CHAPTER 1

PART - A

Ferroelectrics are materials that display an electric polarization in

the absence of an external applied electric field, together with the property
that the direction of the polarization may be reversed by an electric field.
The study of the properties of ferroelectric materials and the attempt to
understand the nature of the ferroclectric state constitute the field of
ferroelectricity.

Any macroscopic collection of matter is, to a very high degree of

accuracy, electrically neutral, there being very nearly equal numbers of
positive and negative charge, tlowever, the positive and negative charges
are not necessarily distributed in a homogeneous or symmetric manner. If
the "centers of gravity" of the positive and negative charge distributions
within a volume of material do not coincide, the material is said to have an
electric dipole moment. The net dipole moment per unit volume is defined
as the spontaneous polarization [I].

1.1 General properties of ferroelectrics

1.1.1 Polarization
Polarization (P) is the electric dipole moment per unit volume, and
is related to dielectric displacement D through the linear expression
D = P + E ~ Ewhere
, e0 (usually called as the permitivity of Free space)
equals 8.854 i10-I2 Coulomb/volt. In ferroelectric materials both D and P
are non-linear functions of E and can depend on the previous history of the
material. When the term cOE in the above expression is negligible
compared to P (as in the case of most ferroelectric materials), D is nearly
equal to P. The different types of polarization are

Electronic : a displacement of the electronic cloud with respect to

the nucleus.
Ionic : separation of positive and negative ions in the crystal.
Orientational : alignment of permanent dipoles (molecules).
Space-charge : free electrons are present, but are prevented from
moving by barriers such as grain boundaries.

1.1.2 Spontaneous polarization

Spontaneous polarization (P,) is defined as the magnitude of
polarization within a single ferroelectric domain in the absence of an
external electric field. Spontaneous polarization is a fundamental property
of all pyroelectric crystals, but it is reversible and reorientationable in
ferroelectrics only. Most ferroelectric phase originate from a non-polar
prototypic phase and all of the polarization is reorientationable. However,
if the prototypic phase is polar, only a proportion of the total spontaneous
polarization may be reoriented, the reorientationable or reversible process
is commonly called as the spontaneous polarization. The magnitude of Ps
in a single crystal is directly related to the atomic displacements that occur
in ferroelectric reversal and may be calculated from the atomic positions
within the unit cell if known. Designating Ai as the component of the
atomic displacement vectors joining the ith atom positions in the original
and reversed orientations along the direction of P,, Zi the effective charge
on the ith ion and V as the unit cell volume, then P,=(IRV)AiZi. The
spontaneous polarization of single-domain materials usually lie within the
range 10"and 10~Coulomblcm~~
[2]. The spontaneous polarization can be
calculated directly from charge density obtainable by X-ray diffraction
structural measurements, corrected for the charge transferred across unit
cell boundaries.

1.1.3 Ferroelectric domain

A ferroelectric crystal consists of regions of homogeneous
polarization that differ only in the direction of polarization. These regions

Fig. 1.1. Domain configurations (simplified) encountered in

ferroelectric crystals (a) First class (b) Second class.

are called ferroelectric domains. Ferroelectrics of the first class consists of

domains with parallel and antiparallel polarization as shown in Fig. I. l (a),
whereas ferroelectrics of the second class can assume much more
complicated domain configurations as shown in Fig, l.t(b). The region
between two adjacent domain is called domain wall. Within this wall, the
spontaneous polarization changes its direction. The wall between
antiparallel domains is probably only a few lattice spacing thick, whereas
the wall between domains polarized at right angle to each other is probably
thicker [3].

1.1.4 Ferroelectric hysteresis

Using the domain concept, the occurrence of hysteresis in the
polarization 'P' versus field 'E' relationship can be explained as follows.
Considering a crystal, which initially has an overall polarization equal to
zero, i.e., the sum of the vectors representing the dipole moments of the
individual domain vanishes. When an electric field is applied to the
crystal, the domains with polarization components along the applied field
direction grow at the expense of the "antiparallel" domains; thus the
polarization increases. When all domains are aligned in the direction of
the applied field the polarization saturates and the crystal becomes a single
domain. A further increase in the polarization with increasing applied field
results from "normal polarization" effects; rotation of domain vectors may
also be involved if the external field does not coincide with one of the
possible directions of spontaneous polarization. The extrapolation of the
linear part AB (saturation value of polarization) to zero external field as
shown in Fig. 1.2, gives the spontaneous polarization Ps (since in the
relation between electric displacement, field and polarization
D = P + E ~ EIn) the above relation the polarization arises from both the

polarization of the material in the presence of a field PE = XE and from the

spontaneous alignment of dipoles in the ferroelectric Ps. The spontaneous
polarization and its dependence on temperature or on other external
conditions that might be imposed can be measured by displaying the
hysteresis loop on an oscilloscope screen. When the applied field for a
crystal corresponding to point B in Fig. 1.2 is reduced, the polarization of
the crystal decreases, but for zero applied field there remains the remanent
polarization (Pr) where Pr refers to the crystal as a whole.

Fig.l.2. A hysteresis loop illustrating the coercive field Ec, the

spontaneous polarization Ps, and the remanent polarization Pr.

In order to remove the remanent polarization (P,), the polarization of the

crystal must be reversed and this occurs only when a field in the opposite
direction is applied. The field required to make the polarization zero is
called the coercive field Ec [4].
1.1.5 Phase transition
The ferroelectric properties of a ferroelectric disappear above a
critical temperature Tc; this temperature is called the ferroelectric Curie
temperature diagrams for a fernelectric has one or more very sharp
maxima where E can reach values of several thousands. The temperatures
at which these maxima occur are called Curie temperatures. Above the
Curie temperature E obeys the Curie Weiss law
E = C I ( T - 8 ) + cO (1.1)
where C is a constant and 0 is a characteristic teniperature which is usually
some degrees smaller than the transition temperature Tc, E, is a constant
contributed by the electronic polarization. In the vicinity of the transition
temperature, EO may be neglected since it is of the order of unity and

1.Z A simple-model Hamiltonian

The value of simple models is well established in solid-state physics.
From a theoretical point of view it is of great value to deduce the simplest
Hamiltonian which appears to embody the essential features of a co-
operative phenomenon and to study its statistical behavior. In such a way
the basic properties of an idealistic system can be probed in considerable
details using relatively sophisticated techniques, and subsequent
experimental deviations from this fundamental behavior can be interpreted
in terms of the deviations of a Hamiltonian from the model. The basic-
model Hamiltonian which describes structural phase transitions (and
ferroelectrics in particular) takes advantage of the fact that such transitions
are often associated with the rearrangement of only a few atoms in the unit
cell, whereas the positions of others are relatively undisturbed. Thus for a
simple model it is essential to take account only of these particular co-
ordinates while treating the rest of the crystal lattice as a heat bath. The
basic Hamiltonian describing a solid is conventionally written as
H = H (ion) + H (electron) + H (electron-ion) (1.2)
where H (ion) describes a collection of ions interacting via a potential
which depends only on the positions Ri, Rj,....., of the ion centres, H
(electron) describes the valence electron motion, and tl (electron - ion) is a
suitably chosen potential which represents interactions between valence
electrons and ion cores. In this approximation the electrons are treated as
if they respond so quickly to the motion of the ions that their state is
always just a function of the ionic co-ordinates. H (electron-ion) can be
written as a potential energy contribution E (Ri, Ri, ...) to an effective
Hamiltonian for ionic motion alone in the form
Heff(ion)= Z (p2i 12mi) + U (Ri, R,,...) + E(Rj. Rj,...) (1.3)
where, the first and second terms are respectively the kinetic and
potential energy of the lattice of ion cores themselves. The next
approximation is usually to assume E(Rj, Rj, ...) to be independent of
electron configuration or, in the subsequent statistical calculation of
temperature. In this way the potentials U and E can be combined into a
resultant effective ion-ion potential V(Ri, Rj,...). This is in fact quite a
good approximation even for narrow-band-gap materials since, by the
exclusion principle, most valence electrons are always in the same state
even when thermal electronic excitations occur. On the other hand the
effects of valence to conduction band thermal excitations are by no means
negligible when band gaps are small. Neglecting any temperature
dependence of E, effective ionic Harniltonian is given by
 Heff(ion) =Z (p2 /2mi ) + V(Rj, Rj,...) (1.4)
The potential V in eqn. (1.4) is an effective ionic potential. It is
nevertheless the sum of two distinct parts, one of which is physically of
electronic origin. Thus Hamiltonian eqn. (1.4) does not describe the
motion of rigid ions but, to use the shell-model includes a relative motion
of rigid core and valence-shell electrons for each ion. This distortion has
traditionally been described as resulting from the electronic polarizability
of ions in the local field of their neighbours. According to the adiabatic
principle electronic and ionic energies are essentially independent. Thus
from eqn. (1.4) any lattice-driven instabilities responsible for structural
transitions can be found out. The companion effective electronic
Hamiltonian involving H (electron), and dependent on instantaneous ion
configuration via H (electron-ion), is not to be forgotten. Phase transitions
in crystal lattices usually involve some particular type of co-ordinates.
This is perhaps best illustrated by examples: the rotations of BO6
octahedra in antidistortive perovskite (ABO3) transitions, the displacement
of B with respect to the 0 6 oxygen cage in ferroelectric perovskite
transitions, the coupled proton-lattice motion in hydrogen-bonded
ferroelectrics, etc.. Very often a single 'local mode' of appropriate
symmetry is sufficient for an adequate description, and accordingly local
canonically conjugate momentum and displacement co-ordinates for each
unit cell can b e defined.
The non-kinetic part of the effective ionic Hamiltonian is most
conveniently expressed in terms of the displacements qlb of the hth ion in
the Ith unit cell from a fixed reference configuration representing the
averaged positions of the ions in the high-temperature phase. 'Local-
mode' generalized momentum xl and displacement 61 variables can be
defined in terms of the 'local-ion' momentum plb and displacement qlb co-
ordinates by the equations
Plb = mb 'Jlb 'Il, 91b = Ulb 51 (1.5)
XI = Z ulb Pibl 61 = 2 mb"lb~qlh (1.6)

where Cb runs over all ions in the lth cell, mh is the mass of the bth

ion, and (mb)li2 ulb is the bth component of the normalized local
eigenvector and satisfies
Z mi, ulb ulb= I (1.7)
Thus 61 is the scalar amplitude of that local motion which broadly
characterizes the phase transition. If this mode should be degenerate then it
becomes necessary to define 61 as a vector of appropriate dimension. The
dynamics of the ionic system (eqn. 1.4) is now described in the local-mode
approximation by the model Hamiltonian

H =Z x2[ + V (51, 52, ....,SN) (1.8)

2
in which N labels the number of unit lattice cells in the macroscopic
crystal. The potential V of eqn. (1.8) can now be decomposed into a sum
of single-cell contributions V(c1) and an intercell interaction part. The

latter can be written as a sum of bilinear two-body interactions ulll51511.

The final form of the model Hamiltonian is therefore

H =x {id
I
+V(~I))-;Z
I
Z ~II~FI~I
I'
 In this model the local potential function V(5) can be anything from
quasi-harmonic to deep double-well form, and the interaction potential ulll
can be short range and/or long range in character. With this flexibility the
Hamiltonian eqn. (1.8) can be used to describe in a fairly realistic way a
whole range of polar and non-polar displacive, tunnel-mode, and order-
disorder transitions [ 5 ] .

1.3 Phenomenological theory for ferroelectrics

The inclusion of non - linear terms in the thermodynamic equations of
state allows for a description of possible divergencies in response. With
the independent variables used to describe thc simple elastic dielectric this
divergence can be with respect to elastic response (ferroelasticity) or to
dielectric response (ferroelectricity). Linear coupling to other properties
(e.g. Elastic, thermal, other dielectric, or even magnetic) will. of course,
produce associated anomalies in other responses but. at least in principle,
one can usually examine the theoretical consequences of removing such
couplings. In thermodynamic theory one stops inquiring as to the origin
of the driving mechanism, but merely identifies the primary pair of
conjugate variables (D, and E, for ferroelectrics) and defines lowest-order
non-linearities in the equation of state involving them. Results are
conventionally obtained using the assumption that the relevant equation of
state can be represented by a polynomial and that the same polynomial can
be used above the phase transition anomaly.
An equivalent and more usual starting point is a polynomial
thermodynamic potential expressed in terms of deviations from the
prototype state D, = x, = 0. Normally the most convenient potential is the
elastic Gibbs function GI expressed as a function of temperature, stress and
displacement. The simplest conceivable theoretical situation results with
the assumption that D, = D is directed along one of the crystallographic
axis only (i.e. spontaneous polarization occurs along this direction and the
applied field is restricted to this direction also), that all stresses are zero
and that the non - polar phase is centrosymmet&. None of these
restrictions is essential, but with them the free energy GI can be expressed
in the very simple polynomial form
Gi = ( a 1 2 ) ~+ ~(y14)4+ ((616) D' (1.10)
Where energy is measured from the non-polar phase and the polynomial
can be terminated rather arbitrarily at D~ for mathematical simplicity. The
coefficients a, y, (6 are in general temperature dependent, but simple
examples of first-and second-order ferroelectric transitions can be
described assuming y and 6 independent of temperature. With this
assumption the order of the transition depends on the sign of y ((6 being
necessarily positive in eqn. (1.10) for stability reasons as D -+ m ) [5].

1.4 Applications of ferroelectric materials

1.4.1 Applications of ferroelectric ceramics
A capacitor consists of a dielectric material sandwiched between two
electrodes. The total capacitance for this device is given by
C =(E,E,A)I t (1.11)
Where ' C' is the capacitance, E, is the permittivity of free space, &,is
the relative dielectric permittivity, 't' is the distance between the
electrodes, and 'A' is the area of the electrodes. To get a high volumetric
efficiency (capacitance per unit volume) the dielectric material between the
electrodes would have a large dielectric constant, a large area and a small
thickness. BaTiO, based ceramics having a perovskite type structure
show dielectric constant values as high as 15,000 as compared to 5 or 10
for common ceramic and polymer materials. The high dielectric constant
BaTiO3 ceramic based disk capacitors are simple to make and have
captured more than 50% of the ceramic capacitor markpt.
The volumetric efficiency can be further enhanced by using
multilayer ceramic (MLC) capacitors. Each individual dielectric layer
contributes capacitance to the MLC capacitor as electrodes terminate in a
parallel configuration. Hence the effective equation for capacitance
becomes,
C = (n co~,A)It (1.12)
Where n is the number of dielectric layers. The advances in tape casting
technology have made it possible to make dielectric layers, 20pm thick.
This, combined with the use of a high dielectric constant ceramic like
BaTiO,, allows large capacitance values to be achieved in relatively small
volume capacitor devices.
MLC capacitors are made by the tape casting process. A slurry with
a suitable binderisolvent system is first made from the dielectric ceramic
powder. Thin green sheets of the ceramic are then made by the tape
casting process. An ink consisting of an electrode (Pd, Ag-PD, etc.) and
organics isscreen printed on the dielectric sheets and then hundreds of
sheets are stacked one on top of the other. A low pressure at a
temperature between - 50 and 70°C is applied to laminate the sheets. The
binder removal is accomplished by heating the green body very slowly to a
temperature of - 300-400' C. The MLC capacitor is then sintered at a
high temperature depending on the type of ceramic. After applying the
terminations for the internal electrodes of the MLC capacitors, the
capacitor is mounted on the electrode substrate by soldering.
The critical steps in the fabrication of the MLC capacitors include
the formation of thin uniform green sheets from the ceramic slurry. Any
non - uniformity in the thickness could lead to dielectEjc breakdown during
operation of the device. The co-firing of the ceramic tapes and metal
electrodes should be compatible and should not lead to any reactions
between the two. The high dielectric constants of the relaxors materials
make them ideal for MLC capacitor applications. Relaxors have various
advantages over BaTiOl based materials including higher dielectric
constants (E, ~25,000,a maximum E, of =15,000 for BaTiO, based
ceramics). In addition, the temperature dependence of permittivity is much
low for relaxor materials, with no microstructure control needed for getting
a shallow peak as in BaTiO,. Much finer grain sizes (Slpm)can be used
for relaxor ferroelectrics as compared to BaTiO, (2-5 pm ) to get high
dielectric constant with low temperature dependence. The electric field
dependence of the dielectric properties is also lower for relaxor dielectrics.
In spite of various advantages of relaxor materials over BaTi03 based
dielectrics, they have some inherent problems which have to be overcome
before large scale production takes place. The dielectric losses for relaxors
are the highest just the Curie point. The permittivity is also is also highly
frequency dependent in this temperature range. The processing of PbO
based relaxor powders leads to the formation of a low dielectric constant
pyrochlore phase, leading to a decrease in the relative permittivity of the
dielectric. Also the relaxor ceramics are mechanically weak and it is
difficult to reproduce the same properties from batch to batch, and control
of the lead rich atmosphere is needed for proper sintering of the relaxor
material [ 6 ] .

1.4.2 Applications of ferroelectric thin films

Ferroelectric thin films have attracted attention b r applications in
many electronic and electro - optic devices. Some of the important
ferroelectric materials being used for making thin films include the
perovskite type materials such as BaTiO,, PbTiOj, Pb (ZrxTil.x)03,and Pb
(ZrxTil.,)03. Some of the non-perovskite materials being include BiTij012
and (Pb,Ba)NbzOa, Applications of ferroelectric thin films utilize the
unique dielectric, piezoelectric, pyroelectric sand electro- optic properties
of ferroelectric materials. Some of the important electronic applications
of ferroelectric thin films include nonvolatile memories, thin film
capacitors, pyroelectric sensors, and surface acoustic wave (SAW)
substrates. The electro-optic devices include optical waveguides and
optical memories and displays.

1.4.3 Ferroelectric memories

Semiconductor memories such as dynamic random access memories
(DRAMS) and static random access memories (SRAMs) currently
dominate the market. However, the disadvantage of these memories is that
they are volatile, i.e. the stored information is lost when the power fails.
The non-volatile memories available at this include complementary metal
oxide semiconductors (CMOS) with battery backup and electrically
erasable read only memories (EEROMs). These non-volatile memories are
very expensive. The main advantages offered by ferroelectric random
access memories (FeRAMs) include non-volatile and radiation hardened
compatibility with CMOS and GaAs circuitry, high speed (3011s cycle time
for read /eraselrewrite) and high density (4pm)2 cell size) [7,8].
The fabrication of ferroelectric thin films is done by three methods
including physical vapour deposition (PVD), chemical vapor deposition
(CVD) and sol-gel processing. In PVD,precursors +the desired film
composition are vaporized and deposited on the substrate by one of the
sputtering techniques (i.e, r.f magnetron sputtering, pulsed vapor
deposition (PLD) etc,) methods, molecular beam epitaxy (MBE)and laser
ablation [6,9].

1.4.4 Electro-optic applications

The requirements for using ferroelectrlc thin films for electro-optic
application include an optically transparent film with a high degree of
crystallinity. The electro-optic thin film devices are of two types; one in
which the propagation of light is along the plane of the film (optical
waveguides) and the other in which the light passes through the film
(optical memory and displays).

1.4.5 Ferroelectric thin film waveguides

An optical waveguide controls the propagation of light in a
transparent material (ferroelectric thin film) along a certain path. For the
waveguide to work properly, the refractive index of the film should be
higher than that of the substance. For light to propagate in the waveguide,
the thin film should be optically transparent. This can be achieved by
fabricating the film under clean conditions and aiming for a fine grain size
with ultra-phase purity and high density. A great deal of work has been
done on making ferroelectric thin film waveguides from LiNbO3 and Li
(Nb,Ta,)3 using LPE and MBE methods. Lead Zirconate Titanate (PZT)
and Lead Lanthanum Zirconate Titanate (PLZT) thin films are even better
candidates for optical waveguide applications because of their large
electro-optic coefficients [lo, 1I].

1.4.6 Ferroelectric thin film optical memory displays

Ferroelectric thin films may replace the use of PLZT bulk ceramics
for optical memory and display applications. The advantages offered by
thin films for display applications include a simplification of the display
device design and lower operating voltages as compared to PLZT ceramic
devices. Optical memories using PLZT thin films will also need lower
operating voltages. The material requirements for thin film optical
memory and displays include large electrwptic coefficients andlor strong
photosensitivities for the film. PZT and PLZT thin films show a lot of
promise for these optical applications[6].

1.4.7 High- capacity memories

The use of optical techniques for both recording and read-out offers
the capability of high-capacity, random-access memories without serious
interconnection problems. The ideal storage medium allows insitu erasure
information (for updating the memory), a resolution of about 1000
linesimm, long storage time, high recording sensitivity, high signal to noise
ratio and linear exposure characteristics. Holographic storage with a
resolution of 500 linesimm has been accomplished with BirTi30,e
ferroelectric-photoconductor sandwich structures.
 PART - B
1.5 Structural classification of ferroelectric materials
The types of ferroelectric materials, discussed here, have been
grouped according to their structure. The four main types of structures; the
comer sharing oxygen octahedra, compounds containing hydrogen bonded
radicals, organic polymers and ceramic polymer composites.

1.5.1 Corner sharing octahedra

A large class of ferroelectric crystals are made up of mixed oxides
containing corner sharing octahedra of 02.ions, Inside each octahedron is
a cation B ~ 'where 'b' varies from 3 to 6 . The space between the octahedra
are occupied by Ad+ions where 'a' varies from 1 to 3. In prototypic forms,
the geometric centers of the A ", B bi and 02-ions coincide, giving rise to a
non-polar lattice. When polarized, the A and B ions are displaced from
their geometric centers with respect to the 02.ions to give a net polarity to
the lattice. These displacements occur due to the changes in the lattice
structure when phase transitions take place as the temperature is changed.
The formation of dipoles by the displacement of ions will not lead to
spontaneous polarization if there is a compensation of dipoles leading to
net zero dipole moment. The comer sharing oxygen octahedra includes the
perovskite type compounds, tungsten bronze type compounds, bismuth
oxide layer structured compounds, lithium niobate and tantalate.
(i) Perovskites
Perovskite is a family name of a group of materials having the
mineral name of calcium titanate (CaTi03) exhibiting a structure of the
type ABO3. Many fenoelectric ceramics such as Barium Titanate
(BaTiOa), Lead Titanate (PbTiO,), Lead Zirconate Titanate (PZT), Lead
Lanthanum Zirconate Titanate (PLZT), Lead Magnesium Niobate (PMN),
Potassium Niobate (KNbOl), Potassium Sodium Niobate (KxNal.,Nb03)
and Potassium Tantalate Niobate (K(TaxNbl.,)03) have a perovskite type
structure.
(K) Tungsten Bronze type compounds
The tungsten bronze type ferroelectric crystals have a structure
similar to tetragonal tungsten bronze KxW03 (x<l). Lead niobate
(PbNb206) was one of the first crystals of the tungsten bronze type
structure to show useful ferroelectric properties. For lead niobate the BI
and B2 sites are occupied by ~ b ' ' ions. The open nature of the structure as
compared to the perovskite allows a wide range of cation and anion
substitutions without loss of ferroelectricity.
The ferroelectric crystals grown ffom solid solutions of alkali and
alkaline earth niobates have shown great potential for being used as a
material for laser modulation, pyroelectric detectors, hydrophones and
ultrasonic applications. The high Curie point (Tc = 560°C for lead niobate)
of their compounds makes them suitable for high temperature applications.
Another problem associated with this type of materials is the large volume
change due to phase transformation on cooling below the Curie point,
leading to cracking of the ceramic [I 2, 131.

(iii) Bismuth oxide layer structured ferroelectrics

The two most important piezoelectric materials with the ( ~ i 2 0 2 ) ~ '
layer structure are bismuth titanate (Bi4Ti3012) and lead bismuth niobate
(PbBizNb209). The structure of PbBi2Nb409 consists of comer linked
perovskite-like sheets, separated by ( ~ i 2 0 * )layers.
~*
The plate like crystal structure of these compounds leads to highly
anisotropic ferroelectric properties. The ceramics fabricated from the
( ~ i 2 0 2 ) ~layer
' compounds do not have very good piezoelectric properties
because of a very low poling efficiency. The piezoelectric properties have
been shown to be improved by grain orientation during the processing step.
One fabrication method involves the tape casting of plate like Bi4Ti3OI2
and PbBizNb~Oopowders. The powders get aligned during the formation
of the green tape. The orientation is hrther enhanced on sintering. In the
other method, the ceramic is hot forged leading to the orientation of the
grains along the forged direction [14 - 161.

The bismuth oxide layer structured ferroelectrics may become

important ferroelectric ceramics because of their higher stability, higher
operating temperature (Tc = 550-65OUC)and higher operating frequency.
These ceramics are mainly useful for piezoelectric resonators, which are
needed to exhibit a very stable resonant frequency.
(iv) Lithium Niobate and Tantalate
Lithium niobate (LiNbO,) and lithium tantalate (LiTa03) have
similar structures. The LiNb03 structure is actually a variant of the
perovskite structure with a much more restrictive arrangement. The
ferroelectric behavior of LiNb03 and LiTa03 was first discovered in 1949.
Their crystals are very stable with very high Curie points of 1210°C and
620°C for LiNb03 and LiTa03 respectively. These compounds are mainly
used in the single crystal form and have found applications in
piezoelectric, pyroelectric and electro-optic devices [17, 181.
1.5.2 Compounds containing hydrogen bonded radicals
Several water soluble ferroelectrics usually made in the single
crystal form have hydrogen bonded radicals in them. Potassium
dihydrogen phosphate (KH2P04,KDP) has a tetragonuoint group 42m.
Hence this crystal shows only piezoelectricity and no ferroelectric behavior
at room temperature. KDP single crystals exhibit very good electro-optical
and non-linear optical properties [19,20].
Triglycine sulfate (NH2CHzCOOH),H2S04 (TGS) and Rochelle
salt (NaKC4H406.4H20, sodium potassium tantalate tetrahydrate)
are water soluble crystals which are used due to their superiority over
other crystals that large crystals are easily grown from water solution. Yet
these crystals have many deficiencies such as weak ferroelectricity, low
Curie point, poor mechanical properties. For these reasons, KDP, TGS and
Rochelle salt single crystals are being gradually replaced by ferroelectric
ceramics [21,22].

1.5.3 Organic polymers

Polyvinylidene fluoride (PVDF, (CH2-CF2),,) and copolymers of
PVDF with trifluoroethylene {P(VDF-TrFE)) have found applications as
piezoelectric and pyroelectric materials. The piezoelectric and pyroelectric
properties of these polymers are due to the remanant polarization obtained
by orienting the crystalline phase of the polymer in a strong poling field.
Hence the piezoelectric and pyroelectric properties depend on the degree
of crystallinity of the polymer and the ferroelectric polarization of the
crystalline phase [23, 241.
The piezo-polynfers have some properties whch make them better
suited for use in medical imaging applications. The density of these
polymers is very close to that of water and the human body tissues and
hence there is no acoustic impedance mismatch with the body. The piezo-
polymers are also flexible and conformable to any shape. However, there
are also some problems associated with the piezcp$ectric polymers
including the very low dielectric constant (K = 5-10) which could lead to
electrical impedance matching problems with the electronics. The
dielectric losses at high frequencies are very large for these piezo-
polymers. The polymers also have a low Curie point and the degradation
of the polymer starts occurring at low temperatures (70°C-100°C). The
poling efficiency is very low for polymer specimens with large thickness
(>I mm).

1.5.4 Ceramic polymer composites

The drive for piezoelectric composites stems from the fact that
desirable properties could not be obtained from single phase materials such
as piezo-ceramics or piezo-polymers. For example, in an
electromechanical transducer, the desire is to maximize the piezoelectric
sensitivity, minimize the density to obtain good acoustic matching with
water, and make the transducer mechanically flexible to conform to a
curved surface. Neither a ceramic nor a polymer satisfies these
requirements. The requirement can be optimized by combining the most
useful properties of the two phases which do not ordinarily appear
together.

Ferroelectric composites are made up of an active ceramic phase

embedded in a passive polymer. The properties of the composites depend
on the connectivity of the phases, volume percent of ceramic and the
spatial distribution of the active phase in the composite [25]. The concept
of connectivity describes the arrangement of the component phases within
a composite. It is critical in determining the electromechanical properties
of the composite. The density, acoustic impedance, Qlectric constant and
the ferroelectric properties like the electromechanical coupling coefficient
kt change with the volume fraction of the ceramic. Optimum material and
ferroelectric parameters for medical ultrasound applications are obtained
for - 20-25 vol. % PZT ceramic in the composite [26,27].
1.6 Review of Literature: Bismuth Layer Structured Ferroelectrics
Bismuth containing layered perovskites belong to the family of
Aurivillius compounds with a general formula ( B ~ ~ O ~ ) ~ * ( A ~ - I B ~ O ~
consisting of m perovskite unit sandwiched between bismuth oxide layers,
where A and B are the two types of cations that enter the perovskite unit.
A is ~ i ~ ~' , a ~Sr2+,
+ , pb2+, or K";B is ~ i ~ ~' a, ' + ~, b " , MO" or w".
Ferroelectric properties, crystal structure and microstructure are
remarkably influenced by the composition fluctuation in the bismuth layer
structured compounds [28].
Ferroelectrics are technologically important materials particularly,
for their applications in non-volatile ferroelectric random access memory
(FeRAM) [29]. Random access memories (RAMS) based on
semiconductor integrated technology have been a great success; however,
these semiconductor RAMS can retain information only when the power is
on. A serious drawback is that when the power is interrupted all
information is lost (volatile memory). Furthermore, these RAMS are very
sensitive to radiation and this is detrimental for military and space
applications [30]. To be competitive with electrically erasable read-only
memories (EEPROM), ferroelectric memories (FeRAM) which are also
called non-volatile random-access memories (NVRAMs) must be
improved to withstand atleast lo'* eraseirewrite operations.
Lead zirconate titanate (PZT) has been conside@ to be the most
promising ferroelectric material for such applications, but, these materials
show serious degradation in ferroelectirc properties with cumulative
polarization switching known as 'fatigue'. In a memory cell, polarization
must be reversed to read or write data, this makes 'fatigue' a critical
obstacle for their practical applicability [31]. Thus the widespread
application and commercialization of the bismuth-layered perovskite
ferroelectrics have been limited so far by two drawbacks; their rather high
processing temperature and their relatively low remanent polarization.
In comparison with non-layered perovskite ferroelectrics such as
PZT the BLSFs offer several advantages such as fatigue free, lead free,
low operating voltages and most importantly their ferroelectric properties
are independent of film thickness. For FeRAM applications large
remanent polarization (Ps), low coercivity (Ec) accompanied by high Curie
temperature (Tc) are required for better performance and reliable
operation. The properties of the layered perovskite ferroelectrics can be
enhanced by addition or substitution of alternative cations [32]. Partial
substitution in perovskite ferroelectrics may result in the formation of a
heterogeneous microstructure that consist of microdomains which have
slightly different chemical compositions. These microdomains with
various chemical compositions will have different para-ferroelectric
transition temperatures and thus lead to a broadened dielectric peak at the
transition temperature, The para-ferroelectric transition of such
ferroelectrics vary characteristically with frequency. Such ferroelectrics
are known as relaxor ferroelectrics and typical examples include
Pb(MglnNb2i3)03(PMN) and Pb (Zn,;3Nb2~3)03
(PZN). Although common
relaxors have isotropic perovskites structures, dielectric~elaxationhas also
been reported in layered perovskite ferroelectrics [33]. The bismuth oxide
layered perovskite materials, such as SrBi2Nb20s (SBN) and SrBizTalOp
(SBT) for FeRAM applications have attracted an increasing attention
because they possess a reasonable spontaneous polarization which is one
of the key parameters for the information storage application [30].
Strontium bismuth tantalate (SBT) is reported to exhibit fatigue endurance
even after 1012cycles of operation with excellent retention characteristics.
The presence of Bi202 layers has been thought to serve as the shock
absorber for endwing the fatigue of polarization.

1.6.1 Effect of doping in BLSFs

Ferroelectric properties, crystal structure and microstructure are
remarkably influenced by the composition fluctuation in the bismuth layer
structured compounds. Much research has been reported in the open
literature aiming at improving the dielectric and ferroelectic properties of
SBN and SBT. In particular, doping with various metal oxides has been
one of the effective approaches to improve the properties [28]. Generally
doping could be in bismuth oxide layer andlor in perovskite like units (A
A+
or B sites) [34]. Millan et.al, reported the substitution of Rh. in ( ~ i z 0 3 ) ~ '
layers by other cations such as ~ b " , sn2* or ~e~~[35]. Within the
perovskite like units, partial substitution of strontium ions by bismuth ions
would increase the Curit: temperature and improve the dielectric properties
in both SBT and SBN [36].
1.6.2 Doping effect in layer structured SrBilTa109(SBT)ferroelectrics
SBT is orthorhombic structure with the space group A21am. The
dimensions of the rectangular parallelpiped unit c e h are a = 5.531,
b = 5.534, c = 24.984A0. This unit cell contains four SrBi2Ta209formula
units (56 atoms). The primitive cell, the smallest translational unit, is half
as large, containing two formula units (28 atoms), since the A2,am cell is
face centered and contains two lattice points, the (SrTaa07)--perovskite
slab perpendicular to the c axis, is a double layer of comer sharing TaOh
octahedra with Sr atoms in the midplane.
Unit cell structure of SBT is given in Fig. 1.4. The XRD analysis
of SBT doped with vanadium with a composition SrBi2(V,
T~I.,)~OU(SBVT)
revealed that SBVT ceramics consisted of a single
layered perovskite phase with no detectable secondary phase [37]. This
indicates that the layered perovskite structure of SBT was preserved with
partial substitution of tantalum ions by vanadium ions upto 10 at. %. No
appreciable change in lattice constants was observed. The presence of
~ i 0 layers
~ ~ might
' impose a restriction to prevent detectable secondary
phase [37]. This indicates that the layered perovskite structure of SBT
was preserved with partial substitution of tantalum ions by vanadium ions
upto 10 at. %, no appreciable change in lattice constants was observed.
The presence of ~ i 0 layers
~ ~ .might impose a restriction to prevent
possible reduction of lattice constants. The SBT sample showed a typical
paraelectric ferroelectric transition at 305OC with a peak dielectric
constant of -400, whereas SBVT sample had broad peak around 364 OC
with a maximal dielectric' constant 454 [33]. The paraelectric ferroelectric
 OBi @Sr OTa 0 0

Fig. 1.4. Unit cell structure of bismuth layered oxide (SrBizTa209)

transition temperature increased with the reduced frequency. The

dielectric constants at low temperatures are the same regardless of the
frequencies, whereas the dielectric constants at high temperatures varied
markedly. The significantly enhanced dielectric constant at high
temperatures may be due to the mobile charge vacancies. Oxygen vacancy
induced polarization becomes more predominant at high temperatures due
to an increased vacancy concentration, and at low frequencies due to the
inertia of oxygen ions.. Defect induced dielectric relaxation would be
directly related to the concentration of defects [38]. The broadening of
dielectric peak and the shift of para-ferroelectric transition of as-sintered
SBVT ferroelectrics were most likely due to the defect - induced -
dielectric relaxation as reported in the Bi doped SrTiO, system [39] and
not a result of diffised phase transition as that in a relaxor ferroelectric. In
the SBVT system, oxygen vacancies could be generated during the
sintering and due to the presence of the reduced valence state of vanadium
ions.
SBT films synthesized with Sr-deficient and Bi-excess
compositions (Sro8Bi2,2Ta2O9)show significantly improved ferroelectric
properties 1291. Large atomic displacements of ~ i "in the Bi202layer and
of 0(5)~- in the TaOd octahedron play an important role in the large
ferroelectric spontaneous polarization in SBT. Although the displacements
of ~ a ' ' and 02-ions in the TaOb octahedron contribute to the total
polarization oppositely, net polarization caused by the distortion of TaOh
enhances the total ferroelectric polarization. In Sro.RBi2,2Ta209
the lattice
parameters a and b are smaller while c is slightly larger than those in the
stoichiometric sample. The distortion of the TaO6 octahedron is more
pronounced in Sr0.8Bi2.2Ta209
than SBT [40].
It is well known that the ferroelectricity in the Aurivillius phase
originate from the structural distortions in the compounds. It has
suggested that the major cause of spontaneous polarization in the two-layer
material is the displacement of the A -site cation in the perovskite block
along the a direction in the polar space group A2,am and not the
movement of the B - site cation away from the octahedral center [41]. On
the contrary 0 - B - 0 chains which construct perovskite layers, are thought
to be the origin of the polarization in Aurivillius phases [42]. It has been
reported that in Sr- deficient and Bi - excess SBT compositions
(Sr1.~Bi2+~Ta209),
the increasing substitution of ~ i ions
~ for
' sr2' ions gives
rise to cation vacancies at the sr2' site and causes a compressive stress in
the perovskite - like unit, as a result, the mismatch between the peovskite -
like unit and the Biz02 layer distortion in the [Ta06] octahedral compared
with the stoichiometric composition of SBT. Such structural distortions
are reported to lead to the larger ferroelectric spontaneous polarization and
higher Curie temperatures [29]. For the composition Srl.,Bi2+2x/3Ta209
the
distortion in TaOh octahedra gives rise to ferroelectricity while sr2+is a
spacer ion in SBT crystal [43]. A vacant site is creased by replacement of
two ~ i for
~ three
' sr2', assuming oxygen vacancies do not exist. In the
crystal structure, the Ta02 plane in TaOh octahedra is the densest plane in
the unit cell. Thus the substitution of smaller ~ i ions
" for srZ' ions causes
pronounced stresses in the perovskite unit and the layer and results in a
large distortion in the Ta06 octahedra [41].
XRD patterns of powdered Srl.,Ba,Bi2(Nbo,5Ta)209 and
Sro5Bao,sBi2(Nbl.,Ta,)09
where x = y = 0.0, 0.25,0.5, 0.75 and 1.0 clearly
confirmed the formation of a single phase with no impurity phase [21]. At
room temperature the solid solutions were reported to have orthorhombic
structure. A linear increase in both the lattice constants a and c was
observed w i t h _ ~ a doping.
~& With increasing ~ a doping
~ ' there was a
decrease in Curie temperature accompanied by peak broadening suggested
a decrease in the distortion of the oxygen octahedral. The Curie
temperature of Sro,5Bao,5Bi2(Nbl.,Ta,)OP
is found to decrease with
increasing ~ a ' +content [32].
1.6.3 Doping effect in layer structured SrBilNb109(SBN) ferroelectrics
Yun Wu et al. reported that when SBN was doped with vanakum
SrBi2(VxNbl.x)209 there was a decrease in sintering temperature
(),20O0C) [30]. The lowered sintering temperature was partly due to the
low melting point of vanadium oxide ( - 6 9 0 Q and partly due to the
possible formation of an eutectic liquid phase in this multiple oxide
system. There was a slight decrease in crystal lattice constants as the
amount of vanadium increased. The Curie temperature of the layered
perovskites increased from 435°C for SBN to 440°C and 457°C for
strontium bismuth vanadium niobate (SBVN) for 5 at.% and 10 at.%
vanadium doping respectively. The peak dielectric constants at their
respective Curie points also increased approximately by 60%. The
increase in Curie temperature was due to the increase in rattling space [30].
There was a decrease in coercive field, increase in remanent polarization
and a decrease in tangent loss [38]. The dc conductivity of the SBVN
ceramics decreased with an increased doping content at higher temperature
range, while there was an appreciable decrease in activation energy from
1.174 to 1.074eV as the doping increased from 0 to 10 at. % [3 I].
Ezhilvalavan et al. reported that vanadium doping (10 at.%) made
the SBN denser with the reduced level of porosity and a marginal increase
in grain size [28]. The grain size of SBVN ranges from 2-5 mm. The
enhanced dielectric constant at high temperature can be due to mobile
charge carriers which may be related to the oxygen vacancies. Oxygen
vacancy-induced polarization becomes more predominant at elevated
temperatures, they play a significant role in the electric polarization of
perovskite materials. 1' n comparison with the as-sintered SBVN, the
dielectric constant of annealed sample at Tmax is lower than that of the as-
sintered SBVN. Since oxygen vacancy-induced dielectric relaxation is a
diffusion related process, its effect is less significant at low temperatures
[281.
Doping in the A site of SBN by Ca (CSBN) ( C ~ S ~ , . ~ ) B i ~ N band
20g
Ba (BSBN) (BaXSrl.,)Bi2Nb2Opresulted in enhanced dielectric properties
[MI. XRD analyses indicated that single phase layered perovskites were
formed and no secondary phase was detectable. Doping of Ba and Ca
resulted in a linear increase or decrease in lattice constants, depending on
the dopant ionic radii relative to that of sr2.. No apparent differences in
microstructures between SBN and Ca or Ba doped samples was observed.
With vanadium doping, the grain sizes were about 2 to 3 times smaller as
compared to SBN ceramics. When s r 2 in the A site is substituted by a
larger ion ( ~ a " ) ,the Curie temperature decreases. When a smaller ion,
ca2' substitutes sr2-,the Curie temperature increases. This suggests that
with the bismuth layer structural constraint, the introduction of bigger ions
into A site would eventually consume more space despite the lattice
increase. As a result, the rattling space for the ions inside the oxygen
octahedra would be reduced and Curie temperature would decrease. The
dielectric constants of SBN, regardless of doping, consist only of ionic and
atomic polarization, since the ionic radii of doping ions (ca2., ~ a ' .and v")
are different from the original site ions ( s r 2 and ~ b ' ' ) . Increasing the
amount of doping ions would lead to a change of both electronic and ionic
polarization [45]. Both atomic and ionic polarization would decrease with
an increasing amount of vanadium introduced into the system and lead to
reduced dielectric constants. The tangent loss increases with increasing
temperatures particulariy at temperatures higher than 400°C. This is
caused by a higher concentration of charge camers at higher temperatures.
The dc conductivity of SBN at higher temperatures is found to decrease
with A or B site doping [44].
The substitution of tellurium for bismuth and of sodium or
potassium for strontium in Bi2Te,Srl.,Na(K),Nb2O9{@T)(SK)N and
(BT)(SN)N} compounds with an Aurivillius structure has been reported
[46]. This material has a double substitution; a B-site substitution in
bismuth oxide layers and a Sr-site substitution in the perovskite layers. In
these compositions the charge compensation is made at the A-site by the
substitution of Na or K for Sr. For compositions where the substitution is
involved only at the A-site of the perovskite layer, a larger radius of the
substituting ion leads to a lower ferro-paraelectric transition temperature
and to only one maximum in dielectric constant 1471. According to this the
order of the transition temperatures, obtained from E'(T) measurements,
should be Tc (BT)(SK)N < T, (BT)(SN)N < Tc (BSN) [48].
When La is doped in SBN (Lao,lSro,gBi2Nb209
(LSBN)) the Curie
temperature increased from 418OC for SBN to 480°C [49]. However the
dielectric constant at both room temperature and Curie point were found to
decrease with ~ a doping.
~ ' The dc conductivity of LSBN is appreciably
lower than that of SBN. The activation energy, (DE,) of LSBN at the
higher temperature range was found higher than that of SBN and DEj is
assumed to be the energy necessary to create and move vacancies, whereas
the conduction at the high temperature range is likely predominant by
intrinsic defects. DEj of LSBN at higher temperature range found to be
higher than that of SBN is assumed to be the energy necessary to create
and move vacancies. The tangent loss is found to be higher than that of
SBN ferroelectrics. This is due to a combination of the space charge
polarization and domain wall relaxation [49].
Moon et al. studied the ferroelectric properties of substituted
Aurivillius phases SrBizNb2.xM,0s(M = Cr, Mo) [42]. 4jhe XRD data of
doped samples were similar to SBN. All peaks were slightly shifted to the
higher angle in the Cr substituted compounds, while those are almost
unchanged in the Mo substituted compounds. This is due to the difference
of ionic radii between Nb and Cr, and between Nb and Mo substituted
compounds. When Cr or Mo substitutes Nb more than 20 mole%, the
second phase peak begins to arrive. The dielectric constant of
4 7
unsubstituted compound is 120 within the frequency range of 10 -10 Hz.
When substitution is made at 10 mole% the value increases to about 225 in
the case of Mo insertion while it becomes -160 for Cr doping. The XRD
spectra of SBN doped with tungsten with a composition Sr,+,
Bi2.2x/3(WxNbl.x)209
(SBWN) showed the formation of single phase layered
perovskites upto 2.5 at % of W. However XRD spectra of samples
containing more than 5% tungsten showed the formation of secondary
phase [SO]. Compared to vanadium doping, tungsten has a much lower
solubility in SBN, though the ionic radius of tungsten ions is closer to that
of pentavalent-niobium ions than that of the vanadium ions. For all
samples of SBWN there is a sharp transition in Curie points and lowered
dielectric constant. There was an appreciable reduction in the loss tangent
for SBWN compared to SBN. The tangent loss is likely due to space
charge polarization and domain wall relaxation [SO].
1.7 ABi4T4015andA2Bi,TisO,s (A = Ba, Ca, Pb and Sr) ceramics
Crystallographic and microstructural studies on BLSFs have been
made by many workers [47, 51,531, detailed reports on the dielectric and
ac impedance studies are rare. S. Ikegami et a1 [52] repmed the Tc value
as 550°C for SrBi4Ti4015(SBT), while E.C. Subbarao et a1 [53], Maria et al
[54] and S. Kojima et a1 [55] have reported 530°C for the same system.
Jun Zhu et al [56] reported the Curie temperature as 520°C for SBT.
Noguchi et a1 [57] and R.Z.Hou et a1 [58] have reported that there is a
sharp dielectric constant peak at Tc = 520°C for SrBi4Ti4015with the
maximum dielectric constant more than 2300. Szu Hwee Ng, et a1 [59]
reported that the dielectric constant measured at 100kHz, increases from
room temperature and peaks to 2770 at the Curie temperature of -539 "C.
The dielectric behavior of BaBilTi4Ol5reported by D. Ben Jannet et
a1 [60] shows the diffused phase transition with the temperature maximum
(Tm) permittivity shifting from 400 to 420°C as the frequency increases
from 1 kHz to 1MHz. This is a typical behavior of relaxor ferroelectrics,
e.g. PMN [61]. Such a phenomenon has already been observed in several
Ba - bearing Aurivillius compounds [62].
K. Srinivas [63] et al reported the dielectric constant of the poled
Sr2Bi4TiSOIB
sample was measured at a frequency of 10 kHz from room
temperature to 400°C and the values of the dielectric constant were found
to be about 120 and 1150 at room temperature and at the transition
temperature (Tc) respectively. The values of the dielectric loss are fairly
low and remain almost constant upto about 225 "C. Thereafter, there is a
sudden increase in the dissipation factor, with a peak slightly before the
Tc. At the Tc, it reaches a minimum value and thereafter, it begins to
increase once again. B.Aurivillius et a1 studied the ferroelectricity in the
compound Ba2Bi4Ti5018[64] with the tetragonal cell dimensions a = 3.88
A" and c = 50.3A0. R. 2.Hou et a1 [65] reported a sharp dielectric constant
peak at 32j0C for SrZBi4Ti5Ol8
ceramics and for Ba2Bi4TisOlsceramic the
diffused phase transition at -330°C.
References

[l] G.D.Raj, Solid State Physics, Amrnol Publication, New Delhi (1995).
[2] C. Kittel, Introduction to Solid State Physics, Wlley Eastern, New
York (1977).
[3] J. S. Black more, Solid State Physics, Cambridge UniversityPress,
Cambridge (I 985).
[4] H. V. Keer, Principles of Solid State Physics, Wiley Eastern, New
York (1993).
[S] M. E. Lines and A. M. Glass, Principles amd Applications of
Ferroelectrics and Teleated Materials, Claredon Press, Oxford ( 1 997).
[6] S. L. Swartz, IEEE Trans. Electrical Insulation, 25 (1990) 935.
[7] J. F. Scott, L. Kamrnerdiner, M. Paris, S. Traynor, V. Ottenbacher,
A. Shawabkeh and W. F. Oliver, J. Appl. Phy. 64 (1988) 1400.
[8] J. F. Scott and C. A. Araujo, Science, 246 (1989) 1400.
[9] L.I. Maissel and R. Clang, Handbook of Thin film Technology,
Mcgraw Hill, New York (1970).
[lo] A. Baudrant, H. Vial and J. Daval, J. Crys. Growth, 43 (1978) 197.
[ I l l C.E. Land, J. Sm. Ceram. Soc. 72 (1989) 2059.
[I21 G. Goodman, J. Am. Ceram. Soc. 36 (1958) 368.
[I31 M.H. Francombe and B. Lewis, Acta Cryst. 11 (1958) 696.
[14] E.C. Subbarao, J. Am. Ceram. Soc. 45 (1962) 166.
[15] S.L. Swartz, W.A. Schulze, and J.V. Biggers, Ferroelectrics, 38 (1981)
765.
[I61 T. Takenaka and S. Sakata, Jap. J. Appl. Phys. 19 (1980) 31.
[I71 K. Nassau, H.J. Levinstein and G.M. Loiacona, J. Phys. Chem. Solids,
 27 (1966) 983.
[18] (a) S.C. Abraham, W.C. Hamilton, and J.L. Bemstein, J. Phys. Chem.
Solids, 27 (1966) 997 (b) S.C. Abraham, W.C. Hamilton and J.M.
Reddy, J. Phys. Chem. Solids 27 (1966) 997 (c) S.q, Abraham, H.J.
Levinstein, and J.M. Reddy, J. Phys. Chem. Solids, 27 (1966) 1019.
[I91 B.C. Frazer, and R. Pepinsky, Acta Crystallogr. 6 (1953) 273.
[20] G.E. Bacon, and R.S. Pease, Proc. R. Soc. London, A 220 (1953) 397.
[21] S. Hoshino, Y. Okaya and R. Pepinsky, Phys. Rev. B. 115 (1959) 323.
[22] M.I. Kay and R. Kleinberg, Ferroelectrics, 5 (1973) 45.
[23] M.G. Broadhurst, G.T. Davis, J.E. McKinney and R.E. Collins,
J. Appl. Phys. 49 (1978) 4992.
[24] N. Takahashi and A.Odaiyama, Ferroelectrics, 32 (1981) 49.
[25] N. Takahashi and L.J. Bowen, K.A. Klicker and L.E.Cross, Mater.
Engg. 2 (1980) 107.
[26] A.Safari, J.Phys.111 France, 4 (1994) 1129.
[27] V.F. Janas and A. Safari, J. Am.Ceram.Soc. 78 (1995) 1945.
[28] S. Ezhilvalavan, J.M. Xue, John Wang, Mater. Chem. and Phys.75
(2002) 50.
[29] A.B. Panda, A. Pathak, M. Nandagoswami, P. Pramanik, Mat. Sci.
and Engg. B. (2003) 97.
[30] Y. Wu, G. ~ a oJ., Mater. Res. 15 (2000) 1583.
[31] Y. Wu, Chau Nguyen, J. Am. Cer. Soc. B. 4 (2001) 2882.
[32] N. Syam Prasad, K.B.R. Varma, Mater. Res. Bull. 38 (2003) 195.
[33] Y. Wu, Mike J. Forbess, S. Seraji, S.J. Limmer, T.P. Chou, G. Cao,
J. Appl. Phys. 89 (2001) 5647.
1341 P. Millan, A. Ramirea, J. Mater. Sci. Lett. 14 (1995) 1657.
[54] Maria V. Gelfuso, Daniel Thomazini, Jose A. Eiras, J. Am. Ceram.
SOC.,82 (9) (1999) 2368.
[55] S. Kojima R.Imaizumi,S.Hamazaki, M. Takashige, J. of Mol. Struc.,
348 (1995) 37.
[56] Jun Zhu, Xiang-yu Mao, Xiao-bing Chen, Solid State Commn., 129
(2004) 707.
[57] Y. Noguchi, M.Miyayama, T. Kudo, Appl. Phy. Lett. 77 (2000) 3639.
[58] R.Z. Hou, X.M. Chen, Solid State Commn. 130 (2004) 469
[59] Szu Hwee Ng, J.M. Xue, John Wang, Materials chem.. and phys. 75
(2002) 131.
[60] D.Ben Jannet, M. El Maaoui, J.P. Mercurio, kluwer Academic
publishers, Nov. 13, (2003).
[61] G.A. Smolenski, V.A. Isupov, A. 1. Agranovskaya, S.N. Popov, Phys.
Solid State, 2 (1960) 2584.
[62] C. Miranda, M.E.V. Costa, M. Avdeev, A.L. Kholkin,, J.L.Baptista,
Journal of Europian Cer. Soc., 21 (2001 ) 1303.
[63] K. Srinivas, A.R. James, J. Appl. Phys. 86 (1999) 3885.
[64] B.Aurivillius, Physical review, 126, No.3 (1962) 893.
[65] R.Z. Hou, X..M Chen, Journal of materials research, 20 (9) (2005)
2354.