Kao 2004

2 Electric Polarization and
Relaxation
The task of science is both to extend the range of our experience and to reduce it to order. Only by
experience itself do we come to recognize those laws which grant us a comprehensive view of the
diversity of phenomena.
The classical theories have contributed so fundamentally to our knowledge of atomic structure.
Neils Bohr
2.1 Fundamental Concepts F is determined by the magnitudes and the

locations of all other charges, as well as the
We can say that all dielectric phenomena arise surrounding medium. For a single-point
from an electric force which is primarily due to charge with total net charge +Q in free space,
the attraction and the repulsion of electric F can be written as
charges. The action of the force tends to reduce
the potential energy of the whole electrically
stressed system to a minimum. This is a uni-
versal law of nature. Photocopies stick together, where r is the distance between the location of
dust collects on television screens, dirty parti- +Q and a point P, and e0 is the permittivity of
cles adhere to outdoor power transmission line the medium. From Equations 2-1 and 2-2, it can
insulators, etc.: Such phenomena are caused by be seen that the force acting on the charge q can
this force. In the following sections, we shall either be attractive or repulsive, depending
discuss the origin of this force. solely on the relative polarity of Q and q, as
shown in Figure 2-1.
In a neutral atom or molecule, or any neutral
2.1.1 Electric Charge Carriers and particle, the number of electrons is equal to the
Their Motion number of protons. When such a neutral entity
The most important electric charges are elec- loses one electron, it becomes a positive ion
trons and protons, which are the major elemen- (i.e., the entity has one proton which is not
tary particles of any atom of any material. Both neutralized by the electron); when the entity
the electron and the proton are the elementary receives an electron, it becomes a negative ion.
charges; they are equal in magnitude, but oppo- Free ions are just like free electrons and will
site in charge polarity. An electron has a nega- move in the presence of an electric field. A
tive charge -q, and a proton has a positive charged particle moving due to an electric force
charge +q, where q = 1.602 x 10"19 C. will have an acceleration a, based on Newton's
An electric field can be considered a region law, given by
in which a particle with a charge q would expe-
rience an electric force acting on it. This force,
denoted by J7, is given by
Since the charged particle is accelerated in the
T = qF (2-1) direction of the field, it exhibits a kinetic energy
where F is the electric field strength at the point
J-Mv2 = \jdr = jqFdr (2-4)
the particle is placed. The electric field strength
41
42 Dielectric Phenomena in Solids
where M is the mass, v is the velocity of the that produced by a point charge +<2 shown in
particle, (i.e., charge carrier), and r is the length Figure 2-1. In this case, the neutral particle will
of the path along which the particle has trav- be polarized, shifting the normally symmetrical
eled. In a nonuniform field, such as that pro- electron clouds of atoms or orienting existing
duced by a point charge +<2, the velocity of the permanent dipoles, if any, toward the direction
charge carrier varies from point to point. Sup- of the field. Electric polarization processes will
posing that the charge carrier is an electron and be discussed in Section 2.3. This neutral parti-
it is placed at point P at t = 0, v = 0, and r=ru cle, polarized by the field, will become an
then its velocity at point P' where r=r2 can be overall large dipole consisting of two charges
expressed as equal in magnitude but opposite in polarity, +qd
1/2 1/2 and -qd, separated by a small distance 5. At
2qQ enrndr
ar
JFdr point P\ where this dipole is located, the charge
-qd will experience an attractive force given by
k 1/2 (2-5)
2ne0m V rx f\=qF = -qd Q (2-6)
Ane0r2
It can be seen from Equations 2-4 and 2-5 that
the velocity and the kinetic energy of an elec- and the charge +qd will also experience a force,
tron or a negative ion increase as it travels which is repulsive, given by
toward a convergent field, and decrease when Q
it travels in a divergent field. j2=qF = +qd-
Suppose now we place a neutral particle with
a permittivity e2 in a nonuniform field such as In this case, JFI is larger than JF2 by an amount
equal to
Aj = l£-Jil = 9-fifl
Am„ (r + Sf
(2-8)
q<,Q 8
3
2n£„ r
since r is much larger than 5.

This force always tends to move the particle
from the weak field region to a strong field
region. It should be noted that, apart from this
force, there is another force arising from the
release of some stored potential energy of the
system: in the present case, the permittivity of
the particle e2 > £0> This force also helps to
move the particle to a strong field region
because this action would reduce the potential
energy of the system. This is why dust is
attracted to television screens and dirty parti-
cles are collected on the surface of power trans-
Figure 2-1 Nonuniform electric field F produced by a mission line insulators, which may lead to a
single point charge +Q, and the force between +Q and
various charged particles.
very harmful effect. This force depends on the
• Electron with a charge -q. relative magnitude of the permittivity of the
Q Negative ion with a charge -q. particles and that of the medium. We shall
© Positive ion with a charge +q. discuss this force further in a later section.
(-4) Dipole formed by a polarized neutral particle with Now let us consider the case with a constant
a net negative charge —qd on oiie end and a net
positive charge of +qd on the other at a separation
uniform electric field F. Such a uniform field
of<5. can be produced simply by connecting a steady
=> The direction of the force. DC voltage supply across two parallel metallic
Electric Polarization and Relaxation 43
+Q increases with t, the time required to travel a

unit distance decreases as x increases from 0.
Suppose it takes td time for the carrier to travel
D to x = d from x = 0, then the velocity of the
carrier at x = d can be expressed as
t t vd=atd (2-12)
Electron Ion
x=0 Thus, the average velocity can be written as
-Q
_._ atd (2-13)
Figure 2-2 Uniform electric field F produced inside a 2
capacitor by the uniformly distributed charge +Q on plate
A and -Q on plate B, the separation between two parallel
Since ^ = j / y , the average velocity can be
plates being d and the area of each plate being A. expressed in the form
1/2
fqFd\
plates and charging this system as a capacitor. (2-14)
KIMJ
After charging, the supply is disconnected and
removed from the system, leaving positive The mobility of the carrier is defined as the
charge +Q on plate A and a negative charge velocity per unit electric field strength, so the
-Q on plate B, as shown in Figure 2-2. The average mobility of the carrier u is the average
electric field created by the charges on the velocity per unit field strength. Thus,
plates can be considered to be uniform in _ _ v _ qd 1/2
the region between the plates if the edge effect (2-15)

U
~~F~ IMF
is ignored. This field F can be written as
The average velocity of the carrier in this case
F=Q_ = Î_ = 2_ (2-9) is proportional to (F)1/2, while the average
eA e e mobility of the carrier is inversely proportional
where e is the permittivity of the medium, A is to (F)m.
the area of the plates, as is the surface charge Now we will consider the case for solids,
density on the plates, and D is the electric flux such as semiconductors or dielectric materials.
density in the medium, which is equal to os if In this case, there are about 1022 to 1023 atoms
the medium is free space. Now, if we place an or molecules per cm3. It can be imagined that
electron or a negative ion at a point very close in solids, a charge carrier will not move freely
to x = 0, as shown in Figure 2-2, this charge as in free space. It will suffer many, many col-
carrier will move freely from x = 0 to x = d if lisions with phonons and impurities during its
the medium is free space with £ = e0. This movement from x == 0 to x = d in the same elec-
carrier will be accelerated along the field trode configuration shown in Figure 2-2. Based
direction without encountering any collision. on Equation 2-9, for the same charges on
Thus, we can write plates A and B, F will be smaller by a factor of
£
J = qF = Ma — = £r. An electron (or a hole) will react with
(2-10)
the lattice vibrations (i.e., phonons) and other
M
imperfections in the material, and its quantum
where M is the mass of the carrier, and M = m state is governed by Pauli's exclusion principle.
if it is an electron. The acceleration a is equal To take into account the effects of the inertia
to dv/dt, thus the velocity of the carrier can be due to all the interactions and to make the
expressed as motion of electrons or holes to follow the clas-
v = I adt =at sical Newton's law, we use the effective mass
(2-H)
Jo ra* instead of the conventional electron mass
where a is constant since F is constant, but v m0 in free space, which is practically equal to
increases linearly with traveling time t. Since v the rest mass of the electron because, for most
cases, the electron velocity is much smaller which is proportional to (T)m. It is usually
than the light velocity, so the relativistic effect much larger than the drift velocity. The temper-
can be ignored. ature dependence of drift velocity or mobility is
The electric field exerts on the electron an mainly due to the temperature dependence of r.
electric force qF, but the electron encounters a The collision time T is sometimes called the
friction force (called an inertia force or a mean free time or relaxation time. It is really the
viscous force) due to the collisions with time between collisions. Based on the quantum
phonons and imperfections (structural defects mechanics, an electron has a wave character.
and chemical defects or impurities). Assuming Thus, a wave passing through a perfect periodic
that this friction force, which opposes the lattice should travel freely without scattering (or
electric force, has the form m*v/r, where v is collision).1 However, the lattice is far from
the drift velocity of the electron and r is the perfect periodicity because thermal agitation
collision time, then based on Newton's law, causes lattice vibrations (or atomic vibrations),
we can write which generate phonons, and also various
defects (structural and chemical) in the material
* dV
T7 *
m* — = qF - m* —
V
(2-16) form scatterers to collide with electrons. In
dt X general, the probability of electron collision
The friction force tends to retard the motion of with phonons increases with increasing temper-
the electron. In the steady state condition, dvldt ature, while the probability of electron collision
= 0, and Equation 2-16 reduces to with defects decreases with increasing temper-
ature. That the mobility decreases with increas-
V = ?LF = HF (2-17) ing temperature is mainly due to the decrease of
r with temperature.
where ]X is called the electron mobility, which
Ions are much larger in size than electrons.
is defined as the drift velocity per unit field
It can be imagined that it would be much more
strength
difficult for ions to move in solids. Small ions
may be able to move by jumping from one
V = dv/dF = ^- (2-18)
interstitial site to the next, but large ions must
Electrons in a solid also have their thermal move from one lattice site to the next, which
velocity vthj due to thermal agitation. The direc- must be vacated for their occupation. Thus, the
tion of thermal velocity is random, and hence probability of ion movement depends on the
the random motion of electrons does not con- probability of the creation of a vacancy next to
tribute to electrical conductivity. Drift velocity the moving ion. For large ions, the activation
leads the electrons to move in the direction of energy for their movement involves several eV.
the electric field and hence contributes to the Donors or acceptors in semiconductors or insu-
electrical conductivity, which is defined as the lators, when ionized, become positive or nega-
electrical current density J per unit electric field tive ions, respectively. These ions, for most
strength. Thus, the electrical conductivity a is practical cases, cannot move at normal tem-
given by perature. Charge carrier transport processes
will be discussed in Ionic Conduction and
J qnv Electronic Conduction in Chapter 7.
a = - = -— = qjun (2-19)
F F
where n is the concentration of electrons. 2.1.2 Electromechanical Effects
Both thermal velocity and drift velocity are
temperature dependent. Thermal velocity is In an electrically stressed system, there is
given by always a force tending to reduce the stored
potential energy of the system. For a simple
1/2
capacitor, shown in Figure 2-2, the force tends
(2-20)
'{3m* J to increase the capacitance in order to reduce
the potential between the two charged plates so Case i (a): The field in the media par-
as to reduce the potential energy, since the allel (or tangential) to the
decrease in V would result in a decrease in boundary, as shown in
potential energy, which is equal to QV/2 or Figure 2-3(a). For this case,
CV2/2. If the medium inside the capacitor is air, the tangential fields in the
such a force would tend to attract any substance two media are equal.
with a permittivity larger than e0 into the capac-
F2t=Flt (2-22)
itor to replace the air medium. In the following
discussion, it is assumed for mathematical sim- Case i (b): The field in the media
plicity that the dielectric system is isotropic and normal (or perpendicular) to
linear, and heat losses and other effects, such as the boundary, as shown in
thermal agitation and gravitation, are ignored. Figure 2-3(b). For this case,
If the initial charges in the system are fixed, the relation of the fields in
then any mechanical work done for the reduc- both sides is given by
tion of the potential energy of the system must
be at the expense of the stored energy in the F - £ l
•Fln (2-23)
system. The change of the stored potential £2
energy AU in the system, due to the introduc- Case i (c): The field inclined with an
tion of a dielectric body of volume V2 and per- incident angle 0X incident on
mittivity e2 into a dielectric system of volume the boundary, as shown in
Vi and permittivity £x, can be evaluated by Figure 2-3(c). For this case,
we have
AU = ^j(e2-el)FlF2dV (2-21)
Flt=FlsmOu Fln=Flcos$l
where F\ is the electric field in the region of the
dielectric medium in the system of permittivity F2t = F2sin
i 02, F2n=F2cos02
£1 where a dielectric body of permittivity £2 is Based on Equations 2-22
subsequently introduced, and F2 is the electric and 2-23, we have
field in the dielectric body itself.2 The stored
potential energy of the system after the intro- tan0! _
duction of the body into the system is decreased tan 02 £2
when AU is positive, and increased when AU

This is the law of refraction
is negative. Using Equation 2-21, we have
(see Wave Theory in
derived the electromechanical forces resulting
Chapter 3).
from the change of the stored energy of the
system for the following cases. 3 2. A dielectric sphere with a radius a. For this
Before presenting our analysis of the cases, case, the field inside the sphere is given by
it is desirable to review briefly some formulas
3a
about the field distribution inside and outside a F2=- (2-24)
2£i + £2
dielectric body of permittivity £2 placed in a
medium of permittivity £l9 which is stressed in F2 < Fx for £2 > £i, as shown
a uniform static electric field Fx. The field in Figure 2-3 (d) and F2 > Fx
inside the dielectric body F2 depends on the for £2 < eu as shown in
geometric shape of the body. The formulas for Figure 2-3(e).
some common configurations shown in Figure A dielectric ellipsoid with its major axis 2c
2-3 are summarized as follows. 4-6 and 2 equal minor axes 2a and 2b, i.e., a
1. Medium 1 of permittivity ex and medium 2 prolate spheroid with c> a = b.
of permittivity £2 separated by a plane Case iii (a): The field in the medium par-
boundary. allel to the major axis. For
Sphere with e2 > £1 Sphere with % < £ i

(d) (e)
Spheroid with field parallel Spheroid with field normal

to major axis 2c and % > e i to major axis 2c and % > e i
(f) (g)
Figure 2-3 Electric field distribution in two media separated by a boundary, (a) Tangential field parallel to a plane
boundary, (b) normal field perpendicular to a plane boundary, (c) inclined field with an incident angle 0, incident on a plane
boundary, (d) a sphere of radius a having permittivity £2 in a medium having permittivity £\ with £2> £\, (e) same as (d)
but with £2 < £\, (f) a spheroid of major axis 2c having permittivity e2 in a medium of permittivity £\ with e2 > £1 a n d the
field parallel to the major axis 2c, and (g) same as (f) but with the field normal to the major axis 2c.
this case, the field inside the Case iii (b): The field in the medium
spheroid is given by normal to the major axis.
For this case, the field inside
(2-25) the spheroid is given by
ex-{£{-e2)Gu
F2=- (2-27)
where Gn is given by £x-{£x-£2)GL
ds where G± is given by
âc p as 26)
2 K
" 2 Jo (s + cy\s +a ) ds
G± = .r ™ (2. 28)
Jo (s + a ) / (5 + c 2 )
2 3 2 K
1
In this case, G/f < - , F2 < F{
In this case, G > 1/3, F2 < Fx
for e2 > £1, as shown in for £2 > £ b as shown in
Figure 2-3(f). Figure 2-3(g).
With these formulas, we can now analyze the tivity. Because of the presence of this force, we
following cases. would expect the occurrence of the following
electromechanical effects:
The Force Acting on the Boundary • If material 1 is air with ex = £b, and material
between Two Different Dielectric 2 is a dielectric solid or liquid with £2 > &u
Materials there is always a pulling force to pull mate-
At the boundary between two different materi- rial 2 into the space of air. Since the force is
als 1 and 2, the electric flux is usually refracted, proportional to F2, a big mechanical impact
and the tangential components of the electric on the dielectric solid may occur when the
field Fu and F2t, as well as the normal compo- insulation system is subjected to a high
nents, Fin and Flm follow the relationship based voltage transient or switching surge.
on Equations 2-22 and 2-23, which are • If material 1 is a dielectric fluid, such as
Flt=F2t (2-29) transformer oil or hydrocarbon liquids with
£i > e09 and material 2 is a particle with e2 <
Fln=e2F2n/£l (2-30) £i, then under a nonuniform electric field,
where ex and e2 are, respectively, the permittiv- such as that shown in Figure 2-1, the parti-
ities of materials 1 and 2. According to Equa- cle may be ejected from the strong field
tion 2-21, there is a force acting on the region because the force given by Equations
boundary with the tendency of expanding mate- 2-33 and 2-34 is more dominant than the
rial 2 into material 1 if e2 > £i- The decrease in force given by Equation 2-8. In this case,
stored energy due to the expansion of a small these two forces are present but, in opposite
volume dV of material 2 into material 1, due to directions. For example, in the strong field
the tangential component of the field, is then region near the high voltage windings of a
power transformer, in which transformer oil
</tf,=|(£2-ei)/ft/V (2-31) is generally used for both heat transfer and
insulation purposes, we would expect that
Thus, the force acting normally on the bound- any gas bubbles formed in the transformer
ary is oil would be ejected from the strong field
region, and any particles with e2 > ex and
djt =V(dUt) = ±(e2-el)VF?tdV (2-32) conductive particles, particularly moisture-
absorbed particles, would be attracted to the
since dTJdV = Vpt, where pt is the mechanical strong field region, resulting in many insu-
stress, which is the mechanical force per unit lation problems in power transformers.
area. Therefore, the mechanical stress due to
the tangential components of the field can be The Force Acting on Conductor Surfaces
written as
The attractive force between two parallel metal
plates with opposite electric charges, in contact
P,=|(£2-ei)*2' (2-33) with a dielectric material of permittivity £, will
bring about deformation within the volume of
Similarly, it can easily be shown that the the dielectric material in order to reduce the
mechanical stress due to the normal compo- stored energy of the system. The decrease in
nents of the field is stored energy resulting from a reduction of the
volume of the material by a small amount of dV
due to this compressive force can be written as
Both pt and pn are normal to the boundary, dU = -eF2dV (2-35)
tending to cause the dielectric material having 2
a higher permittivity to move into the space where F is the electric field strength produced
occupied by the other, having a lower permit- by the charges on the metal plates. Following
the same derivation procedure, the compressive tion of Equation 2-25 into Equation 2-21 gives
pressure acting on the dielectric material is the change in stored energy in the electrically
stressed dielectric fluid due to the presence of
Pc=-eF2 (2-36) a spheroidal bubble or globule, which is
2
This mechanical stress will result in a com- AUe =
frfe-fr) \F2(-Kr2 (2-38)
x
£x-{£i-e2)G\ V3
pression of the dielectric material, and its effect
on the breakdown strength of polymers has Thus, the work done at the expense of the
been reported by several investigators. 7 ' 8 For stored energy for the deformation of the bubble
example, a dielectric material, such as an insu- or globule is
lating polymer, having a relative permittivity of
4 stressed at a field of 2MV cm -1 , may suffer a AU = AUe-AUs
compressive pressure of about 273 newtons £ 1 (£ 2 -£ 1 ) 2 (1"3G)
(2-39)
cm" 2 . This pressure, which increases propor- (2£ 1 H-e 2 )[£ 1 -(£ 1 -e 2 )G]J
tionally with permittivity and the square of 4 J
applied electric field, may be sufficient, espe- -/rr
cially under transient overvoltage conditions, to
When AU is positive, the stored energy is
cause reduction of thickness or fracture of the
further reduced after the bubble or globule is
insulation, followed by dielectric breakdown.
degenerated into a spheroid. For AU > 0, G
must be less than 1/3, according to Equation 2-
The Force Elongating a Bubble or a 26. This corresponds to a prolate spheroid,
Globule in a Dielectric Fluid implying that the spherical bubble or globule is
elongated along the direction of the applied
A small bubble of gaseous or vapor phase or a
field, irrespective of whether the permittivity
globule of liquid phase of permittivity £2 in a
of the globule is larger or smaller than that
dielectric fluid of permittivity ex always takes
of the fluid.
on a spherical shape because of surface tension.
The force involved in the elongation can be
In the presence of an electric field Fx in the
derived as follows: By assuming that the
dielectric fluid, the field strength inside a spher-
volume of the bubble or globule is unchanged
ical bubble or globule is given by Equation 2-
during elongation (we make this assumption
24. Thus, by substituting Equation 2-24 into
purely for mathematical simplicity; there
Equation 2-21, we would obtain the change in
should be some change during the elongation
stored energy in the electrically stressed dielec-
process), we can write
tric fluid due to the introduction of a bubble or
globule, which is r3=ca2 (2-40)
3ex{e2~£x) When the sphere is slightly deformed by

AC/, = F?\±»> (2-37)
2e{ + e2 elongation of the major axis from the original
where r is the radius of the spherical bubble or 2r to 2(r + dc), then we have
globule, and Fx is the field strength in the region c = r + dc (2-41)
of the fluid which the bubble or globule subse-
quently occupies. It is assumed that a spher- and hence
oidal shape is a good approximation to that into
3dc
which a spherical bubble or globule may be -1 (2-42)
c2 (r + dcf r
expected to deform, and that the volume of the
bubble or globule remains unchanged while its the terms of higher powers in the binomial
shape deforms from a sphere into a spheroid. expansion being ignored. Substituting Equation
As the field strength inside a spheroidal bubble 2-42 into Equation 2-26 and then integrating it,
or globule is given by Equation 2-25, substitu- we obtain
Consider a nonuniform field, such as that pro-

(2-43)
3 r duced by a single point charge Q given by
Equation 2-2; the dielectrophoretic force can be
Substitution of Equation 2-43 into Equation 2-
expressed as
39 gives the decrease in stored energy due to a
bubble or globule with its major axis elongated 3£1(£2-£1) 4 Q2
Td —Tea3
from 2r to 2(r + dc) 2£i + €2 3 4n2£2r5
(2-47)
gi(g2-gi)2 T*(* 3
\dr?
d(AU) = - F?\ -Ttr —
(2eê2)2} V3 ,/ r This force depends on the distance from the
(2-44) point charge r and is opposed by a viscous drag
of the medium, which, according to Stokes's
Thus, the elongation force is law, is 6nr\ av, where r\ is the viscosity of the
d(AU) medium and v is the velocity of the particle.
Te Hence, the velocity of the particle in such a
dc
(2-45) nonuniform field can be approximately evalu-
gi(g2-gi)2 îV4
—;rr ated by
,(2£ 1 + £ 2 ) 2 3
The elongation phenomenon was first predicted 3g1(g2-g1) 2tf_ Q2

(2-48)
L 2ex + e2 47C2e2r5
based on a simple analysis by Kao,3'9 and later 97]
verified experimentally by Kao10 and further The dielectrophoretic effect may be used to
analyzed rigorously by Garton and Krasucki.11 produce pumping action of nonconducting
Electric breakdown of dielectric liquids has liquids, to cause separation of the components
been attributed to the formation and subsequent in suspension in a dielectric fluid, to cause pre-
elongation of vapor bubbles in the liquids.12 cipitation, or to produce mixing.13
The Dielectrophoretic Force

The Electrostriction Force
Unlike electrophoresis, in which the motion of
Electrostriction is defined as the elastic defor-
particles depends on the charge of the particles,
mation of a dielectric material under the force
dielectrophoresis is defined as the motion of
exerted by an electric field. A dielectric mate-
neutral particles in a nonuniform electric field,
rial can be thought of as consisting of dielectric
depending only on the force acting on polarized
particles (atoms or molecules) uniformly dis-
particles. The dielectrophoretic force tends to
tributed throughout a vacuum space. Thus, the
draw the particles whose permittivities are
local field acting on each particle is higher than
larger than that of the dielectric medium from
the apparent or measured field by an amount
the weak field to the intense field region, and to
resulting from the polarization field of the sur-
reject those with lower permittivities from the
rounding polarized particles. This local field is
intense field to the weak field region in the
given by
same manner described in Section 2.1.2. On
the assumption that the particles are spherical
in shape, the change in stored energy due to the loc (2-49)
3e0
presence of spherical particle of radius a and
permittivity £2 in a dielectric medium of where £ and e0 are the permittivities of the
permittivity ex and stressed at a field F\ can be dielectric material and vacuum, respectively.
calculated from Equation 2-37. Thus, the (See Section 2.5.) The polarization involves the
dielectrophoretic force can be written as displacement of electric charges, shifting the
originally symmetrical charge distribution to an
?d=V(AUd) nonsymmetrical one. This implies that the
'3^(62-gQ' (2-46) polarization tends to elongate the particles in
-itar
2£i+£2 the direction of the field, i.e., to cause the
material to expand in the direction parallel to Hence, the torque exerted on the body is
the field and to contract in the direction per-
pendicular to the field. This process is accom- T_d(AUT)
panied by a slight decrease in volume of the dO

material and hence a slight increase in density 27rca2£l(£2-£l)2(Gy-Gx)sm20
of the material. Suppose a material of volume " 3[£i ~(£i -£2)GX][£{ - ( d ~£2)Gy]
V suffers a slight decrease in volume by dV and (2-53)
hence a slight increase in density by dp, due
to elongation; then from Equation 2-21, the where
decrease in stored energy of this piece of mate- ds
rial due to the decrease in volume by dV is JO ( + c2Y?2(s + a2)
2 Jo
1
s
(€-e0)(e+2e0)'
d{AU)v =
l£o
F2dV (2-50) (fj 21- K
ll/2
Therefore, the hydrostatic pressure tending to

contact the material is 2lV2 1
d(AU)v (e-e0)(e + 2e0) + log

-[•-(%:)'] (2-54)
(2-51) 1/2
Pv = 1+
dV 6e0 [l-W
It should be noted that most solids are incom- _ca r°° ds
pressible because the particles (atoms or mole- (s + c2)V2(s + a2)2
'"~2~Jo7^7
cules) are rigidly bonded together. For slightly
compressible materials—such as some liquids,
polymers, or ceramics—electrostriction phe-
jH'/JT
213/2
nomenon may be detectable. Also, the hydro- M
static pressure may become significant for a 2-|l/2]
materials with high dielectric constants.
+ log-
1- H /J (2-55)
a
1+ H /c)}
The Torque Orienting a Solid Body Fx is the field strength in the region of the
dielectric medium where the solid body is
If a body is not symmetrical about its center, it
subsequently placed, and £x and £2 are the
will experience a torque, tending to orient itself
permittivities of the medium and the body,
in such a manner that the stored energy in the
respectively.
electric field will become minimal. For sim-
The torque tends to orient the major axis of
plicity, a prolate spheroidal shape is chosen as
the body to the direction of the field and is inde-
the approximation to the general shape of a
pendent of the relative magnitudes of £x and £2-
solid body, and its principal axes, which
The magnitude of the torque depends on the
follow the relationship c > a = b, form a rec-
size and the position of the body. It is zero when
tangular coordinate frame, x directing the major
the major axis of the body is perpendicular to
axis 2c, y and z, the minor axes 2a and 2b. If
the field, and is maximal when it is at the angle
one of the minor axes is originally in the direc-
of 45° to the field. The torque exerted on
tion of the field, the decrease in stored energy
the solid particles may accentuate the effect
when its major axis orients and causes an
of impurities on the dielectric breakdown of
angular displacement 0, from Equations 2-21
liquids, but on the other hand, it may be used
and 2-25, is
as a means to determine the permittivity of a
27cca2£l(£2 -£if(Gy -Gx)sin20 dielectric sample of known shape.
AUT =
3[£! -(ex -e2)Gx][el -(£! -£2)Gy]' n These six electromechanical forces may be
(2-52) appreciable when the impressed electric fields
are large. The directions of these forces are produced by A. If we now connect the conduc-
independent of the directions of the fields, tor B to ground by a conducting wire, electrons
implying that unidirectional and alternating will flow to B from the ground to make up the
fields will produce the same effects. deficit of electrons on the positive-charge side,
as shown in Figure 2-4(b). This is equivalent to
saying that some positive charge flows out from
2.1.3 Electrostatic Induction B to ground. Thus, the negative charge on the
Before discussing electric polarization and side close to A is called the bound induced
relaxation, it is desirable to review briefly the charge, while the positive charge on the oppo-
electrostatic induction phenomenon. site side is called the free induced charge. When
Suppose that we have a positively charged the free induced charge disappears, both the
insulator A with the charge distribution, as magnitude and the distribution of the negative
shown in Figure 2-4(a). If we now place an bound charge will change in such a way that the
uncharged conductor B near the charged insu- electric field inside B remains zero. Now, if the
lator A, the electrostatic field produced by A ground connecting wire is cut off and the con-
will shift the free electrons in B toward the ductor B is removed from the field produced by
surface close to A, resulting in the formation of A, then the conductor B becomes an isolated
a negative charge on one side and a positive negatively charged conductor, and the bound
charge on the opposite side, where there is a induced charge becomes free to move to redis-
deficit of electrons, as shown in Figure 2-4(a). tribute itself on the surface in such a way that
This phenomenon is referred to as electrostatic the electric field inside B is zero and the elec-
induction. The electric field inside B is zero tric flux outside B is perpendicular to the
because this field is composed of two compo- surface of B, because any tangential compo-
nents: one is the original field from A, which nents of the electric field would cause the
should always exist inside B, and the other is movement of surface charges. The charge dis-
the field produced by the induced negative and tribution is dependent solely on the geometric
positive charges in B, which is opposite to that shape of the conductor B. It should be noted
Bound Charge Free Charge

Uncharged Conductor
F=0
Charged Insulator Uncharged Conductor Charged Insulator
(a) (b)
Dipoles
D
Charged Insulator
Insulator
(c)
Figure 2-4 Electrostatic induction: (a) positively charged insulator A and uncharged conductor B (isolated), (b) positively
charged insulator A and uncharged conductor B (grounded), and (c) positively charged insulator A and isolated insulator D.
that any change in charge distribution in con- polarization and relaxation processes are dis-
ductor B due to different arrangements does not cussed in the following sections.
affect the charge distribution in the charged
insulator A, simply because charges in insula-
tor A cannot freely move. However, if A is a 2.2 Electric Polarization and
charged conductor instead of a charged insula-
Relaxation in Static Electric Fields
tor, then the induced charge in conductor B
will also cause a change in charge distribution Electric polarization refers to a phenomenon of
in charged conductor A, in order to make the the relative displacement of the negative and
electric field inside both A and B be zero. positive charges of atoms or molecules, the ori-
A conductor, or a conducting material, entation of existing dipoles toward the direction
generally refers to the material consisting of of the field, or the separation of mobile charge
a great many mobile free charge carriers, such carriers at the interfaces of impurities or other
as a metal in which the concentration of free defect boundaries, caused by an external elec-
electrons (mobile charge carriers) is of the tric field. A detailed account of various polar-
same order as that of atoms, i.e., about 1022 to ization processes will be given in Section 2.3.
1023cm-3, and electrolytes in which ions are the Electric polarization can also be thought of as
charge carriers. In those conducting materials, charge redistribution in a material caused by
the electrostatic induction phenomenon pre- an external electric field. The work done for
vails. But in other materials, such as semicon- the charge redistribution and the energy loss
ductors and dielectric materials, the number of involved in the redistribution process require
mobile charge carriers is far less than the number an energy supply. Where does this energy come
of atoms. Although an electric field would cause from? We can say that the whole polarization
the movement of mobile charge carriers to process is performed at the expense of the
produce space charge polarization in a manner potential energy released from this process,
similar to electrostatic induction, space charge because the total potential energy of the system
polarization usually plays an insignificant role in an electric field is smaller after electric polar-
because it involves only a small number of ization than before it.
mobile charge carriers, as compared to electric We shall apply Gauss's law to explain some
polarization, which involves all atoms. dielectric phenomena resulting from electric
Now, suppose the uncharged conductor B, polarization. As discussed in Gauss's Law in
shown in Figure 2-4(a), is replaced by an iso- Chapter 1, Gauss's law simply states that the
lated uncharged insulator D of the same shape, electric flux outward from a volume is equal to
shown in Figure 2-4(c). In this case, the field the total net charge enclosed inside it. We use
produced by the charged insulator A will not a simple system consisting of two metal plates
induce charge in D because there are practically parallel to each other with an area A each and
no mobile free charge carriers in insulator D. a separation d, which is much smaller than the
Instead, the field will polarize the material by linear dimension of the plates, so that the fring-
shifting slightly the normally symmetrical dis- ing effect at the edges can be ignored by
tribution of electron clouds of atoms and by ori- approximation. Suppose we introduce a posi-
enting dipolar molecules toward the direction tive charge +Q on the upper plate and a nega-
of the field to form dipoles. So, each atom tive charge -Q of the same magnitude on the
or molecule contributes a tiny dipole to form a lower plate. This can be easily done by con-
big dipole, as shown in Figure 2-4(c). If we necting a steady DC voltage supply across the
remove the insulator D after it has been com- plates and charging the system as a capacitor,
pletely polarized, the polarized insulator D will and then disconnecting the supply from the
undergo relaxation and gradually become system as soon as the charge has been
depolarized, due to thermal agitation. Electric accumulated to the desired value Q. These
charges on the plates will create a potential dif- +Q

ference between the plates V that is propor- • ^ t t + t t t t t + t t t t t + t t t f >
tional to Q. Thus, we can write
T.—>
Plate A
r—V"The volume containing
VaQ D0 = £0Fo F
o free surface charges
(2-56) of density oso
V =Q or Q = CV
Plate B
where C is the proportionality constant, gener- \
ally called the capacitance. By denoting the -Q
surface charge density on the plates as <JS, the
charge Q can be expressed as Figure 2-5 The electric field F0 and the associated elec-
tric flux density D0 produced by the free surface charges of
Q = C(Fd) = asA (2-57) density oso on the plates in vacuum space (free space).
where F is the electric field strength, which is

simply equal to V/d. From Equation 2-57 we
£o = " (2-59)
can express C in the form F0
Q osA A where F0 is the electric field strength in the

C= e— vacuum space, which is equal to Q/V0 and V0 is
Fd d
the potential between two plates created by the
From this we obtain charge Q\ and £0 is the permittivity of vacuum
or free space, which is practically the same as
£=- (2-58)
the permittivity of dielectric gases, such as air,
at normal pressure and temperature. The per-
Where £is the permittivity of the material filled
mittivity of gases, depending on the molecular
between the two plates, which is equal to the
structure, has only negligibly small deviation
ratio of — . In the following sections, we shall from the ideal value of e0 at normal pressure
and temperature.14
discuss this ratio for three general cases.
2.2.2 Conducting Materials

2.2.1 Vacuum Space A conducting material, or a conductor, refers to
By true vacuum space, or free space, we mean a material containing many mobile free charge
that in the space there are no detectable parti- carriers. For example, in a piece of copper or
cles, although there may be a hypothetical sodium, each atom contributes one free elec-
ether, thought of as the medium for the propa- tron, making the total electron concentration
gation of electromagnetic waves in free space. equal to the concentration of atoms, i.e., 1022 to
However, a true vacuum is not available on the 1023cm"3. These electrons move freely and ran-
earth. The best vacuum system that can be pro- domly and distribute themselves with statistical
duced by today's technology can achieve only uniformity in the bed of regularly arranged pos-
a vacuum of about 10-14 torr, in which there are itive ions. If such a piece of metal with the
still about 300 particles per cm3. Assuming that thickness of d-2s were inserted into the space
the space between the plates is free space, aso between the two charged plates, with a small
is equal to the electric flux density D0 based on gap s to separate the plates and the metal con-
Gauss's law, as shown in Figure 2-5. ductor, then the charge on the plates would
So we can write cause the electrons in the conductor to move
toward the surface close to the positively
O so *-*o Ô"O
charged plate but could not leave the surface,
or resulting in a nonuniform distribution of the net
charge. This creates a net negative space charge

£s = - (2-60)
on one surface and a net positive space charge
on the other in a manner similar to the elec-
In metal, the permittivity is °o under static fields
tronic polarization of a neutral atom. Thus, in
only. Under a time-varying field or an alternat-
this case, we can say that charge Q on plate A
ing field, the charges induced on the metal
induces a negative charge equal in magnitude
surfaces may not follow instantaneously the
but opposite in polarity on one surface of the
time-varying field. In this case, the permittivity
metal close to plate A. Similarly, a positive
is not infinite, but a finite value. We shall
charge is induced on the other surface of the
discuss this phenomenon further in later
metal by the negative -Q on plate B. So, in the
sections.
vacuum space between plate A and the metal
surface, or between plate B, and the other
surface of the metal, we have 2.2.3 Dielectric Materials
One of the important electrical properties of
dielectric materials is permittivity (or relative
e =—— permittivity, which is generally referred to as
F
1
o the dielectric constant). For most materials, the
which is the same as that as given by Equation dielectric constant is independent of the elec-
2-59. The response in the vacuum space is tric field strength for fields below a certain
exactly the same as that described in Section critical field, at or above which carrier injection
2.2.1. into the material becomes important. The
However, inside the metal there is no elec- dielectric constant depends strongly on the fre-
tric flux based on Gauss's law, as shown in quency of the alternating electric field or the
Figure 2-6, and hence, F = 0. rate of the change of the time-varying field. It
+ + + + + + + + +
+ + + + + + + + +
+ + + + + + + + +
+ + + + + + + + +
+ + + + + + + + +
++++++++++++i
Positive Ion Free Electron Net Charges on
Distribution Distribution the Surfaces of
Metallic Conductor
The volume containing a positive (a)
charge on the plate surface of
density QIA and an induced
negative charge on the conductor
surface of density -QIA, so the
total net charge is zero.
*• + + + + + *
Vacuum space
Metal F=0
++++++++++++++++++++++
s Vacuum space
T
(b)
Figure 2-6 (a) Induced net charges on the surfaces of a metallic conductor in the presence of a static electric field, and
(b) the charge on the plate with the surface charge density <7S inducing a charge equal in magnitude but opposite in polar-
ity on the surface of the conductor, resulting in zero electric field inside the conductor. The vacuum space s is exaggerated
for clarity.
also depends on the chemical structure and the The volume containing free surface
imperfections (defects) of the material, as well charge of density <ys - ab and bound
surface charge of density ab, which
as on other physical parameters including is for compensating the polarization
temperature and pressure, etc. This chapter is surface charge of the dielectric material.
devoted to the discussion of electric polariza-
tion and relaxation under various conditions. I +S+ffl +BB+S+E + + +BB N
N ;
A dielectric material is made up of atoms or J§fo&&®®&&®-ffir
molecules that possess one or more of five basic
types of electric polarization: e„F
1. Electronic polarization ©©© ©©©©©©
2. Atomic or ionic polarization -B-B-B-B-B-B-B-B-B-
3. Dipolar polarization Figure 2-7 The surface charge of density Gs consisting of

4. Spontaneous polarization two portions: the bound charge Gb and the free charge os -
Gb. The free charge portion produces the electric Field F
5. Interface or space charge polarization. and the electric flux density of EoF, while the bound charge
Each type of polarization requires time to portion produces polarization P. + and - denote the free
positive and negative charges, respectively, and El and
perform; this is why the degree of the overall B denote the bound positive and negative charges,
polarization depends on the time variation of respectively.
the electric field. In the present section, we con-
sider only the polarization under static fields. in Figure 2-7. This portion of charge is termed
In the first place, we consider a perfect dielec- the bound charge density ob. By denoting the
tric material. The so-called "perfect" material number of atoms (or molecules) per unit
implies that inside the material no mobile volume of the material as N> and the average
charge carriers (electrons or ions) are present. separation between charges +qd and —qd of each
If a piece of such a dielectric material is put dipole (each atom or molecule produces one
near the system shown in Figure 2-5, the dipole) as S, we can write
system will tend to attract this piece of material
into the vacuum space between the two plates ab = N{qd8) (2-61)
in order to reduce the potential energy of the The other portion os - cb is the free surface
system. charge density, which acts in exactly the same
Suppose now that the piece of the material is manner as that in a vacuum, creating an elec-
inside the space between the metal plates with tric flux density D0 = e0F. Thus, we can write
the original surface of density os unaltered, as
shown in Figure 2-7. This causes the potential as = (os-ob)+<yb
between the plates produced by the original = D = D0 + P (2-62)
charge Q on the plates to decrease to a smaller = esF = e0F+(es-e0)F
value. In fact, with Q on the plates remaining
constant, the ratio of the electric field in free Therefore, polarization P can be expressed as
space F0 to that filled with the dielectric mate- p = (es- £0 )F = <7b = polarization
rial F is the so-called dielectric constant or rel- _ Bound charge
ative static permittivity, esr = esle0. - = N{qdd)
Surface area
With the dielectric material, one portion of _ Number of Induced Dipole Moments
as is used to compensate the polarization
Volume
charges on the surfaces of the material in
= N(u)
contact with the metal plates. This portion of ab
(2-63)
is bound at the locations with its charge oppo-
site in polarity and equal in magnitude to the where (u) is the average value of the dipole
polarization charges of the material, as shown moment, which is given by
(u) = (qdS) = aFioc (2-64) It should be noted that this discussion is

limited to the phenomena under a static field.
where a is called the polarizability and is a This means that the charges +Q and -Q already
scalar quantity if the constituent particles (atoms put on the plates are constant and that the
or molecules) in a material are spherically sym- dielectric material has been inserted into the
metrical in shape and (u) is in the direction of vacuum space for a time period long enough for
F. However, if the particles (atoms or mole- all types of polarization to settle down to their
cules) are not spherically symmetrical in shape, steady state values. Now, supposing that it takes
particularly those with permanent dipoles, then a negligibly small amount of time for the inser-
(u) is not exactly in the direction of F. In this tion of the dielectric material into the vacuum
case, a becomes a tensor. Throughout this book, space, we can assume that at time t = 0 the
we are interested mainly in physical concepts material is already inside the vacuum space and
and, therefore, to maintain clarity and simplic- starts to be polarized by the field created by the
ity rather than to involve lengthy mathematics, charges on the metal plates. The polarization
we consider only homogeneous materials and due to the elastic displacement of electron
electric fields that are independent of space clouds of the particles (atoms and molecules)
coordinates, although they may vary with time. requires a very little time, while the polariza-
However, it should be noted that (u) depends on tion involving the movement of the particles,
the local field at which the individual particle is such as the orientation of permanent dipoles or
polarized. This field Fhc is different from the the migration of charge carriers (electrons or
applied external field F; therefore, by making ions), requires a much longer time to perform.
(u) proportional to F, the polarizability a must All types of polarization encounter some inertia
be related to the local field. From Equations 2- counteracting the change and, therefore, in-
63 and 2-64, we can also write volve some dielectric loss, as shown in Figure
a = elZe^L (265)
2-8.
N In the system being considered, Gs on the
plates is constant, but the permittivity starts to
We can also say that the polarizability a of a rise from e = e0 and the corresponding field to
particle in a dielectric material is induced by the decrease from F = F0 at t = 0. It takes a very
local field. We will see this point more clearly little time for e to increase to £«, and for F to
in Section 2.3. It is sometimes convenient to use decrease to E^ due to the polarization con-
the relative permittivity esr (i.e., the dielectric tributed by the elastic displacement of electron
constant) to describe the dielectric properties clouds of atoms or molecules (electron and
atomic polarization). The time involved in this
polarization is of the order of 10"14 to 10~13
second. We shall discuss e* further in Section
Na 2.6. However, for the permittivity to increase
where X is called the electric suscepti-
ve from £oo to its steady state value es and for the
bility, which can also be written as corresponding field to decrease from FU to its
final value Fs, a considerably larger amount of
, P time is required, because this change is caused
by the polarization associated with the inelas-
Bound surface charge density
_ on the plates
tic movement of particles such as the sluggish
collective orientation of dipoles or the migra-
Free surface charge density on the plates
tion of charge carriers to form space charges
(2-67)
near the electrodes or grain boundaries. Ignor-
X is the electric counterpart of the magnetic ing the small loss due to elastic displacement
susceptibility described in Magnetization in of electron clouds, the transition from P^ -
Chapter 1. (e^r - l ^ F ^ to Ps = (esr - 1)£0FS involves some
Polarization due to
inelastic movement
of charged particles
Polarization due to (a)

elastic displacement
of electron clouds of
particles
(b)
Figure 2-8 (a) The variation of the susceptivity er - 1 and the electric field F with time, and (b) the polarization P *
(sr- 1) e0F and the dielectric loss with time, where Px = (£« - 1) BQF^ and Ps = (eSr - 1) e0Fs.
energy loss, which must be consumed to over- After the overall polarization process is com-
come the inertia resistance. This energy loss is pleted, we short-circuit the two metal plates.
called the dielectric loss, which always accom- Under this short-circuiting condition, the
panies time-varying electric fields. At static charges on plate A and plate B will be immedi-
fields, it appears only during the transient ately neutralized, leaving only the polarized
period in which the electric field across the particles in the dielectric material to be gradu-
dielectric material is time-varying, although the ally depolarized. It would take some time,
charges on the metal plates are constant. This depending on the environmental temperature,
loss per unit time will become zero when the for all polarized particles to become completely
polarization has reached its final steady value depolarized, particularly for those whose polar-
e = es. The loss is proportional to the PF ization involves the orientation of permanent
product; thus the loss per unit time becomes dipoles or the migration of charge carriers.
maximal when the PF product is maximal, as After the short-circuit current, due to the neu-
shown in Figure 2-8(b). tralization of the charges on the metal plates, a
reverse depolarization current will follow, due or

to the depolarization of the polarized particles.
dCTs
The thermally activated depolarization process (Pd) = (2-68)
dt . £s J
will be discussed in Charge Storage Involving
Dipolar Charges in Chapter 5, which deals with using the boundary condition os (t = o) = as (o),
electrets. the solution of Equation 2-68 gives
In reality, there is no perfect dielectric mate-
os(t) = os{o)e •t/td (2-69)
rial. All real dielectric materials contain, more
or less, some mobile charge carriers. If such a where rd = pes is the dielectric relaxation time
real material has the total charge carriers n per and p is the resistivity of the material. xd is the
unit volume, each carrying a charge q and time required for the originally induced charge
having an average mobility un, then the con- to decay to 36.7% of its original value. xd can
ductivity of this material a is qunn. For a dielec- be very large for materials with a very high
tric material with low conductivity, we can use resistivity (i.e., low conductivity). So, it is quite
an equivalent circuit, shown in Figure 2-9, to dangerous to touch an electrically disconnected
describe the response of this material under an capacitor, because some charges may still
electric field produced by two charged metal remain on the electrodes even though the
plates similar to that shown in Figure 2-7. As capacitor may have been de-energized for
mentioned, the parameters e and Rd are derived some time. For example, if es = 3e0 and p =
directly from the polarization process. Thus, at 1018Q - cm, then rd = 2.65 x 105 seconds = 3
t = 0, just after the insertion of the material into days. This means that it may take many days
the vacuum space, e -» £*> and Rd -> ©o, and at for the charges to be completely neutralized.
t = ts, e —> es and Rd —> «>. The value of the
resistance Rd is finite only during the period 0
< t < ts. But the leakage current through the
2.3 The Mechanisms of
leakage resistance Rc will continue to drain the
Electric Polarization
charge from plate A to plate B, making as decay
with time. By expressing the voltage between The polarizability defined by Equation 2-64
the two plates as implies that if (u) is proportional to the local
field Flocal at which the particles are polarized,
then (u) is expressed as (w) = aFhcai and the
dt polarizability a depends only on the mecha-
cx« = Q/A nism of polarization and can be defined simply
as the average dipole moment per unit field
strength of the local field Fiocah which is differ-
ent from the externally applied field F. But if
(u) is expressed as (u) = aeffF, i.e., (u) is pro-
/ , portional to F, in this case, aeff should also be
Rr
defined similarly to the effective dipole
moment per unit field strength of the applied
field F. However, aeff depends not only on
the mechanism of polarization, but also on the
factor of FtocIF. We shall consider first the
mechanisms responsible for the polarizability,
Figure 2-9 The equivalent circuit of a real dielectric a, and the local field later.
material. Rc = pd is the leakage resistance of the material
There are three major mechanisms of elec-
where p is the resistivity; C = eld is the capacitance per unit
area, where e varies with time; Rd is the resistance per unit tric polarization. At moderate electric fields
area derived directly from the polarization process and (i.e., at fields much lower than the inner atomic
varies with time. or molecular fields), and for materials with a
very low conductivity (i.e., the concentration of polarization involves the inelastic movement of
charge carriers inside the materials is so low particles, and its interaction is an intermolecu-
that its effect can be neglected), these mecha- lar phenomenon; hence, orientational polariz-
nisms prevail. ability is strongly temperature-dependent.
At higher fields, carrier injection becomes
• Electronic polarization (also called optical important. For materials consisting of a high
polarization): The electric field causes defor- concentration of charge carriers (i.e., with a
mation or translation of the originally sym- high conductivity), polarization due to the
metrical distribution of the electron clouds migration of charge carriers to form space
of atoms or molecules. This is essentially the charges at interfaces or grain boundaries
displacement of the outer electron clouds becomes important. This type of polarization is
with respect to the inner positive atomic called space charge polarization.
cores. Thus, the total polarizability of material a
• Atomic or ionic polarization: The electric comprises four components
field causes the atoms or ions of a poly-
a = ae+ai+a0+ ad (2-70)
atomic molecule to be displaced relative to
each other. This is essentially the distortion where ae, (%, a0, and ad are the polarizabilities
of the normal lattice vibration, and this is due to electronic, atomic, orientational, and
why it is sometimes referred to as vibra- space charge polarizations, respectively. For
tional polarization. ferroelectric materials, there is also a compo-
• Orientational polarization: This polarization nent, called spontaneous polarization, under
occurs only in materials consisting of mole- certain conditions. The following sections
cules or particles with a permanent dipole discuss the mechanisms responsible for these
moment. The electric field causes the reori- polarizations.
entation of the dipoles toward the direction
of the field.
2.3.1 Electronic Polarization (Also Called
Both the electronic polarization and the atomic Optical Polarization)
polarization are due mainly to the elastic dis- There are two ways to calculate electronic
placement of electron clouds and lattice vibra- polarizability: the classical approach and the
tion within the atoms or molecules. Their quantum-mechanical approach. We will first
interaction is an intramolecular phenomenon, discuss the classical approach.
and the restoring force against the displace-
ment is relatively insensitive to temperature, so
electronic and atomic (or ionic) polarization Classical Approach
processes are only slightly dependent on tem- The polarizability ae depends on the atomic
perature. However, orientational polarization is number of the atom and predominantly on the
a rotational process, which encounters not only number of electrons in the outermost shell. The
the resistance due to thermal agitation, but also noble gas atoms, such as He, Ne, Ar, Kr, Xe,
that due to the inertia resistance of the sur- and Ra, whose shells are completely filled
rounding molecules, giving rise to mechanical (effectively screening the nucleus from the
friction. The rotation of a dipole in a material influence of the applied fields), have the lowest
is like a small ball, or any body, rotating in a polarizabilities for their atomic numbers. The
viscous fluid. Under an external force, it tends Group I elements, such as H, Li, Na, K, Rb, and
to change from its original equilibrium state to Cs, with only one electron in the outermost
a new, dynamic equilibrium state, and when the shell, have the highest polarizabilities for their
force is removed, it then relaxes back to its atomic numbers, possibly due to the ease of
original equilibrium state. This process is gen- polarization of a single electron in the outer-
erally referred to as the relaxation process. This most orbit.1516 Electronic polarizability can be
The dipole moment is given by

ue = aeFloc = Zq8 = 47ie0R3Fl0C (2-75)
Therefore, the electronic polarizability is
ae = 4ne0R3=3e0Va (2-76)
where Va is the volume of the atom.
ae is proportional to the volume of the atom.
For the hydrogen atom, R is about 0.50 A,
ae = 1.57 x 10"24£ocm3. For Floc = lO'VcnT1,
8 = 10"14cm. The displacement distance 8 is
extremely small, only of the order of two-
millionths of the radius of the atom. The sus-
ceptibility Xe = N(u)/e0Fl0C = N4nR? = 1.57 x
10~247V. For the ideal gas under normal condi-
tions (0°C and 760 mm Hg), the number of par-
Figure 2-10 The displacement of the electron cloud ticles per unit volume is 2.687 x 1019cm~3,
relative to the nucleus of positive charges Zq due to the which is also known as the Loschmidt number,
polarization by the local field Ftoc.
and xe = 1.57 x 10~24 x 2.687 x 1019 = 4.22 x
10"5. These numerical data give us a feeling for
the orders of magnitude involved in the micro-
measured in monoatomic gases. Suppose that
scopic parameters. There is another classical
the electron cloud of charges -Zq is uniformly
method that can be used for the calculation of
distributed in a sphere of radius R and that its
ae, i.e., using Bohr's model with the electrons
center of gravity originally coincided with that
revolving around a circular orbit instead of
of the nucleus, and suppose that it is displaced
using a uniform electron cloud density up to
by the field to a distance 8 from the center of
radius R as just shown.
the nucleus, as shown in Figure 2-10. There is
This displacement of the electrons At, result-
a coulombic force tending to bring the electron
ing from electronic polarization, always gener-
cloud back to its original position.
ates an elastic restoring force tending to bring
According to Gauss's law, the coulombic
the electrons back to their equilibrium position.
force is only exerted on the electron cloud that
This restoring force is proportional to the dis-
does not surround the positive nucleus charges
placement Ax, resulting from the interaction of
+Zq. This portion of the electron cloud is con-
the electron with the bare nucleus and with
tained in the sphere of radius 8 and it is
other electrons screening the nucleus. The elec-
(2 71> tron orbiting the nucleus is like a harmonic
o-zgfijp-w* - oscillation with a natural frequency co0. Thus,
the motion of the electrons is governed by the
Thus, the coulombic force is following equation:
I = z^s^=(Zq)2d/4Ke^ (2.72)
m ^ = -y&x-ZqFloc (2-77)
This force must balance the displacement force,
where m is the electron mass, q is the electronic
which is
charge, Z is the number electrons involved, Fhc
Td=ZqFloc (2-73) is the local field acting on the atoms, and y is
the force constant. By assuming, for simplicity,
By setting Equation 2-72 equal to Equation 2-
that when the centroid of the negatively
73, we obtain
charged electrons is displaced by AJC from
8 = 47te0R3Fl0C/Zq (2-74) that of the positively charged nucleus, the
coulumbic force is the restoring force, which and hence the electronic polarizability is
can be expressed as
(Zq
«.= . 2 l .- (2-86)
Zq (Zq)2 m(G)i-CQz) + jpcQ
>Zq Ax = yAx
4ne0(Ax)2 (T)'- Ane0R:
In static fields, co= o9 and the static electronic
(2-78)
polarizability becomes
In this case, the restoring force constant is
(Zqf
oce • (2-87)
... (Zqf (2-79) mcOn
4ne0R3
Based on Bohr's model of an electron
The restoring force can also be expressed in revolving around a circular orbit about the
terms of the natural oscillation frequency (0o by nucleus, the potential energy of the electron is
the following relation: given by
yAx = mco2Ax (2-80)
E = h(Q0 = - mq (2-88)
Thus, {47te0fh2
1/2 and the radius of the ground-state orbit of
-© (2-81) Bohr's atom is given by
However, there must be some loss or damping (47T£0)h2

R= (2-89)
mechanisms to limit the forced oscillation. The mq2
oscillation of an electron is equivalent to an
where fi = hlln and h is Planck's constant.
oscillating dipole and would radiate energy
When Z = 1, by expressing the parameters m
according to the electromagnetic theory of radi-
and co0 in terms of R and e09 and substituting
ation. Thus, radiation can be considered one of
them into Equation 2-87, we obtain
the damping mechanisms. Therefore, a term
dx
ft—, representing the retarding force, should ae = = 4TT£0R3 (2-90)
at ma>2
be included in Equation 2-77. By including this
Amazingly, Equation 2-90 turns out to be the
term and substituting Equation 2-81 into Equa-
same as Equation 2-76, although the two equa-
tion 2-77, we obtain
tions are derived on the basis of different
d2Ax assumptions. For the hydrogen atom, ae = 1.57
dt2
- + met)2Ax - -P^-m, (2-82)
x 10- 24 £ o cm 3 . Taking q = 1.6 x 10~19C and m =
where j8 is the damping coefficient. Damping 9.11 x 10~31kg, we have estimated the value of
mechanisms will be discussed further in (00 from Equation 2-90; it is 4.5 x 1016 radians
Section 2.6. If the field is an alternating field sec"1.
with the frequency (0 expressed as The average electronic polarizability is
obtained by the summation of the contributions
Floc = Fo(loc)e-^ (2-83)
of all atoms divided by the number of atoms.
then the solution of Equation 2-82 gives the net We can consider only the contributions of the
displacement of the electrons by the applied outermost electrons of each atom, and ignore
field, which is the negligibly small contribution of inner elec-
trons. From Equation 2-66, the electronic sus-
ZqFiol
Ax = - 2 2 (2-84) ceptivity is given by
m(o) -co ) + jP(0
The induced dipole moment is Nae (Zqf
Xe=~ (2-91)
mcOo
(ZqfFloc
— (XeFloc — 0CeUff)F
m(co2-CQ2) + Jl3(D and the relative permittivity, or dielectric con-
(2-85) stant, is
N (Zqf Single atoms in gases may be considered

£r = 1 + ^ = 11 + — i - ^ - (2-92)
e0 meal spherical in shape, but molecules, each com-
prising two or more atoms, are generally not
the natural frequency (00 should vary from atom spherical. In this case, the geometric shape does
to atom. However, for gases, the separation play an important role in determining ae. We
between atoms is large enough, and it is quite will choose a diatomic molecule with a prolate
acceptable to assume that the local field Floc, at spheroidal shape to demonstrate the calculation
which the atom is polarized, is equal to the of the electronic polarizability of this molecule.
applied field F and that the interaction of the We assume that the two atoms are identical;
outermost electrons between atoms is negligi- each has an electron polarizability ae, which is
bly small and can be completely ignored. So, in proportional to its volume, as given in Equation
this case, we can assume that each atom has the 2-76. There are two extreme possible arrange-
same co0 and hence, the same ae. Equation 2-92 ments, as shown in Figure 2-11 (a) and (b): the
indicates that the dielectric constant increases molecular axis ab is parallel to the applied field
with the increase of the density of atoms. For a (abllF), or the axis ab is perpendicular to F
hydrogen atomic gas, N is about 2.7 x 1019 (ablF*). For the case ab//F, the dipole of each
atoms cm"3 at the standard temperature and atom consists of two components, one due to
pressure. Thus, the dielectric constant for the the applied field F and the other due to the field
hydrogen atomic gas is er = 1 + 2.7 x 1019 produced by the other polarized atom, which is
x 1.57 x 10"24 = 1 + 4.24 x 1(T5. N of gases is in the same direction as F. Thus, we have
much smaller than that of solids, which is of the
order of 1022 to 1023cnT3. Furthermore, in F+-
ii//
(2-93)
un=at
solids, the interaction of outermost electrons 2ne0ab
between atoms cannot be ignored. Rearrangement of Equation 2-93 gives
It should be noted that only in inert gases,
such as He, A, etc., the quantum states of the out- ae •F = a'eF (2-94)
«// =
ermost shell are completely filled, so the inert \-aej2KE0ab
gases are atomic gases. In other gases, including
where ab is the distance between the two atoms
hydrogen, oxygen, etc., the gases are formed by
and a'e is the effective polarizability of each
molecules such as H2, and 0 2 instead of H or O.
atom. Thus, the electronic polarizability of this
In this case, the particles are no longer spherical diatomic molecule is
in shape, but prolate ellipsoidal or spheroidal,
and we shall see that geometric shape does play 2at
an = 2a'e • =->2ae (2-95)
an important role in determining ae. l-2(R/abf
F • F •
Major Axis
(a) (b)
Figure 2-11 (a) The major axis ab of a spheroidal molecule parallel to the applied field F, (b) the major axis ab of a spher-
oidal molecule perpendicular to F, and (c) the major axis ab of a spheroidal molecule at an angle 6 with respect to F.
In this case, a„ is larger than twice the polariz- in liquids or gases, they will tend to align their
ability of a single atom. major axes with the applied field. It should be
For the case ablF, the dipole of each atom noted that the induced dipoles in the molecules
also consists of two components, one due to the due to electronic polarization also involve the
applied field F, and the other due to the field orientation of the molecules in a manner similar
produced by the other polarized atom, but this to the orientation of the permanent dipoles.17
field is in the opposite direction to F. Thus, we The field strength produced by the nucleus
have charges and experienced by the electrons is
ZqlAKSoR2 is larger than 109V/cm. This field
u± =ae uL (2-96) strength is far larger than the applied field. This
may be the reason why the induced dipole
moment for electronic polarization is propor-
Rearrangement of Equation 2-96 gives
tional to the applied field and independent of
ae frequency, because the time required for this
u± =l + ae/47ie0ab-F
: = a'e'F (2-97)
polarization to occur is of the order of 10"15 sec
(a)"1). This type of polarization is also called
where o^'is the effective polarizability of each
optical polarization, because it does not vary
atom. Thus, the electronic polarizability is
until the frequencies are in the optical range
2ae (visible region).
a± = 2a'e'= z=—<2ae (2-98) Equations 2-76 and 2-87 are, of course, only
\ + (R/ab)
a very rough approximation, because we have
In this case, a± is smaller than twice the polar- used a model that ignores the complications
izability of a single atom. Obviously, a/f > aL. involving quantum mechanics. We have men-
If the molecular axis is neither parallel nor tioned that in gases we can assume quite rea-
perpendicular to F, but has an angle 6 with sonably that the interaction between outermost
respect to F, as shown in Figure 2-11(c), the electrons of atoms can be ignored and that
total dipole moment is the local field is equal to the applied field.
However, in a condensed phase (i.e., in liquids
U=U// + UL
(2-99) or solids) the interaction between electrons
= a„F„+a±F± of different atoms becomes very important.
—> According to Pauli's exclusion principle, there
and the dipole moment in the direction of F is
should be many values of co0, since each elec-
uF = UII cos 0 + u±sin 0 tron is allowed to occupy only its own state.
(2-100) Therefore, the electric susceptibility should be
= («// cos2 0 + Oisin2 6)F
obtained by the summation of the contributions
The potential energy of a dipole induced by the of all outermost electrons of all atoms in a piece
field is therefore of dielectric solid. The electric susceptibility
can be written as
W = -±(a//F?/+a1F21)
N i
f(Zqf Floe
2 2 2 •=27 F (2-102)
= --(«//cos 0 + a ± sin 0)F (2-101)
= -•£[«// -(«// -<*±)sin20]F2 where N denotes the total number of atoms

per unit volume. Furthermore, because the
Equation 2-101 indicates that when 6= 0 or n, surrounding polarized atoms will modify the
the potential energy is minimal, corresponding applied field, in this case each atom is polarized
to the most stable position; when 0 = 7t/2, W is in the so-called local field, which is generally
maximal, corresponding to the most unstable larger than the applied field. We shall discuss
condition. If the molecules are free to rotate, as the local field in Section 2.5.
Quantum Mechanical Approach an isolated atom, may be applicable to gases

The motion of electrons in an atom is governed because a gas may be considered an assembly
by quantum mechanical rules. On the basis of of isolated atoms with negligible interaction
the quantum mechanical approach, electron among them. Equation 2-105 is not applicable
cloud density is assumed to vary from zero at to solids, where the binding force between
the center of the atom to zero at far distance, atoms affects the movement of valence elec-
with a peak around the Bohr radius R. The electrons, and also the energy levels are no longer
tronic polarizability, derived by van Vleck in discrete but form continuous bands. However,
1932, is given by cojo is directly related to Eg = Ec- Ev, where Ec
and Ev are, respectively, the energy levels of the
«.(») = - f - r M (2-103) conduction and the valence band edges, and
Eg is the forbidden energy gap. Qualitatively,
m \coto-co J
we can think from Equation 2-106 and would
where co^ = (E{ - E0)lfi is the proper oscilla-
expect that the smaller the forbidden gap, the
tion frequency for two energy levels, E0 denotes
larger is the electron polarizability and hence,
the energy level of the ground state, Ex denotes
the static dielectric constant, as shown in the
the energy level of one of the allowed excited
following table20'21:
states, and /i 0 represents the strength of the
oscillator associated with the coupling between
the wave function in the ground state and that
Crystal C Si Ge
in the allowed excited state.18,19 Its value is of
the order of unity. It is amazing that Equation Eg{eV) 5.2 1.1 0.7
2-103, based on the quantum mechanical 5.68 12 16
approach, is similar to Equation 2-86, based on
the classical approach. By setting o> = 0, we
have the static electronic polarizability It should be noted that in crystalline solids
with covalent bonds, such as Si and Ge, the cal-
«,(*)= *1 A (2-104) culation for ae is quite involved,1819 since the
m coto
wave functions in the crystal are quite different
which is again similar to Equation 2-87. If an from those in isolated atoms because of the
atom has several excited levels, then Equation involvement of valence electrons in the binding
2-103 must be written in generalized form, as of atoms to form a solid. However, in ionic
crystals, such as NaCl, the valence electrons are
^ ) =^X-AT (2-105) strongly localized in each ion. In this case, the
electronic polarizability can be calculated
and simply by ^ #,-£*«, where Nt and aei are,
respectively, the concentration of /th ions and
<*» = - X ^ f (2-106) the electronic polarizability of each /th ion. For
™ j#> «>jo
example, for NaCl ionic crystal, the energy
where cojo = (Ej - E0)lh and j refers to the 7th band gap Eg = 7 eV, the values of electronic
excited level. polarizability21'22 for N+ and CI" are, respec-
Nonpolar solids are generally elemental tively, 0.20 x 10"40 and 2.65 x 10"40 farad m2.
solids, and they consist of only one kind of From Equation 2-66, we have calculated the
atom, such as diamond (C), silicon (Si), and relative permittivity (dielectric constant) to be
germanium (Ge). In these materials, there are about 2.28 by assuming the concentration of
no permanent dipoles or ions. The only polar- NaCl molecules N = 4x 1028nT3 and a = (0.20
ization they have is electronic polarization. In + 2.65) x 10"40 farad m2. This value is very
general, Equation 2-105 for ae, derived on the close to the experimental value of 2.25. Note
basis of the displacement of electron clouds in that the static dielectric constant of NaCl
crystals is 5.62. The difference between 5.62 will consider only a simple case to illus-
and 2.28 is due to the contribution of ionic trate the basic concept of atomic polarization.
polarizability, which will be discussed in the Suppose we have a molecule consisting of
following section. Similarly, in molecular two atoms: A and B. Atom A is ready to
solids where the atoms or molecules are bonded give part of its valence electrons to atom B
by weak van der Waals forces, each particle in order to make the outermost shells of
(atom or molecule) can be considered a quasi- both atoms more fully completed. This ten-
isolated particle. In this case, the electronic dency is, in fact, the force that produces the
polarizability of the solid can also be calculated so-called ionic bond. In this case, atom A is
by ]T Ntaeh where Nt is the concentration of ith more electropositive and atom B more elec-
i tronegative. NaCl is a good example, in which
particles and aei is the electronic polarizability atom A is Na and atom B is CI. The one elec-
of each ith particle. For condensed matter tron in the outermost shell of Na is not com-
(liquids or solids) made of single inert (or pletely given to CI (Na gives, on average, about
noble) atoms, such as He, neon, or argon, the 78% of the electron to CI). However, CI
total electronic polarizability can be simply receives some of the valence electron from
calculated by Nae. In this case, N is just the con- Na and becomes a negative ion; Na becomes a
centration of atoms in the material. positive ion.
Let us consider a linear chain of ions A and
ions B placed at equal distances along the x
2.3.2 Atomic or Ionic Polarization direction, as shown in Figure 2-12(a). In
(Vibrational Polarization) thermal equilibrium and in the absence of an
A dielectric material consisting of polyatomic electric field, the positive ions A at x2n, x2n+2,
molecules usually has electronic polarization x2„-2, etc., and the negative ions B at x2n+i, x2n+3,
and ionic polarization, and in some cases, also x2n-u etc., will always undergo lattice vibra-
orientational polarization when it is subjected tions, but their interatomic distance is x2n+i - x2n
to an electric field. In general, there are two = a, on average. Now if a step-function electric
groups of ionic solids. One group does not field is applied in the x direction, the electron
possess permanent dipoles, such as NaCl, clouds immediately shift to the left, and it takes
which forms a simple cubic lattice so that the only about 10"15sec to produce electronic
lattice symmetry and the overall charge neu- polarization, as shown in Figure 2-12(b). In
trality ensure that electric dipoles formed by about 10"13sec after the application of the
each ion pair everywhere cancel each other. field, the positive ion at x2n and the negative
The other group possesses permanent dipoles, ion at x2n+x tend to attract and move toward each
because the crystal lattice in this case is less other, making the interatomic distance kxx =
symmetrical, as with HC1. In fact, the internal Xin+i -x2n<a and Ax2 = x2n-x2n_i > a . The same
field at the positive ion sites is generally dif- tendency prevails in other ions. The displace-
ferent from that at the negative ion sites. In ment in both the electron clouds and the ions
general, most ionic solids are asymmetrical and themselves produces electronic polarization, as
the electronegativities of both ions are differ- well as ionic polarization, as shown in Figure
ent. They possess permanent dipoles, but these 2-12(c).
dipoles are foreseen in the solid state and Displacement of the atoms (i.e., ions) from
cannot be aligned by an electric field. This is their equilibrium sites by Ax will generate a
why in most ionic solids belonging to this force that tends to bring them back to their
group, the permanent dipole moments do not original thermal equilibrium sites. Within the
contribute to the polarizability in the solid state approximation of harmonic oscillation, the
although the materials possess them. elastic restoring force is proportional to the dif-
Quantitative analysis of the ionic polariza- ference between displacements of neighboring
tion is quite involved.22"24 In this section, we ions. Thus, from Figure 2-12(c), we can write
x x x x x
2n-2
cir-c
2n-1
cir-\
2n 2n+1 2n+2
M*
V- a
H > < a —4«—a—4*— c
MB MA MB MA
— a >« a
MB
>< i
MA MB
w-CPCp (P (p (p Cp (j) (p
<b>-°G G) <3> G) 0 G> O O
i i ^ i i i i i
I I i i i i i i
« M G G OO OO OG Ax1 Ax2
+ZqF |oc - Z( 7 F loc
<a | >a
Figure 2-12 Illustrating a linear chain of atoms A with mass MA and atoms B with mass Mfl placed at equal spacing a
along the JC direction, (a) In the absence of an electric field F = 0, (b) in the presence of an electric field F > 0, only elec-
tron clouds are shifted to the left, and (c) also in the presence of an electric field F > 0, but in this case the positive ions
and the negative ions tend to move toward each other, resulting in ionic polarization, which involves both the displacement
of electron clouds and the movement of ions.
the equations of motion for ions MA and MB as corresponding to the equilibrium sites, and the
follows: second term is the elastic restoring force
d2x2n
M, 2 =7i[(X2n+l-X2n)-a] — • Ax = YjAx (2-110)
dt (2-107) 47ue0a3 '
-7i[(X2n-X2n-l)-a]
In general, ZA = ZB = Z. For NaCl, Z = 0.78. So
~ Yi[X2n+l + X2n-l — 2*2/i J
the force constant is
Mi dt2 = 7i[(X2n ~ X2n-l ) ~ a]

Yi =
2zy (2-111)
-Yi[(xln-\-X2n-2)-a\
4ne0a3
=
Yi[x2n + x2n-2 "~ 2*2/i-l J Equations 2-107 and 2-108 are equations of
(2-108) phase waves and therefore have the solutions of
phase wave form
where y is the restoring force constant, which
X =
is different from y i n Equation 2-79. y can be 2n Xi exp[jk(2n)a] (2-112)
calculated on the basis of only the electrostatic
x2n-i =xnexp[jk(2n-\)a] (2-113)
interaction between the ion and its nearest
neighbors. Suppose, for simplicity, the positive where xt and xn are periodic functions of time
and the negative ions have net charges ZAq and and the exponential is a phase factor. These
-ZBq, respectively, and the net displacement equations indicate that the phase of the elastic
from their equilibrium sites is Ax, then the wave changes by 2ka from one ion to the
coulombic force between the two ions is nearest ion of the same kind, such as from MA
at x2n-2 to another MA at xln. k is the wave
ZAZBq2 ZAZBq2 | 2ZAZBq2 ^ number, which may be taken as 2nlX with X as
47te0(a-Axf 4ne0a2 4jte0a3 the wavelength of the elastic wave. For lattice
(2-109)
vibrations, in the visible and infrared regions,
since Ax « a. We retain only the first two X is of the order of 3 x 10~ 5 cm, which is con-
terms. The first term is the force of interaction siderably larger than the interatomic distance of
about 3 x 10~8cm. So we can say that ka is (Zq)2Floc

Hi = ZqAx -
extremely small and is much smaller than unity Mr(co2-co2) + jPco
for the frequency range associated with the (2-120)
- (XiFloc - &i(eff)F
lattice waves. Substituting Equations 2-112 and
(2-113 into Equations 2-107 and 2-108, and Thus, the ionic (or atomic) polarizability is
bearing in mind that exp (jka) is equal to cos (Zqf
ka (which can be practically considered as OCi = (2-121)
Mr(co2-0)2)+jPco
coska — 1), we obtain
In static fields, (O = 0 the static ionic polariz-
ability becomes
M, = -2yi(xI -xu) (2-114)
~df
(Zqf
CCi = (2-122)
d2xH Mrcol
Mi = -2yi(xII-xI) (2-115)
dt2 The expression for at is similar to that for ae.
Under an electric field, the equations of It should be noted that the distinction between
motion for ions should include both the elastic electron and ionic polarizations is not sharp,
force and the electric force, as well as the because a displacement of ions is always
damping or retarding force. Therefore, the accomplished by a displacement of electrons.
above equations must be modified to Generally, it is much easier to measure the
electronic polarizability than the ionic polar-
d2Xi izability. Usually, the electronic polarizability
M/ = -2yi(xI-xII)
IF (2-116) is determined by measuring the refractive index
d(Xl XlI in the visible or ultraviolet region, and the ionic
-p - KzqFloc polarizability is then determined by extrapolat-
at
ing the refractive index spectrum to frequencies
d2xH much lower than the visible region—generally,
M, ••-2yt(x„-x,)
(2-117) in the infrared region, in which both ae and
_pd{xII-xI)_ZqFix (Xi are the dominant polarizabilities. The time
at required for electronic polarization is about
Equation 2-117 x MA minus Equation 2-116 x 10" 15 sec, and that required for the ionic polar-
MB yields ization is about 10~13sec, simply because ions
are heavier than electrons by more than 103
, M d2Ax _ . n dAx ^ ^ //-i 11 ox times. This is why the resonances for these two
Mr—-r = -2yiAx-p— ZqFloc (2-H8) polarizations occur in different frequency
dr dt regions.
where Mr = MAMB/(MA + MB) is the reduced
mass and Ax = Axn - AJCZ is the relative dis-
placement of positive and negative ions. Fol- 2.3.3 Orientational Polarization
lowing the same principle for Equation 2-81,
To begin, it may be desirable to show why some
we can express yt in terms of the natural oscil-
molecules possess permanent dipole moments
lation (or lattice vibration) frequency (0o as 2yt
and some do not. For a molecule with two
= Mr(020. If the applied electric field is an alter-
atoms A and B, atom A gives some of its
nating field with the frequency co, as given by
valence electrons to atom B, then atom A
Equation 2-83, then the solution of Equation
becomes a positive ion and atom B becomes a
2-118 gives
negative ion. This ionic bond molecule obvi-
ZqFloi ously possesses a permanent dipole moment,
Ax = - (2-119) which is the product of the charge of the portion
Mr(co2-co2)+jPco
of the valence electrons transferred from atom
The induced ionic dipole moment is A to atom B and the interatomic distance. If a
great many such molecules form an ionic will orient to align the permanent dipole
crystal, such as an NaCl crystal, the vector sum moments along the direction of F. This process
of all the dipole moments vanishes, because is referred to as orientational polarization,
they tend to cancel each other. So, the crystal which occurs only in dipolar materials pos-
itself is not a dipolar material. For some mate- sessing permanent dipoles. It is obvious that
rials, a molecule consists of three atoms in the orientation of a molecule involves the energy
form AB2 or A2B. For example, C0 2 is in the required to overcome the resistance of the sur-
form of AB2. In this molecule, the bonding rounding molecules, so the orientation process
structure is symmetrical, so the centroids of the is strongly temperature dependent.
positive and the negative charges are coinci- In general, nonmagnetic dielectric materials
dent, thus the resultant dipole moment is zero, can be considered paraelectric materials, the
as shown in Figure 2-13(a). Similarly, mole- counterpart to paramagnetic materials, in which
cules consisting of similar atoms, such as the induced electric susceptibility x > ° *s
H2, 0 2 , and N2, carry no permanent dipole always positive. There is no counterpart to
moments. However, H 2 0 is in the form of A2B, diagmagnetic materials in dielectric materials,
but in this case, the bonding structure is asym- in which % should be negative. In paraelectric
metrical, so the centroid of the negative charge materials, ae and at can be considered practi-
is not coincident with that of the positive cally independent of temperature for the normal
charge, thus resulting in a net permanent dipole temperature range in most applications,
moment, i.e., the vector sum of the two indi- because the electronic structure does not
vidual dipole moments gives a resultant dipole change in the normal temperature range.
moment, as shown in Figure 2-13(b). However, the orientational polarizability a0 is
In the presence of an electric field F, the strongly temperature dependent. Figure 2-14
molecules carrying a permanent dipole moment shows the total polarization of some vapors as
a function of temperature. The data are from the
references.25-28 It can be seen that C2H2 and
molecules with a similar symmetrical and
linear structure, such as CH4, CC14, and C0 2 , do
not possess permanent dipole moments and
therefore are not dipolar. Those with an asym-
u up = 0 u
metrical structure are dipolar, such as H 2 0, in
No permanent
dipole moment which two OH bonds make an angle of 105°,
(a) resulting in the formation of a permanent dipole
moment. This is why the total polarization in
H—O —H which the orientational polarization is domi-
nant decreases with increasing temperature, as
Centroid of ^/_~ +*~^\^^ Centroid of shown in Figure 2-14. It should be noted that
+ charges /^r'^*^yf^N) ~cnar9es
for gases at normal temperature and pressure,
susceptibility is of the order of 10-4 to 10~3. The
^/<I><^-- 0=105° difference in susceptibility reflects the differ-
u < ^JV u ence in the number of atoms per unit volume.
up = 2us\n(d/2) However, for solids, susceptibility is generally
With permanent much higher than unity.
dipole moment
Molecules having a permanent dipole mo-
(b)
ment experience a torque in an electric field,
tending to orient themselves to the field direc-
Figure 2-13 The formation of dipole moments for two
typical molecules: (a) molecule C0 2 having no net perma- tion, but thermal agitation tends to put them
nent dipole moment and (b) molecule H 2 0 having net per- back into random orientation. However, the
manent dipole moment. ensemble of molecules will gradually attain a
F • F • F •
e—©
J*
©
(a) (b) (c)
Figure 2-15 The orientation of a dipole: (a) stable posi-

tion, (b) unstable position, and (c) orienting to the field
direction.
2.5 3.0 3.5

1/7-(K- 1 )x10 3
Figure 2-14 The susceptibility (er - 1) as a function of

temperature for some polar and nonpolar gases at the
normal pressure of 1 atmosphere.
Figure 2-16 The calculation of the number of dipoles

statistical quasi-equilibrium. The method of having the inclination of their axes to the x-axis between 0
calculating orientational polarizability was first and e+d6.
developed for permanent magnetic moments in
paramagnetic materials by Langevin29 and later
applied to permanent dipole moments in dielec-
tric materials by Debye.30'31 located between 6 and 0 + d 6, as shown in
Supposing that the permanent dipole Figure 2-16, is
moment u0 of the molecule is not affected by
the applied electric field and temperature, dN = N = —Nsm0d0 (2-124)
and that the density of molecules in the gas Aitr1 2
medium as an ensemble is so small that the With an applied field F and the effect of tem-
dipole-dipole interaction is very small com- perature, Equation 2-124 is changed to the
pared to the thermal equilibrium energy kT; the following form, according to Boltzmann's
moment of the permanent dipole in the direc- statistics,
tion of the applied field F, as shown in Figure
2-15, can be written as dN = Aexp(-U/kT)-sm6dO (2-125)
{uF) = uo(cos0) (2-123)
where U is the potential energy of the dipole at
where 0 is the angle between the dipole 6, which is given by
moment and the applied field, and < > repre-
U = u„Fcos 0 (2-126)
sents the statistical ensemble average.
At zero applied field, the number of dipoles Thus, the average dipole moment in the field
having an inclination of their axes to the axes direction is
\u0 cos OdN L(z) = (cos0) = 1-1/z (2-132)

{uF) = l—1
However, as FIT increases, the dipole of the
)dN
molecule becomes more parallel to the applied
f exp(u0Fcos 0/kT) cos 0 sin a/0
Jo field. In practical and most cases, u0FlkT« 1.
U
= o Tn
Thus, substitution of Equation 2-131 into Equa-
exp(woFcos0/£7>in&/0 tion 2-123 gives
Jo
= uo(cos0)
(2-127) ujF
(uF) = (2-133)
3kT
By introducing
and the orientational polarizability is
y = cos6 (2-128)
z = u0F/kT (2-129) a0=- 3kT (2-134)
we obtain
In general, orientational polarizability is much
larger than electronic and atomic polarizabili-
f exp(yz))Yfy
<COS0> = ^ ties at normal conditions, since ae and (% are
(2-130) practically independent of temperature (i.e.,
\ iQxp(yz)dy their temperature dependence at normal con-
= cothz-l/z = L(z) ditions is not significant) but a0 is strongly
The function L(z) is called the Langevin func- temperature dependent. Thus, a0 can be easily
tion, which is shown in Figure 2-17 as a func- distinguished from ae and at by the tempera-
tion of z = u0FlkT. ture dependence measurement of er.
At low values of z, L(z) varies almost There are two factors that have not been
linearly, with z following the relation taken into account in the derivation of a0. One
L(z) = {cos0) = uoF/3kT (2-131) is the effect of electron spins, which cause spin
paramagnetism and may affect the results. To
For example, a dipole moment u0 = 10"8 qC - take this effect into account, the quantum
cm at a field of 106V/cm and a temperature of mechanical approach must be used. The other
300 K gives u0FlkT =* 0.4, which is smaller than factor is that the permanent dipole moment u0
1. There is no possibility for < cos 0 > to reach for multiatomic molecules is not independent
unity at room temperature. Only at very low of temperature because in this case, the perma-
temperatures may it approach unity at high nent dipole moment is the result of several
fields. For z « 1, Equation 2-131 is reasonably moments, and their internal orientation is
accurate, and for z » 1, L(z) can be estimated dependent on their individual activation ener-
by gies and hence on the temperature.
In solids, the dipoles do not rotate freely as
do dipoles in liquids or gases. The rotation of
dipoles in solids is constrained to a few discrete
orientations, which are influenced by the crys-
talline field determined by the interaction of the
dipole with neighboring ones. Thus, to describe
the dielectric constant in terms of the dipole
moments u0 of individual dipoles (molecules),
it is necessary to know the crystal structure of
the material in the solid state.
The potential energy of a dipole in a crys-
talline solid depends on the direction of the
Figure 2-17 The Langevin function. dipole relative to the crystal axes. In other
Solid Liquid
40 Solid Liquid
20 • h<-
30 -
16 [-
12
20 Melting
point
8
10
4 -
^ m.p.
n
^r i i i i I I I
0
270 280 290 300 80 120 160 ^ 200
+ T(K) + T(K)
(a) (b)
Figure 2-18 Variation of dielectric constant (measured at low frequency 5 kHz) with temperature for (a) nitrobenzene
(QH5CO3), and (b) hydrogen sulfide (H2S).
words, the potential energy is directly related to constant of H2S in the solid state increases with
the crystalline field acting on the dipole and decreasing temperature, but at T0 = 103 K, er
is therefore temperature dependent. In some drops abruptly from about 20 to about 3, indi-
solids, dipoles at temperatures below the cating that the dipoles at T0 are frozen and
melting point are frozen, and their rotation is become immobile, resulting in a0 -» 0. The
inhibited even under an electric field. In such experimental results are from Smyth and
solids, the contribution of a0 to the total polar- Hitchcock.32'33
ization is zero, ie., a0 = 0, as in nitrobenzene To explain why H2S and some other materi-
(C6H5C03), shown in Figure 2-18(a). The als, such as HC1, still have the temperature
dielectric constant decreases abruptly from the dependence of a0 in the solid state within a
value of 35 (in liquid state) to a value of about certain range of temperatures between Tm and
3 (in solid state) at 278 K, the melting point of T09 we can assume that the average potential
this material. For T< 218 K, only ae and at con- energy of a dipole due to short-range interac-
tribute to er. For T > 278 K, the material is in tion, which creates the crystalline field, has a
the liquid state; ae, ah and a0 all contribute to profile similar to that shown in Figure 2-19.
en and er decreases with increasing temperature The bottom of the potential wells represents
following Equation 2-134. However, in some the equilibrium state of the dipoles in which the
other solids, the dielectric constant is still potential energy of the dipoles reaches the
increasing with decreasing temperature at tem- minimum value. In the absence of an electric
peratures below the melting point, i.e., in the field, a dipole has the same probability of ori-
solid state. This rise in dielectric constant con- enting to the right or to the left, so there is no
tinues until a critical temperature T0, at and net polarization. But when the material is
below which the dipoles become frozen. H2S is present in an electric field, the well at 6 = 0 is
a typical example of this kind of material. lowered by an amount of u0F, and that, at 0 =
Figure 2-18(b) shows that, for temperatures 7t, is deepened by the same amount of u0F, as
between the melting point Tm (188K) and the shown by the dashed line in Figure 2-19. The
critical temperature T0 (103 K), the dielectric direction of the dipoles orienting to the right
The term exp(-2u0F/kT) can be expanded into

a series because u0F « kT. This leads to
2u„F 1 (2u0F
exp(-2u0F/kT) =1 — + -
kT 2K kT
1 (2u0F
3!V kT + ...
By keeping only the first two terms, we obtain
-> 2u2F
Uo(parallel to F) = =^- (2-139)
kT
This leads to the polarizability
Figure 2-19 Schematic diagram showing the potential
(2 i40)
energy of a dipole as a function of the orientation angle 6 in
the solid state. Solid curve—in the absence of an electric
field and dashed curve—in me presence of an electric field.
*4HF -
which is in the same form as that of Equation
2-134 derived on the assumption that the
corresponds to the orientation of the dipoles dipoles rotate freely in the material. However,
toward the direction of the applied field, and the difference between these two equations
that to the left corresponds to the orientation of explains that at T = 7m, a0 in the solid state is
the dipoles opposite to the field. Denoting the larger than in the liquid state, as shown in
probability of the dipoles at 0 = n orienting to Figure 2-18(b).
the left as w, then 1 - w is the probability of the The transition from one form of the solid
dipoles at 6 = 0 orienting to the right, and state at To < T < Tm to the form at T < T0 can
according to the Boltzmann statistics, we can be considered the transition from a disordered
write state to an ordered state. At T = 0, all dipoles
are in a completely ordered state. That at T= 0,
= exp[-2u0F/kT] (2-135) a0 = 0 implies that the number of dipoles point-
1-w ing in one direction is equal to that pointing in
This leads to the opposite direction, so the net polarization
vanishes. As the temperature increases, the
exp(-2u()F/kT) number of dipoles in the ordered state will tend
(2-136)
l + exp(-2u0F/kT) to move to the disordered state. Considering
In general, u0F is much smaller than kT under that the dipoles have their lowest potential
most practical conditions. For uaF « kT, Equa- energy in the completely ordered state, there
tion 2-136 can be simplified to exists a potential difference U(T) between two
equilibrium states, as shown in Figure 2-20(a).
w= -exp[-2u0F/kT] (2-137) It can be imagined that the probability for a
dipole having its direction at 6 = 0, termed the
Based on this model, the percentage of a dipole right direction, is w, then a lower probability
moment u0 in the direction opposite to the (zero at T = 0) for the dipole having its direc-
applied field is w, than the percentage of tion at 6 = 7T, termed the wrong direction, is 1
the dipole moment in the direction parallel to - w, as shown in Figure 2-20(a). Similarly,
the field is 1 - w. Thus, we can write another dipole may behave oppositely; its
_^ —>
• direction at 0 - 0 is the wrong direction, and
u0(parallel to F) = u0{\ - w) - u0(w) that at 6 = n is the right direction,28,34,25 as
= u0(l-2w) shown in Figure 2-20(a). We can use the same
= u0[l- QXp(-2u0F/kT)]
approach for the polarization at T0 < T< Tm\ the
(2-138) probability for a dipole in the wrong direction
is 1 - w, which is governed by the following A qualitative picture of U{T) as a function of

relation: T, as shown in Figure 2-20(b), is sufficient to
explain the temperature dependence of £r at T
w = exp[-U(T)/kT] (2-141) < T0. Since U(T)» kT, the energy of the dipole
1 —w due to the applied field u0F is negligibly small
Hence, we have compared to U(T); therefore, the field F does
not affect the order-disorder transition for T <
exy[-U(T)/kT] T0. According to Equation 2-142 and Figure 2-
w l + exp[-U(T)/kT] (2-142)
20(c), w is very small for T <T0 and becomes
The calculation of U(T) is very involved, even 1/2 at T= T09 at which U(T) = 0, as shown in
by means of approximation methods.28'36 We Figure 2-20(b). This means that w = 111 for T
shall not give details about the calculation > T0, implying that the probability for the
here. dipoles pointing to the right or to the left is
(a)
w
U(T)
(b) (c)
Figure 2-20 (a) Potential energy profiles of two different types of dipoles in the lattice, (b) U(T) as a function of tem-
perature, and (c) the probability W as a function of temperature.
equal. Under this situation, the applied field There is no counterpart to diamagnetization in
will start to play an important role in causing dielectric materials, with % in negative values.
polarization, as discussed earlier for T between Spontaneous polarization occurs in materials
T0 and Tm. whose crystalline structure exhibits electrical
order. This implies that spontaneous polariza-
2.3.4 Spontaneous Polarization tion occurs only in single crystals or crystallites
Electronic polarization is always present in in polycrystalline materials with a noncen-
atoms or molecules in all kinds of materials; trosymmetic structure, because only in a non-
ionic polarization is present only in materials centrosymmetic structure does the centroid of
made of two or more different kinds of atoms the negative charges not coincide with that of
that form ions due to the sharing of the valence the positive charges. In ferroelectric materials,
electrons of one or more atoms with the others. electric polarization occurs spontaneously due
For both electronic and ionic polarizations, the to a phase transition at a critical temperature
dipole moments are induced by electric fields, called the Curie temperature, Tc, without the
so they are classified as induced dipole help of an external electric field. At and below
moments. In short, both electronic and ionic Tc, the crystal undergoes a phase transition,
polarizations are due to the translation (or usually from a nonpolar cubic structure to a
deformation) of the valence electron clouds polar structure. BaTi0 3 is a typical example of
from their original thermal equilibrium state a ferroelectric crystal. At T > Tc, the BaTi0 3
to a new equilibrium state. These types of crystal assumes a cubic structure in which the
polarization are only slightly dependent on centroid of the negative charges and that of the
temperature because they are intramolecular positive charges coincide, so the molecule does
phenomena. However, orientational polariza- not form a dipole moment. However, at T < Tc,
tion occurs only in the materials composed of the cubic structure is slightly distorted, result-
molecules with an asymmetrical structure in ing in a slight displacement of Ba2+ and Ti4+
which the centroid of the negative charge ions. This displacement, though very small
(mainly electrons) and that of the positive (only about 0.15 A), is enough to cause the cen-
charge (mainly nuclei) are not coincident, so troid of the negative charges to be different
they possess permanent dipole moments in the from that of the positive charges, thus forming
absence of external fields. The directions of an electric dipole moment. Each unit cell
these permanent dipole moments are randomly carries a reversible electric dipole moment
distributed in the material. An electric field will spontaneously oriented parallel to the dipole
cause them to reorient toward the direction of moment of neighboring cells. This chain-
the field, resulting in orientational polarization. reaction process is referred to as spontaneous
The net polarization will return to zero after the polarization.
removal of the external field because thermal The build-up of such dipole moments point-
agitation tends to randomize the alignment. ing along one crystal axis will gradually form
This is why polarization decreases with in- a domain and will gradually increase the free
creasing temperature. energy of the system. This process cannot con-
However, there is another kind of polariza- tinue; the domain will stop growing when it has
tion, called spontaneous polarization. By reached a certain size, and another domain with
analogy to magnetization, electric polarization dipole moments pointing in the opposite direc-
can be grouped into two major polarizations: tion will be formed in order to reduce the free
energy of the system. In a single crystal or a
• Paraelectric polarization, which includes crystallite, there are many domains with
mainly electronic, ionic, and orientational moments pointing in various directions. But
polarizations, with % always positive the vector sum of the dipole moments of all
• Ferroelectric polarization, with % very large, domains vanishes. Each domain can be consid-
similar to ferromagetization ered as a large dipole. Under an external
electric field, all of these randomly arranged

domains tend to move toward the direction of
the field, resulting in a net total spontaneous
polarization. Upon the removal of the field,
spontaneous polarization does not vanish
but remains inside the material. The field-
polarization relation forms a hysteresis loop
similar to the hysteresis loop for ferromagnetic
materials. This topic will be discussed in more
detail in Ferroelectric Phenomena in Chapter 4.
2.3.5 Space Charge Polarization

The induced, orientational, and spontaneous
polarizations discussed previously are due to
the bound positive and negative charges within
the atom or the molecule itself, which are linked Figure 2-21 A double potential well illustrating hopping
polarization due to the hopping of the charged particle over
intimately to each other and which normally the barrier from one well to the other.
cannot be separated. However, electric polar-
ization may also be associated with mobile and
trapped charges. This polarization is generally semiconductors are essentially due to the
referred to as space charge polarization, Pd. This hopping process.37 Depending on the width and
occurs mainly in amorphous or polycrystalline the height of the potential barrier, a charged
solids or in materials consisting of traps. Charge particle on one site may hop or tunnel to the
carriers (electrons, holes, or ions), which may other site. A simple, double-well potential
be injected from electrical contacts, may be barrier is shown in Figure 2-21.
trapped in the bulk or at the interfaces, or may In thermal equilibrium, the time-averaged
be impeded to be discharged or replaced at the probability for a charged particle (for example,
electrical contacts. In this case, space charges a negatively charged particle) to hop from site
will be formed, field distribution will be dis- A to site B, leaving a positive charge in site A
torted, and hence, the average dielectric con- and creating a negative charge in site B to form
stant will be affected. In the following sections, a dipole, is
we shall consider two possible ways in which ft(A-*) = Cexp(-EA/kT) (2-143)
space charge polarization may result.
where C is a constant and EA is the activation
energy for the hopping transition. Similarly, the
Hopping Polarization probability for the charged particle to hop in the
In a dielectric material, localized charges (ions reverse direction from site B to site A is
and vacancies, or electrons and holes) can hop
from one site to the neighboring site, creating P ^ =C e x p ^ " ^ ) (2-144)
so-called hopping polarization.16 These charges
are capable of moving freely from one site to where AE0 is the difference between EA and EB,
another site for a short time, then becoming which is positive when EA > EB, zero when EA
trapped in localized states and spending most = EB, or negative when EA < EB. An applied field
of their time there. Occasionally, these charges changes the potential barrier profile, making the
make a jump surmounting a potential barrier to hopping of the particle from site B to site A
other sites. In fact, the movement of ions or much easier than from site A to site B (for the
vacancies in ionic crystals and the movement case shown in Figure 2-21). Thus, the proba-
of electrons and holes in glasses and amorphous bilities under an applied field become
P{A^B) = Cexp[-(£A + aAE')/kT] (2-145) If qrF/kT « 1, p' can be approximated to

and Pf - po(A-*B)Po(BÂ)(qrFcos0/kT) (2-154)
P(BÂ) = Cexp{~[EA -(\-a)&E']lkT} the random hopping dipole moment is

(2-146) uh=qr (2-155)
where AE' = AE - AE0 is the potential barrier The hopping dipole moment may be considered
height difference created by the appliedfieldF, similar phenomenally to the orientational
a(AE') is the increase in potential barrier height dipole moment, but in nature, they are differ-
in site A, and (1 - a)AE' is the decrease in ent. The orientational dipole moment refers to
potential barrier height in site B. Obviously, the the permanent dipole moment formed by bound
magnitude of AE' depends on the direction of charges within the particle, while the hopping
the vector r (the axis joining site A and site B) dipole moment is the moment formed by the
relative to F. Therefore, AE' can be written as transition of a separate charged particle from
AE' = AE-AE0 = qFrcosO (2-147) one potential well to another potential well.
Following the same concept in treating the
This implies that AE' - qFr when 0 = 0 , i.e., the orientation of permanent dipoles (Equation 2-
vector r is in the field direction; AE' = 0 when 123), the hopping dipole moment in the direc-
0= 7t/2, i.e., the vector r is perpendicular to the tion of the applied field can be written as
field direction. In this case, the applied field
(uhF) = uh(p'cosd) (2-156)
does not modify the original potential barrier
height. In terms of probabilities, we can write where the factor p means that it is the field-
induced excess probability p that produces
P(A^B) = Po(A^B) ~ P' (2- 148) the hopping polarization. Using Equation 2-
and 127, we have
\pcos OdN
P(B->A) = Po(BÂ) + P' (2-149) (p'cosO) = A—*
\dN
The applied field causes a decrease in hopping
probability from site A to site B by an amount \ p„(A->B)Po(BÂ)(qrFcosO/kT)
Jo
p' and an increase in hopping probability from
site B to site A by p'. From Equations 2-143 and x(cos0)-Nsm0d0
2-147 we have
f - N sin GdO
2**^- = exp(AE0/kT) (2-150) = Po^B)PoÂ)[fE)
Po(A-^B)
(2-157)
^z£l = exp(AE/kT) = exp[( AEa + AE')/kT]
P(A^B)
Substitution of Equation 2-157 into Equation
(2-151)
2-156 gives
Assuming that the particle must be located
either in site A or in site B, then {UhF) - „Jrr Po(A-*B)Po(BÂ) (2-158)
3kT
Po(A-*B) + Po(B->A) =
P(A-*B) + P(BÂ) ~ 1
Thus, the hopping polarizability is
(2-152)
q2r2
From Equations 2-148 and 2-152, we obtain 0Ch = - r — PO(A^B)PO(BÂ) (2-159)
P - Po(A^B)Po(B-*A) where PO(A-*B)PO(B^>A) denotes the ensemble
[ l-exp(grFcosfl/A:r) } average of the product of these two probabili-
X
11 - Po(BÂ){QXip(qrF cosO/kT) - 1 ] J ties, because EA and EB may vary from site to
(2-153) site.
Interfacial Polarization r = (elrd2 + e2rdx )e0 l(oxd2 + o2dx) (2-166)

The space charge, or interfacial polarization, is
Y can also be written as
produced by the separation of mobile positively
and negatively charged particles under an Y = o/d + jco£re0/d (2-167)
applied field, which form positive and negative
space charges in the bulk of the material or By comparing Equation 2-167 with Equation
at the interfaces between different materials. 2-163, we obtain the overall dielectric constant
These space charges, in turn, modify the field and conductivity of this composite dielectric as
distribution. follows:
Suppose that we have a composite dielectric
1 ( oxo2d \TX + T2 + (Q2TXT2T-T
specimen comprising two parallel sheets of £r=-
£0 v oxd2 + G2dx) 1 + o)2r2
different materials, and that this dielectric
(2-168)
specimen is inserted into the space between two
parallel metallic plates of unit area, as shown (Jx<72d M - 0)2TxZ2 + 0)2TxT + C02T2T
in Figure 2-22. In this case, under AC fields the (7 =
oxd2 + G2dx l + coh2
admittance is (2-169)
YyY, Under static DC fields, co = 0. Thus, the static
Y =- (2-160)
Yx + Y2 overall dielectric constant and conductivity are
in which
oxo2d rx + r2-z
F — (2-170)
Yx=(<jx + jco£Xr£o)/dx (2-161) oxd2 + o2dx £0
Y2=(a2 + jcQe2r£o)/d2 (2-162) and

where eXr and e2r are the dielectric constants, Gx oxo2d
and (72 are the conductivities, and dx and d2 are (2-171)
<Jxd2 + 0*2^1
the thicknesses of sheets 1 and 2, respectively.
Substitution of Equations 2-161 and 2-162 into At high frequencies, the space charges cannot
Equation 2-160 gives follow the change of the field and hence do not
produce space charge polarization. By setting
{
(\-(02TxT2~\ corx +m2^ co —> oo? we obtain
2
+ (D TXT +A + fi)3T1T2T
GXG2
2
+ G) T2T v -aw J £Xr£2rd
Y =- £ —- (2-172)
Gxd2 + G2dx ,2*-2
1 + ftTT £Xrd2 + £2rdx
CToo = Gs (TiT 2 + TiT + T 2 r ) / T 2 (2-173)
(2-163) We can consider e^ to be mainly the contribu-

tion of electronic and atomic polarizations.
where
Thus, the total polarization is
Ti=elr£0/ax (2-164)
P = (£rs-l)£0F = NccF (2-174)
r2=e2re0/(J2 (2-165)
and the polarization due to electronic, atomic,
and orientational contributions is
1 P' = (£*» + £ro - 1)£0F = Nia^ + a0 )F

1 T" CT 1 ,£ 1
-1
eue0/d^ (2-175)
fi: C72, £2
|
cr 2 /c/ 2
_X
e £ /c/
2r 0 2 Therefore, P - P' is the polarization due to the
space charge contribution. Thus,
Pc = P-P' = (£rs -£r„-£ro)£oF

Figure 2-22 A composite dielectric material between two
(2-176)
parallel metal plates of unit area. = N(a-a00-a0)F
and the space charge polarizability is dielectric constant should be independent of

frequency and equal to the static dielectric
ac =a-aoo-a0 = (ers - eroo - ero)e0 / N
constant. According to Maxwell's relation,
(2-177)
the refractive index of such materials can be
where N is the number of molecules per unit written as
volume of the composite dielectric. It should
n = (ersT (2-178)
be noted that if this composite dielectric also
involves hopping polarization, then the space The total polarizability is
charge polarizability ad= ah + ac is the com-
cc = cce (2-179)
bination of both the hopping and the interfacial
polarizabilities. Since these two types of polar-
ization involve the movement of charged parti- Polar Materials
cles, there is no easy experimental method to
In materials of this class, an electric field will
separate these two different mechanisms.
cause elastic displacement of the valence elec-
tron clouds, as well as elastic displacement of
the relative positions of ions, so such materials
2.4 Classification of Dielectric
have both electronic and ionic polarization. The
Materials material may be composed of molecules, and
Dielectric materials may be classified into two each of the molecules is made of more than
major categories: nonferroelectric (or normal one kind of atom without permanent dipole
dielectric or paraelectric) materials and ferro- moments. Examples of such materials are ionic
electric materials. The former may be divided crystals, including alkali halides, some oxides,
into three classes, as the following sections paraffins, benzene, carbon tetrachloride, etc.
show. The appreciable absorption occurs at two
resonant frequencies: one in the optical fre-
quency region (corresponding to electronic
2.4.1 Nonferroelectric Materials polarization) and the other in the lower reso-
(Normal Dielectric or Paraelectric nance frequency—(the infrared region corre-
Materials) sponding to ionic polarization). In this case, the
In materials of this category, electric polariza- total polarizability is
tion is actuated by external electric fields.
Based on the mechanisms of electric polariza- a = ae+ai (2-180)
tion, we have three classes: nonpolar, polar, and
dipolar materials.
Dipolar Materials
The materials of this class have all three fun-
Nonpolar Materials damental polarizations: electronic, ionic, and
In materials of this class, an electric field can orientational. Thus, the total polarizability is
cause only elastic displacement of electron
cc = cce + 0Cj + cc0 (2-181)
clouds (mainly valence electron clouds), so
they have only electronic polarization. Such Materials whose molecules possess a perma-
materials, which are generally referred to as nent dipole moment belong to this class. An
elemental materials, consist of a single kind of electric field will cause spatial orientation of the
atom, such as silicon (Si), diamonds (C), inert permanent dipoles, resulting in orientational
elements in gas, liquid, or solid phase. For these polarization. The orientation process occurs
materials, the appreciable absorption occurs predominantly in liquids and gases, and in
at the resonance frequency 0)0, which is in the some solids within a certain range of tempera-
visible-to-ultraviolet region. For frequencies tures, such as solid hydrochloric and sulfuric
lower than the resonance frequency, the acids. However, in the solid state, there exists
a critical temperature at and below which all ferroelectric materials are given in Ferroelectric
dipoles are frozen in and lose their capability to Phenomena in Chapter 4.
contribute orientational polarizability. But for
most materials, the dipoles are frozen in below
the melting point of the material. 2.5 Internal Fields
It should be noted that dielectric materials,
Only in gases or in dilute phases, in which the
in general, are not single crystals; they are
interaction between atoms or molecules can
either amorphous or polycrystalline, containing
be neglected, the internal or local field, Fhc,
a large quantity of various traps. Furthermore,
acting on the atomic or molecular sites, can be
they are not nonconductive and always involve
assumed to be the same as the applied field F.
charge carriers (electrons, holes, or both) in-
However, the local field in condensed phases
jected from electrical contacts. Therefore, in
(solids and liquids) is higher than F because of
this case, the total polarizability should include
the polarization of the surrounding particles. In
the space charge polarizability
this section, we shall briefly discuss local fields
a = ae+ai+a0+ad (2-182) for nondipolar and dipolar materials.
where ad in the space charge polarizability
includes ah + ac. Local Fields for Nondipolar Materials
To understand the concept of local fields,
imagine a small sphere of dielectric material
2.4.2 Ferroelectric Materials removed from around the site of the atom or
A ferroelectric material is normally in single molecule to form a cavity, as shown in Figure
crystalline or polycrystalline form and pos- 2-23. The local field Fhc is composed of four
sesses a reversible spontaneous polarization components
over a certain temperature range. There is a crit-
Floc=F0 + F1 + F2 + F3 (2-183)
ical temperature, called the Curie temperature,
which marks the transition from the ordered to where F0 is the externally applied field in the
the disordered phase. At this temperature, the cavity, which is assumed to be a vacuum, acting
dielectric constant may reach values three to on A and is related to the applied field in the
four orders of magnitude higher than in the dis- bulk F by
ordered phase. The order-disorder phase tran-
D = £0F+P = e0F0 (2-184)
sition involves the displacement of atoms so
that crystals or crystallites exhibiting ferroelec- or
tric phenomena must be noncentrosymmetric.
This implies that a phase transition will induce F0=F+P/e0
a mechanical strain, tending to change not only Fi is the depolarization field resulting from
the volume and the shape of the material body, polarization charges (bound charges) on the
but also the optical refractive index. Thus, fer- surface of the specimen, which is given by
roelectric materials exhibit not only ferro-
electric phenomena, but also piezoelectric, F^-P/eo (2-185)
pyroelectric, and electro-optic effects, which In fact, F0 + Fi is equal to the applied field F.
can be used for many technological applica- F2 is the Lorentz field due to polarization
tions. In general, ferroelectric materials also charges on the inside surface of the spherical
have electrically induced polarizations, but cavity,6'27 which is given by
these are negligibly small compared to sponta-
neous polarization. For most practical cases F2=^-^F (2-186)
dealing with ferroelectric phenomena, electri-
cally induced polarization can be ignored. More F3 is the field of the adjacent dipoles due to
details about the properties and applications of the molecules inside the spherical cavity. In
<a> (+) m o fflffl (+)

Dipole
Pcosd
dA
4/renr2
Pfcos20c//4
4/re0r2
Pcos202;zr2sinfloffl
4;re0r2
:(£ r -1)F/3
9 9 99 999
Figure 2-23 The model for the calculation of the Lorentz field.
general, we assume that the cavity is ideally rial. Substitution of Equation 2-190 into Equa-
formed without disturbing the state of polar- tion 2-189 gives the molar polarization in terms
ization of the surrounding material and that of polarization per mole
the fields contributed by all dipoles inside the
£r-l M
cavity tend to compensate each other. This (2-191)
assumption leads to er + 2 p 3£0
This is the Clausius-Mossotti equation ex-
F3=0 (2-187)
pressed in terms of M and p. Using either
Thus, the local field can be written as Equation 2-189 or Equation 2-191, a can be
determined because all other parameters are
Floc=F0 + Fl + F2 + F3=^LF (2-188) generally known or can easily be measured. It
should be noted that the Clausius-Mossotti
equation can be used only for nonpolar materi-
The Clausius-Mossotti Equation als because F3 cannot be assumed to be zero for
From Equations 2-63, 2-64, and 2-188, we polar materials, which will be discussed in later
obtain sections. It should also be noted that when
3£0
er-\ Ncc (2-189) approaches unity, er approaches infinity, and
3£0
when p is larger than a certain critical value, er
This is the well known Clausius-Mossotti equa- tends to be negative. This phenomenon is gen-
tion. The number of atoms or molecules per erally called the Clausius-Mossotti catastro-
unit volume N is related to the number of atoms phe, indicating that Equations 2-189 and 2-191
or molecules per mole (Avogadro's number N0) are reasonable only for a low a, a low p, and
by nondipolar materials.
N0= — = 6.023 x 1023 per mole (2-190)

P The Lorentz-Lorenz Equation
where M and p are, respectively, the atomic or The Clausius-Mossotti equation can be used
molecular weight and the density of the mate- for dipolar materials in high-frequency AC
fields in w h i c h orientational p o l a r i z a t i o n c a n n o t cavity of radius a, it produces a field that polar-

follow t h e t i m e - v a r y i n g fields b u t electronic izes the cavity's surrounding molecules. The
a n d a t o m i c polarizations a r e still present. F o r reaction field Fr is the field created inside the
optical f r e q u e n c i e s , o n l y electronic p o l a r i z a t i o n cavity by the charges of these polarized mole-
is d o m i n a n t , a n d i n this c a s e er = n2, w h e r e n is cules on the cavity surface. Thus, in the absence
the refractive index. I n t h e optical frequency of the applied field F = 0, the effective dipole
range, Equations 2-189 and 2-191 can b e moment of a dipolar molecule in a condensed
rewritten as dipolar material is
n2-l Na Ueff = U0 + OCFr (2-196)

2
(2-192)
n + 2~ 3e0
where u0 is the permanent dipole moment of an
n -1 M N0a individual molecule and a is the total polariz-
(2-193)
n2 + 2 p 3e0 ability of the material. The permanent dipole
moment and Fr are in the same direction. This
T h e s e e q u a t i o n s are s o m e t i m e s called t h e
reaction field is given by
Lorentz-Lorenz equation because they were
first d e r i v e d i n d e p e n d e n t l y b y H . A . L o r e n t z i n 2ueff f er-\
Fr = = fue1 (2-197)
H o l l a n d a n d L . L o r e n z i n D e n m a r k i n 1880. 2 7 47ie0a3 V2£r + 1
A t h i g h frequencies, w h i c h are so h i g h that
n o orientational p o l a r i z a t i o n c a n o c c u r b u t still where
l o w e n o u g h t o i n c l u d e b o t h t h e electronic a n d £r-l\ 2N(er-\
a t o m i c p o l a r i z a t i o n s , er is t e r m e d eroo. I n this f =-
2xe0a3 V 2er +1J 3e0 V 2er +1
case, e^ - n2 is t h e c o n t r i b u t i o n of a t o m i c (2-198)
polarization. W e c a n also e s t i m a t e t h e orienta-
3 -1 6
tional c o n t r i b u t i o n . T h e electronic a n d a t o m i c since N = (4/&z /3) . T h e n , f r o m E q u a t i o n s 2 -
polarizability c a n b e e s t i m a t e d from 196 a n d 2 - 1 9 7 , w e o b t a i n
W0
eroo-\ _ N{ae+a,i) U (2-199)
(2-194) eff =
£roo + 2 3e0
l-af
W i t h a n a p p l i e d field F, t h e field i n s i d e t h e
T h e total polarizability c a n b e o b t a i n e d b y
e m p t y s p h e r i c a l cavity is g i v e n b y
m e a s u r i n g er u n d e r static D C fields. T h u s , s u b -
tracting E q u a t i o n 2 - 1 9 4 f r o m E q u a t i o n 2 - 1 8 9
3er
gives t h e polarizability c o n t r i b u t e d b y orienta- F( = F = hF (2-200)
2er + \
tional p o l a r i z a t i o n if b o t h t h e h o p p i n g a n d t h e
s p a c e c h a r g e p o l a r i z a t i o n s a r e absent. T h i s g i v e s where
\t>rs c-roo/ Na0 3er

(2-195) h= (2-201)
(ers + 2)(eroo + 2) 9e0 2er + \
w h e r e ers d e n o t e s t h e static dielectric c o n s t a n t . T h u s , t h e effective i n d u c e d d i p o l e m o m e n t of a

dipolar molecule inside the cavity due to the
c a v i t y field Ft a n d t h e r e a c t i o n field Fr i n t h e
2.5.2 The Reaction Field for Dipolar direction of t h e a p p l i e d field F, as s h o w n i n
Materials Figure 2-24, becomes
In Section 2.5.1, we derived an expression for ueff=a(Fi + fu'eff)

the local field based on the assumption that
there is no contribution to the field from the or
molecules inside the spherical cavity. If a ah
dipolar molecule is located at the center of the
U
eff = (2-202)
l-ocfJ
F=0
Fj=hF 4>ff and u^

= fu0W-af)
Figure 2-24 Schematic diagrams illustrating the concept of the reaction field Fn the cavity internal field Fh the effective
dipole moment of a polar molecule ueff, and the effective induced dipole moment of a polar molecule u'eff in a condensed
polar material.
The total dipole moment in the cavity is the «o | n , ah

COS0 + T zF
sum of the permanent dipole moment and the LU-osr 1-cf
induced dipole moment in the direction of F
xexp hup FcosO kT 27tsm0d0
(UF) = - l-cf
U
F = Ueff COSO + U'e* hu0
Jexpl \Fcos0 /kT 2xsm0d0
Polarized by the Polarized by the cavity LU-cgr
reaction field at field at applied field F h ul
\F
F =0 l-qf a + 3(\-cf)kT
(2-203) (2-206)
To obtain the mean value of wF, we use the
Boltzmann statistics with a weight factor The Debye Equation
exp
w(ey , in which W(0) is the activation Assuming that ah = ac = 0 (or ad - 0) and
kT expressing the total polarizability as
energy for the rotation of a dipole toward the
field direction and is given by a = ae+GCi + (2-207)
3kT
W(e) = ut/f.F = -hu«F™e (2-204) substitution of Equation 2-207 into Equation

kT 2-189 gives
Since only ueff must rotate but not u'eff, which is
er-\ N
already in parallel to F, we have = — ae+a, + - = H (2-208)
er + 2 3e0{ 3kTj
j(uejrcosO + u'eJf) This is called the Debye equation. The molar
x _ x exp(hueffFcosO/kT)dQ. polarization can be expressed in a form similar
\Uf) — —f (2-205)
to Equation 2-191
J Qxp(huejfFcosO/kT)dQ.
where d£l = 27rsin OdO is the solid angle cor-

\er + 2) p 3e0\ 3kT)
responding to dO. Following the same approach
(2-209)
for L(z) (Equation 2-127) and substituting
Equations 2-199 and 2-202 into Equation 2- In fact, the Debye equation is similar to the
205, we obtain Clausius-Mossotti equation; the only differ-
ence is that the former introduces a0 = u2J3kT At zero and low frequencies, all types of polar-
into the equation. Both equations were derived ization are taking place, so a = ae + aa + a0
on the basis of the internal local field rather (hopping and space charge polarizations are
than the reaction field. The Debye equation is assumed to be absent). However, in the high-
in poor agreement with experiments for dipolar frequency fields, in which orientational polar-
liquids and solids, possibly due in part to ization does not occur, u0 = 0. Then we have
the fact that for dipolar liquids and solids, the
3erc
Lorentz internal field is not applicable. The
Onsager equation improves on the Debye equa- 2ereo + l (2-212)
tion by using the reaction field instead of the _2Nf
/->/- = eroo-\
Lorentz internal field. " 3e0 v2£roo + l
Theoretically, the molar polarization de- The polarization becomes
duced from the Debye equation should be a
constant at a fixed temperature and independ- P~=(eroo-l)e0F = N{uFoo) (2-213)
ent of pressure. This means that Equation 2-191 Hence, we obtain
should have the same value for the material
whether in the gaseous phase or in the liquid e^-1 _ iV(ae+af)
(2-214)
phase. Experimentally, this is not true for £^ + 2 3£ 0
dipolar liquids. However, for gases in which This equation is, in fact, the same as the
er - 1 « 1, the Debye equation gives a rea- Clausius-Mossotti equation. By subtracting
sonably accurate prediction. Based on the Equation 2-213 from Equation 2-211, we
Debye equation, (ae + ai) and u0 can be deter- obtain
mined by measuring er at different tempera-
tures. Furthermore, the Debye equation gives (g» ~ £r~ )(2en + eroo) = Nul (2-215)
reasonable results for diluted solutions of en(e^ + 2f 9e0kT
dipolar molecules in nondipolar solvents. This is the well known Onsager equation.38 This
In general, dipolar materials can be classified equation enables the computation of u0 of a
into three groups based on the permanent dipole dipolar molecule from ers of a pure dipolar
moments of their molecules: liquid or solution, if its density and e^ are
known. The Onsager equation was first derived
• Weakly dipolar materials in which u0 < 0.5
on the assumption that the molecules are spher-
debye
ical in shape and that there are no interactions
• Medium dipolar materials in which 0.5 between the molecules (or between the dipoles).
debye < u0 < 1.5 debye However, u0 of nonassociating organic com-
• Strongly dipolar materials in which u0> 1.5 pounds (with weak interactions between mole-
debye cules) agrees reasonably with those computed
from Onsager equation. For associating organic
The Onsager Equation compounds (with strong interactions between
In the static DC or low-frequency fields, we molecules), the difference between the meas-
have ured and the computed values of u0 is large,
indicating that Onsager's model is inadequate
/* —> /L =
3E
for associating organic compounds.39 This is
2£r5 + l mainly due to the following reasons:
(2-210)
3£0v2e„ + l • The model assumes that the surrounding
medium of any molecule is a continuum.
On the basis of the definition of the polariza-
This means that it does not take into account
tion, we obtain
the interactions between the molecule and
P = (ers-l)e0F = N(uF) (2-211) its neighbors. In other words, only long-
range interaction is considered; short-range into account. This is true only for systems
interaction energy has been neglected. of nonassociating molecules. Therefore, the
• The model does not take into consideration Clausius-Mossotti, Debye, and Onsager equa-
the electrostatic interaction between nearest tions can apply only to systems of nonassociat-
dipoles. ing molecules and, as a result, they are not
satisfactory for condensed systems, such as
• The model does not take into account the
polymeric materials.
nonpolar molecular interactions, which may
Kirkwood40 developed his theory of static
be significant in some liquids and solids.
polarization by taking into account the short-
Several investigators6'24,28 have attempted to
range interaction between molecules in the
modify the Onsager equation by using an
liquid state. The Kirkwood equation is
ellipsoid instead of a spheroid in the shape
of the cavity, and also by taking into account
the anisotropy of the polarizabilities of the
molecules. However, their modified models 9ers 9eA 3kT)
do not improve much. The improvement (2-216)
must take into account the intermolecular
where g is the correlation factor. The correlation
interactions, and to this end, we must use the
factor is a measure of the local ordering in the
statistical-mechanical approach.
material, g = 1 indicates that the average dipole
moment of a finite spherical region around one
Statistical-Mechanical Approaches and the reference molecule, which is held fixed in an
Kirkwood Equation infinite size of the material specimen, is about
Both Debye and Onsager used a statistical equal to the moment of the fixed molecule. This
method to solve only part of the polarization implies that the location of one dipole does not
problem by considering the whole dielectric influence the positions of the other dipoles. If
medium as a continuum where there is no a fixed dipole tends to make the neighboring
correlation between molecules (molecular dipoles line up in a parallel direction, then the
interactions) in deformational (or induced) effective moment is larger than u0 and hence
polarization and in orientational polarization. g > 1. Conversely, if the fixing dipole tends to
Statistical-mechanical approaches to the polar- line up with the neighboring dipoles in an
ization problem, taking into account the inter- antiparallel direction, then g < 1,34 If the refer-
action between nearest dipoles, were first used ence molecule (the fixing dipole) is surrounded
by Kirkwood.40 He developed a general method by z neighboring molecules and cosO is the
of solving the problem that had some influence average of cos O made between the reference
on all the subsequent polarization theories. molecule and one of its z nearest neighbors, then
Kirkwood had taken into account the short- g may be expressed as39'40
range correlation in orientational polarization,
but he ignored the correlation in deformational g = l + zcos0 (2-217)
polarization between molecules. In Kirk wood's
final equation, the terms due to deformational The correlation factor is supposed to character-
polarization were added empirically. Later, ize the molecular interaction and the short-
Frohlich28 developed a more rigorous expres- range structure. Up to the present, no good
sion, taking into account the correlation method is available for determining the value of
between molecules in both orientational and g accurately. In general, it is roughly estimated
deformational polarizations by means of a from the measured value of the dielectric con-
statistical-mechanical approach. The equations stant or the dipole moment. This is one of the
derived by these and other investigators include major shortcomings of the Kirkwood equation.
a correlation factor g, which is related to the Furthermore, Kirkwood used the approximate
effects of molecular interactions. When g = 1, formula for the internal field, which also makes
no correlation between molecules is taken Kirkwood's equation inaccurate.
The Frohlich Equation within the same polymer chain, and cos0," is
In the model of Frohlich,28 both the short-range the average of cos $ made between the refer-
interaction between molecules and the defor- ence unit / and a kth unit in the polymer chains
mational polarization are taken into account. that do not contain the reference unit i.39 It is
Frohlich also considered a spherical region also assumed that the polymer chain contains x
within an infinite, homogeneous dielectric con- repeating dipole units and the polymer has a
tinuum of static dielectric constant ers. This high molecular weight, so the contribution to
region contains dipoles whose behavior is gov- the overall polarization made by the end groups
erned by statistical-mechanical laws. Frohlich can be neglected.
also treated the internal and reaction fields in a In contrast to Frohlich's approach, Harris
macroscopic manner, which is more realistic and Adler41 considered a macroscopic sphere
than the Onsager model. The general Frohlich immersed not in an infinite macroscopic dielec-
equation is tric continuum but in a vacuum, in order to
simplify the evaluation of the energy of
(en -£roo)(2£rs + £roo) _ Ngu2 the interaction between the specimen and the
2 (2-218)
ers(eroo + 2) 9£0kT field. Their results predicted dielectric con-
where stants larger by a factor roughly equal to
(2g r5 +l)(g roo +2)
(2-219) — - — . However, because of many
g = l + £cos<fe
3(2e r ,+£ roo )
shortcomings to their approach, several inves-
Nv is the number of the dipoles within the spher- tigators42"46 have proposed different approaches
ical region of volume V, and thus, N = NJV is in order to correct them.
the number of dipoles per unit volume, and Buckingham45 has pointed out that the
cos0o>/ is the average of cosO?> made between Onsager relation between ueff and u0 is a suf-
the reference molecule / and the jth nearest ficiently good approximation for many polar
neighbor. liquids. By taking into account the discrete
Equation 2-218 becomes identical with the nature of the particles surrounding the refer-
Onsager equation (Equation 2-215) when g = 1. ence molecule and the short-range interactions,
However, the definition of u in the Frohlich Buckingham, in his derivation for the dielectric
equation differs from that in the Onsager constants, arrived at the same expression as the
equation.28 If ers» eroo, the Kirkwood equation Frohlich equation (Equation 2-218).
and the Frohlich equation do not differ As mentioned earlier, the statistical theories
significantly. of polarization may enable the determination of
In polymeric liquids or solids, the polymer the effective dipole moment and the correlation
chains are entangled. The appropriate way to factor, which characterize the molecular inter-
deal with such molecules is to choose a small, actions. It has been shown that the value of the
repeating unit of a chain as a basic dipole unit; effective dipole moment in polymers depends
each of these basic units contributes equally on the internal rotation in polymer chains,
to the average polarization of a macroscopic which is greatly hindered due to the presence
sphere of the dielectric in an applied field. of strong molecular interactions.47 A study
Based on this consideration, the Frohlich equa- of the dielectric polarization of polymers cer-
tion can be used for polymers, provided that g tainly will provide useful information about
is replaced by molecular interaction in polymers. As regards
polymers, it is not possible to eliminate the
g' = l + 5/x>s$i + JT cos#£ (2-220) interaction between polar groups belonging to
the same chain; even polymers are dissolved in
a nonpolar solvent with an infinite dilution. In
where cos0," is the average of cos<$* made alcohol, g is large (g > 3 for butyl alcohol at
between the reference unit / and the y'th unit 0°C) because of the high degree of correlation
between the alcohol molecules due to the
existence of hydrogen bonds, which leads to the dynamic response, we can use either the time-
parallel orientation of dipole moments. Replac- domain approach or the frequency-domain
ing the OH radical with a halogen in butyl approach. These two approaches are equally
alcohol leads to a big change in g. For example, powerful methods for studying dielectric
g = 0.76 for butyl bromide and g = 0.85 for phenomena. From the viewpoint of measur-
butyl chloride at 0°C. The value of g < 1 indi- ing techniques, the time-domain approach is
cates that the correlation in orientation makes simpler than the frequency-domain approach,
the dipole moments antiparallel and cancel one but from the viewpoint of data analyses, the
another. The study of g as a function of tem- time-domain approach is more complex. In the
perature and pressure in polar substances, and time-domain approach, we measure the time-
as a function of the concentration in polar solu- dependent polarization immediately after the
tion, is important to understanding molecular application of a step-function electric field, or
interactions with respect to structural changes. we measure the decay of the polarization from
an initial steady state value to zero after the
sudden removal of an initial polarizing field.
2.6 Electric Polarization and This decay is generally referred to as dielectric
Relaxation in Time-Varying relaxation. In the frequency-domain approach,
Electric Fields we mainly measure the dielectric constants at
various frequencies of alternating excitation
Atoms, molecules, and ions are so small that
fields. Both approaches should be intimately
even a macroscopically tiny region in a solid
connected and should yield, in principle, the
contains very large numbers of such particles.
same results.
The discrete nature of matter, and the behavior
and interaction of those particles, can be man-
ifested through their response to time-varying 2.6.2 Complex Permittivity
electric fields with wavelengths comparable to When a time-varying electric field is applied
the distances between particles (interparticle across a parallel-plate capacitor with the plate
distances). In condensed matter—solids and area of one unit and a separation of d between
liquids—the interparticle distances are of the the plates, then the total current is given by
order of a few angstroms; electric fields with
wavelengths of this order would be in the
region of x-rays, which have energies capable
of ionizing the particles. In this section, we will where J is the conduction current and £* is
deal with the dynamic response of the dielec- defined as the complex permittivity, which is
tric material involving electric fields with introduced to allow for dielectric losses due to
wavelengths much larger than the interparticle the friction accompanying polarization and
distances, so the dielectric material can be orientation of electric dipoles. This may be
treated as a continuum. written as
e* = e-je' = {er-je'r)e0 (2-222)
2.6.1 The Time-Domain Approach and
the Frequency-Domain Approach in which er is called the dielectric constant and
£r the loss factor. An arbitrary time-varying
In the previous section, we discussed the
field can be resolved into sinusoidal AC fields
static response of dielectric materials under
of a spectrum of frequencies by means of the
static electric fields with co = 0. However, the
Fourier transformation. For simplicity, we con-
dynamic response under time-varying electric
sider the applied field to be monochromatic and
fields provides much more information about
a sinusoidal function of angular frequency co,
the structure of the material and its dielectric
which can be expressed as
behavior for fundamental studies, as well as
for technological applications. To measure the F = FmexpO'<ar) (2-223)
Substitution of Equations 2-222 and 2-223 into 2.6.3 Time-Dependent Electric

Equation 2-216 yields Polarization
JT=oF+jo)(e - je')F = (a + (oe')F+jcoeF In general, the time required for electronic and
(2-224) atomic polarization and depolarization is very
short (<10"12sec). This deformational polariza-
where a is the electric conductivity of the tion process is also referred to as the resonance
material. The first term on the right is a loss process because it involves vibrating modes.
component due to the inelastic scattering of Resonance of a vibrating system occurs when
conducting charge carriers with scatterers an excitation field oscillates at a frequency
during their migration, which is present at all close to the natural frequency of the system.
frequencies, including co = 0 (DC fields); the The time required for orientational, hopping, or
second term is also a loss component due to space charge polarization and depolarization is
the friction in the polarization processes, which quite long and varies in a wide range, depend-
disappears when co = 0 and increases with co; ing on the dielectric systems; such polarization
and the third term is a lossless component processes are sometimes referred to as relax-
which is, in fact, the displacement current. For ation processes because they involve a relax-
dielectric polymers, a is normally extremely ation time. A relaxation phenomenon occurs
small, and in most cases the contribution of the when restoring action tends to bring the excited
first term can be neglected. By ignoring the first system back to its original equilibrium state.
term, the tan 8 is given by Ignoring, for simplicity, hopping and space
charge polarization, the total polarization of an
tand = £'r/£r (2-225)
arbitrary dielectric system is
which is generally called the loss tangent, as P = Pe + Pi + Po (2-228)
shown in Figure 2-25. If £rl£r « 1 then
Since the response time for electronic and
idxid = e'r I er — 8 (2-226) atomic polarization is so short and can be
assumed to be practically constant for all fre-
This is generally called the loss angle.
quencies from 0 to about 1012Hz, we can group
The instantaneous energy absorbed per
these two polarizations as P^ = Pe + Pf =
second per cm3 by the material is given by
(eroo - l)£0F. These two types of polarization
JT(t)F(t). Thus, on average, the amount of
can be considered to follow instantaneously the
energy per cm3 per second absorbed by the
excitation field F without time lag; in other
material is
words, Poo and F can be considered to be in
, 2K phase. Thus, we can write
W = — f JT{t)F(t)d(cot)
(2-227) P = P . + P0 = (£roo - l)£0F+(e„ - £roo)£aF
= CQ£'re0F*/2 (2-229)
coeF
oF coe'F
Figure 2-25 The response of a parallel plate capacitor with the plate area of one unit and its equivalent circuit under an
alternating electric field.
It is P0 that has a time lag with F; therefore, Po{t) = (£rs -era)e0F[\ -exp(-f/T„)] (2-232)
there is a phase difference between P0 and F.
Supposing that the orientational polarization In this case, dP0{i)ldt gives the polarization
takes time to respond to the applied excitation rate. The approximate time required for polar-
field F, and that at the removal of the excitation ization is shown in Figure 2-27.
field making F = 0 at t = 0, the polarization will Supposing that during the time interval be-
decay at a rate proportional to its change from tween u and u + du, an excitation field F(u) is
its equilibrium state (as shown in Figure 2-26), applied to the dielectric system, and that F = 0
then we can write for t < u and t > u + du (as shown in Figure 2-
28), then P0(t) will take time to respond and will
dPo(t)= P0(t)
(2-230) change for t > u. As soon as P0{t) reaches the
dt T„ value of P0(u + du) at u = t + du, the polariza-
where ra is the macroscopic relaxation time. tion will decay gradually. During the polariza-
Using the boundary condition that at t = 0, P0 tion period u < t < u + du, the change of the
= (ers - eÊoF, the solution of Equation 2-230 polarization can be written as
is dP()(t-u)
po(t) = (ers-£roo)£0Fexp(-t/T0) (2-226)
1 - exp •A \dF{u)
ra ) \
So, dPa(t)/dt gives the depolarization rate for
(2-233)
this case, in which the excitation field is a step-
function, F = constant up to t = 0. Similarly, if in which 1 - exp[-(f - U)/T0] is a response
a step-function excitation field F is applied to function. The total P consists of two parts: P^,
the dielectric system at t = 0, then Pa(i) = 0 at which can follow the excitation field immedi-
t = 0, and P()(t) increases with time. Following ately, and P0{t\ which is governed by Equation
the same method, we obtain 2-233. Thus, we can write
P~ + P,
o — • t " o — • t
Figure 2-26 Time dependence of polarization and depolarization under a step-function electric field F.
c
o
0 .N
c £*
o Q. O
O)"c0 CO Q_
c N
CO c
Q- i=
O CD ©00©
5 o
o ~ Xo Q_
o ©©©©
ariza
entat
©©©0
'^ O
ODL
I ©©00
c
o
c "co
O N
10" 5 1
Time (second)
Figure 2-27 The variation of different types of polarization with time under a step-function electric field F.
dP(t-u) = dP00(t-u) + dP0(t-u) CaseA

= (eroo - \)e0dF{u) + (ers - eroo) If F is a step-function electric field with F = 0
t—u at t = 0~ and F = F at t = 0+, then from Equa-
xe0 1 - exp \dF(u) tion 2-236 we have
(2-234)
P = (erco~l)e0F
(2-237)
According to the superposition principle, the + (ers -erao)£0F[l-exp(-t/z0)]
total polarization at time n s a superposition of
all increments dP, so we have The variation of P with time t is shown in
Figure 2-26.
P = {eroo-\)e0F{t)
t—u
+ (ers-£r~K£ 1 - exp \dF(u)
Case B
(2-235) If F is a sinusoidal AC field with
Integrating by part, we obtain
F = Fm coscot = Re[Fm exp(jox)] (2-238)
P = (eroo-l)e0F(t)
where Re refers to the real part of Fm exp(jcot),
+ (ers - erao)e0 f — ^ exp[-(t - U)/T0]du Poo can follow the field immediately, but P0 has
JO ^
a time lag, i.e., P0 has a phase shift. To analyze
(2-236)
this case, we must assume that P has reached
where exp[-(£ - U)/T0] is a decay function that its dynamic steady state at t = 0, so that the
tends to approach zero at t —> °o. bottom limit of the integral in Equation 2-236
In the following cases, we shall discuss two must be changed to -<*>, where P = 0. From
general cases. Equation 2-236 we obtain
F(t)
1 \ /
k
Po Poo
P(t) 1r 1
r
1
\. Po(t)
p~ ^jS
1 ' 1
u u+ du -• t
Figure 2-28 The time response of P(t) to a delta function electric field F(u) of strength F(u) within the time period of
u < t < u + du.
j exp(jcot)exp[-(t - u)jxa]du
P = (eroo-\ )e() Fm cos cot + (ers -eroo )e() — Re\
To
exp(jcot)
• (eroo - \)e()Fm cos cot + (ers -eroo)e()FmRe\
L \-jan„ J
(2-239)
/ ,x ^ (£rs -e )e()Fm coscot (ers ~eroo)cor0£0Fm sin cot
= (eroo-\)eaFm cos cot + ^- roo :——r~, + ^~
\ + CD2TI \ + co2xl
In plase with the Lag by K/2 from the
applied field applied field
The component in phase with the applied field co » \lx0 the dipoles cannot follow the field
is the lossless component, while the component variation; hence, the polarization gradually
with nil out of phase with the applied field decreases to zero as co increases. The loss corn-
is the loss component. It is obvious that for ponent, in fact, represents the dielectric losses
long time that D has been settled to be a peri-

odic function of time, we can write
D-erooe0Fm cos cot
= (ers - eroo )e0Fm J T(x) cos cot cos coxdx
+ j T( JC) sin cot sin coxdx

(2-243)
By expressing the AC field in the form of
Figure 2-29 The variation of P„, Pu and P2 with Equation 2-238, D can be considered to be the
frequency. real part of the product £*F. Thus,
D = Re[e?e0F] = Re[(£r - j£fr)£0Fm expO'oif)]
in the form of energy absorption. This term is = £0Fm [£r cosfttf+£r'sin o#]
maximum at co0 = l/r0 and decreases gradually (2-244)
as co increases or decreases from the point co0 =
\IT0. Using Poo, Pu and P 2 to denote the three Comparing Equation 2-243 with Equation 2-
components, Equation 2-234 can be written as 244, it follows that
P = (Poo + Pi)coscot + P2 sincot (2-240) (£r - £roo) = (£rs - £roo)J T(x) cos coxdx
(2-245)
These components as functions of frequency
are shown in Figure 2-29. e'r = (srs - £roo) [ r(x) sin coxdx (2-246)
The variation of the dielectric parameters,
such as P and £*, with frequency is known as These two equations indicate that the real and
the dispersion of the dielectric material, and it imaginary parts of complex permittivity depend
is one of the most important properties of all on the same decay function T(x). According to
materials. the Fourier theorem, T(x) can be expressed in
the following form:
2.6.4 Kramers-Kronig Relations
= 2_ r ^ H . ^ ^ (2_247)
Equation 2-231 can be used to derive the rela- 7F J ° £ r v -£ r o o
tions between the dielectric constant er(co) and
the loss factor e'r(oS). These relations are known or
as the Kramers-Kronig relations.48'49 By writing
2 p e'rW
r(u) .
the decay function in a more generalized T(x) = -fQ sinuxdx (2-248)
form T(t - u) to replace —exp[—(f — M)/T 0 ],
These two equations must be equal, so £r must
Equation 2-231 can be written as be related to £'. Substituting Equation 2-248
into Equation 2-245, we obtain
p = (eroQ-\)e0F(t)
+ [ (ers -£roo)e0F(u)T(t-u)du £r(w)-£roo = — j J £'(u)smuxdu\coscoxdx
Jo
Since D = D0 + P = £0F{t) + P, we have = —I [sinuxcoscoxdx]£r(u)du

D = £roo£0F(t)+(ers - eroo )e0 £ F(u)T(t - u)du 2 r~ e'(u)u ,
=— 0 ?du
(2-242) 7tJo U2 -CO2
(2-249)
By introducing a new variable x = t - u and
assuming that the material has been subjected Similarly, substitution of Equation 2-247 into
to the AC field F(t) = Fmcos cot for a sufficiently Equation 2-246 yields
By expressing Fmcoscot = Fm exp(jcot),

f£(<B) = ^-r[er(iO-gr-]
J0
,*" ,<fo (2-250)
z Z Fm sin cot can be expressed as
7T JO r/ — />)
Equations 2-249 and 2-250 are called the Fm sin cot = Fm exp[j(cot - n/2)]
Kramers-Kronig relations. These relations are (2-252)
= -jFm exp(jcot)
very useful under certain situations because
they enable the determination of the value of and Equation 2-239 can be rewritten as
one parameter (er or er) from the other param-
(£rs-£roo)
eter (£' or er) at any frequency. This is particu- £roo-l + \£0Fm QXp(jCOt)
\ + C02T20
larly important when, for some reason, one of
(£rs-£roo)C0Tt
the parameters cannot be measured. These \£0Fm exp(jcot)
relations are valid for any type of polarization, \ + co2xl
(2-253)
and from them a complete spectrum of er and
£' over a wide range of frequencies can be From Equations 2-68 and 2-222, P can also be
obtained—provided that one of the parameters expressed as
(£r or e') can be measured throughout the same
P = (£?-\)£0F
frequency range. (2-254)
= [(£r -1) - j£r]£0Fm exp(jcot)
If we set co = 0 for the case of DC fields,
Equation 2-249 becomes By comparing Equation 2-253 with Equation
2-254, we have
du dco
7TJo U KJo co P —£
e* = £r -je'r = £roo + (2-255)
2 r° 1 +jC0To
= — £\co)d(£nco)
p
+ £n~
£r -= £roo (2-256)
or 1 +co2xl
_ ( £ r , - £ r c ~)C0To
[)£/(co)d(£nco) = -(£rs -£r„) (2-251) F' • (2-257)
\ + co2xl
This implies that the total area under the curve and
of er vs tnco is simply related to ers - e^ the
£r (£rs-£roo)C0T()
extreme values of the dielectric constants, and t x
tan o - — = - (2-258)
is independent of their dispersion mechanisms. £r
We shall return to this interesting feature in the Equations 2-255 through 2-257 are generally
Section 2.7, which deals with the distribution called the Debye equations for dynamic polar-
of relaxation times. ization with only one relaxation time T0.31'5051
These equations are based on the assumption
2.6.5 Debye Equations, Absorption, and that the decay function is exponential. Similar
Dispersion for Dynamic Polarizations to the situation for static polarization, Debye
Strictly speaking, no material is free of dielec- equations are satisfactory only for the condition
tric losses, and therefore, no material is free £rs - £roo < 1, which can be fulfilled only in
of absorption and dispersion. This implies that dilute solutions because the derivation of
there is no material having frequency- Equations 2-255 through 2-257 has not taken
independent £r and £'. In fact, the dispersion into account the interaction between par-
in £r and £r is an intrinsic property of all dielec- ticles. er and e'r are temperature dependent
tric materials, and all other properties have to through the temperature dependence of (ers -
coexist with it. In this section, we shall first eroo) and r0.
show the derivation of the Debye equations We can easily find the frequency at which £r
and then discuss the absorption and dispersion d£r
and tan S are maximal by setting — - = 0 and
phenomena. dco
d(tan<5) eliminating cor0 from these two equations, we

= 0. The maximum value of £' occurs
dco obtain
at co0 when
£rs + £™
CO0T0=1 (2-259) £r-- + £'/ (2-267)
At this frequency, £n s'r and tan 8 are given by

Of course, only the semicircle of Equation
2-267, over which £' is positive, as shown in
£r\(0=(0
o
(2-260) Figure 2-31, has physical significance.
The major disadvantage of this method for
t,rs c,ro
£r I co=co ~~ £(max) "~ "
()
(2-261) finding £r or £' is that the frequency, which is
the independent variable and is the parameter
Cl-C C-^5 most accurately measured, is not explicitly
tan£|ft (2-262)
£„ + £roo shown in the Argand diagram. It should also
be noted that the derivation of Equations 2-255
However, the value of tan 5 at co0 is not
through 2-257 is based on the following as-
maximal. The maximum value of tan 8 occurs
sumptions for simplicity: The local field is the
at G)s when
same as the applied field F\ the conductivity
G)8T0=(ers/er„f2>l (2-263) of the materials is negligible; and all dipoles
have only one identical relaxation time r0. The
At this frequency, tan 8 is given by
effects of these assumptions on £r and £' will be
£ —£ discussed later in this section.
tan S\ ^cos = tan <5(max) = —^ ^ - (2-264) Debye equations emerge directly from the
Kramers-Kronig relations. They show explic-
The values of en £', and tan 8 as functions of co
itly the frequency dependence of £r and £'. The
are shown in Figure 2-30.
type of dispersion shown in Figure 2-30 is
Equations 2-256 and 2-257 can be written as
generally considered the normal dispersion. All
types of polarization may be grouped into
(2-265)
two major regimes: the resonance regime and
the relaxation regime. Polarizations associated
er m0 (2-266) with vibrations of electrons (i.e., electronic
£ -£roo 1 + (IT To
rs
or optical polarization) or with vibrations of
These two equations are the parametric equa- atoms or ions (i.e., atomic or ionic polarization)
tions of a circle in the £r - e'r plane. By belong to the resonance regime because in the
e
/-(max)- (£rs e
/-°°V2
(o=0
O=1/T0\ V
co5=(ers/eroo) 2 (T 0 )" 1 (ers+£roo)/2
Figure 2-30 er, s'r and tan 5 as functions of (0 for cases Figure 2-31 Argand diagram of er - e'r relation for cases
with negligible contribution of a due to carrier migration. with only one relaxation time T0.
polarization a resonance will occur when the valid in dilute gases. In Section 2.5, we dis-
frequency of the excitation field is close to the cussed the internal field or the Lorentz field
natural frequency of the vibration or oscillation (Equation 2-188). By taking into account the
system. Polarizations involving the movements local field based on the Lorentz field, Equations
of charges either by orientation (i.e., orienta- 2-255 through 2-258 and other equations are
tional polarization) or through the migration of still valid if T0 is replaced with T'0, which is
charge carriers (i.e., hopping or space charge given by17
polarization) belong to the relaxation regime
because during the polarization or depolariza- *o= " T 0
(2-271)
tion processes, a relaxation phenomenon occurs e^ + 2
due to the time required for the charge carriers In general, the measured relaxation time based
to overcome the inertia arising from the sur- on the Cole-Cole plot is higher than the real
rounding medium in order to proceed in their relaxation time, possibly due to the local field
movement. The variation of er and e'r with effect. Note that the local field given by Equa-
frequency shown in Figure 2-30 illustrates tion 2-193 is not applicable to condensed polar
schematically the typical dispersion behavior polymer materials, as explained in Section
for polarizations in the relaxation regime. 2.5.2.
However, in the resonance regime, the
optical constants, i.e., the complex refractive
index n *, which consists of the refractive index The Effects of DC Conductivity
n and the extinction coefficient (or called the If the DC conductivity oris not negligibly small,
absorption index) k, can be expressed in terms then c will contribute to the imaginary part of
of the complex dielectric constant as complex permittivity. The total complex per-
mittivity becomes
n* = n- jk
(2-268)
= (e?y/i=(er-jeT £* = £-J\ 1
£ +— £r-j\£r + \£o
CO (0£o
Rearrangement of Equation 2-268 leads to
or
er=n2-k2 (2-269)
£7=£r-j£r-J — (2-272)
e'r = Ink (2-270) 0)
The frequency dependence of er and e' in the Taking the DC conductivity into account, the
optical frequency range, i.e., f > 10 n Hz, and Debye equation (Equation 2-255), for example,
the dispersion of er and e'r (or n and k) are anom- becomes
alous. So this type of dispersion is sometimes
referred to as anomalous dispersion. Figure 3- £? -£r~=- -J-C0£ (2-273)
1 + JCQT0
32 in Chapter 3 shows schematically this anom- o
alous dispersion phenomenon, and Figure 3-33 and Equations 2-257 and 2-258 become
shows the variation of er and e'r with frequency
for various types of polarization in both the , (£rv-£rooWr„ o (2-274)
£r = — - — H
resonance and the relaxation regimes. For 1 + G)2Z2 (0£o
more details, see Absorption and Dispersion (0£o(£rs -£roo)(0To + (l + C02T2 )<J
in Chapter 3. tan<5 =
Q)£0(£rs+£rooC02T2)
(2-275)
The Effects of the Local Field When COT0 « 1, Equations 2-274 and 2-275
reduce to
So far in Section 2.6, all equations derived are
based on the assumption that the local field is
£r=- (2-276)
the same as the applied field. This, of course, is (0£fl
the basis of the suggestion of Cole and Cole,

tan<5 = - (2-277)
C0£o£rs Equation 2-282 can be written as
When C0To « 1, Equations 2-274 and 2-275 £r-£r» l + (wT) 1-a sina/r/2
reduce to - £_ 1 + 2(my-a sin an/2 + (m)2{l-a)
(2-283)
(2-278)
2 e0 (cor)1 a
cosan/2
g 0 ( g r j - g r , ) + 20T 0 £rs - groo 1 + 2( COT) ' sin an J 2 + (coT)2(l-a)
1 01
tan<5 = (2-279) (2-284)

£o(£rs+£r~)
When C0To » 1, Equations 2-274 and 2-275 When a= 0, Equations 2-283 and 2-284 reduce
reduce to back to the Debye equations. The dispersion
curve with a > 0 is broader than that for a single
£r
c-rs £p>
(2-280) relaxation time but still symmetrical about COT
COT0 de'r
£0(£rs-£roo) + OT0
= 1. By setting = 0, we can find that the
tan 8 = (2-281) dco
C0To£n£roo
maximum loss also occurs at COT = 1. By elim-
The variation of e'r and tan 8 with co, includ- inating cor from Equations 2-283 and 2-284, we
ing the effect of DC conductivity, is shown in obtain
Figure 2-32, and the Argand diagram of er - £' (£rs+£r„y , (£rs -£roo) an
relation showing this effect in Figure 2-33. £r- +
2 2 2 .
&rs £r°° an
-sec —
2.6.6 The Cole-Cole Plot 2 2 (2-285)
Debye equations (Equations 2-256 and 2-257) This equation represents a circle with the center
based on a single relaxation time do not suffice
^ rs & r°z i j f-]^
to describe the relaxation phenomena for most at -tan- an(
dielectric materials, e.g., polymers. In this case,

a distribution of relaxation times is necessary radius of an This is shown in
-sec-
to interpret the experimental data. In Section 2 2
2.7, we shall discuss the approach based on a Figure 2-34. The Cole-Cole plot exhibits a
distribution of relaxation times. However, to depressed semicircle because of a > 0. Some
take into account the effect of a distribution of experimental results agree well with Equations
relaxation times, Cole and Cole 52 have pro- 2-283 and 2-284 with a > 0.39
posed that an Argand diagram, in which e'r is Debye's er - £' semicircle is based on a = 0,
plotted as a function of en can be constructed which represents the materials' having only one
following the empirical relation single relaxation time. In general, there exists
a distribution of relaxation times in all solid
£rs £r»
F —F = F — F -J£r materials because there are always some
I+U<D*OT nonuniformities in local domains, which would
(2-282)
alter the individual dipoles or charges in addi-
where a is the parameters with 0 < a < 1. tion to existing dipoles in the material. If the
In fact, many materials, particularly polymers relaxation times are due to several different
with long chains, have a broader £rr- tnco dis- mechanisms, then the £r - £' arc will be asym-
persion curve and a lower maximum loss than metrical.53'54 Several investigators53"56 have put
would be expected from the Debye semicircle forward suggestions to modify the Cole-Cole
shown in Figure 2-31. 39 In such cases, the er - empirical equations. Some of the equations are
E\ arc will fall inside the Debye semicircle. On listed below for comparison purposes:
Dominant
o- Effect
Negligible
a Effect
Negligible
o- Effect
co=(ers/erS/2ô
Figure 2-32 e'r and tan 5 as functions of co, taking into account the effect of DC conductivity.
equation will reduce to the Debye equation for

j8 = 1. But, with (5 < 1, the er - e'r arc will become
asymmetrical, as shown in Figure 2-35(a). It has
been found that several materials obey the
Davidson-Cole equation, such as glycerol tri-
acetate55 and Pyralene (a fluid mixture of
chlorinated diphenols in chlorobenzene).17
However, the Havriliak-Negami equation gives
a much better fit to most experimental results if
a and (3 are properly chosen. The er - e'r arc for
a = 2/3 and /? = 1/2 is shown in Figure 2-35(b).
In the Havriliak-Negami equation, the parame-
ters a and /? are not based on the physics of
the dielectric polarization, although the modifi-
Figure 2-33 The effect of DC conductivity on the er - e'r cation of the original Cole-Cole equation
arc. (A) a= 0, (B) o= ox > 0, (C) a2 > Oi, (D) <73 = a2.
empirically may make the equation better fit
experimental results. The modification does not
lead to a better understanding of the physics
behind the distribution of relaxation times.
Fuoss and Kirkwood56 have also put for-
ward a semi-empirical equation, called the
Fuoss-Kirkwood equation, to relate only the
imaginary part £' of the complex dielectric con-
stant with frequency, which is given by
2e'r, max)
£r{(0) = (2-288)
(<0T)A+(COT)"
Figure 2-34 The Cole-Cole plot for dielectric material
with a set of relaxation times.
where X is the parameter with the value 0 < A
< 1, and £'(maX) is the maximum value of the loss
factor when COT = 1. On a logarithmic fre-
Debye equation quency scale, the efr - (£nco) relation exhibits a
symmetrical bell-shaped curve about a central
£ —£ = - (2-255) frequency co = 1/T, which is broader than the
1 + jCOT
corresponding Debye curve (Equation 2-257),
Cole-Cole equation but very similar in shape to, although not iden-
tical with, that based on the Cole-Cole equa-
~£roo=- (2-282) tion (Equation 2-284). The difference can be
l + (jC0T)l-a
seen by comparing these two equations.
Davidson-Cole equation Cole57'58 has discussed the correlation between
these two equations in some detail.
VT.V tro
£ —£ oo =
"
vrs v-roo
(2-286)
(1 + W
Havriliak-Negami equation 2.6.7 Temperature Dependence of
Complex Permittivity
£ ~ £ oo
£* _ £ ^ = (2-287) Complex permittivity is a complex function of
[i+o^y-i both the frequency and the temperature, apart
All of these equations depend on the values of from the dependence of other parameters such
a and (5 chosen within the ranges 0 < a < 1 and as pressure, etc. We have briefly discussed
0 < (} < 1. Obviously, the Davidson-Cole the frequency dependence of er and £'. The
(a)
£r
(b)
Figure 2-35 The £r - e'r arcs based on (a) the Davidson-Cole equation for p = 1/2, and (b) the Havriliak-Negami equa-
tion for a = 1/3 and £ = 1/2.
temperature dependence of er and £' is mainly {ers - ErJ) for nondipolar materials is not tem-
through the temperature dependence of the perature dependent, then the area beneath the e'r
relaxation time which can, for a single relaxa- vs \IT curve depends only on the activation
tion time, be expressed as energy H and does not depend explicitly on fre-
quency, based on the following equation:
T = rhexp(H/kT) (2-289)
where %h is a pre-exponential factor and H is
the activation energy.28 By replacing %, with a
I>l --(£n.-£roo)
HV2
- tan COT
temperature-dependent T, the Debye equations (2-292)

(Equations 2-256 and 2-257) can be written as For tan-1 cor < nil, the area beneath the £'
vs \IT curve is practically independent of
£r = S'roo +
el-el- (2-290) frequency.
l + co2r2hQxp(2H/kT)
However, for dipolar materials with dipolar
cozh Qxp(H/kT) molecules, (eTrs - £Tr0o) is temperature dependent.
£;=(sT-el).
\ + co2r2hexp(2H/kT) According to the theories of Onsager, Frohlich,
(2-291) and Debye (see Section 2.5.2), we can write
where eTrs and ejL are the value of ers and £roo at
temperature T. It can be seen from these two (eT ~eT ) = — (2-293)
\*-TS Woo/
equations that the dielectric constant er and the
loss peak £^maX) decrease with increasing tem- where A is a constant. In this case, the area
perature, and the loss peak shifts toward higher beneath the £' vs 1/7 curve is more compli-
temperatures, as shown in Figure 2-36. If cated, and it is given by39'59
£r
^
1 A_ T2>T,
V ^x C02 > COj
W
/ \ -/ TT
\>^72>71
(02 > 6>1
N
£'r / A '
CO >-
(a) (b)
®1 r\ C02 > tf)-|
\f\ /'V
\ \
\ \ / 1
/ 1
>v S
^s,^^ ^ ^
I ' 1 "*• —
(c)
Figure 2-36 (a) er and e'r as functions of co for fixed temperatures Tx and T2 with T2>TU (b) £r and £' as functions of tem-
perature for fixed frequencies coi and coi with (Oi > cou and (c) tan 8 as a function of temperature for fixed frequencies C0\
and o>2 with C0i> cox. For the same frequency, Tmax for £'(max) is higher than T'max for tan£(max).
rAi
Jo ' ^7" J:,«%J=e* Croo )
nk
~2H
(2-297)
Ak2 KinCOT y (-l) r (0)T) 2 r + 1

For COT< 1, Equation 2-297 differs from Equa-
H (2r + l) 2 tion 2-292 only in the (eTrs - eTr„) term. Thus, the
(2-294)
area beneath the £' vs \IT curve is still practi-
By denoting J max as the temperature at which e'r cally independent of the frequency, as shown in
is maximum, then Figure 2-36. It should be noted that for the same
co, T'max, for tan 8 to be maximum, is lower than
(2-295) ^max for £' to be maximum.
x
max The same concept for temperature depend-
ence of er and e'r can be applied to other empir-
^Tmax=rhexp(H/kTmax) (2-296)
ical equations—the Cole-Cole equation, the
In general, 6)Trmax differs little from unity. 60 Davidson-Cole equation, and the Havriliak-
Thus, for the frequency range normally encoun- Negami equation. The measurement of e'r as a
tered in practice (cor < 1), Equation 2-294 can function of frequency at a fixed temperature
be approximated to is equivalent to the measurement of e'r as a
function of temperature at a fixed frequency. microwave fields of frequencies of the order of

Both methods should provide similar informa- 10 GHz (10 x 10 9 Hz) or higher may become
tion provided that there is no change in the spec- complex. Only at AC fields of frequencies low
imen and the DC conductivity is negligible over enough for the current capable of following
the temperature range for the measurements. the field, the conductivity would have the value
appropriate to that at a DC field of the same
magnitude. If the total field applied to a semi-
2.6.8 Field Dependence of Complex
conductor specimen is
Permittivity
From Equations 2-68 and 2-224, it can be seen F = FDC + FAC cos cot (2-300)
that if the relationship between the polariza-
with FAC « FDC. For AC fields not high enough
tion P and the electric field F is linear and the
to heat the charge carriers, the current may no
electrical conductivity o is constant, then the
longer follow the field at frequencies for which
dielectric constant er and the loss factor e'r
cor is comparable to unity, where r can be taken
should be independent of electric field. This is
as the mean free time between collisions.
true only when the total field during the meas-
Obviously, r depends on both the temperature
urement of these parameters is low. However,
and the applied field magnitude; it decreases
when the specimen is biased with a strong field,
with increasing temperature and with increas-
a may not be constant or the P-F relationship
ing field magnitude. So we can describe the sit-
may become nonlinear, then er and e'n measured
uation by expressing the AC conductivity as a
with a small-signal AC voltage, will be field-
complex conductivity <7*(co) with the real part
dependent. Since the total current density in a
G(CQ) representing the in-phase conductivity,
material specimen under an AC field is given
which means the current capable of following
by
the field, and the imaginary part &{co) repre-
dF senting the out-of-phase conductivity (nil
JT = J + e* — = oF+j(er - je' )e0F lagging the field). For a simple band structure,
dt (2-298)
the AC conductivity at low AC fields is given
= (<7 + e're() )F+jer£a F
by
the complex relative permittivity, including the
electrical conductivity cr, can be expressed as nq (AEaverm)
O*((0)-
m*L(AEflW(l + c>2T£))
mm(AEaverm) (2-301)
E?=er-j£rT=er-j\er + C0£ (2-299)
o
-J-,
= <j(a>) + jo]((D)
where J is the conduction current, e'rT is the total
loss factor, including the loss factor due to the where AEave is the average energy of the carrier,
loss involved in the polarization processes and n is the carrier concentration, q is the electronic
the loss due to the electrical conduction involv- charge, m* is the effective mass of the carrier,
ing the movement of charge carriers. In the fol- and Tm is the momentum relaxation time, which
lowing sections, we shall discuss the effects of is the average time required for a component
strong electric fields on complex permittivity of the carrier momentum in any direction to be
for three categories of materials: semiconduct- reduced to Me of its value. 61,62 The T mentioned
ing, ferroelectric, and insulating. above can be considered to be identical to Tm.61
The symbol ( ) indicates the average values to
be taken over the carrier distribution in the
Semiconducting Materials specimen. The out-of-phase current {nil lag-
In semiconductors, the concentration of free ging FAC) may be thought of as a contribution
charge carriers and hence the conductivity are to the displacement current, so that we can
high. The value of the conductivity <7 at write the dielectric constant and the loss factor
û
1 o ^
ÂT7*3£~A°~^>~C^0
15.5
15 - w^
300 K
p78K AAV°AAaAAA°°*o
10 15.0 A
c
o A
O 5 -I c
CO
o o
O 14.5 a^
o
o
A*
0
b 14.0 I
O
Apparent Dielectric
A
A Constant of Intrinsic
CD
-5 CO Germanium
CO
g> 13.62 A
"co
E
co 13.5
R
I A
co • -10 "co I Observed Dielectric
E Constant
CO A
13.0 I
° Dielectric Constant
-15
Deduced from
<> A Absorption Data
.on 12.5
2.5 5.0 7.5 10 12.5 15 2 3 4
Electric Field (kV/cm) Electric Field (kV/cm)
(a) (b)
Figure 2-37 The variation of small-signal dielectric constant with DC bias field: (a) for «-Ge specimen of low-field resis-
tivity of 1.9 ohm-cm at 300 K and small-signal frequency of 70 GHz, and (b) for rc-Ge specimen of low field resistivity of
4.7 ohm-cm at 300 K and small signal frequency of 35 GHz.
for semiconductors at high frequency AC fields of mm(mm < 0.1). However, at T = 78°K and
as FDC - 0, Tm becomes larger, making cozm close
to unity. In this case, the carrier contribution
£rT(co) = £r(co)+ (2-302) becomes large enough to result in a large neg-
C0£o
ative value of £rT. The DC bias field tends to
£'rT(CQ) = e'r(G)) +
via) (2-303) reduce the value of Tm and hence the value of
CD£n C0Tm, thus reducing the carrier contribution to
Since a\(o) is negative, the contribution of the the dielectric constant, as shown in Figure 2-
carriers to the dielectric constant is also nega- 37(a). Similar results have also been observed
tive. This implies that a\co) tends to decrease for ft-type germanium specimens with low-field
the dielectric constant, and it may even make resistivity of 4.7 ohm-cm and the small-signal
the dielectric constant negative for high carrier frequency of 35 GHz, as shown in Figure 2-
concentrations. 37(b). The results are from Gibson, Granville,
Using a small-signal AC field FAC cos cot and Pagie.62'64 Obviously, the loss factor £rT is
superimposed on a DC bias field FDC, as given also field dependent through the field depend-
by Equation 2-300, the field dependence of the ence of rm, as indicated in Equations 2-301 and
dielectric constant for n type germanium spec- 2-303.
imens with low-field resistivity of 1.9 ohm-cm
and the small-signal frequency of 70 GHz is Ferroelectric Materials
shown in Figure 2-37(a). The results are from For ferroelectric materials, the applied electric
Conwell, Fowler, and Zucker.62'63 At T = 300°K field has a marked effect on complex permit-
and FDC = 0, erT does not lie much below en tivity. In Thermodynamic Theory in Chapter 4,
which is about 16, because of the small value we mention that the application of a suitable
DC field can shift the Curie point temperature the space charge polarizations for simplicity, the
of BaTi0 3 to a higher temperature, converting major types of the polarization are electronic
the paraelectric state to the ferroelectric state; polarization (Pe), atomic polarization (P,-), and
this converted ferroelectric state will return to orientational polarization (P0). For Pe and Ph the
the paraelectric state when the electric field is polarization-field relationship should remain
removed. However, dielectric constant is field linear even at high fields. But for P0, the polar-
dependent in the paraelectric state at tempera- ization-field relationship is linear only at low
tures above the Curie point temperature Tc. fields and becomes nonlinear at high fields.
Typical results for triglycine sulphate (TGS) Dipolar molecules in a material are similar to
and potassium dihydrogen phosphate (KDP) balls with a positive charge on one end and a
are shown in Figure 2-38. In the measurements negative charge of the same magnitude on the
of the effects of an applied electric field, the other end in a viscous fluid. They are in motion
electric field used to induce the polarization due to thermal agitation. When an electric field
P was a 200 Hz sinusoidal field, while for the is applied across a material specimen between
measurements of the dielectric constants a two electrodes, these dipolar molecules will
small-signal AC field was used and kept con- orient toward the direction of the field, with the
stant and low at a fixed frequency of 50 kHz. end of positive charge facing the negative elec-
The results shown in Figure 2-38(a) for TGS trode (cathode) and the end of negative charge
are from Triebwasser,65 and those in Figure 2- facing the positive electrode (anode), resulting
38(b) for KDP are from Baumpartner.66 in orientational polarization. The magnitude
When no electric field is applied to the spec- of the polarization depends on the angle made
imen, the polarization P is equal to zero. When between the dipole axis and the field direction,
a finite electric field F is applied, it induces a which, in turn, depends on the applied field
finite value of the polarization P. For T> Tc and strength and the temperature. The degree of
F = 0, the dielectric constant follows closely the polarization is governed by the Langevin func-
Curie-Weiss relation (see Chapter 4) tion, as discussed in Section 2.3.3. It should be
noted that molecules are generally not spherical
er=—— (2-304) in shape, but rather ellipsoidal. So, even
T-Tc in nondipolar materials, the molecules are not
where C is known as the Curie constant. In fer- dipolar but are ellipsoidal in shape; the mole-
roelectric materials, there exists an intrinsic cules will still orient under an applied electric
bias, which produces just the same effects on field in order to reduce their potential energy. In
the dielectric constant as an applied bias.67 fact, a molecule can be considered a configura-
Since the applied electric field along the ferro- tion built up of a number of charges; it will be
electric axis is opposite in sign to the intrinsic distorted in an elastic system under an external
bias, the net dielectric constant decreases with electric field. Including its distortion, any mol-
increasing applied field, as shown in Fig. 2.38. ecule can be represented by an ellipsoid with
three principal polarizabilities related to the
Insulating Materials three perpendicular axes. If such a molecule is
Insulating materials generally refer to the mate- placed in a field, the axis of highest polarizabil-
rials with a large bandgap, a low concentration ity will orient toward the direction of the field
of free charge carriers, a small carrier mobility, to reduce its potential energy to a minimum.
and hence an extremely small conductivity. Therefore, there will exist an orientation effect
Therefore, for these materials we can ignore in a manner similar to the dipolar molecules (see
the effects of electrical conuctivity a because it Classical Approach in Section 2.3.1).
would be negligibly small. The field depend- It can be imagined that in a high electric field,
ence of the complex permittivity is associated all individual molecules will be largely oriented,
mainly with the nonlinear field dependence of so that there can be little extra polarization from
the polarization. Excluding the hopping and the further orientation of dipolar or nondipolar
1.2
10000
c
&
w
c
o
O
0
b
8000
12000
18000
J_ _L J_ 100
0.1 0.2 0.3 0.5 1.0 2.0 3.0 5.0 10.0
•T C (°C)
(b)
Figure 2-38 The variation of small-signal dielectric constant with DC bias field, with the field applied along the ferro-
electric axis at temperatures above the Curie point temperature Tc for (a) triglycine sulphate (TGS) [(NF^CF^COOH^t^SOJ,
for which er(F) is normalized to er(F - 0), and (b) potassium dihydrogen phosphate (KDP) [KH2P04] plotted as a function
of temperature at various DC bias fields.
molecules when the field is further increased. in which z is

Under this situation, the polarization becomes
saturated; this phenomenon is known as z = ~kT (2-307)
dielectric saturation. Several investigators 68-72
have observed experimentally that the static Here, we must use the local field Floc instead of
dielectric constant of dipolar solution decreases the applied field F. The local field is always
as the applied electric field is increased, and that larger than F by a local field correction factor
the amount of the decrease in dielectric constant B, so Equation 2-302 can be rewritten as
is approximately proportional to the square of
u0BF (2-308)
the field strength. It should be noted that in order z = - kT
to observe the dielectric saturation phenome-
non, the material used must be in fluid state so At high fields (HF) for which z > 1, we must
that molecules have freedom to orient; in a mate- include higher power terms of L(z). For sim-
rial in solid state, the molecules are frozen, plicity, we take only the terms up to z3. Thus,
losing their ability to move. Several investiga- the polarization can be written as
tors have studied this phenomenon theoretically P(HF) - [(£roo - 1 ) + £ro(HF)\£oF
and proposed equations to elucidate quantita-
BulF
tively the behavior of the dielectric constant at = N aeBF+aiBF +
2>kT (2-309)
high fields. The first was Debye. 31 Later, van
Vlect, 18 ' 72 Bottcher, 6 and several others 74-79 also uQ (u0BF
put forward their approaches based on their 451 kT
assumptions. All of the theoretical equations are Assuming that for the same high field, the
derived on the basis of some assumptions, but dielectric constant remains the same as that
none of them is very satisfactory. It is obvious obtained at low fields, based on the linear part
that the tremendous local fields existing around of the Langevin function shown in Figure 2-17,
ions and dipoles in the solution must have large the polarization can be written as
effects on the dielectric behavior. An accurate
method to calculate such local fields is still not P = [(eroo-\)+ero]eaF
available. We shall not include the discussion of BulF (2-310)
= N\ aeBF+atBF+
various analyses for this dielectric saturation 3kT
phenomenon here. The reader interested in this
The difference between P(HF) - P in Equation
topic is referred to the references cited above. In
2-309 and Equation 2-310 gives the difference
the following, we shall present a simple analy-
between Em{HF) and Ew due to the effects of high
sis with the aim of showing the physical concept
fields. Thus, we have
of this phenomenon.
In a dielectric solution under a high electric A P - P(HF) — P — [Ero(HF) ~ Ero\eoF
field, the average dipole moment in the direc- Nu0fu0BF\ 3 (2-311)
tion of the field is given by [see Equations 45 { kT J
(2.123), (2.129) and (2.130)]
and the decrease in dielectric constant as
{uF) = u0L(z) (2-305)
Nu40F2
Aer • B3 (2-312)
where L(z) is the Langevin function, which can 45e0k3T3
be expressed in the form of a series as Depending on the local field correction factor
B, A£r is proportional to the square of the
L(z) = coth z — applied field and inversely proportional to T3.
z If the local field, based on Lorentz's approach,
= (2-306) is used, B is given by
_z + 3 45 + 945
Srs + 2
= B= (2-313)
3 45 + 945
If the reaction field, based on Onsager's protect the circuit; it also gives a chance to
approach, is used, B is given by adjust the amplifier to a null position before
recording the transient current. It is important
B=3(ers-e:)(2e+er„)
to make the time constant of the amplifier,
ers(eroo + 2f
which depends on the stray capacitance in shunt
The basic difference among the different with Ru much smaller than the time during
expressions put forward by other investigators which the transient current is flowing. The
lies mainly in the different expressions for B. specimen has the guard and the guarded elec-
However, this simple analysis, though not accu- trodes, the outer guard electrode being con-
rate, does provide a way to understand this nected to ground to eliminate surface leakage
phenomenon. currents from the specimen. The charging or
discharging current is measured as a voltage
appearing across Rt by means of a DC ampli-
fier. The voltage drop from point A to ground
2.7 Dielectric Relaxation
is made zero by a negative feedback in the
Phenomena
amplifier circuit, which produces a voltage
Using the time-domain approach, measure- across R2 equal and opposite to that across Ru
ments of the growth or the decay of the polar- thus making the applied step voltage across the
ization when a step-function electric field is specimen only. The step voltage, and the charg-
suddenly applied or removed from the speci- ing and discharging current as function of time,
men will yield information about the relaxation are schematically shown in Figure 2-39(b), in
behavior equivalent to that from the measure- which I0 is the steady DC component of the
ments of er and e'r as functions of frequency charging current and the width of the step
based on the frequency-domain approach. voltage is 63 seconds.
In the previous section, we discussed the rela-
tion of the dielectric constant and absorption,
with frequency based on the frequency-domain 2.7.1 The Hamon Approximation
approach. In the present section, we shall use In principle, the equivalent frequency-domain
the time-domain approach to deal with relax- response of a dielectric material can be
ation phenomena. In general, the time-domain obtained from a time-domain response by a
response provides conspicuous information Fourier transformation which extracts all the
about the nonlinearity of the dielectric behav- separate harmonic components from the charg-
ior simply by varying the amplitude of the ing or the discharging current. For a linear
applied step-function field. dielectric material in a capacitor of unit area in
The basic experimental arrangement for cross-section and d in thickness, as shown in
the measurements of the time-domain response, Figure 2-39(a), the complex dielectric constant
(i.e., the transient charging or discharging can be expressed as
current, resulting from the application or the
removal of a step DC voltage) is shown ef(co) = e^ + J (f)(t)e-j(otdt -j-^- (2-315)
schematically in Figure 2-39(a). o ®^°
This arrangement was originally used by where o0 is the steady DC electrical conductiv-
Williams.80 The circuit is self-explanatory. The ity and 0(0 is the decay function of the tran-
switch Si has two positions: one for turning on sient current J(t).s\ Thus, (j)(t) can be written
the step DC voltage to start the flow of the as
charging current, the other for short-circuiting
the specimen to allow the discharging current
to flow after the specimen has been fully m =^ (2-316)
charged to a steady-state level. The switch S2 is e0F
used to short-circuit Rx to provide a path for where F is equal to VId. Note that 0(0 does not
surge currents for a very short period of time to include the contribution to the observed tran-
-V1-
/////// Specimen
A*
Amplifier
s
R,f 2
B li
Meter
(a)
Step Voltage
f=0 f=63sec
10 20 30 40 50 60 120
Time (sec)
(b)
Figure 2-39 Schematic diagrams illustrating (a) the basic experimental arrangement for the measurements of the charg-
ing and the discharging current resulting from the application and the removal of a step voltage, and (b) the step voltage
and the charging and the discharging current as functions of time. I0 is the steady DC component of the charging current.
sient current J(t) from the steady DC conduc- For a simple relaxation process involving only
tivity of the specimen. By considering only the one single relaxation time based on the Debye
relaxation component of the transient current, theory, the decay function is
the complex dielectric constant should not
</>(0 = Aexp(-//r o ) (2-318)
include the term o0l(Q£0, although it is part of
the total loss factor. Thus, we have where
A = (ers-eroo)/z0 (2-319)
jmt
£*((») = er(co) - j€r(co) = e^ + ^{t)e~ dt
However, for relaxation processes involving
(2-317) a distribution of relaxation times, as in poly-
mers, </)(t) would be a complicated function of The basic criterion for the Hamon approxima-
time, and an explicit expression for the spectra tion is that at a certain value of t in Equation 2-
of er(co) and £'r(co) cannot be deduced by 324 £' (or) value of t in Equation 2-324, £'r(co)
analytical integration of Equation 2-317. Of at any desired value of co can be obtained
course, if 0 ( 0 can be determined experimen- simply by the following expression
tally over the entire range of time, numerical
J(t)
solution of Equation 2-317, with the aid of a £r{co) = (2-325)
C0£oF
computer, would provide information about
£r(co) and eXco). This criterion implies that
However, the frequency-domain response for
r(l-r)cosr^/2
the case with a distribution of relaxation times l
{coty
can be expressed by empirical equations (see
Equations 2-282 and 2-286 through 2-288), or
hinting that the time-domain response can also atf = [ r ( l - r ) c o s r ; r / 2 ]Vr (2-326)
be expressed by an empirical equation such as
Hamon found that for 0.1 < r < 1.2, the right-
j(t) = Arr (2-320)
hand side of Equation 2-326 is approximately
From Equation 2-316, the decay function can constant at a value of 7z/5 = 0.63 at an accuracy
be written as of about 3 % . Based on the Hamon approxima-
tion, the loss factor can be expressed simply as
(2-321)
£oF J(t = 0.63/CQ)
£'{co)- (2-327)
CO£„F
in which B = Ale0F. At a fixed constant tem-
perature, the constant A (and hence the con- This equation indicates that the observed tran-
stants B and r) can be determined for a specific sient current J(t) at t = 1 sec and 10 sec corre-
dielectric material by a direct fitting of Equa- sponds to the frequency / = col2n= 0.6312m =
tion 2-320 to the experimental data of the 0.1/t = 0.1 Hz for t = 1 sec, and 10" 2 Hz for t =
charging or the discharging current. Substitu- 10 sec. That means that we can obtain £'r(co) at
tion of Equation 2-321 into Equation 2-317, / = 0.1 Hz and 10~2Hz using the value of J(t) at
followed by integration, yields t = 1 sec and at 10 sec, respectively. The dielec-
tric constant £r(co), based on the Hamon approx-
er{(0) = £«, + BcorlT{\ - r)sin rn/2, 0 < r < 1
imation, can be deduced from Equations 2-322
(2-322)
through 2-324. It is
rl
£'r(oo) = Bco ' T(l-r)cosr7i/2, 0<r<2
£r(co) = £^ + £'r((o)tanm/2, 0.1<r<l
(2-323)
(2-328)
16 39
where T(l - r) is the gamma function. ' If B
The Hamon approximation is not rigorous,
and r are known, £r{(0) and £'(co) at any desired
since Equation 2-321 does not hold in practice
value of co can be obtained from Equations 2-
over the entire range of times. Williams 83 has
322 and 2-323.
shown that the Hamon approximation can be
To simplify the calculation, Hamon 82 has
derived more rigorously for low-frequency long
proposed an approximation method, generally
relaxation time or high-frequency short relax-
referred to as the Hamon approximation. Equa-
ation time for dielectric materials following the
tion 2-323 for £'r(co) may be rearranged to the
Cole-Cole empirical relation (see Equation
form
2-282). Later, Hyde 84 treated this time-
r
Br \ 'r(l-r)cosr7r/2l frequency transformation problem based on the
£r((o) = :
CO [ (cot)" \ data of the time-integral of the transient current,
(2-324)
fr(l-r)cos«r/2" i.e., the charge Q(t). The measurement of the
(oe„F I (ox)-' charge on the capacitor plates rather than the
current through the specimen has some advan- //for dipole orientation and hence the relaxation
tages in principle. The decay function for Q(t) time T, as well as the preexponential factor
can be obtained directly, and Q(t) is usually Th based on Equation 2-289. In polymers, the
bound at short times, while J(t) involves a much molecules are complex; the orientation of the
longer time to reach a steady level though the polar group related to the link segment along a
current is much easier to measure. Hyde's polymer chain may involve many relaxation
approach may be more accurate, but it involves times. For example, in polyvinyl chloride, the
more computations. Using in-line computation, degree of rotation of the polar groups about the
Hyde's method can provide a rapid means of C-C axis depends on the position and the angle
determining £r(co) and e'r(co) over the frequency of the section of the C-C links.
range of 10"4 to 106Hz.84 The Fourier transfor- If a system has more than one type of dipole,
mation technique has been extended to obtain the relaxation times will be distributed. Sup-
very high frequency (108 to 1010Hz) data. The posing that N is the number of dipoles of one
experimental method is usually referred to as type per unit volume, than we can write
time-domain spectroscopy. It involves the use
N = N0f(r) (2-329)
of a waveguide and very short-duration pulses
with fast rise times, the reflected or the trans- where N0 is the total number of dipoles of all
mitted pulses being observed with a sampling different types per unit volume,/(r) is the dis-
oscilloscope.85 tribution function of the relaxation time Tor the
However, the Hamon approximation is probability density of T. Thus, J{t)dr represents
useful for a rapid appraisal of the dielectric the probability density of/(r) in the interval
losses and the dielectric constant from the between r and r + dr. So, we can write
polarization or the depolarization current in
practical dielectric materials, such as polymers. ]f(r)dz = l (2-330)
2.7.2 Distribution of Relaxation Times Taking into account the distribution of relax-
Dipoles (permanent or induced) are generally ation times, the Debye equations, (Equations 2-
present in inorganic, organic, and biological 255 through 2-258) become
materials. Orientational polarization plays a
f(*)dr
decisive role in dielectric phenomena in materi- fc-r C r o o i \Crs fcj-oo/J (2-331)
als. The behavior of a polar molecule in a mate-
rial is similar to a body of ellipsoidal shape in a (orf(T)dT
£'r={£rs ~£r~)\-^ (2-332)
viscous fluid. If these bodies in the fluid are not
all identical in size, then their orientations will
(orf(r)dT
involve more than one relaxation time. Since K?n £r~)) l + (o2z 2~
molecules are generally ellipsoidal in shape, the tan o = — =
friction coefficients of the three axes are differ- £ +{e _£ x fmdr
/W*
( n rOo)
ent; hence, three different relaxation times ~ {\ + C02T
may exist. Many reasons exist for the relaxa- (2-333)
tion times to be distributed in solids, the most If r is distributed, then the peak value of £'
obvious being the presence of inhomogeneities. decreases and the spread width of e'r increases,
It is likely that not all dipoles in a solid are as shown in Figure 2-40.
situated in the same environment, so some are The area beneath the curve in Figure 2-40
more free to rotate than others. Even in a single can be expressed as
crystal, dipoles may find certain orientations
,dto
more favorable than others and certain transi- S= \e'rd{ln(o) = \e,
0)
tions between orientations easier than others. (2-334)
The variation of such local transition probabili-
ties reflects the variation of the activation energy i a i 1+oV
times makes £^max) decrease, Fuoss and

Kirkwood56 have proposed a parameter X with
0 < X < 1 to modify Equation 2-257 to take into
account the effect of the distribution of relax-
ation times empirically. They have changed
Equation 2-257 to the following form:
X(ers-eroo)
£'r =
(corf + (on)''' (2-338)
1
= — (£rs - eroo)Sech[X£n(cor)]
When cor = 1, the maximum value of the loss

€nft) = €n(//r 0 ) factor, termed £r(max)> occurs. It is
£-r(max) ^"y^rs ^r°°) (2-339)
Figure 2-40 The e'r - €n (O curves (A) involving only one
relaxation time, and (B) involving a set of relaxation times, Substitution of Equation 2-339 into Equation
each having its Debye-type loss peak. 2-338 gives the expression exactly the same as
Equation 2-288—the Fuoss-Kirkwood equa-
tion. When X = 1, Equation 2-338 returns to the
By changing the integration order, we have Debye equation in the case of only one relax-
ation time. So X < 1 implies that the relaxation
dco
S = (ers ^ ) J T / ( T > / T f — 2 ^ - (2-335) processes involve more than one relaxation
time. Thus, the smaller the value of A, the
Since higher the number of different relaxation times,
the wider the spread of e'r((D), and hence the
7 dco _ K
smaller the value of ^(maX). This implies that the
distribution function/(r) covers a wide range of
frequencies.
J/(T>/T = 1
To find/(r), we can rewrite Equation 2-332
to the form86
Equation 2-330 reduces to
^ f e - O l ^ (2-340)
S = -(ers-eroo) (2-336)
j=l
This equation implies that the area is not related If the summation is represented by a matrix M,
to the distribution function f(x). This means then £r = Mf(r), where the elements of the
that the area beneath the s'r - (nco curve for any matrix are
number of relaxation times is identical irre-
spective of the dispersion mechanisms of T and Ay — (£rs £roo) (2-341)
irrespective of/(r). The quantity T+coTcj.
So the distribution function f(f) can be
Aer =ers-e^=- f e'rd{lnco) = — (2-337) expressed in terms of the inverse matrix as
n K
i
/(T) = Af"1e; (2-342)
is called the relaxation strength, which is an
important physical parameter in relaxation If £' is known, f(f) may be determined.
processes. However, this method is suitable for the con-
Since/(T) is unknown, we cannot determine tinuous distribution of the relaxation times.
£r{(0) and s'Xco) from Equations 2-331 and 2- Great errors may result if the relaxation process
332. Because the distribution of relaxation involves only a few, discrete relaxation times.
If all dipoles follow Debye's relaxation In polymers, there are mainly two types of
model,/(T) may be expressed as dielectric relaxation: dipolar segmental relax-
ation and dipolar group relaxation. The dielec-
/(T) = -exp(-f/r) (2-343) tric relaxation processes of these two types
T exhibit loss peaks on the curves of the temper-
If the distribution of the relaxation times ature or frequency dependence of er. In amor-
assumes a Gaussian type of distribution, then phous polymers, the most intensive peaks of er
we have or tan 8 occur in the region of the transition
from the glassy to the rubbery (high elastic)
/ ( T ) = (b/7t2T)txV{-[b£n(T/t0)]2} (2-344)
state, i.e., near the glass transition temperature
where b is a constant and r0 is the central and Tg. Dielectric losses caused by dipolar segmen-
the most probable relaxation time. The empiri- tal relaxation are associated with the micro-
cal relation developed by Cole and Cole52 cor- Brownian motion of segments in polymeric
responds to the distribution function52'56 chains, while dielectric losses caused by dipolar
group relaxation are associated with the local-
ized movement of molecules. Several distinct
2[cos h(£nr/r0)+cos Xn\
dielectric relaxation processes are usually pre-
Based on Equation 2-289, we can write sent in solid polymeric materials. They can be
observed by means of the thermally stimulated
^=e4^=exra (2 346)
- relaxation technique.94 As the temperature is
raised, the mobility of various types of mole-
The distribution function for activation ener- cules becomes successively enhanced, making
gies for dipole orientation G(H) can be written it easier for dipolar orientation to occur.
as Usually, dielectric relaxation processes are
labeled by a, )3, 7 5, and so on, beginning at
G(H)d(AH) = f(z)d(£nr/z0) (2-347)
the high-temperature end.
This expression represents the fraction of Figure 2-41 illustrates schematically the a,
dipoles having activation energy for orientation j8, and 7 peaks on the curve of the loss factor
between H-d (AH)/2 and H + d (Atf)/2. er as a function of temperature at a fixed
A great deal of work has been carried out frequency. In this convention, the dipolar
in studying the distribution functions for the segmental relaxation corresponds to the a
relaxation times and their associated activation relaxation, and the dipolar group relaxation
energies. However, the molecular processes processes correspond to the j8, 7 <5, relaxations.
leading to such various distributions are still not The high-temperature a relaxation is mainly
well understood. We still rely on some empiri-
cal approaches. It is possible that a collective
and cooperative mechanism may be involved.
2.7.3 The Relation between Dielectric

Relaxation and Chemical Structure
This is a huge topic, and we shall not include a
detailed review of it in this chapter. There are
several excellent reviews dealing with this
topic, and the reader who is interested in more Temperature *•
details is referred to these reviews.26'39'87"94
Here, we shall briefly discuss the relation Figure 2-41 Schematic illustration of the loss factor e'r as
between dielectric relaxation and chemical a function of temperature showing the a, ft and 7 loss
structure, taking polymers as an example. peaks.
associated with the micro-Brownian motion phous polymers, or in amorphous regions of

of whole chain (segmental movement). The crystalline polymers, because crankshaft rota-
dielectric loss peak in the region of /J relaxation tion can only occur about two collinear bonds.
is mainly due to the movement of side groups This condition is not fulfilled in crystalline
or small units of main chains. The relative regions, where the sequences of methylene
strength of the a and /J dielectric relaxation groups form mainly transconformations. Relax-
depends on the degree of the orientation of ation processes at very low temperatures have
dipolar groups through the limited mobility a quantum nature.93
allowed by the /? process before the more diffi- To study the dielectric relaxation processes,
cult, but more extensive, mobility for the a apart from the measurement of the thermally
process comes into play. There is a partitioning stimulated relaxation as a function of tempera-
of the total dipolar alignment among the molec- ture, we need also the data from isothermal
ular rearrangement processes. scans of the dielectric constant er and the loss
The mechanisms for the p process may factor e'r as functions of frequency, so that the
depend on the nature of the dipolar group con- effective dipole moments and activation ener-
cerned and its position on the polymeric chain. gies may be obtained. Typical results from such
Among the most important mechanisms are the measurements for polyvinyl chloride in the
following91'92: a relaxation region are shown in Figure 2-42.
The results are from Ishida.97 Such plots are
• Relaxation of a side group about a C-C sometimes referred to as dielectric spectra.
chain, From a series of such plots, the relaxation times
• Conformational flip of a cyclic unit involv- can be determined for the individual relaxation
ing the transition from one chain form to processes as functions of temperature.
another, altering the orientation of a polar The a relaxation peak atT=Tg of an amor-
substituent, phous polymer is much narrower than the P
• Local motion of a segment of the main relaxation peak. Also, the temperature depend-
chain, since the small segment of a (CH2)„ ence of the a process is much steeper than that
chain can have such motion without involv- of the P process, indicating that a greater
ing the rest of the chain. thermal activation energy is required for the
motion. The a relaxation process near Tg is
At lower temperatures, there is a 7 relaxation largely dependent on free volume. Molecular
peak due mainly to the movement of small structure greatly affects the glass transition
kinetic units of the main or the side chains, temperature and hence the dielectric relaxation
which consist of a sequence of several carbon times. Thus, a bulky side group would cause the
atoms, such as the movement of several CH2 or Tg to decrease, thus preventing the chains from
CF2 groups. Another possible mechanism for packing together tightly. Tg can also be delib-
the /relaxation is due to the crankshaft rota- erately reduced by doping a plasticizer into the
tion95 below the glass transition temperature. polymer, such as diphenyl doped into polyvinyl
The /relaxation process has been observed in chloride, which affects the a relaxation process
polyethylene, aliphatic polyamides, polyester, because of the decrease of Tg.9S
and some polymethacrylates containing linear The effect of the degree of crystallinity on
methylene chains in side branches.95 The pre- the dielectric properties of polymers has been
sence in the polymeric chain of runs of three or extensively studied. The so-called crystalline
more CH2 units, each of which is linked to polymers always consist of many fully crys-
immobile groups, would lead to the crankshaft talline and fully amorphous regions. So, the
rotation, thus resulting in /relaxation in various overall dielectric relaxation processes may be
polymers.93'96 However, the available experi- regarded as a superposition of the relaxation
mental evidence indicates that /relaxation due processes in fully crystalline and in fully amor-
to crankshaft rotation is possible only in amorphous regions in the polymer. Usually, the
2 3 4 5 6 2 3 4 5 6
logf(Hz) log/(Hz)
(a) (b)
Figure 2-42 (a) The dielectric constant er and (b) the loss factor e'r as functions of frequency at temperatures in the a
relaxation region for polyvinyl chloride.
amorphous regions make the major contribu- space network formed by chemical bonds. They
tion to both er and e'r. As a result, er and e'r will are sometimes called the network polymers.
decrease with an increasing degree of crys- There are many ways to cause the formation
tallinity. Such a dependence of the dielectric of a cross-linked network. Polyethylene under-
properties on the degree of crystallinity is goes cross-linking under the action of ionizing
usually observed at temperatures higher than Tg radiation. Cross-linking can greatly restrict
of the amorphous regions. When the tempera- certain kinds of molecular movement and
ture is lowered further to the level at which r hence can affect the dielectric relaxation
becomes so large that cm » 1, then the dielec- processes. In general, dielectric losses will
tric constant becomes dominated by e^. Since diminish with an increasing degree of cross-
the value of e^ for a number of nonpolar (e.g., linking. This tendency appears more promi-
polyethylene and polytetrafluoroethylene) and nently in the region of the transition from the
polar (e.g., polyamids) polymers differs little, glassy to the rubbery state. Thus, as the degree
at very low temperatures, the dielectric con- of cross-linking increases, the magnitude of the
stants of these polymers are almost identical. principal a relaxation peak decreases.
Thus, when the relaxation spectra are frozen at Obviously, the incorporation of any impuri-
low temperatures, the difference between non- ties into the polymer in order to modify the
polar and polar polymers will disappear— dielectric properties will also affect the relax-
at least, as far as the value of er is concerned. ation processes and hence the values of er and
This phenomenon has been experimentally £'. Examples include plasticizer to lower the
observed.93 glass transition temperature and polar impuri-
Polymers may exist with linear, branched, ties to increase the dielectric relaxation for non-
and cross-linked chain structures. Linear poly- polar polymers.
mers usually have long chains composed of a
large number of repeating units, such as poly- References
ethylene and polytetrafluoroethylene. Branched
1. L. Brillouin, Wave Propagation in Periodic
polymers have side chains attached to the main Structures, (Dover, New York 1953).
chain, such as polymethyl methacrylate. Cross- 2. J. A. Stratton, Electromagnetic Theory,
linked polymers have a three-dimensional (McGraw-Hill, New York, 1941), p. 113.
3. K.C. Kao, Brit. J. Appl. Phys., 12, 629 27. A.R. Von Hippel, Dielectrics and Waves,
(1961). (John Wiley and Sons, New York, 1954).
4. W.R. Smythe, Static and Dynamic Electric- 28. H. Frohlich, Theory of Dielectrics,
ity, (McGraw-Hill, New York, 1950). (Clarendon, Oxford, 1958).
5. J.H. Jeans, The Mathematical Theory 29. P. Langevin, J. Physique, 4, 678 (1905);
of Electricity and Magnetism, (Cambridge Ann. Chim. Phys., 5, 70 (1905).
University Press, Cambridge, 1925). 30. P. Debye, Physik Zeit., 13, 97 (1912).
6. C.J.F. Bottcher, Theory of Electric Polar- 31. P. Debye, Polar Molecules, (Dover, New
ization, (Elsevier, Amsterdam, 1978). York, 1945).
7. R. Cooper and A.A. Wallace, proc. Phys. 32. C.P. Smyth and C.S. Hitchcock, J. Amer.
Soc, B66, 1113(1953). Chem. Soc, 55, 1290 (1933).
8. K.H. Stark and C.G. Garton, Nature, 176, 33. C.P. Smyth and C.S. Hitchcock, J. Amer.
1225 (1955). Chem. Soc, 56, 1084 (1934).
9. K.C. Kao, "The Bubble Theory for Electri- 34. N.E. Hill, "Theoretical Treatment of Per-
cal Breakdown of Liquid Dielectrics," mittivity and Loss" in Dielectric Properties
A.I.E.E. Conference Paper No. CP 60-84 and Molecular Behaviour, Ed. N.E. Hill,
(American Institute of Electrical Engineers, W.E. Vaughan, A.H. Price, and M. Davies
New York, January 1960). (Van Norstrand Reinhold, London, 1969).
10. K.C. Kao, Nature, 208, 279 (1965). 35. J.C. Slater, Insulators, Semiconductors
11. C.G. Garton and Z. Krasucki, Proc. Royal and Metals, (McGraw-Hill, New York,
Soc. (London), 208A, 111 (1964). 1955).
12. K.C. Kao and J.B. Higham, J. Electrochem. 36. EC. Nix and W. Shockley, Rev. Mod. Phys.,
Soc, 108, 522 (1961). 10, 1 (1938).
13. H.A. Pohl, Dielectrophoresis, (Cambridge 37. N.F. Mott and E.A. Davis, Electronic
University Press, Cambridge, (1978), and Processes in Non-Crystalline Materials,
J. Appl. Phys., 29, 1182(1958). (Clarendon, Oxford, 1974).
14. R.H. Cole, "Theories of Dielectric Polariza- 38. L. Onsager, J. Am. Chem. Soc, 58, 1486
tion and Relaxation", in Progress in (1936).
Dielectrics, Vol. 3, (Heywood, London, 39. N.G. McCrum, B.E. Read, and G. Williams,
1961) pp. 47-100. Anelastic and Dielectric Effects in Poly-
15. R.E. Teachout and R.T. Pack, Atomic Data, meric Solids, (John Wiley and Sons, New
3, 195 (1971). York, 1967).
16. A.K. Jonscher, Dielectric Relaxation in 40. J.G. Kirkwood, J. Chem. Phys., 7, 911
Solids, (Chelsea Dielectric Press, London, (1939).
1983). 41. F.E. Harris and B.J. Adler, J. Chem. Phys.,
17. R. Coelho, Physics of Dielectrics, 21f 1031 (1952).
(Elsevier, Amsterdam, 1979). 42. H. Frohlich, Physica, 22, 898 (1956).
18. J.H. van Vleck, Theory of Electric and Mag- 43. J.J. O'Dwyer, J. Chem. Phys., 26, 878
netic Polarizabilities, (Oxford Univeristy (1957).
Press, Oxford, 1932), p. 203. 44. B.K.P. Scaife, Proc. Phys. Soc. (London),
19. N.F. Mott and I.N. Sneddon, Wave Mechan- B70, 314 (1957).
ics and Its Applications, (Clarendon, 45. A.D. Buckingham, Proc Roy. Soc.
Oxford, 1948), p. 166. (London), A238, 235 (1956); also, Trans.
20. A.J. Dekker, Electrical Engineering Faraday Soc, 52, 1035 and 1551 (1956).
Materials, (Prentice Hall, Englewood Cliffs, 46. R.H. Cole, J. Chem. Phys., 27, 33 (1957).
NJ, 1961). 47. G.P. Mikhailov, in Physics of Non-
21. A.J. Dekker, Solid State Physics, (MacMil- Crystalline Solids, (North Holland,
lan, London, 1963). Amsterdam, 1965), p. 270.
22. I. Pirenne and E. Kartheuser, Physica, 20, 48. H.A. Kramers, Atti del Congresso Inter-
2005 (1964). nazionali dei Fisici, Como 2, 545 (1927),
23. M. Born and K. Huang, Dynamical Theory Collected Set Papers, (North Holland,
of Crystal Lattices, (Clarendon, Oxford, Amsterdam, 1956), p. 333.
1954). 49. R. de L. Kronig, J. Optical Society of
24. B.K.P. Scaife, Principles of Dielectrics, America, 12, 547 (1926).
(Clarendon, Oxford, 1984). 50. P. Debye and W. Ramm, Annalen der
25. R.J.W. LeFevre, Dipole Moments, Physik, 28, 28 (1937).
(Methuen, London, 1953). 51. P. Debye, The Collected Papers of Peter
26. C.P. Smyth, Dielectric Behavior and Struc- J. W. Debye, (Wiley Interscience, New York,
ture, (McGraw-Hill, New York, 1955). 1954).
52. R.H. Cole and K.S. Cole, J. Chem. Phys., 9, 76. A. Anselm, Acta Physicochim, U.S.S.R., 19,
341 (1941); ibid., 10, 98 (1942). 400 (1944).
53. D.W. Davidson and R.H. Cole, J. Chem. 77. J.B. Hasted, D.M. Ritson, and C.H. Collie,
Phys., 18, 1417 (1950); ibid., 19, 1484 J. Chem. Phys., 16, 11 (1948).
(1951). 78. S. Lielich, in Dielectric and Related
54. S. Havriliak and S. Negami, J. Polymer Sci., Molecular Processes, Vol. 1, (The Chemical
Pt. C. 14, 99 (1966); also, Polymers, 8, 161 Society, Burlington House, London, 1972),
(1967). (1972).
55. A.M. Rad and P. Bordewijk, Rec. Trav. 79. G.P. Jones, ibid., Vol. 2 (1975).
Chim., 90, 1055 (1971). 80. G. Williams, Polymers, 4, 27 (1963).
56. R.M. Fuoss and J.G. Kirkwood, J. Amer. 81. M.F. Manning and M.E. Bell, Rev. Mod.
Chem. Soc, 63, 385 (1941). Phys., 12, 215 (1940).
57. R.H. Cole and P.M. Cross, Rev. Sci. Instru., 82. B.V Hamon, Proc. IEE, (London), 99, 151
20, 252 (1949). (1952), and also Austral. J. Phys., 6, 304
58. R.H. Cole, J. Chem. Phy., 23, 493 (1955). (1953).
59. B.E. Read and G. Williams, Polmyers, 2, 83. G. Williams, Trans. Faraday Soc, 58, 1041
239 (1961); also, Trans. Faraday Soc, 57, (1962); ibid., 59, 1397 (1963).
1979 (1961). 84. P.J. Hyde, Proc. IEE. (London), 117, 1891
60. A. Schallamach, Trans. Faraday Soc, 42A, (1970).
495 (1946). 85. A. Suggett, in Dielectric and Related Mole-
61. C. Herring, Bell System Tech, J., 34, 237 cular Processes, Vol. 1 (Chemical Society,
(1955). London, 1972), p. 100.
62. E.M. Conwell, High Field Transport in 86. D.R. Uhlmann and R.M. Hakim, J. Phys.
Semiconductors, (Academic Press, New Chem. Solids, 32, 2652 (1971).
York, 1967). 87. N.E. Hill, W.E. Vaughan, A.H. Price, and
63. E.M. Conwell, V.J. Fowler, and J. Zucker, M. Davies (Eds.), Dielectric Properties and
Report on Contract AF19 (604)-5714 (Air Molecular Behaviour, (Van Nostrand
Force Cambridge Res. Center, Cambridge, Reinhold, London, 1969).
Massachusetts, 1960). 88. A.M. North, Chem. Soc. Rev., 1,49 (1972).
64. A.F. Gibson, J.W. Granville, and E.G.S. 89. P. Hedvig, Dielectric Spectroscopy of Poly-
Paige, J. Phys. Chem. Solids, 19, 198 mers, (Adam Hilger, Bristol, 1977).
(1961). 90. P.E. Karasz (Ed.), Dielectric Properties of
65. S. Triebwasser, IBM J. Res. Develop., 2, Polymers, (Plenum, New York, 1972).
212 (1958); also, Phys. Rev., 118, 100 91. A.R. Blythe, Electrical Properties of
(1960). Polymers, (Cambridge Universtiy Press,
66. H. Baumpartner, Helv. Phys. Acta, 23, 651 Cambridge, 1979).
(1950); ibid., 24, 326(1951). 92. I.I. Perepechko, An Introduction to Polymer
67. E. Fatusso and W.J. Merz, Ferroelectricity, Physics, (Mir Publisher, Moscow, 1981),
(North Holland, Amsterdam, 1967). English Translation by A. Beknazarov).
68. S. Ratnowsky, Vorhandl deut Physik. Ges., 93. I.I. Perepechko, Low Temperature Proper-
75, 497 (1913). ties of Polymers, (Mir Publisher, Moscow,
69. J. Herweg, Z. Physik, 3, 36 (1920). and Pergamon Press, Oxford, 1980).
70. J. Herweg and W. Potzsch, Z. Physik, 8, 1 94. P. Braunlich, Thermally Stimulated Relax-
(1922). ation in Solids, (Springer-Verlag, Berlin,
71. J. Malsch, Physik Z., 29, 710 (1928). 1979).
72. G.P. Jones and M. Gregson, Total 95. T.F. Schatzki, J. Poly. Sci., 57, 496 (1962);
Dielectric Saturation Observed in a Dipolar also, Meeting of Am. Chem. Soc. Div.
Solution, Chem. Phys. Lett., 4, 13 (1969). Polymer Chem., Atlantic City, September
73. J.H. van Vleck, J. Chem. Phys., 5, 556 1965, Polymer Preprints, 6, 646 (1965).
(1937). 96. B.J. Wunderlich, J. Chem. Phys., 7, 10 and
74. A.D. Buckingham, J. Chem. Phys., 25, 428 2429 (1962).
(1956). 97. Y. Ishida, Kolloid. Z., 168, 29 (1960).
75. J.J. O'Dwyer, Proc Phys. Soc. (London), 98. R.M. Fuoss, J. Am. Chem. Soc, 63, 378
A64, 1125 (1951). (1941).

Kao 2004

Uploaded by

Copyright:

Available Formats

Kao 2004

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

Kao 2004

Uploaded by

Copyright:

Available Formats

2 Electric Polarization and

2.1 Fundamental Concepts F is determined by the magnitudes and the

since r is much larger than 5.

+Q increases with t, the time required to travel a

the region between the plates if the edge effect (2-15)

when AU is positive, and increased when AU

Sphere with e2 > £1 Sphere with % < £ i

Spheroid with field parallel Spheroid with field normal

3ex{e2~£x) When the sphere is slightly deformed by

Consider a nonuniform field, such as that pro-

The elongation phenomenon was first predicted 3g1(g2-g1) 2tf_ Q2

The Dielectrophoretic Force

panied by a slight decrease in volume of the dO

Therefore, the hydrostatic pressure tending to

d(AU)v (e-e0)(e + 2e0) + log

Bound Charge Free Charge

charges on the plates will create a potential dif- +Q

where F is the electric field strength, which is

Q osA A where F0 is the electric field strength in the

2.2.2 Conducting Materials

charge. This creates a net negative space charge

3. Dipolar polarization Figure 2-7 The surface charge of density Gs consisting of

(u) = (qdS) = aFioc (2-64) It should be noted that this discussion is

Polarization due to (a)

reverse depolarization current will follow, due or

The dipole moment is given by

However, there must be some loss or damping (47T£0)h2

N (Zqf Single atoms in gases may be considered

= -•£[«// -(«// -<*±)sin20]F2 where N denotes the total number of atoms

Quantum Mechanical Approach an isolated atom, may be applicable to gases

Mi dt2 = 7i[(X2n ~ X2n-l ) ~ a]

about 3 x 10~8cm. So we can say that ka is (Zq)2Floc

Figure 2-15 The orientation of a dipole: (a) stable posi-

2.5 3.0 3.5

Figure 2-14 The susceptibility (er - 1) as a function of

Figure 2-16 The calculation of the number of dipoles

\u0 cos OdN L(z) = (cos0) = 1-1/z (2-132)

The term exp(-2u0F/kT) can be expanded into

is 1 - w, which is governed by the following A qualitative picture of U{T) as a function of

electric field, all of these randomly arranged

2.3.5 Space Charge Polarization

P{A^B) = Cexp[-(£A + aAE')/kT] (2-145) If qrF/kT « 1, p' can be approximated to

P(B^A) = Cexp{~[EA -(\-a)&E']lkT} the random hopping dipole moment is

Interfacial Polarization r = (elrd2 + e2rdx )e0 l(oxd2 + o2dx) (2-166)

Y2=(a2 + jcQe2r£o)/d2 (2-162) and

CToo = Gs (TiT 2 + TiT + T 2 r ) / T 2 (2-173)

(2-163) We can consider e^ to be mainly the contribu-

1 P' = (£*» + £ro - 1)£0F = Nia^ + a0 )F

Pc = P-P' = (£rs -£r„-£ro)£oF

and the space charge polarizability is dielectric constant should be independent of

<a> (+) m o fflffl (+)

N0= — = 6.023 x 1023 per mole (2-190)

fields in w h i c h orientational p o l a r i z a t i o n c a n n o t cavity of radius a, it produces a field that polar-

n2-l Na Ueff = U0 + OCFr (2-196)

\t>rs c-roo/ Na0 3er

w h e r e ers d e n o t e s t h e static dielectric c o n s t a n t . T h u s , t h e effective i n d u c e d d i p o l e m o m e n t of a

In Section 2.5.1, we derived an expression for ueff=a(Fi + fu'eff)

Fj=hF 4>ff and u^

The total dipole moment in the cavity is the «o | n , ah