AMPERE Newsletter Issue 94 December 18, 2017

Microwave Characterization of Nanostructured Material by

Modified Nicolson-Ross-Weir Method
J. Y. Wang1, J. Tao1*, L. Severac2, D. Mesguich2, Ch. Laurent2
1
LAPLACE, Toulouse Université, CNRS, INPT, UPS, INP-ENSEEIHT, 2 rue Camichel, F-31071 Toulouse cedex 7, France
2
CIRIMAT, Toulouse Université, CNRS, INPT, UPS, Université Paul-Sabatier, 118 route de Narbonne, F-31062
Toulouse cedex 9, France
* E-mail: [email protected]

Introduction reactive energy by modeling it by a lumped

reactive element, deduced from a 3D electro-
Nanostructured materials and nanocomposites
magnetic simulation of a simple junction. This
have specific properties thanks to their nanoscale
technique has been validated on several samples of
and are serious candidates in mechanical,
known dielectric materials, and will be applied to
electrical, chemical and environmental
a sample of nanostructured materials.
applications. Among their many physical
properties, their capacity to interact with
Material
electromagnetic waves is particularly interesting as
a potential source of new functionalities1-3. Random composite materials combining a
In this paper, the simultaneous ferromagnetic metal within an oxide matrix have
characterization of the permittivity and the demonstrated their potential as double negative
permeability of the samples of a new nano- materials (DNM)1-3. DNM exhibit both negative
structured material will be presented in the electric permittivity and negative magnetic
frequency band 40 MHz and 1600 MHz. First, the permeability leading to unique properties in terms
Nicolson-Ross-Weir algorithm (NRW)4,5 will be of wave propagation with high breakthrough
applied to the scattering parameters measured with potential, especially in fields such as optics (super-
a vector network analyzer on a microstrip lenses) or electromagnetic cloaking (invisibility
transmission line cell on which a rectangular slab cloaking)7. Such composites are bulk 3D materials
sample is added. The effective complex with isotropic electromagnetic properties
permittivity and permeability of central part originating from their composition and
consisting of the plate on a reference planar line microstructure.
will be deduced. This central part will then be An Fe-Al2O3 composite powder was
modeled as a section of multilayer planar prepared by a method involving the selective
transmission line including the sample to be reduction of an oxide solid solution. First, the
characterized. The permittivity and permeability of mixed oxalate (NH4)3Al0.8Fe0.2(C2O4)3,nH2O is
the materials will be determined by an inverse obtained by precipitation in alcohol. Heat-
modeling method based on the use of the full-wave treatment in air (at 400°C for 2 hours) leads to the
modified transverse resonance method (MTRM)6. decomposition of the oxalate and the formation of
The weak point of the NRW algorithm lies in the amorphous (Al0.8Fe0.2)2O3 solid solution8.
the fact that the reactive electromagnetic energy in Heat-treatment in H2 (1100°C, 5 hours) then
the discontinuities regions is not taken into produces the reduction of the ferric ions to metallic
account. This reactive energy contributes to iron, forming the Fe-Al2O3 composite powder9.
modifying the effective length of the central The iron content in the composite is equal to
section, thus introducing a potentially large source 12.0 vol.%, well below percolation threshold
of error. We propose an iterative procedure to evaluated at about 30 vol.%. The specimen is thus
minimize the error induced by the presence of this an insulating material. The so-obtained powder

35
AMPERE Newsletter Issue 94 December 18, 2017

was consolidated by spark plasma sintering (SPS) delay also exists compared to the microstrip line
(1400 °C, 3 min, 50 MPa) in order to obtain pellets without SUT. The scattering parameters will be
8 mm and 50 mm in diameter and 1.5 mm thick measured by using a vector network analyzer
(PNF²-Toulouse, Dr. Sinter 2080, SPS Syntex Inc., between two SMA connectors (Fig. 2(b)). The use
Japan). Then, the 50 mm pellet was cut into of thru-reflect-line (TRL) calibration procedure
rectangle 28.5 x 21.5 mm for the electromagnetic leads to the determination of S parameters between
measurements. A typical FEG-SEM image the two ends of SUT loaded section.
(recorded in backscattered electron chemical
composition mode to enhance contrast) of a cross- If each part of measurement cell can be considered
section of the Fe-Al2O3 composite reveals that the as fully loaded transmission lines, Nicholson-
Fe particles (appearing as white dots on the image) Ross-Weir (NRW) algorithm permits the
are homogeneously dispersed into the Al2O3 deduction of wide frequency range permittivity
matrix (Fig. 1). The diameter of the Fe particles and permeability of SUT4,5. This measurement
was measured on hundreds of particles on such method is a non-resonant one, and as others non
images and the distribution (inset in Fig 1) lies in resonant methods the broadband measurement will
the range 0.2 µm to 5.7 µm in diameter, with a be effectuated but the measurement results will be
median value of 0.7 µm. less accurate compared to resonant methods,
particularly with regard to dielectric and magnetic
losses.

(a)

Fe particle diameter (µm)

(b)

Figure 1: FEG-SEM image of a cross-section of the 12 vol.%

Fe-Al2O3 composite and the corresponding Fe particle
diameter distribution.

Description of measurement cell Figure 2: The quasi-TEM measurement setup in (a) a

schematic view, and (b) an image of the real structure
In this paper, we have opted for the multilayer with an alumina plate.
microstrip line structure in which the sample under
test (SUT) is placed on the top of a reference Extraction method of material properties
microstrip line (Fig. 2). The SUT loaded section is
inserted between two know reference microstrip Once the S parameters obtained by the vector
lines. Both characteristic impedance and network analyzer in the microstrip cell loaded with
propagation constant are different from those of a thin rectangular SUT, the determination of the
reference microstrip line due to the different complex permittivity and permeability of the
permittivity and permeability of the SUT. material pass through an inverse modeling
Reflection occurs at two extremities of SUT. A procedure. This procedure includes a comparison

36
AMPERE Newsletter Issue 94 December 18, 2017

of the measurement results and numerical In most of the cases, the reference TEM line
simulation with trial values of the relative complex has very low losses, hence ε1 ≅ ε 0ε r1 and µ1 ≅ µ0 .
permittivity and permeability. An optimization By noting the central line as fully filled with
algorithm is used then to adjust the trial values in material to be measured,
order to minimize the deviation between the
simulated and measured S parameters. ε 2 = ε 0 ( ε r 2 '− jε r 2 ") = ε 0ε r 2 (1 − j tan δ ε ) ,
Generally speaking, the numerical simulation
involves a 3D electromagnetic simulation µ2 = µ0 ( µr 2 '− j µr 2 ") = µ0 µr 2 (1 − j tan δ µ )
software, such as commercial HFSS or CST, or
the unknown relative permittivity and permeability
other electro-magnetic modeling methods
will be given by
developed by university researchers, such as the
µ r 2 ε r1 1 + Γ
multimodal variational method (MVM)10. The 3D = ,
εr2 1− Γ
simulations require significant computer resources
in RAM and computing time. Here we propose an − j ⋅ arg ( P ) − ln P
alternative solution with the modified version of ε r 2 µr 2 =
k0 d
Nicolson-Ross-Weir (NRW) algorithm, which
combines the original method and a 2D electro- where k0 = 2πf µ0ε 0 is the propagation constant
magnetic formulation for the determination of the
in free space.
fundamental characteristics of an equivalent
transmission line.
B. Application of the NRW formulation for multilayer
planar structures
A. NRW algorithm for fully filled transmission line structure
The explicit relations between the propagation
In the case of homogeneous transmission lines with constant, the characteristic impedance of a TEM
the middle section fully filled with a different mode of fully-filled transmission line, and the
material of length d, taking the two ends of the parameters of the constituent material, no longer
central section as reference planes, the parameters exist in the cases of multilayer planar structures. If
S are given by11 for the conventional lines, such as simple
microstrips or coplanar lines, the approximate
S11 =
(1 − P ) Γ
2

, S21 =
(1 − P ) P ,
2
explicit formulations exist, for the others, a
1 − P2Γ2 1 − P2Γ2 numerical simulation tool is necessary, especially
for the multilayer structures6. A series of design
where P = e −γd is the propagation exponent of the curves will be obtained in a given frequency range,
TEM mode in the fully filled central section, and corresponding respectively to the effective relative
Γ = (Z − Z 0 ) (Z + Z 0 ) is the reflection coefficient, permittivity and permeability, and the
characteristic impedance, with each curve obtained
where Z and Z0 correspond respectively to the
with a trial value for sample's permittivity and
characteristic impedance of the central section and
permeability. Theses curves constituent a data base
the reference line. The direct inversion of this
for the interpolation purpose during the inverse
system of equations leads to the following
modeling procedure.
relationships,
An example is shown here with a square
alumina plate of a 50.8-mm length and a 0.5-mm
Γ = X + s1 X 2 − 1, s1 = ±1 ,
thickness. The characteristic of the substrate of
V1 − Γ reference microstrip line are shown in Table 1. In
P= = P e j arg ( P ) , our measurement system, the other parameters of
1 − V1Γ the reference microstrip line are the substrate
(
where V1 = S11 + S 21 and X = 1 − S 21 + S11
2 2
) (2S
11 ). thickness, h = 1.524 mm, and the central conductor

37
AMPERE Newsletter Issue 94 December 18, 2017

width and thickness, w = 4.257mm and t = 35µm, of frequency, and of the estimated relative
respectively. The cell has been designed to operate permittivity in the case of alumina sample, were
between 200 MHz and 1.6 GHz, when the relative simulated. The new estimated values of the sample
permittivity of the material does not exceed 15. were compared to those derived directly from the
The use of MTRM6 on this structure with trial NRW formulations. These new values better
relative permittivity varying between 9 and 12 lead correspond to those expected in the range between
to the effective relative permittivity as shown in 9 and 11 as expected (Fig. 4).
Fig. 3(a). Once the measurements carried out by a
vector network analyzer, the effective relative
permittivity of the central section were obtained by (a)
2.86
12
application of the NRW algorithm. The results are

estimated effective relative permittivity

2.84

given in Fig. 3(b). Finally, the extracted relative 2.82

11
2.8
permittivity is shown in Fig. 3(c). 2.78

2.76 10
Table 1. Frequency dependent parameters of an AD255C 2.74

substrate 2.72
9
2.7

Property 1 MHz 10 GHz 2.68

2.66

Relative permittivity 2.55 2.55 0.2 0.4 0.6 0.8 1

Frequency (GHz)
1.2 1.4 1.6

Loss tangent 0.0011 0.0014

(b) 3

C. Modification of the NRW formulation with discontinuity

r
meaured effective relative permittivity ε

effect correction
2.5
The determination of the reflection coefficients in
the preceding formulation depends only on the
relative characteristic impedances between the 2
reference microstrip line and the partial loading by
the alumina sample. This is true in the absence of
the higher order modes on either side of the 1.5
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
junction of the two lines. This is the case for Frequency (GHz)

homogeneous TEM lines. However, the planar

case with introduction of the sample above the 15

central stripline naturally leads to the presence of (c) 14

r

evanescent modes. This will introduce reactive

estimated relative permittivity ε

13

energy accumulation in the vicinity of the

12
discontinuities planes. In accordance with the
nature of reactive energy, the length of the central 11

section seen by fundamental mode will be different 10

from the actual length. Instead of the actual value 9

d, it is necessary to introduce deff as a function of
the frequency. 8
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Frequency (GHz)
For each frequency, after the simulation by
one of the numerical methods10, the equivalent Figure 3: Characteristics of multilayered microstrip line:
length will be deduced as a function of the (a) Estimated effective relative permittivity by TRM, with
geometry and the relative permittivity value trial relative permittivity varying between 9 to 12.
(b) Measured effective permittivity.
proposed for the material to be measured. The
(c) Extracted sample's relative permittivity.
evolution of the corrected length dl1 as a function

38
AMPERE Newsletter Issue 94 December 18, 2017

15 5.6
(a) 5.55
14
5.5
13
Estimated relative permittivity

5.45

12 5.4

εr
5.35
11
5.3

10 5.25

9 5.2

5.15
8
5.1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
7 Frequency (GHz)
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Frequency(GHz) 0.12

Figure 4: Comparison between new (+) and old (o) (b) 0.1

estimation values. 0.08

Application for nanostructured sample

0.06

tanδ
0.04

Several Fe-Al2O3 composite samples were

0.02
prepared. Figure 5 shows a rectangular plate cut
from the 50 mm pellet and two 8 mm pellets. All 0

were metallized on one side. The rectangular plate -0.02

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Frequency (GHz)
was the subject of our research here, in order to
extract its complex permittivity and complex Figure 6: Real (a) and imaginary (b) parts of the effective
relative permittivity.
permeability. The Fe-Al2O3 composite plate
dimensions are 1.47-1.50 mm thickness, 28.51 mm
1.01
width, and 21.54 mm length. (a)
1

0.99

0.98
υr

0.97

0.96

0.95

0.94
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Frequency (GHz)

0.02

(b) 0.015

0.01

0.005

0
tanµ

-0.005

Figure 5: Fe-Al2O3 composite prepared by spark plasma -0.01

sintering (SPS). -0.015

-0.02

After applying the NRW algorithm to the -0.025

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Frequency (GHz)
measurement performed with the rectangular Fe-
Al2O3 composite, the raw results of the complex Figure 7: Real (a) and imaginary (b) parts of the effective
effective permittivity and permeability are shown, relative permeability.
respectively, in Figs. 6 and 7.

39
AMPERE Newsletter Issue 94 December 18, 2017

By using the inverse modeling method Conclusions

proposed in the previous section, we have deduced
A modification of the Nicholson-Ross-Weir
the estimated values of the complex permittivity,
formulation has been proposed for the
as well as its complex permeability, both given in
simultaneous extraction of the permittivity and the
Figs. 8 and 9. The SUT shows significant losses
permeability of a material in a multilayer planar
and exhibits magnetic properties.
transmission line configuration. The new
formulation was successfully tested on a thin plate-
(a) like alumina sample. The application of the
5.6
technique to a Fe-Al2O3 composite material
5.4
allowed the determination of the electrical and
r
extracted relative permittivity ε

5.2

5
magnetic properties of this material. Significant
4.8
losses have been observed due to the magnetic
4.6
nature of the sample.
4.4

4.2 For further reading

4
1. Zhang, Z., R. Fan, et al., J. Mater. Chem., 2013, 1, 79-85
3.8
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 2. Shi, Z., R. Fan, et al., J. Mater. Chem., 2013, 1, 1633-
Frequency (GHz)
1637
0.2
3. Shi, Z., R. Fan, et al., Adv. Mater., 2012, 24, 2349-2352
(b) 0.18
4. Nicolson, A.M., G. F. Ross, IEEE Trans. on IM, 1970,
19, 377–382.
0.16
5. Weir, W. B., Proc. IEEE, 1974, 62, 33-36.
extracted loss factor tanδ

0.14
6. Tao, J., IEEE Trans. MTT, 1992, 40, 1966–1969.
0.12
7. Pendry, J.B., D. Schurig, et al., Science, 2006, 312,
0.1
1780-1782
0.08 8. Devaux, X., Ch. Laurent, et al., Nanostruct. Mater.,
0.06 1993, 2, 339-346
0.04 9. Laurent, Ch., A. Rousset, et al., J. Mater. Chem., 1993,
0.02 3, 513-518
0
10. Tao, J., et al., 23th EuMC, 1993, 662-664.
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Frequency (GHz) 11. Rothwell, E. J., et al, PIER, 2016, 157, 31-47.

Figure 8: Real part (a) and loss factor (b) of the relative
permittivity extracted by the inverse modeling method. About the Authors
3
Jingyi Wang is currently a PhD student in
2.8 LAPLACE-ENSEEIHT-INP, Toulouse, France.
2.6 She received her B.Sc. and M.Sc. degrees
r
extracted relative permeability υ

2.4 (in 2010 and 2013, respectively) in test

2.2 and measurement technology and
2 Instrumentation from School of Mechano-
1.8 electronic Engineering, Xidian University,
1.6
Xi’an, China. Since 2014, Ms. Wang has
1.4
been working on computational electromagnetics,
1.2
supported by the China Scholarship Council. Her main
research is about modelling and applications of numerical
1
0.2 0.4 0.6 0.8 1
Frequency (GHz)
1.2 1.4 1.6
methods in electromagnetics, including 2D and 3D analyses
of the coaxial and micro-strip measurement cells.
Figure 9: Relative permeability extracted by the inverse E-mail: [email protected]
modeling method.

40
AMPERE Newsletter Issue 94 December 18, 2017

Junwu Tao was born in Hubei, China, in David Mesguich is Assistant Professor in
1962. He received his B.Sc. degree in Materials Chemistry at Toulouse
electronics from the Radio Engineering University (Université Paul Sabatier). He
Department, Huazhong (Central China) is part of the Nanocomposites and
Universitry of Science and Technology, Carbon Nanotubes team working at the
Wuhan, China, in 1982; the Ph.D degree Interuniversity Center for Materials
(with honors) from the Institut National Research and Engineering (CIRIMAT)
polytechnique of Toulouse, France, in 1988, where he works on nanopowders including nanocarbon
and the Habilitation degree from the University of Savoie, synthesis by Catalytic Chemical Vapor Deposition and
France, in 1999. From 1983 to 1991, Dr. Tao was with the preparation of ceramic- and metal-matrix nanocomposites
electronics laboratory of ENSEEIHT, Toulouse, France, where and carbides, both as powders and bulk materials sintered
he worked on the application of various numerical methods by SPS.
to 2- and 3-D problems in electromagnetics, and on the
design of microwave and millimeter-wave devices. From E-mail: [email protected]
1991 to 2001 he was with the microwave laboratory (LAHC)
at the university of Savoie, Chambéry, France, where he was
an associate professor in electrical engineering and involved
in the full-wave characterization of discontinuity in various Christophe Laurent is a Full Professor of
planar waveguides, and in nonlinear transmission line Materials Chemistry at Toulouse
design. Since September 2001 he is a full professor at the University (Université Paul-Sabatier).
Institut National Polytechnique of Toulouse, where he is He is currently serving as Director of
involved in the numerical methods for electromagnetics, CIRIMAT, the Interuniversity Center for
microwave and RF components design, microwave and Materials Research and Engineering,
millimeter-wave measurements, and microwave power where he formerly headed the
applications. Nanocomposites and Carbon
Nanotubes team (1998-2015). He got his BSc and MSc
E-mail: [email protected] degrees in Chemistry, and his Doctor in Materials Sciences
degree (PhD) at Toulouse University. Currently Prof.
Laurent’ researches focus on the synthesis of carbon
Laura Severac is a PhD student in nanotubes and graphene (notably the selectivity on the
Materials Sciences at Toulouse University number of walls/layers), ceramic- and metal-matrix
(Université Paul-Sabatier), currently nanocomposites, and spark plasma sintering. He has
working at the Interuniversity Center for published more than 110 papers in peer-reviewed journals.
Materials Research and Engineering
(CIRIMAT), in the Nanocomposites and E-mail : [email protected]
Carbon Nanotubes team. She got her MSc
in Materials Sciences at Toulouse
University. Her research focuses on the preparation of
ceramic-matrix composites using spark plasma sintering.
E-mail : [email protected]

41