, to
find the advantage of the proposed signal processing method, it was compared to
traditional envelope analysis. The detection results of the test were shown by the power
spectrum and comparison of the harmonics level of the rotating speed. Modal test and
zooming by a microscope were performed to prove the reason of the other faults. And the
crack was seeded by wire cutting with 0.5 mm depth. The shaft was coupled with motor
and non-drive-end was left 6.5 mm by lifting tool. During rotating the shaft, AE signals
were acquired by AE sensor with 5MHz sampling frequency and 0.5 seconds storing time.
The AE signals were transformed by FFT to create the power spectrums, and in the
spectrums several peaks were occurred by the crack growth. Along the growth of the
crack, the characteristic of the power spectrum was changed and displayed different
frequencies.
2. Signal processing method
Envelope analysis typically refers to the following sequence of procedure: (1) band-pass
filtering (BPF), (2) wave rectification, (3) Hilbert transform or low-pass filtering (LPF) and
(4) power spectrum. The purpose of the band-pass filtering is to reject the low-frequency
high-amplitude signals associated with the ith mechanical vibration components and to
eliminate random noise outside the pass-band. Theoretically, in HFRT (High Frequency
Resonance Technique) analysis, the best band-pass range includes the resonance of the
bearing components. This frequency can be found through impact tests or theoretical
calculations involving the dimensions and material properties of the bearing. However, it is
very difficult to predict or specify which resonant modes of neighboring structures will be

Machinery Faults Detection Using Acoustic Emission Signal

173
excited. It will be costly and unrealistic in practice to find the resonant modes through
experiments on rotating machinery that may also alter under the different operational
conditions. In addition, it is also difficult to estimate how these resonant modes are affected
in the assembly of a complete bearing and mounting in a specific housing, even if the
resonant frequencies of individual bearing elements can be tested or calculated theoretically
(Misiti et al., 2009). Therefore, most researchers decide on the band-pass range as on option.
To recover the disadvantage of this option, wavelet analysis is included in the process of
traditional envelope analysis in this paper.
Wavelet theory (Burrus et al., 1997) is introduced that is a tool for the analysis of transient,
non-stationary, or time-varying phenomena. Wavelet analysis is also called wavelet
transform. There are two kinds of wavelet transform: continuous wavelet transform (CWT)
and discrete wavelet transform (DWT). CWT is defined as the sum over all time of the signal
multiplied by scaled, shifted versions of the wavelet function. To use CWT, one signal can
be decomposed into a series of small waves belonging to a wavelet family. The wavelet
family is composed of scaling functions, (t) deduced by father wavelet and wavelet
functions, (t) deduced by mother wavelet. The scaling function can be represented by the
following mathematical expression:
( )
/2
,
( ) 2 2
j j
j k
k t k =
where j represents the scale coefficient and k represents shift coefficient. Scaling a wavelet
simply means stretching (or compressing) it. Shifting a wavelet simply means delaying (or
hastening) its onset. Mathematically, delaying a function f(t) by k is represented by f(t + k).
Similarly, the associated wavelet function can be generated using the same coefficients as
the scaling function.
( ) ( )
/2
,
2 2
j j
j k
t t k =
The scaling functions are orthogonal to each other as well as the wavelet functions as shown
in the following equations:

( ) ( )
( ) ( )
2 2 0 for all
0
t k t l dt k l
t t dt

=
=

Using an iterative method, the scaling function and associated wavelet function can be
computed if the coefficients j and k are known.
For many signals, the low-frequency content is the most important part. It is what gives the
signal its identity. The high-frequency content, on the other hand, imparts flavour or nuance
that is often useful for singular signal detection. In wavelet analysis, we often speak of
approximations and details. The approximations are the high-scale, low frequency
components of the signal. The details are the low-scale, high-frequency components. A
signal can be decomposed into approximate coefficients a
j,k
, through the inner product of the
original signal at scale j and the scaling function.
( ) ( )
, , j k j j k
f t t dt


174
Similarly the detail coefficients d
j,k
can be obtained through the inner product of the signal
and the complex conjugate of the wavelet function.
( ) ( )
, , j k j j k
d f t t dt

However, CWT takes a long time due to calculating the wavelet coefficient at all scales and
it produces a lot of data. To overcome such a disadvantage, we can choose scales and
positions based on powers of two the so-called dyadic scales and positions then wavelet
analysis will be much more efficient and just as accurate. Such an analysis is obtained from
the discrete wavelet transform (DWT). The approximate coefficients and detail coefficients
decomposed from a discredited signal can be expressed as
( )
( )
( )
( ) [ ]
( )
( )
( )
( ) [ ]
, , , 1 , 1 ,
0
, , , 1 , 1 ,
0
N
j k j k j k j k j k
k k
N
j k j k j k j k j k
k k
t t dt g k
t t dt h k

+ +
=
+ +
=
= =
= =

The decomposition coefficients can therefore be determined through convolution and
implemented by using a filter. The filter, g[k], is a low-pass filter and h[k] is a high-pass filter.
The decomposition process can be iterated, with successive approximations being
decomposed in turn, so that one signal is broken down into many lower resolution
components. This is called the wavelet decomposition tree as shown in Fig. 1.

Fig. 1. Wavelet decomposition tree
DWT has a de-noise function and a filter effect focused on impact signal. To make up the
weak point of BPF of the envelope analysis, DWT was intercalated on typical envelope
analysis, between BPF and wave rectification exactly. The signal by DWT will be
separated to different band widths by decomposition level and adapted to the signal with
impact.
For more complicated signals which are expressible as a sum of many sinusoids, a filter can
be constructed which shifts each sinusoidal component by a quarter cycle. This is called a
Hilbert transform filter. Let H
t
{x} denotes the output at time t of the Hilbert-transform filter
applied to the signal x. Ideally, this filter has magnitude 1 at all frequencies and introduces a
phase shift of -/2 at each positive frequency and +/2 at each negative frequency. When a


175
real signal x(t) and its Hilbert transform y(t) = H
t
{t} are used to form a new complex signal
z(t)= x(t)+jy(t), the signal z(t) is the (complex) analytic signal corresponding to the real signal
x(t). In other words, for any real signal x(t), the corresponding analytic signal z(t) =
x(t)+jH
t
{x} has the property that all negative frequencies of x(t) have been filtered out
(Douglas & Pillay, 2005). Hence, the coefficients of complex term in the corresponding
analytic signal were used for FFT.
Fig. 2 shows an analytic signal of the Hilbert transform for envelope analysis. The solid line
is a time signal and the dash is its envelope curve. A high frequency signal modified by
wavelet transform is modulated to a low frequency signal with no loss of the fault
information due to envelope effect. According to that, the fault signals in the low frequency
region can be detected using the analytic signal. That is an important fact for the proposed
signal processing method. Therefore, the proposed signal processing method in this paper is
an envelope analysis with DWT and using the coefficients of the complex term in Hilbert
transform.

Fig. 2. Analytic signal (dash) of the envelope effect
Furthermore, to reduce the noise level in the power spectrum, the spectrum values were
presented as the mean value of each day. Fig. 3 shows the power spectrums of the two
different signal processing method. Fig. 3(a) is from envelope analysis, and Fig. 3(b) shows
the envelope analysis intercalated DWT using Daubechies mother function between BPF
and wave rectification. In Fig. 3, the DWT has an effect the amplifying sidebands peaks,
especially about gear mesh frequencies, so the peaks of the harmonics of the rotating speed
(f
r
) and gear mesh frequencies (f
m
) are bigger than another, and we can check up them easily.
Therefore, in the following result, the power spectrum through envelope analysis with DWT
will be shown.


176

Fig. 3. The comparison of Power spectrums in Envelope analysis with/without DWT
3. Gearbox
3.1 Test-rig
The test-rig employed for this investigation consists of one identical oil-bath lubricated
gearbox, 3 HP-motor, rigid coupling, tapper-roller bearing, pinion, gear, control panel and
break system, as seen Fig. 4. The pinion was made from steel with heat treatment, the
number of teeth is 70, and diameter is 140 mm. The gear was made from steel, but it was
produced without any heat treatment process during manufacturing. The number of gear
teeth is 50 and diameter is 100 mm, and module is 2 mm for the gear and pinion,
respectively.

Fig. 4. Test-rig
(a) Envelope analysis
(b) Envelope analysis with DWT


177
A simple mechanism that permitted a break of disk-pad type to be rotated relative to each
other was employed to apply torque to the gear. Contact ratio (Pinion/Gear) of the gears
was 1.4. The motor used to drive the gearbox was a 3-phase induction motor with a
maximum running speed of 1800 rpm respectively and was operated for 15 days with
1500rpm. The torque on the output shaft was 1.2 kNm while the motor was in operation,
and other specifications of the gearbox are given as in Table 1.

Gear Pinion
No. of teeth 50 70
Speed of
shaft
25.01 rev/s
Meshing
frequency
1250 Hz 1750 Hz
Bearing (NSK HR 32206J)
No. of
rolling
element
17 Type
Defect Freq.
(f
d
)
Fault Freq.
(f
d
X f
r
)
Diameter of
outer race
62 mm
BPFO
BPFI
FTF
BSF
8.76 Hz
11.24 Hz
3.84 Hz
0.44 Hz
219.3 Hz
281.38 Hz
96.13 Hz
11.01 Hz
Diameter of
inner race
30 mm
BPFO : ball pass frequency of outer race
BPFI : ball pass frequency of inner race
FTF : fundamental train frequency
BSF : ball spin frequency
Table 1. Specification of gearbox and bearing
3.2 Acquisition system and test procedures
AE sensors used in this paper are a broadband type with a relative flat response in the range
frequency from 10 kHz to 1 MHz. They are placed on the right side of the gearbox cases near
the coupling in the horizontal direction at the same height with the shaft center (Fig. 4).
AE signals are pre-amplified by 60 dB and the output from the amplifier is collected by a
commercial data acquisition card with 10 MHz sampling rate during the test. Prior to the
analog-to-digital converter (ADC), anti-aliasing filter is employed that can be controlled
DAQ software. And Table 2 is shown the detail specifications of the data acquisition system.
Before the test, attenuation test on the gearbox components was taken in order to
understand the characteristics of the test-rig. The gearbox was run for 30 minutes prior to
acquiring AE data for the unload condition. Based on the sampling rate of 10 MHz, the
available recording acquisition time was 2 sec.


178
2 Channel
AE
system on
PCI-Board
18-bit A/D conversion
10M samples/s rate
(on one channel, 5M samples/s on 2 AE channels)
AE Sensor
(Wideband
type)
Peak sensitivity [V/bar] : -62 dB
Operating frequency range : 100-1,000 kHz
Resonant freq. [V/ bar] : 650 kHz
Directionality : 1.5 dB
Preamplifier
Gain
Wide dynamic range < 90 dB
Single power/signal BNC or optional separate power/signal BNC
20/40/60 dB selectable gain
Table 2. Specifications of data acquisition system
3.3 Experiment result and discussion
In general, the misaligned gear which almost always excites higher order f
m
harmonics is
shown as in Fig. 5. Often, only small amplitudes will be at the fundamental f
m
, but much
higher levels will be at 2 f
m
and/or 3 f
m
. The sideband spacing about f
m
might be 2 f
r
or even
3 f
r
when gear misalignment problems are involved. When significant tooth wear occurs, not
only will sidebands appear about f
m
, but also about the gear natural frequencies. In the case
of those around f
m
, the amplitude of the sidebands themselves is a better indicator for wear
than the amplitude of f
m
.

Fig. 5. Spectrum indicating misalignment of gear
As for significant gear eccentricity and/or backlash, these problems display the following
characteristics:
- Both eccentricity and backlash excite the gear natural frequencies as well as f
m
. They
also may generate a number of sidebands about both the natural and gear mesh
frequencies.
- If a gear is eccentric; it will modulate the natural frequency and gear mesh frequencies,
both of which will be sidebanded around the f
r
of the eccentric gear. An eccentric gear
can generate significant forces, stresses and vibration if it is forced to bottom out with
the meshing gears. (James & Bery, 1994)
In the results of the envelop analysis with DWT, the high harmonics of f
m
occurred by strong
wearing phenomena caused by misaligned teeth. In the power spectrum (Fig. 6), 25Hz (f
r
)
and its harmonics are generated and 11.32Hz was the ball pass frequency of inner race (BPFI
[f
d
]). In Fig. 6(c) and (d), the center dash line is shown for f
m
and 2 f
m
, and their side lines are
the sidebands with difference 25Hz (f
r
).


179

Fig. 6. Power spectrum of the second day

Fig. 7. Peak level trend among days

Fig. 8. Gear tooth weaned by misalignment
(a) Harmonics of f
r
(b) Harmonics of f
m


180

Fig. 9. Modal test result
In condition monitoring for general rotating machinery, the harmonics (2 f
r
, 3 f
r
, 4 f
r
) of f
r
are
occurred higher than f
r
when the misalignment was happened. According to the phenomena
of misalignment as shown in Fig.5, high level harmonics of f
r
were generated such as in Fig.
6(b), and 2 f
r
was always bigger than f
r
as shown in Fig. 7(a). The level of 2 f
m
from second to
thirteenth day was higher than or similar to f
r
as shown in Fig. 7(b). Thus, it is easily catching
up to the misalignment that occurred in this test rig. However, it might be that faults of this
system are not only misalignment but also resonance trouble, looseness, bearing fault, etc.
Wearing effect by misalignment pollutes the lubrication oil. In Fig. 8, it could be found by
the worn teeth and the spots near the pieces of gear teeth. The dripped pieces from the
unloading surface raised the wearing effect on the loading surface, and then the gap
between gear and pinion was increased. In addition, we could know that the impact marks
on the unloading surface (Fig. 8(b)) were generated by misalignment; the impacting force
was strong in the initial condition. In this way, the gear teeth were seriously damaged as in
Fi g. 8. In Fig. 6(c) and (d), the sidebands are created on wide-spread frequency range near f
m

and 2 f
m
. That is similar to a state excited by impact force. To confirm the natural frequencies
of the test-rig, modal test was fulfilled. The result of the modal test as in Fig. 9 show that f
m

and 2f
m
exist on the exiting frequency range. On the other hand, partial frequency bands
close to f
m
and 2 f
m
were excited by the impact force, but it is not an exact natural frequency
because the phase did not shift enough. Therefore, the peaks near f
m
and 2 f
m
were amplified
and have many sidebands of f
r
and 11.32Hz (BPFI [f
d
]). Therefore, it is considered that
excessive backlash occurred. Moreover, Fig. 10(a) shows the zooming power spectrum of
Fig. 6(a) focused on f
r
harmonics. We could clearly know that if the sidebands were caused
by BPFI [f
d
], then the inner race had some kind of fault. To find out the fault, the surface of
the bearing inner race was carried out and viewed by a microscope with 100X zoom as
shown in Fig. 11. Small spots were found on the surface, and small cracks were found out on
(a) Phase
(b) Frequency response function


181
the spots. However, this trouble was not seeded and existed from the initial condition. Thus,
it is as assumed that the problem happened in assembly and/or was caused by
misalignment.

(a) Power spectrum of the AE signal using envelope analysis with DWT(suggested)
(b) Power spectrum of the AE signal using traditional envelope analysis
(c) Power spectrum of the signal from accelerometer

Fig. 10. Sidebands BPFI in the first day power spectrum
To identify the sensing ability of the AE, vibration signal was acquired through
accelerometer and compared with the AE signal. Also, to find the advantage of the proposed
signal processing method, it was compared to traditional envelope analysis.
The power spectrum of the AE signal using traditional envelope analysis is shown in Fig.
10(b), and the power spectrum using the vibration signal by accelerometer is displayed on
Fig. 10(c). The vibration signal was treated by the same method with AE signal. The
harmonics of f
r
are generated, and 2 f
r
for detecting the misalignment is created and can be
found in all spectrums (Fig. 10) but the power spectrum of the AE signal, Fig. 10(a) and (b),
can explicitly display the defect frequencies as compared to the accelerometer signal (Fig.
10(c)). For example, in Fig. 10(c), the sidebands of BPFI are not easily found because of the
higher level of noise in the low frequency range below f
r
than in the AE signal with or
without DWT.


182

Fig. 11. Zooming of the inner race surface of the fault bearing

Frequency [Hz] Traditional Method Proposed method
13.709 1X-BPFI 0.2146 0.2459
25.034 1X 0.8727 0.9525
36.360 1X+BPFI 0.1936 0.3196
38.750 2X- BPFI 0.0952 0.1435
50.070 2X 1.0000 1.0000
61.393 2X+ BPFI 0.1277 0.2470
63.181 3X- BPFI 0.0970 0.1560
75.102 3X 0.3044 0.5465
86.427 3X+ BPFI 0.1699 0.2352
99.540 4X 0.4269 0.4616
110.866 4X+ BPFI 0.1062 0.1675
124.574 5X 0.2312 0.3253
135.899 5X+ BPFI 0.0714 0.1443
Table 3. Ratio of peaks versus the maximum peak in respective spectrum
According to the above results, we can understand that the AE signal can detect the fault
more easily than accelerometers and can be used in the condition monitoring system for
early detection fault. Moreover, as shown in Table 3 which is the ratio of peaks versus the
maximum peak in the respective spectrum, the peak levels of the harmonics of f
r
and
sidebands caused by BPFI are highly generated in the proposed signal processing method
(Fig. 10(a)) than the traditional method. This can lead good feature values to evaluate the
condition of the machinery. Therefore, the power spectrum of the proposed envelope
analysis using AE signal can be shown the clean result with harmonics and sidebands and is
a better technique for condition monitoring system.
4. Cracked rotor
4.1 Experiment system
Test rig consisted of a motor, a flexible coupling, rolling element bearings (NSK6200), three
steel bearing housings, a lifting tool and a cracked shaft. The transverse crack was seeded by


183
wire-cutting with 0.5 mm depth on the shaft made from SM45C. As shown in Fig. 12, the
crack was positioned at 5mm near to the second drive-end bearing, and the non-driven end
of the shaft was left 6.5 mm with bearing housing by the lifting tool.
AE signal was acquired by an AE sensor and transferred to amplifier, analog-filter, DAQ
board and HDD of a desktop. AE sensor is a wideband type with a relative flat response
in the range frequency range from 100 kHz to 1 MHz. AE signals were pre-amplified with
60 dB and the output from the amplifier was collected by a commercial data acquisition
card with 5 MHz sampling rate during the test. The signals were stored 0.5sec by every
30sec until the shaft was fractured, and the rotating speed of the motor was 600rpm
(10Hz).

Fig. 12. Experiment system
4.2 Test result and discussion
The operating speed was 600rpm, and the initial radial load for 160N was employed. The
radial load was a variable parameter because it was applied by keeping the lifting distance
with 6.5mm of the non-drive end of the shaft, and the test terminated on a fracture of the
shaft. Fig. 13 shows the observations of continuous feature values as mean value, RMS, peak
value and entropy estimation. In information theory, uncertainty can be measured by
entropy. The entropy of a distribution is the amount of a randomness of that distribution.
Entropy estimation is two stage processes; first a histogram is estimated and thereafter the
entropy is calculated. Here, we estimate the entropy of AE signal with using unbiased
estimated approach. Fig. 13(a), relatively high level of AE activity was noted from 18
minutes, and it was increased until 60 minutes. But in RMS, Peak and Entropy estimation,
the levels were kept to 18minutes beside a peak in around 9 minutes, since that these were
continuously decreased to 70 minutes and were increased with hunting with several minute
intervals until the fracture.
Interestingly observations of the AE waveform, sampled at 5 MHz showed changing
characteristics as a function of time. Fig. 14 shows a contour map of the peaks level of each
frequency with time. Rotating speed (9.5Hz) and 3rd harmonic of rotating speed (28.6Hz,
3X) dominated while the test as shown in Fig. 14(b). It is normally known that 3X is caused
by misalignment of the bearings created by the loading system for this test (Hatch & Bently,


184
2002). However, the harmonic component (3X) was kept the level in the wavelet level 6(Fig.
14(b)), but it was increased from 30 minutes in the wavelet level 4(Fig. 14(a)). Additionally,
1X started increasing earlier than 3X as shown in Fig. 14(a).

0 20 40 60 80
-2.4
-2.2
-2
-1.8
-1.6
Time [min]
A
m
p
l
i
t
u
d
e
0 20 40 60 80
0
0.05
0.1
0.15
0.2
Time [min]
A
m
p
l
i
t
u
d
e

(a) Mean value [V10
-4
] (b) RMS value [V]
0 20 40 60 80
0
0.1
0.2
0.3
0.4
0.5
Time [min]
A
m
p
l
i
t
u
d
e
0 20 40 60 80
-4
-3
-2
-1
0
Time [min]
A
m
p
l
i
t
u
d
e

(c) Peak value [V] (d) Entropy estimation value
Fig. 13. Shaft test results; run-to-failure
31Hz and 62Hz that were the harmonics of the fundamental train frequency (FTF or cage
noise) of the bearing were continuously occurred. Cage noise can be generated in any type
of bearing and the magnitude of it is usually not very high. Characteristics of this noise
include: (1) it occurs with pressed steel cages, machined cages and plastic cages. (2) It occurs
with grease and oil lubrication. (3) It tends to occur if a moment load is applied to the outer
ring of a bearing. (4) It tends to occur more often with greater radial clearance. In Fig. 14,
62Hz was continuously detected; 31Hz was detected with 3X. In a general bearing system,
the amplitude of the bearing fault frequency is depended on the load grade and is increased
along the growing load grade. However, 31Hz of the case noise of this test was not followed
the load scale because loading force for this test was decreased with the crack growth. So, it
was shown that 31Hz was related with the crack growth.
According to this result, we could know that the reason of 3X (28.6Hz) was the moment load
by the loading system; 31Hz was connected with the crack growth. Therefore, the peak
levels around 3X and 31Hz was excited by the two frequencies and was increased with the
crack growth.
To clear more the characteristic of the crack growth, in addition, PAC energy value was
observed. In acoustic emission technology, PAC-Energy is a 2-byte parameter derived from
the integral of the rectified voltage signal over the duration of the AE hit (or waveform),


185
hence the voltage-time units (V
2
sec). So, PAC energy value was determined by an
integral of the square sum total of the transferred time signal in each wavelet level. Fig. 15
shown the energy level of every wavelet levels along time, and its value was transferred to
logarithmic value because of too low resolution in linear scale.
Fig. 15 shows the energy trend of wavelet level 1 to 8. Many peaks were created while the
test in wavelet level 1(Fig. 15(a)), a high peak was created around 9 minutes existed in
wavelet level 1 to 4. Wavelet level 2, 4, 5 and 6 shows a similar trend after 10minutes. The
energy level was slowly increased with time until about 30 minutes, and then it was
increased fast until 35 minutes (additionally, it was considered that this increasing was
related with the growth of the 3X and 31Hz in Fig. 14(a)), after that it was decreased a little for
10 minutes. And it was hunted with every several minutes about 4 minutes until close fracture.
In this trend, we had considered of two phenomena, the high peak and the period hunting.
We could mind that the high peak was related with initial crack growth. Because it was
shown as follows,
- The increasing ratio of the energy was changed after it was happened.
- The mean value was began fast change (Fig. 13(a)).
- Generated in high frequency range (Fig. 15(a) ~ (d)).
The period hunting was clearly occurred and displayed in wavelet level 2, 4 ~ 8. It was
considered that it could indicate a state of the final stage of the fracture in the rotating shaft
because of follows,
- It had a period about 4 minutes.
- It was displayed in lower frequency range than the high peak (Fig. 15(e) ~ (h)).
- Its level was increased along near to the fracture.

(a) Wavelet level 4

(b) Wavelet level 6
Fig. 14. Peak level trend by frequency


186

0 20 40 60 80
0.8
1
1.2
1.4
x 10
-7
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]
Energy Trend Curve, Level = 1

0 20 40 60 80
0
0.5
1
1.5
2
x 10
-6
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]

(a) Level 1 (b) Level 2

0 20 40 60 80
0
0.5
1
1.5
x 10
-5
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]
0 20 40 60 80
0
0.5
1
x 10
-4
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]

(c) Level 3 (d) Level 4

0 20 40 60 80
0
1
2
3
4
x 10
-4
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]
0 20 40 60 80
0
2
4
6
8
x 10
-5
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]

(e) Level 5 (f) Level 6

0 20 40 60 80
0
2
4
6
8
x 10
-4
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]
0 20 40 60 80
0
0.5
1
1.5
2
x 10
-3
Time [min]
E
n
e
r
g
y

[
m
V
2
*
s
e
c
]

(g) Level 7 (h) Level 8

Fig. 15. PAC energy trend of each wavelet level


187

Fig. 16. PAC-Energy level of total wavelet level
To compare absolutely the energy of each wavelet level, all of PAC energy was displayed on
a 3D graph with time and wavelet levels as shown in Fig. 16. In wavelet level 5, after
approximately 30 minutes, a large transient rise in PAC energy level was observed and this
AE activity gradually observed frequently after 60 minutes until the shaft was fractured. In
addition, a peak created at 9 minutes was indicated. In here, we could consider that the
frequency range of wavelet level 5 could be shown a good relationship between the PAC
energy and the crack growth of the middle and final stage. Even so, the trend of the wavelet
level 7 and 8 was not clearly connected with the others because the frequency range of
wavelet level 7 and 8 was lower than the useful frequency range (100kHz to 1MHz) of AE
sensor for this research.
Therefore, the AE signal caused by the crack growth was generated on the whole ultra-
sound frequency range; the initial crack could be detected using the PAC energy on wavelet
level 1 to 4. In addition, it could be presented on wavelet level 5 until the fracture of the
shaft. In the frequency domain, it was shown that the harmonic components of the rotating
speed and bearing cage frequency were excited by the crack growth, especially on the 3X
(28.6Hz) and 31Hz.
5. Conclusion
In this paper, a signal processing method for AE signal by envelope analysis with discrete
wavelet transforms is proposed. For the detection of faults generated from a gear system


188
using the suggested signal processing, a gearbox was installed in the test rig system.
Misalignment was created by twisted case caused by arc-welding to fix the base and bearing
inner race fault is generated by severe misalignment. To identify the sensing ability of the
AE, vibration signal was acquired through accelerometer and compared to the AE signal.
Also, to find the advantage of the proposed signal processing method, it was compared with
traditional envelope analysis.
According to the experiment result, AE sensor can detect the fault earlier than an
accelerometer because of high sensitivity and in the power spectrum, the harmonics of the
rotating speed and the gear mesh frequency clearly occurred. Misalignment was observed
and bearing faults were also detected in the early fault stage. The proposed envelope
analysis is worked to evaluate the faults and indicated the faults frequencies, rotating speed,
sideband of BPFI, gear mesh frequency and harmonics, explicitly.
For the detection of the crack growth on the shaft, a cracked shaft was installed on the test
rig, and the crack was seeded by wire-cutting with 0.5 mm depth. The cracked shaft was
lifted 6.5 mm by the lifting tool. The AE signals were transformed by FFT to create the
power spectrums, and in the spectrums several peaks were occurred by the crack growth.
Along the growth of the crack, the characteristic of the power spectrum was changed and
displayed different frequencies.
In the power spectrum, it was shown that the harmonic components of the rotating speed
and bearing cage frequency were excited by the crack growth as shown in the Fig. 6,
especially on the 3X (28.6Hz) and 31Hz. And the AE signal caused by the crack growth is
generated on the whole ultrasonic frequency range; the initial crack could be detected using
the PAC-Energy on wavelet level 1 to 4, and after that, it could be presented on wavelet
level 5 until the fracture of the shaft. Therefore, in this paper, it could be shown that the
crack growth in rotating machinery is able to be considered and to be detected; in addition,
PAE-Energy can be used to detect the early detection of the crack.
Therefore, the proposed signal processing method that is the envelope analysis intercalated
DWT using Daubechies mother function between BPF and wave rectification can be shown
to provide better result than traditional envelope analysis.
6. Acknowledgment
This work has been supported by Basic Science Research Program through the National
Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and
Technology (2011-0013652) and the 2nd Phase of Brain Korea 21.
7. References
Burrus, C.S., Gopinath, R.A. & Guo, H. (1997). Introduction to wavelet and wavelet transforms:
A Primer, Prentice-Hall, ISBN 0134896006, Upper Saddle River, NJ.
Douglas, H. & Pillay, P. (2005). The impact of wavelet selection on transient motor current
signature analysis, Proceedings of 2005 IEEE International Conference on Electric
Machines and Drives, ISBN 0780389875; 978-078038987-8, San Antonio, TX, May
2005.
Hatch, C.T. & Bently, D.E. (2002). Fundamentals of Rotating Machinery Diagnostics, Bently
Pressurized Bearing Press, ISBN 0-9714081-0-6, Minden, NV.


189
James, E. & Bery, P.E. (1994). IRD Advancement Training Analysis II: Concentrated Vibration
Signature Analysis and Related Condition Monitoring Techniques, IRD Mechanalysis
Inc., Columbus, Ohio.
Kim, Y.H., Tan, Andy. C.C., Mathew, J., Kosse, V. & Yang, B.S. (2007). A comparative study
on the application of acoustic emission technique and acceleration measurements
for low speed condition monitoring, 12th Asia-Pacific Vibration Conference, Hokkaido
Univ. Japan, August 2007.
Kim, Y.H., Tan, Andy. C.C., Mathew, J. & Yang, B.S. (2005). Experimental study on incipient
fault detection of low speed rolling element bearings: time domain statistical
parameters, 12th Asia-Pacific Vibration Conference, Hokkaido Univ. Japan, August
2007.
Li, H., Zhang, Y. & Zheng, H. (2009). Gear fault detection and diagnosis under speed-up
condition based on order cepstrum and radial basis function neural network,
Journal of Mechanical Science and Technology, Vol. 23, No. 10, (October 2009), pp.2780-
2789, ISSN 1738494X.
Mba, D. & Bannister, R. H. (1999). Condition monitoring of low-speed rotating machinery
using stress waves: Part 1 and Part 2, Journal of Process Mechanical Engineering. Vol.
213, No. 3, (1999), pp.153-185, ISSN 0954-4089(Print), 2041-3009(Online)
Mba, D., Cooke, A., Roby, D. & Hewitt, G. (2003). Opportunities offered by acoustic
emission for shaft-seal rubbing in power generation turbines; a case study.
Conference sponsored by the British Institute of NDT. International Conference on
Condition Monitoring, ISBN 1901892174, Oxford, UK, July 2003.
Mba, D., Cooke, A., Roby, D. & Hewitt, G. (2004). The detection of shaft-seal rubbing in
large-scale power generation turbines with acoustic emission; Case study,
Proceedings of the Institution of Mechanical Engineers, Part A: Journal of Power and
Energy, Vol. 218, No. 2, (March 2004), pp.71-81, ISSN 0957-6509.
Misiti, M., Misiti, Y., Oppenheim, G. & Poggi, J.M. (2009) Wavelet Toolbox TM 4 users guide,
The MathWorks. Inc., Retrieved from <http://www.mathworks.com >
Ronnie, K.M. & V. K. Eric, (2005). Nondestructive testing handbook Volume 6; Acoustic Emission
Testing, American Society for Nondestructive Testing, ISBN 978-1-57117-137-5,
Columbus, Ohio.
Sato, I. (1990). Rotating machinery diagnosis with acoustic emission techniques, Electrical
Engineering of Japanese, Vol. 110, No. 2, (1990), pp.115-127, ISSN 04247760.
Sentoku, H. (1998). AE in tooth surface failure process of spur gear, Journal of Acoustic
Emission, Vol. 16 No. 1-4, (August 1998) pp.S19-S24, ISSN 0730-0050.
Sheen, Y.T. (2008). An envelope detection method based on the first-vibration-mode of
bearing vibration, Measurement: Journal of the International Measurement
Confederation, Vol. 41, No. 7, (August 2008), pp.797-809, ISSN 0263-2231.
Sheen, Y.T. (2010). An envelope analysis based on the resonance modes of the mechanical
system for the bearing defect diagnosis, Measurement: Journal of the International
Measurement Confederation, Vol. 43, No. 7, (August 2010), pp.912-934, ISSN 0263-
2241.
Shiroishi, J., Li, Y., Lian, S., Danyluk, S. & Kurfess, T. (1999). Vibration analysis for bearing
outer race condition diagnostics, Journal of Brazilian Society of Mechanical Science,
Vol. 21, No. 3, (September 1999), pp.484-492, ISSN 0100-7386.


190
Singh, A., Houser, D. R. & Vijayakar, S. (1996). Early detection of gear pitting, American
Society of Mechanical Engineers, Vol. 88, (1996) pp.673-678, ISSN 15214613.
Siores, E. & Negro, A.A. (1997). Condition monitoring of a gear box using acoustic emission
testing, Material Evaluation, Vol. 55, No. 2, (February 1997), pp.183-187, ISSN
00255327
Tan, C.K. & Mba, D. (2005). Limitation of acoustic emission for identifying seeded defects in
gearboxes, Journal of Non-Destructive Evaluation, Vol. 24, No. 1, (March 2005), pp.11-
28, ISSN 0195-9298.
Tan, C.K., Irving, P. & Mba, D. (2007). A comparative experimental study on the diagnostic
and prognostic capabilities of acoustics emission, vibration and spectrometric oil
analysis for spur gears, Mechanical Systems and Signal Processing, Vol. 21, No. 1,
(January 2007), pp.208-233, ISSN 0888-3270
Wu, J.D., Hsu, C.C. & Wu, G.Z. (2009). Fault gear identification and classification using
discrete wavelet transform and adaptive neuro-fuzzy inference, Expert Systems with
Applications, Vol. 36, No. 3, (April 2009), pp.6244-6255, ISSN 0957-4174.
Wu, J.D. & Chen, J.C. (2006). Continuous wavelet transform technique for fault signal
diagnosis of internal combustions engines, NDT&E International, Vol. 39, No. 4,
(June 2006), pp.304-311, ISSN 0963-8695.
Yang, Y., Yu, D. & Cheng, J. (2007). A fault diagnosis approach for roller bearing based on
IMF envelope spectrum and SVM, Measurement: Journal of the International
Measurement Confederation, Vol. 40, No. 9-10, (November 2007), pp.943-950, ISSN
0263-2231.
9
Compensation of Ultrasound Attenuation in
Photoacoustic Imaging
P. Burgholzer
1,2
, H. Roitner
1,2
, J. Bauer-Marschallinger
1,2
,
H. Grn
2
, T. Berer
1,2
and G. Paltauf
3

1
Christian Doppler Laboratory for Photoacoustic Imaging and Laser Ultrasonics,
2
Research Center for Non Destructive Testing (RECENDT),
3
Institute of Physics, Karl-Franzens-University Graz
Austria
1. Introduction
Photoacoustic imaging is a non-destructive method to obtain information about the
distribution of optically absorbing structures inside a semitransparent medium. It is based
on thermoelastic generation of ultrasonic waves by the absorption of a short laser pulse
inside the sample. From the ultrasonic waves measured outside the object, the interior
distribution of absorbed energy is reconstructed. The ultrasonic waves, which transport
information from the interior to the surface of the sample, are scattered or absorbed to a
certain extent by dissipative processes. The scope of this work is to quantify the information
loss which is equal to the entropy production during these dissipative processes and thereby
to give a principle limit for the spatial resolution which can be gained in photoacoustic
imaging. This theoretical limit is compared to experimental data. In this book chapter state-
of-the-art methods for modeling ultrasonic wave propagation in the case of attenuating
media are described. From these models strategies for compensating ultrasound attenuation
are derived which may be combined with well-known reconstruction algorithms from the
non-attenuating case for photoacoustic imaging.
Section 2 gives a short description of photoacoustic imaging, especially photoacoustic
tomography, and the available image reconstruction algorithms to reconstruct the interior
structure from the detected ultrasound signal at the sample surface. Beside small point-like
detectors also large detectors, so called integrating detectors are used for photoacoustic
tomography. The latter ones require different image reconstruction algorithms. Spatial
resolution is an essential issue for any imaging method. Therefore we describe the
influencing factors of the resolution in photoacoustic tomography.
Section 3 is dedicated to acoustic attenuation. The spatial resolution in photoacoustic
imaging is limited by the acoustic bandwidth. To resolve small objects shorter wavelengths
with higher frequencies are necessary. For such high frequencies, however, the acoustic
attenuation increases. This effect is usually ignored in photoacoustic image reconstruction
but as small objects or structures generate high frequency components it limits the
minimum detectable size, hence the resolution. Several models for acoustic attenuation,
especially used for ultrasound propagation in biological tissue, are compared with
experimental data.


192
Section 4 describes two different attempts to compensate this acoustic attenuation: either to
include the compensation directly in the image reconstruction, e. g. in a modified time
reversal method, or to calculate first the acoustic signal without attenuation from the
measured attenuated signal and then perform the conventional photoacoustic image
reconstruction. As any compensation of acoustic attenuation is mathematically an ill-posed
problem both methods need regularization to prevent small measurement fluctuations from
growing infinitely high in the reconstructed image. The possible degree of compensation
depends on the size of these fluctuations. On the other hand acoustic attenuation is a
dissipative process that causes entropy production equal to a loss of information, which
cannot be compensated by any compensation algorithm. Therefore one can use the entropy
production caused by acoustic attenuation to determine the minimal fluctuations in the
measurement data, which turn out to be equal to thermal fluctuations. In statistical physics
this fact is well known as the fluctuation-dissipation theorem, but the information
theoretical background as a starting point to derive this theorem was not mentioned before
in the literature.
Section 5 uses stochastic processes to understand theoretically how information can be lost
and its connection to entropy production. Therefore, the measured pressure signal is treated
as a random variable with a certain mean value as a function of time and certain fluctuations
around that value. First for the simple model of a damped harmonic oscillator, it is shown
how information is lost during a dissipative process and to what extent we can reconstruct
the original information after some time. Then attenuated acoustic waves can be treated in a
similar way: the spatial Fourier transform of the pressure wave can be described by a similar
stochastic process as the damped harmonic oscillator only in a higher dimension. Each
wave vector is represented by a damped oscillator of different frequency.
Thinking about acoustic attenuation as a stochastic process helps to understand how
entropy production and loss of information work on a microscopic scale. Beside a better
theoretical insight the stochastic view on the acoustic wave answers a very important
question: which is the best compensation method and the corresponding practicable spatial
resolution in photoacoustic imaging? This question can be answered without taking
fluctuations on a microscopic scale into account: the entropy production, which can be
calculated from macroscopic mean values, is set equal to the information loss.
2. Photoacoustic imaging
In 1880, Alexander Graham Bell discovered that pulsed light striking a solid substrate can
produce a sound wave, a phenomenon called the photoacoustic effect (Bell, 1880). Practical
imaging methods based on this effect have been developed and reported the last decade (Xu &
Wang, 2006). Today, photoacoustic imaging, which is also referred to as optoacoustic imaging
or, when using microwaves instead of light for excitation, as thermoacoustic imaging, is
attracting intense interest for cross-sectional or three-dimensional imaging in biomedicine.
In photoacoustic imaging, short laser pulses are fired at a sample and the absorbed energy
causes local heating (Fig. 1). This heating causes thermoelastic expansion and thereby
generation of broadband elastic pressure waves (ultrasound) which can be detected outside
the sample, for example by a piezoelectric device or by an optical detector. Two methods are
used for photoacoustic imaging: photoacoustic microscopy uses focused ultrasonic detectors
and the sample is imaged by scanning the focus through the sample. In photoacoustic
tomography (PAT) an unfocused detector is used which detects the pressure from the

Compensation of Ultrasound Attenuation in Photoacoustic Imaging

193
ultrasound wave arriving from all different locations of the source. A map or image of the
photo-generated pressure distribution in the sample can be made by collecting the
ultrasound at many different locations and processing it using a suitable algorithm e.g. by a
filtered backprojection algorithm or by a time reversal algorithm.
Only if the pulse is short enough, thermal expansion causes a pressure rise proportional to
the locally absorbed energy density. Short enough means that the pressure wave does not
run out of the smallest structure which should be resolved in the photoacoustic image
during the pulse time. This so called stress confinement is therefore fulfilled if the sound
velocity multiplied by the pulse time of the laser is small compared to the spatial resolution
one wants to achieve in imaging. Another constraint is the thermal confinement which is
fulfilled if the heat induced in a structure by the absorbed laser pulse does not diffuse out of
this structure during the time of the laser pulse. As heat diffusion is usually slower than the
propagation of sound the thermal confinement is fulfilled if stress confinement is fulfilled.

Fig. 1. Photoacoustic Imaging the spatial resolution is determined by excitation,
propagation, and detection of the acoustic wave. (a) Thermoelastic generation of acoustic
wave by laser light (arrows indicate scattered photons): excited pressure is proportional to
the absorbed optical energy density, if laser pulse is short enough to satisfy thermal and
stress confinement. (b) Propagation of ultrasonic wave to sample surface: frequency
dependent acoustic attenuation causes entropy production and therefore a loss of
information. (c) Detection of ultrasonic wave: bandwidth and size of detector limits spatial
resolution
Any photons, either unscattered or scattered (see arrows in Fig. 1), contribute to the
absorbed energy as long as the photon excitation is relaxed thermally. Therefore PAT
visualizes the product of the optical absorption distribution and the local light fluence.
Using a Nd:YAG laser and an optical parametric oscillator (OPO) light pulses from the
infrared to the visible regime can be selected with a repetition rate from 10 Hz up to 100 Hz.
Some high speed PAT systems can even go up to 1000 Hz. The pulse duration in the
nanosecond range enables a theoretical resolution of several microns in tissue (sound
velocity similar to water at approx. 1500 m/s). For biomedical applications the light energy
should not exceed 20 mJ/cm
2
in the visible spectral range and 100 mJ/cm
2
in the near
infrared light.


194
If acoustic attenuation and shear waves (in liquid and soft tissue) are neglected, the acoustic
pressure p as a function of time and space obeys the equation
( ) ( ) ( )
2
2 2
1
, , ,
p
p t p t t
c t C t

=

r r r (1)
where is the three-dimensional Laplace operator, c the sound velocity, the thermal
expansion coefficient,
p
C the specific heat capacity and ( , ) H t r the deposited energy per
time and volume (heating function) caused by the absorption of the electromagnetic
radiation in the sample.
For short electromagnetic pulses ( , ) ( ) ( ) H t A t = r r , where ( ) A r is the energy density of the
absorbed electromagnetic radiation and ( ) t the Dirac delta function. Then the acoustic
pressure ( , ) p t r solves the homogeneous scalar wave equation with the initial conditions
( ) ( ) ( )
2
0
, 0 ( )
p
p p c C A A = = r r r r and / ( , 0) 0 t p = r . The initial pressure
0
p at time
0 t = is therefore directly proportional to the absorbed energy density A with the
dimensionless constant , the Grneisen coefficient.
As shown in Fig. 1 (c) bandwidth and size of the detectors for collecting the ultrasound
signals are important for the resolution of this imaging modality. A photoacoustically
generated ultrasound signal is a broadband signal and contains frequencies in the range
from kilohertz up to a few megahertz. Conventional point like piezo elements such as
known from arrays for medical ultrasonic imaging have their maximum sensitivity close to
their center frequency and can detect frequencies only within a certain bandwidth around
this frequency. Therefore high frequencies are not detected which correspond to small
structures and are necessary for image reconstruction with a high spatial resolution. Other
approaches are necessary for high resolution photoacoustic imaging. A hydrophone could
be one solution (Wang, 2008), or the utilization of optical point like detection as
demonstrated e.g. by (Zhang et al., 2009) or (Berer et al., 2010).
Point like detectors show a limit in achievable resolution by their size. The smaller the point
detector the better is the spatial resolution. Unfortunately thermal and other fluctuations
increase for a smaller detector, which results in a reduction of resolution. A totally different
approach is the use of so called integrating detectors which are at least in one dimension
larger than the object. This way the drawback of finite dimensions of point like detectors can
be overcome. Such an integrating detector for photoacoustic imaging was introduced by
(Haltmeier et al, 2004). They showed the mathematical proof of integrating area and line
detectors and introduced new reconstruction methods which are necessary when using such
a detector. First measurements using an integrating detector were shown by (Burgholzer et
al., 2005). The first integrating detector was an area detector which was bigger than the
object in two dimensions. For sufficient data for 3D image reconstruction the area detector
had to be scanned around the object tangential to the surface of a sphere. This detector
movement is difficult to realize. Hence the idea of the integrating line detector was
developed. A fragmentation of the area detector into an array of line detectors results in an
easier setup with only one rotation axis for the object and a linear motion of the integrating
line detector (Fig. 2).
An integrating line detector is a line which has at least a length 8 * D , where D is the
diameter of a circle enclosing the sample and tangentially touching the line detector
(Haltmeier et al., 2004). The line detector integrates the pressure along the line on a


195
cylindrical surface with the radius c t where c is the speed of sound in the medium and t
the time. Thus, integrating line detectors arranged in an array around the sample, e.g. in a
circle, measure projection images of the object in a first measurement and reconstruction
step. By rotating the sample and measuring such projection images from different angles it
is possible to reconstruct a 3D image (Fig. 2). Three dimensional image reconstruction from
a set of projection images requires only the application of the inverse Radon transform.
Therefore 3D imaging using integrating line detectors is not computationally intensive
compared with other algorithms which reconstruct a 3D image from a set of signals
acquired from point like detectors.

Fig. 2. Line detectors around a sample rotated on one axis. Either one line detector is
scanning around the object or a detector array is used
Since the first measurement results from (Burgholzer et al, 2005) the integrating line detector
was further improved in sensitivity and spatial resolution. Several types of such line
detectors were implemented. One premature approach was a line made of PVDF (piezo foil)
which provides high sensitivity but with the drawback of directivity. The next consequential
step was an optical line detector realized by an interferometer. A laser beam that is part of
an interferometer measures variations of the refractive index induced by the acoustic
pressure (elasto-optic effect) (Paltauf et al., 2006). Optical detectors offer a broadband
characteristic and due to the circular shape of a laser beam or light guiding fiber there is no
such directivity like when using a film of piezo foil. Two main approaches can be
distinguished: free-beam implementations and the use of fiber-based interferometers.
Independent of the realization different types of the interferometer can be used, e.g. a Mach-
Zehnder or a Fabry-Perot interferometer which is in general more sensitive than the first one.
(Paltauf et al., 2006) presented measurements using a free-beam Mach-Zehnder
interferometer. They used a focused laser beam as line detector. When placing the object
next to the beam waist of the focused laser beam the best spatial resolution due to the
smallest beam diameter could be achieved (Paltauf et al., 2008).
(Grn et al, 2010) implemented fiber-based line detectors. The advantages of fiber-based line
detectors are the easy handling and the small and constant beam diameter in the fiber. A
small diameter of the laser beam is necessary for a good spatial resolution. The smaller the


196
diameter of the detecting part the better is the spatial resolution. A typical single mode fiber
for near infrared has a core diameter of 9 microns; single mode fibers for the visible range of
detection wavelength have typically about 6 microns. Due to the constant diameter along
the whole line detector this type of integrating line detector is dedicated for the imaging of
big samples. After first implementations of a Mach-Zehnder and a Fabry-Perot
interferometer in glass fibers now polymer fibers are used. Due to the much better
impedance matching of polymer fibers to the surrounding water, their sensitivity is higher
than for glass fibers, where approximately 2/3 of the incoming signal is reflected before
reaching the core (Grn et al., 2010). Furthermore the Youngs modulus is much lower in
polymer fibers than in glass fibers for which reason the deformation of a polymer fiber is
bigger than of a glass fiber applying the same pressure wave. As the strain optic coefficients
are in the same order, this results in an enhanced change of refractive index and thus to
higher signal amplitudes in the polymer fiber (Kiesel et al., 2007).
(Nuster et al, 2009) did a comparison of the different implementations of an integrating line
detector. At this stage of development the free-beam Mach-Zehnder interferometer was the
most sensitive integrating line detector. But these measurements showed some new
approaches how the fiber-based line detector could be made more sensitive, e.g. by building
up a Fabry-Perot interferometer in a single mode polymer fiber.
The next step after developing a sensitive line detector, no matter of which approach is the
most sensitive one, is the creation of an array of many integrating line detectors, e.g. 200
detectors arranged in a curve around the sample. This way one could acquire all data for a
projection image within one excitation pulse of the laser.
3. Acoustic attenuation
The imaging resolution in photoacoustic imaging is limited by the acoustic bandwidth and
therefore by the laser pulse duration as mentioned above, but also by the attenuation of the
acoustic wave on its way to the sample surface, and finally by the bandwidth and size of the
ultrasonic transducer. The acoustic attenuation can be substantial for high frequencies. This
effect is usually ignored in reconstruction algorithms but can have a strong impact on the
resolution of small objects or structures within objects. Stokes could already show in 1845
that for liquids with low viscosity, such as water, the acoustical absorption increases by the
square of the frequency (Stokes, 1845). If we do not take acoustic attenuation for
photoacoustic image reconstruction into account, especially small structures (corresponding
to shorter wavelengths and therefore higher frequencies) appear blurred (La Riviere et al.,
2006). To what extent this blurring can be compensated by regularization methods (as
performed by (La Riviere et al., 2006)) and how much information is lost due the
irreversibility of attenuation is investigated in this chapter.
For thin layers (1D), small cylinders (2D), and small spherical inclusions (3D) the effect of
attenuation is simulated and experimental results for several types of tissue are given. For
photoacoustic tomography a new description of attenuation seems to be useful: like for a
standing wave in a resonator the wave number is real but the frequency is complex. The
complex part of the frequency is the damping in time. The resulting pressure wave as the
solution of the wave equation is of course the same as by decomposing into plane waves
with complex wave number. But with the complex frequency description acoustic attenuation
can be included in all k-space methods well known in photoacoustic tomography just by
introducing a factor describing the exponential decay in time (Roitner & Burgholzer, 2011).


197
Acoustic attenuation is an irreversible process and therefore the wave equation is not
invariant to time reversal. Several important reasons for acoustic attenuation have been
reported: viscosity, heat conduction, relaxation processes and chemical reactions. Stokes
derived the scalar wave equation

2
2 2
1
0
p p
p
c t t

+ =

(2)
under the assumption of adiabatic conditions (thus neglecting the loss due to heat
conduction) (Stokes, 1845). This equation can also be found in (Shutilov, 1988) and can
generally be derived from a relaxation behavior of pressure and density (Royer &
Dieulesaint, 2000), where the density change follows the pressure change after a relaxation
time . If is further expressed in terms of viscosity and specific heat, this equation is also
known as the thermoviscous equation. Eq. (2) describes acoustic attenuation which is
approximately proportional to the square of the frequency. Other wave equations can
describe a more general power law frequency dependence of the attenuation of the form
0
( )
y
= (0 < y < 3). (Szabo, 1994) has suggested adding the loss term ( ) ( ) , L t p t r to
the wave equation in order to account for such attenuation behavior:
( ) ( ) ( ) ( )
2
2 2
0
1
, , , 0 p t p t L t p t
c t
+ =
r r r (3)
Other models for acoustic wave propagation in acoustic media have been proposed also by
(Nachman et al., 1990) and (Treeby & Cox, 2010).
Measurement and simulation of broadband acoustic attenuation:

Fig. 3. Experimental set-up to measure broadband high frequency acoustic attenuation in
tissue with ultrasound generated by short laser pulses (a) and by piezoelectric transducers
(b). In both cases a piezoelectric transducer receives the ultrasonic signals


198
Fig. 3 shows the set up to determine frequency dependent acoustic attenuation in tissue by a
single transmission experiment. High frequency ultrasonic pulses are either generated by a
pulse laser (pulse duration 6 ns) heating up the surface of a silicon wafer (Fig. 3a) or by a
piezoelectric transducer (Fig. 3b). The resulting planar-like ultrasound waves propagate
through a fat tissue with varying thickness and through distilled water as a coupling
medium to a piezoelectric transducer. The ultrasonic attenuation is determined by
comparing transmission measurements of the investigated samples and distilled water. In
Fig. 4 Fourier transformed attenuation results for subcutaneous fat of pig, human blood and
olive oil are shown. A power law
0
( )
y
f f = can be applied to those substances, where f
denotes the frequency (Bauer-Marschallinger et al., 2011).

Fig. 4. Attenuation as a function of frequency for three biological substances (double
logarithmic scale)
The attenuated planar ultrasound waveform can be calculated by a number of available
simulation methods. In a comprehensive study (Roitner et al., 2011) we compared measured
waveforms for two fat thicknesses (3.2 mm and 6.2 mm) to simulated waveforms obtained
with three current simulation methods. The methods using the Matlab toolbox by (Treeby &
Cox, 2010b) or the relation (4) by (LaRiviere et al., 2006) rely on a frequency power law
absorption and are described in detail below. A third method by (Nachman et al., 1990)
assumes multiple molecular relaxation processes as the cause of attenuation. These
processes are characterized by their contributions ( 1 )
n
n N = to the isothermal
compressibility of the tissue and the relaxation times ( 1 )
n
n N = of their vibrational


199
energies. With these parameters a complex wave number is obtained involving a
generalized compressibility
1
( ) ( ) exp( / ) ( )
N
n
n
n n
t t t H t
=
= +
where
1
N
n
n

=
=
and
( ) H t is the Heaviside step function. denotes the usual compressibility calculated with
the formula
2 1
( ) c

= which holds in liquids as well as in soft biological tissues of density
. Stokes equation (2) is seen to be the special case of a single relaxation mechanism 1 N = ,
1
= .
All three simulation methods produce waveforms in good agreement to the measured
waveform. Of course, the approximation will become less accurate with longer propagation
distance in the absorbing medium.
Simulation results may also be validated if we recall that the frequency domain transfer
function of the pressure signal propagating through a layer with complex wavenumber
( ) K and thickness d equals exp(iK( ) ) d . So the simulated water-fat-water waveform
may be obtained from the measured water-only waveform by inverting water attenuation
over the fat thickness and then applying fat attenuation over the same fat thickness.
4. Compensation of acoustic attenuation
When taking acoustic attenuation into account, the wave equation, e.g. (2) or (3), is not
invariant to time reversal any more. First order terms in time t or higher order odd terms
change sign if time is reversed. The equations then do not behave well any more, noise is
amplified exponentially and regularization methods have to be used for solving these time
reversed equations. Using such regularized methods the spatial resolution, that is limited by
the frequency-dependent damping, is improved (Burgholzer et al., 2007). To prevent high-
frequency noise from growing exponentially, Fourier spectral methods (Trefethen, 2000) in
space have been used. They utilize the spatial Fourier transform to calculate the Laplacian
and therefore allow for incorporating a damping of higher frequencies when calculating the
time reversal. A similar algorithm to compensate acoustic attenuation step by step was
proposed by (Treeby et al., 2010c)

.
In the second strategy, attenuation is compensated in one step where an approximation to
the un-attenuated measured signal is calculated from the attenuated measured signal by
solving an appropriate integral equation (La Riviere et al., 2006); then an ordinary
reconstruction algorithm for photoacoustic tomography is applied as in the absence of
attenuation.
The Matlab Toolbox by Treeby et al. implements a state-of-the-art algorithm to simulate
ultrasound wave propagation (the direct problem) and image reconstruction (the inverse
problem) for PAI. For the direct and inverse problems arbitrary source distributions and
detector geometries, variable medium densities and sound speeds and an arbitrary constant
medium absorption are supported in 1D-3D. The algorithm is first-order finite difference in
time and pseudospectral in space (working in spatial Fourier space = k-space). The
linearized Euler equations in a fluid
1
p
t
u
,
t
u together with an
adiabatic equation of state
2
0
p c = are solved up to time T for the sound velocity vector u ,
pressure p and density . In the case of absorbing media the equation of state is extended to


200
2 1 2 1
0
( , )
( , ) ( , ) ( ( , ))
y y
t
p t c t k k t
t

= + +
k
r r k F , where is related to attenuation and to
dispersion (Treeby & Cox, 2010) and the inverse Fourier transform (IFT) in d spatial
dimensions is defined by
1

( ) ( ( )) (2 ) ( )exp( )
d
d
R
f f f i d

= =
r k k k r k F . With this extension of

the equation of state frequency-domain ultrasound absorption power laws of the form
0
( )
y
= can be supported.
For the inverse problem, the same algorithm as for the direct problem is applied, but with
zero initial conditions ( , 0) 0 p = r , ( , 0) = u r 0 and time-varying Dirichlet boundary conditions
on detector surface points
S
r in the form of the time-reversed sensor data
( , ) ( , )
S meas S
p t p T t = r r . Absorption is compensated by inverting the sign of but leaving
unchanged. Additionally, a regularization filter is applied in k-space that suppresses the
higher frequencies.

p
i
x
e
l

N
r
.
pixel Nr.

200 300 400 500 600
200
300
400
500
600
a

pixel Nr.
p
i
x
e
l

N
r
.

200 300 400 500 600
200
300
400
500
600
b

pixel Nr.
p
i
x
e
l

N
r
.

200 300 400 500 600
200
300
400
500
600
c

pixel Nr.
p
i
x
e
l

N
r
.

200 300 400 500 600
200
300
400
500
600
d

Fig. 5. Reconstruction of nine-circle phantom: (a) initial pressure distribution, (b)
reconstruction Matlab toolbox with taper ratio r = 0.5, (c) reconstruction Matlab toolbox with
taper ratio r = 0.75, (d) reconstruction one-step compensation plus fast series algorithm


201
A Tukey (tapered cosine) window is chosen for this filter. This window is parameterized by
its cutoff-frequency in k -space (here
4
2 10 / rad m , corresponding to 2 15MHz in -
space) and its taper ratio r . The ratio controls the filter roll-off: raising r suppresses more
high-frequency content in the measurement noise. In Fig. 5a we display a complex phantom
consisting of nine circles of different sizes and source strengths, surrounded by a circular
array of detectors (which could represent line detectors in 3D). This phantom was
reconstructed first with taper ratio 0.5 r = (Fig. 5b) and then with taper ratio 0.75 r = (Fig.
5c). The absorption parameters were an exponent of 1.5 y = and
1.5
0
3 /( ) dB cm MHz = ,
and the measurement noise consisted only of numerical noise.
Note that in Fig. 5b ring-like artifacts are produced during reconstruction, which are much
reduced in Fig. 5c because there the regularization filter is more restrictive. With taper ratio
1 r = they disappear altogether. If a simpler phantom is used, e.g. just a single circle, taper
ratio 0.5 r = does not produce any artifacts. To sum up, complex, fine-structured objects
exhibit more signal content and hence more measurement noise at higher frequencies. Then
the regularization filter must be parameterized more restrictively than for compact objects.
The formula used for calculating and compensating attenuation in one step was first
presented by (LaRiviere et al., 2006) and reads

0
0
( , ) ( , )exp( ( ) )
( )
att S ideal S
p p t ic K t dt
c K

r r (4)
This relation yields the temporal Fourier transform of the attenuated pressure at a given
point
S
= r r (and, via an IFT, the attenuated pressure itself) if the ideal (i.e. un-attenuated)
pressure is known over time at that point. This can be exploited for the direct problem
where the evaluation involves only two nested numerical integrations. For the inverse
problem, the left-hand side of (4) is known and the relation represents an integral equation
to be solved for ( , )
ideal S
p t r . Since this solution is very sensitive to measurement noise, i.e.
noise in the spectrum of ( , )
att S
p t r , regularization is necessary, for instance in the form of
truncated singular value decomposition (SVD, see e.g. La Riviere et al., 2006).
If the detector geometry is regular, fast reconstruction algorithms using series expansions in
eigenfunctions of the Laplacian on the regular domain are available for the lossless case
making the whole reconstruction with compensation much faster than reconstruction with
the mentioned toolbox algorithm. In Fig. 5d we display a reconstruction of the nine-circle
phantom using this procedure. To apply the SVD, a certain noise level in temporal Fourier
space must be assumed. It is desirable to base this assumption on physical reasons - this will
be discussed below. Here we assume a white Gaussian noise with a standard deviation of
0,1% of the largest amplitude in the spectrum.
Fig. 5d shows that the reconstruction with the one-step compensation is almost of the same
quality as the reconstruction with the Matlab toolbox algorithm. However, computation
time is reduced by a factor of 10 in comparison to the toolbox algorithm.
We already observed that the quality of reconstructions depends crucially on a good
analysis of the amplitude and spectral distribution of the measurement noise. On the source
side, advance knowledge of the type of object to be imaged is helpful (more bulky and
compact, or more fine-structured). On the detector side, noise is created by fundamental
physical processes as well as by technological limitations.


202
First, for any material volume V at temperature T pressure fluctuations of variance
( ) /
B
Var p k T V = (Landau&Lifshitz, 1980) will occur where is the adiabatic
compressibility and
B
k is the Boltzmann constant. So if detectors are e.g. thin foils with
small volume, this may play a role since noise amplitudes grow like
1/2
V

.
Second, the fluctuation-dissipation theorem states that sound absorption processes in media
create pressure fluctuations. If the absorption is described by Stokes equation (2), a
straightforward calculation yields a power spectral density of these pressure fluctuations
2
2 2 2
2
( )
)
B
P
k T
S

=
(1+
.
Third, noise is created by the analog-to-digital conversion process during electronic
acquisition of the pressure signal. If this quantization noise is assumed to be white, its
power spectral density at sampling frequency
s
f will be constant and equal
2
/12
s
q f
(Widrow & Kollar, 2008) where q is the quantization interval.
We have given only an outline of the considerations that have to be taken into account for a
good estimation of the measurement noise. In a concrete application the relative importance
of the mentioned contributions will depend on sizes and shapes of objects and detectors, the
physics of the ultrasound propagation medium (compressibility, temperature) and the
properties of the measurement devices.
5. Stochastic processes for modeling acoustic attenuation
As for any other dissipative process the energy of the attenuated acoustic wave is not lost
but is transferred to heat, which can be described in thermodynamics by an entropy
increase. This increase in entropy is equal to a loss of information, as defined by Shannon,
and no compensation algorithm can compensate this loss of information. This is a limit
given by the second law of thermodynamics. But can this limit be found in the algorithms
for compensation of acoustic attenuation given in the previous section? Fluctuations of the
measured pressure are amplified exponentially during the compensation and therefore
we suspect a relation between the entropy production caused by acoustic attenuation and
the fluctuations. Indeed such a relation is known in statistical physics as the fluctuation-
dissipation theorem and is due to (Callen & Welton, 1951), (Callen & Green, 1952), and
(Greene & Callen, 1952). It represents in fact a generalization of the famous (Johnson, 1928)
(Nyquist, 1928) formula in the theory of electric noise.
In this section we use the theory of non-equilibrium thermodynamics presented by S.R. de
Groot and P. Mazur (De Groot & Mazur). More about random variables and stochastic
processes can be found e.g. in the book about Statistical Physics from (Honerkamp, 1998).
An introduction to stochastic processes on an elementary level has been published by.
(Lemons, 2002), also containing On the Theory of Brownian Motion by (Langevin, 1908).
An introduction to Markov Processes is given by (Gillespie, 1992).
In this section the measured pressure signal is treated as a time-dependent random variable
with a mean value and a variance as a function of time. To be able to use the results of some
model stochastic processes given in literature (Ornstein-Uhlenbeck process or damped
harmonic oscillator) for a model of photoacoustic reconstruction we have changed the initial
conditions: instead of a defined initial value (with zero variance) we have taken the
stochastic process at equilibrium before time zero and at a time zero a certain perturbation
has been applied to the process (e. g. a rapid change in momentum for the damped


203
harmonic oscillator called kicked damped oscillator in this chapter). Reconstruction of the
size of this perturbation at time t=0 from the measurement after a time t shows how the
information about the size of this kick at t=0 gets lost with increasing time if dissipative
processes occur. This change of the initial conditions has a significant advantage: it turns out
that the variance stays constant in time, while the mean value is a function of time. This
facilitates the calculations of the entropy production and of the information loss due to the
stochastic process.
Gauss-Markov processes and entropy
In this section we shall assume that the time varying stochastic processes will have Gauss-
Markov character. In doing so we do not wish to assume that all dissipative macroscopic
processes considered belong to this specific class of Gauss-Markov processes. It may,
however, be surmised that a number of real phenomena may, with a certain approximation,
be adequately described by such processes (De Groot & Mazur). The advantage of
specifying more precisely the nature of the processes considered is that it enables us to
discuss, on the level of the theory of random processes, the behavior of entropy production
and of information loss.
Following the theory of random fluctuations given e.g. by (De Groot & Mazur), we take as a
starting point equations analogous to the Langevin equation used to describe the velocity of
a Brownian particle:
( )
d
t
dt
= +
M (5)
The components of the vector are the random variables ( 1, 2,..., )
i
i n = , having zero mean
value at equilibrium. The matrix Mof real phenomenological coefficients is independent of
time. The vector ( ) t represents white noise, which is uncorrelated at different times. The
distribution density of turns out to be an n-dimensional Gaussian distribution:

1
1
( ( )) ( ( ))
2
1
( , )
(2 )
T
t t
n
f t e

(6)
with the mean value ( ) t and the covariance matrix , which is usually also a function of
time but for the initial conditions used later on the covariance matrix will be constant in
time. is the determinant of the covariance matrix. If the initial value of is given by a
given
0
, the mean value at a later time t will be:

0
( )
t
t e
=
M
(7)
By an adequate coordinate transformation of the matrix Mcan be diagonalized: the
eigenvalues of M are the elements of the diagonal matrix.
According to the second law of thermodynamics the entropy of an adiabatically insulated
system must increase monotonously until thermodynamic equilibrium is established within
the system. Then the entropy will be set to zero and the entropy at a time t is:

( , )
( ) ( , )ln
( , )
B
f t
S t k f t d
f t
=
(8)


204
with the Boltzmann constant
B
k . For a constant covariance matrix the above integration
results in:

1
1
( ) ( ) ( )
2
T
B
S t k t t
= (9)
Before modeling the attenuated acoustic wave as a Gauss-Markov process we give two
simple examples: the Orstein-Uhlenbeck process with only one component of as a model
for the velocity of a Brownian particle and the damped harmonic oscillator with two
components of .
Example: kicked Ornstein-Uhlenbeck process
If the random vector in eq. (5) has only one component we get the Langevin equation

( )
( ) ( )
dv t
v t t
dt
= + (10)
which was used to describe Brownian motion of a particle. The random variable v is the
particle velocity, v is the viscous drag, and is the amplitude of the random
fluctuations. The Langevin equation governs an Ornstein-Uhlenbeck process, after
(Uhlenbeck & Ornstein, 1930), who formalized the properties of this continuous Markov
process. Now we assume that initially we have thermal equilibrium with zero mean
velocity. At time zero the particle is kicked which causes an immediate change in velocity of
0
v . Following eq. (7) the mean value ( ) v t shows an exponential decay:

0
( )
t
v t e v

= (11)
The variance of the velocity ( ) Var v is
2
2 and is constant in time. In Fig. 6 time and
velocity are scaled to be dimensionless and the standard deviation (square root of the
variance) of the velocity is normalized.
From eq. (9) the information loss equal to the entropy production till the time t after the kick
is:

2
2
( ) ( )
B
S t k v t
= (12)
On the other hand the entropy production known from thermodynamics is the dissipated
energy Q , which is the kinetic energy of the Brownian particle of mass m, divided by the
temperature T:

2
( )
( )
2
Q mv t
S t
T T
= = (13)
The thermodynamic entropy production in eq. (12) has to be equal to the loss of information
in eq. (13), and therefore we get for the variance of the velocity:

2
2
kT
m
= (14)


205
Eq. (14) has been derived previously by the equipartition theorem: the equilibrium energy
associated with fluctuations in each degree of freedom is k
B
T/2. We have used the equity of
entropy production and information loss. Eq. (14) states a connection between the strength
of the fluctuations, given by
2
, and the strength of the dissipation . This is the
fluctuation-dissipation theorem in its simplest form for uncorrelated white noise.

Fig. 6. Points on a sample path of the normalized kicked Ornstein-Uhlenbeck process
defined by the Langevin eq. (10). The solid lines represent the mean, and mean standard
deviation of the scaled velocity coordinate. At the time t=0 a value of v
0
=10 has been added
to the scaled velocity. After some time the information of the amplitude gets more and more
lost due to the fluctuations
It is instructive to determine the least square estimator (Honerkamp, 1998) for the initial
velocity
0
v . If we write for the estimated initial velocity
r
v
( ) ( )
r
v R t v t = (15)
we calculate ( ) R t by minimizing the mean error
2
0
( )
r
v v :

2 2 2
0
( )
2
t
t
e
R t
e v
=
+
(16)
This gives the Tikhonov regularization with
2 2 2
0 0
2 ( ) v Var v v = as regularization
parameter. The inverse square root of the regularization parameter
0
( ) v Var v can be
interpreted as signal-to-noise-ration (SNR) of ( ) v t .
Example: kicked harmonic osciallator
For modeling of acoustic waves one needs in addition to the dissipation also an oscillating
term. For pure oscillation without damping we have no loss of information (Fig. 7).


206

Fig. 7. Points on a sample path of the kicked harmonic oscillator without damping. The
momentum used to kick the oscillator can be reconstructed without loss of information at
any time after the kick if is known
The stochastically damped harmonic oscillator combines the oscillatory and the diffusive
behavior and therefore it is a good starting point to model attenuated acoustic waves. The
equations of motion are (using the momentum p instead of the velocity v ):

( ) 1
( )
dx t
p t
dt m
= (17)

2
0
( )
( ) ( )
dp t
p t m x t
dt m
= + (18)
These equations can be combined using a two dimensional random vector ( , ) x p = and
were solved already by (Chandrasekhar, 1943) for definite initial conditions (0) x and (0) p .
Again we have changed the initial conditions to an oscillator with zero mean values kicked
by an initial momentum
0
p . In Fig. 8 the damping is chosen to be
0
3 m = .
Using the fluctuation-dissipation theorem
2
2 kT = one obtains for the distribution
function

2 2 2
0
1 1
( ( )) ( ( ))
2 2
0
1
( , , )
2
p p t m x x t
mkT kT
f x p t e
kT

= , (19)
where ( ) x t and ( ) p t are the solutions of the ordinary (non-stochastic) damped harmonic
oscillator. Then one gets for the information loss from eq. (9)

2
2 2
0
( ) 1 1 1
( ) ( ( ) ) ( )
2 2
pot kin
p t
S t m x t E E
T m T
= + = + , (20)
which is equal to the entropy from thermodynamics, where E
pot
+E
kin
is the total energy of
the harmonic oscillator (sum of the potential and kinetic energy). This fact confirms again
that the fluctuation-dissipation theorem for the damped harmonic oscillator
2
2
B
k T =
can be derived from the equity of entropy production and information loss.


207
-4 -2 0 2 4 6 8 10
-6
-4
-2
0
2
4
6
8
10
scaled time (t /m )
s
c
a
l
e
d

o
s
c
i
a
l
l
a
t
i
o
n

(
x

s
q
r
t
(
2

m

)
/

)
-4 -2 0 2 4 6 8 10
-10
-5
0
5
10
15
scaled time (t /m )
s
c
a
l
e
d

m
o
m
e
n
t
u
m

(
p

s
q
r
t
(
2

/

m
)
/

)

Fig. 8. Points on a sample path of the normalized kicked damped harmonic oscillator (eq.
(17) and (18)). The solid lines represent the mean, and mean standard deviation of the
scaled oscillation (left) and momentum (right). At the time t=0 a value of p
0
=10 has been
added to the scaled momentum. After some time the information about the value of p
0
gets
more and more lost due to the fluctuations
For the mean value of the momentum ( ) p t one gets from eq. (7):

2
0
sin( )
( ) [cos( ) ]
2
t
m
t
p t p t e
m
= (21)
Compared to the Ornstein-Uhlenbeck process (eq. (11)) the mean value has not only an
exponential decay in time but also an oscillation with a frequency
2 2 2
0
/(4 ) m = . As
mentioned above (Fig. 7) the oscillating term does not change the entropy and no
information is lost. In the average only the exponential decay causes production of entropy
and the information about the value of
0
p gets more and more lost due to the fluctuations.
The entropy production which is proportional to the total energy is shown in Fig. 9.

-4 -2 0 2 4 6 8 10
0
10
20
30
40
50
scaled time (t /m )
e
n
e
r
g
y

Fig. 9. Total energy of the damped harmonic oscillator (solid line) shows in the average an
exponential decay with the time constant /m (dashed line)
Attenuated acoustic wave as a stochastic process
In photoacoustic imaging, the laser pulse at a time t = 0 generates an initial pressure
distribution ( )
0
p r (see section 2). For numerical calculations we use a discrete space
j
= r r
(j=1,..,N), where
j
r are N points on a cubic lattice with a spacing of r within the sample


208
volume V. At a time t the pressure distribution ( )
j
p t can be represented by a Fourier series
(Barret et al. 1995), including the time t = 0 with the initial pressure distribution:

2
1
( ) ( , ) ( ) ( ) ( ) ( )
k
N
i
j j k k j k
k
p t p t p t with e D

=
= = =
r
r r r r (22)
( ) D r is a support function which is one within the sample volume V and zero outside.
k

are integer points on an infinite 3D lattices. From eq. (1) we get (if acoustic attenuation can
be neglected):

2 2 2 2

( ) cos( ) (0) 4
k k k k k
p t t p with c = = . (23)
Therefore we have only an oscillating term as shown in Fig. 7, but in higher dimensions, and
no information is lost.
For an attenuated acoustic wave instead of the wave equation (1) we have used the Stokes
equation (2) or the wave equation (3), giving for the spatial Fourier components
( )
k
p t not
only an oscillating term but also an exponential decay:

( ) [ sin( ) cos( )]
k
t
k k k k k
p t A t B t e

= + (24)
where the phase factors
k
A and
k
B can be derived from the initial conditions and
k
and
k
are functions of
k
. Like for the damped harmonic oscillator this exponential decay
causes the loss of information and can be modeled by a Gauss-Markov process. The
information content in Fourier space is the same as in real space (Fig. 10). Therefore we can
describe the information loss in the attenuated acoustic wave by the same model as for the
damped harmonic oscillator, only in higher dimensions, as for each wave-vector-index k an
oscillator is needed.

Fourier space
k-space
Real space
) (
0
r p
t
k k k k k
k
e t B t A t p

+ = )] cos( ) sin( [ ) (
FT IFT
) 0 (
k
p
) , ( t r p
Time t

Fig. 10. The initial pressure distribution just after the laser pulse is Fourier transformed (FT).
The time evolution of the Fourier series coefficients can be described similar to the mean
value of a stochatsically damped harmonic oscillator. The pressure distribution after a time t
is then calculated by an inverse Fourier transform (IFT)
One dimensional (1D) example: photoacoustic signal of a 0.2 mm thin layer in glycerin:
The effect of acoustic attenuation for the reconstructed image is similar in 1D, 2D, and 3D
(Burgholzer et al., 2010a and 2010b). As in 1D the reconstructed image is just the shifted


209
measured signal, the effect of attenuation and of its compensation can be directly seen. The
photoacoustic signal of a 0.2 mm thin absorbing layer in glycerin is calculated by using the
scheme of Fig. 10 after a time of 4.5 microseconds. In glycerin the acoustic pressure can be
described well by the Stokes equation with a relaxation time of 244 picoseconds (Shutilov,
1988).
For an attenuated acoustic wave instead of the wave equation (1) we have used the Stokes
equation (2) or the wave equation (3), giving for the spatial Fourier components ) ( t p
k
not
only an oscillating term but also an exponential decay:
Putting eq. (24) into the Stokes equation (2) one gets:

2 2 2 2 2 2 2
1
4 (1 )
4
k k k
c c = and
2 2
k k
c = (25)
The initial pressure distribution is a one dimensional square pulse corresponding to an
absorbing layer of thickness 2a with a = 0.1 mm. In Fourier space we get:

1
(0) sin( ) 2
ikr
k k
p e ka withk
k

= = (26)
The calculated signal after a time of 4.5 microseconds is shown in Fig. 11 (dashed line). The
dashed dotted line shows the signal without attenuation (no relaxation time) and therefore
represents the ideal reconstructed image.

Fig. 11. Simulation result as an example for Stokes equation: it is shown how an initial
square pulse would look after a time of 4.5 microseconds in glycerine (=244ps) (dashed
line). The dashed dotted line shows the pulse without attenuation (=0), which correspond
to the reconstruction of the initial pressure pulse, and the solid line is the SVD
reconstruction of the initial pressure pulse (see text)
For compensation of the acoustic attenuation we use the time reversed process of the Gauss-
Markov process from Fig. 10. Similar to the stochastically damped oscillator the entropy
production for the acoustic wave is set equal to the loss of information from the Gauss-


210
Markov process with standard deviation s
k
and is approximated by an exponential decay in
time (Fig. 9):

2
2
1 1
( ) exp( 2 ) (0)
2
B k k
k k
S t k t p
s

(27)

2
2 2
0 0
1 1
( ) exp( 2 ) (0)
2
k k
k k
Q
S t t p
T c T s

(28)
which gives a variance of
( )
k
p t in the spatial Fourier space (k-space) independent from k:

2 2
0 0 k B
s k c T = (29)
In real space this gives the thermodynamic fluctuations. Or the other way around, the
thermodynamic fluctuations have a quantity such that the information loss of the Gauss-
Markov process is equal to the dissipated heat divided by the temperature. Using the
thermodynamic fluctuations for a detector size of 1 cm
2
and the SVD regularization from
section 4 the compensated
( )
k
p t is calculated. By applying a subsequent inverse Fourier
transform (IFT) one gets the compensated pressure profile in real space (solid line in Fig.
11).
6. Summary and conclusions
Acoustic attenuation is modeled as a stochastic process: this helps to understand how
thermodynamic entropy production and the decrease of information, which is transported
in the acoustic wave, are closely connected on a microscopic scale. This theoretical insight
enables to answer an important question in photoacoustic imaging: what is the highest
possible compensation of attenuation and therefore the best spatial resolution one can
achieve?
We could show that for thermal fluctuations the information loss of the reconstructed image
is equal to the entropy production due to attenuation of the acoustic wave. Therefore it is
sufficient to calculate the entropy production from the macroscopic mean values and it is
not necessary to take the fluctuations of the pressure into account.
The size and locations of detectors in photoacoustic imaging should be optimized to get the
best resolution and sensitivity. Up to now in such models it is not assumed that the pressure
is a random variable, which favors small point like detectors. Taking thermal fluctuations
and other noise into account will help to get a more realistic model for detectors and the
reconstructed images from measured signals with these detectors.
7. Acknowledgments
This work has been supported by the Christian Doppler Research Association, by the
Federal Ministry of Economy, Family and Youth, by the Austrian Science Fund (FWF)
project numbers S10503-N20 and TRP102-N20, by the European Regional Development
Fund (EFRE) in the framework of the EU-program Regio 13, and the federal state Upper
Austria.


211
8. References
Barret, H. H., Denny, J. L., Wagner, R. F. & Myers, K. J. (1995). Objective assessment of
image quality II, J. Opt. Soc. Am. A 12(5) 834-852
Bauer-Marschallinger, J., Berer, T., Roitner, H., Grn, H., Reitinger, B. & Burgholzer, P.
(2011). Ultrasonic attenuation of biomaterials for compensation in photoacoustic
imaging, Proc. SPIE 7899, 789931
Bell, A. G. (1880). On the production and reproduction of sound by light: the photophone,
American Journal of Science 20, 305324
Berer, T., Hochreiner, A., Zamiri, S. and Burgholzer, P. (2010) Remote photoacoustic imaging
on solid material using a two-wave mixing interferometer, Opt. Lett. 35, 4151-4153
Burgholzer, P., Hofer, C., Paltauf, G., Haltmeier, M. & Scherzer, O. (2005). Thermoacoustic
tomography with integrating area and line detectors, IEEE Trans. Ultrason.
Ferroelectr. Freq. Control 52, 15771583
Burgholzer, P., Grn, H., Haltmeier, M., Nuster, R. & Paltauf, G. (2007). Compensation of
acoustic attenuation for high resolution photoacoustic imaging with line detectors,
Proc. of SPIE Vol. 6437 643724-1
Burgholzer, P., Roitner, H., Bauer-Marschallinger, J. & Paltauf, G. (2010a). Image
Reconstruction in Photoacoustic Tomography using Integrating Detectors
accounting for Frequency-Dependent Attenuation, Proc. of SPIE Vol. 7564 756423-1
Burgholzer, P., Berer, T., Gruen, H., Roitner, H., Bauer-Marschallinger, J., Nuster, R. &
Paltauf, G. (2010b). Photoacoustic Tomography using Integrating Line Detectors
Invited, Journal of Physics: Conference Series, 214(1).
Callen, H. B. & Welton, T. A. (1951), Phys. Rev. 83, 34
Callen, H. B. & Greene, R. F. (1952). Phys. Rev. 86, 702
Chandrasekhar, S. (1943). Stochastic Problems in Physics and Astronomy, Reviews of Modern
Physics 15, 1-89
De Groot, S.R. & Mazur, P. (1984) Non-Equilibrium Thermodynamics, Dover Publications, Inc.,
New York
Gillespie, D. T. (1992). Markov Processes, Academic Press, New York
Greene, R. F. & Callen, H. B. (1952). Phys. Rev. 88, 1387
Grn, H., Berer, T., Nuster, R., Paltauf, G. & Burgholzer, P. (2010). Three-dimensional
photoacoustic imaging using fiber-based line detectors, Journal of Biomedical Optics,
15(2), 021306-1 - 021306-8
Haltmeier, M., Scherzer, O., Burgholzer, P. & Paltauf, G. (2004). Thermoacoustic computed
tomography with large planar receivers, InverseProbl. 20, 16631673
Hansen, P. C. (1987). Rank-deficient and discrete ill-posed problems: Numerical aspects of linear
inversion, SIAM, Philadelphia
Honerkamp, J. (1998). Statistical Physics, Springer-Verlag
Johnson, J. B. (1928). Thermal Agitation of Electricity in Conductors, Phy. Rev. 32, 97
Kiesel, S., Peters, K., Hassan, T. & Kowalsky, M. (2007). Behavior of intrinsic polymer optical
fibre sensor for large-strain, Meas. Sci.Technol. 18, 31443154
Landau, L. D. & Lifshitz, E. M. (1980). Statistical physics, Part 1, Pergamon Press, Oxford
Langevin, P. (1908). Sur la thorie du mouvement brownien, Compets rendus Acadmie des
Sciences (Paris) 146, 530-533. Translation by Anthony Gythiel, published in
American Journal of Physics 65, 1079-1081 (1997)


212
La Rivire, P. J., Zhang, J. & Anastasio, M. A. (2006). Image reconstruction in optoacoustic
tomography for dispersive acoustic media, Optics Letters. 31(6), 781-783
Lemons, D. S. (2002). An Introduction to Stochatic Processes in Physics, The Johns Hopkins
University Press
Nachman, A. I., Smith III, J. F. & Waag, R. C. (1990). An equation for acoustic propagation in
inhomogeneous media with relaxation losses, J.Acoust.Soc.Am. 88, 1584-1595
Nuster, R., Gratt, S., Passler, K., Grn, H., Berer, T., Burgholzer, P. & Paltauf, G. (2009).
Comparison of optical and piezoelectric integration line detectors, in Biomedical
Optics: Photons Plus Ultrasound: Imagingand Sensing 2009, edited by A. A. Oraevsky
and L. H. Wang, Proc. SPIE 7177, 71770T
Nyquist, H. (1928). Phys. Rev. 32, 110
Paltauf, G., Nuster, R., Haltmeier, M. & Burgholzer, P. (2006). Photoacoustic tomography using
a Mach-Zehnder interferometer as acoustic line detector, Appl. Opt. 46, 33523358
Paltauf, G., Nuster, R., Passler, K., Haltmeier, M. & Burgholzer, P. (2008). Optimizing Image
Resolution in Three-Dimensional Photoacoustic Tomography With Line Detectors,
in Biomedical Optics: Photons Plus Ultrasound: Imaging and Sensing 2008, Proc SPIE,
6856
Roitner, H. & Burgholzer, P. (2011). Efficient modeling and compensation of ultrasound
attenuation losses in photoacoustic imaging, Inverse Problems 27, 015003
Roitner, H., Bauer-Marschallinger, J., Berer, T. & Burgholzer, P. (2011). Experimental
Evaluation of ultrasound attenuation losses in photoacoustic imaging, submitted
to J.Acoust.Soc.Am.
Royer, D. & Dieulesaint, E. (2000). Elastic Waves in Solids I, Springer
Shutilov, V. A. (1988). Fundamental Physics of Ultrasound, Gordon and Breach Science
Publishers
Stokes, G. G. (1845). On the theories of the internal friction of fluids in motion, and of the
equilibrium and motion of elastic solids, Trans. Cambridge Philos. Soc. 8, 287-319
Szabo, T. L. (1994). Time domain wave equations for lossy media obeying a frequency
power law, J.Acoust. Soc. Am. 96, 491500
Treeby, B. E. & Cox, B. T. (2010). Modeling power law absorption and dispersion for acoustic
propagation using the fractional Laplacian, J.Acoust. Soc. Am. 127 , 2741-2748
Treeby, B.E. & Cox, B. T. (2010b). k-wave: Matlab toolbox for the simulation and reconstruction
of photoacoustic wave fields, J. Biomed. Optics 15(2), 021314-1 - 021314-12
Treeby, B.E., Zhang, E. Z. & Cox, B. T. (2010c). Photoacoustic tomography in absorbing
acoustic media using time reversal, Inverse Problems 26, 115003
Trefethen, L. M. (2000). Spectral Methods in Matlab, SIAM, Philadelphia
Uhlenbeck, G. E. & Ornstein, L. S. (1930). On the Theory of Brownian Motion, Phys. Rev. 36,
823-41
Wang, L. V. (2008). Prospects of photoacoustic tomography, Medical Physics 35, 57585767
Widrow, B. & Kollar, I. (2008). Quantization noise, Cambridge Books, Cambridge
Xu, M. & Wang, L. V. (2006). Photoacoustic imaging in biomedicine, Review of Scientific
Instruments. 77(4), 041101-041122
Zhang, E. Z., Laufer, J. G., Pedley, R. B. & Beard, P. C. (2009). In vivo high-resolution 3D
photoacoustic imaging of superficial vascular anatomy, Phys. Med. Biol. 54, 10351046
10
Low Frequency Acoustic Devices for
Viscoelastic Complex Media Characterization
Georges Nassar
Universit de Lille Nord de France
France
1. Introduction
The evolution in consumer expectations in terms of quality and safety in the agro-industry
has led to the need to develop new methods of investigating product quality and the
processes involved. Many fields of production still rely too much on the know-how of the
operators who, with their experience acquired over time, have become key players in the
company. In addition, the manufacturing of quality food products frequently relies on
artisanal know-how that is difficult to industrialise and often synonymous of high
production losses, therefore prohibitive costs. In contrast, so as to limit costs, the industrial
production process is often associated with poorer quality. The objective evaluation of
product quality involves the development of methods and sensors adapted to the product or
the manufacturing process.
Indeed, beneath an apparent simplicity, agro-industry products have complex physical
properties linked to elasticity, viscosity and plasticity. One of the major difficulties lies in the
complexity of the processes which depend on numerous physical parameters. The matter is
subjected to numerous mechanical, thermal or chemical treatments thus migrating towards
viscoelastic or even plastic properties that are more difficult to quantify.
The originality of the approach adopted consists in the study and set up of an ultrasonic
measuring device associated with its electronic environment in order to reply to a specific
need due to the complexity of the physico-chemical phenomena involved. A global
approach to this problem is very tricky as the physical properties of the media evolve
significantly throughout processing.
We thus focused on the development of sensors and methods of characterisation dedicated
to different phases of the industrial processes. Two very closely linked aspects were
therefore studied targeting product characterisation and process control.
Work has been carried out to develop acoustic and ultrasonic instrumentation designed to
monitor the change in state of the matter (liquid-gel transition and product cohesion), then
to monitor the evolution of its elastic properties. The process control applications concern
the development of a very low frequency, non-destructive monitoring method to reply to
the specificities of the physical properties of the matter.
In this document, we report the scientific approach highlighting the design of the ultrasonic
sources which dispenses with classic design through the choice of specific resonance modes
for the sensors. Their design aims at promoting low frequency resonance in a relatively
small scale composite structure. This sensor technology was adapted according to the


214
frequency chosen to study the change in physical state of the media and to monitor the
evolution of the acoustic properties of products that are often heterogeneous.
Several approaches were used to optimise this technology: an analytical approach to
determine the sensor's first vibratory mode which was consolidated by a numerical study,
then confirmed and validated experimentally.
The application is based on two points:
The physical or physico-chemical phenomenon linked to the transition phase and the
sol-gel transition in the dairy field; the opaqueness and the fragility of this type of gel
justifies the importance of quantifying the metrological parameters such as
measurement accuracy, stability over time and mechanics of the sensor implemented.
The interaction of the sensor with its environment in a process. A bivariate study
(sensor/propagation medium) was carried out in order to select the required geometry
to ensure that the sensor is adapted to its environment: loaded metallic plates subjected
to mechanical stress and heated to a temperature of around 100C (plate heat
exchanger) or strongly absorbent media (fermenting bread dough...). This part of the
study consisted in finding a good compromise between the geometry of the sensor, its
location in the overall system and the required sensitivity.
2. Context
Many biochemical industrial activities involve very complex physicochemical phenomena
which enable products to be processed. These products often go through a variety of states
during processing.
Processing uses energy from chemical reactions (e.g. enzymatic), thermal or mechanical
energy, or even a combination of these different forms of energy. Physical modifications can
also occur (incorporation of air in the matter). All these forms of energy are often
combined within the same process and it is difficult to quantify the contribution of each in
the product processing phenomena.
Due to the complexity of the processes (several processing stages, multiplicity of forms of
energy) and the products (viscoelastic matter, visco-elasto-plastic), associated with the
legitimate concern of not interfering with the process, few measurements have been carried
out during the various stages of processing. The temperature, pressure and flow are often
monitored during processing even though they are not necessarily correlated to the desired
properties of the product under development.
This is why we chose to develop acoustic sensors adapted to the constraints imposed by
either the product or the process.
Among the essential parameters sought-after, rheological measurements are often
determining in terms of the consumers perception of the qualities of the end product such
as texture, viscosity, elasticity.... Several techniques of investigation exist but the majority
are laboratory applications and are difficult to adapt for in-line controls. The difficulty thus
arises of using multiple techniques to obtain a more complete characterization of the process
and the interpretation of the data obtained with these analysis techniques.
Furthermore, the quality control of the processed products also involves evaluating the
performance of the process. The temperature, pressure and flow are of course part of the
characteristics measured for the control and/or closed-loop control of the various stages of
product processing. However, in the case of complex processes they are difficult to correlate
to the final properties of the product.

Low Frequency Acoustic Devices for Viscoelastic Complex Media Characterization

215
The processes can also evolve over time. This is the case with heat exchangers for which the
performance varies over time due to fouling. Only preventive maintenance leading to
additional production costs can ensure stable performances of the process over time. The
development of sensors integrated in the process to provide information on the evolution of
the performance remains essential.
This work presents a selection of studies which have led to the development of low
frequency acoustic sensors specifically adapted to monitor changes in the physical state of
complex media and the process: fragile gel, highly heterogeneous or highly absorbent
media, media with complex rheological behaviour...
Several cases were studied:
A low frequency acoustic sensor adapted to the characterization of complex products
using an omni-directional source in the case of media undergoing a change in physical
state;
continuous homogeneous medium: sol-gel transition,
complex heterogeneous medium: transition from a suspension of particles in a
liquid to a cohesive visco-elasto-plastic solid.
A very low frequency acoustic sensor used to monitor the response of a medium
subjected to mechanical vibrations. Such technology is designed to study the processing
phenomena of a highly absorbent product such as bread dough during fermentation.
Finally, the identification of the needs and constraints imposed by certain environments
(temperature, hygiene, attenuation...) have led to the combination of these types of
technology to monitor a process (e.g. fouling of plate heat exchangers, search for an
optimum point in the kneading phase...). By taking into account the coupling of the sensor
with its environment this technique can, in certain cases, exploit the noise emitted by the
process itself, as in kneading for example.
In this work, we chose to illustrate the potential of low frequency acoustic methods on
applications from the agri-foodstuffs sector. These same states can also be found in the
pharmaceutical and cosmetics industries, the aviation industry, the medical field as well as
in material chemistry.
The methodology implemented can be divided into several phases:
Analysis of the product and/or process
Definition and/or optimisation of the appropriate method and a sensor meeting the
various constraints :
Analytical study
Numerical modelling
Experimental validation of the sensor
Validation of the application
3. Sensors suitable for studying media of which the physical properties
evolve over time
3.1 Monitoring changes in the physical state of the matter
The analysis of the different stages in the formation of macromolecular networks is of major
importance, since understanding the structure and properties (physical or chemical) of gels
requires the understanding of the process of organization. In many physical, chemical or
biological processes, the union of small separate elements to form aggregates of different
sizes and further macroscopic phases makes connectivity an essential characteristic of this


216
type of process. Many models have been proposed to explain the phenomenon of
aggregation. The most important ones are those of Flory (Flory, 1953), Stockmayer
(Stockmayer, 1943), Case (Case, 1960), Gupta (Gupta et al., 1979), Eichinger (Eichinger,
1981), Allsopp (Allsopp, 1981) and San Biagio (San Biagio et al., 1990). In most cases the
phenomenon is described by the classical theory as a particular case of percolation and the
two-dimensional growth of the network according to Caylays tree. Other studies including
those of De Gennes (De Gennes, 1989) and Stauffer (Stauffer, 1981, 1985) describe the
phenomenon of random aggregation and the problems of percolation and gelation.
However, the different characteristics of the macromolecular chain-making system can be
evaluated according to Clerc (Clerc et al., 1983), using for example a Monte-Carlo
simulation, predicting the influence of the characteristics of the starting solution and the
gelation conditions on the structure and the arrangement of the masses.
In fact, the gelation process is a transition from an entirely soluble system to a
heterogeneous two-phase system: composed of an insoluble entity (infinite-size
macromolecule) and a soluble phase. This transition is accompanied by radical changes in
some physical properties of the medium. Below the gelation point, the viscosity of the
medium increases and the medium ceases to flow by developing an elasticity.
To study this phenomenon, several physical measurement techniques exist i.e. optical,
thermal, rheological and acoustic (Nassar, 1997). However, sampling and sensitivity to a
limited range of physical properties are often drawbacks. Consequently, different
techniques are required to explore an entire process with the difficulty of bringing together
the heterogeneous data provided by these techniques. This is, for example, the case of
optical methods which are penalized by the opacity of the substances analyzed as well as
the size of the molecules formed in relation to the wavelength. Thermal methods are
insensitive to the mechanical characteristics of the medium. The fragility of some gels (milk
gel) limits rheological techniques. In many cases, several analytical techniques exist, but
they are only used in the laboratory.

Fig. 1. Basic principle
To develop further instrumentation in order to understand and to quantify the modification
process of media in real conditions, a low-frequency ultrasonic technique using sensors with


217
extremely pointed ends that act as point sources was examined. The application of this
technology using two near-field coupled sensors to explore the relationship between the
physical properties measured during the evolution of the time of flight of the wave and the
structural changes during matter formation (Figure 1) was investigated.
The sensors were near-field coupled through the medium to be characterised. Such
disposition privileges the Signal/Noise ratio and avoids the loss of acoustic pressure which
is inversely proportional to the ray of the spherical wave. In a metal or ceramic solid all the
waves are generated simultaneously, but in the media we are concerned with, the
dominating longitudinal wave is the fastest and is relatively simple to exploit.
The advantage of these sensors is that they can be adapted to the measurement
configuration envisaged according to the nature of the wave and the appropriate resonance
mode.
3.2 Study of a low-frequency ultrasonic device
The aim of the study was to define and develop optimal ultrasonic instrumentation to
understand the phenomenon and quantify the viscoelastic properties of changing media.
The usual ultrasonic characterization techniques are generally based on the use of a resonant
piezoelectric transducer in thickness mode. As the resonant frequency of a transducer is
inversely proportional to its size, it is greater for low frequencies around 100 KHz. Some
researchers like Degertekin (Degertekin & Khury-Yakub, 1996a, 1996b, 1996c), Shuyu
(Shuyu, 1996, 1997) and Nikolovski (Nikolovski & Royer, 1997) used this physical principle,
but associated a tapered volume with the ceramic components to concentrate the mechanical
energy.
The aim of this part of the work was to obtain a low frequency acoustic point source to
generate a spherical wave in the medium. To do this a different procedure from that
traditionally used in classic sensor design was implemented. A new technique was used
which consisted in setting in resonance the entire mechanical structure of a reduced-size
unit through the contact of an extremely pointed end with the material to be analyzed. In
order to behave like an acoustic point source, the size of the point was smaller than the
wavelength in the medium.
The first part presents a theoretical analysis of low-frequency ultrasonic resonators, beaming
a spherical wave in the medium. The choice of a triangular shaped resonator and its
mechanical behavior will be assessed and the study completed by a numerical approach
based on the application of the finite elements method to characterize all the resonator
vibration modes and visualize the corresponding distortions when the structure is excited.
As the analytical results were in good agreement with the numerical results, they were
applied to the whole triangular-shaped sensor to validate the findings experimentally. The
resonance mode frequencies determined by the numerical calculation were then correlated
with the electrical impedance measurements.
3.3 Study and design
3.3.1 Analytical approach
For a possible analytical analysis, the structure of a standard ultrasonic sensor is based on a
simple triangle shape (Figure 2).
The propagation of longitudinal waves in the triangular part of the sensor was studied to
determine the resonance frequency of the elongation mode and the velocity amplification


218
ratio between the ends. The analysis is based on an extension of Ensmingers (Ensminger,
1960) theory.
According to figure 2, the x section is written:
( ) ( )
1 1 1
S e x ThenS e x x /x = = + (1)

l
2
l
1

l
x
1
L
e
x
0

Fig. 2. Basic analytical shape
Ensminger studied the propagation of a wave in extensional mode in a cone with no loss of
which the lateral dimensions were short in comparison with the length. In the case of a
triangular shape, this equation takes the following form:

2 2
2 2
1
1
0
( )
v v
v
x x x x c

+ + =
+
(2)
Where:
v is the velocity of the particles, is the pulsation and c is the velocity of the longitudinal
wave in the material making up the vibratory element.
On the basis of the dimensions given in figure 3, the solution to this differential equation
leads to an approximate velocity amplification ratio between the two extremities (Nassar,
1997); |v(0) /v(L)| = 1/ 0.46 = 2.16 for a resonance frequency: f = 60 KHz.

10 mm
32 mm
l2
l
V
1

Fig. 3. Basic shape. Triangular sensor of thickness e = 1 mm, L = 32 mm,
1
= 2 mm and
2
= 16 mm


219
3.3.2 Numerical approach
While an analytical study can only take into consideration one particular mode of vibration
of the triangular part of the sensor, a numerical study based on the finite elements method
can determine all the vibrating modes of these parts as well as those of the realised sensor.
For a real structure; whole sensor included a binding rod and a triangular truncated part
(Figure 4), the displacement differential equations were solved with a continuous regime,
taking into account the boundary conditions at the surfaces. The materials were defined by
Youngs modulus E, Poisson's coefficient and density . The results presented below were
applied without loss and they were compared to the characteristics of the longitudinal mode
determined by the analytical calculations. This comparison was also made for the triangular
sensor which was studied as a whole.
For our study, the ANSYS analysis software was used. The sensors used were made
essentially of piezoelectric material. A source of excitation was engraved in the general
structure of the vibrating element (triangle part), providing mechanical continuity without
any break (Figure 4a). This type of engraving was considered as it has been demonstrated
(Nassar, 1997) that for the same longitudinal mode, the amplification ratio at the ends is 71
times bigger when there is one engraved source providing mechanical continuity with the
vibrating element so that one source can impact the structure by gluing (Figure 3b).

Fig. 4. From left to right: (a) engraved source; (b) embedded source; (c) elongation mode
Table 1 show the resonance frequencies and the vibration velocity transformation ratio
(|v(0) /v(L)|) and table 2 present the difference rate of this ratio according to the position of
the excitation source

Longitudinal resonance
frequency Hz
Velocity ratio v(0)/v(L)
at the extremities
Analytical calculation 60132 2.16
Numerical calculation 57611 2.32
Electrical impedance measurements 59 kHz
Table 1. Comparative study. Analytical and numerical approach

sensors Embedded source Engraved source
Frequency (Hz) 57729 57611
Standardised velocity ratio 1.4 10
-2
1.00
Table 2. Impact of the nature and the location of the source


220
The results show a good correlation between the frequencies determined by the calculations
and those determined numerically or using impedance measurements.
A significant increase in the amplitude of vibration was observed resulting from the design
of the electrode on an active element.
3.4 Application for monitoring changes in state
3.4.1 Pointed sensors for sol-gel transition
The milk gelation can be considered as an aggregation of different sized molecules
(Walstra& Vliet, 1986; Fox, 1989; Dalgleish, 1993). This model was explored for several
reasons: the available knowledge, the experimental conditions that are known and relatively
easy to conduct, the complex medium with the physical properties of liquid and gel states in
close contact.
As the reaction progresses, the average mass of each aggregate increases and the number of
molecules in the medium tested decreases. The aggregation process results in a giant
macromolecule defining the gel.
This process was examined using two identical ultrasonic sensors near-field coupled
through the medium to be characterized. The working frequency was 60 kHz.
Figure 5 presents a schematic diagram of the measuring device. The emitter is driven by a
sharp electrical pulse lasting 15 s. These conditions provide a longitudinal vibration mode
at the end of the sensor which behaves like a point source. This phenomenon generates a
divergent ultrasonic wave in the medium, one part of which was measured using a receiver
located at a constant distance from the transmitter by the first the zero-crossing of the wave.
The propagation of the wave in the medium is more or less a compressional wave, as
suggested by the time of flight corresponding to a velocity of 1600 m/s in reconstituted milk
samples at 25 C.

Sharp electrical
pulse
Transmiter
Ta
Receiver
Water flux
Electric signal
Calculator
Electrical and mechanical protection
Sensor

Fig. 5. Diagram of the measuring cell. Tus is the temperature of the product at sensor
level, Tcw is the temperature of the container walls and Ta is the ambient temperature


221
3.4.2 Ultrasonic monitoring of gelation: measurement of the variation in the time-of-
flight of the wave
3.4.2.1 Measurement stability
The time of flight dt(ns) of the signal, measured in distilled temperature-controlled water at
300.1C (reference medium) remains stable. The precision obtained was 1ns over a global
reply time of 10s, given a relative precision of 10
-4
per measurement at the zero-crossing
point
3.4.2.2 Measurement reliability in gelation process
The reliability of a measurement system resides in its reproducibility and its faculty to
follow all the stages of the gelation process. In the standard conditions using 12 grams of
skimmed milk powder dissolved in 100 ml of distilled water and in accordance with the
literature (Nol & al., 1989), the sol-gel transition clotted between 15 and 16 min. At a
regulated ambient temperature at 30 C similar to the one of the medium under test, as
shown in figure 6, the time of flight of the signal decreased indicating an increase in the
mechanical resistance of the product. This variation had not reached a plateau value,
indicating that the medium was still changing.

0
5
10
15
20
25
30
35
40
0 20 40 60 80 100 120
Time (min)
V
a
r
i
a
t
i
o
n

o
f

t
i
m
e

o
f

f
l
i
g
h
t

d
t

(
n
s
)
Estimation of clotting time
Dispersion of the measurements

Fig. 6. Typical curves of the gelation process of two media prepared in the same conditions
at a reaction temperature of 30.1 C observed using the ultrasonic technique. The difference
between the two curves provides a quantitative estimation of the global dispersion of the
ultrasonic measurement due to the electronics and sample preparation
Figure 6 also provides a qualitative estimation of the reproducibility of the ultrasonic
measurements. The curves show the progress of the action of the rennet in two media
prepared in the same conditions. The maximum dispersion of the measurements was 5 ns
due to the electronic parts and the milk reconstitution process.


222
3.4.2.3 Evolution of the molecular network
Figure 7 presents the variations in the time of flight resulting from the variation of the
ultrasonic wave velocity during the milk gelation process at different ambient temperatures.

0
10
20
30
40
50
60
0 20 40 60 80 100 120 140
Time (min)
V
a
r
i
a
t
i
o
n

o
f

t
i
m
e

o
f

f
l
i
g
h
t

d
t

(
n
s
)
Ta = 20 C
Ta = 17 C
Ta = 23 C
Ta = 30 C

Fig. 7. Ultrasonic sensor responses during milk gelation at 30.1 C, but at different ambient
temperatures (from 17 C to 30 C)
The curve observed at an ambient temperature of 30 C, similar to the temperature of the
milk, was similar to those obtained by measurements using other physical methods
(McMahon & Brown, 1984). When the ambient temperature was significantly different from
the product temperature, ultrasonic curves showed a specific pattern.
According to Dalgleish (Dalgleish, 1982), gelation process was defined like a transition
phenomenon in which a soluble suspension made of macromolecules (liquid phase)
becomes insoluble when a giant mass forms. We can assume that basic macromolecules are
synthesized by linking monomers (building units) via covalent bonds. This chemical
reaction is due to the presence of functional groups of the monomers that are able to form
chemical bonds with other functional groups of the monomers (Mercier & Marechal, 1993).
The network formation occurs if the functionality of the units is greater than two (Mercier &
Marechal, 1993). As the reaction progresses, the conversion status of the system is
characterized by the connectivity rate p (Flory, 1953; Stockmayer, 1943). Below the gelation
threshold, the viscosity of the medium increases as the connectivity rate p approaches the
critical advancement rate pc. This phenomenon is known as a critical connectivity transition.
Above the threshold, the medium ceases to flow and the gel phase develops some elasticity.
This phenomenon introduces structural changes in the physical properties and more
particularly in the mechanical behavior of the medium, thus resulting in the transition from
a liquid state to a viscoelastic solid state.
To illustrate this process schematically, let us consider an initial solution containing units
that can link together. At the beginning, the medium behaves like a viscous solution in a sol
phase due to the presence of a single type of finite-size masses: in this case, p is low (Figure


223
8b). When p increases, bigger and bigger masses are formed (Figure 8c). For a certain critical
value of p, pc, a giant chain appears (continuous connectivity of the space from one side to
the other A B ; Figure 8d) defining the gel point. Above the threshold pc, the medium in
the gel phase has the macroscopic behavior of a viscoelastic solid. For p = 1 all the units
belong to the giant mass (Figure 8e).

p < pc
p pc
A
B
p = pc AB
(c)
(d)
p > pc
p 1
p = 1
(e)
p << pc Initial product
Rennet product
(chymosin)
Additive
(a) (b)

Fig. 8. Schematic representation of the sol-gel transition. a) Initial phase, b) Suspension of
molecules of finite sizes, c) Agglomeration and formation of macromolecules of large
masses, d) Critical connection phase: p = pc, e) Network continuity connection to give a
single giant macromolecule, the gel
As the properties of a gelling medium are proportional to the reaction progress, it is possible
to represent the behavior of the viscoelasticity in terms of the connectivity rate p according
to the following cases (Figure 8):
For p < pc, the system is a liquid whose viscosity increases as the gel point approaches.
For p = pc, elastic behavior appears.
For p > pc, the medium becomes a solid gel whose elasticity increases with p.
In order to relate the theoretical aspect to the experimental results, the following curve
(Figure 9) has been divided into five different stages. The phenomenon describes the
gelation process when the ambient temperature was different from the product
temperature. It was undetectable when these temperatures were the same.
According to figure 9, stage (a) could be interpreted as a proteolytic phase characterized
by the appearance of two polypeptides resulting from the effect of the rennet product. This
stage is followed by the formation of aggregates of finite size (Figure 8b). This molecular
reorganization might be related to the change of slope of the curve (stage b) reflecting a
decrease in the time of flight, which means an increase in the velocity in the sol medium.


224
28.2
28.4
28.6
28.8
29
29.2
29.4
29.6
29.8
30
30.2
0
5
10
15
20
25
30
0 10 20 30 40 50 60 70
M
e
d
i
u
m

t
e
m
p
e
r
a
t
u
r
e

v
a
r
i
a
t
i
o
n

(
C
)

a
t

t
h
e

s
e
n
s
o
r
s

l
e
v
e
l
V
a
r
i
a
t
i
o
n

o
f

t
i
m
e

o
f

f
l
i
g
h
t

d
t
(
n
s
)
Time (min)
Ultrasonic measurement Thermal measurement

Fig. 9. The specific ultrasonic response during milk gelation when the ambient temperature
was different from the temperature of the milk
The temperature inside the medium remained constant during this stage and resulted from
the propagation of heat by pure and free convection. The formation of more or less
voluminous masses in the medium (stage (c)) induced the transition from a viscous state to a
viscoelastic state, slowing down the free convection. This led to a slight temperature
decrease in the medium at the level of the sensors (T
US
in Fig. 4) to reach a new equilibrium
where heat was mainly transmitted through the container walls (regulated temperature:
T
cw
C), by conduction. The changes in the medium during this stage could be interpreted in
the following manner:
1. Due to a thermal conduction phenomenon, a slight temperature gradient appears in the
medium, between the container walls (T
cw
= 30.1C) and the center of the vat, at the
location of the measuring point; T
US
C (following on the ambient temperature). The
temperature decrease induced an increase of the time of flight.
2. The time of flight decreased as the reaction progressed. This decrease can be attributed
to the development of an elastic modulus resulting from the formation of
macromolecules, changing the medium from a viscous liquid state to a viscoelastic solid
state. The phenomenon was expressed physically by the evolution of the connectivity
rate p towards its critical value pc .
These two phenomena make stage (c) a competition between :
1. An increase of the time of flight resulting from a decrease in temperature.
2. A decrease in the time of flight resulting from the appearance of an elastic component
in the changing medium.
The connectivity rate p, reached a critical value pc in stage (d), at the maximum of the
curve when the existence can be assumed of a giant macromolecular chain linking the two
extreme sides of the considered space (Figure 8d). During this stage, the mechanical aspect
S
t
a
g
e

(
e
)

S
t
a
g
e

(
a
)

S
t
a
g
e

(
b
)

S
t
a
g
e

(
c
)

S
t
a
g
e

(
d
)


225
of the medium could dominate the remaining part of the reaction, due to the weak thermal
variation ( 2 C) resulting from the difference in temperature between the medium and the
environment (ambient temperature).
During stage (e), the gel strengthened. The gel was stronger when the ambient temperature
was very close to the temperature medium (Figure 7). This phenomenon can be shown
experimentally by an increase in dt(ns) resulting from a decrease in time of flight, whereas
theoretically it was explained by the evolution of the connectivity rate p towards 1 following
the establishment of continuous connections of finite size masses on the giant molecule, the
gel.
3.4.3 Monitoring the formation dynamics of the cohesion forces in a fractionated
medium: measurement of the variation in the wave amplitude
A key step often met in agro-industry processes is the formation of the matrix of the final
product. In cheese-making, this phase involves the cohesion of the elements making up the
medium. Generally, it is the conversion of the matter from a heterogeneous state (made up
of overlapping grains) to a homogenous state. In this particular case, the cohesion of the
curd grains, essential step in the process, varies according to the process conditions as well
as the enzymatic and bacterial activities in the medium. It is thus necessary to take these into
account in the description of the cohesion.
Analysis of the medium during the cohesion process
During draining, very different physical states are involved in the conversion of the curd
grains from a heterogeneous medium to a more homogenous medium. It is therefore
difficult to describe the interaction between the ultrasonic wave and the curd grains during
draining using just one physical model. So, for a better evaluation of the different phases in
the processing of the medium:
1. From moulding and for a very short period of time the grains are touching and cohesive
links begin to form between the contact surfaces, thus forming a skeleton containing
connected porosities through which the whey continues to drain (Figure 10a).
2. When the whey evacuation channels become blocked (Figure 10b), this phase is
described by the multilayer model by Brekovskikh (Brekovskikh, 1980). It is equivalent
to a material made up of layers of grains and whey of which the thickness is equivalent
to the size of the grains as well as whey evacuation channels. The layers of whey
become thinner and thinner until they disappear (Figure 10c), producing a homogenous
medium (final phase). This approximation is valid insofar as the main signal beam is
confined to a narrow area of the medium, which is the case in our measurements.
Indeed, the zone of interest is comparable to the size of the grains (Figure 11).
The evolution of the medium throughout the entire draining phase in the mould was
described using these two models: the outflow of the whey and the cohesion of the grains.

a b c

Fig. 10. Evolution of the medium over time


226

Fig. 11. Estimation of the propagation zone of the main signal beam
Heated to a constant temperature of 35 C, the experimental mould was instrumented with a
transmitter and two ultrasonic receivers spread out so as to integrate the signals transmitted
through two paths presenting enough interfaces between the grains. This thus reduced
measurement dispersion linked to the random number of interfaces (Figure 12).

Transmitter

Receiver1 Receiver2
Controlled pressure
Scales
Thermometer
Whey ejected

Fig. 12. On the left: The experimental measuring device with a horizontal cross-section of the
mould at sensor level and on the right: omni-directional, optimised sensor in elongation
mode at 246 kHz
Figure 13 shows that the amplitude of the ultrasonic signal is a parameter that is sensitive to
variations in the properties of the medium under investigation.

Uncontrolled rapid
evacuation of the whey
Constant ratio between
deformation and stress
(incompressible
medium)
Cohesive
Time

Fig. 13. Typical curve reflecting the cohesion phenomenon as seen by the variation in the
ultrasonic amplitude


227
3.5 Case of highly absorbent matter
The characterisation of media using ultrasounds is often limited by the heterogeneous
nature of the matrix which can, in the case of cosmetic, pharmaceutical and agro-food
products, be viscoelastic and heterogeneous (foam or emulsion for example). Wave
attenuation in such media is mainly due to viscous absorption and scattering from
heterogeneities. The higher the frequency the greater the attenuation, hence the necessity to
analyse the media using low frequencies in order to characterise the evolving matter.
The search for a compromise between the analysis frequency and the volume of the medium
to be characterised led us to propose specific sensor geometries associated with specific
excitation conditions.
In order to manage this constraint, a very low frequency acoustic technique was adapted so
as to communicate sufficient energy to a particularly absorbent sample. This was achieved
by mechanical excitation caused by a shock. An electrical image of this excitation is obtained
using a second identical sensor used as the synchronisation reference (Figure 14).

Fig. 14. Schematic diagram
3.5.1 Principle of the sensor proposed
The sensor proposed is illustrated in figure 15. In the shape of a thin disc, its structure is
made up of a ring with an embedded piezoelectric disc.
A structure like this offers the advantage of being able to work at resonances lower than
those of the piezoelectric disc and thus several resonance modes can be used of which the
main ones are flexion and radial modes. This type of sensor also offers a large area of contact
with the medium studied, which, in the case of soft and aerated materials, can be
advantageous.
The resonance modes of a circular structure, notably those of a disc or a thin ring, have been
studied for many years by several authors (Aggarwal, 1952a, 1952b; Moseley, 1960; Vogel &
Skinner, 1965; Leissa, 1969; Blevins, 1979; Irie & al., 1984; Lee & Singh, 1994) . The main
modes of resonance of a disc are radial modes in the disc plane and flexion modes outside
the disc plane (Tables 3, 4 & 5). The tables show a good correlation between the theoretical,
numerical and experimental analyses.


228

Fig. 15. Set-up of the sensor proposed

Flexion
modes
(s, n)
Theoretical
(Hz)
FEM (Hz)
(1.0) 1798 1794
Table 3. Illustration of the deformations. Amplitude perpendicular to the disc plane Flexion
mode (1.0) for a frequency of 1.8 kHz (theoretical and using finite elements)

Radial modes Analytical calculation (kHz) FEM (kHz)
35.5 35.5
92.6 92.7
Table 4. The resonance frequencies of the first two radial modes for a free aluminium disc
(R=5cm; h=2mm) obtained analytically and numerically using finite elements


229
Radial modes
Frequency (kHz)
1 35.9
2 90
Flexion modes
(1.0) 1.6
(2.0) 6.5
Table 5. Resonance frequencies of the first radial and flexion modes for the composite sensor
3.5.2 Application for monitoring fermenting bread dough
The objective of this application was to establish the links between the product evolution
kinetics and the acoustic characteristics measured.
From a practical point of view, impulse excitation was used in this system. The excitation
was obtained by a controlled mechanical impact (rod of an electromagnet), thus exciting the
disc used for the synchronisation. The vibration induced in the dough is received by a
receiver disc identical to that of the synchronisation disc (Figure 16).

Fig. 16. Experimental measuring device
A metrological study of the measuring device carried out using standard samples (for
example a pocket of water at 25C) showed that the standard deviation of the amplitude and
the velocity was approximately 2%. Signal acquisition was carried out over 3 hrs.
3.5.3 Dynamic monitoring of the fermentation process of bread dough
After controlled kneading of the dough, the measurement chamber was placed in an
enclosure in order to control the temperature and humidity. The acoustic values studied
were the variation of the time-of-flight and the wave amplitude on reception.


230
Figure 17 shows the variations in these two values. It can be noted that the critical points
and phases appear simultaneously on the two curves.

Fig. 17. Evolution of the standardised amplitude and the relative signal delay on reception
during the fermentation phase
Where:

r
is the time necessary to reach a relatively stable zone,
T
r
reflects the period of stability during which the relative delay reaches its maximum
and remains relatively constant,
t
M
is the maximum relative delay. It is linked to the gas fraction contained in the
dough and therefore the extensibility of the latter.

a
is the period during which the amplitude of the signal decreases before reaching a
plateau,
T
a
is the period of stability of the amplitude,
A
S
is defined as being the amplitude of the signal during the period of stability.


231
A repeatability study was carried out to estimate the dispersion of the parameters (delay
and amplitude). Several tests were performed under the same operating conditions. The
standard deviation of the measurements of these parameters was around 3%.
Table 6 summarises the variations in the characteristic parameters observed on the curves
according to the evolution in the temperature

20 27 34
r
(min) 165 105 60
T
r
(min) 145 70 55
t
M
(s) 380 385 374
a
(min) 160 95 55
T
a
(min) 130 75 50
A
S
(%) 40 43 44
Table 6. Parameters relating to the variation in temperature
It can be noted that the maximum relative delay is relatively constant (approximately 380s)
for the three products made under the same operating conditions. This parameter seems to
be independent of the temperature, which is in agreement with the hypothesis that it varies
according to the gas fraction contained in the matter and the elastic properties of the matrix.
4. Acoustic sensor for in-line monitoring of a manufacturing process
In certain industrial processes it is often difficult to access useful information in real-time
due to the conditions imposed on the mechanical and thermal parameters, pressure,
hygiene..., conditions which require a specific installation of the sensor with regard to its
environment. The difficulty thus arises of an integration taking into account both the
process constraints and the acoustic constraints. This is the case of a plate heat exchanger
which can be considered as a typical example in this category (Figure 18).

Fig. 18. Standard plate heat exchanger
4.1 Sensor selection criteria
For the exchanger, the sensor selected is not cumbersome and is sensitive over a
temperature range reaching over 100C (Figure 19). The excitation and synchronisation


232
modes remain the same as the previous case (disc sensor). The principle of the measurement
is to excite a vibration mode in one or several plate exchangers and to analyse the evolution
under the effect of fouling by measuring the response of the plates using a receiver.
A bivariate system-sensor study enabled the geometry of the latter to be defined over the
same vibration frequency range as the system (exchanger).

Sensor
Exchanger plate
Electromagnet soliciting a
reference sensor
Receiver

Fig. 19. Positioning of the sensors on an exchanger plate
4.1.1 Sensor excitation mode
In order to monitor the evolution of the damping of the plate modes due to fouling of the
exchanger, it is necessary to excite these modes with enough energy to preserve the signal-
noise ratio (of the signal received) after going through the exchanger.
A mechanical shock is the only way of producing enough energy for local excitation.
The frequency response obtained by modal analysis in the absence of structural constraints
is given in the first column in table 7. This column gathers the different modes specific to the
structure studied. Some correspond to simple, longitudinal or transversal displacements,
others to more complex displacements (flexions, torsions...).

Mode Frequency (Hz) - numerical Frequency (Hz) - experimental
1 1683 1586
2 2387 2894
3 3557 3639
4 5422 5330
5 5734 6639
6 7417 7390
Table 7. The first 6 modes specific to the sensor


233
The second column shows the modal frequencies obtained from the analysis of the
impedance of the sensor mounted on a heat exchanger.
The mean standard deviation between the frequencies obtained by modal analysis and those
obtained experimentally is 5 %. The good correlation between these results indicates that the
numerical modelling provides a good estimation of the resonance frequencies of the sensor.
4.1.2 Excitation by mechanical shock: estimation of the frequency range
The mechanical excitation in question is ensured via the core of an electromagnet.
As an indication, figures 20a and 20b show the temporal and frequency responses of the
sensor.

Fig. 20a. Temporal response of a mechanical shock

Fig. 20b. Spectral response associated with the shock


234
The curves show the temporal response and the frequency range of the sensor following a
stress induced by a mechanical shock of short duration. The experiments carried out on the
overall system (sensor & exchanger) in real configuration show that the temporal response is
maximum 4 ms and its frequency response is around a central frequency of approximately 4
kHz.
4.2 Application
4.2.1 Fouling mechanism
Heat exchanger fouling is a dynamic process. The phenomenon continues to evolve,
generally until equilibrium is reached or cleaning is required. The period of fouling can vary
from a few hours to several months.
Mller (Mller-Steinhagen & Middis, 1989) looked at five stages in the process of the
appearance and development of particulate fouling:
The initiation, which corresponds to the time necessary before fouling, can be observed
on a clean surface. The duration depends on the nature of the deposit, the initial state of
the surface (material, roughness) and the temperature of the wall.
The denaturing of the product (protein, organic matter...) under the effect of heat and
the surrounding parameters (pH...), their aggregation and transport within the vicinity
of the wall.
The adhesion of the particles transported to the wall, controlled by surface adhesion
forces (Van der Waals, electrostatic...) and cohesion of the deposit. It has been shown
that the particles can adhere to a clean surface or adhere to other particles already
deposited.
The dislodging of deposited particles, caused by hydrodynamic forces which exert
shear stress on the deposit.
The aging of the deposit over time results in changes in its structure which can either
weaken or consolidate it.
Generally, the initiation phase is rarely taken into account in particulate fouling models. The
mechanisms that govern the deposit of particles are generally presented as being the
transport of the particles to the surface, then the "adhesion" to the wall and finally the
possible dislodging of the particles.
4.2.2 Results
Before studying the phenomenon of fouling, the metrological variation of the measurement
system was taken into account according to the main technological parameters:
Variation in temperature at constant flow.
Variation in flow at constant temperature.
Variation in viscosity at constant temperature and flow.
This phase is essential in order to separate the interferences of acoustic values generated by
the fouling phenomenon from those linked to the technological conditions of the exchanger
and its environment.
The curves in figure 21 show the evolution of the energy of the acoustic signals as well as
the pressure drop in the system as a function of the process time.
The "Power" curve shows the damping effect linked to the load on the plate caused by
fouling.


235

Fig. 21. Evolution of the power of the acoustic signal received during the fouling test and
cleaning
In conclusion, this work concerned the monitoring of fouling using acoustics. By adopting a
multi-stage experimental protocol we have been able to show that the variation in the
acoustic signal can be used to predict variations in the pressure drop as well as the state of
fouling in the plate heat exchanger under very specific operating conditions.
Finally, this study illustrates an example of a non-intrusive acoustic technique for the local
monitoring in real time of the fouling of plate heat exchangers. The results show that it is
possible to follow the relative kinetics of the state of fouling in each zone of the exchanger
with the right choice and positioning of the sensors.
5. Conclusion
This chapter has proposed a synopsis of all the work that has led to the development of
novel low frequency sensors. By using structural resonance modes excited by a transducer,
these sensors present the advantage of having small sized sources with regard to the
acoustic wavelength generated. These sensors are omni-directional but can nevertheless
present significant contact areas with the medium to be characterised. This is the case for
sensors developed for the characterisation of gels. The close contact of the elements set in
resonance with the medium enables phenomena linked to changes in state to be monitored
easily. Various applications have led us to develop sensors with very different geometries
and which are optimised with the application in mind.
Indeed, for each need expressed, the approach consisted in optimising not only the
geometry of the sensors but also their optimum position according to the problem posed.
Three different cases were thus studied:
identical near-field coupled sensors, through the medium to be characterised. They
were used for monitoring the evolution of the ultrasonic values to characterise a sol-gel
transition or the cohesion kinetics of a medium. For certain applications, the sensors are
immersed in the medium. This direct immersion is essential for characterising fragile
media.


236
a low frequency receiver associated with an excitation of the medium via a mechanical
shock in the case of very absorbent and scattering media. A second identical sensor is
used for the synchronisation of the acquisitions thus reducing, by standardisation, the
scattering of the values measured. The mechanical shock produces significant vibratory
energy over a broad frequency range.
finally, the sensors were coupled to heat exchanger plates in order to characterise
fouling. This work has shown the interest of using acoustic sensors to monitor
processes, providing an often local and dynamic response to the evolution of the
performances of the process.
The work carried out provides a solid base of knowledge on ultrasound-complex media
interactions. This knowledge could be put to good use in the development of sensors and
integrated ultrasonic methods and their applications in the analysis and monitoring of local
properties.
6. References
Aggarwal R. R., (1952a). Axially Symmetric Vibrations of a Finite Isotropic Disk. I, Journal of
acoustical society of America, Vol. 24, N0. 5, pp. 463-467
Aggarwal R. R., (1952b). Axially Symmetric Vibrations of a Finite Isotropic Disk. II, Journal
of acoustical society of America, Vol. 24, N. 6, pp. 663-666
Allsopp, M. W. (1981). The developement and importance of suspension PVC morphology,
Pure an applied chemistry, Vol. 53, pp. 449-465.
Blevins R. D., (1979). Formulas for natural frequency and mode shape, Van Nostrand
Reinhold Co., ISBN 0-4422-0710-7, New York, USA
Brekhovskikh, L.M., (1980). Waves in layered media, Academic Press, ISBN 0-12-130560-0,
New York, USA
Case, L. C. (1960). Molecular distributions in polycondensations involving unlike reactants.
VII. Treatment of reactants involving nonindependent groups, Journal of polymer
science, Vol. 48, pp. 27-35
Clerc, J. P. ; Giraud, G. ; Roussenq, J. ; Blanc, R. ; Carton, J.P. ; Guyon, E. ; Ottavi, H. &
Stauffer, D. (1983). La Percolation: modles, simulation analogiques et numriques,
Annales de Physique, Vol. 8, Masson, Paris, France.
Dalgleish, D. G. (1982). Developments in Dairy Chemistry, edited by P. F. Fox (Applied
Science, London,), Vol. 1, Chap. 5, ISBN 0-8533-4142-7, London, United kingdom
Dalgleish, D.G. (1993). Cheese: Chemistry, Physics and Microbiology, General Aspect, 2nd
ed., Vol. 1, p. 69, Fox, P.F., Chapman & Hall, ISBN 0-1226-3652-X, London, United
kingdom.
De Gennes, P. G. (1989). Scaling Concepts in Polymer Physics, Cornell University Press,
Ithaca, ISBN 0-8014-1203-X, New York, USA
Degertekin, F. L. & Khury-Yakub, B.T. (1996). Hertzian contact transducers for non-
destructive evaluation, Journal of acoustical society of America, Vol. 99, pp. 299-308
Degertekin, F. L. & Khury-Yakub, B.T. (1996). Lamb wave excitation by Hertzian contacts
with applications in NDE. IEEE Transactions on Ultrasonics Ferroelectrics and
Frequency, Vol. 44, N. 4, pp. 769-778
Degertekin, F. L. & Khury-Yakub, B.T. (1996). Single mode lamb wave excitation in thin
plates by Hertzian contacts, Applied physics letters, Vol. 69, N. 2, pp. 146-148


237
Eichinger, B. E. (1981). Random elastic networks. I. Computer simulation of linked stars,
Journal of chemical physics, Vol. 75, pp. 1964-1979
Ensminger, D. E. (1960). Solid cone in longitudinal half-wave resonance, Journal of
acoustical society of America, Vol. 32, pp. 194-196
Flory, P. J. (1953). Principles of polymer chemistry, Cornell University Press, Ithaca &
London, ISBN 0-8014-0134-8, New York, USA
Fox, P. F. (1989). Proteolysis during cheese manufacture and ripening. A review, Journal of
dairy science, Vol. 72, pp. 1379-1385
Gupta, S. K.; Kumar A. & Bhargava, A. (1979). Molecular weight distribution and moments
for condensation polymerization of monomers having reactivity different from
their homologues, Polymer, Vol. 20, pp. 305-310
Irie T., Yamada G. & Muramoto Y., (1984). Natural frequencies of in-plane vibration of
annular plates, Journal of sound and Vibration, Vol. 97, N. 1, pp. 171-175
Lee M. & Singh R., (1994). Analytical formulations for annular disk sound radiation using
structural modes, Journal of acoustical society of America, Vol. 95, N. 6, pp. 3311-
3323
Leissa A. W., (1969). Vibration of plates, NASA SP-160, U.S. Government Printing Office,
Washington, D.C.
McMahon D. J. & Brown R. J., (1984). Enzymic coagulation of caseine micelles: a review,
Journal of dairy science, Vol. 67, pp. 919-929
Mercier, J. P. & Marechal, E. (1993). Chimie des Polymres 1st ed. Presses polytechniques et
universitaires romandes, Lausanne, Chap. 1, 3, 8. Lausanne, Swiss.
Moseley D. S., (1960). Contribution to the Theory of Radial Extensional Vibrations in Thin
Disks, Journal of acoustical society of America,Vol. 32, N. 8, pp. 991-995
Mller-Steinhagen H. & Middis J., (1989). Particulate fouling in plate heat exchangers, Heat
Transfer Engineering, Vol. 10, N. 4, pp. 30-36
Nassar, G. (1997). Etude et Optimisation d'un Dispositif Ultrasonore De Suivi en Ligne des
proprits viscolastiques, Doctoral dissertation, Valenciennes University-France.
Nikolovski, J. P. & Royer, D. (1997). Local and selective detection of acoustic waves at the
surface of a material, IEEE Ultrasonics Symposium, pp. 699-703, ISBN 0-7803-4153-
8, Toronto, Ontario, Canada, October 5-8, 1997
Nol, Y. ; Flaud, P. & Quemada, D. (1989). Traitement Industriel des Fluides Alimentaires
Non Newtoniens, Tome II, Actes du 2me Colloque la Baule, La Baule, France,
September 11-13, 1989, pp. 215-224.
San Biagio, P. L.; Bulone, D.; Emanuele, A.; Madonia, F.; Di Stefano, L.; Giacomazza, D.;
Trapanese, M.; Palma-Vittorelli, M. B.; & Palma, M.U. (1990). Spinodal demixing,
percolation and gelation of biosttural polymers, IUPAC 10th Int. Symp. on
Polymer Networks, Vol. 40, pp. 33-44, Jerusalem, Israel, December, 1990
Shuyu, L. (1996). Study on the longitudinal-torsional composite mode exponential ultrasonic
horns, Ultrasonics, Vol. 34, pp. 757-762
Shuyu, L. (1997). Study on the longitudinal-torsional composite vibration of a sectional
exponential horn, Journal of acoustical society of America, Vol. 102, pp.1388-1393
Stauffer, D. (1981). Can percolation theory be applied to critical phenomena at gel point?,
Pure an applied chemistry, Vol. 53, pp. 1479-1487
Stauffer, D. (1985). Introduction to Percolation Theory, Taylor & Francis Ltd., ISBN 0-7484-
0253-5, London, United Kingdom


238
Stockmayer, W. H. (1943). Theory of molecular size distribution and gel formation in
branched chain polymers, Journal of chemical physics, Vol. 11, pp. 45-55
Vogel S. M. & Skinner D. W., (1965). Natural frequencies of transversely vibrating uniform
annular plates, Journal of applied mechanics, Vol. 32, pp. 926-931
Walstra, P. & Vliet, V. (1986). The physical chemistry of curd making, Netherlands. milk
dairy journal, Vol. 40, pp. 241-259
11
Modeling of Biological Interfacial Processes
Using ThicknessShear Mode Sensors
Ertan Ergezen et al.
*

School of Biomedical Engineering, Health and Sciences, Drexel University, Philadelphia
USA
1. Introduction
Biological interfaces and accompanying interfacial processes constitute one of the most
dynamic and expanding fields in science and technology such as biomaterials, tissue
engineering, and biosensors. For example, in biomaterials, the bio-interfacial processes
between biomaterials and surrounding tissue plays a crucial role in the biocompatibility of
the layer (Werner, 2008). In tissue engineering, cellular adhesion plays an important role in
the regulation of cell behavior, such as the control of growth and differentiation during
development and the modulation of cell migration in wound healing, metastasis, and
angiogenesis (Hong et al., 2006). Performance of a biosensor is highly dependent on
interfacial processes involving the sensor sensing interface and a target analyte. Therefore,
quantitative information on the novel and robust immobilization of detector molecules is
one the most important aspects of the biosensor field (Kroger et al., 1998).
Thickness shear mode (TSM) sensors have been used in a variety of studies including
interfacial biological processes, cells, tissue and properties of various proteins and their
reaction (Cote et al., 2003). Phenomena such as cell adhesion (Soonjin et al., 2006.),
superhydrophobicity (Sun et al., 2006, Roach et al., 2007), particle-surface interactions
(Zhang et al.,2005), organic and inorganic particle manipulation (Desa et al., 2010) and
rheological and interfacial properties of blood coagulation (Ergezen et al. 2007) were studied
using TSM sensors. Due to the high interfacial sensitivity of TSM sensors, it has been shown
that cell motility can be monitored by analyzing the noise of the TSM sensor response
(Sapper et al., 2006). It has also been demonstrated that the number of motile sperm in a
semen sample can be assessed in real-time using a flow-chamber integrated with a thickness
shear mode sensor (Newton et al., 2007).
1.1 Quantification of Thickness Shear Mode (TSM) sensor response
The TSM sensor response is affected by the complex nature of the interface. Its response is
influenced by the geometrical and material properties of the interacting surfaces such as
surface roughness (Cho et al., 2007), hydrophobicity (Ayad and Torad, 2009), interfacial

*
Johann Desa, Matias Hochman, Robert Weisbein Hart, Qiliang Zhang, Sun Kwoun,
Piyush Shah and Ryszard Lec
School of Biomedical Engineering, Health and Sciences, Drexel University, Philadelphia
USA


240
slippage (Zhuang et al., 2008), coverage area (Johanssmann et al., 2008), sensitivity profile
(Edvardsson et al., 2005) and penetration depth of the shear acoustic wave (Kunze et al.,
2006).
Various theoretical models have been developed for quantitative characterization of the
TSM sensor response to interfacial interactions. Nunalee et al (2006) developed model to
predict of the TSM sensor response to a generalized viscoelastic material spreading at the
sensor surface in a liquid medium. Cho et al (2007) created a model system to study the
viscoelastic properties of two distinct layers, a layer of soft vesicles and a rigid bilayer.
Urbakh and Daikhin (2007) developed a model to characterize the effect of surface
morphology of non-uniform surface films on TSM sensor response in contact with liquid.
Hovgaard et al (2007) have modeled TSM sensor data using an extension to Kevin-Voigt
viscoelastic model for studying glucagon fibrillation at the solid-liquid interface. Kanazawa
and Cho (2009) discussed the measurement methodologies and analytical models for
characterizing macromolecular assembly dynamics.
The physical description based on a wave propagation concept in a one-dimensional
approximation has been proven as the best model of thickness shear mode (TSM) sensors.
The fundamentals have been published in several books (Rosenbaum, 1998). Martin et al.
have (1994) applied this background to sensors by using Mason's equivalent circuit to
describe the thickness shear mode sensor itself and transmission lines as well as lumped
elements for viscoelastic coatings, semi-infinite liquids etc.. Follow-up papers have
introduced a more straightforward definition of the elements of the BVD-model (Behling et
al, 1998) as well as several additional approximations, e.g. based on perturbation theory, to
derive less complex equations, have suggested a simplified notation to separate the mass
from so-called nongravimetric effects, or have applied the transmission line model to several
subsystems (Voinova et al, 2002) for demonstration of specific situations just to call some
examples. More recent papers deal with deviations from the one-dimensional
approximations, e.g. by introducing generalized parameters by deriving specific solutions
e.g. for surface roughness or with discontinuity at boundaries.
TSM sensors combined with the theoretical models mentioned above were used to
determine the properties of liquids (Lin et al., 1993), high protein concentration solutions
(Saluja et al., 2005), and thin polymer films (Katz et al., 1996).
For viscoelastic layers, their mechanical impedance depends upon the density, thickness,
and the complex shear modulus of the loading. Identification of the all the system
parameters from the impedance measurements has been very challenging and uncertain
without a priori knowledge of the thicknesses and/or some of the material properties
(Lucklum et al. 1997).
Furthermore, Kwoun (2006) showed the beneficial features of the multi-resonance operation
of the TSM (called as multi-resonance thickness shear mode) sensor to study the formation
of biological samples, specifically collagen and albumin, on the sensor surface. In this work,
it was demonstrated that the different harmonic frequency clearly showed the different
characteristics of mechanical properties, especially shear modulus, of the biological sample.
Although this work was one of the pioneer studies to demonstrate the strengths of the
MTSM measurement technique, it is limited as it is a semi-quantitative method. Exact values
of mechanical properties of anisotropic collagen and albumin samples were not able to be
defined due to complexity of the non-linear simultaneous equations of the model. An
improved MTSM technique combined with an advanced data analysis technique was
proposed by Ergezen et al (2010). A new approach merging the multi-harmonic thickness

Modeling of Biological Interfacial Processes Using ThicknessShear Mode Sensors

241
shear mode (MTSM) measurement technique and genetic algorithm-based data analysis
technique has been used. This novel method was utilized to solve two unmet needs:
1. Identification of all four parameter by using the MTSM sensors single harmonic
response results in an under-determined problem. The MTSM sensor response enables
the identification of two parameters by providing imaginary and real components of
the mechanical impedance. In other words, there are fewer equations than the
material/geometrical parameters of the interface, therefore, the stochastic method is the
only approach that can address this problem mathematically. In this project it was
shown that combination of the MTSM measurement technique and the genetic
algorithm-based data analysis technique (called as MTSM/GA technique) was used to
solve this under-determined problem. It was reported for the first time, a novel approach
that enables determining all four parameters, which define the response of the MTSM
technique.
2. Most of the biological interfaces constitute multi-layer structures. Multi-layer modeling
of biological interfacial processes was proposed by several researchers and by us
(Wegener et al., 1999, Ergezen et al., 2007). In contrast, there has been very limited
(Lucklum et al., 2001) theoretical study and no experimental studies based on the
MTSM sensor for quantitative characterization of multi-layer biological processes. It was
reported, for the first time, the most comprehensive theoretical and experimental study for
quantitative characterization of multi-layer biological interfacial processes.
A new approach merging the multi-harmonic thickness shear mode (MTSM) sensor and a
data extraction technique based on stochastic global optimization procedure has been
proposed. For this purpose, the MTSM/GA technique is being developed and calibrated
with a polymer layer (having known properties). This was then used to estimate the
properties of a protein layer with unknown properties adsorbed to the MTSM sensor
surface. It was demonstrated that this new method has the potential to be a novel tool for
quantitatively characterization of interfacial biological layers.
2. Theory
2.1 Multi-Harmonic Thickness Shear Mode (MTSM) sensor
Piezoelectric MTSM sensors transmit acoustic shear waves into a medium under test, and
the waves interact with the medium. Shear waves monitor local properties of a medium in
the vicinity of the sensor and of the medium/sensor interface (on the order of nm - m);
thus, they provide a very attractive technique to study interfacial processes. Measured
parameters of acoustic waves are correlated with medium properties such as interfacial
mass/density, viscosity, or elasticity changes taking place during chemical or biological
processes.
The shear acoustic wave penetrates the medium over a very short distance. The square of
the depth of penetration of an acoustic shear wave in MTSM sensor is related to medium
viscosity, elasticity, density and the frequency of the wave (please see Appendix IA.)
(Kwoun et al. 2006). Figure 1a shows the acoustic wave penetrating the adjacent medium
and Figure 1b shows that the depth of penetration decreases at higher harmonic frequencies
in a semi-infinite medium.
Therefore, by changing the frequency, one can control the distance at which the wave
probes the medium. Multi-harmonic operation of MTSM sensor will enable to control the
interrogating depth into the biological processes. Therefore it will provide a more in depth


242
characterization of the biological interfacial processes. For example, it was suggested that
cell adhesion on extra cellular matrix should be modeled as a multi-layered structure
(Wegener et al. 2000). Therefore MTSM sensors can provide information about mechanical
and structural properties of the biological processes from different depths (slicing the
medium).

Fig. 1. a) Acoustic wave penetrating into the medium b) depth of penetration decreases at
higher harmonic frequencies
It should be noted that it was assumed that the medium is semi-infinite and the mechanical
properties are not frequency dependent in fig. 1.
2.2 Electrical response of MTSM sensor
The MTSM sensor is a piezoelectric-based sensor which has the property that an applied
alternating voltage (AC) induces mechanical shear strain and vice versa. By exciting the
sensor with AC voltage, standing acoustic waves are produced within the sensor, and the
sensor behaves as a resonator. The electrical response of the MTSM sensor in air over a wide
frequency range is shown in figure 2, where S
21
is the magnitude response of the MTSM
sensor (|S
21
|=20log(100/(100+Z
t
)), Z
t
=total electromechanical impedance of the MTSM
sensor (Rosenbaum 1998). As an example, the magnitude and phase responses of MTSM
sensor are presented at the first (5 MHz), third (15 MHz), fifth (25 MHz) and seventh (35
MHz) harmonics in air.

Fig. 2. A typical a) frequency vs. magnitude response and b) frequency vs. phase response
characteristic and the associated resonance harmonics for the MTSM sensor, spanning a
wide frequency range (5 MHz to 35 MHz). (Insets) Magnified view of magnitude and phase
response at 5 MHz


243
An example of the MTSMs magnitude response in the vicinity of the fundamental resonant
frequency is given below (figure 3a). When the TSM sensor is loaded with a biological
media, there will be a shift in resonant frequency and a decrease in the magnitude. These
changes can be correlated with changes in the mechanical and geometrical properties of the
medium such as thickness, viscosity, density and stiffness. Depending on the changes at the
interface of the sensor surface-medium interface, a positive and/or negative shift can be
seen in the frequency response (Figure 3b).

Fig. 3. (a)Demonstration of a typical qualitative frequency-dependent response curve for the
MTSM sensor in the vicinity of the resonant frequency; n = harmonic number,
Rn
=Initial
maximum magnitude, f
Rn
=Initial resonant frequency, (b) In the case of both positive and
negative frequency shifts throughout the experiment,
Rn
I
,
Rn
II
=Instantaneous maximum
magnitudes of loaded MTSM sensor at time t
1
and t
2
respectively, f
Rn
I
,

f
Rn
II
=Instantaneous
resonant frequencies of the loaded MTSM sensor at time t
1
and t
2
respectively (Inlet)
resonant frequency and magnitude are monitored as a function of time
2.3 MTSM/GA data processing technique
This section will be structured in the following manner; first, the general structure of a
genetic algorithm will be explained. Second, advantages of genetic algorithm over other
techniques will be discussed. Finally, implementation of MTSM-GA technique for
determination of material parameters will be explained.
Principles of operation of a genetic algorithm (GA)
Basic definitions of GA terms are defined in Appendix IB. Genetic algorithm (GA) is based
on the genetic processes of biological organisms (figure 4). GA works with a population of
individuals, each representing a possible solution to a given problem. Each individual is
assigned a fitness score according to how good a solution to the problem it is. The highly-fit
individuals are given opportunities to reproduce, by cross breeding with other individuals
in the population. This produces new individuals as offspring, which share some features
taken from each parent.
Comparison of GA to other data processing techniques
Complex models are ubiquitous in many applications in the fields of engineering and
science. Their solution often requires a global search approach. Therefore the objective of
optimization techniques is to find the globally best solution of models, in the possible


244
presence of multiple local optima. Conventional optimization and search techniques
include; (1) gradient-based local optimization method, (2) random search, (3) stochastic hill
climbing, (4) simulated annealing, (5) symbolic artificial intelligence and (6) genetic
algorithms. The detailed information on each technique and comparisons to Genetic
Algorithms (GA) are already explained by Depa and Sivanandam (2008). Here, the aim is
not to analyze these techniques in detail but to show the suitability of GA as a parameter
estimation algorithm. As discussed by Depa and Sivanandam, some of the advantages of
GA over other techniques are: (1) it is good for multi-mode problems, (2) it is resistant to
becoming trapped in local optima, (3) it performs well in large-scale optimization problems,
(4) it handles large, poorly understood search spaces easily. These advantages match with
the requirements for an optimization technique to be applied in this application. Therefore
GA was chosen as an optimization technique and successfully combined with the MTSM
technique.

Fig. 4. Flow chart of a genetic algorithm
Structure of the MTSM/GA technique
The structure of MTSM-GA technique is presented in figure 5. As seen from the figure, there
are two inputs to the GA, namely; range of variables and MTSM sensor response. GA
outputs the determined values of the variables by using GA functions such as crossover,
mutation and fitness evaluation. In the following sections, initially, the inputs to the GA will
be explained. Then the structure of GA and its internal functions will be presented.
MTSM sensor response
The first input to the GA is the MTSM sensor response. Both magnitude and phase
responses were continuously monitored during the experiments (see materials and methods
section). Then the specific points on these responses such as resonant frequency, maximum
magnitude, minimum phase, frequency at minimum phase, and phase at maximum
magnitude were input to GA for calculating the fitness score for each individual. The
changes in these target points were calibrated with the diwater/glycerin changes.
Selection of the ranges for variables
The next step of the technique is to set the ranges for the variables (chromosomes). These
ranges represent the bounded space within which the GA will search for solutions. The


245
ranges should be reasonable for each parameter in order to determine accurate solutions.
For example, for a Newtonian liquid the stiffness is 0, therefore one should not set the range
to be between 1e5 N/m
2
and 1e7 N/m
2
.

If this were done the algorithm will not converge to
a solution because of the inappropriate choice of ranges.

Fig. 5. Basic structure of MTSM/GA technique
As shown by Kwoun (2006), the viscoelastic materials can be divided in to four regimes,
namely; liquid like, soft rubber, hard rubber and solid like. As seen from table 1, the
viscosity values might change between 0.001 and 0.1 kg/m.s and stiffness value changes
between 0 1e9 N/m
2
. Typical range of density values for a polymer was determined to be
between 1000 1400 kg/m
3
.

Phase (kg/m.s) C (N/m^2)
Liquid like 0.001 0.01 0-1e5
Soft Rubber 0.01 0.1 0-1e5
Hard Rubber 0.01 0.1 1e5 1e7
Solid Like - 0.1 1e7 1e9
Table 1. Four regimes of a viscoelastic system
Genetic Algorithm and its internal functions
This section will be divided into three sections. First, the GAs main parameters such as
number of populations, crossovers, mutation rates and genes per chromosome will be
analyzed. Then the fitness function of the GA will be explained. Finally, the technique
combination of sub-spacing and zooming to determine the values for four variables will be
presented.
Selection of GA parameters
Different combinations of the GA parameters were evaluated. Here, the combination that
gives the best result is presented. Each variable was represented by a binary chromosome
that contains 16 genes. A random population of 100 individuals was generated. Tournament


246
selection was implemented for selection of individuals for mutation and crossover. In order
to carry out the crossovers the entire population is divided into groups of 5 individuals
each, these groups are randomly selected. From each group, the individual with the highest
fitness together with another individual of this group are selected for crossover. The two
selected individuals are the parents and yield two offspring. Both the parents and the
offspring pass to the next generation. This idea was implemented in order to reduce the
selection pressure.
The crossover between the parents is a simple one meaning that a random crossover point is
selected and two kids genome are formed with the left and right genes of the crossover
point of each parent. A relatively high mutation probability (0.5) is present in order to avoid
local minimum, otherwise all the individuals might end up having the same genome and
this genome corresponding to a not optimal solution. Also elitism was implemented to
assure that the best individual of a generation survives to the next generation. This ensures
that the algorithm keeps the best solution until a better one is found.
Fitness function
One of the most important parts of a genetic algorithm is the fitness function. The fitness
function must reflect the relevant measures to be optimized. This function evaluates the
function being searched for the set of parameters of each member of the population. The
output of the fitness function is a vector that contains the fitness for each member of the
population. This vector helps in the selection of individual for generating new offspring or
individuals that will be included in the new generated population.
The approach used, in this study to model biolayers on a MTSM sensor, is Mason's
transmission line model (please see Appendix C). This model is a one-dimensional model
that describes the electrical characteristics of an acoustic structure wherein, each layer of
load can be represented as a T-network of impedances.
Once the initial population is created the algorithm randomly generates a population
(includes 100 individuals) chosen from the ranges of the variables (the section titled
selection of the ranges for variables). Then each individual was input to fitness function
(transmission line model). The error between the model (transmission line model) and the
experimental results were compared by using the following equation:
2 2 2 2 2 2
100
1
Re Rt Re Rt Me Mt Me Mt ARe ARt ARe ARt
fit_ func
( ( ) ( f f ) (P P ) ( f f ) ( ) ( f f ) )
=
+ + + + + +

The denominator of this function represents the difference between the model and the
experimental data (we use the plus one in order to avoid the eventual division by zero). In
this project, rather than fitting the whole magnitude and phase curve, certain points such as
R
= maximum magnitude, f
R
= resonant frequency, P
M
= minimum phase, f
M
= resonant
frequency at minimum phase,
AR
= minimum magnitude, f
AR
= anti-resonant frequency has
been compared between the model and the experimental results. Subscript e indicates
experimental results and subscript t stands for theoretical model. This function is
monotonously increasing with the kindness of the solution provided by the genetic
algorithm. The algorithm was terminated at after 500 generations.
Set-up of the Genetic Algorithm
Acoustic impedance seen at the sensor/film interface is derived from transmission line
theory (Martin and Frye 1991). Surface mechanical impedance is related to density and


247
thickness of the film, and complex modulus (= G
I
+ jG
II
). Therefore there are four
independent variables to define the surface acoustic impedance. The MTSM sensor response
contributes two parameters by providing real and imaginary part of mechanical impedance.
Hence using single harmonic response results in an under-determined problem. Genetic
optimization technique has been applied to under-determined problems to obtain
approximate solutions with satisfactory accuracy (Wang and Dhawan, 2008). Here genetic
algorithm has been improved by combining sub-space and zooming techniques. It was
shown that this combination provides very good approximation with less than 1% error.
First, sub-spacing method was applied. This method gives a quick idea of where the
solution can be and also it decreases algorithm running time dramatically (Garaia and
Chaudhurib, 2007). Therefore the solution space was divided in 10 sub-spaces. Genetic
algorithm was run 5 times in each subspace. Each subspaces convergence performance was
evaluated. The sub-space with the best fitness score was considered to be the candidate
solution space. It was observed that the candidate sub-space had a distinct convergence
performance compared to the others. This method dramatically increased the efficiency of
GA by eliminating the irrelevant solution spaces.
Secondly, GA was run 100 times (this number was chosen to have 95% confidence level and
10% confidence interval statistically). The termination criterion for each run was 500
generations. After 100 runs, it was observed that, for two out of four variables, observed
points having a uniform distribution (skewness < 0.5) were accumulating around one
number in a narrow range (in 20% of candidate solution point). The average value of the
observed points was also equal or very close (<5%) to solution (theoretically shown).
Therefore GA was always able to converge to the most likely values for two out of four
variables after these two steps (from our observations, mostly stiffness and thickness, and
sometimes, viscosity and thickness). It was shown theoretically that one can always put these
numbers, and calculate the other two variables with the error of less than <15% at this step.
Then zooming method was applied to reduce the search space around the candidate
optimum solution point. Several zooming methods have been developed for different
applications (Ndiritu and Daniel, 2001, Kwon et al. 2003). In this project, the GA was run 30
times, and then the new range was set to be between maximum and minimum numbers of
the 30 points. This zooming continued until the error was less than 1% for all variables. This
error was achieved after 6 zooming.
These results showed that the MTMS/GA technique combined with sub-spacing and
zooming methods can be applied successfully to approximate the solution with good
accuracy for this under-determined problem.
3. Materials and methods
The MTSM/GA technique first experimentally tested with the polymer SU8-2002 layer spin
coated on sensor surface. The determined properties of the layer were compared with the
values obtained from literature. The technique was then applied to obtain the mechanical
and geometrical properties of a protein layer adsorbed on gold layer. The methods and
chemicals used in the experiments are described below.
a. Deposition of the thin polymer film
The SU 8-2002 (MicroChem) polymer solution was spin coated on MTSM sensor by using
the following procedure. First, the gold electrode surface of TSM sensors was cleaned using


248
Piranha solution (one part of 30% H2O2 in three parts H2SO4). After 2 min exposure time,
the sensors were rinsed with distilled water. The surface was dried in a stream of nitrogen
gas. The SU 8 2002 sample was dispensed on MTSM sensor surface and sensors were spin
coated for 40 seconds. The sensors were then soft baked for 1 min at 95
o
C. The SU 8-2002
films were exposed to UV light for 4 seconds under 25 mJ/cm
2
. This was followed by 1 min
hard baking on hot plate at 95
o
C.
b. Antibody adsorption on MTSM sensor surface
The reference measurements were taken for air and phosphate buffer saline (PBS). Next, the
sensors were exposed to rabit-immunoglobulin G (IgG) (50 g/ml) suspended in diwater
(Fisher Scientific, pH: 5.34, Cat No: 25555-CM) for 50 minutes to allow IgG coating of the
sensor surface by adsorption.
c. Characterization of geometrical properties of the thin film
The thicknesses of the SU 8 2002 films were determined by using optical profilometer
(Zygo Inc. Model #: NV6200). For the thickness measurements, a very small portion of
MTSM sensor surface was not exposed to UV light. After the films were developed, the SU
8-2002 layer was removed from this portion. To obtained different thicknesses of film layer,
1:1 solution of SU8-2002 and cyclopentanone (Acros Organics) was prepared.
The surface topography of the film layer was measured using atomic force microscopy
(AFM). The prepared samples were placed on a glass slide installed on the atomic force
microscope (Bioscope; Veeco), that was mounted on the inverted fluorescence microscope
(TE2000; Nikon, Melville, N.Y.). Measurements were made using contact mode with a scan
rate of 2 Hz.
d. Measurement system and MTSM sensor data analysis technique
A 14 mm diameter, 0.33 mm thick, 5 MHz quartz crystal with deposited 7 mm gold
electrodes was placed in a custom fabricated brass sensor holder (ICM). The sensor holder
was connected to a Network Analyzer (NA) (HP4395A). A LabView program on a personal
computer was used to control the network analyzer and collect the data at 5, 15, 25 and 35
MHz. The experiments were done in room temperature (24
o
C1
o
C). Magnitude and phase
responses of MTSM sensor were monitored during the experiments (figure 6). The sampling
rate was 30 seconds. Each experiment was repeated three times.

Fig. 6. a) Magnitude and b) phase responses of MTSM sensor


249
4. Results and discussions
Initially, two different thicknesses of SU8 2002 layers were spin coated on sensor surface and
changes in the frequency and magnitude responses were monitored at 5, 15, 25 and 35 MHz.
The thicknesses of the layers were measured by using optical profilometer (fig. 7a). The
average thicknesses of the layers were 192025 nm and 77050 nm respectively. Surface
topography of the SU8 - 2002 layers was measured by using AFM (fig. 7b). The average
roughness of the layer was 20 nm and no cracks on the surface were observed.

a)

-0.5
0
0.5
1
1.5
2
2.5
0 0.2 0.4 0.6 0.8
Distance (mm)
T
h
i
c
k
n
e
s
s

(
u
m
)
Gold level
Sample B
Sample A
b)

Fig. 7. A) Thickness measurements from optical profilometer sample a. SU8-2000 solution
sample b. 1:1 dilution of SU8-2002 and cyclopentanone B) Surface topography of SU 8 layer
a. Determination of mechanical and geometrical properties of SU8 layer of 1.92 m
thickness
First set of experiments were performed by spin coating 2 m thick SU 8 - 2002 layer on
sensor surface. The MTSM/GA determined properties are presented in table 2. The average
thickness of the polymer layer determined to range from 2080 nm to 2140 nm among the
harmonics. Although these values are slightly higher than the value (192025 nm) obtained
in control experiments, they are still in less than 10% experimental errors. The variation
between the frequencies for density value was also very small, ranging from 1240 to 1253
kg/m
3
. These numbers correlate well with the literature value of 1200 kg/m
3
(Jiang et al.,
2003) for SU8.

MTSM Frequency MTSM/GA Results Profilometer Jiang et al. [38]
(MHz) d(nm) (kg/m
3
) d (nm) (kg/m
3
)
5 212060 125310

192025

1200
15 214050 124611
25 2080110 124050
35 208060 124028
Table 2. Comparison density and thickness values of SU 8-2002 layer determined using
MTSM/GA sensor at 5, 15, 25 and 35 MHz with profilometer and Jiang et al. (Jiang et al.
2003)
SU8 layer
Gold layer


250
The frequency dependent shear modulus of SU 8-2002 layer obtained using the MTSM/GA
is presented in table 3. Both loss and storage modulus varies with the operating frequency.
These extracted values were compared with the values obtained by Jiang et al (2003) (table
3). Jiang et al calculated the shear modulus of SU8 layer by using the impedance-admittance
characteristics of the equivalent circuit models of loaded and unperturbed TSM sensors
operating at 9 MHz.

MTSM
Frequency
MTSM/GA Results
Jiang et al.[38]
( at 9 MHz)
(MHz) G
I
(N/m
2
) G
II
(N/m
2
) G
I
(N/m
2
) G
II
(N/m
2
)
5 (4.552.12) x10
7
(1.890.26) x10
5

7.80e7

2.00e5
15 (2.330.18)x10
8
(1.000.03) x10
6

25 (3.820.52)x10
8
(4.690.52) x10
6

35 (5.810.71) x10
8
(6.490.18) x10
6

Table 3. Comparison of determined G
I
and G
II
values of SU8 layer using MTSM/GA at 5, 15,
25 and 35 MHz and Jiang et al (2003)
As seen in table 3, the values obtained by Jiang et al. fall between the values obtained using
the MTSM/GA method for 5 and 15 MHz. The small variation in the G
I
and G
II
may be due
to difference in the film preparations. Alig et al. (1996) has shown that variations in film
preparation methods can affect the mechanical properties of the polymer layers.
b. Determination of mechanical and geometrical properties of SU8 layer of 0.770 m
thickness
The second set of experiments was done with the ~770 nm thick SU 8-2002 layer on MTSM
sensor. As seen from the table 4, the thickness of the layer determined using the MTSM/GA
method correlates well with the expected thickness for each harmonic (less than 10% error).
Furthermore the results vary only 10 nm between the harmonics. Similarly, determined
values for density were consistent between the harmonics, which is around ~1200 kg/m
3
.

MTSM Frequency MTSM/GA Results Profilometer Jiang et al.[39]
(MHz) d(nm) (kg/m
3
) d (nm) (kg/m
3
)
5 82045 118040

77050

1200
15 82020 119030
25 81035 119050
35 81052 121335
Table 4. Determined density and thickness values by MTSM/GA or 770 nm thick SU8 layer
at 5, 15, 25 and 35 MHz
The initial losses before coating were -0.53 dB and -2.5 dB for 5 and 35 MHz respectively.
The losses increase to -0.59 dB for 5 MHz and -4.18 dB for 35 MHz. As seen from these
results, the losses remain relatively low when the thickness of the layer was decreased to 770
nm in contrast to the phenomenon observed when the film thickness was 2 m. For 2 m


251
film thickness, the losses increase to -1.9 dB and -11.5 dB at 5 MHz and 35 MHz respectively,
while initial loses were similar to what observed for 770 nm film thickness.
The shear modulus values determined via the MTSM/GA technique are presented. Both
loss and storage modulus were decreased compared to the values obtained when film
thickness was 2 m (figure 8). It has been shown that the scale effect on the mechanical
properties of the polymers might be the reason for the decrease in the values (Liu et al, 2009,
Luo et al, 2003).

Fig. 8. a) Storage and b) loss modulus as a function of harmonic frequency for 770 nm and
1920 nm thick SU8 layer at 5, 15, 25 and 35 MHz (error bars are smaller than symbols when
not visible)
c. Determination of mechanical and geometrical properties of an antibody layer
Third set of experiments were done by adsorbing an antibody layer on MTSM sensor
surface under static conditions at 5, 15, 25 and 35 MHz. Antibodies play crucial importance
in many applications such as biosensing (Hanbury et al. 1996) and drug delivery (Morrison
et al., 1995). The sensor surface was saturated with antibody to form a uniform protein layer
on the surface. Change in the frequency and magnitude responses at 15, 25 and 35 MHz are
presented in figure 9. At the fundamental frequency (5 MHz), high fluctuations observed in
sensor response are likely due to insufficient energy trapping as described by others (Li et
al. 2004).

Fig. 9. Time response of A. resonant frequency and B. maximum magnitude responses of
MTSM sensor to antibody binding at 15, 25 and 35 MHz


252
The properties of the medium were determined at t
1
= 10 and t
2
= 70 minutes. At t
1
= 10, the
system is modeled as MTSM sensor loaded with semi-infinite Newtonian medium (
DIwater) (fig 10A). The height of the column (2 mm) was much higher than the penetration
depth of the acoustic wave at 5 MHz (~250 nm in DI water).
At t
2
= 10 min., the MTSM/GA determined properties of the layer at 15, 25 and 35 MHz are

presented in table 5. The variations in the determined thickness values were very high
(ranging from 300 nm to 5 m due to the fact that column height was much larger than the
penetration depth. Solution range for thickness values was set to be between 1 nm to 10 m
in genetic algorithm. Thus any thickness value larger than the penetration depth will satisfy
the solution because the MTSM sensor is not sensitive to the changes beyond the penetration
depth. However, the solutions were always higher than penetration depth as expected. Due
to the high fluctuations in thickness values, it was not presented here. In contrast the
solutions for
1
,
1
and C
1
match with the literature values very well. (Literature values are
1
= 1000 kg/m
3
,
1
= 0.001 kg/m.s and C
1
= 0 N/m
2
at room temperature (Greczylo and
Deboswka 2005)).

Fig. 10. Physical model for MTSM sensor system at A) t=10 and B) t=70

MTSM Frequency Density (kg/m
3
) C
I
(N/m
2
) (kg/m.s)
15 10065 (2.001.00) x10
2
(1.05.004) x10
-3

25 10032 (5.003.00) x10
2
(1.080.03) x10
-3

35 10044 (1.001.00) x10
2
(1.060.04) x10
-3

Table 5. Determined properties for semi-infinite Newtonian medium layer by MTSM/GA at
15, 25 and 35Hz
At t = 70 min., the physical model is presented in fig 10b. A viscoelastic layer (protein layer)
with finite thickness and semi-infinite Newtonian medium were loaded on MTSM sensor.
The properties for diwater layer were entered into the algorithm as known variables and the
unknown properties (
p,
p,
C
1
and d
p
) of viscoelastic layer were determined using the
MTSM/GA method. The results are presented in table 6. The thickness of the layer was
determined to range from 10.3 to 11 nm for the harmonics. This number correlates well with
the values presented by the other researches. Westphal et al (Westphal and Bornmann 2002)
calculated the height of antibody layer as 9.2 nm. Furthermore Liao et al (Liao et al 2004)
measured the average height of the antibody layer as 10.13.3 nm.


253
The density of the antibody layer was also determined by the MTSM/GA to be 103014
kg/m
3
. This density value is close to the water density in which the antibodies were
suspended. Hook et al. (2002) considered the density of antibody layer as 1050 kg/m
3
when
the antibodies were not attached to gold surface. After the cross-linking, the density value
was1300 kg/m
3
, this is closer to the density value of dry protein. Voros (2004) also showed
that the wet density of antibody layer is significantly different than the dry protein density
value due to the solvent present in the adsorbed proteins. Therefore we believe that the
determined value of the density is in a reasonable range.

MTSM Frequency Thickness (nm) Density (kg/m
3
) G
I
(N/m
2
) G
II
(N/m
2
)
15 110.3 105010 (5.200.5) x10
4
(4.800.58) x10
5

25 10.40.6 108012 (5.000.13) x10
4
(9.501.40) x10
5

35 10.30.4 104014 (5.600.12) x10
4
(1.520.31) x10
6

Table 6. Determined properties for antibody layer by MTSM/GA at 15, 25 and 35 MHz
As seen from the table 6, the adsorbed antibody layer has low storage modulus (<1e5
N/m
2
), and relatively higher loss modulus. While storage modulus was same for each
harmonic, loss modulus changed with frequency. It has been experimentally shown that the
adsorbed protein layers on TSM sensor, such as antibody, vesicles and cells do not behave
like rigid and thin films (Voinova et al, 2002). Therefore the linear relationship between
resonant frequency shift and mass deposition is not observed. Saluja et al. (2005) indicated
low concentrations (less than 60 mg/ml) of antibody suspension behave like Newtonian
medium. But it should not be expected that the properties of adsorbed layer will not be the
same as the properties of antibody suspension. The effect of the binding between protein
layer and gold layer should be considered. No literature value was found for direct
comparison. Therefore we believe that MTSM/GA technique will lead to development of a
quantitative tool for study of biological interfacial processes.
5. Conclusions
It was shown that MTSM sensor combined with genetic algorithm can be used to extract
mechanical and geometrical properties of biological layers. The developed technique was
first experimentally tested with SU8-2002 polymer layers with known properties having two
different thicknesses. It was shown that the developed technique was successfully
determined the mechanical and geometrical layers of thin polymer layers. MTSM/GA
technique was then applied to extract the properties of antibody layer coated on MTSM
sensor. The obtained data support our hypothesis about use of MTSM/GA technique can be
a powerful tool for quantitative characterization of interfacial biological interfacial
processes.
6. Acknowledgments
We are thankful to Dr. Moses Noh for providing supplies and micro-fabrication facilities for
polymer coating.


254
7. Appendix I
A. The depth of penetration of a shear wave ()
The depth of penetration of a shear wave () in a Newtonian medium is given by the
equation shown below:

1/4
2
2 2
1
1
arctan
( ) 2
n
m m
n n
m n m m
Sin
C C

=

+

(2)
m
= density of medium (kg/m
3
),
m
= viscosity of medium (kg/m.s),
C
m
= stiffness of medium (N/m
2
),
= angular frequency (rad/s) and
n = harmonic number
B. Basic terminologies of a genetic algorithm
Individual: A solution to the problem is called an individual.
Population: The total number of solutions is called population.
Chromosome: Each individual has a number of chromosomes that represent each parameter
(i.e. variables to be determined) of the problem.
Genes: Each chromosome contains a fixed number of genes, the number of genes per
chromosome determine the resolution of the total solution. The number of genes per
chromosome is mostly determined by the broadness of the range in which each
chromosome lies.
Fitness: Every individual has to be weighed according to its fitness. The individual fitness
value determines its survival and breeding probability. A higher fitness individual has
higher probability of survival.
C. Masons transmission line model
As seen in fig. 11, the biological process consist of a piezoelectric layer (MTSM sensor) and a
non-piezoelectric biological layer. In this model, each layer of load can be represented as a
T-network of impedances.

Fig. 11. Mason model representation of non-piezoelectric layers loaded on piezoelectric plate


255
8. Appendix II (Symbols)
F
1
= input force (N)
F
2
= output force (N)
v
1
= input particle velocity (m/s)
v
2
= output particle velocity (m/s)
A = area of active electrode of MTSM (m
2
)
k = propagation constant (m
-1
)
d = thickness (m)
Z = acoustic impedance (acoustic ohm)
I = current (C)
C
0
= static capacitance of MTSM (F)
: transformer ratio
9. References
Alig et al, 1996. Ultrasonic shear wave reflection method for measurements of the
viscoelastic properties of polymer films. Rev. Sci. Instrumen. 68. 1536-1542
Ayad M. M., Nagy L. Torad. 2009. Alcohol vapours sensor based on thin polyaniline salt
film and quartz crystal microbalance. Talanta. 78. 1280-1285
Behling C, Ralf Lucklum and Peter Hauptmann. 1997. Possibilities and limitations in
quantitative determination of polymer shear parameters by TSM resonators.
Sensors and Actuators A: Physical V. 61, pp. 260-266
Cho N. J., J. Nelson DAmour, Johan Stalgren, Wolfgang Knoll, Kay Kanazawa, Curtis W.
Frank. 2007. Quartz resonator signatures under Newtonian liquid loading for
initial instrument check. Journal of colloid and Interface Science. 315. 248-254
Cote G. L., R. M. Lec and M. Pishko. 2003. Emerging biomedical sensing technologies and
their applications. IEEE Sensors Journal. 3. 251-266.
Desa J., Zhang Q., Ergezen E., and Lec. R., 2010 Microparticle manipulation on the surface of
a piezoceramic actuator. Measurement Science and Technology. 21 (10) 105803
Dylkov M.S., Sanzharovskii A.T., Zubov P.I. 1966. The effect of thickness on the strength of
polymer films. Mekhanika Polimerov. 2. 940-942
Edvardsson M., Michael Rodahl, Bengt Kasemo, and Fredrik Ho1o1k. 2005. A Dual-
Frequency QCM-D Setup Operating at Elevated Oscillation Amplitudes. Anal.
Chem. 77. 4918-4926.
Ergezen E, Appel M, Shah P, Kresh JY, Lec RM, Wootton DM, Real-Time Monitoring of
Adhesion and Aggregation of Platelets using Thickness Shear Mode (TSM) Sensor,
Biosensors and Bioelectronics, V. 23, #4, pp. 575-82, 2007.
Ergezen E. 2010 Multi-Resonant Thickness Shear Mode (MTSM) Measurement Technique
for Quantitative Characterization of Biological Interfacial Processes Drexel
University, PhD Thesis
Fredriksson, C., S. Kihlman, M. Rodahl, and B. Kasemo. 1998. The piezoelectric quartz
Crystal mass and dissipation sensor: a means of studying cell adhesion. Langmuir,
14, 248-251.
Galipeau D.W., Vetelino J.V., Lec R. M. and Freger C.1991. The Study of Polyimide Film
properties and Adhesion Using a Surface Acoustic Wave Sensors, ANTEC 91,


256
Conference Proceedings, Society of Plastic Engineers and Plastic Engineering, Montreal,
pp. 1679-1984.
Gautam Garaia and B.B. Chaudhurib.2007. Adistributed hierarchical genetic algorithm for
efficient optimization and pattern matching. Pattern Recognition. 40. 212-228
Greczylo T. and Deboswka E. 2005. Finding Viscosity of liquids from Brownian motion at
students laboratory. European Journal of Physics. 26. 827-833
Hanbury C. M., Miller G. W. and Harris B. R. 1996. Antibody characteristics for a continuous
response fiber optic immunosensor for theophylline. Biosensors and Bioelectronics.
11. 1129-1138.
Hong S., Ertan Ergezen, Ryszard Lec, Kenneth A. Barbee. 2006. Real-time analysis of cell
surface adhesive interactions using thickness shear mode resonator. Biomaterials, 27,
5813-5820
Hook F., Larsson C., Fant C., 2002. Biofunctional Surfaces Studied by Quartz Crystal
Microbalance with Dissipation Monitoring. Encyclopedia of Surface and Colloid
Science. 774-790.
Jiang L., Hossenlopp J., Cernosek R., Josse F. 2003. Characterization of Epoxy Resin SU-8
Film Using Thickness-Shear Mode (TSM) Resonator. Proceedings of IEEE
International Frequency Control Symposium. 996-982
Johannsmann D., Ilya Reviakine, Elena Rojas, and Marta Gallego. 2008. Effect of Sample
Heterogeneity on the Interpretation of QCM(-D) Data: Comparison of Combined
Quartz Crystal Microbalance/Atomic Force Microscopy Measurements with Finite
Element Method Modeling. Anal. Chem. 80. 8891-8899.
K. Kanazawa and Nam-Joon Cho. 2009. Quartz Crystal Microbalance as a Sensor to
Characterize Macromolecular Assembly Dynamics. Journal of Sensors.
doi:10.1155/2009/824947
Katz A. and Ward M. D. 1996. Probing solvent dynamics in concentrated polymer films with
a high frequency shear mode quartz resonator. J. Applied Physics. 80. 4153
D. Kroger, A. Katerkamp, R. Renneberg and K. Cammann. 1998. Surface investigations
onthe development of a direct optical immunosensor. Biosens. Bioelectron. 13, 1141
1147.
M. Kunze, Kenneth R. Shull, and Diethelm Johannsmann. 2006. Quartz Crystal
Microbalance Studies of the Contact between Soft, Viscoelastic Solids. Langmuir. 22.
169-173
Kwoun S., 2006. A Multi-Resonant Thickness Shear Mode (MTSM) Sensor for Monitoring
The Formation of Biological Thin Films. PhD Thesis, Drexel University, pp. 125.
Li J, Thielemann C, Reuning U, Johannsmann D. 2004. Monitoring of integrin-mediated
adhesion of human ovarian cancer cells to model protein surfaces by quartz crystal
resonators: evauation in the impedance analysis mode. Biosens Bioelectron. 20. 1333
7.
Liao W., Wei F., Qian X. M., Zhao S. X. 2004. Characterization of protein immobilization on
alkyl monolayer modified silicaon (111) surface. Sensors and Actuators B. 101. 361-
367
Lin Z., Yip M. C., Joseph S. I., Ward D. M. 1993. Operation of an ultrasensitive 30 MHz
quartz crystal microbalance in liquids. Anal. Chem. 65. 1546-1551.
Liu M., Sun J., Bock C., Chen Q. 2009. Thickness dependent mechanical properties of
Polydimethylsiloxane. J. Micromech. Microeng. 19. 1-4


257
Lucklum R. and Hauptman P. 1997. Determination of polymer shear modulus with quartz
crystal resonators. Faraday Discussions. 107. 123-140.
Lucklum R., Peter Hauptmann. 2000. The quartz crystal microbalance: mass sensitivity,
viscoelasticity and acoustic amplification. Sensors and Actuators B 70 _2000. 3036
Luo C., Schneider T., White R., Currie J., Paranjape M. 2003 A simple deflection testing
method to determine Poissions ratio for MEMs applications. J. Micromech.
Microeng. 13. 129-133
Mads Bruun Hovgaard, Mingdong Dong, Daniel Erik Otzen, and Flemming Besenbacher.
2007. Quartz Crystal Microbalance Studies of Multilayer Glucagon Fibrillation at
the Solid-Liquid Interface. Biophysical Journal. 93. 2162-2169
Martin S.J and G.C. Frye. 1991. Polymer film characterization using quartz resonators.
Ultrasonic. Symp. 393-398.
Morrison L.S. and Shin S. U. 1995. Genetically engineered antibodies and their applications
to drug delivery. Advance Drug Delivery Reviews. 15.147-175.
Nam-Joon Cho, Kay K. Kanazawa, Jeffrey S. Glenn, and Curtis W. Frank. 2007. Employing
Two Different Quartz Crystal Microbalance Models To Study Changes in
Viscoelastic Behavior upon Transformation of Lipid Vesicles to a Bilayer on a Gold
Surface. Anal. Chem. 79. 7027-7035.
Ndiritu J.G. and Daniell T. M. 2001. An improved genetic algorithm for rainfall-runoff
model calibration and function optimization. Mathematical and Computer Modeling.
33. 695-706.
Newton M. I., Evans, C.R Simons, Hughes DC. 2007. Semen quality detection using time of
flight and acoustic wave sensors. Applied Physics Letters. 90 (15) 154103
F. N. Nunalee, Kenneth R. Shull, Bruce P. Lee, Phillip B. Messersmith. 2006. Quartz Crystal
Microbalance Studies of Polymer Gels and Solutions in Liquid Environments. Anal.
Chem. 78. 1158-1166.
Roach, P; McHale, G; Evans, CR, Shirtcliffe NJ, Newton MI. 2007. Decoupling of the liquid
response of a superhydrophobic quartz crystal microbalance. Langmuir, 23, 9823-
9830
Rosenbaum J.F. 1998. Bulk acoustic Wave Theory and Devices. Boston, MA: Artech House
Publishers
Saluja A., Kalonia S. D. 2005. Application of Ultrasonic Shear Rheometer to Characterize
Rheological Properties of High Protein Concentration Solutions at Microliter
Volume. Journal of Pharmoceutical Sciences. 94. 1161-1168
Sapper A., Wegener J., Janshoff A., 2006. Cell Motility Probed by Noise Analysis of
Thickness Shear Mode Resonators. Anal. Chem. 78, 5184-5191.
Sivanandam S. N. and Deepa S. N. 2008. Introduction to Genetic Algorithms. Springer Berlin
Heidelberg. New York. 34-35
Song Wang and Atam P. Dhawan. 2008. Shape-based multi-spectral optical image
reconstruction through genetic algorithm based optimization. Computerized Medical
Imaging and Graphics. 32. 429441
Su H and Thompson M. 1996. Rheological and interfacial properties of nucleic acid films
studied by thickness shear mode sensor and network analysis. Can. J. Chem., 74,
344-358


258
Sun K., R. M. Lec, Cairncross R. A., Shah P., Brinker C. J. 2006. Characterization of
Superhydrophobic Materials Using Multiresonance Acoustic Shear Wave Sensors.
IEEE Transactions on Ultrasonics Ferroelectrics and Frequency Control. 53. 1400- 1403
Szabad, Z. Sangolola, B. and McAvoy, B. 2000. Genetic optimisation of manipulation
forces for co-operating robots. IEEE International Conference on Systems, Man, and
Cybernetics. 5. 3336-3341
Urbakh M. and Leonid Daikhin. 2007. Surface morphology and the quartz crystal
microbalance response in liquids. Colloids and Surfaces A. Physicochemical and
Engineering Aspects. 134. 75-84
Voinova M., Jonson M., Kasemo B., 2002. Missing Mass effect in biosensors QCM
Application. Biosensors and Bioelectronics. 17, 835-841
Voros J. 2004. The Density and Refractive Index of Adsorbing Protein Layers. Biophysical
Journal. 87. 553-561
Yang D., Huang C., Lin Y., Tsaid D., Kao L., Chi C. Lin C., 2003 Tracking of secretory
vesicles of PC12 cells by total internal reflection fluorescence microscopy. Journal of
Microscopy, 209, 223-227
Young-Doo Kwon, Soon-Bum Kwon, Seung-Bo Jin and Jae-Yong Kim. 2003. Convergence
enhanced genetic algorithm with successive zooming method for solving
continuous optimization problems. Computers and Structures. 81. 1715-1725
Wegener, J, et al., 2000. Analysis of the composite response of shear wave resonators to the
attachment of mammalian cells. Biophys. J. 78, 28212833.
Werner C. 2008. Interfacial Phenomena of Biomaterials. Polymer Surfaces and Interfaces.
Dresden, Germany. Springer-Verlag Berlin Heidelberg. pp. 299
Westphal S. and Bornmann A. 2002. Bimolecular detection by surface plasmon enhanced
Ellipsometry. Sensors and Actuators B. 84. 278-282
Zhang Q, Desa J., Lec R., Yag G., and Pourrezaei K. 2005, Combination of TSM and AFM for
Investigating an Interfacial Interaction of Particles with Surfaces. Joint IEEE
International Frequency control Symposium (FCS) and Precise Tie and Time Interval
(PTTI) Systems and Applications Meeting. 4490454
Zhuang H., Pin Lu, Siak Piang Lim, and Heow Pueh Lee. 2008. Effects of Interface Slip and
Viscoelasticity on the Dynamic Response of Droplet Quartz Crystal Microbalances.
Anal. Chem. 80. 7347-7353
12
Analysis of Biological Acoustic Waves by
Means of the PhaseSensitivity Technique
Wojciech Michalski
1
, Wojciech Dziewiszek
2
and Marek Bochnia
2

1
Technical University of Wrocaw,
2
Medical University of Wrocaw
Poland
1. Introduction
The analysis of hearing mechanisms and research on the influence of various internal
(pathologies, ageing) and external (trauma, vibration, noise) factors on sound perception are
usually done using acoustic waves induced in the external ear canal. Stimuli which have
been used for this purpose are: clicks, tone bursts, half-sine-waves, single tones or pairs of
tones. The Corti organs responses to the external stimuli have either an electric or acoustic
character. In the former case, these are cochlear microphonics (CMs) picked up from the
surface or from the inside of the cochlea, which are usually used as an indicator of damage
to the organ of Corti in animals. In the latter case, these are acoustic waves that appear in the
external ear canal as a result of stimulation. The acoustic waves have an important clinical
value. Taking into account the presence of nonlinear distortions in the cochlea, the waves
that appear after stimulation with a pair of tones are called distortion product otoacoustic
emissions (DPOAE).
In studies on CMs, the origin of stimulating waves is often a single earphone (controlled by
a generator of defined, often periodical, electrical signals) placed in the external auditory
canal. In studies on DPOAE, a probe with two miniature earphones and one microphone is
placed in the external auditory canal. The earphones are controlled by two generators of
frequencies f
1
and f
2
and the microphone converts the returned DPOAE wave with a
combination frequency, e.g.
3 1 2
2 f f f = , into an electrical signal.
The acoustic wave which induces CM signals is usually a periodic wave, while the waves
inducing DPOAE signals consist of two pure tones. Thanks to the easy access to the
output(s) of the generator(s) the phase-sensitive detection (PSD) technique can be used to
measure both CM and DPOAE signals. Very weak (even below single microvolts) CM and
DPOAE signals originating from the unimpaired cochlea can be measured in this way.
Thanks to this technique signals obscured by other disturbing sources (even thousand times
larger) can be measured accurately. This is possible because the phase-sensitive detector
singles out the input signal with a specific reference frequency while signals with
frequencies different from the reference are rejected. The fundamentals of this technique and
its measuring potential are described in section 2.
In section 3, the authors own experiments aimed at determining the effect various factors
on the electrical function of the Corti organ are described. The factors include: vibration


260
(3.3), ototoxic medicines (3.4) and laser beams used in ear microsurgery (3.6). The role of the
signal phase in the measurements is given special attention.
In section 4, experiments involving acoustic waves being nonlinear products of the Corti
organ are presented. The individual subsections describe the way in which the PSD
technique is applied (4.1), compare the latter with the previously used methods (4.2) and
discuss the authors own experiments in which the phase-sensitive technique is employed to
measure DPOAE signals (4.3) and to measure simultaneously DPOAE signals and CMDP
(cochlear microphonic distortion product) signals (4.4). It is shown that the phase of DPOAE
signals plays an essential role in otoacoustic emission studies.
All the experiments described in sections 3 and 4 were carried out on coloured guinea pigs,
each weighing 500-650 g, being under general ketamine/xylazine (15 mg/kg and 10 mg/kg
body weight, respectively) anaesthesia. A Homoth measuring probe was placed in the
external auditory meatus of the animals. The probe contained two mini earphones and a
standard microphone. Prior to the measurements the probe had been graduated in a
Brel&Kraej artificial ear 4144, using a measuring amplifier 2607 made by the same
company. Permission to carry out the experiments had been given by the Bioethical
Committee in Wrocaw.
Section 5 presents the final conclusions and discusses the future of the phase sensitive
detection technique in investigations into the function of the cochlea exposed to various
hazards.
2. Phase-sensitive detection technique
2.1 Fundamentals
The measurement apparatus based on the phase-sensitive detection technique is called a
lock-in amplifier or a lock-in nanovoltmeter. Lock-in measurements require a frequency
reference which should be strictly connected with a fixed frequency of the function
generator used in the experiment. The reference signal can be either a square wave or a
sinusoid. A block diagram of a typical lock-in amplifier is shown in fig. 1.

INPUT
AMPLIFIER
LOW PASS
FILTER
BANDPASS
FILTER
PHASE-SENSITIVE
DETECTOR
REGULATED
PHASE SHIFTER
REFERENCE
TRIGGER
OUTPUT
AMPLIFIER
Signal input
Reference
input
Output

Fig. 1. Block diagram of lock-in amplifier with single phase-sensitive detector
Let us assume that the input signal can be described as:
( )
1 0
cos ,
sig sig
V A t = + (1)
and the reference signal as:

Analysis of Biological Acoustic Waves by Means of the PhaseSensitivity Technique

261
( )
2 0
cos
ref ref
V A t = + (2)
The two signals have different amplitudes, frequencies and initial phases. At the phase-
sensitive detector inputs there are signals with unchanged frequencies, but with different
amplitudes and phases: ( )
1 1 01
cos
sig sig
V A t = + and
( )
1 2 01
cos .
ref ref
V A t = + The output
of the PSD is simply the product of the two sine waves
( ) ( )
1 1 01 1 2 01
cos cos
PSD sig ref
V A t A t = + +
( ) ( ) ( ) ( )
1 1 1 2 01 01 1 2 01 01
0, 5 cos cos
sig ref
A A t t
( ( ( = + + + + +

(3)
At the output of the PSD there are two signals: a slow-changing signal with differential
frequency ( )
1 2
and a signal with overall frequency ( )
1 2
+ . If the PSD output signals
are passed through a low pass filter, the fast AC signal will be removed. When the
frequencies of the two signals are approximately equal (
1 2
), the filtered PSD output is a
slowly changing DC signal proportional to the signal amplitude and cos(
01
-
01
). When
1 2
= , the filtered signal is exactly a DC signal. By adjusting the phase of the reference
signal one can make (
01
-
01
)

equal to zero, in which case only B
sig
can be measured. This is
true if both initial phases
0
and
0
do not change over time, otherwise cos(
01
-
01
)

will
change over time and
out
V of the lock-in amplifier will not be a DC signal.
The phase dependency of the output voltage of the lock-in amplifier with one PSD unit can
be eliminated by adding a second PSD multiplying the same measured signal by a reference
signal shifted by 90
0
. A block diagram of the lock-in amplifier with a double PSD is shown
in fig. 2.

R

Reference
input
LOW PASS
FILTER
R and
CALCULATOR
Signal input
INPUT
AMPLIFIER
BANDPASS
FILTER
PHASE
SENSITIVE
DETECTOR
PHASE
SHIFTER
REFERENCE
TRIGGER
PHASE
-SENSITIVE
DETECTOR
90

PHASE
SHIFT
LOW PASS
FILTER

Fig. 2. Block diagram of lock-in amplifier with double PSD
The output of the second PSD, filtered by the low pass filter, is proportional to the signal
amplitude and ( )
01 01
sin . Now there are two outputs: ( )
1 01 01
cos
sig
X A = and
( )
1 01 01
sin .
sig
Y A = When
( ) ( ) ( ) ( )
2 2
1 01 01 1 01 01 1
cos sin
sig sig sig
R A A A = + = (4)


262
the phase dependency is removed. Phase difference (
01
-
01
) can be measured according
to
( )
1
01 01
tan Y X

= (5)
The first lock-in amplifiers were based on analogue technology. The measured signal and
the reference were analogue voltage signals and they were multiplied in an analogue PSD.
The results of multiplication were filtered through a multistage RC filter. In such lock-ins
the reference signal phase at the PSD input had to be manually adjusted to the phase of the
measured signal so that ( )
01 01
cos 1. = It was technically difficult to perform
measurements by means of such lock-ins and it was practically impossible to register the
amplitude and phase changes of the measured signals. Digital technology made it possible
to build lock-in amplifiers in which both signal and reference inputs were multiplied and
filtered digitally. Dual phase-sensitive detection eliminated the need for manual phase
adjustments and enabled the simultaneous measurement of signal amplitude and phase.
Such simultaneous measurements can be performed in real time, practically without any
delay to the inducing signal. It also became possible to register short (below 0.1s) and slow
changes in amplitude and phase over time.
2.2 Application of double-phase detection
The PSD technique offers greater measuring possibilities owing to the fact that:
1. the signal fed to the examined object may have various periodical waveforms,
2. the examined object can be linear or nonlinear,
3. the reference signal frequency can be equal to the frequency of the signal being
delivered to the examined object, but it also can be an integral multiplicity (or
submultiplicity) of this frequency.
4. two coherent signals can be introduced to the examined (usually nonlinear) object; the
reference signal can be used at a frequency that is a linear combination of the
frequencies of the inducing signals.
The basic experimental setup is shown in fig.3.
The generator used in the setup has two synchronous outputs. One of them (sync. output)
supplies a TTL signal. Depending on the frequency of the reference signal one can measure
the first harmonic, higher harmonics and subharmonics. In switch position 1 (fig.3), the
reference signal is taken directly from the generators synchronous output whereby the first
harmonic can be measured. In switch position 2, the synchronous signal is multiplied by
integral number n whereby the n-harmonic can be measured. In order to measure the n-
subharmonic the generators synchronic output must be divided by integral number n (the
switch in position 3).
In the simplest case, the waveform of the signal directed to the examined object is
sinusoidal. When the examined object is linear, using the PSD technique one can very
precisely (with an accuracy of 1 nanovolt) measure the electrical response of the object. If
the signal is a simple square wave or another periodical wave with frequency f, the
examined linear object does not change the signal spectrum and the filtered PSD output is
a DC signal proportional to the root mean square (rms) of the first component of the
signal.


263
FUNCTION
GENERATOR

measured signal
amplitude
phase
LOCK-IN
AMPLIFIER
MEASURED
NONLINEAR
OBJECT
MULTIPLIIER DIVIDER
2 3 1
TTL output
switch

Fig. 3. Basic experimental set-up with lock-in amplifier for measuring first harmonic (1),
higher harmonic (2) and subharmonic (3)
The situation becomes more complicated when the examined object is nonlinear. The signal
spectrum at the PSD input differs from the one at the generator output (the nonlinear object
changes the input signal spectrum). This is true for all the signal waveforms, including the
sinusoidal one. Then the lock-in amplifier measures the rms of both the first harmonic and
the n-harmonic (n-subharmonic). When harmonic or subharmonic distortion is measured,
the function generator supplies a pure sinusoidal signal without any harmonics. The basic
experimental setup shown in fig. 3 was used to carry out experiments described in section 3.
2.3 Nonlinear object testing with two synchronous signals
The double phase-sensitive detection technique and modern digital technologies offered
new possibilities of examining nonlinear objects. An example of the measuring systems
which have been developed is shown below (fig. 4). The main component of the setup is a
digital sinus generator of three signals with synchronous frequencies. Two of the signals are
delivered to the examined object while the third one serves as a reference signal. The
frequency of the third signal is a linear combination of the frequencies of the other two
signals.
Let us assume that the output-input function for the examined object can be described by
the formula:

3
,
out in
V C V = (6)
where C is a constant value. The input signal is the sum of signals with different amplitudes,
phases and frequencies and so:
( ) ( )
3
1 1 01 2 2 02
cos cos
out
V C A t A t ( = + + +

(7)
After trigonometric conversions it is possible to receive a signal frequency spectrum at the
objects output. The frequencies, amplitudes and phases of the particular spectral
components (assuming that the examined object does not change its phase relations, i.e. it is
characterized by pure resistances) are shown in table 1.


264

LOCK-IN
AMPLIFIER DIGITAL SINUS
GENERATOR OF THREE
SYNCHRONOUS
FREQUENCIES

signal of f1 frequency
MEASURED
NONLINEAR
OBJECT
reference of f3 =kf1mf2 frequency
complex signal
signal of f2 frequency
amplitude
phase

Fig. 4. Block diagram of experimental setup for more complicated studies of nonlinear
objects

Frequencies of
spectral
components
Amplitudes of
spectral components
Phase of
spectral
components
1
( )
2 2
1 1 2
1, 5 0, 5 A C A A +

01
2
( )
2 2
2 2 1
1, 5 0, 5 A C A A +

02
3
1

3
1
0,75C A 3
01
3
2

3
2
0,75C A 3
02
2
1
+
2
2
1 2
0,75C A A 2
01
+
02

1
+2
2
2
1 2
0,75C A A
01
+ 2
02

2
1
2
2
1 2
0,75C A A 2
01

02

1
2
2
2
1 2
0,75 C A A
01
2
02

Table 1. Exemplary spectrum at output of object with 3rd-order nonlinearity,
tested by pair of pure tones
The amplitude of each of the spectral components can be measured using this technique if a
proper reference signal frequency is selected. The interpretation of phase shifts between the
particular spectral components is much more complicated and requires taking into account
the phase shifts introduced by the examined object. Moreover, the phase shifts introduced
by the examined object may be a function of frequency and so they may be different for each
spectral component. The technique was used to examine nonlinear distortions during the
stimulation of the cochlea by a pair of pure tones. The results of this research are presented
in section 4.
3. Using PSD technique to study cochlear potentials
3.1 Measuring techniques
Cochlear potentials are biopotentials of the inner ear. They are described as electrical signals
arising in response to the acoustic stimulation (usually by a click or a tone) of the organ of


265
Corti. For the first time they were registered by Wever and Bray in 1930 (Wever &Bray,
1930). The discovery of the signals made it much easier to examine the function of the inner
ear and made it possible to assess the impact of various external and internal factors on this
function. It is widely believed that cochlear microphonics (CMs) are generated mainly by
outer hair cells (OHCs). Therefore it seems reasonable to use CMs as an indication of the
OHC function. On the basis of measurements performed over a long period (e.g. a few
weeks or months) one can assess if given hearing damage is temporary or permanent. The
CM signal originating from different places in the human ear (or the animal ear) can be
recorded. In humans CMs are usually picked up from the round window during surgical
procedures performed on patients with various hearing pathologies. There are much fewer
reports describing the reception of CM signals from the promontory or the ear canal near
the eardrum. The past and present studies of the mechano-electrical cochlear function
(based on the reception of CMs) are conducted mainly on animals, using: in vivo
preparations of anaesthetized animals with positive Preyers reflex, in vitro preparations of
the cochlea or in vitro preparations of the hair cells. As regards the research into the impact
of various external and internal factors on the hearing organs, the in vivo studies seem to be
most clinically valuable.
In the 1930s and 1940s CMs were measured at the cochleas round window (Wever & Bray,
1930). In most animals the round window is relatively easily accessible and so measuring
electrodes were usually placed on it or in its direct proximity. In the first years after the
discovery of inner ear potentials, CM signals were measured by a single active probe.
Several years later the first mapping of CMs on the cochlear surface was described
(Thurlow, 1943). It was probably the first attempt ever to place the probe so close to the
source of cochlear microphonics. CM potentials are continued to be measured at the
cochleas round window today (Brown, 2009). This measuring technique was not
abandoned after the introduction of very sensitive (but invasive) procedures (Tasaki et al.,
1952). Tasaki monitored CMs using a pair of active intracochlear electrodes in the basal turn
(one electrode in scala tympani, the other in scala vestibuli). The electrodes were connected
to a balanced differential amplifier. The reference electrode was placed on the neck muscles.
This enables the measurement of the potentials very close to the organ of Corti and
eliminates the auditory nerve potentials. The largest drawback is the mixing of perilymph
and endolymph when the probe is introduced.
A new recording technique has been described by Carricondo (Carricondo at al., 2001). In
this technique, CM potentials are recorded by subcutaneous electrodes in animals or by
surface electrodes in humans. Two active electrodes are placed on the mandibular muscles
while the reference electrode is located on the heads vertex. All the three electrodes are
connected to a differential amplifier. The signal coming from the amplifiers output is
filtered and subsequently averaged through in-phase synchronization with the sound
stimuli.
3.2 History of CM studies in Wroclaw
In 1960 at the Wroclaw University of Technology an oscillograph was built and used as part
of an experimental setup for registering cochlear microphonics (CMs). In the following years
Jankowski and Giedanowski started a series of experiments on animals first on cats, later
on guinea pigs (Jankowski et al., 1962). A Biopotentials Research Workshop was founded,
where biopotentials were measured after damage to the inner ear or the skull, in acoustic


266
trauma, in hypothermia or hypoxia, after the administration of certain drugs, and so on.
Figure 5 shows a schematic of the experimental setup.

GENERATOR
ACOUSTIC
TRANSDUCER
PREAMPLIFIER AMPLIFIER
ANALOGUE
SCOPE
electric and acoustic screen
ear
active electrode

Fig. 5. Schematic of experimental setup used by Jankowski & Giedanowski
The experimental animals (under urethane anaesthesia, which does not diminish CM
voltages) underwent ear surgery: the bones were exposed and drilled until the round
window was revealed. Platinum electrodes (platinum wires 0.1 mm in diameter, coated with
PMMA) were used for CM measurements. The bare end (not coated with PMMA) of the
platinum wire was brought into contact with the round window membrane (without
damaging it). An injection needle was used as the other electrode. It was inserted into the
muscles around the surgical wound. An example of an CM oscillogram from Ziemskis
work (Ziemski, 1970) is shown in fig. 6.

Fig. 6. Example of CM oscillograph record (stimulus tone parameters: f =4096 Hz, acoustic
pressure level - 60dB)
In the middle of the 1990s a new surgical approach, making it possible to expose the whole
cochlea in guinea pigs, was proposed. Skin was cut from the occipital part of the skull,
around the angle of the mandible up to the place of about 0.5 cm above the angle of the
animals snout. The temporal muscles were removed, the mandible partially resected and
the stylo-hyoid muscle cut. Bones were drilled, leaving the tympanic membrane intact. In
the 1990s the present authors used the phase-sensitive detection technique to measure CMs.
A patent was applied for in 1996 and the technique was patented in December 2000 (Patent
PL no. 180060). A report on the use of a similar method was published by Kobayashi
(Kobayashi, 1997). A schematic of the first setup used to measure CMs by the PSD technique
is shown in fig. 7.


267
external ear
reference
SINUS
GENERATOR
MEASUREMENT OF
OF CM POTENTIAL
RMS BY ANALOGUE
LOCK-IN
platinum
electrode
small headphone
cochlea
sine output
REGULATED
RESISTANCE
DAMPER

Fig. 7. Schematic of first setup used to measure CMs by phase-sensitive detection technique
Thanks to the new surgical approach combined with the phase-sensitive detection technique
it became possible to create a map of CM potential amplitude and phase distributions on the
cochlear surface. For this purpose, the active electrode would be fixed in six different points
on the cochleas surface (on the apex, one point at the third turn, two points at the second
turn and two points at the basal turn). When the electrode was fixed in any of the six points,
the frequencies of the stimulating acoustic wave would be successively selected from the
measurable range. Three different acoustic wave pressures (60 dB, 80 dB and 95 dB) would
be set for each of the measuring frequencies. This means that 18 different amplitude-phase
(A-P) values of the CM signal were measured in each of the six points on the cochleas
surface. In this way 18 different A-P distribution patterns would be obtained for the guinea
pig cochlea. Six of them (three for f
.
= 260 Hz and three for f = 8 kHz) are shown in fig. 8.
Phase in each measuring point is related to phase on apex at 60 dB.
It was found that CMs had a different phase and amplitude in the different parts of the
cochlea, which was due to the fact that each of the six measuring points was located at a
different distance from the CM sources inside the organ of Corti.

440 V/+32
o

235 V/+15
o

68 V/0
o

19 V/46
o

9 V/28
o

1,5 V/60
o

31,5 V/+60
o

14,5 V/+32
o

3,2 V/+20
o

f = 260 Hz
115V/+56
o

53 V/+40
o

11 V/+20
o

13,5 V/+4
o
7,4 V/+3
o

1,5 V/-5
o

40 V/-97
o

9,5 V/-34
o

2 V/-17
o

9,5V/-81
o

2 V/-76
o

1,1 V/-84
o

8 V/-3
o

10 V/+12
o

3,8 V/-12
o

4,5 V/-46
o

3 V/-20
o

1,45 V/-53
o

7 V/+4
o

10 V/+60
o

3,7 V/0
o

13 V/-115
o

4,3 V/-122
o

0,9 V/-112
o

21,5 V/-109
o
11 V/-119
o

4,1 V/-116
o

f = 8 kHz
apex apex
reference
phase
reference
phase

Fig. 8. CM potential amplitude and phase in six different points on cochlea surface,
measured at 260 Hz and 8 kHz. At each of six points there are three amplitude/phase
values: upper for 90 dB, central for 80 dB and lower for 60 dB


268
In order to unify CM measurements, two distinctive and universal measurement points
were established: the cochleas apex for the frequencies of 260, 500, 1000 and 2000 Hz and
the cochleas base for 4000 and 8000 Hz. Phase in each measuring point is related to phase
on apex at 60 dB.
3.3 Influence of whole-body vibration on inner ear
Vibration is one of the most widespread injurious factors in the environment of civilized
man (Palmer et al., 2000a, 2000b). The energy absorbed can have a pathological effect on all
the tissues and organs of the body, although the consequences of exposure to vibration do
not present a uniform clinical picture (Jones, 1996; Seidel & Heide, 1986). Because all
machines and vibration devices also produce noise, usually the combined effect of the two
factors is examined (Castelo Branco, 1999). There is a prevalent view that mechanical
vibrations exert only a weak, additionally traumatic influence on the hearing organ (Seidel,
1993). Several experimental investigations into the harmfulness of vibration were carried out
on animals (Hamernik et al., 1980, 1981). Changes in the hearing organ most often would be
found in the hair cells (Rogowski, 1987). This made us undertake our own research in the
1990s. In order to determine the impact of long-term general vibration on the inner ear it
was necessary to: 1) design and built noiseless vibration apparatus, 2) subject several groups
of animals to general vibration (defined by controlled parameters over different periods of
time) and 3) evaluate selected parts of the organ of hearing, using norms based on values
derived from a control group.
In order to ensure proper experimental conditions, i.e. sinusoidal (10 Hz) vertical (5 mm)
shaking, a device consisting of an electric impulse generator, a power amplifier and an
impulse exciter was built (fig. 9). Experiments were carried out on young, coloured guinea
pigs of both sexes weighing 240-360g. Fifty six animals with the normal Preyer reflex and
without otoscopically detectable changes were used. The control group (group m0)
consisted of 20 of the animals and served to establish functional and morphological norms.
In order to avoid changes due to aging being interpreted as the effects of vibration, the
control group was examined after a seven-month stay (6+1 months = duration of the longest
experiment + a rest) in an animal house. The study group consisted of 36 guinea pigs
divided into two subgroups of 18 animals each. Each subgroup was subjected to vibration
over different periods, i.e. 30 (group m1) and 180 (m6) days. These were in fact respectively
22 days (5 days/week, 6 hours/day = 132 hours) and 132 days (792 hours). After the
experiment and a one-month (30 day) rest, the animals which were in good general
condition and without otoscopically detectable changes were qualified for functional and
morphological investigations.
Cochlear microphonics were measured under urethane anaesthesia, using the PSD
technique and the setup schematically shown in fig. 3 (the switch in position 1). CMs were
picked up from the apex of the cochlea for the frequencies of 250 Hz, 500 Hz, 1 kHz and 2
kHz and from the region of the round window for 4 kHz and 8 kHz, using a platinum
needle electrode. For the two study groups and the control group, a total of 6048 data values
were taken for the bilaterally examined pulse wave frequencies (260 Hz-8 kHz) and
intensities (55 dB-95 dB).
The results of the CM measurements were subjected to statistical analysis. The aim was to
find out whether the experiment had any influence on CMs and, if so, what that influence
was. The questions asked were: 1) are there statistically significant differences between the


269
CM voltages obtained from the control groups and the study groups, and 2) are there
statistically significant differences in the CM voltages obtained within the study groups?
The CM values obtained from the healthy animals showed considerable individual
differences, and their distribution showed neither normalcy nor log-normalcy. Therefore all
the experimental samples were examined using non-parametric tests. The K-S Lilliefors test
showed: 1) for control group m0 compared with study groups m1 and m6, a significant
decrease in CMs for the frequencies of 260 Hz, 1 kHz and 2 kHz, and 2) for m1 compared
with m6, a decrease in CM for the frequencies of 260 Hz and 2 kHz. The Kruskall-Wallis test
confirmed the results of the K-S Lilliefors test as regards the location and nature of the
changes.

Fig. 9. Cage with animals exposed to vibrations
The results of the investigations indicated possible greater damage to the hair cells in the
forth and third turnings of the cochlea. Further morphological examinations were needed to
verify this observation. After the bilateral CM measurements the animals were decapitated
and samples were prepared for SEM examinations of the sensorial epithelium. The samples
were examined and photographed using a scanning DSM 950 microscope. The influence of
general vibration on the organ of Corti was assessed on the basis of the condition of the hair
cells, taking into consideration their disorganization, deformation, mutual adhesion and any
reduction in the number of cilia.
SEM examinations were carried out on 20 cochleae from the control group animals and on
all the animals in the two study groups. In the healthy animals, the sensorial epithelium was
found to be normal in every case, but in each of the study groups the above mentioned
damage was observed. It usually occurred in the OHC region of the apex, and its extent
gradually increased in the direction of the cochleas base (up to the second turning). OHC3
was found to be most susceptible to vibratory trauma. Cell damage decreased from the
circumference to the modiolus, and the OHCs showed considerably greater resistance to
vibration (fig.10). Undoubtedly, the observed damage to the sensorial epithelium resulted
from mechanical vibration, and its severity clearly increased with the duration of the


270
experiment. Consequently, the mechanism of deterioration in hearing in all the frequency
ranges (especially at low and average frequencies) in persons subjected to whole-body
vibration could be discovered by analyzing the observed changes.

Fig. 10. Group M6, 4
th
cochlear turning: numerous lesions of hair cells and damage to
Hensens cells
3.4 Studies of gramicidin ototoxicity
Polypeptide antibiotics are used in a variety of clinical situations. Their molecules contain a
specific chain of aminoacids and a non-aminoacidic part (e.g. fatty acids in polymyxins or
glycopeptide in vancomycin). They are generally effective against Gram-positive bacteria,
except for polymyxins which are effective against Gram-negative bacteria. They act by
disrupting the selective permeability of bacterial cellular membranes. Despite their long
history, polymyxins have had a limited clinical use due to the large number of side effects.
Currently they are used primarily for topical treatment (Wadsten at all, 1985).
Since no descriptions of the effects of the systemic administration of gramicidin on the inner
ear could be found in the literature, the authors decided to examine CMs and to compare the
ototoxic effects after the systemic and topical administration of gramicidin. Also the inner
ear of animals which received i.m. injections of gramicidin were examined using a DSM 950
scanning electron microscope (Bredberg at al., 1970; Davis, 1983) .
The research was conducted on 70 young, coloured guinea pigs. All the animals showed the
positive Preyer reflex and no pathologies under otoscopic examinations. The experimental
animals (G) were divided into 5 subgroups, depending on the drug administration mode
and the administered dose. Each experimental subgroup (G1-G5) consisted of 8 randomly
chosen animals. Subgroups G1-G3 received respectively 2, 5 and 10 mg of gramicidin/kg
i.m., once per day, for 14 consecutive days. The animals from subgroups G4 and G5 were
administered a 0.25% and 10% solution of gramicidin suspended on a haemostatic sponge
placed on the round window.
The control group (K) consisted of 30 animals randomly divided into 2 subgroups (K1 and
K2). The animals in control subgroup K1 were injected with normal saline solution once per
day for 14 consecutive days. The animals in subgroup K2 were administered normal saline


271
solution placed on the round window. One day after the last injection (the 15
th
day of the
study) electrophysiological measurements were carried out on the animals in subgroups G1-
G3 and K1. Then their cochleae were removed for SEM examinations. In the case of the
animals belonging to subgroups G4, G5 and K2, CM measurements were performed after
removing the haemostatic sponge from both ears and allowing the round windows with
their surroundings to dry (Gale & Ashmore, 1977).
Cochlear microphonics (CMs) were investigated under urethane anaesthesia, using the PSD
technique and the setup schematically shown in fig. 3 (the switch in position 1). CMs were
picked up from the apex of the cochlea for the frequencies of 260 Hz, 500 Hz, 1 kHz and 2
kHz and from the region of the round window for 4 kHz and 8 kHz by means of a platinum
needle electrode. As regards study subgroups G1-G5 and control subgroups K1 and K2, a
total of 7560 data values were taken for the examined frequencies (260 Hz-8 kHz) and
intensities (55 dB-95 dB). The results of the CM measurements were subjected to statistical
analysis (the t-Student test).
Gramicidin administered systemically in a dose of 2 mg/kg led to a significant (38%) decline
in CM voltage in K1 subgroup animals for the frequencies of 260 Hz and 2 kHz. For the
other frequencies the drop in CMs amounted to about 15%, except for the 4 kHz at which a
slight improvement was observed for sound levels between 55 and 70 dB. A significant drop
in CMs was observed in subgroup G2 at 2 kHz and sound levels above 70 dB. At 95 dB the
decline in CMs was 30% larger than in the G1 animals. The changes in the G2 animals
relative to G1 were even more significant at 500 Hz, 1 kHz and 8 kHz. The animals receiving
10 mg/kg of gramicidin showed lower CMs than the ones registered in all the examined
frequency ranges for control subgroup K1. The largest drop was registered at 2 kHz (31%
lower than in the K1 control subgroup). The smallest changes were observed at 8 kHz. In
subgroups G1-G3, the largest differences in CMs were observed at 4 kHz for all the sound
levels.

Fig. 11. Group K1, 2nd cochlear turn: unchanged sensory epithelium


272
In the animals receiving topical 0.25% gramicidin solution (G4), a significant drop in CMs
(in comparison with control K2) was observed at 1 kHz and 2 kHz. In group G5 (where the
animals were administered 10% gramicid in solution on the round window) a drop in CMs
was observed also at 4 kHz and 8 kHz. At low sound levels the largest falls in CMs were
observed in subgroup G4.
In the G1 and G2 animals no damage to the sensory epithelium was found under SEM. The
destruction of cochlear hair cells occurred in the G3 animals. The changes were most visible
in OHC3 cells in the cochleas third turning.
To sum up, the systemic administration of gramicidin leads to greater disruptions of the
bioelectric functions of the inner ear than local, topical administration (Linder at al., 1995).

Fig. 12. Group G3, 3
rd
cochlear turn: numerous lesions in OHC3 cells and structural changes
in cilia
3.5 CM amplitude and phase changes caused by changes in intensity of stimulating
acoustic wave
Another important improvement in CM measurement came with the introduction of a lock-
in amplifier with double phase-sensitive detection. In December 2003 a device for the phase-
sensitive measurement of inner cochlea microphonic potentials was registered at the Patent
Office. It was patented in November 2010. The device can measure harmonic, subharmonic
and linear distortion products of the cochlea after dual-tone stimulation. Figure 13 shows a
schematic of the measuring device.
The amplitude and phase of CMs in a given point on the surface of the cochlea depend on
the intensity (L) and frequency (f) of the sound. When the frequency is fixed, the two CM
potential parameters (amplitude and phase) depend on only parameter L. Typical changes
in amplitude and phase over time registered at two different acoustic wave frequencies (260
and 8000 Hz) for the same guinea pig are shown in fig. 14. For this data, graphs of CM
potential rms and phase depending on the level of sound intensity are shown in Fig. 15.


273
external ear
platinum
electrode

sine output REGULATED
AMPLIFIER
STANFORT
LOCK-IN SR830
small headphone
cochlea
INTERFACE
COMPUTER
RECORDING
rms of CM
phase of CM

Fig. 13. Experimental setup for measuring CM potentials

recording time in seconds
-35
-30
-25
-20
-15
-10
0 100 200 300 400 500 600
-35
-30
-25
-20
-15
-10
-5
0
5
10
15
20
25
30
35
Phase
95dB
90 dB
55dB
60dB
65dB
70dB
75 dB
80 dB
85 dB
90dB
95 dB
-5
0
5
10
15
20
25
30
35

C
M

r
m
s

[
V
]

C
M

p
h
a
s
e

[
d
e
g
]

phase
0 100 200 300 400 500 600
-250
-200
-150
-100
-50
0
50
100
150
200
250
-250
-200
-150
-100
-50
0
50
100
150
200
250
frequency of exciting
acoustic wave - 260 Hz
95 dB
55dB
60 dB
65 dB 70 dB
75 dB
80 dB
85 dB
90 dB
95 dB
recording time in seconds
phase
frequency of exciting
acoustic wave 8 kHz
rms rms

C
M

r
m
s

[
V
]

C
M

p
h
a
s
e

[
d
e
g
]

Fig. 14. Exemplary changes in CM rms and phase depending on sound intensity (sound
levels were changed by 5 dB every 50 seconds)
Cochlear microphonic potentials are believed to be generated by the outer hair cells (OHCs).
The latter are situated in three rows on the basilar membrane. All the OHCs have tiny
strands (numbering about a hundred) called stereocillia. The apex of each single
stereocillium lies in the tectorial membrane. In the resting state the stereocillia of each single
cell form a conical bundle. During the acoustic excitation of the cochlea the stereocillia may
dance about wildly. This alternating motion causes the channels in the stereocillia to open
and close, providing a route for the influx of K
+
ions. The upper part of the OHCs acts as a
resistor whose resistance changes according to the mechanical movements of the stereocillia.
Changes in this resistance cause changes in extra-cellular currents. The measured CM
potential is the result of the flow of extra-cellular currents through the input resistance of the
lock-in amplifier.
The place theory suggests that a tone of a defined frequency excites mainly the OHCs
located on the basilar membrane in a place specific for the given frequency (CF). The OHC
electrical activity picked up from a given place on the cochlea surface is the vector sum of
the extra-cellular currents generated by the particular OHC cells belonging to the given CF
area (probably oval in shape). As the excitation wave intensity increases, extra-cellular


274
currents are generated by an increasing number of OHC cells within the same CF area,
which results in an increase in CM amplitudes. The phase changes registered then probably
correspond to the shifts of the centre of the extra-cellular currents within the CF area.

55 60 65 70 75 80 85 90 95
0
50
100
150
200
250

C
M

r
m
s

[
V
]
A
55 60 65 70 75 80 85 90 95
0
10
20
30
40
50
55 60 65 70 75 80 85 90 95
-10
0
10
20
30
40
50
60

C
M

p
h
a
s
e

[
d
e
g
]
tone intensity [dB]
f = 260 Hz f = 8 kHz
55 60 65 70 75 80 85 90 95
-5
0
5
10
15
20
25
30
35
tone intensity [dB]

Fig. 15. Output-input characteristic obtained from traces shown in Fig. 14
3.6 Changes in amplitude and phase of CM potentials as result of laser irradiation
A focused laser beam can be a precise surgical scalpel. Perkins was the first to describe the
use of a laser (an argon laser to be precise) in the surgical treatment of otosclerosis (Perkins,
1980). Since that time several kinds of laser (Ar, KTP, CO
2
, Er) have been used in ear
microsurgery. Vollrath and Schreiner were the first to use the rms of cochlear microphonics
to estimate the effect of the argon laser beam on the electrical response of the cochlea in
guinea pigs (Vollrath & Schreiner, 1982). The PSD technique enables the recording of the
simultaneous changes in amplitude and phase of the CM potential during laser irradiation.
The information about cochlear activity acquired in this way is more detailed.
Studies of the effect of Ar laser irradiation on the electrical activity of the cochlea have been
described by us in several papers. We used the double PSD technique to record CM
potentials prior to, during and after argon laser irradiation of the cochlea in guinea pigs. The
goal of the studies was to determine safe laser parameters for argon laser stapedotomy,
taking into account changes in not only the rms of CM potentials but also in their phase. In
our experiments we used a CW argon laser with adjusted output power (0.1 3.0 W). An
electronically controlled mechanical chopper was used to obtain laser light pulses differing
in their parameters (the duration of a single laser pulse, the time interval between the
successive pulses, the number of pulses in a series). Via a 200 m optical lightguide the laser
pulses would be delivered to the cochlear bone (near the round window) of an
anaesthetized guinea pig with the surgically opened bulla. Exemplary traces selected from
many different recordings are shown in fig. 16.


275
recording time [s]
-60
-40
-20
20
40
60
80
100
120
140
160
0 60 120 180 240 300
0
p
h
a
s
e

[
d
e
g
]

r
m
s

[
V
]

rms of CM
phase of CM
p
h
a
s
e

[
d
e
g
]

r
m
s

[
V
]

-80
-60
-40
-20
20
40
60
80
100
120
140
160
180
0 60 120 180 240 300
0
rms of CM
phase of CM
1
2
3 4
recording time [s]
5

Fig. 16. Changes in rms and phase of CM potentials evoked by 80 dB acoustic wave of 1 kHz
frequency during Ar laser pulse irradiation of 0.27 W (left) and 0.48 W (right) peak power.
Irradiation parameters: 1 single pulse of 0.5 s duration, 2 - single pulse of 0.5 s duration, 3 -
single pulse of 1 s duration, 4 two pulses of 1s duration with 1s interval between them, 5
single pulse of 0.5s duration
It was found that in each registration the phase and amplitude of CM potentials changed
during a laser pulse. The characteristic of the phase changes is always the same and
diminishes relative to the initial (prior-to-irradiation) phase (in fig. 16 the initial phase was
assumed to be equal to -30
0
). The character of changes in CM rms depends on the peak
power of the pulses used. Two characteristic peak power levels: P
1
and P
2
can be
distinguished. When the peak power of the pulses is lower than P
1
, laser irradiation results
in a small increase in CM rms. This may be due to the slight increase in the temperature of
the cochlea and to a biostimulating effect. After the first peak, but still below the second one
(P
2
), a sharp drop (even down to zero) in CM rms occurs. The drop is temporary and the
cochlea quickly recovers its initial activity. Beyond P
2
, changes in the electrophysiological
activity of the cochlea are irreversible. As for today, the observed changes in the phase of
CM potentials are hard to explain. It remains unknown why low-level laser radiation
activates other groups of OHC cells in the CF area.
4. Double PSD technique in studies of DPOAE
4.1 Evoked otoacoustic emission
Evoked otoacoustic emissions (EOAE) are acoustic waves present in the external auditory
canal after the cochlea is stimulated with an acoustic excitation wave. Depending on the
excitation, different kinds of emission can be distinguished. If the stimulating signal is
constant, then the emission is called simultaneous evoked otoacoustic emission (SEOAE).
When pulse stimulating (clicks) sounds are used and the emission is registered between
the clicks, the emission is called transiently evoked otoacoustic emission (TEOAE). If dual-
tone stimulation (by two sinusoidal waves with respectively frequencies f
1
and f
2
and
levels L
1
and L
2
) is used, then the emission is called distortion product otoacoustic emission
(DPOAE).


276
Otoacoustic emission was predicted by Gold as early as in 1948 (Gold, 1948). Thirty years
later Kemp published a paper in which he described experiments proving the existence of
this phenomenon (Kemp, 1978). He used clicks of 0.2 ms duration at a repetition rate of
16/s. In-between the successive pulses he recorded (with an electret microphone) acoustic
wave pressure fluctuations at the outlet of the external acoustic canal. By applying an
averaging procedure to the two-minute recordings he was able reduce the noise level to
0 dB SPL and reveal the backward signal which originated from the cochlea stimulated by
the click. A few hundreds of works on this subject have been published since the first paper
by Kemp. New experimental data are reported but their interpretations are not always
explicit and mutually consistent. Despite the fact that the DPOAE mechanism is not yet fully
understood, DPOAE signal estimation is a method of testing the human peripheral auditory
function. The method is widely used in newborn hearing screening tests.
The presence of components which are absent in the stimulating acoustic wave is distinctive
of DPOAE. The components result from the mechanical activity of the organ of Corti and are
transmitted in the reverse direction through the middle ear and the tympanic membrane.
Among the few possible products of cochlear nonlinearity, the acoustic wave f
3
=2 f
1
f
2
is
most widely examined because of its highest acoustic pressure level.
All the DPOAE acoustic waves are studied after their transduction into electric signals by a
microphone. The microphone must be of high sensitivity and with a linear dynamic reserve
(about 80 dB). The same requirements apply to the input preamplifier and the lock-in
voltmeter amplifier since the measured DPOAE electrical signals cannot result from
measuring system nonlinearity. The microphone placed in the external auditory canal
transduces acoustic waves into electrical signals: both primary tones of 60-70 dB and reverse
DPOAEs of 0 20 dB. Also floor noise occurs in the external ear canal. The apparatus used
for measuring DPOAE must eliminate all undesirable signals with frequencies different than
the frequency of the signal to be measured.
Otoemissions are examined after they have been converted in very accurate electric
microphones. The biggest problem faced when examining DPOAEs is their extremely low
level in comparison with the excitation waves. The difference may reach 30-60 dB. The
phase-sensitive detection of DPOAE is therefore very useful. A basic experimental setup for
measuring DPOAE signals is shown in fig. 17.

GENERATOR OF
THREE SYNCHRONIC
SINUSOIDAL SIGNALS
PC
LOCK-IN
AMPLIFIER
reference
rms
phase
DPOAE
f2
f1
f3 =2f1-f2
earphone 1
earphone 2
microphone

Fig. 17. Basic experimental setup for measuring DPOAE signals, using double PSD technique


277
The main unit of the experimental setup is a generator of three synchronous sinusoidal
signals. The three frequencies are synchronized by a 18 MHz clock, whereby weak DPOAE
signals can be measured using the PSD technique. The DPOAE response is measured by
means of a probe which contains two miniature earphones and a low noise microphone.
From the generator, pure tones with frequencies f
1
and f
2
are fed to the earphones. The two
primary tones are digitally synthesized. The amplitudes, phases and frequencies of the tones
are regulated by a PC with dedicated software. The software also enables the acquisition of
the amplitude and phase of the DPOAE signals during measurements.
4.2 Previous techniques of measuring DPOAE
As mentioned earlier, DPOAE signals are of very low level, even if they are evoked in an
unimpaired ear. Several signal processing techniques for the measurement of DPOAE
signals under a large amount of noise and primaries of 60-70 dB have been developed.
Initially, the Fast Fourier Transform (FFT) was used as the main signal processing tool for
improving the signal-to-noise ratio in order to better estimate the level of DPOAE signals. In
this method, the signals are first divided into data blocks and then averaged over time. For
better reduction of the overall background noise long measurement time is required, which
increases the amount of recorded data to be averaged. The FFT method requires about 10
seconds of block data. During long DPOAE recording, transient artefacts (e.g. talking, head
movements) may occur, which when averaged together with the measuring signal may
degrade the accuracy of the signal. Besides, the averaging method is incapable of measuring
rapid changes of DPOAE signals.
In the first decade of the 21st century several novel methods of measuring DPOAE signals
were developed (Ziarani & Konrad, 2004; Li et al., 2003). In comparison with the
conventional methods, the new methods offer a shorter measurement time, which is of
significance for clinical examinations. In addition, these methods are more immune to
artefact and background noise. Thanks to the new methods it is possible to continuously
record DPOAE signals. Besides offering the above advantages, the double PSD technique
enables the simultaneous measurement of amplitude and phase of DPOAE signals. The two
DPOAE parameters can be measured in a very short time, even below 10 ms.
In screening protocols typically a few pairs of primary tones with fixed acoustic levels are
used and the responses are analyzed one after another (sequentially). In order to reduce the
examination time the multiple-tone pairs method can be employed. In this method, DPOAE
signals are evoked simultaneously by three or four pairs of two-tones. This method reduces
measurement time but has a limited use .
4.3 DPOAE measurement using double PSD technique
Before the PSD technique was introduced to measure DPOAE signals it had been assumed
that the amplitude and frequency of DPOAE signals depended on four acoustic parameters
of the stimulating signals (primaries), i.e. the amplitude and frequency of each of the two
signals. The DPOAE phenomenon itself is investigated according to the procedure described
below. First the f
2
/f
1
ratio (usually 1.22) and the stimulating signal intensity levels (e.g. L
2
/L
1

= 60dB/65dB) are fixed. For the frequency of one of the stimulating waves (usually f
2
)
several discrete values are set while the frequency of the other wave is changed in small
steps, e.g. 1/3 octave-bands centred around the fixed f
2
(Wagner at al., 2008). When
examining the effect of different internal (e.g. age, gender) and external (e.g. industrial


278
noise, medicines) factors, DPOAE is measured in the same stimulation conditions before
and after the stimulus acts. Most experimental works in this field describe measurements of
solely the amplitude of DPOAE signals. Some works also dealt with the phase of DPOAE
signals, but it was measured in an indirect way, using signal processing methods. The
phase-sensitive technique enables the simultaneous measurement of the amplitude and
phase of DPOAE signals, with no need to use complex signal processing methods. The
measurement takes place in real time.
Figure 18 shows an exemplary record of the simultaneous changes in the amplitude and
phase of DPOAE signals caused by changes in the acoustic parameters of the stimulating
waves. The recording was made in real time using the measuring setup shown in fig. 17. In
the whole course of recording the combination frequency (f
3
) remained constant at 3749 Hz
while the other parameters were changed every 20 seconds in a specified sequence. The
whole 980 second long recording time had been divided into seven 140 long time intervals
in which parameter k = f
2
/f
1
assumed the consecutive values: 1.10, 1.15, 1.20, 1.25, 1.39, 1.35,
1.40. The following seven combinations of stimulating wave levels: 1 - (55 dB, 55 dB), 2 - (55
dB, 60 dB), 3 - (60 dB, 55 dB), 4 - (60 dB, 60 dB), 5 - (65 dB, 60 dB), 6 - (60 dB, 65 dB) and 7-
(65 dB, 65 dB) were fixed for each value of parameter k. In each of the combinations, the dB
SPL of primary f
1
is in the first place. During the 980 second long recording the parameters
of the primaries were changed 49 times in total.

0 140 280 420 560 700 840 980
-200
-160
-120
-80
-40
0
40
80
120
160
200
recording time [s]
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
0 140 280 420 560 700 840 980
0
10
20
30
40
50
60
70
80
k=1,10
k = 1,20
k=1,15
k = 1,25 k = 1,30
k = 1,35
k = 1,40

D
P
O
A
E

p
h
a
s
e

d
e
g
]

D
P
O
A
E

r
m
s

[
V
]

Fig. 18. Simultaneous changes in rms and phase of DPOAE signals, caused by fixed
sequence of changes in parameters of primaries. Numbers 1 7 denote following
combinations of primary frequencies
1dB 2dB
/ L L levels: 1 - (55 dB, 55 dB), 2 - (55 dB, 60 dB), 3
- (60 dB, 55 dB), 4 - (60 dB, 60 dB), 5 - (65 dB, 60 dB), 6 - (60 dB, 65 dB) and 7 - (65 dB, 65 dB).
The same combinations of levels were used for each value of parameter k=f
2
/f
1


279
The measurements showed that each change in the value of one of the parameters of the
primaries results in a change of both the amplitude and phase of the DPOAE signal.
Moreover, the character of the changes depends on the on the ontogenetic traits.
Thanks to the phase-sensitive technique one can determine the effect of the initial phase of
each of the primaries on the amplitude and phase of DPOAE signals. For this purpose a
generator of three synchronous sinusoidal signals was incorporated into the setup shown in
fig. 17. The generator offers the possibility of fixing not only the amplitude and frequency of
each of the primaries, but also the initial phase of each of the signals.
Five parameters of the primaries, i.e. the amplitude and frequency of each of the signals and
the initial phase of one of the signals were fixed. The sixth parameter, i.e. the initial phase of
the second primary was changed in a range of 0 360 degrees. The phase was changed in
steps of 22.5 degrees. Exemplary measurements are shown in figs 19 and 21. Each of the
figures comprises four panels. On the left side of each of the figures there are two panels
showing experimentally determined changes in the amplitude and phase of DOPAE signals,
caused by changes in the initial phase of primary f
1
(fig.19) or primary f
2
(fig(21). The data
were obtained for combination frequency f
3
= 3749 Hz, parameter k = f
2
/f
1
= 1.25, intensity
levels L
1
/L
2
= 65/55dB and the zero initial phase of primary f
1
(fig.19) or f
2
(fig.21).
The graphs in the panels on the right side of each of the figures were plotted on the basis of
formulas (8) (10), but the values of some of the constants in the formulas were matched to
obtain agreement with the experimental traces.

Fig. 19. Simultaneous changes in amplitude (upper panels) and phase (lower panels) of
DPOAE signals, caused by changes in initial phase of primary f
1
, obtained from experiment
(left) and theoretically (right) (details in text)
0 30 60 90 120 150 180 210 240 270 300 330 360
-180
-150
-120
-90
-60
-30
0
30
60
90
120
150
180
D
P
O
A
E

p
h
a
s
e

[
d
e
g
]
phase of f
1
primary [deg]
0 30 60 90 120 150 180 210 240 270 300 330 360
0
2
4
6
8
10
12
14
16
18
20
22
24
D
P
O
A
E

r
m
s

[
u
V
]
0 30 60 90 120 150 180 210 240 270 300 330 360
-180
-150
-120
-90
-60
-30
0
30
60
90
120
150
180
D
P
O
A
E

p
h
a
s
e

[
d
e
g
]
phase of f
1
primary [deg]
0 30 60 90 120 150 180 210 240 270 300 330 360
0
2
4
6
8
10
12
14
16
18
20
22
24
D
P
O
A
E

r
m
s

[
u
V
]


280
Currently, it is generally believed that the DPOAE signal induced in the external acoustic
canal by a double-tone is composed of two backward travelling waves (e.g. Knight
&Kemp, 2000). The primary wave arises in the place where the two regions (CF
1
and CF
2
)
characteristic of frequency f
1
and f
2
overlap (but much more closer to CF
2
). The wave
propagates in the basilar membrane towards both the cochleas base and its apex . The
wave directed towards the apex bounces off in the region characteristic of frequency f
3
(CF
3
)
and propagates towards the base. Thus two waves with the same frequency f
3
, but shifted in
phase relative to each other, propagate towards the cochleas base. Depending on the
difference between the two waves, destructive or constructive amplitude interference
occurs.
There are two different theories in the literature, concerning how the waves propagate
backward from their generation places (He at all, 2007). According to one theory, the two
waves propagate as compression waves to the cochlear base via the cochlear fluids.
According to another theory, the two waves are transverse waves slowly propagating along
the basilar membrane. Currently the prevailing view is that two backward waves, being
transverse waves in the basilar membrane, arise in the cochlea excited by two tones. Taking
into consideration this view and the previously determined sites where the backward waves
arise, the schematic shown below (fig. 20) was drawn.

f2 f1
f
3
=2f
1
-f
2
basilar membrane
apex
region of
overlap
distortion component
reflection component
oval
window

Fig. 20. Schematic diagram of source of two backward travelling waves whose interference
produces DPOAE wave in auditory canal
The resultant wave near the oval window can be written as
( ) ( )
3 10 1 2
cos 2
w m
A t A t = + + + , (8)
where

2
1
1 2 cos ,
m m r
A A K K = + + (9)

1
sin
atan
cos
r
r
K

| |
=
|
+
\ .
(10)
A
1m
the amplitude of the primary wave,
K = A
2m
/A
1m
,
10

- the initial phase of the primary wave,


281
1 2
, - the phases induced by the initial phases of the primaries,
r
- a phase difference between the primary and secondary wave, due to the path length
distance.
It follows from formula (9) that the amplitude of the resultant wave does not depend on the
the initial phases of the primaries, and the phase of the resultant wave:

10 1 2
2 = + + (11)
(directly measured by the lock-in amplifier) is a linear function of the stimulating waves
phase. However, experimental results do not corroborate the above dependence. The
amplitude of the DPOAE signal turns out to be a function of the initial phases of the
stimulating signals, and the measured phase is only approximately a linear function of the
initial phases (panels on the left side of figs 19 and 21). If it is assumed that angle
r
changes
in the same way as the initial phase of the primary, full agreement between the experimental
traces and the ones determined from formulas (8) and (10) is obtained. This is shown in the
panels on the right side of figs 19 and 21. The graphs on the right side of fig. 19 were plotted
on the basis of formulas (8) and (10), assuming
1
2 ,
r
= while the graphs on the right side
of fig. 21 were plotted assuming
2
.
r
=

Fig. 21. Simultaneous changes in amplitude (upper panels) and phase (lower panels) of
DPOAE signals, caused by changes in initial phase of primary f
2
, obtained in experiment
(left) and theoretically (right) (details in text)
0 30 60 90 120 150 180 210 240 270 300 330 360
-180
-150
-120
-90
-60
-30
0
30
60
90
120
150
180
D
P
O
A
E

p
h
a
s
e

[
d
e
g
]
negative phase of f
2
primary [deg]
0 30 60 90 120 150 180 210 240 270 300 330 360
0
2
4
6
8
10
12
14
16
18
20
22
24

D
P
O
A
E

r
m
s

[
u
V
]
0 30 60 90 120 150 180 210 240 270 300 330 360
-180
-150
-120
-90
-60
-30
0
30
60
90
120
150
180
D
P
O
A
E

p
h
a
s
e

[
d
e
g
]
phase of f
2
primary [deg]
0 30 60 90 120 150 180 210 240 270 300 330 360
0
2
4
6
8
10
12
14
16
18
20
22
24

D
P
O
A
E

r
m
s

[
u
V
]


282
It follows from formula (9) that the ratio of the maximum amplitude of the DPOAE wave
(A
wmax
) to the minimum value (A
wmin
) amounts to

max 2 max min
min 1 max min
1
1
w m w w
w m w w
A K A A A
K
A K A A A
+
= = =
+
(12)
For the measurement conditions for which the traces shown in figs 19 and 21 were
determined, it was calculated from formula (12) that K = 0.34 and K = 0.27 when
respectively initial signal phase f
1
and f
2
is changed. This means that about 11.5% and 7.3%
of the primary wave energy is reflected from region CF
3
in respectively the former and
latter case.
The preliminary measurements shows that there is a certain mechanism in the Corti organ,
which is responsible for the fact that a change in the phase of one of the stimulating signals
(i.e. phase modulation) causes the amplitude modulation of the DPOAE signal. Further
research is needed to explore this mechanism, but already at this stage one can say that the
PSD technique proposed by the authors will play a major role in the exploration of this
mechanism.
4.4 Simultaneous measurements of DPOAE and CMDP, using double PSD technique
Much of the experimental research reported in the world literature indicates that the main
source of CM signals and DPOAE waves are OHCs. Thanks to the use of phase-sensitive
detection in the measurement of each of the signals one can observe the simultaneous
changes in the amplitude and phase of the two signals, resulting from changes in the
parameters of the primaries. The measuring setup used for this purpose is shown in fig. 22.
The setup incorporates two patents developed by the authors.

INTERFACE
LOCK-IN
AMPLIFIER
No. 1
reference
rms
DPOAE
f
2
f
1
earphone
earphone
microphone
LOCK-IN
AMPLIFIER
No. 2
platinum
electrode
cochlea
GENERATOR OF
THREE SYNCHRONIC
SINUSOIDAL SIGNALS
PC
phase rms
phase f
3
=2f
1
-f
2
CMDP

Fig. 22. Experimental setup for simultaneous measurement of amplitude and phase of
DPOAE and CMDP signals


283
Two lock-in amplifiers, one for measuring the rms and phase of the DPOAE signal
(amplifier No. 2) and the other for measuring the rms and phase of the CMDP signal
(amplifier No. 1), have been incorporated into the setup. The CMDP signal is the distortion
product in cochlear microphonics. The same signal (with combination frequency f
3
) from the
generator is fed to the reference inputs of each of the amplifiers. The reference input of lock-
in No.2 can also be successively fed signals with stimulation frequencies f
1
and f
2
and
combination frequency f
3
and the rms and phase of three CM signals with different
frequencies can be measured. Measurements made in this way may provide a fuller picture
of the cochlea functions.
The above setup was used to measure changes in the rms and phase of both DPOAE and
CMDP signals, caused by changes in the excitation parameters. Four anaesthetized guinea
pigs with the positive Preyer reflex were subjected to the experiments. Recordings were
made at the following combination frequencies: 1312, 1875, 2671, 3749 and 5342 Hz. The
parameters of the primaries were changed as in sect. 4.3, i.e. one of the parameters of the
primaries was changed every 20 seconds in a specified way. Exemplary traces recorded for
frequency f
3
= 1875 Hz are shown in fig. 23.

Fig. 23. Simultaneous changes in rms and phase of both DPOAE and CMDP, induced by
changes in parameters of primaries
980
0
140 280 420 560 700 840
-210
-140
-70
0
70
140
210 0
10
20
30
40
50 -210
-140
-70
0
70
140
210 -20
-10
0
10
20
p
h
a
s
e

o
f

C
M

[
d
e
g
]
recording time [s]
k=1,10 k=1,15 k=1,20 k=1,25 k=1,30
r
m
s

o
f

C
M

[
u
V
]
p
h
a
s
e

o
f

D
P
O
A
E

[
d
e
g
]
k=1,35 k=1,40
r
m
s

o
f

D
P
O
A
E

[
d
B
]
1 3 5 7
2 4 6
1 3 5 7
2 4 6
1 3 5 7
2 4 6
1 3 5 7
2 4 6
1 3 5 7
2 4 6
1 3 5 7
2 4 6
1 3 5 7
2 4 6


284
Nearly 100% correlation between the DPOAE rms and the CMDP rms was found, i.e. when
after a change in one of the parameters of the primaries the DPOAE rms increased, then
CMDP rms would also increase. Unfortunately, there was no such correlation between the
phases of the two signals. This situations is well illustrated by the records of the changes,
shown in fig. 23.
5. Conclusion
Practically all the ways of measuring biological acoustic waves in the cochlea, in which the
phase-sensitive detection technique can be applied, have been described. Exemplary
experimental results coming from many different measurement cycles carried out by the
authors on guinea pigs in the last nearly 20 years were presented to demonstrate the
measuring possibilities offered by the PSD technique. The latters main advantage is that
very weak (even below the ambient noise level) electrical signals can be measured in a very
short time (in the order of milliseconds). A minor limitation of this technique is that it is
applicable to objects to whose input periodical signals are fed from the outside.
In many investigations into the electrophysiological function of the cochlea it is essential not
only to simultaneously measure the amplitude response, but also the phase response to the
stimulation. This is undoubtedly another advantage of the PSD technique.
Much more difficult than the measurement of the phase is the interpretation of its changes.
As for now, it is not always possible to interpret the observed changes, which particularly
applies to DPOAE. This phenomenon has been known for over 30 years, but it still has not
been fully explored. The great worldwide interest in this subject is reflected in the large
number of publications devoted to it. The interest stems from the fact that for many years
DPOAE measurements have been part of hearing screening tests during which DP-grams
are recorded. This especially applies to newborns and people with mental disabilities, in
which cases it is impossible to record audiograms. Besides gaining an insight into the nature
of the DPOAE phenomenon, it is essential to determine the correlation between the DP-
gram and the audiogram. In the authors opinion, the phase-sensitive detection technique
represents a new tool for investigating electrophysiological phenomena in the cochlea and it
will contribute to the better understanding of the phenomena taking place in this organ.
6. References
Bredberg, G., Lindeman, H. H., Ades H. W., West R. & Engstrom H. (1970). Scanning
electron microscopy of the organ of Corti. Science, Vol. 170, No.960, pp. 861-863,
ISSN Print 0036-8075
Brown, D.J., Hartsock, J. J., Gill, R.M., Fitzgerald, H.E. & Salt, A.N. (2009). Estimating the
operating point of the cochlear transducer using low-frequency biased distortion
products. J. Acoust. Soc. Am., Vol. 126, No. 4, pp.2129 2145, ISSN Print 0001-4966
Carricondo F., Sanjuan-Juaristi J., Gil-Loyzaga P. & Poch-Brotto Joaquin. (2001). Cochlear
microphonic potentials: a new recording technique. The Annals of otology, Rhinology
& Laryngology, Vol.110, No. 6, pp. 565 573, ISSN Print 0003-4894
Castelo Branco (1999). The clinical stages of vibroacoustic disease. Aviation, Space, and
Environmental Medicine , Vol. 70, Suppl 3, A32-39, ISSN 0095-6562
Davis, H. (1983). An active process in cochlear mechanics. Hearing Research, Vol. 9, No,1, pp.
79-90, ISSN Print 0378-5955


285
Gale, J. A. & Ashmore J. F. (1997). An intrinsic frequency limit to the cochlear amplifier.
Nature, Vol. 389, No.6646, pp.63-66, ISSN 0028-0836
Gold, T. (1948). The physical basis of the action of the cochlea. Proceedings the Royal of Society,
Vol.135, No.881, pp. 492-498, ISSN 1471-2954
Hamernik, R.P., Henderson, D., Coling D. & Slepecky N. (1980). The interaction of whole
body vibration and impulse noise. J. Acoust. Soc. Am., Vol. 67, No.3, pp. 928-934,
Hamernik, R.P., Henderson, D., Coling, D. & Salvi R. (1981). Influence of vibration
on asymptotic threshold shift produced by impulse noise. Audiology, Vol.20, No.3.
pp.259-269, ISSN Print 0020-6091
Hamernik, R.P., Henderson, D., Coling, D. & Salvi R. (1981). Influence of vibration on
asymptotic threshold shift produced by impulse noise. Audiology, Vol. 20, No.3.
pp.259-269, ISSN 1499-2027
He, W., Nuttall, A.L. & Ren, T. (2007). Two-tone distortion at different longitudinal locations
on the basilar membrane. Hear. Res., Vol. 228, No.1-2, pp.112-122, ISSN Print 0378-
5955
Jankowski, W. Giedanowski, J. & Birecki W. (1962). Effect of some vasoconstrictor drugs on
the microphonic potential of the cichlea. The Polish OtoLaryngology, Vol.16, pp. 321
329, PL ISSN 0030-6657
Jones, C.M.(1996). ABC of work related disorders. Occupational hearing loss and vibration
induced disorders. Occupational. Br Med J , Vol. 313, No. 7051, pp. 223-226, ISSN
0959-8138
Kemp, D.T. (1978). Stimulated acoustic emissions from within the human auditory system.
J. Acoust. Soc. Am., Vol. 64, No.5, pp. 1386-1391, ISSN Print 0001-4966
Knight, R.D. & Kemp D.T. (2000). Indications of different distortion product otoacoustic
emission mechanisms from a detailed f
1
, f
2
area study, J. Acoust. Soc. Am., Vol. 107,
No.1, pp. 457-473, ISSN Print 0001-4966
Kobayashi, T., Rong Y., Chiba T., Marcus C.D., Ohyama K. & Takasaki T. (1997). Ototoxic
effect of erythromycin on cochlear potentials in the guinea pig. Ann. Otol. Rhinol.
Laryngol, Vol.106, No.7, pp. 599-603, ISSN Print 0003-4894
Li, X., Wodlinder, H. & Sokolov Y. (February 2003). A new method for measuring DPOAEs
and ASSRs, Available from
http://www.hearing review.com/issues/articles/2003-02 05.asp
Linder, T.E., Zwicky, S. & Brndle P. (1995). Ototoxicity of ear drops: a clinical perspective.
Am. J. Otol., Vol. 16, No. 5, pp. 653-657, ISSN Print 0192-9763
Palmer, K.T., Griffin M.J., Bendall H., Pannett B. & Coggon D.(2000a). Prevalence and
pattern of occupational exposure to hand transmitted vibration in Great Britain:
findings from national survey. Occup. Environ. Med. , Vol. 57, No.4, pp. 218-228,
ISSN Print 1351-0711
Palmer, K.T., Griffin M.J., Bendall H., Pannett B. & Coggon D. (2000b). Prevalence and
pattern of occupational exposure to whole body vibration in Great Britain: findings
from national survey. Occup Environ Med, Vol. 57, pp. 229-236, ISSN Print 1351-0711
Perkins, R. (1980). Laser stapedotomy for otosclerosis. Laryngoscope, Vol.90, No.2, pp.228-
241, ISSN 1531-4995
Rogowski, M. & Chodynicki, S. (1987). Einflu von Vibrationen und Gentamycin auf das
Gehrorgan des Meerschweinchens. HNO-Prax , Vol. 12, pp.219-223


286
Seidel, H., Heide R. (1986). Long-term effects of whole-body vibration: a critical survey of
the literature. Int Arch Occup Environ Health, Vol.58, No.1, pp. 1-26, ISSN Print 0340-
0131
Tasaki, I., Davis, H. & Legouix J.P. (1952). The space-time pattern of the cochlear
microphonics (guinea pig) as recorded by differential electrodes. J. Acoust. Soc. Am.,
Vol. 24, No.5, pp. 502 519, ISSN Print 0001-4966
Thurlow, W.R. (1943). Studies in auditory theory:II The distortion of distortion in the inner
ear. Journal of Experimental Physology: General, Vol.32, No.4, pp. 344 350, ISSN
0096- 3445
Wadsten C.J., Bertilsson C.A., Sieradzki H. & Edstrm S. (1985). A randomized clinical trial
of two topical preparations (framycitin/gramicidin and oxytetracycline /hydro-
cortisone with polymyxin B) in the treatment of external otitis. Arch
Otorhinolaryngol Vol. 242, No. 2, pp. 135-139. ISSN Print 0302-9530
Wever, E.G. & Bray, C.W. (1930). Auditory nerve impulses, Science, Vol.71, No.1834, p.215,
ISSN Print 0036 8075
Vollrath, M., Schreiner, Chr.(1982). Influence of argon laser stapedotomy on cochlear
potentials I: Alteration of cochlear microphonics (CM). Acta Otolaryngol, suppl 385,
pp.1-31, ISSN Print 365-5237
Ziarani, A.K. & Konrad A. (2004). A novel metod of estimation of DPOAE signals. IEEE
Transactions on biomedical engineering, Vol.51, No.5, pp.864-868, ISSN 0018 9294
Ziemski, Z. (1970). Ototoxity of selected organic solvents of industrial plastics in
experimental animals. Papers of Medical University in Wroclaw, Vol.15, No.1, pp.59-
128
13
Photoacoustic Technique Applied to
Skin Research: Characterization of Tissue,
Topically Applied Products and
Transdermal Drug Delivery
Jociely P. Mota, Jorge L.C. Carvalho, Srgio S. Carvalho and Paulo R. Barja
UNIVAP
Brazil
1. Introduction
The photoacoustic (PA) effect basically consists in the production of acoustic waves due to
the absorption of modulated (or pulsed) radiation by a sample. Graham Bell discovered the
PA effect in 1880, when he noticed that the incidence of modulated light on a diaphragm
connected to a tube produced sound. Thereafter, Bell studied the PA effect in liquids and
gases, showing that the intensity of the acoustic signal observed depended on the absorption
of light by the material.
In the nineteenth century, it was known that the heating of a gas in a closed chamber
produced pressure and volume changes in this gas. However, there were many different
theories to explain the PA effect. Rayleigh said that the effect was due to the movement of
the solid diaphragm. Bell believed that the incidence of light on a porous sample expanded
its particles, producing a cycle of air expulsion and reabsorption in the sample pores. Both
were contested by Preece, who pointed the expansion/contraction of the gas layer inside the
photoacoustic cell as cause of the phenomenon. Mercadier explained the effect conceiving
what we call today thermal diffusion mechanism: the periodic heating of the sample is
transferred to the surrounding gas layer, generating pressure oscillations.
The lack of a suitable detector for the PA signal made the interest in this area decline until
the invention of the microphone. Even then, research in this field was restricted to
applications in gas analysis up to 1973, when Rosencwaig started to use the PA technique in
spectroscopic studies of solids and, together with Gersho, developed a mathematical model
for the generation of the PA signal in solid samples the Rosencwaig-Gersho (RG) Model
(Rosencwaig & Gersho, 1976).
In condensed matter samples, one of the most important mechanisms for PA signal
generation is the thermal diffusion, classically described by the RG model. According to this
model, the (modulated) radiation absorbed by condensed matter samples is converted into
heat, causing temperature modulation in the surrounding atmosphere. This eventually
produces the mechanical effect of periodic expansion and contraction originating sound
waves that can be detected by a microphone.
Since the publishing of the RG model and, soon after that, of the generalized theory for the
PA effect by McDonald and Wetsel (1978), the PA technique has already proved its


288
relevance in a large number of very different fields, from the polymerization of dental resins
(Balderas-Lopez et al., 1999) to photosynthesis studies (Malkin & Puchenkov, 1997; Herbert
et al., 2000).
1.1 Objectives
The purpose of this chapter is to present applications of the PA technique in skin research,
both in the characterization of skin itself and in transdermal drug delivery studies. The basic
experimental setup for such studies will be briefly presented, aiming to help those who may
be interested in developing similar studies. Emphasis will be done to in vivo measurements,
because of its importance in this field. Our objective is to show the usefulness of the PA
technique in the biomedical field, particularly in skin research; finally, perspectives for
future work in this field will be presented.
2. Photoacoustic measurements
2.1 Basic experimental setup
Figure 1 presents one scheme for a basic photoacoustic experimental setup.

Fig. 1. Example of a basic photoacoustic experimental setup (scheme)
The experimental scheme in Figure 1 shows a (typically mechanical) chopper positioned in
front of the light source, in order to modulate the radiation that comes into a
monochromator (utilized in PA spectrocopy measurements). Light absorbed by the sample
generates acoustic waves inside the PA cell; the PA signal is captured by a microphone
(inside the PA cell) that sends it to the lock-in amplifier (also connected to the chopper, to
receive information on the frequency modulation). The lock-in amplifier is connected to a
microcomputer for data acquisition. In vivo, skin measurements are performed with an
open-ended PA cell, in which it is the sample itself that closes the chamber.
2.2 Measurements as a function of time
The PA signal depends on the optical and thermal properties of the sample, which may
vary with time due to different factors. When a sample undergoes changes in its
(Optical Cable, optional)
Light
source
Chopper Monochromator
PA Cell +
Sample
Lock-in Amplifier
Data Acquisition
System
Photoacoustic Technique Applied to Skin Research:
Characterization of Tissue, Topically Applied Products and Transdermal Drug Delivery

289
composition or structure (as it occurs during the polimerization of a dental resin, for
instance), the propagation of heat inside the sample is also modified, thereby altering the
PA signal.
We must also mention the possibility of performing photosynthesis studies using PA
measurements as a function of time (W.J. Silva et al., 1995). When PA measurements are
performed in photosynthesizing samples as plant leaves, the PA signal presents, in
addition to the photothermal component, a photobaric component, resulting from the gas
exchanges associated to the photosynthesis process (Acosta-Avalos et al., 1996). This
allows the study of the so-called photosynthetic induction, that is, the increase of the net
photosynthetic rate that occurs when a plant is shifted from darkness to light (Sui et al.,
2011).
As stated by Bodzenta et al. (2002) in their work on PA detection of drug diffusion into a
membrane, PA measurements give the possibility for investigations in relatively long time
periods. This makes the PA technique suitable for the monitoring of dehydration processes
(Lopez et al., 2005) and of changes occurring in time in biological tissues such as skin. It is
possible to study, for example, the kinetics of transdermal drug delivery through the
analysis of PA measurements as a function of time. One example will be presented at the
section 4 of the present chapter.
2.3 Studies on the modulation frequency: depth profile
In thermally thick samples (as skin tissue), only the light absorbed within the first thermal
diffusion length (
T
) of the sample/tissue contributes to the PA signal (Rosencwaig, 1980).
As the thermal diffusion length depends on the modulation frequency (f) of the incident
light by the relation
=
T
f

(1)
where is the thermal diffusivity of the sample, it is possible to perform depth-profile
studies, with the evaluation of the penetration depth of a product (or even a microorganism)
in tissue. The possibility of performing depth-profile studies is particularly interesting in the
characterization of multilayer systems (as skin itself).
The frequency dependence analysis of the PA signal can also be employed in the
determination of the thermal properties (thermal diffusivity, thermal effusivity) of a sample
or material (Balderas-Lopez & Mandelis, 2001), including biological tissues as porcine skin
(Gao et al., 2005; Qiu et al., 2008).
2.4 Measurements as a function of the wavelength: Photoacoustic spectroscopy
Photoacoustic spectroscopy (PAS) is already incorporated to the roll of useful photothermal
techniques since the 1980s (Rosencwaig, 1980; Vargas & Miranda, 1988). Besides the
possibility of rendering depth-profile analysis in multi-layered samples, PAS presents at
least two additional advantages over other spectroscopy techniques: i) as transmitted and
reflected light do not interfere in PAS measurements, it is a more direct technique,
representing a direct measurement of the light absorption by the sample; ii) it allows the
study of optically opaque and highly scattering samples (which could not be analyzed by
conventional optical spectroscopy).


290
In PAS measurements, the emission spectra of the light source is typically obtained through
measurements using black carbon powder (or other black material) as the sample, with all
the remaining measurements being normalized with respect to the lamp spectrum.
PAS can also be employed in skin research. In 2004, Benamar and co-workers presented a PAS
study on the effect of dihydroxyacetone, frequently employed for artificial tan. Measurements
were carried out in the presence and absence of dimethylisosorbide (a solvent for
dihydroxyacetone), on excised human skin. By monitoring the PAS signal intensity with time
in the UV (300-400nm) range, these authors demonstrated that dihydroxyacetone in
combination with dimethylisosorbide enhances the process of tanning (Benamar et al., 2004).
Recently, Melo et al. (2011) applied PAS to evaluate the penetration rate of Helicteres
gardneriana extract, topically applied for anti-inflammatory purposes. Experiments were
conducted ex vivo in mice. Croton oil was applied into both mouses right and left auricles to
induce inflammatory response, and the left auricle was treated with the extract. The strong
anti-inflammatory effect observed for the Helicteres gardneriana extract was associated with
the deep percutaneous penetration observed for the formulation, according to PA data
(Melo et al., 2011).
2.5 Photoacoustic imaging and tomography
Photoacoustic imaging is based on the production of acoustic waves following irradiation by
a short pulse of light whose absorption generates local heating and transient thermoelastic
expansion (Balogun et al., 2009). According to Beard (2009), haemoglobin represents the
most important source of endogenous contrast in PA imaging. This makes the technique
particularly indicated to studying tissue abnormalities as tumors and other diseases related
to changes in the structure and oxygenation status of the vasculature (Beard, 2009).
Recently, Hu and Wang (2010) presented PA tomography as a method combining high
spatial resolution and optical absorption contrast, important in microvascular imaging and
characterization. Reviewing the major embodiments of PA tomography (microscopy,
computed tomography and endoscopy), they have analyzed the methods employed in
different studies, including hemodynamic monitoring, determination of hemoglobin
concentration, evaluation of oxygen saturation, studies of blood flow and tumor-vascular
interaction.
Besides being applied to soft tissues, PA imaging can also be employed to hard tissues. Li
and Dewhurst (2010) have applied a PA imaging system with a near-infrared (NIR) pulsed
laser to obtain images from both soft tissue and post-mortem dental samples. They have also
performed simulations (based on the thermoelastic effect) to predict initial temperature and
pressure fields within a tooth sample, observing that values are maintained below the
corresponding safety limits. In this way, the results presented by Li and Dewhurst show
that the PA technique can be sucessfully applied to image both soft and hard tissues.
3. Photoacoustic measurements and the characterization of skin
Biological materials are sometimes difficult to study employing conventional techniques
that require previous preparation of the samples, because these materials can have its
properties significantly altered by preparation processes as solubilization, for example. The
PA technique does not require previous preparation; it can be described as a non-invasive
technique that allows even in vivo measurements.

291
In general, biological tissues can be characterized as highly scattering samples; however, this
is not a problem for PA mesurements, in which the signal is based in the direct absorption of
radiation. As pointed by Cahen and co-workers (1980), the relative insensitivity to scattered
light of the PA signal makes such measurement an attractive way to measure biological
samples in vivo. These features explain the potential of the PA technique in the study of
opaque materials and complex biological systems such as skin. PA measurements can be
employed to determine the absorption characteristics of the skin itself or topically applied
products, as well as kinetic changes related to transdermal drug delivery.
Skin diseases can also be studied through PA measurements. In 2010, Swearingen et al.
developed a PA methodology to determine the nature of skin lesions (pigmented and
vascular) in vivo, which is important because misdiagnosis may even lead to cancerous
lesions not receiving proper medical care. These authors irradiated skin with two laser
wavelengths (422 and 530nm), with the relative response at these two wavelengths
(422nm/530nm) indicating whether the lesion is pigmented or vascular, due to the distinct
absorption spectrum of melanin and hemoglobin (Swearingen et al., 2010).
3.1 Skin type classification
Skin type classification is important not only for medical or clinical purposes, but also for
pharmaceutical and cosmetic industries, following the idea that an objective, precise
characterization of skin could be useful in the design of new topically applied products and
in defining more specific skin treatments according to each skin type.
However, in dermatology, there is still no universal agreement about the best method for
classifying skin, as even the widely accepted method proposed by Fitzpatrick (1988)
defining the so-called skin phototypes is based in clinical, subjective analysis.
More recently, Baumann (2006a, 2006b) proposed a new skin type classification, according
to which 16 different skin types are defined from the combination of four parameters, as
skin can be characterized as: i) pigmented or nonpigmented; ii) dry or oily; iii) sensitive or
resistant; and iv) wrinkled or tight. Baumanns skin typing is based on an extensive
research, performed with 1400 volunteers. However, it relies essentially on the response of
volunteers to a questionnaire; therefore, it does not fulfill per se the need of an objective
classification, which would require experimental evaluation.
PA measurements have a potentially important role to play in an experimental approach to
skin type classification. In 2000, Schmidt and co-workers conducted non-contacting, in vivo
PAS measurements in skin (performed in 50 volunteers), in the VIS-NIR range, seeking an
objective determination of pigmentation, blood microcirculation and water content of
human skin (Schmidt et al., 2000). According to these authors, strong spectral variations
observed within the same skin type are probably based on the natural variability of human
skin and in the subjective clinical evaluation of the skin type; nevertheless, PAS results
obtained show good correlation between PA data and (clinically evaluated) skin type,
indicating that skin type determination could indeed be performed through the analysis of
PA measurements.
3.2 Skin pigmentation analysis employing photoacoustic measurements
In 2004, Viator and co-workers proposed a method for the determination of the epidermal
melanin content employing a PA probe using a Nd:YAG (neodymium, yttrium, aluminum,
garnet) laser at 532nm (Viator et al., 2004). Ten human subjects with skin phototypes IVI


292
were tested using the PA probe and visible reflectance spectroscopy (VRS); melanin content
was evaluated through each of these methods, and a good linear fit (r
2
=0.85) was obtained
for the plot of PA x VRS.
Pigmentation skin level can also be evaluated through simple, direct PA measurements
employing non-laser light sources. Actually, PA measurements have been performed at the
Laboratory of Photoacoustic Technique Applied to Biological Systems (FASBio), at UNIVAP
(Brazil). The objective of such in vivo measurements was to classify different skin types
according to the amplitude of the PA signal, which can be associated to the corresponding
pigmentation level of the skin.
In the following subsections, we present this straightforward PA approach to skin
characterization according to the level of pigmentation, employing PA measurements in
volunteers. Experimental results are compared both to Fitzpatrick and Baumann clinical
skin type evaluations.
3.2.1 Materials and methods
The PA setup employed in such in vivo skin measurements consisted of a 250W tungsten
halogen lamp as light source (with wavelength range 400nm<<700nm and light intensity of
about 20W/m
2
), a mechanical chopper (SRS, model SR540), a lock-in amplifier (SRS, model
SR530) and a microcomputer for data acquisition.

Fig. 2. Experimental setup with volunteer positioned for in vivo skin measurement
The double faced PA cell employed, with an electret microphone, was developed at
UNIVAP; sensitivity was 15 mV/Pa for the frequency employed in skin measurements,

293
17Hz. The electret microphone structure was described by Marquezini et al. (1990). The PA
cell has a cylindrical body and two opposite, parallel faces (one is closed by a thin glass
layer and the other, by the sample itself). For the modulation frequency employed (17Hz),
the thickness of the skin layer under study is about 30m.
PA measurements were recorded as a function of time (200 readings for each measurement,
in 0.5s intervals, up to a total of 100s per measurement). During measurements, one face of
the PA cell was closed with a thin transparent window, while the forearm of the volunteer
was gently pressed against the opposite face.
Figure 2 shows the PA experimental setup employed for in vivo skin measurements at the
FASBio/UNIVAP, with a volunteer positioned for measurement.
Measurements were performed in 57 female volunteers, between 20 and 30 years-old.
Initially, each volunteer answered a questionnaire according to their daily routine associated
to skin care; volunteers were also clinically evaluated and, as a result, they were classified
according to skin phototype, following Fitzpatrick classification (Fitzpatrick, 1988).
Before measurements, the skin area to be evaluated was cleaned with cotton embedded in
alcohol 70%. The PA signal was then recorded for the inner and outer faces of both
forearms. Volunteers were then classified according to the respective PA signal amplitude,
and this classification was compared to the phototype classification.
3.2.2 Results and discussion
Initially, a comparison between the PA signal amplitude of the inner and outer faces was
performed, showing a highly significant statistical difference (paired t-test, p<0,005), with
higher PA amplitude being observed for the outer face of the forearm. This result can be
attributed to the higher pigmentation level of the skin region continuously exposed to solar
radiation, demonstrating that skin constitution and aspect are clearly influenced by the level
of sun exposure.
After clinical evaluation of the volunteers for skin phototype (following Fitzpatrick
classification), PA results were grouped according to the phototype of each volunteer.
Results are presented in Table 1.

Skin phototype (Fitzpatrick) PA signal amplitude (mV)
II 1.26 0.05
a

III 1.59 0.09
ab

IV 1.70 0.10
ab

V 1.80 0.10
b

Table 1. PA signal amplitude (mV) for the inner face of the forearm, for each skin phototype
(average standard error). Different indexes (a, b) indicate significant statistical difference
(comparison among groups: p=0.009, ANOVA)
Comparison among phototype groups was performed and significant statistical difference
was verified (as we can see in Table 1), showing that the PA signal level (amplitude) for the
inner forearm tends to scale with skin phototype, as defined by Fitzpatrick.
Afterwards, the PA signal amplitude for each volunteer (average values for the inner face of
the forearm) allowed the division of the volunteers in two groups, pigmented (P) and
non-pigmented (NP), following the Baumann proposal. As the average PA signal


294
amplitude obtained for all measurements (inner face of the forearm) was 1.5mV, this was
the cutoff value adopted for separating the volunteers into P (for PA signal amplitude
above 1.5 mV) and NP (under 1.5 mV). Table 2 shows the division of each (clinically
evaluated) phototype group into the (experimentally evaluated) P and NP groups. In this way,
the PA technique allowed the comparison between two different skin classification forms.

Skin phototype NP group (%) P group (%)
II 93 7
III 46 54
IV 31 69
V 20 80
Table 2. Distribution of the volunteers of each phototype in the NP and P groups, according
to the PA signal level
Table 2 shows that phototype II is highly related to the NP group, while phototypes IV and
V concentrate in the P group. Phototype III appears in both groups, showing the variability
of elements inside this classification.
The simple methodology presented here and the corresponding results obtained open
perspectives for an objective, experimental classification of skin types, based upon PA
measurements. Additional work in this field is currently being performed at
FASBio/UNIVAP (Brazil).
4. Transdermal drug delivery
4.1 Topical application of drugs: advantages and requirements for evaluation
Topical application of drugs is known as an interesting alternative route to oral and
intravenous administration, both aiming to systemic effects and local action, offering
advantages such as ease of administration and lack of first-pass effect (Aqil et al., 2007).
However, in studies of transdermal drug delivery (penetration of substances into skin), one
must employ non-invasive techniques, in order to avoid second order effects that would at
least bring difficulties to the interpretation of the results. In vivo measurements are
particularly desirable in such studies, because the response of excised skin can be affected
by dehydration, and the response of artificial skin differs significantly from that of in vivo
skin at least in some cases as when we talk about in-depth processes, in which even blood
circulation may play a role.
As mentioned earlier in the present chapter, the PA technique can be applied without
previous preparation of the samples and even for in vivo measurements; as such,
transdermal drug delivery studies have been performed employing the PA technique in
order to obtain the penetration rate of a wide range of different products topically applied to
skin (Bernengo et al., 1998; Hahn et al., 2001; Savateeva et al., 2001; Pedrochi et al., 2005;
Truite et al., 2007).
4.2 Substrates for transdermal drug delivery studies
Besides measurements in human skin, penetration rates of topically applied products are
frequently evaluated through measurements performed in animal skin tissue. In this case,

295
rabbit and pig skin are, by far, the most employed alternatives, because of the similarity to
human skin.
As pointed by Simon & Malbach (2000), physiological and anatomical similarities between
man and pig make this animal a good model for man in biomedical research. The correlation
of quantitative data between pig skin and human skin can be frequently classified as very
good (Benech-Kieffer et al., 2000); therefore, pharmacological (and even toxicological) skin
research is often based on the knowledge of pig skin absorption and percutaneous
permeation.
Recently, Nicoli et al. (2008) employed qualitative and quantitative analysis of stratum
corneum lipids and permeation experiments to analyze the utilization of rabbit ear skin in
transdermal permeation studies, using pig ear skin as a reference. Their results showed that
the stratum corneum of both rabbit ear skin and pig ear skin present similar thickness.
Probably due to its higher lipophilicity, rabbit ear skin was less permeable to hydrophilic
compounds; however, the permeability to progesterone was comparable between isolated
pig epidermis and rabbit ear skin. Nicoli and co-workers conclude that the rabbit ear skin
can be sucessfully employed in skin permeatin studies.
4.3 Photoacoustic evaluation of topically applied products
Different pharmaceutical formulations for a topically applied drug may present very
different transdermal delivery ratios, depending on the product composition (excipients
usually play a major role in the penetration kinetics of topically applied products). These
penetration rates may be evaluated through the analysis of the time-dependence of the PA
signal after topical application of a given product in skin. This methodology can also be
applied to the evaluation of sunscreens, that may be characterized in terms of their
(photo)stability after topical application (in this case, the lower the rate, the better the
product).
Gutierrez-Juarez et al. (2002) employed PA measurements in the analysis of substances
topically applied to the human skin. To fulfill this purpose, these authors utilized a double-
chamber PA cell; the absorption determination was obtained through the measurement of
the thermal effusivity of the binary system substanceskin. The model employed by
Gutierrez-Juarez and co-workers (that assumes that the effective thermal effusivity of the
binary system corresponds to that of a two-phase system) was experimentally applied to
study different topically applied substances, in different parts of the body. The
corresponding relative concentrations of substances as a function of time were determined
by fitting a sigmoidal function (for ketoconazol and sunscreen) or an exponential function
(for nitrofurazona, vaseline and vaporub) to the experimental data.
Pedrochi and co-workers (2005) employed PAS measurements to evaluate the penetration
rate of different sunscreens into human skin in vivo. Their results showed that the
diminution rate of the sunscreen amount in the skin surface depends on the form of the
product: sunscreens in cream form tend to present faster reduction after application in skin.
This leads to the conclusion that sunscreens in gel form are more adequate (presenting
longer protection against UV radiation).
Another transdermal drug delivery study is the work of Truite et al. (2007), which employed
PAS measurements in the ex vivo determination of the penetration rate of different
phytotherapic formulations (with and without salicylic acid) for treatment of vitiligo.
Measurements were performed as a function of time in rabbits. PA depth monitoring


296
showed that both formulations propagated through the skin up to the melanocytes region,
leading the authors to suggest that the delivery of the active agents may occur even without
the use of queratinolitic substances (that are not really recommended, since they are known
to induce side effects in animals).
The PA technique can be employed to study the penetration kinetics of topically applied
products not only as a function of product composition, but also according to the
application method. Phonophoresis is the utilization of ultrasound (US) waves to enhance
the delivery of topically applied substances (Byl, 1995). In physiotherapy practice,
phonophoresis is one of the various strategies developed to overcome the skins resistance
in transdermal drug delivery, enhancing skin permeability (Duangit et al., 2011).
In the last years, comparative studies between massage and phonophoresis (in its
different modes) as application methods for different anti-inflammatories have been in the
front line of research at the FASBio/UNIVAP; in the experiments, transdermal drug
delivery has been evaluated through PA measurements as a function of time after topical
application of different drugs in the forearm region, using manual massage or
phonophoresis.
Results indicate that different products present distinct absorption times (depending on the
vehicle employed, for example); the application method also affects the typical time constant
of drug penetration into skin, though not for all tested formulations.
In the following subsections, we present one experiment performed at FASBio/UNIVAP in
which the penetration kinetics of the pharmaco Cordia verbenacea DC (Acheflan) in the
human skin was evaluated through PA measurements as a function of time for each of the
application methods: massage and phonophoresis.
4.3.1 Materials and methods
The pharmaco Cordia verbenacea DC (Acheflan) is a topic usage anti-inflammatory
medication widely employed in medicine, having alpha-humulen and trans-caryophyllen as
active agents. Our experiment aimed to evaluate the penetration kinetics of Acheflan in the
human skin (massage versus phonophoresis) through in vivo PA measurements.
The survey was conducted in 10 volunteers (four men and six women) aged between 18 and
30 years. The following inclusion criteria were adopted: (i) absence of ulcers or any change
in dermatology distal forearm and wrist; (ii) not being allergic to any component of the
formula topically applied; (iii) absence of metal implants in the wrist or forearm; (iv)
absence of stomach pain complaints; and (v) not being pregnant.
The protocol for cleaning prior to drug application consisted of cleaning the skin area to be
evaluated (region near the distal forearm and right ulnar artery) with cotton soaked in 70%
alcohol. The area of topical application was then demarcated and Cordia verbenacea DC
(essential oil 5.0mg/g) was applied by rubbing the head of the ultrasound therapy
equipment configured for continuous mode and intensity of 1.2W/cm
2
, but switched off, for
five minutes. This procedure was repeated on the opposite forearm with the ultrasound
therapy equipment turned on, also for five minutes.
PA measurements employed the same experimental system described in section 3.2.1.
During measurements, the volunteers were positioned adjacent to the assembly with an
aluminum foil (60m thick) sealing the PA cell and the distal forearm positioned in direct
contact with it, as proposed by Bernengo et al. (1998). For an aluminum foil with this
thickness, the cutoff frequency is approximately 7kHz; in the present study, the modulation

297
frequency employed was 17Hz, so that the foil can be considered as thermally thin
(Rosencwaig, 1980).
Each measurement series consisted of 40 readings (one each two seconds, for a total time of
80 seconds), repeated 10 times for each application form, with rest intervals of 100 seconds
between successive series (total time of 30 minutes for each volunteer and application
method). The software SISCOMF (developed at UNIVAP) was employed for data
acquisition.
In order to analyze the typical time constant for the penetration of the applied drug, the
experimental curves were fitted by the Boltzmann equation:

1 2
2 ( t t )/dt
0
A A
PA(t) A
1 e

= +
+

(2)
The Boltzmann curve is a S-shaped curve in which A
1
is the initial signal amplitude, A
2
the
final signal amplitude, t
0
is the half-absorption time and dt, the time interval such as 67% of
the penetration process occurs between t
0
-dt e t
0
+dt.
Analysis of the PA data was performed with the aid of the software Microcal Origin 7.5
(employed for the generation of the fitting curves); statistical analysis was performed with
GraphPad Instat 3.0.
4.3.2 Results and discussion
Figure 3 shows an example of PA data measurements as a function of time fitted by a
Boltzmann curve.

0 5 10 15 20 25
6,0
6,5
7,0
7,5
8,0

P
A

a
m
p
l
i
t
u
d
e

(
m
V
)
Time (minutes)

Fig. 3. Example of a Boltzmann curve (dashed line) fitting PA data obtained for one of the
volunteers after phonophoresis application of the pharmaco Cordia verbenacea DC (Acheflan)


298
The experimental results obtained (average values for each parameter of the fitting curve)
are summarized in Table 3; interpretation of such results must consider that the initial
amplitude (A
1
) of the PA signal corresponds to the system formed by the applied drug+skin,
while the final amplitude (A
2
) corresponds to the skin only, with the product having
penetrated beyond the layer responsible for the generation of the PA signal (about 30m, in
the present case).

Application Method A
1
(u.a) A
2
(u.a) t
0
Dt t
s
(t
0
+ dt)
Massage 0,66 0,04 0,70 0,02 7 2 4 2 11 2
Phonophoresis 0,65 0,03 0,72 0,04 5 1 1,3 0,3 7 1
Table 3. Average values ( standard error) for A
1
and A
2
(in arbitrary units), t
0
and dt (in
minutes), for each of the application methods (N=10)
In order to understand if penetration was effective, first of all, it is imperative to evaluate if
the difference between A
1
and A
2
is statistically significant. Therefore, initially a paired t-test
was carried on to verify if there was significant difference between A
1
and A
2
for each
application method employed (indicating significant penetration of the applied product).
This was verified for both application methods; however, this difference is more evident for
phonophoresis application, in which the difference between the initial and the final signal
presents p=0,011 (p=0,066 was found for massage application).
Statistical tests were also employed in the comparison between the two application methods
(massage and phonophoresis); no statistical significance was found for t
0
and dt.
Considering t
S
= (t
0
+ dt) as the total effective penetration time for the epidermal layer under
study, the results obtained (average error, N=10) are 11(2) minutes (for massage
application) and 7(1) minutes (for phonophoresis application). The paired t-test for this
parameter shows p=0,073.
Experiments performed at FASBio/UNIVAP show that the form of application can
influence the kinetics of transdermal drug delivery, depending on the applied product. In
the experiment presented here, significant penetration has been reported for both forms of
administration (massage and phonophoresis); PA measurements showed that effective
penetration is at least more evident after phonophoresis application, when compared to
massage application.
5. Conclusion and perspectives
The possibility of performing in vivo studies brings great value to the use of the PA
technique in experimental skin research. The potential and relevance of PA measurements
in this field have been shown by a large range of experiments being performed by different
research groups around the world actually, the examples here presented must be seen as a
sample of what has been done.
PA measurements in vivo are able to detect alterations in skin pigmentation. Even in skin
regions normally protected from sun exposure, the PA signal level tends to follow clinical
classification; actually, the use of the PA technique goes one step ahead, allowing
comparative and quantitative research through simple, direct measurements.
Topical drug application has been employed in the treatment of many pathological
processes; its efficiency is associated to the efficiency of transdermal drug delivery. PA

299
measurements have been sucessfully employed in transdermal drug delivery studies,
allowing a quantitative analysis of the kinetics and effectivity of drug delivery. Different PA
experiments point to the fact that gel formulations tend to be more adequate for topical use.
Depending on the topically applied product, the form of application can also be determinant
in the kinetics of transdermal drug delivery.
Measurements already performed indicate various perspectives for future research, such as:
i) the determination of the skin oiliness level; ii) analysis of skin lesions; iii) studies on the
photostability of sunscreens and even determination of the sun protection factor (SPF) of
sunscreens (through PAS measurements); and iv) further studies on formulation and
application form of a wide range of topically applied products.
6. Acknowledgment
The authors acknowledge Fapesp and CNPq for financial support of biomedical research
being developed at FASBio/UNIVAP, So Jos dos Campos (SP), involving the
characterization of human skin and transdermal drug delivery of topically applied products.
7. References
Acosta-Avalos, D.; Alvarado-Gil, J.J.; Vargas, H.; Fras-Hernndez, J.; Olalde-Portugal, V.;
Miranda, L.C.M. (1996) Photoacoustic monitoring of the influence of arbuscular
mycorrhizal infection on the photosynthesis of corn (Zea mays L.). Plant Science,
v.119, n.1-2, pp.183-190; doi:10.1016/0168-9452(96)04454-8
Aqil, M.; Abdul, A.; Yasmin, S.; Asgar, A. (2007) Status of terpenes as skin penetration
enhancers. Drug Discovery Today (Oxford), v.12, n.23/24, pp.1061-1067.
Balderas-Lopez, J.A.; Moreno-Mrquez, M.M.; Martnez, J.L.; Snchez-Sinencio, F. (1999)
Thermal characterization of some dental resins using the photoacoustic phase lag
discontinuites. Superfcies y Vacio, v.8, pp.42-45.
Balderas-Lopez, J.A.; Mandelis, A. (2001) Thermal diffusivity measurements in the
photoacoustic open-cell configuration using simple signal normalization
techniques. Joumal of Applied Physics, v.90, n.5. pp.2273-2279.
Balogun, O.; Regez, B.; Zhang, H.F.; Krishnaswamy, S. (2009) Real time, full-field imaging of
photoacoustic generated signals for biomedical applications, ICPPP15 - Book of
Abstracts, 15th International Conference on Photoacoustic and Photothermal Phenomena,
pp.77, Leuven, Belgium, July 19-23, 2009.
Baumann, L. (2006a) New Skin Typing System. Skin & Aging , v.14, n.2, pp.60-64.
Baumann, L. (2006b) The Skin Type Solution. Bantam Books, ISBN-10: 0-553-80422-7, New
York, NY.
Beard, P. (2009) High resolution spectroscopic photoacoustic imaging for characteristic
tissue structure and function, ICPPP15 - Book of Abstracts, 15th International
Conference on Photoacoustic and Photothermal Phenomena, pp.79, Leuven, Belgium,
July 19-23, 2009.
Benamar, N.; Laplante, A.F.; Lahjomri, F.; Leblanc, R.M. (2004) Modulated photoacoustic
spectroscopy study of an artificial tanning on human skin induced by
dihydroxyacetone. Physiological Measurement, v.25, n.5, pp.1199-1210.
Benech-Kieffer, F.; Wegrich, P.; Schwarzenbach, R.; Klecak, G.; Weber, T.; Leclaire, J.;
Schaefer, H. (2000) Percutaneous Absorption of Sunscreens in vitro: Interspecies


300
Comparison, Skin Models and Reproducibility Aspects. Skin Pharmacology and
Physiology, v.13, pp.324-335; doi: 10.1159/000029940
Bernengo, J.C.; Gasquez, C; Falson-Rieg, F. (1998) Photoacoustics as a tool for cutaneous
permeation studies. High Temperature-High Pressure, v.30, n.5, pp.619-624.
Bodzenta, J.; Bukowski, R.J.; Christ, A.; Pogoda, T. (2002) Photoacoustic detection of drug
diffusion into a membrane: theory and numerical analysis. International Journal of
Heat and Mass Transfer, v.45, n.22, pp.4515-4523; doi:10.1016/S0017-9310(02)00155-2
Byl, N.N. (1995) The Use of Ultrasound as an Enhancer for Transcutaneous Drug Delivery:
Phonophoresis. Physical Therapy, v.75, n.6, pp.539-553.
Cahen, D.; Bults, G.; Garty, H.; Malkin, S. (1980) Photoacoustic in life sciences. J. Biochemical
Biophysical Methods, v.3 (5), 1980, pp.293-310.
Duangjit, S.; Opanasopit, P.; Rojanarata, T.; Ngawhirunpat, T. (2011) Characterization and In
Vitro Skin Permeation of Meloxicam-Loaded Liposomes versus Transfersomes.
Journal of Drug Delivery, v.2011, ID 418316; doi:10.1155/2011/418316
Fitzpatrick, T.B. (1988) The validity and practicality of sun-reactive skin types I through VI.
Archives of Dermatology, v.124, pp.869871.
Gao, C.M.; Zhang, S.Y.; Zhang, Z.; Shui, X.J. (2005) Thermal diffusivity of porcine tissues
characterized by photoacoustic piezoeletric technique. Journal of Physics IV (France),
v.125, pp.777-779; doi:10.1051/jp4:2005125179
Gutierrez-Juarez, G.; Vargas-Luna, M.; Cordova, T.; Varela, J.B.; Bernal-Alvarado, J.J.; Sosa,
M. (2002) In vivo measurement of human skin absorption of topically applied
substances by a photoacoustic technique. Physiological Measurement, v.23, pp.521-
532.
Hahn, B.D.; Neubert, R.H.H.; Wartewig, S.; Lasch, J. (2001) Penetration of compounds
through human stratum corneum as studied by Fourier transform infrared
photoacoustic spectroscopy. Journal of Controlled Release, v.70, n.3, pp.393-398.
Herbert, S.K.; Han, T.; Vogelmann, T.C. (2000) New applications of photoacoustics to the
study of photosynthesis. Photosynthesis Research, v.66, n.1-2, pp.13-31; doi:
10.1023/A:1010788504886
Hu, S.; Wang, L.V. (2010) Photoacoustic imaging and characterization of the
microvasculature, Journal of Biomedical Optics, v.15, pp.011-101;
doi:10.1117/1.3281673
Li, T.; Dewhurst, R.J. (2010) Photoacoustic imaging in both soft and hard biological tissue.
Journal of Physics: Conference Series, v.214, 012028; doi: 10.1088/1742-
6596/214/1/012028
Lopez, T.; Picquart, M.; Aguilar, D.H.; Quintana, P.; Alvarado-Gil, J.J.; Pacheco, J. (2005)
Photoacoustic monitoring of dehydration in sol-gel titania emulsions. Journal of
Physics IV (France), v.125, pp.583-585; doi:10.1051/jp4:2005125134
McDonald, F.A.; Wetsel G.C. (1978) Generalized theory of the photoacoustic effect, Journal of
Applied Physics, v.49, pp.2313.
Malkin, S.; Puchenkov, O.V. (1997) The photoacoustic effect in photosynthesis. In Progress in
Photothermal and Photoacoustic Science and Technology: Life and Earth Sciences
(Mandelis, A., and Hess, P., editors), SPIE, ISBN 0-8194-2450-1, Washington, USA.
Marquezini, M.V.; Cella, N.; Mansanares, A.M.; Vargas, H.; Miranda, L.C.M. (1990) Open
photoacoustic cell spectroscopy. Measurement Science & Technology, v.2, pp.396-401.

301
Melo, J.O.; Pedrochi, F.; Baesso, M.; Hernandes, L.; Truiti, M.; Baroni, S.; Bersani-Amado, C.
(2011) Evidence of Deep Percutaneous Penetration Associated with Anti-
Inflammatory Activity of Topically Applied Helicteres gardneriana Extract: A
Photoacoustic Spectroscopy Study. Pharmaceutical Research, v.28, n.2, pp.331-336;
doi: 10.1007/s11095-010-0279-3
Nicoli, S.; Padula, C.; Aversa, V.; Vietti, B.; Wertz, P.W.; Millet, A.; Falson, F.; Govoni, P.;
Santi, P. (2008) Characterization of Rabbit Ear Skin as a Skin Model for in vitro
Transdermal Permeation Experiments: Histology, Lipid Composition and
Permeability. Skin Pharmacology and Physiology, v.21, pp.218-226; doi:
10.1159/000135638
Pedrochi, F.; Sehn, E.; Medina, A.N.; Bento, A.C.; Baesso, M.L.; Storck, A.; Gesztesi, J.L.
(2005) Photoacoustic Spectroscopy to Evaluate the Penetration Rate of Three
Different Sunscreens into Human Skin in vivo. Journal of Physics IV (France), v.125,
pp.757-759; doi:10.1051/jp4:2005125174
Qiu, P.F.; Zhang, S.Y.; Shui, X.J. (2008) Photoacoustic study of thermal properties of
biological tissues detected by PVDF film transducer. European Physics Journal -
Special Topics, v.153, pp.487-490.
Rezende, D.V.; Nunes, O.A.C.; Oliveira, A.C. (2009) Photoacoustic Study of Fungal Disease
of Aai (Euterpe oleracea) Seeds. International Journal of Thermophysics, v.30, n.5,
pp.1616-1625; doi: 10.1007/s10765-009-0655-6
Rosencwaig, A.; Gersho, A. (1976) Theory of the photoacoustic effect with solids. Journal of
Applied Physics, v.47, pp.64-69.
Rosencwaig, A. (1980) Photoacoustics and Photoacoustic Spectroscopy. John Wiley & Sons, ISBN
0-471-04495-4, New York, USA.
Savateeva, E.V.; Karabutov, A.A.; Oraevsky, A.A. (2001) Proceedings of SPIE 4256, pp.61-69.
Schmidt, W.D.; Fassler, D.; Zimmermann, G.; Liebold, K.; Wollina, U. (2000) Non-contacting
diffuse VIS-NIR spectroscopy of human skin for evaluation of skin type and time-
dependent microcirculation. Progress in biomedical optics and imaging, v.1, n.31,
pp.91-102.
Silva, W.J.; Prioli, L.M.; Magalhes, A.C.N.; Pereira, A.C.; Vargas, H.; Mansanares, A.M.;
Cella, N.; Miranda, L.C.M.; Alvarado-Gil, J.J. (1995) Photosynthetic O
2
evolution in
maize inbreds and their hybrids can be differentiated by open photoacoustic cell
technique. Plant Science, v.104, n.2, pp.177-181; doi:10.1016/0168-9452(94)04026-D
Simon, G.A.; Maibach, H.I. (2000) The Pig as an Experimental Animal Model of
Percutaneous Permeation in Man: Qualitative and Quantitative Observations An
Overview. Skin Pharmacology and Applied Skin Physiology, v.13, pp.229-234; doi:
10.1159/000029928
Sui, X.; Sun, J.; Wang, S.; Li, W.; Hu, L.; Meng, F.; Fan, Y.; Zhang, A. (2011) Photosynthetic
induction in leaves of two cucumber genotypes differing in sensitivity to low-light
stress. African Journal of Biotechnology, v.10, n.12, pp. 2238-2247.
Swearingen JA, Holan SH, Feldman MM, Viator JA. (2010) Photoacoustic discrimination of
vascular and pigmented lesions using classical and Bayesian methods. Journal of
Biomedical Optics, v.15, n.1, 016019; doi: 10.1117/1.3316297
Truite, C.V.R.; Philippsen, G.S.; Ueda-Nakamura, T.; Natali, M.R.M.; Dias Filho; B.P.; Bento,
A.C.; Baesso, M.L.; Nakamura, C.V. (2007) Photochemistry and Photobiology, v.83,
pp.1529-1536; doi: 10.1111/j.1751-1097.2007.00197.x


302
Vargas, H.; Miranda, L.C.M. (1988) Photoacoustic and related photothermal techniques.
Physics Reports, v.161, n.2, pp.43-101.
Viator, J.A.; Komadin, J.; Svaasand, L.O.; Aguilar, G.; Choi, B.; Nelson, J.S. (2004) A
Comparative Study of Photoacoustic and Reflectance Methods for Determination of
Epidermal Melanin Content. Journal of Investigative Dermatology, v.122, pp.1432-
1439.
14
AcousticGravity Waves in the Ionosphere
During Solar Eclipse Events
Petra Kouck Knov and Zbyek Mona
Institute of Atmospheric Physic, Czech Academy of Sciences
Czech Republic
1. Introduction
Terrestrial atmosphere shows a high variability over a broad range of periodicities, which
mostly consists of wave-like perturbations characterized by various spatial and temporal
scales. The interest for short time variability in ionospheric attributes is related to the role
that ionosphere plays in the Earth's environment and space weather. Acoustic-gravity
waves (AGWs), waves in the period range from sub-seconds to several hours, are sources of
most of the short-time ionospheric variability and play an important role in the dynamics and
energetics of atmosphere and ionosphere systems. Many different mechanisms are likely to
contribute to the acoustic-gravity wave generation: for instance, excitation at high latitudes
induced by geomagnetic and consequent auroral activity, meteorological phenomena,
excitation in situ by the solar terminator passages and by the occurrence of solar eclipses.
During solar eclipse, the lunar shadow creates a cool spot in the atmosphere that sweeps at
supersonic speed across the Earths atmosphere. The atmosphere strongly responds to the
decrease in ionization flux and heating. The very sharp border between sunlit and eclipsed
region, characterized by strong gradients in temperature and ionization flux, moves
throughout the atmosphere and drives it into a non-equilibrium state. Acoustic-gravity
waves contribute to the return to equilibrium. At thermospheric heights, the reduction in
temperature causes a decrease in pressure over the totality footprint to which the neutral
winds respond. Thermal cooling and downward transport of gases lead to neutral
composition changes in the thermosphere that have significant influence on the resulting
electron density distribution. Although the mechanisms are not well understood, several
studies show direct evidence that solar eclipses induce wave-like oscillations in the acoustic-
gravity wave domain.
Many different mechanisms are likely to contribute to wave generation and enhancement at
ionospheric heights. Hence, it is difficult to clearly separate or differentiate each contributing
agent and to decide which part of wave field belongs to the in situ generated and which part
comes from distant regions. First experimental evidence of the existence of gravity waves in
the ionosphere during solar eclipse was reported by Walker et al. (1991), where waves with
periods of 3033 min were observed on ionosonde sounding virtual heights.
1.1 Ionospheric sounding
As the solar radiation penetrates Earths atmosphere it forms pairs of charged particles.
Under a normal day-time conditions the ionization solar flux increases immediately after


304
sunrise, reaches maximum around local noon and decreases again till sunset. Under such
conditions concentration of charged particles significantly grows in the atmosphere and forms
atmospheric plasma called ionosphere. Due to the composition of the neutral atmosphere
together with the changing efficiency of the incoming solar radiation, ionosphere is stratified
into the layers denoted D, E, F1 and F2. After sunset, electrons and ions recombine rapidly in
the D, E and F1 layer. Due to slower recombination processes of atomic ions that dominate at
heights approximately above 150km altitude, F2 layer remains present all the night. Special
stratification Es, sporadic E layer, occurs sometimes at heights of E layer (Davies, 1990).
Ionosphere significantly affects propagation of the electromagnetic waves. According to a
frequency of the wave with respect to a concentration of the ionospheric plasma, wave
propagates through the medium or it is reflected. Electromagnetic waves with frequency
lower than plasma frequency of the particular plasma parcel are reflected, which allows to
estimate plasma frequency. Higher frequency waves propagate through plasma. An
instrument called ionosonde (or digisonde) transmits electromagnetic wave of a defined
frequency and detects it after reflection from the ionosphere. Typical ionosonde sounding
range is 1 MHz 20 MHz. For each sounding wave ionosonde records time of flight on the
path transmitter - reflection point receiver. Time of flight is simply converted into a
virtual height
virtual
.c
h
2
= that corresponds to wave propagation in the vacuum (c stands

here for speed of light). Virtual height is equal or higher than the corresponding real height.
The output of the measurement is height-frequency characteristics called an ionogram. Real
height electron concentration profiles can further be inverted from ionograms using for
instance programs POLAN (Titheridge, 1985) or NHPC (Huang and Reinish, 1996).
Ionosphere represents inhomogeneous and anisotropic medium which leads to a wave
splitting into an ordinary and extraordinary wave modes. Hence, two reflection traces are
recorded by the ionosonde (as seen on ionograms in Figure 1). However, the extraordinary
mode is not further used for electron concentration profile inversion.
Figure 1 shows typical day-time and night-time ionograms recorded by a digisonde in the
observatory Pruhonice. Together with the ionograms there are plots of the real height
electron concentration derived by NHPC routine. On the day-time electron concentration
profile, three ionospheric layers E, F1 and F2 are present while on the night-time profile
there is only F2 layer detected by the ionosonde.
Sequences of ionograms are widely used for analyses of variability of atmospheric plasma
ranging from detection of rapid changes with periods of minutes to the study of long-term
trends.
2. Basic theory of AGWs in the Earths atmosphere
Most of the wave-like oscillations in the atmosphere can be described/parametrized using
basic acoustic-gravity wave theory in the atmosphere. Details can be found, for instance, in
works of Davies (1990), Bodo et al. (2001), Hargreaves (1982), Yeh & Liu (1974) among
others. Here, we show brief derivation of the dispersion relation that any wave motion of
the AGW type must satisfy. In a plane-stratified, isothermal atmosphere under gravity that
is constant with height, two frequency domains exist in the atmosphere where atmospheric
waves can propagate, acoustic and gravity wave. Atmosphere represents compressible gas
that once compressed and then released would expand and oscillate about its equilibrium
state. Its oscillation frequency is known as an acoustic cut-off frequency

AcousticGravity Waves in the Ionosphere During Solar Eclipse Events

305

=
a
c
2H
(1)
where c is speed of sound

= c gH
(2)
is the ratio of specific heats at constant pressure and constant volume, g is the gravitational
acceleration, and H is the scale height. For diatomic gas ~ 1.4.

(a)

(b)
Fig. 1. Typical day-time (a) and night-time ionogram (b) measured by digisonde DPS 4 in
the Observatory Pruhonice. On both plots, there is real hight electron concentration (solid
line with error bars) provided as obtained by the NHPC routine.


306
Single element of fluid, parcel of the atmosphere, at height z with density which is
displaced in the vertical by z to a place where its density changes to +, remains in
pressure equilibrium with its surroundings. Displacement takes place adiabatically. This is
valid when the motion is so slow that sound waves with speed

dp
c
d
=
(3)
where p stands for pressure can traverse the system faster than the time-scale of interest and
the motion is so fast that the entropy is preserved. The parcel is no longer in equilibrium
and starts to oscillate about its equilibrium height with buoyancy frequency.
The buoyancy force which acts on the parcel is balanced by inertial force (Newtons second
law):

( ) =
2
2
d
z g
dt
(4)
where is the difference between internal and external densities.
Internal and external are derived as:

( )

= =
2
2 int ernal
g
p/c z
c
(5)
which is due to compressibility of the fluid within the membrane and

( )

=
external
d
z
dz
(6)
is the change of background density at new position due to inhomogeneous nature of the
atmosphere. Taking both the contributions of we get

( ) ( )

= +

2
2 2
2
d d
z g ln g c z
dt dz
(7)
which can be recast into

( ) + =
2
2
B 2
d
z z 0
dt
(8)
where

( )

= +

2 2
B
d
g ln g c
dz
(9)
If
2
B
>0, the solution is oscillatory and the fluid parcel will oscillate with characteristic
buoyancy frequency
B
called Brunt-Vaisala frequency.
More convenient form used for atmosphere is following:

( ) = +
2 2 2 2 2
B
1 g c g /c dc /dz
(10)
This approximation is valid in the atmosphere-ionosphere system of our interest.


307
In isothermal atmosphere
B
reduces to

( ) =
2 2 2
B
1 g /c
(11)
In the terrestrial atmosphere the buoyancy period depends on the height. The height
variance of the acoustic cut-off and buoyancy frequencies in the isothermal atmosphere is
shown in Figure 2.

Fig. 2. Height dependence of acoustic cut-off period t
a
and Brunt-Vaisala period t
B
that
represent limits dividing periods into acoustic and gravity wave domains. Period domain
between acoustic cut-off and Brunt-Vaisala represents region where no AGW propagates.
Wave motion in the atmosphere can be described using mass conservation (continuity
equation), and equation of motion:

+ + =

. u u. 0
t
(12)

+ = +

u
u. u p g
t
(13)
where pressure gradients and gravity are the only forces causing the acceleration.
Oscillation takes place adiabatically

+ = +

p
u. p p u.
t t
(14)
where , p, and u
are parameters of the atmosphere density, pressure, ratio of specific

heats, and velocity.
t
B

t
a


308
Applying the perturbation approach we are searching for wave-like solutions for the
perturbation quantities. Further simplification comes from the assumption that the
background state is of constant temperature T in which p
0
/
0
must be a constant.

=
2
0 0
p / c /
(15)
Then the system (12), (13) and (14) reduces to:

+ =
0 0 z
'
. u' u /H 0
t
(16)

+ =
0
u
p' 'g 0
t
(17)

=

' '
0 0 z 0 0 z
p' '
p u /H p u /H
t t
(18)
where index 0 denotes stationary (non fluctuating) component and the apostrophe denotes
perturbation. These are the basic governing equations for the gravity waves. For a non-
trivial solution the following prescription of the dispersion relation must be satisfied:

( )
=
4 2 2 2 2 2 2 2 2 2
a x g z
k c c k 0
(19)
From disperse relation, it is evident that between buoyancy frequency and acoustic cut-off
frequencies one cannot have both k
x
and k
z
real. Figure 2 shows two period domains with
border limits of acoustic cut-off period and buoyancy period.
An attenuation or growth in the wave amplitude must occur in either the vertical or the
horizontal directions. We suppose that there is no variation in amplitude in horizontal
directions so that k
x
is purely real and k
z
has an imaginary component. At frequencies
exceeding acoustic cut-off
a
, expression (19) becomes simple and the waves may be termed
as ACOUSTIC WAVES. At frequencies smaller than Brunt-Vaisala frequency where gravity
plays an important role, the waves are called GRAVITY or INTERNAL GRAVITY WAVES.
Brunt-Vaisala frequency and acoustic cut-off frequency divide the frequency spectrum into
two domains in which
g

forms the high frequency limit for one class <
g
normally called
internal gravity waves and
a

is the low frequency limit for another class >
a
called the
acoustic waves. A gap in the frequency spectrum exists between
g

and
a
where no
internal waves can propagate.
Important approximations can be obtained under the assumption
z
k 1 2H >> and
g
<<
then:

( )
=
2 2 2 2
z g x
k / k
(20)
These approximations apply to much of the observed gravity waves. From (20) we see that
the angle of ascent of the phase

is:

= = =
z x g g
tg k /k / /
(21)


309
The motions of the air parcels are, in general, ellipses in the plane of propagation and have
components transverse to the direction of wave propagation. The ratio of the horizontal
displacement
to its vertical displacement
is:

x
2
2
g
a z
2
x
ck
ck
i
ck
1
(22)
On the frequencies just above the acoustic cutoff the air motion is essentially vertical. With
acoustic waves on high frequencies the motion is radial as in sound waves. The motion is
circular with horizontal propagation at a frequency
a
2 / . Gravity wave propagation is
limited to angles between

1
min
g
sin
,

1
max
g
sin
(23)
The sense of rotation of the air for gravity waves is opposite than for acoustic waves. As
approaches its asymptotic values the air motion becomes linear and transverse to the
direction of propagation. Air parcel rotation is clockwise in case of acoustic waves while
anticlockwise in case of gravity waves. Energy vector lies in the same quadrant as direction
of propagation of acoustic waves. Energy flows up when phase travels down and vice versa
in case of gravity waves propagation. This is important property since it accounts for the
observed downward phase propagation when the source is below the level at which a
disturbance is observed.
The horizontal u
x
and vertical u
z
components of the packet velocity are obtained from
disperse relation:

( )
( )
( )

=

=

2 2 2
x g
x
2 2 2 2
a
2 2
z
z
2 2 2 2
a
c k
u
2 c k
c k
u
2 c k
(24)

Due to coupling between neutral and charged components the initial wave-like oscillation in
the neutral atmosphere induces wave-like perturbation in the ionosphere. Perturbation in
the ion production is the most effective when solar ionizing rays are nearly in alignment
with the initial wave front. Perturbations in the neutral atmosphere may cause perturbations
in chemical processes. Presence of AGW influences the ionisation rate through changes in
the local neutral density and temperature, and through changes in the ionisation radiation
absorption (Hooke, 1970).
3. AGW in the ionospheric plasma
Acoustic-gravity waves are always present in the Earths atmosphere. AGWs arise from
many natural sources like convection, topography, wind shear, moving solar terminator,
earthquakes, tsunami, etc. Increase in wave-like activity is associated also with human


310
activity including coordinated experiments or unwilling accidents. AGWs influence on the
upper atmosphere is not yet understood enough. They produce a great amount of variability
and contribute to the background conditions in a specific parcel of the atmosphere. Gravity
waves propagating from lower laying atmosphere have been long regarded as a very
important source of the energy and momentum transfer in the upper atmosphere (Hines,
1960). The breaking of the upward propagating waves affects wind system, generates
turbulence and heats the atmospheric gas.
Waves that reach upper atmosphere produce travelling atmospheric disturbances (TAD) or
travelling ionospheric disturbances (TID) and even form the ionospheric inhomogenities
which grow and finally break into the plasma instabilities observed by radar techniques that
might cause scintillation of the communication signals propagating through the ionosphere.
From the observation it is evident that the thermosphere is continuously swept by the
acoustic-gravity waves. Statistically, the waves show a moderate preference for southward
travel, with this preference being reduced or shifted to southeastward travel during
disturbed times (Oliver et al., 1997). Experimental studies show that AGW activity in the
ionosphere slightly increases during dawn and dusk periods of the day (Galushko et al.,
1998; Somsikov & Ganguly, 1995; Sauli et al., 2005 among others). Influence of infrasonic
waves generated by ground experimental sources on the ionosphere was reported for
instance by Rapoport et al. (2004).
Solar eclipse represents well defined source of the AGW in the atmosphere and
ionosphere systems. During solar eclipse event, solar ionization flux decreases producing
well-defined cool spot in the atmosphere that moves through the Earths atmosphere.
Moving source in the atmosphere can emit both acoustic and gravity waves. Supersonic
motion of the source forms wave field with bow wave. Both acoustic and gravity waves
can be radiated in association with supersonic motion in the atmosphere. When the source
is moving within atmosphere with subsonic velocity only gravity waves can be emitted
(Kato et al., 1977).
4. Solar eclipse event signatures in the ionospheric plasma
It has been proposed by Chimonas and Hines (1970) that solar eclipses can act as sources for
AGWs. The lunar shadow creates a cool spot in the atmosphere that sweeps at supersonic
speed across the Earth. The sharp border between sunlit and eclipsed regions, characterized
by strong gradients in temperature and ionization flux, moves throughout atmosphere and
drives it into a non-equilibrium state. Earth atmosphere shows variable sensitivity to the
changes of ionization flux.
4.1 Experiments
Solar eclipse event represents phenomenon that can be precisely predicted, hence many
observational campaigns are organised around the world. Effects of the solar eclipses on the
ionospheric plasma are studied by mean of GPS techniques, radars, vertical ionospheric
soundings etc. Study limitations lay mainly in the fact that there are no identical solar
eclipse events. Moreover, solar eclipse induced effects are easily to be mixed with effects
caused by geomagnetic field variations, diurnal changes of the ionosphere, seasonal
variability of the atmosphere/ionosphere etc. In the upper atmosphere, AGWs can be
observed either directly as neutral gas fluctuations or indirectly as induced ionospheric


311
plasma variations. Despite intensive research many questions in the problem of the
generation and propagation remain to be understood.
Studies by Fritts and Luo (1993) suggest that perturbations generated by the eclipse induced
ozone heating interruption may propagate upwards into the thermosphereionosphere
system where they have an important influence. Temperature fluctuations and electron
density changes propagate as a wave, away from the totality path, cf. Muller-Wodarg et al.
(1998). By means of vertical ionospheric sounding, Liu et al. (1998) detected waves excited
during solar eclipse event at F1 layer heights and their generation and/or enhancement
attributed to changes of temperatures and variations of the height of the transition level for
the loss coefficient and the height of the peak of electron production. Studies reported by
Farges et al. (2001) suggest a longitudinal diversity of the disturbances with respect to pre-
noon and postnoon phases. Xinmiao et al. (2010) reported synchronous oscillations in the Es
and F layer during the recovery phase of the solar eclipse. Ivanov et al. (1998) found that
during solar eclipse with maximum obscuration of about 70% the F-region electron density
decreased by 6-8% compared to a control day and detected travelling ionospheric
disturbances. Additionally, they detected strong variations in the difference group delays
with a period about 40 minutes associated with the start and end of the eclipse. Oscillations
in the ionosphere, similar to gravity waves, were observed following some solar eclipse
events (Chimonas and Hines, 1970; Cheng et al., 1992; Liu et al., 1998; Sauli et al., 2006).
Investigation of the latitudinal dependence of NmF2 (the maximum electron density of the
F2 layer) indicated that the strongest response was at middle latitudes (Le et al., 2009). The
response of the sporadic-E (Es) layer also differed in each solar eclipse event. A remarkable
decrease in Es layer ionization was observed during the eclipse of 20 July 1963 (Davis et al.,
1964). Enhancement of Es layer ionization has also been reported and it has been suggested
that it is related to internal gravity waves generated in the atmosphere during the solar
eclipse (Datta, 1972).
4.2 Processes induced by solar eclipse
During the solar eclipse, on the time scale shorter than day-night change, the ionosphere
reconfigures itself into a state similar to that of night situation. Photochemical ionization
falls heavily almost to a night-time level. With the decreasing solar flux, atmospheric
temperature falls in the moon shadow creating a cool spot with well defined border. Then
the increasing solar flux starts ionization processes and warms the atmosphere again to
daytime level.
Such changes in the ionization cause variation in the reflection heights, decrease/increase in
electron concentration at all ionospheric heights, decrease/increase in the total electron
content, rising/falling of the layer height. Such effects are characteristic for the processes
during sunrise/sunset in the ionosphere. However, supersonic movement of the eclipsed
region represents a key difference from the regular solar terminator motion at sunrise and
sunset times. These changes in the neutral atmosphere and ionosphere induced by solar
eclipse force the evolution of the ionospheric plasma toward a new equilibrium state. The
return to equilibrium is likely accompanied by the eclipse induced wave motions excited in
the atmosphere. Any moving discontinuity of gas parameters such as temperature, pressure
etc. will generate transit-like waves. In the upper ionosphere, waves can be generated by a
strong horizontal electron pressure gradient. Possible mechanisms contributing to the wave
generation in the region of solar terminator are in detail discussed by Somsikov & Ganguly
(1995).


312
Solar eclipse induces changes in all atmospheric regions extend from the upper atmosphere
down to ground level. Despite the low magnitude of the eclipse induced effects at ground
level, Jones et al. (1992) reported wave-like oscillation related to eclipse on the
microbarometer pressure records. The cooling effect of the Moons shadow may induce the
powerful meridional airflow in the atmosphere, which accelerates the ionized clouds in the
Es layer and forms the wind shear to raise the observed Doppler frequency shift and foEs
values, respectively (Chen et al., 2010).
5. Solar eclipse observed by vertical ionospheric sounding in midlatitudes
Vertical sounding measurements provide local information on the electron density
distribution of the bottomside ionosphere. Electron concentration in the plasma and its
corresponding plasma frequency are related via following equation:

2
2
p
0
Ne
f
4 m
=
(25)
where f
p
denotes plasma frequency and N, e,
0
and

m stand for the electron
concentration, the charge of electron, permittivity of free space, and the mass of the
electron, respectively.
This section summarizes experimental results from the midlatitude ionospheric observatory
Pruhonice (50N, 15E). At the observatory, the vertical sounding measurements were
performed with ionosonde IPS 42 KEL Aerospace till the end of year 2003. Then this older
equipment was replaced by digisonde DPS 4. Special campaigns of rapid sequence
soundings were organized in order to study in detail ionospheric behavior during partial
solar eclipses of 11 August 1999, 4 January 2011 and annular solar eclipse 3 October 2005. All
three analyzed events were characterized by low geomagnetic activity; hence they represent
a good occasion to observe mostly solar eclipse induced effects in the ionosphere. However,
inconclusive results of the solar eclipse observations rise from the fact that different solar
eclipses produce different plasma motions. Indeed, the travel cone geometry and its angular
effects on the magnetized plasmas are different for each eclipse.
Solar eclipse of 11 August 1999 (as a total seen in place as close as 200 km from the
measurement point) represents so far the event of the highest solar disc coverage observed
in the Observatory Pruhonice. Figure 3 depicts sequence of raw ionograms measured during
this event by IPS 42 KEL Aerospace equipment. The ionograms were recorded with the
cadence of 1 minute. On the ionograms there is clearly seen that the eclipse event affects
whole electron density profile. Critical frequencies in the E and F layer decrease before
maximum disc occultation and then increase again. The electron density decrease in the E
layer is much stronger than in the F layer due to different dominant type of the
recombination. Electron density fall and increase occur simultaneously with occultation
and de-occultation of the solar disc in the E and F1 layer while the F2 layer electron
density reacts with slight delay. There are special structures of the spread F type
developed on the profile after beginning of the solar disc occultation (clearly seen on
ionograms at 9.14 UT and 9.16 UT). Shape of the F layer is affected as well. Unfortunately,
effects in the F1 region cannot be discussed here in details because F1 layer is blanketed
by strong sporadic E layer during part of the solar eclipse.


313

Fig. 3. Sequence of raw ionograms measured by the ionosonde KEL Aerospace IPS 42 at the
observatory Pruhonice. During the special campaign ionograms were recorded with one-
minute resolution in order to study rapid ionospheric changes during the solar eclipse.


314
Detail analysis of electron concentration by mean of spectral analysis reveals that within
oscillation of electron concentration there occur several clear wave-like oscillations. It has
been shown by Sauli et al. (2007) that wavelet spectral analysis is very convenient approach
for such wave detection. The advantage of the wavelet based analysis is identification of the
structure occurrence time which helps to associate particular wave-like structure to the
agent. Figure 4 shows estimated wave parameters for selected structure that is coherent
through all studied heights. Parametrization of the wave-like structure is based on AGW
approximation described in Section 2. From Figure 4 it is evident that wave originates at
height of about 200 km and propagates upward and downward from the source region.

Fig. 4. Parameters of acoustic-gravity wave structure detected within ionospheric plasma
during solar eclipse event 11 August 1999 (Sauli et al., 2007). Panels: wave vector (a), phase
velocity (b), packet velocity (c), wave number (d), energy (e) and phase (f) angles. For the
vectors of first row, the correspond to the measured (full squares) and computed (empty
symbols) z-components, the correspond to the horizontal components while the
are
related to the modulus.
Another representation of the rapid changes in the ionospheric plasma is shown on the
profilogram (Figure 5) measured during solar eclipse 3 October 2005 by DPS 4. Decrease in
the plasma frequency at all heights is well developed. Within plasma frequency oscillation,
several wave coherent structures were found that can be attributed to the eclipse event.
These structures occur in the plasma at the maximum of the eclipse and after the event. In
all cases we detected a component of upward energy progression. Due to the occurrence
time and low geomagnetic activity the detected wave-like oscillations in the ionospheric
plasma are likely signatures of bow shock and possibly waves excited by cooling of ozone in
the lower laying atmosphere. Estimated velocities for one particular structure are shown in
Figure 6.


315

Fig. 5. Profilogram (height-time-plasma frequency development) during solar eclipse 3
October 2005 as measured by DPS 4. Ionograms were measured every 2 minutes. All
ionograms were manually scaled and inverted into true-height profiles using True Height
Profile Inversion Tool NHPC.

Fig. 6. Parameters of acoustic-gravity wave structure detected within ionospheric plasma
during solar eclipse event 3 October 2005 (Sauli et al. 2007). Panels: wave vector (a), phase
velocity (b), packet velocity (c), wave number (d), energy (e) and phase (f) angles. For the
vectors of first row, the correspond to the measured (black) and computed (empty) z-
components, the correspond to the horizontal components while the
are related to the

modulus.


316
Result of the annular eclipse is significantly different from the case of the total eclipse event
of 1999 where the dominant AGW activity took place at the beginning of eclipse. The
atmospheric cooling and decrease in radiation flux during an annular solar eclipse is not as
strong as during a total eclipse and the ionospheric response occurs with time delay.

Fig. 7. Virtual reflection heights of plasma frequency in the range 4.2 - 4.3 MHz derived from
raw ionograms. From up to bottom: day before, day of eclipse, day after eclipse. Vertical
lines in middle panel depict beginning and end of the eclipse. Time resolution is different
for day of solar eclipse (2 min) and days before/after (15 min).
In Figure 7 and Figure 8, there are plots of virtual reflection height variations at single
frequency during three consecutive days, day of solar eclipse event and one day before and
after the event. Variation of the reflection height during eclipse event of 3 October 2005 does
not differ much from the corresponding variation during reference time span day before
and day after. Wave-like oscillations excited by solar eclipse are of comparable magnitude
as those induced by other sources preceding and consecutive day. On the contrary, clear
difference in reflection height oscillation during reference days and solar eclipse event is
perfectly seen in Figure 8. Records of virtual heights at fixed frequency from January 4, 2011
present strong ionospheric response which is exhibited as periodic changes in reflection
height. Sharp changes in the reflection height develop immediately after the beginning of
the solar disc occultation and last till the end of eclipse event. Higher wave-like activity
remains remarkable whole day. In this partial solar eclipse event, wave-like oscillations can
be very probably attributed to the solar eclipse.
Strong decrease in electron concentration in practically whole electron profile as well as the
wave-like changes were observed during and after August 11, 1999 and January 4, 2011.
Wave-like activity develops immediately after the start of the solar disc obscuration during


317
partial solar eclipse. During annular solar eclipse, significant acoustic-gravity wave type
bursts develop around and after maximum phase of the eclipse.

Fig. 8. Virtual reflection heights of plasma frequency range 3.4 3.5 MHz derived from raw
ionograms. From up to bottom: day before, day of eclipse, day after eclipse. Vertical lines in
middle panel depict beginning and end of the eclipse. Time resolution is different for day of
solar eclipse (5 min) and days before/after (15 min).
6. Conclusion
Acoustic-Gravity waves play important role in the dynamic of the upper atmosphere.
Vertical ionospheric sounding represents powerful tool that allows us to monitor acoustic-
gravity wave activity in the ionosphere. Ionospheric observation of such a strong event as
solar eclipse gives us an opportunity to better understand processes of creation and
dissipation of the AGW in the area of the ionosphere. Although the acoustic-gravity waves
are always present in the area of our interest, sharp temporally well-defined changes of
solar flux during the solar eclipse give us a possibility to define sources of AGW.
It is rather uneasy to unambiguously assess causality between the solar eclipse events and
the detected wave structures in the ionospheric plasma. Difficulties result from the fact that
there are no two exactly identical solar eclipse events and from limitations of sounding
techniques. Despite the fact that various AGW sources have been identified, many others
remain to be found. Amongst irregular AGW bursts, regular increase in AGW activity were
found to occur around sunrise and sunset hours, excited by Solar Terminator movement.
Most of other sources (meteorological systems, geomagnetic and solar disturbances, etc.)
and corresponding wave-like oscillations contribute to the irregular patterns of AGW
activity observed in the ionospheric plasma.


318
As the solar eclipses, analyzed in the Section 5, occur sufficiently long time after the sunrise
hours, one can assume that none of the reported waves are induced by solar terminator.
During the analyzed sounding campaigns, no wave coming from auroral zone was
expected, due to the quiet geomagnetic and solar activity. Additionally, meteorological
analysis shows that meteorological systems very probably did not influence the ionosphere
during studied events by means of AGW. The acoustic-gravity wave activity increases after
a notably larger delay for the annular solar eclipse compared to the total solar eclipses:
waves are found during the maximum phase of the eclipse only for the former while they
occur during the initial phase for the latter. This discrepancy in gravity waves
generation/occurrence can likely be explained by differences in the terrestrial atmosphere
cooling: the border between sunlit and eclipsed region is much sharper in the case of total
eclipse. Analyzing wave propagations, we observe predominantly upward propagating
structures. The wave structure, that propagate upward and downward from the source
region located around 200 km height, was created during an exceptional case related to the
Solar eclipse of 11 August 1999.
7. References
Altadill, D., Sole, J.G. & Apostolov, E.M. (2001). Vertical structure of a gravity wave like
oscillation in the ionosphere generated by the solar eclipse of August 11, 1999.
Journal of Geophysical Research, 106 (A10), 2141921428, ISSN 0148-0227.
Bodo, G., Kalkofen, W., Massaglia, S. & Rossi, P. (2001). Acoustic waves in a stratified
atmosphere. III. Temperature inhomogenities. Astronomy & Astrophysics, 370, pp.
1088-1091, ISSN 0004-6361.
Chen, G., Zhao, Z., Zhou, C., Yang, G. & Zhang, Y. (2010). Solar eclipse effects of 22 July
2009 on Sporadic-E. Annales Geophysicae, 28, 353357, ISSN 0992-7689.
Cheng, K., Huang, Y.N. & Chen, S.W. (1992). Ionospheric effects of the solar eclipse of
September 23, 1987, around the equatorial anomaly crest region. Journal of
Geophysical Research, 97, A1, 103111. ISSN 0148-0227.
Chimonas, G. & Hines, C.O. (1970). Atmospheric gravity waves induced by a solar eclipse.
Journal of Geophysical Research, 75, 4, pp. 875, ISSN 0148-0227.
Datta, R.N. (1972). Solar-eclipse effect on sporadic-E ionization. Journal of Geophysical
Research, 77, 1, 260262.
Davies, K. (1990). Ionospheric radio. Peter Peregrinus Ltd., ISBN 0 86341 186 X, London,
United Kingdom.
Davis, J.R., Headrick, W.C. & Ahearn, J.L. (1964). A HF backscatter study of solar eclipse
effects upon the ionosphere. Journal of Geophysical Research, 69 (1), 190193. ISSN
0148-0227.
Farges, T., Jodogne, J.C., Bamford, R., Le Roux, Y., Gauthier, F., Vila, P.M., Altadill, D., Sole,
J.G. & Miro, G. (2001). Disturbances of the western European ionosphere during the
total solar eclipse of 11 August 1999 measured by a wide ionosonde and radar
network. Journal of Atmospheric and Solar-Terrestrial Physics, 63, 9, pp. 915924, ISSN
1364-6826.
Fritts, D.C. & Luo, Z. (1993). Gravity wave forcing in the middle atmosphere due to reduced
ozone heating during a solar eclipse. Journal of Geophysical Research, 98, pp. 3011
3021, ISSN 0148-0227.


319
Galushko, V.G., Paznukhov, V.V., Yampolski, Y.M. & Foster, J.C. (1998). Incoherent scatter
radar observations of EGW/TID events generated by the moving solar terminator.
Annales Geophysicae, 16, pp. 821-827, ISSN 0992-7689.
Hargreaves, J.K. (1982). The upper atmosphere and solar-terrestrial relations. An
introduction to the aerospace environment. Van Nostrand Reinhold, ISBN 0 521
32748 2, Cambridge, United Kingdom.
Hines, C.O. (1960). Internal atmospheric gravity waves at ionospheric heights. Canadian
Journal of Physics, 38, pp. 14411481, ISSN 1208-6045.
Hooke, W.H. (1968). Ionospheric irregularities produced by internal atmospheric gravity
waves. Journal of Atmospheric and Solar-Terrestrial Physics, 30, pp. 795-829, ISSN 1364-
6826.
Huang X. & Reinish B.W (1996). Vertical electron density profiles from the digisonde
network. Advances in Space Research., 18, pp. 121 129, ISSN 0273-1177.
Ivanov, V.A., Ryabova, N.V., Shumaev, V.V., Uryadov, V.P., Nosov, V.E., Brinko, I.G. &
Mozerov, N.S. (1998). Effects of the solar eclipse of 22 July 1990 at mid-latitude path
of HF propagation. of Atmospheric and Solar-Terrestrial Physics, 60, 10, pp. 1013-1016,
ISSN 1364-6826.
Jones, B.W., Miseldine, G.J. & Lambourne, R.J.A. (1992). A possible atmospheric-pressure
wave from the total solar eclipse of 22 July 1990. Journal of Atmospheric and
Terrestrial Physics, 54, 2, pp.113-115, ISSN 1364-6826.
Kato, S., Kawakami, T. & St. John, D. (1977). Theory of gravity wave emission from moving
sources in the upper atmosphere. Journal of Atmospheric and Terrestrial Physics, 39,
pp. 581588, ISSN 1364-6826.
Le, H., Liu, L., Yue, X., Wan, W. & Ning, B. (2009). Latitudinal dependence of the
ionospheric response to solar eclipses. Journal of Geophysical Research, 114, A07308,
ISSN 0148-0227.
Liu, J.Y., Hsiao, C.C., Tsai, L.C., Liu, C.H., Kuo, F.S., Lue, H.Y. & Huang, C.M. (1998).
Vertical phase and group velocities of internal gravity waves derived from
ionograms during the solar eclipe of 24 October 1995. Journal of Atmospheric and
Solar-Terrestrial Physics, 60, pp. 16791686, ISSN 1364-6826.
Muller-Wodarg, I.C.F., Aylward, A.D. & Lockwood, M. (1998). Effects of a mid-latitude solar
eclipse on the thermosphere and ionosphere - a modeling study. Geophysical Reseach
Letters, 25, 20, pp. 37873790, ISSN 0094-8276.
Oliver, W.L., Otsuka, Y., Sato, M., Takami, T. & Fukao, S. (1997). A climatology of F region
gravity wave propagation over the middle and upper atmosphere radar. Journal of
Geophysical Research, 102, A7, pp.14,499-14,512, ISSN 0148-0227.
Rapoport, V.O., Bespalov, P.A., Mityakov, N.A., Parrot, M. & Ryzhov, N.A. (2004).
Feasibility study of ionospheric perturbations triggered by monochromatic
infrasonic waves emitted with a ground-based experiment. Journal of Atmospheric
and Solar-Terrestrial Physics, 66, pp. 1011-1017, ISSN 1364-6826.
Reinisch, B.W., Huang, X., Galkin, I.A., Paznukhov, V. & Kozlov, A. (2005). Recent advances
in real-time analysis of ionograms and ionospheric drift measurements with
digisondes, Journal of Atmospheric and Solar-Terrestrial Physics 67, pp. 1054-1062,
ISSN 1364-6826.


320
Sauli, P., Abry, P., Boska, P. & Duchayne, L. (2006). Wavelet characterisation of ionospheric
acoustic and gravity waves occuring during solar eclipse of August 11, 1999. Journal
of Atmospheric and SolarTerrestrial Physics, 68, pp. 586-598, ISSN 1364-6826.
Sauli, P., Roux, S.G., Abry, P. & Boska, J. (2007). Acousticgravity waves during solar
eclipses: Detection and characterization using wavelet transforms. Journal of
Atmospheric and Solar-Terrestrial Physics, 69, pp. 24652484, ISSN 1364-6826.
Sauli, P., Abry, P., Altadill, D. & Boska, P. (2005). Detection of the wave-like structures in the
F-region electron density: two station measurement. Studia Geophysica & Geodetica,
50, pp. 131-146, ISSN 0039-3169.
Somsikov, V.M. & Ganguly, B. (1995). On the formation of atmospheric inhomogenities in
the solar terminator region. Journal of Atmospheric and SolarTerrestrial Physics, 57,
12, pp. 1513-1523, ISSN 1364-6826.
Titheridge J.E., 1985. Ionogram Analysis with the Generalised Program POLAN. UAG
Report-93, 1985 (http://www.ips.gov.au/IPSHosted/INAG/uag_93/uag_93.html).
Walker, G.O., Li, T.Y.Y., Wong, Y.W., Kikuchi, T. & Huang, Y.N. (1991). Ionospheric and
Geomagnetic effects of the solar eclipse of 18 march 1988 in East-Asia. Journal of
Atmospheric and SolarTerrestrial Physics, 53, 12, pp. 2537, ISSN 1364-6826.
Xinmiao, Z., Zhengyu, Z., Yuannong, Z. & Chen, Z. (2010). Observations of the ionosphere
in the equatorial anomaly region using WISS during the total solar eclipse of 22 July
2009. Journal of Atmospheric and SolarTerrestrial Physics, 72, pp. 869875, 1364-6826.
Yeh, K.C. & Liu, C.H. (1974). Acoustic-gravity waves in the upper atmosphere. Reviews of
Geophysics and Space Physics, 12, 2, pp. 193-216, ISSN 8755-1209.
Part 3
Acoustic Waves as Manipulative Tools

15
Use of Acoustic Waves for
Pulsating Water Jet Generation
Josef Foldyna
Institute of Geonics of the ASCR, v. v. i., Ostrava
Czech Republic
1. Introduction
The technology of a high-speed water jet cutting and disintegration of various materials
attained considerable growth during the last decades. Continuous high-speed water jets are
currently used in many industrial applications such as cutting of various materials, cleaning
and removal of surface layers. However, despite the impressive advances made recently in
the field of water jetting, substantial attention of number of research teams throughout the
world is still paid to the improvement of the performance of the technology, its adaptation to
environmental requirements and making it more beneficial from the economic point of view.
An obvious method of the water jetting performance improvement is to generate jets at
ultra-high pressures. The feasibility of cutting metals with pure water jets at pressures close
to 690 MPa was investigated already in early nineties of the last century (Raghavan & Ting,
1991). Such a high pressure, however, induces extreme overtension of high-pressure parts of
the cutting system which has adverse effect on their lifetime.
An alternate approach, as shown in this chapter, is to eliminate the need for such high
pressures by pulsing the jet. It is well known that the collision of a high-velocity liquid mass
with a solid generates short high-pressure transients which can cause serious damage to the
surface and interior of the target material. The liquid impact on a solid surface consists of
two main stages (see Fig. 1). During the first stage, the liquid behaves in a compressible
manner generating the so-called water-hammer pressures. These high pressures are
responsible for most of the damage resulting from liquid impact on the solid surface. The
situation shortly after the initial impact of the liquid on the solid surface is illustrated in Fig.
2. After the release of the impact pressure, the second stage of the liquid impact begins.
Once incompressible stream line flow is established, the pressure on the central axis falls to
the much lower Bernoulli stagnation pressure that lasts for relatively long time.
The force distribution on liquid jet impact on the solid surface can be summarized as
follows: initially a small central area of the first contact is compressed under a uniform
pressure. The magnitude of the impact pressure p
i
on the central axis is given by
p
=
:p
1
c
1
p
2
c
2
p
1
c
1
+p
2
c
2
(1)
where v is the impact velocity and
1
,
2
and c
1
, c
2
are the densities and the shock velocities in
the liquid and the solid, respectively (de Haller, 1933).

Acoustic Waves From Microdevices to Helioseismology 324

Fig. 1. Two stages of liquid impact on a solid target

Fig. 2. Initial stage of impact between a water drop and a solid target with the contact edge
moving faster than the shock velocity in the liquid. The liquid behind the shock envelope is
compressed and the target beneath this area subjected to high pressure
The magnitude of the impact pressure is independent of the geometry of the drop (Thomas
& Brunton, 1970), but the duration of the pressure is affected by the size and shape of the
drop. For a sphere or cylinder the corresponding radius or half-width of the contact area R
exposed to this pressure is given by
R =
r:
c
1
(2)
where r is the radius of curvature of the drop or cylinder (liquid mass) in the region of
contact (Bowden & Field, 1964).
The initial area of contact grows as the impact continues; there is very little reduction in
pressure on the surface until appreciable outward flow begins. The outward flow of the
liquid becomes possible when the limit of the compressible deformation of the liquid is
exceeded. The limit is given by
:
c
1
= sin [
(3)
where is the liquid/solid interface angle see Fig. 2 (Hancox & Brunton, 1966).
v
2
v
v
v
c
1
c
1
c
2
v
r
v
r
c
2
p
i
p
i
p
s
1 stage
st
2 stage
nd
Target
c
1
c
2
c
2
Shock envelope
Water drop
Contact edge
Target

Use of Acoustic Waves for Pulsating Water Jet Generation 325
At this stage there is a rapid fall in pressure along the periphery of contact. As the outward
flow continues, the water-hammer compression at the centre of impact is relieved until the
maximum pressure acting on the surface is the central stagnation pressure for the
incompressible flow. The stagnation pressure is given by
p
s
=
1
2
p
1
:
2

(4)
When the liquid begins to flow away from the point of impact, there is evidence that the
velocity of this tangential flow may be as much as five times the impact velocity (Thomas &
Brunton, 1970). The velocity increase is thought to be connected with the shape of the head
of the jet. It has been observed that an increase in velocity along the surface occurs only in
cases where the jet head is inclined at an angle to the surface. Since spherical drops (and/or
spherical heads of a train of pulses of pulsating jet) always provide a sloping interface to a
plane solid surface it might be expected that high radial velocities will occur on impact.
Therefore, there are additional shear forces associated with the high speed flow across the
surface acting on the surface in addition to the normal forces. The shear forces acting on a
roughened surface are large enough to cause local shear fractures, even in high strength
materials (Hancox & Brunton, 1966).
Exploitation of above described effects associated with water droplet impingement on solids
in a high-speed water jet cutting technology should lead to considerable improvement of its
performance, better adaptation to more and more demanding environmental requirements,
and consequently to more beneficial use of the technology also from the economical point of
view. Generating sufficiently high pressure pulsations in pressure water upstream the
nozzle exit enables to create a pulsating water jet that emerges from the nozzle as a
continuous jet and it forms into a train of pulses at certain standoff distance from the nozzle
exit. Such a pulsating jet produces all of the above mentioned effects associated with water
droplet impingement on solids. In addition, the action of pulsating jet induces also fatigue
stress in the target material due to the cyclic loading of the target surface. This further
improves the efficiency of the pulsating liquid jet in comparison with the continuous one.
Thus, destructive effects of the continuous high-speed water jet can be enhanced by the
introduction of high-frequency pulsations in the jet, i.e. by generation of pulsating water
jets.
Recently, a special method of the generation of the high-speed pulsating water jet was
developed and tested extensively under laboratory conditions. The method is based on the
generation of acoustic waves by the action of the acoustic transducer on the pressure liquid
and their transmission via pressure system to the nozzle. The high-pressure system with
integrated acoustic generator of pressure pulsations consists of cylindrical acoustic chamber
connected to the liquid waveguide. The liquid waveguide is fitted with pressure liquid
supply and equipped with the nozzle at the end. The acoustic actuator consisting of
piezoelectric transducer and cylindrical waveguide is placed in the acoustic chamber (see
Fig. 3). Pressure pulsations generated by acoustic actuator in acoustic chamber filled with
pressure liquid are amplified by mechanical amplifier of pulsations and transferred by
liquid waveguide to the nozzle. Liquid compressibility and tuning of the acoustic system are
utilized for effective transfer of pulsating energy from the generator to the nozzle and/or
nozzle system where pressure pulsations transform into velocity pulsations. The acoustic
generator can be used for generation of both single and multiple pulsating water jets (e.g.

rotating) using commercially available cutting heads and jetting tools. Laboratory tests of
the device based on the above mentioned method of the pulsating liquid jet generation
proved that the performance of pulsating water jets in cutting of various materials is at least
two times higher compared to that obtained using continuous ones under the same working
conditions.

Fig. 3. Schematic drawing of the high-pressure system with integrated acoustic generator of
pressure pulsations
However, further improvement of the apparatus for acoustic generation of pulsating liquid
jet requires thorough study oriented at determination of fundamentals of the process of
excitation and propagation of acoustic waves (and/or high-frequency pressure pulsations)
in liquid via high-pressure system and their influence on forming and properties of
pulsating liquid jet.
Problems related to the generation and propagation of pressure pulsations with frequency
in the order of tens of kHz in liquid under pressure of tens of MPa and subsequent
discharge of the liquid influenced by the pulsations through the orifice in the air (producing
pulsating liquid jet with axial velocity in the order of hundreds meters per second) were not
investigated in detail so far. Only partial information on this topic can be found in
publications dealing with processes of a fuel injection for combustion in diesel engines (see
e.g. Pianthong et al., 2003 or Tsai et al., 1999) and/or underwater acoustics (Wong & Zhu,
1995).
Therefore, the research on pulsating water jets was focused recently on the study of
fundamentals of the process of excitation and propagation of acoustic waves (high-
frequency pressure pulsations) in liquid via high-pressure system and their influence on
forming and properties of pulsating liquid jet as well as on the visualization of the pulsating
jets and testing of their effects on various materials. Results obtained in above mentioned
areas so far are summarized in following sections.
2. Acoustic wave propagation in high-pressure system with integrated
acoustic generator
The efficient transfer of the high-frequency pulsation energy in the high-pressure system to
longer distances represents one of the basic assumptions for generation of highly effective
pulsating water jets with required properties. To achieve that goal, the amplification of
pressure pulsations propagating through the high-pressure system is necessary. The
amplification can be accomplished by properly shaped liquid waveguide that is used for the
pulsations transfer to the nozzle. In addition, maximum effects will be obtained if the entire
high-pressure system from the acoustic generator to the nozzle is tuned in the resonance. To
Acoustic actuator Liquid
waveguide
Acoustic
chamber
Pressure
liquid
supply
Pulsating
jet
Nozzle

be able to study theoretically process of generating and propagation of pressure pulsations
in the high-pressure system, both analytical and numerical models of the system with
integrated acoustic generator were developed.
2.1 Analytical solution
The analytical solution of both pressure and flow oscillation waveforms in the conffuser-
shaped tube with circular cross-section is based on linearized Navier-Stokes equations and
wave equation for propagation of pressure wave. The wave equation incorporates both the
standard kinematical viscosity and the kinematical second viscosity that is related to the
liquid compressibility. Therefore, the irreversible stress tensor
ij
, on the basis of which the
wave equation is derived, can be written as follows:
L
]
= 2pc
]
+ o
]
_ 0(t )c
kk
()J
t
0

(5)
where the function (dynamic second viscosity) is related to the voluminous memory, and
c
ij
represents the tensor of deformation velocity.
In the frequency domain (), equation (5) can be written in simplified form verified
experimentally:
L
]o
= 2pc
]o
+ o
]
k
c
kko

(6)
whereby
ij
represents Kronecker delta, and dynamic viscosity. It is obvious from (6) that
the dynamic second viscosity is frequency dependent. The kinematical second viscosity is
then defined using following formula:
=
k
p

(7)
where p represents liquid density.
2.1.1 Wave equation
If one considers linearized Navier-Stokes equations, the wave equation for pressure function
can be written using the Laplace operator in the following form:
o
2
p
ot
2
2y
o
ot
(p) _ p
-1
(t )
o
o
(p)
t
0
J :
2
p = u

(8)

where is kinematical viscosity, p pressure, t time and v speed of sound in water,
respectively.
If Laplace transformation for zero initial conditions is applied in (8), following equation can
be obtained:
s
2
o s|2y + (s)]o :
2
o = u
(9)
where s represents parameter of the Laplace transformation according to time ((s) is the
Laplace function of the second kinematical viscosity), and, at the same time, following is
valid:

I{p(t)] = o(s)
(10)
If following expression is denoted :
2
= s
2
|:
2
+ (2y +)]
-1
(11)
then it can be written
2
o +o = u
(12)
In the frequency domain it is valid that (i) = k.
The solution of (12) can be performed by the implementation of spherical coordinate system
(r, , ), see Fig. 4. Now, the wave equation can be written in the following form:

Fig. 4. Implementation of spherical coordinate system

o
2
o
or
2
+
2
r
oo
or
+
1
r
2
o
2
o
ou
2
+
1
r
2
cotg
oo
ou
+
1
r
2
1
sin
2
u
o
2
o
o
2
+
2
o = u

(13)
Lets assume the solution of (13) as a product of functions:
o = Z(r)w(cosu)()
(14)
Then, individual particular integrals can be expressed as follows:
p
= A
n
cosn +B
n
sinn
(15)
w
p
(cosu) = H
nm
P
m
n
(cos) +N
nm
m
n
(cos)
(16)
where P, Q are special Legendre polynomials:
Z
p
=
1
r
|F
m
[
(r)+0
m
(r)]

(17)
=
1 + 4m(m +1)
2

(18)
2.1.2 Transfer matrix
The objective is to determine transfer matrix P that can be used in solving pressure and
flow pulsations in hydraulic systems in conffuser-type tubes. For this purpose, it is
convenient to introduce the mean velocity of the liquid c
r
in a direction of r using
following formula:
c
=
1
2u
0
_ c
(r, , u)Ju
u
0
0
(19)
The solution can be simplified by the assumption that the flow is rotary symmetrical. It can
be derived under the above mentioned assumption that functions P
m
n
,
m
n
will be
streamlined to the following form: P
m
0
,
m
0
. Further, considering that the pressure function p
varies only a little with respect to the angle , the following relation for the mean velocity c
r

can be written based on Navier-Stokes equations:
oc
ot
=
2u
0
p
op
or

2u
0
p
2
v
2
o
2
p
orot
(20)
The continuity equation and component c
r
in Navier-Stokes equations expressed in the
spherical coordinate system were used in the above mentioned derivation. Withal, effects of
dynamic viscosity were neglected. If we will keep considering zero initial conditions, it can
be written after the Laplace transformation (20):
sw
= o
oo
or
; I{c
(t)] = w
(s)

(21)

o =
2u
0
p
_1 +
s
v
2
]

(22)
If all assumptions of the solution are considered, following can be written for Laplace
images of both the pressure function and the velocity c
and with respect to (21):

o =
1
r
|F[
0.5
(r) + 0
0.5
(r)]

(23)
m = u; =
1
2

(24)
w
=
o
sr
_F
o[
0.5
(r)
oy
+0
o
0.5
or
_

(25)
If we introduce for r = r
0
the state vector
u
1
= |w
(r
0
, s), o(r
0
, s)]
(26)
and for r the state vector
u
1
= |w
(r, s), o(r, s)]

(27)
the dependence in locations r and r
0
can be expressed by means of the transfer matrix:
u(r, s) = Pu(r
0
, s)
(28)
Then, the matrix P will be derived from (26) and (27) by the elimination of integration
constants F, G. If we designate:
o =
o
r
_
o[
0.5
or
(r
0
)
0
(r
0
)
o
0.5
or
(r
0
)[
0.5
(r
0
)_
(29)
following relation can be written for matrix P:
P =
1
or
0
r
o
o[
0.5
or
(r) o
o
0.5
or
(r)
[
0.5
(r)
0.5
(r)

0
(r
0
) o
o
0.5
(r
0
)
or
[
0.5
(r
0
) o
o[
0.5
or
(r
0
)
(30)
In the frequency domain, s = i is substituted.
Both pressure and flow pulsations of hydraulic systems with conffuser-shaped tubes can be
solved on the basis of the transfer matrix (30). Individual elements of the transfer matrix are
dependent on values of the speed of sound and the second viscosity. Values of both these
quantities depend on the static pressure and the value of second viscosity depends also on
the frequency. The values can be determined experimentally using the transfer matrix.
2.1.3 Application of the transfer matrix
The transfer matrix derived in the previous section can be used in solving transmission of
pressure and flow pulsations in complex hydraulic systems. Such a system can consist of
cylindrical and confusser-shaped sections; the system can also be bifurcated.

The advantages of use of the transfer matrixes for determination of both pressure and flow
oscillation waveforms in the hydraulic system are illustrated on the model of the conffuser-
shaped tube with the circular cross-section. The tube consists of series of coaxial cylinders
with various diameters connected with cone frustums and filled with water at a pressure of
30 MPa. The acoustic generator of pressure pulsations located at the end of the largest
diameter cylinder vibrates at the frequency of 20 kHz. The cylindrical nozzle is situated on
the other end of the tube. The length of the largest diameter cylinder L
1
(and thus also total
length of the tube L) can be changed (see Fig. 5).

Fig. 5. Schematic drawing of the conffuser-shaped tube with circular cross-section
Effects of acoustic generator on the pressurized fluid in the tube are simulated by mean
velocity:
c
= c
0
c
ot
; w
0
=
1
s i

(31)
The cylindrical nozzle at the end of the conffuser is represented by the linear hydraulic
resistance, simulated by following equations:
p = zc
; o = zw
(32)
where represents the discharge coefficient.
Based on the transfer matrixes, a simulation model with the above described geometrical
configuration was elaborated. Firstly, the natural frequency of the hydraulic system in
question was determined using the simulation model. Next, the frequency was applied as
oscillation frequency of the acoustic generator and propagation of a generated forced
pressure waveform in the hydraulic system was investigated. Fig. 6 illustrates calculated
forced pressure waveforms in the simulated geometry related to the phase angle.
It can be seen from the Fig. 6 that oscillations of the acoustic actuator generate a standing
wave in the hydraulic system. The standing wave converts to the travelling wave at the area
close to the nozzle exit. It is obvious from the presented solution that a properly designed
conffuser-type tube can amplify the amplitude of pressure pulsations at the exit of the
pulsating nozzle.
The presented analytical solution of the problem of the acoustic (pressure) wave
propagation in the high-pressure system can be used for the determination of the optimal
geometrical configuration of the high-pressure system to operate in the resonance mode.
The solution can also be used in design of a transmission line for efficient transfer of high-
frequency pulsation energy to longer distances in high-pressure systems for generation of
pulsating high-speed water jets.


Fig. 6. Amplitudes of forced pressure waveforms in the simulated geometry related to the
phase angle calculated from the analytical model
2.2 Numerical model
Computational Fluid Dynamics (CFD) models of selected geometrical configurations of the
high-pressure system with integrated acoustic generator were created using CFD code
ANSYS CFD to simulate numerically the influence of operating and configuration
parameters of the acoustic generator and transmitting line on the generation and

propagation of acoustic waves (pressure pulsations) in high-pressure system and properties
of pulsating jet. The high-pressure system consisted of cylindrical acoustic chamber, liquid
waveguide provided with high-pressure water supply and the nozzle. To simplify the
model, acoustic actuator was substituted by vibrating wall of the acoustic chamber. The
fluid flow in the model was solved as 3-D turbulent compressible unsteady flow of water.
Water compressibility was taken into account in the numerical model using so called user
defined function (UDF). The UDF covers calculations of both water density and speed of
sound in water related to pressure:
p =
p
c]
1
p
K
=
p
c]
1
p p
op
K
(33)
o = 2. 1u
-6
p + 14S2 (S4)
where is water density [kg.m
-3
],
ref
is reference water density under normal conditions
(998,2 kg.m
-3
), p and p
op
are real and operating pressures [Pa], K represents bulk modulus of
water (2.2 . 10
9
Pa) and a is speed of sound in water [m.s
-1
] determined experimentally.
The numerical simulation of a high-pressure system equipped with an acoustic generator
was verified by the measurement of pressure pulsations in the high-pressure system
upstream to the nozzle exit using dynamic pressure sensors. The pressure waveform in the
numerical model was recorded at the same location as the pressure sensor was installed
during the laboratory measurement. It was found out that numerical model provides
information on the pressure waveform in high-pressure system that is in relatively very
good agreement with experimental measurement. Comparison of results of numerical
simulation and measurement also proved that the numerical model is able to simulate
influence of geometry changes on the amplitude of dynamic pressure accurately and thus
also to simulate pressure wave propagation and transmission in the high-pressure system.

Fig. 7. Standing wave amplitudes along longitudinal axis of the high-pressure system
After the verification of plausibility of results of numerical simulation by the laboratory
measurement, the model was used in studying of the process of propagation and
transmission of acoustic waves in the high-pressure system from the acoustic actuator to the

nozzle. An example of the behaviour of amplitudes of standing wave along the longitudinal
axis of high-pressure system can be seen in Fig. 7. Figure 8 illustrates forced pressure
waveforms in the simulated geometry related to the phase angle.

Fig. 8. Calculated forced pressure waveforms in the simulated geometry related to the phase
angle recorded along the longitudinal axis of high-pressure system from numerical model.
(Scale indicates amplitude of dynamic pressure in MPa.)

Results obtained from the numerical simulation correspond to results obtained from the
analytical one even if the numerical model used is not physically accurate in the close
vicinity of the nozzle outlet where cavitation occurs. Cavitation model was not implemented
in the numerical model with respect to the computational speed. Results of numerical
modelling clearly indicate that the geometrical configuration of high-pressure system
influences significantly propagation and transmission of pressure pulsations from the
acoustic actuator to the nozzle. The amplitude of pressure waves increases towards the
nozzle outlet due to the proper shaping of the liquid waveguide its frustums act as
mechanical amplifiers of the acoustic waves. At the same time, the amplitude of pressure
pulsations close to the nozzle outlet (where it has crucial influence on the pulsating jet
generation) changes significantly with respect to the geometrical configuration of the high-
pressure system.
3. Visualization of pulsating jet
The use of visualization plays an important role in the study of behaviour of pulsating water
jet. It enables not only the examination of characteristics of the jet such as mean velocity and
break-up length of the pulsating jet but also to study the morphology and processes of
formation of the pulsating jet and development of pulses in the jet. Furthermore, the
visualization can be used to validate results obtained from numerical simulation of the
process of generation of pulsating jets using CFD methods.
An original method of visualization of pulsating water jets based on the application of
stroboscopic effect was elaborated for the above mentioned purposes. The method enables
to obtain visual information not only on instantaneous structure of the pulsating jet but also
on the mean structure of the jet. In addition, the stroboscopic effect allows observing process
of formation of pulsating water jet by the naked eye. Special stroboscope for the pulsating jet
visualization was developed where the frequency of stroboscope flashing is controlled by
the frequency of pressure pulsations in the high-pressure system measured upstream from
the nozzle exit. An example of the mean structure of pulsating jet with pulsating frequency
of 20 kHz can be seen in Fig. 9. Exposure time of the photograph was 1/1000 s and the
frequency of stroboscope flashing was about 20 kHz, therefore the figure represents
superposition of 20 images of pulsating jet frozen by the stroboscope flashing.

Fig. 9. The mean structure of the pulsating water jet generated at 30 MPa (illumination by
the stroboscope)
An instantaneous structure of the pulsating water jet with the frequency of 20 kHz was
studied using the high-speed camera LaVision VC-HighSpeedStar 5 equipped with image
amplifier LaVision HighSpeed IRO. The record rate of the high-speed camera was 35 000
frames per second and the gate was set to 1s. In addition, visualization of instantaneous
structure of the pulsating water jet was also performed using Particle Image Velocimetry
(PIV) system consisting of LaVision Imager Intense camera and New Wave Research laser,

Model Solo 120 with pulse duration 3-5 ns; the optical system was used to produce 1 mm
thick sheet of light. An example of the pulsating jet visualized by high-speed camera can be
seen in Fig. 10, the same jet visualized by PIV system is presented in Fig. 11.

Fig. 10. The instantaneous structure of the pulsating water jet generated at 30 MPa (high-
speed camera)

Fig. 11. The instantaneous structure of the pulsating water jet generated at 30 MPa (PIV
system)
The fan (flat) pulsating water jet visualization was performed using the pulsed laser New
Wave Research and digital camera Nikon D70s. Figure 12 shows the morphology of fan
pulsating water jet generated at a pressure of 20 MPa.

Fig. 12. The instantaneous structure of the fan pulsating water jet generated at 20 MPa
(illumination by pulsed laser, camera Nikon D70s)
4. Effects of pulsating water jets on materials
Effects of pulsating water jets with the frequency of 20 kHz were tested on various types of
materials, such as metals, rocks and concrete. Tested materials were exposed to the action of

diverse types of jets: single round and fan pulsating jets as well as rotating pulsating jets.
The effects of pulsating jets were evaluated in terms of cutting depth, rate of mass-loss or
volume removal rate respectively and compared with the effects of continuous water jets
under the same operating conditions.
Obtained results show clearly the supremacy of pulsating water jets over continuous ones in
terms of their effects on material. Figures 13 to 15 illustrate the effects of various types of
both pulsating and continuous jets on metal, rock and concrete samples. Differences in the
surface structures created by pulsating and continuous water jets on individual materials
are clearly visible in the above mentioned figures. An example of erosion effects of pulsating
fan water jets (generated at various pressures) on aluminium samples at variable stand-off
distance is presented in Figure 16.

Fig. 13. Comparison of effects of pulsating (P) and continuous (C) water jets on samples of:
a) mild steel (pressure 40 MPa, nozzle dia. 1.98 mm, traverse speed 0.03 m.min
-1
, standoff
distance 140 mm), b) brass (pressure 40 MPa, nozzle dia. 1.98 mm, traverse speed 0.03
m.min
-1
, standoff distance 140 mm), c) duralumin (pressure 50 MPa, nozzle diameter 1.45
mm, traversing speed 0.05 m.min
-1
, standoff distance 60 mm) and d) basalt (pressure 50
MPa, nozzle diameter 1.45 mm, traversing speed 1.0 m.min
-1
, standoff distance 40 mm (P)
and 20 mm (C))
P C
a)
P C
b)
P C
c)
P C
d)


Fig. 14. Comparison of effects of rotating pulsating (A) and rotating continuous (B) water
jets on concrete sample (V
r
removed volume, pressure 30 MPa, nozzle diameter 2x1.47
-1
, standoff distance 40 mm (A) and 20 mm (B))

Fig. 15. Comparison of effects of fan pulsating (A) and fan continuous (B) water jets on
concrete samples (V
r
removed volume, pressure 30 MPa, equivalent nozzle diameter 2.05
-1
, standoff distance 40 mm)
A B
V = 22 cm
r
3
V = 4
r
cm
3

Results of the measurement of surface roughness characteristics on surfaces created by fan
pulsating jets indicate that the characteristics are strongly influenced by both the standoff
distance and the operating pressure. An example of the influence of a standoff distance on
arithmetic mean roughness (SRa) and average maximum height roughness (SRz) can be seen
in Figures 17 and 18, respectively. It should be pointed out that surface roughness (both SRa
and SRz) produced by the pulsating fan water jet out of the range of optimum standoff
distances (where the pulsating jet acts as a continuous jet) correspond to those produced by
continuous jets reported by Kunaporn et al. (2009). On the other hand, the fan pulsating
water jet produces surfaces with much higher values of the surface roughness (up to 20
times higher) within the optimum range of standoff distances (where the pulses are well-
developed in the jet) compared to continuous jets.

Fig. 16. Top: Pulsating fan water jets generated by nozzle with equivalent of diameter 1.10 mm
and spraying angle of 10 at various operating pressures. Scale on the left side of photographs
represents standoff distance in millimeters. Dots indicate the range of standoff distances where
maximum erosion effects of pulsating fan water jet occur. Bottom: Erosion effects of the above
pulsating fan water jets on duralumin samples. Scale on the right side of photographs
indicates standoff distance in millimeters; scale on the bottom indicates width in millimeters
1u
2u
Su
4u
Su
6u
7u
8u
-1u u 1u
1u
mm mm mm mm
2u
Su
4u
Su
6u
7u
8u
-1u u 1u
1u
2u
Su
4u
Su
6u
7u
8u
-1u u 1u
1u
2u
Su
4u
Su
6u
7u
8u
-1u u 1u
20 MPa 30 MPa 50 MPa 70 MPa

Results of the research of effects of pulsating jets on various materials obtained so far
indicate that the pulsating jets can be used advantageously for the removal of surface layers
of materials and/or rough cutting. However, further research will be necessary to be able
to use the pulsating water jets in applications of precise cutting.

Fig. 17. Influence of a standoff distance on arithmetic mean roughness (SRa) of the surface
created by the action of pulsating fan water jet generated by the fan jet nozzle with
equivalent of diameter 1.10 mm and spraying angle of 10

Fig. 18. Influence of a standoff distance on average maximum height roughness (SRz) of the
surface created by the action of pulsating fan water jet generated by the fan jet nozzle with
equivalent of diameter 1.10 mm and spraying angle of 10
0
20
40
60
80
100
120
10 20 30 40 50 60 70 80
S
R
a

[
m
]
Standoff distance [mm]
20 MPa
30 MPa
50 MPa
70 MPa
0
200
400
600
800
1000
10 20 30 40 50 60 70 80
S
R
z

[
m
]
Standoff distance [mm]
20 MPa
30 MPa
50 MPa
70 MPa

5. Conclusions
Presented results of the analytical solution and numerical simulation of the transmission of
acoustic waves in high-pressure system represent the first step in gaining knowledge
regarding processes of generation and propagation of high-frequency pressure pulsations in
the liquid under high pressures and their influence on forming and morphology of
pulsating liquid jets.
Results obtained from the visualization of pulsating water jets are used in studying of the
characteristics of the jets and to verify results obtained from numerical simulation of the
process of generating and forming of pulsating water jets. Laboratory and pilot tests of
effects of pulsating water jets on various materials showed clearly the potential of pulsating
jets to improve the performance of water jetting technology significantly.
It can be concluded that the research presented in the paper contributed to better knowledge
of processes occurring in areas of generation and propagation of high-frequency pressure
pulsations in the liquid under high pressure, their influence on forming and morphology of
pulsating water jets and effects of the jets on materials. However, it is still necessary to
further study problems of the efficient transfer of the high-frequency pulsation energy to
longer distances in the high-pressure system. This will enable creation of the highly effective
pulsating liquid jet with required properties.
6. Acknowledgements
The chapter has been done in connection with project Institute of clean technologies for
mining and utilization of raw materials for energy use, reg. no. CZ.1.05/2.1.00/03.0082
supported by Research and Development for Innovations Operational Programme financed
by Structural Founds of Europe Union and from the means of state budget of the Czech
Republic. Presented work was also supported by the Academy of Sciences of the Czech
Republic, project No. AV0Z30860518. Author is thankful for the support.
7. References
Bowden, F. P., & Field, J. E. (1964). The brittle fracture of solids by liquid impact, by solid
impact, and by shock. Proceedings of the Royal Society of London. Series A,
Mathematical and Physical Sciences, Vol. 282, No. 1390, pp. 331-352
de Haller, P. (1933). Untersuchungen ber die durch Kavitation hervorgerufenen
Korrosionen. Schweizerische Bauzeitung, Vol. 101, No. 21& 22, pp. 243-246,260-264
Hancox, N. L., & Brunton, J. H. (1966). The erosion of solids by the repeated impact of liquid
drops. Philosophical Transactions of the Royal Society of London. Series A, Mathematical
and Physical Sciences, Vol. 260, No. 1110, pp. 121-139
Kunaporn, S., Chillman A., Ramulu, M., & Hashish, M. (2009). Effect of waterjet formation
on surface preparation and profiling of aluminum alloy. Wear, Vol. 265, No. 1-2,
pp. 176-185
Pianthong, K., Zakrzewski, S., Behnia, M., & Milton, B. E. (2003). Characteristics of impact
driven supersonic liquid jets. Experimental thermal and fluid science, Vol. 27, No. 5,
pp. 589-598

Raghavan, C. & Ting, E. (1991). Hyper pressure waterjet cutting of thin sheet metal.
Proceedings of 6th American Water Jet Conference, pp. 493-504, ISBN 1-880342-00-6,
Houston, Texas, August, 1991
Thomas, G. P., & Brunton, J. H. (1970). Drop impingement erosion of metals. Proceedings of
the Royal Society of London. Series A, Mathematical and Physical Sciences, Vol. 314, No.
1519, pp. 549-565
Tsai, S. C., Luu, P., Tam, P., Roski, G., & Tsai, C. S. (1999). Flow visualization of Taylor-mode
breakup of a viscous liquid jet. Physics of fluids, Vol. 11, No. 6, pp. 1331-1341
Wong, G. S., & Zhu, S. (1995). Speed of sound in seawater as a function of salinity,
temperature and pressure. Journal of the Acoustical Society of America, Vol. 97, No. 3,
pp. 1732-1736
16
Molecular Desorption by LaserDriven
Acoustic Waves: Analytical Applications
and Physical Mechanisms
Alexander Zinovev, Igor Veryovkin and Michael Pellin
Argonne National Laboratory
USA
1. Introduction
Analytical mass-spectrometry (MS) is a powerful, widely-used tool for materials analysis,
helping to make progress in materials and environmental sciences, chemistry, biology,
astrophysics, etc (Dass 2007). Often the sample to be studied (analyte) is a solid requiring: a)
volatilization/desorption of the analyte atoms/molecules and b) their consequent
conversion to the charged particles (ionization) prior to mass analysis. The last two decades
have seen revolutionary advances in these techniques (Dass 2007) and the use of direct laser
irradiation to achieve volatilization is one of the wide-spread methods (Lubman 1990)
These pulsed laser-based techniques for the desorption/emission of the atoms, molecules
and ions from the surface of solids has benefitted from fundamental study of the process
beginning with the invention of the lasers (Honig and Woolston 1963) . A short laser pulse
hitting a solid absorbing surface delivers high energy in a small volume inducing a
variety of state changes. One consequence is the evaporation/desorption of surface atoms
and molecules could be used for further analysis by MS technique. However, the
increasing use of MS methods in analytical chemistry of organic and biomolecules
revealed that this direct desorption process had significant drawbacks for the analysis of
molecular solids. Most importantly, the high energy density produced during irradiation
results in not only surface heating but also in excitation of internal vibrational and
electronic states of desorbed molecules leading to their partial or even complete
fragmentation (Lubman 1990). This difficulty was overcome for many samples by the
development of Matrix Assisted Laser Desorption and Ionization (MALDI), which by
imbedding the analyte in a specialized UV absorbing molecular solid (the matrix)
allows UV lasers to both desorb and ionize large organic and biomolecules without
significant fragmentation (Cole 2010). Because MALDI combines both of the needed initial
processes (desorption and ionization) it very quickly following the pioneering publication
(Karas, Bachmann et al. 1985) became a key analytical tool. MALDI is now one of
principle research tools in proteomics (Cole 2010) and its discovery was recognized with
the Nobel Prize in chemistry in 2002.
Despite the success of the MALDI technique current active areas of research include
quantification and analysis in the low mass region. Application of MALDI to analyte
quantification while possible requires careful attention to matrix/analyte sample


344
preparation, a detailed understanding of the crystallization process with regard to
the analyte, and careful many spot analyses (to find the sample signal average
which often varies by orders of magnitude as a function of laser position). (Duncan, Roder
et al. 2011)
This desire to find a discriminative, sensitive and more easily quantifiable alternative to
MALDI has lead us to re-examine another molecular desorption method that doesnt require
the use of a matrix. The first observation of this method of molecular desorption (later called
Laser Induced Acoustic Desorption, LIAD) belongs (to the best of our knowledge) to B.
Lindner (Lindner and Seydel 1985). This observation has been followed by several studies
including (Golovlev, Allman et al. 1997) where the abbreviation LIAD was introduced and
by (Perez, Ramirez-Arizmendi et al. 2000) who applied the technique to several classes of
MS problems.
LIAD has rather simple experimental layout: an analyte is deposited onto the front surface
of thin metal foil (the substrate), which is irradiated from the back (i.e. the side opposite to
both the analyte and the mass spectrometer) by a pulsed laser beam with power density
insufficient to pierce the foil. Such irradiation results in the volatilization of the analyte.
The volatilization is largely in the form of neutral molecules that can be utilized for
further MS analysis using an appropriate post-ionization method. The method seems to be
relatively insensitive to the sample preparation method. Commonly the sample
preparation requires only evaporating a drop containing a few nano-moles of the analyte.
Remarkably little desorption induced fragmentation is seen when a suitable soft
ionization method such as VUV photoionization can be found. It is also useful to note that
while the molecular signal depends on drive laser intensity the fragmentation observed
varies only weakly.
The advantages of LIAD have been demonstrated by many researchers; however, the
mechanism is not well understood. The first desorption mechanism was proposed
(Golovlev, Allman et al. 1997). It was supposed that because the metal is opaque and
completely blocks the direct interaction between the drive laser light and adsorbed, front-
side molecules, the only possible way of energy transfer is the mechanical. In this model, the
interaction of laser pulse with metal foil backside resulted in formation of acoustic waves
that move through the foil inducing a front surface oscillation motion. The molecules that
are sitting on this surface desorb due to a simple shake-off mechanism similar to those
that we use to remove dust particles from our clothes by shaking it.
A difficulty with this model arises when considering the relatively strong surface binding
energy experienced even by physisorbed molecules. In order to be efficiently desorbed from
the surface, surface molecules need to achieve initial kinetic energies exceeding their surface
binding energies (typically in the range of 0.05 - 0.5 eV for physically adsorbed molecules
(Adamson and Gast 1997)). This corresponds to velocities of a few hundred m/s for
molecules with masses of a few hundred atomic mass units. Unfortunately, acoustic
vibrations have mass transfer velocities much lower than the speed of sound, and in elastic
deformation mode, this velocity does not exceed a few m/s (Landau and Lifshits 1987).
While laser-driven acoustic wave generation in metals is very well studied problem
(Hutchins 1985), the physics of their generation in metal foils is crucial to understanding the
need for development of new desorption mechanisms. In the next paragraphs we will give a
brief theoretical overview and will present our experimental results on laser-driven acoustic
waves in thin metal foils.
Acoustic Waves: Analytical Applications and Physical Mechanisms

345
2. Laser-driven acoustic waves in thin metal foils
The interaction of pulse laser beam with metal surface is very complex phenomenon but our
specific interest is in formation of the acoustic waves in irradiated material. To generate an
acoustic wave a time dependent stress needs to be applied to the solid. A laser pulse is an
excellent tool to generate this kind of the stress. There are two principal mechanisms of
laser-induced stress formation in the solids: a) thermal stress, resulted from the non-uniform
heating of the irradiated surface by the laser beam; b) mechanical stress due to mechanical
impulse transferred from the leaving plasma plume formed on the surface during laser
ablation. The parameters of the acoustic waves generated for these two processes are
slightly different and will be discussed in details later. Here we will use acoustic wave
theory but one should note that its use is applicable only when the magnitude of the
applied stress is small in comparison with the Youngs modulus of the material. Large
applied stresses can cause the development of the shock waves a phenomenon with
different characteristics than acoustic waves (Menikoff 2007). Shock waves have been also
hypothesized to be the driver of the LIAD phenomenon, and, as such cannot be entirely
excluded from consideration, especially in some extreme cases. Nevertheless, during the last
decade, laser-driven acoustic waves emerged in the literature as the prime suspect in the
LIAD case.
2.1 Acoustic waves in metal foils
The general governing equation for the generation of elastic waves in solids can be derived
by combining the equation of motion and the Hookes law. In general case it is a differential
tensor equation, which interconnects the stress tensor, applied to the body, the displacement
of the bodys elemental volumes (strain), and their elastic properties. In order to analyze the
data in detail, the appropriate stress tensor needs to be determined, and the corresponding
set of partial differential equations for strain and stress must be resolved (Pollard 1977). One
can simplify this analysis by taking into account specifics of the experiments. For thin foils
with h/R
0
<<1 (where R
0
is the radius of the target foil, and h is its thickness), a round thin
plate approximation can be applied to describe and analyze this problem (Smith 2000). The
rise of the strain due to laser heating and the consequent development of the plasma plume
can be considered as an external driving force. Generally speaking, the thin plate equation is
a differential equation of the forth order,

2
2 2 2
2
2 2 2 2
1 1
2 ( , )
| | | |
+ + + =
| |

\ . \ .
L
v F r t
r r r r t t

(1)
where =(r,t) is the surface displacement in the z direction (perpendicular to the sample
surface), F(r,t) is the external driving force caused by the laser irradiation, v
L
is a parameter
depending on the material density, , thickness, h, and flexural rigidity, D. Furthermore,
/ =
L
v D h , and ,
3 2
( /12 (1 )) = D N h (2)
where is the Poisson ratio, N is Youngs modulus and is the oscillation decay constant. A
vibrating thin plate is one of the most frequently analyzed mathematical problems, and its
detailed analysis can be found elsewhere (McLachlan 1951; Smith 2000). In the discussion
given below, we will remain in the framework of the analysis given by (Smith 2000).


346
Assuming separate solutions for the radial and tangential terms, Eq. (1) can be converted
into a system of three differential equations of the second order. Being primarily interested
in the foil displacement in normal (z ) direction to the surface at the epicenter (r=0) and
assuming that the external driving force F(t) is a pulse function lasting a specific time , we
can separate the variables in the Eq. (2) for t>>. Under the assumption of harmonic motion
for all modes, the surface displacement can be expressed as
,
1
( , ) ( )

=
m n
i t t
r t r e

, where
m,n
is the vibration frequency. The general governing equation can then be written in the
following form:

2 2
2
, 1 2 2
1
0
| |
+ =
|

\ .
n m
n
k
r r r r
(3)
where n is an integer number,
, ,
/ =
n m n m L
k v and v
L
is the wave velocity as in Eq.(2).
Under the assumption that the oscillations are harmonic and decay exponentially, the
following solution for the surface displacement can be obtained (McLachlan 1951):

( )
0 , ,
( , ) ( ) ( ) exp( ) exp( ) = +
n n n m n n m n
r t J k r I k r i t t (4)
where J
n
and I
n
are Bessel functions and is a constant. The term describing the angular
dependence of the oscillations is omitted in Eq.(4). The fact that foils used in LIAD
experiments are typically glued or welded on their perimeter corresponds in our analysis to
the situation when edges of the round plate are fixed (i.e. non-vibrating), and is described by
the following boundary conditions:

0
0
( , )
( , ) 0, 0 = =
d R t
R t
dt
, (5)
that lead to an equation, whose solutions j
n,m
have tabulated values (Smith 2000). The
corresponding vibration frequencies,
n,m
, can be then expressed as

2 2
, 4 2
, , 4 2 2
0 0
12 (1 )
=

n m L
n m n m
j h v N
j
R R

(6)
Because values of are small compared to the first term under the square root sign in Eq.(6),
it is a reasonable assumption that the frequency is proportional to the square root of the
ratio of Youngs modulus, N, to the density of the foil material, . As described by Eq. (6), in
the steady-state regime (driving force F(t)=0 for t>>) the frequency spectra and decay times
of the oscillation will remain the same while the laser intensity is varied, and only the
amplitude should change.
For short times, (t<), the approach to solving Eq.(1) is to assume that the external force is a
delta-function in space F(r,t)=F(t)(r) (point source), allowing the solution of Eq.1 to be
expressed in the form
( , ) ( ) ( ) = r t g r t (7)
where g(r) and (t) represent the spatial and the time dependencies of the final solution,
respectively. This approximation should help to develop a clearer understanding of physical
problems related to the laser generation of acoustic waves in thin foils.

347
Using Eq. (7), Eq. (1) can be split into two independent equations, with the equation for (t)
having the form

2
2
2
2 ( ) + =
n
d d
F t
dt dt

, (8)
and the equation for g(r) being a non-uniform Bessel equation with the (r)- function in the
right side of it.
The exact solution of Eq. (8) can be expressed as the convolution of the Green function of the
problem (8) and real time shape of F(t). The governing equation for the Green function will
be Eq.(8) with -function in the right side. It can be easily derived by applying the Laplace
transform to the Eq.(8), subsequently solving the obtained linear equation in the s-space and
returning back to the time space with using the inverse Laplace transform. As a result of
these procedures, one can finally obtain the following equation

( )
2 2
2 2
( ) sin ( ) ( )

(
=
(

}
t t
n n
o
n
e
t t F d

, (9)
and the complete solution of Eq.(8) can then be obtained as the sum of the components of
Eq.(9) over n.
Thus, the generation of acoustic waves in thin foils can be described by Eq.(9) which
strongly depends on driving force F(t) after whose cease the vibration evolves into decaying
harmonic oscillations (Eq.(4)) with frequencies defined by Eq.(6). It is apparent that maximal
surface velocities can be achieved only at the initial stage of acoustic wave generation when
t<. The application of Eq.(9) to analysis of laser-driven acoustic vibrations is complicated by
the lack of the exact knowledge of the time profile of the driving force F(t). Depending on
regime of the surface irradiation, this force may be of different origins and, accordingly,
have strongly varying magnitudes and time profiles. The appropriate mechanisms will be
discussed in the following section.
2.2 Generation of the acoustic waves by laser pulses
2.2.1 The action of laser pulse on the metal surface: heating and plasma generation
Metals subjected to pulsed laser irradiation absorb energy within a very thin surface layer
(the skin-depth for most metals is less than 10
-7
m) so that the temperature of the irradiated
surface can rise extremely fast. For moderate laser intensities (below the plasma formation
threshold) and a Gaussian-shaped laser beam, the maximum surface temperature can be
estimated using a well-known expression (Prokhorov, Konov et al. 1990)

1/2
max
max 1/2
(0) 2.15
( )
=
AI
T
c

, (10)
where T
max
(0) is the maximum surface temperature at z=0 (the z direction is orthogonal to
the target surface), A is the laser radiation absorption coefficient, I
max
is the peak laser power,
is the laser pulse duration, and c, , and are the specific heat, density and thermal
conductivity of the corresponding metal, respectively.
After cessation of the laser pulse, the adsorbed energy continues to diffuse into the bulk
metal and along the metal surface, resulting in a temperature increase underneath the


348
irradiated spot and outward from the spot along the surface. Temperature evolution at any
moment, t>, and for any position, z>0, proceeds according to the following equation
(Prokhorov, Konov et al. 1990):

1/2
1/2 1/2 max
1/2 1/2
2
( , ) [ ( ) ( ) ( )]
2( ) 2[ ( )]
=

AI z z
T z t t ierfc t ierfc
t t

(11)
where is the thermal diffusivity of the metal, which can be expressed as / = c .
The function ierfc(x) is given by

1/2 2
( ) {exp( ) (1 ( ))}
= ierfc x x x erf x (12)

where
2
0
2
( ) exp( ) =
}
x
erf x d

Eqs. (10) and (11) are the solutions of the one-dimensional heat diffusion equation and are
valid only if the laser beam size, r
0
, is significantly greater than both the foil thickness h and
the thermal diffusion length l
th
calculated as l
th
=(t)
1/2
.
The strong rise of the surface temperature given by Eq.(10) results in the surface melting and
evaporation, as well as in plasma plume formation (ablation regime) (Miller and Haglund
1998). Despite the fact that laser plasma generation and evolution have been the focus of
numerous studies, no general mechanisms exist that describe the plasma recoil pressure on
the surface for a broad range of laser intensities (Phipps, Turner et al. 1988), due to the
complexity of the phenomenon. For GW/cm
2
peak laser powers, hot and dense plasma is
formed in the vicinity of the surface, which can screen the surface and prevent laser
radiation from reaching it. In this case, the ablative pressure very weakly depends on the
target material parameters (Phipps, Turner et al. 1988) and has a sub-linear dependence on
laser intensity. In a semi-regulating, one-dimensional plasma model (which can be applied
to our case as a simplified, first-order approximation), this equation is written, as follows
(Gospodyn, Sardarli et al. 2002) :

8 3/4 1/4
,max
7.26 10 ( )

a
P I , (13)
where I is expressed in GW/cm
2
, in microns, in nanoseconds and P
a,max
in Pa. While Eq.
(13) was derived for an aluminum target in vacuum and for a supercritical plasma density, it
exhibits only a weak dependence on the atomic mass, A, of the material irradiated (A
-1/8
)
(Gospodyn, Sardarli et al. 2002) and may be applicable to specific experiments only as an
upper limit estimate. For lower laser intensities (<1 GW/cm
2
), the plasma plume
transmittance strongly varies with laser intensity (Song and Xu 1997), depending upon the
plasma density and temperature. In this case the evaporated surface material is ionized only
partially and the total mass of the evaporated atomic cloud are exponentially increasing
with the surface temperature and, hence with the laser intensity (Murray and Wagner 1999).
2.2.2 Thermal and plasma driven acoustic waves in metal foils
The temperature rise, as heat is transported into the solid causes linear thermal expansion
resulting in the development of thermoelastic acoustic waves in the irradiated metal. Eqs.
(10) and (11) can be used to determine the driving force which produces the waves. In

349
accordance with the general theory of thermal stresses in thin plates (Boley and Weiner
1960), a non-uniform heating of the surface is equivalent to a negative loading pressure and
may be expressed as

/2
2
/2
( , )
1

(
= (
(

}
h
T r
h
N
P T r z zdz
, (14)
where the temperature distribution over z is described by the Eq. (11) and is the linear
thermal expansion coefficient.
If the loading force is negative (i.e., directed backwards, towards the heating laser beam, Eq.
14), it is not surprising that an initial depression observed in the foil surface is opposite to
the heating laser beam. Similar results have been reported in the literature (Scruby 1987) for
thicker metal samples where the thin plate approximation was not applicable. Maximum
amplitudes and shapes of the observed depression vary for different metals and are defined
by both the temperature profile (Eq. (11)) and by elastic properties of the material.
In the case of plasma formation, the situation becomes more complex. The amplitude of the
driving force can be estimated using an expression similar to Eq.(13), whereas the time
profile of generated stress pulse is the subject of experimental study (Krehl, Schwirzke et al.
1975). Laser plasmas formation and their interaction with the surface is very complex and
multi-variable problem, which can only be solved in the framework of some model
assumption (Mora 1982). This is why direct experimental studies of laser-driven surface
vibrations should be an essential part of any acoustic wave related desorption phenomena.
2.2.3 Experimental observations of laser-generated acoustic waves in thin foils
Experimental studies of acoustic waves in solids due to pulsed laser irradiation have started
with the advent of such lasers (White 1963). A great collection of experimental results and
theoretical analyses of acoustic wave generation in solids driven by laser pulses has been
accumulated since (Hutchins 1985), and these studies continue at present (Xu, Feng et al.
2008). Regrettably, there is a very limited data set, which could be used to interpret of LIAD
experiments. To prove (or disapprove) the shake-off hypothesis of molecular desorption,
direct measurements of thin foil surface velocities in back-side irradiation geometry are
required. The scarcity of such data motivated us to setup a series of our own experiments
aiming at measurements of thin foils vibrations under typical LIAD conditions.
Experimental approaches to this problem are well known and described in the literature
(Scruby and Wadley 1978; Royer and Dieulesaint 2000). Nevertheless, we will briefly
describe below our system, in order to create a better stage for presentation and discussion
of the original results.
2.2.3.1 Experimental technique: optical and electrical methods
One of the most popular and widely used methods to studies of acoustic waves is based on
non-contact optical measurements. Figure 1a shows the experimental setup for
measurements of surface displacement using interferometry-based approach. A He-Ne laser
(1) (Melles-Griot, 543 nm, 0.5 mW) was used as the light source for a Michelson
interferometer. It consisted of a beam splitter (5), an etalon and steering mirrors (3, 4), a
focusing lens (6), a target (7), an imaging lens (9), an aperture (13), a focusing lens (14) and a
photomultiplier (15). The target was back-irradiated by a pulsed laser (12) through a fused
silica lens (8). Laser beam parameters were measured by intersecting the laser flux with two


350
partially reflecting (8%) quartz plates (10, 11) directing reflected beams onto a fast
photodiode (16) and an energy meter (17), respectively. A wedge-type optical attenuator
(18) was used to balance the Michelson interferometer shoulders and to increase the contrast
of the resulting interference pattern. Focusing lenses (6) and (8) were mounted on three-axis
translation stages. In this arrangement, both the acoustic-wave generating and diagnostic
laser beams could be independently focused and translated to different points on the target
surface. A lens (9) formed the magnified interference pattern in the plane of an aperture (13)
whose diameter was selected to be equal to the width of the dark band of the lowest
interference order. A second lens (14) was used to collect the light, which passed through
the aperture (13), and to direct it to the photomultiplier (PMT, 15).
The anode of the PMT was terminated with a 50 Ohm load to allow optical signals with rise
times as short as 5 ns to be measured. This capability was confirmed by demonstrating that
the shape of a 12 ns, NdYAG laser pulse was identical when measured by using this
detection system and by a high-speed avalanche photodiode. The measurements bandwidth
was limited by the PicoScope 3206 oscilloscope (200 MHz bandwidth and 200 Ms/s
sampling rate), which was used for signal acquisition. The digitized PMT signal was
transmitted to a PC via USB port and stored for further processing. The oscilloscope was
triggered by a pulse from the fast photodiode (17). The measured lag between the trigger
signal and the PMT signal was less than 40 ns. The maximum signal amplitude,
corresponding to the peak-to-valley ratio of the interference pattern, was 80 mV. The
minimum detectable signal was 5 mV at signal-to-noise ratio of about 3, which corresponds
to a surface displacement of approximately 25 nm. However, because of the strong electrical
noise generated by the Q-switch of the laser, the smallest surface displacement detectable in
this series of experiments was about 40 nm.

4 13 14
1
2
3
5
6
7
8
9
10 11 12
15
16
17
18

d
1
2
3, 3
4
5
6
7
8
9
10
11, 12

Fig. 1(a,b). Schematic drawing of the experimental setup for laser-driven acoustic wave
studies. (a) Interferometry method; (b) Capacitance method
The same target irradiation scheme as described above was also used for the capacitance
transducer measurements (Fig.1b). In contrast to the previous approach, a metal pin was
placed in front of the target. This pin served as the second plate of a capacitor whose first
plate was the target. These two capacitor electrodes were separated by a small gap d,
typically about 100 m. Both the target and the pin were fixed in optical mounts that
allowed alignment in the plane of the sample surface. In addition, the mount of the pin was
placed on a translation stage (9), driven by a picomotor, which could move the target (with
the precision of 1 m) in the direction orthogonal to the target surface. The pin had a
diameter of 3 mm and its end was polished flat. In general, the design of this detector is

351
similar to that described in Ref.19. The pin was connected to the input of a miniature charge
amplifier (7) powered by a constant (20 mA) direct current (DC) supply (8) and connected to
the oscilloscope (10). The bandwidth of the charge amplifier was about 2 MHz, which
provided a signal rise time of less than 400 ns. To increase the overall sensitivity of the
detector, a positive bias potential of 100 V was applied to the target.
The sensitivity of the transducer to surface displacement can easily be expressed in terms of
a planar capacitor

2
=
VS
q d
d
(15)
where q is the change in the electric charge, V is the applied voltage, S is the surface area of
the pin tip, d is the width of the gap between the electrodes and d is the change of the
width. The sensitivity of the charge preamplifier was 10 mV/pC, which corresponded to a
minimal detectable signal of about 5 mV (thus yielding reasonably good signal-to-noise
ratio). With an applied bias potential of -100 V, the estimated detection limit of the
transducer-based sensor was about 5 nm.
2.2.3.2 Experimental results: displacement and surface velocity of thin foils
Foils with various thicknesses (from 12.5 m up to 100 m) made from different materials
were used in our experiments. Materials were selected to span the range from soft metals
(Au, Al and Ni) to refractory metals (W, Mo and Ta) and semiconductors (Si). For each
experiment, the front surface of the sample was mechanically polished to roughness of less
than 0.250 m (RMS). After polishing, the foils were glued with silver epoxy to the rim of a
hollow quartz cylinder (8 mm outside diameter, 8 mm height, and 0.5 mm wall thickness).
The epoxy was cured for 2 hours in an oven at temperature of 100 C. Due to the differences
between thermal expansion coefficients of the foil materials and quartz, the foil stretched
over the top of the quartz cylinder once the assembly cooled to room temperature. The
tension was not very strong (according to our estimates, the total radial force did not exceed
1 N), and therefore, the foils in our experiments may be considered as supported at the
edges. The silver epoxy also provided a conductive path between the sample and the
instrument by placing a silver epoxy track along the quartz cylinder side.
Lasers generating both ultraviolet (UV) and infrared (IR) light were used for target
irradiation. The UV light was generated by an ArF excimer laser (EX10-300, GAM, Inc.)
having a wavelength 193 nm and a pulse duration of 15 ns. The output pulse energy could
be varied from 0.4 to 4 mJ by using neutral density optical filters. The laser radiation was
focused onto the backside of the target (opposite to the surface displacement sensors) using
a fused silica lens with a focusing distance of f=300 mm. The irradiated spot on the target
had rectangular dimensions of 100500 m, which corresponded to a UV laser power
density in the range of 50500 MW/cm
2
. For IR irradiation of the target, a Q-switched
Nd:YAG laser (Continuum) was used. This IR light had a wavelength of 1064 nm, a 12 ns
pulse duration and an output energy in the range of 115 mJ/pulse. The IR laser beam had a
Gaussian profile and produced a spot on the target surface with a nominal diameter of 500
m, corresponding to peak power density of 40600 MW/cm
2
.
Waveforms representing foil oscillations were measured over the time range from 5 s to 5
ms. For times much greater than the laser pulse duration (t>>), a decaying quasi-harmonic
oscillations were observed for all materials. The measured time dependence of the
displacement of the foils irradiated by laser pulses with different intensities exhibited


352
qualitatively similar behavior, although amplitudes, frequencies and decay times varied.
These results suggest that each foil behave as a mechanical system able to oscillate in a free-
running mode after the external force is removed. Fast Fourier Transform (FFT) analysis
applied to the measured data has shown that the frequency spectra consisted of discrete
lines (modes) appearing in the range of 10100 kHz.
In contrast to the steady-state regime (t>>) when different foils oscillated very similarly, the
initial moment (t<) of the evolving oscillation was distinctly different for each foil. Fig. 2
presents the time dependence of the surface displacement for different metals at low
irradiation intensities (50 MW/cm
2
) for early times in the range up to 50 s. For all
measurements at low intensities, the initial displacement is found to be negative, indicating
that the surface is first depressed (i.e. towards to the driving laser beam and,
correspondingly, away from the detector). Increasing the laser intensity leads to the initial
surface vibration waveform displacement changing from negative to positive. This is due to
the recoil pulse which occurs when material is ablated from the irradiated surface due to
plasma formation. Fig. 3 shows the time dependences of the displacement of Ta foil surface
for different laser irradiation intensities. The plasma formation threshold for this Ta foil was
~220 MW/cm
2
, as determined by the observation of the plasma plume glow in a separate
experiment. The FFT analysis of the experimental data at higher laser intensities revealed
that the frequency spectrum of the foil oscillations in the non-steady-state regime contains
much higher frequency components than found for the steady-state case. After a few tens of
microseconds, the high frequency components disappear as the foil oscillation become
harmonic as described by Eq.(4).

-1.0x10
-5
0.0
1.0x10
-5
2.0x10
-5
3.0x10
-5
4.0x10
-5
5.0x10
-5
-20
-15
-10
-5
0
5
10
D
i
s
p
l
a
c
e
m
e
n
t
,

n
m
Time, s
Ni
Ta
Au
Al

Fig. 2. Time dependent displacement for different metal foils at low laser intensities (50
MW/cm
2

The obtained results are in good agreement with time dependencies of surface displacement
measured under slightly different experimental conditions (Hutchins 1985). This fact clearly
demonstrates that laser-driven acoustic waves in thin metal foils have no experimental
peculiarities distinguishing them from well-known acoustic wave mechanisms. This result
allowed us to calculate surface velocities using the measured surface displacements (Fig. 4).
As one can see from Fig.4, these velocities are indeed in the range of meters per second.

353
-1.0x10
-5
0.0 1.0x10
-5
2.0x10
-5
3.0x10
-5
4.0x10
-5
5.0x10
-5
-100
-50
0
50
100
150
200
D
i
s
p
l
a
c
e
m
e
n
t
,

n
m
Ti me, s
100 MW/cm
190 MW/cm
320 MW/cm
430 MW/cm
510 MW/cm
2
2
2
2
2

Fig. 3. Time dependent displacement of Ta foil (12.5 m thick) at different laser intensities

0.0
5.0x10
-6
1.0x10
-5
1.5x10
-5
-0.2
0.0
0.2
0.4
0.6
0.8
V
e
l
o
c
i
t
y
,

m
/
s
Time, s

Fig. 4. Time dependence of Ta foil (12.5 mm thick) surface velocity at 400 MW/cm
2
driving
laser intensity
This picture clearly confirms the statement above that the mass transfer velocity (or surface
displacement velocity, in terms of our experiment) is much slower than the speed of sound
in metals. In its turn, this result supports our hypothesis that the vibrational motion of the
foil surface cannot serve as direct cause of molecular desorption, and that the real physical
mechanism of LIAD is not as simple as mechanical shake-off.
3. Desorption of the molecules from back-irradiated thin metal foils
3.1 Laser desorption in modern MS: methods and applications
As shown in the previous sections, laser induced desorption phenomena play important
role in modern MS. The primary role of the laser beam there is to deliver high energy
density into some (typically small) volume of the analyte. Due to the local overheating this
volume is volatilized forming hot and dense vapor plume, which might be partially ionized.
This ionization phenomenon can be considered as a great advantage of laser desorption
because there is no need for an additional ionization step, so that the desorbed ions may be
directly analyzed by a mass-spectrometer. At the same time, this can be a significant
drawback, because, due to collisions in the plume, organic molecules may fragment to the
point that their mass analysis becomes meaningless (Miller and Haglund 1998). While using
UV lasers in MS analyses of organic materials often produced encouraging results, it is well
recognized in the literature that general mechanism that is applicable to all organic solids


354
at all UV wavelengths does not exist(Srinivasan and Braren 1989). The introduction of
MALDI gave strong indication that many problems, associated with laser desorption MS
might have been solved. However despite popularity of MALDI in MS analyses of proteins,
lipids and many other organics (Schiller, Suss et al. 2007), this method cannot be considered
as universal because it requires to identify efficient matrix substances for different organic
species, and often to develop specialized sample preparation protocols. And, regrettably,
MALDI MS cannot be used to directly characterize mixtures of unknown molecules. This is
why the search for more versatile and universal methods of molecular desorption/ionization
is still on in the analytical mass spectrometry community. From this perspective, the ability of
LIAD to volatilize different kinds of organic molecules without noticeable (or, very often,
without any) fragmentation has attracted strong interest among researchers.
3.2 Laser-induced acoustic desorption
As described above, the acronym LIAD was suggested in the work conducted by the Chens
team (Golovlev, Allman et al. 1997) where the hypothesis about the acoustic wave nature of
the desorption process was expressed. However, LIAD remained just an interesting
observation until Kentamaas team of researchers from Purdue University (Perez, Ramirez-
Arizmendi et al. 2000) took on it and demonstrated successful applications of LIAD for the
MS analysis of different organic species like cytosine, guanine, thimidine and some others.
This work was followed by the series of studies where the applicability of LIAD to the MS
analysis of a wide range of organic samples has been demonstrated. The LIAD
volatilization method was successfully coupled with Fourier transform ion cyclotron
resonance mass spectrometer and alanylglycine (Reid, Tichy et al. 2001), saturated
hydrocarbons (Campbell, Crawford et al. 2004), polyethylene (Campbell, Fiddler et al.
2005) and even petroleum distillates (Crawford, Campbell et al. 2005) were analyzed. The
great advantages of this technique were the ability to efficiently volatilize various
organics and the simplicity of its incorporation into different classes of MS instruments,
such as Linear Quadrupole Ion Trap MS (Habicht, Amundson et al. 2010) and Time-of-
Flight MS (Zinovev, Veryovkin et al. 2007). Various approaches for the ionization of the
desorbed molecules were successfully employed, among them: electron impact and
chemical ionization (Crawford, Campbell et al. 2005), single-photon ionization (SPI)
(Zinovev, Veryovkin et al. 2007), as well as Elecro- Spray Ionization (ESI) (Cheng, Huang
et al. 2009). Moreover, the ability of the LIAD process to non-destructively eject from solid
surfaces not only single molecules but also larger intact biological particles, such as
viruses (Peng, Yang et al. 2006) and 1 m size tungsten particles (Menezes, Takayama et
al. 2005) have been demonstrated.
In our opinion, the wider spread of LIAD among analytical MS applications is now limited
by the lack of an adequate theoretical concept able to explain the existing observations and
to predict optimal experimental conditions for future measurements. The mechanical
shake-off model was proposed only as a qualitative explanation of observed desorption
process, and as such was never used to obtain any quantitative agreement between the
observable LIAD parameters and the generated acoustic waves. Moreover, to date, there
was no work published in the literature, which would be devoted to systematic studies of
physical parameters of the molecules desorbed by LIAD. Since we have attempted such as
study, we feel it would be beneficial for the research community if we describe here briefly
our own experimental methods and experimental results on LIAD.

355
3.3 Energy and velocity distributions of desorbed molecules
Because the dominant fraction of the desorbed flux in LIAD are neutral molecules, it is very
important to select an appropriate ionization method for the molecules as well as the type of
their mass analysis technique. Single-photon ionization (SPI) is well suited for
characterization of this phenomenon (Pellin, Calaway et al. 2001) because of its ability to
efficiently ionize the desorbing flux with minimal fragmentation. SPI occurs following
absorption of a single photon whose energy exceeds the ionization potential (IP) of the
molecule of interest, creating a cation. For many molecules, particularly those with aromatic
rings to stabilize the cation, fragmentation from photoionization is minimized, and thus the
state of the initial LIAD flux can, in principle, be revealed(Lipson and Shi 2002). Currently,
the shortest wavelength of commercially available energetic lasers suitable for SPI is 157 nm
(F
2
laser), which corresponds to the photon energy of 7.9 eV. This energy limits the range of
species that can be ionized by SPI to atoms and molecules with IPs less than 7.9 eV.
Therefore, organic dyes were chosen as analytes for this LIAD study due to their low IPs,
ability to form stable cations, and high photon absorption coefficients. We selected dyes that
have IPs in the range from 5 eV to 7 eV and can be easily ionized by the F
2
-laser radiation.

Ionizing laser
beam
Desorbing
laser beam
Sample
carousel
Ion optics
entrance
S
Substrate foil
Sample
mount
dS
Desorbed
molecules cloud

Fig. 5. Schematic drawing of target assembly and laser irradiation pathways. On the right
the enlarged view of desorption/ionization scheme is displayed
A time-of-flight mass spectrometer (TOF MS) with a combined LIAD/SPI ion source was
employed in our studies of the LIAD phenomenon. The experiments were conducted under
ultra high vacuum conditions, with the residual gas pressure in the sample chamber less
than 310-
7
Pa. The schematic drawing of the target assembly in our instrument is shown in
Fig. 5. The sample was mounted on one of six sample holders that were supported by a
hexagonal carousel. This carousel was driven by an ultrahigh-vacuum compatible motion
stage with closed-loop precision of better than 50 nm. The sample holders were secured on
the carousel via three 30 mm long alumina ceramic insulators and connected (using vacuum
feedthroughs) to a high-voltage pulser unit, which provided voltages necessary for the
operating of TOF MS instrument. Each new sample was inserted into the UHV chamber
through a vacuum loadlock. Using the motion stage, the sample was then positioned in the
focal plane of the TOF MS source optics. The ion optics and operational principles of our
instrument are described in more detail elsewhere (Veryovkin, Calaway et al. 2004). A


356
dielectric mirror with 98% reflection at 248 nm was mounted in the center of the carousel, in
order to deliver the laser beam to the back side of the sample. Note that we will use the
convention that the front of the sample is the side facing the ion source and TOF, while the
back side is the opposite. For desorption, an excimer KrF laser with wavelength 248 nm
(EX10/300 GAM Laser Inc.) was used. The output energy of the laser pulse could be varied
between 0.5 - 5 mJ by adjusting the laser discharge voltage and by an additional attenuation
with a set of neutral optical filters. The driving laser beam was focused on the target back
surface into a spot of rectangular shape ~200800 m
2
by the fused silica lens with focusing
distance of 500 mm. The laser pulse duration was 7 ns, producing a peak power density on
the irradiated surface ranging from 50 to 500 MW/cm
2
. These laser intensities are close to
those used in most of LIAD experiments (Perez, Ramirez-Arizmendi et al. 2000; Campbell,
Crawford et al. 2004; Campbell, Fiddler et al. 2005; Crawford, Campbell et al. 2005) taking
into account that the reflection coefficient in the UV is normally less than it is at visible
wavelengths. Post-ionization of the desorbed molecules was performed with an F
2
laser,
with output energy of 2 mJ/pulse, and pulse duration of 10 ns. The F
2
laser beam was
focused just above the front target surface, with a waist of 4002000 m
2
, using of a
combination of MgF
2
spherical and cylindrical lenses. The F
2
laser radiation power density
in the focal plane was ~10 MW/cm
2
, which assured the saturation of the photoionization
process for the investigated molecules (as verified by a laser power study).
For comparison with LIAD, direct laser desorption (LD) mass spectra were measured for the
same samples, also using the F
2
laser for post-ionization. To this end, an N
2
laser (337 nm
wavelength, 100 J/pulse energy and 7 ns pulse duration) was focused onto the target front
surface using an in-vacuum Schwarzschild optical microscope (Veryovkin, Calaway et al.
2004). The beam spot size on the surface was about 50 m in diameter.
The delay between the driving KrF (or N
2
) laser pulses and the ionizing F
2
laser could be
precisely controlled and varied from 0 to 1000 s. The desorbed molecules that move away
from the surface could therefore be ionized at a precisely defined moment in time and
volume in space above the target after the desorption event, with the photoions then
analyzed by the TOF MS. This approach allowed us to measure mass spectra for the
(postionized) desorbed neutral molecules and determine their velocity distribution. Each
mass spectrum was the sum of 128 individual acquired spectra. To prevent the rise of the
average foil temperature due to adsorption of laser power, the repetition rate of the laser
pulses was maintained at 8 Hz.
Foils from different materials with different thicknesses were used in the experiments. The
foil preparation procedure was the same as described in paragraph 2.2.3.1. Before applying
the analyte to the top surface of the foil, each substrate was cleaned in methanol-acetone
solution (1:1) in an ultrasonic bath (10 minutes).
Organic dyes rhodamine B, fluorescein, methylanthracene (MA), coumarin-522 (N-Methyl-
4-trifluoromethylpiperidino3,2-gcoumarin), and BBQ (4,4-Bisbutyloctyloxy-p-quaterphenyl)
were used as received (Eastman Kodak). The dyes were dissolved in methanol (for MA and
BBQ, mixed xylenes were also used as solvent), and then the resulting solution (about 10
-3
M) was used for sample preparation. One l of the analyte solution was pipetted onto the
foil surface, and then the quartz cylinder-foil assembly was spun at 4500 rpm for 30 seconds
to coat the analyte uniformly over the surface. During spin-coating, a significant part of the
solution (90% or more) was taken off the surface and, surface concentrations of the analyte
could be estimated to be less than 0.5 nM/cm
2
. After the sample preparation, the foil was
introduced into the instrument via the loadlock for analysis.

357
In good agreement with the previously published results on LIAD, we detected strong and
stable desorption signals from foils with thickness of 12.5 m (Fig. 6). Thicker foils (25 m)
produced relatively weak signals for the range of acoustic wave driven laser intensities used in
our experiments. For comprehensive experiments, a Ta foil with 12.5 m thickness was chosen
as optimal, not only because high desorption signals were detected from it, but also for its
good mechanical strength, high melting point, and durability under powerful laser irradiation.
The TOF mass spectra of different organic dye molecules desorbed from the front of the
back-irradiated Ta foil surface and ionized by the 157 nm laser radiation are shown in Fig.6.
This figure shows three major features, (1) all analytes display large parent molecular ion
signals, (2) all spectra display a small number of peaks, with a few or none in the mass range
below 100 Da., (3) the number of fragment ion peaks is specific to each molecular analyte. In
order to characterize the desorption process in terms of the corresponding molecular
fragmentation, a parameter , can be defined as a ratio of the sum of intensities of the fragment
peaks A
f
to the parent molecular peak intensity A
p
, / =
f p
A A . As will be shown below,
this parameter depends on laser intensities that drive the acoustic waves. For SPI , the
parameter characterizes not only the peculiarities of the desorption process, but it also
generally depends on the photoionization cross-section of the parent molecule and specifics of
its photofragmentation such as possible decay channels and their activation energies.

0 200 400 600 800
443
399
Rhodamine B
O N N
+
CH3
CH3
C H3
C H3
C
O
OH
0 200 400 600 800
Coumarin
281
N O O
CH
3
H
CF
3
0 200 400 600
Anthracene
192
H
H
H
CH
3
0 200 400 600 800
675
413
520
BBQ
O
C
6
H
13
C
4
H
9
O
H
13
C
6
H
9
C
4

Fig. 6. Mass spectra of organic dyes in LIAD experiments
For rhodamine B, the parameter grew linearly with KrF laser intensity (Fig. 7), and a
similar behavior was also detected for fluorescein, although values of were different ( =0.8
for fluorescein at 300 MW/cm2 of KrF laser intensity). On the contrary, the same
experiments conducted for BBQ, surprisingly revealed that the parameter decreased with
increasing desorption laser power. This discrepancy in parameter dependency may
indicate that the relationship between these two observations is not a trivial one. This raises
the question of whether the desorption and fragmentation phenomena are driven by the
same fundamental process. We note that fragmentation is not intrinsic to LIAD. In our
experiments, we observed some indications of fragmentation for only three analytes out of
five. The analysis of existing data from the literature shows that in some cases
fragmentation was observed even with soft ionization of the desorbed molecules whereas in
other cases fragmentation was very small or completely absent. A detailed study of


358
fragmentation in LIAD has not been done yet and the presented here results are the first
attempt to quantify this characteristic of LIAD.

200 250 300
0.2
0.3
0.4
0.5
Desorbing laser intensity, MW/cm

2

Fig. 7. Power dependence of fragmentation parameter

2.5 3.0 3.5 4.0 4.5 5.0
10
3
10
4
10
5
Rhodamine B
Coumarin
BBQ

S
i
g
n
a
l

i
n
t
e
n
s
i
t
y
,

a
.
u
.
Desorbing laser energy, mJ

Fig. 8. Power dependence of total desorption yield for rhodamine B molecules
An important characteristic of any desorption phenomenon is its yield. This is why in order
to understand the basic processes driving the phenomenon, one has to identify external
parameters that have the strongest effect on the desorption yield and then measure a
dependence of the yield for each parameter. In the case of LIAD, the dependence of key
peak intensities in the mass spectra and the parameter on the driving laser intensity
appears to be the most important for understanding this phenomenon. The overall
desorption yield for all studied analytes strongly increased with desorption laser intensity
(within our experimental range), displaying for most peaks approximately exponential
dependency. Figure 8 demonstrates this dependence clearly. Plotted on a semi-logarithmic
scale, these dependences appear linear, with different slopes for each analyte.
For the characterization of physical nature of the desorption processes, the knowledge of
velocity distributions of desorbed neutral molecules is extremely important (Levis 1994). As

359
mentioned above, time-of-flight mass spectra of such molecules can be measured using the
laser post ionization (LPI) technique (Spengler, Bahr et al. 1988). This experimental
arrangement makes it possible to determine the distribution of the desorbed neutral
molecules over their translational velocities (in the direction normal to the substrate surface)
by varying the time delay between desorbing and post-ionizing laser pulses. Raw
experimental data in this case are the dependencies of the observed signal on the laser delay
time, called below as signal-vs-delay dependencies. In order to be able to directly compare
LIAD and LD results, the energies of the desorbing N
2
(LD) and KrF (LIAD) lasers were
adjusted such that the output molecular ion signals reached about the same intensities for
the same gains of the TOF MS detector.
In good agreement with the previous experiments (Perez, Ramirez-Arizmendi et al. 2000),
dramatic differences between LIAD and LD in widths of the signal-vs-delay dependencies
and in positions of their maximums have been observed in these measurements. The most
tempting explanation for this experimental finding was that the mean velocities of desorbed
molecules in LIAD and LD processes were very different. Unfortunately, no actual energy or
velocity distributions of desorbed molecules in LIAD process have been measured
experimentally and reported in the literature to date. For the first time, this gap in
knowledge can now be filled by processing our experimental data from the signal-vs-delay
dependencies and converting them into kinetic energy distributions corresponding to our
experimental conditions.
Our experiment geometry (Fig.5) is rather common, and the detailed discussion of the
method as well as the appropriate conversion equations can be found elsewhere (Young,
Whitten et al. 1989; Balzer, Gerlach et al. 1997). Typical values of the distance S and the
thickness of ionization volume dS were set to 3 and 0.4 mm, respectively. While, as
mentioned above, the value of S could be varied in our experiments between 1 mm and 5
mm, in order to achieve optimal compromise between the energy/velocity resolution and
the signal-to-noise ratio, most of our data were obtained at S=3 mm. At this distance, the
relative velocity resolution was dv/v=dS/S=0.4/3=0.13 and, correspondingly, the energy
resolution was dE/E=0.18. A typical velocity distribution obtained for rhodamine B is
demonstrated on Fig.10. It is apparent that the desorbed molecules in LIAD are very slow.
The average velocities of rhodamine B and BBQ molecules were found to be 5912 m/s and
479 m/s, respectively. For methylanthracene and coumarin522 molecules, the measured
average velocities were 7620 m/s and 7018 m/s. It is important to recognize that these
LIAD-desorbed molecules are much slower than what could be expected assuming the
thermal mechanism of LIAD. In order to fit these rhodamine B data with Maxwellian
distributions (dashed curve on Fig.10), physically unrealistic low temperatures of about 100
K, were required. On the other hand, the measured velocities are much higher than that of
the laser-induced acoustic motion of the foil surface in normal direction, which, measured in
the same experimental arrangement, did not exceed 1 m/s. This observation caused serious
doubts on the validity of the shake-off mechanism considered by many as the most likely
cause of LIAD. The measurements of velocity distribution of species desorbed in LD
geometry were used for direct comparison with LIAD, and as the proof of validity of our
experimental procedure. A smaller insert plot on Fig.10 demonstrates the velocity
distribution of desorbed rhodamine B main fragment (399 amu) obtained in the LD
irradiation scheme. The mechanisms of the velocity distribution formation in the LD process
are well-known and discussed in many reviews (Levis 1994). According to a commonly
used procedure described in (Natzle, Padowitz et al. 1988), this distribution can be fit by a


360
two-temperature bi-modal velocity distribution: dashed lines in the insert plot represent
spectral components with different temperatures, and the solid line corresponds to their
sum. Being in a good agreement with general LD regularities, these results also confirm the
validity of the experimental procedure used for both LD and LIAD.

Fig. 9. The comparison of signal-vs-delay for LIAD and LD desorbed rhodamine B molecules
Velocity distributions were measured also at different fluences of the driving KrF laser. On
Fig. 10, two distributions corresponding to laser fluences of 2.3 J/cm
2
and 3.4 J/cm
2
are
plotted. Within the limits of accuracy of our measurements, no change of average velocity
has been detected, which suggests that both thermal and shake-off models are not
applicable for the explanation of LIAD process.

0 50 100 150 200 250 300
0.00
0.25
0.50
0.75
1.00
1.25
Rhodami ne D, LE
Rhodami ne B, HE
d
N
/
d
V
,

n
o
r
m
a
l
i
z
e
d
Velocity, m/s

0 500 1000
0.0

d
N
/
d
V
,

r
e
l
.
u
n
i
t
s
Velocity, m/s

Fig. 10. Velocity distribution for rhodamine B molecules at different LIAD desorption laser
intensities. On insert there is the same distribution for LD regime
In order to calculate the mean energies of desorbed molecules, we used the described above
approach and took into account that the Jacobian of this variable transformation was given
by S
2
/t
3
(Balzer, Gerlach et al. 1997). The mean energies of molecules in LIAD experiments
(as well as their mean velocities) showed no apparent trend with the increase of desorbing
laser fluences keeping the average value at about 9 meV for rhodamine B, 9.5 meV for BBQ,
6.5meV for coumarine 522 and 7.5meV for methylanthracene. Typical energy distributions

361
measured for these molecules are shown on Fig.11. Dashed line represents Maxwell
distribution for rhodamine B corresponding to the 80 K temperature. This temperature was
chosen to obtain the best fit at the distributions maximums. At the same time, high energy
tails of the experimental energy distributions strongly deviate from the exponential law,
which is apparent with the double logarithmic scale in Fig.11, and reveal behavior close to
the power dependence.

10
-4
10
-3
10
-2
10
-1
10
-6
10
-5
10
-4
10
-3
10
-2
10
-1
10
0
Ant hracene
Rhodami ne B (HE)
Rhodami ne B (LE)
BBQ
Coumari n
d
N
/
d
E
,

n
o
r
m
a
l
i
z
e
d
Energy, eV

Fig. 11. Energy distribution of different organic dyes molecules. Dashed line represents
equilibrium Boltzmann distribution at T=100 K
4. Is desorption process in LIAD really driven by acoustic waves?
4.1 Shake-off mechanism versus thermal origin of molecular desorption
The shake-off mechanism of molecular desorption in LIAD for a long time was considered
as the only sensible explanation of the observed phenomenon. However, as was shown
above, it contradicts both with general physical considerations and with the experimentally
measured parameters of acoustic vibrations of the surface as well as with the observed
energy and velocity spectra of desorbed molecules.
An important consequence of the backside irradiation is the heating of the front side.
Considering this effect, we should keep in mind that this heating process in LIAD is
distinctly different from the case of direct front-side laser irradiation (LD). In the latter case,
due to the small value of skin depth in metals, the rate of the temperature rise is extremely
high. In case of backside irradiation of the foils, the front surface temperature is governed by
the heat conduction through the metal foil, which makes the heating rate much slower than
that for the front side irradiation geometry. For one-dimensional heat conduction problem,
the specific time of the temperature rise is defined by the heat propagation time
2
/ = l ,
where is the thermal diffusivity of the metal and l is the foil thickness. For the foil
thicknesses typical for LIAD (~10 m) and the thermal diffusivity of the most metals 0.2
0.7 cm
2
s
-1
, has the values in the range of a few s. Our numerical calculations using the
heat conduction equation showed that for Ta foil with the thickness of 12.5 m and the
driving laser fluence 3.5 J/cm
2
(corresponding under conditions of our experiments to the
peak laser intensity of about 500 MW/cm
2
), the front surface temperature rise is 375 K,
which reaches its maximum 1.75 s after the laser desorbing pulse ceases. If experiments


362
start with room temperatures, the peak foil surface temperatures then can reach 668673 K.
Because the melting point of rhodamine B is just 438 K, the thermal origin of the LIAD
process can, in fact, come into the focus of our consideration. However, other experimental
results obtained in this work, make this mechanism very unlikely. These results are:
1. rather small fragmentation of molecules in LIAD, compared to that for LD;
2. distinct differences in velocity and energy distributions between LIAD and LD;
3. slow velocities and low kinetic energies of desorbed molecules compared to those
required by the thermal mechanism;
4. the apparent independence of mean energies of the desorbed molecules in LIAD on the
driving laser fluence (also observed by by Kenttamaa et al. in their recent work (Shea,
Petzold et al. 2007).
Moreover, the mechanical shake-off mechanism is also in contradiction with the
observation (4), because the amplitudes and velocities of laser generated acoustic waves
should increase with the driving laser fluencies. Thus we can conclude that both
mechanisms of the direct energy transfer (acoustic waves and heat conduction) cannot serve
as the primary explanation of the LIAD phenomenon, and apparently more complicated
processes are involved here.
4.2 Stress and strain of the foil surface due to the laser irradiation
It is a well-known fact that a film deposition on a substrate surface results in many cases in
the residual mechanical stress (due to the lattice parameters mismatch between the film and
the substrate) and thus in some excessive potential energy stored in the film. This stress can
be produced by two ways: one is the growth stress and the other is the induced (or extrinsic)
stress (Freund and Suresh 2003). An external impact, such as acoustic or thermal wave
generated by the laser irradiation of the substrate, should initiate the reconstruction of the
film, which can result in releasing this excess energy or, possibly, generating an additional
extrinsic stress. In both cases, it can cause formation of cracks in the film so that intermolecular
bonds at the edges of the cracks can break. As a result of this crack formation process, excited
electronic states can form at the crack edges and induce desorption of the molecules.
Despite a great variety of mechanisms of stress formation and evolution do exist, a
quantitative description of this process is possible only in limited cases even for simple
adsorbent-adsorbate systems. Accurate modeling of the film behavior requires precise
knowledge of the thermal and mechanical properties of the film material and their
interaction with substrate surface. For most organic materials, such data are unavailable,
and thus only limited estimates may be done, based on the little amount of data published
in the literature (Bondi 1968). Moreover, the deposited films of organic materials tend to
consist of a mixture of micro-crystals with different crystallographic orientation and,
therefore, there are many grain boundaries, defects, dislocations in such films that could
significantly change their local mechanical properties.
A simplified numerical estimate of the stress energy, related to the thermal mismatch of the
substrate foil and the analyte film on its top could be done, as follows. An equi-biaxial
stress resulting from acoustic vibration of the back-irradiated metal foil may be estimated
using the following expression (Boley and Weiner 1960; Freund and Suresh 2003)

(1 )
=

f
m
E h
R
(16)

363
where E is the Young's modulus, v is the Poisson ratio, h
f
is the thickness of organic film and
R is the radius of curvature of the foil surface. From the other side, the stress, resulting from
thermal mismatch has the following value (Boley and Weiner 1960)
( ) /(1 ) =
t s f
T E , (17)
where
s
and
f
are thermal expansion coefficients of the substrate and the film respectively,
T is the temperature increase. There is not much information in the literature on thermo-
mechanical parameters for molecular crystals but based on existing data for anthracene
(Bondi 1968) we can estimate the order of the generated stress values. Under the assumption
that
f
=2.8 10
-4
K
-1
,
s
= 6.3 10
-6
K
-1
, E= 13 GPa, v0.25, and T=100 K, we can obtain
t
=485 MPa. At the same time, the stress associated with acoustic vibrations is much lower:
m
1.7 kPa (taking into account that maximal value of R is approximately 1 m, and the film
thickness never exceeded 10
-7
m,). This is a negligible value, in comparison with
t
, which
means that the thermal mismatch stress is the principal reason for cracks formation. The
estimate of internal energy, stored in thermally strained organic films can be done with
using of following expression (Boley and Weiner 1960)

2
2 2
0 2
(1 )
( ) 2
2 (1 )
+
=

s f f
E
G T r h
(18)
The average energy per analyte molecule can easily be calculated

2 2
2
2
( )
(1 )
2 (1 )

+
=

s f
a
A
T M
E
g
N

(19)
Here M is the molar mass, is the specific gravity and N
A
is the Avogadro number. It is
interesting to note that g
a
does not depend on the analyte island size but strongly depends
on the thermal and mechanical parameters. Again with the use of existing data for
anthracene we can estimate the value of g
a
, and for T=100 K we will get g
a
=0.025 eV . This
is not enough to break intermolecular bonds but when thermally induced stress exceeds a
critical value, the film can start to fracture and the stored energy is released in a small
volume in the vicinity of the stress cracks. Due to the strong spatial nonuniformity of
thermo-mechanical properties of molecular crystals (Bondi 1968) the physical mechanisms
involved in this process are very complicated in nature; therefore we can give only a
qualitative picture of this phenomenon. Presumably, the cracks are formed along grain
boundaries, defects and interfaces. The increase of the desorption laser intensity that causes
a rise in T and, in accordance with Eq. 18, an increase in energy G, results in the formation
of additional cracks. Some part of this excess energy can be then converted into the
increased free surface energy, other part into electronic excitations, but because these
processes are obviously non-adiabatic, the crack formation most likely will be accompanied
by breaking intermolecular bonds and forming new desorption sites.
4.3 Proposed mechanisms of the molecules desorption
The desorption process itself appears to be the most obscure part of the LIAD phenomenon.
The formation of the electronically excited states on the surface due their mechanical
fracture is considered to be the main physical nature of triboemission (Nakayama, Suzuki et
al. 1992), also known as Kramer effect (Oster, Yaskolko et al. 1999; Oster, Yaskolko et al.


364
2001). The essence of this effect is the emission of charged particles and photons initiated by
surface distortion (in particular by mechanical deformation, scratching, bending, etc.) and
do not connected with thermal excitation. The luminescence of thin metal discs, irradiated
from back side by the laser pulses (Abramova, Shcherbakov et al. 1999; Abramova, Rusakov
et al. 2000) can also serve as evidence of formation of excited electronic states by the laser-
driven stress in thin foils.
The mechanism of molecular desorption due to such laser-generated stress has been
proposed earlier by Vertes (Vertes and Levine 1990; Vertes 1991) for MALDI, and was based
on the thermal stress generation in the layer of organic film deposited on solid substrates.
One should notice, however, a clear differences in the physical conditions between these
sample volatilization methods. For MALDI, the absorption of the laser pulse energy occurs
in an optically and thermally dense film, which experiences thermal stress due to its non-
uniform and fast heating. In the case of LIAD, the laser radiation is absorbed by the back
side of the metal foil substrate, opposite from where the sample was deposited. The amount
of energy transmitted through the metal foil to the analyte layer on the front site should be
so strongly attenuated, compared to the direct (front side) irradiation, that it cannot directly
be sufficient for desorbing molecules with velocities observed in our experiments. On the
other hand, a typical average thickness of the analyte film in LIAD can be estimated to be on
the order of several molecular layers. Because of this, the specific density of energy stored in
each analyte island due to intrinsic stress can be high, and during laser irradiation of the
back side of the foil, the laser irradiation can simply trigger the release of this energy and to
induce molecular desorption event.
The other experimental fact that could help in the interpretation of LIAD phenomenon is the
similarity of the energy and velocity spectra of desorbed neutral molecules in case of LIAD
and Electron Stimulated Desorption (ESD) (Young, Whitten et al. 1989). The primary
mechanism of ESD is supposed to be the formation of the repulsive states in the surface due
to electron excitation.
Thus we could hypothesize that in case of LIAD, due to complete opaqueness of metal foil
for laser radiation, the only channel for energy transfer is the formation of acoustic and
thermal field by laser pulse impact. As shown above, the result can be the mechanical
distortion of the analyte film followed by film cracking and delamination and the
consequent formation of excited and repulsive states electronic for analyte molecules.
This will increase the number of desorption sites and finally the total number of desorbed
molecules as observed in the nonlinear laser intensity dependence. But because the
formation of the any individual cracks is defined only by the intermolecular bonding forces
in the vicinity of the crack, the translational kinetic energy of desorbed molecules should
still remain independent of driving laser intensity, also matching our observations and other
recent LIAD publications.
5. Conclusion
Traditionally, the name LIAD combine all desorption phenomena taking place when an
opaque target is irradiated from the back side, ignoring differences in experimental
conditions. In our opinion, it is not correct. As it was demonstrated in this work, under some
experimental conditions (most commonly used in many present studies), the physical origin
of the observed desorption phenomenon is not (and could not be) connected with the
acoustic waves generated in the foil and most likely is defined by the film stress and

365
cracking due to thermal and mechanical mismatch of the analyte and substrate. Therefore,
ironically, the acronym LIAD in this case does not correctly reflect the physical nature of the
process.
From the other hand, we cannot exclude that some strong change of the experimental
conditions can also change the relationship between various physically possible
mechanisms of molecular desorption, similarly to the case of direct laser desorption, when
the thermal mechanism dominates under wide range of conditions and makes other possible
mechanisms undetectable. One can expect, for example, that strong increase of laser power
density (10 GW/cm
2
and above) will cause the corresponding increase of the foil
temperature and formation of hot and dense plasma plume near its surface facing the laser.
In some such cases and strongly depending on the foil material properties, the material
motion could evolve from elastic into plastic regime of deformation. The acoustic wave
relation could not then apply, and the velocity of the surface linear motion may strongly
increase. Large and heavy objects weakly bound to the substrate surface may be then
kinematically removed from it (shaken-off). This could indeed serve as an explanation of
observations from recent experiments where the desorption of intact viruses and biological
cells were reported (Peng, Yang et al. 2006). However, this regime cannot be connected with
generation of the acoustic waves but most likely corresponds to a physically different mode
of shock-wave generation (Menikoff 2007). One cannot exclude that under some
experimental LIAD conditions only these shock-wave induced phenomena can be
responsible for molecular desorption, particularly in the experiments where emission of ions
was detected. In one of such pioneering experiments (Golovlev, Allman et al. 1997), where
the laser generated pressure pulse was apparently much stronger than that in our work,
because of confined ablation conditions (Fabbro, Fournier et al. 1990), the emission of both
electrons and ions was observed. This may have been connected with significant surface
disruption at the microscale generated by shock waves.
From another standpoint, the backside irradiation of very thin films (about two or three
hundred nanometers), which also could be called LIAD, has demonstrated domination of
the thermal mechanism in the desorption process (Ehring, Costa et al. 1996). It is clear that
the thermal equilibrium between front and back sides in such thin films can establish within
the time interval of tens of nanoseconds, and the absolute temperature difference between
front and back sides is negligibly small. Thus, at some laser power densities, the front side
temperature could reach the melting temperature of the metal that would be enough to
cause an efficient thermal desorption of the most organic molecules.
To conclude, the variety of desorption and emission phenomena observed on the front side
of thin metal foils whose back sides are subjected to pulsed laser irradiation, combined
under the general name of LIAD, could, in fact, have a number of different physical origins
depending on specific experimental conditions. According to our observations conducted at
moderate laser power densities (0.1 - 1 GW/cm
2
) and foils thicknesses of about 5 20 m,
that appear to be the most commonly used conditions in LIAD experiments, the
predominant desorption mechanism is connected with the reorganization of the deposited
analyte film and the consequent breaking of molecular bonds on the edges of these cracks.
6. Acknowledgment
This work is supported by the U.S. Department of Energy, BES-Materials Sciences, under
Contract DE-AC02-06CH11357, by UChicago Argonne, LLC.


366
7. References
Abramova, K. B., A. I. Rusakov, et al. (2000). Photon emission from metals under fast
nondestructive loading. J. Appl. Phys. 87(6): 3132-3136.
Abramova, K. B., I. P. Shcherbakov, et al. (1999). Emission processes accompanying
deformation and fracture of metals. Phys.Solid State 41(5): 761-762.
Adamson, A. W. and A. P. Gast (1997). Physical Chemistry of Surface. New York, John
Wiley &Sons, Inc.
Balzer, F., R. Gerlach, et al. (1997). Photodesorption of Na atoms from rough Na surface. J.
Chem. Phys. 106(19): 7995-8012.
Boley, B. A. and J. H. Weiner (1960). Theory of thermal stresses. New York,, Wiley.
Bondi, A. (1968). Physical properties of molecular crystals, liquids and gases. New York,
London, Sydney, Hohn Wiley & Sons.
Campbell, J. L., K. E. Crawford, et al. (2004). Analysis of saturated hydrocarbons by using
chemical ionization combined with laser-induced acoustic desorption/Fourier
transform ion cyclotron resonance mass spectrometry. Analytical Chemistry 76(4):
959-963.
Campbell, J. L., M. N. Fiddler, et al. (2005). Analysis of polyethylene by using
cyclopentadienyl cobalt chemical ionization combined with laser-induced acoustic
desorption/Fourier transform ion cyclotron resonance mass spectrometry.
Analytical Chemistry 77(13): 4020-4026.
Cheng, S. C., M. Z. Huang, et al. (2009). Thin-Layer Chromatography/Laser-Induced
Acoustic Desorption/Electrospray Ionization Mass Spectrometry. Analytical
Chemistry 81(22): 9274-9281.
Cole, R. B. (2010). Electrospray and MALDI mass spectrometry: fundamentals,
instrumentation, practicalities, and biological applications. Hoboken, N.J., Wiley.
Crawford, K. E., J. L. Campbell, et al. (2005). Laser-induced acoustic desorption/Fourier
transform ion cyclotron resonance mass spectrometry for petroleum distillate
analysis. Anal. Chem. 77(24): 7916-7923.
Crawford, K. E., J. L. Campbell, et al. (2005). Laser-induced acoustic desorption/Fourier
transform ion cyclotron resonance mass spectrometry for petroleum distillate
analysis. Analytical Chemistry 77(24): 7916-7923.
Dass, C. (2007). Fundamentals of contemporary mass spectrometry. Hoboken, N.J., Wiley-
Interscience.
Duncan, M. W., H. Roder, et al. (2011). Quantitative matrix-assisted laser
desorption/ionization mass spectrometry. Briefings in Functional Genomics 7(5): 355-
370.
Ehring, H., C. Costa, et al. (1996). Photochemical versus thermal mechanisms in matrix-
assisted laser desorption/ionization probed by back side desorption. Rapid
Communications in Mass Spectrometry 10(7): 821-824.
Fabbro, R., J. Fournier, et al. (1990). Physical Study of Laser-Produced Plasma in Confined
Geometry. Journal of Applied Physics 68(2): 775-784.
Freund, L. B. and S. Suresh (2003). Thin Film Materials. Stress, Defect Formation and Surface
Evolution. Cambridge, Cambridge University Press.
Golovlev, V. V., S. L. Allman, et al. (1997). Laser-induced acoustic desorption of electrons
and ions. Applied Physics Letters 71(6): 852-854.

367
Gospodyn, J. P., A. Sardarli, et al. (2002). Ablative generation of surface acoustic waves in
aluminum using ultraviolet laser pulses. Journal of Applied Physics 92(1): 564-571.
Habicht, S. C., L. M. Amundson, et al. (2010). Laser-Induced Acoustic Desorption Coupled
with a Linear Quadrupole Ion Trap Mass Spectrometer. Analytical Chemistry 82(2):
608-614.
Honig, R. E. and J. R. Woolston (1963). Laser-induced emission of electrons, ions and neutral
atoms from solid surfaces. Applied Physics Letters 2(7): 138-139.
Hutchins, D. A. (1985). Mechanisms of Pulsed photoacoustic generation. Can. J. Phys. 64:
1247-1263.
Karas, M., D. Bachmann, et al. (1985). Influence of the Wavelength in High-Irradiance
Ultraviolet-Laser Desorption Mass-Spectrometry of Organic-Molecules. Analytical
Chemistry 57(14): 2935-2939.
Krehl, P., F. Schwirzke, et al. (1975). Correlation of Stress-Wave Profiles and Dynamics of
Plasma Produced by Laser Irradiation of Plane Solid Targets. Journal of Applied
Physics 46(10): 4400-4406.
Landau, L. D. and E. M. Lifshits (1987). Fluid mechanics. New York, Pergamon.
Levis, R. J. (1994)."Laser-Desorption and Ejection of Biomolecules from the Condensed-
Phase into the Gas-Phase. Annu. Rev. Phys. Chem. 45: 483-518.
Lindner, B. and U. Seydel (1985). Laser Desorption Mass-Spectrometry of Nonvolatiles
under Shock-Wave Conditions. Analytical Chemistry 57(4): 895-899.
Lipson, R. H. and Y. J. Shi (2002). Ultraviolet Spectroscopy and UV lasers. New York-Basel,
Marcel Deccer, Inc.,.
Lubman, D. M. (1990). Lasers and mass spectrometry. New York, Oxford University Press.
McLachlan, N. W. (1951). Theory of vibrations. New York, Dover Publications.
Menezes, V., K. Takayama, et al. (2005). Laser-ablation-assisted microparticle acceleration
for drug delivery. Applied Physics Letters 87(16): -.
Menikoff, R. (2007). Empirical Equations of State for Solids. Shock Wave Science and
Technology Reference Library. Y. Horie. Berlin, Springer. 2: v. <1-3>.
Miller, J. C. and R. F. Haglund (1998). Laser ablation and desorption. San Diego, Academic Press.
Mora, P. (1982). Theoretical-Model of Absorption of Laser-Light by a Plasma. Physics of
Fluids 25(6): 1051-1056.
Murray, T. W. and J. W. Wagner (1999). Laser generation of acoustic waves in the ablative
regime. Journal of Applied Physics 85(4): 2031-2040.
Nakayama, K., N. Suzuki, et al. (1992). Triboemission of Charged-Particles and Photons
from Solid-Surfaces during Frictional Damage. J. Phys. D: Appl. Phys. 25(2): 303-308.
Natzle, W. C., D. Padowitz, et al. (1988). Ultraviolet-Laser Photodesorption of No from
Condensed Films - Translational and Internal Energy-Distributions. Journal of
Chemical Physics 88(12): 7975-7994.
Oster, L., V. Yaskolko, et al. (1999). Classification of exoelectron emission mechanisms.
Physica Status Solidi a-Applied Research 174(2): 431-439.
Oster, L., V. Yaskolko, et al. (2001). The experimental criteria for distinguishing different
types of exoelectron emission mechanisms. Physica Status Solidi a-Applied Research
187(2): 481-485.
Pellin, M. J., W. F. Calaway, et al. (2001). Laser Post Ionization for Quantative Elemental
Analysis. ToF-SIMS: Surface Analysis by Mass Spectrometry. J. Vickerman and D.
Briggs, Surface Spectra Ltd. and IM Publications 375.


368
Peng, W. P., Y. C. Yang, et al. (2006). Laser-induced acoustic desorption mass spectrometry
of single bioparticles. Angew. Chem., Int. Ed. 45(9): 1423-1426.
Perez, J., L. E. Ramirez-Arizmendi, et al. (2000). Laser-induced acoustic desorption/chemical
ionization in Fourier-transform ion cyclotron resonance mass spectrometry. Int. J.
Mass Spectrom. 198(3): 173-188.
Phipps, C. R., T. P. Turner, et al. (1988). Impulse coupling to targets in vacuum by KrF, HF
and CO
2
single-pulse lasers. Journal of Applied Physics 64(3): 1083-1096.
Pollard, H. F. (1977). Sound waves in solids. London, Pion.
Prokhorov, A. M., V. I. Konov, et al. (1990). Laser Heating of Metals. Bristol, Philadelphia,
New York, Adam Higler.
Reid, G. E., S. E. Tichy, et al. (2001). N-terminal derivatization and fragmentation of neutral
peptides via ion-molecule reactions with acylium ions: Toward gas-phase Edman
degradation? Journal of the American Chemical Society 123(6): 1184-1192.
Royer, D. and E. Dieulesaint (2000). Elastic waves in solids. Berlin ; New York, Springer.
Schiller, J., R. Suss, et al. (2007). Maldi-Tof Ms in Lipidomics. Frontiers in Bioscience 12: 2568-2579.
Scruby, C. B. (1987). An introduction to acoustic emission. Journal of Physics E (Scientific
Instruments) 20(8): 946-953.
Scruby, C. B. and H. N. G. Wadley (1978). Calibrated Capacitance Transducer for Detection
of Acoustic-Emission. Journal of Physics D-Applied Physics 11(11): 1487-1494.
Shea, R. C., C. J. Petzold, et al. (2007). Experimental investigations of the internal energy of
molecules evaporated via laser-induced acoustic desorption into a Fourier transform
ion cyclotron resonance mass spectrometer. Analytical Chemistry 79(5): 1825-1832.
Smith, S. T. (2000). Flexures: elements of elastic mechanisms. Amsterdam, Gordon & Breach.
Song, K. H. and X. Xu (1997). Mechanisms of absorption in pulsed excimer laser-induced
plasma. Applied Physics a-Materials Science & Processing 65(4-5): 477-485.
Spengler, B., U. Bahr, et al. (1988). Postionization of Laser-Desorbed Organic and Inorganic-
Compounds in a Time of Flight Mass-Spectrometer. Anal. Instrum. 17(1-2): 173-193.
Srinivasan, R. and B. Braren (1989). Ultraviolet-Laser Ablation of Organic Polymers.
Chemical Reviews 89(6): 1303-1316.
Vertes, A. (1991). Laser Desorption of Large Molecules: Mechanisms and Models. Methods
and Mechanisms for Producing Ions from Large Moleculres. K. G. Standing and W. Ens.
New York, Plenum Press.
Vertes, A. and R. D. Levine (1990). Sublimation Versus Fragmentation in Matrix-Assisted
Laser Desorption. Chem. Phys. Lett. 171(4): 284-290.
Veryovkin, I. V., W. F. Calaway, et al. (2004). A new time of flight instrument for quantative
surface analysis. Nucl. Instrum. Methods Phys. Res., Sect. B 219-220: 473.
White, R. M. (1963). Generation of elastic waves by transient surface heating Journal of
Applied Physics 34(12): 3559-&.
Xu, B. Q., J. Feng, et al. (2008). Laser-generated thermoelastic acoustic sources and Lamb waves
in anisotropic plates. Applied Physics a-Materials Science & Processing 91(1): 173-179.
Young, C. E., J. E. Whitten, et al. (1989). Electron-Stimulated Desorption of Neutrals from
6063 Aluminum - Velocity Distributions Detected by 193 Nm Non-Resonant Laser
Ionization. Surface and Interface Analysis 14(10): 647-655.
Zinovev, A. V., I. V. Veryovkin, et al. (2007). Laser-driven acoustic desorption of organic
molecules from back-irradiated solid foils. Analytical Chemistry 79(21): 8232-8241.
17
Excitation of Periodical Shock Waves in
SolidState Optical Media (Yb:YAG, Glass)
at SBS of Focused LowCoherent Pump
Radiation: Structure Changes,
Features of Lasing
N.E. Bykovsky and Yu.V. Senatsky
Lebedev Physical Institute, Russian Academy of Sciences, Moscow
Russia
1. Introduction
During several last decades much attention was paid to the processes that occur in solid-
state optical media under the interaction with high-power focused laser radiation. A great
number of studies were devoted to the phenomena of optical breakdown, structure changes,
stimulated scatterings, generation of hypersonic waves in transparent dielectrics under the
action of nanosecond (ns) and picosecond (ps) laser pulses (Manenkov & Prokhorov, 1986;
Nelson et al., 1982; Ready, 1971; Robinson et al., 1984; Stuart et al., 1995). Recent interest in
these studies was stimulated by the appearance of lasers with femtosecond (fs) pulses
(Gordienko et al., 2010; Merlin, 1997; Sakakura et al., 2007).
An experimental study of a small region with high pressure and temperature gradients
formed in a medium at focusing high-power laser radiation had been performed, as a rule,
outside the laser cavity. In our experiment (Basiev et al., 2004), a region with such properties
happened to be formed directly in the 2-mirror laser cavity, when Yb:YAG samples were
pumped by the focused wide-band (0,89-0,95 m) radiation from a pulsed LiF:

F
2
+
color
center laser (ccl). Thus, in contrast to many studies on ytterbium lasers, conditions for
generation in Yb-doped samples in this experiment had been distinguished by the very high
intensity (over 1 GW/cm
2
) of the pump, which moreover had a low coherence. Experiments
on pumping of Yb-doped and non-doped samples of different optical media (YAG, glass,
LiF et al.) by powerful low-coherent radiation from LiF:

F
2
+
ccl were continued in
subsequent papers (Bykovsky, 2005, 2006; Bykovsky & Senatsky, 2008a,b, 2010). At
intensities I 1 GW/cm
2
interaction of ccl pump radiation with the medium in the focal
region was essentially nonlinear. The interaction of ccl pulses with samples was
accompanied by excitation of stimulated Brillouin scattering (SBS) and stimulated Raman
scattering (SRS) of pump radiation. The scattering generated hypersonic waves of high
amplitude, which were converted into a periodic sequence of shock waves with sharp
pressure jumps on their fronts propagating along the direction of pump. Pressure jumps
were so large that they caused a phase transition in an optical medium, which was observed
near the sample surface in the form of small domains with spatial modulation of the
refractive index caused by the interference of hypersonic waves.


370
Under ccl pumping due to heat release and generation of intensive hypersonic waves a
region with strong temperature, pressure and refractive index gradients and at the same
time with a high-level of inversion was formed in the focus of the pump laser in Yb-doped
materials. Despite the strong optical inhomogeneity of the medium, Yb lasing in 10-15 ns
pulses was observed in Yb:YAG (with 20% concentration of Yb
3+
ions) and Yb:glass (with
10% Yb
3+
concentration) during the action of the 20-30 ns pump pulse and after it (due to
inversion remaining in the medium). During the SBS of pump radiation the hypersonic
wave spatial structure served as a resonator for Yb lasing. Shock waves (with phonon
energies up to 1000 cm
-1
) affected the generation dynamics. The Yb lasing was
distinguished by some specific characteristics such as a surprisingly wide spectrum (up to
50 nm) and a high directivity of the emission. In addition to the wide-band generation on
the shock-waves grating there was also observed Yb lasing on resonator modes. After the
end of the pump pulse another sequence of shock waves diverging outward the focal
region affected the build-up of generation between 2 mirrors in the cavity. The line
spectra of Yb generation in the resonator contained twisted spectral lines with structures
of small-scale spots.
Description of these unusual phenomena observed under the interaction of short intense
pump laser pulse of low coherence with optical media and their explanation are presented
in this chapter. The optical scheme and the parameters of LiF:F
2
+
ccl are considered. The
features of SBS and SRS and the appearance of the periodical shock waves in the optical
medium at low-coherent pumping are discussed. The interpretation to observed specific
optical damage is given. The mechanism of generation of broadband, high-directional short
laser pulses in the spatial structure of thin layers with inversion produced in the region of
propagation of intense hypersonic waves in the medium is discussed. Conditions for
generation in a 2 mirror resonator containing active medium with a strong refractive index
gradient are considered. The interpretation of the observed twisted lines with small-scale
structures in generation spectra as well as temporal, spatial-angular characteristics of Yb
lasing in the resonator is given.
2. LiF: F
2
+
color center pump laser
Optical pumping of Yb-doped materials can be performed only into the single Yb
3+
ion
absorption band at
2
F
7/2
-
2
F
5/2
transition near 0.9 m (Figs.1, 2). In this connection, the use of
flash lamps as sources of broadband radiation for pumping an Yb-doped medium is
ineffective. At the present moment, the most effective and widely used sources of Yb-doped
materials pumping are semiconductor laser diodes operating within the spectral range near
0.94 m. Along with semiconductor diodes solid-state laser pump sources have been used to
investigate Yb-doped active media. Cr:LiSAF, Ti:Sa, Nd:YAG (0.94 m transition) lasers
have been used to pump Yb:S-FAP and Yb:YAG (Bykovsky et al., 2000; Kanabe et al., 2000;
Marshall et al., 1997, as cited in Bykovsky & Senatsky, 2008b).
In our work a LiF:

F
2
+
color center laser (ccl) was used to pump an Yb:YAG crystal (Basiev et
al., 2004). The lithium fluorine color center (LiF:

F
2
+
,
LiF:

F
2
-
) lasers are the sources of
radiation in the near IR (0.8-1.3 m) , and they effectively convert the neodymium and ruby
laser radiation into this spectral range (Basiev et al., 1982). Possible room temperature
operation, high conversion efficiency (up to 30%), and a large generation tuning range (more
than 1000 cm
-1
) make such lasers very attractive for certain practical applications. The
Excitation of Periodical Shock Waves in SolidState Optical Media (Yb:YAG, Glass)
at SBS of Focused LowCoherent Pump Radiation: Structure Changes, Features of Lasing

371
pumping of Yb-doped samples by the ccl had been, undoubtedly, inferior to laser diodes
pumping in efficiency. However, at focusing ccl radiation in the Yb-doped medium the
pump power densities typical of semiconductor diodes could be easily exceeded. This had
been, of course, of interest for research. Therefore, just the ccl had been used later in
experiments on study lasing in Yb-doped samples as well as nonlinear interaction of pump
radiation with optical materials. The possible generation region of LiF:

F
2
+
ccl extends from
0.83 to 1.1 m (Basiev et al., 1982) and completely covers the absorption band of Yb:YAG.
In particular experimental conditions the actual emission band of ccl depends on the
selective properties of the resonator, and usually covers only part of the noted wavelength
range.

Fig. 1. Energy level diagram of Yb
3+
ions in a Yb:YAG crystal. Ovals combine the Stark
components of levels with rapid ( 10
-12
s) thermal relaxation (Krupke, 2000)

Fig. 2. Absorption and luminescence spectra of Yb
+3
ions in the YAG crystal (1 mm plate,
Yb
+3
ions concentration 20%)


372
Figure 3 illustrates the scheme of the ccl with elements of radiation transport and
diagnostics. A LiF:

F
2
+
crystal (40206 mm) was placed in the resonator formed by a plane
mirror M
1
( 100% reflection at 0.9 m) and a glass plate M
2
. The length of the ccl resonator
made 30 cm. The LiF:

F
2
+
crystal was pumped through glass plate by a ruby laser operating
in the single shot regime with pulse duration 30 ns and the energy up to 1 J. A multimode
radiation at the wavelength of 0.694 m was focused into the LiF:

F
2
+
crystal by a lens L
1

with the focal length F
1
=500 mm. The ccl multimode radiation was, in its turn, focused on
the studied samples by a lens L
2
with the focal length F
2
=120 mm. The Yb-doped samples
were placed into a compact two-mirror resonator.

Fig. 3. Scheme of the experimental setup: (1) active element of a ccl; (2) Yb-dopped plate in
the resonator with mirrors M8 and M9; (3) calorimeter; (M1, M2) ccl resonator mirrors; (M3 -
M7) steering mirrors for ccl and Yb laser radiation; (L1 L2) lenses; (PD1, PD2) photodiodes
The ruby laser and ccl pulses were registered with the help of photodiodes and a two-
channel oscilloscope, and the laser energy was measured by calorimeters. The ccl radiation
spectra were analyzed by the STE-1 spectrograph operating in the near IR region. The ccl
emitted pulses of 20-30 ns duration and the energy of 100-150 mJ. For the most cases, the ccl
pulse shape repeated the shape of the ruby laser pulse, Fig. 4a. Since the ccl resonator
round-trip time made 2 ns, then during the ruby laser pumping the radiation made not
more than 15 round trips in the ccl cavity. Due to high amplification in the LiF:

F
2
+
medium
the ccl pulse was formed in several round trips inside the resonator. Though the LiF:

F
2
+

crystal was cut at a Brewster angle the ccl radiation was weakly polarized.
Figure 4b presents the densitograms of LiF:

F
2
+
laser spectrum. The LiF:

F
2
+
laser emission
was observed within the range 0.89-0.95 m. The lines of an argon spectral lamp were used
as the wavelengths markers. Large spectrum width of ccl and short time of radiation
development in the resonator speak about low coherency of the ccl emission. The ccl
multimode radiation divergence was 210
-3
rad. This allowed focusing the ccl pump at the
sample into a 250m spot. Moving samples along the axis of the focused pump beam one
could change the size of the focal region within the limits of 2501000 m, and the power
density in the medium within the range of 0.55 GW/cm
2
. Such a range of power density

373
variation allowed one to carry out both experiments on Yb lasing (see Section 4-6) and
experiments on nonlinear interaction of ccl radiation with optical materials (see Section 3).
In some of experiments, the ccl energy density was close to the damage threshold of
Yb:YAG, glass and other studied materials. The material being damaged, the further
experiments used a fresh part of a sample.

Fig. 4. Oscillograms of the ruby laser (1) and ccl (2) pulses (a); densitograms of the LiF:

F
2
+

laser spectrum with the argon spectral lamp reference lines (b)
3. Interaction of ccl radiation with optical media
The experiments on nonlinear interaction of ccl radiation with optical media have been
mainly performed using non-doped samples out of the cavity, Fig. 5. The ccl radiation was
focused at samples by L
2
, F=120 mm, and, so, within the sample thickness of 1-3 mm the
pump power density changed insignificantly. The lens was tilted at 10
0
to the direction of ccl
beam so that reflections from lens surfaces would not come back to the ccl resonator. As the
samples there were used plates and slabs with polished surfaces made of the following
materials: crystalline quartz (101020mm), Yb:YAG crystal (1-2 mm thick plates), YAG
crystal (4,53030 mm), calcite (2 mm plates), LiF crystal (5 mm plate), 2 mm plexiglas
plates, glass cube (202030 mm). The ccl radiation was directed onto samples at a normal
or at some angle (including the Brewster angle) to the samples surface. The ccl pulse energy
coming to samples varied within the range from 50 to 120 mJ.
In all materials a strong scattering of radiation was registered under the action of a ccl pulse.
A diagnostic complex consisting of photodiodes, oscilloscopes, calorimeters and a
spectrograph was arranged on the stand (Fig.5) in order to study the scattering of the low
coherent wideband (0,89-0,95 m) radiation of LiF:

F
2
+
ccl. Due to very large spectrum width
(tens of nm) and complicated space-angular structure of the scattered radiation components,
one comes across difficulties in obtaining the spectral data on the scattered radiation, and it
was not done. Strong scattering from the ccl focusing region in a wide angular range (tens of
degrees) in forward and opposite directions (relative to the ccl beam) was observed for all


374
samples at the ccl power density 1 GW/cm
2
. The scattering intensity grew with the
growth of the ccl intensity. The strongest scattering was observed at an angle of 180
0
, i.e.
back to the ccl aperture. The scattering was also observed at 90
0
to the ccl beam. Using
photodiodes FD with 1 ns time resolution, the shape and duration of the ccl pulse and
scattered radiation were registered by two-channel oscilloscopes of 1ns and 4ns resolution.
To collect the scattered radiation to photodiodes the spherical mirrors were used, Fig. 5. The
registered time and space-angular characteristics of scattering evidenced the rise of
stimulated scattering (SBS and SRS) of wideband low-coherent ccl radiation in the samples.

Fig. 5. Scheme of the experiment on the study of nonlinear phenomena in the crystal and
glass samples: R1, R2 LiF:

F
2
+
ccl mirrors; R3 R5 ccl radiation transport mirrors; R6 R7
mirrors to collect scattered radiation at diagnostic system; L1, L2 lenses; FD photodiodes;
1 LiF:

F
2
+
active element; 2 sample; 3 calorimeter; D diaphragm
Figure 6 shows oscillograms of the ccl and scattered radiation pulses. A series of
oscillograms in Fig.6 (left) illustrates the SBS of ccl pump radiation in YAG. Strong back
scattering typical of SBS was observed in the YAG sample, placed at different positions to
the pump. The duration of backscattered pulses was always smaller than the pump pulse
duration (2030 ns) and constituted 815 ns. At the oscillogram of the ccl pulse which
passed the sample and the diaphragm D, one observes a pump cut-off typical for SBS,
Fig. 6 (left, I, b). This pulse cut-off was connected with the reflection of the ccl beam
incident onto the sample under the occurrence of a periodic grating formed by hypersonic
waves.
Figure 6 (right) illustrates oscillograms of scattering in the YAG and crystalline quartz
samples in forward and backward directions, as well as at 90
0
to the ccl beam. Mirror R6
(Fig.5) focused the backward scattering at angles from 6
0
to 43
0
, and mirror R7 - forward
scattering at 5
0
to 25
0
in respect to the pump beam. The photodiode FD 6 recorded the 90
0
scattering. The data obtained on the scattering of wide-band low-coherent ccl radiation
allow one to consider the pulses observed as different forms of the stimulated scattering.
The scattering backward to the ccl aperture proves to be the SBS of pump, as noted above.
The forward scattering at large angles may be attributed to the anti-Stokes component of
SRS, and the backward scattering at large angle is due to the Stokes component of SRS.
Scattering at 90
0
, apparently, is due to the SBS and SRS of pump in the transverse direction.

375
Profiles of the scattered pulses observed in the oscilloscope, Fig.6 (right) illustrate the
dynamics of SBS and SRS in samples and the influence of different components of
stimulated scattering on each other. It should be noted that the pump intensities at which
SBS and SRS were observed in our experiments (0.5 GW/m
2
), had been almost an order
lower than those typical of scattering at a coherent pumping.

Fig. 6. (Left) SBS of ccl radiation in YAG slab (4,53030mm). Oscillogams of SBS pulses (a)
and ccl pulses behind the diaphragm D (b) for different orientations of the slab: I - pumping
along the slab long side; II - pump beam along the short side; III - sample at Brewster angle
to the pump. (Right) Oscillograms of scattering for YAG (I) and crystalline quartz (II): a
SBS pulse; bscattering at 90
0
to the pump; canti-Stokes SRS signal, dStokes SRS signal
After the action of ccl pulses of 4-5 GW/m
2
intensity onto samples multiple marks of such
an action in the form of volume and surface damages, and changes of the material structure
were observed. Microphotographs of structural changes were collected, data were
systematized. The material of transparent dielectrics was found to be damaged in different
ways. Cracks in the medium, local regions of structure changes, tracks of self-focusing, and
some other types of optical damage were observed near the surface and in the depth of a
sample in the direction of the pump beam. Figures 7-9 demonstrate microphotographs of
structural changes in YAG and LiF samples. One can observe plane formations of several
micron thicknesses with a sharp boundary, Figs.7, 9. Most of such objects were concentrated
near the sample surface at depths < 200 m. Fig.8 demonstrates cracking at the LiF crystal
surface. All these figures are illustrative for the energy release in the subsurface layers of
samples. The optical damage in transparent dielectrics caused by high-power radiation was
observed and studied for a few decades. The structure changes observed in the ccl
experiments were compared with literature on the interaction of highly coherent radiation
with transparent media. It was found that the low-coherent ccl radiation causes structural
changes of some new kinds, which are not described in literature, for example, flat regions
with structural changes bordered by a sharp boundary. The effects of nonlinear interaction
of ccl pump radiation with transparent optical media (stimulated scattering, structural
changes) were interpreted under the assumption that the SBS of low-coherent ccl radiation
causes a succession of hypersonic shock waves. The formation of hypersonic shock waves at


376
elastic nonlinearity at SBS was considered theoretically (Polyakova, 1966, 1968). But those
results have not been so far supported by SBS experiments using the coherent pumping.

a b
Fig. 7. Structure changes in YAG plate in 2 different ccl shots (a, b): left sample surface,
right- 7-10 m under the surface. Wavy lines on the photos are supposed to be caused by the
hypersonic waves interference inside the sample, connected with the aberration (coma) of
the focusing lens, L
2
. Scale ~ 11,8 m/mark

Fig. 8. Cracking at the LiF crystal surface. Scale ~ 11,8 m/mark

a b c
Fig. 9. Cross-like cracking having an additional plane formation with a sharp boundary
inside the LiF crystal (a); ellipse-like cracking in the YAG crystal (b); a group of rectangular
damages in the LiF crystal (c). Scale ~11,8 m/mark
Structural changes in samples subjected to ccl pulses allow one to suppose the effect of
hypersonic shock waves. Actually, a microphotograph of Fig.7a shows a region of the
medium located at the depth of 7-10 m under a YAG plate surface, where one can clearly
see a quasi-periodic structure of stripes imprinted into an oval region with a sharp
boundary. Figure 7b demonstrates another region, where a similar structure with step 10-15
m is shown. These structures happened to be visible thanks to the refractive index

377
gradient. However, such structures could not be formed in a medium as a result of the
interference of low-coherent ccl radiation. At the same time, these structures could be the
result of interference of intensive hypersonic waves propagating at a small angle to each
other in the region of SBS of pump radiation. There are some other experimental data, which
support a hypothesis of origin of intensive hypersonic waves in the medium. For example, a
sharp boundary outlining structural changes in an YAG crystal, Fig.7 is the evidence of
matter displacement deep down caused by the motion of dislocations in a surface region
probably due to the phase transition in the low-lying layers in the region of hypersonic
waves propagation. Note, that no apparent disturbances of the sample surface were
observed. It is seen that some elements at the surface inside the oval region are near the
same as they are outside: the pre-radiation scratches across the region boundary being
continued, Fig.7.
To interpret the phenomena observed (stimulated scattering, structural changes) we
consider a physical picture of the ccl radiation interaction with the medium, and a possible
mechanism of the hypersonic waves buildup. The pump radiation spectral width of 600
permits one to suppose that a ccl pulse represents a sequence of ultra-short pulses (USP) of
fs durations distributed chaotically within the 2030 ns envelope of the pump pulse. Energy
of a single USP is estimated as 0.11 J. Power density of separate ultra-short spikes could
exceed 10
10
W/m
2
. The interaction of ps and fs pulses with optical media was intensively
studied, see for example (Nelson et al.,1982). The interaction of a single fs pulse (10
13
W/m
2

intensity) with transparent dielectrics (quartz, LiF, CaF
2
, and others) produced generation of
coherent optical phonons in the THz region leading to Raman scattering of probe

pulses
with ps delay (Gordienko et al., 2010; Merlin, 1997). The hypersonic waves (GHz region)
were excited in liquids and solid-state media due to SBS on a space grating at the
interference of two identical laser beams (10
10
W/m
2
intensity, ps durations) propagating at
an angle to each other (Nelson et al., 1982; Robinson et al., 1984). In our conditions samples
were irradiated by a succession of the chaotically distributed USP with a spectral range,
which may be characterized by the central wavelength
and the wave vector k

p
= 2/
.
The appearance of shock waves at low-coherent pumping and their absence at the coherent
pumping may be due to different conditions of SBS of pump radiation for those two cases.
As is known, the SBS of coherent pump is developed on density fluctuations, i.e. hypersonic
waves of the thermal origin (Ritus, 1982; Starunov & Fabelinsky, 1969). Aside from thermal
fluctuations, there exist the stationary inhomogeneities in any solid-state medium, such as
micro-inclusions, lattice inhomogeneities, dislocations. The stationary inhomogeneities may
be the source of excitation of hypersonic waves and SBS. In fact, the interference of the
incident and reflected pump radiation near inhomogeneity gives rise to a standing light
wave with a spatial period
/2n
p
where n
p
is the refractive index of the medium at
. The
standing wave amplitude is determined by the coefficient of the pump radiation reflection
from an inhomogeneity. The standing wave causes modulation of the refractive index,
which gives rise to a spatial structure with the period =
/2n
p
, which is resonant to the
incident pump radiation and can be considered as the seed of the hypersonic wave. This
structure starts moving due to the absorption of the momentum of light at scattering of
pump photons. However, the stationary inhomogeneities do not exert influence on the
development of hypersonic waves at the quasi-continuous coherent pumping, because the
constant generation of the perturbations near the inhomogeneity is a drawback to the
development of hypersonic waves and SBS.


378
At the pulsed low-coherence pumping, not only fluctuations but stationary inhomogeneities
may be the source of SBS and hypersonic waves as well. Let an interaction of a single USP (a
spike of pump) with the radiation reflected by an inhomogeneity produces a spatial
structure with the period =
/2n
p
in a medium at the distance corresponding to the spike
coherence length (for YAG, n
p
1.82, 0.25 m). For USP with durations ranging 0.01-0.1
ps the coherence length makes 5-50 m. If the spike duration t
s
is less than half period of
the hypersonic wave t
sp
< T/2 (T =
/2n
p
v
s
50 ps, v
s
510
5
m/s -sound velocity in YAG)
then after the spike end the periodical perturbations should propagate in the direction of
pump and backward in the form of two travelling hypersonic waves, whose lifetime in the
solid-state medium is 10
-9
s for the GHz frequencies (Ritus, 1982; Starunov & Fabelinsky,
1969). In this time interval, the hypersonic waves may fall under influence of many other
USP pulses, which follow the first spike. The pump radiation affects the hypersonic waves,
and as a result it is partially scattered with changing frequency due to the Doppler shift both
to the Stokes and anti-Stokes sides. In the anti-Stokes case an atom moving towards pump is
decelerated by absorbing momentum from a counter propagating pump photon. That stops
eventually the counter-propagating hypersonic wave. At Stokes scattering, however,
momentum from a co-propagating pump photon is transferred to an atom in the direction of
its motion, supporting the hypersonic wave propagation. Therefore, only one from the two
hypersonic waves, i.e. the wave moving along the pump direction is sustained. Herewith
the scattered light wave at the Stokes frequency
s
propagates towards the pump and
stimulates emission (at frequency
s
) of the excited atoms, which absorbed a pump photon
(at frequency
p
). The difference of the photons energy, =
-
s
, determines the
amplification of the hypersonic wave, co-propagating the pump. At the same time this wave
is extended towards the pump source. In other words, the scattered light wave deflates
energy from the pump to the co-propagating hypersonic wave and extends it toward the
pump source by half of the coherence length of the USP. The following next pump spikes
running over the hypersonic wave should extend it up to the sample input surface.
Because the spatial structure of the co-propagating hypersonic wave, which is resonant with
pump, reflects effectively the incident radiation, the maximum photon energy and
momentum are transferred to atoms in a leading edge of this structure facing the pump
source. The energy is accumulated there as a result of a multiple run of USP pulses on a
hypersonic wave, which stops near the surface unable to propagate further. Thus, the
hypersonic wave amplitude rises. This provides jumps of pressure at the wave fronts, which
make the sound velocity higher at the maxima and the leading fronts steeper. The
hypersonic wave is transformed into a periodic shock wave due to this growing
anharmonicity. Thus concentration of acoustic energy near the input surface of the sample
occurs.
A proposed model of shock waves formation at SBS of low-coherent ccl radiation makes it
possible to interpret the experimental data on stimulated scattering and structure changes in
the medium. A directed action of high-energy phonons causes structural changes near the
sample surface, as shown on microphotographs of Figs.7-9. Release of acoustic energy and
the rise of tension in a sub-surface layer may be evidenced by cracking of a LiF crystal
surface, Fig.8. Other evidences may be the flat regions of structural changes, Figs.7, 9. The
analysis of experimental data on lasing in the Yb-doped samples (see section 4-6) made it
possible to estimate pressure at the shock wave fronts (tens of GPa). The pressure above 10
GPa is higher than the pressure of phase transitions in some materials. One may assume that

379
structural changes of YAG crystal with a sharp boundary, Fig.7, 9 are rather due to phase
transitions in a medium.
It should be noted that propagation of a hypersonic wave produced at some inhomogeneity
deep in the medium may be stopped inside the sample by a barrier in the form of a
dislocation, a crack, and the like. The wave would be partially stopped by the barrier. Its
acoustic energy is released in the sample near the side of the barrier. In the case of acoustic
energy exceeding the crystal lattice deformation threshold or the phase transition threshold
there occurs local splitting of medium, shift of the barriers wall. In this way one can explain
the observed structural changes in crystals occurring along cleavage planes, Fig.9. Figure 9a
demonstrates a typical cross-like cracking of a LiF crystal having an additional plane
formation with a sharp boundary. Figure 9b shows an ellipse-like splitting in the YAG
crystal. Sharp boundaries in a region of structural changes may be interpreted as phase
transition boundaries in the medium. Figure 9c illustrates a group of rectangular damages,
which seems to be due to the concentration of acoustic energy on defects (dislocation walls)
inside the sample. The formation of tensions around damages, which seen in the form of
blurred fringes, Fig.9c, is also the evidence of a considerable release of energy in that region.
The comparison of mechanisms for hypersonic wave formation at SBS of the coherent and
incoherent pump may be supplemented by the following considerations. The efficiency of
the pump energy transfer into the acoustic energy proves to be higher in the case of the low-
coherence pumping. At similar pump energy the lightacoustic wave interaction proceeds
more effectively when pumping by USP with higher intensities and at small distances (USP
coherence lengths) as compared with the case of a high-coherence radiation. Rather large
amplitude of light wave scattered on stationary inhomogeneities (as compared to
fluctuation scattering) and high intensity of the low-coherence spikes (compared to average
level of the pump intensity) should make the SBS threshold lower and stimulate formation
of the intensive hypersonic waves. This explains the appearance of intensive hypersonic
waves at low-coherence pumping. At the coherent pumping, the hypersonic wave seeded
by a small-dimensional fluctuation of pressure, needs large distances for its development,
and cannot grow to a shock wave. Note, that at the low-coherence pumping, the
hypersonic wave may be caused even by a large-scale inhomogeneity, like a plane back
surface of a sample. In that case, the produced hypersonic wave has a wide wave front.
Even in a thin sample (1-2mm) this may cause structural changes of medium (phase
transitions) comparable with dimensions of the pump spot.
Sharpening of the hypersonic wave fronts provides the appearance of high-frequency
harmonics up to the optical phonons excitation. Moreover, as steepness of the shock waves
increases, the momentum transferred to waves from successive USP pumping increases too,
thus exciting oscillations of atoms (coherent phonons) with frequencies inversely
proportional to the USP duration. The USP of the pump being scattered on coherent
phonons (like probe pulses in works Gordienko et al., 2010; Merlin, 1997) give rise to the
stimulated Stokes or anti-Stokes scattering. As optical phonons are concentrated on shock-
wave fronts, hence the stimulated scattering is linked to fronts, Fig.10. Due to a large shift of
the Stokes wavelength (
s
), the SRS may be amplified only at the angle to the pump: cos =
n
s
p
/n
p
s
, Fig.10. The anti-Stokes SRS is developed in the opposite direction. The respective
directions and angles of scattering are illustrated in Fig.10. It should be noted that high-
intensity optical phonons excited at shockwave fronts stir up the molecules at amplitudes
and frequencies close to the possible limit in the crystal lattice, and hence make easier
structural changing of the medium, i.e. its phase transition to a more dense state.


380

Fig. 10. Illustration to Stokes (
s
) and anti-Stokes (
as
) scattering at hypersonic shock wave
fronts: angle between Stokes and pump waves; , - angles between anti-Stokes and
pump: n
s
p
/n
p
s
= cos; n
s
as
/n
as
s
= sin = cos; - = and = 90
- ( ), = 90 - ( + ).
4. Experiments on pumping Yb-doped media by LiF: F
2
+
ccl
Experiments on pumping Yb-doped media were performed using a setup of Fig. 3. Samples:
plane parallel 2-mm-thick Yb :YAG (20% of Yb) or 3-mm Yb-doped phosphate glass (10%
of Yb) plates were mounted in a resonator of length L 20 mm formed by plane mirrors M8
and M9. Samples were pumped by pulses of focused ccl radiation through mirror M8 with
the reflectance 100% at 1 m transmitting 80%-90% of pump radiation. The pump beam
propagated close to the normal to the resonator mirrors. The reflectance of mirror M9 at 1
m was 30% (for Yb: YAG) or 70% (for Yb: glass). The energy, spectrum, shape of Yb-laser
pulses were registered by calorimeters, photodiodes, and STE-1 spectrograph. Experiments
were performed at room temperature in the single-shot regime. In most experiments
samples in the resonator were exposed to the ccl radiation of energy 60 - 70 mJ. The focused
ccl beam produced the radiation intensity distribution I (r) (r-radius of the excited region) in
the subsurface layers of samples with the maximum at the pump beam axis. The minimal
size 2r of the focal spot on the sample surface was 250 m. The ccl intensity was varied
within 0.5 - 5 GW cm
-2
by moving the lens. Lasing in a Yb :YAG crystal was observed at
pump intensities exceeding 0.5 GW cm
-2
, Yb laser energy did not exceed 0.5 mJ. Figures 11
and 12 show Yb-lasing oscillograms recorded for one shot. The first of a series of laser pulses
appeared during the pump pulse, while the next pulses appeared after the end of the pump
and were delayed by tens and hundreds of ns, up to 1.2 microsecond (s). Spectra were
recorded by STE-1 on an IR film or on the image converter. The optical scheme of the STE-1
imaged (1:1) the vertical slit of 100m width on the film. At uniform illumination of the slit
by the monochromatic radiation of a spectral lamp the film registered separate narrow
vertical lines, Fig.4b. In case of Yb-lasing we observed much more complex spectra, Figs 11-
14. The lasing spectra were recorded within an angle of 10
-2
rad, which was determined by

381
the slit height and a distance (1m) from the sample. Laser radiation going along the
resonator axis was directed to the lower part of the slit. No focusing of radiation on the slit
was performed. Figs 11, 12 demonstrate lines in the region of Yb transitions in YAG at 1.03
and 1.05 m. At low pump intensities line lasing appeared first at 1.05 m. As the pump
intensity was increased, line lasing was observed simultaneously at 1.03 and 1.05 m (Fig.
11), and at the maximum pump it was observed at 1.03 m, Fig.12. Fig.11 demonstrates a
noticeable inclination of the spectral line at 1.03 m from the vertical direction. This means
that the lasing wavelength in this line was varied along the slit height. Structure of small-
scale (50-200m) spots was observed in 1.05 m line, Fig.11. Along with narrowband lasing,
broadband unstructured lasing was observed virtually in each pulse. Spectral bands up to
20 nm width were extended to the blue (up to 1.0 m) and red (up to 1.06 m) parts of the
spectrum far beyond the regions near 1.03 and 1.05 m in which Yb :YAG lasing is usually
observed. As pump was increased, the broadband spectra were shifted to the blue
(Fig. 12).

Fig. 11. Oscillograms of pump (1) and Yb:YAG laser (2) pulses, and the lasing spectrum in
the 1.03 1.06-m region

Fig. 12. Oscillograms of pump (1) and Yb:YAG laser (2) pulses and lasing spectra in 1.03
1.06-m region for 3 ccl pulses with energies 100, 125, and 150 mJ (from left to right)


382

Fig. 13. Lasing spectra of Yb: glass in the 1.00 1.05-m region
Lasing in the Yb: glass (Figs 13,14) appeared at pump intensities 4-5 GW cm
-2
and was
accompanied, as a rule, by the local damage of samples. The lasing energy did not exceed
0.1 mJ. The oscillograms of the glass lasing (similar to the Yb :YAG lasing) exhibit several ns
pulses delayed with respect to the pump, Fig.14. The spectrograms demonstrate line (in
1.02-1.05 m region) and broadband (1.00-1.06 m) lasing spectra. Structures of small-scale
spots are clearly visible in Yb:glass line spectra, Fig.14. The total spectrum of two broad
bands of the Yb:glass emission exceeded 50 nm, Fig. 13. The width of stripes of broadband
lasing in vertical direction was about 1mm, Fig.13. The line and broadband lasing spectra
both in glass and Yb:YAG were recorded from the slit parts located at different heights, Figs
11-14. Note, that spectra of Yb lasing were recorded when the slit was in the far-field
diffraction zone with respect to the position of the radiation source. So, the observed narrow
stripes and small spots of generation, Figs 11-14, indicated a high directivity of radiation.
The angular divergence of radiation, estimated from the size of spots on the slit, did not
exceed usually 10
-3
-10
-4
rad. At the same time, the angular divergence caused by diffraction
from the lasing region with r 100 m should be 10
-2
rad. Thus, the angular directivity of
Yb-lasing could exceed the diffraction limit by one- two orders of magnitude, i.e. the
generation should occur in highly directed beams of small transverse size.
At pump intensities from 0.5 to 5 GWcm
-2
used in our experiments, we observed intense
pulses of scattered ccl radiation in Yb-doped crystal and glass samples similar to
observations of scattering (SBS and SRS) in non-doped samples described in the part 3.
Consider the features of the inversion distribution in the Yb-doped medium upon SBS of
pump radiation. The dynamic grating of hypersonic waves produced in samples near the
surface played the role of an additional external mirror for the ccl and efficiently reflected
incident radiation, preventing its propagation inside a crystal or glass sample. The
penetration depth of hypersonic waves into the medium is l 1/, ( = 10
2
- 10
3
cm
-1
is the
hypersonic waves attenuation coefficient) (Ritus, 1982; Starunov & Fabelinsky, 1969). Thus,
the excitation region of SBS (and SRS) in the medium was a surface layer of thickness l < 100
m. The heat release caused by the dissipation of energy of hypersonic waves and optical
phonons excited in the medium upon SRS of ccl radiation occurred in this layer. Inversion
was also mainly produced in this thin layer. The maximum energy density stored in Yb
:YAG containing 20% of Yb (N
0
= 2.9 x10
21
cm
-3
) upon excitation of all Yb
3+.
ions to the
2
F
5/2

metastable level was estimated as 500 J cm
-3
. Even for such an extremely large energy in

383
the volume of an Yb :YAG crystal cylinder of diameter 250 m and length 100 m, the
energy stored in inversion does not exceed, according to estimates, 2.5 mJ, which
corresponds to the low level of Yb-lasing energy in experiments. A part of the pump
energy was absorbed by Yb ions beyond the region of SBS excitation, and the other part
escaped from the sample. So, the population of Yb ions on the metastable level decreased
over the sample length along the pump beam. Only a small fraction (estimated as 10%) of
the ccl output energy incident on samples was spent to excite ytterbium. A grater part (up
to 70%) of the ccl energy was transformed to stimulated scattering due to nonlinear
interaction with the medium and was spent to form intense hypersonic waves and optical
phonons accompanied by heat release in the medium. The thermal energy required for
heating and melting of the above-mentioned Yb :YAG crystal cylinder was estimated as 30
mJ. At the same time, the thermal energy in the medium during the formation of
inversion and generation and due to stimulated scattering did not exceed, according to
estimates, 10 mJ, i.e. it was insufficient for melting of the material. This suggests that
structural changes in the surface layer of samples (Fig.7-9) could be produced by intense
hypersonic waves appearing during the SBS of pump radiation. Thus, the Yb-doped
medium during the SBS of pump proved to be divided into a layer of thickness 100 m
with the very high inversion and the main part of the sample to which pump radiation
hardly penetrated.

Fig. 14. Oscillograms of Yb: glass laser pulses and spectra in the 1.00 1.04-m region
Figures 11-14 demonstrate the integrated (per pulse) temporal pictures and broadband and
narrowband lasing spectra of ytterbium. These two types of emission correspond to
different Yb-laser pulses appearing at different stages of the development of lasing. We can
distinguish two such stages: the first one is the development of Yb-lasing during irradiation
of a sample by the pump pulse, and the second one is Yb-lasing after the end of the pump
pulse. At the first stage of duration t
1
= 2030 ns, during the action of a ccl pulse onto the
sample the generation could occur without a resonator, just due to the feedback at the
dynamic hypersonic wave grating. The broadband unstructured spectra correspond to this
stage. At the second stage of duration t
2
up to 1.2 s there was generation in the Yb laser
resonator which was characterized by line spectra, Figs. 11-14.


384
5. Features of Yb- lasing: broadband spectra, data interpretation
We attribute broadband lasing in the crystal and glass to Yb-laser pulses appearing at the
first stage. Consider the conditions of inversion formation and lasing at the first stage by
using a model assuming the appearance of shock hypersonic waves in the medium during
the SBS of pump radiation. Excitation of Yb ions to the
2
F
5/2
metastable level and formation
of hypersonic wave grating (Fig. 15) occurred simultaneously in a thin layer of the medium.
The grating period is related to the pump wavelength
p
and the refractive index of the
medium n
p
by the expression =
p
/(2n
p
) (Ritus, 1982; Starunov & Fabelinsky, 1969). By
assuming that
p
0.92 m, we obtain for YAG 0.25 m. The period of the hypersonic
wave is T=
p
/(2n
p
v
s
), where v
s
is the sound speed. For the glass, we have T 70 ps, and for
the YAG crystal, T 50 ps. Intensive hypersonic waves strongly affected the spatial
distribution of inversion. Indeed, conditions were produced for excitation of high-frequency
vibrations of the medium (optical phonons) and SRS on the fronts of shock waves. The
scattering parameters show that a broad phonon spectrum was excited upon SRS, including
phonons with energies 1000 cm
-1
near the high-frequency boundary of the phonon
spectrum in the medium (Gorelik, 2007). The energy of such phonons is sufficient for the
population of the Stark components of the
2
F
7/2
level lying above the ground level of Yb
(Fig.1). The redistribution of the population of Stark levels occurs during several ps. Due to
the phonon population on the fronts of shock waves, the inversion for transitions from the
2
F
5/2
level to the Stark components of the
2
F
7/2
level was decreased, preventing the
development of lasing. At the same time, due to the rapid decrease of pressure and dynamic
cooling of the medium, which occurred behind the fronts of shock waves (Fig. 15), the
population of the Stark components of the
2
F
7/2
level was rapidly depleted, resulting in the
inversion jump. As a result, a structure of thin (<0.3 m) periodical layers with the high
inversion was established in the medium for a short time (t
inv
< T). During the propagation
of a hypersonic wave, the inversion regions were displaced. After the time T, the same
inversion distribution over layers again reestablished. Thus, upon the SBS and SRS of pump
radiation, amplification could appear at transitions between the
2
F
5/2
level and Stark
components of the lower
2
F
7/2
level, including transitions in the short-wavelength region (<
1.03 m), which are not observed usually in Yb- lasers.

Fig. 15. Illustration to conditions for inversion formation and stimulated emission in the
region of shock hypersonic waves in the active medium. The dependence of the pressure
profile P in the hypersonic wave on the coordinate x along the pump direction is shown. The
dashed straight line indicates the pressure at which inversion is produced. The possible
relations between the pump,
p
and laser,
g
wavelengths satisfying condition (1) are given

385
The formation of inversion in the region of propagation of hypersonic waves produces the
conditions for generation of short radiation pulses in the medium. Indeed, a distributed
feedback (DFB) laser (Kogelnik & Shank, 1971) appeared in fact in our experiments in the
hypersonic wave grating region. Such a laser operates without an external resonator and can
emit ps pulses (Bor & Muller, 1986; Katarkevich et al, 1996; Kogelnik & Shank, 1971). Unlike
dye DFB lasers based on a photo-induced sinusoidal grating in the medium, the high-
contrast DFB structure was formed under our conditions due to modulation of the medium
parameters (refractive index, inversion) during the propagation of a sequence of intense
hypersonic waves. The possibility of creating a DFB laser on a hypersonic wave grating was
discussed already in pioneering paper (Kogelnik & Shank, 1971). The length of the DFB
structure in the region of SBS is 100 m; in this case, the travel time t of a photon in the
structure is much shorter than the round-trip transit time in the Yb- laser resonator ( =
2L/c 100 ps). The possibility of lasing using the hypersonic wave grating without the
resonator was confirmed by experiments. Lasing during the ccl pump pulse was observed
both in samples mounted inside the resonator and in samples without the resonator, in
particular, in samples mounted at an angle to the pump beam axis. The generation of ps
pulses in dye DFB lasers on a stationary grating pumped by ns pulses occurs due to self-Q-
switching mechanism: `blowing away' of the gain grating by the structure emission itself
(Bor & Muller, 1986; Katarkevich et al, 1996). Unlike this, upon the SBS of ccl pump
radiation, a DFB structure moving in the medium appears with period which can depend
on the pump intensity (pressure in the medium). Under these conditions, the Q-switching in
the DFB structure can be determined, apart from the mechanism considered above, by the
rapid movement of the grating in the medium and by variations in its parameters as well.
Pulses of ps duration, t < t
inv
should be generated in our experiments successively in these
periodic inversion layers during their movement. Besides, the change in should lead to
scanning of the lasing spectral band over the gain profile in the medium. The temporal
picture of Yb- lasing at the first stage (during pump) was not investigated in detail due to
the lack of recording equipment with a proper time resolution. The shape and duration of
Yb-laser pulses were determined by the resolution of the oscilloscope (1- 4 ns).
By considering the conditions for appearing of stimulated emission in the moving layer
structure, we can interpret features of broadband lasing spectra of the glass and Yb :YAG
crystal (Figs 11-14). The Yb emission appears at wavelengths
g
>
p
. This emission is
amplified along the normal to the layer structure when the path difference between the
layers involved in lasing is equal to the quantity multiple of
g
. This synchronism condition
between hypersonic waves and lasing can be written in the form

2
g p
g p
k m
n n

= (1)
Here, n
g
is the refractive index of the medium for
g
; and k and m are integers. It is important
that the refractive index depends on pressure P in the medium: n
g
, n
p
~ P. Pressure depends
on the intensity of hypersonic waves and the heat release in the medium upon pumping.
The maxima of the spatial structure of the laser field are located in regions with a decreased
density, behind the fronts of shock waves, Fig. 15. Here n
g
can take values lower than the
refractive index n
0
at the normal pressure. At the same time, the refractive index n
p
on the
fronts of shock waves can be larger than at the normal pressure. Under these conditions for
n
g
< n
p
, relation (1) is fulfilled for k = 1 for small values of m (m > 2, m = 3, 4..., see, Fig. 15). It


386
follows from the relation =
p
/(2n
p
) that, as the intensity of hypersonic waves and heat
release increase (with increasing P and n
p
) at the fixed
p
, the period of the grating
decreases. This in turn leads to the blue shift of
g
. Correspondingly, as P is decreased, the
wavelength
g
should shift to the red. Experiments confirm these conclusions. The shift of
broadband spectra with changing the pump intensity is illustrated in Fig. 12. One can see
that the spectrum of Yb:YAG shifts to the blue by 10 nm with increasing the ccl pulse energy
from 100 to 150 mJ. Note that the shift of the emission region of DFB lasers with changing of
the pressure was observed in dye lasers (Bor & Muller, 1986).
The inhomogeneous distribution of the pump with the maximum at the beam axis gave rise
to the transverse gradient of pressure P(r) in the region of propagation of hypersonic waves.
Because of this gradient, for regions with different pump intensities (at the beam axis and its
periphery) relation (1) is fulfilled for different values of m. These values can correspond to
the regions of the broadband spectrum shifted in the wavelength scale and over the slit
height, Figs 11-14. In the axial region with the maximum pressure (r 0, the lower part of
spectrograms), lasing was developed in the short-wavelength part of the spectrum. The
spectra in regions with the lower pump (r > 0) are shifted to the red (Figs 11-14).
The regions of the broadband spectrum should be located on the wavelength scale in
accordance with possible transitions in Yb: glass and Yb:YAG. According to the energy level
diagram presented in Fig. 1, the emission of Yb:YAG in the spectral region under study (Figs
11, 12) can be determined by transitions between the two lower components of the
metastable
2
F
5/2
level with energies 10327 and 10624 cm
-1
and three components of the
2
F
7/2

level with energies 565, 612, and 785 cm
-1
. In each spectral region, lasing in the DFB
structure was built up in the active medium within the part of the gain line corresponding to
the Yb transition which was resonant for the particular period of hypersonic waves. Thus, a
large width of lasing spectra of Yb-doped materials (Figs 11 - 14) could be caused not only
by the generation of short pulses in the DFB structure upon emission of thin layers of the
active medium but, as explained above, by the shifts of
g
with changing the pump intensity
as well. This is a picture of lasing in Yb-doped media in the region of the propagation of
longitudinal hypersonic waves at SBS of pump radiation. Thus, the data on Yb broadband
lasing confirm the concept of intense (shock) hypersonic wave formation at powerful low-
coherent pumping of the optical media.
The interpretation of data on the sub-diffraction angular divergence of Yb:YAG and
Yb:glass generation had required the introduction of a new concept for the photon, which
means its existence not in the form of a traveling wave, but with fixed positions of
maxima, minima, and nodes along the photon propagation direction (Bykovsky, 2006;
Bykovsky & Senatsky, 2008a,b, 2010). It was assumed that the photon is a combination of the
two pinpoint dipoles, which move jointly with light velocity and rotate with frequency in
the opposite directions. The phase of the photon and its field are connected with the phase
of this rotation (Bykovsky, 2006). Such an approach helps to describe the effects observed.
The field produced by the photon motion along the coordinate x is described by the
function:
0 0
sin( ) E E kx = , where k
2
= , and is the distance, at which the photon

changes its phase by 2; E
0
the field amplitude,
0
- the initial phase. The new concept
does not contradict to the properties of the well-known electromagnetic traveling wave:
the traveling wave may be constructed from separate photons following each other along
the same direction and having a continuous relative phase shift from 0 to 2. If we assume
that the excited atoms in a medium irradiate photons along the same direction and with the

387
initial phases +/2 or -/2 only, and that emitted photons are correlated by the relation
=t, where is the distance between the neighbor photons and t - the interval between
the moments of their emission, then such a combination of photons should form a traveling
wave. This assumption follows from an equation describing a field of a neighbor photon
irradiated from a point distanced by :
0 0
sin[ ( ) ] E E k x x = . Taking into account that
k x kc t t = = we obtain:
0 0
sin( ) E E kx t = - the traveling wave relation. The
traveling wave amplitude increases in proportion to the number of equal-phase photons. To
provide the traveling wave formation it is necessary that the size of a homogeneous region
with inversion, from which a large number of photons is spontaneously irradiated into 4,
exceeded the wavelength . Organization of photons in the form of the traveling wave makes
it possible to deplete inversion with maximum efficiency, as it is taken off continuously from
all points on the path of the wave propagating in the active medium. If the inversion is non-
homogeneous and structured with the period /2, such a medium will emit photon in trains,
whose maxima should spatially coincide with the inversion maxima. Photons having phases
differing by 2n form a photon train described by the function:
0 0
sin( 2 )
n
E E kx n = . That
should not be a traveling wave. `The instant photograph' of the filed distribution in a photon
train should represent the contrast periodic structure of antinodes and nodes and can
noticeably differ from `the instant photograph' of a traveling sinusoidal wave. However, the
summary field produced by many photon trains, which are correlated by the relation =t,
becomes traveling. This traveling is like the Christmas-tree garland, whose running lights
are due to the time shift of neighboring lamps burning moments.
Because the field produced by the photon is described by a harmonic function, the field
superposition of the photon trains propagating at different angles and with different phases
gives rise to the wave effects: interference and diffraction. The diffraction is connected with
spatial confinement of the light wave front and caused by the interference of photons with
different phases contained in the aperture of a beam. The intensity peaks are formed in the
directions defined by the beam wave front profile, dimension and shape of a confined
region, and the light wavelength. By the angular width of these peaks one may judge about the
light beam divergence. The diffraction divergence of the traveling wave arises because the
traveling wave restricted by an aperture contains within this aperture at the length of one
spatial period the photons with different phases ranging from 0 to 2. In general the phase
dispersion of photons distributed in the interval x along the beam direction can be estimated
as = 2x/. When the beam is limited by an aperture of diameter d, the direction to the first
minimum of the far-field intensity distribution (half the angle of the diffraction divergence)
can be written in the form ~ /(2d). In the case of a traveling plane monochromatic wave,
the phase dispersion in the beam can achieve . This gives the known estimation ~ /d of
the divergence angle for a beam limited by an aperture of diameter d. If the phase dispersion
of photons is decreased, the diffraction divergence angle decreases proportionally to . The
group of such phased photons will form a beam with the sharp angular radiation pattern,
which was observed in our experiments. Formation of beams with sub diffraction divergence
occurs, apparently, in experiments on super-resolution microscopy, when the size of the
fluorescent or light scattering object along the direction of beam does not exceed the light
wavelength (Galbraith, C.& Galbraith, J., 2011; Tychinskii, V., 2008).
Unlike a usual laser with the extended active medium in the resonator, in our experiments
radiation was formed in a structure consisting of thin inversion layers. The regions with
reduced pressure behind the front of shock waves, shown in Fig. 15, represent as if the


388
`instant photograph' of the spatial arrangement of such a structure. The photons with the
wavelength resonant with the structure,
g
/n
g
= m, will occupy different spatial positions in
the `photograph' with respect to the layers of thickness l
inv

g
/n
g
. And only a small part of
the total number of photons in the `resonator' of our DFB laser will coincide in position and
phase with the inversion maxima in the medium and, therefore, will be amplified efficiently
during the lifetime t
inv
< T of the layers of the given spatial configuration, forming a beam on
the basis of photon trains. During the movement of the inversion layers structure (following
the movement of the hypersonic wave), new groups of phased photons will be formed, and
their frequency will shift with changing the structure parameters. In such a way the model
considered above explains the appearance of the sharp angular radiation pattern of the
broadband lasing of ytterbium in the crystal and glass. In the case of a `thick' sinusoidal
grating, photons occupying different positions in space (which is equivalent to the large
phase dispersion within the photon train) are amplified. This leads to the increase in the
radiation divergence. Thus, when a periodic spatial grating of thin inversion layers is
produced in the active medium, the high-directional stimulated radiation of an ensemble of
excited atoms can be observed in the optical range.
6. Features of ytterbium lasing: line spectra, data interpretation
Line spectra observed in the crystal and glass relate to the second stage of Yb-lasing. The
second stage is the Yb-lasing in the resonator after the end of pump due to inversion
remained in the medium. Of special interest the details of Yb spectra are. Twisted, inclined
and structured lines in spectra were observed in the region 1.031.05 m, Figs.11-14,16.
Figures 11, 13, 14 showed an inclination of some lines to the short-wavelength side of the
spectrum in respect to the vertical position of the slit. Nearly all line spectra showed quasi-
periodic structures of small-scale (50200 m) spots (Figs. 11, 14, 16). Twisted lines
consisting of separate spots are shown at Fig.16. This picture indicates a wave-like change in
the generation wavelength in separate lasing spots along the height of the slit.

Fig. 16. Twisted lines with multiple spots of Yb:glass generation at several transitions near
1.04 m; time delay from the pump pulse 50 ns

389
Consider the conditions for the development of lasing in a resonator at the end of pump,
under disappearance of longitudinal hypersonic waves and relaxation of the excitation
region in the active medium. Lasing developed in the resonator in the presence of the region
of strong optical inhomogeneity near the active element surface- a thin layer with the
inhomogeneous inversion distribution and gradient of the refractive index, n produced by
pressure, temperature, and density gradients. To the rest part of the sample the pump
penetrated weakly. The refractive index profile was formed during the action of the pump
on a medium and it also changed after the end of pump during the relaxation of the excited
region. This profile affected the development of Yb-lasing. An experimental study of a non-
stationary refractive index profile (so-called transient lens) arising under short laser pulse
interaction with a medium is a rather complicated task. Here we discuss in a qualitative
form processes that affected formation of the index profile, n(r) assuming that pumping was
symmetrical over the azimuth angle. Consider again a part of the medium in the form of a
cylinder of 250 m diameter and 100 m length, which fits the size and configuration of the
focal region at tight ccl focusing. The heat load occurred in such a cylinder during inversion
formation, Yb-lasing, and mainly, due to pump radiation scattering. The heat load was
estimated to be <10 mJ, which allows for the average thermal energy density in the cylinder,
< 2 10
3
J/cm
3
. Estimations for the average pressure, P
av
and temperature, T
av
just after the
end of pump give P
av
1 GPa and T
av
10
3
K. In actuality, due to the bell-shaped profile of
the pump beam, P(r) and T(r) profiles with maxima exceeding averaged values arose in the
medium. The maximum pressure near the axis of the excitation region at the moment near
the end of pump pulse may be estimated by using the value P
pt
, at which a phase transition
occurs. Structure modifications similar to phase transitions had been observed in Yb:YAG
and Yb:glass samples near the axis, where hypersonic waves of maximum intensity
propagated (section 3). We use data for glass P
pt
10 GPa, n/P =-0.13 10
5
bar
1
, and
n/T = 6 10
6
K
1
(Alcock & Emmony, 2002; Koechner, 2006; Mak et al., 1990). At P(0)
P
pt
the drop of n(r) along r from axis to periphery of the cylinder due to the drop in pressure
in the medium is n(P
pt
) = n/P P
pt
= 0.13. Changes of n due to temperature gradients
are significantly less. For example, at T = 10
3
K, n(T) = n/T T = 6 10
3
. For YAG
n/T 8 10
6
K
1
(Koechner, 2006; Mak et al., 1990) and n(T) 8 10
3
.Thus, the main
contribution to the n(r) profile was made just by pressure. The n(r) profile defines a sign,
optical power and aberrations of a transient lens in the medium. Optical power of the lens
after the end of pump can be estimated at the assumption that the index had been changed
by n 0.1 at r 150 m. That gradient corresponds to appearance in the medium of a
positive lens with diameter 2r 300 m and f 1cm. A simplified scheme for the case of
lasing in the resonator with a transient lens is given in Fig. 17.
The evolution of profiles P(r) and n(r) after pumping was determined by unloading of the
high-pressure region in the focal volume. A problem of unloading of a small cylindrical or
spherical region in a solid-state optical medium, where the energy had been stored after ns
laser pulse irradiation, had been considered in several studies (Bullough & Gilman, 1966;
Conners & Thompson, 1966; Sharma & Rieckhoff, 1970). A typical process of relaxation is
the propagation of an elastic dilatational wave across the medium. The time of traveling, t
tr

of an elastic wave with the sound velocity across the excitation region can be evaluated by t
tr

=r/v
s
. For the glass, at r
p
= 250 m and v
s
= 4.5 10
5
cm/s, t
tr
50 ns. The propagation of the
dilatational wave along radial direction outward the center of the pumped region had led to
the pressure profile deformation. With the drop of pressure at the axis, the gradient dn/dr


390
decreased accordingly. So, in the course of relaxation there occurred smoothing of profiles
P(r) and n(r). It should be noted that profile n(r) could include a small-scale modulation as
well (imitated at Fig.17). The sources for such modulation might be pressure and density
perturbations that had been arising in a medium at propagation of hypersonic waves during
and after pumping. Note, that typical time of thermal relaxation of the medium in a
cylindrical region with radius r
p
is = r
p

2
/4 t
tr
. Here r
p
is the radius of the pumped region
and is the coefficient of the temperature diffusivity. For Yb:YAG at r
p
= 250 m, = 0.046
cm
2
s
1
(Koechner, 2006 ), 3 ms. Hence, the time of thermal relaxation of medium is
several orders higher than dynamic unloading. That is, roughly, a picture of formation and
evolution of the refractive index profile in the medium in the region of ccl focusing defined
by a non-uniform heat load and dynamics of the pressure profile during pumping and
under the medium relaxation. So, a sequence of acoustic waves diverging in radial direction
outward the focal region should affect the build-up of Yb-lasing in the resonator.

Fig. 17. Scheme of the experiment on excitation of Yb-doped medium in the resonator by
focused ccl radiation: 1 - active element; 2, 3 - mirrors; 4 - spectrograph STE-1 located at
distance S1m from the resonator; 5 - a picture imitating spectra registered; angle
between a generation trajectory and resonator axis. The dimensions of the resonator with
active element are exaggerated relative to the scale of the scheme. Profiles of pump beam
intensity, I(r) and pressure in the medium, P(r) are out of scale as well
The interpretation of data on the Yb-lasing in the resonator has required also the
introduction of a new concept for the photon electromagnetic field distribution in space,
considered in section 5. This new approach allows the existence of light beams of an
aperture d and with a sub-diffraction angle of divergence. Within the framework of
this new concept the picture of Yb lasing in the resonator looks as follows. The location of
line spectra in the upper part of the slit evidences that Yb radiation emergent from the
resonator was deflected from the axis, i.e., off-axis oscillations were built up in the resonator.
For the arrangement of mirrors and active element given in our experiment off-axis lasing
could be developed over trajectories with reflections at angles
1
,
2
at the mirror (3), Fig.
17. Such trajectories located at distances r
pt
< r < r
p
from the axis, should contain curvilinear
segments in the area of the optical inhomogeneity. We suppose that beams with aperture of
several (small generation channels) constituted lasing along these off-axis trajectories. At
the high value of dn/dr and the very high level of inversion, in each of the channels the
selection of emitted photons (in accordance with their phases) took place in such a way that
only a group of photons being approximately in the same phase
0
overcame the generation
threshold. The photons different by phase from
0
were scattered due to diffraction and left
the channel aperture. In the given channel with a coordinate r a group of photons of the

391
same phase made several round trips in the resonator during the time t', and a beam of the
high directivity with the generation wavelength
g
corresponding to one of the longitudinal
modes was formed. So, generation in different channels over the pumped area developed at
their own frequencies, and this radiation, reaching the slit at different points gave us an
integrated picture of lasing beams and the spectrum in the form of lines composed of small
spots, which dimensions characterized beam divergence in each of the channels.
For Yb-doped plates placed in the middle of the resonator with L = 20 mm and for
trajectories located in plates at r < 500 m, angles should lie inside the spectrograph
registration angle 10
2
rad. Beams of generation left resonator at angles and reached the
slit. The greater was the distance r between the generation trajectory inside the sample and
the resonator axis, the greater was the shift of the corresponding generation beam spot
upward in vertical direction on the slit, Fig. 17. So, under the propagation of beams of high-
directivity from resonator at angles , one, actually, observed at the slit a projection (a
magnified image) of a spatial distribution of generation channels in the active medium
over the radial coordinate. Since data on the Yb emission intensity distribution over the
azimuth angle were absent, one can speak only about partial mapping on the slit of the
spatial distribution of generation channels in the lasing area. The magnification coefficient k
of such an image may be estimated from comparison of dimensions of the possible
generation area (r 500 m) and the height of line spectra at the slit (1 cm), k 20. With
account of k factor, to observed structures of small-scale spots in line spectra should
correspond sources of Yb emission with dimensions of several wavelengths only.
Structures of multiple spots in Yb:YAG and Yb:glass lines (Figs. 11, 14, 16) may find
explanation when one takes into consideration the specific spatial configuration of the field
of thermo-elastic stresses in the area of ns laser pulse focusing into the medium. The quasi-
periodic, alternating in sign, oscillating character of the amplitude of the tangential
component of stress in dependence on the radial coordinate inside a small spherical region
of a solid-state optical medium at ns laser pulse focusing was ascertained (Conners &
Thompson, 1966; Sharma & Rieckhoff, 1970). In spherical ring zones round the centre of the
focal region the tangential tensile stresses in the medium are consequently replaced by
compressing stresses, and then, again, by the tensile ones, etc. The spatial period of such
oscillations calculated for the case of 70 mJ laser pulse focusing into the glass constituted 30
m (Sharma & Rieckhoff, 1970). In conditions of our experiment one can expect the
occurrence of oscillating (in space and time) profiles of thermo-elastic stresses in Yb-doped
media within the area of ccl radiation focusing as well. Alternating in sign stresses should
result in a small-scale modulation at the profile n(r), Fig.17. The oscillation of the n(r) profile
should stimulate the Yb generation in ring zones, which may fall into separate generation
channels. The structures of bends consisting of generation spots, Figs.16 may be considered
as a kind of an image of the distribution over the radial coordinate of tangential stress
peaks in the focal region projected with a magnification on the slit. It is possible to say that
such images were taken by the high-speed photography method. The illumination for this
high-speed photography came from the Yb laser pulse itself. The exposure time of a single
frame corresponded to the duration of the generation pulse. Time delay between the pump
pulse and the shooting moments makes 50ns (Figs. 16) and over 300 ns (Fig.14). Splitting
of a single bend into several generation spots with wavelength shifts (Fig.16) is in agreement
with the model of generation channels. The shift of a spot to the long-wavelength side
corresponds to increase of pressure (stresses) in the medium, and the shift to the short-


392
wavelengthto decrease of pressure. In the course of medium relaxation, stress amplitudes
and, correspondingly, spot wavelength shifts should have been reduced. At 50ns delay
(twisted lines, Fig.16) spots have noticeable wavelength shifts which correspond to high
stress amplitudes. Structures at moments over 300 ns after pump (Fig.14) correspond to a
smoothed picture of n(r) profile. Note, that structures recorded with such big delays indicate
the continued acoustic ringing in the medium. Estimated from Fig. 11, 14, 16 (at k = 20)
period of stress spatial oscillations in YAG and glass varies from 15 to 40 m and is in a
qualitative agreement with the calculations (Sharma & Rieckhoff, 1970). Acoustic vibrations
frequencies corresponding to these values constitute 10
8
10
9
Hz. Attenuation of phonons at
these frequencies at a room temperature for YAG is smaller than 0.1 db/s and for glass
about 10 cm
1
(Dutoit, 1974; Zhu et al., 1991). These data confirm that ringing of the
unloading medium in the focal region in YAG and glass may continue over several s.
The Yb:YAG line spectra near 1,03 m were registered usually with 1050 ns delays after the
end of ccl pulse and even together with the trailing edge of the ccl pulse. Observations of
line spectra near 1,03m, emitted soon after the pump pulse, like Fig.11, reveal noticeable
inclination of spectral lines. This bending means that the wavelength of lasing at the same
longitudinal mode (the number of nodes is preserved) changes from the centre to periphery
of the excited region in the active medium. This corresponds to the development of lasing in
some sites of the medium with a pressure (refractive index) gradient from the beam centre
to its periphery. It can be easily shown that
g
shifts to the blue if the refractive index
gradient decreases from the beam axis to its periphery, Fig. 11. The pressure drop P in the
medium after the end of pump can be estimated from this line spectral shift. The frequency
q
of the longitudinal mode of the resonator with the number of wavelengths
q
over the
resonator length 2L equal to q is described by the expression

0
( ) ( )
q
cq
L l n l n r
=
+
(2)
where l is the longitudinal size of the optical inhomogeneity of radius r; n
0
is the averaged
refractive index outside the nonlinearity region; and n(r) is the refractive index in the
nonlinear region. It follows from (2) that the change in the mode frequency
q
during the
displacement along the radius from r
1
to r
2
is

1 2
0 1 0 2
[ ( ) ( )]
[( ) ( )][( ) ( )]
q
cq l n r n r
L l n l n r L l n l n r

=
+ +
(3)
By substituting the expression for
0
2[( ) ( )]
( )
q
L l n l n r
q
r
+
= into (3) and assuming that
ln(r)<Ln
0
, we obtain the dependence of the change in the refractive index, n on r

0
( )
q
q
r Ln
n
l
(4)
Here,
q
(r) is the wavelength shift along the radius. For
q
(r) 1.4 10
-7
cm,
q
10
-4
cm, L
2 cm, n
0
1, and l 10
-2
cm, the change in the index is n 0.28. By assuming that the
change in the index is produced only by the change in pressure along the radius and using

393
the value of dn/dP for glass (Alcock & Emmony, 2002), we obtain P 30 GPa. So, this
estimates the pressure, which arises at the axis of the focal region after the end of pump.
Similar estimations are possible for the pressure jumps, which arise during the propagation
of dilatational waves outward the center of the focal region. Knowing a value of the
wavelength shift in a bend at Fig.16,
q
(r), one can estimate changes of the index, n and
pressure, P in the medium, using the expression (4). As seen from Fig.16, for glass at
q
(r)
0.3x10
7
cm,
q
10
4
cm, L 2 cm, n
0
1, l 10
-2
cm, n 0.05, and P 5 GPa. The
estimated P considerably exceeds the glass fracture strength < 0.1 GPa (Sharma &
Rieckhoff, 1970), which was measured usually for applied static load. The role of tensile
stresses in laser damage of transparent dielectrics was discussed in many publications
(Koldunov et al., 2002; Sharma & Rieckhoff, 1970; Strekalov, 2000). It was considered that
due to oscillating character of the stress amplitude in the focal area there must be observed
laser damage of the medium in the form of periodically spaced spherical rings (Sharma &
Rieckhoff, 1970; Strekalov, 2000). It is known that material strength sharply grows under
pulsed load as compared to the static load. Under the high-speed deformation (in the ns
range), strength of material becomes comparable to the theoretical limit-tens GPa (Kanel et
al., 2007). So a fast periodic change of the stress sign in the medium should decelerate the
development of material destruction in the form of rings. This is the reason why the laser
damage in the form of multiple rings usually was not observed in many experiments when
ns pulses of laser radiation were focused into the volume of transparent dielectrics. The
damage in the form of rings was not observed in our experiments as well. One of the few
observations of the multiple ring damage (Martinelli, 1966) relates to the case when glass
samples were exposed to focused free-running laser radiation. Anyway, according to
calculations (Sharma & Rieckhoff, 1970) oscillating profile of thermo-elastic stresses should
occur in the region of ns laser pulse focusing. The presented material provides experimental
data which confirm in the quality form the calculated (Sharma & Rieckhoff, 1970) picture of
thermo-elastic stresses distribution and elastic wave propagation across the medium.
7. Conclusion
The study of the interaction of powerful ns pulses of low-coherence radiation of the LiF:

F
2
+

color center laser (ccl) with optical materials (Yb:YAG, glass, et al.) was carried out. Efficient
SBS of low-coherence pump, accompanied by SRS and formation of hypersonic waves
reaching the intensity of shock waves were found. A physical model of excitation of SBS and
hypersonic waves at scattering of ultrashort pulses of low-coherence pump at stationary
inhomogeneities in optical materials is presented. It is shown that ns laser pulse, whose
duration is much higher than its inverse spectral width, causes SBS much more efficient
than a pulse of high coherence with the same duration and energy. Unlike SBS of a coherent
radiation caused by a pressure fluctuation, scattering of low-coherence pump may be
caused by any stationary inhomogeneities in a medium: cracks, dislocations, micro-
inclusions, or just by a plane back surface of a sample. An effective energy contribution of
light pulses into hypersonic waves on a small coherence length near the input surface of a
sample leads to their transformation into a periodic succession of high-pressure shock
waves, which results in structure changes of a crystal lattice (phase transition) in that region.
The appearance of structural changes in optical materials that are specific to the interaction
of powerful pulses of low coherence radiation with matter was found. The mechanisms of
structural changes based on the action of intense hypersonic waves were considered.


394
Nanosecond pulses of Yb lasing in the region 1.00-1.06 m with the spectral width up to 20
nm in Yb :YAG and 50 nm in the Yb: glass samples were observed. The divergence of the
broadband laser radiation (10
-3
-10
-4
rad) was one or two orders of magnitude smaller than
the diffraction limit respectively to the source of Yb radiation in a sample. The mechanism of
generation of broadband laser pulses of short duration and high directivity in the spatial
structure of thin layers with inversion produced in the region of the propagation of intense
hypersonic waves in the medium is discussed. The interpretation of experimental data is
based on a new concept of the spatial distribution of the electromagnetic field of a photon
not in the form of a traveling wave but with the field structures located in fixed positions
along the photon propagation direction. The new approach allows the existence of light
beams of an aperture d and with a sub-diffraction angle of divergence. Such beams
must consist of groups of phase-synchronized photons with a small phase difference
distanced by an interval . This synchronized group is no longer a traveling wave, its
angle of divergence is defined by the phase difference of photons in a group.
Spectral lines in 1.031.05 m region structured by 50200 m spots as well as lines with
inclinations were found at Yb lasing in a resonator. Structures of multiple spots in spectra
reflect the specific spatial configuration of the field of thermo-elastic stresses in the
unloading region of ccl pulse focusing after the end of the pump. Inclinations of spectral
lines reflect the pressure gradient from the center to the periphery of the region of ccl
focusing. Basing on inclinations of spectral lines the pressure in the region of shock
hypersonic wave propagation was estimated. Estimations show that for some of the studied
media the pressure values may exceed the phase transition threshold.
8. Acknowledgment
Authors thank O. Yaremchuk for the help in preparing this article for publication.
9. References
Alcock, R. & Emmony, D. (2002). Sensitivity of reflection transducers J. Appl. Phys., Vol. 92,
No.3 (August 2002), pp. 1630-1643, ISSN 0021-8979.
Basiev, T. et al. (1982). Solid-state tunable lasers based on color centers in ionic crystals
Bulletin of the Academy of Sciences of the USSR, ser. phys., Vol. 46, No.8, pp.145-154.
Basiev, T.T, Bykovsky, N.E, Konyushkin, V.A & Senatsky, Yu.V. (2004). Use of a LiF colour
centre laser for pumping an Yb:YAG active medium. Quantum. Electron. Vol.34,
No.12, pp. 1138-1142, ISSN 0368-7147.
Bor, Z. & Muller, A. (1986). Picosecond distributed feedback dye lasers IEEE J. Quantum
Electron., 22 (8), 1524-1533. ISSN: 0018-9197.
Bullough, R. & Gilman, J.J. (1966). Elastic Explosions in Solids Caused by Radiation. J. Appl.
Phys. Vol.37, pp.2283-2288 , ISSN 0021-8979.
Bykovsky, N. & Senatsky, Yu. (2008a). Broadband collimated generation in YAG:Yb crystal
and ytterbium glass under LiF:F2+color center laser pumping. Laser Phys. Lett.
Vol.5, Iss. 9. pp. 664670, ISSN 1612-2011.
Bykovsky, N.& Senatsky, Yu. (2008b). Spectra, temporal structure, and angular directivity of
laser radiation of a Yb:YAG crystal and ytterbium glass pumped by low-coherence
radiation from a F2+:LiF colour centre laser. Quantum Electron. Vol.38, No.9, pp.
813-822, ISSN 0368-7147.

395
Bykovsky, N.E. & Senatsky, Yu.V. (2010). Twisted lines and small-scale structures in
generation spectra of Yb-doped media excited by focused radiation of LiF:F 2+
color center laser. Laser Physics Vol.20, No.2, pp. 478-486, ISSN 1054-660X.
Bykovsky, N.E. (2005). Preprint FIAN No. 16. (http://ellphi.lebedev.ru/12/pdf16/pdf).
Bykovsky, N.E. (2006). Preprint FIAN No. 36. (http://ellphi.lebedev.ru/17/pdf36/pdf).
Conners, G. & Thompson, R. (1966). A Continuum Mechanical Model for Laser-Induced
Fracture in Transparent Media. J. Appl. Phys. Vol.37, pp.3434-3441, ISSN 0021-8979.
Dutoit, M. (1974). Microwave phonon attenuation in yttrium aluminum garnet and
gadolinium gallium garnet. J. Appl. Phys. Vol.45, pp.2836-2841, ISSN 0021-8979.
Galbraith, C. & Galbraith, J. (2011). Super-resolution microscopy at a glance Journal of Cell
Science Vol.124, 10, pp.1607-1611 ISSN: 0021-9533
Gordienko,V., Mikheev P. & Potemkin F. (2010). Generation of coherent terahertz phonons
by sharp focusing of a femtosecond laser beam in the bulk of crystalline insulators
in a regime of plasma formation.JETP Letters, Vol.92, No.8, pp.502-506, ISSN 0021-
3640.
Gorelik, V. (2007). Optics of globular photonic crystals. QuantumElectron, Vol.37(5), pp. 409-
432, ISSN 0368-7147.
Kanel, G., Fortov, V. & Razorenov, S. (2007). Shock waves in condensed-state physics. Phys.
Usp. Vol.50, pp.771-791, ISSN: 0038-5670.
Katarkevich, V., Kurstak V., Rubinov, A., & Efendiev T. (1996). Kinetics of the operation of a
distributed-feedback dye laser with nanosecond excitation. Quantum Electron.,Vol.
26, pp.1061- 1064, ISSN 0368-7147.
Koechner, W. (2006). Solid State Laser Engineering, 6th ed., Springer, Berlin.
Kogelnik, H., & Shank, C. (1971). Stimulated emission in a periodic structure. Appl. Phys.
Lett., Vol.18 (4), pp.152-155, ISSN 0003-6951.
Koldunov, M., Manenkov, A. & Pokotilo, I. (2002). Mechanical damage in transparent solids
caused by laser pulses of different durations. Quantum Electron., Vol.32. No.4,
pp.335340, ISSN 0368-7147.
Krupke W.F. (2000). Ytterbium solid-state lasers. The first decade IEEE J. Sel. Top. Quantum
Electron., Vol.6 (6), pp.1287 1296, ISSN1077-260X.
Mak, A & Soms, L. (1990). Nd:Glass Lasers, Nauka, Moscow.
Manenkov, A. & Prokhorov, A. (1986). Laser-induced damage in solids. Sov. Phys. Usp.,
Vol.148, pp.104122. ISSN: 0038-5670.
Martinelli, J. (1996). Laser-Induced Damage Thresholds for Various Glasses. J. Appl. Phys.,
Vol.37, pp.1939-1941, ISSN 0021-8979.
Merlin R. (1997). Generating coherent THz phonons with light pulses. Solid State Commun.,
Vol.102, pp.207-220, ISSN 0038-1098.
Nelson K.A., Miller R.J.D. & Fayer M.D., (1982). Optical generation of tunable ultrasonic
waves. J. Appl. Phys., Vol.53, pp.1144-1149, ISSN 0021-8979.
Polyakova A.L. (1968). Elastic Nonlinearity in Stimulated Mandel'shtam-Brillouin
Scattering. JETP Letters, Vol. 7, Iss. 2, pp. 57-59, ISSN: 0021-3640.
Polyakova, A. (1966). Nonlinear Effects in a Hypersonic Wave. JETP Letters, Vol.4, Iss. 4,
p.90-92, ISSN: 0021-3640.
Ready, J. F. (1971). Effects of High Power Laser Radiation, Academic, New York, London.


396
Ritus, A. (1982). Study of Mandelshtam-Brillouin light scattering in crystals and glasses
respecting to problems of quantum electronics and fiber optics. Trudy FIAN 137, pp.
3-80.
Robinson M.M., Yan Y.-X., Gamble E.B., Williams Jr., L.R., Meth J.S, & Nelson K.A. (1984).
Picosecond impulsive stimulated Brillouin scattering: Optical excitation of coherent
transverse acoustic waves and application to time-domain investigations of
structural phase transitions. Chem.Phys. Lett., Vol.112, pp.491-496, ISSN 0009-2614.
Sakakura, M., Terazima, M., Shimotsuma, Y., Miura, K. & Hirao, K, (2007) Observation of
pressure wave generated by focusing a femtosecond laser pulse inside a glass Opt.
Exp. 15, 5674-5686, ISSN 1094-4087.
Sharma B.S. & Rieckhoff, K.E. (1970). Laser-induced dielectric breakdown and mechanical
damage in silicate glasses. Can. J. Phys. Vol.48(10), 11781191, ISSN 0008-4204.
Starunov V.S. & Fabelinskii I.L. (1970) Stimulated Mandelshtam-Brillouin scattering and
stimulated entropy (temperature) scattering of light. Sov. Phys. Usp., Vol.13, pp.428
428. ISSN: 0038-5670.
Strekalov, V. (2000).Mechanical damage of transparent dielectrics by focused laser radiation
J. Tech. Phys. Lett. Vol.26, No. 24, pp.19-22. ISSN: 0320 - 0116 .
Stuart, B, Feit, M, Rubenchik, A., Shore, B. & Perry, M. (1995). Laser-Induced Damage in
Dielectrics with Nanosecond to Subpicosecond Pulses. Phys. Rev. Lett. Vol.74,
pp.22482251, ISSN 0031-9007.
Suguira, H., Ikeda, R., Kondo, K. & T. Yamadaya, T. (1997). Densified silica glass after shock
compression. J. Appl. Phys., Vol.81, pp.1651-1656, ISSN 0021-8979.
Tychinskii, V. (2008). Super-resolution and singularities in phase images Phys.Usp. Vol.178,
11, pp. 1205-1214. ISSN: 0038-5670.
Zhu, T., Maris, H., & Tauc, J. (1991). Attenuation of longitudinal-acoustic phonons in
amorphous SiO
2
at frequencies up to 440 GHz. Phys. Rev. B Vol.44, pp.42814289,
ISSN 1098-0121.Dictionary -
18
An Optimal Distribution of Actuators
in Active Beam Vibration
Some Aspects, Theoretical Considerations
Adam Braski
Rzeszow University of Technology
Poland
1. Introduction
The reduction of the effects of mechanical vibration fall into the of vibration isolation, design
for vibration or vibration control (de Silva, 2000). The vibration control is subdivided into two
group: passive control and active one. The core of the vibration control is to detect the level of
vibration in a system and to counteract the effects of the vibration, so it needs two devices.
Hence, the passive devices do not require external power for their operation. Hence, passive
control is relatively simple, reliable and economical. But it has limitations namely, the
control force depends entirely on the natural dynamics and it may not be adjust on line.
Furthermore, in a passive device, there is no supply of power from an external source. It
leads to the incomplete control, particularly in complex and high-order systems.
The shortcomings of passive control can be overcome using an active one. In this case, the
system response is directly sensed on line and on that basis, the specific control actions are
applied to any locations of the system. But the active control needs external power, namely
to apply control forces to vibrating system through actuators and to measure vibration
response using sensors.
Two different types of actuators can be applied (Shimon et al., 2005). The first, inertial
actuators, make up a piezoelectric material to vibrate large masses. Their vibrations are used
to counteract the vibrations of the structure (Jiang et al., 2000). The advantages and
disadvantages are enumerated in above reference.
The second type of actuators is a layer of smart or intelligent materials. The sensors also
belong to these materials; together they are wellknown as piezoelectric elements
(Tylikowski & Przybyowicz, 2004). It was shown that these elements can offer excellent
potential for an active vibration reduction of the structure vibrating with low frequencies
(Croker, 2007; Fuller at al, 1997; Hansen & Snyder, 1997; Kozie, 2006; Przybyowicz, 2002;
Wiciak, 2008). As a general, piezoelectric elements are glued to the host structure. It makes
the advantage, namely their incorporating into the structure is that the actuating mechanism
becomes part of the structure. Both sensors and actuators are relatively light, compared to
the structure, and can be made in arbitrary shape. The disadvantage is that they once
bonded and they cannnot be used again. In recent years the measure of the vibration with
the sensors are replaced by touch less measures. For this reason, hereafter in research the
sensors are omitted and only second type actuators will be considered. Nowadays actuators


398
are used to very original structures for example to the satellite boom (Moshrefi-Torbati et al.,
2006) or to sun plate (Qiu et al., 2007).
To make the reduction more effective, many problems should be solved.
dynamic effects (mass loading and stiffness) of the actuators on the structure vibration
(Charette et al., 1998; Gosiewski & Koszewnik, 2007; Hernandes et al., 2000; Q. Wang &
C. Wang, 2001)
dynamic effects of the glue (between actuators and structure) on the structure vibration
(Pietrzakowski, 2004; Sheu et al., 2008).
actuators geometric-technical features (Frecker, 2003; Hong et al., 2007; Wang, 2007),
orientation of the actuators on the structure (Bruant et al., 2010; Ip & Tse, 2001; Qiu et
al., 2007),
appropriate actuators distribution on the structure (Bruant et al., 2010),
others, but they play a minor part.
Reviewing the literature, it appears that the actuators distribution play a major part. Now, a
question arises about an optimal distribution of actuators. In the recent year, a great number
of papers has been published on this subject. It is obvious that there are a lot of optimization
techniques; an excellent survey is given in (Bruant et al., 2010). Two main approaches are
distinguished to this problem.
First of them is the coupling of the optimization of actuators/sensors locations and controller
parameters. In this case the following criterions are taken into account for the optimization:
quadratic cost function of the measure error and the control energy (Bruant et al., 2001),
maximization of dissipation energy during the control (Yang, 2005),
spatial
2
H norm of the closed-loop transfer matrix from the disturbance to the
distributed controlled output (Liu et al., 2006),
simultaneous simple H
controller (Guney & Eskinat, 2007).

As can be seen, the optimization criterions are dependent on the choice of controllers.
Therefore, the optimal location obtained using one controller may not be a suitable choice
for another one.
At the latter approach, the optimal location is obtained independently of the controller
definition. In this case, the following criterions are used:
maximization controllability/observability criterion using the gramian matrices (Bruant
& Proslier, 2005; Jha & Inman, 2003),
modal controllability index based on singular value analysis of the control vector
(Dhuri, & Seshu, 2006),
maximization of the control forces transmitted by the actuators to the structure (Q.
Wang & C. Wang, 2001),
using the
2
H norm (Halim & Reza Moheimani, 2003; Qiu et al., 2007).
In the quoted references, it was not provided the actuators distribution in explicite; only the
general rules (criterions) were formulated. However, this problem was partially solved; it
was proved in (Braski & Szela, 2007; Braski & Szela, 2008; Szela, 2009; Braski & Szela
2010; Braski & Lipiski, 2011) that the most effective actuators distribution was on the
structure sub-domains with the largest curvatures; such distribution was called quasi-
optimal one. As the research object, a right-angled triangle plate with clamped-free-free
boundary conditions was taken into account. The quasi-optimal distribution was deduced
based on the heuristic reasons and the conclusions were confirmed only numerically.
Furthermore, the problem was solved merely for the separate modes.
An Optimal Distribution of Actuators in
Active Beam Vibration Some Aspects, Theoretical Considerations

399
Basing on the quasi-optimal distribution of the actuators, the protection beam vibration is
achieved (Braski et al., 2010; Braski & Lipiski, 2011). In this case always the separate
modes were considered. The problem was solved based on heuristic reasons and was
confirmed analytically. In the latest own research, the results presented in (Braski et al.,
2010) were substantiated analytically (Braski & Lipiski, 2011).
In this chapter, the above attitude to the optimal actuators distribution is continued and
extended. First at all, the optimal problem is formulated. For this purpose, the optimization
criterion is defined. It is assumed that a measure of the vibration reduction is a reduction
coefficient (Szela, 2009; Braski & Szela 2010) and here it becomes the objective function.
This attitude is quite similar to the maximization of the control forces transmitted by the
actuators to the structure (Q. Wang & C. Wang, 2001).
Dynamics effects of the glue and actuators are also considered. Furthermore, the solution of
active vibrations reduction is derived for general solution, not only for separate modes.
Since analytical solution was attained with separation of variables method, first of all the
modes of the problem are derived. Next, the orthogonality condition of the modes is derived
too.
The simple supported beam is chosen as the research object. The study of beams is very
important in a variety of practical cases, noteworthy, the vibration analysis of structures like
bridges, tall buildings, and so on. Loosing a bit on generality, it is considerably easier to
realize the aim of the paper. It is assumed that the beam is excited with evenly spread and
harmonic force. The material inner damping coefficients of all elements of the research
system are taken into account. It seems that all main factors having the influence on the
beam vibration were considered.
To solve the problem analytically, a few simplifications are made. Namely, the energy
provided to the system is in the form of voltage applied to the surface of the actuators.
Assuming that the charge is homogenously distributed, as a result of piezoelectric effect, the
actuators interact with the beam with moments for couple of forces homogenously
distributed along the actuators edges. Next, these moments are replaced with the couple of
forces and finally, they are counteracted the vibrations.
All problems were considered only theoretically; no calculations are run. It seems that
presented considerations will be the base to many numerical simulations and experiments.
To the authors knowledge, the theoretical description of the optimal actuators distribution
on even simple structure like the beam, up to now have not been brought up.
2. Active beam vibration reduction with additional elements
In this problem, the additional elements make the concentrated masses and actuators and all
constitute the mechanical set beam-actuators-masses. Adding actuators (and the glue at the
same time) is the technical necessity but they introduce to the mechanical set the additional
dynamics effects namely, local stiffness and concentrated masses. As far as concentrated
masses are concerned, adding them is substantiated as follows. The proposed optimal
distribution of the actuators needs asymmetrical beam vibrations and these ones may be
ensured by at least one concentrated mass.
2.1 Uniform beam vibration with damping
There are four theories (models) for the transversely vibrating uniform beam (Han et al.,
1999): Euler-Bernoulli, Rayleigh, shear and Timoshenko. The first of them, called the


400
classical beam theory, is applied here. It is simple and provides reasonable results for
formulated problem.

u
x
f(x)
0

Fig. 1. The geometry of the simple supported beam
Let be the beam as depicted in Fig. 1. The Bernoulli-Euler equation governs transverse
vibration (or bending or lateral vibration) of the beam has a following standard form
(Kaliski, 1986; Pietrzakowski, 2004),
( )
4 4 2
t t
EJ D u EJ D D u SD u f + + = (1)
where u u(x, t) = beam deflection at the point x and the time t , f f(x,t) = load force,
4 4 4
D (.) (.) / x = ,
t
D (.) (.) / t = ; hereafter the rest symbols are jointly explained.
To solve Eq. (1) explicitly, four boundary conditions, at the ends of the beam, are needed. In
general, boundary conditions represent displacement, slope, moment and shear
respectively. Here, it is assumed that the beam is simple supported, then both displacement
and the bending moment equal zero
u(0, t) 0, =
2
D u(0, t) 0 = (2)
u( , t) 0, =
2
D u( , t) 0 = (3)
To solve over determined problem, one needs to know initial conditions. But here, the
harmonic steady state plays a major part, so that the initial conditions are omitted.
2.2 Beam vibration with concentrated masses
To solve the intended problem, Eq. (1) must be rounded out. First of all, to obtain
asymmetric modes and consistently asymmetric general vibration, a few concentrated
masses are added to the beam (Low & Naguleswaran, 1998; Majkut, 2010; Naguleswaran,
1999). They are marked by { }
r
m , and their distribution is described with set of coordinates
{ }
r
x , see Fig. 2, hence

r r 1 1 2 2 r r
r
m (x x ) m (x x ) m (x x ) ... m (x x ) ... = + + + +
(4)
where
r
r 1, 2,..., n = , (.) Diracs delta function.

0

x
x
2
x
r
x
1
m
2
m
r
m
1
... ...

Fig. 2. Distribution of the concentrated masses

401
Furthermore, the dynamic effects of the actuators and glue on the beam vibration are
introduced. The location and length of separate actuators, and the glue layers
simultaneously, are denoted commonly with coordinates { }
s
x and { }
s
respectively and
they are arranged as depicted in Fig. 3.

0
x
x x
2
2
x
s
s ... ...
1
1

Fig. 3. Distribution of actuators and glue layers
For simplicity, let { } P E, J, h, , S, = means the physical and geometrical parameters of the
beam, actuators and glue, i.e. {Young's modulus, surface moment of inertia, thickness, mass
density, surface of the rectangular cross-section, inner damping factor} respectively.
Furthermore all parameters are supplemented with following index
{ } { } b, a,g [b]eam,[a]ctuator,[g]lue = = , for example S bh

= means the surface of the
rectangular cross-section, b beam / glue layer width. Moments of inertia are calculated
relatively of y -axis, see Fig. 4, where the neutral axis displacement d is neglected , hence
3
b b
J (bh ) /12 = ,
3 2
g g g b g
J (bh ) /12 S (h 2 h 2) = + + ,
3 2
a a a b g a
J (bh ) /12 S (h 2 h h 2) = + + + .

y
b
h
a
h
b d
h
g

Fig. 4. Cross-sections of the set beamactuatorglue
The parameters of the set beam-actuators-glue may be written as

0
b s 1s 2s b s
s s
P P P H(x x ) P P H = + = +

(5)
where
s
s n 1, 2,..., = ,
s a g
P P P = + ,
0
1s 2s 1s 2s
H H(x x ) H(x x ) H(x x ) = = ,
1s
H(x x )
Heaviside step function in point
1s
x and so on, { } { }
1s 2s s s s s
x , x x 2, x 2 = + .
For
s
n aktuators (
s
n glue layers) and
r
n concentrated masses, Eq. (1) takes the form
( ) ( ) ( ) ( ) + + + + + +

u D H S S u D D H J E J E u D H J E J E
2
t
s
0
s s b b t
4
s
0
s s s b b b
4
s
0
s s b b

2
r r t
r
m (x x )D u f + =
(6)
The Eq. (6) may be written down quite similar like Eq. (1), namely
( ) ( )
4 4 2
t r t
EJ D u EJ D D u S D u f + + + = (7)


402
where hereafter
0
b b s s
s
EJ E J E J H , = +

0
b b b s s s
s
EJ E J E J H = +

,

0
b b s s
s
S S S H , = +
r r r
r
m (x x ) =
(8)
On the ground of the EJ , S and
r
form, Eq. (7) can not be understood in a classical
manner. To solve it, some methods may be applied. One of them is presented in (Ercoli &
Laura, 1987; Kasprzyk & Wiciak, 2007; Majkut, 2010); another attitude may be found in
(C.N. Bapat & C. Bapat, 1987) and it is applied here.

0
x
x
s
x
r
x
1s
m
r
x
2s
s

Fig. 5. Geometry of the set beamone actuatorone mass
At the latter attitude, the beam is divided into some uniform elements. The division may not
be coincidental. To clearly explain this problem, for simplicity consider a set beam-one
actuator (and glue)-one concentrated mass, Fig. 5. The division is imposed out of the change
of physical properties namely, properties of the actuators (and glue) and concentrated
masses. So, the beam is divided into
j
j 1, 2,..., n 4 = = elements. All elements may be
considered separately and the solution to Eq. (7) can be expressed as

j
j
u(x, t) u (x, t) =

(9)
where
j
u (x, t) is the solution on j -element and it is fulfilled the following equation

( ) ( )
4 4 2
j j j j j j t j j j r t j j
E J D u E J D Du S D u f + + + = (10)
To find
j
u (x, t) with the separation of variables method, the eigenvalues and eigenfunctions
for each element are needed.
2.3 Eigenvalues and eigenfunctions problem
In this problem it is assumed that ( , ) 0
j
f x t = and
j
0 = , hence based on Eq. (10) one obtains

4 2
j j j j j t j
E J D u S D u 0 + = (11)
where
j j
E J and
j j
S may be different on the separate elements, but here, as depicted in
Fig. 5, is

1 1 3 3 4 4 b b
E J E J E J E J , = = =
2 2 b b a a g g
E J E J E J E J = + + (12)

1 1 3 3 4 4 b b
S S S S , = = =
2 2 b b a a g g
S S S S = + + (13)
The boundary conditions for the j element consist of boundary conditions of the problem
and coupling conditions between neighboring elements. The concentrated mass
r
m is

403
considered in coupling conditions between third and forth elements and therefore it is
omitted in Eq. (11).
Let the solution be represented by a product of spatial and temporal functions

j j
u (x, t) X (x) T(t) = (14)
Substituting (14) into (11) gives

4 2
j j j j j j t
E J D X T S X D T 0 + = (15)
or

4
2
j j j 2 t
j j j
E J D X
D T
S X T

= = (16)
hence

4 4
j j j
D X X 0 = (17)

2 2
t
D T T 0 + = (18)
where the dispersion relationship is given by

2
j j 4 2
j
j j j
S
E J
= = (19)
The Eq. (17) is very important and the solution to it is

j j 1 j j 2 j j 3 j j 4 j
X (x) A K ( x) B K ( x) C K ( x) D K ( x) = + + + (20)
where Krylov functions are defined as, (Kaliski, 1986),
( )
1
K (z) ch(z) cos(z) 2 , = + ( )
2
K (z) sh(z) sin(z) 2 = ,
( )
3
K (z) ch(z) cos(z) 2 , = ( )
4
K (z) sh(z) sin(z) 2 = + (21)

0
x
u
2
u
4
u
3
u
1
e
2
e
4
e
3
e
1
0 0 0 0
m
r

Fig. 6. Geometry of the set beam-one actuator-one mass in local coordinates
The boundary conditions in local coordinates,
j
x [ 0, e ] , to the separate j -element have the
form, Fig. 6,
boundary conditions at the left end of the 1stelement

1
X (0) 0, =
2
1
D X (0) 0 = (22)


404
coupling conditions between 1st and 2ndelements
1 1 1 2 2
X ( e ) X ( 0), =
1 1 1 2 2
DX ( e ) DX ( 0) = ,

2 2
1 1 1 1 1 2 2 2 2
E J D X ( e ) E J D X ( 0), =
3 3
1 1 1 1 1 2 2 2 2
E J D X ( e ) E J D X ( 0) = (23)
coupling conditions between 2nd and 3rdelements
2 2 2 3 3
X ( e ) X ( 0), =
2 2 2 3 3
DX ( e ) DX ( 0) = ,

2 2
2 2 2 2 2 3 3 3 3
E J D X ( e ) E J D X ( 0), =
3 3
2 2 2 2 2 3 3 3 3
E J D X ( e ) E J D X ( 0) = (24)
coupling conditions between 3rd and 4thelements
3 3 3 4 4
X ( e ) X ( 0), =
3 3 3 4 4
DX ( e ) DX ( 0), =
2 2
3 3 3 3 3 4 4 4 4
E J D X ( e ) E J D X ( 0) =
and
3 2 3
3 3 3 3 3 r 3 3 3 4 4 4 4
E J D X ( e ) m X ( e ) E J D X ( 0) + =
or

3 2 3
3 3 3 3 3 r 4 3 4 4 4 4
E J D X ( e ) m X ( 0) E J D X ( 0) = + (25)
boundary conditions at the right end of the 4thelement
=
4 4 4
( ) 0, X e =
2
4 4 4
( ) 0 D X e (26)
Since
1 2 3 4
then, to calculate them, the Eq. (19) must be used. It is convenient to
express { }
2 3 4
, , as a function
1
, hence

4 4 4 4 2
1 1 2 2 3 3 4 4
= = = = (27)
or
( )
4 4
2 1 1 2
= , ( )
4 4
3 1 1 3
= , ( )
4 4
4 1 1 4
= (28)
Substituting Eq. (20) into boundary conditions (22) it appears that
1
A 0 = ,
1
C 0 = . In the
same way, the rest of conditions given by Eqs. (23) (26) lead to the set of algebraic
equations and it may be written in the matrix form
= A x 0 (29)
The matrix A is too large, to presented it in explicit form. Hence, its elements fall into
blocks so that the matrix A can be written as

(
(

(
=
(

(
(

1 2
1 2
1 2
1
A A 0 0
0 B B 0
A
0 0 C C
0 0 0 D
(30)

405
In the current boundary problem, the separate blocks take the form

2 4
1 3 1 1 2
2 2 2
1 1 1 4 1 1 1 2 2 2 2
3 3 3
1 1 1 1 1 1 1 3 2 2 2
1 2 3 4
2 2 2 3 2 4 2 1
2 2 2 2
2 2 2 3 2 2 2 4 2 2 2 1 2 2 2 2
3
2 2
K K 1 0 0 0
K K 0 0 0
E J K E J K 0 0 E J 0
E J K E J K 0 E J 0 0
K K K K
K K K K
E J K E J K E J K E J K
E

(
=
(

(

1 2
1
A A
0 B
0
3 3 3
2 4 2 2 2 1 2 2 2 2 2 2 2 3
J K E J K E J K E J K
(
(
(
(
(
(
(
(
(
(
(
(
(

(31)

3
2
3 3 3
3
3 3 3
1 2 3 4
3 2 3 3 3 4 3 1
2 2 2 2
3 3 3 3 3 3 3 4 3 3 3 1 3 3 3 2
3 3 3 3
3 3 3 4 3 3 3 1 3 3 3 2 3 3
1 0 0 0
0 0 0
0 0 E J 0
0 E J 0 0
K K K K
K K K K
E J K E J K E J K E J K
E J K E J K E J K E J
(
=
(

(

1
1 2
1
B
B B
0 C
0
3 3
K
(
(
(
(
(
(
(
(
(
(
(
(
(

(32)

4
2
4 4 4
2 3
r 4 4 4
1 2 3 4
3 4 1 2
1 0 0 0
0 0 0
0 0 E J 0
m E J 0 0
K K K K
K K K K

(
(
(
(
(
=
(
(

(

(
(
(

(

1
1 2
1
C
C C
0 D
0
(33)
where the symbols in matrices are given by
{ } { }
1 2 3 4
K K , K , K , K
= ,

1 1
K K ( e ),

=
2 2
K K ( e ),

=
3 3
K K ( e ),

=
4 4
K K ( e )

= (34)
The unknowns are collected in column matrix
[ ]
T
1 1 2 2 2 2 3 3 3 3 4 4 4 4
B , D ,A , B , C , D ,A , B , C , D , A , B , C , D = x (35)
To solve of the homogeneous matrix equation (29), one assumes that
1
det ( ) 0 = A and it
gives the set { }
1
, 1, 2,..., n = . Based on Eq. (28) one can calculate { }
2 3 4
, ,

and
finally, based on Eq. (19), the frequency { }
of the system beam-actuator-mass.


406
Now, the unknowns put down in column matrix, Eq. (35), should be determined. Let the
main matrix elements A be written as two suffix quantities A
, where and label the

rows and columns respectively. Let M
be the minor of the A
element. The general

solution to Eq. (29) is

1 2 3
1 1 2 1 2 3
B : D : A : ... ( 1) M : ( 1) M : ( 1) M : ...

+ + +
= (36)
Substituting
{ }
j
and unknowns x to Eq. (20), the eigenfunctions ( modes) assigned
to the j element are obtained. The solution to Eq. (10) is given by

j j j j
j j j
X(x) X (x) X ( x) X (x) X (x)

= = = =

(37)
where
j j
(...) (...)

=

and the separate modes are equal

j j 1 j j 2 j j 3 j j 4 j
X (x) A K ( x) B K ( x) C K ( x) D K ( x)

= + + + (38)
2.4 Orthogonality of modes
Orthogonality condition of the uniform beam modes may be found in (Kaliski, 1986; de
Silva, 2000). First of all, based on twice integration by parts, one has

( )
4 3 2 2 2
0 0 0
X (x)D X (x)dx X (x)D X (x) DX (x)D X (x) D X (x)D X (x)dx

= +
} }

(39)
The separate modes X (x)
, X (x)
, fulfill the following modal equations

4 2
EJ D X (x) SX (x)

= (40)

4 2
EJ D X (x) SX (x)

= (41)
Multiplying above equations by X (x)
and X (x)
respectively, integrate both in range o

integration x [0, ] , use Eq. (39), subtract the second result from the first one, one obtains
(for simplicity an argument (x) is omitted)

( ) ( ) ( )
2 2 3 2 3 2
0
0
S X X dx EJ X D X DX D X X D X DX D X

(
=

}

(42)
For standard boundary conditions, the right-hand-side equals zero.
The procedure outlined above can be used to the problem presented in Fig. 6, but Eq. (39)
must be applied to the separate j element, namely

( )
j j
j
e e
4 3 2
j j j j j j
0 0
e
2 2
j j
0
X (x)D X (x)dx X (x)D X (x) DX (x)D X (x)
D X (x)D X (x)dx

= +
+
}
}
(43)
Considering both boundary conditions of the problem and coupling conditions between
neighboring elements, Eqs. (22)(26), instead of Eq. (42) one has

407
( )
(
1 2 3 4
e e e e
2 2
1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4
0 0 0 0
S X X dx S X X dx S X X dx S X X dx

+ + + +
} } } }

)
r 4 4 4 4
m X (0)X (0) E J

+ =
( ) ( )
3 2 3 2
4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4
X (e )D X (e ) DX (e )D X (e ) X (e )D X (e ) DX (e )D X (e )

(

(44)
Because of Eq. (26), the right-hand-side is zero, hence
( )
(
1 2 3 4
e e e e
2 2
1 1 1 1 2 2 2 2 3 3 3 3 4 4 4 4
0 0 0 0
S X X dx S X X dx S X X dx S X X dx

+ + + +
} } } }

)
r 4 4
m X (0)X (0) 0

+ = (45)

0 j = 2 j = 3
...
... j = 1
m
1
m
3
m
2
...

Fig. 7. Geometry of set with
j
n elements
The orthogonality condition, Eq. (45), may be generalized in a simple way. Let the system
beam-actuators-masses be divided into
j
n elements as depicted in Fig. 7. In this case one has

( )
( )
j j j j j
2 2
j j j j j j j n 1 n n n n
j
j
S X X dx m X (0)X (0) m X (e )X (e ) 0

+
(
+ + =
(

}
(46)
Since the term
2 2

is canceled for = , the general orthogonality condition is given by

( )
j j j j j
j j j j j j j n 1 n n n n
2 j
j
0,
S X X dx m X (0)X (0) m X (e )X (e )
,

+ + =

=
}
(47)
The Eq. (47) in particular case is used in deriving the solution to the forced vibration
problem.
2.5 Forced vibrations with damping
A point departure for further consideration is Eq. (7); for j element one has

( )
4 4 2
j j j j j j t j j j t j j
E J D u E J D D u S D u f + + = (48)
The solution to Eq. (48) is forced vibrations with damping. Let be the load force in the form

j j f
f (x, t) f (x)exp(i t) = (49)
where
1 2
i ( 1) = ,
f
excited frequency.
Applying separation of variables method, the solution to Eq. (48) is assumed as

j jf f
u (x, t) X (x)exp(i t) = (50)
Substituting Eqs. (49) and (50) to Eq. (48) one obtains


408

4 2
j j f jf f jf j
j j
1
(1 i )D X (x) X (x) f (x)
S

+ = (51)
The solution of the above equation is given by

jf j j
X (x) C X (x)

=

(52)
where
j
C

constants,
j
X (x)
Eq. (38).
After some calculation, the constants
j
C

are expressed by

j j j j 2 2
j j j
1 1 1
C I C I
S

= = (53)
where

j 2 2
j j j
1 1 1
C
S
= ,
2 2 2
j j f f
1 1
(1 i )

=
+
,
j j j
j
I f (x) X (x)dx

=
}
(54)
In the end, the problem of the forced j element beam vibration with damping, excited with
the force
j
f (x) is solved; in the harmonic steady state it is given by

f jf j j
j j
X (x) X (x) C X (x)

= =

(55)
In current problem, two form of the forces have the practical meaning namely, the force
with constant amplitude
j0 0
f (x) f = and the force acting at discrete point
ja i
f (x ) . The former
may be interpreted as the spread excitation forced, for example with plane acoustic wave,
but the latter is the control force due to actuators, henceforth

j 0 ja i
f (x) f f (x ) = + (56)
2.6 Interaction between beam and actuators
It is assumed that the actuator is perfectly bonded to the beam surface. Exciting actuator, the
interaction between actuator and the beam is appeared. The interaction process is explained
in (Hansen & Snyder, 1997; Fuller at al, 1997) in detail and references cited therein.
Assuming the spatially uniform actuator, it provides boundary induction solely in terms of
the external line moment distributed along its edges (Burke & Hubbard, 1991; Sullivan et al.,
1996). So, the bending moment in y direction is given by the formula (Hansen & Snyder
1997), Fig. 8,

( )
2 2
x 0 1 2
M M x x x x

= < > < > (57)
where (.) .
1
= > <

and (.) D .
2
= > <

doublet function,
0
M line moment amplitude

31
0 a
a
d
M C V
h
= (58)

409
where
a
C constant depending on geometry and mechanical properties of the actuator and
plate,
31
d piezoelectric material strain constants, V voltage in the direction of
polarization.

M
x
M
x
x
x
1
x
2
a

Fig. 8. External line moments of the actuator
The problem is to determine of the
a
C , because it depends on the analysis method of the
mutual interaction between beam-actuator (Hansen & Snyder 1997; Pietrzakowski, 2004).
Let the static force coupling model is taken into account. If relatively thin actuator compared
with beam thickness is assumed (so uniform normal stress distribution is accepted) and
furthermore by ignoring the neutral axis displacement d , see Fig. 4, the constant
a
C is come
down to the form

( )
( )
b b a a b a
a
b b a a
E h E h h h
C
2 E h E h
+
=
+
(59)
Since the beam vibration equation is the forces equation then to consider the action of
actuator with the beam, moments
x
M are replaced with two couples of forces, Fig. 9,

x a a
M f 2 = (60)

x
f
a
f
a
f
a
f
a
/2
a
a

Fig. 9. External pair of forces of the actuator
Next, the separate forces are considered in Eq. (56).
2.7 Beam vibration reduction through actuators
For the problem presented in Fig. 5, the total load of the beam, described by Eq. (56), is
given by

( )
j 0 js 1s js s js 2s
f (x) f f (x x ) 2f (x ) f (x x ) = + + + (61)
where the symbol
ja
f is replace by
js
f in order to express, in the future, the interaction sum
of actuators and the glue on the beam.
An expression in brackets is the sum of interacting forces actuator-beam. Hence, the integral
j
I

, Eq. (54), for
j
f (x) expressed by above equation is given by


410
( )
j j j 0 j js 1s s 2s j
j j j
I f (x) X (x)dx f X (x)dx f (x x ) 2 (x ) (x x ) X dx

= = + + + =
} } }

0 j js j 1s j s j 2s
j
f X (x)dx f X (x ) 2X (x ) X (x )

( = + +
}
(62)
The expression in square bracket constitutes the second-order central finite difference. Since
the distance between nodes
s
is constant, then the difference can be transformed into

2
j 1s j s j 2s j s 2
s
1
X (x ) 2X (x ) X (x ) D X (x )

( + =

(63)
where

2
j s j s
D X (x ) (x )

= (64)
The
j s
(x )
is the curvature of the mode

j
X (x)
at the point
s
x x = (Braski & Szela, 2007;
Braski & Szela, 2008). The sign of the
j s
(x )
is contractual namely, if the bending of the

beam is directed upwards, the sign is positive and vice versa. Substituting Eq. (64) into
Eq. (62), one obtains

2
j 0 j js s j s j 0 j s
j
I f X (x)dx f (x ) I I

= + = +
}
(65)
Next, substituting Eq. (65) into Eq. (52) through Eq. (53), the reduction vibration is obtained

( )
jf j j j j j 0 j s j jf j
X (x) C I X (x) C I I X (x) A X (x)

= = + =

(66)
Note, that the amplitude
jf
A

is the direct quantity which is liable to the reduction, in
explicit form is

( )
jf j j j j 0 j s
A C I C I I

= = + (67)
At the same time, together with the vibration reduction amplitude
jf
A

, the curvature is
subjected to the reduction and based on Eq. (66) is

2
jf jf jf jf j
D X A

= = =

(68)
Furthermore, the reduction of the
jf
A

leads to the reduction of the shear force
jf
Q (x) and
bending moment
jf
M (x) (Braski et al., 2010; Kaliski, 1986; Kozie, 2006),

jf j j jf j j jf j j jf j
Q (x) E J D (x) E J D (x) E J A D (x)

= = =

(69)

jf j j jf j j jf j j jf j
M (x) E J (x) E J (x) E J A (x)

= = =

(70)
As can be seen, the vibration reduction undergo on the following amplitudes: of the beam
vibration
jf
X (x) Eq. (66), of the shear force
jf
Q (x) Eq. (69) and of the bending moment
jf
M (x) Eq. (70). Hereafter, the notion e.g. shear force reduction is used instead of the
reduction of the amplitude of the shear force, and so on.

411
The Eqs (66), (69) and (70) and may be written commonly

jf jf j jf j
(x) (x) C A (x)

= =

(71)
where

jf jf jf jf
{X , Q , M }, =
j j j j
{X , D , },

=
j j j j j
C {1, E J , E J } = (72)
Omitting for simplicity the index "f " , for entire system beam-actuators one has

j j
j j
(x) (x) (x)
= =

(73)
or

j j j
j
(x) C A (x) CA (x)

= =
(74)
where

( ) ( )
2
0 s s s 0
s
A C I C I f (x ) C I I

= = + = +
(75)
(x) { X(x), D (x), (x)} = (76)
It is appeared from Eq. (75) that the active vibration reduction depends on the following
parameters:

f
excited frequency, it is contained in C
,

s
x distribution of the actuators on the beam,

s
(x ) value of the beam curvature at the point of the actuators distribution,

s
f interacting forces between beam-actuators or more generally mechanical
properties of the actuators,

s
actuators lengths or more generally geometrical properties of the actuators,

s
n number of actuators.
As mentioned above, the optimal actuators distribution described with { }
s
x has an
important meaning and finding of the { }
s
x is the aim of the chapter.
3. Optimal actuators distribution problem
Before the optimization problem will be formulated, any coefficients of the vibration
reduction should be defined.
3.1 Reduction and effectiveness coefficients
Let be the difference between any quantities of the beam vibration

R
(x) (x) (x) = (77)
where (x) ,
R
(x) quantities calculated without and with actuators respectively; (x) ,
R
(x) are given together by Eq. (74), where

0
A C I
= (78)


412
and
( )
R 0
A C I I
= + (79)
The difference (x) is interpreted as the quantity of the vibration reduction and it is the
first measure of this reduction namely, the quantity reduction coefficient.
The second measure of the vibration reduction is defined as

R
(x) (x) (x)
R (x)
(x) (x)

= =

(80)
It is called as the reduction coefficient and it may be expressed in per cent. Note, that if the
reduction coefficient equals one, the vibration reduction is total,
R
(x) 0 = .
An effectiveness of the vibration reduction is defined as a quotient of some vibration
reduction measure by an amount of the energy W provided to the system in order to excite
actuators. Hence, thirst measure of the vibration reduction may be defined by so called the
effectiveness coefficient
E (x) R (x) W

= (81)
The energy W provided to the system is translated into couples of forces, Fig. 9. Therefore,
the energy W may be replaced by forces
R s
s
f 4f =
, hence

R
E (x) R (x) f

= (82)
The Eqs. (77) (82) define the appropriate factors of the vibration reduction at the point x .
In many cases, it is convenient to calculate mean values of these coefficients at whole beam
domain or at the beam sub-domains. First of them is the mean quantity reduction coefficient
and it is defined by the formula
( )
m i R i
i
i
1
(x ) (x )
n
=

i
i 1, 2,..., n = (83)
Consequently, the mean reduction coefficient and the mean effectiveness coefficient are
defined respectively

m m m
R
= ,
m i i
i
(x ) n =
(84)

m m R
E R f

= (85)
The coefficients defined above may constitute the base to formulate the optimization
problem; hereafter the R (x)
is chosen.
3.2 Formulation of the optimization problem
In this chapter, one formulates the following problem: find the optimal actuators
distribution { }
s
x which maximize of the reduction coefficient R (x)
; hence R (x)
is
assumed as an objective function. In this case the maximal value of R (x)
equals one and it


413
means p-reduction; such instance is considered in (Braski et al., 2010; Braski & Lipiski,
2011) and it seems that it is possible only in for separate mode.
Let the energy provided to the actuators be constant and hence, the
s
f is always constant.
Now, for clarity of the disquisition, rewrite the effectiveness coefficient in explicit form

R R
(x) (x) CA (x) CA (x)
R (x)
(x) CA (x)

= =

(86)
Working out on above assumption, the R (x)
will be maximal, if
R
(x) is minimal. Hence,
the optimal condition
;max
R (x) R (x)

= leads to the next condition
R R;min
(x) (x) . Note,
that the
R
(x) depends on the reduction amplitude
R
A . So, the
R R;min
(x) (x) , if the
amplitude
R
A is minimal and instead of the above condition, it leads to

R R;min
A A = (87)
3.3 Heuristic analysis of the optimization problem
Note, that the amplitude
R
A comprises the factor C 0
, but it is constant and this is the

factor
0 R
I I I
+ = which is changed. In practice, instead of the condition (87), the following

condition of the reduction must be fulfilled

2
R 0 0 s s s min
s
I I I I f (x ) I
= + = + =
(88)
For future considerations the sign of
R
I is very important. The vibrations are reduced with
actuators, if the I
is positive, but must be fulfilled the following condition:

R 0
I I I 0
= + ;
R
I 0 = assures the total reduction. If this condition is not fulfilled, the actuators excite
vibrations and thereby they are not accomplished owns role. Note, that the sign of
0
I is
always negative, see Eq. (65). Then, the sign of I
must be positive and it depends on the

signs both forces
s
f and curvatures
s
(x ) .
From physical point of view, the sign of (x) is changed and as established above; it is
positive, if the bending of the beam is directed upwards. Then, for many actuators one has

2 2 2 s 1
1 1 1 2 2 2 s s s
s
I f ( ) (x ) f ( ) (x ) ... f ( 1) (x ) 0
+
= + + + = >
(89)
To obtain the positive sign of
I , the signs of
s
f should alternates; this problem clearly
expressed in the following way

2 2 s 1 2 s 1
1 1 1 2 2 2 s s s
s
I ( )f ( ) (x ) ( )f ( ) (x ) ... ( 1) f ( 1) (x ) 0
+ +
= + + + + = >
(90)
First at all, it is possible if the signs of
s
(x ) and
s
f are the same, namely positive or
negative, and they take their extremes. To fulfill this requirement, the actuators should be
specially distributed on the beam. An idea of description of the sign of
s
(x ) is advance
determined. The value of
s
(x ) is determined by means of the distribution of the actuators;
they are bended at { }
s
x . Hence, the distribution has a great significance; this problem was
solved in (Braski & Szela, 2007; Braski & Szela, 2008; Braski et al., 2010; Braski & Szela
2010; Szela, 2009). Interpreting Eq. (88) through Eq. (90) it is appear that the actuators ought


414
to be bonded on the beam sub-domains in which the curvatures reach their extremum and
consequently the highest and lowest values respectively, see Fig. 10. This is so called quasi-
optimal actuators distribution and it is described with
Q s
x x points, their number is
Q s
n n .

x
f
s
f
s
f
s
f
s
x
s

Fig. 10. Optimal distribution of the actuators
As far as signs and values of { }
s
f are concerned, it was assumed that the added energy
exciting actuators is constant. So, the values { }
s
f of the separate actuators are known and
always are constant, while the sign
s
f springs from the physical interpretation of the
interaction beam-actuator. As can be seen in Fig. 10, the forces
s
2f are placed at the point of
local extreme, namely at the
s
x , with the opposite direction to the bending of the beam
X(x) . At the same time, the forces
s
f on the actuator edges are in the direction of the beam
bending and let assume that this sign of
s
f is positive. Another way, the vectors
s
f and the
beam bending X(x) are in the same direction. In such sign convection, both
s
f and
s
(x ) in
Eq. (90) have the same signs and all terms are positive. Furthermore, the actuators
distribution described with
Q s
x x ensures the maximum of the reduction coefficient.
The heuristic analysis described above was substantiated numerically for the separate beam
and triangular modes and the details may be found in own papers.
3.4 Analytical analysis of the optimization problem
The aim of this section is to work out of the analytical method, which will describe such
distribution of the actuators in order to assure the maximum of the reduction coefficient. It
is expected that the analytical method will confirm the quasi-optimal distribution which has
been found above with heuristic method. Therefore the assumptions are the same like in
heuristic method, namely
s
n ,
s
f and
s
are settled.
Let the distribution of actuators be marked with the set of unknown coordinates
s
{x } for the
moment; that are exactly these coordinates
s Q
{x } {x } of which are looked for. One starts
from Eq. (88), hence

2 2
R 0 0 1 1 1 1 2 2
I I I I f (x ) f (x ) ...
= + = + + + (91)
Since the (x) is the function which changes the sign, it is appropriate to search the points
s
x which assure the extreme
R
I (x) , not minimum only. The function
R
I (x) can have the
extreme only at points
s
x , at which
R
DI (x) is equal to zero or
2 2
1 1 1 1 2 2
f D (x ) f D (x ) ... 0 + + = does not exist (Fichtenholtz, 1999). Because
0
I is constant
than a necessary condition for existing extreme value is

415

R s
DI (x ) 0 = (92)
where
R s R s
DI (x ) DI (x x ) = = and hence

2 2
1 1 1 1 2 2
f D (x ) f D (x ) ... 0 + + = (93)
Because of
2
s s
f 0 then instead of Eq. (93) one has

s
D (x ) 0 = (94)
From the condition (94), a set of stationary points { }
s
x is obtained. The sufficient condition
for existing extreme demands, in order to the function be determined on either side of the
point
s
x and x D ( ) must change sign at this point (turning point); it is sufficient condition
formulated in the first form. This condition is expressed in the other form

2
s
D (x ) 0 (95)
If this condition is not fulfilling, then this point should be omitted.
One still needs to consider the biggest and the lowest values of (x) ; they are in
hypothetical points { }
max min
x , x . In order to find them, the values of (x) at the stationary
points { }
s
x are calculated and they are compared to the values calculated at the end points
of the appropriate interval. In the future consideration, the
s
n points among stationary { }
s
x
and { }
max min
x , x ones, at which (x) takes in turn its absolute values, are taken into account.
The problem of the signs of the
s
(x ) and
s
f is quite the same as in heuristic analysis.
Analytical analysis was applied for p-reduction and for the separate beam modes (Braski &
Lipiski, 2011). As pointed out there, the analytical solution to the optimal actuators
distribution problem confirms the results obtained with heuristic solution.
4. Conclusion
Deriving the shape of (x) , the influence both masses and stiffness of the actuators and glue
on the shape of X(x) , and consistently on the shape of (x) , were omitted; if not, an
adaptation method must be applied. But after determining shape of (x) , all these
parameters were considered.
As can be note, the actuators optimal distribution is attained assuming that the added
energy to excite actuators is constant. It is translated into constant
s
f . Having the optimal
distribution, the reduction coefficient may be improved by adding more energy or in order
words, by increasing
s
f . This way, presented optimal method corresponds to that one
presented in (Q. Wang & C. Wang, 2001), namely maximization of the control forces
transmitted by the actuators to the structure.
Based on theoretical considerations, and numerical ones presented in own papers, the
following conclusion may be formulated.
1. The optimization problem of the actuators distribution assuring the maximal active
vibration reduction of the beam, measured with reduction coefficient, may be solved
both heuristically and analytically. In analyzed problem, it turned out that both
methods give the same results.
2. The following algorithm of analytical method may be worked out:


416
to search of stationary points { }
s
x of the beam curvature,
to search of { }
max min
x , x points of the beam curvature,

Q
n points among stationary { }
s
x and { }
max min
x , x ones, at which (x) takes in turn its
maximum absolute values, are selected, they are denoted by
{ }
Q
x ,
to bond the actuators at the
{ }
Q
x points,
to determine the value of the reduction coefficient,
to increase the value
s
f , through the energy increase which excites actuators, until the
reduction coefficient will attain its maximum.
It seems that proposed optimization method is very simple and may be useful in many
technical problems of active vibration reduction. This work is a starting point for many
computer simulations and experiments.
5. References
Bapat, C.N. & Bapat, C. (1987). Natural frequencies of a beam with non-classical boundary
conditions and concentrated masses, Journal of Sound and Vibrations, Vol.112, No.1,
pp. 177182.
Braski, A. & Szela, S. (2007). On the quasi optimal distribution of PZTs in active reduction
of the triangular plate vibration. Archives of Control Sciences, Vol.17, No.4, pp. 427
437.
Braski, A. & Szela, S. (2008). Improvement of effectiveness in active triangular plate
vibration reduction. Archives of Acoustics, Vol.33, No.4, pp.521-530.
Braski, A.; Borkowski, M. & Szela, S. (2010). The idea of the selection of PZT-beam
interaction forces in active vibration protection problem. Acta Physica Polonica,
Vol.118, pp.17-22.
Braski, A. & Szela S. (2010). Quasi-optimal PZT distribution in active vibration reduction of
the triangular plate with P-F-F boundary conditions. Archives of Control Sciences,
Vol.20, No.2, pp.209226.
Braski, A. & Lipiski, G. (2011). Analytical determination of the PZTs distribution in active
beam vibration protection problem. (in press in Acta Physica Polonica).
Braski, A. & Szela, S. (2011). Evaluation of the active plate vibration reduction via the
parameter of the acoustic field. (in press in Acta Physica Polonica).
Bruant, I.; Coffignal, G.; Lene, F. & Verge, M. (2001). A methodology for determination of
piezoelectric actuator and sensor location on beam structure. Journal of Sound and
Vibrations, Vol.243, No.5, pp.861882.
Bruant, I. & Proslier, L. (2005). Optimal location of actuators and sensors in active vibration
control, Journal Inteligent Material System Structures, Vol.16, pp.197206.
Bruant, I.; Gallimard, L. & Nikoukar, S. (2010). Optimal piezoelectric actuator and sensor
location for active vibration control, using genetic algorithm. Journal of Sound and
Vibrations, Vol.329, pp.1615-1635.
Burke, S.E. & Hubbard, J.E. (1991). Distributed transducer vibration control of thin plate,
J.A.S.A., Vol.90, No.2, pp.937944.
Charette, F.; Berry, A. & Guigou, C. (1998). Dynamic effect of piezoelectric actuators on the
vibration response of a plate. Journal Inteligent Material System Structures, Vol.8,
pp.513524.
Croker, M.J. (2007). Handbook of noise and vibration control, John Wiley & Sons.

417
Dhuri, K.D. & Seshu, P. (2006). Piezo actuator placement and sizing for good control
effectiveness and minimal change in original system dynamics. Smart Material
Structure, Vol.15, pp.16611672.
Ercoli, L. & Laura, P.A.A. (1987). Analytical and experimental investigation on continuous
beam carying elastically mounted masses. Journal of Sound and Vibrations, Vol.114,
No.3, pp.519533.
Fichtenholtz, G.M. (1999). Differential and integral calculus, PWN, Warsaw.
Frecker, M. (2003). Recent advances in optimization of smart structures and actuators.
Journal Intelligent Material System Structures, Vol.14, pp.207215.
Fuller, C.R.; Elliot, S.J. & Nielsen, P.A. (1997). Active control of vibration, Academic Press,
London.
Gosiewski, Z. & Koszewnik, A. (2007). The influence of the piezoelements placement on the
active vibration damping system, Proceedings Active Noise and Vibration Control
Method, pp.6979, Krakow,
Guney, M. & Eskinat, E. (2007). Optimal actuator and sensor placement in flexible structures
using closed-loop criteria, Journal of Sound and Vibrations, Vol.312, pp.210233.
Halim, D. & Reza Moheimani, S.O. (2003). An optimization approach to optimal placement
of collocated piezoelectric actuators and sensors on a thin plate. Mechatronics,
Vol.13, pp.2747.
Han, S.M., Benaroya, H. & Wei, T. (1999). Dynamics of transversely vibrating beams using
four engineering theories, Journal of Sound and Vibrations, Vol.225, No.5, pp.935988.
Hansen, C.H. & Snyder, S.D. (1997). Active control of noise and vibration, E&FN SPON,
London.
Hernandes, J.A.; Almeida, S.F.M. & Nabarrete, A. (2000). Stiffening effects on the free
vibration behavior of composite plates with PZT actuators, Composite Structures,
Vol.49, pp.5563.
Hong, C.; Gardonio, P. & Elliott, S.J. (2007). Active control of resiliently mounted beams
using triangular actuators, Journal of Sound and Vibrations, Vol.301, pp.297318.
Ip, K.H. & Tse, P.C. (2001). Optimal configuration of a piezoelectric path for vibration
control of isotropic rectangular plate, Smart Material Structure, Vol.10, pp.395403.
Jha, A.K. & Inman, D.J. (2003). Optimal sizes and placement of piezoelectric actuators and
sensors for an inflated torus. Juornal Inteligent Material System Structures, Vol.14,
pp.563576.
Jiang, T.Y.; Ng, T.Y. & Lam, K.Y. (2000). Optimization of a piezoelectric ceramic actuator.
Sensors and Actuators, Vol.84, pp.8194.
Kaliski, S. (1986). Vibrations and Waves, PWN, Warsaw.
Kasprzyk, S. & Wiciak, M. (2007). Differential equation of transverse vibrations of a beam
with a local stroke change of stiffness, Opuscula Mathematica, Vol.27, No.2, 245252.
Kozie, M. (2006). Acoustic radiation of plates and shallow shells, PK, Monograph 331, ISS 0860-
097X, Krakow.
Liu, W.; Hou, Z. & Demetriou, M.A. (2006). A computational scheme for the optimal
sensor/actuator placement of flexible structures using spatial H2 measures,
Mechanical Systems and Signal Processing, Vol.20, pp.881895.
Low, K.H. & Naguleswaran, S. (1998). On the eigenfrequencies for mass loaded beams
under classical boundary conditions, Journal of Sound and Vibrations, Vol.215, No.2,
pp.381389.


418
Majkut, L. (2010). Eigenvalue based inverse model of beam for structural modication and
diagnostics. Part I: Theoretical formulation, Latin American Journal of Solids and
Structures, Vol.7, pp.423436.
Moshrefi-Torbati, M.; Keane, A.J.; Elliott, S.J.; Brennan, M.J.; Anthony, D.K. & Rogers, E.
(2006). Active vibration control (AVC) of a satelite boom structure using optimally
positioned stacked piezoelectric actuators, Journal of Sound and Vibrations, Vol.292,
pp.203-220.
Naguleswaran, S. (1999). Lateral vibration of a uniform Euler-Bernoulli beam carrying a
particle at an intermediate point, Journal of Sound and Vibrations, Vol.227, No.1,
pp.205214.
Pietrzakowski, M. (2004). Active damping of transverse vibration using distributed piezoelectric
elements, Monograph 204, PW, ISSN 0137-2335, Warsaw.
Przybyowicz, P.M. (2002). Piezoelectric vibration control of rotating structures, Monograph 197,
PW, ISSN 0137-2335. Warsaw.
Qiu, Z.; Zhang, X.; Wu, H. & Zhang, H. (2007). Optimal placement and active vibration
control for piezoelectric smart flexible cantilever plate, Journal of Sound and
Sheu, W.J.; Huang, R.T. & Wang, C.C. (2008). Influence of bonding glues on the vibration of
piezoelectric fans, Sensors and Actuators A, Vo.148, pp.115121.
Shimon, P.; Richer, E. & Hurmuzlu, Y. (2005). Theoretical and experimental study of
efficient control of vibration in a clamped square plate, Journal of Sound and
de Silva, C.W. (2000). Vibration, Fundamentals and practice, CRC Press.
Sullivan, J.M.; Hubbard, J.E. & Burke, S.E. (1996). Modeling approach for two-dimensional
distributed transducers of arbitrary spatial distribution, J.A.S.A., Vol.99, No.5,
pp.29652974.
Szela, S. (2009). Distribution method of the actuators in an active vibration reduction of the
triangular plate, AGH, Krakow, Ph.D. thesis, in polish.
Tylikowski, A. & Przybyowicz, P.M. (2004). Non-classical piezoelectric materials in vibrations
stability and dumping, PW, Warsaw.
Wang, F. (2007). Shape optimization for piezoceramics, Paderborn, Ph.D. thesis.
Wang, Q. & Wang, C. (2001). A controllability index for optimal design of piezoelectric
actuators in vibration control of beam structures, Journal of Sound and Vibrations,
Vol.242, No.3, pp.507518.
Wiciak, J. (2008). Vibration and Structural Acoustic Control-Selected Aspects, AGH, Monografh
175, ISSN 0867-6631, Krakow.
Yang, Y.; Jin, Z. & Soh, C.K. (2005). Integrated optimal design of vibration control system for
smart beam using genetic algorithms, Journal of Sound and Vibrations, Vol.282,
pp.12931307.
Part 4
Acoustic Wave Based Microdevices

19
Multilayered Structure as a Novel Material for
Surface Acoustic Wave Devices:
Physical Insight
Natalya Naumenko
Moscow Steel and Alloys Institute (Technological University)
Russia
1. Introduction
Since 70-ies, when the first delay lines and filters employing surface acoustic waves (SAW)
were designed and fabricated, the use of SAW devices in special and commercial
applications has expanded rapidly and the range of their working parameters was extended
significantly (Hashimoto, 2000; Ruppel, 2001, 2002). In the last decade, their wide
application in communication systems, cellular phones and base stations, wireless
temperature and gas sensors has placed new requirements to SAW devices, such as very
high operating frequencies (up to 10 GHz), low insertion loss, about 1 dB, high power
durability, stable parameters at high temperatures etc.
The main element of a SAW device is a piezoelectric substrate with an interdigital
transducer (IDT) used for generation and detection of SAW in the substrate. The number of
single crystals utilized as substrates in SAW devices did not increase substantially since 70-
ies because a new material must satisfy the list of strict requirements to be applied in
commercial SAW devices: sufficiently strong piezoelectric effect, low or moderate variation
of SAW velocity with temperature, low cost of as-grown large size crystals for fabrication of
4-inch wafers, long-term power durability, well developed and non-expensive fabrication
process for SAW devices etc. Today only few single crystals are utilized as substrates in
SAW devices: lithium niobate, LiNbO
3
(LN), lithium tantalate, LiTaO
3
(LT), quartz, SiO
2
,
lithium tetraborate, Li
2
B
4
O
7
(LBO), langasite, La
3
Ga
5
SiO
14
(LGS) and some crystals of LGS
group (LGT, LGN etc.) with similar properties.
The SAW velocities in these single crystals do not exceed 4000 m/s, which limit the highest
operating frequencies of SAW devices by 2.5-3 GHz because of limitations imposed by the
line-resolution technology of IDT fabrication. The minimum achievable insertion loss and
maximum bandwidth of SAW devices depend on the electromechanical coupling
coefficient, which can be evaluated for SAW as k
2
2V/V, where V is the difference
between SAW velocities on free and electrically shorted surfaces. The largest values of k
2
can
be obtained in some orientations of LN and LT. Ferroelectric properties of these materials
are responsible for a strong piezoelectric effect. As a result, k
2
reaches 5.7% in LN and 1.2%
in LT, for SAW. For leaky SAW (LSAW) propagating in rotated Y-cuts of both crystals, the
coupling is higher and can exceed 20% for LN and 5% for LT. However, LSAW attenuates
because of its leakage into the bulk waves when it propagates along the crystal surface. As a


422
result, insertion loss of a SAW device increases. Attenuation coefficient depends on a crystal
cut and IDT geometry. For example, in 36 to 48 rotated YX cuts of LT and in 41 to 76 YX
rotated YX cuts of LN, high electromechanical coupling of LSAW can be combined with low
attenuation coefficient via simultaneous optimization of orientation and electrode structure
(Naumenko & Abbott, US patents, 2003, 2004). When these substrates are utilized in
radiofrequency (RF) SAW filters with resonator-type structures, low insertion loss of 1dB or
even less can be obtained. Today such low loss filters are widely used in mobile
communication and navigation systems. The main drawback of these devices is high
sensitivity of the characteristics to variations of temperature because the typical values of
temperature coefficient of frequency (TCF) vary between -30 ppm/C and -40 ppm/C for
LT and between -60 ppm/C and -75 ppm/C for LN.
Contrary to LN and LT, quartz is characterized by excellent temperature stability of SAW
characteristics but low electromechanical coupling coefficient, k
2
<0.15%. Hence, even in
resonator-type SAW filters with very narrow bandwidths, about 0.05%, where the loss of
radiated energy is minimized due to the energy storage in a resonator, the best insertion loss
achieved in a SAW device with matching circuits is only 2.5-4 dB.
In some orientations of LBO, LGS and other crystals of LGS group, zero TCF is combined
with a moderate electromechanical coupling coefficient. However, these crystals have
limited applications in commercial SAW devices because low SAW velocities restrict high-
frequency applications on LGS and LBO dissolves in water and acid solutions, which
prohibits application of conventional wafer fabrication processes to this material and finally
results in an increased cost of SAW devices.
Hence, none of available single crystalline materials provides a combination of large
piezoelectric coupling, zero TCF and high propagation velocity. A strong need in such
material exists today, especially for application in SAW duplexers and multi-standard
cellular phones, where the temperature compensation is the key issue because of necessity
to divide a limited frequency bandwidth into few channels with no overlapping allowed in
a wide range of operating temperatures. As an alternative to conventional SAW substrates,
layered or multilayered (stratified) materials were studied extensively since 80-ies but only
in the last decade some of these structures found commercial applications in SAW devices,
due to the recent successes of thin film deposition technologies and development of robust
simulation tools for design of SAW devices on layered structures.
2. Multilayered structures as materials for SAW devices
As described above, the increasing requirements to the substrate materials, on one side, and
rapid development of thin film deposition technologies, on the other side, gave rise to the
novel class of materials for SAW devices layered or multilayered structures. One or more
films of different materials deposited on a regular substrate can improve its characteristics
significantly. A proper combination of a substrate and a thin film helps to overcome the
limitations of the conventional SAW substrates. In this section, a brief overview of the
layered structures will be given. The examples presented here are currently investigated by
different research groups as promising compound SAW substrate materials or found
already some applications in SAW devices. The focus is made on the recent achievements in
material research mostly motivated by challenges of the rapidly developing market of
communication devices. Based on this overview, a generalized multilayered structure,
Multilayered Structure as a Novel Material
for Surface Acoustic Wave Devices: Physical Insight

423
which includes all described examples, will be derived and a method of investigating this
structure will be presented and then applied to few different structures, for which more
detailed discussion of SAW propagation characteristics will be given.
The first example of a layered structure is a dielectric isotropic silicon dioxide (SiO
2
) film
deposited on one of rotated YX-cuts of LN or LT characterized by a high electromechanical
coupling. Today these structures attract attention of researchers and SAW designers as the
most promising candidates for application in SAW duplexers required in most of popular
mobile phone systems (Kovacs et al., 2004; Kadota, 2007; Nakai et al., 2008; Nakanishi et al.,
2008]. These RF devices must separate the transmitted and received signals in a narrow
frequency interval and in a wide range of operating temperatures, e.g. between -30C and
+85C. Therefore, substrate materials combining high propagation velocity, high
electromechanical coupling and low TCF are strongly required. Due to the opposite signs of
TCF in SiO
2
and LT or LN, a layered SiO
2
/LT or SiO
2
/LN structure allows to obtain the
desired combination of characteristics. When it is utilized in a resonator-type filter, the
electrodes of IDT and metal gratings are commonly built at the interface between SiO
2
and
LN or LT. Besides, a heavy metal, such as copper, is utilized as an electrode material (Nakai
et al., 2008). A layered structure with such electrode configuration is schematically
presented as Type 1 in Fig.1. The location of electrodes at the interface helps to keep a high
electromechanical coupling and combine it with a large reflection coefficient in SAW
resonators. A large metallization thickness effectively reduces resistive losses and results in
a high Q factor of a SAW resonator. Another advantage of using heavy electrodes is a
reduced propagation loss, which is achieved due to the transformation of LSAW into SAW.
SAW devices with low insertion loss of 1-2 dB and low TCF of -7ppm/C, have been
successfully realized on such substrates (Kadota, 2007).
A thickness of SiO
2
film in SiO
2
/LT and SiO
2
/LN structures can vary within a wide range,
from a few percent of a SAW wavelength up to a few wavelengths, to provide the required
combination of electromechanical coupling, TCF and propagation loss. Moreover, SiO
2
film
helps to isolate the working surface of a SAW device from environmental influences and
facilitates packaging of SAW chip.

film
substrate substrate
film
IDT electrodes
Type 1 Type 2

Fig. 1. Two typical structures with one thin film
Another electrode configuration in a structure with one thin film is schematically presented
as Type 2 in Fig. 1, with IDT located on the top surface. It is typical for a piezoelectric film on
a non-piezoelectric substrate or on a substrate with low electromechanical coupling
coefficient. As a piezoelectric film, zinc oxide ZnO is widely used (Kadota & Minakata, 1998;
Nakahata et al., 2000; Emanetoglu et al., 2000; Brizoual et al., 2008). ZnO films are cheap and
provide sufficiently high values of electromechanical coupling. The film deposition


424
technique (e.g. magnetron sputtering) has been well developed for this material. Another
piezoelectric film, which is extensively studied as a promising material for high-frequency
SAW devices is aluminum nitride (AlN) (Benetti et al., 2005, 2008; Fujii et al., 2008; Omori et
al., 2008). It is characterized by chemical stability, mechanical strength, high acoustic
velocity and good dielectric quality. Some other piezoelectric films, like CdS or GaN, were
investigated previously but did not receive as much attention as ZnO or AlN.
A piezoelectric film is usually combined with silica glass, silicon, sapphire or diamond
substrate. Silica glass is cheap, the use of silicon as a wafer enables simple integration of IF
and RF components in one chip, sapphire is characterized by high SAW velocities, up to
6000 m/s, and diamond provides the highest SAW velocities among all materials, up to
11000 m/s, and is being used for high frequency SAW devices in the GHz range. For
example, SAW resonator with center frequency about 4.5 GHz was built on AlN/diamond
structure characterized by SAW velocity about 10000 m/s and k
2
1% (Omori et al., 2008). A
combination of SAW velocity about 5500 m/s and electromechanical coupling about 0.25%
can be obtained in AlN/sapphire structure (Ballandras et al., 2004). To reduce TCF of a SAW
device, ZnO film can be combined with quartz or LGS. For example, nearly zero TCF and k
2

about 1.8% was achieved for SAW in ZnO/quartz structure, via optimization of quartz
orientation (Kadota et al., 2008).
One more structure, which can be referred to the Type 1, recently found application in SAW
devices. It is a thin plate of a piezoelectric crystal, such as LN or LT with thickness 10-15
wavelengths, which is directly bonded to a dielectric or semiconductor wafer. The bonding
technology (Eda et al., 2000) provides excellent contact between the two materials and
allows fabrication of SAW devices with reproducible characteristics on a thin LN or LT plate
bonded to a thick silicon or glass wafer. In these structures, high values of electromechanical
coupling coefficients typical for LN and LT are combined with improved TCFs, due to low
thermal expansion coefficients (TCE) determined by massive silicon, glass or sapphire wafer
(Tsutsumi et al., 2004). An example of bonded wafer will be numerically investigated in
section 4.
The quality of a contact between LN or LT plate and a silicon wafer can be improved if a
thin SiO
2
film is deposited between these materials (Abbott et al., 2005). Such two-layered
structure with IDT on the top surface is schematically shown as Type 3 in Fig. 2. With silicon
as a substrate, SiO
2
as the first film and LN or LT as the second film (plate), this structure
can give the same advantages as LT/Si or LN/Si bonded wafers but with higher quality
contact between the materials. The presence of additional SiO
2
film results in spurious
acoustic modes propagating in a SAW device. These modes deteriorate the device
performance and should be simulated properly to achieve the desired device characteristics.
Another example of the Type 3 structure is a silicon wafer with isotropic SiO
2
as the first film
and ZnO as the second film. Optimization of SiO
2
and ZnO film thicknesses enables
obtaining of a structure with TCF=0 (Emanetoglu et al., 2000). A high frequency SAW device
can be built if SiO
2
and ZnO films are deposited atop of a diamond or a sapphire substrate.
With ZnO as the first film and isotropic SiO
2
as the second film, the preferential location of
IDT electrodes is at the substrate-film interface (Type 4). Alternatively, IDT can be built on
ZnO surface and then buried in SiO
2
overlay (Type 5). For example, Nakahata (Nakahata et
al., 2000) reported on a SAW resonator using SiO
2
/ZnO/diamond structure with two
different electrode configurations (Type 4 and Type 5). Zero TCF, high velocity about 10000
m/s and k
2
1.2% were obtained for shear horizontally (SH) polarized SAW mode. A

425
resonator with center frequency about 2.5 GHz, temperature compensated characteristics
and low insertion loss was fabricated on this structure.
Nakahata (Nakahata et al., 1995) reported one more example of the two-layered structure,
which can be referred to Type 4. It is a ZnO film on a silicon wafer with thin isotropic
diamond layer between them. The following SAW characteristics have been obtained:
velocity V8050 m/s, k
2
1.42 %, TCF0.

2
nd
film

1
st
film

Substrate

2
nd
film

1
st
film

Substrate

Type 4 Type 5 Type 3
2
nd
film

1
st
film

Substrate

IDT

Fig. 2. Three typical structures with two thin films
The examples described above are not aimed at comprehensive survey of layered structures
potentially applicable in SAW devices but demonstrate that a variety of layered structures
can be referred to a few basic types. A unified approach to analysis of acoustic modes in
different layered structures would be beneficial for optimization of SAW devices, because
such approach allows comparing characteristics of the same SAW design built on different
combinations of film and substrate materials.
The simulation of SAW characteristics is an important part of the SAW device design
procedure. In a specified structure, such simulation must take into account orientation of
each material if it is anisotropic, film thicknesses, a thickness and shape of IDT electrodes,
electrode width to pitch ratio etc. Besides, the accurate analysis of all modes propagating in
the investigated structure is required, including the main SAW or LSAW mode and all
spurious modes generated by IDT in the specified frequency interval. A number of spurious
modes grows with a number of layers and increasing of their thicknesses, which makes the
simulation procedure more complicated. Moreover, with increasing film thickness SAW
changes its nature and eventually transforms into a new type of acoustic wave. However,
the characteristics of any acoustic mode change continuously with this transformation.
The variation of film thicknesses within wide range helps to obtain a variety of novel
materials with different combinations of characteristics demanded for SAW devices of
different applications. After a proper combination of materials is selected, the geometrical
parameters of a multilayered structure must be optimized to satisfy the desired electrical
specification, including frequency bandwidth, insertion loss, out-of-band rejection, shape
factor of frequency response or Q factor of a SAW resonator, temperature deviation of
frequency etc. It is a common practice to optimize film and electrode thicknesses and other
geometrical parameters of IDTs simultaneously with orientations of anisotropic materials
included in the layered structure, to achieve the best SAW device performance.
The challenges described above require a robust, fast and universal numerical technique,
which could be applied to different types of multilayered structures, with film thicknesses
varying within wide range and allowing transformation of SAW into boundary waves, plate
modes or other types of acoustic waves. Such technique is described in the next section.


426
3. An advanced numerical technique for analysis of acoustic waves in
multilayered structures
The most popular numerical technique used for simulation of SAW characteristic in multi-
layered structures is Transfer Matrix Method (TMM) (Adler, 1990). It is based on the matrix
formalism suggested by Stroh (Stroh, 1965) for solution of a SAW problem in anisotropic
media. For each material of a multilayered structure, TTM assumes building of a fundamental
acoustic tensor dependent on the material constants and the analyzed orientation. Then the
characteristics of the partial acoustic modes are found as the eigen vectors and eigen values
of the matrix associated with this tensor. Finally, the transfer matrix is calculated, which
characterizes the change of acoustic fields within the analyzed layer. The method is fast and
convenient and does not impose any limitations on the number of layers. However, it is
known to work unstable when the film thickness exceeds 3-5 wavelength because of bad
conditioned matrices built of elements, some of which exponentially decay and others
exponentially grow with film thickness. These elements are associated with incident and
reflected modes. As suggested by Tan (Tan, 2002) and Reinhardt (Reinhardt et al., 2003), a
separate treatment of these two groups of partial modes helps to avoid the instability and
extends the range of the analyzed thicknesses from zero to infinite value.
Another limitation of the previously reported numerical techniques developed for analysis
of SAW in multilayered structures is their focusing on a certain type of acoustic waves,
which is related to a fixed type of a multilayered structure, with analytically defined
boundary conditions at the interfaces and external boundaries. For example, acoustic waves
propagating in a substrate with a thin film of finite thickness and stress-free boundary
conditions at the top surface are different from acoustic waves propagating in the same
combination of materials when the film thickness tends to infinite value. In practice, the
results obtained with different versions of software using fixed boundary conditions often
diverge and do not allow seeing how the wave characteristics change continuously with
variation of film thickness within wide range. For example, the software FEMSDA (Endoh et
al., 1995; Hashimoto et al., 2007, 2008), which is very popular among SAW device designers,
includes separate versions for analysis of SAW in a substrate with a thin film and for
investigation of boundary waves. In the first version, a film thickness providing robust
calculations does not exceed a half-wavelength.
To overcome the described limitation, an advanced numerical technique was developed
(Naumenko, 2009, 2010). It can be applied to a variety of multilayered structures and types
of acoustic waves. The universal character of the software is achieved due to
characterization of the air as a dielectric medium with a very small density and elastic
stiffness constants and treatment of this medium as an example of a dielectric. The same
numerical methods are applied to this and other materials, which compose a layered
structure. With such approach, it is not necessary to fix the stress-free boundary conditions
at the top surface of a thin film or at the boundaries of a finite-thickness plate to find
acoustic waves propagating in these structures. The stress-free boundary conditions are
automatically simulated with high accuracy when the adjacent medium is specified as the
air.
The developed technique refers to a multilayered structure schematically shown in Fig. 3, in
which a metal film or IDT is located between N upper and M lower layers, where M and N
can vary between one and ten or more if necessary.

427

p
Upper
layers
(n=1,N)

Electrodes
or uniform
metal film

Lower
layers
(m=1,M)
Piezoelectric
Air or dielectric
p
Air or substrate
z=h
m

z=0
X
Z

Fig. 3. Schematic drawing of analyzed multilayered structure
Analysis starts from the uppermost or lowermost half-infinite material, in which the wave
structure is calculated. It can be a dielectric, a piezoelectric material or the air. In each
adjacent finite-thickness layer, the transformation of the wave structure is deduced via
separate treatment of incident and reflected partial modes. It means that the reflection and
transmission matrix coefficients replace the transfer matrix to escape numerical noise at film
thicknesses exceeding 3-5 wavelengths. For the structures with few dielectric (isotropic)
films, the variation of the dielectric permittivity within each film characterized by the finite
thickness h and dielectric permittivity
film
is taken into account via the well known
recursive equation (Ingebrigtsen, 1969):
( )
( )
( )
( )
( )
film
film
film
z th kh
z h
z th kh

+

+ =
+

, (1)
where k is the wave number. Analysis of the lower and upper multilayered half-spaces is
considered completed when the wave structure has been determined at z=0 and z=h
m
,
where h
m
is a metal film thickness, and the surface impedance matrices ( )
UP
Z k and
( )
LOW
Z k have been calculated at the upper and lower boundaries of the metal film. Each of
these matrices characterizes the ratio between the vectors of displacements u and normal
stresses T at the analyzed interface,
1
Z

= uT . A piezoelectric material is characterized by
the generalized 4-dimensional displacement and stress vectors, with added electrostatic
potential and normal electrical displacement D, respectively. The matrices ( )
UP
Z k and
( )
LOW
Z k comprise the information about the layers located above and below the metal film
and enable simple formulation of electrical boundary conditions at z=0 and z=h
m
. If the mass
load of metal film is included in the analysis of the upper N layers, then the function of
effective dielectric permittivity (EDP) ( )
s
k can be calculated at z=0 (Ingebrigtsen, 1969;
Milsom et al., 1977). This function relates the electric charge at the surface to the
electrostatic potential ,


428
( )
s
k k = (2)
and can be used for extraction of the velocities and electromechanical coupling coefficients
of SAW and other acoustic modes propagating in a multilayered structure. EDP function
was originally introduced for semi-infinite piezoelectric medium, but it is also an efficient
tool for analysis of acoustic waves in layered structures.
The numerical method described above was extended to a periodic metal grating
sandwiched between two multilayered structures. In this case, the spectral domain analysis
(SDA) of the upper and lower multi-layered half-spaces is combined with the finite-element
method (FEM) applied to simulation of the electrode region. To some extent, the developed
SDA-FEM-SDA technique (Naumenko, 2010) can be considered as an advanced FEMSDA
method. In this case, the function of Harmonic Admittance Y(f,s) (Blotekjear et al, 1973;
Zang et al, 1993) is calculated. Similar to EDP function, Harmonic Admittance relates the
electric charge on the electrodes of the infinite periodic grating Q to the applied
harmonically varying voltage
e
V ,
( )
1
e
Q j Y(f,s) V
= (3)
and depends on frequency f and the normalized wave number, s p / = , where p is a pitch
of the grating. This function can be used for simulation of a SAW resonator and calculation
of its main parameters: resonant and anti-resonant frequencies, reflection coefficient etc.
Also it comprises the information about SAW and other acoustic modes, which can be
generated in the analyzed layered structure, and their characteristics can be extracted from
Y(f,s) .
It should be mentioned that a numerical procedure of finding eigen modes of the
fundamental acoustic tensor can be successfully applied to the air as an example of isotropic
medium and the calculated SAW characteristic do not differ noticeably from that obtained
with stress-free conditions set analytically at the film/air interface. The method and
software SDA-FEM-SDA enable analysis of electrodes composed of few different metal
layers and having a complicated profile, with different edge angles in metal layers. The gaps
between electrodes may be empty or filled with a dielectric material. Due to these options,
some important physical effects can be simulated, such as the effect of a sublayer (e.g.
titanium) often used for better adherence of electrode metal to the substrate or the effect of
nonrectangular electrode profile on a SAW device performance.
The developed numerical technique can be applied to different types of multilayered
structures and different acoustic waves can be investigated, for example,
a. SAW and LSAW in a piezoelectric substrate;
b. SAW and LSAW in a substrate with one or few thin films (e.g. Love modes);
c. plate modes generated by IDT and propagating in a thin plate (e.g. Lamb waves);
d. boundary waves propagating along the interface between two half-infinite media;
e. acoustic waves propagating in a thin piezoelectric plate bonded to a thick wafer.
In addition, a continuous transformation between different types of acoustic waves can be
observed. It gives a physical insight into the mechanisms of wave transformation
with increasing film thickness. An example of wave transformation will be considered in
Section 5.

429
4. Multilayered structures: Examples of analysis
In this section, few examples of application of the developed numerical technique to the
structures of practical importance are presented.
4.1 SiO
2
/42YX LT with Al film at the interface
The first example is a dielectric film on a piezoelectric substrate, which can be referred to the
Type 2 structure shown in Fig.1. The calculated characteristics of LSAWs propagating in
SiO
2
/42YX LT with uniform Al film atop of the structure are presented in Fig. 4.

-50
-40
-30
-20
0 0.05 0.1 0.15 0.2
SiO
2
thickness (wavelengths)

T
C
F

[
p
p
m
/
o
C
]
0
0.001
0.002
0.003
0.004

A
t
t
e
n
u
a
t
i
o
n

[
d
B
/
w
a
v
e
l
e
n
g
t
h
]
4060
4100
4140
4180
V
e
l
o
c
i
t
y

[
m
/
s
]
0
1
2
3
4
C
o
u
p
l
i
n
g

[
%
]
V
OC

V
SC

k
2

OC

SC

TCF
OC

TCF
SC

Al
SiO
2

42YX LT

Fig. 4. Characteristics of leaky SAW propagating in SiO
2
/42YX LT with uniform Al film
(h
Al
=5%) atop of SiO
2
film, as functions of normalized SiO
2
film thickness. OC or SC
electrical conditions are analyzed
SiO
2
is an isotropic dielectric film. LSAW velocities V, attenuation coefficients (in dB/,
where is LSAW wavelength) and TCF are presented as functions of the normalized SiO
2

film thickness, h/. These characteristics have been calculated for the open-circuited (OC)
and short-circuited (SC) electrical conditions in Al film. The finite thickness of a metal film
(h
Al
=5%) was taken into account. The difference between the OC and SC velocities
determines the electromechanical coupling coefficient k
2
, which decreases rapidly with
increasing dielectric film thickness. The behavior of attenuation coefficients depends on the
electrical condition. The functions
OC
(h
SiO2
) and
SC
(h
SiO2
) reach nearly zero values at
h
SiO2
=5% and h
SiO2
=8%, respectively. Therefore, the variation of SiO
2
film thickness can be
used for minimization of propagation losses in a SAW device. Due to the opposite signs of
TCF in SiO
2
film and LT substrate, in the layered structure the absolute value of TCF


430
reduces with increasing film thickness but does not reach zero in the investigated interval of
film thicknesses. However, larger SiO
2
thicknesses are not considered because the
electromechanical coupling coefficient becomes too low for practical applications. Further
improvement of TCF is possible if IDT is located at the SiO
2
/LT interface (Type 1 structure).
Such example will be considered in Section 5.
4.2 ZnO/sapphire: Existence of high-velocity SAW
The next example, ZnO film on a sapphire substrate, is a layered structure with a
piezoelectric film on a non-piezoelectric substrate, which is potentially useful for high-
frequency SAW device applications. Also this example demonstrates that a deposition of a
thin film on a substrate can result in the existence of a high-velocity SAW, which cannot
exist in a crystal without a thin film.
The typical SAW velocities in layered structures using a sapphire substrate are about 5500
m/s. Leaky SAWs, which have higher velocities, propagate with certain attenuation
dependent on orientation of a substrate. Two types of LSAW can exist in crystals and
layered structures: common or low-velocity LSAW, with velocities confined in the interval
between that of the slow quasi-shear and fast quasi-shear limiting bulk acoustic waves
(LBAWs), and high-velocity LSAW with velocities between that of the fast quasi-shear and
quasi-longitudinal LBAWs. The limiting BAW is a bulk wave, which propagates in the
sagittal plane (i.e. the plane, which is normal to the substrate surface and parallel to the
propagation direction of SAW or LSAW) and is characterized by the group velocity parallel
to the substrate surface. Usually a high velocity LSAW is not suitable for SAW device
applications because of its fast attenuation. In some crystals with strong acoustic anisotropy,
a high-velocity LSAW degenerates into the quasi-longitudinal LBAW in selected
orientations, and low-attenuated LSAW can propagate around such orientations
(Naumenko, 1996). For example, such waves exist in some orientations of quartz and LBO.

Fig. 5. (a) Velocity and attenuation coefficients of high-velocity leaky waves propagating in
sapphire, Euler angles (0, -20.3, 0), with ZnO film, as functions of normalized ZnO
thickness, and (b) Displacements as function of depth, for HVSAW existing at h
Zn0
=5.16%

431
Fig. 5, a shows the calculated velocities and attenuation coefficients of high-velocity LSAWs
propagating in ZnO/sapphire structure, when the sapphire orientation is defined by the
Euler angles (0, -20.3, 0). The thickness of a metal film deposited atop of ZnO is not taken
into account. The velocities of the fast quasi-shear and quasi-longitudinal LBAWs in
sapphire are shown as V
FAST SHEAR BAW
and

V
LONGITUDINAL BAW
, respectively. The non-
attenuated SAW solution, which occurs on a high-velocity LSAW branch at h
ZnO
5.16%
(Fig.5, a), is not a quasi-bulk wave described above. Fig.5,b shows the mechanical
displacements, which follow the propagation of this wave. The analyzed solution is a
sagittally polarized surface wave, which attenuates exponentially into the depth of sapphire,
similar to Rayleigh SAW. Such high velocity SAW (HVSAW) can not exist without
perturbation of a free crystal surface, e.g. by deposition of a thin film or a metal grating. The
existence of this type of waves was revealed via numerical analysis of experimental data on
SAW modes in ZnO/SiC structure (Didenko et al., 2000) and confirmed by other examples
of layered structures, which support propagation of these waves, such as ZnO/diamond,
Zno/sapphire etc. (Naumenko & Didenko, 1999).
The HVSAW found in ZnO/sapphire may be attractive for applications in high-frequency
SAW devices because it combines a high propagation velocity exceeding 9000 m/s with
electromechanical coupling about 0.3%. With deposition of a metal film or a periodic metal
grating the wave with attractive properties does not disappear but a combination of cut
angle and ZnO thickness should be optimized properly to provide low LSAW attenuation.
4.3 Al grating on 46YX LT/Si bonded wafer
In this example, SAW modes are investigated in LT/Si structure with a periodic Al grating
atop of LT, which can be obtained experimentally by bonding LT plate to a silicon wafer. In
such structure, the TCF may be dramatically reduced compared to regular LT wafer.

0 2 4 6 8
Normalized LT thickness (wavelengths)
V
e
l
o
c
i
t
y

[
m
/
s
]
Al
46YX LT
Silicon

h
LT

4200

4100

4000

3900

3800
V
SC
in regular LT
V
OC
in regular LT
SC
OC

Fig. 6. Velocities of acoustic modes propagating in 46YX LT plate bonded to silicon wafer,
with OC and SC Al grating, as functions of normalized LT thickness, h
Al
=9%


432
The results shown in Fig.6 were obtained with the software SDA-FEM-SDA because the
effect of a periodic metal grating is different from that of a uniform metal film. When LT
thickness is about 1-2 wavelengths, the velocity of the SAW mode propagating in the
layered structure is nearly the same as in a regular LT substrate with electrode thickness
h
Al
=9% but the wave characteristics are perturbed by interactions with multiple plate
modes, the number of which grows with increasing LT thickness. It should be noted that in
a regular 46YX LT substrate the acoustic wave propagating with nearly the same velocity
has a leaky wave nature. The bonding of LT plate to a silicon wafer results in the
transformation of LSAW into SAW. The leakage of the wave becomes impossible because in
silicon the shear BAW propagates faster than the analyzed SAW mode.
4.4 Al grating on 46YX LT/SiO
2
/Si bonded wafer
The next example differs from the previous one by the additional SiO
2
film between LT and
Si wafer. A silicon dioxide layer is required to be deposited on the LT wafer to enable a
stronger bond to silicon (Abbott et al., 2005). The presence of additional SiO
2
film impacts
the acoustic and electrical properties of a bonded wafer and SAW resonators built on its
surface. The spectrum of acoustic modes propagating in LT/SiO
2
/Si bonded wafer looks
more complicated than in LT/Si structure and depends on SiO
2
and LT thicknesses. In Fig.
7, the velocities of acoustic modes are shown as functions of the normalized SiO
2
film
thickness when LT thickness is fixed, h
LT
=6.

0 1 2 3 4
SiO
2
V
e
l
o
c
i
t
y

[
m
/
s
]
Al
46YX LT

SiO
2

Silicon

h
SiO2

V
SC
in regular LT
V
OC
in regular LT
4200

4100

4000

3900

3800

Fig. 7. Velocities of acoustic modes propagating in 46YX LT/SiO
2
/Si structure with Al
grating, as functions of normalized thickness of SiO
2
film, when h
LT
=6 and h
Al
=9%
4.5 Cu grating buried in SiO
2
,

on 46YX LT/SiO
2
/Si bonded wafer
This numerical example refers to the same bonded structure as described above but
demonstrates the effect of additional SiO
2
overlay deposited over periodic metal grating

433
electrodes. Such layer is aimed at further improvement of the temperature characteristics of
SAW devices.
In this case, Cu electrodes are investigated because this metal provides higher reflection
coefficients in SAW resonators and hence lower insertion loss in RF SAW filters than Al,
when electrodes are buried in SiO
2
overlay. The velocities of acoustic modes are shown in
Fig.8 as functions of the normalized intermediate SiO
2
film thickness, while the thicknesses
of SiO
2
overlay, LT plate and Cu electrodes are fixed, h
OVL
=0.25%, h
LT
=4 and h
Cu
=2.5%,
respectively. The dispersion of SAW velocities in 46LT/SiO
2
/Si (Fig. 8,a) and
SiO
2
/46LT/SiO
2
/Si (Fig. 8,b) structures demonstrates that in practice very accurate
simulation is required to account for all spurious modes, because these modes may affect
the admittance of a SAW resonator.

Cu
46YX LT

SiO
2

Silicon
4
3850
3900
3950
4000
4050
4100
4150
0 0.5 1 1.5 2
SiO
2
V
e
l
o
c
i
t
y

[
m
/
s
]
3850
3900
3950
4000
4050
4100
4150
0 0.5 1 1.5 2
SiO
2
V
e
l
o
c
i
t
y

[
m
/
s
]
SiO
2

46YX LT

SiO
2

Silicon
Cu
h
OVL

(a)
(b)
V
OC
in LT
V
SC
in LT
V
OC
in LT
V
SC
in LT
SC
OC

Fig. 8. Velocities of acoustic modes as functions of normalized thickness of intermediate SiO
2

film, when h
LT
=4 and h
Cu
=2.5%, (a) in 46LT/SiO
2
/Si structure with Cu grating, and (b) in
SiO
2
/46LT/SiO
2
/Si structure with Cu grating atop of 46LT plate and overlay thickness
h
OVL
=0.25%


434
Fig. 9 demonstrates an example of calculated admittance of a periodic Cu grating used with
different multilayered structures described above. In addition to the main SAW mode,
which exhibits resonance and anti-resonance, the multiple spurious modes propagate in the
analyzed layered structures and can disturb a SAW resonator performance. The frequencies
of the spurious responses are very sensitive to the thicknesses of the layers in a
multilayered structure. However, the accurate simulation of these modes enables
optimization of the layered structure to minimize the effect of the spurious modes on a
resonator performance.

1000

100

10

1

0.1

0.01

0.001
0.45 0.46 0.47 0.48 0.49 0.5
Normalized frequency, fp/V
BAW
N
o
r
m
a
l
i
z
e
d

a
d
m
i
t
t
a
n
c
e
LT
LT/Si
LT /SiO2/Si
Resonance
Anti-resonance

Fig. 9. Admittance of infinite periodic grating with Al electrodes (h
Al
=10%), as function of
normalized frequency, when the grating is built on regular 42YX LT substrate, on LT plate
(h
LT
=5.5) bonded to Si substrate and atop of LT/SiO
2
/Si structure, with h
SiO2
=2.
V
BAW
=4214.636 m/s
The effect of spurious modes on a resonator performance can be minimized by the
variation of rotation angle of LT plate. If one of rotated YX cuts of LN is used as a
piezoelectric plate, the insertion loss of a resonator SAW device can be reduced in a wider
bandwidth.
The examples presented in this section illustrate possible applications of the developed
numerical technique and demonstrate that being a part of design tools for SAW device
simulation it can be also an efficient tool for analysis of acoustic modes in multilayered
structures.

435
5. Transformation of acoustic waves in anisotropic layered structures
In isotropic or highly symmetric materials, acoustic waves are characterized by mechanical
displacements either belonging to the sagittal plane, u=(u
1
, 0, u
3
), or normal to this plane,
u=(0, u
2
, 0), i.e. such waves are Rayleigh-type or SH polarized modes. Similar solutions
occur in some symmetric orientations of materials belonging to the lower symmetry classes.
Such solutions have been extensively studied analytically. To the best authors knowledge,
the most comprehensive overview of different types of acoustic waves existing in substrates
with thin films, in thin plates and at the boundary between two half-infinite media was
made by Viktorov (Viktorov, 1967, 1981). Some statements, which refer to acoustic waves
propagating in isotropic structures, are listed below.
a. In isotropic substrate with isotropic thin film, SH-polarized Love waves can propagate
if the shear BAW propagates faster in a substrate than in a film.
b. Along the boundary between two rigidly connected isotropic half-infinite media, the
sagittally polarized Stonely waves can propagate. These waves are usually trapped near
the interface between two media, with penetration depth about one wavelength.
c. SH-polarized boundary waves can exist if additional thin film with lower shear BAW
velocity is added between two half-infinite media.
d. In isotropic thin plates, two types of waves can exist: sagittaly polarized Lamb waves
(symmetric and anti-symmetric modes) and SH-polarized plate modes. With increasing
plate thickness, higher-order modes appear and their number increases.
e. With plate thickness decreasing to zero, the first-order symmetric Lamb mode
degenerates into the longitudinal BAW. With increasing plate thickness, this mode
finally degenerates into two Rayleigh SAWs propagating along the boundaries of the
plate.
f. Higher-order plate modes arise from the shear and longitudinal BAWs at certain cut-off
thicknesses and have a structure of standing waves propagating between two
interfaces.
The layered structures used in SAW devices must include at least one anisotropic material
to provide a piezoelectric coupling of SAW with IDT. Anisotropy results in mixed
polarizations of acoustic modes propagating in thin films, plates and along the interface
between two media. The transformation of each mode with increasing film thickness is
unique and requires separate investigation. Whereas analytical study of such waves is
possible only in some symmetric orientations, the numerical technique presented in this
chapter enables calculation of the wave characteristics and analysis of displacements
associated with different acoustic modes. Its application to multilayered structures can
reveal the mechanisms of wave transformation. The understanding of these mechanisms
helps to select properly the thicknesses of metal and dielectric or piezoelectric layers to
ensure the propagation of a required acoustic wave.
An example of such investigation is presented here. It refers to 42YX LT with SiO
2
film.
Similar structure was considered in Section 4.1, but in the present example a periodic
grating is analyzed instead of a thin metal film and this grating is located at the interface
between LT substrate and SiO
2
film. As a metal of the grating, copper is considered. Such
structure is of great practical importance as potential material for RF SAW devices with
improved temperature characteristics.
The calculated velocities of acoustic modes propagating in LT substrate with copper grating
are shown in Fig.10,a as functions of the Cu electrode thickness. Fig.10,b shows the velocities


436
of acoustic waves propagating in SiO
2
/Cu grating/LT structure, as functions of the
normalized SiO
2
thickness, with Cu thickness fixed, h
Cu
=0.2. The analyzed Cu thicknesses
look too high for application in SAW resonators but provide better insight into the wave
transformation mechanisms, which is a purpose of this investigation.
When the metal thickness is small, two acoustic waves exist in LT, SAW1 and LSAW. With
increasing Cu thickness, the velocities of both modes decrease rapidly and at h
Cu
=0.075
LSAW transforms into the second SAW mode, SAW2. With further increasing of electrode
thickness, two SAW modes interact with each other. To avoid discontinuities in the
characteristics of two SAW modes, these modes are distinguished by their velocities:
V
SAW2
>V
SAW1
.

0 1 2 3
h
Si O2
(wav elengt hs )
2000
3000
4000
5000
0 0.1 0.2
h
CU
(wavelengths)

V
e
l
o
c
i
t
y

[
m
/
s
]
V
BAW1-LT

Boundary waves: BW1 BW2
SAW1
LSAW

h
SiO2

(a) (b)
Cu
LT

V
SBAW-SiO2

V
BAW2-LT

SAW2
SiO
2

Cu
LT
h
SiO2
=h
Cu

High-order LSAW modes
OC

SC
1
st

2
nd

Fig. 10. SAW and leaky SAW velocities in 42YX LT, (a) with Cu film, as functions of film
thickness; (b) with Cu and SiO
2
films, as functions of normalized SiO
2
film thickness
If metal thickness is fixed (h
Cu
=0.2) and the gaps between electrodes are filled with SiO
2
,
the velocities of two SAW modes grow rapidly (Fig.10,b). Another interaction between
SAW1 and SAW2 occurs at h
SiO2
h
Cu
, i.e. when the top surface of the whole structure
becomes flat. With increasing SiO
2
film thickness two SAW modes finally transform into the
boundary waves, BW1 and BW2. The boundary waves propagate with velocities lower than
that of the shear BAW in SiO
2
. The wave BW2 shows electromechanical coupling sufficient
for application in resonator SAW devices, k
2
=3.49%. For this mode, TCF grows from -31
ppm/C in LT substrate up to 10 ppm/C in a layered structure with SC grating and from -
43 ppm/C up to 6 ppm/C with OC grating. At h
SiO2
>0.7, higher-order plate modes arise
from the fast shear limiting BAW in LT.

437

(a)

(c)

(e)

(g)

(b)

(d)

(f)

(h)

Cu SiO
2

LT

h
SiO2
=1

Cu
LT
SiO
2

Cu

LT
u
2

LT
1.5
p=/2
0.2

Fig. 11. Transformation of two SAW modes in 42YX LT with Cu grating with increasing
SiO
2
film thickness; (a), (c), (e), (g) SAW1, (b), (d), (f), (h) SAW2
Fig.11 demonstrates how the acoustic fields associated with the modes SAW1 and SAW2
change with increasing SiO
2
film thickness. OC electrical condition is considered, by way of
example. The components of mechanical displacements in the sagittal plane are revealed as
perturbations of the regular mesh and the values of SH components are presented as
colored diagrams. When Cu thickness is small (Fig.11,a,b), SAW1 is nearly perfect Rayleigh
wave and LSAW is a quasi-bulk SH-type wave slowly attenuating with depth. When
h
Cu
>0.075, LSAW transforms into SAW2, which is also SH-type wave. At Cu thicknesses
about 0.12, both SAW modes are perturbed by interaction between them. In the interval of


438
Cu thicknesses between 0.12 and 0.2, the modes SAW1 and SAW2, which have been
determined as the lower-velocity and higher velocity modes, exchange their polarizations.
After the second interaction, which occurs at h
SiO2
=h
Cu
=0.2, SAW1 and SAW2 turn back into
Rayleigh-type and SH-type waves, respectively. However, at h
Cu
=0.2 (Fig.11,c,d) both
waves still have mixed polarizations. With increasing SiO
2
film thickness, SAW1 and SAW2
transform into the boundary waves BW1 and BW2, respectively (Fig. 11, g, h), with acoustic
waves localized in Cu grating and around it. The boundary waves have mixed
polarizations, which would be impossible in isotropic substrate with isotropic thin film, but
due to specific features of the analyzed LT orientation, BW1 is nearly sagittally polarized
wave and BW2 is nearly pure SH wave. BW2 penetrates deeper into SiO
2
film than into LT
substrate. A numerical analysis reveals that SiO
2
thickness about 1.5 is sufficient for
transformation of SAW into the boundary wave.

(a) (b)

h
SiO2
=3

Cu

LT
p=/2

Fig. 12. Acoustic fields associated with two higher-order modes propagating in 42YX LT
with Cu grating and SiO
2
film when h
SiO2
=3. (a) 1
st
mode; (b) 2
nd
mode
The acoustic fields associated with propagation of the two higher-order modes (Fig.10,b)
have been also investigated. These modes have leaky wave nature. Fig.12 illustrates the
structure of these modes at h
SiO2
=3. The first mode, which exists when h
SiO2
>0.7, has SH
polarization deeply penetrating into SiO
2
(Fig.12,a). With increasing SiO
2
thickness, this
mode degenerates into the SH BAW propagating in SiO
2
. The second mode, which exists at
h
SiO2
>1.6, looks as a combination of SH-type SAW in LT substrate with Cu grating and
sagittally polarized quasi-bulk wave propagating in SiO
2
(Fig.12,b), with the amplitude of
SH polarization component much higher than the amplitudes of two other components.
This example demonstrates the effect of anisotropy on the propagation of acoustic waves in

439
multilayered structures and reveals another application of the numerical technique
described in this chapter. Similar investigation can be performed for other structures of
practical importance or serve to study the wave processes in multilayered structures.
6. Conclusion
In this chapter, some layered and multilayered structures, which look promising as
substrates for modern SAW devices developed for applications in cellular phones,
communication and navigation systems have been overviewed. A universal numerical
technique, which enables fast and accurate analysis of these and other structures have been
presented and, by way of example, applied to some multilayered structures of practical
interest. The physical insight into the mechanisms of SAW transformation with increasing
film thickness in a multilayered structure was provided via simulation of acoustic fields in
one of the structures.
7. References
Abbott B.P., Naumenko N.F. & Caron J. (2005). Characterization of bonded wafer for RF
filters with reduced TCF, Proceedings of IEEE Ultrasonics Symposium, pp.926-929,
Rotterdam, the Netherlands, Sept. 2005.
Adler E. L. (1990). Matrix methods applied to acoustic waves in multi-layers, IEEE Trans.
Ultrason., Ferroelectr., Freq. Control, Vol. 37, No 6, pp. 485-490.
Ballandras S., Reinhardt A., Laude L., Soufyane A., Camou S. & Ventura P. (2004).
Simulations of surface acoustic wave devices built on stratified media using mixed
finite element/boundary integral formulation, J. Appl. Phys., Vol. 96, No 12,
pp.7731-7741.
Benetti M., Cannatta D., Pietrantonio F. D. & Verona E. (2005). Growth of AlN piezoelectric
film on diamond for high-frequency surface acoustic wave devices, IEEE Trans.
Ultrason., Ferroelectr., Freq. Control, Vol. 52 , pp. 18061811.
Benetti M., Cannata D., Di Pietrantonio F., Verona E., Almaviva S., Prestopino G., Verona C.
& Verona-Rinati G. (2008). Surface acoustic wave devices on AlN/single-crystal
diamond for high frequency and high performances operation, Proceedings of IEEE
Ultrasonics Symposium, pp. 1924-1927, Beijing, China, Nov. 2008.
Blotekjear K., Ingebrigsten K. & Skeie H. (1973). A method for analyzing waves in structures
consisting of metal strips on dispersive media, IEEE Trans. Electronic Devices, vol.
20, no. 12, pp. 11331138.
Brizoual L. L., Sarry F. , Elmazria O. , Alnot P. , Ballandras S. & Pastureaud T. (2008). GHz
frequency ZnO/Si SAW device, IEEE Trans. Ultrason., Ferroelectr., Freq. Control, Vol.
55, No 2, pp. 442450.
Didenko I. S., Hickernell F. S. & Naumenko N. F. (2000). The theoretical and experimental
characterization of the SAW propagation properties for zinc oxide films on silicon
carbide, IEEE Trans. Ultrason., Ferroelectr., Freq. Control, Vol. 47, No. 1, pp.179-187.
Eda K., Onishi K., Sato H., Taguchi Y. & Tomita M. (2000). Direct bonding of piezoelectric
materials and its applications, Proceedings of IEEE Ultrasonics Symposium , pp. 299-
309, San Juan, Puerto Rico, Sep. 2000.


440
Emanetoglu N.W., Patounakis G., Muthukumar S. & Lu Y. (2000). Analysis of temperature
compensated modes in ZnO/SiO
2
/Si multilayer structures, Proceedings of IEEE
Ultrasonics Symposium , pp. 325-328, San Juan, Puerto Rico, Sep. 2000.
Endoh G., Hashimoto K. & Yamagouchi M. (1995). SAW propagation characteristics by
Finite Element Method and Spectral Domain Analysis, Jpn. J. Appl. Phys., Vol. 34,
No 5B, pp. 2638-2641.
Fujii S., Kawano S., Umeda T., Fujioka M. & Yoda M. (2008). Development of a 6GHz
resonator by using an AlN diamond structure, Proceedings of IEEE Ultrasonics
Symposium, pp. 1916-1919, Beijing, China, Nov. 2008.
Hashimoto, K. Y. (2000). Surface Acoustic Wave Devices in Telecommunications: Modeling and
Simulation, Springer, ISBN 354067232X, USA.
Hashimoto K., Omori T. & Yamaguchi M. (2007). Extended FEM/SDA Software for
Characterizing Surface Acoustic Wave Propagation in Multi-layered Structures,
Proceedings of IEEE Ultrasonics Symposium, pp. 711-714, Rome, Italy, Nov. 2008.
Hashimoto K., Wang Y., Omori T., Yamaguchi M., Kadota M., Kando H. & Shibahara T.
(2008). Piezoelectric Boundary Waves: Their Underlying Physics and Applications,
Proceedings of IEEE Ultrasonics Symposium, pp. 999-1005, New York, USA, Oct. 2007.
Ingebrigtsen K.A. (1969). Surface waves in piezoelectrics, J. Appl. Phys., Vol. 40, No 7,
pp.2681-2686.
Kadota M. & Minakata M. (1998). Piezoelectric Properties of ZnO Films on a Sapphire
Substrate Deposited by an RF-Magnetron-Mode ECR Sputtering System, Jpn. J.
Appl. Phys., Vol.37, pp.2923-2926.
Kadota M. (2007). High Performance and Miniature Surface Acoustic Wave Devices with
Excellent Temperature Stability Using High Density Metal Electrodes, Proceedings
of IEEE Ultrasonics Symposium, pp.496-506, New York, USA, Oct. 2007.
Kadota M., Nakao T., Murata T. & Matsuda K. (2008). Surface Acoustic Wave Filter in High
Frequency Range with Narrow Bandwidth and Excellent Temperature Property,
Proceedings of IEEE Ultrasonics Symposium, pp. 1584-1587, Beijing, China, Nov. 2008.
Kovacs G., Ruile W., Jakob M., Rosler U., Maier E., Knauer U. & Zottl H. (2004). A SAW
Duplexer with Superior Temperature Characteristics for US-PCS, Proceedings of
IEEE Ultrasonics Symposium, pp. 974-977, Montreal, Canada, Aug. 2004.
Milsom R. F., Reilly N.C.H. & Redwood M. (1977). Analysis of generation and detection of
surface and bulk acoustic waves by interdigital transducers, IEEE Trans. Sonics and
Ultrasonics, Vol. 24, pp. 147-166.
Nakahata H., Higaki K., Fujii S. , Hachigo A., Kitabayashi H. , Tanabe K., Seki Y. & Shikata
S. (1995). SAW Devices on Diamond, Proceedings of IEEE Ultrasonics Symposium,
pp.361-370, Seattle, USA, Nov. 1995.
Nakahata H., Hachigo A., Itakura K., Fujii S. & Shikata S. (2000). SAW Resonators on
SiO
2
/ZnO/Diamond Structure in GHz Range, Proceedings of IEEE Freq. Contr.
Symposium, pp. 315-320, Kansas City, USA, June 2000.
Nakai Y., Nakao T. , Nishiyama K. & Kadota M. (2008). Surface Acoustic Wave Duplexer
composed of SiO
2
film with Convex and Concave on Cu-electrodes/LiNbO3
Structure, Proceedings of IEEE Ultrasonics Symposium, pp.1580-1583, Beijing, China,
Nov. 2008.

441
Nakanishi H., Nakamura H. , Hamaoka Y., Kamiguchi H. & Iwasaki Y. (2008). Small-sized
SAW Duplexers with Wide Duplex Gap on a SiO
2
/Al/LiNbO
3
structure by using
Novel Rayleigh-mode Spurious Suppression Technique, Proceedings of IEEE
Naumenko N. F. (1996). Application of exceptional wave theory to materials used in surface
acoustic wave devices", J. Appl. Phys., Vol.79, pp. 8936-8943.
Naumenko N.F. & Didenko I.S. (1999). High-velocity surface acoustic waves in diamond
and sapphire with zinc oxide film, Appl. Phys. Lett., Vol. 75, No 19, pp. 3029-3031.
Naumenko N.F. & Abbott B.P. (2003). Surface Acoustic Wave Devices Using Optimized
Cuts of a Piezoelectric substrate, US Patent #6,556,104 B2. Int.Cl. H03H 9/64.
Apr.29, 2003
Naumenko N.F. & Abbott B.P. (2004). Surface Acoustic Wave Filter , US Patent 6,833,774 B2.
Int.Cl. H03H 9/16. Dec. 21, 2004.
Naumenko N.F. (2009). Transformation of Surface Acoustic Waves into Boundary Waves in
Piezoelectric/Metal/Dielectric Structures, Proceedings of IEEE Ultrasonics
Symposium, pp.2635-2638, Rome, Italy, Sep. 2009.
Naumenko N.F. (2010). A Universal Technique for Analysis of Acoustic Waves in Periodic
Grating Sandwiched Between Multi-Layered Structures and Its Application to
Different Types of Waves, Proceedings of IEEE Ultrasonics Symposium, San Diego,
USA, Oct. 2010.
Naumenko N.F. & Abbott B.P. (2010). Insight into the Wave Transformation Mechanisms
via Numerical Simulation of Acoustic Fields in Piezoelectric/Metal Grating/
Dielectric Structures, Proceedings of IEEE Ultrasonics Symposium, San Diego, USA,
Oct. 2010.
Omori T., Kobayashi A., Takagi Y., Hashimoto K. & Yamaguchi M. (2008). Fabrication of
SHF range SAW devices on AlN/Diamond-substrate, Proceedings of IEEE
Reinhardt A., Pastureaud T., Ballandras S. & Laude L. (2003). Scattering matrix method for
modeling acoustic waves in piezoelectric, fluid, and metallic multilayers, J. Appl.
Phys., Vol. 94, No 10, pp.6923-6931.
Ruppel C.C.W. & Fjeldy T. A. Eds. (2001). Advances in Surface Acoustic Wave Technology,
Systems and Applications. Volume 2 . ISBN 981-02-4538-6. World Scientific Publishing
Co. Pte Ltd, Singapore.
Ruppel, C. C. W., Reindl, L. & Weigel, R. (2002). SAW devices and their wireless
communication applications, IEEE Microwave Magazine, ISSN 1527-3342 3(2): pp.
6571.
Stroh A. N. (1962). Steady state problems in anisotropic elasticity, J. Math. Phys. Vol. 41., pp.
77-103.
Tan E. L. (2002). A robust formulation of SAW Green's functions for arbitrarily thick
multilayeres at high frequencies, IEEE Trans.Ultrason., Ferroelectr., Freq. Control, Vol.
49, No 7, pp. 929-936.
Tsutsumi J. , Inoue S., Iwamoto Y., Miura M., Matsuda T., Satoh Y., Nishizawa T., Ueda M.
& Ikata O. (2004). A Miniaturized 3x3-mm SAW Antenna Duplexer for the US-PCS
band with Temperature-Compensated LiTaO
3
/Sapphire Substrate, Proceedings of
IEEE Ultrasonics Symposium, pp. 954-958, Montreal, Canada, Aug. 2004.


442
Viktorov, I.A. (1967). Rayleigh and Lamb Waves: physical theory and applications. Plenum Press,
New York.
Viktorov I.A. (1981). Surface sound waves in solids. Moscow: Nauka (in Russian).
Zhang Y., Desbois J. & Boyer L. (1993). Characteristic parameters of surface acoustic waves
in a periodic metal grating on a piezoelectric substrate, IEEE Trans.Ultrason.,
Ferroelectr., Freq. Control, vol. 40, pp. 183-192, May 1993.
20
SAW Parameters Analysis and
Equivalent Circuit of SAW Device
Trang Hoang
Faculty of Electrical-Electronics Engineering, University of Technology, HoChiMinh City
VietNam
1. Introduction
Surface Acoustic Wave (SAW) devices, using interdigital electrodes, play a key role in
todays telecommunication systems and are widely used as electronic filters, resonators,
delay lines, convolvers or wireless identification systems (ID tags).
During the last three decades, demands set by the expansion of the telecommunication
industry and many applications in sensor have resulted in the introduction of a new
generation of the SAW devices. Consequently, the design of high performance SAW devices
requires precise and efficient models, simulation tools. Several methods have been proposed
for modeling, analyzing SAW devices. These include the impulse model, the equivalent
circuit models, the coupling-of-mode (COM) model, P-matrix model, angular spectrum of
waves models [1] and the Scattering Matrix approach that was presented by Coldren and
Rosenberg [2]. While the impulse model is only a first order model, the other models
include second order effects, e.g. reflections, dispersion, and charge distribution effects.
Purely numerical methods have also been and are being developed by many authors [3]-
[35], [41].
In this chapter, the method for calculating the SAW parameters, including modeling and
simulation, is given.
Section 2 gives the calculation of SAW properties and analyses of different SAW device
structures.
Section 3 presents the equivalent circuit of SAW delay line based on Mason model.
The equivalent circuit of SAW delay line based on Couple-Of-Mode theory is presented in
section 4.
Based on section 3 and 4, section 5 shows comparison between using the equivalent circuit
of SAW delay line device based on Mason model and COM theory. This model is useful and
fast model for designing the SAW device.
2. Calculation of SAW parameters
2.1 SAW parameters
The most important parameter for SAW device design is the center frequency, which is
determined by the period of the IDT fingers and the acoustic velocity. The governing
equation that determines the operation frequency is:


444
f
0
= v
SAW
/ (1)
where
is the wavelength, determined by the periodicity of the IDT and v
SAW
is the acoustic wave
velocity . For the technology being used in this research:
= p = finger width 4 (2)
with the finger width (as shown in Figure 1) is determined by the design rule of the
technology which sets the minimum metal to metal distance.
v
SAW
is surface acoustic wave velocity.

Aperature (W)
period ()
Finger
width
Finger
spacing

Fig. 1. IDT parameters
By using matrix method or Finite Element Method (FEM) in section 2.2, the velocity v of
acoustic wave is derived in two cases:
- Wave velocity V
0
is velocity in case of free surface.
- Wave velocity Vs is velocity in case of short-circuit surface.
Therefore, the electromechanical coupling coefficient K is calculated approximately by
Ingebrigtsen [54] as:

2
2
o s
o
V V
K
V
= (3)
By using the matrix method or FEM and approximation of coupling factor as in (3), the SAW
parameters in different structures AlN/Si, AlN/SiO
2
/Si and AlN/Mo/Si are calculated and
analysed in three next sections.
2.2 Matrix method and Finite Element Method (FEM). The choice between them
Matrix method
The SAW propagation properties on one layer or multilayer structure are obtained by using
matrix approach, proposed by J.J.Campbell and W.R.Jones [50], K.A.Ingebrigtsen [54], and
then developed by Fahmy and Adler [31], [32], [33] and other authors [51], [52], [53]. The
numerical solution method is based on characterizing each layer by means of a transfer
matrix relating the mechanical and electrical field variables at the boundary planes. The
boundary conditions for multilayer are based on the mechanical and electrical field
variables those quantities that must be continuous at material interfaces. This matrix method
is used to calculate the wave velocity and therefore, the electromechanical coupling factor. A
general view and detail of this approach are given as follows and also presented in [50]-[53].

SAW Parameters Analysis and Equivalent Circuit of SAW Device

445

substrate
1
st
layer
2
nd
layer

x
x
1
x
3
x
2
h
1
h
2

Fig. 2. Multilayer structure
Constitutive equations:

2 2 2
2
l i
ijkl kij
j k j k
u u
c e
x x x x t

+ =

(4)

2 2
0
l
jkl jk
j k j k
u
e
x x x x

=

(5)
where
, , ,
ijkl ijk jk
c e are elastic tensor, piezoelectric tensor, dielectric tensor and mass density,
respectively, of the considered material.
U is the particle displacement.
is the scalar electric potential.
The boundary conditions are shown in Table 1

Position Mechanical conditions Electrical conditions
x
3
=0 1
st
S
i i
U U =

1
3 3
st
S
i i
T T =
Boundary is open
1
st
S
= ,
1
st
S
D D =
Boundary is short
1
0
st
S
= =
x
3
=h
1
1 2
st nd
i i
U U =
1 2
3 3
st nd
i i
T T =
Boundary is open
1
2
st
nd
= ,
1 2
st nd
D D =
Boundary is short
1
2
0
st
nd
= =
x
3
=h
1
+ h
2
2
3
0
nd
i
T =
Boundary is open
2 2
. .
nd nd
D k =
Boundary is short
2
0
nd
=
Table 1. Boundary conditions


446
where D: electronic displacement,

k
k
D
x

=
(6)
The general solution for U
l
and (1) and (2) may be written as follows:

( ) ( )
3 1
1
exp[ ( )]
n
m m
l m l
m
U C A ik b x x vt
=
= +
(7)
where l =1, 2, 3

( ) ( )
4 3 1
1
exp[ ( )]
n
m m
m
m
C A ik b x x vt
=
= +
(8)
The coefficients C
m
are determined from boundary conditions.
By substituting (7) and (8) in every layer into the boundary conditions, we have general
form
[ ]
1
4
1
1
1
8
2
1
2
8
...
...
0
...
st
st
nd
nd
S
S
C
C
C
H
C
C
C
(
(
(
(
(
(
(
=
(
(
(
(
(
(

(9)
Phase velocity is determined from the condition:
Det(H)=0 (10)
(use approximation to solve (10))
Figure 3 shows the wave velocity of structure AlN/SiO
2
(1.3m)/Si(4m).

Fig. 3. Wave velocity in structure AlN/SiO
2
(1.3m)/Si(4m) with different thicknesses of AlN


447
Finite Element Method (FEM)
In the design procedure of SAW devices, simple models like Equivalent Circuit Model
coming from Smith Model and COM Model as presented above are used to achieve short
calculation time and to get a general view of response of SAW devices. They are a good
approach for designing SAW devices, for getting the frequency response, impedance
parameters and transfer characteristics of SAW device. They could allow the designer to
determine the major dimensions and parameters in number of fingers, finger width, and
aperture. However, they are subjected to some simplifications and restrictions.
Field theory is the most appropriate theory for the design SAW devices as it involves the
resolution of all the partial differential equations for a given excitation. The Finite Element
Model (FEM) is the most appropriate numerical representation of field theory where the
piezoelectric behaviour of the SAW devices can be discretized [45], [46]. Besides, nowadays,
FEM tools also provide 3D view for SAW device, such as COMSOL [47], Coventor [48],
ANSYS [49].
The typical SAW devices can include a lot of electrodes (hundreds or even thousands of
electrodes). In fact, we would like to include as many IDT finger pairs as possible in our
FEM simulations. This would however significantly increase the scale of the device.
Typically finite element models of SAW devices require a minimum of 20 mesh elements
per wavelength to ensure proper convergence. A conventional two-port SAW devices
consisting of interdigital transducers (IDT) may have especially on substrate materials
with low piezoelectric coupling constants - a length of thousands of wavelengths and an
aperture of hundred wavelengths. Depending on the working frequency, the substrate which
carries the electrode also has a depth of up to one hundred wavelengths. Taking into account
that FEM requires a spatial discretization with at least twenty first order finite elements per
wavelength and that an arbitrary piezoelectric material has at least four degrees of freedom,
this leads to 8 x 10
8
unknowns in the three dimensional (3-D) case. Hence, the 3-D FEM
representation of SAW device with hundreds of IDT fingers would require several million
elements and nodes. The computational cost to simulate such a device is extremely high, or the
amount of elements could not be handled by nowadays computer resources.
Fortunately, SAW devices consist of periodic section. M.Hofer et al proposed the Periodic
Boundary Condition (PBC) in the FEM that allows the reduction of size of FE model
tremendously [45], [46].
A good agreement between FEM and analytic method is obtained via the results in case of
SAW with AlN thickness of 4m, wavelength of 8m presented in Table 2.

Matrix method FEM Difference between Matrix
method and FEM (%)
f
0
(MHz) 771.13 775.48 0.56
f
s
(MHz) 770.26 774.57 0.56
v
0
(m/s) 6169.02 6203.87 0.56
v
s
(m/s) 6162.07 6196.54 0.56
K (%) 4.74 4.86 2.4
Table 2. Comparison between matrix method and FEM
From this table, matrix method and FEM give the same results. However, FEM would takes
a long time and require a trial and error to find the results. Consequently, to reduce time,
the matrix method proposed to be used to extract the parameters of SAW devices; FEM is


448
used to get a 3D view and explain some results that can not be explained by equivalent
circuit. This point will be presented in next sections.
The three next sections present and analyse SAW parameters in different structures AlN/Si,
AlN/SiO
2
/Si and AlN/Mo/Si.
2.3 Wave velocity, coupling factor in AlN/Si structure
Figure 4 shows the dependence of Rayleigh wave velocity V
0
and Vs on the normalized
thickness as respect to the wavelength, khAlN of AlN layer in SAW device AlN/Si
substrate, where normalized thickness is defined by:

2 h
kh

= (11)
In this graph, when the normalized thickness of AlN, khAlN is larger than 3, the wave
velocity reaches the velocity of the Rayleigh wave in AlN substrate v
(AlN substrate)
=6169 (m/s).
This could be explained that the wave travels principally in AlN layer when khAlN is larger
than 3, because for low frequency the wave penetrates inside the other layer and this work is
in the case where the wave are dispersive. It is better to be in the frequency range where the
Rayleigh wave is obtained to have a constant velocity.

Fig. 4. Calculated values of wave velocity V
0
and Vs in SAW device AlN/Si substrate
depend on the normalized thickness khAlN of AlN layer

Fig. 5. Calculated values of coupling factor K(%) in SAW device AlN/Si substrate depends
on the normalized thickness khAlN of AlN layer


449
The coupling factor K for this kind of device is shown in Figure 5. When normalized
thickness of AlN layer is larger than 3, the coupling factor K still remain at 4.74% by that the
wave travels principally in AlN layer.
In this configuration, K is at its maximum value of 5.34% when khAlN=0.55.
2.4 Wave velocity, coupling factor in AlN/SiO
2
/Si structure
Wave velocity and coupling factor in structure AlN/SiO
2
/Si are also presented in Figure 6
and Figure 7, respectively.

Fig. 6. Dependence of wave velocity in SAW device AlN/SiO
2
/Si substrate on the
normalized thickness khAlN of AlN layer and khSiO
2

Fig. 7. Dependence of coupling factor K(%) in SAW device AlN/ SiO
2
/Si substrate on the
normalized thickness khAlN of AlN layer and khSiO
2

In this configuration, as results in Figure 6, when 6 khAlN < , with the same thickness of
AlN layer, an increase in thickness of SiO
2
would decrease the wave velocity. When
6 khAlN > , the wave velocity reaches the velocity of the Rayleigh wave in AlN substrate
v(AlN substrate)=6169 (m/s). A same conclusion is formulated also for coupling factor for this
kind of structure, AlN/SiO
2
/Si, in Figure 7; when 6 khAlN > , K remains at the value of 4.7%.


450
To understand the above behavior, we use FEM method to display displacement profile
along the depth of multilayer AlN/SiO
2
/Si. These results obtained from FEM method in
case of khSiO
2
=0.7854, khAlN=5 and khAlN=0.2 are compared as in Figure 8.

Fig. 8. Displacement profile along the depth of the multilayer AlN/SiO
2
/Si, khSiO
2
=0.7854
From Figure 8, we note that wave travels principally in AlN layer for a khAlN value of 5. By
this reason, from a khAlN value of larger than 5, the coupling factor K remains at 4.7% and
wave velocity remains at 6169m/s. For khAlN=0.2, wave travels principally in SiO
2
layer
and Si substrate that are not piezoelectric layer. Consequently, the coupling factor K reaches
the value of 0%.
In conclusion, the values of wave velocity and coupling factor depend on wave propagation
medium, in which constant values of wave velocity and coupling factor indicate a large
contribution of AlN layer, and coupling factor value of near 0% indicates a large
contribution of SiO
2
layer and Si substrate.
2.5 Wave velocity, coupling factor in AlN/Mo/Si structure
For our devices, a thin Mo layer will be also deposited below the AlN layer to impose the
crystal orientation of AlN. Besides this dependence, the Mo layer also has influences on

Fig. 9. Wave velocity AlN/Mo/Si substrate depends on the normalized thickness khAlN
and khMo


451
wave velocity and coupling factor K. These influences are shown in Figure 9 and Figure 10,
respectively.

Fig. 10. Coupling factor K(%) in SAW device AlN/Mo/Si substrate depends on the
normalized thickness khAlN and khMo
From Figure 9, the use of Mo layer would increase the wave velocity with any thickness of
AlN layer and Mo layer. In case of coupling factor K as in Figure 10, the Mo layer, however,
could decrease K when the khAlN is less than 1.02. When the normalized thickness of AlN
layer khAlN is in the range from 1.17 to 2.7, the Mo layer would increase the coupling factor
K. And when the khAlN is larger than 2.7, the Mo has no influence on wave velocity and
coupling factor. The reason of this effect could be explained by the displacement profile in
AlN/Mo/Si structure, as shown in Figure 11 for thickness AlN value of khAlN=2.7. We
could note that when 2.7 khAlN , the first interesting point is that the wave travels
principally in AlN layer and Si substrate, the second one is that the relative displacement
U/Umax in Mo layer will be smaller than 0.5. These points would explain the reason why
when 2.7 khAlN the use of Mo has no influence on wave velocity and coupling factor.

Fig. 11. Displacement profile along the depth of the multilayer AlN/Mo/Si, khAlN=2.7
1.02


452
3. Equivalent circuit for SAW delay line based on Mason model
3.1 Why Equivalent Circuit model is chosen?
Actual devices exist in a three-dimensional physical continuum. Their behaviour is
governed by the laws of physics, chemistry, biology, and electronics. From a general point
of view, the analysis of devices can be carried out by using some equations of laws of
physics, chemistry For example; the analysis of piezoelectric resonators or transducers
and their application to ultrasonic system can be solved by using the wave equation
[36],[37]. But through analysis, equivalent electrical circuit representations of devices can be
extracted. So, they can be readily expressible with Equivalent Electric Circuit. Below is the
presentation of advantages and disadvantages of equivalent circuit.
Advantages:
- There are an immensely powerful set of intellectual tools to understand electric circuits.
- The equivalent circuit approach has distinct advantages over the direct physical,
chemical equations approach (such as direct wave equations approach).
- Many theories, problems of electric circuits have already been solved such as
microwave network theory, integrated circuit etc.
- Electric circuit approach is intrinsically correct from an energy point of view [56].
- A further advantage of electric circuit model is that it permits efficient modelling of the
interaction between the electric and non-electric components of a microsystem. Both the
electrical and mechanical portions of a system are represented by the same means. With
software like Simulink, the block diagram is easily constructed, easily to build a more
complex system but when we would like to connect a mechanical element to electrical
circuits, Simulink can not do that. The analogies between electrical and mechanical
elements are presented clearly by Warren P.Mason [57], [58].
Disadvantages:
- Care must be taken to make sure whether the boundary conditions are compatible with
those used in the original derivation of the equivalent circuit [58].
In many systems, both commercial and industrial, pressure measurement plays a key role.
Since pressure is a normal stress (force per unit area), pressure measurement can be done by
using piezoelectric material which can convert stress into voltage. Equivalent circuits such
as Masons model [36] provide a powerful tool for the analysis and simulation of
piezoelectric transducer elements. Most of the analogous circuits which have appeared in
the literature implement transducers as the circuit elements. This model simulates both the
coupling between the mechanical and electrical systems and the coupling between the
mechanical and acoustical systems [39]. The mechanical, electrical and acoustic parts of
piezoelectric transducer can be varied and analysed about behaviour by implementing
equivalent circuits on computer tools such as Ansoft, Spice, ADS, etc. For IDT composing
of N periodic sections, Smith et al [41] developed the equivalent circuit model based on
Berlincourt et al [40] work about equivalent circuit for Length Expander Bar with parallel
electric field and with perpendicular electric field and based on the equivalent circuit for
electromechanical transducer presented by Mason [36]. Smith model henceforth will be
used to indicate this model. From this model, some models for SAW device in literature
have been implemented. However, these models would include only IDTs [42], [43]. In SAW
pressure sensor, one of sensitive parts is propagation path. It should be included in the
model. The hybrid model based on Smith model for SAW pressure sensor which includes
the IDTs and propagation path have been constructed.


453
Another equivalent model is based on the Coupling-Of-Modes (COM) theory. An excellent
recent review of COM theory used in SAW devices was written by K.Hashimoto [10]. Based
on the COM equations, as the force and voltage analogy can be used, the relationships
between the terminal quantities at the one electrical port and two acoustic ports for an IDT
have been done. K.Nakamura [44] introduced a simple equivalent circuit for IDT based on
COM approach that is developed in section 4.
In conclusion, the equivalent-circuit model is chosen because it can allow fast design. This
allows the designer to determine the major dimensions and parameters in number of
fingers, fingers width, aperture, delay line distance, frequency response, impedance
parameters and transfer characteristics of SAW device.
3.2 Equivalent circuit for IDT including N periodic sections
Based on Berlincourt et al [39] about equivalent circuit for Length Expander Bar with
parallel electric field and with perpendicular electric field and based on the equivalent
circuit for electromechanical transducer presented by Mason [36], Smith and al [41] have
developed the equivalent circuit for IDT composed of N periodic sections of the form shown
in Figure 12.

Fig. 12. Interdigital transducer diagram
One periodic section as shown in Figure 13 (a) can be presented by analogous one-
dimensional configurations: crossed-field model as in Figure 13 (b), and in-line model
as in Figure 13 (c). In crossed-field model, the applied electric field is normal to the
acoustic propagation vector; while in in-line field model, the electric field is parallel to the
propagation vector.
The important advantage of two one-dimensional models is that each periodic section can
be represented by equivalent circuit of Mason, as shown in Figure 14 for crossed-field
model and Figure 15 for in-line field model. The difference between these two equivalent
circuits is that in crossed-field model, the negative capacitors are short-circuited.
Where:

0
4 2

= = (12)


454
With periodic section transit angle
0
2

=

0 2
0
2
s
R
C k
= (13)
R
0
is electrical equivalent of mechanical impedance Z
0
[59]
k: electromechanical coupling coefficient
C
0
=Cs/2 with Cs : electrode capacitance per section
0
is center angular frequency

Fig. 13. Side view of the interdigital transducer and 2 analogous one-dimensional
configurations (a) Actual model, (b) crossed-field model, (c) in-line field model

Fig. 14. Mason equivalent circuit for one periodic section in crossed-field model
(a)
(b)
(c)


455

Fig. 15. Mason equivalent circuit for one periodic section in in-line field model
One periodic section can be represented by the 3-port network [y] matrix. The [y] matrix of
one periodic section for 2 models as follows (see Appendix, section Appendix 1), with
G
0
=R
0
-1
, R
0
is expressd by (13):
- for the crossed-field model:

11 0
0
12
13 0
33 0 0
cot (4 )
sin(4 )
(2 4 )
y jG g
jG
y
y jG tg
y j C G tg

=
=
=
= +
(14)
- for the in-line field model:

2
0
0 0
11 0 2
0
0
0
2
0
0
12 0
0 0
0 0
13 0
0
0
0
33
0
1
sin(2 )
cot cot (2 ) 2
cot (2 )
1
cot
sin(2 )
2
2 cot cot (2 )
2
1
2
2
1
G
C G
y jG g g
C
G
g
C
G
g
C
y jG
G G
g g
C C
tg
y jG
G
tg
C
j C
y
G

(
| |
(
|
| |
(
\ .
=
|
(
| | \ .
(
|
(
\ .
| |
|
\ .
=
| || |

| |
\ .\ .
=
0
tg
C

(15)
In IDT including N periodic sections, the N periodic sections are connected acoustically in
cascade and electrically in parallel as represented in Figure 16.


456

Fig. 16. IDT including the N periodic sections connected acoustically in cascade and
electrically in parallel
Matrix [Y] representation of N-section IDT for two models, crossed-field model and in-
line model are in (16) and (17), respectively (the calculation development is presented in
Appendix, section Appendix 1):
- In "crossed-field" model:

11 0
0
12
13 0
33 0 0
cot (4 )
sin(4 )
(2 4 )
Y jG g N
jG
Y
N
Y jG tg
Y jN C G tg

=
=
=
= +
(16)
- In "in-line field" model:

11
11
12
12
12
13 0
0
0
0
33
0
0
1
2
1
2
2
1
Q
Y
Q
Y
Q
tg
Y jG
G
tg
C
j NC
Y
G
tg
C
=
=
=
(17)
It was shown in the literature that the crossed field model yielded better agreement than the
experiment when compared to the in-line model when K is small. In section 2, K is always
smaller than 7.2%. Besides, in section stated above, the crossed-field model is simpler than
in-line field model in term of equations of all element of [Y] matrix. Consequently, the
crossed-field model is selected henceforth for the calculating, modeling the devices.
3.3 Equivalent circuit for propagation path
The delay line SAW device could be used for pressure sensor application. The sensitive part
of this kind of device will be the propagation path. To model the pressure sensor using


457
SAW, it is necessary to construct the model for propagation path. Based on the equivalent
circuit for electromechanical transducer presented by Mason [36], equivalent circuit of
propagation path is presented as in Figure 17.

Fig. 17. Equivalent circuit of propagation path, based on Mason model
Where

2
l
v
= (18)
with v is SAW velocity, l is propagation length.
3.4 Equivalent circuit for SAW delay line
Due to the piezoelectric effect, an RF signal applied at input IDT stimulates a micro-acoustic
wave propagating on its surface. These waves propagate in two directions, one to receiving
IDT and another to the medium. The approximations as follows are assumed to construct
the equivalent circuit for SAW delay line:
- Assume that the IDT radiates the wave into a medium of infinite extent.
Experimentally, an infinite medium is approximated either by using absorber, such as
wax, polyimide to provide acoustic termination, or by using a short RF pulse
measurement. The condition of infinite medium means that no wave reflects back to
input IDT. This is created for SAW device model by connecting the acoustic
characteristic admittance Y
0
to one terminal of IDT.
- Assume that the wave propagating to receiving IDT has no attenuation during
propagation way between two IDTs. So, the propagation path between two IDTs can be
expressed as the no-loss transmission line.
Based on these two approximations, the [Y] matrix representation of IDT in section 3.2, and
propagation path representation in section 3.3, the SAW delay line can be expressed as
equivalent circuit as in Figure 18.

Fig. 18. Equivalent circuit of SAW delay line, based on Mason model


458
Section 3 gives the equivalent circuit of SAW delay line, including IDT input, IDT output
and propagation path. All of calculation developments are presented in appendix, section 2.
In this appendix, a new equivalent circuit of IDT including N periodic section plus one
finger, which we call it N+1/2, also are developed and presented. Another representation
of SAW delay line is [ABCD] matrix representation which also proposed in appendix,
section Appendix 4. [ABCD] matrix representation has one interesting property that in
cascaded network, the [ABCD] matrix of total network can be obtained easily by
multiplying the matrices of elemental networks.
4. Equivalent circuit for IDT based on the Coupling-Of-Mode theory
The Coupling-Of-Modes formalism is a branch of the highly developed theory of wave
propagation in periodic structure, which has an history of more than 100 years. This theory
covers a variety of wave phenomena, including the diffraction of EM waves on periodic
gratings, their propagation in periodic waveguides and antennas, optical and ultrasonic
waves in multi-layered structures, quantum theory of electron states in metal,
semiconductors, and dielectrics. Theoretical aspects of the wave in periodic media and
applications were reviewed by C.Elachi [4], in which it included theories of waves in
unbounded and bounded periodic medium, boundary periodicity, source radiation in
periodic media, transients in periodic structures, active and passive periodic structures,
waves and particles in crystals. An excellent recent review of COM theory used in SAW
devices was written by K.Hashimoto [10].
A simple equivalent circuit for IDT based on COM approach was proposed by K.Nakamura
[29]. This model would be useful to analyze and design SAW devices. Based on the COM
equations, the relationships between the terminal quantities at the one electrical port and
two acoustic ports for an IDT have been done.
4.1 COM equation for particle velocities
Consider an IDT including N periodic sections with periodic length of L as shown in Figure 19.

Fig. 19. IDT including N periodic sections
The particle velocities v
+
(x) and v
-
(x) of the wave propagating in the +x and x directions in
the periodic structure can be expressed as follows with the time dependence exp(jt) term:
( ) ( )
jkx
v x A x e
+ +
= (19)
( ) ( )
jkx
v x A x e

= (20)


459
Where k is the

wave number

/
SAW
k V = (21)
The amplitude A
+
(x) and A
-
(x) obey the following coupled-mode equations [60]:

2
11 12
( )
( ) ( )
j x j x
dA x
jK A x jK e A x j e V
dx

+
+
= + (22)

2
12 11
( )
( ) ( )
j x j x
dA x
jK e A x jK A x j e V
dx

+
= + (23)
Where V is the voltage applied to the IDT,
is the constant associated with the convention from electrical to SAW quantities,
K
11
and K
12
are coupling coefficients, sum of the coupling coefficient coming from the
piezoelectric perturbation and that coming from the mechanical perturbation.

0
k k = , with
0
2
k
L
= (24)
The solution to (22) and (23) can be expressed as

( )
1 2 0
1 2
( )
j x j x jk x
v x h e ph e q V e

+
= + + (25)

( )
1 2 0
1 2
( )
j x j x jk x
v x ph e h e q V e

= + + (26)
Where the subscripts 1 and 2 indicate the elementary waves with wavenumbers
0 1
k + and
0 2
k + in the +x direction, and the magnitudes h
1
and h
2
, respectively.

2 2
1 2 11 12
, ( ) K K = + (27)

1 11
12
K
p
K

= (28)

11 12
1
q
K K
=
+ +
(29)
4.2 Equivalent circuit for IDT based on COM theory
From the equations (25) and (26), the particle velocities at the both ends of the IDT can be
expressed as:

1 2
(0) v h ph q V
+
= + + (30)

1 2
(0) v ph h q V
= + + (31)

( )
1 1
1 2
( )
j NL j NL
v NL e h e ph q V

+
= + + (32)


460

( )
1 1
1 2
( )
j NL j NL
v NL e ph e h q V

= + + (33)
The upper and lower signs in (32) and (33) correspond to the cases N=i and N=i+0.5,
respectively, where i is an integer. Consequently, the total particle velocities at the two
acoustical ports can be expressed as:
- Particle velocity at port 1 (x=0):

1 1 2
(0) (0) (1 )( ) 2 v v v p h h q V
+
= + = + + + (34)
- Particle velocity at port 2 (x=NL):

1 1
2 1 2
( ) ( ) (1 )( ) 2
j NL j NL
v v NL v NL p e h e h q V

+
( ( = + = + + +

(35)
The two forces at two acoustic ports are considered to be proportional to the difference of v
+

and v
-
. For the simplicity, these forces can be expressed as follows:

1 1 2
(0) (0) (1 )( ) F v v p h h
+
= = (36)

1 1
2 1 2
( ) ( ) (1 )( )
j NL j NL
F v NL v NL p e h e h
+
( = =

(37)
From these equations, h
1
and h
2
are the terms of F
1
and F
2
as follows:

1 1
1 1
2
1 1 2 2 2
(1 )( 1) (1 )( 1)
j NL j NL
j NL j NL
e e
h F F
p e p e

=

(38)

1
1 1
2 1 2 2 2
1
(1 )( 1) (1 )( 1)
j NL
j NL j NL
e
h F F
p e p e

=

(39)
The current I at the electrical ports can be expressed as:

1 2
1 1
1 2
0
1 2
1 2
(1 )( ) 2
(1 ) ( 1) ( 1) 2
NL
j x j x
s
j NL j NL
s
I p h e h e q V dx j NC V
h h
j p e e q NLV j NC V

( = + + + +

(

= + + + +
` (
)
}
(40)
where is the constant associated with the convention from SAW to electrical quantities,
therefore associated with the coupling factor K.
C
s
is the capacitance for one electrode pair.
By substituting equations (38) and (39) in (34), (35) and (40), the following equations can be
obtained:

1 2
(1 ) (1 )
( 2 )
(1 ) (1 )
s
p p
I j NC q NL V F F
j p j p

+ +
= + +

(41)

1 1 2
1 1 1 1
2
1 tan2 1 sin2
p p
v q V F F
p j p j

+ +
= +

(42)


461

2 1 2
1 1 1 1
2
1 sin2 1 tan2
p p
v q V F F
p j p j

+ +
= +

(43)
Where
/2 NL = (44)

1 2
= (45)
From these equations, the matrix as follows can be obtained:

1 1
2 2
(1 ) (1 )
( 2 )
(1 ) (1 )
1 1 1 1
2
1 tan2 1 sin2
1 1 1 1
2
1 sin2 1 tan2
s
p p
j NC q NL
j p j p
I V
p p
v q F
p j p j
v F
p p
q
p j p j

( + +
+
(

(
( (
(
+ +
( (
= (
( (

(
( (

(
+ +
(

(

(46)
In the acoustic wave transducer using piezoelectric effect, the force and voltage analogy can
be used. Therefore, the COM-based circuit of IDT as matrix in (46) can be considered as the
reciprocal circuit. The reciprocity theorem states that if a voltage source E acting in one
branch of a network causes a current I to flow in another branch of the network, then the
same voltage source E acting in the second branch would cause an identical current I to flow
in the first branch. By using this theorem in this case, replacing V and F
1
together, the same
value I requirement leads the following equations:

(1 )
2
(1 )
p
q
j p
+
=
(47)
2 j = (48)
From (46), (47), and (48), the matrix as in (46) becomes:

2
0 0 0
1 1
0 0 0
2 2
0 0 0
2 2 2
1 1
2 tan2 sin2
1 1
2 sin2 tan2
T
j C
j Z j Z j Z
I V
v F
j Z jZ jZ
v F
j Z jZ jZ

(
+
(
(
( (
(
( (
=
(
( (
(
( (

(
(
(

(49)
Where

0
1 1
1
p
Z
p q
= =
+
(50)
2 NL j NL = = (51)

T s
C NC = (52)


462
Consequently, the simple equivalent circuit obtained for IDT with N electrode pairs is
shown in Figure 20:

Fig. 20. Equivalent circuit IDT based on COM theory
4.3 Equivalent circuit for propagation path based on COM theory
In SAW devices, the propagation path should be taken into account. It is necessary to
determine the equivalent circuit for a propagation path of distance l between 2 IDTs. This
propagation path is a uniform section of length l, with a free surface or a uniformly
metallized surface. In this case, K
11
=K
12
=0, and =.
Consequently, from equations (38) and (39), h
1
and h
2
can be expressed as:

2
1 1 2 2 2
1 1
j l j l
j l j l
e e
h F F
e e

=

(53)

2 1 2 2 2
1
1 1
j l
j l j l
e
h F F
e e

=

(54)
And, the particle velocities are expressed as:

0 1
( )
1 1
( )
j k x jkx
v x h e h e
+ +
= = (55)

0 1
( )
2 2
( )
j k x jkx
v x h e h e
+
= = (56)
If the v
1
, v
2
, F
1
, and F
2
are defined as:

1
(0) (0) v v v
+
= + (57)

2
[ ( ) ( )] v v NL v NL
+
= + (58)

1
(0) (0) F v v
+
= (59)


463

2
( ) ( ) F v NL v NL
+
= (60)
Then, by expressing h
1
and h
2
in terms of F
1
and F
2
based on equations (53) and (54), the v
1

and v
2
become as follows:

2
1 1 2 1 2 2 2
1 2
1 1
j l j l
j l j l
e e
v h h F F
e e

+
= + =

(61)

2
2 1 2 1 2 2 2
2 1
1 1
j l j l
jkl jkl
j l j l
e e
v h e h e F F
e e

+
= + =

(62)
Using the relation between complex number and trigonometry, the v
1
and v
2
can be
expressed as follows:

1 1 2 ' ' ' '
0 0
1 1
tan2 sin2
v F F
jZ jZ
= (63)

2 1 2 ' ' ' '
0 0
1 1
sin2 tan2
v F F
jZ jZ
= + (64)
Consequently, the equivalent circuit for propagation path can be represented by the -
circuit of Figure 21:

Fig. 21. Equivalent circuit of propagation path based on COM theory
Based on Mason model, the equivalent circuit of propagation path was presented in Figure
17, which has star form. In Figure 21, the circuit has triangle form. By using triangles and
stars transformation theory published by A.E. Kennelly, equivalent circuit of propagation in
these two figures is the same. Consequently, the approachs that are based on Mason model
and COM theory can get the same equivalent circuit of propagation path.
4.4 Equivalent circuit for SAW delay line based on COM theory
Based on section 4.2 and 4.3, equivalent circuit of SAW delay line based on COM theory is
presented in Figure 22.
In this model, some parameters must to be calculated or extracted. SAW velocity v,
piezoelectric coupling factor K could be calculated from section 2. The periodic length L (or
wavelength ) is determined by design and fabrication.
The parameters K
11
and K
12
are coupling coefficients. They are sum of the coupling
coefficient coming from the piezoelectric perturbation and that coming from the mechanical
perturbation, and their equations for calculation are complicated [55]. Exact equations for


464
K
11
and K
12
were given by Y.Suzuki et al [55], but it seems so complex that their usefulnesses
could be limited. However, from this work of Y.Suzuki et al [55], we propose the K
11
and K
12

could be expressed as follows:
K
11
= O
11
K
2
k
0
(65)
K
12
= O
12
K
2
k
0
(66)
Where k
0
is stated by (24)
0
2
k
L
= and K is piezoelectric coupling factor.

O
11
is so-called self-coupling constant of finger, and O
12
is so-called coupling constant
between fingers. O
12
could also presents the reflective wave between two fingers.

Fig. 22. Equivalent circuit of SAW delay line based on COM theory

Fig. 23. Effect of O
12
on S21(dB), N=50, v
SAW
=5120m/s, =8m, K=0.066453, O
11
=0
Figure 23 shows the effects of O
12
on S21(dB) of SAW device N=50, v
SAW
=5120m/s, =8m,
K=0.066453 when O
11
=0. S
21
is the transmission coefficient in the scattering matrix
representation [28].
(a)


465
Figure 24 shows the effects of O
11
on S21(dB) of SAW device N=50, v
SAW
=5120m/s, =8m,
K=0.066453 when O
12
=0. So, O
11
coefficient shifts the center frequency of SAW device, the
positive value of O
11
reduces the center frequency f
0
of device, the negative on will increase the f
0
.

Fig. 24. Effect of O
11
on S21(dB), N=50, v
SAW
=5120m/s, =8m, K=0.066453, O
12
=0
The effect of K
11
and K
12
could be explained by their measurement method [61]. K
11
could be
derived from the measurement of frequency response, therefore the usefulness of its
calculation could be limited. Meanwhile, K
12
can be extracted from FEM. It is shown in
literature that K
12
depends on the thickness of finger with respect to the wavelength. In our
work, the ratio thickness/wavelength (its maximum value is 300nm/8m) is too small that
its effect can be ignored. In conclusion, in our work, value of K
11
and K
12
are 0.
5. Comparison of equivalent circuit of SAW device based on Mason model
and COM thoery
Figure 25 presents the comparison between hybrid model and COM model in that
O
11
= O
12
=0, distance between 2 IDTs is 50. These models could be the same, except that a

1.58 dB

Fig. 25. Comparison between Hydrid model and COM model (O
11
=O
12
=0)


466
small difference in the peak value of S21 (dB) occurs. This difference could be explained by
using crossed-filed model instead of actual model as in Figure 13.
6. Conclusion
The model used for SAW pressure sensor based on delay line are presented. For usefulness
and reduction of time in design process, the equivalent circuit based on COM model, in
which K
11
, K
12
=0 is proposed to be used.
Acoustic wave properties in different structures of AlN/SiO
2
/Si, AlN/Si, and AlN/Mo/Si
are analyzed. The wave velocity, coupling factor could depend on the wave propagation
medium.
From analyses of these structures, the range in which there is a weak dependence of the
wave velocity, coupling factor on the AlN layer thickness could be known. The SAW
devices should be fabricated in this range to facilitate manufacturing.
For AlN/Si structure, this range is 3 khAlN .
For AlN/Mo/Si, if this kind of SAW device is fabricated in the range from 2.7 khAlN to
facilitate manufacturing, the use of Mo layer is useless. Consequently, to take full
advantage of using Mo layer in term of wave velocity and coupling factor, it should be
required to control the fabrication process carefully to obtain the required AlN thickness
from khAlN=1.02 to khAlN=2.7.
For AlN/SiO
2
/Si, this range is 5 khAlN for khSiO
2
=0.7854, for thicker SiO
2
layer, this
range changes based on Figure 6 and Figure 7. Besides, using SiO
2
layer would reduce
temperature dependence of frequency. To choose the thickness of SiO
2
layer, it would
consider the effect of temperature dependence and analyses of wave velocity, coupling
factor.
7. Appendix: Development of calculation for equivalent circuit of SAW device
7.1 Appendix 1. Equivalent circuit for normal IDT including N periodic sections

Fig. Appendix.1. Mason equivalent circuit for one periodic section in crossed-field model


467

Fig. Appendix.2. Mason equivalent circuit for one periodic section in in-line field model
One periodic section can be expressed by the 3-port network as follows:

Fig. Appendix.3. One periodic section represented by 3-port network, admittance matrix [y]

1 11 12 13 1
2 21 22 23 2
3 31 32 33 3
i y y y e
i y y y e
i y y y e
( ( (
( ( (
=
( ( (
( ( (

(Appendix.1)
By the symmetrical properties of one periodic section (the voltage applied at port 3 will
result in stress of the same value at port 1 and 2), the [y] matrix in (Appendix.1) becomes
(Appendix.2) for Figure Appendix.4 and becomes (Appendix.3) for Figure Appendix.5.

1 11 12 13 1
2 12 11 13 2
3 13 13 33 3
i y y y e
i y y y e
i y y y e
( ( (
( ( (
=
( ( (
( ( (

(Appendix.2)

Fig. Appendix.4.
3-port network
representation of one
periodic section, with
the change of sign
between Y
13
and Y
23
to
ensure that acoustic
power flows
symmetrically away
from transducer
1 11 12 13 1
2 12 11 13 2
3 13 13 33 3
i y y y e
i y y y e
i y y y e
( ( (
( ( (
=
( ( (
( ( (

(Appendix.3)

Fig. Appendix.5.
3-port network
representation of one
periodic section, with
the no change of sign
between Y
13
and Y
23


468
Applying circuit theory, definitions of [y] matrix elements are presented:

2 1
3 3
1 1
2 2
1 1
11 12
0 0
1 2
0 0
1 3
13 33
0 0
3 3
0 0
;
;
e e
e e
e e
e e
i i
y y
e e
i i
y y
e e
= =
= =
= =
= =
= =
= =
(Appendix.4)
And using trigonometric functions as follows:
( )
2
cot
sin(2 )
1 1
cot
sin(2 ) 2
3cos(2 ) 1 sin(4 ) sin(2 )
(4 )
sin(4 ) sin(2 ) cos(2 )
tg g
tg tg g
tg tg
tg tg
tg tg

=
=
+
=
+
(Appendix.5)
The [y] matrix can be obtained for 2 models as follows:
- for the crossed-field model:

11 0
0
12
13 0
33 0 0
cot (4 )
sin(4 )
(2 4 )
y jG g
jG
y
y jG tg
y j C G tg

=
=
=
= +
(Appendix.6)
- for the in-line field model:

2
0
0 0
11 0 2
0
0
0
2
0
0
12 0
0 0
0 0
13 0
0
0
0
33
0
1
sin(2 )
cot cot (2 ) 2
cot (2 )
1
cot
sin(2 )
2
2 cot cot (2 )
2
1
2
2
1
G
C G
y jG g g
C
G
g
C
G
g
C
y jG
G G
g g
C C
tg
y jG
G
tg
C
j C
y
G

(
| |
(
|
| |
(
\ .
=
|
(
| | \ .
(
|
(
\ .
| |
|
\ .
=
| || |

| |
\ .\ .
=
0
tg
C

(Appendix.7)


469
In IDT including N periodic sections, the N periodic sections are connected acoustically in
cascade and electrically in parallel as Figure Appendix.6.

Fig. Appendix.6. IDT including the N periodic sections connected acoustically in cascade
and electrically in parallel
Because the symmetric properties of the IDT including N section like these of one periodic
section, and from (Appendix.2), (Appendix.3), Figure Appendix.4 and Figure Appendix.5,
the [Y] matrices of N-section IDT are represented as follows:

Fig. Appendix.7. The [Y] matrices and the model corresponsive models
Since the periodic sections are identical, the recursion relation as follows can be obtained:
e
1 m
=e
2 m-1
(Appendix.8)
e
3 N
= e
3 N-1
= e
3 N-2
=...= e
3 2
= e
3 1
=E
3
(Appendix.9)
i
1 m
=i
2 m-1
(Appendix.10)
With m is integer number, m=1,2, , N-1, N
The total transducer current is the sum of currents flowing into the N sections.
( ) ( )
( ) ( )
3 3 1 3 2 3 N 1 3 N
13 1 1 13 2 1 33 3 1 13 1 2 13 2 2 33 3 2
13 1 N 1 13 2 N 1 33 3 N 1 13 1 N 13 2 N 33 3 N
I i i . i i
y e y e y e y e y e y e ...
y e y e y e y e y e y e

= + + + +
= + + + +
+ + + +
(Appendix.11)


470
By applying (Appendix.8), (Appendix.9) and boundary conditions (e
11
= E
1
, e
2N
=E
2
),
(Appendix.11) becomes:
I
3
= y
13
e
1 1
-y
13
e
2 N
+ Ny
33
E
3
= y
13
E
1
-y
13
E
2
+ Ny
33
E
3
(Appendix.12)
From Figure Appendix.7, the Y
13
and Y
33
can be expressed as:
Y
13
=y
13
(Appendix.13)
Y
33
= Ny
33
(Appendix.14)
Because the N periodic sections are connected acoustically in cascade and electrically in
parallel, the model as in Figure Appendix.5 should be used to obtain the [Y] matrix of N-
section IDT.
From (Appendix.3) for one section, the i
1
and i
2
can be expressed
i
1
= y
11
e
1
+y
12
e
2
+ y
13
e
3
, i
2
= -y
12
e
1
-y
12
e
2
+ y
13
e
3
(Appendix.15)
Equations (Appendix.15) can be represented in matrix form like [ABCD] form in electrical
theory as follows:
[ ]
2 1
3
2 1
[ ]
e e
K L e
i i
( (
= +
( (

(Appendix.16)
Where
[ ]
11
12 12
2 2
11 12 11
12 12
1 y
y y
K
y y y
y y
(
(
(
=
(
(
(

(Appendix.17)
[ ]
13
12
11 13 12 13
12
y
y
L
y y y y
y
(
(
(
=
( +
(

(Appendix.18)
By applying (Appendix.16) into N-section IDT as in Figure Appendix.6 and using
(Appendix.9), the second recursion relation is obtained as follows:
[ ]
1
3
1
[ ]
m m
m m
e e
K L E
i i

( (
= +
( (

(Appendix.19)
Where m is integer number, m=1,2, , N-1, N
Starting (Appendix.19)(Appendix.19) by using with m=N, and reducing m until m=1 gives
the expression:
[ ] [ ]
0
3
0
N
N
e e
Q X E
i i
( (
= +
( (

(Appendix.20)


471
Where
[ ] [ ]
N
Q K = (Appendix.21)

[ ] [ ]
1
1
0 2
[ ]
N
n
n
X
X K L
X

=
(
= =
(

(Appendix.22)
Solving (Appendix.20) and using the boundary conditions (e
0
= E
1
, i
0
=I
1
) gives:

11 1
1 1 2 3
12 12 12
1 Q X
I E E E
Q Q Q
= + (Appendix.23)
Consequently,

11
11
12
Q
Y
Q
= (Appendix.24)

12
12
1
Y
Q
= (Appendix.25)

1
13
12
X
Y
Q
= (Appendix.26)
The Y
13
is known by (Appendix.13), so (Appendix.26) and matrix [X] dont need to be
solved.
To solve (Appendix.24) and (Appendix.25), matrix [Q] should be solved.
In crossed-field model, matrix [Q] can be represented in a simple form as follows:

[ ]
0
0
cos(4 ) sin(4 )
sin(4 ) cos(4 )
jR
K
jG

(
=
(

(Appendix.27)

[ ]
2
0
0
cos(8 ) sin(8 )
sin(8 ) cos(8 )
jR
K
jG

(
=
(

(Appendix.28)
[ ]
3
0
0
cos(12 ) sin(12 )
sin(12 ) cos(12 )
jR
K
jG

(
=
(

(Appendix.29)
. . . . . . etc. Consequently, matrix [Q] will be given:
[ ] [ ]
0
0
cos( 4 ) sin( 4 )
sin( 4 ) cos( 4 )
N
N jR N
Q K
jG N N

(
= =
(

(Appendix.30)
From (Appendix.24) and (Appendix.35), Y
11
and Y
12
in cross-field model can be expressed:

11 0
cot ( 4 ) Y jG g N = (Appendix.31)

0
12
sin( 4 )
jG
Y
N
= (Appendix.32)


472
In conclusion, matrix [Y] representation of N-section IDT is:
- In "crossed-field" model, from (Appendix.6), (Appendix.13), (Appendix.14),
(Appendix.31) and (Appendix.32):

11 0
0
12
13 0
33 0 0
cot (4 )
sin(4 )
(2 4 )
Y jG g N
jG
Y
N
Y jG tg
Y jN C G tg

=
=
=
= +
(Appendix.33)
- In "in-line field" model, from (Appendix.7), (Appendix.13), (Appendix.14),
(Appendix.24) and (Appendix.25):

11
11
12
12
12
13 0
0
0
0
33
0
0
1
2
1
2
2
1
Q
Y
Q
Y
Q
tg
Y jG
G
tg
C
j NC
Y
G
tg
C
=
=
=
(Appendix.34)
Where [Q] can be calculated from (Appendix.17) and (Appendix.21).
7.2 Appendix 2: Equivqlent circuit for N+1/2 model IDT
In case IDT includes N periodic sections (like in section 3.2 plus one finger (in color red) as
shown in Figure Appendix.8 that we call N+1/2 model IDT.

Fig. Appendix.8. N+1/2 model IDT
The equivalent circuit for this model is shown in Figure Appendix.9 and the matrix [Yd]
representation is shown as in Figure Appendix.10 (letter d stands for different from model
[Y] in section 3.2.


473

Fig. Appendix.9. Equivalent circuit of N+1/2 model IDT

Fig. Appendix.10. [Yd] matrix representation of N+1/2 model IDT
The form of matrix [Yd] is:

[ ]
11 12 13
21 22 23
31 32 33
Yd Yd Yd
Yd Yd Yd Yd
Yd Yd Yd
(
(
=
(
(

(Appendix.35)
The elements of [Yd] matrix for crossed-field model are given as follows:

11 0 2
1
cot (4 )
sin (4 )(cot (2 ) cot (4 ))
Yd jG g N
N g g N

=
`
+

)
(Appendix.36)

0
12
sin(2 )[cot (4 )cos(2 ) sin(2 )]
cos(2 )
sin(4 ) cos(2 ) cot (4 )sin(2 )
jG g N
Yd
N g N

=
`
+

)
(Appendix.37)
2 2
13 0
( 2cot (4 )sin sin(2 ))sin(2 ) 2sin
sin(4 )(cos(2 ) cot (4 )sin(2 )) sin(4 )
tg g N
Yd jG tg
N g N N

+ +
= +
`
+

)
(Appendix.38)

21 0
1
sin(4 )(cos(2 ) cot (4 )sin(2 ))
Yd jG
N g N
=
+
(Appendix.39)

22 0
cot (4 )cos(2 ) sin(2 )
cos(2 ) cot (4 )sin(2 )
g N
Yd jG
g N

=
+
(Appendix.40)


474

2
23 0
2cot (4 )sin (2 ) sin(2 )
cos(2 ) cot (4 )sin(2 )
tg g N
Yd jG
g N

+ +
=
+
(Appendix.41)

31 0
Yd jG tg = (Appendix.42)

32 0
sin(2 ) Yd jG = (Appendix.43)
{ }
33 0 0
(2 1) sin(2 ) (4 1) Yd j C N jG N tg = + + + (Appendix.44)
7.3 Appendix 3: Scattering matrix [S] for IDT
The scattering matrix [S] of a three-port network characterized by its admittance matrix [Y]
is given by [3]:

1
3 3
2 ( ) S Y Y

= + (Appendix.45)
Where
3
is the 3x3 identity matrix.
After expanding this equation, the scattering matrix elements for a general three-port
network are given by the following expressions:

{
}
11 33 11 22 11 22 12 21
13 31 22 21 32 23 32 11 12 31
1
(1 )(1 )
[ (1 ) ] [ ( 1) ]
S Y Y Y Y Y Y Y
M
Y Y Y Y Y Y Y Y Y Y
= + + + +
+ + +
(Appendix.46)

[ ]
12 12 33 13 32
2
(1 ) S Y Y Y Y
M
= + (Appendix.47)

[ ]
13 13 22 12 23
2
(1 ) S Y Y Y Y
M
= + (Appendix.48)

[ ]
21 21 33 23 31
2
(1 ) S Y Y Y Y
M
= + (Appendix.49)

{
}
22 33 11 22 11 22 12 21
13 31 22 21 32 23 32 11 12 31
1
(1 )(1 )
[ ( 1) ] [ ( 1) ]
S Y Y Y Y Y Y Y
M
Y Y Y Y Y Y Y Y Y Y
= + + + +
+ + +
(Appendix.50)

[ ]
23 23 11 13 21
2
(1 ) S Y Y Y Y
M
= + (Appendix.51)

[ ]
31 31 22 21 32
2
(1 ) S Y Y Y Y
M
= + (Appendix.52)

[ ]
32 32 11 12 31
2
(1 ) S Y Y Y Y
M
= + (Appendix.53)

{
}
33 33 11 22 11 22 12 21
13 31 22 21 32 23 32 11 12 31
1
(1 )(1 )
[ ( 1) ] [ ( 1) ]
S Y Y Y Y Y Y Y
M
Y Y Y Y Y Y Y Y Y Y
= + + + +
+ + + +
(Appendix.54)


475
where

3
33 11 22 12 21 23 32 11 12 31
13 31 22 21 32
det( )
(1 )[(1 )(1 ) ] [ (1 ) ]
[ (1 ) ]
M Y
Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y
= +
= + + + +

(Appendix.55)
For model IDT including N identical sections, these equations can be further simplified. In
case of Figure Appendix.7 (b):

11 22
21 12
31 13
23 32 13
Y Y
Y Y
Y Y
Y Y Y
=
=
=
= =
(Appendix.56)
Therefore, S
ij
s take the following form
( ) { }
2 2 2
11 22 33 11 12 13 11 12
1
(1 )(1 ) 2 S S Y Y Y Y Y Y
M
= = + + + + (Appendix.57)

2
12 21 12 33 13
2
(1 ) S S Y Y Y
M
( = = + +

(Appendix.58)

13 31 13 11 12
2
(1 ) S S Y Y Y
M
= = + + (Appendix.59)

23 32 13
S S S = = (Appendix.60)

{ }
2 2 2
33 33 11 12 13 11 12
1
(1 )[(1 ) ] 2 (1 ) S Y Y Y Y Y Y
M
= + + + + (Appendix.61)
Where

2 2 2
33 11 12 13 12
(1 )[(1 ) ] 2 (1 ) M Y Y Y Y Y = + + + (Appendix.62)
7.4 Appendix 4: Equivalent circuit for SAW device base on Mason model, [ABCD]
Matrix representation
7.4.1 Appendix 4.1: [ABCD] Matrix representation of IDT
In SAW device, each input and output IDTs have one terminal connected to admittance G
0
.
Therefore, one IDT can be represented as two-port network. [ABCD] matrix (as in Figure
Appendix.11) is used to represent each IDT, because [ABCD] matrix representation has one
interesting property that in cascaded network, the [ABCD] matrix of total network can be
obtained easily by multiplying the matrices of elemental networks.

Fig. Appendix.11. [ABCD] representation of two-port network for one IDT


476
To find the [ABCD] matrix for one IDT in SAW device, the condition that no reflected wave
at one terminal of IDT, and the current-voltage relations by [Y] matrix in section are used as
follows:

Fig. Appendix.12. Two-port network for one IDT

1 11 12 13 1
2 12 11 13 2
3 13 13 33 3
I Y Y Y V
I Y Y Y V
I Y Y Y V
( ( (
( ( (
=
( ( (
( ( (

(Appendix.63)
And I
1
=-G
0
V
1
(Appendix.64)
From these current-voltage relations, the V
3
and I
3
are given:

2 2
11 12 11 0 0 11
3 2 2
12 13 11 13 13 0 12 13 11 13 13 0
Y Y Y G G Y
V V I
Y Y Y Y Y G Y Y Y Y Y G
+ +
=
+ + + +
(Appendix.65)
2 2 2 2
13 12 13 11 13 0 11 33 13 33 0 11 12 11 0
3 2
0 11 12 13 11 13 13 0
2
11 33 13 33 0
2
12 13 11 13 13 0
( ) ( )( )
( )( )

Y Y Y Y Y G Y Y Y Y G Y Y Y G
I V
G Y Y Y Y Y Y G
Y Y Y Y G
I
Y Y Y Y Y G
+ + + + +
=
+ + +
+
+ +
(Appendix.66)
From (Appendix.65) and (Appendix.66), equivalence between port 3 in Figure Appendix.12
equals to port 1 in Figure Appendix.11, and consideration of direction of current I
2
in Figure
Appendix.11 and Figure Appendix.12, [ABCD] matrix representation for two-port network
of IDT in obtained:

2 2
11 12 11 0
12 13 11 13 13 0
Y Y Y G
A
Y Y Y Y Y G
+
=
+ +
(Appendix.67)

0 11
12 13 11 13 13 0
G Y
B
Y Y Y Y Y G
+
=
+ +
(Appendix.68)
2 2 2 2
13 12 13 11 13 0 11 33 13 33 0 11 12 11 0
0 11 12 13 11 13 13 0
( ) ( )( )
( )( )
Y Y Y Y Y G Y Y Y Y G Y Y Y G
C
G Y Y Y Y Y Y G
+ + + + +
=
+ + +
(Appendix.69)

2
11 33 13 33 0
12 13 11 13 13 0
Y Y Y Y G
D
Y Y Y Y Y G
+
=
+ +
(Appendix.70)
In case of crossed-field model, the [ABCD] can be further simplified:


477

[ ]
sin(4 ) cos(4 )
1 cos(4 ) sin(4 )
N j N
A
tg N j N

=

(Appendix.71)

0
A
B
G
= (Appendix.72)
[ ]
0 0
sin(4 )
(2 cot 4)(cot(4 ) )
1 cos(4 ) sin(4 )
N
D N C Z N j tg
N j N

= + + +

(Appendix.73)

0
1
C G D
B
= + (Appendix.74)
One interesting property of [ABCD] of crossed-field mode is:
AD-BC=1 (Appendix.75)
This means [ABCD] matrix is reciprocal.
In SAW device, the ouput IDT is inverse of input IDT. By the reciprocal property of [ABCD],
the [ABCD] matrix of output IDT can be easily obtained:
A
output
= D
input
(Appendix.76)
B
output
= B
input
(Appendix.77)
C
output
= C
input
(Appendix.78)
D
output
= A
input
(Appendix.79)
in which N is replaced by M (number of periodic sections in output IDT)
Consequently, the [ABCD] matrix of output IDT is:
[ ]
0 0
sin(4 )
(2 cot 4)(cot(4 ) )
1 cos(4 ) sin(4 )
out
M
A M C Z M j tg
M j M

= + + +

(Appendix.80)

[ ]
0
sin(4 ) cos(4 ) 1
1 cos(4 ) sin(4 )
out
M j M
B
G tg M j M

=

(Appendix.81)

[ ]
sin(4 ) cos(4 )
1 cos(4 ) sin(4 )
out
M j M
D
tg M j M

=

(Appendix.82)

0
1
out out
out
C G A
B
= + (Appendix.83)
At the center frequency f
0
, the [ABCD] matrix becomes infinite since =0.5(f/f
0
)= 0.5.
However, [ABCD] elements may be calculated by expanding for frequency very near
frequency f
0
.
By setting:

0
0
2 2 2 2
f f x
f N

= + = + (Appendix.84)


478
Where
0
0
f f
x N
f

= (Appendix.85)
By using the limit of some functions as follows:

0 0
lim[sin(4 )] lim[sin(2 )] 2
x x
N x x

= (Appendix.86)

0 0
lim[cos(4 )] lim[cos(2 )] 1
x x
N x

= (Appendix.87)

0 0
2
lim[ ] lim[ cot( )]
2
x x
x N
tg
N x

= (Appendix.88)
The [ABCD] matrix of input IDT is obtained:

2
4
x j
A
j N
(Appendix.89)

0
2 1
4
x j
B
G j N
(Appendix.90)

0
0 0 0
4
2 4
2
NG
C fC x NG j fC
x j

| |
+
|
\ .
(Appendix.91)

0 0 0 0
2 4 D fC Z x N j fC Z (Appendix.92)
7.4.2 Appendix 4.2: [ABCD] matrix representation of propagation path
Based on equivalent circuit star model of propagation path in section 3.3, [ABCD] matrix
representation of propagation way can be obtained clearly:
cos2
path path
A D = = (Appendix.93)
sin2
path path
B C j = = (Appendix.94)
With
fl
v
= (Appendix.95)
Where l is the length of propagation path between two IDTs.
So, [ABCD] matrix representations of input IDT, propagation way and output IDT are
obtained. They are cascaded as Figure Appendix.13:

Fig. Appendix.13. Cascaded [ABCD] matrices of input IDT, propagation way and output IDT


479
And the [ABCD] equivalent matrix of SAW device is shown in Figure Appendix.14

Fig. Appendix.14. [ABCD] matrix of SAW device
[ABCD] matrix of delay line SAW is

path path device device in in out out
path path device device in in out out
A B A B A B A B
C D C D C D C D
( ( ( (
=
( ( ( (

(Appendix.96)

device in path out in path out in path out in path out
A A A A B C A A B C B D C = + + + (Appendix.97)

B A A B B C B A B D B D D = + + + (Appendix.98)

C C A A D C A C B C D D C = + + + (Appendix.99)

D C A B D C B C B D D D D = + + + (Appendix.100)
Where [ABCD]
in
is calculated from (Appendix.71), (Appendix.72), (Appendix.73) and
(Appendix.74).
[ABCD]
out
is calculated from (Appendix.80), (Appendix.81), (Appendix.82) and (Appendix.83).
[ABCD]
path
is calculated from (Appendix.93) and (Appendix.94).
8. References
[1] C.C.W.Ruppel, W.Ruile, G.Scholl, K.Ch.Wagner, and O.Manner, Review of models for
low-loss filter design and applications, IEEE Ultransonics Symposium, pp.313-324,
1994.
[2] L.A.Coldren, and R.L.Rosenberg, Scattering matrix approach to SAW resonators, IEEE
Ultrasonics Symposium, 1976, pp.266-271.
[3] R.W.Newcomb, Linear Multiport Synthesis, McGraw Hill, 1966.
[4] C.Elachi, Waves in Active and Passive Periodic Structures: A Review, Proceedings of the
IEEE, vol.64, No.12, December 1976, pp.1666-1698
[5] M.Hikita, A.Isobe, A.Sumioka, N.Matsuura, and K.Okazaki, Rigorous Treatment of
Leaky SAWs and New Equivalent Circuit Representation for Interdigital
Transducers, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control,
Vol.43, No.3, May 1996.
[6] L.F.Brown, and D.L.Carlson, Ultrasound Transducer Models for Piezoelectric Polymer
Films, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, Vol.36,
No.3, May 1989.
[7] K.Hashimoto, and M.Yamaguchi, Precise simulation of surface transverse wave devices
by discrete Green function theory, IEEE Ultrasonics Symposium, 1994, pp.253-258.


480
[8] K.Hashimoto, G.Endoh, and M.Yamaguchi, Coupling-of-modes modelling for fast and
precise simulation of leaky surface acoustic wave devices, IEEE Ultrasonics
Symposium, 1995, pp.251-256.
[9] K.Hashimoto, and M.Yamaguchi, General-purpose simulator for leaky surface acoustic
wave devices based on Coupling-Of-Modes theory, IEEE Ultrasonics Symposium,
1996, pp.117-122.
[10] K.Hashimoto, Surface Acoustic Wave Devices in Telecommunications, Modelling and
Simulation, Springer, 2000, ISBN: 9783540672326.
[11] P.M.Smith, and C.K.Campbell, A Theoretical and Experimental Study of Low-Loss SAW
Filters with Interdigitated Interdigital Transducers, IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, Vol.36, No.1, January 1989, pp.10-15.
[12] C.K.Campbell, Modelling the Transverse-Mode Response of a Two-Port SAW
Resonator, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control,
Vol.38, No.3, May 1991, pp.237-242.
[13] C.K.Campbell, P.M.Smith, and P.J.Edmonson, Aspects of Modeling the Frequency
Response of a Two-Port Waveguide-Coupled SAW Resonator-Filter, IEEE
Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, Vol.39, No.6,
November 1992, pp.768-773.
[14] C.K.Campbell, Longitudinal-Mode Leaky SAW Resonator Filters on 64
0
Y-X Lithium
Niobate, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control,
Vol.42, No.5, September 1995, pp.883-888.
[15] C.K.Campbell, and P.J.Edmonson, Conductance Measurements on a Leaky SAW
Harmonic One-Port Resonator, IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control, Vol.47, No.1, January 2000, pp.111-116.
[16] C.K.Campbell, and P.J.Edmonson, Modeling a Longitudinally Coupled Leaky-SAW
Resonator Filter with Dual-Mode Enhanced Upper-Sideband Suppression, IEEE
September 2001, pp.1298-1301.
[17] J.Munshi, and S.Tuli, A Circuit Simulation Compatible Surface Acoustic Wave
Interdigital Transducer Macro-Model, IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, Vol.51, No.7, July 2004, pp.782-784.
[18] M.P.Cunha, and E.L.Adler, A Network Model For Arbitrarily Oriented IDT Structures,
IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, Vol.40, No.6,
November 1993, pp.622-629.
[19] D.R.Mahapatra, A.Singhal, and S.Gopalakrishnan, Numerical Analysis of Lamb Wave
Generation in Piezoelectric Composite IDT, IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, Vol.52, No.10, October 2005, pp.1851-1860.
[20] A APPENDIX. Bhattacharyya, Suneet Tuli, and S.Majumdar, SPICE Simulation of
Surface Acoustic Wave Interdigital Transducers, IEEE Transactions on Ultrasonics,
Ferroelectrics, and Frequency Control, Vol.42, No.4, July 1995, pp.784-786.
[21] C.M.Panasik, and APPENDIX.J.Hunsinger, Scattering Matrix Analysis Of Surface
Acoustic Wave Reflectors And Transducers, IEEE Transactions On Sonics And
Ultrasonics, Vol.SU-28, No.2, March 1981, pp.79-91.
[22] W.Soluch, Admittance Matrix Of A Surface Acoustic Wave Interdigital Transducer,
November 1993, pp.828-831.
[23] W.Soluch, Scattering Matrix Approach To One Port SAW Resonators, IEEE Frequency
Control Symposium, 1999, pp.859-862.


481
[24] W.Soluch, Design of SAW Synchronous Resonators on ST Cut Quartz, IEEE
September 1999, pp.1324-1326.
[25] W.Soluch, Scattering Matrix Approach To One-Port SAW Resonators, IEEE Transactions
on Ultrasonics, Ferroelectrics, and Frequency Control, Vol.47, No.6, November 2000,
pp.1615-1618.
[26] W.Soluch, Scattering Matrix Approach To STW Resonators, IEEE Transactions on
Ultrasonics, Ferroelectrics, and Frequency Control, Vol.49, No.3, March 2002, pp.327-330.
[27] W.Soluch, Scattering Analysis Of Two-Port SAW Resonators, IEEE Transactions on
Ultrasonics, Ferroelectrics, and Frequency Control, Vol.48, No.3, May 2001, pp.769-772.
[28] W.Soluch, Scattering Matrix Approach To STW Multimode Resonators, Electronics
Letters, 6
th
January 2005, Vol.41, No.1.
[29] K.Nakamura, A Simple Equivalent Circuit For Interdigital Transducers Based On The
Couple-Mode Approach, IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control, Vol.40, No.6, November 1993, pp.763-767.
[30] K. Nakamura, and K.Hirota, Equivalent circuit for Unidirectional SAW-IDTs based on
the Coupling-Of-modes theory, IEEE Trans on Ultrasonics, Ferroelectrics, and
Frequency Control, Vol.43, No.3, May 1996, pp.467-472.
[31] A.H.Fahmy, and E.L.Adler, Propagation of acoustic surface waves in multilayers: A
matrix description. Applied Physics Letter, vol. 22, No.10, 1973, pp. 495-497.
[32] E.L.Adler, Matrix methods applied to acoustic waves in multilayers, IEEE Transactions
on Ultrasonics, Ferroelectrics, and Frequency Control, Vol.37, No.6, November 1990,
pp.485-490.
[33] E.L.Adler, SAW and Pseudo-SAW properties using matrix methods, IEEE Transactions
on Ultrasonics, Ferroelectrics, and Frequency Control, Vol.41, No.5, September 1994,
pp.699-705.
[34] G.F.Iriarte, F.Engelmark, I.V.Katardjiev, V.Plessky, V.Yantchev, SAW COM-parameter
extraction in AlN/diamond layered structures, IEEE Transactions On Ultrasonics,
Ferroelectrics, And Frequency Control, Vol. 50, No. 11, November 2003.
[35] M.Mayer, G.Kovacs, A.Bergmann, and K.Wagner, A Powerful Novel Method for the
Simulation of Waveguiding in SAW Devices, IEEE Ultrasonics Symposium, 2003,
pp.720-723.
[36] W.P.Mason, Electromechanical Transducer and Wave Filters, second edition, D.Van
Nostrand Company Inc, 1948.
[37] W.P.Mason, Physical Acoustics, Vol 1A, Academic Press, New York 1964.
[38] S.D.Senturia, Microsystem Design, Kluwer Academic Publishers, 2001, ISBN 0-7923-
7246-8.
[39] W.Marshall Leach, Controlled-Source Analogous Circuits and SPICE models for
Piezoelectric transducers, IEEE Transactions on Ultrasonics, Ferroelectrics, and
Frequency Control, Vol.41, No.1, January 1994.
[40] D.A.Berlincourt, D.R.Curran and H.Jaffe, Chapter 3, Piezoelectric and Piezomagnetic
Materials and Their Function in Tranducers.
[41] W.R.Smith, H.M.Gerard, J.H.Collins, T.M.Reeder, and H.J.Shaw, Analysis of
Interdigital Surface Wave Transducers by Use of an Equivalent Circuit Model, IEEE
Transaction on MicroWave Theory and Techniques, No.11, November 1969, pp.856-864.
[42] C. K. Campbell, Surface acoustic wave devices, in Mobile and Wireless Communications,
New York: Academic, 1998.


482
[43] O.Tigli, and M.E.Zaghloul, A Novel Saw Device in CMOS: Design, Modeling, and
Fabrication, IEEE Sensors journal, vol. 7, No. 2, February 2007, pp.219-227.
[45] M.Hofer, N.Finger, G.Kovacs, J.Schoberl, S.Zaglmayr, U.Langer, and R.Lerch, Finite-
Element Simulation of Wave Propagation in Periodic Piezoelectric SAW Structures,
June 2006, pp.1192-1201.
[46] M. Hofer, N. Finger, G. Kovacs, J. Schoberl, U. Langer, and R. Lerch, Finite-element
simulation of bulk and surface acoustic wave (SAW) interaction in SAW devices,
IEEE Ultrasonics Symposium, 2002.
[47] Online: http://www.comsol.com/
[48] Online: http://www.coventor.com/
[49] Online: http://www.ansys.com/
[50] J.J. Campbell, W.R. Jones, A method for estimating optimal crystal cuts and
propagation directions for excitation of piezoelectric surface waves, IEEE
Transaction on Sonics Ultrasonics. SU-15 (4), 1968, pp.209217.
[51] E.Akcakaya, E.L.Adler, and G.W.Farnell, Apodization of Multilayer Bulk-Wave
Transducers, IEEE Transactions on Ultrasonics Ferroelectrics and Frequency Control,
Vol.36, No.6, November 1989, pp 628-637.
[52] E.L.Adler, J.K.Slaboszewicz, G.W.FARNELL, and C.K.JEN, PC Software for SAW
Propagation in Anisotropic Multilayers, IEEE Transactions on Ultrasonics
Ferroelectrics and Frequency Control, Vol.37, No.2, May 1990, pp.215-223.
[53] E.L.Adler, SAW and Pseudo-SAW Properties Using Matrix Methods, IEEE Transactions
on Ultrasonics Ferroelectrics and Frequency Control, Vol.41, No.6, pp.876-882,
September 1994.
[54] K. A. Ingebrigtsen, Surface waves in piezoelectrics, Journal of Applied Physics, Vol.40,
No.7, 1969, pp.2681-2686.
[55] Y.Suzuki, H.Shimizu, M.Takeuchi, K.Nakamura, and A.Yamada, Some studies on SAW
resonators and multiple-mode filters, IEEE Ultrasonics Symposium Proceedings, 1976,
pp.297-302.
[56] S.D.Senturia, Microsystem Design, Kluwer Academic Publishers, 2001, ISBN 0-7923-
7246-8.
[57] W.P.Mason, Electromechanical Transducer and Wave Filters, second edition, D.Van
Nostrand Company Inc, 1948.
[58] W.P.Mason, Physical Acoustics, Vol 1A, Academic Press, New York 1964.
[59] W.R.Smith, H.M.Gerard, J.H.Collins, T.M.Reeder, and H.J.Shaw, Analysis of
Interdigital Surface Wave Transducers by Use of an Equivalent Circuit Model, IEEE
Transaction on MicroWave Theory and Techniques, No.11, November 1969, pp.856-864.
[60] Y.Suzuki, H.Shimizu, M.Takeuchi, K.Nakamura, and A.Yamada, Some studies on SAW
resonators and multiple-mode filters, IEEE Ultrasonics Symposium Proceedings, 1976,
pp.297-302.
21
Sources of ThirdOrder Intermodulation
Distortion in Bulk Acoustic Wave Devices:
A Phenomenological Approach
Eduard Rocas and Carlos Collado
Universitat Politcnica de Catalunya (UPC), Barcelona
Spain
1. Introduction
Acoustic devices like Bulk Acoustic Wave (BAW) resonators and filters represent a key
technology in modern microwave industry. More specifically, BAW technology offers
promising performance due to its good power handling and high quality factors that make
it suitable for a wide range of applications. Nevertheless, harmonics and 3IMD arising from
intrinsic nonlinear material properties (Collado et al., 2009) and dynamic self-heating (Rocas
et al., 2009) could represent a limitation for some applications.
Driven by the need for highly linear devices, there is a demand for further development of
accurate models of BAW devices, capable of predicting the nonlinear behavior of the
device and its impact on a circuit. Many authors have attempted to model the
nonlinearities of BAW devices by using different approaches, mostly involving
phenomenological lumped element models. Although these models can be useful because
of their simplicity, they are mainly limited to narrow-band operation and they usually
cannot be parameterized in terms of device-independent parameters (Constantinescu et
al., 2008). Another approach consists of extending all the material properties on the
constitutive equations to the nonlinear domain and introducing the nonlinear relations to
the model implementation, which leads to several possible nonlinear sources increasing
model complexity (Cho et al., 1993; Ueda et al., 2008). On the other hand, (Feld, 2009)
presents a one-parameter nonlinear circuit model to account for the intrinsic
nonlinearities. Such a model does not include the self-heating mechanism and can
underestimate the 3IMD by more than 20 dB.
In this work, a model that uses several nonlinear parameters to predict harmonics and 3IMD
distortion is presented. Its novelty lies in its ability to predict the nonlinear effects produced
by self-heating in addition to those due to intrinsic nonlinearities in the material properties.
The model can be considered an extension of the nonlinear KLM model (originally proposed
by Krimholtz, Leedom and Matthaei) (Krimholtz et al., 1970) to include the thermal effects
due to self-heating caused by viscous losses and electrode losses. For this purpose a thermal
domain circuit model is implemented and coupled to the electro-acoustic model, which
allows us to calculate the dynamic temperature variations that change the material
properties. In comparison to (Rocas et al., 2009), this work describes the impact that
electrode losses produce on the 3IMD, presents closed-form expressions derived from the


484
circuit model and validates the model with extensive measurements that confirm the
necessity to include dynamic self-heating to accurately predict the generation of spurious
signals in BAW devices.
2. Nonlinear generation mechanisms
The origin of nonlinearities in BAW resonators has been controversial and there still exists
no consensus (Nakamura et al., 2010). However, recent results point to several underlying
causes which combine in different ways to give rise to a wide range of nonlinear effects
(Rocas et al., 2009). We summarize the nonlinear effects of a stiffened elasticity, and then
address the nonlinearity due to self-heating caused by viscous losses and electrode losses.
We develop a circuit model to describe self-heating effects, and compare the measured
results with simulations. Closed-form expressions for a simple one-layer BAW device model
are then extracted to better understand the nonlinear generation mechanisms.
2.1 Nonlinear stiffened elasticity
Nonlinear elasticity has been proposed as the predominant contribution to the measured
second harmonics and as a potential source of the observed 3IMD products (Collado et al.,
2009) in two-tone experiments.
The approach described in (Collado et al., 2009) starts by considering a nonlinear stress-
strain relation under electric field described by a nonlinear stiffened elasticity c
D
(T) in the
form of the polynomial

2
0 1 2
( )
D D D D
c T c c T c T = + + (1)
where T is the stress. As detailed in (Collado et al., 2009), (1) translates into a nonlinear
distributed capacitance C
d
(v) in the equivalent electric model of the acoustic transmission
line (Auld, 1990), in which the voltage v is equivalent to force. In the equivalent electric
model (1) transforms into:

2
,0 1 2
( ) .
d d
C v C C v C v = + + (2)
Equation (2) leads to the nonlinear acoustic Telegraphers equations which can be used to
obtain the maximum voltage amplitude occurring along a resonating transmission line as
shown in (Collado et al., 2009; Collado et al., 2005). When the device is driven by two tones
at frequencies
l
and
2
, standing waves with maximum force amplitudes V
1
and V
2
are
trapped in the line. Then, as detailed in (Collado et al., 2009), the nonlinear capacitance (2) is
responsible for generating 3IMD signals that result from adding the contributions due to
c
1
D
and c
2
D
:

12 1 2
2 * 2
1 1 L
V A Q V V C

= (3)

12 1 2
2 *
2 2 L
V A Q V V C

= (4)
where
12
= 2
1
-
2
, Q
L
is the loaded quality factor and A
1
and A
2
are constants that depend
on the geometry of the device and on its materials. Identical results would be obtained for
the 3IMD at 2
2
-
1
(which we will denote as
21
).
Sources of ThirdOrder Intermodulation Distortion
in Bulk Acoustic Wave Devices: A Phenomenological Approach

485
2.2 Self-heating
Third-order intermodulation distortion due to dynamic self-heating is a well known process
in microwave power amplifiers (Camarchia et al., 2007; Parker et al., 2004; Vuolevi et al.,
2001) but has received less attention in passive devices (Rocas et al., 2010). What makes it
different from the 3IMD caused by intrinsic nonlinearities is its dependence on the envelope
frequency of the signal. For the particular case of a two-tone experiment, in which the
envelope is a sinusoid, the thermal generation of 3IMD has a low-pass dependence on the
envelope frequency due to the slow dynamics related with heating effects.
Recent results of two-tone 3IMD tests in BAW resonators as a function of the tones spacing
reveal the important impact of self-heating effects in thin-Film Bulk Acoustic Resonators
(FBAR) (Collado et al., 2009; Feld, 2009; Rocas et al., 2008) and Solidly Mounted Resonators
(SMR) (Rocas et al., 2009). Heating produced by viscous damping in the acoustic domain
and by ohmic loss in the electric domain produce local temperature oscillations which affect
the temperature-dependent material properties.
If
1
=
0
-

/ 2 and
2
=
0
+

/ 2 are the input signals for a two-tone test, dissipation
occurs as a result of electric and acoustic losses, and the quadratic dependence of the
dissipated power on the signal amplitude

2
1 0 2 0
cos cos
2 2
d
P V t V t

+ +

(5)
gives rise to several frequency components of the dissipated power:

2 2 2
1 2 1 0
2
2 0 1 2 0
1 2
1 1 1
cos(2 )
2 2 2
1
cos(2 ) cos(2 )
2
cos( ).
d
P V V V t t
V t t VV t
VV t

+ + +
+ +
+
(6)
These frequency components produce temperature variations on the device at the same
frequencies. These temperature variations K() can be written in terms of the dissipated
power and the thermal impedance as (Parker et al., 2004)
( ) ( ) ( ).
th d
K Z P = (7)
It is important to point out that the temperature variation at the envelope frequency (

=
2
-
1
) is the most relevant for the generation of spurious signals because of the low-pass filter
character of the thermal impedance Z
th
(). These slow temperature oscillations induce low
frequency changes of the material properties, and consequently, generate undesired 3IMD.
In addition to being able to calculate the temperature oscillations, we also need to determine
how these oscillations influence the device performance. For the specific case of BAW
devices, there is consensus in assuming that the detuning of BAW devices with temperature
is due to the variation of multiple material properties with temperature (Lakin et al., 2000;
Ivira et al., 2008; Petit et al., 2007). We reflect this in our model by adding a temperature-
dependent term to the stiffened elasticity in (1)

2
0 1 2
( , )
D D D D D
K
c T K c c T c T c K = + + + (8)


486
where K represents the temperature, the equivalent capacitance is

2
,0 1 2
( , ) ,
d d K
C v K C C v C v C K = + + + (9)
where each of the nonlinear terms C
1
, C
2
and C
K
are related to their counterparts c
1
D
,
c
2
D
, c
K
D
respectively, as detailed in Appendix I.
The term C
K
generates 3IMD, whose maximum voltage V
12
can be found in a similar way
as the contribution of C
1
in (3) and C
2
in (4) (see details in Appendix I):

12 1
*
,
T L K d th
V A Q C P Z V

= (10)
where A
T
is a constant that depends on the device geometry and material parameters, Q
L
is
the loaded quality factor, Z
th,
is the thermal impedance (7) evaluated at , and P
d,
is the
frequency component of the dissipated power in (6). Equation (10) describes the 3IMD
signal due to self-heating effects, inside the acoustic transmission line, in terms of the
dissipated power. As detailed in the following sub-sections, the dissipated power is due to
both electric and acoustic loss, thus both effects contribute to the 3IMD in (10).
2.2.1 3IMD due to viscous losses
Viscosity is introduced in the model as a complex elasticity (Auld, 1990), which translates
into a shunt resistance R
d,
in series with the shunt capacitance C
d
in a transmission line
implementation. Appendix II details a model transformation to go from the original R
d,
to
an equivalent model in which the viscosity is implemented as a conductance G
d
in parallel
with the capacitance C
d
. The equivalent model allows for an easier extraction of the closed-
form expressions.
The instantaneous dissipated power due to viscous damping at each position z along the
transmission line of length l (thickness of the piezoelectric layer) is

1 2
2 ,
( )
cos
d
d
P z z
G V V
z l

, (11)
which can be integrated along l to obtain the total dissipated power

1 2
*
,
1
2
d d
P lG V V

= . (12)
Equation (12) can be combined with (10) to obtain the peak 3IMD voltage (V
,12
) due to the
viscous damping

12 1 2
* 2 *
,
1
2
T d L K th
V A lG Q C Z V V

= (13)
2.2.2 3IMD due to loss in the electrodes
There is certain agreement in considering ohmic losses as a significant dissipation
mechanism (Thalhammer et al., 2005) in addition to the viscous damping. As it will be
discussed in section II.B.3, electrodes losses are introduced in the circuit model as parasitic
series resistances at the input and at the output ports, and their values are determined by
fitting the model to the measured scattering parameters in the linear regime. Their

487
contribution to the 3IMD can be calculated by the use of (10) and the power dissipated in the
parasitic resistances P
:

12 1
*
, T L K th
V A Q C P Z V

= (14)
Whereas the parasitic resistance and distributed conductance can be obtained from the
measured scattering parameters, that is, they produce distinguishable measurable effect,
examination of (13) and (14) looks like both self-heating mechanisms produce the same
experimental observable so they may not be distinguishable. This is true if a two-tone
experiment at a fixed frequency is performed, but the two effects have different frequency
dependence that can be distinguished if the central frequency
0
of the 2 tones is swept
while keeping the tones spacing constant. This happens because the frequency pattern of
the dissipation due to ohmic losses is different than that produced by viscous losses, as
shown in Fig 1. This information is extremely useful to validate the model with 3IMD
measurements by looking at the frequency dependence of the 3IMD.
Note that (13) and (14) keep the same definition of thermal impedance Z
th,
. This is because
the electrodes and the piezoelectric layer are thin and made of good thermal conductors, so
that the thermal impedance between those layers is negligible, as will be verified with the
temperature simulations shown in Section III.B.2.

1.8 1.9 2 2.1 2.2 2.3 2.4
Frequency (GHz)
P

(
m
W
)
e
l
e
c
t
r
o
d
e
s
0.8
0
0.64
0.48
0.32
0.16
1.2
0
0.96
0.72
0.48
0.24
P
(
m
W
)
v
i
s
c
o
s
i
t
y
R
e
s
o
n
a
n
c
e
M
a
x
.
S
t
r
e
s
s
A
n
t
ir
e
s
o
n
a
n
c
e

Fig. 1. Simulations of the dissipated power, for an input power of 20 dBm, due to acoustic
viscous damping (solid line) and electrode electric losses (dashed line)
2.2.3 Circuit model with self-heating effects
A circuit model implementation to reproduce thermal effects should be capable of
predicting dynamic temperature variations. To achieve this, we extend the nonlinear KLM
model (Collado et al., 2009) to include the thermal domain (Rocas et al., 2009).
The procedure starts with the one dimensional heat equation along the z direction:

2
2
,
p
d
th th
C
K K P
z k t k

=

(15)


488
where the equivalent distributed parameters can be identified as the volumetric heat
capacitance

, d th p
C C = (16)
and the thermal resistance

,
1
d th
th
R
k
= (17)
with C
p
and k
th
being the material-specific heat capacity and thermal conductivity,
respectively.
With the above-mentioned distributed parameters, a thermal distributed model can be
constructed as a cascade of sections of series resistances and shunt capacitance, where each
section corresponds to a specific thickness and area. Figure 2 shows a segment with R
th
=
R
d,th
z/A and C
th
= C
d,th
Az, where A is the area of the cross-section perpendicular to the z
direction. In such a thermal equivalent circuit the equivalents of voltage and current are the
temperature and heat respectively.

R
th
/2
C
th
R
th
/2
Z direction

Fig. 2. Implementation of a z section of thermal equivalent circuit
The thermal model of a multilayer SMR can be implemented as a cascade of the previously
described sections for each material, as shown in Fig. 3. The boundary conditions are the
ambient temperature, modeled as a voltage source under the substrate, and the parallel
combination of the radiation and convection resistances, terminated with a voltage source at
ambient temperature on the upper side of the device (Larson et al., 2002).

Z direction
...
C
th,Si
T
amb
R
th,1
/2
C
th,1
R
rad
R
conv
T
amb
R
th,1
/2
R
th,Si
/2 R
th,Si
/2
R
th,2
/2
C
th,2
R
th,2
/2

Fig. 3. Thermal model of the upper and lower materials stacks with boundary conditions
As it can be seen from Fig. 3, the thermal impedance seen from any point along the line has
a low-pass filter behavior, which means that for faster variations of the heat source, smaller
temperature variations are produced.
The piezoelectric layer is implemented as a cascade of cells, in which the dissipated power
due to viscous damping is directly coupled to its correspondent thermal cell. A current
source is used because current is the analogue of heat in the thermal domain. The

489
temperature rise is used to modify the distributed acoustic capacitance C
NL
(T,K), as shown
in Fig. 4.

z
L
d
/2
C
NL
(T,K)
G
d
+
F
-
+
F+dF
-
R
th
/2
C
th
L
d
/2
P
d
+
K
-
+
K+dK
-
Thermal
Domain
Acoustic
Domain
R
th
/2

Fig. 4. Implementation of a section of the piezoelectric layer with the acoustic and thermal
domains coupled by the generated heat at G
d
and the temperature K. L
d
is the acoustic
distributed inductance L
d
= Az.

Thermal Domain Extension
... ...
... ...
Nonlinear
KLM model
R
in
R
out
z
Top Layers
(Thermal Domain)
Bottom Layers
(Thermal Domain) Piezoelectric Layer
Electro-Acoustic
Conversion
Z
air
Z
air
Top Layers
(Acoustic Domain)
Bottom Layers
(Acoustic Domain)
T
amb
T
amb
R
rad
R
conv
R
th,1
/2
C
th,1
C
th,Si
R
th,1
/2 R
th,Si
/2 R
th,Si
/2

Fig. 5. Complete circuit model with thermo-acoustic model of the piezoelectric layer, top
and bottom layers, and lossy electrodes. Electric losses, in the electrodes, and viscous losses,
in the piezoelectric layer, produce dissipation that is coupled to the thermal domain to
reproduce temperature rise. The temperature rise is used to change the material properties
On the other hand, the parasitic electrodes losses are implemented by use of a lumped
resistor at the input and output of the modeled device as shown in Fig. 5. As done for the
viscosity, the dissipation in each resistor is coupled to the thermal model as a heat source. In


490
fact, dissipation in the input and output resistors is coupled to the correspondent top and
bottom thermal sections that model the electrodes. The complete model can be seen in Fig. 5,
where a cell of the piezoelectric layer like that in Fig. 4, is highlighted in red.
In the figure above, the electric-acoustic conversion box includes those elements of the KLM
model whose purpose is the electro-acoustic signal conversion (Krimholtz et al., 1970).
Additionally, the material layers above and below the piezoelectric are shown as simplified
blocks for clarity.
2.2.4 Comparison of formulation and nonlinear simulations
We use the circuit model of Fig. 5, with only a piezoelectric layer, to check the accuracy of
the formulation described in the previous section. The circuit model has been simulated,
reproducing a two-tone experiment, with Harmonic Balance techniques by use of a
commercial CAD software. A simple model is implemented making use of 100 cells to
reproduce a 1.25 m thick and 2.3310
-8
m
2
piezoelectric layer with a quality factor of 1800.
The electrodes losses and viscous losses are coupled to a low-pass thermal impedance.

1.8 1.9 2 2.1 2.2 2.3 2.4
P
h
a
s
e
o
f
3
I
M
D
(
d
e
g
r
e
e
s
)
M
a
g
n
i
t
u
d
e

o
f

3
I
M
D
(
d
B
m
)
-80
-180
-100
-120
-140
-160
0
-400
-80
-160
-240
-320
Frequency (GHz)
(a)

1.8 1.9 2 2.1 2.2 2.3 2.4
0
-400
-80
-160
-240
-320
-85
-140
-96
-107
-118
-129
Frequency (GHz)
M
a
g
n
i
t
u
d
e

o
f

3
I
M
D
(
d
B
m
)
P
h
a
s
e

o
f

3
I
M
D
(
d
e
g
r
e
e
s
)
(b)

Fig. 6. Comparison of the magnitude and phase of 2
1
-
2
calculated with equation (13)
(circles) in Fig.6a (viscous losses, no electrode losses) and equation (14) (circles) in Fig.6b
(electrode losses, no viscous losses), vs. simulation with the circuit model (solid lines)

491
In the first set of simulations we keep the tones spacing constant at /2 = 220 Hz and
sweep the central frequency
0
in a 600 MHz range around the resonance frequency, which
is 2.18 GHz. By doing this, we can distinguish the 3IMD produced by viscous self-heating
from that produced by electric self-heating by analyzing the resulting frequency
dependence. In the former case, we do not connect the dissipation in the electrodes to the
thermal domain (Fig.6a), whereas in the latter case we do not connect the dissipation in the
piezoelectric layer to the thermal domain (Fig.6b). The 3IMD frequency dependences are a
direct consequence of the frequency dependences of the dissipated power. More specifically,
a minimum at the anti-resonance frequency appears in Fig. 6.b because there is minimum
current flowing through the electrodes at anti-resonance, which can be used in experimental
measurements to identify different sources of self-heating effects.
In the second set of simulations we keep the central frequency constant at 2.18 GHz and we
change the separation between tones from 100 Hz to 1 MHz. This allows us to reproduce the
low-pass filter behavior of the thermal impedance. Figure 7 shows the results of the second
set of simulations for a wide range of separation between tones when the self-heating effects
are due to viscous losses, where it is clear the low-pass filter behavior of the temperature
induced effects. A very similar plot was obtained for electrode losses, which is not shown
for simplicity.

M
a
g
n
i
t
u
d
e

o
f

I
3
M
D
(
d
B
m
)
P
h
a
s
e
o
f
3
I
M
D
(
d
e
g
r
e
e
s
)
-80
-140
-92
-104
-116
-128
-60
-160
-80
-100
-120
-140
Separation between tones (Hz)
10
2
10
3
10
4
10
5
10
6

Fig. 7. Magnitude and phase of equation (13) (circles) and simulations with the circuit model
(traces) for a wide range of separation between tones
Figures 6 and 7, in addition to giving useful qualitative information about the 3IMD
generation due to the self-heating mechanism, show that the formulation of equations (13)
and (14) is in very good agreement with the simulations, so that these expressions can be
used for a better understanding of the temperature-induced 3IMD in BAW resonators.
3. Experimental results
Four state-of-the-art rectangular Solidly-Mounted Resonators (SMR) from a commercial
manufacturer, with different areas summarized in Table 1, have been measured. The
resonators have a 1.25 m thick aluminum nitride layer and a W - SiO
2
Bragg mirror
(alternating layers of W and SiO
2
), and show quality factors around 1800.


492
From (8) it is clear that several sources, characterized by c
D
1
, c
D
2
and c
D
K
, can generate
3IMD. Therefore, we follow a step-by-step procedure that includes several experiments to
determine which nonlinear source is designed responsible for each observable at each
experiment by use of the circuit model:
We first adjust the linear model to the measured S-parameters of the devices, so that the
electric and viscous losses can be quantified. The procedure consists of a fine tuning of
the material properties.
Second harmonic measurements are performed along the frequency range of interest to
extract the intrinsic nonlinear parameter c
D
1
.
The term c
D
2
also contributes to the 3IMD generation. We use the literature value in
(epkowski et al., 2005) because this contribution cannot be independently extracted
from measurements.
Third-order intermodulation distortion measurements with sweeping the tone spacing
are conducted to quantify the frequency dependence of the thermal impedance, and set
the temperature coefficient of stiffened elasticity c
K
D
accordingly.
3.1 Linear modeling of the devices under test
Broadband one-port S-parameters measurements of the devices have been performed after
an on-wafer OSL calibration. The measurements have been done at a power level of -10 dBm
to ensure the linear regime and are used to fit the linear parameters of the circuit model. The
only differences between the devices are the resonator area and electrode losses.
Electric losses due to the resistivity of the electrodes are modeled as lumped parasitic
resistances, so their values are dependent on the resonator area and have a broadband effect
on the linear device response. Table 1 summarizes the device areas and electric resistances.
On the other hand, acoustic losses due to viscosity only have an observable effect at those
frequencies where there is substantial electro-acoustic coupling, that is around resonance
and anti-resonance. By using the model transformation in Appendix II, the acoustic losses
that fit all devices can be described with the same material viscosity value = 0.033 Nsm
-2
,
what verifies the validity of the linear model.

Resonator Area (m
2
) Electric Resistance R
s
()
A1 6.41e-8 0.28
A2 4.88e-8 0.37
A3 2.33e-8 0.42
A4 1.25e-8 1
Table 1. Tested devices
3.2 Nonlinear characterization
The nonlinear behavior of a BAW resonator arises from different contributions due to
intrinsic nonlinear material properties and self-heating mechanisms.
3.2.1 Intrinsic nonlinearities
Intrinsic nonlinearities due to the stiffened elasticity, as shown in (1), predominate above
other intrinsic nonlinear material contributions. As a consequence, the second harmonic has
the same frequency dependence as the mechanical stress in the piezoelectric layer (Collado

493
et al., 2010). In fact, as done in (Collado et al., 2010), the second harmonic is used to extract a
unique unitless value c
D
1

= 10.5 of the stiffened elasticity that fits the second harmonic for
all the devices.

1.9 1.94 1.98 2.02
-80
-60
-40
-20
0
20
O
u
t
p
u
t

P
o
w
e
r

(
d
B
m
)
Frequency (GHz)

1 2
,
c
D
1
c
D
2

Fig. 8. 3IMD measurements (squares are 2
1
-
2
, circles are 2
2
-
1
) and circuit simulations of
the c
D
1
and c
D
2
contributions to the 3IMD (2
1
-
2
and 2
2
-
1
overlap). Both
measurements and simulations are done for f = 220 Hz. The intrinsic nonlinearities are not
sufficient to explain the measurements. Measurements (squares and triangles) and
simulations (dashed lines) of
1
and
2
are also presented

10
2
10
3
10
4
10
5
10
6
10
7
-70
-60
-50
-40
3
I
M
D

(
d
B
m
)
Tones Spacing (Hz)
c
D
1
c
D
2

Fig. 9. 3IMD measurements (squares are 2
1
-
2
, circles are 2
2
-
1
) and circuit simulations of
the c
D
1
and c
D
2
contributions to the 3IMD (2
1
-
2
and 2
2
-
1
overlap), for several
separations between tones. The intrinsic contributions cannot reproduce the envelope
frequency-dependent 3IMD level


494
The parameter c
D
1
is responsible for second harmonic generation, which in turn mixes with
the fundamental frequencies
1
and
2
, and gives rise to 3IMD. On the other hand, c
D
2

directly generates a certain level of 3IMD distortion. We use the literature value of c
D
2

= -
110
-10
N
-1
m
2
(epkowski et al., 2005).
With the above-mentioned values of intrinsic nonlinearity, simulations of 3IMD are
performed obtaining values which are below the measured levels, as shown in Fig. 8. This
shows that other contributions exist. Figure 9 shows 3IMD measurements at different
envelope frequencies, centered at the frequency where the 3IMD is maxima. The
measurements reveal a strong dependence with the envelope frequency that cannot be
accounted for intrinsic nonlinearities. The dependence of the 3IMD level on the envelope
frequency suggests that the 3IMD is dominated by a thermal effect.
3.2.2 Self-heating
The thermal model, as presented in section II.B.3, is implemented by using the literature
values of thermal conductivity and specific heat for each layer. The materials stack is
composed of more than ten layers. Dissipation on the electrodes and viscous losses are
coupled to the thermal domain by means of current sources. By using the model we can
determine the temperature distribution along the materials stack, even at different envelope
frequencies as shown in Fig. 10. Note that the temperature is almost the same in the
piezoelectric layer and the electrodes, which validates the hypothesis in Section II.B.2, of
negligible thermal impedance between the electrodes and the piezoelectric layer, to obtain
the closed-form expressions.

AlN
SiO
2
0 5 10 15 20 25 30
DC
60 Hz
10 KHz
Temperature Rise (K)
z
SiO
2
W
W
...
Top electrode
Bottom electrode
Silicon

Fig. 10. Simulations of the distribution of the z temperature variation, inside the resonator,
calculated with the circuit model. Simulations are done for resonator A1 at P
in
= 30 dBm, for
different separation between tones
A value of c
D
K
/c
D
0
= -15 ppm/K is found to fit the 3IMD best for all resonators. Figure 11
shows the intrinsic and self-heating 3IMD contributions independently as well as the sum of

495
all contributions, which match the measurements for all the tone separations. The model
reproduces very well the measured 3IMD values except the asymmetry between lower and
higher 3IMD that appears for envelope frequencies around 6 MHz. This asymmetry is
considered to be a consequence of a cancellation between different 3IMD contributions and
is currently under investigation.

Tones Spacing (Hz)
10
2
10
3
10
4
10
5
10
6
10
7
O
u
t
p
u
t

P
o
w
e
r

(
d
B
m
)
20
0
-20
-40
-60
-80

1 2
,
2 -
1 2
2 -
2 1
All
c
D
1
c
D
2
c
D
K

Fig. 11. Measurements (stars, triangles, squares and circles are
1
,
2
, 2
1
-
2
and 2
2
-
1

respectively) and simulations (thin dotted line is
1
and
2
, solid line is 2
1
-
2
and 2
2
-
1

overlapped) of the fundamentals and the 3IMD for different tones spacing for A1. The
dotted line, the dash-dot line and the dashed line are simulations of the 3IMD contribution
due to c
D
K
, c
D
2
and c
D
1
respectively. The solid line is the 3IMD simulation with all the
nonlinear contributions

1.88 1.92 1.96 2 2.04
-90
-80
-70
-60
-50
-40
10
-3
10
-2
10
-1
Frequency (GHz)
Viscosity
Electrodes
3
I
M
D

(
d
B
m
)
D
i
s
s
i
p
a
t
i
o
n
(
W
)
(a)


496
1.88 1.92 1.96 2 2.04
-90
-80
-70
-60
-50
-40
10
-3
10
-2
10
-1
3
I
M
D

(
d
B
m
)
D
i
s
s
i
p
a
t
i
o
n

(
W
)
Frequency (GHz)
(b)

1.86 1.9 1.94 1.98 2.02
-80
-70
-60
-50
-40
10
-3
10
-2
10
-1
3
I
M
D

(
d
B
m
)
D
i
s
s
i
p
a
t
i
o
n
(
W
)
Frequency (GHz)
(c)

-80
-60
-40
-20
1.86 1.9 1.94 1.98 2.02
10
-3
10
-2
10
-1
3
I
M
D

(
d
B
m
)
D
i
s
s
i
p
a
t
i
o
n

(
W
)
Frequency (GHz)
(d)

Fig. 12. Measurements (filled circles are 2
1
-
2
, empty circles are 2
2
-
1
) and simulations
(solid line is 2
1
-
2
, dash-dot line is 2
2
-
1
) of the 3IMD for resonators A1, A2, A3 and A4 in
Fig.12.a, Fig.12.b, Fig.12.c and Fig.12.d respectively. The figures also show the dissipation in
the electrodes (dashed line) and in the piezoelectric layer due to viscous losses (dotted line)

497
Figure 12 shows measurements and simulations of the 3IMD about the frequency range of
interest, for each resonator, by use of the circuit model. In this experiment, the two tones are
swept around the resonating frequencies keeping the separation between tones constant at
100 Hz. The results show good agreement between simulations and measurements above
the nonlinear system baseline level, which is around -60 dBm.
The dashed and dotted lines in Fig. 12 show the simulated dissipated power due to
electrodes losses and viscosity respectively, which have different frequency dependences
according to the maximum electric current and mechanical stress, respectively. 3IMD
measurements for resonators A3 and A4 in Fig. 12 show a peak at the antiresonant
frequency that is underestimated by the simulations. The frequency dependence in that
range points to a possible electric-field contribution to the 3IMD. This contribution is below
the system nonlinear baseline level for resonators A1 and A2, and the area scaling has not
been successfully reproduced by use of an electric-field dependent permittivity or stiffened
elasticity in the acoustic transmission line, so further research is needed.
4. Conclusion
The role of self-heating and material nonlinearities in the generation of 3IMD in bulk
acoustic wave devices has been evaluated through measurements, models and equations.
Self-heating is found to have a very significant contribution to 3IMD and thus thermal
considerations are critical in the device design. The presented circuit model implementation
offers the possibility to predict 3IMD in BAW resonators, given their materials stack and
geometry. With such information one can use the resonator model to accurately predict
3IMD in filters. Further research will be performed to investigate the relation between the
electric-field contribution to 3IMD and the cancellation shown in the measurements. The
development of a 3D equivalent thermal model, to take into account complex heat
dissipation through the substrate, will also be investigated.
5. Appendix I 3IMD equations
At each elemental section, and following a similar process than that described in (Collado et
al., 2009), the nonlinear capacitance acts as an infinitesimal nonlinear current generator at
2
1
-
2
(and 2
2
-
1
), when
1
and
2
are at resonance:

12
1
, *
12
( )
1
cos
2
nl
K
I z
z
j C K V
z l

(18)
where K

=Z
th
(

) P
d
(

) .
Therefore the broadband energy balance all over the acoustic transmission line leads to

12 1
*
,0
0,
1 1
2
1
2
K
L d th
eq L
d
v
C
V j Q P Z V
C Q
C
j
W

(19)
6. Appendix II - Model transformation
Losses are introduced as a complex elasticity by means of the viscous damping coefficient :


498

t
c c
+
(20)
The inverse damping coefficient can also be understood as the conductance per unit length
G
d
=
-1
. With that, the acoustic telegrapher equations, making use of the analogy between the
acoustic and electric domains, can be written as:

d
V
L j I
z

(21)
and

1
.
D
I
j V
z Ac j A

=
+
(22)
The shunt admittance of the acoustic transmission line implementation, given by (22) and in
which Ac
D
=C
d
-1
, is a shunt capacitance in series with a resistance. To transform this to be a
capacitance in parallel with the loss term, we introduce eq. 8 in eq. 22 and expand the shunt
admittance in as a Taylor series. The result is a conductance value in parallel with a
nonlinear capacitance of the form:

2
,0 1 2
( , )
d d K
C v K C C v C v C K = + + + (23)
whose terms are related with the material linear and nonlinear properties as follows:

2 2
,0 d
G C A = (24)

1
1 2
0
( )
D
D
c
C
Ac
= (25)

2
2 3 2
0
( )
D
D
c
C
A c
= (26)

1 2
0
( )
D
K
D
c
C
A c
= (27)
7. Appendix III - Broadband loaded quality factor
The loaded quality factor can be defined as (Russer, 2006)

0
1
L
Q
Q
=
+
(28)
where relates the dissipated power in the acoustic resonator P
res
, that is the acoustic
transmission line, and the externally dissipated power P
ext
as follows:
.
ext
res
P
P
= (29)

499
By circuit analysis of the KLM circuit model, it can be found that is

*
0
*
0
1
Re
1
50
1
Re
1
m
in m
jX
j C
Z jX
j C

+ +

=

(30)
where Z
in
is the input impedance of the device and X
m
is the series reactive term of the KLM
model (Krimholtz et al., 1970). Q
0
in (28) represents the unloaded quality factor, that is
obtained from S-parameters using (Feld et al., 2008)

11
0 2
11
1
S
Q
S
=
(31)
8. References
Auld B. A., Acoustic Fields and Waves in Solids (Krieger, Malabar, Florida), Vol. I, 1990
Camarchia V., Cappelluti F., Pirola M., Guerrieri S. D., Ghione G. 2007. Self-Consistent
Electrothermal Modeling of Class A, AB, and B Power GaN HEMTs Under
Modulated RF Excitation. IEEE Transactions on Microwave Theory and Techniques, vol.
55, no. 9, Sept. 2007, pp. 1824-1831.
Cho Y., Wakita J. 1993. Nonlinear equivalent circuits of acoustic devices. Ultrasonics
Symposium, 1993. Proceedings, IEEE 1993, pp. 867-872 vol. 2, 31 Oct-3 Nov 1993
Constantinescu F., Nitescu M., Gheorghe A. G. 2008. New Nonlinear Circuit Models for
Power BAW Resonators. ICCSC 2008. 4th IEEE International Conference on Circuits
and Systems for Communications, pp. 599-603, 26-28 May 2008
Collado C., Rocas E., Padilla A., Mateu J., OCallaghan J. M., Orloff N. D., Booth J. C., Iborra
E., Aigner R. 2010. First-order nonlinearities of bulk acoustic wave resonators. IEEE
Transactions on Microwave Theory and Techniques, submitted.
Collado C., Rocas E., Mateu J., Padilla A., O'Callaghan J. M. 2009. Nonlinear Distributed
Model for Bulk Acoustic Wave Resonators. IEEE Transactions on Microwave Theory
and Techniques, vol. 57, no. 12, Dec. 2009,pp. 3019-3029
Collado C., Mateu J. and OCallaghan J. M. 2005. Analysis and Simulation of the Effects of
Distributed Nonlinearities in Microwave Superconducting Devices. IEEE Trans.
Appl. Supercond. vol. 15, No. 1, March 2005, pp. 26-39.
Feld D. A. 2009. One-parameter nonlinear mason model for predicting 2nd & 3rd order
nonlinearities in BAW devices. 2009 IEEE International Ultrasonics Symposium (IUS),
pp. 1082-1087, 20-23 Sept. 2009
Feld D. A., Parker R., Ruby R., Bradley P., Shim D. 2008. After 60 years: A new formula for
computing quality factor is warranted. 2008 IEEE International Ultrasonics
Symposium, pp. 431-436, 2-5 Nov. 2008


500
Ivira B., Benech P., Fillit R., Ndagijimana F., Ancey P., Parat G. 2008. Modeling for
temperature compensation and temperature characterizations of BAW resonators
at GHz frequencies. IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency
Control, vol. 55, no. 2, February 2008, pp. 421-430.
Krimholtz R., Leedom D.A., Matthaei G.L. 1970. New equivalent circuits for elementary
piezoelectric transducers. Electronics Letters , vol. 6, no. 13, pp. 398-399, June 1970
Lakin K. M., McCarron K. T., McDonald J. F. 2000. Temperature compensated bulk acoustic
thin film resonators. 2000 IEEE International Ultrasonics Symposium, pp. 855-858 vol.
1, Oct 2000
Larson III J. D., Oshrnyansky Y. 2002. Measurement of effective kt2, q, Rp, Rs vs.
Temperature for Mo/AlN FBAR resonators. 2002 IEEE International Ultrasonics
Symposium, pp. 939- 943 vol. 1, 8-11 Oct. 2002
epkowski S. P., Jurczak G. 2005. Nonlinear elasticity in III-N compounds: Ab initio
calculations. Physical Review B, vol. 72, 245201, 2005, pp 1-12.
Nakamura H., Hashimoto K.-y., Ueda M. 2010. Nonlinear effects in SAW and BAW
components. 2010 IEEE International Ultrasonics Symposium (IUS), Short Course, 11-
14 Oct. 2010
Parker A. E., Rathmell J. G. 2004. Self-heating process in microwave transistors. Workshop on
Applications in Radio Science, Hobart, TAS, Australia, 18-20 Feb. 2004.
Petit D., Abele N., Volatier A., Lefevre A., Ancey P., Carpentier J.-F. 2007. Temperature
Compensated Bulk Acoustic Wave Resonator and its Predictive 1D Acoustic Tool
for RF Filtering. 2007 IEEE International Ultrasonics Symposium, pp. 1243-1246, 28-31
Oct. 2007
Rocas E., Collado C., Orloff N. D., Mateu J., Padilla A., OCallaghan J. M., Booth J. C. 2010.
Passive intermodulation due to self-heating in printed transmission lines. IEEE
Transactions on Microwave Theory and Techniques, accepted for publication.
Rocas E., Collado C., Booth J. C., Iborra E., Aigner R. 2009. Unified model for Bulk Acoustic
Wave resonators' nonlinear effects. 2009 IEEE International Ultrasonics Symposium
(IUS), pp. 880-884, 20-23 Sept. 2009
Rocas E., Collado C., Mateu J., Campanella H., OCallaghan J. M. 2008. Third order
Intermodulation Distortion in Film Bulk Acoustic Resonators at Resonance and
Antiresonance. IEEE MTT-S International Microwave Symposium Digest, pp.1259-
1262, June 2008.
Russer P. 2006. Electromagnetics, microwave circuit, and antenna design for
communications engineering (Artech House), 2006
Thalhammer R., Aigner R. 2005. Energy loss mechanisms in SMR-type BAW devices. 2005
IEEE MTT-S International Microwave Symposium Digest, pp. 4, 12-17 June 2005
Ueda M., Iwaki M., Nishihara T., Satoh Y., Hashimoto K.-Y. 2008. A circuit model for
nonlinear simulation of radio-frequency filters using bulk acoustic wave resonators.
IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, vol. 55, no. 4,
pp.849-856, April 2008
Vuolevi J. H. K., Rahkonen T., Manninen J. P. A. 2001. Measurement technique for
characterizing memory effects in RF power amplifiers. IEEE Transactions on
Microwave Theory and Techniques, vol. 49, no. 8, Aug. 2001, pp. 1383-1389.
22
Shear Mode Piezoelectric Thin Film Resonators
Takahiko Yanagitani
Nagoya Institute of Technology
Japan
1. Introduction
1.1 Shear mode bulk acoustic wave devices and sensors
Acoustic microsensor technique, well known as QCM (Quartz crystal microbalance) or TSM
(Thickness shear mode) sensor, is an effective method to detect small mass loading on the
sensor surface. This sensor can be operated even in liquid by using shear mode resonance.
Therefore, shear mode piezoelectric film resonators are attractive for liquid microsensor
technique such as biosensors and immunosensors.
Shear wave has some unique features compared with the longitudinal wave, for example, it
has extremely low velocity in the liquid. Longitudinal wave velocity in the water is 1492.6
m/s, whereas, shear wave velocity in the water is 20-60 m/s at 20-200 MHz (Matsumoto et
al., 2000). Therefore, shear mode vibrating solid maintains its vibration even in the liquid,
because the difference of acoustic impedance which determines the refection coefficient of
solid / liquid interface is very large in the case of shear wave.
The complex refection coefficient of the interface is given as

l s
l s
Z Z
Z Z
=
+
(1)
where Z
s
and Z
l
are the complex acoustic impedance of solid and liquid.
Complex acoustic impedance can be written as
( ) ( )
1 2
Z R jX c j = + = + (2)
R and X represent the real part and imaginary part of the acoustic impedance and , c and
represent mass density, stiffness constant and viscosity in the medium, respectively.
Acoustic wave equation gives dispersion relation of
( )
2
2

= + |
.
|
\
|
j c j
v
(3)
where v is velocity and is attenuation factor (B. A. Auld, 1973).
According to (2) and (3), acoustic impedance gives

2
2 2 2
v
R
v

=
+
,
2
2 2 2
v
X
v

=
+
(4)


502
Longitudinal and shear wave velocities of water were reported as 1492.6 m/s (Kushibiki et
al., 1995) and 35 m/s (Matsumoto et al., 2000), respectively, at 100 MHz. Attenuations of
longitudinal and shear wave in the water were also measured to be /f
2
= 2.2610
-14

nepers
2
/m (Kushibiki et al, 1995) and /f
2
= 2.1210
-9
nepers
2
/m, (Matsumoto et al., 2000)
respectively. By substituting these values into Eq. (4), the complex longitudinal wave and
shear wave acoustic impedance of the water can be estimated to be 1489000+j800 Ns/m
3

and 14510+j17340 Ns/m
3
at 100 MHz, respectively.
From these values and Eq. (1), when quartz resonator is immersed in water, the reflection
coefficient of acoustic energy
2
in an X-cut quartz vibrating in thickness extensional mode
(Z
s
= 15.2310
-6
Ns/m
3
) is estimated to be only 68 % whereas that in an AT-cut quartz
vibrating in thickness shear mode (Z
s
= 8.79510
-6
Ns/m
3
) is 98 %. This is because an AT-cut
quartz has been used as a QCM or TSM sensor operating in liquid. Sensitivity of the QCM
mass sensor is determined by the ratio of the mass and the entire mass of the vibrating part
in the sensor, at constant sensor active area (Sauerbrey, 1959). Therefore, it is important to
decrease thickness of the vibrating part of sensor. Shear mode thin film is promising for high
sensitivity mass sensor.
1.2 Piezoelectric thin film for shear mode excitation
Piezoelectric thin film, which excites shear wave, is expected to provide higher sensitivity
and IC compatibility, but it is not straightforward. To excite shear wave by standard
sandwiched electrode configuration, polarization axis in the film must be tilted or parallel
to the film plane. Although perovskite ferroelectric films have large piezoelectricity, their
polarization axis is generally normal to the film surface due to the nature of crystal
growth, difficultly of in-plane polarization treatment and domain control. Trigonal
piezoelectric material such as LiNbO
3
, LiTaO
3
and quartz are difficult to crystallize (tend
to form amorphous structure) or to obtain a strong preferred orientation in polycrystalline
film.
6mm wurtzite AlN and ZnO film can be easily crystallized, but they tend to develop their
polarization axis (c-axis) perpendicular to the substrate plane. This c-axis oriented film
cannot excite shear wave in the case of standard sandwiched electrode structure.
Crystalline orientation control for both in-plane and out-of-plane direction is necessary to
excite shear wave. One solution is to use an epitaxial growth technique. However, the
combinations of the shear mode piezoelectric film and substrate are limited due to the lattice
mismatch. a-plane ZnO or AlN/r-plane sapphire (Mitsuyu et al., 1980; Wittstruck et al.,
2003), a-plane ZnO/42 Y-X LiTaO
3
(Nakamura et al., 2000) where c-axis in the film is
parallel to the substrate plane have been reported.
Ion beam orientation control technique (Yanagitani & Kiuchi, 2007c), which enables in-plane
and out-of-plane orientation without use of epitaxial growth, is introduced in the third
section. This technique is a good candidate for obtaining c-axis parallel films which excites
pure shear wave without any excitation of longitudinal wave.
2. Electromechanical coupling properties of wurtzite crystal
Elastic and piezoelectric properties of wurtzite crystals vary with direction due to the crystal
anisotropy. Electromechanical coupling changes as a function of the angle between the c-
axis and the applied electric field direction (Foster et al., 1968; Auld, 1973).


503
The analytical model of a thin film resonator is shown in Fig. 1. The electric field is applied
in the x
3
direction. The c-axis is assumed to lie in the x
1
-x
3
plane and be inclined at an angle
with respect to the x
3
direction.

x
1
x
3
c-axis
u
3
u
1
u
(S)
u
(L)
Substrate
Wurtzite piezoelectric film
x
3
Acoustic wave
Electrode

Fig. 1. Analytical model of a thin film resonator
The physical constants of the crystal in each direction are determined by the transformed
coordinate of each constant tensor. Bonds method (Bond, 1943) for transforming the elastic
and piezoelectric constant tensor is introduced below, which can be applied to the constant
tensor with abbreviated subscript notation. For example, the transformation matrix [a] of a
clockwise rotation through an angle about the x
2
-axis is described by:

[ ]
cos 0 sin
0 1 0
sin 0 cos
a

(
(
=
(
(

(5)
The dielectric constant transforms as
[ ] [ ] [ ] [ ] .
T
a a = (6)
The 66 transformation matrix of coefficients M is defined as
[ ]
2 2 2
2 2 2
2 2 2
2 2 2
2 2 2
2 2 2
xx xy xz xy xz xz xx xx xy
yx yy yz yy yz yz yx yx yy
zx zx zz zy zz zz zx zx zy
yx zx yy zy yz zz yy zz yz zy yx zz yz zx yy zx yx zy
zx xx zy xy zz xz xy zz xz zy xz zx xx
a a a a a a a a a
a a a a a a a a a
a a a a a a a a a
M
a a a a a a a a a a a a a a a a a a
a a a a a a a a a a a a a a
=
+ + +
+ +
zz xx zy xy zx
xx yx xy yy xz yz xy yz xz yy xz yx xx yz xx yy xy yx
a a a a
a a a a a a a a a a a a a a a a a a
(
(
(
(
(
(
(
+
(
( + + +

(7)
Finally, using the above transformation matrix, transformed elastic constant and
piezoelectric constant tensors c and e are obtained:


504
[ ] [ ] [ ] [ ]
T
c M c M = , [ ] [ ] [ ] [ ]
T
e M e M = (8)
In the x
2
axis rotation of a hexagonal (6mm) crystal, the transformed stiffness and
piezoelectric constant tensors c and e are given by
[ ]
11 12 13 15
12 22 23 25
13 23 33 35
44 46
15 25 35 55
46 66
0 0
0 0
0 0
0 0 0 0
0 0
0 0 0 0
c c c c
c c c c
c c c c
c
c c
c c c c
c c

(
(

(
(
=
(

(
(

(

(

,
[ ]
11 12 13 15
24 26
31 32 33 35
0 0
0 0 0 0
0 0
e e e e
e e e
e e e e

(
(
=
(
(

(9)
In case, wave propagation toward x
3
direction is only focused, the term of
1
x and
2
x can be ignored. Thus, the wave motion equation for the x
3
direction is given by
mechanical displacement component u
1
, u
2
and u
3
:

2
31 1
2
3
T u
x t

=

(10a)

2
33 3
2
3
T u
x t

=

(10b)

2
32 2
2
3
T u
x t

=

(10c)
where

1 3
31 55 35 35
3 3 3

= + +

E E
u u
T c c e
x x x
(11a)

1 3
33 35 33 33
3 3 3

= + +

E E
u u
T c c e
x x x
(11b)

2
32 44
3
E
u
T c
x
(11c)
As div D = 0, the electrostatic equation is given by

2 2 2
3 1 3
35 33 33 2 2 2
3 3 3 3
0
S
D u u
e e
x x x x

= + =

(12)
In Eqs. (10)-(12), T
31
and T
33
are stress components, D
3
is the electric displacement, c
33
E
, c
35
E

and c
55
E
are the stiffness constants with constant electric field, e
33
and e
35
are piezoelectric
constants,
33
S
and
35
S
are dielectric constants with constant strain, and is the electric
potential.


505
Equation (10c) describes a pure shear wave with a u
2
displacement component in the x
2

direction and propagates along the x
3
direction with a phase velocity of
44
c . Eqs. (10a)
and (10b) represent a quasi-longitudinal wave and quasi-shear wave. These waves
incorporate u
1
, u
3,
and , which are coupled with each other. It is well known that Eqs (10a),
and (10b) have plane-wave solutions:

1
3
3
exp
u A
x
u B j t
v
C

| | | |
| | | |
=
`
|
| |
\ . )
| |
\ . \ .
(13)
Substituting Eq. (13) into Eqs. (11) and (12), the simultaneous equations are obtained

2
55 35
2
35 33
35 33 33
0
0 0,
D D
D
S
c v c A
c c v B
e e C
| | | |
|
|
=
|
|
| |

\ .
\ .
(14)
where
( )
2
33 33 33 33
,
D E S
c c e = +
( )
35 35 33 35 33
,
D E S
c c e e = + (15)
( )
2
55 55 35 33
.
D E S
c c e = +
A, B and C are all nonzero when the coefficient matrix in Eq. (14) is zero. From this condition,
we obtain the phase velocity v
(L, S)
of a quasi-longitudinal wave and quasi-shear wave:

( )
1
2 2 2
,
33 55 33 55 35
2 2
D D D D D
L S
c c c c c
v

(
| | | | +
(
= +
| |
(
\ . \ .

(16)
Figure 2 shows the calculated results of phase velocity of a quasi-longitudinal wave and
quasi-shear wave for a ZnO crystal as function of the angle between the c-axis and x
3

direction. Physical constants in a ZnO single crystal reported by Smith were used in the
calculation (Smith, 1969).
The general solutions for u
1
, u
3
and are given by

1 1 2
3 3
3 1 2 ( ) ( )
1 2
exp exp
u A A
x x
u B j t B j t
V V
C C

+
| | | | | |
| | | | | | |
= +
` `
| |
| | |
\ . \ . ) )
| | |
\ . \ . \ .
(17)
and

1 2
1 2
B A
A B
= (18)


506
6400
6300
6200
6100
6000
5900
L
o
n
g
i
t
u
d
i
n
a
l

w
a
v
e

v
e
l
o
c
i
t
y

(
m
/
s
)
90 80 70 60 50 40 30 20 10 0
(deg.)
3300
3200
3100
3000
2900
2800
2700
S
h
e
a
r

w
a
v
e

v
e
l
o
c
i
t
y

(
m
/
s
)
Quasi-longitudinal wave
Quasi-shear wave

Fig. 2. Phase velocity of quasi-longitudinal wave and quasi-shear wave for a ZnO crystal as
function of the angle between the c-axis and x
3
direction
is derived from Eqs. (14) and (16). It can be seen that the displacement components of the
quasi-longitudinal wave and quasi-shear wave are perpendicular to each other. From Eqs.
(14) and (16), the angle
L
between the quasi-longitudinal wave displacement u
3
and the x
3

direction and the angle
S
between the quasi-shear wave displacement u
1
and the x
1

direction are given by

1 1
1
tan
L
A
B

| |
=
|
\ .
,
1 2
2
tan
S
B
A

| |
=
|
\ .
(19)
The extensional and shear effective piezoelectric constants e
(L)
eff
and e
(S)
eff
are defined as

( )
35 33
sin cos
L
eff L L
e e e = + ,
( )
35 33
cos sin
S
eff S S
e e e = (20)
Thus, the quasi-longitudinal and quasi-shear-mode electromechanical coupling coefficients
k
(L)
(transformed k
33
) and k
(S)
(transformed k
15
) are

( ) ( ) ( )
2 2 2
( ) ( ) ( )
33
S S
k e e V

= ,
( )
( )
( )
( ) ( )
2 2 2
( )
33
L L
k e e V
+
= (21).
Finally, Figs. 3 (a) and (b) show the calculated angle and the electromechanical coupling
coefficients (k values) of the quasi-longitudinal and quasi-shear waves for the ZnO crystal as
function of the angle (Foster et al., 1968)
From these figures, we can see a relatively large shear-mode electromechanical coupling k
15
= 0.39 at c-axis tilt angle of = 28. Several author reported FBAR (film bulk acoustic
resonator)-type viscosity sensor and biosensor, consisting of c-axis tilted wurtzite films
(Weber et al., 2006; Link et al., 2007; Wingqvist et al., 2007, 2009, 2010; Yanagitani, 2010,
2011a). However, the thickness extensional mode (longitudinal wave mode) also has the
coupling of k
33
= 0.155 and the displacement inclination angle of
S
= 4.1 at angle of = 28.
This indicates that the resonator excites both thickness extensional and shear mode
(longitudinal and shear wave modes), and the shear displacement direction is not
perpendicular to the propagation direction. Larger
S
values may result in energy leakage


507
-5
-4
-3
-2
-1
0
1
2
3
4
5

(
d
e
g
.
)
80 60 40 20 0
(deg.)
(a)
Quasi-thickness extensional mode
Quasi-thickness shear mode

0.4
0.3
0.2
0.1
0.0
k
3
3
o
r

k
1
5
80 60 40 20 0
(deg.)
(b)
Quasi-thickness extensional mode
Quasi-thickness shear mode

Fig. 3. (a) Angle between the wave displacement u

and the x direction and
(b) electromechanical coupling coefficient of the quasi-longitudinal and quasi-shear waves
for the ZnO crystal as function of the angle between the c-axis and x
3
direction
due to mode conversion in the reflection plane. This induces the decrease of Q value. Both of
the no extensional mode coupling and small
S
values of 0.38 can be obtained at = 43,
however, it is difficult to adjust such as large c-axis tilt angle in a large area deposition. One
option is to use a pure-shear-mode ( = 90) resonator to satisfy both the conditions of no
extensional coupling and
S
= 0. Pure shear mode excitation can be achieved by two electric
field-orientation combination. One is to apply the cross-electric field to c-axis parallel film
by sandwiched electrode (Yanagitani et al., 2007d), and the other is to apply the in-plane
electric field to c-axis normal film by IDT electrode (Corso et al., 2007; Milyutin et al., 2008,
2010). Of course, the latter is the easiest way to obtain pure shear mode because deposition


508
technique of c-axis normal film has been well established, but effective electrometrical
coupling is weak (k
eff
=0.04-0.06) (Corso et al., 2007; Milyutin et al., 2008). The former has
large electrometrical coupling (k
15
=0.24) (Yanagitani et al., 2007a), and recently the c-axis
parallel oriented film can be easily obtained by using ion beam orientation control technique
(presented in next section), even in a large area (Kawamoto et al., 2010).
3. Ion beam orientation control technique for shear mode piezoelectric films
3.1 Ion beam orientation control of wurtzite thin film by ion beam irradiation
Polycrystalline films tend to grow in their most densely packed plane parallel to the
substrate plane. Bravais proposed the empirical rule that the growth rate of the crystal plane
is proportional to the surface atomic density. Namely, the lattice plane with higher surface
atomic density grows more rapidly. Curie argued that the growth rate perpendicular to a
plane is proportional to the surface free energy (Curie, 1885).
Ion bombardment during film deposition can modify this preferred orientation of the films.
This is usually explained by a change in anisotropy of the growing rate of the crystal plane
in the grain, which is reflected by the difference in the degree of the ion channeling effect or
ion-induced damage in the crystal plane (Bradley et al., 1986; Ensinger, 1995; Ressler et al.,
1997; Dong & Srolovitz, 1999). For example, during ion beam irradiation, the commonly
observed <111> preferred orientation in a face-centered cubic film changes to a <110>
preferred orientation, which corresponds to the easiest channeling direction (Van Wyk &
Smith, 1980; Dobrev, 1982). In-plane texture controls have also been achieved by optimizing
the incident angle of the ion beam (Yu et al., 1985; Iijima et al., 1992; Harper et al., 1997;
Kaufman et al., 1999; Dong et al., 2001; Park et al., 2005).
In wurtzite films, for example, the surface energy densities of the (0001), (11 2 0) and (10 1 0)
planes of the ZnO crystal are estimated to be 9.9, 12.3, 20.9 eV/nm
2
, respectively (Fujimura
et al., 1993). The (0001) plane has the lowest surface density. Thus, the ZnO film tends to
grow along the [0001] direction. When wurtzite crystal is irradiated with ion beam, the most
densely packed (0001) plane should incur more damage than the (10 1 0) and (11 2 0) planes,
which correspond to channeling directions toward the ion beam irradiation. We can
therefore expect that the thermodynamically preferred (0001) oriented grain growth will be
disturbed by ion damage so that the damage-tolerant (10 1 0) or (11 2 0) orientated grains (c-
axis parallel oriented grain) will preferentially develop instead.
On this basis, in-plane and out-of-plane orientation control of AlN and ZnO films by means
of ion beam-assisted deposition technique, such as evaporation (Yanagitani & Kiuchi, 2007c)
and sputtering (Yanagitani & Kiuchi, 2007e, 2011b) was achieved. c-axis parallel oriented
can be obtained even in a conventional magnetron sputtering technique using a low
pressure discharge ( <0.1 Pa) (Yanagitani et al., 2005) or RF substrate bias (Takayanagi,
2011), which leads ion bombardment on the substrate. Figure 4 shows the XRD patterns of
the ZnO films deposited with various ion energy and amount of flux in ion beam assisted
evaporation (Yanagitani & Kiuchi, 2007c). Table 1 shows the ion current densities in the case
of Large ion flux and Small ion flux in Fig 4. The tendency of the (10 1 0) orientation is
enhanced with increasing ion energy and amount of ion irradiation, demonstrating that the
ion bombardment induced the (0001) orientation to change into a (10 1 0) orientation, which
corresponds to the ion channeling direction.


509
Ion energy A: Large ion flux B: Small ion flux
0.05 keV
0.25 keV
0.5 keV
190 A/cm
2
140 A/cm
2
0.75 keV
220 A/cm
2
130 A/cm
2
1.0 keV
240 A/cm
2
120 A/cm
2
0-5 A/cm
2
30-50 A/cm
2

Table 1. Ion current densities in Large ion flux and Small ion flux

26
24
22
20
18
16
14
12
10
8
6
4
2
0
I
n
t
e
n
s
i
t
y

(
k
c
p
s
)
60 55 50 45 40 35 30 25
2 (deg.)
Without ion
irradiation
0.25 keV
0.75 keV
1 keV
0.5 keV
Ion energy
0.05 keV
0
0
0
2
C
u

1
1
1
1
0
1
1
1
0
1
0
(A) : Large ion flux
(B) : Small ion flux
(A)
(B)
(A)
(B)
(A)
(B)
(A)
(B)
(A)
(B)
(A)
(B)

Fig. 4. 2 scan XRD patterns of the ZnO films deposited without ion irradiation, and with
ion irradiation of 0-1 keV with Large ion flux and Small ion flux (Yanagitani & Kiuchi,
2007c)
Figure 5 shows the XRD patterns of the samples deposited under the conditions that various
RF and DC bias are applied to the substrate. Although any dramatic change in usual (0001)


510
preferred orientation is not occurred in the case of positive or negative DC bias, (0001)
orientation changed to (11 2 0) and (10 1 0) orientation with the increase of RF bias power
which induces the bombardment of positive ion on substrate. Interestingly, the order of the
appearance of the (0001) to (11 2 0) and (10 1 0) corresponds to the order of increasing
surface atomic density, which may be the order of damage tolerance against ion
bombardment.
In order to excite shear wave in the c-axis parallel film, c-axis is required to orient not only
in out-of-plane direction but also in in-plane direction. The ion beam orientation control
technique allows us to control even in in-plane c-axis direction and polarization by the
direction of beam incident direction (Yanagitani et al., 2007d).

I
n
t
e
n
s
i
t
y

(
a
.
u
.
)
70 65 60 55 50 45 40 35 30 25 20
2 (deg.)
(
0
0
0
2
)
0.1
-100 V
-200 V
100 V
(
1
0
1
0
)
(
1
1
2
0
)
Non-bias
150 W
250 W
100 W
50 W
200 W
80 MHz RF bias power:
DC bias voltage:
Film
thickness:
6.9 m
6.9 m
7.5 m
8.8 m
10.0 m
9.8 m
9.8 m
9.6 m
9.8 m
10 kcps

Fig. 5. 2- scan XRD patterns of the samples deposited without bias, with 80 MHz RF bias
of 50 to 250 W, or with -200 to 100 DC bias. All samples were measured at the center of the
bias electrode (Takayanagi et al., 2011)
4. Method for determining k values in piezoelectric thin films
4.1 k value determination using as-deposited structure (HBAR structure)
A method for determining piezoelectric property in thin films is described in this section. In
general, electromechanical coupling coefficient (k value) in thin film can be easily
determined by series and parallel resonant frequency of a FBAR consisting of top electrode
layer/piezoelectric layer/bottom electrode layer or SMR (Solidly mounted resonator)
consisting of top electrode layer/piezoelectric layer/bottom electrode layer/Bragg reflector.
In case thickness of electrode film is negligible small compared with that of piezoelectric
film. k of the piezoelectric film can be written as follows (Meeker, 1996):


511

2
tan
2 2
p s
s
p p
f f
f
k
f f

| |
= |
|
\ .
(22)
where f
p
and f
s
are the parallel resonant frequency and series resonant frequency,
respectively.
However, it takes considerable time and effort to fabricate FBAR structure which have self-
standing piezoelectric layer. It is convenient if k value can be determined from as deposited
structure, namely so-called an HBAR (high-overtone bulk acoustic resonator) or composite
resonator structure consisting of top electrode layer/piezoelectric layer/bottom electrode
layer/thick substrate. Methods for determining the k value of the films from HBAR
structure are more complex than that for the self-supported single piezoelectric film
structure (FBAR structure). Several groups have investigated methods for the determination
of k
t
value from the HBAR structure (Hickernell, 1996; Naik, et al., 1998; Zhang et al., 2003).
One of the easiest ways of k determination is to use a conversion loss characteristic of the
HBAR structure. When the thickness of electrode layers is negligible small compared with
that of piezoelectric layer, capacitive impedance of resonator is equal to the electrical source
impedance, and k value of the piezoelectric layer is smaller than 0.3, conversion loss CL is
approximately represented by k value at parallel resonant frequency (Foster et al., 1968):

10 2
10log
8
s
p
Z
CL
k Z
(23)
where, Z
s
and Z
p
is acoustic impedance of the substrate and piezoelectric layer, respectively.
However, various inhomogeneities sometimes exist in the film resonator, such as non-
negligible thick and heavy electrode layers, thickness taper, or the piezoelectrically inactive
layer composed of randomly oriented gains growing in the initial stages of the deposition.
In this case, the k values of the film can be determined so as to match the experimentally
measured conversion losses (CL) of the resonators with theoretical minimum CL by taking k
value as adjustable parameter. The theoretical CL in this case can be calculated by Masons
equivalent circuit model including electrode layer, film thickness taper and piezoelectrically
inactive layer. This method allows various inhomogeneous effect of film to be taken into
account (Yanagitani et al., 2007b, 2007c).
4.2 Experimental method to estimate conversion loss of HBAR structure
The experimental CL of HBAR can be determined from reflection coefficients (S
11
) of the
resonators, which can be obtained using a network analyzer with a microwave probe. The
inverse Fourier transform of S
11
frequency response of the resonator gives the impulse
response of the resonator in the time domain. In the HBAR structure, the impulse response
is expected to include echo pulse trains reflected from the bottom surface of the substrate,
and the insertion loss of resonator can be obtained from the Fourier transform of the first
echo in this impulse response. This experimental insertion loss IL
experiment
includes doubled
CL in the piezoelectric film and round-trip diffraction loss DL and round-trip propagation
loss PL in the silica glass substrate. Therefore, CL can be expressed as

( )
exp
1
,
2
eriment
CL IL DL PL = (24)


512
where diffraction loss DL can be calculated according to the method reported by Ogi et al.
(Ogi et al., 1995). This method is based on integration of the velocity potential field in the
divided small transducer elements, which allows calculation of the DL with electrode areas
of various shapes. The round-trip propagation loss PL is given as

2
2 ,
s
s
PL d
f
= (25)
where d
s
is the thickness of the substrate,
s
represents the shear wave attenuation in the
substrate, for example,
s
/ f
2
= 19.910
-16
(dBs
2
/m) in silica glass substrate (Fraser, 1967).
4.3 Conversion loss simulation in HBAR by Masons equivalent circuit model
Electromechanical coupling coefficient k can be estimated by comparing an experimental CL
with a theoretical CL of the HBAR. One-dimensional Masons equivalent circuit model is
convenient tool for simulating theoretical CL of the resonator. Generally, in case non-
piezoelectric elastic solid vibrates in thickness mode, its can be described as T-type
equivalent circuit (Fig. 6 (a)) where F
1
and F
2
are force and v
1
and v
2
are particle velocity
acting on each surface of elastic solid. Piezoelectric elastic solid can be represented as the
Masons three ports equivalent circuit which includes additional electric terminal
concerning electric voltage V and current I (Fig. 6 (b)) (Mason, 1964). Here, is
propagation constant, Z is acoustic impedance and d
p
is thickness of elastic solid. To take
account of attenuation of vibration, mechanical quality factor Q
m
is defined as Q
m
= c
r
/c
i

where c
r
and c
i
are real part and imaginary part of elastic constant, respectively. Using
mechanical quality factor Q
m
, propagation constant and acoustic impedance Z are given
as:

( )
{ }
1 1
r m
j
c j Q
=
+
, ( )
{ }
1 1
r m
Z S c j Q = + (26)
where is density of the elastic solid and S is electrode area of the resonator.
Static capacitance C
0
and ratio of transformer
0
in the circuit are given as:

0 11
S
p
S
C
d
= ,
1
2 2
0
15
0 2
15
,
1
p p
p
C v Z
k
d k
(
| |
= (
|
( \ .

(27),
where d is the thickness of the layers,
11
S
is permittivity, and v is the velocity of the shear
wave. Subscript p, e1, e2 and s respectively represent piezoelectric layer, top electrode layer,
bottom electrode layer and substrate. k value affects the equivalent circuit through the ratio
of transformer
0
.
Equivalent circuit for the over-moded resonator structure is given in Fig. 7 by cascade
arranging non-piezoelectric and piezoelectric part as described in Figs. 6 (a) and (b).
Substrate thickness is assumed infinite to ignore reflection waves from bottom surface of the
substrate in this case. When the surface of the top electrode is stress-free, the acoustic input
port is shorted. As top electrode part circuit can be simplified, three-port circuit in Fig. 7 is
transformed to the two-ports circuit shown in Fig. 8 (Rosenbaum, 1988).


513

v
1
v
2
Z
p
tanh (
p
d
p
/ 2) Z
p
tanh (
p
d
p
/ 2)
Z
p
/sinh (
p
d
p
)
F
1
F
2
v
1
v
2
Z
p
tanh (
p
d
p
/ 2) Z
p
tanh (
p
d
p
/ 2)
Z
p
/sinh (
p
d
p
)
- C
0
C
0
i
V
1 :
0
F
1
F
2
(a) (b)

Fig. 6. Equivalent circuit model of (a) non-piezoelectric (b) piezoelectric elastic solid

Piezoelectric
layer
Electrode
layer
Substrate
Electrode
layer
Piezoelectric
layer
Electrode
layer
Substrate
Electrode
layer
Z
p
tanh (
p
d
p
/2)
Z
p
/ sinh (
p
d
p
)
1 :
0
Z
e
tanh (
e2
d
e2
/ 2)
Z
e
/ sinh (
e2
d
e2
)
Z
s
Z
e1
tanh (
e1
d
e1
/ 2)
Z
e1
/ sinh (
e1
d
e1
)
C
0
-C
0
Piezoelectric
layer
Electrode
layer
Electrode
layer
Substrate Air

Fig. 7. Equivalent circuit model of the over-moded resonator structure
It is convenient to derive whole impedance of the circuit by using ABCD-parameters (Paco
et al., 2008) As shown in Eqs. (28)-(32), ABCD-parameters of whole circuit is derived
multiplying each circuit element.

0
0
1 / 0
0
Transformer
F

(
=
(

,
0
0
1 0 1 1 /
1 0 1
Electric port
j C
F
j C

( (
=
( (

,
1
0 1
s
Substrate
Z
F
(
=
(

(28)


514
{ }
1 1 1
1 0
1 /sinh ( )
0 1 1 tanh( /2) tanh( /2) 1
p p p
Piezo Electrode layer
e e e p p p
Z d
F
Z d Z d

+
(
(
= (
(
+
(

1 tanh( /2)
0 1
p p p
Z d (
(

(29)
2 2 2 2 2 2
2 2 2
1 tanh( /2) 1 0 1 tanh( /2)
0 1 sinh( ) / 1 0 1
e e e e e e
Counter electrode
e e e
Z d Z d
F
d Z

( ( (
=
( ( (

(30)

1
0 1
s
Substrate
Z
F
(
=
(

(31)

Piezo Elecrode layer Over moded resonator Electric port Transformer Counterelectrode Substrate
F F F F F F
+
= (32)

Z
e1
tanh (
e1
d
e1
/ 2)
Z
e1
/ sinh (
e1
d
e1
) Z
e1
tanh (
e1
d
e1
)
-C
0
C
0
1:
0
Z
e1
tanh (
e1
d
e1
)
Z
p
tanh (
p
d
p
/ 2)
Z
p
/ sinh (
p
d
p
)
Z
e2
/ sinh (
e2
d
e2
)
Z
e2
tanh (
e2
d
e2
/ 2) Z
s
F
2
v
2
V
I
Top electrode part

Fig. 8. Simplification of equivalent circuit model for over-moded resonator structure
Insertion loss IL is expressed as the ratio of the signal power delivered from a source into
load resistance to the power delivered from a source into the inserted network. IL of the
resonators can be calculated with the following equation using conductance of the electrical
source G
S
(0.02 S), input conductance G
f
, and susceptance B
f
of the circuit model, which can
be derived from ABCD-parameter to Y-parameter conversion of eq. (32):

( )
( )
2
2
2
10
20log .
4
S
f f
S f f
S
G
e G jB
G G B
IL
G

+
`
+ +
)
= (33)


515
Hence the CL is

( )
10
2 2
4
10log .
2
S f
S f f
G G
IL
CL
G G B
= =
+ +
(34)
4.4 k value determination from conversion loss curves
Figure 9 (a) shows the pure shear mode theoretical and experimental CL curves of the c-axis
parallel film HBAR as an example. By comparing experimental curve with theoretical curves

60
50
40
30
20
10
0

C
o
n
v
e
r
s
i
o
n

l
o
s
s

(
d
B
)
900 800 700 600 500 400 300 200 100 0
Frequency (MHz)
k
15
= 0.12
k
15
= 0.16
k
15
= 0.20
k
15
= 0.26
Propagation loss
Experiment
Mason's model
(a)

60
50
40
30
20
10
0

C
o
n
v
e
r
s
i
o
n

l
o
s
s

(
d
B
)
900 800 700 600 500 400 300 200 100 0
Frequency (MHz)
Propagation loss
Experiment
Model including inactive layer

d
n
= 1m
d
n
= 2m
d
n
= 0m
(b)

Fig. 9. Frequency response of the experimental shear mode CL (open circles). (a) The simulated
shear mode CL curves (solid line) in various k
15
values and (b) the curve simulated by the model
including various thickness of piezoelectrically inactive layer (Yanagitani & Kiuchi, 2007c)


516
at minimum CL point (near the parallel resonant frequency), we can determine the k
15
value
of the film. As shown in Fig. 9 (b), effective thickness of the piezoelectrically inactive layer d
n

in the initial stages of the deposition also can be estimated from comparison of the curves.
Figure 10 shows the correlation between k
15
value and crystalline orientation of the film.
FWHM values of -scan and -scan curve of the XRD (X-ray diffraction) pole figure show
the degree of crystalline orientation in out-of plane and in-plane, respectively. Thicker films
tend to have large k
15
values even though they have same degree of crystalline orientation as
thinner one. This kind of correlations and inhomogeneities characterization in wafer can be
easily obtained from as-deposited film structure, by using present k value determination
method.
4.5 Conclusion
In this chapter, shear mode piezoelectric thin film resonators, which is promising for
the acoustic microsensors operating in liquid, were introduced. Theoretical predictions of
electromechanical coupling and tilt of wave displacement as functions of c-axis tilt angle
showed that pure shear mode excitation by using c-axis parallel oriented wurtzite
piezoelectric films expected to achieve high-Q and high-coupling sensor. Fabrication of
c-axis parallel oriented films by ion beam orientation control technique and
characterization of the film by a conversion loss of the as-deposited resonator structure
were discussed.

0.30
0.20
0.10
0.00
k
1
5

700 600 500 400 300 200 100 0
-FWHM -FWHM (deg.
2
)
Single crystal (k
15
=0.26)
12 10 8 6 4 2
Film thickness (m)

Fig. 10. k
15
values of the ZnO piezoelectric layers as a function of multiplication of -scan
and -scan profile curve FWHM values extracted from XRD pole figure (indicating the
degree of crystalline orientation in out-of-plane and in-plane) (Yanagitani et al., 2007b)


517
5. References
Auld, B. A. Acoustic Fields and Waves in Solid. (1973). Vol. 1, pp. 7376, A Wiley-
Interscience Publication.
Bond, W. (1943). The Mathmatics of the Physical Properties of Crystals. Bell System Technical
Journal, Vol. 22, pp. 172.
Bradley R. M.; Harper J. M. E. & Smith, D. A. (1986). Theory of Thinfilm Orientation by Ion
bombardment during deposition. J. Appl. Phys., Vol. 60, No. 12, pp. 41604164.
Corso, C. D.; Dickherber, A. & Hunt, W. D. (2007). Lateral Field Excitation of Thickness
Shear Mode Waves in a Thin Film ZnO Solidly Mounted Resonator. J. Appl. Phys.
Vol. 101, pp. 054514-1054514-7.
Curie, P. (1885). Bull. Soc. Franc. Miner. Crist. Vol. 8 p. 145.
Dobrev, D. (1982). Ion-beam-induced Texture Formation in Vacuum-condensed Thin Metal
Films. Thin Solid Films, Vol. 92, No. 1-2, pp. 4153.
Dong, L. & Srolovitz, D. J. (1999). Mechanism of Texture Development in Ion-beam-assisted
Deposition. J. Appl. Phys. Lett., Vol. 75, No. 4, pp. 584586.
Dong, L.; Srolovitz, D. J.; Was, G. S.; Zhao, Q. & Rollett, A. D. (2001). Combined Out-of-
plane and In-plane Texture Control in Thin Films using Ion Beam Assisted
Deposition. J. Mater. Res., Vol. 16, No. 1, pp. 210216.
Ensinger, W. (1995). On the Mechanism of Crystal Growth Orientation of Ion Beam Assisted
Deposited Thin Films. Nucl. Instrum. Meth. Phys. Res. B, Vol. 106, pp. 142146.
Foster, N. F.; Coquin, G. A.; Rozgonyi, G. A. & Vannatta, F. A. (1968). Cadmium Sulphide
and Zinc Oxide Thin-Film Transducers. IEEE Trans. Sonic. Ultrason., Vol. 15, pp. 28
41.
Fraser, D. B.; Krause, J. T. & Meitzler, A. H.; (1967). Physical Limitations on the Performance
of Vitreous Silica in Highfrequency Ultrasonic and Acoustooptical Devices. Appl.
Phys. Lett. Vol. 11, No. 10, pp. 308310.
Fujimura, N.; Nishihara, T.; Goto, S.; Xu, J. & Ito, T. (1993). Control of Preferred Orientation
for ZnOx Films: Control of Self-texture. J. Crystal Growth, Vol. 130, pp. 269279.
Harper, J. M. E.; Rodbell, K. P.; Colgan, E. G. & Hammond, R. H. (1997). Control of In-plane
Texture of Body Centered Cubic Metal Thin Films. J. Appl. Phys. Vol. 82, pp. 4319
4326.
Hickernell, F. S. (1996). Measurement Techniques for Evaluating Piezoelectric Thin Films.
Proc. IEEE Ultrason. Symp., pp. 235242.
Iijima, Y.; Tanabe, N.; Kohno, O. & Ikeno, Y. (1992). In-plane Aligned YBa
2
Cu
3
O
7-x
Thin
Films Deposited on Polycrystalline Metallic Substrates. Appl. Phys. Lett., Vol. 60,
No. 6, pp. 769771.
Kaufman, D. Y.; DeLuca, P. M.; Tsai, T. & Barnett, S. A. (1999). High-rate Deposition of
Biaxially Textured Yttria-stabilized zirconia by Dual Magnetron Oblique
Sputtering. J. Vac. Sci. Technol. A, Vol. 17, pp. 28262829.
Kawamoto, T.; Yanagitani T.; Matsukawa, M.; Watanabe, Y.; Mori1, Y.; Sasaki, S. & Oba M.
(2010). Large-Area Growth of In-Plane Oriented (11-20) ZnO Films by Linear
Cathode Magnetron Sputtering. Jpn. J. Appl. Phys. Vol. 49 pp. 07HD16-107HD16-4.
Kushibiki, J-I.; Akashi, N.; Sannomiya, T.; Chubachi, N.; & Dunn, F. (1995). VHF/UHF
Range Bioultrasonic Spectroscopy System and Method. IEEE Trans. Ultrason.,
Ferroelect., Freq. Contr., Vol. 42 No. 6, pp. 10281039


518
Link, M.; Weber, J.; Schreiter, M.; Wersing, W.; Elmazria, O. & Alonot, P. (2007). Sensing
Characteristic of High-freqency Shear Mode Resonators in Glycerol Solutions. Sens.
Actuators B, Vol. 121, No.2, pp. 372378.
Matsumoto, Y.; Ujiie, T. & Kushibiki, J-I. (2000). Measurement of Shear Acoustic Properties
of Water using the Ultrasonic Reflectance Method in Pulse-mode Operation. Spring
Meeting of Acoustical Society of Japan (in Japanese)
Meeker, T. R. (1996). IEEE Stabdard on Piezoelectricity (ANSI/IEEE Std. 176-1987). IEEE
Trans. Ultrason., Ferroelect., Freq. Contr., Vol. 43, No. 5, pp. 719772..
Milyutin, E. & Mural, P. (2010). Electro-Mechanical Coupling in Shear-Mode FBAR with
Piezoelectric Modulated Thin Film. IEEE Trans. Ultrason., Ferroelect., Freq. Contr.,
Vol. 58, No. 4, pp. 685688.
Milyutin, E.; Gentil, S. & Mural, P. (2008). Shear Mode Bulk Acoustic Wave Resonator
Based on c-axis Oriented AlN Thin Film. J. Appl. Phys. Vol. 104, pp. 084508-1
084508-6.
Mitsuyu, T.; Ono S. & Wasa, K. (1980). Structures and SAW Properties of Rf-sputtered
Single-crystal Films of ZnO on Sapphire. J. Appl. Phys., Vol. 51, No. 5, pp. 2464
2470.
Naik, B. S.; Lutsky, J. J.; Rief R.; & Sodini, C. D. (1998). Electromechanical Coupling
Constant Extraction of Thin-lm Piezoelectric Materials Using a Bulk Acoustic
Wave Resonator. IEEE Trans. Ultrason., Ferroelect., Freq. Contr., Vol.45, No.1, pp.
257263.
Nakamura, K.; Shoji, T. & Kang, H. (2000). ZnO Film Growth on (0112) LiTaO
3
by Electron
Cyclotron Resonance-assisted Molecular Beam Epitaxy and Determination of Its
Polarity. Jpn. J. Appl. Phys., Vol. 39, No. 6A, pp. L534L536.
Ogi, H.; Hirao, M.; Honda T. & Fukuoka, H. (1995). Ultrasonic Diffraction from a Transducer
with Arbitrary Geometry and Strength Distribution. J. Acoust. Soc. Am. Vol. 98, No.
2, pp. 11911198.
Paco, P.; Menndez, . & Corrales E. (2008) Equivalent Circuit Modeling of Coupled
Resonator Filters. IEEE Trans. Ultrason., Ferroelect., Freq. Contr., Vol. 55, No. 9, pp.
20302037.
Park, S. J.; Norton, D. P. & Selvamanickam, V. (2005). Ion-beam Texturing of Uniaxially
Textured Ni Films. Appl. Phys. Lett., Vol. 87, pp. 031907031909.
Ressler, K. G.; Sonnenberg, N. & Cima, M. J. (1997). Mechanism of Biaxial Alignment of
Oxide Thin Films during Ion-Beam-Assisted Deposition. J. Am. Ceram. Soc., Vol. 80,
No. 10, pp. 26372648.
Rosenbaum, J. F. (1988). Bulk Acoustic Waves: Theory and Devices. Artech House Boston
London.
Sauerbrey, G. (1959) Verwendung von Schwingquarzen zur Wgung Dnner Schichten und
zur Mikrowgung. Z. Physik, Vol. 155, pp. 206222.
Smith, R. T. & Stubblefield, V. E. (1969). Temperature Dependence of the Electroacoustical
Constants of Li-doped ZnO Single Crystals. J. Acoust. Soc. Am., Vol. 46, pp. 105.
Takayanagi, S.; Yanagitani, T.; Matsukawa, M. & Watanabe Y. (2010). A Simple Technique
for Obtaining (11-20) or (10-10) Textured ZnO Films by RF Bias Sputtering, Proc.
2010 IEEE Ultrason. Symp. pp. 10601063.


519
Van Wyk, G. N. & Smith, H. J. (1980). Crystalline Reorientation Due to Ion Bombardment.
Nucl. Instrum. Meth., Vol. 170, pp. 433439.
Weber, J.; Albers, W. M.; Tuppurainen, J.; Link, M.; Gabl, R.; Wersing, W. & Schreiter, M.
(2006). Shear Mode FBARs as Highly Sensitive Liquid Biosensors. Sens. Actuators A,
Vol. 128, No. 2, pp. 8488.
Wingqvist G.; Anderson H.; Lennartsson C.; Weissbach T.; Yantchev V. & Lloyd Spetz A.
(2009). On the Applicability of High Frequency Acoustic Shear Mode Biosensing in
View of Thickness Limitations Set by The Film Resonance. Biosens. Bioelectron., Vol.
24 No. 11, pp. 33873390.
Wingqvist, G. (2010). AlN-based Sputter-deposited Shear Mode Thin Film Bulk Acoustic
Resonator (FBAR) for Biosensor Applications. Surf. Coat. Tech., Vol. 205, No. 5, pp.
12791286
Wingqvist, G.; Bjurstrom, J.; Liljeholm, L.; Yantchev, V. & Katardjiev, I. (2007). Shear mode
AlN Thin Film Electro-acoustic Resonant Sensor Operation in Viscous Media. Sens.
Actuators B, Vol. 123, No. 1, pp. 466473.
Wittstruck, R. H.; Tong, X.; Emanetoglu, N. W.; Wu, P.; Chen, Y.; Zhu, J.; Muthukumar, S.;
Lu, Y.; & Ballato, A. (2003) Characteristic of Mg
x
Zn
1-x
O Thin Film Bulk Acoustic
Wave Devices. IEEE Trans. Ultrason., Ferroelect., Freq. Contr., Vol. 50, pp. 1272
1277.
Yanagitani, T. & Kiuchi, M. (2007c). Control of In-plane and Out-of-plane Texture in Shear
Mode Piezoelectric ZnO Films by Ion-beam Irradiation. J. Appl. Phys., Vol. 102, pp.
044115-1044115-7.
Yanagitani, T. & Kiuchi, M. (2007e). Highly Oriented ZnO Thin Films Deposited by Grazing
Ion-beam Sputtering: Application to Acoustic Shear Wave Excitation in the GHz
Range. Jpn. J. Appl. Phys. Vol. 46, pp. L1167L1169.
Yanagitani, T. & Kiuchi, M. (2011b). Texture Modification of Wurtzite Piezoelectric Films by
Ion Beam Irradiation. Surf. Coat. Technol., in press.
Yanagitani, T., Matsukawa, M., Watanabe, Y., Otani, T. (2005). Formation of Uniaxially (11-2
0) Textured ZnO Films on Glass Substrates. J. Cryst. Growth, Vol. 276, No. 3-4, pp.
424-430.
Yanagitani, T.; Arakawa, K.; Kano, K.; Teshigahara, A.; Akiyama, M. (2010). Giant Shear
Mode Electromechanical Coupling Coefficient k
15
in c-axis Tilted ScAlN Films. Proc.
2010 IEEE Ultrason. Symp., pp. 20952098.
Yanagitani, T.; Kiuchi, M.; Matsukawa, M. & Watanabe, Y. (2007b). Shear Mode
Electromechanical Coupling Coefficient k
15
and Crystallites Alignment of (11-20)
Textured ZnO Films. J. Appl. Phys., Vol. 102, pp. 024110-1024110-7.
Yanagitani, T.; Kiuchi, M.; Matsukawa, M. & Watanabe, Y. (2007d). Characteristics of Pure-
shear Mode BAW Resonators Consisting of (11-20) Textured ZnO Films. IEEE
Trans. Ultrason., Ferroelect., Freq. Contr., Vol. 54, No. 8, pp. 16801686.
Yanagitani, T.; Mishima, N.; Matsukawa, M. & Watanabe, Y. (2007a). Electromechanical
Coupling Coefficient k
15
of Polycrystalline ZnO Films with the c-axes Lie in the
Substrate Plane. IEEE Trans. Ultrason., Ferroelect., Freq. Contr., Vol. 54, No. 4, pp.
701704.
Yanagitani, T.; Morisato, N.; Takayanagi, S.; Matsukawa, M. & Watanabe, Y. (2011a) c-axis
Zig-Zag ZnO Film Ultrasonic Transducers for Designing Longitudinal and Shear


520
Wave Resonant Frequencies and Modes. IEEE Trans. Ultrason., Ferroelect., Freq.
Contr., Vol. 58, No. 5, pp. 10621068.
Yu, L. S.; Harper, J. M. E.; Cuomo, J. J. & Smith, D. A. (1985). Alignment of Thin Films by
Glancing Angle Ion Bombardment During Deposition. Appl. Phys. Lett., Vol. 47, No.
9, pp. 932933.
Zhang, Y.; Wang Z. & Cheeke, J. D. N. (2003). Resonant Spectrum Method to Characterize
Piezoelectric Films in Composite Resonators. IEEE Trans. Ultrason., Ferroelect., Freq.
Contr., Vol. 50, No. 3, pp. 321333.
23
Polymer Coated Rayleigh SAW and STW
Resonators for Gas Sensor Applications
Ivan D. Avramov
Georgi Nadjakov Institute of Solid State Physics, Sofia
Bulgaria
1. Introduction
Polymer coated gas-phase sensors using the classical Rayleigh-type surface acoustic wave
(RSAW) mode have enjoyed considerable interest worldwide over the last two decades [1-
3]. This interest is motivated by their orders of magnitude higher sensitivity and larger
dynamic range compared to bulk acoustic wave (BAW) sensors, fast response times,
excellent overall stability, coming close to that of quartz crystal sensors, and low phase noise
of the sensor system making high-resolution measurements possible [4]. Because of these
features that are difficult to achieve with other technologies, RSAW based gas sensors have
found successful application in a variety of industrial implementations such as electronic
noses, systems for analysis of chemical and biological gases, medical diagnostics,
environmental monitoring and protection, etc. [5-11]. On the other hand, surface transverse
wave (STW) based gas sensors, even though sharing the same operation principle, have not
been studied so extensively yet. The purpose of this article is to present and discuss
systematic experimental data with both acoustic wave modes which will prove that STW
based gas-phase sensors not only successfully compete with their RSAW counterparts but
also complement them in applications where RSAW gas sensors reach their limits.
Successful corrosion proof RSAW sensors using gold metallization for operation in highly
reactive chemical environments will also be presented.
2. Operation principle of RSAW/STW based resonant gas phase sensors
Both RSAW and STW based gas sensitive resonant sensors share the same operation
principle illustrated in Fig. 1. The sensor device typically is a two-port RSAW or STW
resonator on a temperature compensated rotated Y cut of quartz whose geometry has been
optimized in such manner that the resonator retains a well behaved single-mode resonance
and suffers minimum loss increase and Q-degradation after the gas sensitive layer,
(typically a solid, semisolid or soft polymer film with good sorption properties), is deposited
on its surface. On the other hand, the sensor has to have maximum active area in the centre
of its geometry where the magnitude of the standing wave and deformation are maximized.
Thus, strong interaction with the gas adsorbed in the polymer film occurs and maximum
gas sensitivity is obtained. The sensor operation principle according to Fig. 1 is fairly simple.
If a gas-phase analyte of a certain concentration is applied to its surface, gas molecules are
absorbed by the sensing layer until thermodynamic equilibrium is achieved; i. e. the number


522
of adsorbed molecules becomes equal to the number of desorbed ones. Due to adsorption,
the layer becomes heavier and this increases the mass loading on the sensor surface. As a
result of that, the acoustic wave propagation velocity v decreases and causes a concentration
proportional frequency down shift f of the sensors resonance, called sensor signal. The
resonance frequency shift of RSAW gas sensors coated with a polyisobutilene (PIB) polymer
film is shown in Fig. 2 a) and b) for two different concentrations of tetrachloroethilene
vapors. If the vapor concentration is small (0,1% in Fig. 2 a)) then the resonance shifts down
by 83 ppm without degradation in loss or Q. At large concentrations of the gas vapors (0,7%
in Fig. 2 b)), the 550 ppm of observed frequency down shift is accompanied by a 2 dB loss
increase due to the heavy mass loading. However, the sensor device retains a high loaded
Q, (above 2000 in Fig. 2 a) versus >4000 in Fig. 2 b)) and a steep phase slope in a well
behaved single-mode resonance without distortion or excitation of undesired longitudinal
modes.

Fig. 1. Operation principle of RSAW/STW based resonant gas phase sensors

a) b)
Fig. 2. Frequency (upper curves and phase (lower curves) responses of PIB coated RSAW
sensors prior to (right) and after (left) tetrachloroethilene vapor probing at a) 0,1% and b)
0,7% concentration
3. Measurement resolution of RSAW/STW gas phase sensor systems
If a sensor device as the ones from Fig. 2 a) and b) is used as a frequency stabilizing element
in the feedback loop of an oscillator circuit and its frequency f
0
is adjusted at the resonance
Span 1.000 MHz Center 432.551 MHz Span 1.000 MHz Center 432.551 MHz
2:Transmission Phase 90.0 / Ref 0.00
1M1
2
1
2
Meas1:Mkr2 432.513 MHz
-10.498dB
1
2
-32.576
1:Transmission &MLog Mag 1.0 dB/ Ref -10.43 dB
1:
2:
Span 1.000 MHz Center 433.508 MHz Span 1.000 MHz Center 433.508 MHz
2:Transmission &MPhase 90.0 / Ref 0.00
1M1
2M2
1
2
-11.808dB
1
2
-29.038
1:Transmission &MLog Mag 1.0 dB/ Ref -9.75 dB
1:
2:

Polymer Coated Rayleigh SAW and STW Resonators for Gas Sensor Applications

523
frequency of the sensor (see marker positions in Fig. 2 a) and b)) then due to the high Q of
the sensor device, low-noise oscillation with high short-term stability will be obtained. Any
change in gas concentration will alter the resonance frequency and the output frequency f
0

of the sensor oscillator, accordingly. Thus f can be measured with a high precision using a
high-resolution frequency counter, connected to the output of the sensor oscillator. At a
given gas concentration C, measured in parts per million (ppm), the resolution R of the
sensor system, also measured in ppm, will be limited only by the short-term stability of the
sensor oscillator
y
(), also called Allans variation, for the measurement time . The value of
y
() represents the flicker phase noise of the sensor oscillator in the time domain which is dominated
by the actual flicker phase noise of the coated acoustic wave sensor. The resolution R determines
the minimum change in gas concentration that the system can detect and is, therefore, also
called detection limit. It is calculated as follows:

0
[ ( ) ]/
y
R C f f = (1)
To calculate R for a given gas concentration C, according to (1), it is sufficient to measure
y
() of the sensor oscillator for the time interval which is normally 1s for most frequency
counters operating in the typical 0,3 to 1,0 GHz RSAW/STW sensor range with 1 Hz
resolution. Then, according to [12],
y
() can be calculated from a finite number M of
consecutive frequency measurements y
i
of f
0
as:

1/2
1
2
1
1
1
( ) ( )
2( 1)
M
y i i
i
y y
M

+
=

=

, (2)
where i is an integer. In a well stabilized against thermal transients sensor oscillator
typically 20 to 50 consecutive measurements of f
0
are enough to calculate
y
() with
sufficient accuracy for practical sensor applications.
4. Chemosensitive layers for RSAW/STW based gas sensors
The correct choice of the sensing layer suitable for the chosen acoustic mode is the key to
proper sensor operation and good sensitivity and dynamic range [13, 14]. A sensing layer is
considered as good if it has an excellent adhesion to the surface of the acoustic device for
proper interaction with the acoustic wave, can easily adsorb and restlessly desorb large
amounts of probing gases without chemically reacting with them, has good temperature
stability and low ageing and does not change its sensitivity and sorption characteristics over
thousands of measurement cycles. It is also desirable that the layer provides some selectivity
to a certain chemical compound, i. e. it absorbs larger amounts of that compound than other
compounds. Finally, the layer should not significantly degrade the Q, loss and the shape of
the resonance after deposition onto the acoustic device.
Because of their complicated net structure, many polymers feature excellent physical
sorption, as required for reproducible sensor performance and this makes them appropriate
for gas sensing applications [15-19]. If some of them have also appropriate viscoelastic
properties for good interaction with the RSAW or STW mode, then they will provide the
required performance of the acoustic wave sensor, accordingly. Layers with appropriate
viscoelastic properties are those that follow the deformation of the surface as a result of the
wave propagation without causing significant propagation loss and conversion of the


524
acoustic energy into undesired modes that decay into the bulk of the substrate and may
cause degradation of sensor performance.
An important parameter of the sensing film, except for its viscoelastic properties is its
solidness. On one hand, the parameter solidness determines the sorption properties of the
film and the amount of gas that the layer can accommodate before saturation is reached. On
the other hand, it determines the way in which the polymer film interacts with the acoustic
wave. Therefore, the film solidness will determine the sensitivity, dynamic range and
detection limit of the sensor. Based on their solidness, there are three types of polymer films
that are appropriate for RSAW/STW sensors:
a. Solid polymer films. In fact, these films are solid as glass. That is why, they are often
called glassy polymer films and have a stiffness value close to that of the sensors
quartz substrate that they are deposited on. If used with the STW mode, due to the
lower propagation velocity, these solid films trap the wave energy to the substrate
surface and the acoustic wave propagates with low loss. That is why, solid films work
much better with the STW mode than with the RSAW one. When their thickness becomes
too high, a second slightly faster mode, called Love mode gets excited and multimoding
occurs. Solid polymer films feature surface sorption and become easily saturated by the
adsorbed gas but on the other hand, they feature very fast response times and are very
sensitive if the sensor is operated far below saturation. That is why they are appropriate
for high resolution measurements at low gas concentrations, (typically below 0,1%). A
typical representative of the solid polymer family is the hexamethyldissiloxane (HMDSO),
obtained in a glow-discharge plasma polymerization process [19].
b. Soft polymer films. These films are soft and elastic just like rubber. That is why they are
referred to as rubbery or jelly-like films. Typically, they are deposited using spin
coating or more advanced techniques such as airbrush or electro spray methods that
provide good control over film thickness and uniformity. Since these soft polymers
provide profound bulk sorption, they are capable of adsorbing large amounts of gas
and are appropriate for measurements at high gas concentrations, (typically above
0,1%). They are well tolerated by the RSAW mode but do not work so well with STW.
The reason is that they cause energy leakage of the STW into the bulk of the soft layer
which results in increased loss and Q-degradation of the sensor resonator. Polymers
like polyisobutilene (PIB), poly-(2-hydroxyethylmethacrylate) (PHEMA) and poly-(n-
butylmethacrylate) (PBMA) are often used in RSAW based gas sensors.
c. Semisolid polymer films. These light and highly elastic films are also typically obtained in a
plasma polymerization process [17, 18] for good reproducibility of the film parameters
and have a structure very similar to polystyrene, the material used in plastic bags. They
are highly resistant to almost all aggressive chemicals such as acids, bases and organic
solvents and this makes them appropriate for environmental sensing applications. They
are well tolerated by both, the RSAW and STW mode and often feature sensitivities
comparable to those of the soft polymer films. The two semisolid films used in this study
are styrene (ST) and allylalkohol (AA) synthesized in a plasma polymerization reactor.
5. Comparative characteristics of polymer coated RSAW and STW gas
sensors operating at the same acoustic wave length
To identify the advantages and disadvantages of the STW mode versus its RSAW
counterpart on quartz for gas sensor applications it is necessary to compare the sensor


525
performance of both modes under identical real-life conditions. Such a performance
comparison would be correct only if it is carried out with sensor devices of both modes
operating on the same acoustic wave length for the following reason: If both types of
devices are fabricated on the same piezoelectric material and cut orientation (AT-cut quartz
in this case), use the same device geometry, are coated with the same sensing layer of the
same thickness and are probed with identical gases and concentrations, then the only factors
responsible for the differences in electrical and sensor performance would be the type of
motion for each mode, (elliptical for the RSAW and shear horizontal for the STW) and the
way the acoustic wave interacts with the sensing layer. The results presented in the next
sections were performed with RSAW and STW sensors whose electrical characteristics in the
uncoated state are summarized in Table 1.
5.1 Electrical performance of STW/RSAW sensor resonators coated with solid and
semisolid sensing layers
The frequency and phase responses of the STW and RSAW sensor resonators from Table 1
prior to and after coating with the solid HMDSO are compared in Fig. 3. After film
deposition, the frequency of the RSAW device shifts down by about 1,5 MHz (3500 ppm), its
insertion loss increases by 5,7 dB and the loaded Q decreases from 6000 to about 2000.

Acoustic wave mode STW RSAW
Acoustic wave length 7,22 m 7,22 m
Sensor resonator frequency 433 MHz 701 MHz
Device insertion loss 5-7 dB 6-7 dB
Loaded Q-factor 3000-4000 5000-6000
Side lobe suppression > 8 dB >12 dB
Metallisation Al Al
Table 1. Electrical characteristics of the uncoated STW/RSAW sensor resonators used in the
comparative studies
In addition, the RSAW device retains a well behaved single-mode resonance with excellent
side lobe suppression as required for stable operation of the sensor oscillator. The STW
device shows a different behavior. Its frequency shifts down by 4 MHz (6100 ppm) which
accounts for about 2 times higher relative mass loading sensitivity than its RSAW
counterpart. The insertion loss increases by just about 3 dB versus 5,7 dB for the RSAW
mode which implies that the STW mode tolerates solid films better in terms of loss increase.
On the other hand, excitation of a second higher-order Love wave mode [20] about 7 MHz
higher than the main STW mode at 697 MHz is observed. Since a 180 deg. phase reversal at
this Love mode occurs, (see the lower data plot in Fig. 3 b)), it is not very likely to degrade
the performance of the sensor oscillator. A more serious problem, however, is the distortion
at the main STW mode that indeed can cause the sensor oscillator to jump onto an adjacent
peak during the measurement. That is why, coating STW sensor resonators with excessively
thick solid films as the 190 nm HMDSO from Fig. 3 should be stopped before distortion and
multiple peak behavior on the main STW mode occurs. As far as the higher-order Love
mode at 704 MHz is concerned, we have noticed that its gas sensitivity is orders of
magnitude lower than the STW mode on the right side. This lack of sensitivity is explained
by the fact that the Love mode scatters its energy into the bulk of the sensing layer [20].


526
p
MKR( 250): 433.86MHz
MAGTD ( ) -7.49dB 5dB/ -31.75dB
PHASE ( ) 159.4deg 100deg/ 320.0deg
CF: 433.86MHz SPAN: 10MHz

MKR( 250): 701.33MHz
MAGTD ( ) -5.40dB 5dB/ -30.90dB

p
MKR( 250): 431.34MHz
MAGTD ( ) -13.18dB 5dB/ -31.75dB

MKR( 219): 697.08MHz
MAGTD ( ) -8.38dB 5dB/ -32.80dB

a) b)
Fig. 3. Frequency (upper curves) and phase responses (lower curves) of the a) RSAW and b)
STW sensor resonators from Table 1 prior to (upper plots) and after (lower plots) 190 nm
HMDSO solid film deposition
5.2 Electrical performance of STW/RSAW sensor resonators coated with soft polymer
films
A similar comparison between both acoustic wave modes was performed by coating the
devices from Table 1 with the soft polymer film PIB using the micro drop deposition
method. The data obtained shows quite the opposite tendency compared to the solid film
behavior from Section 5.1. The STW devices suffered a 5 dB increase in insertion loss and
rather distorted frequency responses even at fairly thin soft layers. Only a moderate
frequency down shift of 1330 ppm was obtain as a result of film coating. As evident from the
frequency responses in Fig. 2 the RSAW devices were found to provide a much better
performance at the same film thickness. They retain a high loaded Q and low insertion loss,
as well as an undistorted single-mode resonance. These data imply that RSAW sensors will
work better with soft polymer films while the STW mode will provide better performance
with solid films as long as they are not excessively thick to cause distortion.
6. A practical method for film thickness optimization of RSAW/STW gas
sensors coated with solid and semisolid sensing layers
The most important step in designing practical RSAW/STW resonant sensors is the
selection of an optimum thickness of the sensing layer. It should be selected in such manner


527
that maximum sensor sensitivity and dynamic range are obtained at minimum degradation
of the electrical resonator performance (insertion loss, loaded Q, side lobe suppression and
distortion) as required for stable low-noise operation of the sensor oscillator. A very efficient
method for film thickness optimization using a controlled plasma deposition of the
semisolid polymer Parylene C is described in [21]. This material has viscoelastic properties
very similar to practical solid and semisolid layers but has the unique feature that it can
polymerize directly on the surface of the acoustic devices at room temperature, thus
avoiding undesired thermal frequency drifts. Also the actual deposition is performed in a
chamber separate from the plasma reactor where the devices are protected from the high
electric fields of the main plasma generator. This allows direct measurement of their
frequency and phase responses with a network analyzer in the process of film deposition
while the film thickness is measured with a quartz crystal microbalance (QCM). The results
from a Parylene C deposition on a 433 MHz RSAW resonator and real time measurements of
its electrical characteristics are shown in Fig. 4 for a polymer thickness ranging from 0 to 700
nm. From this measurement it is possible to identify the thickness range in which optimum
sensor performance is expected and to extract information on the behavior of important
sensor parameters in the process of polymer coating as follows:
the down shift of the resonant frequency versus film thickness;
the loss increase with film thickness;
the loaded Q decrease with film thickness;
the behavior of the adjacent longitudinal modes in the process of deposition;
the thickness range over which the sensor demonstrates maximum mass sensitivity of
frequency while retaining good electrical performance. In Fig. 4 this optimum thickness
range is between 100 and 300 nm with an average of 200 nm.

Fig. 4. Parylene C coating behavior of a 433 MHz RSAW sensor resonator in the 0 to 700 nm
polymer thickness range
As evident from Fig. 4, in the optimum thickness range (100 to 300 nm) a maximum linear
frequency shift, (maximum mass sensitivity), is accompanied by just about 7 dB loss
increase. The Q remains high as shown by the sharp resonance while the first longitudinal
mode on the left side of the resonance remains suppressed by at least 12 dB. If the sensor is


528
intended for high-resolution measurements at low gas concentrations, then a thickness close
to 100 nm should be chosen due to the highest Q and lowest loss. If measurements at higher
gas concentrations are expected then a 300 nm thickness may be more appropriate since the
thicker layer may adsorb larger amounts of gas without film saturation.
6.1 Critical thickness in RSAW/STW based sensor resonators coated with solid and
semisolid sensing layers
As shown in the previous sections, the sensing layer does not only shift the resonant
frequency down, increase the loss and decrease the loaded Q as a result of mass loading but
it also influences the longitudinal modes supported by the resonator geometry that appear
on the left side of the main resonance. In the uncoated resonator these modes are well
enough suppressed (typically by 5 to 15 dB) and do not cause any problems when the
resonator is operated in an oscillator circuit. As soon as a sensing layer is deposited on the
surface, it will change the phase conditions along the device topology and this will cause the
adjacent longitudinal modes to arise in magnitude at the expense of the main resonance.
This situation gets worse at thick solid films for both, the STW and the RSAW mode. At a
certain thickness which we call critical thickness the magnitude of the first adjacent low-
frequency longitudinal mode on the left becomes equal to the magnitude of the higher-
frequency main resonance. This creates a potential for instability in the sensor oscillator
stabilized with this sensor since it can easily jump from the main resonance onto the left
longitudinal mode during gas probing which will ruin the measurement. The critical
thickness situation is illustrated in Fig. 5 a) and b) for a STW and a RSAW device from Table
1, accordingly, in the process of Parylene C deposition as described in Section 6. As evident
from Fig. 5 a), at thickness values above 185 nm the first longitudinal mode on the left starts
rapidly growing until its magnitude becomes equal to the main resonance. The critical
thickness at which this happens is about 350 nm. At this thickness also a strong Love mode
excitation on the right is observed. The RSAW device in Fig. 5 b) reaches its critical thickness
at about 650 nm. From these data we can draw the conclusion that the devices from Table 1
can be usable as Parylene C coated sensors as long the film thickness is lower than 300 nm
and 600 nm for the STW and RSAW devices, respectively. Comparing the coating behavior
of both modes in Fig. 5 we see that the STW mode retains a much better behaved resonance
than its RSAW counterpart until the critical thickness is reached. At that thickness the STW
device has a loss of 14 dB (Fig. 5 a)), versus 35 dB for the RSAW device (Fig. 5 b)). Therefore,
the STW mode tolerates solid and semisolid sensing films much better than the RSAW one
and is more appropriate for operation with such films in practical gas sensors.
7. Gas sensing characteristics of RSAW/STW resonant sensors coated with
solid and semisolid chemo sensitive films
In the following sections we present results from gas probing experiments on RSAW/STW
sensors coated with solid HMDSO and semisolid ST and AA films. Four different chemical
agents at different vapor concentrations are used for gas probing as follows:
Dichloroethane 6500 ppm
Ethylacetate 17600 ppm
Tetrachloroethylene 2650 ppm
Xylene 1400 ppm


529
The purpose of the gas probing tests is to identify which acoustic wave mode provides
better performance in real-world gas sensing conditions.

Critically coated
SAW devices

a) b)
Fig. 5. Critical thickness in a) STW and b) RSAW devices in the process of Parylene C
coating
7.1 Computer controlled automatic system for gas probing measurements
The block diagram of the computer controlled system for measuring the gas sensing
characteristics of the RSAW/STW polymer coated sensors is shown in Fig. 6. For correct
comparison of the gas probing performance of both acoustic wave modes four pairs of
devices (one RSAW and one STW sensor in each pair, coated with the same polymer to the
same film thickness and in the same deposition process) are mounted in open TO 92
packages and placed in the sensor head which can accommodate a total of eight sensors.
Each device is connected to one of the 8 sensor oscillator circuits in the head. During gas
probing each of the 8 oscillators is turned on for a short period of time to take the
measurement. The oscillators are operated one at a time and multiplexed consecutively to
avoid possible injection locking. Their output frequency is down converted to an
intermediate frequency in the 4-9 MHz range by means of a stable heterodyne reference
oscillator to allow fast high-resolution measurements with a reciprocal frequency counter.
The chemical compounds 1 through 4 used for gas probing are vapors from the 4 liquid-
phase analytes in the 4 containers. A permeation cell is placed on top of each container to
allow a defined vapor pressure which is controlled by the rotation speed of the pump. By a
switch block of valves the vapors of each analyte are then consecutively fed to the sensor
head where they interact with the sensors. After the measurements at each analyte are
completed the sensors are flushed with dry air passing through a silica gel integrator which
provides also a homogenous air and gas flow. The entire system is controlled by a computer
which performs measurements in probe-flush cycles over time and provides real-time
sensor data on the computer screen. The data in Fig. 7 is the gas probing performance of a
700 MHz STW styrene coated sensor probed with dichloroethane vapors at 6500 ppm
concentration in 100 s probe-flush cycles. It should be noted that prior to this measurement
the sensor was probed with a different compound (xylene) for 62 hours and 40 minutes.
Note the excellent reproducibility of the noise free sensor signal with a magnitude


530
f=160KHz over time indicating that prolonged xylene treatment has not had any influence
on the sensor performance. This is a clear indication that styrene has very good physical
sorption properties to a variety of gas phase compounds.

Fig. 6. Block diagram of the automated system for simultaneous gas sensitivity
measurements on eight sensor devices

0 100 200 300 400 500 600 700 800 900
0
20k
40k
60k
80k
100k
120k
140k
160k
STW Sensor J7 after Xylene treatment for 62 h. 40 min.
Di-chloro ethane probing at 6500 ppm concentration
F
r
e
q
u
e
n
c
y

s
h
i
f
t

w
i
t
h

p
r
o
b
i
n
g

(
H
z
)
Time (s)

Fig. 7. Gas sensing performance of a 700 MHz styrene coated STW sensor probed with
dichloroethane at 6500 ppm concentration in 100 s probe-flush cycles


531
7.2 Gas sensitivity comparison of RSAW/STW sensor resonators coated with solid
HMDSO films
This study aims at finding out which of both acoustic wave modes provides better gas
sensitivity when coated with solid HMDSO films and what is the optimum film thickness at
which maximum sensitivity is achieved. For this purpose, 5 pairs of RSAW/STW devices
according to Table 1 were coated at 5 different HMDSO thicknesses (50, 100, 190, 280 and
350 nm) each, in the same plasma deposition process for each pair. Figure 8 compares the
gas sensing characteristics of both modes gas probed with tetrachloroethilene at 2650 ppm
concentration. The results from all gas probing experiments on the 5 pairs of devices are
summarized in Table 2. In this table the sensitivity factor is the ratio between the relative
sensitivities (in ppm) for the two devices of each pair. It is given for each of the 5 film
thicknesses and shows which mode is more sensitive and at which thickness. From the data
in Fig. 8 and Table 2 the following important practical conclusions can be drawn:
1. The HMDSO coated sensors have very short response times and reach adsorption-desorbtion
equilibrium just a few seconds after the gas flow is applied. We attribute this behavior to the
surface sorption of the HMDSO which is typical for solid sensing polymers.

Sensor/Compound Dichloroethane
6500 ppm
Ethylacetate
17600 ppm
Tetrachloroethylene
2650 ppm
Xylene
1400 ppm
700 MHz STW
50 nm HMDSO
2 KHz
(2.9 ppm)
2.8 KHz
(4 ppm)
3.5 KHz
(5 ppm)
2.4 KHz
(3.4 ppm)
433 MHz RSAW
50 nm HMDSO
1.5 KHz
(3.5 ppm)
4 KHz
(9.2 ppm)
3 KHz
(6.9 ppm)
2.2 KHz
(5 ppm)
Sensitivity factor
(STW/ RSAW)
0.82 0.43 0.72 0.68
700 MHz STW
190 nm HMDSO
3 KHz
(4.3 ppm)
4.8 KHz
(6.9 ppm)
7.5 KHz
(10.7 ppm)
3.7 KHz
(5.3 ppm)
433 MHz SAW
190 nm HMDSO
1.8 KHz
(4.2 ppm)
3.8 KHz
(8.8 ppm)
3.5 KHz
(8.1 ppm)
2.5 KHz
(5.8 ppm)
Sensitivity factor
(STW/RSAW)
1.02 0.78 1.32 0.91
700 MHz STW
280 nm HMDSO
8.3 KHz
(11.9 ppm)
8.5 KHz
(12.1 ppm)
7 KHz
(10 ppm)
4 KHz
(5.7 ppm)
433 MHz SAW
280 nm HMDSO
3.5 KHz
(8 ppm)
6.5 KHz
(15 ppm)
6 KHz
(14 ppm)
4.3 KHz
(10 ppm)
Sensitivity factor
(/RSAW)
1.49 0.81 0.71 0.57
700 MHz STW
350 nm HMDSO
11 KHz
(15.7 ppm)
14 KHz
(20 ppm)
15 KHz
(21.4 ppm)
9.5 KHz
(13.6 ppm)
433 MHz SAW
350 nm HMDSO
1.8 KHz
(4.2 ppm)
2.3 KHz
(5.3 ppm)
5.1 KHz
(11.8 ppm)
4.2 KHz
(9.7 ppm)
Sensitivity factor
(STW/RSAW)
3.74 3.77 1.81 1.4
700 MHz STW
100 nm HMDSO
11 KHz
(16 ppm)
20 KHz
(29 ppm)
37 KHz
(53 ppm)
9 KHz
(13 ppm)
Table 2. Gas sensitivity comparison of RSAW vs. STW devices coated with solid HMDSO
film of 5 film thicknesses. Data on the 100 nm coated SAW device are not available


532
Compound
Concentration
Dichloroethane
6500 ppm
Ethylacetate
17600 ppm
Tetrachloroethylene
2650 ppm
Xylene
1400 ppm
700 MHz STW,
100 nm HMDSO
11 KHz
(16 ppm)
20 KHz
(29 ppm)
37 KHz
(53 ppm)
9 KHz
(13 ppm)
433 MHz RSAW,
280 nm HMDSO
3.5 KHz
(8 ppm)
6.5 KHz
(15 ppm)
6 KHz
(14 ppm)
4.3 KHz
(10 ppm)
Sensitivity
factor
STW/RSAW
2.0
STW/RSAW
1.93
STW/RSAW
3.79
STW/RSAW
1.3
Table 3. Gas sensitivity comparison of the RSAW and STW sensors coated at their optimum
HMDSO thickness values

0 100 200 300 400 500 600 700 800
0
5k
10k
15k
20k
25k
30k
35k
40k
HMDSO coated 700 MHz STW resonators
Tetra-chloro ethylene probing at 2650 ppm concentration
A: 50 nm
B: 190 nm
C: 280 nm
D: 350 nm
E: 100 nm
F
r
e
q
u
e
n
c
y

s
h
i
f
t

w
i
t
h

p
r
o
b
i
n
g

(
H
z
)
Time (s)
a)
0 50 100 150 200 250 300 350 400 450 500 550 600 650
0
1k
2k
3k
4k
5k
6k
7k
HMDSO coated 433 MHz SAW resonators
Tetra-chloro ethylene probing at 2650 ppm concentration
A: 190 nm
B: 50 nm
C: 280 nm
D: 350 nm
E: uncoated
F
r
e
q
u
e
n
c
y

s
h
i
f
t

w
i
t
h

p
r
o
b
i
n
g

(
H
z
)
Time (s)
b)
Fig. 8. Tetrachloroethilene probing data of HMDSO coated a) STW and b) RSAW sensors at
50, 100, 190, 280 and 350 nm film thicknesses
2. Starting from very thin films (50 nm in this case) and increasing the thickness, the gas
sensitivity increases in both the RSAW and STW devices, accordingly, until a thickness value is
reached at which maximum sensitivity is achieved. This optimum thickness value is different for
both modes (100 nm for the STW and 280 nm for the RSAW mode at the wavelength of 7,22 m
in this case). Further increase in film thickness beyond the optimum thickness value only reduces
the relative frequency sensitivity and increases the loss of the gas sensor.
3. The optimum thickness values for both modes are far below critical thickness and are well within
the thickness ranges in which the sensor devices demonstrate high mass sensitivity while
retaining low insertion loss, high Q and a well behaved single-mode resonance in the Parylene C
coating experiment from Fig. 5. Therefore, the practical film thickness optimization method
described in Section 6 is well suited for identifying the optimum film thickness at which
maximum gas sensitivity should be expected.
4. The relative sensitivities for both acoustic wave modes at their optimum thickness values,
summarized in Table 3, demonstrate a 1,3 to 3,8 times higher sensitivity to all 4 gases of the
STW mode versus its RSAW counterpart operating at the same acoustic wavelength. This
suggests that the STW mode is much more appropriate for operation with solid chemo
sensitive polymers.


533
7.3 Gas sensitivity comparison of RSAW/STW sensor resonators coated with
semisolid styrene (ST) and allylalcohol (AA) films
A similar comparative study was performed also when pairs of RSAW/STW devices from
Table 1 were coated at three different thicknesses of the semisolid ST and AA polymer films
in a glow discharge plasma reactor. Since no equipment was available to measure the
thickness of semisolid layers directly we used the deposition time in seconds for each layer
as a measure of the layer thickness. In this study the deposition times for all three
thicknesses were 10, 15 and 20 s for the films with the lowest, medium and highest
thicknesses, accordingly. The results from these gas probing tests are summarized in Tables
4 and 5 for the ST and AA coated sensors, accordingly. Note that in these experiments we
had to reduce the concentrations of all four probing gases by a factor of 4 to avoid saturation
of the sensing layers due to the much higher adsorption capacity of the ST and AA films
compared to HDMSO. When comparing the data from Tables 4 and 5 an interesting
behavior is observed. Styrene coated STW devices are up to 3 times more sensitive than their
ST coated RSAW counterparts while with the AA coated sensors we see the opposite
behavior - the AA coated RSAW devices are up to 3,6 times more sensitive than their AA
coated STW counterparts. We attribute this behavior to the fact that AA is the softest of the
three polymers we used in this work and this material behaves much more like a soft
polymer than a semisolid one. This implies that the RSAW mode might be more suitable for
operation with soft sensing films than the STW mode.

Compound/
Concentration
Tetrachloroethylene
(630 ppm)
Dichloroethane
(1550 ppm)
Ethylacetate
(4190 ppm)
Xylene
(330 ppm)
Acoustic mode STW RSAW STW RSAW STW RSAW STW RSAW
Deposition time
10 s/styrene
119
ppm
74
ppm
100
ppm
92
ppm
94
ppm
76
ppm
71
ppm
44
ppm
Deposition time
15 s/styrene
254
ppm
85
ppm
206
ppm
171
ppm
190
ppm
95
ppm
140
ppm
65
ppm
Deposition time
20 s/styrene
239
ppm
108
ppm
219
ppm
201
ppm
206
ppm
122
ppm
111
ppm
76
ppm
Sensitivity factor
10s / 15s / 20s
STW/RSAW
1.61 / 3.0 / 2.21
STW/RSAW
1.09 / 1.2 / 1.09
STW/RSAW
1.24 / 2.0 / 1.69
STW/RSAW
1.61 / 2.15 / 1.46
Table 4. Gas sensitivity comparison of semisolid ST coated RSAW and STW devices

Compound/
Concentration
Tetrachloroethylene
(630 ppm)
Dichloroethane
(1550 ppm)
Ethylacetate
(4190 ppm)
Xylene
(330 ppm)
Acoustic mode STW RSAW STW RSAW STW RSAW STW RSAW
Deposition time
20 s/AA
12.3
ppm
20.8
ppm
22.8
ppm
66.4
ppm
32.6
ppm
69.9
ppm
7.1
ppm
25.2
ppm
Deposition time
25 s/AA
14.3
ppm
25.9
ppm
23.1
ppm
83.1
ppm
31.5
ppm
90.5
ppm
8.3
ppm
28.4
ppm
Sensitivity factor
20s / 25s
RSAW/STW
1.69 / 1.81
RSAW/STW
2.91 / 3.18
RSAW/STW
2.14 / 2.87
RSAW/STW
3.55 / 3.42
Table 5. Gas sensitivity comparison of semisolid AA coated RSAW and STW devices


534
For the semisolid film coated RSAW/STW sensors we can make the following conclusions:
1. Semisolid sensing films improve gas sensitivity of the RSAW and STW modes dramatically
compared to the solid films. ST coated devices demonstrate one to two orders of magnitude
higher relative gas sensitivities compared to HMDSO coated ones.
2. As observed with the solid film, also semisolid layers seem to have an optimum film thickness at
which maximum gas sensitivity is achieved. Here these optimum thicknesses are achieved at 15 s
for ST and 20 s for AA with the STW mode, while the RSAW mode needs somewhat higher
optimum thicknesses 20 s for ST and 25 s for the AA film.
3. There is a significant film type dependent difference in gas sensitivities for both modes. ST
provides much better gas sensitivity compared to AA and this is attributed to both the different
sorption properties and different viscoelastic properties of the films which determine how the
wave interacts with the film and the gas sorbed in it.
4. The RSAW mode operates better with the relatively soft AA than with the semisolid ST.
8. Noise and measurement resolution (detection limit) of RSAW/STW
resonant sensors operating with gas sensing polymer layers
As shown in Section 3, if the sensor resonator is connected in the feedback loop of a sensor
oscillator whose short-term stability over the time has been measured as
y
(), then
measuring the sensor signal f which is the response of the sensor oscillator to the gas with
the concentration C , the measurement resolution R, also called detection limit of the sensor
system, can readily be calculated with Equation (1). As an example, let us determine the
measurement resolution of the sensor system with which the dichloroethane measurement
at C=6500 ppm (6,5x10-
3
) from Fig. 7 was performed. With
y
() measured as 1,17x10
-9
/s at
the oscillator frequency f
0
=700 MHz (7x10
8
Hz), for the sensor resolution over the
measurement time =1 s we obtain R=33,3 ppb (parts per billion). This means that this
sensor system can detect changes in the dichloroethane concentration as small as 33 ppb.
Such high sensor resolutions are extremely difficult to achieve with other sensor
technologies. They are attributed to the fact that RSAW/STW resonant sensors retain
excellent resonance characteristics, low loss, high Q and low flicker phase noise when coated
with solid and semisolid chemo sensitive polymer films at optimum thickness.
The data in Tables 6 and 7 represent the detection limits of the styrene coated RSAW and
STW sensors, respectively, during the measurements on the 4 gas-phase analytes used in
this work. The
y
() values for the sensor oscillators were measured as 5,6x10
-9
/s and
3,04x10
-9
/s, respectively. The Resolution factor in Table 7 indicates the resolution
improvement of the STW mode versus its RSAW counterpart in these measurements. In

Compound/
Concentration
Tetrachloroethylene
(630 ppm)
Dichloroethane
(1550 ppm)
Ethylacetate
(4190 ppm)
Xylene
(330 ppm)
Acoustic mode/fo RSAW/433 MHz
Dep. time/polymer 20 s/styrene
Sensor signal f 47 KHz
(108 ppm)
87 KHz
(201 ppm)
53 KHz
(122 ppm)
33 KHz
(76 ppm)
y
(1s) 5,610
-9
/s
Resolution R 32,5 ppb 43,2 ppb 192 ppb 24 ppb
Table 6. Detection limits of the styrene coated RSAW sensors at the 4 gas-phase analytes


535
Compound/
Concentration
Tetra-chloro
ethylene (630
ppm)
Di-chloro
ethane
(1550 ppm)
Ethyl acetate
(4190 ppm)
Xylene
(330 ppm)
Acoustic mode/fo STW/700 MHz
Dep. time/polymer 15 s/styrene
Sensor signal f 178 KHz
(254 ppm)
144 KHz
(206 ppm)
133 KHz
(190 ppm)
98 KHz
(140 ppm)
y
(1s) 3,0410
-9
/s
Resolution R 7,5 ppb 22,9 ppb 67 ppb 7,2 ppb
Resolution factor
(STW/RSAW)
-1

4,3

1,9

2,9

3,3
Table 7. Detection limits of the styrene coated STW sensors at the 4 gas-phase analytes
Table 7 this resolution improvement is by a factor of 1,9 to 4,3. We attribute this advantage
of the STW mode to the following factors:
The styrene coated STW sensors feature lower flicker noise values than their RSAW
counterparts. This is evident from the
y
() measurement (5,6x10
-9
/s vs. 3,04x10
-9
/s for the
RSAW and STW sensors, respectively). This suggests that the STW devices tolerate the styrene
film better than the RSAW ones;
The STW mode features better relative gas sensitivity than its RSAW counterpart at the same
acoustic wave length and type of semisolid polymer film (styrene) as evident from Table 4.
8.1 External factors that may degrade measurement resolution in practical
RSAW/STW sensor systems
The data in Tables 6 and 7 represent the physical detection limits that can be achieved with
practical RSAW/STW based sensor systems. If all other factors that may have a negative
effect on the measurements are excluded then the only limiting quantity to the measurement
resolution remains the electrical flicker phase noise of the sensor oscillator represented by its
Allans variation
y
(). Unfortunately, in practical sensor systems there are several external
factors that may seriously degrade sensor resolution and therefore care should be taken to
eliminate them or to reduce their influence to acceptable values. The closer the system is
brought to its
y
() limit, the better it has been designed.
The major factors that may degrade sensor resolution in practical sensor systems will be
briefly discussed next.
a. Gas flow homogeneity.
This is one of the key disturbances that may seriously degrade sensor noise and should
be eliminated first. If the gas flow applied to the sensor head from Fig. 6 is not
homogeneous then the sensor devices will sense a variable gas concentration during the
probe-flush cycles. Since the measurement cycle is much longer than the time over
which inhomogeneities occur, amplitude fluctuation of the sensor signal at or close to
equilibrium will occur. This situation is illustrated in Fig. 9 showing results from a
tetrachloroethilene measurement with the setup from Fig. 6 when the integrators are
removed. This causes a serious turbulence of the gas flow in the system which results in
strong noise levels on top of the sensor signals. Even a periodicity on the noise signal is
observed which is attributed to the pump rotation. Fortunately, gas flow


536
inhomogeneities are easily eliminated. After the integrators are placed back at the air
inlet and outlet a noise free measurement similar to the one from Fig. 7 is obtained.

0 100 200 300 400 500 600 700
0
5k
10k
15k
20k
25k
30k
35k
40k Tetra-chloro etylene probing on HMDSO coated
433 MHz SAW devices with both integrators removed
190 nm
50 nm
350 nm
280 nm
uncoated
F
r
e
q
u
e
n
z

s
h
i
f
t

(
H
z
)
Time (s)

Fig. 9. Tetrachloroethilene probing with 5 HMDSO coated RSAW sensors with the
integrators from the setup in Fig. 6 removed
b. Gas saturation of the sensing films.

0 5 10 15 20 25 30
0
50k
100k
150k
200k
250k
300k
350k
Ethyl acetate measurement at 17600 ppm concentration (701 MHz STW)
ST/10s
ST/15s
ST/20s
ST/25s
AA/10s
AA/15s
AA/20s
AA/25s
F
r
e
q
u
e
n
c
y

s
h
i
f
t

w
i
t
h

p
r
o
b
i
n
g

(
H
z
)
Time (min)
a)
0 5 10 15 20 25 3
0
10k
20k
30k
40k
50k
60k
70k
80k
90k
100k
110k
120k
130k
140k
150k
Ethyl acetate measurement at 4190 ppm concentration (701 MHz STW)
ST/10s
ST/15s
ST/20s
ST/25s
AA/10s
AA/15s
AA/20s
AA/25s
F
r
e
q
u
e
n
c
y

s
h
i
f
t

w
i
t
h

p
r
o
b
i
n
g

(
H
z
)
Time (min)
b)
Fig. 10. Ethylacetate probing on ST and AA coated STW sensors at a) 17600 and b) 4190 ppm
vapour concentration
Saturation of the sensing films occurs when gas concentrations become so high that
sorption limit of the layer is reached. A situation like this during an ethylacetate
measurement at 17600 ppm concentration is illustrated in Fig. 10 a). In this case strong
peaks of overpressure in the analyte container as well as noise and distortion on the
sensor signals are observed as a result of film saturation. When the gas concentration is
reduced by a factor of 4 to 4190 ppm, the sensor signals become much more uniform
and noise fluctuations are greatly reduced. The behaviour in Fig. 10 a) has the following
explanation. When the films become saturated dynamic equilibrium is disturbed and a


537
very turbulent adsorption-desorbtion process takes place, the films get lighter and
heavier in a stochastic way and this generates noise on top of the sensor signals. Once
the gas concentration is reduced, equilibrium occurs and the adsorption-desorbtion
process returns back to normal (see Fig. 10 b)).
c. Adsorption-desorbtion noise (ADN) in soft film coated RSAW sensors operated far below gas
saturation.
When RSAW sensors are coated with soft polymer films featuring profound bulk
sorption these films can accommodate large amounts of gas without being driven into
saturation. The larger the amount of adsorbed gas, the more turbulent the adsorption-
desorbtion process at equilibrium becomes even though the film is operated far below
its saturation limit. This results in ADN evident in Fig. 11 which presents results from
tetrachloroethilene probing on RSAW/STW sensors coated with the soft PIB film. ADN is
visible on top of all sensor signals regardless of how strong they are. It should be noted
that ADN levels depend entirely on the sorption characteristics of the soft polymer films
and this makes elimination of this type of noise very difficult. In practical sensor systems
one should either cope with ADN or, in critical situations, a different type of polymer film
with lower ADN should be used. For example, the magnitude of the sensor signals in Fig.
11 is comparable with those in Fig. 7 where ST was used and no measurable ADN levels
were observed. Therefore, results as the ones from the PIB film coated sensors from Fig. 11
but free of ADN could readily be obtained with ST coated sensors.

0 5 10 15 20 25 30 35 40
0
10k
20k
30k
40k
50k
60k
70k
80k
90k
100k
110k
120k
130k
140k
150k
160k
170k
Tetra-chloro ethylene probing on PIB coated SAW and STW devices at 433 MHz
STW 4x
STW 5x
STW 6x
STW 7x
SAW 7x
SAW 6x
SAW 5x
SAW 4x
F
r
e
q
u
e
n
c
y

s
h
i
f
t

w
i
t
h

p
r
o
b
i
n
g

(
H
z
)
Time (min)

Fig. 11. Tetrachloroethilene probing on RSAW/STW sensors coated with the soft PIB
polymer film and operated far below gas saturation
9. Corrosion proof RSAW resonant sensors using gold electrode structure
Typically, RSAW/STW sensors are fabricated with Al electrode structure using a well
established photolithographic process. Al metallization is cheap and provides excellent
electrical performance in almost all SAW devices fabricated to date. However, a major
problem occurs if SAW devices with Al metallization are coated with sensing layers and
used as gas sensors. Very often, the chemical gas-phase compounds to be detected form
corrosive bases and alkalis with the ambient humidity and attack the Al electrode structure


538
by entering into a chemical reaction with the Al film. The problem is further aggravated by
the presence of the sensing layer which, by absorbing large amounts of gas, greatly increases
the concentration of the aggressive analyte that comes in contact with the Al electrodes. As a
result of corrosion, the sensors suffer irreversible performance degradation, provide
inconsistent data and even dye within a limited number of measurement cycles. The
solution to that problem is the implementation of SAW devices with corrosion proof gold
(Au) metallization that can successfully stand severe corrosion attacks by chemically
reactive substances. The design of RSAW sensor resonators with Au electrode structure is
not so straight forward as with Al metallization. Due to the fact that Au has a 7 times higher
density than Al and is much softer several side effects, such as excitation of strong SSBW
modes and transverse waveguide modes occur that can cause serious loss and Q-degradation
as well as distorted characteristics at the main resonance. However, by careful selection of the
Au thickness and choosing proper device geometry, these side effects can be kept under
control and very good resonance characteristics appropriate for gas sensor applications can be
achieved [22], [23]. In the next sections we will discuss the performance of RSAW sensors with
Au electrode structure intended for operation as gas sensors in highly reactive chemical
environments. These sensors were designed to replace their predecessors using the
problematic Al electrode structure in a practical sensor system operating at 433 MHz.
9.1 Electrical performance comparison of Au vs. Al RSAW sensor resonators
Generally two types of resonator devices are used in a practical resonator system two-port
resonators (TPR) featuring a single-mode resonance and coupled resonator filters (CRF) that
have a two-pole resonance achieved with a coupling grating in the centre of the resonant
cavity. As evident from Fig. 12 a) and b), the CRF devices have twice the phase slope of a
TPR in their filter pass bands and generally provide better stabilization of the sensor
oscillator than the TPR, especially in measurements at high gas concentrations. In a real-
world sensor system, the sensor oscillator is designed to operate on the right CRF resonant
mode since the left one vanishes when the device is coated with a polymer film [6].
Typically, the sensor oscillator provides stable oscillation on the right mode and never
jumps onto the left one since a 180 deg. phase reversal makes oscillation impossible at that
mode (see Fig. 12 b)). The frequency and group delay responses of the Al RSAW devices
previously used in the sensor system are shown in Fig. 12 c) and d) while a) and b) represent
the electrical performance of their Au substitutes designed in [23]. The insertion loss, Q and
group delay data from these devices at resonance are compared in Table 8. Es evident from
that data and also from Fig. 12 the Al and Au devices feature very similar frequency
responses, insertion loss and loaded Q values and the replacement of the Al devices with
their Au successors was made without any changes or adjustments of the sensor circuitry.
The Au devices are slightly inferior to the Al ones in terms of loss and loaded Q. This is
attributed to the loss of energy as a result of the heavy Au loading on the quartz surface.

Parameter / Device (433 MHz) Al-CRF Au-CRF Al-TPR Au-TPR
Unmatched insertion loss [dB] 6.5 10.5 6.5 7.5
Group delay (50 load) [s] 4.01 3.44 3.94 2.83
Loaded Q 5450 4430 5350 3890
Unloaded Q 10400 6190 10160 6730
Table 8. Comparison of the insertion loss, Q and group delay data at resonance of the
uncoated RSAW sensors with Al and Au metallization characterized in Fig. 12


539
TPR_Au device
MKR( 250): 437.3758MHz
MAGTD ( ) -7.56dB 5dB/
PHASE ( ) -7.7deg 100deg/

Al - TPR device
MKR( 250): 432.436MHz
DLY ( ) 3.94us 2us/

a) c)
CRF_Au device
MKR( 299): 410.45MHz
MAGTD ( ) -10.45dB 5dB/
PHASE ( ) -45.0deg 100deg/

Al - CRF device
MKR( 250): 432.566MHz
DLY ( ) 4.01us 2us/

b) d)
Fig. 12. Frequency (upper curves) and phase/group delay (lower curves) responses of
RSAW single mode two-port resonators (upper row) and coupled resonator filter (lower
row) using (a) and (b) Au and (c) and (d) Al electrode structure
9.2 Chemo sensitive polymer films and deposition methods used
This section compares the new Au vs. the old Al sensor devices in their sensor
characteristics to find out if the replacement causes any performance degradation of the
sensor system the Au devices were expected to operate in. To check the sensor performance
we again used two types of polymer films: (A) solid Parylene C to simulate coating
behaviour with solid and semisolid films as described in Section 6 and (B) a soft polymer
called poly[chlorotrifluoroethylene-co-vinylidene fluoride] (PCFV) to test sensor
performance at high gas concentrations. Since Parylene C coating was discussed in Section 6
already, here we will briefly discuss a relatively novel soft polymer deposition method,
which is called electro spray method and is described in [24] in detail. We applied it
successfully to all 4 devices from Fig. 12 to obtain very uniform high-quality soft PCFV films
for reproducible sensor performance. According to this method, the holder with the SAW
devices mounted on it, spins in a cloud of very small liquid-phase polymer droplets coming
out from a narrow capillary tube and directed by an electrostatic field towards the sensor
surface. The droplets settle onto the device surface, stick together and form a uniform film.
Its thickness depends on the deposition time. Since the SAW device loss increases with film
thickness, a certain insertion loss value, as necessary for optimum sensor performance, can
be obtained simply by adjusting the deposition time. Except for excellent control over film
thickness and uniformity [24], major advantage of this polymer coating method is that, even


540
at 433 MHz, the droplets are much smaller than the acoustic wavelength of about 7 m at
this frequency. Because of the small droplet size, the electro spray films cause much less
propagation loss for the SAW, compared to films of the same type and thickness, deposited
in an older airbrush coating technique. In contrast to electro spray films, airbrush coatings
have a rough textured structure and the droplet size typically exceeds the acoustic
wavelength. The two optical microscope pictures in Fig. 13 a) and b) compare the film
structures of PCFV deposited with the two methods on identical 433 MHz RSAW devices.
Fig. 13 a) shows part of the electrode structure and bus bars at the centre of the electro spray
coated SAW device. For better visibility of the textured film structure obtained in an
airbrush technique, the picture in Fig. 13 b) has been taken with a slightly higher
magnification and shows part of the reflector with some free substrate area on which the
large drops are clearly visible.

a) b)
Fig. 13. Comparison of two identical RSAW devices PCFV coated with (a) the electrospray
and (b) the airbrush method
9.3 Polymer coating behaviour of Au vs. Al sensor resonators
The three dimensional data plots in Fig. 14 compare the Parylene C coating behaviour of the
Au and Al CRF devices from Fig. 12 b) and d) accordingly with a film thickness varying
from 0 to 700 nm. In both cases the magnitude of the left longitudinal mode decreases with
thickness until it disappears completely at about 300 nm for the Au and 450 nm for the Al
device. Above these thickness ranges both devices demonstrate a smooth and well behaved
single-mode resonance. In the 300 to 500 nm range the Au CRF reaches its highest mass
sensitivity, accompanied with a gradual increase in insertion loss while the loss of the Al
device decreases more rapidly. The critical for the system operation loss value of 20 dB is
reached at 370 nm vs. 450 nm for the Au and Al devices, respectively.
An identical comparative Parylene C coating test, (not shown here), was performed also
with the Au vs. Al TPR devices from Fig. 12 a) and c). In these tests again the Au devices
were found to tolerate the solid Parylene C better than their Al counterparts. The increase in
device insertion loss with Parylene C thickness for all four tested devices from Fig. 12 is
shown in Fig. 15. At film thicknesses up to about 180 nm, all devices yield the same loss
increase. Above 200 nm the loss behaviour starts to diverge. The two Al devices keep the
same insertion loss up to about 550 nm thickness. Above 200 nm thickness, the loss of the
Au devices increases at a much lower rate indicating that these devices can tolerate up to


541
40% thicker solid films than their Al counterparts for the same amount of loss increase.
Finally, Fig. 16 compares the frequency sensitivities of the four tested devices with Parylene
C film thickness which is also an indication of their gas probing sensitivity with solid films.
Up to about 300 nm thickness, the sensitivity slope of the devices is nearly identical with a
small advantage of the Al TPR device, followed by the Al CRF. Above 300 nm, the
sensitivity slope of the Al TPR device increases but in view of its strong loss degradation, its
sensitivity advantage gets lost. The other three devices keep a nearly constant sensitivity
slope up to 500 nm thickness. Below the practical 370 nm thickness at which the critical for
this particular system 20 dB of loss is reached for the Al devices, the sensitivity of all four
devices differs by less than 20%. This difference is insignificant for practical sensor systems.

a)

b)
Fig. 14. Parylene C coating behaviour of the RSAW CRF devices from Fig. 12 using (a) Au
and (b) Al metallization


542
0 50 100 150 200 250 300 350 400 450 500 550 600
0
5
10
15
20
25
30
35
TPR_Au
CRF_Al
CRF_Au
TPR_Al
I
n
c
r
e
a
s
e

i
n

i
n
s
e
r
t
i
o
n

l
o
s
s

[
d
B
]
Parylene thickness [nm]

Fig. 15. Insertion loss behaviour of the Au and Al devices from Fig. 12 vs. Parylene C
thickness

0 50 100 150 200 250 300 350 400 450 500
-16.0M
-14.0M
-12.0M
-10.0M
-8.0M
-6.0M
-4.0M
-2.0M
0.0
CRF_Au
TPR_Au
CRF_Al
TPR_Al
F
r
e
q
u
e
n
c
y

s
h
i
f
t

[
H
z
]
Parylene thickness [nm]

Fig. 16. Frequency (mass) sensitivity behaviour of the Au and Al devices from Fig. 12 vs.
Parylene C thickness
The soft PCFV polymer coating experiments were performed on the two TPR devices from
Fig. 12 a) and c) since they have about the same amount of loss prior to coating, as required
by the sensor system. Since the devices are mounted on a spinning holder, monitoring of
their electrical performance in the process of electro spray deposition is not possible. That is
why we recorded the frequency and phase responses of each device prior to and after the
deposition to obtain the frequency shift and loss increase as a function of the deposition time
which we used as a measure for the thickness of the soft PCFV film. The data plots in Fig. 17
illustrate the PCFV coating behaviour of an Au TPR device in a 7,5 minutes deposition time.
As a result of film loading the device loss increases by about 8 dB to 17,9 dB while its
frequency shifts down by 3 MHz. The coated device on the left retains a well behaved
single-mode resonance with a smooth phase response in the resonance region. The loss
increase vs. thickness proportional deposition time for the Al and Au devices is shown in
Fig. 18. For the Au device this dependence is linear while the Al device shows a small loss
increase up to about 10 min. of deposition time and after that its loss degrades very rapidly.
We attribute this behaviour again to the difference in Au vs. Al densities. Once the Au
device has been optimized for operation under the heavy Au film load, it tolerates much


543
better the polymer film which is much lighter than Au. Finally, Fig. 19 compares the
frequency sensitivity of both devices to increased PCFV thickness. The curve for the Au
device is steeper meaning that its mass sensitivity is higher than its Al counterpart.

Fig. 17. Frequency responses (upper plots) and phase responses (lower plots) of an Au TPR
device before (data on the right) and after (data on the left) 7.5 minutes of PCFV deposition
using the electro spray method

0 2 4 6 8 10 12 14 16 18
0
5
10
15
20
Au TPR data points
Linear fit of the Au TPR data
Al CRF data points
Exponential fit of the Al CRF data
L
o
s
s

i
n
c
r
e
a
s
e

i
n

d
B
Coating time in minutes

Fig. 18. Loss increase vs. deposition time for Au and Al devices electro spray coated with
soft PCFV film


544
0 2 4 6 8 10 12 14 16 18
-6
-5
-4
-3
-2
-1
0
Au TPR - PCFV coating data points
Al CRF - PCFV coating data points
F
r
e
q
u
e
n
c
y

s
h
i
f
t

i
n

M
H
z
Coating time in minutes

Fig. 19. Frequency downshift vs. deposition time for Au and Al devices electro spray coated
with soft PCFV film
9.4 Gas probing behaviour of PCFV coated Au and Al TPR and CRF RSAW sensors
To compare the gas sensitivities of Au vs. Al sensors, pairs of devices of the same type
according to Fig. 12 were PCFV coated in the same electro spray deposition method and
probed with cooling agent, octane and tetrachloroethylene at different concentrations.
Cooling agent and octane were among those gases that the sensors were intended to operate
with in a specific application. After coating, the Au/Al pairs were selected to have nearly
the same loss increase as a result of PCFV deposition to simulate identical mass loading. The
gas probing results for 4 different loss increase values (thicknesses) are summarized in Table
9. As expected from the mass sensitivity data in Fig. 19, the Au devices demonstrate higher
gas sensitivity than their Al counterparts. Another important conclusion evident from Table
9 is that soft polymer coating also requires an optimum film thickness for maximum gas
sensitivity. In this case the optimum PCFV thickness is achieved when both the Au and Al
devices are coated to about 6 dB loss increase, (numbers in bold in Table 9).

Probing gas
Loss increase values
for each device pair
Cooling agent
40000 ppm
concentration
Octane
1100 ppm
concentration
Tetrachloroethylene
1000 ppm
concentration
PCFV coated device Al CRF Au TPR Al CRF Au TPR Al CRF Au TPR
Al_8/Au_8,7 dB 12,5 kHz 15 kHz 13 kHz 15 kHz 23 kHz 25 kHz
Al_6,2/Au_5,5 dB 11 kHz 21 kHz 14 kHz 19 kHz 26 kHz 28 kHz
Table 9. Summary of the gas probing performance of Au vs. Al device pairs PCFV coated to
nearly identical loss increase values in the same electro spray deposition process
10. Summary and conclusions
This chapter has highlighted important practical aspects for the design and operation of
chemical gas detection systems using STW and RSAW resonant devices. It has been shown
that both acoustic wave modes provide excellent gas sensitivity and low detection limits,


545
down to a few ppb when coated with solid, semisolid and soft polymer sensing layers.
Furthermore, the RSAW and STW modes do not only compete but rather complement each
other in different measurement tasks. The STW mode operates better with solid and
semisolid sensing films featuring surface sorption and is better suited for high-resolution
measurements at low gas concentrations (<1%) while the RSAW mode tolerates much better
thick soft sensing layers with profound bulk sorption that operate better at high gas
concentrations (>1%). Carefully designed RSAW sensors with Au metallization provide an
excellent corrosion proof substitute of their Al counterparts when operated in highly
reactive gas-phase environments, thereby greatly increasing system reliability and
measurement reproducibility over time and a large number of measurement cycles. All gas
sensors, regardless of acoustic wave mode, design, metallization and type of sensing
polymer requires a careful thickness optimization to provide highest gas sensitivity,
maximum dynamic range and lowest detection limit.
11. Acknowledgments
The author wishes to gratefully acknowledge Dr. E. Radeva from the Georgi Nadjakov
Institute of Solid State Physics, Bulgarian Academy of Sciences in Sofia, Bulgaria for expert
preparation of the HMDSO films used in this study as well Professor Shigeru Kurosawa and
his research associates from the National Institute of Materials Chemistry in Tsukuba, Japan
for the deposition of the semisolid ST and AA films. Special thanks are directed to Dr.
Michael Rapp and his research team at the Research Centre Karlsruhe in Germany for the
opportunity to perform a substantial part of this work at those laboratories.
12. References
[1] R. M. White, Acoustic sensors for physical, chemical and biochemical applications, Proc.
IEEE 1998 International Symposium on Frequency Control, pp. 587-594.
[2] R. M. White, Surface acoustic wave sensors, Proc. IEEE 1985 Ultrasonics Symposium, pp.
490-494.
[3] H. Wohltjen, Mechanism of operation and design considerations for surface acoustic
wave device vapour sensors, Sens. Act., vol. 5, p. 307, 1984.
[4] R. Chung, R. A. McGill and P. Matthews, Phase noise characterization of polymer
coated SAW-gas sensors: Implications for the performance of an oscillator circuit,
Proc. 1997 IEEE Int. Freq. Control Symp., pp. 169-174.
[5] S. J. Martin, G. C. Frye, J. J. Spates, and M. A. Butler, Gas sensing with acoustic devices,
Proc. IEEE 1996 Ultrasonics Symposium, pp. 423-434.
[6] M. Rapp, J. Reibel, S. Stier, A. Voigt, and J. Bahlo, SAGAS: Gas analyzing sensor systems
based on surface acoustic wave devicesAn issue of commercialization of SAW
sensor technology, in Proc. IEEE Int. Freq. Contr. Symp., 1997, pp. 129132.
[7] E. J. Staples, Dioxin/furan detection and analysis using a SAW based electronic nose,
Proc. IEEE 1998 Ultrasonics Symposium, pp. 521-524.
[8] E. J. Staples, T. Matsuda, and S Viswanathan, Real Time Environmental Screening of Air,
Water and Soil Matrices Using a Novel Field Portable GC/SAW System,
Environmental Strategies for for the 21st Century, Asia Pacific Conference, pp. 8-10
April 1998.


546
[9] United States Patent No. 5,289,715, Vapour Detection Apparatus and Method Using an
Acoustic Interferometer.
[9] G. C. Frye, S. J. Martin, R. W. Cerenosek, K. B. Pfeifer, and J. S. Anderson, Portable
acoustic wave sensor systems, in Proc. IEEE Ultrason. Symp., 1991, pp. 311316.
[10] G. C. Frye and S. J. Martin, Dual output acoustic wave sensor for molecular
identification, in Proc. IEEE Transducers, 1991, pp. 566-569.
[11] T.Wessa, S. Kueppers, G. Mann, M. Rapp, and J. Reibel, On-line monitoring of process
HPLC by sensors, Organ. Process Res. Develop., vol. 4, no. 2, pp. 102106, 2000.
[12] D. A. Howe, D.W. Allan, and J. A. Barnes, Properties of Signal Sources and
Measurement Methods, U.S. Dept. Commerce, NIST Tech. Note 1337, 1990.
[13] D. S. Ballantine, Jr. and H.Wohltjen, Elastic properties of thin polymer films
investigated with surface acoustic wave devices, in Chemical Sensors and
Microinstrumentation, R.W. Murry, R. E. Dessy,W. R. Heinemann, J. Janata, and W.
R. Seitz, Eds. Washington, DC: Amer. Chem. Soc., 1989, pp. 222236.
[14] J. W. Grate and E. T. Zellers, The fractional free volume of the sorbed vapor in modeling
the viscoelastic contribution to polymer-coated surface acoustic wave vapor sensor
responses, Anal. Chem., vol. 72, no. 12, pp. 28612868, July 1, 2000.
[15] F. L. Dickert and A. Haunschild, Sensor materials for solvent vapor detection-donor-
acceptor and host-guest interactions, Adv. Mater., vol. 5, no. 12, pp. 887895, Dec.
1993.
[16] N. Barie, M. Rapp, and H. J. Ache, UV crosslinked polysiloxanes as new coating
materials for SAW devices with high long-term stability, Sens. Actuators B, Chem.,
vol. B46, pp. 97103, 1998.
[17] H. Yasuda, Plasma Polymerization. New York: Academic, 1985, p. 11 and 294.
[18] C. Hamann and G. Kampfrath, Glow discharge polymeric film: Preparation, structure,
properties and applications, Vacuum, vol. 34, pp. 10531059, 1984.
[19] E. I. Radeva, Thin Plasma-Polymerized Layers of Hexamethyldisiloxane for
Acoustoelectronic Humidity Sensors, Sensors and Actuators, B, vo1.44, pp.275-278,
1997.
[20] G. Kovach, G. W. Lubking, M. J. Vellekoop, and A. Venema, Love waves for (bio)
chemical sensing in liquids, in Proc. IEEE Ultrason. Symp., 1992, pp. 281285.
[21] I. D. Avramov, M. Rapp, A. Voigt, U. Stahl and M. Dirschka, Comparative studies on
polymer coated SAW and STW resonators for chemical gas sensing applications,
Proc. 2000 IEEE International Frequency Control Symposium, pp. 58-65.
[22] I. D. Avramov, A. Voigt and M. Rapp, "Rayleigh SAW Resonators Using Gold Electrode
Structure for Gas Sensor Applications in Chemically Reactive Environments",
Electronics Letters, 31-st March 2005, Vol. 41, No. 7, pp. 450-452.
[23] I. D. Avramov, Design of Rayleigh SAW resonators for applications as gas sensors in
highly reactive chemical environments, Proc. 2006 IEEE International Frequency
Control Symposium, 5-7 June 2006, Miami, Florida, USA, pp. 381-388.
[24] F. Bender, L. Waechter, A. Voigt and M. Rapp, Deposition of High-Quality Coatings on
SAW Sensors using Electrospray, IEEE Sensors Conference, 2003, Paper ID 1022 (on
CD).
24
Ultrananocrystalline Diamond as Material for
Surface Acoustic Wave Devices
Nicolas Woehrl and Volker Buck
University Duisburg-Essen and CeNIDE, Duisburg
Germany
1. Introduction
Diamond is one of the most promising materials for SAW applications due to the highest
sound velocity and thermal conductivity. Unfortunately single crystals and epitaxial CVD-
films are expensive and beyond that conventional CVD grown microcrystalline diamond
films feature large facet structures with high roughness inapplicable for this application.
Ultra-Nanocrystalline diamond (UNCD) films grown in a microwave plasma enhanced
chemical vapor deposition (MPECVD) system on Si substrates possess a smooth surface
making it an ideal material for SAW applications. Moreover, due to its nanocrystalline
structure, the film properties of the UNCD material can be tailored in a wide range to adjust
them to the specific needs of a SAW filter. For this task a profound understanding of the
growth process of UNCD and the dependency of the film performance from the film
properties is needed. In addition, a simple and quick method to characterize the properties
of the UNCD films is necessary. Laser-induced SAW pulse method, which is fast and
accurate, is demonstrated to measure the mechanical and structural properties of the UNCD
films. AFM measurements were done to correlate the SAW pulse method results with the
surface roughness of the deposited films.
Another advantage of the UNCD films is, that highly C-axis textured aluminum nitride
(AlN) films can be grown directly on UNCD films by DC-sputtering. Using this technique a
feasibility study for SAW devices has been successfully performed.
2. Saw filters
SAW devices are electromechanical products commonly used in high frequency applications
such as filters, oscillators and transformers and are based on the transduction of acoustic
waves. SAW filters are now widely used in mobile telephones applications for filtering
and provide significant advantages in performance, cost, and size over other filter
technologies such as quartz crystals (based on bulk waves), LC filters, and waveguide
filters by offering a high degree of frequency selectivity with low insertion loss in compact
size (Campbell, 1989). In SAW filters an Interdigital Transducer (IDT) that is attached to a
piezoelectric material converts electrical signals to a mechanical wave. The piezoelectric
effect and the electric field generated by the IDT are distorting the crystal close to its
surface.


548
The oscillations of the crystal lattice can than add up by constructive wave interference and
superimpose to a surface wave before being converted back to an electrical signal by further
electrodes.
The passing frequency of a SAW filter can be calculated by

0
0
= f
v
(1)
where represents the wavelength of the acoustic surface wave (corresponding to twice the
distance of the fingers of the IDT) and v
0
is the crystal-dependent velocity oft the surface
wave. The operation frequency of a SAW device is closely related to the spacing of the
interdigital transducer (IDT) that is significantly limited by the photolithograph capability
(Springer et al., 1999). Thus one way to achieve higher passing frequencies is to use crystals
with a higher speed of sound, such as sapphire (Caliendo, 2003), SiC (Takagaki, 2002) or
diamond (Yamanouchi et al., 1989)(Nakahata et al., 1992).
3. Diamond as SAW material
The material with the highest speed of sound is diamond with 18000 m/s. Besides the high
speed of sound, diamond features other remarkable properties such as high thermal
conductivity and high hardness to name only a few. Due to its extraordinary properties
natural and HPHT diamond is used for a long time as a material for tools, especially for
grinding or sawing of rocks. Since the 1980s the microcrystalline diamond deposited by thin
film technology is increasingly used. One major problem with the microcrystalline diamond
films deposited in CVD processes especially for microelectronics and micromechanical
applications with their decreasing structural sizes - is the high surface roughness (Malshe et.
al., 1999). Moreover, high surface roughness results in large propagation loss, reducing the
applicability of the material. Although Sumitomo Electrics developed SAW filters and
resonators with various bandwidth in the 2-5 GHz range it turned out that the polishing of
the rather rough CVD diamond surface was too expensive and time consuming due to the
chemical inertness and highest hardness of diamond and the SAW filters were never
produced in an industrial scale (Fujimori, 1996). Even if one solution to this problem was
demonstrated by using the unpolished nucleation side of freestanding CVD diamond
(Lamara et. al., 2004) this idea never went into production.
Another drawback of the microcrystalline diamond films is that the homogeneous
deposition of such films on substrates with a high aspect ratio is difficult because the films
consist of relatively large crystals.
4. Nanocrystalline diamond as SAW material
The growing interest in nanotechnology and nanostructured materials has encouraged the
research of diamond films with reduced grain size. By reducing the grain size those films
feature rather unique combinations of properties making them potential materials for
emerging technological developments such as Nano/Micro- Electro-mechanical Systems
(N/MEMS) (Auciello et. al., 2004) (Hernandez Guillen, 2004), optical coatings, bioelectronics
(Yang et. al., 2002), tribological applications (Erdemir et. al., 1999) and also surface acoustic
wave (SAW) filters (Bi et. al., 2002).

Ultrananocrystalline Diamond as Material for Surface Acoustic Wave Devices

549
The nanostructured films differ from the microcrystalline films in grain size and in
roughness of the surface as shown in Fig. 1.

Fig. 1. Morphological comparison of microcrystalline diamond film (upper picture) and
UNCD film (lower picture). The scale bar in the upper picture corresponds to 1 m while
the scale bar in the lower picture corresponds to 5 m
The terms nanocrystalline (NCD) and ultrananocrystalline diamond (UNCD) were coined
by the Argonne National Laboratory group that performed the pioneering work in this field.
These terms were introduced to establish a differentiation to the microcrystalline diamond
films that differ not only in film properties but also in the way they are deposited. The
technology developed at Argonne National Laboratory started from deposition of hydrogen
free plasmas using fullerenes in Ar (Ar/C
60
) and was thereafter extended to hydrogen
diluted plasmas using Ar/CH
4
and gas mixtures containing only about 1 % hydrogen
(either added intentionally or through the thermal decomposition of CH
4
) (Gruen, 1999).
UNCD is grown from Argon-rich plasma giving it a very fine and uniform structure with
grain sizes between 2 and 15 nm (Auciello et. al., 2004). The grain sizes are independent


550
from film thickness due to the high secondary nucleation of new growth sites during the
whole deposition that is not taking place in the standard growth of diamond. This can be
shown within the experimental errors when measuring the Youngs modulus (GPa) as a
function of the deposition time (Shen et. al., 2006).
UNCD consists of pure sp
3
crystalline grains that can be separated by atomically abrupt (0.5
nm) grain boundaries or embedded in an amorphous 3D matrix. By reducing the grain size
of microcrystalline diamond films the amount of material between the grains is increased.
This matrix in the films can contribute to a large fraction of the overall film, sometimes
exceeding 10 % of the total volume, giving those films a great proportion of non-diamond or
disordered carbon (Auciello et. al., 2004]. But also values down to 5 % sp
2
-bonded carbon
have been reported and determined by UV Raman spectroscopy and synchrotron based
near-edge X-ray absorption fine structure measurements (NEXAFS) (Gruen, 1998).
In fact the overall volume and structure of the film matrix significantly determine the
properties of nanocrystalline diamond films giving another degree of freedom for the
material. The well-aimed use of an amorphous matrix for nanocrystalline diamond grains
leads to an enormous field of new materials, because a whole class of carbon based materials
(diamondlike carbon, DLC) can be used as matrix that may contain carbon solely (a-C) or
carbon and hydrogen (a-C:H) as well as other components like metals (Me-C:H);
additionally other dopants like silicon, oxygen, halogens or nitrogen may be added with
considerable effect on the film properties. By combining soft matrix properties with the hard
diamond crystals on the nanoscale it is possible to combine hard with elastic properties and
get a material that is hard and tough at the same time. With tailoring the mechanical stress
in the films or the coefficient of thermal expansion it was possible to tailor yet other very
important mechanical properties for the application of UNCD films by adjusting the overall
matrix fraction to the film volume (in the case of a 3D matrix surrounding the nanocrystals)
(Woehrl & Buck, 2007) (Woehrl et. al., 2009).
Thus, when comprehensively characterizing UNCD films, one also has to analyze the matrix
properties. Since the carbon atoms in the matrix have no crystalline configuration and are
indeed amorphous, conventional techniques known from the analysis of amorphous carbon
films can be used.
5. Deposition of UNCD films
It is well accepted that the initial nucleation is one decisive factor for the subsequent CVD
diamond film growth. While a low nucleation density can lead to van-der-Drift growth
known as the survival of the largest high initial nucleation leads to shorter coalescence
time and lower surface roughness. Due to the fact that substrate pre-treatment can
significantly increase initial nucleation, the pre-treatment is an important process step
already predetermining the film properties (Liu & Dandy, 1995). Three effective seeding
methods are known: Mechanical scratching of the substrate surface (see e.g. (Buck &
Deuerler, 1998)), enhancing the nucleation by applying a bias voltage to the substrate in the
early stages of deposition (Yugo, 1991), and ultrasonically activating the substrate in a
suspension containing diamond powder (Lin et. al., 2006)(Sharma et. al., 2010). Nucleation
densities of 10
10
cm
-2
or more were achieved with either of these methods. The latter method
was used for the substrates in this work mainly because of the good reproducibility and
uniformity even with larger substrates. Details on the pre-treatment and the deposition
parameters used for the UNCD films deposited in this work are given below.


551
Ultra-Nanocrystalline diamond (UNCD) films were synthesized by microwave plasma
enhanced chemical vapour deposition technique using a 2.45 GHz IPLAS CYRANNUS
I-6
plasma source. The nanocrystalline films were deposited from an Ar/H
2
/CH
4
plasma.
As standard substrates in this work (100) oriented double side polished silicon wafers with a
thickness of 425 m were used. The substrates were usually cut from a wafer to a size of
about 20 x 20 mm.
To enhance the nucleation of diamond the substrates were ultrasonically scratched for 30
min with a scratching solution consisting of diamond powder (~ 20 nm grain size), Ti
powder (~ 5 nm particle size) and Ethanol in a weight percent ratio of 1:1:100 (wt%).
Afterwards the substrates were ultrasonicated for 15 min in Acetone to clean the surface
from any residues of the scratching solution (Lin et. al., 2006) (Buck, 2008). After the
substrate pre-treatment they were immediately installed into the vacuum chamber placed
on top of the molybdenum substrate holder and the recipient was pumped down to high-
vacuum.
The plasma is ignited at ca. 1 mbar pressure with a process gas mixture of hydrogen ( 3 %)
in argon ( 97 %) and a MW-power of 1 kW. After the ignition the pressure was slowly
increased up to the deposition pressure (typically 200 mbar) during a 30 min period. This 30
min step is due to two reasons: Firstly the substrate surface is cleaned by the etching effect
of the plasma. Secondly the temperature of the substrate is slowly increased in the process
of the rising pressure. In doing so the substrate is already close to the targeted deposition
temperature before switching to the deposition parameters and introducing the carbon
carrier gas into the chamber. During the whole process of increasing the pressure the MW
power coupling into the plasma is adjusted to the changing conditions. After reaching the
desired deposition pressure the carbon carrier gas was introduced therewith starting the
deposition process.
The nanocrystalline films shown here were deposited at a pressure of 200 mbar from an
Ar/H
2
/CH
4
plasma. To investigate the influence of the hydrogen admixture on the
properties of the deposited films, the percentage of hydrogen in the process gas was varied
between 2 % and 8 % as shown in Table 1.
The MW-power was kept constant at 1 kW and the films were deposited for 5 h.

Pressure 200 mbar
Gasflow 400 sccm
H
2
fraction 2 % - 8 %
CH
4
fraction 0,8 %
Ar fraction 91,2 % - 97,2 %
MW-power 1 kW
Deposition time 5 h
Substrate Pretreated Silicon (100)
Table 1. CVD Deposition Parameters
6. Morphology of the films
The deposition parameters were systematically varied to investigate the influence on film
structure and film properties with special attention to the speed of sound and the roughness
of the films as most important properties for the application as SAW filters. Because of that


552
the main focus was on deposition parameters that influence the diamond grain size and
matrix. It is expected that both are directly influencing the elastic modulus of the films and
thus the speed of sound. One important parameter that is influencing the crystal size is the
admixture of hydrogen in the process gas. The higher the hydrogen fraction the bigger the
crystals grow. (Woehrl & Buck, 2007)
In previous publications it was suggested that different species for the nucleation on the one
hand and the growth of diamond grains on the other hand exist. The ratios of these species
determine the macroscopic structure of the growing films by influencing the rate of
secondary nucleation and therefore the matrix density and the grain size of the growing
crystals. A higher amount of the nucleation species leads to smaller crystals and more
material between the grains. A higher amount of growth species allows the grains to grow
faster (thus a higher growth rate) suppressing the secondary nucleation. In the literature, C
2

was suggested to be the nucleation species (Gruen, 1999) as strong emission of the C
2
dimer
could be found in the plasmas used for the deposition of fine-grained UNCD films. On the
other hand the CH
3
radical is generally believed to be the growth species of diamond films
(May & Mankelevich, 2008). Without taking part in the discussion concerning specific
details of growth and nucleation species, previously published data can be interpreted in a
way that these two competitive processes determine the structure of the deposited films.

Fig. 2. Morphology of UNCD films deposited with different hydrogen admixtures. The scale
bars in all three pictures correspond to 2 m
The atomic force microscope (AFM) is a scanning probe type microscope that offers a
resolution of less than a nanometer that is by a factor of 1000 better than the optical


553
diffraction limit. The AFM consists of a cantilever with a sharp tip with a radius of curvature
in the order of nanometers at its end that is used to scan the sample surface. When the tip is
brought close to the surface, atomic forces between the tip and the sample lead to a
deflection of the cantilever. The deflection of the cantilever is then measured by a laser that
is reflected from the cantilever onto an array of photodiodes. In comparison to the scanning
electron microscope (SEM) that is measuring a two- dimensional image of a sample not
necessarily corresponding to the morphological features, the AFM provides a true three-
dimensional topographical image of the surface giving information about the roughness of
the investigated surface. While specimens measured in SEM needs to be conducting and are
therefore often coated with a thin metal film (e.g. gold) irreversibly alter the film properties,
AFM measurements do not require such special treatments. While the SEM can easily
measure an area in the order of square millimeters with a depth of field on the order of
millimeters the AFM is usually restricted to a maximum scanning area of around 150 m
2

with a depth of field in the order of micrometers. Another characteristic that has to be
considered for high resolution AFM is the fact that the quality of an image is limited by the
radius of curvature of the probe tip and can lead to image artifacts. (Sarid, 1991)

Fig. 3. AFM measurements of UNCD samples deposited with 2,5 % H
2
(left) and 6 % H
2

(right). Both images cover a 5 x 5 m area
Fig. 2 shows UNCD films deposited with different admixtures of hydrogen to the process
gas. It is clearly seen that the hydrogen is influencing the morphology of the deposited
films. In fact the crystals are larger and the surface is rougher at hydrogen admixtures of 7%
compared to the films deposited at lower admixtures.
Fig. 3 shows an AFM measurement of 5 m thick UNCD films on a Si substrates. The
measured area on the sample was 25 m
2
. The RMS-roughness (root-mean-squared
roughness) of the surface is measured to be R
q
= 21.1 nm for the sample deposited at 2.5 %
H
2
(left picture) and R
q
= 51.3 nm for the sample deposited at 6 % H
2
(right picture).
SEM as well as AFM measurements show that higher hydrogen admixture in the process
gas lead to larger diamond crystals and rougher surfaces.
The RMS-roughness measurements as a function of the hydrogen admixture are shown in
Fig. 4.


554

Fig. 4. RMS-roughness measurements as a function of hydrogen admixture in process gas
7. Influence of nitrogen admixture on morphology
An especially appealing field of application for UNCD is nitrogen doped semiconducting
films. UNCD films are usually insulating, but n-doping is easily possible by admixture of
nitrogen to the process gas. (Gruen, 2004)
To investigate the influence of the nitrogen admixture in the plasma on the film properties,
more films were deposited at a pressure of 200 mbar with admixtures of nitrogen from 0 %
to 7.5 %.

Fig. 5. High resolution SEM measurement of a UNCD film deposited with 2.5 % hydrogen
and 2.5 % nitrogen admixture. The scale bar shown corresponds to 1 m


555
High-resolution SEM pictures were taken to investigate the influence of hydrogen and
nitrogen admixture on the morphology of the films. Fig. 5 shows a film deposited with 2.5 %
hydrogen and 2.5 % nitrogen in the plasma. The diamond grains appear to be very fine.
Increasing the nitrogen admixture to 7.5 % and keeping the hydrogen admixture at 2.5 %
changes the shape of the diamond grains. They appear to be needle-shaped as shown in Fig.
6. These measurements show that the nitrogen admixture can influence the shape of the
diamond grains.

Fig. 6. High resolution SEM measurement of a UNCD film deposited with 2.5 % hydrogen
and 7.5 % nitrogen admixture. The scale bar shown corresponds to 200 nm
It is expected that the change in the crystal shape will have a strong influence on the
propagation speed of sound in the material giving yet another degree of freedom when
designing the material for specific applications.
8. SAW pulse technique
The low surface roughness of UNCD films on the one hand and the high speed of sound in
single crystalline diamond on the other hand are making UNCD a very promising material
for SAW application. Yet the decisive question is whether the abundance of grain
boundaries in the films or the amorphous matrix surrounding the grains will change this
picture by e. g. damping the excellent propagation characteristics of surface acoustics waves.
The laser-induced SAW pulse method is capable of measuring the SAW-related (i.e.
mechanical and structural) properties of thin films (Weihnacht et. al, 1997) (Schenk et. al.,
2001) and was used in this work. The applicability of this method for investigating the film
properties of polycrystalline diamond films was demonstrated in previous publications
(Lehmann et. al., 2001). This method allows measuring all necessary material constants and
the wave excitation and propagation parameters decisive for the performance of the SAW
material. The biggest advantage of this method is, that it is not necessary to prepare a
piezoelectric layer or patterning an interdigital transducer (IDT) structure on the surface,
and that rather thin films can also be measured without being disturbed by effects from the
Si substrate. The method is a fast and accurate way to measure acoustic wave propagation


556
effects in thin film systems (Schneider et. al. 1997). Pioneering work on utilizing surface
acoustic waves as a tool in material science has been done by P. Hess, a general overview
can be found in (Hess, 2002).
A somewhat different setup has been used in this work and is schematically shown in Fig. 7.
This setup is commercially available at Fraunhofer IWS Dresden
1
.
A pulsed laser beam (N
2
-laser at 337.1 nm, 0.5 ns pulse duration) is focused on the substrate
by a cylindrical lens to excite a line-shaped broadband SAW pulse via a thermo-elastic
mechanism. A piezoelectric PVDF polymer foil, pressed onto the sample surface by a sharp
steel wedge (width around 5 m), is used as a broadband sensor for detecting the SAW
pulse propagated along the surface of the thin film system. SAW propagation measurements
are performed for different propagation lengths between a few mm and some cm. The signal
will then be amplified, digitized by an oscilloscope and converted to complex valued spectra
(i.e. amplitude and phase spectra) by a fast Fourier transform algorithm. By doing so for
different well-known propagation lengths on the one hand the SAW phase velocity
dispersion can be determined accurately from the accompanying phase spectra. Knowledge
of the velocity dispersion of a film system is decisive, because it gives the possibility to
recover the materials parameters (e.g. elastic constants, mass density and film thickness). To
derive the elastic properties, a theoretical approach, modeling the films as an isotropic layer
but taking into account the anisotropy of the silicon substrate, was fitted to the measured
dispersion data. The fact that we have a specimen that consists of a film on top of a substrate
introduces a length scale, and thus generates the observed dispersion effect from that the
elastic and mechanical properties can be derived.

Fig. 7. Principle of SAW pulse technique
A measurement of the SAW phase velocity as a function of frequency as well as the fitted
data is shown in Fig. 8. The phase velocity increases with frequency in the case of diamond
on silicon substrate (anomalous dispersion or stiffening case), because the smaller
wavelengths, propagating predominantly in the film, have higher phase velocity.

1
LAWave (http://www.iws.fhg.de/projekte/062/e_pro062.html)


557

Fig. 8. Measured velocity dispersion and fitted data
Beyond that the damping of the amplitude spectra with increasing propagation length can
deliver an estimation of SAW propagation losses due to scattering at thin film imperfections.

Fig. 9. E-modulus as a function of hydrogen admixture
As expected the elastic modulus is higher (the material is stiffer) for higher admixtures of
hydrogen (Fig. 9). This can be explained by the larger diamond crystals and a smaller


558
contribution of amorphous matrix and the fact that the elastic modulus of the amorphous
matrix is significantly lower than the modulus of the diamond grains. While the elastic
modulus for diamond is around 1220 GPa the elastic modulus of the deposited UNCD films
can reach ca. 65 % of this value.
9. Influence of nitrogen and oxygen on mechanical properties
The influence of the nitrogen admixture on the elastic modulus of the deposited films was
measured by nanoindentation.
The films that were deposited with additional nitrogen are less stiff compared to films
where no additional nitrogen was used. The elastic modulus of the UNCD films deposited
with 2.5 % nitrogen in the plasma was measured to be around 370 GPa and increasing the
nitrogen admixture even higher to 7.5 % in the plasma resulted in UNCD films with values
for the elastic modulus as low as 100 GPa. Thus it was shown that UNCD films deposited
with additional nitrogen are unsuitable for the application as SAW device.
An opposite trend can be found when oxygen is used as admixture to the process gas. It was
shown that the Youngs modulus can be increased up to 950 GPa (ca. 75 % of single
crystalline diamond). The reason can be found in the effective etching of sp
2
-bonded carbon
by the oxygen in the plasma and thus bigger diamond crystals (Shen et. al., 2006).
10. Feasibility study
As a feasibility study SAW resonators with sputtered AlN film as piezoelectric transducer
have been produced. Fig. 10 shows the concept of the fabricated AlN-UNCD layered SAW
resonator.

Fig. 10. Schematic Structure of AlN-UNCD layered SAW resonator with golden IDT patterns
shaped by photolithography
In the previous chapters it was shown that UNCD films are very suitable as basic material
for SAW applications. It was shown that the addition of hydrogen on the one hand
improves the elastic constants (towards the value of diamond single crystals), and on the
other hand increases the roughness (to values of microcrystalline diamond films), which
leads to large propagation loss. Thus a compromise has to be made. The process parameters
used for this feasibility study are given in table 2.


559
Pressure 240 mbar
Gasflow 400 sccm
H
2
fraction 2 %
CH
4
fraction 0,8 %
Ar fraction 97,2 %
MW-power 1 kW
Table 2. Deposition conditions
In order to induce a surface acoustic wave in the UNCD material, a piezoelectric layer is
necessary. AlN was chosen for this feasibility study due to being the material with the
highest phase velocity (6700 m/s) among piezoelectric materials (Ishihara et. al., 2002). The
applicability of AlN thin films on various CVD diamond substrates was demonstrated
before (Chalker et. al., 1999).
AlN is an intrinsic piezoelectric material; the wurtzite structure is thermodynamically stable.
Several methods for deposition of AlN-films have been reported e.g. MOCVD (Tsubouchi &
Mikoshiba, 1985), MBE (Weaver et. al., 1990) and reactive DC or RF sputtering (Akiyama et.
al., 1998)(Karmann et. al., 1997). Reactive sputtering processes have the advantage of low
substrate temperatures (Dubois & Muralt, 2001)(Naik et. al., 1999)(Tait & Mirfazli,
2001)(Assouar et. al., 2004). Here, magnetron sputtering processes was chosen, for being a
common and reliable industrial process.
However, highly (002) oriented films with smooth surfaces are required. Thus deposition
parameters (power, pressure, N
2
ratio and substrate temperature) have to be systematically
optimized to reach this goal. The influence of oxygen on the film structure was
demonstrated before (Vergara et. al., 2004) showing that a low residual gas pressure is
crucial for the desired film properties. Therefore a vacuum chamber with turbo molecular
pump and a load lock system was used in this work to assure clean conditions. By that,
highly oriented AlN films with very smooth surface were deposited on UNCD films that
turned out to possess good piezoelectric properties. (Lee et. al., 2007). DC power was 300 W
at a pressure of 0.4 Pa and 50 sccm N
2
gas flow at 300C. The film thickness of the AlN films
was ca. 3.5 m and structure, morphology and bonding structure were characterized by X-
Ray diffractrometry (XRD), scanning electron microscopy (SEM), atomic force microscopy
(AFM), Raman spectroscopy (Renishaw, RA100) and NEXAFS in synchrotron technique.
On top of the AlN film a gold film was deposited by sputtering which was shaped by
conventional photolithography. The resonator consists of a central IDT with reflectors at
each side (Fig. 11).
The produced SAW Resonators were analyzed due to their performance. Thickness of
UNCD as well as AlN have been systematically varied (2 m to 6.2 m for UNCD, 1.4 m to
3.5 m for AlN). It was measured that with increasing thickness of AlN and UNCD films the
resonance frequency increases as well and the resonance peak become clearer. The increase
of resonance frequency and thus of SAW velocity is due to reduced influence of the low
SAW velocity of the Si substrate. The clearer resonance peak means larger coupling
coefficient, which is due to the relative thickness of AlN piezoelectric layer increasing.
Furthermore the influence of the IDT pair number on the SAW resonator performance was
investigated (100 Pairs to 200 Pairs). It was measured that the resonance frequency and the
resonance strength kept almost the same while doubling the IDT pair numbers.


560
This feasibility study indicates that the SAW velocity and coupling coefficient only depend
on the relative thickness of ALN and UNCD films, but are not affected by IDT pattern.

Fig. 11. Schematic Pattern design of SAW Resonator. The actual device consists of significant
more lines
11. Acknowledgment
The authors like to thank Dr. Dieter Schneider at Fraunhofer IWS Dresden for the E-
modulus measurements of the UNCD films and Hanna Bukowska, University Duisburg-
Essen for the AFM measurements.
12. References
Akiyama M., Xu C.-N., Nonaka K., Shobu K., Watanabe T., Thin Solid Films, 315 (1998) 62
Assouar M. B., El Hakiki M., Elmazria O., Alnot P., Tiusan C., Diamond Relat. Mater. 13
(2004) 1111
Auciello O., Birrell J.,. Carlisle J. A, Gerbi J. E., Xiao X., Peng B., Espinosa H. D., J. Phys.:
Condens. Matter 16 (2004) R539
Bi B., Huang W. -S., Asmussen J., Golding B., Diamond Relat. Mater. Vol. 11, Issues 3-6 (2002)
677-680
Buck V., Deuerler F., Diamond Relat. Mater. 7 (1998) 1544
Buck V., J. Opto. Adv. Mater. 10 (2008) 85
Caliendo C., Appl. Phys. Lett. 83 (2003) 4851
Campbell C., Academic Press (1989)
Chalker P. R., Joyce T. B., Johnston C., Crossley J. A. A., Huddlestone J., Whitfield M. D.,
Jackman R. B., Diamond Relat. Mater. 8 (1999) 309
Dubois M.-A., Muralt P., J. Appl. Phys., 89 (2001) 6389
Erdemir A., Fenske G. R., Krauss A. R., Gruen D. M., McCauley T., Csencsits R. T., Surf.
Coat. Techn., 120-121 (1999) 565- 572


561
Fujimori N., Recent Progresses in Electrical Application of Diamond, 7th European Conference
on Diamond, Diamond-like and Related Materials, 8-13 September 1996
Gruen D. M., MRS Bulletin, Vol. 23, No. 9 (1998) 32
Gruen D. M., Annu. Rev. Mater. Sci. 29 (1999) 211
Gruen D. M., Krauss A. R., Auciello O. H., Carlisle J. A., (2004) US Patent No. 6793849
Hernandez Guillen F. J., Janischowsky K., Ebert W., Kohn E., Phys. Stat. Sol., A, Appl. res.,
vol. 201, no11 (2004) 2553-2557
Hess P., Physics Today, March issue (2002) 42
Ishihara M., Nakamura T., Kokai F., Koga Y., Diamond Relat. Mater. 11 (2002) 408
Karmann S., Schenk H. P. D., Kaiser U., Fissel A., Richter W., Mater. Sci. Eng., B50 (1997) 228
Lamara T., Belmahi M., Elmazria O., Le Brizoual L., Bougdira J., Rmy M., Alnot P., Diamond
Relat. Mater. 13 (2004) 581
Lee Y. C., Lin S. J., Buck V., Kunze R., Schmidt H., Lin C. Y., Fang W. L., Lin I. N., Diamond
Relat. Mater. 17 (2008) 446
Lehmann G., Schreck M., Hou L., Lambers J., Hess P., Diamond Relat. Mater. 10 (2001) 686
Lin I. N., Lee Y. C., Lin S. J., Lin C. Y., Yip M. C., Fang W., Diamond Relat. Mater., Vol 15 No
11-12 (2006) 2046
Liu H., Dandy D. S., Diamond Chemical Vapor Deposition, Nucleation and early growth
stages, Noyes Publications (1995)
Malshe A. P., Park B. S., Brown W. D., Naseem H. A., Diamond Relat. Mater. 8 (1999) 1198
May P. W., Mankelevich Yu. A., J. Phys. Chem. (2008)
Naik R. S., Reif R., Lutsky J. J., Sodini C.G., J. Electrochem. Soc., 146 (1999) 691
Nakahata H., Hachigo A., Shikata S., Fujimori N., IEEE Ultrason. Symp. Proc. (1992) 377
Sarid D., Scanning Force Microscopy, Oxford Series in Optical and Imaging Sciences, Oxford
University Press, New York (1991)
Schenk H. P. D., Feltin E., Vaille M., Gibart P., Kunze R., Schmidt H., M. Weihnacht, E.
Dogheche, Phys. Stat. Sol., A Appl. Res. 188 (2) (2001) 537
Schneider D., Schwarz T., Scheibe H.-J., Panzner M., Thin Solid Films 295 (1997) 107
Shen Z. H., Hess P., Huang J. P., Lin Y. C., Chen K. H., Chen L. C., Lin S. T., J. Appl. Phys., 99
(2006) 124302
Sharma R., Woehrl N., Barhai P. K., Buck V., J. Opt. Adv. Mater., Vol.12 Iss. 9 (2010) 1915
Springer A., Hollerweger F., Weigel R., Berek S., Thomas R., Ruile W., Ruppel C., Guglielmi
M., IEEE Trans. Microwave Theor. Tech. 47 (1999) 2312
Tait R. N., Mirgazli A., J. Vac. Sci. Technol. A, 19 (2001) 1586
Takagaki Y., Santos P.V., Wiebicke E., Brandt O., Schoenherr H.P., Pliig K., Appl. Phys. Lett.
81 (2002) 2538
Tsubouchi K., Mikoshiba N., IEEE Trans. Sonic Ultrason., SU-32 (1985) 634
Vergara L., Clement M., Iborra E., Sanz-Hervs A., Garca Lpez J., Morilla Y., Sangrador J.,
Respaldiza M. A.
Weaver W., Timoshenko S. P.,. Young D. H, Vibration Problems in Engineering, John Wiley &
Sons., (1990)
Weihnacht M., Franke K., Kaemmer K., Kunze R., Schmidt H., IEEE Ultrason. Symp. Proc.
(1997) 217
Woehrl N., Buck V., Diamond Relat. Mater. 16 (4-7) (2007) 748
Woehrl N., Hirte T., Posth O., Buck V., Diamond Relat. Mater. 18 (2009) 224
Yamanouchi K., Sakurai N., Satoh T., IEEE Ultrason. Symp. Proc (1989) 351


562
Yang W. et al., Nature Materials 1, (2002) 253257
Yugo S., Kanai T., Kimura T., Muto T., Appl. Phys. Lett. 58, (1991) 1036
25
Aluminum Nitride (AlN)
Film Based Acoustic Devices:
Material Synthesis and Device Fabrication
Jyoti Prakash Kar
1
and Gouranga Bose
2

1
Department of Electronics Engineering, University of Tor Vergata, Rome
2
Department of Applied Electronics and Instrumentation Engineering,
Institute of Technical Education and Research, Bhubaneswar, Orissa
1
Italy
2
India
1. Introduction
Enormous growth has taken place in electronics, especially in the field of RF
communications towards the beginning of 21
st
century and continuously striving for better
communication performance. Presently, the key concerns of RF communications is
bandwidth, in the range of low/medium GHz range, to avoid frequency crowding,
especially for wireless communication mobile handsets and base stations (Kim et al., 2004).
In addition, reduction in signal loss, low power consumption, scaling down device size,
reduction in materials and fabrication costs, and packaging of the device are main issues
today. Some of these issues can be resolved, if the new generation of electroacoustic devices
can be monolithically integrated with integrated circuit (IC). Conventional electroacoustic
devices, used in the communication e.g. Surface Acoustic Wave (SAW) and Bulk Acoustic
Wave (BAW) based systems, are widely used for todays wireless communication. These
devices are typically made on a single crystal piezoelectric substrate such as quartz, lithium
niobate, and lithium tantalate (Assouar et al., 2004). Unfortunately, these substrates based
electroacoustic devices are made separately and then it is wired with the signal processing
chip, which has several limitations, in particular low acoustic wave velocity and high
frequency device fabrication. To resolve these two core issues, thin film materials based
electroacoustic devices are actively under consideration [Bender et al., 2003]. Where, a
crystalline film is grown on a particular substrate, especially silicon wafer and
electroacoustic device is made out of crystalline film. Thus, the electroacoustic device can be
integrated with the signal processing circuit. Apart from the silicon wafer as a base material
for crystalline film deposition, a variety of other substrates are also explored for academic
and technology interests. Furthermore, to get electroacoustic devices of better quality in
terms of high frequency and high quality factor (Q), the piezoelectric property of the film is
also exploited with different type of device concept called Micro-Electro-Mechanical
Systems (MEMS). Thin film bulk acoustic resonators (TFBAR) comes under this MEMS
devices, where the crystalline film is made to resonate at RF frequency. These MEMS


564
devices have smaller size, lower insertion loss and higher-power handling capabilities than
conventional SAW devices (Lee et al., 2004).
Generally, thin piezoelectric films, such as aluminum nitride (AlN), zinc oxide (ZnO) and
lead zirconium titanate (PZT) are used for high frequency acoustic devices (Loebl et al.,
2003; Yamada et al., 2004; Schreiter et al. 2004). AlN has higher SAW velocity, lower
propagation loss, and higher thermal stability in comparison to ZnO; whereas, PZT thin
films need selective substrates for deposition and thereafter, needs post-deposition poling to
get specific cystal orientation. Thus, AlN seems to have edge over the ZnO and PZT films
for electroacoustic devices. The critical factor of piezoelectric AlN thin film is its crystal
orientation and morphology. Furthermore, to integrate with the signal processing chip, it is
also essential that AlN film should be compatible to the complementary metal oxide
semiconductor (CMOS) fabrication processes. In addition, AlN being a dielectric material, it
can be used as an insulating material in integrated circuits as well as a piezoelectric material
in electroacoustic device. Thus, it is imperative to study the presence of electrical charges
and the nature of generation of defects in the AlN film along with its morphology. Usually,
there are four types of electric charges present in the insulating film; namely, bulk charges
(Q
in
) and interface (D
it
) charges, fixed charges (Q) and mobile charges (Q
m
). In present IC
processing, the presence of fixed charges (Q) and mobile charges (Q
m
) are eliminated upto a
large extent. Furthermore, the bulk charges (Q
in
) and interface (D
it
) charges are reduced
further by the optimization of growth parameter and the post-deposition treatments.
Reduction in the bulk charge (Q
in
) and interface charge (D
it
) density is most essential in
cantilever beam based MEMS resonator, otherwise the electrostatic force produced by the
these charges may stuck cantilever beam on the substrate (Luo et al., 2006). Most of the
MEMS are made out of single crystal silicon substrate utilizing well-matured IC fabrication
technology. This poses a challenge to be compatible with a new generation of functional
materials. Apart from the electrical charges, the selective etching of piezoelectric materials
and silicon for electroacoustic device fabrication is a key technology.
2. Properties of AlN film
AlN is a III-V family compound having hexagonal wurtzite crystal structure with lattice
constants a = 3.112 and c = 4.982 (Yim et al., 1973). In this structure, each Al atom is
surrounded by four N atoms, forming a distorted tetrahedron with three Al---N
(i)
(i = 1, 2,3)
bonds named B
1
and one Al---N
0
bond in the direction of the c-axis, named B
2
. The bond
lengths of B
1
and B
2
are 1.885 and 1.917 , respectively. The bond angle N
0
---Al---N
i
is
107.7 and that for N
1
---Al---N
2
is 110.5 (Xu et al., 2001).
AlN has gained ground in semiconductor industry because of its unique electrical,
mechanical, piezoelectric and other properties (Table 1). Some of these noteworthy
properties are wide bandgap, high thermal conductivity, high SAW velocity, moderately
high electromechanical coupling coefficient, high temperature stability, chemical stability to
atmospheric gases below 700 C, high resistivity, low coefficient of thermal expansion (close
to Si), high dielectric constant and mechanical hardness (Xu et al., 2001; Strite et al., 1992;
Wang et al., 1994). Its high thermal conductivity (about 100 times that of SiO
2
and roughly
equal to that of silicon) and electrical insulating property can prove to be a good dielectric
layer for a new generation of integrated circuit devices, particularly in metal insulator
semiconductor (MIS) devices. High heat dissipation of AlN can significantly enhance device
lifetime and efficiency. AlN film with (002) preferred orientation (c-axis) has maximum
Film Based Acoustic Devices: Material Synthesis and Device Fabrication

565
piezoelectricity among all other orientations of its crystal structure (Naik et al., 1999).
Furthermore, its lattice matching is near to that of silicon and thus less stress is expected to
be generated at the AlN/silicon interface. Owing to these properties, AlN films have
received great interest as an electronic material for thermal dissipation, dielectric and
passivation layers for ICs, acoustic devices, resonators and optoelectronic devices.

Bandgap 6.2 eV, direct
Thermal conductivity 2.85 Wcm
-1
K
-1

Coefficient of thermal expansion 4-510
-6
K
-1

Refractive index 1.8-2.2
Dielectric constant 8.5
Electrical resistivity 10
11
-10
13
.cm
SAW velocity 6000 m/sec
Melting point 2490 C
Hardness 9 Mhos
Table 1. Properties of AlN
3. Synthesis of AlN film
Depending on the intended application, various techniques have been implemented for
synthesizing AlN films; namely, molecular beam epitaxy (MBE), reactive evaporation,
pulsed laser deposition (PLD), chemical vapour deposition (CVD) and sputtering. Among
these techniques, sputtering has the advantage of low-temperature deposition, ease of
synthesis, less expensive, non-toxic, good quality films with a fairly smooth surface [Kar et
al., 2006; Kar et al., 2007]. In addition, sputtering technique has also CMOS process
compatibility. In sputtering technique, plasma is created between the two electrodes by
applying high voltage in low pressure. The plasma region contains, positive ions, electrons
and neutral sputtering gas, thus the plasma behaves like a conducting medium. Usually,
argon gas is used as a sputtering gas. The material that is to be sputtered is called target and
it is fixed to the negatively charged electrode. The other electrode is called anode, which is
grounded so that the ratio of the target to anode area is significantly reduced. This electric
configuration of the sputtering system makes high electric field at the target and that
enhances the rate of sputtering. During sputtering process, the energetic ions strike the
target and dislodge (sputter) the target atoms. These dislodged atoms travel through the
plasma in a vapour state and stick to the surface of wafers, where they condense and form
the film. AlN film can be deposited either by directly using the AlN target or by sputtering
of aluminum metal in presence of argon and nitrogen gas. The sputtered aluminum atoms
react with the nitrogen gas and form AlN film. This process of film deposition is called
reactive sputtering deposition. The sputtering parameters are required to be optimized for
desired morphological and electrical properties. These deposition parameters are mainly
sputtering pressure, wafer to target distance, sputtering power and wafer temperature. AlN
film deposition by reactive sputter deposition technique requires nitrogen as a reactive gas,


566
where it is introduced into the sputtering chamber along with inert argon gas. Argon ions
produced in the plasma due to sputtering power and thereafter they strike to the
aluminum target and sputter aluminum atoms. These aluminum atoms react with
nitrogen and form AlN compound and that deposit on the wafer. Hence, the gas flow
ratios need to be optimized. To increase sputtering rate, magnets are placed under the
aluminum target, so that magnetic field and the electric field are perpendicular to each
other. This configuration of sputtering system is called magnetron sputtering technique.
In the magnetron sputtering, electrons travel in spiral motion in the plasma region. This
increases the collision of electrons to neutral argon atoms significantly and that increases
argon ions in many folds, thus sputtering rate becomes high.
AlN film can be deposited by DC (direct current) and RF (radio frequency, 13.56 MHz)
magnetron sputtering modes. In the DC mode of sputter deposition, the target material
must be conductive, so that plasma can sustain. If trace of impurity is present in the system,
the surface of the aluminum target becomes contaminated and target poisoning takes place.
On the other hand, RF sputtering has the major advantages to produce good quality film,
high deposition rate and less chance of target poisoning. For these reasons, RF sputtering
technique is preferred than the DC sputtering technique. To obtain well oriented crystalline
AlN films for SAW and MEMS structures, the RF sputtering parameters need to be
optimized. The sputtering parameters are: RF power, substrate temperature, sputtering
pressure, nitrogen concentration and target-substrate distances (D
ts
). AlN films are
deposited on CMOS IC compatibility silicon (100) wafer by the RF reactive magnetron
sputtering. The change in morphological and electrical properties of the AlN films with the
growth parameters are reported in following section.
3.1 RF power
Amorphous AlN film is found at lower RF sputtering power (100 W), but films became (002)
oriented at a sputtering power of 200 W. Further increase of RF power to 400 W, a
significant increase in (002) orientation has taken place. This is due to the increase of kinetic
energy of atoms that leads to atomic movements on the substrate surface as a result of
higher RF power. These newly arrived surface atoms are called ad-atom. Higher
sputtering power increases the AlN grain size that leads to increase in surface roughness as
shown in scanning electron microscope (SEM) images (Fig. 1) (Kar et al., 2009).

Fig. 1. SEM micrographs of AlN films deposited at (a) 200 W, and (b) 300 W

567
3.2 Substrate temperature
The structural and morphological properties of the deposited AlN films are strongly
dependent on the kinetics of the sputtered atoms arrived at the substrate. The kinetics of
sputtered atoms depends on the sputtering parameters. For instance, substrate temperature
increases the ad-atom mobility and changes the film morphology significantly. One such
illustrations of morphological change with temperature are seen from the X-Ray diffraction
(XRD) studies. It is clearly seen from the XRD studies that the c-axis oriented AlN (002)
peaks become prominent at moderate temperature range (200300 C), but degrades
significantly at 400 C (Kar et al., 2006). This can be attributed to the structural disorderness
resulting from the incorporation of impurity atoms at higher temperature (Wang, 2000). The
amount of contamination depends on the sputtering deposition system and process related
factors, such as base pressure, temperature, gas purity and the partial pressure of moisture, etc.
(Naik et al., 1999). Furthermore, smaller grain size with smoother surfaces is observed at lower
deposition temperature, and that increases with temperature (Fig. 2). A possible reason may be
that the smaller grains grow and merge to form bigger grain, due to the higher thermal energy.

Fig. 2. SEM micrograph of AlN films deposited at (a) 100 C, and (b) 400 C
3.3 Sputtering pressure
The variation in crystal orientation with different sputtering pressure are observed from the
XRD studies, where the intensity of (002) orientation increases with the deposition pressure
and attained a maximum value at 610
-3
mbar. On further increase to a deposition pressure
of 810
-3
mbar, the (002) crystal orientation of the AlN film is changed abruptly to the (100)
orientation with lesser intensity. The deposited atoms may have altered their direction,
energy, momentum and mobility due to the decrease in mean free path of the atoms with
sputtering pressure. The hexagonal wurtzite structure of AlN has two kinds of AlN bond
named as B
1
and B
2.
These bonds B
1
and B
2
together correspond to (110) and (002) planes,
where B
1
corresponds to (100) plane. The formation energy of B
2
is relatively larger that of B
1

(Cheng et al., 2003). Hence, the energy required for sputtering species to orient along c-axis
is larger than the other possible planes. At low pressure, sputtering species possess enough
energy to form hexagonal wurtzite crystalline structure on substrate surface. It is also
reported that the surface roughness of the film increases with the increase in deposition
pressure. The grain size is increased till 610
-3
mbar deposition pressure and then it reduced
to 80 nm at 810
-3
mbar (Kar et al., 2006). In addition, inhomogeneous patterns on the
surface are also observed at this higher pressure (Fig. 3). It is also observed that the AlN film


568
has changed its orientation with less Al-N bond density and reduction of grain size at 810
-3

mbar sputtering pressure. Hence, it is inferred that the structural disorder and/or the change
in the Al-N bond density/angles must have taken place at this particular sputtering pressure.

Fig. 3. SEM micrograph of the AlN films deposited at (a) 210
-3
mbar, and (b) 810
-3
mbar

Fig. 4. SEM micrograph of AlN films deposited at (a) 20 % N
2
, and (b) SEM image of AlN
film for D
ts
of 5 cm
3.4 Gas flow ratio
At lower nitrogen concentration, the intensity of (100) peak is relatively more prominent
than (002), but the trend reverses with higher nitrogen concentration (Kar et al., 2006). At
80% N
2
, a highly oriented (002) peak is observed without trace of (100) orientation. Lower
argon and higher nitrogen gas concentration results slower aluminum sputtering rate. If the
time interval for the arrival of Al species at the wafer surface is slower, the Al atom gets
enough time to react with N
2
. This increases the probability of Al-N bond formation and
bonded Al-N molecules get more time to adjust themselves along (002) orientation on the
substrate. On the other hand, at higher argon concentration, Al does not get enough time for
complete nitridation due to higher sputtering rate. In addition, faster arrival of the Al at the
substrate surface results not only in a poor AlN bond, but also provides less time for the
newly formed AlN to arrange itself along c-axis. A surface texture of smaller grain size,
smoother, homogeneous and dense granular microstructures has been observed at higher
concentrations of nitrogen. This indicates a low surface mobility of the ad-atoms at high

569
nitrogen concentration. In contrast, bigger grain size with increased roughness is observed at
lower nitrogen concentration, where the newly formed smaller grain merges together with a
previously formed grain and becomes bigger in size. The size and distribution of the micro-
grains is quite uniform at 80% nitrogen concentration. At lower nitrogen concentrations, Ar
+

ions transfer more energy to the Al target during bombardment, generating more aluminum
atoms that make clusters with incomplete nitridation of aluminum on the wafer surface. This
leads to formation of fewer bonds, a poor c-axis orientated and a rough film (Fig 4 (a)).
3.5 Target-substrate distance
The kinetics of the sputtered species arriving at the substrate controls the ad-atom mobility
and atomic rearrangement that governs the microstructure of the film. From the XRD
studies, it is observed that the intensity of c-axis orientation of the film decreases with
increase in target to substrate distance D
ts
(Kar et al., 2008). At shorter D
ts
, the Ar ions travel
almost normal to the target due to the high electrical field and knock out Al atoms around
perpendicular to the target. Because of short deposition path, the probability of collisions of
the Al atom with gas atoms is low. Therefore, a good quality film is obtained at lower D
ts
(5
cm). On the other hand, at larger D
ts
, the chances of Al collision with gas molecules is
increased. In this process Al atoms lose its kinetic energy significantly as well as alter
deposition angles. These randomly arriving Al atoms, with lesser energy, cause self-
shadowing effects and reduce atomic migration that leads to generation of voids in the film
(Lee et al., 2003). The grain size of the AlN film increases with D
ts
. For lower D
ts
, smaller
grain with minimum surface roughness is observed (Fig. 4(b)) and a coarser grain is found
at the highest D
ts
(8 cm). Surface roughness of the synthesized AlN films are also increases
with D
ts
. The kinetic energy of deposited species is considered to be a major factor for the
grain size and the surface roughness of the film.
3.6 Variation of electrical properties with sputtering parameter
The AlN film can be used as a dielectric layer in IC; hence, the electric charges are essential
to study with the sputter deposition parameters. Electric charges like Q
in
and D
it
are highly
governed by the sputter deposition parameters. A decrease in the Q
in
is observed with
sputtering power, where as D
it
is found to be minimum at moderate RF power. At higher
temperature, better electrical properties in the bulk as well as the interface of sputtered AlN
films are reported; this is mainly due to the formation of bigger grain size and its associated
effects. It is reported that the defects produced by stress, voids and incorporation of gases
are main responsible cause for the monotonic increase in Q
in
. The D
it
has a minimum value
at 610
-3
mbar sputtering pressure. The Q
in
and D
it
increases with nitrogen concentration.
This will have a deleterious effect for silicon-based devices at higher nitrogen concentration.
Rise in the Q
in
and D
it
with the increase in D
ts
is also reported. It is seen that at larger D
ts
, the
morphological as well as the electrical properties of the AlN films deteriorates, whereas, at
shorter D
ts
the quality of the film comes out to be better (Kar et al., 2007). Apart from the
electric charges, it is observed that better crystallinity posses AlN films of higher dielectric
constant.
4. Post-deposition annealing effect
AlN film may see high temperature, if AlN film is monolithically integrated during IC
fabrication. Post-deposition heat treatment significantly affects the morphology and electric


570
charges of AlN film. The post-deposition thermal treatments (annealing) of AlN film are
generally carried out by two distinguished modes; namely, Rapid Thermal Annealing (RTA)
and conventional furnace annealing.
4.1 Rapid thermal annealing (RTA) process
The XRD studies show that the intensity of c-axis (002) orientation increases upto annealing
temperature of 800 C in nitrogen ambient and then it marginally decreases at 1000 C (Kar
et al., 2005). The shift in XRD diffraction peaks is reported at higher temperatures, which
may be due to the generation of stress. Granular worm-like nanostructures are found in as-
deposited AlN films (Fig. 5), whereas cracks are observed for annealing at 1000 C. A short
duration of heat pulse by RTA is barely sufficient to modulate the film surface, but not
enough to activate the grains of the AlN films to merge themselves to form bigger grains.
Appearances of cracks are due to the stress developed in the film. The thermal coefficient
mismatch between the AlN and silicon substrate may be generated from the fast ramp up
and ramp down annealing heat cycle during rapid thermal annealing. The position and
density of cracks depend strongly on the defects, dislocations and the structural relaxation
of grain boundaries. The surface roughness is considerably increased for the film annealed
at higher temperatures due to the surface modulation.

Fig. 5. SEM micrograph of AlN films RTA processed at (a) as-deposited, and (b) 1000 C
4.2 Furnace annealing
The intensity of the (002) peak increases with furnace annealing temperature, where the
atoms acquire adequate activation energy to become (002) oriented. Sometimes, many of the
atoms may not be at the crystal lattice site in the as-deposited AlN film, which causes the
lattice strain and the formation of microvoids. During conventional furnace annealing,
atoms get enough time to acquire sufficient kinetic energy and occupy relative equilibrium
positions that minimize the lattice strain and microvoids, which results in a better crystalline
film. Furthermore, the furnace annealing process minimizes the dislocations and the other
structural defects and forms a better stoichiometric material. From the SEM micrographs, it
is observed that the granular worm-like textures grow bigger in size with increased surface
roughness as a result of annealing (Fig. 6). The possible reasons for increase in the grain size
and the surface roughness may be due to atomic migration in the film towards the lower
surface energy with annealing temperature (Kar et al., 2009).

571

Fig. 6. SEM micrograph of AlN films annealed at (a) as-deposited, and (b) 800 C
4.3 Effect of annealing on electrical properties
In both types of annealing, the Q
in
is increased with temperature. The probable reasons for
the rise in the Q
in
may be due to the generation of trap centres with annealing in the
nitrogen ambient. In RTA, the D
it
is found to be strongly dependent on the annealing
temperature and it significantly reduced

at 600C. With furnace annealing, the D
it
decreases
with increase in annealing temperature.
5. Growth of AlN films on different substrates
Many a times, AlN films are made on an insulator (SiO
2
) for isolation or it is deposited over
the metallic electrodes for thin film resonators (TFR). In future, AlN film on high speed
semiconductor substrates such as GaAs, InP can be exploited for high speed signal
processing and Micro-Opto-Electro-Mechanical Systems (MOEMS) applications. Hence,
integration of AlN films on GaAs and InP substrates for a new generation of high-speed
devices/subsystems, especially for telecommunications, and radar applications are
required. Growth and surface morphology of a deposited film depends not only on the
kinetics of the arriving species at the substrate, but also on the nature of the substrates
chosen, even if they belong to the same family. In addition, substrate orientation, thermal
conductivity and thermal expansion coefficients play vital roles in film growth and its
morphology. C-axis oriented AlN films are deposited on Si and SiO
2
/Si substrates by RF
reactive magnetron sputtering, where the degree of orientation decreases with increase in
oxide thickness. The surface roughness of the films deposited on SiO
2
/Si is higher. AlN
films are also deposited on GaAs and InP substrates by reactive magnetron sputtering
technique under identical deposition conditions. c-axis (002) oriented films are observed on
GaAs substrates; whereas, AlN (100), (002) and (102) oriented peaks are seen in case of InP
substrates. Surface morphology of the films deposited on Si and InP substrates seems to be
similar, but the films on InP are little rougher with the development of nano-pores. AlN
films, grown on GaAs substrates, forms bump like structures (Kar et al., 2009), which may
be due to thermal and/or lattice mismatch. It is important to note that the crystallinity and
stochiometry of the initial layer of AlN film also plays a significant role in the creation of
defects and mismatches (Ahmed et al., 1992). Crystal orientation of AlN films is also a strong
function of the bottom metal electrodes. AlN films deposited on metals (Al, Cu, Cr, Au) are c-
axis oriented, whereas the films deposited on Al and Cu are rough with larger grains.


572
6. AlN film based acoustic device
SAW and BAW are two import types of acoustic wave devices used in RF comunication.
These devices can be realized either on a solid substrate/film or through micromachined
suspended beam structures. In SAW devices, an elastic wave travels on the surface of a
piezoelectric material and displaces the atoms about their equilibrium positions at the
interface of piezoelectric film and solid substrate. The neighboring atoms at the interface
then produce restoring forces to bring the displaced atoms back to their original positions.
SAW can be generated by placing two inter digital transducers (IDT) either sides of the
substrate. These IDTs have alternating periodic fingers (Fig. 7). RF signal is applied one of
these alternating polarity fingers (IDT) that produces elastic mechanical wave in the
substrate. This wave travels along the substrate and also collected by placing another IDT on
the piezoelectric material, some distance away from the first IDT. The second IDT collects
the RF signal, which can be retransformed into the electric signal. As the elastic-mechanical
wave has the speed of acoustic wave, it introduces the delay of signal by 10
3
order. This is
the prime use of SAW device. The periodicity p, (centre-to-centre spacing between
neighbouring IDT fingers of same polarity) becomes the wavelength of the acoustic wave,
and dictates its frequency f = v/p, where f and v represent acoustic wave frequency and
acoustic propagation velocity, respectively.

Fig. 7. Schematic diagram of SAW device
Thin Film Bulk Acoustic Resonator (FBAR) device consisting of a piezoelectric material
sandwiched between two electrodes and is acoustically de-coupled from the surrounding
medium. FBAR devices, using AlN piezoelectric with thickness ranging from tenth of
micrometers to several micrometers, resonate in the cellular bands of cell phones and other
wireless applications. On applying voltage across the electrodes, the piezoelectric thin film
undergoes a shear deformation, and a BAW resonance occurs in the AlN film due to
coherent reflection at the top and bottom boundaries of the metal film or plate electrodes.
The frequency of resonance is dependent on the physical structures; hence, desired resonant
frequency can be obtained by tailoring physical dimension of the structure. For RF
frequency, physical dimension of resonators can be realized by using MEMS technology.

573
MEMS resonators are comprised of a microscale mechanical element, which converts
mechanical to electrical signal and vice versa. One of the prominent resonator structures is
MEMS cantilever, which is based on thin piezoelectric films. Film resonates when an ac
voltage is applied across the film. Resonator can be made without piezoelectric material
(electrostatic, capacitive resonator), but it suffers with large resistance, in the range of M,
and depends on driving voltage. On the other hand, piezoelectric resonators have smaller
resistance of the order of K and are more suitable for UHF device applications. (Lakin,
1999; Quandt et al., 2000; Humad et al., 2003). In addition, the output is easier to sense in a
piezoelectric resonator. Furthermore, a piezoelectric resonator has certain advantages over
the electrostatic resonator (capacitive resonators), such as low current consumption and
lower actuation voltages (Olivares et al.; 2005). But the quality factor (Q) of piezoelectric
resonator is smaller than that of a capacitive resonator. The quality factor of any resonator is
proportional to the decay time, and is inversely proportional to the bandwidth around
resonance. Higher Q represents higher frequency stability and accuracy capability of the
resonator (De Los Santos, 1999).

Fig. 8. XRD pattern, and AFM image (inset) of sputtered AlN film for SAW
6.1 Evaluation of AlN films through SAW devices
Higher RF power (400 W) and nitrogen concentration (80%), moderate substrate
temperature (200 C) and sputtering pressure (610
-3
mbar), lower target-substrate
distance (5 cm) is suitable for the growth of smooth, highly c-axis oriented AlN film with
better electrical properties. A c-axis (002) oriented peak is recorded at 2 value of 36.1
(Fig. 8). The atomic force micrograph of the film shows dense microstructure with
continuous grain growth (inset of Fig. 8). This kind of film is suitable for SAW devices. In
a typical case, each IDT consisted of 25 pairs of fingers/electrodes with 30 m centre-to-
centre spacing between the two neighbouring fingers comprising a pair (p/2). The width
of each finger/electrode is designed to be 15 m (p/4) with each of 6.0 mm length and 5.0
mm overlap, producing a SAW filter with an acoustic wavelength of 60 m (Kar et al.,
2009). The SAW device parameters are: AlN film thickness = 0.92 m, acoustic wavelength
= 60 m, SAW velocity = 5058 m/sec, electromechanical coupling coefficient (K
2
) = 0.34%.
X 0.500 m/div
Z 55.000 nm/div


574
Time response, due to acoustic frequency alone, is found after the gating out the response
due to electromagnetic feed through. For the centre to centre distance between the two
IDTs (d) = 7 mm, the main SAW signal centred at time (delay) = d/ V
SAW
= 7/5058 = 1.384
s is obtained. The central acoustic frequency (f
0
)

response after the gating out is observed
at 84.304 MHz (Fig. 9).

Fig. 9. (a) Optical image of AlN based SAW devices, and (b) Measured response of the SAW
device
6.2 Fabrication of AlN film based MEMS
Anisotropic etching of silicon is a key technology for fabrication of various three-
dimensional structures such as thin membranes and microbridges for MEMS. Generally,
anisotropic silicon etching is done by potassium hydroxide (KOH) or ethylenediamine
pyrochatechol (EDP) etchant (French, 2001; Ni et al., 2005). Another technique, which is
more versatile, CMOS process compatible and nontoxic, provides better selective etching
using doped tetramethylammonium hydroxide (TMAH) etchant (Biswas et al., 2006). The
characteristics of these three etchants are listed in Table 2. Many AlN and Al based MEMS
structures are also isolated from silicon by silicon dioxide. Hence, their selective etching is
very important. Diluted tetramethylammonium hydroxide (TMAH, 5 wt %), doped with
silicic acid (30.5 g/l) and ammonium persulphate (5.5 g/l), is suitable for CMOS silicon
microprocessing. To protect Al and AlN, silicic acid has been chosen instead of pure silicon
powder, because silicic acid dissolves quickly in TMAH solution. Ammonium persulphate
(AP) is also added to the above-mentioned solution to reduce the surface roughness of
etched silicon. The etch rate of silicon in doped TMAH is found to be 50 m/hour. During
silicon etching the Al, AlN and SiO
2
films are used as mask layers (Fig. 10). Low etch rates of
Al and AlN (18-30 nm/hour) as well as SiO
2
(2.5 nm/hour) are found to be suitable for
MEMS applications. Probable reason for low etch rate of Al may be the formation of a
passivating layer during TMAH etching (Fujitsuka et al., 2004). Dilute TMAH is a well
known etchant for AlN film (Kim et al., 2004) and Al as well. But doped TMAH shows
significantly lower etch rates for AlN and Al, which are exploited for AlN based suspended
microstructures. Fig. 10 (d) depicts suspended Cr/AlN/Cr/SiO
2
cantilevers fixed at one
end, where in one of the microstructures is lifted up because of the stress (Kar et al., 2009).

575
Property KOH EDP TMAH
Si etch rate (100), m/h 150 30-35 40-60
Etch quality high high medium
Selectivity (111)/(100) 1:30-100 1:20 1:10-50
Under-etch rate 0.5-1.5 1.4-1.5 0.2-1.7
CMOS compatible no yes yes
Selectivity PECVD SiO
2
/Si 1:100-300 1:10,000 1:100-1000
Selectivity PECVD SIN/Si 1:10,000 ---------- 1:150-200
Attack of aluminum high medium low with Si
Etch stop boron dope boron dope boron dope
Toxicity low high low
Long-term stability high low medium
Cost low high medium
Table 2. Characteristics of important wet etchants used for silicon micromachining (French,
2001)

Fig. 10. Micrographs of etched silicon (a) AlN/Si, (b) Al/Si, (c) SiO
2
/Si, (d) AlN based
suspended microstructures
Al
Cavity depth = 125 m
(b)
AlN
Cavity depth = 75 m
(a)
SiO
2

Cavity depth = 150 m
(c)
(d)


576
7. Conclusion
This chapter focuses on the study of RF sputtered AlN films in view of ICs, acoustic devices
and MEMS applications. It is divided into two distinct parts; growth, characterization and
optimization of device worthy AlN film (mostly on silicon), and demonstration of acoustic
and MEMS device applications. The morphological and electrical properties of RF sputtered
AlN films are studied with sputtering power, deposition temperature, sputtering pressure,
gas flow ratio and target to substrate spacing (D
ts
). Higher RF power (400 W) and nitrogen
concentration (80%), moderate substrate temperature (200 C) and sputtering pressure
(610
-3
mbar), lower target-substrate distance (5 cm) is suitable for the growth of smooth,
highly c-axis oriented film with better electrical properties. Post-deposition (RTA and
furnace) annealing has a significant impact on the morphology as well as the electrical
properties. The RTA processed AlN films have relatively high c-axis (002) orientation films
at 800 C, where microcracks are appeared during RTA process at 1000 C. Bulk charge
density is increased with annealing temperature both types of annealing. A significant
reduction in interface charge density is found at 600C with RTA process, whereas it
decreases with furnace annealing temperature. AlN films are deposited and characterized
on different substrates by RF reactive magnetron sputtering. On SiO
2
/Si substrates, (002)
orientation is deteriorated and surface roughness of the films is increased with the increase
in oxide thickness. c-axis (002) oriented films are observed on Si and GaAs substrates,
whereas AlN (100), (002) and (102) oriented peaks are seen on InP substrates. AlN films,
deposited on GaAs substrate, show bump like structures. c-axis oriented AlN films are also
observed on metallic films. The AlN films, deposited on Al and Cu, are found to be rough
with larger grains. Piezoelectric nature of RF deposited AlN films is ascertained from the
performance of a SAW device. This device is centred around a frequency of 84.3 MHz and
acoustic phase velocity is inferred to be 5058 m/ sec with K
2
of 0.34 %. The TMAH solution,
doped with an appropriate ratio of silicic acid and ammonium persulphate, is developed for
micromachining of AlN based structures. The etch rate of silicon is around 50 m/hour. On
the otherhand, the doped solution has negligible impact on Al, AlN and SiO
2
films. The
growth of highly (002) oriented AlN films, post deposition process and micromachining
method will provide an appropriate platform for the fabrication of futuristic electronic
devices.
8. References
Ahmed, A.U., Rys, A., Singh, N., Edgar, J.H. & Yu, Z.J. (1992). The Electrical and
Compositional Properties of AlN-Si Interfaces, J. Electrochem Soc, Vol. 139(4), pp.
1146-1151.
Assouar, M.B., Hakiki, M.E., Elmazria, O., Alnot, P. & Tiusan, C. (2004). Synthesis and
microstructural characterisation of reactive RF magnetron sputtering AlN films for
surface acoustic wave filters, Diamond and Related Materials, Vol. 13, pp. 1111-1115.
Belyanin, A.F., Boulov, L.L., Zhirnov, V.V., Kamenev, A.I., Kovalskij, K.A. & Spitsyn, B.V.
(1999). Application of aluminum nitride films for electronic devices, Diamond and
Related Materials, Vol. 8, pp. 369-372.
Bender, S., Dickert, F.L., Mokwa, W. & Pachatz, P. (2003). Investigations on temperature
controlled monolithic integrated surface acoustic wave (SAW) gas sensors, Sensors
and Actuators B, Vol. 93, pp. 164-168.

577
Biswas, K., Das, S., Maurya, D. K., Kal, S. & Lahiri S.K. (2006). Bulk micromachining of
silicon in TMAH-based etchants for aluminum passivation and smooth surface,
Microelectronics Journal, Vol. 37(4), pp. 321-327.
Cheng, H., Sun, Y. & Hing, P. (2003). The influence of deposition conditions on structure
and morphology of aluminum nitride films deposited by radio frequency reactive
sputtering, Thin Solid Films, Vol. 434, pp. 112-120.
De Los Santos, H.J. (1999). Introduction to Microelectromechanical (MEM) Microwave
Systems, Artech House, pp. 83.
French, P.J. (2001). Integration of silicon MEMS devices: Materials and processing
considerations, Smart Materials Bulletin, January, pp. 7-13.
Fujitsuka, N., Hamaguchi, K., Funabashi, H., Kawasaki, E. & Fukada, T. (2004). Silicon
anisotropic etching without attacking aluminum with Si and oxidizing agent
dissolved in TMAH solution, Sensors and Actuators A, Vol. 114, pp. 510-515.
Humad, S., Abdolvand, R., Ho, G.K. & Ayazi F. (2003). Micromechanical Piezo-on-Silicon
Block Resonators, in Proc. IEEE International Electron Devices Meeting (IEDM03),
Washington DC, Dec., pp. 957-960.
Kar, J.P., Mukherjee, S., Bose, G., Tuli, S. & Myoung, J.M. (2009), Impact of post-deposition
annealing on the surface, bulk and interface properties of RF sputtered AlN films,
Materials Science and Technology, Vol. 25, pp. 1023-1027.
Kar, J.P., Bose, G., Tuli, S., Myoung, J.M. & Mukherjee, S. (2009). Morphological
investigation of AlN films on various substrates for MEMS applications, Surface
Engineering, Vol. 25, pp. 526-530.
Kar, J.P., Bose, G., Tuli, S., Dangwal, A. & Mukherjee, S. (2009). Growth of AlN films and its
process development for the fabrication of acoustic devices and micromachined
structures, Journal of Materials Engineering and Performance, Vol. 18, pp. 1046-1051.
Kar, J.P., Mukherjee, S., Bose, G. & Tuli, S. (2008). Effect of inter-electrode spacing on
structural and electrical properties of AlN films, Journal of Materials Science:
Materials in Electronics, Vol. 19, pp. 261-265.
Kar, J.P. (2007). Growth and characterization of aluminum nitride (AlN) films for electroacoustic
and MEMS applications, Ph. D. thesis, Indian Institute of Technology Delhi, India
Kar, J.P., Bose, G. & Tuli, S. (2006). A study on the interface and bulk charge density of AlN
films with sputtering pressure, Vacuum, Vol. 81, pp. 494-498.
Kar, J.P., Bose, G. & Tuli, S. (2006). Correlation of electrical and morphological properties of
sputtered aluminum nitride films with deposition temperature, Current Applied
Physics, Vol. 6, pp. 873-876.
Kar, J.P., Bose, G. & Tuli, S. (2006). Influence of nitrogen concentration on grain growth,
structural and electrical properties of sputtered aluminum nitride films, Scripta
Materialia, Vol. 54, pp. 1755-1759.
Kar, J.P., Bose, G. & Tuli, S. (2005). Influence of rapid thermal annealing on morphological
and electrical properties of RF sputtered AlN films, , Materials Science in
Semiconductor Processing, Vol. 8, pp. 646-651.
Kim, H.H., Ju, B.K., Lee, Y.H., Lee S.H., Lee J.K. & Kim, S.W. (2004). Fabrication of
suspended thin film resonator for application of RF bandpass filter, Microelectronics
Reliability, Vol. 44, pp. 237-243.
Lakin, K. (1999). Thin film resonators and filters, IEEE Ultrasonics Symposium, pp. 895-906.


578
Lee, H.C., Park, J.Y., Lee, K.H. & Bu, J.U. (2004). Preparation of highly textured Mo and AlN
films using a Ti seed layer for integrated high-Q film bulk acoustic resonators, J.
Vac. Sci. Technol. B, Vol. 22(3), pp. 1127-1133.
Lee, S.H., Yoon, K.H., Cheong, D. S. & Lee, J.K. (2003). Relationship between residual stress
and structural properties of AlN films deposited by r.f. reactive sputtering, Thin
Solid Films, Vol. 435, pp. 193-198.
Loebl, H.P., Klee, M., Metzmacher, C., Brand, W., Milsom, R. & Lok, P. (2003). Piezoelectric
thin AlN films for bulk acoustic wave (BAW) resonators, Materials Chemistry and
Physics, Vol. 79, pp. 143-146.
Luo, J.K., Lin, M., Fu, Y.Q., Wang, L., Flewitt, A.J., Spearing, S.M., Fleck, N.A. & Milne, W.I.
(2006).MEMS based digital variable capacitors with a high-k dielectric insulator,
Sensors and Actuators A, Vol. 132 (1), pp. 139-146.
Naik, R.S., Reif, R., Lutsky J.J. & Sodini, C.G. (1999). Low-Temperature Deposition of Highly
Textured Aluminum Nitride by Direct Current Magnetron Sputtering for
Applications in Thin-Film Resonators, J. the Electrochemical Society, Vol. 146(2), pp.
691-696.
Ni, H., Lee H.J. & Ramirez, A.G. (2005). A robust two-step etching process for large-scale
microfabricated SiO
2
and Si
3
N
4
MEMS membranes, Sensors and Actuators A, Vol.
119, pp. 553-558.
Olivares, J., Iborra E., Clement M., Vergara, L., Sangrador, J. & Sanz-Hervas, A. (2005).
Piezoelectric actuation of microbridges using AlN, Sensors and Actuators A, Vol.
123124, pp. 590-595.
Quandt, E. & Ludwig, A. (2000). Magnetostrictive actuation in microsystems, Sensors and
Actuators A, Vol. 81, pp. 275-280.
Schreiter, M., Gabl, R., Pitzer, D., Primig, R. & Wersing, W. (2004). Electro-acoustic
hysteresis behaviour of PZT thin film bulk acoustic resonators, J. the European
Ceramic Society, Vol. 24, pp. 1589-1592.
Strite, S. & Morko, H. (1992). GaN, AlN and InN: A review, J. Vac. Sci. Technol. B, Vol. 10(4),
pp. 1237-1266.
Wang, H.H. (2000). Properties and preparation of AlN thin films by reactive laser ablation
with nitrogen discharge, Modern Physics Letters B, Vol. 14, 523-530.
Wang, X.D., Jiang, W., Norton, M.G. & Hipps, K.W. (1994). Morphology and orientation of
nanocrystalline AlN thin films, Thin Solid Films, Vol. 251(2), pp. 121-126.
Xu, X.H., Wu, H.S., Zhang, C.J. & Jin Z.H. (2001). Morphological properties of AlN
piezoelectric thin films deposited by DC reactive magnetron sputtering, Thin Solid
Films, Vol. 388, pp. 62-67.
Yamada, H., Ushimi, Y., Takeuchi, M., Yoshino, Y., Makino, T. & Arai, S. (2004).
Improvement of crystallinity of ZnO thin film and electrical characteristics of film
bulk acoustic wave resonator by using Pt buffer layer, Vacuum, Vol. 74, pp. 689-692.
Yim, W.M., Stofko, E.J., Zanzucchi, P.J., Pankov, J.I., Ettenberg, M. & Gilbert, S.L. (1973).
Epitaxially grown AlN and its optical band gap, J. Appl. Phys., Vol. 44(1), pp. 292-
296.
26
Surface Acoustic Wave Devices
for Harsh Environment
Cinzia Caliendo
Istituto dei Sistemi Complessi, ISC-CNR, Area della Ricerca Roma 2, Rome
Italy
1. Introduction
There is an increasing demand of electronic components for aerospace, aircraft industries,
sensors, automotive, chemical and material processing applications, to name just a few, able
to operate reliably and for long time at high-temperature. Measurements reliability requires
the electronic components to be placed directly inside the extreme environment, and to
withstand temperatures of several centigrade degrees with lifetimes of several hours. The
device mounting and packaging, but first of all the device materials must be stable with the
working temperature, otherwise temperature-induced stress may result in devices failures.
Electroacoustic devices based on surface and bulk acoustic wave (SAW and BAW)
technology must satisfy the requirements of low cost, high frequency, high-Q, low loss,
large piezoelectric coupling and zero temperature coefficient of delay (TCD) to be key
devices in the communication and sensor fields. The temperature stability of the
piezoelectric crystal is an essential characteristic because of its direct link with the
temperature sensitivity of the electroacoustic device operation frequency. The high
operation frequency is an essential characteristic for SAW and BAW devices to be used in
mobile phones, cordless headphones, alarm and security systems, military equipment,
sensors, etc. The temperature stability and the high operation frequency demands can be
met through a proper choice of the piezoelectric substrate crystal cut, new piezoelectric
materials and/or multilayer configurations. The use of temperature stable cuts of single
crystal bulk piezoelectric materials or temperature compensated multilayers represents two
possible solutions to the temperature stability requirement. The use of high-resolution
lithography techniques and/or of high SAW velocity materials is required in order to
extend the upper limit of the electroacoustic device frequency range. Submicron feature
sized interdigital transducers (IDTs) are required to implement GHz range SAW devices on
slow piezoelectric materials, while micron feature sized IDTs can still be used on fast
materials, since the SAW device centre frequency, f = v/, depends on both the phase
velocity of the propagating medium, v, and on the acoustic wavelength , being the IDTs
period p = /2. Conventional piezoelectric substrates, such as quartz, lithium niobate
(LiNbO
3
), and lithium tantalate (LiTa0
3
) crystals, cannot be used above 500C. Quartz ST cut
is a temperature stable material but it shows an alpha-beta transition at 573C, which causes
the loss of piezoelectricity, and results in a non-operable device. SAW devices implemented
on LiNb0
3
have been studied for a temporary usage at 400C [1]; however the LiNbO
3

acoustic wave properties are highly dependent on temperature since it is a pyroelectric


580
material and has a TCD as high as 75 ppm/C. LiTaO
3
shows properties, such as a low
Curie temperature (607C), sensitivity to temperature variations (TCD = 22 ppm/C for the
X-112Y cut) and a strong pyroelectricity, which limit its operation at elevated temperatures.
Piezoelectric bulk single crystals such as GaPO
4
, LGS (La
5
Ga
3
SiO
14
) and its isomorphs
(called LGX family group) substrates are widely investigated for the realization of SAW-
based devices able to work at high temperature. LGS belongs to the trigonal class 32 group
as quartz but it has no - transitions and can operate up to its melting temperature of
1470. It shows zero or very low TCD cuts with zero power flow angle and higher
electromechanical coupling coefficient than that of quartz [2]. Langasite based SAW devices
are not suitable for operation in the GHz range as a consequence of their low phase velocity
and high acoustic losses (from 1 to 0.01 dB/wavelength [3]). GaPO
4
has twice the sensitivity
of quartz and many its physical constants are stable up to about 900C, but the accessible
frequencies are limited to values of 1 GHz as a consequence of quite high acoustic losses.
The technology of thin piezoelectric films (such as AlN) offers the opportunity of combining
the properties of the substrate with those of the film: thus a composite arrangement of fast
materials with opposite sign TCDs and a proper design of the electroacoustic configuration
enable achieving a thermally stable SAW device operating in the GHz range.
Aluminium nitride (AlN) is a piezoelectric material that shows interesting properties, such
as excellent thermal conductivity (180W/mK), low coefficient of thermal expansion (CTE,
4.1 x 10
-6
C
-1
), and good resistance to thermal shock and caustic chemicals [4], that make it
useful as protective coating and guarantee the stability of the AlN-based devices when they
are in contact with extreme environments. It is currently being investigated due to its
promising potentialities for high-temperature, high-power, and high-frequency electronics.
It has demonstrated to be an ideal candidate for packaging SiC-devices for high-temperature
applications [5] thanks to its CTE that closely matches those of Si (3.5 x 10
-6
C
-1
) and SiC (3.7
x 10
-6
C
-1
), high electrical resistivity, high mechanical strength, and chemical inertness.
Reactively sputtered AlN films have been used as an effective encapsulant for GaN [6] at an
annealing temperature of 1100C substituting the standard dielectric encapsulants, such as
SiO
2
and Si
3
N
4
, that are not viable at so high temperatures. AlN maintains its piezoelectricity
up to 1200C in vacuum and shows very high BAW and SAW velocities (6000 m/s and
11300 m/s for transversal and longitudinal BAWs propagating along the z direction, 5607
m/s for SAWs propagating in the z plane) that make it the ideal candidate for microwave
electroacoustic devices implementation. Furthermore, AlN can be grown in thin film form
onto non piezoelectric substrates by techniques as simple as the rf reactive sputtering. Both
the structural properties of the substrate and the experimental sputtering parameters (such
as the reactive gas flow rate, the partial pressure of reactive and inert gasses, the substrate
temperature, the rf power, and the substrate-target distance) affect the morphological and
structural properties of the sputtered thin films. The requirements for a suitable substrate
include also a thermal coefficient-of-expansion compatible with that of the film, high-
temperature stability, machinability, good adherence of the AlN film: among the available
substrates, silicon, platinum and sapphire satisfy these requirements. Al
2
O
3
, substrates have
a wide range of industrial applications as structural ceramic and optical materials. Al
2
O
3
is
extensively used as a high temperature, corrosion resistant refractory material due to its
hardness, chemical durability, abrasive resistance, mechanical strength, and good electrical
insulation. Al
2
O
3
shows good thermal conductivity (24 W/mK), high SAW velocity (in the
range 5555 to 5706 m/s in the c-plane), positive TCD (70 ppm/C) and a CTE that closely
matches that of AlN. The AlN/Al
2
O
3
based multilayers can withstand temperatures up to

Surface Acoustic Wave Devices for Harsh Environment

581
900C, thus allowing the realization of high frequency, temperature compensated dispersive
electroacoustic devices for high temperature applications. Electroacoustic devices
implemented on Si substrates offer the opportunity to integrate the device with the
surrounding electronic circuitry on the same chip. Moreover, the opposite TCD of Si ( 30
ppm/C) and AlN ( -30 ppm/C) allows the realisation of zero-temperature-coefficient
acoustic devices at the proper film thickness to be used as sensors and actuators where low
loss, low thermal drift, high sensitivity and high signal-to-noise ratio are demanded [7, 8]. Pt
is the material of choice for metallic components that have to withstand oxidation, thank to
its high temperature coefficient of resistance: Pt can be grown in thin film form and both the
IDTs and ground electrodes can be easily defined by lift off technique.
In the present chapter the sustainability of Pt and AlN films on sapphire and Si substrates
for high temperature applications is assessed.
2. AlN-based SAW devices
Bulk piezoelectric single crystals, such as LGS and GaPO
4
,

can be used for the
implementation of non dispersive SAW devices, such as delay lines, filters and resonators,
and the SAW propagation characteristics, such as phase velocity, electroacoustic coupling
efficiency K
2
and TCD, depend on the crystal cut and SAW propagation direction, as well as
on the geometry of the IDTs . The SAW propagation is excited by IDTs located at the free
surface of the piezoelectric substrate and directly exposed to the surrounding environment,
as shown in figure 1.

Fig. 1. SAW delay line on a piezoelectric substrate
AlN can be grown in thin film form onto non piezoelectric substrates, such as silicon or
sapphire, thus allowing the realization of dispersive electroacoustic devices. Moreover, if the
AlN film is sandwiched between the IDTs and the ground electrode, four piezoelectric
coupling configurations can be obtained by placing the IDTs at the substrate/film interface
or at the film surface, with and without the floating electrode opposite the IDTs. These four
structures will be mentioned hereafter as substrate/film/IDT (SFT),
substrate/IDT/film/metal (STFM), substrate/IDT/film (STF) and substrate/metal/film/IDT
(SMFT), respectively. Figure 2a shows the top view of a dispersive SAW delay line, and figure
2b shows the cross sections of the four coupling configurations.
When the IDTs are located at the substrate/film interface, the piezoelectric film plays the
role of both the acoustic wave transductor and protective layer of the underlying IDTs.
IDTs
Piezoelectric substrate


582
These four configurations show frequency dispersive SAW propagation characteristics, that
are no longer solely determined by the geometry of the IDTs, the crystals cut, and the SAW
propagation direction, but also by the film thickness and the electrical boundary conditions.
For SAW propagating along layered structures, the achievable K
2
value is sometimes larger
than that of the individual piezoelectric materials; it

is frequency dispersive and depends on
the type and orientation of the piezoelectric material, and it is drastically affected by the
location of the IDTs and counter electrode with respect to the piezolectric layer. As an
example, figures 3a and 3b show the K
2
vs film thickness to wavelength ratio, h/, for SAW
propagation along zx- and zy-Al
2
O
3
/AlN for the four coupling configurations: the highest
K
2
values obtainable are about 0.50 and 0.67% for STF x and y propagation (at h/ 0.65
and 0.60), 0.49 and 0.64 % for STFM x and y propagation (at h/ 0.67 and 0.62), being 0.3%
the AlN K
2
value.

Fig. 2a. Dispersive SAW delay line

Fig. 2b. The four electroacoustic coupling configurations
SFT
STF
SMFT
STFM


583

Fig. 3a. The K
2
vs h/ for SAW propagating along zx-Al
2
O
3
/AlN for the four coupling
configurations

Fig. 3b. K
2
vs h/ for SAW propagating along zy-Al
2
O
3
/AlN for the four coupling
structures
Figures 4a and 4b show the K
2
vs h/ for SAW propagation along Si/Pt/AlN and
Si/AlN/Pt, being the Pt thickness the running parameter.


584

Fig. 4a. The K
2
vs AlN h/ for SAW propagation along Si/Pt/AlN, for different Pt thickness
values normalized to the acoustic wavelength

Fig. 4b. The K
2
vs AlN h/ for SAW propagation along Si/AlN/Pt, for different Pt thickness
values normalized to the acoustic wavelength
In figure 4a the highest K
2
values obtainable are in the range 0.5 to 0.63% at h/ = 0.55 for
Pt h/ = 10
-8
to 10
-1
. In figure 4b the highest K
2
values obtainable are in the range 0.5 to


585
0.52% at h/ = 0.555 for Pt h/ = 10
-8
to 10
-3
. The theoretical data shown in figures 3 and 4
have been evaluated using the PC SAW software developed by Mc Gill University [9]. The
theoretical K
2
has been approximated as
f m
ph ph
f
ph
v v
2
v
where
f
ph
v and
m
ph
v

are the SAW
phase velocities along the free and electrically short-circuited surfaces of the AlN film. The
phase velocity
m
ph
v is obtained by the insertion of a perfectly conductive and infinitesimally
thin film at the interfaces where the IDTs and the ground plane are located in each of the
four coupling structure. The physical data relative to the elastic, piezoelectric and dielectric
constants of AlN film are extracted from [10] and [11] and refer to single crystal AlN thin
films grown on the basal plane of Al
2
O
3
(0001) by metalorganic vapor deposition. The TCD
of the bulk piezoelectric crystal depends only on the crystal cut and the SAW propagation
direction, while that of a layered structures is frequency dispersive. Because the SAW
penetration depth inside the propagating medium is about one wavelength, in a layered
medium, for h/ << 1, the most of the SAW energy is confined to the substrate, while, as h
increases with respect to , more and more of the SAW energy is confined to the film. For
small AlN film thickness (h/ < 1) the TCD value of the multilayer corresponds
approximately to that of the substrate. With increasing the AlN film thickness respect to the
acoustic wavelength (h/ 1) the TCD reaches the AlN TCD value. If the film and the
substrate show opposite sign TCD values, there will be a h/ value at which TCD = 0
ppm/C (the temperature compensated point, TCP). This h/ value represents the film
thickness for which the two opposite sign TCDs of the film and of the substrate equilibrate
to form a thermally compensated structure. Thus high-frequency, enhanced coupling, and
thermally compensated elctroacoustic devices can be designed at the proper AlN films
thickness values [12].
3. Materials and methods
Highly c-axis oriented AlN films were grown at 180C and at 200C by rf reactive sputtering
technique on the polished surface of (0001) oriented single crystal Al
2
O
3
substrate, on
SiO
2
/Si(100) and Pt/SiO
2
/Si(100) substrates. The AlN deposition process parameters were
the following: gas atmosphere of 100% of N
2
, high purity (99.999%) 4 diameter Al target
disc, RF power 200 watt, background vacuum 510
-8
Torr and pressure during the deposition
process 310
-3
Torr. Before starting the sputtering process, a 30 minute pre-sputtering was
performed. The substrate temperature was held at 180 C during the deposition process. The
optimized sputtering parameters ensure AlN films showing a high adhesion to the
substrate, c-axis orientation, a columnar growth, smooth surface, and high piezoelectricity;
the films are also uniform, stress-free and extremely adhesive to the substrates. The Pt
sputtering process parameters were the following: substrate temperature 200C, gas
atmosphere of 100% of Ar, high purity (99.99%) 4 diameter Pt target disc, RF power 150
watt, background vacuum 10
-7
Torr and pressure during the deposition process 510
-3
Torr.
The deposition process of both Pt and AlN films is performed subsequently without
breaking the vacuum in order to avoid any oxidation effects of the layers. Then the obtained
samples were heated at 900C in air at ambient pressure by a quartz tube furnace, for
different lengths of time. The cold (20C) sample was abruptly put inside the furnace pre-
heated at 900C and the annealing time was measured from the set temperature was
reached; then the sample was removed from the furnace and brought abruptly to room


586
temperature. A temperature ramp of 1C/s was measured by a thermocouple after the
insertion of the cold sample inside the furnace. The furnace tube was not hermetically sealed,
so ambient air was present during the loading and unloading of samples [14]. The structural
properties of annealed Pt and AlN films were investigated by by X-ray diffraction
measurements (XRD) a Seifert XRD 3003P performed on a a Seifert XRD 3003P
diffractometer operating in the Bragg-Brentano geometry using Cu-K radiation ( = 1.5418
) and the diffracted intensities were collected in -2 scan mode in the range 20 <2 < 80
with step size 0.04. The reflection peaks of the diffractograms were compared with the
standards of the JPCDS database. The crystallite size D was calculated from the Scherrer
formula
0.9
D
B cos
where is the wavelength, B the -2 full width at half maximum

(FWHM) of the (0002) peak in rad and the Bragg angle. The c and a lattice parameters of
the AlN and Pt films were calculated from the angular position of the AlN (002) and Pt (111)
diffraction peaks of the -2 scan and compared to the value from ref. 13 for Pt and powder
AlN (a = 3.9231, c = 4.979). Since the electrical resistivity of the thin conducting films
influences the device characteristics (such as insertion loss and Joule heating), the electrical
resistance and the surface morphology of the outer Pt electrode were investigated at room
temperature after each thermal annealing. The annealing effects on the piezoelectric
constant d
33
of the AlN films were also estimated [14].
3.1 Pt/AlN/Pt/SiO
2
/Si
AlN films, 3.15 m thick, were sputtered on bare and Pt (2200 thick) -covered SiO
2
/Si(100)
substrates, being 2 m the silicon oxide thickness; a Pt film (2200 thick) was sputtered
on the AlN free surface and then the Pt/AlN/Pt/SiO
2
/Si multilayers were heated at 900C
in air for lengths of time ranging from 1 to 32 hours. Figure 5 shows the XRD patterns of the
as grown and annealed multilayers [14].

Fig. 5. XRD patterns of Pt/AlN/Pt/SiO
2
/Si structures: the running parameter represents the
annealing time


587
The piezoelectric AlN film is c axis oriented perpendicularly to the growth plane: in all the
samples the AlN (002) and (004) peaks, at 36 and at 76, are visible even after 32 hours
annealing. The peak at 2 40 corresponds to the Pt film strongly oriented along the (111)
plane; with increasing the annealing time, a very small stress-induced shift in the Pt (111)
peak position can be observed, while the same peak becomes narrower indicating a growing
(111) fiber texture. As a consequence of the temperature-induced improvement of the Pt
local epitaxy, the alignment precision of the AlN film crystalline planes also improves. The
XRD data of the outer Pt film showed two peaks at 2 40 and 46.3 corresponding to the
(111) and (200) orientations, and also a small platinum oxide (220) peak, at 66, clearly
visible after the 1
st
annealing, that does not increase in percentage after the successive
thermal cycles [14]. The AlN and Pt FWHM of the 2 scan before and after the
annealing [14] resulted improved, starting from 0.391 and 0.415 of the as grown AlN and
Pt films, up to 0.24 and 0.28 after 32 hours annealing. The c and a lattice parameter of
AlN and Pt films of the as-grown samples, calculated from the angular position of the
(002) and (111) diffraction peaks, are respectively larger (4.997 ) and smaller (3.914)
than the values reported for bulk single crystal AlN and Pt [13], as a consequence of the
lattice mismatch between the two materials. Figures 6 and 7 show the FWHM of the
AlN(002) and Pt(111) peaks, and the c and a lattice parameters of AlN and Pt films vs the
annealing time.

Fig. 6. FWHM of the AlN(002) and Pt(111) peaks vs annealing time
After the first annealings the AlN c parameter relaxes to the bulk while the Pt a parameter
slightly decreases. Further annealings result in a permanent in-plane compressive stress for
both the AlN and Pt films. The AlN(002) and Pt(111) peaks FWHM decreases with
increasing the annealing time indicating a decrease of inhomogeneous strain distribution.


588

Fig. 7. c and a lattice parameters of AlN and Pt films vs the annealing time
The measurement of the longitudinal piezoelectric coefficient d
33f
of the AlN film was done
at ambient temperature on the same type multilayer without the outer Pt films, before and
after the thermal annealing, with a method described in ref. 15 and based on the direct
piezoelectric effect: a longitudinal acoustic wave perturbs the sample via a special probe and
the electrical voltage induced in the piezoelectric film is measured. The probe consisted of a
metal rod in contact with a Pb(Zr,Ti)O
3
(PZT)-based low frequency transducer that was
connected to a pulse generator (pulse width 0.1-1.0 ns) to produce longitudinal bursts
propagating along the metal rod. The contact between the rod and the piezoelectric film
surface resulted in the application of a stress on the surface of AlN films. Stress-induced
electrical charges were collected at the piezoelectric film surfaces by electrodes (the metal
rod and the conducting substrate) and observed on a scope. The piezoelectric strain
constant, d
33f
, of the tested films was evaluated comparing the film response with the
response of a thin single crystal reference sample, whose d
33
was known. All the tested films
showed to be piezoelectric with a difference in the d
33f
obtained values not appreciable with
this measurement technique because of an error of about 15-20 %. The estimated mean value
is in the range from 6.2 and 7.4 pC/N, for both the as grown and all the annealed samples:
these values well agree with the corresponding value reported in the available literature [16]
that is about 6 pC/N.
The sheet resistivity of the Pt layer deposited on the AlN/Pt/SiO
2
/Si multilayer free surface
was measured at ambient temperature by the four point method, before and after the
thermal annealing. It was observed that the sheet resistance values decreases with
increasing the annealing time, starting from 0.6 /sq, refereed to the unannealed samples,
to 0.5 /sq, referred to the 32 hours annealed samples [14]. Scanning electron microscopy
(SEM) investigations revealed an average grain size increased with increasing the annealing
lasting. The as deposited Pt films have small grain size (about 400 nm), as shown in figure 8,
and this high density of grain-boundary affects the high resistivity of the films.


589

Fig. 8. SEM photo of the as deposited Pt film

Fig. 9. SEM photo of the 32 hours annealed Pt film
Moreover, the as-grown films contain a number of structural defects that anneal out when
heat treatment is carried out and, as a result, the films resistivity decreases. The Pt
annealing results in the relaxation of intrinsic stresses, as well as in the redistribution of
structural imperfections: grains coalescence take place and the film sheet resistivity


590
decreases. SEM photo of the 32 hours annealed sample, shown in figure 9, demonstrates that
recrystallization of Pt surface occurs [14].
The unannealed outer Pt film was silvery while the annealed films were matt: this color
change can be explained with the increase of the Pt grain size. The adhesion strength of the
Pt films was good enough to pass a rudimentary tape test with a transparent tape after
each anneal as well.
3.2 c-AlN/(0001)-Al
2
O
3

Highly c-axis oriented AlN films were grown by rf reactive sputtering technique on the
polished surface of (0001) oriented single crystal Al
2
O
3
substrate. Then the as grown films,
0.26 to 4.7 m thick, were thermally annealed at 900C in air for 1 to 18 hours and the AlN
structural characteristics were evaluated after each thermal cycle. References are available in
the literature concerning the high thermal annealing (HTA) of AlN on Si performed at 700
1200 C for 2 to 12 hours in controlled atmosphere (in oxygen or nitrogen flux) or in high
vacuum; AlN films on Al
2
O
3
were heated at 950 C in air for 30 minutes [20] to 1 hour [21],
at 900 to 1200 C for 10 s in flowing N
2
[22], and at 800 C for 90 minutes in air [23]; bulk AlN
was annealed in oxygen at 900 1150 C for 6 hours [24]. In the present work, the HTA of
the AlN films on Si and on Al
2
O
3
substrates have been performed up to 32 and 18 hours,
respectively. No damage was observed on the surface of the AlN/Al
2
O
3
film even after 18
hours annealing: the AlN films were still clear, uniform, and extremely adhesive to the
substrate. The impact of the annealing on the films structural properties was investigated by
XRD before and after undergoing the thermal annealing. The D and c parameter of the AlN
films thermally annealed for different time periods, are listed in tab. 1.

AlN thickness
(m)
Time
(hours)
FWHM
-2
(deg)
D
()
c
()
4.7

0 0.384 218 4.995
4 0.281 297 4.977
8 0.289 288 4.978
12 0.290 290 4.978
4

0 0.270 540 4.976
1 0.269 548 5.024
3 0.264 555 5.026
5 0.262 593 5.027
9 0.261 611 5.037
18 - - -
1.5

0 0.289 252 4.978
4 0.332 288 4.984
8 0.350 281 4.989
10 0.410 303 5.001
0.26

0 0.500 220 4.986
3 0.422 198 5.048
4 - - -
Table 1. The AlN (002) FWHM, the D and c parameter of the AlN films thermally annealed
for different time periods


591
Figures 10, 11 and 12 show the diffraction patterns of the AlN films, 4.0, 1.5 and 4.7 m
thick, annealed for different time.

Fig. 10. XRD pattern of the AlN, 4 m thick, annealed for 1 (a), 3 (b), 5 (c), and 9 (d) hours

Fig. 11. XRD pattern of the AlN, 1.5 m thick, annealed for 4, 8 and 10 hours


592

Fig. 12. XRD pattern of the AlN, 4.7 m thick, annealed for 4, 8 and 12 hours
The AlN (0002) peak at approximately 2 = 36 of the as deposited films, 0.26 to 4.7 m
thick, showed a FWHM (from -2 scan) in the range 0.5 to 0.27. Both the as grown and
the annealed samples exhibit a strong peak at 2 42 due to the (0006) reflection of the
Al
2
O
3
substrate, and one small peak at 76, corresponding to the (0004) reflections of the
wurtzite AlN structure, not shown in figures 10 to 12. The 2
nd
order peaks of AlN(0002) and
Al
2
O
3
(00001) are clearly visible at 32.4 and 37.5. After 4 hours at 900C in air, the (0002)
peak of the thin AlN film (0.26 m thick) broadens completely as well as that of the 4 m
thick AlN film after 18 hours annealing. No other AlN phases are present, nor AlN-oxide
traces are evident in the spectra after the annealing; only a decrease in the 2 value of the
(0002) peak is observed, resulting in an increase in the c lattice parameter whose values are
larger than the bulk lattice parameter [13], indicating the presence of compressive stress in
the surface plane. This lattice elongation, perpendicular to the growth plane, is usually
associated with the intrinsic compressive stress caused by the lattice mismatch between the
AlN and Al
2
O
3
, (a
sapphire
-a
AlN
)/a
sapphire
, that is approximately equal to 30%. Our previous
results [25] have shown that the c-axis of the as grown AlN films on sapphire relaxes to the
bulk value with increasing the film thickness: an interface layer, strained because of the
large lattice mismatch, is formed on the sapphire surface and is followed by a columnar AlN
layer which runs through the entire thickness of the film. The thinner the AlN film and more
is strained and unable to survive to the HTA.
4. Conclusions
Highly c-axis oriented AlN films and thin film stacks of Pt/AlN/Pt were sputtered at 180C
on (0001)Al
2
O
3
substrates and at 200C on oxidized Si substrates. The multilayers were
heated at 900 C in air up to 32 hours to test their resistance to high temperature. The


593
structural investigation of the annealed films showed that the thermal annealing improved
the crystal quality of the AlN films sandwiched between the Pt films, as confirmed by the
decreased FWHM of the rocking curves, and the film piezoelectric d
33f
coefficient resulted
unaffected by the temperature. The study of the electrical, morphological and structural
characteristics of the Pt electrode revealed a dense surface with a hillock-free morphology,
confirming that Pt is the material of choice when a high oxidation resistance is required for
metallic components within devices operating at elevated temperatures. The AlN films on
sapphire show a lattice elongation, perpendicular to the growth plane, that is usually
associated with the compressive stress caused in the growth plane by the lattice mismatch
between the film and the substrate. After the first few hours annealing, the FWHM of the
(002) AlN peak decreases showing an improvement in the film texture; further annealing
results in the FWHM broadening, whose magnitude depends on the film thickness. The
structural investigations of thin (0.26 m) and thick (from 2 to 4.7 m) annealed AlN films
revealed that the film behaviour in harsh environment is strongly affected by the film
properties. Thick films, whose structure is more relaxed than the thin one, is able to survive
to high temperature without suffering significant deterioration for longer annealing times
than the thin one. The obtained results confirm that, during the HTA, the Pt film on the
substrate surface is protected by the AlN film, while the Pt film directly exposed to the
ambient conditions acts as a protective layer with respect to the AlN film; thus the AlN-
based STFM coupling configuration to be implemented on sapphire or silicon substrates is
an attractive alternative to langasite and GaPO
4
for the development of microwave
electroacoustic devices able to work at high temperatures.
5. Acknowledgements
The author wishes to thank Mr. P.M. Latino for his technical support in the development of
the technological processes and the HTA.
6. References
[1] Hauser, R. Reindl, L. Biniasch, J., Ultrasonics Symposium, Vol. 1, 192 (2003).
[2] M. Pereira da Cunha, M.P. Saulo de A. Fagundes, Ultrasonics Symposium, Vol. 1, 283
(1998).
[3] Maurcio Pereira da Cunha, Eric L. Adler and Donald C. Malocha, Ultrasonics
Symposium, Vol.1 169 (1999).
[4] Kar, J.P., Mukherjee, S., Bose, G., Tuli, S., Myoung, J.M., Materials Science and Technology,
25 1023 (2009).
[5] Zhigang Lin and Yoon, R.J., International Symposium on Advanced Packaging Materials:
Processes, Properties and Interfaces, 156 (2005).
[6] C. Zolper, D. J. Rieger, and A. G. Baca S. J. Pearton and J. W. Lee R. A. Stall, Appl. Phys.
Lett. 69, 538 (1996).
[7] S.M.Middelhoek, S.A.Audet, Silicon sensors, Academic Press, London 1989.
[8] M.J.Vellekoop, E.Nieuwkoop, J.C.Hsaartsen, A.Venema, Ultrasonics Symposium,Vol. 1,
375 (1981).
[9] E.L. Adler, G.W. Farnell, J. Slaboszewicz, C.K. Jen, Ultrasonics Symposium, Vol. 1, 103
(1982).
[10] K. Tsubouchi, K. Sugai, N. Mikoshiba, Ultrasonics Symposium, Vol. 1, 375 (1981).


594
[11] J.G. Gualtieri, J.A. Kosinski, A.Ballato, Ultrasonics Symposium, Vol. 1 403 (1992).
[12] C. Caliendo, Appl. Phys. Letters 92 103501 (2008).
[13] JPCDS cards No. 25-1133 Powder Diffraction File, Joint Committee on Powder
Diffraction Standards, ASTM, Philadelphia, PA, 1967, Card 25-1133
[14] C. Caliendo, P. M. Latino, Thin Solid Films 519 (2011), pp. 6326-6329
[15] C.K.Xu, V.N.Umashev, I.B.Yakovkin, Sov. Phys. PTE6, 192 (1986).
[16] Landolt-Bornstein, Numerical Data and Functional Relationships in Science and Technology,
Group III: Crystal and Solid State Physics, (Springer-Verlag Berlin, Heidelberg-New
York, 1979), Vol. 11 (1979).
[17] C.-Y. Lin, F.-H. Lu, J. Eur. Ceram. Soc. 28 691 (2008).
[18] E. A. Chowdhury, J. Kolodzey, J. O. Olowolafe, G. Qiu, G. Katulka, D. Hits, M. Dashiell,
D. van der Weide, C. P. Swann, K. M. Unruhm, Appl. Phys. Lett. 70 2732 (1997).
[19] F. Jose, R Ramaseshan, S. Dash, S. Bera, A. K. Tyagi, B. Raj, J. Phys. D: Appl. Phys. 43
075304 (2010).
[20] T. Aubert, O. Elmazria, B. Assouar, L. Bouvot, M. Oudich, Appl. Phys. Lett. 96 2035031
(2010).
[21] S.-K. Tien, C.-H. Lin, Y.-Z. Tsai and J.-G. Duh, J. Alloys Compd. 489 237 (2010).
[22] Z. Gu, J.H. Edgar, S.A. Spekman, D. Blom, J. Perrin, J. Chaudhuri, J. Electron. Mater. 34
1271 (2005).
[23] T. Aubert, M. B. Assouar, O. Legrani, O. Elmazria, C. Tiusan, S. Robert, J. Vac. Sci.
Technol. A 29 021010 (2011).
[24] B. Liu, J. Gao, K.M. Wu, C. Liu, Solid State Commun. 149 715 (2009).
[25] C. Caliendo, P. Imperatori, J. Appl. Phys. 96, 2610 (2004).
27
Applications of InFiber AcoustoOptic Devices
*
C. Cuadrado-Laborde
1,2
, A. Dez
1
, M. V. Andrs
1
,
J. L. Cruz
1
, M. Bello-Jimenez
1
, I. L. Villegas
1,3
,
A. Martnez-Gmez
3
and Y. O. Barmenkov
1,3

1. Introduction
Nowadays, in-fiber acousto-optic devices are increasingly used as frequency shifters,
multiplexers, modulators, and tunable filters. They can be easily spliced into optical fiber
systems, and the consequent low insertion loss, make them an attractive alternative to bulk
optics devices. Our group, established at the Institute of Materials Science, Department of
Applied Physics, of the Valencia University (ICMUV, Valencia, Spain), has been involved in
this field for the last ten years, and our fabrication facilities allow us the development of
new in-fiber acousto-optic devices for novel applications in different fields such as sensors,
microwave photonics, lasers, and optical communications. Our aim here is to present the
great potential shown by in-fiber acousto-optic devices for different photonic applications.
Although this chapter is focused in our latest developments, the reader will also find
discussed the work done by other research groups in the field. This chapter is divided in
two main sections: Section 2 is focused on novel applications of acousto-optic fiber devices
based on flexural acoustic waves, and Section 3 is focused on applications of the acousto-
optic devices based on the interaction of longitudinal acoustic waves with fiber Bragg
gratings (FBG). Finally, our conclusions are shown in Section 4.
2. Applications of acousto-optic devices based on flexural acoustic waves
In a standard single mode fiber, when a flexural acoustic wave propagates along an optical
fiber a periodic perturbation is introduced in its refractive index, and it can induce coupling
between the fundamental mode guided by the core and the modes supported by the
cladding (Kim et al., 1997). This acousto-optic interaction can be seen as the dynamic
counterpart of a long period grating (LPG). LPGs are usually fabricated by creating a
periodic perturbation of the refractive index by UV radiation; this of course fixes its spectral
characteristics. On the other hand, when the perturbation is introduced by an acoustic wave,
its period and strength can be controlled through the frequency and amplitude of the
acoustic wave, respectively. Thus, the spectral properties of the optical device can be
controlled dynamically through the characteristics of the acoustic perturbation. The optical

*1
Departamento de Fsica Aplicada y Electromagnetismo, ICMUV,
Universidad de Valencia, Burjassot, Valencia, Spain
2
CONICET La Plata, Buenos Aires, Argentina,
3
Centro de Investigaciones en ptica, Len, Guanajuato, Mexico


596
coupling is resonant in wavelength; it takes place at the optical wavelength that verifies the
phase-matching condition between the beat length of the two optical modes and the acoustic
wavelength. At the output of the acousto-optic device, only the light that remains guided by
the core mode is transmitted. Thus, the coupling of power from the fundamental mode to a
cladding mode results in the appearance of an attenuation notch in the spectrum, whose
amplitude is fixed when travelling acoustic waves are used. The characteristics of the
fundamental flexural acoustic mode of an optical fiber fulfill the above requirements for the
development of efficient in-fiber acousto-optic devices. In subsection 2.1 we will describe a
Q-switched fiber laser where cavity loss modulation is reached by using this operation
principle. As an interesting alternative, if standing instead of travelling acoustic waves are
used, light transmitted at the resonance wavelength will experience an amplitude
modulation at twice the frequency of the acoustic wave. In subsection 2.2 we describe a
mode-locked fiber laser based on the use of this operation principle.
2.1 Q-switching by intermodal acousto-optic modulation in an optical fiber
Q-switched fiber lasers have attractive applications in different fields, such as in remote
sensing and material processing. The mechanism of short pulse emission is based on the
modulation of the Q factor of the cavity, which can be done either passively or actively
(Siegman, 1986). In the former the setups are simpler, but the repetition rate only varies with
the pump power of the medium gain. Further, they usually show long-term instability, and
frequently the amplitude of the pulses is randomly modulated in time (Vicente et al., 2004).
On the other hand, active Q-switching is independently and accurately controlled by an
electrical signal, which triggers the modulator. The bulk approach, such as electro-optic (Kee
et al, 1998) or acousto-optic modulators (lvarez-Chvez et al., 2000), is not adapted to
compact fiber laser systems required nowadays; further they have large optical coupling
losses and stringent alignment requirements. For these reasons, the all-fiber approach is of
permanent interest, being advantageous in terms of cost, loss, packaging, robustness, and
simplicity. One solution widely investigated has been enabled by the use of FBGs as cavity
mirrors. In this way, the tuning of the wavelength of one of the FBGs has been used to
achieve active Q-switching (Andrs et al., 2008). Among the typical tuning methods, we
could mention the stretching of fiber Bragg gratings by magnetostrictive materials (Prez-
Milln et al., 2005a; Andersen et al., 2006), by piezoelectric actuators (Imai et al., 1997; Russo
et al., 2002), or by the interaction of longitudinal acoustic waves with the FBG (Delgado-
Pinar et al., 2006; Cuadrado-Laborde et al., 2007). Other technique used for active Q-
switching relied in the cavity loss modulation by the core-to-cladding mode-coupling in
optical fibers induced by travelling flexural acoustic waves (Huang et al., 2000; Zalvidea et
al., 2005). As opposed to the last two referred works which were focused at the erbium
lasing wavelength, here we experimentally demonstrate the possibility of using this
modulating technique at the technologically relevant ytterbium lasing wavelength (Tuchin,
1993). Very few actively Q-switched ytterbium-doped lasers with an all-fiber configuration
have been reported (Andersen et al., 2006). The main difficulties arise from a relatively long
time response and a limited modulation depth of all-fiber amplitude modulators. Here, we
present the experimental results that we have obtained in a research work focused on the
exploitation of in-fiber acousto-optics. In our proposed Q-switched fiber laser, when the
acoustical signal is switched-off, the optical power losses within the cavity are reduced, and
then a laser pulse is emitted (Villegas et al., 2011b).


597

Fig. 1. Q-switched fiber-laser setup; the acousto-optic modulator is defined by the elements
inside the dashed line; AD stands for acoustic dumper
The setup used for our Q-switched fiber laser is schematically illustrated in Fig. 1. The gain
was provided by 0.65 m of a heavily-doped ytterbium-doped single-mode fiber Nufern
SM-YSF-HI, cut-off wavelength of 860 70 nm, numerical aperture of 0.11, and fiber
absorption of 250 dB/m at 975 nm. The active fiber was pumped through a WDM coupler
by a pigtailed laser diode emitting at 980 nm, providing a maximum pump power of 110
mW. The in-fiber acousto-optic modulator (AOM) was spliced between FBG
1
Bragg
wavelength at 1064 nm, FWHM of 0.23 nm, and 99.6 % of maximum reflectivity and FBG
2

Bragg wavelength at 1064 nm, FWHM of 70 pm, and 44 % of maximum reflectivity,
defining in this way a Fabry-Perot cavity. The AOM in turn is composed of an RF source, a
transversal-mode piezoelectric disk, an aluminum horn, and a tapered single-mode optical
fiber Fibercore SM980 of low numerical aperture (0.13-0.15) . The optical fiber was tapered
down by the fusion and pulling technique using a travelling flame, which produces a taper
waist with a uniform diameter of 76 m and 0.1 m length, for this specific case. The tip of the
aluminum cone with the piezoelectric disc fixed to its base, was glued to an uncoated
section of fiber near the taper, see Fig. 1. Finally, the optical fiber in the AOM was
acoustically dumped in both extremes; see Fig. 1, in order to prevent unwanted acoustical
reflections.
When an RF signal is applied to the piezoelectric disc, a travelling flexural acoustic wave is
launched through the taper. If the acoustic wavelength matches the beat-length between the
fundamental mode guided by the core and one of the optical modes supported by the
cladding, then light coupled to the later remains in the cladding downstream of the taper,
being finally absorbed by the fiber coating (Birks et al., 1994). Thus, the coupling of power
from the fundamental mode to a cladding mode results in the appearance of an attenuation
notch in the spectrum. When the acoustic frequency is varied, the periodicity of the
perturbation also does, and hence the phase-matching condition is shifted to a different
optical wavelength. Figure 2(a) shows the tunability of the notches caused by the coupling
between the fundamental core mode and the first three cladding modes LP
1m
. The selected
operating point was at an optical wavelength of 1064.1 nm for an RF signal applied to the
piezoelectric of 825 kHz. These measurements were made by illuminating the taper with a
broadband light source and detecting the light transmitted through the taper with an optical
spectrum analyzer. As an example, Fig. 2(b) shows one of the measured transmission
spectra for an applied voltage to the piezoelectric of 26 V and a frequency of 885 kHz. The
transfer of optical power from the fundamental core mode to one of the cladding modes


598
behaves periodically as a function of the acoustic power, which in turn is a function of the
applied voltage to the piezoelectric. Figure 2(c) shows this effect for the selected operation
point marked in Fig. 2(a). As it can be observed, the transmittance decays by a maximum of
12 dB for an applied voltage to the piezoelectric of 50 V (peak-to-peak measurement).
Beyond this point, further increment in the applied voltage raises the transmittance again.
Therefore, and bearing in mind its use as a Q-switching device, cavity loss modulation
between 0 dB and 12 dB can be achieved by applying to the piezoelectric a sinusoidal signal
at the frequency of 825 kHz fully-modulated by a rectangular signal, see the inset of Fig.
2(c). This modulation produces onoff periods of the acoustic wave travelling down the
fiber, which results in a modulation of the cavity losses at the resonance wavelength. In
passing, we should mention that several tapers were fabricated in order to better optimize
this laser system, by using Fibercore SM980 of high numerical aperture (0.17-0.19), and
Nufern SM-YSF-LO (a moderately ytterbium-doped fiber), and in turn with different taper
waists. Despite this, the minimal decay in transmittance was always around the reported
values, i.e. between 10 and 16 dB.

Fig. 2. (a) Resonant optical wavelengths as a function of the acoustic frequency for the first
three mode-couplings LP
01
-LP
1m
. (b) Typical transmission notch caused by the first mode-
coupling by applying to the piezoelectric a sinusoidal signal at 885 kHz and 26 V. (c)
Transmittance at the selected operation point marked in (a) i.e. 1064.1 nm and 825 kHz as
a function of the applied voltage to the piezoelectric (solid scatter points), the curve
represents a theoretical fitting according to a sin
2
function
The switching time is one of the key parameters for any modulator intended to be used as a
Q-switching device in a fiber laser. Preferably, it should be as short as possible. For this
reason we measured the temporal response of this device, by detecting the transmitted light,
while simultaneously registering the modulating signal in an oscilloscope. It takes 25 s to
increase/decrease the transmitted optical power through the taper when the voltage applied
to the piezoelectric is switched-off/on, respectively. This time corresponds reasonably well
with the time it takes a flexural acoustical wave to travel down the taper, i.e. 0.1 m / 3764
m/s 26 s. This value is short enough to allow the use of this modulator as a Q-switching
device in a fiber laser, as it will be demonstrated in the following section. In principle, and for a


599
given taper waist, shorter switching times could be achieved by decreasing the interaction
length. Unfortunately, this simultaneously increases the acoustical power needed to reach the
same optical coupling power, therefore there is a trade-off between both parameters.

Fig. 3. (a) Output optical power in CW emission as a function of the pump power.
(b) Emission spectrum
In this setup, both gratings are permanently tuned to the same wavelength by using two
translational stages, one for each FBG; see the setup in Fig. 1. In this way, for zero voltage
applied to the piezoelectric, this laser emits in CW, since then the cavity losses are minimal,
see the transmittance in Fig. 2(c) for 0 V. Figure 3(a) shows the optical power in CW
emission as a function of the pump power. There is a pump power threshold of 56 mW,
about which the laser starts to emit, reaching a maximum output power of 3.4 mW, which in
turn is determined by the maximum pump power available in our setup (110 mW). Figure
3(b) shows the spectrum of CW emission; its linewidth is of 0.21 nm at a center optical
wavelength of 1064.1 nm.
Now, we discuss the Q-switched operation of this laser. To this end, we modulated the cavity
losses by applying to the piezoelectric a fully-modulated sinusoidal signal at the frequency of
825 kHz and 50 V. A rectangular wave was used to modulate the RF voltage that generates the
acoustic wave. At a given frequency of the modulating signal, we found always a maximal
duty cycle i.e. the fraction of time that the signal is in its high level able to perform Q-switch
correctly. If we decrease this duty cycle, then the cavity would stay in its high Q state longer,
and more than one Q-switched pulse would be emitted in each time slot. The Q-switch
repetition rate becomes determined by the frequency of the modulating signal; with this
configuration we reached continuous tuning of the Q-switch repetition rate in the range 1-10
kHz. Figure 4(a) shows, as an example, a Q-switched optical pulses train at 1 kHz, together
with the corresponding modulating signal, for a pump power of 59 mW. Figure 4(b) shows a
detail of a single Q-switched optical pulse of the train shown in Fig. 4(a) with a time width
(FWHM) of 3.72 s. The pulses have a quasi-Gaussian profile; the fitting by this function is also
shown in Fig. 4(b). The effect of pump power on the Q-switched pulses, for different repetition
rates in the range 1-10 kHz, is shown in Fig. 5. For each Q-switching frequency there is a pump
power threshold. Above threshold, the peak power increases with pump power, and there is a
corresponding reduction of pulse width. For a given frequency there is also a pump power
level beyond which extra pulses appear, the curves are truncated at that point. In order to


600
overtake this limitation, a modulator with improved modulation depth is required. In
addition, our laser produces easily multiple pulse emission, which is a well-known problem in
actively Q-switched lasers with relatively large time responses. A large time response makes
critical the adjustment of the duty cycle of the modulation voltage when the pump power is
increased, particularly at low repetition rates. Consequently, a faster switching response of the
acousto-optic modulator is required in order to improve the operation of the laser.

Fig. 4. (a) Modulating signal (above) at 1 kHz repetition rate together with the generated Q-
switched laser output (below). (b) Detail of a single Q-switched pulse of (a) (scatter points)
together with its corresponding fitting by a Gaussian function (solid curve)

Fig. 5. Peak power and time width in Q-switched operation as a function of the pump
power, (a) and (b), respectively, for several repetition rates
2.2 Mode-locking by intermodal acousto-optic modulation in an optical fiber
Mode-locking lasers have a vast number of applications, ranging from the
telecommunications industry to medical surgery (Bonadeo et al., 2000; Haus, 2000; Yu et al.,
2000; Schaffer et al., 2003). The shortest pulses are obtained by passive schemes. However,
they also have some drawbacks, among them, they are generally unstable, they have higher


601
timing-jitter between pulses and usually self-starting cannot be assured. An alternative to
using passive schemes that reduces jitter and allows for synchronization is to use an active
scheme. To this end, a modulator is driven at the cavitys fundamental repetition rate;
therefore, timing-jitter is reduced since each pulse is triggered by the modulator, which in
turn is accurately driven by an electrical signal. In this way, in active mode-locking one has
the possibility to synchronize a laser to an external clock or electronic signal (French, 1995).
Among all these lasers, the all-fiber solution has been the subject of intensive research due to
their compact structure, free alignment problems, and low cost, between other merits.
Earlier and subsequent active modelocked fiber lasers were not strictly all-fiber lasers, since
they currently included a bulk modulator (Geister & Ulrich, 1988; Phillips et al., 1989a;
Hudson et al., 2005). This not only induces extra losses into the laser cavity but of course
also destroys the benefits of an all-fiber configuration. On the contrary a strictly all-fiber
modulator can overcome these limitations and it can be used for actively mode-lock rare-
earth-doped silica fiber lasers (Phillips et al., 1989b; Culverhouse et al., 1995; Jeon et al., 1998;
Costantini et al., 2000; Myrn & Margulis, 2005; Cuadrado-Laborde et al., 2009a, 2010a,
2010b; Bello-Jimnez et al., 2010, 2011). Table 1 shows the major achievements towards the
obtaining of all-fiber actively mode-locked lasers. As it can be observed, acousto-optic
devices have been by far the preferred technique, with exception of the Myrn and
Marguli`s work relying in a electro-optic effect. Our group was involved in works described
in the last two rows of this table, which will be discussed in detail in this chapter.
In Jeon`s et al. work (1998), mode-locking is achieved by frequency shifting based on
travelling flexural acoustic waves in a two-mode optical fiber; such a frequency shifter could
be considered an improved version of the one reported in (Kim et al., 1986). Unlike Jeon
(1998) and Kim (1986), here we propose using a standard single-mode optical fiber in which
core-to-cladding mode-coupling is induced by a standing flexural acoustic wave. In this
way, the device does not work as a frequency shifter, but as an amplitude modulator, which
is used to actively mode-lock a fiber laser. In the following, we demonstrate that this simple
acousto-optic modulator proves to be particularly well-suited for active mode-locking
purposes (Bello-Jimnez et al., 2010, 2011).
The setup of the actively modelocked fiber ring laser is schematically illustrated in Fig. 6.
The medium gain was provided by 4.36 m of erbium-doped fiber (EDF) containing 300 ppm
Er
3+
, with a cut-off wavelength of 939 nm, and a numerical aperture of 0.24. The active fiber
was pumped through a WDM by a 980 nm pigtailed laser diode, providing a maximum
pump power of 600 mW. Next and following a clockwise direction it was inserted the
acousto-optic modulator between two polarization controllers (PCs), followed by a delay
line fusion-spliced to port one of a polarization-independent three-port optical circulator
(OC). The OC not only forces the unidirectional operation within the ring but
simultaneously incorporates the FBG filtering in the cavity by the second port. Finally, the
ring cavity is closed by connecting port three of the OC to the WDM. The output of this laser
is obtained by transmission through the FBG Bragg wavelength at 1549.5 nm, FWHM of
0.45 nm, and 50 % of maximum reflectivity. In this way, the entire laser is spliced. The
AOM in turn is composed of an RF source, a piezoelectric disk, an aluminum conical
horn, and a 0.2 m long standard optical fiber stripped of its polymer coating; in order to
prevent the attenuation of the acoustic wave. The horn is attached to the piezoelectric, and
it focuses the vibrations into the fiber through its tip, which is glued to the uncoated fiber.
In order to allow the generations of a standing flexural acoustic wave, the uncoated
optical fiber of the AOM was firmly clamped at one extreme, whereas in the other it was
damped, see Fig. 6.


602
Year Group Mode-locking mechanism Major achievements
1989 Phillips et al.
Acousto-optically induced intra-
fiber phase modulation
First all-fiber actively mode-locked
laser, 200 ps time width pulses
1995
Culverhouse et
al.
Acousto-optically induced
frequency shifting in an all-fiber
four-port null coupler
18 ps time width pulses (chirped)
1998 Jeon et al.
mode-coupling in a two-mode
optical fiber
Harmonic mode-locking, 3.8 ps
time width pulses
2000
Costantini et
al.
Acousto-optically induced intra-
fiber phase modulation
Visible up-conversion laser, 550 ps
time width pulses
2005
Myrn and
Margulis
Electro-optic effect in an all-fiber
MachZehnder interferometer
1 ns time width pulses
2009
Cuadrado-
Laborde et al.
Acousto-optic super-lattice
effect in a fiber Bragg grating
630 ps un-chirped time width
pulses. First to demonstrate
double-active all-fiber Q-switching
and mode-locking
2010
Bello-Jimnez
et al.
mode-coupling in a single-mode
optical fiber
34 ps time width pulses
Table 1. Major achievements in the development of all-fiber actively mode-locked lasers

Fig. 6. Modelocked fiber-laser setup; where OC stands for optical circulator, EDF for erbium
doped fiber, PC for polarization controller, FBG for fiber Bragg grating, and WDM for
wavelength division multiplexer. The acousto-optic modulator is defined by the elements
inside the dashed line
Figure 7(a) shows the modulation depth as a function of light wavelength for an acoustic
frequency of 2.37315 MHz. The voltage applied to the piezoelectric was 18 V (whenever we
refer to voltages, it is a peak-to-peak measurement). The measurement was performed by


603
illuminating the AOM with a tunable laser and detecting the transmitted light at each
wavelength. The modulation is maximal at the phase-matching wavelength of 1551 nm, and
symmetrically decreases for longer and shorter wavelengths. It should be emphasized the
high modulation depth achieved in this technique, reaching a maximum of 72 %, together
with the relatively broad operative bandwidth of 1.5 nm (FWHM) i.e. 187 GHz at the
operation wavelength. Moreover, the insertion loss is as low as 0.7 dB, corresponding to the
maximum transmittance value attained, which is 86 %. When the acoustic frequency is varied,
the periodicity of the perturbation also does, and hence the phase-matching condition is
shifted to a different wavelength. Figure 7(b) illustrates this effect; the rate of change for the
wavelength shift is of 0.169 nm/kHz; the maximum modulation reached at each specific
operation point is also shown to the right. The differences between the modulation depths at
different acoustic frequencies are mainly originated from the combination of the non-flat
frequency response of the piezoelectric and the acoustic fiber resonator.

Fig. 7. (a) Modulation depth as a function of the optical wavelength for an RF signal applied
to the piezoelectric of 2.37315 MHz and 18 V. (b) Resonant optical wavelengths as a function
of the frequency of the RF signal, its corresponding modulation depth is also shown to the
right ordinate
Now we discuss the operation of the modelocked laser shown in Fig. 6. The delay line
length must be selected to match the round-trip time with the reciprocal of the optical
modulation frequency, which in turn is twice the electrical frequency applied to the
piezoelectric. The selected operation point for the piezoelectric is the same as in Fig. 7(a), i.e.
2.37315 MHz, and then the cavity length results in 43.7 m. However, fine tuning is always
necessary to match the frequency of the modulation signal to the inverse of the round-trip
time of the cavity. A translation stage was also used for matching the reflection band of the
FBG to the operative wavelength (1551 nm). Regarding the dispersion of this laser; we
measured the dispersion of each fiber and component within the cavity by the frequency-
domain modulated-carrier method. The laser cavity is a mixed-dispersion fiber ring with an
average cavity dispersion of 1 ps/nm/km (Bello-Jmenez et al., 2010). We measured also
both the dispersion of the OC and WDM, both resulted normal also with a group delay per


604
wavelength unit of 0.3 ps/nm and 0.04 ps/nm, respectively. In this way, we can assure
that this laser works under the normal dispersion regime. Figure 8(a) shows the RF signal
used to drive the piezoelectric together with the modelocked train of pulses generated (50
GHz bandwidth oscilloscope). As it can be observed, the frequency of the optical train (4.756
MHz) is twice the frequency of the signal used to drive the piezoelectric (2.37315 MHz). The
optical pulses were best fitted by a sech
2
function; neither Gaussian nor Lorentzian functions
improved the fitting. This is an expected profile for the optical pulses in this type of
dispersion-mixed cavities, in which both kind of dispersion coexist, i.e. normal and
anomalous (Blanger, 2005). On the other hand, according to active mode-locking theory,
the time width of the optical pulse can be expressed as (Kuizenga & Siegman, 1970):

0
4
2 ln2 1
,
m m m
g
f f
(1)
where f
m
is the modulation frequency, g
0
is the gain of the active medium, f
m
is the
modulator bandwidth, and
m
is the modulation depth. As a rough estimation, let us
compare these results with the results reported in Cuadrado-Laborde et al. (2009a), where
optical pulses of 780 ps were obtained by using also AM mode-locking, but with a different
modulator and setup. Thus, by comparing them, here there is a higher modulation depth
(7), modulation bandwidth (240), and gain (3), but a lower modulation frequency (0.5)
where the additional assumption that gain scales with the EDF length was made.
Therefore, there is a narrowing factor for the time width of (3/7)
1/4
(0.5240)
1/2
= 7.410
2
.
In this way, comparatively, it should be possible to reach optical pulses around 7.410
2
780
ps 58 ps in this laser, which could be taken roughly as a lower limit for this configuration.
One key element, out of the preceding discussion, is the FBG bandwidth, which acts as a
filter see the setup in Fig. 6 whose bandwidth must be selected narrower than the AOM
bandwidth; otherwise modes not amplitude modulated would interfere in the mode-locking
process. This evidently limits the effective bandwidth and raises the minimum pulse width
able to be reached with this configuration, since modulator bandwidth and time width are
inversely related. The optical pulses peak power and temporal width as a function of the
pump power are shown in Fig. 8(b). The time width as well as the peak power
monotonically increases with the pump power; at the lowest pump powers, pulses as
narrow as 95 ps were obtained with a pump power of 110 mW; see the inset of Fig. 8(b).
Pump powers below 110 mW precludes mode-locking; whereas the upper limit is reached
by the maximum pump power available in our setup.
Next, we analyzed the dependence of the pulses parameters with the applied voltage to the
piezoelectric, see Fig. 9(a). Both the transmittance and modulation depth are functions of the
applied voltage to the piezoelectric through a periodic relationship. However, we can
consider that the relationship is linear within the range of voltages of Fig. 9(a). When the
voltage increases, the modulation also does, and pulses become narrower, as expected from
AM mode-locking theory, see Eq. (1). Finally, it is of practical interest to quantify the
maximum allowable detuning (
max
), i.e. the maximum difference between the reciprocal
of the cavity round trips time and modulation frequency that sustain mode-locking.
Figure 9(b) shows the variation of the pulses parameters under frequency detuning,
resulting in
max
= 300 Hz at half the maximum peak power. According to theory of
detuning in AM mode-locking (Li et al., 2001), the maximum allowable detuning can be
expressed as:


605

max
,
4
m
c m
n f L
=
(2)
with c the speed of light in vacuum, L is the cavity length, and n the modal effective index.
Thus, the minimum locking range is attained when the whole modulation bandwidth is
available, by replacing we obtain
max
= 42 Hz. This value represents a lower limit in this
setup, since we are not using the whole modulator bandwidth due to the reflection
bandwidth of the FBG. Thus, a higher measured locking range is an expected result.

Fig. 8. (a) Voltage signal used to drive the piezoelectric and mode-locked train of pulses
generated at 4.75 MHz repetition rate with 271 mW of pump power (dashed and solid
curves, respectively). (b) Time width (FWHM) and peak power of the optical pulses as a
function of the pump power (solid and open scatter points, respectively). The inset shows a
single pulse of 95 ps time width

Fig. 9. (a) Time width (FWHM) and peak power as a function of the applied voltage; for a
fixed frequency of 2.3731 MHz (open and solid scatter points, respectively). (b) Same as
before, but as a function of the frequency detuning, for a fixed voltage of 26 V. In both cases,
a pump power of 350 mW was used
Next, we carried out an experimental study of this actively mode-locked all-fiber ring laser,
looking towards an improvement in its performance. To this end, we slightly modified the
setup shown in Fig. 6, by replacing the FBG used before with a near 100 % reflective FBG,
with a FWHM bandwidth of 0.3 nm, which should be compared with the preceding FBG
with FWHM of 0.45 nm, and 50 % of maximum reflectivity. Together with this, the EDF


606
length was shortened up to 2.85 m. As a result of these changes, output light pulses were
obtained by one of the ports of a 3 dB coupler incorporated within the ring (Bello-Jmenez et
al., 2011). With this new configuration, output light pulses were obtained with a maximum
peak power of 380 mW and pulse width of 90 ps. By comparing the narrowest pulses
obtained with this configuration with the narrowest pulses reported before (95 ps) where the
same type of AOM was used, we find nearly no difference. However, from the laser operation
point of view, there is an important improvement, since the present arrangement is rather
more stable, and the laser can be adjusted more easily. We attribute this improvement to the
FBG used, which has a flat spectral reflectivity, making less critical the spectral matching
between the FBG reflectivity with the resonant dip of the AOM. On the contrary, the FBG used
before, was of nearly the same FWHM but with a lower reflectivity (50 %). As a consequence,
any spectral detuning resulted in a larger variation in reflectivity, which in turn induced a
higher difficulty to stabilize the laser operation. Next, we analyzed the change in the pulse`s
parameters as the EDF length was varied, see Fig. 10 (a). Changes in the EDF length induced a
slight shortening in the output light pulses, which is in accordance with Eq. 1, if we assume
that gain scales with the EDF length. By this trend, the shortest light pulses (65 ps) were
obtained at 2.2 m of EDF length. Further narrowing by this trend was not possible, since no
mode-locked output pulses were observed for EDF lengths shorter than 2.15 m.

Fig. 10. (a) Time width (FWHM) and peak power as a function of the EDF length, using a FBG
with 0.3 nm bandwidth. Using a 0.7 nm bandwidth: (b) single mode-locked pulse at 4.73 MHz
repetition frequency, together with its corresponding fitting by a sech
2
function; (c) optical
spectrum of the laser; (d) time width and peak power versus the frequency detuning
An important parameter of our laser arrangement is the FBG bandwidth, which acts as a
spectral filter. Its bandwidth has to be selected narrower than the AOM bandwidth; in order
to insure that the modes are properly amplitude modulated. For this reason we used first a
0.3 nm bandwidth FBG, which fulfills this requirement, since the AOM bandwidth is of 1.5
nm. However, it can be easily understood that a larger number of modes locked i.e.
amplitude modulated produce the narrowest mode-locked train of pulses, which is in
accordance with Eq. 1. In this respect, there is margin for improvement, since the bandwidth
of the FBG used is far to be close to the AOM bandwidth. For this reason we replace the flat-
top FBG with another FBG, which also has a flat-unit reflectivity, but a broader optical


607
bandwidth of 0.7 nm. Fig. 10(b) shows the shortest pulse obtained with this configuration,
by using an EDF length of 2.45 m at a pump power of 265 mW. The measured temporal
width was 34 ps (FWHM), with a maximum peak power of 1.4 W. For this pulse the spectral
linewidth was measured to be 110 pm (i.e. 13.7 Ghz at 1551.9 nm), using a 50 pm resolution
optical spectrum analyzer, see Fig. 10(c). On the other hand, a Fourier-transform limited
sech
2
pulse should have a spectral linewidth of 9.3 GHz. From the comparison between this
last value and the linewidth measurement, we conclude that the optical pulses of our mode-
locked laser could have some moderate degree of chirp. Finally, we present the variation of
the pulse parameters as a function of the frequency detuning for this configuration, see Fig.
10(d); the measured allowable frequency detuning was of 8 Hz. According to theory of
detuning in AM mode locking, the maximum allowable detuning is inversely proportional
to the modulator optical bandwidth, in this case replaced by the FBG bandwidth, see Eq. 2.
Since the later has been increased noticeably, i.e. from 0.3 nm to 0.7 nm, a reduction in the
allowable detuning is an expected result.
3. Applications of acousto-optic devices based on the interaction of
longitudinal acoustic waves with fiber Bragg gratings
When a travelling axially-propagating longitudinal acoustic wave is launched through a
FBG, the periodic strain field of the acoustic wave perturbs the grating in two different
ways. First, the average index changes in response to the stress-optical effect and second, the
otherwise uniform Bragg grating pitch changes being modulated by the acoustical signal
(Russell & Liu, 2000). As a consequence of both effects the reflectivity changes, the main
features of these changes depend on the ratio between the acoustical wavelength and the
grating length (Andrs et al., 2008). Thus, we can distinguish two well-different situations:
the long-wavelength and the short-wavelength regimes. In the first, the Bragg grating pitch
is homogeneously perturbed along its length. The successive cycles of compression and
expansion generated by the longitudinal wave will shift periodically in time the spectral
response of the grating as a whole to longer and shorter wavelengths (Cuadrado-Laborde et
al., 2007; Andrs et al., 2008). On the contrary, in the short-wavelength regime the acoustic
wave generates many compressed and expanded sections, which gives rise to a superstructure
within the grating. In this case, the spectral response of the original FBG shows new and
narrow reflection bands symmetrically at both sides of the original Bragg wavelength (Liu et
al., 1997, 1998). The position and strength of these sidebands can be controlled by varying the
frequency and voltage applied to the piezoelectric, respectively. Subsections 3.1 and 3.2 show
the use of these effects to modulate the Q-factor of a fiber laser and as an active mode-locker;
i.e. long and short-wavelength regimes, respectively. Subsection 3.3 shows the use of the
acousto-optic interaction in a FBG in the short-wavelength regime as a tunable photonic true-
time-delay line based on the group delay change of the light reflected from the grating
sidebands. Finally, when the acoustic perturbation is not a harmonic wave but a single pulse,
its passage through the grating creates a defect in the FBG, which can be used to actively Q-
switch a distributed feedback fiber laser, which is discussed in subsection 3.4.
3.1 Q-switching of an all-fiber laser by acousto-optic interaction in a fiber Bragg
grating in the long-wavelength regime
As an example of the use of the long-wavelength regime in FBGs, we present an all-fiber
acousto-optic modulator suitable for Q-switching applications (Cuadrado-Laborde et al.,


608
2007). It consists of a short-length fiber Bragg grating modulated by a standing longitudinal
acoustic wave, whose wavelength is much longer than the grating length. Periodic
stretching and compression of the grating due to the elastic wave, causes that the Bragg
wavelength is harmonically, continuously and repeatedly tuned in time over a given
wavelength range. Figure 11 illustrates the operation principle. In the long-wavelength
regime, the acoustic wavelength is much larger than the grating length.
However, the amplitude of the acoustic waves that can be achieved in a realistic
arrangement is rather small to produce a significant perturbation of the FBG. We found that
this limitation could be overcome by using an acoustic cavity. Thereby, if the grating is
placed in the proper position within the acoustic cavity, a low-frequency standing longitudinal
elastic wave will produce periodic stretching and compression of the grating. As a
consequence, the reflection band will shift periodically in wavelength around the original
Bragg wavelength, sweeping a given wavelength range in a harmonic way. The elastic wave
may also introduce some chirp in the grating due to non-uniformity of the strain induced
along the grating. To achieve efficient modulations this effect must be minimized. Thus, the
grating length must be short compared with the acoustic wavelength, and the positioning of
the grating within the cavity must match an anti-node of the standing acoustic wave.

Fig. 11. Interaction of acoustic waves in the long-wavelength regime and FBGs; T is the
period of the acoustic wave, is the instantaneous pitch of the Bragg grating, and
0
is the
pitch of the Bragg grating in absence of acoustic wave. The effect on the reflection band is
shown qualitatively. The standing acoustic wave stretches and compresses the grating
periodically. During the time that the grating is stretched, the reflection band of the grating
shifts towards longer wavelengths, whereas it shifts towards shorter wavelengths when it is
compressed
Figure 12(a) shows a diagram of the laser setup, together with a detail of the acousto-optic
modulator. The Fabry-Perot cavity becomes defined by FBG
1
(which also acts as modulator)
and FBG
2
. A piezoelectric disk (15 mm in diameter and 2 mm thickness) was used to
generate the longitudinal acoustic waves, which were launched to FBG
1
through a silica
horn. The base of the horn was glued to the piezoelectric and the tip reduced to ~125 m
and then fusion-spliced to FBG
1
. At a distance of 210 mm from the piezoelectric disk, the
fiber was clamped allowing standing acoustic waves to be created in the fiber section
between the horn and the clamp. FBG
1
was written in the core of a photosensitive fiber by


609
UV irradiation using a phase-mask technique. The photosensitive fiber was previously
tapered to enhance the acousto-optic interaction; the taper waist was 50 mm long, 90 m in
diameter, and the transition lengths were 13 mm. The grating was recorded in the taper
waist (uniform diameter section); the FBG
1
length was 22 mm. The highest frequency used
in our experiments was 61 kHz, and according to the velocity of sound in silica fibers for
longitudinal elastic waves (5760 m/s), the wavelength for this frequency is 94 mm. Thus, we
can verify that FBG
1
length is short enough to guaranty a standard long-wavelength regime.
An apodization profile was introduced also during the grating inscription to reduce side-
lobes in the spectrum. Figure 12(b) shows the transmission spectrum for both FBGs, the
bandwidth of FBG
1
was 0.15 nm and the reflectivity was higher than 99.9%. On the short-
wavelength side, the transmission drops sharply (~ 0.72 dB/pm). The side-lobes have been
reduced entirely on this side, while a significant side-lobe still remains on the long-wavelength
side of the transmission band. On the other hand, FBG
2
acted as the output coupler, its Bragg
wavelength was 1543.27 nm, the reflectivity 50%, and the bandwidth was 40 pm.

Fig. 12. (a) Laser setup, the dashed line defines the acousto-optic modulator. (b)
Transmission spectra of the two fiber Bragg gratings
Next, we discuss the modulators performance. We detect the intensity of the light reflected
by the mid-reflection point of the short-wavelength band-edge, when an AC electric signal
of a given frequency was applied to the piezoelectric. To this end a tunable laser diode was
used. The modulation of the grating caused by the elastic wave generates periodic
modulation of the light intensity reflected by the grating. An example is shown in Fig. 13(a).
At low voltage, a sinusoidal modulation was observed since the grating shifts little and the
laser reflects at the band-edge at any time during the period of the electric signal. As the
electric signal amplitude increases, the grating shifts further and then, the laser reflects also
at the top and bottom of the band-edge, generating a square signal. We observed strong
acoustic resonances at frequencies of 18 kHz, 37 kHz and 61 kHz. The peak-to-peak
amplitude was measured for these frequencies as a function of the voltage applied to the
piezoelectric, see Fig. 13(b). The most efficient frequency was 61 kHz, and the responses at
18 kHz and 37 kHz were quite similar. Taking into account the results shown in Fig. 13(b)
and the width of the band-edge, 42 pm, the modulation efficiency (i.e. wavelength shift per
unit volt) was estimated, giving 7 pm/V at 18 kHz and 37 kHz, and 16 pm/V at 61 kHz.
The switching speed is one of the key parameters of any modulator intended to be utilized
as Q-switching element in a fiber laser, being desirable to be as high as possible. In this
modulator, the switching time depends on the AC voltage amplitude, and it becomes
shorter as the voltage is increased, as shown in Fig. 13(a). Switching times of the order of a


610
few microseconds were achievable. Finally, although the acoustic generator was designed
for the excitation of longitudinal waves, we also observed flexural waves in the resonator
when large voltages were applied to the piezoelectric. A large number of flexural standing
waves, spaced fractions of kHz, were observed in the range from 0 to 100 kHz. These
resonances gave rise to light modulated at double frequency of the electrical excitation and
exhibited relatively low amplitude. The aforementioned modulator uses an acoustic
resonator, which enhances the interaction between acoustic waves and the FBG. The
fundamental mechanical resonance at 18 kHz had a Q factor of 1300. With respect to an un-
clamped configuration based on traveling acoustic waves, the use of an acoustic cavity
increases the modulation efficiency, but it restricts the operation of the modulator at the
resonant frequencies of the cavity. A modulator based on traveling acoustic waves would
allow, in principle, tuning continuously the modulation frequency, but a large voltage
would be required and important thermal effects would damage the acoustic transducer.

Fig. 13. (a) Light reflected from the band-edge of the grating when an AC electric signal of
61 kHz and peak-to peak amplitude of 4.6 V (green) and 14 V (blue) were applied to the
piezoelectric. (b) Light modulation amplitude as a function of voltage for three different
resonant frequencies

Fig. 14. (a) CW emission spectrum at 135 mW pump power; the inset shows the output CW
power versus pump power. (b) Pulse train at 18 kHz repetition rate; the inset shows a single
light pulse measured with a 1 GHz bandwidth photodetector, pump power of 80 mW


611

Fig. 15. (a) Pulse peak power as a function of the applied voltage to the piezoelectric (75 mW
pump power, 18 kHz repetition rate). (b) Pulse peak power and pulse width as a function of
pump power, for repetition rates of 18 kHz (blue) and 37 kHz (green). Voltage applied to the
piezoelectric 24 V
The gain of the Q-switched all-fiber laser was provided by 0.3 m of an erbium-doped fiber
containing 1000 p.p.m. Er
3+
, with a cut-off wavelength of 965 nm, and a numerical aperture
of 0.23. The active fiber was pumped through a WDM coupler by a pigtailed laser diode
emitting at 979 nm, providing a maximum pump power of 140 mW. The FabryPerot cavity
length was of 1.2 m. A translation stage was used for tuning the reflection band of FBG
2

closer to the short-wavelength band-edge of FBG
1
, allowing full overlapping between both
bands with moderate voltages being applied to the piezoelectric. In this arrangement, the
longitudinal standing elastic wave causes periodic optimization of the Q factor of the laser
cavity and, as a result, strong pulses are emitted at the repetition frequency of the acousto-
optic modulation, when the fiber is pumped. The CW features of the fiber laser are shown in
Fig. 14(a). The laser line was narrower than 20 pm (resolution limit of our measurement
system), a signal-to-noise level as high as 70 dB was obtained. The efficiency and the
threshold were 3 % and 25 mW, respectively. Figure 14(b) shows a pulse train emitted by
the laser at a repetition rate of 18 kHz. At high pump powers each Q-switched pulse breaks
into a train of narrower, 8 ns wide, sharp pulses (see inset). The frequency of this
modulation, 83 MHz, coincides with the frequency mode-spacing of the laser cavity, which
indicates beating between longitudinal cavity modes. We observed that the modulation
amplitude of this high frequency component was affected by variations in pump level and
repetition rate. This apparent passive mode-locking response has been previously reported
repeatedly in both, passively and actively Q-switched fiber lasers (Philippov et al., 2004;
Andersen et al., 2006). Figure 15(a) shows the peak power of the Q-switched pulse as a
function of the applied voltage to the piezoelectric. Three regions can be clearly
distinguished. For low voltages applied to the piezoelectric there is not emission, since the
wavelength shift of FBG
1
is not enough to overlap the reflection band to FBG
2
. As the
applied voltage increases, there is a Q-switched pulse per period. For voltages close to the
lower limit, the overlapping between both gratings is only partial and the laser is only a
small amount above threshold, so weak pulses are emitted. As the voltage is increased, the
gratings overlap further and the peak power increases. A steady state is reached when the
voltage amplitude is high enough to obtain complete overlapping between the two reflection
bands. If the applied voltage is increased further, the cavity stays in the high Q state longer


612
and more than one Q-switched pulse is emitted per period. When changing the pump power
and/or repetition rate, the upper and lower AC voltage limits vary, but the trend is preserved.
Finally, the effect of pump power on the Q-switched pulses, for repetition rates of 18 and 37
kHz, is shown in Fig. 15(b). The pump power threshold at each frequency was 27 mW and 53
mW, respectively. Above threshold, the peak power increases with pump power, and there is
a corresponding reduction of the pulse width. No evidences of peak power saturation were
observed within the pump power range of our experiments. Pulses of 1.6 W peak power and
172 ns width were obtained at 18 kHz repetition rate and 135 mW pump power.
3.2 Mode-locking and Q-switching mode-locking by acousto-optic interaction in a
fiber Bragg grating in the short-wavelength regime
In this subsection we discuss another mechanism developed by our group to mode-lock an
all-fiber laser (Cuadrado-Laborde et al., 2009a, 2010a). This is based on the acousto-optic
super-lattice modulation, and it is an example of the interaction of longitudinal acoustic
waves and Bragg gratings in the short-wavelength regime (Liu et al., 1997, 1998). In the
following, we start by reviewing the behavior of the acousto-optic super-lattice modulator,
subsection 3.2.1. Then, we show the use of this device as mode-locker, by showing two
different mode-locked lasers, either by driving the acousto-optic modulator by standing or
travelling acoustic waves, subsections 3.2.2 and 3.2.3, respectively. The possibility to reach
simultaneous Q-switching and mode-locking is also discussed in these subsections.
3.2.1 The fiber Bragg grating based acousto-optic modulator
The spectral response of the original FBG in this acoustical regime shows new and narrow
reflection bands symmetrically at both sides of the original Bragg wavelength (Liu et al.,
1997, 1998). The position of these sidebands can be controlled by varying the frequency at a
slope of 0.15 nm /MHz and 0.30 nm /MHz for the first and second order sideband,
respectively. The strength of the reflection bands, on the other hand, can be controlled
independently by varying the voltage applied to the piezoelectric. Since these sidebands can
be regarded as weak ghosts of the strong permanent Bragg grating, its FWHM bandwidth is
that of a weak Bragg grating of the same length (Liu et al., 1997). Figure 16 shows the setup
for a typical reflectivity measurement on an AOSLM. The AOSLM in turn is composed of an
RF source, an electrical RF amplifier, a piezoelectric disk, a silica horn, and a FBG. The tip of
the silica horn was reduced by chemical etching to the same diameter of FBG 125 m and
subsequently fusion-spliced to the fiber. The uniform and non-apodized grating was written
in photosensitive fiber using a doubled argon laser and a uniform period mask; the FBG was
120 mm long. The reflection properties of the AOSLM were investigated by illuminating the
FBG through an optical circulator with a broadband light source, and detecting the reflected
light with an optical spectrum analyzer (OSA). Figure 17(a) shows the reflectivity of the
unperturbed FBG i.e. without electrical signal applied to the piezoelectric and with an
electrical signal applied to the piezoelectric of 4.55 MHz and 16 V (whenever we refer to
voltages throughout this chapter, it is a peak-to-peak measurement). The presence of the
sidebands symmetrically positioned around the Bragg wavelength is clearly discernible.
Further, these sidebands are produced either by standing or travelling acoustic waves.
However, the light reflected by the sidebands in each case behaves differently. When
travelling acoustic waves are used by dumping the end of the FBG opposite to the silica
horn for example with a drop of oil, the light reflected on these sidebands is completely
downshifted or upshifted by the frequency of the acoustical signal, depending if the


613
reflection was in the long or short-wavelength sideband, respectively (Liu et al., 1997, 1998).
Together with this and as a function of the instantaneous phase of the acoustic signal, there
is also present an amplitude modulation at the frequency of the acoustical signal. Figure
17(b) shows the measurement performed by tuning a laser diode to the center of the short-
wavelength sideband and measuring the reflected light. As expected, the amplitude
modulation is at the same frequency of the electrical signal used to drive the piezoelectric.
On the other hand, when standing acoustic waves are used by clamping the end of the FBG
opposite to the silica horn the sidebands raise and fall at twice the frequency of the
electrical signal (Cuadrado-Laborde et al., 2009a). Further, in principle, for a perfect
acoustical reflection, the light reflected by the sidebands does not experiment any Doppler
shift, as opposed to the previous case when travelling acoustic waves are used. Figure 17(c)
shows the measurement performed by tuning a laser diode to the center of the short-
wavelength sideband and measuring the reflected light. Now the optical modulation
frequency is two times the acoustical signal frequency. To summarize, an AOSLM can be
driven in two different regimes, either by using travelling or standing acoustic waves. In
both cases we can use it as an amplitude modulator. However in the first case travelling
acoustic waves the light reflected on the sidebands is modulated at the same frequency of
the acoustical signal, whereas in the second case standing acoustic waves it is modulated
at two times this frequency, respectively.

Fig. 16. Setup for the characterization of the acousto-optic super-lattice modulator; OSA stands
for optical spectrum analyzer, FBG for fiber Bragg grating and RF for radio-frequency

Fig. 17. (a) Reflectivity of the fiber Bragg grating with and without electrical signal applied
to the piezoelectric (4.55 MHz and 16 V peak-to-peak). (b) Optical signal reflected by the
short-wavelength sideband when travelling acoustic waves are used (lower trace) and RF
voltage applied to the piezoelectric (4.11 MHz and 16 V, upper trace). (c) Same as in (b), but
for standing acoustic waves (4.55 MHz and 16 V, dotted curve)


614

Fig. 18. Setup of the mode-locked all-fiber laser by using an acousto-optic supper-lattice
modulator (AOSLM) as mode-locker
3.2.2 All-fiber actively mode-locked laser with a fiber Bragg grating based acousto-
optic modulator driven by standing acoustic waves
Now, we show the use of the amplitude modulation induced by standing acoustic waves in
the AOSLM to mode-lock an all-fiber laser (Cuadrado-Laborde et al., 2009a, 2010a). The
setup proposed for the standard mode-locking laser is schematically illustrated in Fig. 18.
The gain was provided by an erbium-doped fiber (EDF) containing 300 parts per million
(ppm) of Er
3+
, with a cut-off wavelength of 939 nm, and a numerical aperture of 0.24. The
active fiber was pumped through a WDM coupler by a pigtailed laser diode emitting at 976
nm, providing a maximum pump power of 160 mW. The acousto-optic super-lattice
modulator and a short delay line followed by a second fiber Bragg grating FBG
2
were
fusion-spliced at each end of the active fiber. FBG
2
(10 mm long and with a Bragg
wavelength of 1530.2 nm) was written with a uniform period in a photosensitive fiber using
a doubled argon laser and a uniform period mask. The Fabry-Perot cavity was established
once the reflection band of FBG
2
is made to match the short wavelength sideband of the
FBG
1
of the AOSLM, i.e. 1530.5 nm, by straining it with a translational stage (see Fig. 18).
The delay line length must be selected to match the round-trip time with the reciprocal of
the optical modulation frequency, which in turn is twice the electrical frequency applied to
the piezoelectric. Since the selected piezoelectric operation point for the piezoelectric was
4.55 MHz, this results in a cavity length of 11.4 m. However, fine tuning of the electrical
frequency was required to achieve mode-locking.
Figure 19(a) exemplifies the laser behavior showing the train of optical pulses generated, at
a frequency rate of 9 MHz (50 GHz bandwidth oscilloscope). Figure 19(b) shows a single
optical pulse; the timing jitter was measured to be 40 ps (RMS). The pulses of this laser are
best fitted by a sech
2
function, which is in accordance with earlier amplitude modulated
mode-locking theory. Further, the dispersion measurements of the different fibers we used
to form the laser cavity show that both kind of dispersion coexist, from the anomalous
dispersion of the delay line to the normal dispersion of the EDF. For this type of dispersion-
mixed cavities, the analytical steady-state solution is expected to be also of the sech
2
type
(Belanger, 2005), which adds further evidence in the same direction. The emission linewidth
was measured using a high resolution optical spectrum analyzer (BOSA-C, Aragn
Photonics, resolution of 80 fm). Figure 19(c) shows the spectrum of the optical pulses shown
in Fig. 19(b). The high resolution of the optical spectrum analyzer permits a direct
observation of the individual cavity modes separated by 9 MHz, see the inset of Fig. 19(c).
The measured 3-dB linewidth results in 2.8 pm, i.e. 360 MHz at 1530.5 nm. On the other side,
the optical pulses have a temporal width of 780 ps, so according to the Fourier-transform-


615
limited relation for a sech
2
pulse, time-bandwidth product of 0.315 (Wada et al., 2008), its
bandwidth cannot be lower than 380 MHz. From the comparison between this last value
and the linewidth measurement, we conclude that the optical pulses of our modelocked
laser are transform-limited, i.e. un-chirped. A brief digression is in order here, the Fourier-
transform-limited relation for Gaussian pulses with a time-bandwidth product of 0.44
(Kuizenga & Siegman, 1970; Wada et al., 2008), would result in a much higher linewidth
(564 MHz), which is in contradiction with our spectral measurements. As a consequence, the
fitting of the temporal pulses, see Fig. 19(b), the application of the Fourier-transform-limited
relations, and dispersion measurements together with previous theoretical work, seems to
corroborate a sech
2
nature for the emitted pulses.

Fig. 19. (a) Voltage signal applied to the piezoelectric at 4.55 MHz and 16 V, and its
corresponding modelocked train of pulses generated at twice this frequency with a pump
power of 160 mW. (b) A single optical pulse and its corresponding fitting by a sech
2

function. (c) Optical spectrum of the output light pulses shown in (b); the inset shows the
cavity modes distant by 9 MHz

Fig. 20. (a) Peak power and pulse width of the optical pulses as a function of the pump
power. (b) Peak power and pulse width of the output light pulses as a function of the
voltage applied to the piezoelectric, for a pump power of 160 mW. (c) Pulse width of the
output light pulses, as a function of the EDF length, for a fixed pump power of 160mW,
additionally, the average dispersion of the cavity is also shown for each EDF length


616
Now we will show the variation in the pulse`s parameter as typical variables are changed.
First we analyze the optical pulses peak power and temporal width as a function of the
pump power, which is shown in Fig. 20(a). A smooth variation of the pulse width with
pump power can be observed, before reaching the gain saturation. A higher available pump
power would allow additional axial modes to contribute, increasing the spectral bandwidth,
and thereby decreasing the pulses width. Next, we analyzed the variation in the pulse`s
parameters as a function of the modulation voltage. The intensity of the reflection sidebands
of the AOSLM can be dynamically controlled by varying the voltage applied to the
piezoelectric as we explained before in subsection 3.2.1. This voltage controls the amplitude
of the propagating acoustic wave and the amplitude of the standing wave, as a function of
the reflection at the clamp. Thus, as opposed to mode-locking by bulk acousto-optic or
electro-optic modulators, here the reflectivity and the modulation depth are intrinsically
linked, i.e. when the applied voltage increases, both the reflectivity and the modulation
amplitude increase. Figure 20(b) shows the peak power and time width, respectively, as a
function of the voltage applied to the piezoelectric. At lower reflectivities and modulation
amplitudes, the peak power diminishes, whereas the time width increases. According to Eq.
(1), the pulse width and the modulation depth are inverse parameters, and in turn the later
rises with the applied voltage. However, the pulses shorten very slowly with the increased
modulation strength; hence this shortening procedure is generally discouraged (Kuizenga &
Siegman, 1970). One can conclude that further increase of the voltage might improve the
pulse parameters, but it is not likely to lead to a great enhancement in terms either of peak
power or temporal width. We analyzed also the influence that dispersion has on the pulse`s
parameters, when the average dispersion of the cavity is changed. The dispersion for each
type of optical fiber used in the setup was measured by the frequency-domain modulated-
carrier method. The resulting average dispersion of this cavity is normal and its value is 1.2
ps/nm/km. Next, the optical fiber of the delay line the Corning LEAF fiber was replaced
by a normal dispersion fiber Fibercore SM980 leading to a lower (normal) overall
dispersion of 5.8 ps/nm/km. With the new delay line, a higher pump reaches the EDF,
since a better effective area compatibility is insured throughout the different fibers of the
system in fact, unlike LEAF fiber, the SM980 insures single-mode propagation of the pump
power. As a result, output pulses have higher peak power and shorter time width than in
the previous configuration. In this way, optical pulses with temporal width and peak power
of 640 ps and 160 mW, respectively, were obtained. These pulse parameters represent an
improvement of 18% and 28% with respect to the previous configuration. Finally, in order to
reverse the sign of the average dispersion of the cavity, we used an SMF28 optical fiber for the
delay line, resulting in an average anomalous dispersion of 9.8 ps/nm/km. In this case, we
did not observe mode-locking lasing. One reason for this could be motivated by the large
average dispersion introduced within the cavity, since modelocked fiber lasers usually works
with close-to-zero average dispersion cavities. Finally, the change on the pulse parameters was
measured as a function of the EDF length ranging from 0.5 to 3.3 m, but keeping a constant
cavity length of 11.4 m, by adjusting each time the length of LEAF fiber. The average
dispersion remains normal regardless of the amount of EDF used, ranging from 0.3
ps/nm/km up to 4.5 ps/nm/km. The temporal pulse width as a function of the EDF length
is shown in Fig. 20(c), for a fixed pump power of 160 mW. This figure includes the average
dispersion of the cavity as a function of the EDF length. A direct relationship can be observed
between the EDF length, the dispersion, and the time width of the pulses, with a minimum of
630 ps obtained for an EDF length of 1 m. This kind of interplay between gain and time width


617
agrees with Eq. (1). Further narrowing of the optical pulses could not be reached by this trend,
since no mode-locking lasing was observed with 0.5 m of EDF and the available pump power,
as a result of insufficient cavity gain. Finally, we also developed the ytterbium-doped fiber
version of this laser, with equally satisfactory results (Villegas et al., 2011).

Fig. 21. (a) Q-switched mode-locked all-fiber laser setup when the AOSLM is driven by
standing acoustic waves. (b) Reflection spectrum of the AOSLM when an RF signal of 4.55
MHz and 16 V is applied to the piezoelectric (left ordinate). Transmittance of FBG
2
(solid
curve) when the magnetostrictive device is switched-off; the dotted curve shows the new
spectral position of FBG
2
when the magnetostrictive device is switched-on (right ordinate)
Until now we have discussed the standard mode-locking operation. Now we demonstrate
the possibility to simultaneously Q-switch and mode-lock, actively and independently. By
combining both operations in a single laser, a superior performance is achieved with higher
peak powers compared with ordinary mode-locked lasers, but almost the same pulse width
as in a mode-locked laser is retained. Thus, the peak power of the central pulses of the
mode-locked train, underneath the Q-switched envelope, can be greatly enhanced. The
increased peak power provided by the Q-switching technique can be advantageous in
applications such as wavelength conversion or super-continuum generation. It is worth to
say, that any passive mechanism is used in this setup, neither for Q-switching nor for mode-
locking. To this end, we design a solution for actively Q-switching, while simultaneously
mode-locking, preserving the all-fiber configuration. This solution is based on fiber Bragg
grating fast modulation using a magnetostrictive device. This feature provides a direct
control of the repetition rate and a fully-modulated train of modelocked pulses. To the best
of our knowledge, this was the first doubly active Q-switching mode-locking strictly all-
fiber laser presented (Cuadrado-Laborde et al., 2009b). The setup of our doubly-active Q-
switching mode-locking all-fiber laser is shown in Fig. 21(a). It is basically the same as it was
described before for the standard mode-locking and shown in Fig. 18, with the same length
of EDF and delay line, except by the magnetostrictive device controlling the FBG
2
, see Fig.
21(a). It is composed of a 15 mm long (1 mm
2
cross-section) magnetostrictive rod of
Terfenol-D bonded to the FBG
2
. The rod and the fiber were placed inside a small coil driven
by an electronic circuit designed to drive square current pulses with amplitudes up to 260
mA of any required duty cycle. Figure 21(b) shows the spectral positions of both FBGs i.e.
the FBG of the AOSLM and FBG
2
, which helps understand the operation principle of this
laser also. When the magnetic pulses generated in the solenoid stretch FBG
2
, its central
wavelength is brought to match the short-wavelength sideband of FBG
1
for a short period of
time, which results in an increased Q-value. In this way by modulating the coil current with
a given Q-switched frequency, the Q-factor is actively modulated at the same frequency. The


618
magnetostriction has a low pass frequency response and, consequently, presents the
advantage to permit a continuous tuning of both the Q-switched repetition rate and the duty
cycle of the modulation pulses. The dependence of the wavelength shift with the coil current
was measured by illuminating FBG
2
with a tunable laser source and detecting the reflected
light. A quasi-linear behavior was observed, as a result of using moderate magnetic fields.

Fig. 22. (a) Q-switched mode-locked train of pulses at a Q-switching frequency of 500 Hz. (b)
Single Q-switched pulse enveloping a train of 1214 modelocked pulses; the inset shows a
single modelocked pulse. (c) Single Q-switched pulse with no mode-locked. (d) Mode-
locking operation of the laser without Q-switching. In all cases, the pump power was 70mW.
(e) Energy of the Q-switched mode-locked pulses as a function of the pump power for a Q-
switching repetition rate of 500 Hz
Now we discuss the Q-switching mode-locking laser operation. The laser emission
wavelength was in 1530.55 nm, since the overlapping between the short-wavelength
sideband of FBG
1
and the shifted position of FBG
2
takes place at this wavelength, see Fig.
21(b). Figure 22(a) exemplifies the Q-switched mode-locked laser behavior showing the train
of optical pulses generated for a pump power of 70 mW. The optical Q-switched mode-
locked pulses are 2 ms apart, since the repetition rate of the train of voltage pulses applied to
the current driver was of 500 Hz. Figure 22(b) shows a single Q-switched envelope with a
FWHM of 550 ns; it has between 12 and 14 fully-modulated modelocked pulses, with the
expected temporal separation of 110 ns. However, the individual short pulses are broadened
due to the limited bandwidth and sampling rate of the oscilloscope. In order to measure the
FWHM of the individual short pulses, a 50 GHz sampling oscilloscope was used. In this
case, the large difference between the Q-switching and mode-locking frequencies make
difficult to trigger properly the oscilloscope. Even so, relatively good traces were recorded
as the one depicted in the inset of Fig. 22(b). As we discussed before when we presented the
standard mode-locking regime, the tuning sensitivity between the RF signal applied to the
piezoelectric and the cavity round trip is not too critical in this setup. Only when the
detuning is considerably e.g. by a few kHz, the pulse clearly deteriorates. Figure 22(c)
shows the Q-switched pulse obtained by avoiding the mode-locking pulse formation, just by
detuning the applied voltage to the piezoelectric by 10 kHz, in this way only the Q-
switching operation is allowed within the cavity. As expected, the Q-switched pulse


619
reproduces the waveform of the envelope of the Q-switched mode-locked pulses shown in
Fig. 22(b), but with a much lower peak power (by a factor of 410
-3
) and an increased
temporal width. The transition from the fully-modulated Q-switched mode-locked pulses to
the pure Q-switching operation i.e. from Fig. 22(b) to 22(c) is progressive as we detuned
the voltage applied to the piezoelectric. Finally, Fig. 22(d) shows the behavior of this laser
when turning off the active Q-switching, i.e. standard mode-locking operation of the laser.
To this end, a DC current was driven to the coil of the magnetostrictive device, in order to
achieve a stationary overlap with the short wavelength sideband of the AOSLM. The trace
of Fig. 22(d) was recorded with the 1 GHz oscilloscope. In this case, the thermal effects
produced by the DC current made difficult an optimum and stable adjustment of FBG
2
. The
comparison of Figs. 22(b) and 22(d), illustrates the improved performance of the laser in
terms of peak power when both Q-switching and mode-locking are operating. The energy of
a Q-switched train of pulses, as that reported in Fig. 22(b), as a function of the pump power
is shown in Fig. 22(e). This energy was measured directly with a pyroelectric detector. At
high pump powers, the energy of the Q-switched mode-locked train of pulses reach gain
saturation. Thus, a peak power higher than 250 W can be calculated for the central pulses of
a train with energy of 0.65 J, assuming that the mode-locked pulses are 1 ns width. This
result demonstrate a dramatic enhancement in comparison with the peak power achieved
when the laser was operated in standard mode-locking regime (Cuadrado-Laborde et al.,
2009a), since the ratio is higher than 210
3
.

Fig. 23. (a) Oscilloscope traces of the sinusoidal electrical signal applied to the piezoelectric at
4.1 MHz repetition rate and 16 V (upper trace), and the train of optical pulses generated (lower
trace); a pump power of 230 mW was used. The inset shows the detail of a single pulse of the
optical train, with a FWHM of 720 ps and a peak power of 500 mW. (b) Peak power and pulse
width (FWHM) of the optical pulses as a function of the pump power (solid and dotted curves,
respectively). (c) Pulse width of the optical pulses when the applied voltage to the piezoelectric
is varied, and when the frequency of the electrical signal is varied
3.2.3 All-fiber actively mode-locked laser with a fiber Bragg grating based acousto-
optic modulator driven by travelling acoustic waves
The setup of the mode-locking laser driven by travelling acoustic waves is essentially the
same described in Fig. 18, but with two important differences. First, the pigtailed laser diode
emitting at 976 nm provides now a higher maximum pump power of 410 mW. Second,
the modulation frequency is now at the same frequency of the acoustical signal (Cuadrado-
Laborde et al., 2010b), since now we use travelling acoustic waves, see Subsection 3.2.1.
Then, the delay line length is now much larger, because we must match the round-trip time


620
with this frequency, which is lower. Since the selected piezoelectric operation point was of
4.1 MHz; then it results in a cavity length of 25 m compare this with the 11.4 m of cavity
length used before. Apart from the 1.4 m of EDF, the cavity was constructed entirely with
Fibercore SM980 fiber, this result in an average dispersion of 4.9 ps/nm/km (i.e. normal).
Figure 23(a) exemplifies the laser behavior showing the sinusoidal electrical signal applied
to the piezoelectric at a frequency rate of 4.1 MHz and the train of optical pulses generated
at the same frequency. The inset shows a detail of a single optical pulse with a temporal
width (FWHM) of 710 ps. Once mode-locking was reached, the polarization controllers were
adjusted in order to obtain the minimum pulse width.
The emission linewidth was measured using a classical heterodyne configuration
(Galtarossa et al., 1993). To this end, a tunable laser was used as local oscillator, with a 100
kHz linewidth. The output of this laser was superimposed to the optical pulses of the
modelocked laser, through a 1550 nm 50/50 coupler. The beat signal at the coupler output
was detected with a 45 GHz bandwidth photodetector, and analyzed with a 2.5 GHz
oscilloscope. This results in a linewidth for the mode-locked laser of 560 MHz for a 900 ps
(FWHM) pulse, measured directly in the spectrum at 3 dB. Again, we found that the
output pulses of this laser are best fitted by a sech
2
function rather than with a Gaussian
function. From the Fourier-transform-limited relation for a sech
2
pulse, its time-bandwidth
product cannot be lower than 0.315 (FWHM) (Wada et al., 2008). Since we obtain 560 MHz
900 ps = 0.504, we conclude that the optical pulses of our modelocked laser could have some
moderate degree of chirp.
According to Eq. (1), the pulse duration is limited by a variety of factors in an active
modelocked laser. From these parameters, we believe the narrow spectral bandwidth of the
AOSLM plays a key role in this setup. As we explained before, these sidebands can be
regarded as weak ghosts of the strong permanent Bragg grating, its FWHM bandwidth is
that of a weak Bragg grating of the same length (Liu et al., 1997), namely = 1.39
2
/(Ln
0
),
where L is the fiber length and n
0
is the modal effective index. For the grating used in this
chapter, with L = 120 mm, at = 1530 nm, this translates into = 6 pm, which is equivalent
to 770 MHz at the operation wavelength of this laser (1530.5 nm). As a rough estimation,
since the cavity modes are distant 4.1 MHz, it is easily seen that this AOSLM only is able to
lock a few percent of the axial modes available by the medium gain, namely 770 MHz /4.1
MHz = 192 modes (FWHM). If we assume for the output pulses of this laser a sech
2

envelope, then this parameter alone determines a lower limit for the pulse width around
0.315 /770 MHz = 410 ps. Therefore, we believe that narrower pulses can be reached by
broadening the sidebands of the AOSLM in order to lock additional axial modes. The optical
pulses peak power and temporal width as a function of the pump power are shown in Fig.
23(b). A smooth variation of the pulse width with pump power can be observed, before
reaching the gain saturation. Fig 23(c) shows the time width as a function of the frequency
detuning. The behavior was asymmetric, i.e. it depends if the detuning was positive or
negative. Once mode-locking is reached and the PC adjusted to get the minimum pulse
width, a small detuning in either direction does not modify the pulse width e.g. up to a few
tens of Hertz, as expected for amplitude modulation mode-locking (Kuizenga & Siegman,
1970). However, when the detuning is considerably higher, a positive detuning
continuously broadens the time width, whereas for a negative detuning, the mode-locking is
rapidly lost, and the pulses drop out. We believe this asymmetry could be caused by the
non-flat frequency response of the piezoelectric. We conclude this subsection by analyzing
the behavior of this laser when the voltage applied to the piezoelectric changes. Figure 23(c)


621
shows the influence on the time width of the optical pulses with the voltage applied to the
piezoelectric. For higher voltages, the time width decreases; further narrowing by this trend
appears to finish when the reflectivity of the sideband approaches the maximum (in
addition, 20 V was the maximum voltage provided by the electrical amplifier used to drive the
piezoelectric). Within the range of voltage available in our experiments, both the reflectivity
and the modulation amplitude increase continuously. This is consistent with Eq. (1), by which
the time width is inversely related with both, the modulation depth and the reflectivity.

Fig. 24. (a) Voltage signal applied to the piezoelectric and Q-switched mode-locked train of
pulses generated for a 1 kHz Q-switching repetition rate; the inset shows a detail of the
burst signal. (b) Single Q-switched mode-locked pulses enveloping a train of modelocked
pulses for different burst cycles; the inset shows a single-shot capture of a single
modelocked pulse with a FWHM of 680 ps. (c) Energy of the Q-switched mode-locked
pulses as a function of the cycles contained in the burst for a fixed Q-switching frequency of
500 Hz and a pump power of 270 mW
The most significant advantage of using travelling instead of standing acoustic waves is
determined by the possibility to actively Q-switch this modelocked laser simply by
launching to the grating a burst acoustical wave (Cuadrado-Laborde et al. 2010b). Therefore,
the setup is exactly the same as the reported above for the standard mode-locking regime
driven by travelling acoustic waves. The burst signal consists of an integer number of
sinusoidal acoustical cycles N. Figure 24(a) shows the burst voltage signal applied to the
piezoelectric together with the Q-switched mode-locked train of pulses generated, at a
frequency rate of 1 kHz. Figure 24(b) shows several Q-switched mode-locked pulses as a
function of N. When N is low, Q-switched mode-locked is not reached. For higher N, Q-
switched mode-locked is reached and the emission is allowed. For example, the left bottom
corner of Fig. 24(b) shows a single Q-switching envelope with a FWHM of 1.4 s for N = 25;
it has between 16 and 18 fully-modulated modelocked pulses, with the expected temporal
separation given by 1/4.1 MHz = 244 ns. The inset shows a single-shot capture of a
modelocked pulse within this train with a temporal width of 680 ps. As observed, the Q-
switched mode-locked pulses of this laser have excellent inter-pulse characteristics. The
energy of a Q-switched mode-locked train of pulses, as a function of N is shown in Fig.
24(c). This energy was measured directly with a pyroelectric detector. As expected, for a
longer burst signal, the energy of the Q-switched mode-locked train of pulses increases,
until it reaches saturation. Thus, a peak power higher than 200 W can be calculated for the
central pulses of a train with energy of 0.68 J, assuming that the modelocked pulses are 680
ps width. This result clearly demonstrates the degree of enhancement in comparison with


622
the peak power achieved when the laser was operated in the standard mode-locking regime,
since the ratio is higher than 410
2
.
3.3 Photonic true-time delay-line by acousto-optic interaction in a fiber Bragg grating
in the short-wavelength regime
A photonic true-time-delay (PTTD) line is an optical device that permits to vary the group
delay of an optical signal. The ideal PTTD line should exhibit fast tuning within a broad
range of time delay values. PTTD lines attract much attention due to an increasing number
of applications such as the control of phased array antennas (Liu et al., 2002; Perez-Milln et
al., 2004; Italia et al., 2005; ), tunable microwave band-pass filters (Capmany et al., 2005),
buffering and packet synchronization (Li et al., 2007; Caucheteur et al., 2010). The
advantages of PTTD lines are their broad bandwidth, high frequency operation and
immunity to electromagnetic interference. In addition, fiber-based PTTD lines are readily
compatible with fiber systems, robust, compact and lightweight, and they present low
insertion loss.
Recently several types of fiber-based PTTD lines were proposed. For example, an optically
controlled PTTD line based on stimulated Brillouin scattering demonstrates a fast time
response within a relatively small range of the time delay (230 ps were reported using a
fiber of 3.5 km length) (Zadok et al., 2007). Another example is the tunable PTTD line based
on a uniform or chirped FBGs perturbed by heating or mechanical stress (Ortega et al., 2000;
Liu et al., 2002; Perez-Milln et al., 2004), which permits simple experimental realization and
easy tuning. However, such system has a very slow response (below 1 ms). In a number of
PTTD lines tunable light sources are required to adjust the time delay (Liu et al., 2003; Blais
et al., 2009), but this approach is not easily scalable when a large number of independent
delays need to be adjusted.
Here we discuss a tunable PTTD line based on longitudinal ultra-sound modulation of a
FBG written in a standard fiber similar to SMF-28. This device operates with a fixed optical
carrier, demonstrates a wide time delay range (400 ps) and a fast response time [s range,
see Delgado-Pinar et al. (2006)].

Fig. 25. Experimental setup. SL: semiconductor laser; PC: polarization controller; EOM:
electro-optical amplitude modulator; FBG: fiber Bragg grating; PZT: piezoelectric
transducer; SH: fused silica horn; PD: photodetector
The setup of the PTTD line is shown in Fig. 25. Light from a single-frequency semiconductor
laser (SL) modulated by an electro-optical amplitude modulator (EOM) with a frequency of f
m

= 400 MHz is sent through the optical circulator to a long uniform FBG written in a standard
photosensitive fiber; the RF modulation depth is chosen to be comparatively low (~ 10%),
which maintains a practically linear regime of light modulation. The FBG is 12 cm long, with a


623
Bragg wavelength of 1530.8 nm and reflectivity of 99.99% (40 dB transmission attenuation
measured at the Bragg wavelength) that corresponds to a coupling coefficient = 0.442 cm
-1
.
In the experimental arrangement, a longitudinal acoustic wave is generated by a piezoelectic
(PZT) disk attached to a heat sink. This wave is launched along the single-mode optical fiber
containing the uniform FBG using a fused silica horn (SH). The PZT is driven by Generator 1
at the PZT resonance frequency (2.08 MHz in our case). The light reflected by the FBG is
registered by a photodetector connected to a vector voltmeter. AC voltage from Generator 2
drives the EOM; it is also used as a reference signal for the RF phase measurements.
The operation principle of such a PTTD line is based on the reflection of narrow-line light by
one of the FBG side bands generated by the longitudinal acoustical wave. This wave
modulates both the period and the refractive index of the grating producing thereby a
sinusoidal chirp that moves along the fiber synchronously with the acoustic wave, and will
be described by an effective phase modulation of the grating with amplitude . Under
these conditions, the refractive index for the light mode propagating through the single-
mode fiber core (so called LP
01
mode) is written as follows:

( ) ( )
( ) ( )
( ) ( )( )
0 0 0
2
1
1
2 1
1
( , ) cos cos ( )cos
2 cos ( 1) ( )cos 2
2 sin ( 1) ( )cos 2 1 ,
s
m
m s
m
m
m s
m
n z t n n Kz k z t n nJ Kz
n Kz J m k z t
n Kz J m k z t
=
= + + = +

+ +

+ +

(3)
where n
0
is the effective refractive index for LP
01
mode, n is the grating amplitude, K = 2/
is the unperturbed grating wavenumber ( is the grating period), z is the distance along the
fiber, is the phase modulation amplitude, k
s
is the acoustic wavenumber, is the angular
acoustic frequency, m is a natural number corresponding to the order of the FBG side band,
and J
m
is the Bessel function of the m-th order. From Eq. 3 one can see that the longitudinal
acoustic wave produces a series of the traveling FBGs with amplitudes depending on .
Using the coupled-wave theory (Erdogan, 1997), one can obtain the following set of the
coupled-wave equations for two contra-propagating optical waves (one probe wave, A
l
,
and one reflected wave, B
l
) falling into the l sideband induced by the acoustic wave and
interacting with the perturbed grating described by Eq. 3:

( 1) l l
l l l l
dA
i A i B
dz

= + , (4)

( 1)
( )
l l
l l l l
dB
i B i A
dz

= + , (5)
where l is the sideband number,
l
= (
0
lk
s
/2) = 2n
0
(
1

0
1
) l/
s
is the
detuning from the l-sideband Bragg wavelength,
0
is the Bragg wavelength of the
unperturbed FBG, is the probe light wavelength, and
0
are the light wavenumbers that
correspond to and
0
, and
l
=
0
J
l
() is the coupling coefficient for the l-th sideband (
0

is the coupling coefficient of the unperturbed FBG).
The analysis based on the theory developed by Barmenkov et al. (2006, 2010) permits one to
write the equations for the sideband reflectivity, R
l
, and the sideband effective length,
eff
l
L
,
in the following form:


624
( ) ( )
2
tanh
l l
R L

= , (6)
( )
( )
1
tanh
2
2atanh
eff l
l l
l
l
L R
L L
R

= = , (7)
where L is the FBG length. The formula for the group delay is found as

0
2
eff
l l
n L c

= . (8)
From the last three equations one can conclude that the sideband diffraction efficiency, the
effective length and the group delay may be controlled by the amplitude of the phase
modulation of the grating induced by the acoustic wave, which, in turn, depends on the
acoustic wave magnitude. The sideband Bragg wavelength can be tuned by adjusting the
acoustical frequency within the PZT resonance in the case of a slight tuning, or by replacing
the PZT with another one having the necessary resonance frequency.
The diffraction efficiency of the FBG sidebands can be measured using a wide-spectrum
LED connected directly to the circulator port 1, and an optical spectrum analyzer (OSA)
connected, in turn, to the port 3 instead of a photodetector (see Fig. 25); The Generator 2 is
switched off. Fig. 26(a) shows the FBG spectra at different AC voltages applied to the PZT.
At low voltages, the 1 sidebands have amplitudes much higher than that of the high-order
sidebands. Reflectivity of the 1 sidebands did not reach 100% because the OSA resolution
was not enough to resolve comparatively narrow reflection peaks (3 dB spectrum width is
about 40 pm). Note that a proper FBG design based on the grating apodization and chirping
should permit increasing the sideband width.
As it is seen from Fig. 26 (b), the dependence of the +1 sideband efficiency on the voltage
applied to the PZT is in good agreement with Eq. (6), assuming that the amplitude of the
phase modulation of the grating is proportional to the PZT voltage; a slight difference
observed at high voltage amplitudes (> 4V) could be explained by a small heating of the
PZT, which slightly changes the PZT electromechanical coupling factor. Note that the 1
sideband allows operation of AOSLM at low PZT voltages in comparison with high-order
sidebands, the 1 sideband has the same properties as the +1 sideband.
The RF modulation envelope phase,
+1
, is related directly to the +1 sideband group delay
+1
and the effective length
1
eff
L
+
:

1 1 0 1
2 4
eff
m m
f n L f c
+ + +
= = (9)
Thus, the dependence of
+1
on +1 sideband efficiency is the basic feature that permits to
implement a dynamic control of the group delay and the effective length of the grating by
means of the AC voltage. The same is for the 1 sideband.
Fig. 27(a) shows the experimental relationship between
+1
and the amplitude of the AC
voltage applied to the PZT. One can see that the phase decreases as voltage amplitude
increases, which is explained by decreasing of the group delay and the effective length.
Figure 27(b) plots these two parameters versus the +1 sideband diffraction efficiency. The
experimental data show a good agreement with Eqs. (7) and (8). Since the optical path
between the EOM and the photodetector was relatively long (the fiber length was ~ 2 m),
producing an additional phase shift into the measured
+1
values, the experimental data
were corrected by a phase offset (the right scale in Fig. 27(a) that permits to compensate the


625
fiber length excess. The correcting parameter was chosen so that the +1 sideband effective
length is equal to a half of FBG physical length at low sideband efficiency (low PZT voltage),
as the Eq. (7) predicts.

Fig. 26. (a) Spectra of FBG perturbed by ultra-sound wave at different PZT voltages (the
voltage values are shown in the upper right corner); l indicates the sideband number. The
OSA resolution is 20 pm. (b) Dependence of the +1 sideband efficiency on PZT voltage.
Circles: experimental data; solid line: fitting
From Fig. 27(b) one can conclude that the group delay and the effective length for the +1
sideband depend on its efficiency that is controlled electrically by the AC voltage applied to
the PZT. In the experiments, the group delay was adjustable from 150 ps to 550 ps, i.e.,
covering a range of 400 ps. Within the 3 dB range of the sideband efficiency, the group delay
could be continuously adjusted from 150 to 450 ps (a range of 300 ps). The implementation
of an automatic SL power control would permit to adjust the group delay at constant optical
power, compensating for the reflectivity changes of the sideband. The response time of this
PTTD line is approximately 20 s and is determined by the acoustic wave speed and the
grating length.

Fig. 27. (a) Dependence of the RF modulation envelope phase of light reflected from the FBG
+1 sideband on amplitude of voltage applied to PZT. The left scale: uncorrected values, the
right scale: corrected values. (b) FBG +1 sideband effective length and group delay versus
the diffraction efficiency. Symbols: experimental data; solid line: curve calculated by Eq. (7).
In both figures different symbols correspond to different experimental series
0 1 2 3 4 5 6 7 8
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
U
n
c
o
r
r
e
c
t
e
d

+
1
,

r
a
d
.
C
o
r
r
e
c
t
e
d

+
1
,

r
a
d
.

(a)
PZT voltage, V
0.0 0.2 0.4 0.6 0.8 1.0
0
1
2
3
4
5
6
7
(b)
E
f
f
e
c
t
i
v
e

l
e
n
g
t
h
,

c
m
+1 sideband efficiency
0.0 0.2 0.4 0.6 0.8 1.0
0
100
200
300
400
500
600
G
r
o
u
p

d
e
l
a
y
,

p
s


626
In summary, the experimental and theoretical study of the phase and group delay response
of a FBG modulated by a longitudinal acoustic wave permits the implementation of
electrically-tuned PTTD line controlled by the AC voltage applied to the piezo-electric
transducer that generates the acoustic wave. The proposed PTTD line permits to vary the
group delay in the range of 400 ps.
3.4 Q-switching in a distributed feedback fiber laser
When the acoustic perturbation is not a harmonic wave but a single pulse, its passage
through the FBG creates a defect which can be used to control the Q-factor in a DFB all-fiber
laser. Single-mode narrow-linewidth lasers are of great interest as they have a wide
applicability in fields such as high-resolution interferometry, distributed Brillouin sensing,
optical coherent communications, etc. DFB lasers based on semiconductor technology have
linewidths typically in the MHz domain; due to the short cavity length and low thermal
dissipation its output spectrum is typically broad. Fiber lasers usually operate in a multi-
mode regime and have a broad spectrum; to make efficient single-mode narrow-linewidth
fiber lasers different approaches have been implemented. Fiber rings lasers fulfill the
requirement and produce narrow-linewidth outputs; most importantly, they can be made
widely tunable. However, they are more complex and because of their long cavities
susceptible to mode hopping. Distributed Bragg reflector (DBR) fiber lasers is another
option, however, temperature stabilization is still required to prevent mode hopping and, in
the short laser cavity, the pump wave absorption is low, as a consequence the DBR fiber
laser is not efficient and external optical amplifiers are generally needed (Babin et al., 2007).
Distributed feedback fiber lasers can overcome part of these problems. They have the
simplest, robust and compact design providing operation without mode hopping. Its
fabrication is relatively simple, and involves the writing of a grating structure i.e. a fiber
Bragg grating with ultraviolet light into an active fiber. Single-mode pump leads to an
alignment free resonator with optimum overlap of pump and signal light. For these FBG-
based DFB lasers, the distributed reflection occurs in the grating when a phase shift has been
generated within it. A number of techniques have been proposed for this; however, statics
phase shifts only allow CW operation, which translates into low power emission. Recently,
some approaches have been reported to obtain single-frequency pulsed all-fiber lasers,
based on active Q-switching of DFB fiber cavities. The pulsed operation of a DFB laser is
interesting because for certain applications (e.g. Brillouin sensing), it is not only required a
narrow linewidth but also a peak power about a minimum threshold, which otherwise
could not be reached by a CW operation (Cuadrado-Laborde et al., 2008).
Figure 28(a) shows the scheme of the proposed DFB all-fiber laser. The FBG was 100 mm
long, and was written in a 1500 ppm erbium hydrogen-loaded fiber (codoped with
germanium and aluminum) of the same length using a doubled argon laser and a uniform
period mask. The FBG shows more than 30 dB attenuation at the Bragg wavelength of
1532.45 nm and a 3 dB bandwidth of 88 pm. The FBG was pumped through a 980/1550 nm
WDM with a 980 nm semiconductor laser, providing a maximum pump power of 130 mW.
A square shape rod of a magnetostrictive material (Terfenol-D, 15 mm long and 1 mm
2

section) was bonded outside the FBG to a free section of fiber at 88 mm from the center of
the grating, and placed inside a small coil, see Fig. 28(a).
The acoustic pulse generated by using a magnetostrictive device has a superior performance
as compared with piezoelectric devices which have a frequency-dependent response


627
characterized by strong mechanical resonances. On the contrary, a magnetostrictive device
has a lower frequency range but its frequency response is basically flat. The
magnetostrictive rod can be bonded directly over the FBG, but this has some detrimental
consequences. First, any external actuator bonded directly on the fiber is likely to exhibit
long-term instabilities (Andrs et al., 2008). Second, once the magnetostrictive rod is fixed to
the fiber, any temperature change generates a differential expansion between the fiber and
the magnetostrictive rod, producing a local static perturbation of the FBG that may cause
CW emission. Because of this, we avoided these drawbacks by attaching he magnetostrictive
rod outside the FBG and generating a dynamic defect in the FBG through an acoustic
axially-propagating pulse. Thereby, when a pulse of electric current is applied to the coil,
the small rod lengthens and stretches the section of fiber attached to it, generating in this
way a longitudinal acoustic pulse that propagates toward the FBG. The pulse propagating
along the grating generates a phase shift opening a transmission peak within the reflection
band of the grating; as a consequence a high Q resonance is produced and a laser pulse is
emitted (Prez-Milln et al., 2005b; Delgado-Pinar et al., 2007). Otherwise, if no perturbation
is present within the FBG, there is no efficient feedback for the optical signal, and the laser
emission is not allowed. Since the defect is not fixed, but induced by a travelling acoustic
pulse with a time-varying position along the FGB, one could think that this might have
important consequences on the spectral position of the transmission peak. However, this is
not the case, as it was recently demonstrated (Andrs et al., 2008). The spectral position of
the resonance is constant, no matter the actual spatial location of the acoustic pulse,
although some short transients are produced when the pulse overlaps the extremes of the
FBG. This property insures that the laser will emit with a narrow linewidth, preserving one
of the most attractive properties of DFB fiber lasers. The transmission properties of the
passive FBG i.e. without pumping interacting with the acoustic pulses were investigated
by illuminating the FBG with a tunable laser at the Bragg wavelength and detecting the
reflected signal. When the coil current is zero, the reflectance is maximal; but when a current
pulse of 220 mA and 5 s temporal width is applied to the coil, a transmission peak opens
the reflection band, being as narrow as 2 s (FWHM), see Fig. 28(b). Figures 29(a) to 29(c)
shows the voltage pulses applied to the current driver, the backward output train of the
DFB laser, a detail of the optical pulse, and a single current pulse applied to the coil,
respectively. The observed delay time between the voltage signal and the optical pulses is
mainly due to the distance that pulses have to travel from the magnetostrictive device to the
FBG, see Fig. 28(a).

Fig. 28. (a) Q-switched distributed feedback-fiber laser setup. (b) Reflection at the Bragg
wavelength when an acoustic pulse travels along the FBG


628

Fig. 29. Q-switched DFB behavior at 10 kHz repetition rate and 80 mW of pump power. (a)
Emitted optical train pulses and voltage signal applied to the current driver, (b) detail of a
single optical pulse of the train, and (c) detail of a single current pulse applied to the coil

Fig. 30. Peak power and pulse width for backward and forward outputs as a function of the
coil current for two different Q-switched repetition rates: 500 Hz (a, b) and 2 kHz (c, d),
respectively, and 55mW of pump power
One key characteristic in this setup is its versatility, which is given by the possibility of
selecting a variety of peak powers and time widths just by varying the coil current. Figure
30 shows the peak power and pulse width for backward and forward outputs as a function
of the coil current for two different repetition rates: 500 Hz and 2 kHz, Figs. 30(a-b) and 30(c-


629
d), respectively. There is a low threshold value for the coil current (I), bellow this value; there
is not laser emission, since the Q value is not high enough (i.e. I < 150 mA). Above this value,
the laser emission is allowed and one optical pulse per cycle is emitted; at higher electric
currents (i.e., I > 230 mA) there are two optical pulses per current pulse. At different repetition
rates the peak power and time width values change but the general trend is preserved. The
effect of pump power on the optical pulses, for repetition rates ranging from 200 Hz to 20 kHz,
is shown in Figs. 31(a,b) and 31(c,d), for the backward and forward outputs, respectively. A
fixed coil current of 200 mA amplitude was used in all cases. At each different frequency, there
is a corresponding pump power threshold. Above threshold, the peak power increases with
pump power and there is a corresponding reduction of the pulse width. Pulses of 800 mW
peak power and 32 ns time width were obtained at 500 Hz repetition rate and 46 mW pump
power for the backward output. If pump power is high enough, the laser emits more than one
pulse per cycle, defining in this way an upper limit also. It can be observed the differences in
peak powers between backward and forward outputs, being higher in the first case. Temporal
widths and peak power jitters were measured to be below 5 %.

Fig. 31. Peak power and pulse width for the backward (a, b) and forward (c, d) laser
outputs, as a function of the pump power, for several repetition rates
The emission linewidth was measured, using a classical heterodyne configuration (Galtarossa
et al., 1993). A tunable laser was used as local oscillator with spectral linewidth of 100 kHz. The
output of this laser was superimposed to the optical pulses of the DFB laser (backward
output), through a 1550 nm 20 dB coupler. The beat signal at the coupler output was detected
with a 1 GHz bandwidth photodetector, and analyzed with a 500 MHz bandwidth
oscilloscope. From the spectrum of the beating signals, resulted a DFB laser linewidth of 6
MHz. The optical pulses from the DFB all-fiber laser had a temporal width of 80 ns, so
according to the time-frequency uncertainty principle (Agrawal, 2001), its bandwidth cannot
be lower than 5.5 MHz. From the comparison between this last value and the linewidth
measurement, we conclude that the optical pulses of our DFB fiber laser are transform-limited.
Finally, as an example of application of this compact laser source, we demonstrate the
possibility to generate Brillouin scattering for sensing purposes. Brillouin scattering


630
essentially refers to the scattering of a light wave by an acoustic wave (Agrawal, 2001).
When a coherent pulse of light propagates through a medium, part of its energy is
backscattered due to a non elastic interaction with the acoustic phonons. This back-scattered
light is composed of a frequency down-shifted Stokes light and an up-shifted anti-Stokes
light, whose spectral positions are dependent on temperature and strain of the fiber, in this
way allowing its use as a sensing mechanism (Culverhouse et al., 1989a, 1989b; Bao et al.,
1995; Parker et al., 1998; Y. Li et al., 2003). Figure 32(b) shows the backscattered light
spectrum after illuminating a 10.5 km length optical fiber spool (Corning SMF-28) with the
backward output of our DFB all-fiber laser (4 kHz repetition rate and 74 mW pump power).
The Brillouin spectrum was registered with an optical spectral analyzer (resolution of 20
pm). The extreme of the fiber optic spool was terminated with a matching refractive index
liquid (n = 1.46). The central (highest) peak corresponds to the proper laser beam reflections
after successive connections and splices together with Rayleigh scattering. Peaks
symmetrically positioned at both sides correspond to the Brillouin backscattering by Stokes
and anti-Stokes processes (Parker et al., 1998). The measured Brillouin shift results in 88 pm
(i.e. 11.24 GHz at 1532.4 nm), which correspond to the expected value in this fiber (Agrawal,
2001). Fiber optics distributed temperature, and/or strain, sensors have becoming very
attractive for applications requiring sensing lengths of many kilometers, principally due to
its inexpensiveness and availability. Optical fiber based distributed sensor systems normally
make use of the principle of optical time domain reflectometry (Parker et al., 1998).
Therefore, an optical pulse is launched into one end of the fiber system and the variation of
the scattered light is detected as a function of time, giving in this way information of
temperature or strain as a function of distance. A key requirement in this measurement
system is a stable light source with a narrow enough spectral linewidth. In addition, for time
domain reflectometry applications, the sensor spatial resolution proportionally depends on
the optical pulse width, so it must be considered also. In order to fulfill all these
requirements, solid state lasers with external cavities plus amplifiers and amplitude
modulators are currently used in these systems (Parker et al., 1998). Here we propose using
this compact all-fiber pulsed light source as a relatively simpler alternative.

Fig. 32. Schematic diagram of the setup used for the Brillouin backscattering measurements.
Brillouin spectra at room temperature for a 10.5-km length Corning SMF-28 optical fiber
spool, for a backward output at 4 kHz repetition rate and 74 mW pump power


631
4. Conclusions
In this chapter we have discussed the use of acoustic waves to control all-fiber devices for
different applications. In Section 2 we focused on novel applications of acousto-optic fiber
devices based on flexural acoustic waves. As an example of application, we described in
subsection 2.1 an actively Q-switched ytterbium-doped strictly all-fiber laser. Q-switching
modulation was achieved by intermodal modulation induced by flexural acoustic waves
travelling in a tapered optical fiber. Q-switched light pulses at 1064.1 nm were successfully
obtained at repetition rates in the range 1-10 kHz, with pump powers between 59 mW and
88 mW. Best results were for laser pulses of 118 mW peak power, 1.8 s of time width, with
a pump power of 79 mW, at 7 kHz repetition rate.
Next, in subsection 2.2 we described an actively mode-locked fiber ring laser. As mode-locker
was proposed an acousto-optic modulator driven by standing flexural acoustic waves, which
couples core-to-cladding modes in a standard single-mode optical fiber. Among the
remarkable features of the modulator, we mention its high modulation depth (72%), broad
bandwidth (187 GHz), easy tunability in the optical wavelength, and low insertion losses (0.7
dB). The narrowest optical pulses obtained were of 95 ps time width, 21 mW peak power,
repetition rate of 4.758 MHz, and 110 mW of pump power. We also characterized this laser as
a function of the RF voltage that controls the modulator, the length of the active fiber, and the
optical bandwidth of an intracavity filter implemented with a fiber Bragg grating. Best results
were for output light pulses of 34 ps time width, and 1.4 W peak power.
In Section 3 we focused on novel applications of acousto-optic devices based on the interaction
of longitudinal acoustic waves with fiber Bragg gratings. From the comparison of the acoustic
wavelength with the periodicity of the FBG, two well-different situations could be
distinguished: the long-wavelength and the short-wavelength regimes. As an example of
application of the former, we described in subsection 3.1 an in-fiber resonant acousto-optic
modulator suitable for Q-switching applications. The modulator consists of a short-length FBG
modulated by a long-wavelength standing longitudinal acoustic wave. The acoustic wave
shifts periodically in time the reflection band of the grating along a given wavelength range;
by using this modulator, we demonstrated an actively Q-switched all-fiber laser. Output light
pulses of 1.6 W peak power and 172 ns time width were obtained at 18 kHz repetition rate.
Next, in section 3.2 and 3.3 we focused on applications of acousto-optic devices based on the
interaction of longitudinal acoustic waves with fiber Bragg gratings in the short-wavelength
regime. Thus, in subsection 3.2 we proposed the use of an acousto-optic supper-lattice
modulator as mode-locker. Since an acousto-optic supper-lattice modulator can be driven in
two different ways either by standing or traveling acoustic waves; we discussed first the
construction of a mode-locked laser driven by standing acoustic waves. In this configuration
we obtained transform-limited optical light pulses of up to 120 mW peak power and 780 ps
pulse width generated at a fixed repetition rate of 9 MHz, with an emission linewidth of 2.8
pm at 1530.5 nm. We also study the influence of different parameters on the mode-locking
process such as: frequency detuning, EDF length, amplitude modulation, and dispersion. In
this case, narrower pulses were obtained at higher modulation depths, normal dispersion,
and shorter lengths of active fiber. Best results were reached when the laser was optimized
according to these variables (160mW peak power and 630 ps pulse width). On the other
hand, we showed that by slightly modifying the cavity, it is possible to operate the laser in a
double-active Q-switching mode-locking regime. It is worth to mention that this approach
was unique, being this double-active all-fiber laser the first of its kind. To this end, we
attached a magnetostrictive rod to the output FBG to modulate the Q factor of the Fabry


632
Perot cavity. Fully modulated Q-switched mode-locked trains of optical pulses were
obtained for a wide range of pump powers and repetition rates. For a Q-switched repetition
rate of 500 Hz and a pump power of 100 mW, the laser generates trains of 1214 modelocked
pulses of about 1 ns each, within an envelope of 550 ns, an overall energy of 0.65 J, and a
peak power higher than 250 W for the central pulses of the train. Then, we discussed the
construction of a mode-locked laser when the mode-locker is driven by travelling acoustic
waves. In this case, the modulation frequency is half the frequency obtained when standing
acoustic waves are used. Optical pulses were obtained of 530 mW peak power, 700 ps pulse
width, at a repetition rate of 4.1 MHz. The variation of the pulses parameters under
frequency detuning and applied voltage was also studied. Finally, we demonstrated that it
is not necessary to modify the setup in order to reach double active Q-switching and mode-
locking, when travelling acoustic waves were used to drive the mode-locker. In this case the
commutation between mode-locking and Q-switching mode-locking is remarkably simple; it
just needs use of a different electrical signal to drive the piezoelectric of the mode-locker, i.e.
from a sinusoidal to a burst-sinusoidal electrical signal. In this case, fully modulated 1025
modelocked pulses around 700 ps each within a Q-switching envelope around 1 s and a
maximum overall energy of 0.68 J were obtained.
In subsection 3.3 we shown another example of application of acousto-optic devices based
on the interaction of longitudinal acoustic waves with fiber Bragg gratings in the short-
wavelength regime. Thus, we carried out an experimental and theoretical study of the phase
and group delay response of the acousto-optic supper-lattice modulator. The phase
properties of the first sidebands permit the implementation of electrically-tuned photonic
true-time delay line controlled by the AC voltage applied to the piezoelectric transducer that
generates the acoustic wave. The proposed photonic true-time delay line permits to vary the
group delay up to 400 ps.
Finally, when the acoustic perturbation is not a harmonic wave but a single pulse, its
passage through the FBG creates a defect which can be used to control the Q-factor in a DFB
all-fiber laser, subsection 3.4. Thus, we shown a single-mode, transform-limited, actively Q-
switched distributed-feedback fiber laser. Optical pulses of 800 mW peak power, 32 ns
temporal width, and up to 20 kHz repetition rates were obtained. The measured linewidth
demonstrates that these pulses are transform limited: 6 MHz for a train of pulses of 10 kHz
repetition rate, 80 ns temporal width, and 60 mW peak power. Efficient excitation of
spontaneous Brillouin scattering was demonstrated.
In summary, photonic devices can benefit highly of a strictly all-fiber configuration which
provides them a series of attractive advantages. Among all the proposed in-fiber solutions,
devices controlled by acoustic waves have been by far the most employed, especially in
mode-locked lasers, providing a broad range of alternatives. The recent advances in
acoustically controlled photonic systems positioned them as a promising candidate for
commercially available systems in the near future.
5. Acknowledgments
This work has been financially supported by the Ministerio de Ciencia e Innovacin and the
Generalitat Valenciana of Spain (projects TEC2008-05490 and PROMETEO/2009/077,
respectively). C. Cuadrado-Laborde acknowledges the Secretara de Estado de Universidades e
Investigacin del Ministerio de Ciencia e Innovacin (Spain) and ANPCyT (project PICT 2008-
1506, Argentina).


633
6. References
Agrawal, G. P. (2001). Nonlinear Fiber Optics, Academic Press, New York, 2001.
lvarez-Chvez, J. A.; Offerhaus, H. L.; Nilsson, J.; Turner, P. W.; Clarkson, W. A.;
Richardson, D. J. (2000). High-energy, high-power ytterbium-doped Q-switched
fiber laser. Optics Letters, 25 (1): 37-39 Jan 1 2000.
Andersen, T. V.; Prez-Milln, P.; Keiding, S. R.; Agger, S.; Duchowicz, R.; Andrs, M. V.
(2006). All-fiber actively Q-switched Yb-doped laser. Optics Communications, 260 (1):
251-256 Apr 1 2006.
Andres, M. V.; Cruz, J. L.; Dez, A.; Prez-Milln, P.; Delgado-Pinar, M. (2008). Actively Q-
switched all-fiber lasers. Laser Physics Letters, 5 (2): 93-99 Feb 2008.
Babin, S. A.; Churkin, D. V.; Ismagulov, A. E.; Kablukov, S. I.; Nikulin, M. A. (2007). Single
frequency single polarization DFB fiber laser. Laser Physics Letters, 4 (6): 428-432 Jun 2007.
Bao, X.; Dhliwayo, J.; Heron, N.; Webb, D. J.; Jackson, D. A. (1995). Experimental and
theoretical-studies on a distributed temperature sensor-based on Brillouin-
scattering. Journal of Lightwave Technology, 13 (7): 1340-1348 Jul 1995.
Barmenkov, Y. O.; Zalvidea, D.; Torres-Peir, S.; Cruz, J. L.; Andrs, M. V. (2006). Effective
length of short Fabry-Perot cavity formed by uniform fiber Bragg gratings. Optics
Express, 14 (14): 6394-6399 Jul 10 2006.
Barmenkov, Y. O.; Cruz, J. L.; Dez, A.; Andrs, M. V. (2010). Electrically tunable photonic
true-time-delay line. Optics Express, 18 (17): 17859-17864 Aug 16 2010.
Belanger, P. A. (2005). On the profile of pulses generated by fiber lasers. Optics Express, 13
(20): 8089-8096 Oct 3 2005.
Bello-Jimnez, M.; Cuadrado-Laborde, C.; Sez-Rodriguez, D.; Dez, A.; Cruz, J. L.; Andrs,
M. V. (2010). Actively mode-locked fiber ring laser by intermodal acousto-optic
modulation. Optics Letters, 35 (22): 3781-3783 Nov 15 2010.
Bello-Jimnez, M.; Cuadrado-Laborde, C.; Dez, A.; Cruz, J. L.; Andrs, M. V. (2011).
Experimental study of an actively mode-locked fiber ring laser based on in-fiber
amplitude modulation. Applied Physics B, in press (2011).
Birks, T. A.; Russell, P. S. J.; Pannell, C. N. (1994). Low-power acoustooptic device based on a
tapered single-mode fiber. IEEE Photonics Technology Letters, 6 (6): 725-727 Jun 1994.
Blais, S.; Yao, J. P. (2009). Photonic true-time delay beamforming based on superstructured
fiber Bragg gratings with linearly increasing equivalent chirps. Journal of Lightwave
Technology, 27 (9): 1147-1154 May 1 2009.
Bonadeo, N. H.; Knox, W. H.; Roth, J. M.; Bergman, K. (2000). Passive harmonic mode-
locked soliton fiber laser stabilized by an optically pumped saturable Bragg
reflector. Optics Letters, 25 (19): 1421-1423 Oct 1 2000.
Capmany, J.; Ortega, B.; Pastor, D.; Sales, S. (2005). Discrete-time optical processing of
microwave signals. Journal of Lightwave Technology, 23 (2): 702-723 Feb 2005.
Caucheteur, C.; Mussot, A.; Bette, S.; Kudlinski, A.; Douay, M.; Louvergneaux, E.; Megret,
P.; Taki, M.; Gonzalez-Herraez, M. (2010). All-fiber tunable optical delay line. Optics
Express, 18 (3): 3093-3100 Feb 1 2010.
Costantini, D. M.; Limberger, H. G.; Lasser, T.; Muller, C. A. P.; Zellmer, H.; Riedel, P.;
Tunnermann, A. (2000). Actively mode-locked visible upconversion fiber laser.
Optics Letters, 25 (19): 1445-1447 Oct 1 2000.
Cuadrado-Laborde, C.; Delgado-Pinar, M.; Torres-Peir, S.; Dez, A.; Andrs, M. V. (2007).
Q-switched all-fibre laser using a fibre-optic resonant acousto-optic modulator.
Optics Communications, 274 (2): 407-411 Jun 15 2007.


634
Cuadrado-Laborde, C.; Perez-Milln, P.; Andres, M. V.; Dez, A.; Cruz, J. L.; Barmenkov, Y.
O. (2008). Transform-limited pulses generated by an actively Q-switched
distributed fiber laser. Optics Letters, 33 (22): 2590-2592 Nov 15 2008.
Cuadrado-Laborde, C.; Dez, A.; Delgado-Pinar, M.; Cruz, J. L.; Andres, M. V. (2009a). Mode
locking of an all-fiber laser by acousto-optic superlattice modulation. Optics Letters,
34 (7): 1111-1113 Apr 1 2009.
Cuadrado-Laborde, C.; Diez, A.; Cruz, J. L.; Andres, M. V. (2009b). Doubly active Q
switching and mode locking of an all-fiber laser. Optics Letters, 34 (18): 2709-2711
Sep 15 2009.
Cuadrado-Laborde, C.; Diez, A.; Cruz, J. L.; Andres, M. V. (2010a). Experimental study of an
all-fiber laser actively mode-locked by standing-wave acousto-optic modulation.
Applied Physics B-Lasers and Optics, 99 (1-2): 95-99 Apr 2010.
Cuadrado-Laborde, C.; Diez, A.; Cruz, J. L.; Andres, M. V. (2010b). Actively Q-switched and
modelocked all-fiber lasers. Laser Physics Letters, 7 (12): 870-875 Dec 2010.
Culverhouse, D. O.; Farahi, F.; Pannell, C. N.; Jackson, D. A. (1989a). Potential of stimulated
Brillouin-scattering as sensing mechanism for distributed temperature sensors.
Electronics Letters, 25 (14): 913-915 Jul 6 1989.
Culverhouse, D. O.; Farahi, F.; Pannell, C. N.; Jackson, D. A. (1989b). Stimulated Brillouin-
scattering - a means to realize tunable microwave generator or distributed
temperature sensor. Electronics Letters, 25 (14): 915-916 Jul 6 1989.
Culverhouse, D. O.; Richardson, D. J.; Birks, T. A.; Russell, P. S. J. (1995). All-fiber sliding-
frequency Er
3+
/Yb
3+
soliton laser. Optics Letters, 20 (23): 2381-2383 Dec 1 1995.
Delgado-Pinar, M.; Zalvidea, D.; Diez, A.; Perez-Millan, P.; Andres, M. V. (2006). Q-
switching of an all-fiber laser by acousto-optic modulation of a fiber Bragg grating.
Optics Express, 14 (3): 1106-1112 Feb 6 2006.
Delgado-Pinar, M.; Diez, A.; Cruz, J. L.; Andres, M. V. (2007). Single-frequency active Q-
switched distributed fiber laser using acoustic waves. Applied Physics Letters, 90
(17): Art. No. 171110 Apr 23 2007.
Erdogan, T. (1997). Fiber crating spectra. Journal of Lightwave Technology, 15 (8): 1277-1294
Aug 1997.
French, P. M. W. (1995). The generation of ultrashort laser pulses. Reports on Progress in
Physics 58 (2): 169-267 Feb 1995.
Galtarossa, A.; Nava, E.; Valentini, G. (1993). Single-Mode Optical Fiber Measurement:
Characterization and Sensing, Ed. G. Cancellieri, Artech Pub. 1993.
Geister, G.; Ulrich, R. (1988). Neodymium-fiber laser with integrated-optic mode locker.
Optics Communications, 68 (3): 187-189 Oct 1 1988.
Haus, H. A. (2000). Mode-locking of lasers. IEEE Journal of Selected Topics in Quantum
Electronics, 6 (6): 1173-1185 Nov-Dec 2000.
Huang, D. W.; Liu, W. F.; Yang, C. C. (2000). Q-switched all-fiber laser with an acoustically
modulated fiber attenuator. IEEE Photonics Technology Letters, 12 (9): 1153-1155 Sep
2000.
Hudson, D. D.; Holman, K. W.; Jones, R. J.; Cundiff, S. T.; Ye, J.; Jones, D. J. (2005). Mode-
locked fiber laser frequency-controlled with an intracavity electro-optic modulator.
Optics Letters, 30 (21): 2948-2950 Nov 1 2005.
Imai, T.; Komukai, T.; Yamamoto, T.; Nakazawa, M. (1997). A wavelength tunable Q-
switched fiber laser using fiber Bragg gratings. Electronics and Communications in
Japan Part II-Electronics, 80 (11): 12-21 Nov 1997.


635
Italia, V.; Pisco, M.; Campopiano, S.; Cusano, A.; Cutolo, A. (2005). Chirped fiber Bragg
gratings for electrically tunable time delay lines. IEEE Journal of Selected Topics in
Quantum Electronics, 11 (2): 408-416 Mar-Apr 2005.
Jeon, M. Y.; Lee, H. K.; Kim, K. H.; Lee, E. H.; Oh, W. Y.; Kim, B. Y.; Lee, H. W.; Koh, Y. W.
(1998). Harmonically mode-locked fiber laser with an acousto-optic modulator in a
Sagnac loop and Faraday rotating mirror cavity. Optics Communications, 149 (4-6):
312-316 Apr 15 1998.
Kee, H. H.; Lees, G. P.; Newson, T. P. (1998). Narrow linewidth CW and Q-switched erbium-
doped fibre loop laser. Electronics Letters, 34 (13): 1318-1319 Jun 25 1998.
Kim, B. Y.; Blake, J. N.; Engan, H. E.; Shaw, H. J. (1986). All-fiber acoustooptic frequency
shifter. Optics Letters, 11 (6): 389-391 Jun 1986.
Kim, H. S.; Yun, S. H.; Kwang, I. K.; Kim, B. Y. (1997). All-fiber acousto-optic tunable notch
filter with electronically controllable spectral profile. Optics Letters, 22 (19): 1476-
1478 Oct 1 1997.
Kuizenga, D. J.; Siegman, A. E. (1970). FM and AM mode locking of homogeneous laser .1.
Theory. IEEE Journal of Quantum Electronics, QE 6 (11): 694-& 1970.
Li, Y. H.; Lou, C. Y.; Han, M.; Gao, Y. Z. (2001). Detuning characteristics of the AM mode-
locked fiber laser. Optical and Quantum Electronics, 33 (6): 589-597 Jun 2001.
Li, Y. Q.; Zhang, F. C.; Yoshino, T. (2003). Wide-range temperature dependence of Brillouin
shift in a dispersion-shifted fiber and its annealing effect. Journal of Lightwave
Technology, 21 (7): 1663-1667 Jul 2003.
Li, X. W.; Peng, L. M.; Wang, S. B.; Kim, Y. C.; Chen, J. P. (2007). A novel kind of
programmable 3(n) feed-forward optical fiber true delay line based on SOA. Optics
Express, 15 (25): 16760-16766 Dec 10 2007.
Liu, W. F.; Russell, P. S. J.; Dong, L. (1997). Acousto-optic superlattice modulator using a
fiber Bragg grating. Optics Letters, 22 (19): 1515-1517 Oct 1 1997.
Liu, W. F.; Russell, P. S. J.; Dong, L. (1998). 100% efficient narrow-band acoustooptic tunable
reflector using fiber Bragg grating. Journal of Lightwave Technology, 16 (11): 2006-
2009 Nov 1998.
Liu, Y. Q.; Yang, J. L.; Yao, J. P. (2002). Continuous true-time-delay beamforming for phased
array antenna using a tunable chirped fiber grating delay line. IEEE Photonics
Technology Letters, 14 (8): 1172-1174 Aug 2002.
Liu, Y. Q.; Yao, J. P.; Yang, J. L. (2003). Wideband true-time-delay beam former that employs
a tunable chirped fiber grating prism. Applied Optics, 42 (13): 2273-2277 May 1 2003.
Myren, N.; Margulis, W. (2005). All-fiber electrooptical mode-locking and tuning. IEEE
Photonics Technology Letters, 17 (10): 2047-2049 Oct 2005.
Ortega, B.; Cruz, J. L.; Capmany, J.; Andres, M. V.; Pastor, D. (2000). Analysis of a
microwave time delay line based on a perturbed uniform fiber Bragg grating
operating at constant wavelength. Journal of Lightwave Technology, 18 (3): 430-436
Mar 2000.
Parker, T. R.; Farhadiroushan, M.; Feced, R.; Handerek, V. A.; Rogers, A. J. (1998).
Simultaneous distributed measurement of strain and temperature from noise-
initiated Brillouin scattering in optical fibers. IEEE Journal of Quantum Electronics, 34
(4): 645-659 Apr 1998.
Prez-Milln, P.; Torres-Peir, S.; Mora, J.; Dez, A.; Cruz, J. L.; Andres, M. V. (2004). Electronic
tuning of delay lines based on chirped fiber gratings for phased arrays powered by a
single optical carrier. Optics Communications, 238 (4-6): 277-280 Aug 15 2004.


636
Prez-Milln, P.; Dez, A.; Andres, M. V.; Zalvidea, D.; Duchowicz, R. (2005a). Q-switched
all-fiber laser based on magnetostriction modulation of a Bragg grating. Optics
Express, 13 (13): 5046-5051 Jun 27 2005.
Prez-Milln, P.; Cruz, J. L.; Andres, M. V. (2005b). Active Q-switched distributed feedback
erbium-doped fiber lasers. Applied Physics Letters, 87 (1): Art. No. 011104 Jul 4 2005.
Philippov, V. N.; Kiryanov, A. V.; Unger, S. (2004). Advanced configuration of erbium fiber
passively Q-switched laser with Co
2+
: ZnSe saturable absorber. Ieee Photonics
Technology Letters, 16 (1): 57-59 Jan 2004.
Phillips, M. W.; Ferguson, A. I.; Hanna, D. C. (1989a). Frequency-modulation mode-locking
of a Nd
3+
-doped fiber laser. Optics Letters, 14 (4): 219-221 Feb 15 1989.
Phillips, M. W.; Ferguson, A. I.; Kino, G. S.; Patterson, D. B. (1989b). Mode-locked fiber laser
with a fiber phase modulator. Optics Letters, 14 (13): 680-682 Jul 1 1989.
Russell, P. S. J.; Liu, W. F. (2000). Acousto-optic superlattice modulation in fiber Bragg
gratings. Journal of the Optical Society of America A-Optics Image Science and Vision, 17
(8): 1421-1429 Aug 2000.
Russo, N. A.; Duchowicz, R.; Mora, J.; Cruz, J. L.; Andres, M. V. (2002). High-efficiency Q-
switched erbium fiber laser using a Bragg grating-based modulator. Optics
Communications, 210 (3-6): 361-366 Sep 15 2002.
Schaffer, C. B.; Garcia, J. F.; Mazur, E. (2003). Bulk heating of transparent materials using a
high-repetition-rate femtosecond laser. Applied Physics A-Materials Science &
Processing, 76 (3): 351-354 Mar 2003.
Siegman, E. (1986). Lasers, University Science, Mill Valley, CA, 1986.
Tuchin, V. V. (1993). Lasers and fiber optics in biomedicine. Laser Physics, 3 (4): 767-820 1993.
Vicente, S. G. C.; Gmez, M. A. M.; Kiryanov, A. V.; Barmenkov, Y. O.; Andrs, M. V. (2004).
Diode-pumped self-Q-switched erbium-doped all-fibre laser. Quantum Electronics,
34 (4): 310-314 Apr 2004.
Villegas, I. L.; Cuadrado-Laborde, C.; Abreu-Afonso, J.; Dez, A.; Cruz, J. L.; Martinez-
Gmez, M. A.; Andrs, M. V. (2011a). Mode-locked Yb-doped all-fiber laser based
on in-fiber acoustooptic modulation. Laser Physics Letters, 8 (3): 227-231 Mar 2011.
Villegas, I. L.; Cuadrado-Laborde, C.; Dez, A.; Cruz, J. L.; Martinez-Gmez, M. A.; Andrs,
M. V. (2011b). Yb-doped strictly all-fiber laser actively Q-switched by intermodal
acousto-optic modulation. Laser Physics, in press 2011.
Wada, K.; Fujita, J.; Yamada, J.; Matsuyama, T.; Horinaka, H. (2008). Simple method for
estimating shape functions of optical spectra. Optics Communications, 281 (3): 368-
373 Feb 1 2008.
Yu, C. X.; Haus, H. A.; Ippen, E. P.; Wong, W. S.; Sysoliatin, A. (2000). Gigahertz-repetition-
rate mode-locked fiber laser for continuum generation. Optics Letters, 25 (19): 1418-
1420 Oct 1 2000.
Zadok, A.; Raz, O.; Eyal, A.; Tur, M. (2007). Optically controlled low-distortion delay of
GHz-wide radio-frequency signals using slow light in fibers. IEEE Photonics
Technology Letters, 19 (5-8): 462-464 Mar-Apr 2007.
Zalvidea, D.; Russo, N.A.; Duchowicz, R.; Delgado-Pinar, M.; Dez, A.; Cruz, J. L.; Andrs,
M. V. (2005). High-repetition rate acoustic-induced Q-switched all-fiber laser.
Optics Communications, 244 (1-6): 315-319 Jan 3 2005.
28
Surface Acoustic Waves and
NanoElectromechanical Systems
Dustin J. Kreft and Robert H. Blick
University of Wisconsin Madison
Department of Electrical and Computer Engineering
U.S.A.
1. Introduction
Surface acoustic waves (SAW) follow the industrial trend of reducing the size, enhancing
the speed, while enhancing the efficiency of energy coupling. Integrating this with micro-
electromechanical systems (MEMS) and nano-electromechanical systems (NEMS) offers a
wide variety of applications such as touch screens, gas and biological sensors, and
embedded RFID devices. With modern lithographic techniques, allowing the fabrication of
smaller SAW devices, we now use SAWs to probe the mechanical interactions of nano
structures. In particular, SAWs can be used to actuate NEMS which gives rise to many
interesting phenomena including anomalous acoustoelectric currents, shock waves in
suspended devices, and few electron transport, to only name a few, (Beil et al., 2008;
Talyanskii et al., 1997). Today, SAWs are also used to generate a quantized current for use as
a current standard. In practice two counter-propagating SAWs are used to observe a
quantized acoustoelectric current. This leads to population and depopulation of discrete
states (Kataoka et al., 2007). In the following we want to give an overview of the state of the
art of applying SAWs to nanomechanical devices. We will also give a brief introduction to
recent nanoelectromechanical systems with integrated low-dimensional electron gases,
which have the potential to reveal insights into quantized acoustoelectric states.
2. Fabrication
The focus for generating SAWs in this chapter will involve the fabrication of interdigitated
transducers (IDT). An IDT is simple in concept but can be very involved when fine tuning a
structure for engineering applications. Such topics as electronic impedance matching to RF
lines, effects of bulk waves in contrast to SAWs, and increasing bandwidth will not be
covered; though, this is simply a shortened list of things to consider when designing a
proper IDT for engineering applications, they do fall outside the scope of this chapter.
Nevertheless, another fabrication step we will consider is the use of acoustic waveguides for
acoustic impedance matching of IDTs to nanomechanical devices.
2.1 Interdigitated Transducer Design
The main equation to consider when designing an IDT is:


638
v f = (1)
Where v is the velocity of sound in the material, is the SAW wavelength or pitch of the
IDT, and f is the frequency of the propagating SAW. The pitch of the IDT fingers is the same
as the SAW wavelength, , which will propagate across the sample. Fig. 1 shows a simple
IDT schematic along with a scanning electron micrograph (SEM) image and optical image.

(a) (b)

(c) (d)

Fig. 1. Schematics and scanning electron microscope images of IDTs. (a) Schematic of
bidirectional IDT, (b) schematic of unidirectional IDT through the use of a reflection grating,
(c) SEM image of a bidirectional IDT on GaAs with a pitch of 4 m and a center frequency of
~715 MHz. The scale bar is 10 m. (d) Optical image of device with an IDT on both ends and
a nanomechanical device placed in the center
In Fig. 1 the arrows indicate the SAW propagation directions. It can be easily seen that the
SAW will propagate in both directions away from the IDT. A singly propagating SAW
direction can be achieved by placing a reflection grating on one end of the SAW. The
reflection grating will have the same geometry as the IDT; that is, the same finger spacing
and width. The grating distance from the IDT is ~ /2. The IDT fingers are typically chosen
with evenly spaced fingers, where the spacing between the fingers is equal to the finger
width. The finger width and finger spacing is /4 in this scenario and gives a metallization
ratio of = 0.5, which will generate only odd harmonics with no response of the third
harmonic (Campbell, 1998). The bandwidth of an IDT is defined as BW% = f / N
Pairs
. Where f,
again, is the center frequency and N
Pairs
is the number of finger pairs of the IDT. The IDT in
Fig. 1a has five pairs of fingers and is symmetric about a center point along the axis of SAW
propagation. It is typical to fabricate two IDTs, one on each end of the device, so that their
propagating waves can interfere either constructively or destructively across the center
region of the device; this is typically the region where a Quantum Point Contact (QPC) or
other structures resides. This can be seen in Fig. 1d where in the center a double quantum
dot (DQD) is placed and the IDTs are placed to the left and right.

Surface Acoustic Waves and NanoElectromechanical Systems

639
2.2 Acoustic Waveguide Design
Another detail of fabrication is the use of waveguides for the SAW. A waveguide for SAWs
is a pattern that is fabricated onto the device so the SAWs can be guided into a certain
region allowing a stronger SAW signal or amplitude; it is analogous to a coplanar
waveguide for RF signals. When fabricating IDTs, and a device as a whole, using
waveguides does not have much of a use in the areas of RFID, cellular delay lines,
sensors, and other non-region specific devices. However, this chapter will focus on NEMS
and the use of SAWs to interact with these devices. Since the NEMS device is orders of
magnitude smaller than the IDT aperture very little SAW power can interact with the
NEMS structure; it can be beneficial to include a waveguide to focus the SAW onto the
nanostructure.
Waveguides can be modeled using standard acoustic, or horn, waveguides with some minor
tuning. There are several shapes which can be used and each shape offers its own benefits
regarding a particular application. We will start by first mentioning the base equation for
hyperbolic horns used in speaker systems, these horns can also be referred to as hyperbolic-
exponential horns or hypex horns (Salmon, 1946).

2
t
A A [Cosh(kx) T Sinh(kx)] = + (2)

-
-
- -
2 1
2 2
0
2
(1 )
1 1
v
c c c
t
f f T f T
z i
A f f f

= +

(3)
Eq. 2 gives the wave front area expansion of the wave as propagating through the
waveguide or horn. A
t
is the area of the throat, or base, of the waveguide, T is a factor
describing the shape of waveguide; for T = 1 the waveguide becomes exponential in shape,
and becomes canonical as T , k is given as (2f
c
)/v; where v is the velocity of sound, and
f
c
is the cutoff frequency of the waveguide. For hyperbolic and exponential waveguides
there is no transmission below the cutoff frequency, f
c
, since at this point the impedance is
purely reactive. Eq. 3 gives the impedance of the waveguide; all variables are defined
previously with the addition of
0
which is the density of the material.
Fig. 2 contains three SEM images of an exponential waveguide with a tube waveguide in the
center. A tube waveguide is simply a small or narrow SAW delay line. The material is a
GaAs/AlGaAs heterostructure containing a two-dimensional electron gas (2DEG) and a
sacrificial layer. We use a 2DEG because the high electron mobility makes this an ideal
candidate for nanostructures, such as quantum dot (QD) and DQD systems. The waveguide
was defined using electron-beam lithography to open up an etch region through the PMMA.
The material was then wet etched with H
3
PO
4
:H
2
O
2
:H
2
0 into the sacrificial layer. Then
diluted HCl was used to remove the sacrificial layer which resulted in a suspended region in
the center and around the waveguide. This suspended region is visible as a darkened
shadow indicating a height difference due to strain relaxation of the material. The
suspended region forms the actual nanomechanical device. Since this device contains a
DQD in the center region of the structure it is beneficial to use a waveguide to ensure a
higher SAW power density coupled into that region. Since the IDT aperture of this device is
50 m and the DQD region is roughly 600 nm at the widest QPCs.


640
(a)

(b)

(c) (d)

Fig. 2. SEM images of an exponential waveguide with a cutoff frequency of 75 MHz and
throat width of ~1.5 m. The darker region around the waveguide shows that the area is
suspended, which was achieved by an HCl etch to remove the sacrificial layer below the
2DEG, (a), (b) and (c). (d) Schematic of waveguide, center region is the throat (pipe
waveguide) with width W
t
and length L
t
. The outer curves are the exponential portion of the
waveguide
3. Surface Acoustic Waves in Two Dimensional Electron Gases
3.1 Surface Acoustic Wave Basics
When applying an RF signal to the IDTs an alternating electric potential is seen by the
electrodes, this will create an electric field distribution within the piezoelectric substrate, see
Fig. 3a. This field distribution, in turn, will cause a mechanical deformation of the material
through the inverse piezoelectric effect. As stated before, this mechanical deformation will
propagate away from the IDTs and continue to move along the surface of the device, see Fig.
3d. Here we neglect other forms of waves being produced such as bulk waves.
Once the RF signal of the IDT is coupled into the piezoelectric substrate a Rayleigh-wave is
produced. The elliptical wave propagating (Fig. 3b, d) along the surface can be described
quite accurately by a traveling wave:

- - k|y| i( t kz)
U |U|e e
= (4)
The electric potential being created by the propagating wave can also be described in a like
manner:

- - k|y| i( t kz)
| |e e
= (5)


641
In both Eqs. 4 and 5, is the angular frequency and k is the phase constant. The SAW
penetrates into the depth of the material by about one wavelength, ~. This value is different
in the suspended region. Hence, the actuation of nanomechanical resonators is considerably
enhanced. Another property of the SAW is that the electric field created from the induced
electric potential does not terminate at the surface of the material but can extend beyond by .

(a) (b) (c) (d)

Fig. 3. (a) Side view of IDT with an applied voltage, the electric field couples into the
piezoelectric substrate causing a deformation. (b) The elliptical motion of the Rayleigh-
wave. (c) Axis used for reference in equations as it applies to the orientation of the
piezoelectric substrate. (d) Side view of piezoelectric material as it is deformed causing a
SAW to propagate
The set of base equations used for describing SAW phenomena are listed.

i
i
E
x
(6)

ij ijkl kl nij n
T c S e E = + (7)

k l
kl
l k
1 u u
S
2 x x

= +

(8)
Eq. 6 is the electric field intensity that is produced from the deformed piezoelectric material
from the SAW. Eq. 7 is the piezoelectric mechanical stress and Eq. 8 is the linear strain
displacement.
3.1.1 Attenuation
SAW attenuation can be described by the following equations (Wixworth et al., 1989):

2
eff s M
2
s M
K /
k
2 1 ( / )

=
+
(9)

2
eff
2
s M
v K 1
k
v 2 1 ( / )
=
+
(10)
Here the attenuation occurs because part of the longitudinal electric field of the propagating
wave couples into the electrons of the 2DEG. This not only causes a current to flow but pulls
power from the SAW due to ohmic losses. This attenuation is described by Eq. 9. A SAW
velocity shift is also observed due to the piezoelectric stiffening of the substrate, see Eq. 10
(Wixworth et al., 1989). Below are the recreated graphs from (Wixworth et al., 1989) to show
the relationship of the attenuation and sound velocity shift due to a change in conductivity.


642
(a) (b)
Fig. 4. (a) The SAW attenuation in units of K
eff
2
, see Eq. 9. (b) The change in SAW velocity in
units of K
eff
2
, see Eq. 10
Where k is the SAW wave vector, K
2
eff
is the effective piezoelectric coupling coefficient,
s
is
the 2DEG sheet conductivity, and
M
= v
0
(
1
+
2
). Again, where v
0
is the sound velocity and
1
and
2
are the dielectric constants of the piezoelectric substrate and half space above it.
3.2 Acoustoelectric Current
As the SAW propagates across the material it creates two types of currents, one is the
normal acoustoelectric current and the other being the anomalous acoustoelectric current.
The normal acoustoelectric current is created by electrons being dragged across the
material and can be described by Eq. 11 where I is the current, n is the number of electrons, e
is the charge of an electron, and f is the frequency of the SAW or RF signal to the IDT. This
current always flows in the direction of the SAW and is produced as a DC current despite an
oscillating RF signal being applied to the IDTs. Eq. 11 shows that at higher frequencies the
normal acoustoelectric current becomes quantized.
I nef = (11)

2 /
zz z
0
a
j(z) (z, t)E (z, t)dt
2

(12)
Once this deformation occurs the energy bands in the material bend as well causing the
electrons to fall into the created quantum wells and are dragged along with the SAW, see
Fig. 3d. As the frequency increases the wavelength, and pitch of the IDT, decreases causing
fewer energy states to be available within the wells. This idea is also implemented to
generate QDs within the SAW, (Barnes et al., 2000).
The anomalous acoustoelectric current is produced from the deformation of the material
and flows as a DC current. The difference is that the anomalous current always flows in one
direction regardless of which IDT, left or right, produced the SAW. The current is smaller
than the normal acoustoelectric current and is detected by different methods. The
anomalous acoustoelectric current can be obtained by sweeping the RF signal of the IDTs
and typically appears at an off-center frequency. Since the normal acoustoelectric current is
much smaller, the anomalous current can be more easily detected, as shown in Fig. 5b. One
such method for direct detection of the anomalous acoustoelectric current is to apply an RF
signal to both the left and right IDTs, while phase locking the two signals (Beil et al., 2008).
By phase locking the IDT signals a standing SAW can be created. Thus, a surface
deformation is produced without a net direction of propagation. As seen in Fig. 5a the phase


643
is shifted between the two IDTs as the current is measured. At a phase difference of 180 the
current tends towards zero.

(a)

(b)

(c)

(d)

Fig. 5. Results for a suspended beam that is 1.2 m long and 300 nm wide. (a) Anomalous
acoustoelectric current as a result of phase adjusting two counter propagating SAWs. (b)
Acoustoelectric current vs. applied frequency of the IDTs. There is always a negative current
value to the left and is indicative of the anomalous current. (c) Formation of a shock wave
seen by its Sinc(x) shape. The devices response jumps to a higher order mode and returns to
Sin(x). (d) Derived anomalous acoustoelectric current amplitude as a result of the calculated
SAW amplitude. All images are taken from Beil et al., 2008, Copyright (2008) by The
American Physical Society
This zero current occurs because the two SAWs interfere destructively and create a nearly
smooth surface with no deformation. At 0, and also 360, the current is at a maximum since
the two SAWs interfere constructively allowing a maximum in the surface deformation.
The anomalous acoustoelectric current can be used to probe the SAW amplitude of a
suspended beam or nanostructure. The accompanying graphs show the acoustoelectric
current from the device. It has been shown that the anomalous acoustoelectric current
through the device in relation to the SAW amplitude can be described by Eqs. 12-14; where
Eq. 12 is the general equation for an anomalous acoustoelectric current, 13 is the derived
equation, and 14 is the strain equation of the beam.

2 0 z4
an zzzy zy
GaAs
e
j (L /2) ( S (L /2) )
2
(13)
( ) - -
2
zy t SAW 3 2
6z 6z
S (z, t) 2A ( ) Cos( /2) Cos(k L /2)
L L

= +

(14)


644
Where
0
is the unperturbed conductivity,
GaAs
is the permittivity of GaAs, is a tensor
which describes the effect of strain S, and A
t
is the transverse component of the SAW.
An interesting phenomenon that occurs is the formation of shock waves in the suspended
structure. From Figs. 5a and 5c it is seen that the device exhibits a Sin(x) type behavior in the
anomalous acoustoelectric current at low RF powers. Once the RF power starts to increase it
can be seen that the shape of the current starts to exhibit more of a Sinc(x) shape. This
transition from a linear to a non-linear response indicates the formation of a shock wave in
the suspended beam. As the power increases, the beam jumps into the next higher order
mode and the current trace returns back to a linear response again. The shock wave
formation is an indication that the beam will transition from one mode to another.
3.3 Surface Acoustic Waves in Magnetic Fields
We can use SAWs to probe the characteristics of a 2DEG under the presence of a magnetic
field. Since a high mobility 2DEG is subject to Shubnikov-de Haas (SdH) oscillations, which
creates changes in the conductivity, and SAWs are more sensitive to a conducting plane or
surface, and the conductance of that plane, makes this combination an ideal candidate to
investigate quantum effects of the 2DEG. The real interesting features are seen when
integral Landau level filling factors are observed which causes a drop in conductivity of the
2DEG. At these quantized values the SAW responds strongly to the conductivity, .
In Fig. 6 it can be seen that the SdH oscillations and the acoustoelectric current oscillations
are nearly identical. The SdH measurement was measured using a standard four-point lock-
in technique, where as the acoustoelectric current was created by a SAW and was taken
from two ohmic contacts on opposite sides of the sample. The peaks and valleys of the two
measurements line up, for the most part, but there is a small offset. This offset is due to the
fact that the SAW attenuation is not a linear with respect to the 2DEG conductivity, or
magneto conductivity in this case, see Eq. 9.

Fig. 6. A magnetic field, of maximum value 9 T, was applied perpendicular to the surface of
the device. The device is shown in figures 1d and 2a-2d and contains a 2DEG 40 nm below the
surface. Upper trace is the Shubnikov-de Haas oscillations measured using standard lock-in
techniques and normalized to 713.4 . The lower trace is the acoustoelectric current generated
from a SAW with -12 dBm applied to the right IDT at 1.488 GHz and is normalized to 10.47 nA


645
Special attention has to be given to the acoustic current trace of Fig. 6 around 3 4 Tesla.
There is a splitting of the peaks that occurs. This information does not appear in the SdH
oscillation and is only isolated to a SAW effect. During this splitting the conductivity of the
2DEG is very low and
s
becomes much smaller than
M
, see Eq. 9. When this happens a
maximum in SAW attenuation will occur which results in a reduced SAW amplitude. The
center of a SAW current split is a minimum in the 2DEG conductivity; further details are
discussed in Wixworth et al., 1989.
From Fig. 6 there are two observations that need to be explained. One is the spike-like
feature or discontinuity around the 8.5 T mark of the acoustic trace. This is caused from
SAW reflections on the sample. This spike-like feature can be seen on other data plots as
well with SAW currents. The final feature is the negative acoustic current that was
measured. Since at this point the SAW was highly attenuated, causing little or no current to
flow, caused the sample itself to heat. The negative current may be a combination of a
thermal current and a small offset in the measuring equipment.
4. Surface Acoustic Waves in Quantum Electronics
SAWs are used to produce a current in low dimensional electronic systems and NEMS. This
can be used in various applications and for numerous different designs. The information
presented in this section will deal with piezoelectric materials with an embedded two
dimensional electron gas (2DEG); mainly GaAs/AlGaAs heterostructures unless
otherwise stated. These measurements are carried out at liquid helium temperatures or
lower, 4.2 K.
4.1 Quantum Point Contacts and Low Dimensional Channels
SAWs can be used to create a quantized current based upon Eq. 7. The advantage of the
SAW inducing a quantized current is that this process can be used to populate and
depopulate QDs and DQDs at higher frequencies then what can be used by applying an
oscillating signal to the source-drain of the 2DEG. When using a SAW as the current source
the acoustoelectric current can be pinched off in a Coulomb blockade just like a standard
source-drain current can be. This pinch off is done via a quantum point contact (QPC) or a
set of QPCs, which can be used to form QDs; see Fig. 7.
4.1.1 Quantum Point Contact Fabrication
Fabricating QPCs is done with the same methods as fabricating IDTs, see Sec. 2. Since QPCs
are have small dimensions it is most common to use Electron Beam Lithography, or e-beam
lithography, to create the structures. The exact dimensions of the QPC pair depends on what
works best for the user and there is no set rules for design like there is for IDTs. When
viewing Fig. 7 it can be seen that there are five sets of QPCs. A single QPC is seen as an
electrode with another electrode opposite its position. All of the QPCs shown have a few
common features; the tip of the QPC is small when compared to the rest of the electrode and
the gap between the electrode tips is small as well. The tip is small so the electric field being
emitted from the QPC is very localized, and the majority of the electrode is made wider so
that it covers a wider portion of the 2DEG so the electrons are repelled. The gap between
electrodes is small so pinch-off can be achieved with small voltages, more on this in Sec.
4.1.2.


646
4.1.2 Quantum Point Contact Operation
QPCs work by applying a negative voltage to them, this negative voltage produces an
electric field that penetrates into the 2DEG causing electrons to be repelled at or around the
QPC region; this can be seen if Fig 7(c). Here the black region represents the metallic
electrode while the blue region represents the electric field and the green region represents
the 2DEG or substrate. It can be taken as a good estimate that the electric field penetrates
into the 2DEG from the QPC electrode at a 45 angle. As a larger negative voltage is applied
to the QPC the effective screening electric field becomes larger causing the electron path
within the 2DEG to become more constricted until single electron transport is achieved; this
leads to a Coulomb Blockade, see Sec. 4.2. Then finally, the current is completely pinched
off, see Figs 9 and 10.

(a) (b)

(c) (d)

Fig. 7. Two SEM images of the same DQD structure on top of a suspended tube section of an
acoustic waveguide. The DQD is made up of five pairs of QPCs. The material is a
GaAs/AlGaAs heterostructure with an embedded 2DEG about 40 nm below the surface. (a)
Top view with a scale bar of 200 nm, (b) angled view with a scale bar of 100 nm. (c)
schematic view of QPC with applied voltage, the blue area represents the electric field
surrounding the QPC and penetrating into the 2DEG (green). (d) Tunneling of electrons
across potential barrier created by applied voltage from the QPC
The QPC creates a tunneling barrier which separates the source and drain regions of the
sample. As the voltage magnitude becomes larger the strength, or height, or the tunneling
barrier is increased and the width of the barrier becomes larger. When plotted a step like


647
feature can be seen which corresponds to a single conductance step which has a value of G =
2e
2
/h, where h is Planks constant and e is the charge of an electron. Now the temperature
must be low so the thermal energy, E = k
b
T, of the background is smaller than the tunneling
energy needed for the electrons to jump across the barrier, where k
b
is the Boltzmann
constant. As seen in Figs 9 and 10 this step like feature can be seen by doing an I-V
measurement. By changing the temperature of the system the phonon energy is increased
and causes scattering events to increase, or increase the electron-phonon interaction, and the
steepening or smoothing of the step like feature is a direct measure of this.
4.1.3 Usage of Quantum Point Contacts and Surface Acoustic Waves
The use of QPCs offers a benefit of determining which SAW mode(s) are propagating in the
sample. Different SAW modes, such as bulk, longitudinal, and transverse with propagate at
different frequencies due to the fact that they have different sound velocities, see Eq. 1.
Another factor, which affects the sound velocity, is the propagation direction of the SAW
with respect to the crystal orientation of the material. In Fig. 8 a QPC had an applied voltage
of -0.8 V, which puts the QPC into pinch-off mode. Since it is in pinch-off higher RF power is
required to create a sufficiently strong SAW that will overcome the potential barrier. As the
power is increased from -18 dBm to -10 dBm, three peaks emerge as transferring current
through the tunnel barrier. From Eq. 7 we can calculate the electron count to be 6, 3, and 2
for RF powers of -10 dBm, -12 dBm, and -14 dBm, respectively (some rounding is taken into
account, due to thermal errors in measurement).

Fig. 8. A frequency sweep of varying RF powers while the center QPC of the sample in Fig. 4
is held at -0.8 V. The first peak is at 840 MHz with a current of 540 pA and a velocity of 3,368
m/s, the second peak at 1.005 GHz with a current of 472 pA and a velocity of 4,020 m/s, the
third peak is at 1.095 GHz with a current of 1.098 nA and a velocity of 4,380 m/s
Now the three peaks represent different SAW modes. The highest frequency peak of Fig. 8
of 1.095 GHz represents a longitudinal wave with an acoustic velocity of 4,380 m/s and an
angle of about 10 off from the (110) direction (Kuok et al., 2001). This small angle variation
is due to a small misalignment during the lithography process. When viewing the lower
peak of 840 MHz at a velocity of 3,368 m/s, this coincides with a fast transverse wave with,


648
again, a 10 difference between the SAW direction and the (110) GaAs crystal orientation
(Kuok et al., 2001).
4.2 Coulomb Blockade of Acoustoelectric Current
When looking over Fig. 9, we see that single steps are observed just before the total
acoustoelectric current becomes zero. Focusing on Fig. 9a, we see that the last step is at
about 0.5 nA. When solving with Eq. 11 the first step yields an answer of n = 1, i.e. a single
electron is being transferred. We can look at all of the remaining steps and solve for them as
well which will reveal integer numbers for n and will increase by 1 for each step, as is
expected. Fig. 9a also shows that an applied voltage across the source and drain contacts, or
ohmics, has very little effect on the final quantized acoustoelectric current. We do see,
however, that a larger, or smaller, gate voltage is required for final pinch-off in the system
but this is due to the small offset in the Fermi energy because of the applied bias. The only
real difference is the shift from negative current to a positive current which is easy explained
by the fact that the source-drain bias is producing a current that is opposite in direction,
when V
ds
< 0 mV, to that of the acoustoelectric current. Another aspect is the RF power
dependence on the quantized current. When viewing Fig. 9b there is a similar trend to that
of Fig. 9a. The small change in RF power has only a small effect on the acoustoelectric
current. This is because the RF power is directly proportional to the SAW amplitude and we
do not really identify any significant difference in the number of electrons being transferred,
or Coulomb steps, until a drop of about 2 dBm.
So far an acoustoelectric current behaves in the same way as a standard source-drain bias
current when a QPC is near pinch off. There is, however, another effect which can arise. As
seen in Fig 9c a negative current arises and still exhibits the step like behavior. This negative
current is the negative anomalous acoustoelectric current. This is said to be produced as an
effect of SAW reflections, this is mostly seen in a two IDT system. The second IDT acts as a
reflector much as the same way a reflection gradient is used for a unidirectional SAW, see
Fig 1b. This can cause a standing wave to occur in the sample in such a way that the
reflections effectively reduce the potential of the SAW and cause fewer electrons to be
transported.
With slight phase shifts added the standing wave with respect to the QPC a net negative
potential in the energy landscape can exist on one side of the QPC which will cause the
current to change direction. These steps are best observed for QPCs that are long when
compared to the SAW wavelength. This makes it possible to observe the acoustoelectric step
current without applying a magnetic field (Shilton et al, 1996). As mentioned earlier, the
high mobility of the 2DEG will screen the acoustoelectric current but this effect is not as
prevalent when inside of a long QPC channel. The current screening is reduced around the
QPC region and thus electrons can be transported through the channel. A long channel can
allow ballistic transport of the electron causing it to shuttle from one side of the QPC to
another. Since the screening is minimal, the electron will be dragged through the channel
and the current steps of Fig. 9 can be observed.
Another phenomenon is oscillations in the acoustoelectric current once approaching the
QPC pinch-off limit. As described in Shilton et al, 1996, and Maa & Galperin, 1997, the
acoustoelectric current oscillations occur at the same positions as the Coulomb steps. This
oscillation is described as interference effects near the QPC at high frequencies which may
be attributed to impurities in the 2DEG channel. The theory presented suggests that these


649
oscillations are due to state transitions, both propagating and reflecting (Maa & Galperin,
1997). The SAW may reflect from the QPC and create interference patterns; these patterns
will affect the electron transmission probability through the QPC simply by changing the
potential landscape. This mechanism is also sensitive to scattering events near the QPC,
where energy quanta are emitted and absorbed between two waves. Others have
contributed to the theory and more is being added.

(a) (b)

(c)

Fig. 9. QPC pinch-off of acoustoelectric current: (a) pinch-off with varying source-drain
voltages applied at the SAW, changes are in 1 mV increments (Talyanskii et al., 1997)
Copyright (1997) by The American Physical Society. (b) Pinch-off with different SAW power
levels, the left most trace is 7 dBm and decrements by 0.2 dBm (Talyanskii et al., 1997)
Copyright (1997) by The American Physical Society. (c) A negative anomalous
acoustoelectric current for different gate voltages and SAW powers; f = 1007.426 MHz,
V
ds
= 0 V, T = 1.7 K, (Song et al., 2010)
Fig. 10 shows the oscillations of the acoustoelectric current as it is being driven through a
QPC (orange trace) as it approaches pinch-off. It can be seen that the step like features
observed in Fig. 9 are have been replaced by oscillations. These oscillations are negative, or
on the lower cycle, when the current from the source-drain bias is flat or non-changing. We
then see that as the current starts to decrease the acoustoelectric current has a positive value,
or is on the upper cycle of the oscillation. The inset graph of Fig. 10 shows the entire sweep
of the QPC gate voltage. As one conduction channel pinches off at the V
g
= -0.65 V we see
that there is a large, nearly single, oscillation in the acoustoelectric current. Since SAWs are


650
very sensitive to 2DEGs we get what seems to be an amplified signal when compared to a
source-drain bias. This can be used to identify information that may otherwise have been
too weak or nearly washed out from thermal effects.

Fig. 10. Acoustoelectric current oscillations as the QPC nears pinch-off. The data plot has
been normalized to show the relationship between the acoustoelectric current oscillations
(orange trace) and the pinch-off when using a voltage bias across the source-drain electrodes
(red trace). The acoustoelectric current trace has been normalized to -8.832 nA and the
source-drain bias trace has been normalized to 4.692 nA. Measurements were taken at 4.2 K
with the sample shown in Fig. 7. Inset plot shows entire gate voltage sweep range
4.3 Population and Depopulation of Quantum Dots
SAWs have been used widely as a mechanism to control the population and depopulation of
QDs and DQDs. The high speed, or frequency, of operation combined with quantized
current production make SAWs suitable for use with quantum systems. In a traditional QD
system the gate and source-drain voltages are changed to allow single electron population
and depopulation. This has several drawbacks; one being it is hard to decouple the dot from
the surrounding environment due to the precise control of the voltages. There is always
some noise either in the system or from the electronics. Another problem is if one gate is
changed then the others must be changed to ensure a constant electron count in each dot, if
applicable.
A SAW can be used as an alternative and has gotten more attention lately for the use in a
QD or DQD system. The biggest reason for the attention is that SAWs can be used to interact
with QD systems at much higher frequencies then what has been previously achieved by
pulsing the gate electrodes. Since the acoustoelectric current can be well defined a QD
system can use very high gate voltages to ensure no leakage current. We will take a look at
the work done in Kataoka et al., 2007. This work describes very well the use of SAWs as a
way to populate and depopulate a QD.


651
When the QD needs to be populated, the center plunger gate voltage value is lowered;
causing the Fermi energy to lower below the Fermi energy of the surrounding 2DEG. The
gate electrodes are still held at a high voltage value so the electrons cannot tunnel in or out
of the QD, see Fig. 11e. The IDT is pulsed and as the SAW enters the QD region it changes
the tunneling barrier created by the electrodes. The barrier is changed just enough that a
single electron can tunnel into the dot, this is population; see Figs. 11a-b. The IDT is pulsed
many times since the tunneling barrier is so high that the additional pulses are needed to
increase the tunneling probability of the electron.

Fig. 11. Results from the paper Kataoka et al., 2007, Copyright (2007) by The American
Physical Society. (a) and (b) The population of the QD, the e line is an electron entering the
dot. (c) and (d) Depopulation of the QD. (e) Schematic of SAW for population and
depopulation of the QD
Likewise, as shown in Figs. 11c-d, the QD is depopulated by a similar mechanism. Here the
plunger gate voltage is raised, made to be more negative, which causes the Fermi energy of
the QD to be raised in comparison to the 2DEG Fermi energy outside of the dot, Fig. 11e. At
this time the IDT is pulsed and the SAW enters the QD region. The SAW alters the tunneling
barrier created by the gate electrodes when the potential of the SAW is superimposed onto
the barrier, see Eq. 5. When the barrier is raised nothing happens since this will simply
ensure the electron stays in the system, but when the negative cycle of the SAW
superimposes with the barrier it decreases it which causes the electron to tunnel out.
This method of operation has proven, quite nicely, that SAWs and high potential QPCs offer
a more robust method of single electron population and depopulation. As shown in Figs.
11a-d the QD can maintain the electron confinement for a long period of time which can be
difficult in a traditional setup due to noise reduction of the measurement electronics. The
IDTs can be accessed from an outside system such as an antenna, for example, and interact
with the QD allowing the system another degree of freedom from the traditional closed
electronics. However, this may come at the cost of additional noise reduction equipment
and filtering.
5. Conclusion
This chapter has shown all of the basic properties and uses of SAWs in nano-structures and
nano-systems. The reader is shown the parameters required for fabrication, theory of
operation, real results, and application. The use of SAWs is quite limitless in the area of


652
NEMS. As people continue to refine the research through experiment and by continuing to
add to the theory more applications will emerge.
6. Acknowledgment
We like to thank the National Science Foundation for support under a NIRT grant (No.
ECCS-0708759) and the Air Force Office for Scientific Research for support under a MURI
grant (No. FA9550-08-1-0337).
7. References
Barnes, C. H. W., Shilton, J. M., & Robinson, A. M. (2000). Quantum computing using
electrons trapped by surface acoustic waves. Physical Review B, Vol. 62, No. 12,
(September 2000), pp. 8410-8419, ISSN 0163-1829
Beil, F. W., Wixforth, A., Wegscheider, W., Schuh, D., Bichler, M., & Blick, R. H. (2008).
Shock Waves in Nanomechanical Resonators. Physical Review Letters, Vol. 100, No.
2, (January 2008), pp. 026801 1-4, ISSN 0031-9007
Campbell, C. (1998). Surface Acoustic Wave Devices for Mobile and Wireless Communications,
Academic Press, Inc, ISBN 0-12-157340-0, San Diego, CA, USA
Kataoka, M., Schneble, R. J., Thorn, A. L., Barnes, C. H. W., Ford, C. J. B., Anderson, D.,
Jones, G. A. C., Farrer, I., Ritchie, D. A., & Pepper, M. (2007). Single-Electron
Population and Depopulation of an Isolated Quantum Dot Using a Surface-
Acoustic-Wave Pulse. Physical Review Letters, Vol. 98, No. 4, (January 2007), pp.
46801-1 46801-4, ISSN 0031-9007
Kuok, M., Ng, S. C., & Zhang, V. L. (2001). Angular dispersion of surface acoustic waves on
(001), (110), and (111) GaAs. Journal of Applied Physics, Vol. 89, No. 12, (June 2001),
pp. 7899-7902, ISSN 0021-8979
Maa, F. A. & Galperin, Y. (1997). Acoustoelectric effects in quantum constrictions. Physical
Review B, Vol. 56, No. 7, (August 1997), pp. 4028-4036, ISSN 0163-1829
Salmon, V. (1946). A New Family of Horns. Jounal of the Acoustical Society of America, Vol. 17,
No. 3, (January 1946), pp. 212-218, ISSN 0001-4966
Shilton, J. M., Mace, D. R., Talyanskii, V. I., Galperin, Y., Simmons, M. Y., Pepper, M., &
Ritchie, D. A. (1996). On the acoustoelectric current in a one-dimensional channel.
Journal of Condensed Matter Physics, Vol. 8, (March 1996), pp. L337-L343, ISSN 0953-
8984
Song, L., Chen, S. W., He, J. H., Zhang, C. Y., Lu, C., & Gao, J. (2010). The anomalous
negative acoustoelectric current in single-electron transport. Solid State
Communications, Vol. 150, (2010), pp. 292-296, ISSN 0038-1098
Talyanskii, V. I., Shilton, J. M., Pepper, M., Smith, C. G., Ford, C. J. B., Linfield, E. H.,
Ritchie, D. A., & Jones, G. A. C. (1997). Single-electron transport in a one-
dimensional channel by high-frequency surface acoustic waves. Physical Review B,
Vol. 56, No. 23, (December 1997), pp. 15180-15184, ISSN 0163-1829
Wixforth, A., Scriba, J., Wassermeier, M., Kothaus, J. P., Weimann, G., & Schlapp, W. (1989).
Surface acoustic waves on GaAs/Al
x
Ga
1-x
As heterostructures. Physical Review B,
Vol. 40, No. 11, (October 1989), pp. 7874-7887

Acoustic Waves - From Microdevices To Helioseismology

Uploaded by

Copyright:

Available Formats

Acoustic Waves - From Microdevices To Helioseismology

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Acoustic Waves - From Microdevices To Helioseismology

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What topics are covered in the book?

What topics are covered in the book?

v m ( = 1, 2, 3) containing one tangent and two conical points of degeneracy

c a conical degeneracy never split or disappear, but can only shift.

c , however not real

. The dynamic displacement field ( , ) t u r in such medium is

Q are the real symmetric matrices

to the usual acoustic tensor

A and the three corresponding complex eigenvalues

v ( = 1, 2, 3), i.e. the three phase

v as functions of the direction m:

A as eigenvectors of the symmetric matrix Q i - Q

= A . Instead, the normalizing

v being linear in small viscosity

and therefore small. Let us decompose the

A on the two components:

A may be expressed in leading approximation as a

A which are almost orthogonal to

v determined by eqns. (11) and (12) may be

A may be decomposed in the basis

have completely the same symmetrical structure. It is well

Q = m m (20) must have the same eigenvectors.

Q A A of the singular term in eqn. (15) vanishes together with

m (47) path-traces the contour around the

m through the self-intersection line,

(Re ) (Re ) / c v = P u u m . (63)

m moves through all

m and the real phase

m , the elliptic polarization

= U at the ellipse in polar

U of these radius-vectors at the ellipse depend on the azimuth

. As is seen from (83), the angular velocities differently behave in

have the period

for the series of observation points

. This will renew a motion of the ray velocities

are found from (92):

. The operational momentum equation now contains in the right-hand

Fig. 1. Coordinate frame

(x, t) = g (x, t) (38)

yields N = 1, so the F-expansion method gives

UdeVries equation in warm Plasma, Computat. Methods in Sci.

Pa with a turbomolecular pump backed by a rotary pump. Ar-gas

Pa under continuous pumping condition.

[42]. Applying the electrical

= defines the average bulk density of the

= g for the ionic (protonic) mass. Again e

of solar gravity at some radial position

= and apply the necessary condition for the maximization of

= , let us once spatially

that matches with g e 2

) and minimum value of non-

) at the SSB. The value of the electrostatic potential gradient

= . From this observation one can easily estimate that

= . This corresponds to a quasi-hydrostatic type of the SSB

= , which occurs at about ~3.5

= (i.e. ~1 kV). Similarly, Fig. 8c depicts the minimum