4 Ferroelectrics, Piezoelectrics,

and Pyroelectrics

The whole of science is nothing more than a refinement of everyday thinking. Science is the attempt
to make the chaotic diversity of our sense experience correspond to a logically uniform system of
thought.
Albert Einstein

4.1 Introductory Remarks The prefix piezo- in the word piezoelectrics

is derived from a Greek word, piezein, meaning
The term ferroelectrics arose by analogy with pressure. Piezoelectrics are materials in which
ferromagnetics, mainly because they have electricity can be generated by an applied
similar characteristics: under electric fields for mechanical stress or a mechanical stress can be
ferroelectric phenomena and under magnetic produced by an applied electric field. This
fields for ferromagnetic phenomena. The prefix interconvertible behavior was first discovered
ferro- derived from ferum, which means iron in by Pierre and Jacques Curie in 1880 in certain
Latin. The term is perfect for ferromagnetics, crystals, such as quartz, zinc blends, tourma-
since all ferromagnetic phenomena are associ- line, and Rochelle salt.4'5 The term piezoelec-
ated with the special type of spin arrangement tricity has been used by scientists since 1881 to
of the iron atoms. But in ferroelectrics there are distinguish the piezoelectric phenomena from
no iron atoms, so the prefix does not mean iron. electrostriction. The piezoelectric phenomena
Rather, it implies the similarity in characteris- occurs in both the ferroelectric and the nonfer-
tics to ferromagnetics. Like ferromagnetics, roelectric states.2'3
ferroelectrics exhibit a spontaneous electric The prefix pyro- in the term pyroelectrics
polarization below the Curie temperature, a means heat in Greek. Thus, pyroelectricity
hysteresis loop, and an associated mechanical means heat-generated electricity. This effect is
strain. However, ferroelectrics differ from fer- also convertible. This implies that heat can be
romagnetics in their fundamental mechanisms generated by electricity resulting from the
and also in some of their applications. change of the state of electric polarization, such
In Europe, ferroelectrics are sometimes called as electrothermal and electrocalorie effects.
Seignette electrics. This term is somewhat Pyroelectric phenomena were in fact discov-
misleading, because Seignette did not discover ered much earlier. Theophrast discovered the
the ferroelectric phenomena. Instead, in 17th- pyroelectric phenomena in tourmaline in 314
century Rochelle, France, he discovered BC, and the word pyroelectricity has been used
Rochelle salt (potassium-sodium tartrate- since 1824, first by Brewster.6 The theory of the
tetrahydrate, KNaC 4 H 4 0 6 -4 H 2 0), a colorless pyroelectric effect was first formulated by W.
crystalline compound with an orthorhombic Thomson (Lord Kelvin) in 1878 and further
structure. At that time, the material was used developed by Voigt in 1897 and later by Born,
as a laxative. More than 200 years later, in 1921, based on his famous crystal lattice dynamics in
Valasek discovered the ferroelectric phenomena 1921 and 1928.7
in the same material.1"3 Ferroelectrics were In general, all materials undergo a small
discovered much later than ferromagnetics. change in dimension when subjected to an
However, there are now more than 1,000 solid external force, such as an applied electric field,
materials possessing ferroelectric properties. a mechanical stress, or a change in temperature

213
214 Dielectric Phenomena in Solids

(heat). Depending on the material structure, electric as well as pyroelectric effects. There is
such a small change in dimension may result a subgroup within these 10 classes that pos-
in a change in electric polarization and hence sesses spontaneous polarization and reversible
give rise to the occurrence of the ferroelectric, polarization; this subgroup can exhibit all three
piezoelectric, or pyroelectric effects. It can be effects—ferroelectric, piezoelectric, and pyro-
imagined that materials exhibiting these effects electric. In fact, the ferroelectric effect is an
must be polar and have an electrical order, empirical phenomenon distinct from piezoelec-
implying that they must be crystals or poly- tric and pyroelectric effects in that it exists
crystalline materials composed of crystallites. A with a reversible polarization. The relationship
crystal or a crystallite must have a definite between polarization behavior and crystal
chemical composition, with the molecules made structure is shown in Figure 4-1.
up of positive ions (atoms sharing part of their The 32 point groups are subdivisions of
valence electrons with others) and negative ions seven basic crystal systems, which are based on
(atoms receiving part of electrons from others) the degree of symmetry. In the order of ascend-
occupying lattice sites to constitute a crystal ing symmetry, these seven basic crystal systems
structure lattice. The smallest repeating unit of are triclinic, monoclinic, orthorhombic, tetrag-
the lattice is called the unit cell, and the specific onal, trigonal (rhombohedral), hexagonal, and
symmetry of the unit cell determines whether cubic. For a more detailed description of these
the crystal exhibits ferroelectric, piezoelectric, seven systems, see references 8 and 9. Based
pyroelectric, or electro-optic effects. on the optical properties, these systems can be
On the basis of the symmetry elements of classified into three major optical groups, as
translational position and orientation, there are shown in Table 4-1. The optical biaxial group
230 space groups. Ignoring translational repe- refers to the crystals in which the molecules are
tition, these 230 groups break down into 32 not spherical, with the long axis longer than the
classes, known as the 32 point groups. Point axes perpendicular to it. Depending on the
groups are based on orientation only.8'9 Any structure of the crystal, if the index of refrac-
point may be defined by coordinates JC, y, and tion in the direction along the long axis is dif-
z, with respect to the origin of symmetry. A cen- ferent from that along the direction of the short
trosymmetric crystal is a crystal in which the axis, the crystals are optically biaxial and
movement of each point at x, y, z, to a new point exhibit birefringence. If the indices of refrac-
at -x, -y, -z, does not cause a recognizable tion in the two axes are identical, then the
difference. This implies that centrosymmetric crystals are optically uniaxial. However, some
crystals are nonpolar and thus do not possess
a finite polarization or dipole moment. Of the Table 4-1 The seven basic crystal systems.
32 classes (or point groups), 11 classes are
centrosymmetric and 21 classes are non- Crystal System Optical Group Order of Symmetry
centrosymmetric, possessing no center of sym-
metry. The latter is the necessary requirement Triclinic Biaxial Very Nonsymmetric
for the occurrence of piezoelectricity. Monoclinic Biaxial
However, one of the 21 classes, though
classified as the noncentrosymmetric class, Orthorhombic Biaxial
creasi

possesses other combined symmetry elements, Tetragonal Uniaxial

thus rendering no piezoelectricity. So, only 20 c
classes of noncentrosymmetric crystals would Trogonal Uniaxial
(Rhombohedral)
exhibit piezoelectric effects. In 10 of these 20
classes, polarization can be induced by a Hexagonal Uniaxial 1 r
mechanical stress, while the other 10 classes
Cubic Optically Very Symmetric
possess spontaneous polarization, so they are Isotropic
permanently polar and thus can have piezo-
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 215

32 Classes
Crystal Symmetry Point Groups

21 Classes 11 Classes
Noncentrosymmetric Groups Centrosymmetric Groups

1 Class 10 Classes
Possessing Other Symmetric Piezoelectric Effect
Elements (Nonpolar)
(Polarized under Mechanical Stress)

10 Classes
Piezoelectric Effect
Pyroelectric Effect
(Spontaneously Polarized)

Subgroup of the 10 Classes

Ferroelectric Effect
Piezoelectric Effect
Pyroelectric Effect
(Spontaneously Polarized
with Reversible Polarization)

Figure 4-1 Classification of crystals showing the classes with piezoelectric, pyroelectric, and ferroelectric effects.

crystals with their constituent molecules being of polymers, there may exist a possibility
close to spherical particles in shape; then in that the combined collective and coordinate
such cases the index of refraction is identical in effects of the structure would satisfy the
all directions and the crystals are optically requirements for the occurrence of piezo-
isotropic. electric, pyroelectric, and even ferroelectric
Apart from the crystals with the structures phenomena. Polyvinylidene fluoride (PVDF)
just described, some synthetic and biological polymer is a good example that complies with
polymers with different structures also exhibit such requirements under certain conditions.
piezoelectric, pyroelectric, and ferroelectric However, polymers are potential materials
phenomena.10-15 In view of the complex molec- suited particularly for many practical applica-
ular, crystalline, and morphological structures tions because of their easy processability into
216 Dielectric Phenomena in Solids

thin, tough, and flexible films. We shall discuss properties are also very small. However, there
polymers having these properties later. In the are a number of crystals with a nonsymmetri-
present chapter, we shall deal only with cal structure (see Figure 4-1) that exhibit a large
ferroelectric, piezoelectric, and pyroelectric polarization, with the dielectric constant up to
effects. Electro-optic effects were discussed in 105, under certain conditions. Obviously, such
Chapter 3. a large magnitude of polarization has attracted
many researchers to study it theoretically and
to develop various practical applications.
4.2 Ferroelectric Phenomena A ferroelectric crystal shows a reversible
spontaneous electric polarization and a hys-
Ferroelectricity is one of the most fascinating
teresis loop that can be observed in certain tem-
properties of dielectric solids. Materials
perature regions, delimited by a transition point
exhibiting ferroelectric properties must be
called the Curie temperature, Tc. At tempera-
either single crystals or polycrystalline solids
tures above Tc, the crystal is no longer ferro-
composed of crystallites; they must also
electric and exhibits normal dielectric behavior.
possess reversible spontaneous polarization. In
Ferroelectric materials usually, but not always,
this section, we shall discuss the various fea-
exist in a nonpolar state at temperatures above
tures of ferroelectrics, the mechanisms respon-
Tc, and have anomalously high dielectric con-
sible for the appearance of these features, and
stants, especially near the Curie temperature.
ferroelectric materials and their applications.
Typical dielectric constant-temperature and
polarization-temperature characteristics are
4.2.1 General Features shown in Figure 4-2. The dielectric constant
The polarization induced by an externally increases very rapidly to a very high peak value
applied field in normal dielectric materials is at Tc. The anomalously high value of er in the
very small, with the dielectric constant usually neighborhood of Tc is generally referred to as
less than 100, and its effects on other physical the anomalous value. At T > Tc, anomalous

_L
Tc T Tc T Tc T
Barium Titanate KDP Rochelle Salt

(a) rc=i2o°c. (b) rc = ~i50°c. (C)7- C = 24°C.

Figure 4-2 Schematic illustration of the variation of the dielectric constant £, and the spontaneous polarization of P5 with
temperature for three typical ferroelectric crystals: (a) Barrium titanate (BaTi03) with TC = 120°C, (b) Potassium dihydro-
gen phosphate (KDP, KH2P04) with TC = -150°C, and (c) Potassium Sodium tartrate-tetrahydrate (Rochelle Salt,
KNaC4H406-4H20) with Tc = 24°C.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 217

behavior follows closely the Curie-Weiss where A is the area of the specimen. V is the
relation applied voltage, which is usually an AC signal
voltage of low frequencies. Thus, the applied
field across the specimen is F = Vc/d = (V -
V0)/d.
where C is known as the Curie constant. In fact, A typical hysteresis loop is shown schemat-
anomalous behavior always appears near any ically in Figure 4-4. When the field is small, the
transition point between two different phases, polarization increases linearly with the field.
even at T below Tc. At the transition points, This is due mainly to field-induced polariza-
there are anomalies not only in the dielectric tion, because the field is not large enough to
constant and polarization, but also in piezo- cause orientation of the domains (portion OA).
electric and elastic constants and specific heat, At fields higher than the low-field range, polar-
because of the change in crystal structure. ization increases nonlinearly with increasing
Ferroelectrics have reversible spontaneous field, because all domains start to orient toward
polarization. The word spontaneous may mean the direction of the field (portion AB). At high
that the polarization has a nonzero value in the fields, polarization will reach a state of satura-
absence of an applied electric field. The word tion corresponding to portion BC, in which
reversible refers to the direction of the sponta- most domains are aligned toward the direction
neous polarization that can be reversed by an of the poling field. Now, if the field is gradu-
applied field in opposite direction. The sponta- ally decreased to zero, the polarization will
neous polarization Ps usually increases rapidly decrease, following the path CBD. By extrapo-
on crossing the transition point and then lating the linear portion CB to the polarization
gradually reaches a saturation value at lower axis (or zero-field axis) at E, OE represents the
temperatures. The most prominent features of spontaneous polarization Ps and OD represents
ferroelectric properties are hysteresis and non- the remanent polarization Pr. The linear
linearity in the relation between the polariza- increase in polarization from Ps to Pp is due
tion P and the applied electric field E The mainly to the normal field-induced dielectric
simplest method for measuring spontaneous polarization. Pr is smaller than Ps because when
polarization is the Sawyer and Tower method,16 the field is reduced to zero, some domains may
as shown in Figure 4-3, in which C is the capac- return to their original positions due to the
itance of the ferroelectric specimen and C0 is
the standard capacitor. The voltage across C
should be sufficiently large to render a satura-
tion in polarization, so V0 should be propor-
tional to the polarization charge, V0 = AP/C0,

Figure 4-3 The Sawyer-Tower method for the measure- Figure 4-4 Schematic diagram of a typical ferroelectric
ment of the polarization-electric field (P-F) characteristics. hysteresis loop.
218 Dielectric Phenomena in Solids

strain situation, thus reducing these domains' to form a square loop, as shown in Figure
contribution to the net polarization. 4-5(a), while in most ceramics the loop is
For most ferroelectric materials, the compo- rounded, as shown in Figure 4-5(b), because of
nent due to the normal field-induced dielectric the more sluggish reversal, which is due partly
polarization is very small compared to the to the axes of the unit cells in the randomized
spontaneous polarization; therefore, for most arrangement of the nonuniform crystallites.
applications, this component can be ignored. Ferroelectric materials exhibit ferroelectric
The magnitude of the difference between Pp properties only at temperatures below Tc
and Ps in Figure 4-4 is exaggerated for the because they are polar; at temperatures above
purpose of clear illustration. The field required it, they are not polar. Obviously, the shape of
to bring the polarization to zero is called the the hysteresis loop depends on temperature.
coercive field Fc (portion OR on zero polariza- Figure 4-6 shows the shape of the hysteresis
tion axis). Fc depends not only on temperature, loop for Rochelle salt at two different temper-
but also on the measuring frequency and the atures. The loop becomes gradually diminished
waveform of the applied field. When the field at T > Tc, eventually degenerating to a straight
in the opposite direction decreases to zero, line at T much larger than Tc, when the ferro-
the polarization is reversed, indicating that electric behavior disappears completely.
domains (see Section 4.2.4) have already been However, some ferroelectric materials can be
formed before poling and that the motion of the driven from the paraelectric state to a ferro-
domain walls results in the change of direction electric state at T > Tc by an applied field larger
of polarization. The hysteresis arises from the than a certain critical value Ft. In other words,
energy needed to reverse the metastable dipoles the applied electric field larger than Ft tends to
during each cycle of the applied field. The area shift the Curie point to a higher temperature.
of the loop represents the energy dissipated The higher the temperature above Tc, the higher
inside the specimen as heat during each cycle. the field required to induce ferroelectricity. If a
In general, the hysteresis loop is measured with ferroelectric material is subjected to a large AC
AC fields at low frequencies, 60 Hz or lower, to field at a temperature slightly higher than Tc,
avoid heating the specimen. two small hysteresis loops may appear, as
In general, ferroelectricity is harder to shown schematically in Figure 4-7. A double
demonstrate in polycrystalline materials com- hysteresis loop has been observed in BaTi03 at
posed of crystallites, such as ceramics, than in a temperature a few degrees above Tc}1,n For
a single crystal because of the random orienta- the transition from the tetragonal structure to
tion of crystallites. This is why in some single the orthorhombic structure at the transition
crystals the polarization reverses quite abruptly temperature of 5°C for BaTi0 3 at an AC field

-0.50

-800 -400 0 400 800

F(\ZcrrT1)

Figure 4-6 Temperature dependence of the hysteresis

Figure 4-5 Schematic diagrams of (a) the "square" and loop for Rochelle salt with Tc = 24°C. (A) T = 19°C < Tc,
(b) the "rounded" hysteresis loops. and (B) T = 42°C > Tc.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 219

higher than Fh a triple hysteresis loop, consist- ization P, which is the dipole moments per unit
ing of one main loop and two small loops, has volume and is also equal to the surface charge
also been observed.19 density on the electrode. During polarization, a
Using the circuit shown in Figure 4-3, the current j = dP/dt will flow through the speci-
voltage across C is proportional to the polar- men per unit area. Suppose that the wave shape
of the switching signal voltage is as shown in
Figure 4-8(a) for two cases: one with the
voltage (or field) increasing linearly with time
and the other with the field decreasing linearly
with time. In this case, the resulting current
flowing through the specimen, for a normal
linear dielectric, is constant, as shown in Figure
4-8(b). But if the specimen is a ferroelectric
material, the current will be a short pulse, as
shown in Figure 4-8(c). This is one of the
simplest ways to distinguish ferroelectric from
nonferroelectric materials.
The size of the unit cell and the force of ions
in the lattice of a crystal are temperature
dependent. As the temperature changes, there
exists a critical temperature (i.e., the transition
temperature), at which a particular structure
becomes unstable and tends to transform to a
more stable one. Although the transition
Figure 4-7 Schematic diagram showing a typical double involves only small ionic movements, they
hysteresis loop at T slightly higher than Tc and at F larger cause a marked change in properties. A change
than a critical value Ft. in dimensions of a crystal or a crystallite will

(a) Waveform

(b) Dielectric
PI

(c) Ferroelectric
P\ JL.
T~

Figure 4-8 Schematic diagrams showing (a) the waveforms of the linearly time-varying electric field (b) the correspon-
ding current j flowing through the loss-free dielectric material, and (c) the corresponding current j flowing through the
ferroelectric material.
220 Dielectric Phenomena in Solids

create internal stresses, particularly at the coupling forces between the outer electrons of
boundaries between crystallites in polycrys- neighboring magnetic ions.
talline materials, such as in ceramics. Under Since the dielectric constant of ferroelectric
certain conditions, the magnitude of such materials is extremely high near the transition
stresses may be large enough to cause internal temperature, the polarization induced in the
cracks. paraelectric (nonpolar) region at T > Tc by an
Transition from one crystal structure to applied electric field along the ferroelectric axis
another usually involves a change in entropy goes gradually over into the spontaneous
and volume. In general, there are two different polarization region upon cooling below Tc. The
orders of transition. When the spontaneous effect of this field tends to shift Tc to a higher
polarization goes from zero to a finite value, or temperature, as shown in Figure 4-9. The polar-
from one value to another, the change in polar- ization-temperature curve shifts to a higher
ization may be continuous or discontinuous. If temperature as the applied field is increased.
there is discontinuity in the change of the polar- However, the concept of the Curie temperature
ization, the transition is referred to as a first- cannot be applied to ferroelectric materials
order transition, as in BaTi0 3 and KNb0 3 . See under the electrically biased condition.18'20 In
Figure 4-9(a) and (c). In this case, the entropy general, if the direction of the applied field is
changes at a constant temperature (e.g., T= Tc\ parallel to the polar direction of any phase, it
and consequently the latent heat also changes. favors the existence of that phase and extends
If the change of the polarization is continuous, its temperature range. At Tc, the ferroelectric
the transition is referred to as a second-order phase is favored by an applied field; therefore,
transition, as in KH 2 P0 4 and Rochelle salt. See the Curie temperature is raised.
Figure 4-9(b) and (d). In this case, there is no A change in crystal structure and crystal
change in entropy and latent heat at the transi- volume caused by any external forces, such as
tion temperature. It is interesting to note that at electric fields, mechanical stresses, or hydro-
the Curie temperature, the transition in ferro- static pressures, will affect the phase transition.
magnetic materials does not involve changes in A shear stress is essentially equivalent to an
crystal structure but only a small change in the electric field along the ferroelectric axis and

T lowering

T returning up

(c) (d)

Figure 4-9 Schematic diagrams illustrating the effect of applied electric field on (a) first order transition and (b) second
order transition, and the shift of the transition point when the temperature is returning up (increasing), (c) Tc shifts to a
higher temperature for the first order transition, and (d) there is no shift for the second order transition.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 221

therefore would have a similar effect on the • Alloys of lead, zirconium, and titanium
polarization-temperature curve, as shown in oxide (abbreviated PZT, alloys of PbO,
Figure 4-9. It should be noted that in Equation Zr0 2 , and Ti0 2 )-type ferroelectrics
4-1 the Curie temperature Tc coincides with the • Polyvinylidene fluoride (abbreviated PVDF,
actual transition temperature at which sponta- -(CH2-CF2)„-type ferroelectrics.
neous polarization begins to rise if the transition
is of the second order. For first-order transitions,
Tc is generally slightly lower than the actual BaTi03-Type Ferroelectrics
transition temperature by several degrees (typi- BaTi0 3 belongs to the family of AB03 Perov-
cally -10 K). Because the difference is very skite mineral (CaTi03) structures, in which A
small, and for general uses, Tc is usually con- and B are metals. The total charge of the A and
sidered the actual transition temperature. B positive ions must be +6, and A and B must
When the temperature decreases through Tc, be of quite different sizes; the smaller ion, with
the polarization rises from zero continuously a larger charge, must be a transition metal. For
over several degrees of temperature if the tran- BaTi0 3 , Ti is a 3d transition element and has
sition is of the second order. However, when the the d orbital for electrons to form covalent
polarization rises abruptly from zero at Tc fol- bonds with its neighbors. The radius of Ti4+
lowing the first-order transition, then the rise ion is about 0.68 A, and that of Ba2+ is about
involves a change of the latent heat, implying 1.35 A. These ions form nice octahedral cages,
that the process is equivalent to supercooling. with the O 2- ions held apart. At temperatures
When the temperature is reversed—that is, higher than the Curie temperature (>120°C),
increases through the transition point—the Ti4+ stays in the cage, rattling around it to make
process becomes superheating, and a collective the unit cell maintain a symmetrical cubic
or cooperative action among the dipoles and structure, as shown in Figure 4-10(a).
surrounding ions tends to hold the spontaneous However, the structure of the unit cell is tem-
polarization until a slightly higher temperature perature dependent. At a certain transition tem-
is reached, as shown in Figure 4-9(c). A similar perature, the particular structure of the unit cell
phenomenon occurs at any transition point if becomes unstable and must transform to a more
the transition is of the first order. (See Figures stable one. So, at the Curie temperature Tc, the
4-11 and 4-12.) But this phenomenon does not octahedral cages distort and the positive ions
occur if the transition is of the second order, as move to off-center positions. The crystal takes
shown in Figure 4-9(d). a tetragonal form, resulting from the stretching
of the cubic unit cells along one edge, as shown
in Figure 4- 10(b). In fact, the Ba2+ ions shift
0.05 A upward from their original position in
4.2.2 Phenomenological Properties
the cubic structure; Ti4+ ions shift upward by
and Mechanisms
0.1 A, and the O2" ions downward by 0.04 A to
This section discusses the phenomenological form the tetragonal structure. As a result of the
behaviors and their mechanisms of six typical ion shifts, the centroid of the positive charges
ferroelectric materials: no longer coincides with the centroid of the
• Barium titanate (BaTi03)-type ferroelectrics negative charges; therefore, the unit cells
become permanently polarized and behave as
• Potassium dihydrogen phosphate (abbrevi-
permanent dipoles, leading to spontaneous
ated KDP, KH 2 P0 4 )-type ferroelectrics
polarization.
• Potassium-sodium tartrate tetrahydrate The direction of the displacement can be
(Rochelle salt, KNaC4H406*4H20)-type reversed by a sufficiently high electric field of
ferroelectrics opposite polarity. This possibility of dipole
• Triglycine sulphate (abbreviated TGS, reversal distinguishes ferroelectric materials
(NH2CH2COOH)3H2S04)-type ferroelectrics from nonferroelectric ones. At temperatures
222 Dielectric Phenomena in Solids

p
a m3o
• o•
• ; * &

o Ba O Ti

(a) The unit cell of BaTi03 (b) The ion displacements due
to the cubic-tetragonal
distortion in BaTi0 3

/ ^ /

a
aJ
a

Rhombohedral, Stable Orthorhombic, Stable Tetragonal, Stable Cubic, Stable

below -90°C between 5°C & -90°C between 120°C&5°C above 120°C

(c) The temperature dependence of the structure of the unit cell

Figure 4-10 Schematic diagrams showing (a) the unit cell of BaTi03, (b) the ion displacement due to the cubic-
tetragonal distortion in BaTi0 3 , and (c) the temperature dependence of the structure of the unit cell.

below Tc and between 120°C and 5°C, the always along the direction of the unit cell's
structure is tetragonal and the polar axis (i.e., elongation, that is, the stretching direction. This
the direction of the spontaneous polarization) is also referred to as the ferroelectric polar axis.
is along the c-axis of the unit cells, in which BaTi03-type ferroelectrics are nonpiezoelec-
c > a. At about 5°C, the tetragonal unit cells tric in the unpolarized state. The cubic symme-
undergo a transition for a higher stability to the try implies that, depending on the temperature,
orthorhombic structure, which is formed by spontaneous polarization may occur along
stretching the cell along the face-diagonal several axes. For example, there is a set of
direction with the polar axis also along the three (100) axes, a set of six (110) axes, and a
same direction. This structure will remain set of four (111) axes. The three sets are non-
stable for temperatures between 5°C and equivalent.
-90°C. Similarly, at around -90°C, a rhombo-
hedral structure, formed by stretching the unit
cell along the body-diagonal direction, Dielectric Constants
becomes preferred, as shown in Figure 4-10(c). Dielectric constants are usually measured
The direction of spontaneous polarization is with small AC signals. At low frequencies,
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 223

piezoelectric deformation can follow the l-Rhombohedral-f-Orthorhombic -|— Tetragonal —(-Cubic

change of the applied electric field, and the 10,000

crystal can be considered mechanically uncon-
strained. In this case, the dielectric constant
measured is under the free crystal condition.
However, at frequencies higher than the piezo-
electric resonance frequency of the crystal,
piezoelectric deformation cannot follow the
change of the applied AC field due to inertia
effects, and the dielectric constant measured is -200 -160 -120 -80 -40 0 40 80 120
under the clamped crystal condition. In non- Temperature (°C)
piezoelectric crystals, such as BaTi0 3 (which is
nonpiezoelectric in the unpolarized state), the Figure 4-11 Dielectric constant of BaTi0 3 as a function
of temperature measured along the c-axis and along the
dielectric constant is practically independent of
a-axis (perpendicular to the c-axis).
frequency at temperatures near or higher than
Tc. But in piezoelectric crystals, such as KDP a multiaxial ferroelectric, and its basic mole-
and Rochelle salt, the dielectric constant is cular structure, as shown in Figure 4-10(a), is
strongly frequency-dependent and may differ an octahedral arrangement of oxygen ions
by several orders of magnitude, depending on around a central Ti4+ ion. Above the Curie point,
the frequency used for the measurement. the 774+ ion is rattling around from its symmet-
Obviously, the frequency dependence of er of rical position. Each shift will create a dipole
BaTi0 3 is expected to be different in the various moment along that shifting direction, but
phases. In the tetragonal phase, en measured this ion can also shift in the opposite direction,
at low frequencies under the free crystal con- due mainly to thermal agitation, which is
dition, is expected to be higher than that randomized.
measured at frequencies higher than the piezo- However, when a small AC signal field is
electric resonance frequency under the clamped applied to the crystal along one axis for the
crystal condition. At still higher frequencies, an measurement of en the ions will be inclined to
additional drop in er may occur, possibly due to shift along the direction of the applied field,
dielectric relaxation. In general, the frequency thus producing net dipole moments, provided
range used for measurements of er is from zero that the driving force from the field is larger
to microwave frequencies in which the crystal than the thermal agitation force. This can
can still be considered a free crystal. The happen only at a temperature close to, but
dielectric constant-temperature (e-T) charac- above, the Curie point. This shift obviously
teristics for BaTi0 3 under the free crystal con- depends on the local field and the temperature
dition are shown is shown in Figure 4-11. The the ions experience. The local field, derived on
data are from Merz.1718 The special features the basis of Lorentz's internal field, is for non-
of this e-T relationship are discussed as polar materials (see Internal Fields in Chapter
follows. 2). At T > Tc, however, BaTi0 3 would become
For temperatures above the Curie tempera- polar, due to the shifting of the ions under an
ture Tc, the crystal assumes a cubic structure applied field. This mechanism is different from
and is in the paraelectric unpolarized state. The atomic polarization in ionic crystals and also
dielectric constant decreases with increasing from the orientational polarization in dipolar
temperature, following closely the Curie-Weiss materials in which permanent dipoles are
relation given by Equation 4-1. X-ray analysis present. So far, there is no satisfactory method
has indicated that the dipole moment is associ- available to calculate the local field and the
ated with the distortion of the molecular system average dipole moment in ferroelectric materi-
and the cooperative alignment of the dipoles, als for T > Tc. By assuming that the ionic shift-
resulting in spontaneous polarization. BaTi0 3 is ing in producing dipole moments is equivalent
224 Dielectric Phenomena in Solids

to the dipole orientation (although in fact, they field) could be very rewarding. A thermody-
are quite different), we can borrow Debye's namic approach to this phenomenon will be
formula originally derived for dipolar materi- discussed in Section 4.2.3.
als. Then, the average dipole moment per mol- At Tc, the dielectric constant for BaTi0 3 is
ecule in the direction of the applied field can be typically of the order of 10,000, and the ions
written as are moving into a difference phase, correspon-
ding to spontaneous polarization. Conse-
<Mr> = ^ o C a l (4-2) quently, an applied electric field will be able to
produce relatively large shifts. This action
where \i0 is the dipole moment of the perma- results in a big change in dipole moment, cor-
nent dipole in the material and Fiocai is the local responding to a high dielectric constant. When
field. Again, we borrow the formula for local T is lowered just below Tc, the structure of the
field based on Lorentz's internal field, which is unit cell changes from the cubic form to a
given by tetragonal one, as shown in Figure 4-10(b) and
(c). Once the structure begins to change or
F**=*±1F (4-3) spontaneous polarization to arise, the net dipole
moment in the direction of the applied AC field,
Since the polarization is given by and hence the dielectric constant, will start to
decrease, as shown in Figures 4-11 and 4-12.
P = (er-l)e0F = N(iiiF) (4-4)
At T < Tc, the dielectric constant is strongly
where N is the number of molecules per unit anisotropic in the ferroelectric phase for all
volume, then from Equations 4-2 through 4-4 we crystals. For multiaxial ferroelectrics such as
obtain the famous Clausius-Mossotti equation: BaTi0 3 , the dielectric behavior, which is
isotropic in the cubic nonpolarized phase,
becomes strongly anisotropic in the polarized
phase. The dielectric constant measured along
Rearrangement of Equation 4-5 yields the the principal ferroelectric axis (oaxis in the
dielectric susceptibility tetragonal structure) is lower than that normal
to it (i.e., a-axes). This big difference can be
j 3N(tio)2/9e(>k _ 3Td
r observed only in single-domain crystals.1718
* T - N(ju0)2/9e„k T-Td
This is readily understandable because the
(4-6)
spontaneous polarization along the c-axis is sat-
because er » 1. This equation is similar to urated, while the polarization along the a-axes
Equation 4-1. This indicates that er diverges as is not. For uniaxial ferroelectrics, such as KDP
T approaches 7C, and that the system becomes
unstable and must make a transition to a new
C 301 1 1 1 1 1 1 1 1 1 1 1 r
phase. The divergence of er at T = 7C, based on
Equation 4-6, is generally referred to as the
Clausius-Mossotti or polarization catastrophe.
It is important to note that Td = N(ju0)2/9£0k is
definitely not Tc and cannot be used to evaluate
Tc, nor can 3Td be used to estimate the Curie
constant C for ferroelectric materials, simply
because for T > Tc, there are no permanent
dipoles as such in BaTi03. It appears that Temperature (°C)
research in the derivation of the Curie-Weiss
equation based on the ion-shifting mechanism Figure 4-12 Spontaneous polarization and coercive field
(which is similar to ionic polarization but not of a BaTi0 3 single crystal as functions of temperature
quite the same and has a different form of local measured along the tetragonal [001] direction (c-axis).
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 225

and Rochelle salt, the dielectric constant meas- Landauer,23 the rate of the reversal of the polar-
ured along the principal ferroelectric axis is ization P can be expressed as
much larger than that normal to it, simply
because the ions shift only along one principal ^ = f(P)cxp[-a/F(t)] (4-7)
ferroelectric axis and do not shift in any direc- at
tion other than the principal axis. For example, where the function f(P) is related to the switch-
for KDP, the dielectric constant along the prin- ing rate. For most hysteresis loop measure-
cipal axis is about 1100, while that normal to it ments, the applied field varies sinusoidally with
is only 12. For Rochelle salt, the corresponding time (i.e., F(t) = F0sin($t), and a is a parame-
dielectric constants along and normal to the ter depending on temperature. Equation 4-7
principal axis are 1000 and 10, respectively. implies that the coercive field depends on the
In general, pure crystal specimens have a rise rate of the applied field. For slow rates, the
lower Tc than the impure ones. Also, the elec- crystal can have enough time to reverse its
trical conductivity of a material has a screening polarization so that the reversal can be com-
effect on the cooperative force that leads to the pleted before the peak field is reached. For fast
ferroelectric behavior. Other ferroelectric mate- rates, however, the crystal may not have suffi-
rials, which are similar to BaTi0 3 in their chem- cient time to complete the reversal process. In
ical structure (PbTi03, KNb0 3 , KTa0 3 , etc.), this case, a larger Fc is needed to bring Ps to
have similar dielectric and structural properties zero. In other words, Fc is not a constant phys-
to those of BaTi0 3 , except that their Curie tem- ical parameter; it depends on both the value of
peratures and Curie constants are different. F0 and the frequency co of the applied field.24,25
The coercive field depends also on the thick-
ness of the ferroelectric specimens. The thicker
Spontaneous Polarization and Coercive Field the specimen, the smaller the coercive field.
The temperature dependence of the sponta- The experimental results follow the thickness-
neous polarization Ps and the coercive field Fc dependence relation26
for BaTi0 3 single crystals is shown in Figure
Fc=Fco + G/d (4-8)
4-12. The data are from Merz.17,18 The meas-
urements were made along the c-axis of the where Fco is the coercive field for very thick
tetragonal structure, that is, along the (001) specimens, d is the specimen thickness, and G
direction.18'21 The Ps value along the (110) is a constant. Based on the experimental results
face-diagonal direction of the orthorhombic for BaTi0 3 , G turns out to be 2.2 V. Polariza-
structure can be obtained by multiplying the tion reversal is a process of nucleation and
corresponding value along the c-axis by V2". growth of antiparallel domains, which is thick-
Similarly, the Ps value along the (111) body- ness dependent.22,27 The coercive field can be
diagonal direction of the rhombohedral struc- considered the field at which the reversal
ture can be obtained by multiplying the process can proceed easily.
corresponding value along the c-axis by V3. Before ending this section, it is worth men-
The transition at Tc is of the first order for tioning that upon repeated switching of the
single, pure BaTi0 3 crystals. However, if the polarization of the crystals, notably BaTi0 3 ,
crystals are not pure, the expected first-order spontaneous polarization becomes clamped or
transition may become a second-order transi- decreases, and the coercive field increases. This
tion, and the value of Ps would also be lowered, phenomenon is referred to as the fatigue or
due to the hindering of the motion of domain decay of the crystal,28,29 and it has been
walls by impurities or other defects. attributed to the build-up of the space charge
The coercive field Fc is temperature depend- near the crystal surfaces and the interaction of
ent, as shown in Figure 4-12. Fc depends also this space charge with domain walls. The
on the time required for the polarization to degree of the fatigue depends on several factors
reverse its direction. According to Merz22 and including the ambient atmosphere, the elec-
226 Dielectric Phenomena in Solids

trode material, and the waveform of the applied the intrinsic conductivity at room temperature
switching field. The decay is more rapid with is thousands of millions of times smaller than
pulsed fields than with low-frequency AC the real electrical conductivity of BaTi03,
fields.28-30 Ferroelectric ceramics exhibit also which is of the order of 10-14 (ohm-cm)-1 and
fatigue behavior. Apart from the factors previ- of the n-type. So, the contribution from the
ously mentioned, pores, impurities, and micro- intrinsic condition is negligible and the elec-
cracking may also play important roles in this tronic conductivity must be extrinsic (see
phenomenon. Electronic Conduction in Chapter 7).
The displacement current observed during For this finite n-type electronic conductivity,
the reversal of spontaneous polarization is often we must find the source of the electrons. The
accompanied by transient pulses. Such noise- mechanism of electrical conduction for BaTi0 3
like pulses were first observed in ferromagnetic is similar to that for titanium oxide Ti0 2 (rutile)
materials by Barkhausen in 1919, and the or other oxides, such as ZnO, etc. It is almost
analog of this phenomenon in ferroelectric impossible to obtain stoichiometric BaTi03.
materials still bears his name. The processes of Crystal specimen fabrication processes are
domain nucleating or coalescing produce sharp never perfect. There is always a possibility for
discontinuities in the change of polarization. the four-valent Ti4+ ions to reduce to the three-
These are picked up from the electrode as valent Ti3+ or even to the two-valent Ti2+ ions.
Barkhausen pulses. This phenomenon has been Each titanium atom can give four electrons to
studied by many investigators.31-33 In BaTi0 3 , oxygen atoms, and each oxygen atom can take
there are typically 105 to 106 pulses during a only two electrons. If the oxide is oxygen-defi-
complete polarization reversal, and the average cient (or metal-rich), then not all the titanium
pulse height is of the order of 10"12 coulomb. atoms have shed four electrons. Some must be
The total number of pulses of all sizes does left in a lower valency state of Ta3+ or Ta2+ in
not vary with the applied field, but pulse order to maintain an overall neutrality of the
heights increase with increasing field. Similar system. This situation results in a few Ta3+ or
features of this phenomenon appear in other Ta2+ in a sea of Ta4+, and a few of oxygen vacan-
ferroelectric materials, including TGS, cies in the lattice of O 2- ions. The Ta3+ or Ta2+
Rochelle salt, etc. Experiments have revealed ions act like electron donors, ready to provide
that Barkhausen pulses are associated with the electrons for electronic conduction, while O 2-
nucleation and growth of domains. However, ions may also diffuse via the vacancies for ionic
this phenomenon provides an additional avenue conduction. So, the electrical conductivity con-
for the study of polarization reversal processes sists of electronic conductivity cn and ion con-
under various conditions. ductivity (Jj. Thus,
G = Gn+Gj= q[inn + qUiiti (4-9)

Electrical Conductivity and where un and n are, respectively, the mobility

Breakdown Strength and the concentration of electrons; and ut and
In oxide crystals, bonds are formed by the nt are, respectively, the mobility and the con-
complete transfer of the electrons from the centration of vacancies.
metal atoms to the nonmetal atoms, such as The electrons on Ta3+ or Ta2+ ions may hop
oxygen atoms. Each atom then becomes an ion to Ta4+ ions, contributing to an extrinsic n-type
with an inert gas atomic structure. These ions conduction. For pure BaTi0 3 single crystals,
are so stable that it is very difficult to remove electronic conduction is dominant for tempera-
electrons from them or create holes in them. tures up to 500°C, but ionic conduction
This implies that their energy band gaps are becomes significant at about 100°C. It is inter-
large. The energy band gap for the Perovskite esting to note that the activation energy for
oxides, such as BaTi03, is around 3 to 4eV. Even electronic conduction in BaTi0 3 is consider-
assuming electron mobility of lOOcirr^VV1, ably smaller in the polarized state than in the
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 227

paraelectric state, possibly due to the big dif- axis (up or down along the c-axis) and undergo
ference in dielectric constant, polarization, and only one phase transition from a paraelectric
lattice parameters between these two states. into a spontaneously polarized state. The tran-
It should be noted that there is also a possi- sition does not show a thermal hysteresis,
bility of the oxide being oxygen-rich (or metal- implying that the transition is of the second
deficient). In this case, Ta4+ can give four order. In the paraelectric unpolarized state, the
electrons to oxygen atoms, but not all oxygen crystal assumes a tetragonal structure at T above
atoms can have two electrons each; some Tc, with the ferroelectric axis along the c-axis.
oxygen atoms can get only one each and The crystal is piezoelectric because it belongs
become CT ions. The 0~ ions act as acceptors, to the noncentrosymmetric point group. At T
creating holes in the crystals. These holes can below Tc, an orthorhombic phase exists, which
migrate from O" to O2" (equivalent to electrons is ferroelectric with spontaneous polarization
hopping from 02~ to O"). Cations may also due mainly to the displacement of K, P, and O
diffuse via vacancies, but the activation energy ions in the direction of the polar c-axis.34 The
for cation diffusion in oxide crystals is always structure of KDP is quite complicated. Basi-
high. Thus, the extrinsic p-type conduction is cally, it consists of two interpenetrating body-
dominant. However, both electronic conduction centered lattices of P0 4 tetrahedra and two
(n-type or p-type) and ionic conduction occur interpenetrating body-centered K lattices, with
the P0 4 and K lattices separated along the
in BaTi0 3 ; the ratio of —- depends strongly c-axis. Every P0 4 is linked to four other P 0 4
Oi
on the purity and the fabrication process of the groups by hydrogen bonds, which lie perpendi-
material, as well as the temperature. cular to the c-axis. This arrangement gives rise
The breakdown strength of BaTi0 3 single to an ionic configuration of K+(H2P04)". The
crystals is of the order of 0.5MVcm_1. For distribution of proton density shifts from the
BaTi0 3 ceramics, it is about 0.1 MVcm"1. Gen- center of the bond at T above Tc to a localized
erally, ceramics have a lower breakdown asymmetric position below Tc. The direction of
strength, partly because of their porosity. It this shift reverses when the spontaneous polar-
should be noted that a high poling field may ization is reversed.34'35
sometimes cause electrical breakdown, because The dielectric constant-temperature charac-
the poling time is usually quite long (e.g., one teristics are shown in Figure 4-13, the data
hour). High field stressing coupled with long being from Busch,36 and the spontaneous polar-
stressing time may be sufficient to overcome the ization-temperature characteristics in Figure
interdomain stresses in BaTi03. If impurities,
pores, and other defects are present in speci- 106r
mens, as in ceramics, the breakdown strength
will be strongly dependent on specimen thick- 10 5 h
ness, electrode area, and temperature. For more
details about the breakdown processes, see
Electrical Breakdown in Chapter 8.

KH2P04-Type Ferroelectrics
Potassium di-hydrogen phosphate (KH2P04) is
generally referred to as KDP crystals, with a 1 I i i i i i
50 100 150 200 250 300
tetragonal structure at room temperature. Ferro- Temperature (°K)
electrics of the KH 2 P0 4 type exhibit a much
simpler behavior than other ferroelectrics Figure 4-13 Dielectric constants of KH 2 P0 4 as a function
because they belong to the uniaxial ferroelectric of temperature measured along the c-axis and along the
family. These materials polarize along only one a-axis.
228 Dielectric Phenomena in Solids

4-14, the data being from Von Arx and Bantle.37 KNaC4H40^4H20 (Rochelle Salt)-Type
The Curie temperature Tc is +150°C (123 K). At Ferroelectrics
T > Tc, the dielectric constant decreases with Rochelle salt is a potassium-sodium tartrate-
temperature, following closely the Curie-Weiss tetrahydrate (KNaC4H406»4H20) and belongs
law with the Curie constant C being 3250°C.36'37 to the family of uniaxial ferroelectrics. This
At T < Tc, the dielectric constant along the c- was the first material in which Valasek discov-
axis drops sharply (but not abruptly) due to the ered the ferroelectric phenomenon in 1921, and
dielectric saturation effect, which results in a from which he also introduced the term Curie
decrease of dielectric constant with increasing temperature or Curie point for ferroelectrics.
spontaneous polarization. The dielectric con- It is interesting to describe briefly the ferro-
stant along the a-axes also shows an anomaly at electric properties of this material. Rochelle salt
7C, possibly due to the effect of dielectric coop- has a rather poor mechanical strength and low
erative action among the ions.38 The dielectric disintegration temperature; it is also apt to
constant along the oaxis is dependent on the absorb water. The crystal decomposes at a low
applied DC biasing field. The bias tends to temperature of about 55°C. However, the most
lower the dielectric constant and also to shift the interesting property of Rochelle salt is that it
peak of the dielectric constant toward higher exhibits two Curie points, one at +24°C, called
temperatures.39 the upper Curie temperature, and the other at
Figure 4-14 shows that the spontaneous -18°C, called the lower Curie temperature. The
polarization begins to rise at Curie point con- nonpolar paraelectric phases for T> +24°C and
tinuously, so the transition is of the second for T < -18°C are orthorhombic. The polar fer-
order. The coercive field Fc is temperature roelectric phase, occurring at the temperature
dependent. It is interesting to note that at a tem- range between -18°C and +24°C, is mono-
perature of about -213°C, (60 K) there is a clinic. The dielectric constant, measured along
sharp break. At 120K > T > 60 K, Fc is practi- the a-axis with a small signal field of 1 kHz as
cally independent of temperature. But at T < a function of temperature, is shown in Figure
60 K, Fc increases rapidly with decreasing tem- 4-15.
perature.40 This phenomenon is probably There is no thermal hysteresis at the transi-
caused by the hindered motion of domain walls tion points, indicating that the transition is of
at low temperatures. the second order. For temperatures from +24°C
There are a number of crystals isomorphous
with KH 2 P0 4 that have a similar dielectric
behavior, including KH2As04, RbH2P04, 10 4
CsH2P04, etc. In this KDP family, hydrogen
bonds play a vital role in the ferroelectric
behavior. 10 3

10 2

10

100 105 110 115 120 125 -200 -160 -120 -80 -40 40
Temperature (°K) Temperature (°C)

Figure 4-14 Spontaneous polarization of KH 2 P0 4 as a Figure 4-15 Dielectric constant of Rochelle salt as a
function of temperature. function of temperature.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 229

to about +50°C, the dielectric constant obeys the stable ferroelectric phase and giving a
the Curie-Weiss law, with the Curie constant C higher spontaneous polarization, as shown in
2240°C. For the temperature range from -18°C Figure 4-16.41'42
to -28°C, the dielectric constant again follows
the Curie-Weiss law, with a Curie constant of
(NH2CH2COOH)3*H2S04 Type
1180°C. The decrease of the dielectric constant
Ferroelectrics
in the ferroelectric region is due to the effect of
spontaneous polarization saturation, similar to Triglycine sulphate (TGS) is a uniaxial ferro-
BaTi0 3 and KH 2 P0 4 . The temperature depend- electric material, and its chemical formula
ence of spontaneous polarization is shown in is (NH2CH2COOH)3*H2S04. This material
Figure 4-16. The coercive field for Rochelle exhibits ferroelectric properties at room tem-
salt depends on the specimen thickness and the perature. The Curie temperature is 49°C. At
magnitude and the frequency of the applied T > Tc, the crystal structure is centrosymmetric
field, similarly to other ferroelectric materials. and in the nonpolar paraelectric state with a
The data given in Figures 4-15 and 4-16 are monoclinic symmetry. At T < Tc, the crystal
from Hablutzel.41 becomes polar and belongs to the polar group
Rochelle salt is like KDP; the hydrogen bond of the monoclinic system.39 Ferroelectric
plays an important role in the ferroelectric direction is along the b-axis and the transition
behavior. This idea has led to the study of a is of the second order. The dielectric con-
deuterated Rochelle salt, prepared by dissolv- stant-temperature characteristics measured at
ing ordinary Rochelle salt in highly concen- 1kHz and IV cm-1 are shown in Figure 4-17.
trated D 2 0 to make it thoroughly desiccated at The dielectric constant along the polar b-axis
higher temperatures, in order to replace hydro- shows a prominent anomaly in the vicinity of
gen with deuterium, the isotope of hydrogen Tc. The dielectric constants along the a- and c-
(heavy hydrogen consisting of a neutron and a axes are practically independent of tempera-
proton in the nucleus). After this process, the ture, simply because this is a uniaxial polarized
Rochelle salt becomes a deuterated compound material. For T> Tc, er follows the Curie-Weiss
with a chemical formula of KNaC4H2D206. law with a Curie constant C of 3200°C. The
4D 2 0. This compound has the Curie points at temperature dependence of the spontaneous
-22°C and +35°C, thus widening the region of polarization is shown in Figure 4-18. The
experimental data are from Hoshino et al.43
Some crystals are isomorphous with TGS, such
as triglycine selenate (NH2CH2COOH). H 2 Se0 4

b-axis

-120 -80 -40 0 40

-30 -20-10 0 10 20 Temperature (°C)
Temperature (°C)
Figure 4-17 Dielectric constant of triglycine sulfate as a
Figure 4-16 Spontaneous polarization of Rochelle salt function of temperature measured along a-axis, b-axis, and
and deuterated Rochelle salt. c-axis.
230 Dielectric Phenomena in Solids

4.5 r is shown in Figure 4-19.44 The Curie tempera-

• Triglycine Fluoberyllate ture Tc depends on the composition of the
o Triglycine Sulfate
A Triglycine Selenate
system.
In this phase diagram, there are six major
phase boundaries:
1. The cubic paraelectric-tetragonal ferro-
electric (P-FT) phase boundary
2. The cubic paraelectric-rhombohedral fer-
roelectric (HT) (P-FR(HT)) phase boundary
3. The tetragonal ferroelectric-rhombohedral
ferroelectric (HT) (FT - FR(HT)) phase
boundary
4. The rhombohedral ferroelectric (HT)-
rhombohedral ferroelectric (LT) (FR(HT)-
0 20 40 60 80
FR(LT)) phase boundary
Temperature (°C)
5. The rhombohedral ferroelectric (HT)-
Figure 4-18 Spontaneous polarization of triglycine orthorhombic antiferroelectric (FmJTrAF^)l
sulfate, triglycine selenate, and triglycine flueboryllate as a phase boundary
function of temperature.
6. The rhombohedral ferroelectric (LT)-
orthorhombic antiferroelectric (FR(LT)-AF0)
phase boundary
(abbreviated TGSe) and triglycine fluoberyllate
(NH2CH2COOH)3H2BeF4 abbreviated TGFB), The phase is cubic paraelectric, implying that
have similar properties to TGS with different an applied field can distort the system and
Curie temperatures, as shown in Figure 4-18. change it from unpolarized to a polarized state,
but as soon as the field is removed, the system
will quickly relax back to its original unpolar-
Alloys ofPbO, Zr02, and Ti02 (PZT alloys)- ized state due to thermal agitation. Figure 4-19
Type Ferroelectric Ceramics has some prominent features worth mentioning.
PZT alloys belong to the family of ceramics,
which are alloys of lead oxide, zirconium • The Curie temperature Tc for all composi-
oxide, and titanium oxide (PbO, Zr0 2 , and tions is higher than that for BaTi03. The
Ti0 2 ). The familiar Perovskite structure of main reason is that the Pb2+ ion has two outer
BaTiOi also appears in PZT alloys because electrons beyond its last full shell, but the
there are equal numbers of divalent (Pb2+) and Ba2+ ion does not. These outer electrons can
tetravalent (Zr4+ and Ti4+) cations. PZT alloys contribute to covalent bonding with neigh-
are in fact the alloys of two components: boring oxygen ions. With this extra bonding,
PbZr0 3 and PbTi03. This is why PZT alloys are a greater thermal energy or a higher temper-
similar to BaTi0 3 in their ferroelectric proper- ature is required to convert the polarized
ties. Thus, by controlling the composition and state into an unpolarized state. This results
microstructure of the PZT alloys, it is possible in a higher Tc for the PZT system.
to tailor their properties to suit particular • The morphological boundary between the
applications. The general chemical formula is tetragonal and the rhombohedral phases
PbZri_^Ti^03, with x as a fraction from 0 to 1. occurs at x around 0.5. Since the PZT system
The phase diagram of ferroelectric ceramics is is used mostly in piezoelectric devices, the
normally plotted in phase transition tempera- PZT system with the composition near the
ture as a function of the composition. A typical morphological boundary has an advanta-
equilibrium phase diagram for the PZT system geous crystal morphology. The efficiency
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 231

PbZr03 Percentage of PbZr03 in PbZr03-PbTi03 System

100 90 80 70 60 50 40 30 20 10 0
500 F

[220h \ »

p
s IJ
Tetragonal, Stable

FT

Orthorhombic, Stable

0 10 20 30 40 50 60 70 80 90 100
Percentage of PbTi03 in PbZr03-PbTi03 System PbTi03

Figure 4-19 The equilibrium phase diagram of the PZT system and microstructure of the unit cell for different phases.

of the conversion between electrical and • The Curie temperature in pure PbZr0 3 is
mechanical energy depends strongly on 230°C. Below this temperature, the crystal
crystal morphology. The piezoelectric cou- is in an antiferroelectric state, with the struc-
pling coefficient climbs to a maximum value ture changing to orthorhombic one. But at
at x around 0.5 because the shear transfor- or near the Curie temperature, the anti-
mation between the two types of structures is ferroelectric phase assumes a tetragonal
much easier. This is because there are six structure. The mechanism that causes this
easy directions of polarization for the difference in structure at or near Tc is not
tetragonal unit cell and eight easy directions certain; possibly it is due to the presence of
of polarization for the rhombohedral unit foreign impurities.47
cell, which provide much easier mutual
It is interesting to review the ferroelectric prop-
conformation.
erties of lead zirconate PbZr0 3 in the PZT
• For most PZT alloys, the rich Zr region with system with x = 0. The dielectric constant-
x below 0.1 is attractive because they have temperature relation for PbZr0 3 ceramic is
superior properties. In this region, the Curie shown in Figure 4-20. The dielectric constant
temperature Tc is lower. At T < 200°C, there reaches a peak at the Curie temperature 230°C.
exists a boundary between the ferroelectric Above Ta the structure is cubic and er follows
phase and the antiferroelectric phase. the Curie-Weiss law, with the Curie constant C
• The rhombohedral ferroelectric phase is sub- equal to 1.6 x 105 degrees. The results are from
divided into high temperature FR{HT) and low Roberts.48 Although the dielectric anomaly is
temperature FR(LT) phases. Simple rhom- similar to that in BaTi0 3 , no hysteresis loop has
bohedral unit cells appear in both high- been observed for T < Tc, implying that there is
temperature and low-temperature regions. no net spontaneous polarization. This indicates
The basic difference between FR(HT) and FR{LT) that the phase for T < Tc is antiferroelectric.49
is possibly due to the difference in the direc- In other words, all adjacent domains are oppo-
tions of the spontaneous polarization.45'46 sitely polarized, so there is electrical ordering
232 Di«ielectric Phenomena in Solids

60

3000

T 40
E
o
2000
U° 20

w IX
1000
200 210 220 230
Temperature (°C)

Figure 4-21 The critical electric field Fcrit required to

100 200 300 induce ferroelectricity in PbZr0 3 as a function of tempera-
Temperature (°C) ture. Insert: the occurrence of a double hysteresis loop at
the critical field.
Figure 4-20 Dielectric constant of ceramic PbZr0 3 as a
function of temperature.
to produce a powdered mixture of very fine
particle size. The chemical substance is then
but no net polarization (see Sections 4.2.3 and formed by calcining (fiercely heating) this
4.2.5). mixture. This material is then ground up and
However, in the vicinity of the transition artifacts made by compacting and sintering.
temperature, a strong electric field can induce a Ceramics are quite different from single
ferroelectric phase. This means that the field crystals. Ceramics are formed by many ran-
can shift the phase boundary between AF0 and domly arranged crystallites (grains) with
FR, with the tendency of extending the ferro- unavoidable microvoids and impurities trapped
electric region. A double hysteresis loop for in between. In fact, they are imperfect poly-
pure PbZr0 3 has been observed at temperatures crystalline materials. So a poly crystalline
below Tc.50 The ferroelectricity is induced in the ceramic that has not been subjected to static
antiferroelectric phase when the applied elec- field stressing would behave like a nonpolar
tric field reaches a critical value. A similar material, even though the crystallites compos-
double hysteresis loop occurs in BaTi03, but in ing it are polar. However, ferroelectric ceram-
that case the ferroelectricity is induced in the ics can be transformed into polar materials by
cubic paraelectric unpolarized region by a suf- an externally applied static electric field. This
ficiently high electric field and at a temperature process is referred to as the poling process,
slightly above Tc. Obviously, the origins of this which is reversible. To pole a specimen, the
double-loop behavior for these two cases are specimen must be brought to the saturation of
different. In PbZr03, the ferroelectric phase the spontaneous polarization by a strong field
induced by the field has rhombohedral symme- and then allowed to relax back to the remanent
try. The critical field is temperature dependent, condition. Poling is usually carried out at tem-
as shown in Figure 4-21. The data are from peratures as hot as the Curie threshold temper-
Sawaguchi and Kittaka.50 ature allows, and this poling condition is held
It is difficult, if not impossible, to produce a for some hours. However, the Curie tempera-
PZT alloy P b Z r ^ T i A with 0 < x < 1 in single ture is much lower than the ceramic sintering
crystal form. The PZT alloys are ceramics, temperature, at which decent rates of transport
which are a solid solution. Generally, the three between grains can occur, so it is inevitable that
oxides—PbO, Zr0 2 , and Ti02—of selected poled ceramics always carry some residual
purity and proportion, are ball-milled together internal stress. A poled ceramic can be depoled
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 233

by the application of electric fields or mechan- larization with time. However, the magnitude
ical stresses. of the poling field and the poling temperature
A ceramic system consists of randomly orig- are important factors in determining the extent
inated grains (crystallites) intimately bonded of the dipole alignment and hence the resulting
together with grain sizes generally ranging ferroelectric properties. In actuality, the align-
from l/Mn to 20 /urn. There are always pores ment is never complete. Depending on the type
(microvoids) and foreign impurities unavoid- of the crystal structure and the poling condition,
ably present in the material during the fabrica- the thoroughness of poling can be quite high,
tion process. It is obvious that both porosity and approaching 83% (for the tetragonal structure)
grain size affect dielectric and ferroelectric and 91% (for the orthorhombic structure) of
properties. The important factor leading to the that for single crystals.51 The aging process
effect is the variation in the dimensions of the may be accelerated by exposing the poled
grains and hence in the internal stresses during ceramics to
the change of temperature through the Curie
• Strong depoling fields
point. The random directions of the crystallo-
graphic axes of the crystallites limit the extent • High mechanical stresses
to which spontaneous polarization can be • Temperatures close to the Curie point
developed. Due to the inhibition of the motion
A common rule of thumb to reduce the rate of
of domain walls by internal strains and other
depolarization (aging) is that the poled ferro-
defects (such as impurities), the saturation
electric materials should not be exposed to
value of the spontaneous polarization of
temperatures higher than one half of the Curie
BaTi0 3 in ceramic structure is about half of the
temperature, since heat is the major factor that
value for BaTi0 3 single crystals. The hystere-
causes depolarization. However, aging is a
sis loop of BaTi0 3 single crystals with a single
complicated process that may be caused by
domain is a square loop, like the one shown in
several different mechanisms, such as
Figure 4-5(a), having a spontaneous polariza-
tion Ps of about 26jUC cm"2 and a coercive field • Gradual equilibration of domains toward the
Fc of about lkVcm -1 . The hysteresis loop for minimization of elastic and dielectric free
BaTi0 3 ceramics is like a round loop, like the energy
one shown in Figure 4-5(b), having Ps of about • Segregation of impurities and vacancies on
1 jiiCcrrf2 and Fc of about 4kVcmf1. Of course, domain walls and grain boundaries
the latter figures depend to a great extent on the
• Ordering of impurities and vacancies inside
preparation of the ceramic specimens.
the ferroelectric domains with respect to the
One unusual feature of ferroelectrics is the
polar axis52
aging phenomenon. A real crystal is never an
ideal dielectric. It always contains imperfec- Note that PZT ceramics are usually used with
tions, including vacancies, dislocations, impu- a dopant, modifier, or other composition in
rities, etc., which disturb the uniformity of the solid solution in order to improve or modify the
polarization and depolarization. A ceramic properties of the basic PZT system for specific
material contains even more imperfections. The applications. A detailed discussion of many
properties of ceramics change with time, even different compositional systems is beyond the
in the absence of external electric fields, scope of this chapter. However, a few signifi-
mechanical stresses, or temperature changes. cant examples will demonstrate the importance
This aging phenomenon is associated with the of additives in extending the variety of the
continuous changing of the internal structure, applications of the PZT system.
which with time tends to reach a more stable For the case of oxygen-rich PZT, the con-
arrangement—in other words, to lower the duction is mainly p-type extrinsic due to the
internal free energy. In fact, any poled ferro- holes produced by O" ions. To reduce the p-type
electric ceramic is in a state of incessant depo- conduction, we can use donor dopants, such as
234 Dielectric Phenomena in Solids

Nb5+ or La3+, replacing Pb2+ in order to provide dopants can improve this situation.61,62 For the
electrons required to convert 0~ ions to O2" PbZro.975Tio.o2503 system doped with lwt%
thus reducing the conductivity. It is important Nb205, the phase boundary at x = 0.025 occurs
to reduce the conductivity because high con- at room temperature. In general, the external
ductivity makes it difficult to apply a high field electric field tends to broaden the region of the
for poling for a long period of time. The addi- ferroelectric state, and the mechanical com-
tives are usually compensated by A-site vacan- pression extends the antiferroelectric region.
cies. (The Perovskite structure is expressed in Thus, the AF-F boundary can be made to shift
the general formula AB03. For BaTi0 3 , the either by an applied field or by a mechanical
vacancies created by missing Ba2+ are referred compression. This implies that the stored
to as the A-site vacancies. Similarly, the vacan- energy due to spontaneous polarization in the
cies created by missing Ti4+ are B-site vacan- ferroelectric phase can be released by a shock
cies.) In this case, these additives will enhance compression wave to convert the ferroelectric
domain reorientation and hence increase spon- phase to an antiferroelectric phase. This princi-
taneous polarization, dielectric constant, and ple can be applied for the generation of high-
piezoelectric constant; they will also lower the energy electric pulses.63
coercity and reduce the aging effect.53
One important compositional system, gener-
ally referred to as the PLZT system, is the PZT PVDF [(C H 2 - C F2)n]-7ype
system modified by the lanthanum. This system Ferroelectric Polymers
can be formed by the complete miscibility of The chemical formula for polyvinylidene fluo-
PbO, Zr0 2 , Ti0 2 , and La 2 0 3 in a solid solution; ride (PVDF) is (CH2-CF2)n. PVDF crystallizes
the amount of La can be adjusted on the basis from the melt below 150°C into spherulitic
of the general chemical formula Pb^La^ structures.64,65 In the melt, the polymer chains
(Zr^TiJ i_y403 with the fraction y ranging from have randomly coiled shapes. Configurational
0 to 1. Usually, the PLZT system is formulated defects can crystallize into regular conforma-
so that La3+ ions replace mainly the A-site tions when cooled from the melt. This is
vacancies created by missing Pb2+ ions. In this accomplished by the torsional bond arrange-
case, the chemical formula can be expressed as ments, in order to minimize the potential
Pbi-^La^Zr^TiJO^ Selected compositions energy of the chains. The favorable arrange-
within this system offer high electromechanical ments are those having substituents at 180°
coupling coefficients for piezoelectric trans- between each other (called transor t) or having
ducers.54"56 The PLZT system has also excellent substituents at ±60° (called gauche* or g1), the
electro-optic properties.57,59 Depending on its actual torsional angles commonly deviating
composition, the PLZT system exhibits one of from these values. There are three known con-
three major types of electro-optic characteris- formations of PVDF: all-trans, tg+tg~, and
tics: memory, linear, or quadratic. Each of these tttg+tttg~. The first two are by far the most
three types can be used for different electro- common and most important ones,65"67 and they
optic devices.60 are illustrated schematically in Figure 4-22.
Another example is the Nb5+-doped PZT In the all-trans conformation, all dipoles are
system. It is well known that for the aligned in the same direction normal to the
Pb(Zr!__,Ti,)03 system, a Zr-rich system (i.e., chain axis. This yields the most highly polar
small x)—for example with x - 0.05—exhibits conformation. The tg+tg~ conformation is also
an extensive phase boundary between the fer- polar but has components of the dipole
roelectric Fr and the antiferroelectric AF0 moments parallel and perpendicular to the
phases, as shown in Figure 4-19. However, for chain axis.
x < 0.05, it becomes difficult to sinter to obtain Depending on the temperature and the con-
a high-density ceramic, compared to common dition at which the material crystallizes, for
PZT with higher x values. It is found that Nb5+ most common polymorph PVDF produced by
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 235

(a) All-trans (b) tg + tg-

Figure 4-22 Schematic illustration of the two most common crystalline chain conformations in PVDF (a) all-trans con-
formation and (b) tg+tg~ conformation. The arrows indicate the projection of the CF2 dipole directions on planes defined by
the carbon backbone.

melt-solidification, the unit cell consists of two has a large dipole moment (u0 = 7.06 x
chains in the tg+tg~~ configuration and possesses 10~30C-m per monomeric unit).65 The degree
dipole moments with components parallel and of crystallinity for both a-phase and j8-phase
normal to the chain axis. So, the dipole PVDF is approximately 50%. In the as-drawn
moments of the two chains in the unit cell are /J-PVDF film specimens, the dipole moment is
antiparallel and the crystal does not exhibit preferentially oriented along the stretched ± y
spontaneous polarization. In this case, the direction in the film plane, since mechanical
unpolar phase of PVDF is referred to as the a- stretching (drawing) causes a breakdown of the
phase. However, if the PVDF crystallizes at a original spherulite structure into an array of
temperature below 130°C and at a high pres- crystallites whose molecules are oriented in the
sure68 or in conjunction with a special epitaxial direction of the stretching force. If electrodes
technique,69 the morphology of PVDF is are then deposited on the film surfaces and a
changed to a structure oriented to the c-axis, strong poling field of about O.SMVcnT1 is
and the unit cell consists of two all-trans chains applied across the film specimen, the dipoles
packed with their dipoles pointing in the same will reorient to align predominantly along the
direction. This crystal structure renders PVDF poling field in z direction perpendicular to the
the most highly polar phase, which is generally film plane.
referred to as the j8-phase. The chains take on Ferroelectricity of PVDC has been estab-
a planar zigzag (tttt) conformation. The chain lished on the basis of the measured dielectric
236 Dielectric Phenomena in Solids

(a)

(b)

-1 0
Electric Field (MV cm"1)

Figure 4-23 The variation of (a) dielectric constant and (b) polarization of /3-phase PVDF films with applied electric field.
• Initial poling, O subsequent poling cycles.

constant of CF2 and the corresponding polar- This material has been widely used for piezo-
ization as functions of temperature, as shown in electric transducers and pyroelectric devices.
Figure 4-23. The experimental data are from
Takahashi, Date, and Fukada.70 A hystere-
sislike behavior of PVDF is good evidence that 4.2.3 Thermodynamic Theory
PVDF is a ferroelectric material. The Curie Classical thermodynamics deals mainly with
transition involves mainly intramolecular the macroscopic behavior of materials. There
changes of polar orientation through the intro- are laws of thermodynamics that define the
duction of g± bonds that alter the polar all-trans relationships between properties. To gain an
conformation to a somewhat tg± and tt arrange- understanding of the phase transitions and
ment. As a result of this change, the Curie related effects, we need to understand the
temperature for PVDF for the transition from concept of free energy. One of the well known
the ferroelectric phase to a paraelectric phase thermodynamic functions is the Helmholtz
would be around 205 °C—about 20°C above its free-energy function A, which is given by
melting point.71
A = U-TS (4-10)
Although PVDF is ferroelectric, its dielectric
constant is relatively small, around 10 to 15. where U is the internal energy, S is the entropy,
But it exhibits strong piezoelectric and pyro- and T is the temperature. There is also another
electric effects compared to other polymers, as well-known function, called the Gibbs free-
first discovered by Kawai15 and Bergman et al.72 energy function, which is given by
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 237

G = U-TS + pV (4-11) statistical mechanics, the entropy of a system is

defined as
including apV term. Most experiments in solids
are performed at atmospheric pressure or pres- S = k\ogq (4-14)
sures not far from it, which is so small that it where q is the number of microstates of the
makes the pV term negligibly small compared system. Disorder can be interpreted as the
to other terms. Thus, by ignoring the pV term, number of ways in which the particles inside
the Gibbs function becomes a system can be arranged microscopically, so
each different way can be considered a
G0=U-TS = A (4-12) microstate or a thermodynamic state. Suppose
The basic principle is the concept of the that an alloy consists of two components: A and
minimum free energy for the most stable B. The total number of these two components
system. If a system is allowed to have several is N per unit volume, of which n are of A type
alternative states, it will choose the one with the and the remaining N-n are of B type. Since the
lowest free energy. To clarify the meaning of A components can be interchanged themselves
this principle, we apply it to a simple phenom- in n! different ways and the B components in
enon. Based on Equation 4-12, a system can (N - n)f different ways, the number of differ-
have the free energy at a lowest possible value ent arrangements for N components (i.e., the
if the internal energy U can be made as small number of microstates in the system) is
as possible and the entropy S as large as possi-
ble. The internal energy U in a solid is primar-
ily the potential energy, which is usually a
negative quantity, implying that for a minimum Heat is one form of energy and energy cannot be
potential energy all atoms (or ions) must rest at created, it can only be converted from one form
their lattice sites, that is, each atom (or ion) is into other, different forms. The first law of ther-
resting at the bottom of its potential well. But modynamics states that the increase in internal
this is a well ordered arrangement with very energy dU is equal to the sum of the heat put in
low entropy. Such an arrangement does not lead the system dQ and the work put in the system.
to a minimum free energy. Not all atoms (or When a solid body is subjected to an external
ions) are at the bottom of the potential wells electric field and mechanical stress, the internal
due to the thermal agitation, so the degree of energy will change accordingly. We can write
order varies accordingly. At a given tempera- dU = TdS + Xdh + FdP (4-16)
ture T, the value of U and the degree of order
S always tend to balance each other in order to where X and h are, respectively, the mechani-
reach a stable equilibrium state in which the cal stress and the strain; and F and P are,
free energy is the lowest. This is also why all respectively, the electric field and the polariza-
atoms (or ions) in a solid are oscillating tion. F and P are vectors, and they have three
(vibrating) around their lattice sites. components in x, y, and z directions. X and h
Statistically, entropy is a measure of disor- are also vectors and have six components in x,
der. The higher the disorder (or the lower the v, and z, as well as shear directions on the plane
degree of order), the larger is the entropy. Based normal to each of x, y and z, so the energy state
on thermodynamics, the entropy can be defined of a crystal is specified by the values of nine
as variables plus the temperature.
Thermodynamics can be used to describe the
dS = dQ/T = CpdT/T (4-13) macroscopic behavior of the ferroelectric
system by including external work done on the
where dQ is the amount of heat absorbed by the system. If the work put in the system is due to
system in a reversible process, and Cp is the external mechanical stresses and electric fields,
specific heat at constant pressure. Based on then the Gibbs function can be written as
238 Dielectric Phenomena in Solids

G = G0 + JiX,hl + JtFJPj (4-17) does not contain terms of odd powers of P

because crystals such as BaTi0 3 in the unpo-
The independent variables in the measurements larized phase have a center of inversion sym-
of normal ferroelectric properties are the tem- metry. Supposing that the applied poling field
perature T, the mechanical stress X, and the is in the z direction (c-axis of the cubic-tetrag-
polarization P. Thus, it is convenient to discuss onal structure), then the polarization is also
the G function as G (7; X, P). In the absence of along the same z direction. By setting Py =
mechanical stresses, X = 0 and at a constant T, Px = 0 and letting Pz- P for simplicity, Equa-
Gibbs function can be simplified to tion 4-19 can be simplified to

G(7,P) = G ( D + X T ^ (4-18) G(T,P) = G0(T)+^-g2P2 +\g<PA +^geP6

0
J ^^rj^o
1 4 o
(4-20)
where Ga(T) is the Gibbs function under the
unstressed and unpolarized conditions. The where g2, gt, g& etc., are new temperature-
thermodynamic approach can be used to dependent coefficients. For the isothermal case,
explain many ferroelectric properties. The fol- the electric field acting on the ferroelectric
lowing sections offer two transition processes material, expressed in terms of P, can be
as examples. obtained by

Ferroelectric Transition F=^ = g2P+g4P* + g6P5... (4-21)

It should be noted that in Equation 4-18 both
erj and P, are temperature dependent and their At and near the Curie point, g2 can be approx-
behavior is anomalous and nonlinear near the imated with a linear function of temperature.73
transition point. Therefore, the behavior of the Thus, g2 may be written as
ferroelectric system can best be studied by g2=p(T-Tc) (4-22)
expanding the energy in powers of the polar-
ization. The Gibbs function for Perovskite-type where /? is taken as a positive constant and Tc
ferroelectrics, formulated on the basis of the may be equal to or close to the Curie transition
form of a power series by Devonshire73 and temperature. The nature of the transition
further developed by Huibregtse and Young,74 depends on the signs of the coefficients: g2, g*
is given by g6, etc. The simplest case is that all coefficients,
except g2i are positive, and P is continuous at
G(7\ P) = G0(T) + ±f2(Pl + P2y + P2Z) the transition point, implying that the transition
is of the second order.
To find the values of P for which G is at the
+ \fA{P*X+PAy+Pl)
lowest value (minimum free energy), we dif-
ferentiate Equation 4-20 with respect to P and
set it equal to zero, that is, the values of P at
+ ~flu{P6X+P"y+Pt) F = 0.
o
g2P+g4P3+g6P5=0 (4-23)
1 i
+ ^fu2[Pl{P y+P})+Pl(Pt + P X)
For simplicity, we ignore terms with the powers
+ Pl(P$ + P*)\ of P equal to and higher than 5. Then we have
the values of P for the lowest G value as
+ \fmPiPyPl + -.-
o
P=0
(4-19)
and
where f2, U fn, fm> fm, fm, etc., are
temperature-dependent coefficients. The series P = ±(g2/g4)l/2 (4-24)
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 239

The plot of G - G0 as a function of P for three direction opposite to the polarization may
cases—g2 > 0, g2 = 0 and g2 < 0—is shown in enable the Ti^ ions to pass over the potential
Figure 4-24(b). It can be seen that for g2 = 0 barrier between the two wells and reverse the
(i.e., when T = Tc), P changes gradually from direction of the polarity.
zero at T = Tc to a finite value at T < Tc. The If some coefficients other than g2 are nega-
transition in BaTi0 3 is of the first order, but the tive, the situation becomes more complicated.
energy of the Ti4+ ions in terms of their posi- For simplicity, assume that only g4 is negative
tions along the c-axis takes the form of two and ignore all terms with powers of P higher
potential wells. An applied electric field in the than 6, then Equation 4-20 can be simplified to

92

(a) First-Order Transition

92

(b) Second-Order Transition

Figure 4-24 G-G0 as a function of polarization at transition temperature and at temperatures just above and just below
T c for (a) first-order transition and (b) second-order transition.
240 Dielectric Phenomena in Solids

dition that the free energy of the polar and the

G = G0 + ^g2Pi--\gA\Pi + ^gePb (4-25) nonpolar phases are equal. This leads to
L 4 O

For the minimum values of G, we have

^m-Tc)P2-]-\gAPA + \geP6=^ (4-31)
dG 2 4 o
3 5
--g2P~\g4\P + geP =0 (4-26)
dP At the same time, the applied field is set equal
to zero. This gives
The solution of Equation 4-26 yields
5
P=0 F = ^ = (T{-Tc)P-\gt\P' + g()P =0
and (4-32)
From Equations 4-31 and 4-32, we obtain
P = ±{^"[1*41 ± ( N 2 - 4^ 2 ) ,/2 ]} 1/2 (4-27)
"3fM N1 ' /2 (4-33)
p=±\
Figure 4-24(a) shows the value of G - G0 as a 4 Us
function of P for g2 > 0, g2 = 0 and g2 < 0. For 3 M2 (4-34)
g2 = 0 (i.e., T= Tc\ we have P = ±(|g4|/g6)1/2. T^-Tc =
16 &J
This implies that P changes discontinuously
This indicates that the Curie temperature is
from P = -(\g4\/g6)m or P = +(\g4\/g6)m to P =
lower than the actual transition temperature by
0 abruptly at 7 = 7C, and that the transition is of
the first order. 17DS4I7&6]. For BaTi0 3 , Tt-Tc = 7.7°C.30
If the poling field is not high, we can use lo
first-order approximation by ignoring all terms
Thermal Hysteresis
with the powers of P higher than 2 in either
On heating the crystal through the Curie
Equation 4-20 or Equation 4-25. Then we have
point, the polarization changes discontinuously
from P = 0 in the ferroelectric phase to P = 0
F = ^ = g2P = p(T-Tc)P (4-28)
dP in the paraelectric nonpolar phase. In the first-
order transition, the crystal suddenly loses all
and P for T > Tc can be expressed as
the energy associated with the polarization at
P = (er-\)e0F = X£<,F (4-29) the Curie point, indicating a big change in its
latent heat.
Thus, from Equations 4-28 and 4-29, we obtain
We can calculate the actual transition tem-
the Curie-Weiss relation
perature T2, above which the polar phase does
1 not exist in the absence of electric field. In
(4-30)
x = -Pe0(T-Tc) T-Tc Equation 4-27, for the minimum free-energy
condition, there are several distinct values ofP,
where C = (j8e0)_1- This equation is similar to
for which G is a minimum. Above T2 there is
Equation 4-1.
only one value of P, corresponding to the
This thermodynamic approach can also be
condition
used to explain many ferroelectric phenomena
related to the first-order transition. Some of the Is4l2-4g60(r2-:rc) = o (4-35)
common phenomena follow.
Thus,

The Difference between the Curie T2 = Tc + \g4\2/4Pg6 (4-36)

Temperature and the Actual
Transition Temperature In this case the Curie temperature is lower
The actual transition temperature Tx and the than the actual transition temperature by
corresponding value of the spontaneous polar-
hgrf/Pgel For BaTi0 3 , T2 - TC = 10°C,30
ization can be calculated by imposing the con-
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 241

indicating T2>TX. In ferroelectrics with a first- 20 r

order transition, such as BaTi03, the transition
temperature is always at a slightly lower value
when the crystal is cooling from the high
temperature than when it is heating from the
low temperature. This phenomenon is generally
referred to as thermal hysteresis.

Electric-Field Dependence of the 3 4 5

Curie Temperature F(kVcrrr1)
Based on Equation 4-17, the differential
Figure 4-25 Hysteresis of BaTi0 3 above the Curie point
form of the Gibbs free-energy function can be
at three different temperatures.
written as
dG = -SdT + £ hidXt + ]T PjdFj (4-37)
^ _P ^ 2 = 4 pg4|V/2
(4-42)
dF S pP J3p{g6)
Here we choose T,X,F instead ofT,X,P as the
independent variables. Under a constant stress For BaTi0 3 , dT/dF = 1.72 x KT^CcmV- 1 .
condition with the applied field along one direc- This value agrees well with the experimental
tion, Equation 4-37 can be simplified to value, which is 1.43 x lO-^CcmV" 1 . 30
The above argument indicates mainly that at
dG = -SdT + PdF (4-38) temperature T> Tc, the crystal can be forced by
At the Curie point, the free energy in the ferro- an applied field to make a transition into a fer-
electric phase and that in the paraelectric non- roelectric state and then return to the paraelec-
polar phase are equal. So for dG = 0, we have tric nonpolar state when the field is reduced,
producing a double hysteresis loop, as shown
dT P in Figure 4-25 for BaTi0 3 . The data for Figure
(4-39)
dF~ S 4-25 are from Meyerhofer.75
In this case, p and S are, respectively, the spon-
taneous polarization and the entropy at T = Tc. Antiferroelectric Transition
The discontinuous change in polarization at the A crystal is antiferroelectric if it exhibits para-
Curie point causes a discontinuous change in electric behavior, (i.e., the dielectric constant
entropy and hence the latent heat typical of increases with decreasing temperature above
a first-order transition. The entropy can be the transition point) but does not show a net
expressed as spontaneous polarization and the dielectric con-
stant is considerably low below the transition
(4-40) point. A typical example is the PbZr0 3 crystal
(see Figure 4-20). This indicates that the
By differentiating Equation 4-25 with respect transition involves a different mechanism for
to T, and, for simplicity, ignoring all the terms dielectric polarization. X-ray analysis has
with powers of P equal to or higher than 4 revealed the existence of a superstructure
(because the coefficients of these high-power below the transition point. The superstructure
terms do not vary much with temperature), we can be described in terms of two superlattices
obtain having equal but opposite polarization; above
the transition temperature, the two superlattices
(4-41) are crystallographically equivalent and are
dTJxp 2 dT 2H unpolarized.
Substituting Equation 4-33 for P into Equation In the previous section, we analyzed ferro-
4-41, we obtain electric phenomena based on an appropriate
242 Dielectric Phenomena in Solids

thermodynamic free energy function G, which the value of the coefficients. Normally, the
is expressed in powers of the polarization P. coefficient g2 decreases with decreasing tem-
According to Kittel,76 the same approach can be perature, and the antiferroelectric state is stable
applied to an antiferroelectric crystal if G is
at low temperatures. If the term —XP+P- is
expressed in powers of the polarization Pa and
Pb of the two superlattices. It should be noted positive and sufficiently large, then one kind
that, unlike P for ferroelectrics (which can of the state (e.g., antiferroelectric state) will
be detected), Pa and Pb for antiferroelectrics prevent the formation of the other (e.g., ferro-
are not macroscopically observable quantities. electric state).
However, following the same procedure for The sharp decrease of the dielectric constant
Equation 4-20, we can write below the transition point is due to the satura-
tion effect of Pa and Pb. Antiferroelectrics can
G(Pa,Pb,T) = G0(T) + f(P2a+Pi) + gPaPb
be piezoelectric. Some antiferroelectric materi-
+ h(PAa+PAa) + k{P6a+P6a)+... als may also become ferroelectric under the
(4-43) application of a suitable electric field.77 It is not
Whether the ferroelectric state (Pa = Pb), the so obvious that domains would be formed in
antiferroelectric state (Pa = -Pb), or the para- antiferroelectrics. However, optical observation
electric state (Pa + Pb = 0) has the lowest has revealed the domain structure of PbZr0 3
free energy depends on the values of the and NaNb0 3 antiferroelectrics. The domains
coefficients. are nonpolar; therefore, their patterns should
To deal with the antiferroelectric transition, not be affected by external electric fields. For
it is convenient to transform Pa and Pb into the more details about antiferroelectric phenom-
following parameters: ena, see references.38,78
A two-dimensional picture illustrating the
basic difference among the ferroelectric, anti-
ferroelectric, and ferrielectric states is given in
Figure 4.26. In the ferroelectric state, there is
We can consider P+ as the total polarization and
polarization in the vertical direction and no
P_ as a measure of the antiferroelectric state.
polarization in the horizontal direction; in the
Substitution of Equation 4-44 into Equation 4-
antiferroelectric state, there is no net polariza-
43 gives tion in either the vertical or the horizontal direc-
tion; in the ferrielectric state, the crystal could
G(P+,P^T) = G0(T) + ^g2P2++^g4Pl be antiferroelectric in the vertical direction and
ferroelectric in the horizontal direction.
o However, ferrielectric behavior is not really
similar to its ferrimagnetic counterpart. It is a
+ ^P2-+\g'4P< (4-45) very strange behavior. An example of ferrielec-
tric behavior is thiourea [SC(NH)2]. The rever-
+ \g'6Pt + ... sal of the polarization in this crystal is related
to relative displacements of entire molecules,
+ -AP+2P_2 + ... rather than to the motion of single atoms or ions
in the lattice, and its polarization phenomenon
where the coefficients g2, g4, g6, g'2, g\, can be considered ferrielectric.79
ge, . . . and A, etc., are related to the coefficients
f, g, h, k, etc.
Detailed analysis of antiferroelectric behav- 4.2.4 Formation and Dynamics
ior is beyond the scope of this chapter. In of Domains
general, an antiferroelectric state exists if G has A single crystal or a system that is ferroelectric
a minimum for a finite value of P_. Obviously, will undergo spontaneous polarization below
this condition depends strongly on the sign and the Curie temperature. If all the dipoles of the
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 243

- + + - + -
t t t t M M \ / \ /

t t t t 1 t J t \ / \ /

t t t t J t 1 t \ / \ /

t t t t M \ t \ / \ /

(a) (b) (c)

Figure 4-26 Schematic illustration of the polarization in (a) ferroelectric structure, (b) antiferroelectric structure, and (c)
ferrielectric structure.

polarization are pointing in one direction, the directions, so the polarization will be in oppo-
electrostatic energy of the system is extremely site directions in adjacent domains by twinning.
large and the system becomes very unstable. In this case, the domain walls are called the
Naturally, a system always tends to minimize 180° walls, as shown in Figure 4-27(a).
its potential energy by forming the so-called For multiaxial ferroelectrics such as BaTi0 3 ,
domains, which are actually regions in the domains can have more than two dipole direc-
crystal, each containing a large number of tions. For example, in the tetragonal phase,
dipoles all aligned in the same direction. polarization is along the oaxis, that is, the elon-
Domains are arranged in such a way that the gated axis of the cubic structure. But in the fer-
polarization of different domains will compen- roelectric phase, there are six easy directions
sate for each other and the net polarization of deriving from the cubic structure, that is, the c-
the whole crystal along any direction will axis can be in the ±x, ±y, and ±z directions. This
vanish. Domains are separated by domain implies that there are three pairs of antiparallel
walls, which are the loci of the dipole orienta- directions along the cube edges and that
tion from one direction of the domain to domains can have six directions of polarization.
another of the neighboring one. A single crystal This gives rise to different types of domain
may contain many domains with zero net polar- walls: Those separating antiparallel dipoles are
ization, but it can be changed to a single crystal 180° walls, and those separating dipoles at right
with a single domain by a strong poling elec- angles to each other are 90° walls, as shown in
tric field, in which the dipoles of all domains Figure 4-27(b) and (c). Similarly, for the ferro-
must point in the same direction. In other electric phase with the unit cell in rhombohe-
words, all domains will join to form a large dral structure, the polarization is along the
domain. For polycrystalline materials such as body-diagonal directions. In this case, there are
ceramics, single domain cannot be achieved eight easy directions for spontaneous polariza-
because the crystal axes of the grains (crystal- tion. The angle between adjacent domains
lites) in the material are randomly arranged, becomes 70.5° and 110°, so the domain walls
which cannot be altered by the electric field. are called the 70.5° and 110° walls. Some
When a phase transformation begins, the typical domain patterns in crystals are illus-
domains will be nucleated at several places trated schematically in Figure 4-27.
within the crystal, and the nuclei of domains The electrostatic energy decreases with an
will grow along the ferroelectric axes until the increase in the number of domains formed due
transformation of the new phase in the whole to the decrease in the depolarizing field.
volume is completed. Ferroelectric domains are However, the domain formation process does
closely analogous to ferromagnetic domains. In not continue indefinitely, because a certain
uniaxial ferroelectrics such as TGS crystals, the amount of energy is stored in the walls between
domains can have only two possible dipole different domains. For ideal crystals, when the
244 Dielectric Phenomena in Solids

1 i
(a)

(b)

(c)

Figure 4-27 Schematic illustration of some domain patterns: (a) antiparallel domains with 180° walls, (b) domains with
180° and 90° walls, and (c) mixture of domains in c and a directions {a is perpendicular to c).

domain wall energy has increased to a certain nonconductive and contain various defects, the
critical value, at which it just balances the domain configuration is hardly stable at any
decrease in energy of the depolarizing field, the temperature because the charges induced by
total potential energy of the crystal becomes spontaneous polarization are partially compen-
minimal. Under this condition, the domain sated by the conducting carriers in the materi-
configuration has reached an equilibrium and als, and the uniformity of the polarization and
stable state at the temperature considered. But the depolarizing field is disturbed by various
for real crystals that are not completely defects. Because of such defects and the
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 245

conductivity of real crystals, the domain pat-

terns do not always correspond to the absolute
minimum of the free energy. This implies that
the domain configuration is only metastable,
resulting in the aging effect.
The calculation of domain wall energy is
very complicated, but the width of the domain
walls is directly related to the total energy of
the domain walls. The total wall energy is the (a) (b)
sum of the energy due to dipole interaction and
Figure 4-28 Schematic illustration of the difference
the anisotropy energy.22'39 Thus, the total wall
between (a) the ferroelectric and (b) the ferromagnetic 180°
energy £/wall can be written as domain walls. The width of the domain walls is about
several lattice constants for ferroelectrics and about several
hundred lattice constants for ferromagnetics.
f/Waii=4 + ^ (4_46)
o
where 5 is the wall width and A and B are func- only several lattice constants. For ferromag-
tions of the lattice constant, the spontaneous netic materials, the domain walls are a hundred
strain and the elastic constant. By minimizing times wider because the interaction energy of
the wall energy magnetic dipoles is large and the elastic energy
is small.
dU^_ = Q (4_47)
In general, strain energy is the dominant
dd
component in the ferroelectric domain wall
This leads to energy budget, and the 180° and 90° walls are
8 = (A/Bf2 (4-48) thus distinctly different in behavior. It is easy
for the 180° walls to move under the influence
Merz22 has estimated the wall width to be of of an applied electric field. The motion of the
the order of several lattice constants and wall 180° wall merely requires the cations to move
energy of the order of 10"6 joule cm-2, in con- cooperatively toward the other end of their
trast to the ferromagnetic domain walls whose cages without involving the rotation of the
width is of the order of several hundred lattice polarization vectors or any change in the shape
constants. In a ferromagnetic domain wall, the of the crystal. This implies that the domains
magnetization vectors must turn over gradually with the polarization in the direction of the
from one direction to the opposite one, which applied field expand sideways by moving the
involves a change in exchange energy and walls toward the domains with the polarization
anisotropy energy. This is why the ferromag- opposite to the field direction, making them
netic domain walls are much wider than the shrink and eventually vanish. In other words,
ferroelectric domain walls. The difference the two antiparallel domains will join or coa-
between the ferroelectric and the ferromagnetic lesce to form a big domain, with all dipoles
domain walls is shown in Figure 4-28. aligned in the field direction.
In Equation 4-46, the first term A/8 is the For the reversal of the polarization, an
energy from dipole interaction due to the fact applied field should be normally higher than the
that the polarization changes from +P to -P in coercive field Fc, and the time required for the
the wall, and the second term B8 is the energy complete reversal depends on the magnitude of
from the stress causing the strain changes from the applied field following the relation
+ strain to - strain in the wall. In ferroelectric
ts=kF~n (4-49)
materials, the first term (due to sideways inter-
30
action of adjacent dipoles) is small, whereas the where k and n are constants. It should be noted
elastic energy due to the changes of the strain that when the applied field is small, the rever-
is large. This is why the domain-wall width is sal may still occur, but a much longer switch-
246 Dielectric Phenomena in Solids

ing time is required. This implies that the so- surface) and grow in the direction of the applied
called coercive field Fc does not have a clear field with very little sideways motion of the
physical meaning, as fields much smaller than walls. However, it can be imagined that the
Fc still switch on the reversal, if the applied nucleation of new domains becomes more dif-
field is held for a sufficiently long time. For low ficult because more nuclei are already present,
fields, ts follows an exponential law is °° exp due to more electro-mechanical interaction,
(a/F), rather than a power law, given in Equa- particularly in multidomain crystals.
tion 4-49. Sideways motion of 90° walls in an applied
The switching time is directly related to the electric field has been observed.39'81 Since
velocity of the motion of domain walls. At high the 90° wall width is only about several lattice
fields (>lkVcm _1 ), the sideways wall velocity constants, the activation energy for the wall
follows the relation80 motion is only a small fraction of the total
wall energy. The critical field for the 90° wall
v, = hFm (4-50)
motion is smaller than that for the nucleation
where h and m are positive constants. It can be and the growth of new domains. However, the
seen that v^ is approximately proportional to the motion causes a change in the shape of the
reciprocal of the switching time ts. However, crystal. For a more detailed discussion of
when two 180° walls are sufficiently close, they the formation, structure, and dynamic behavior
do not move toward each other and there of domains, see references 22, 30, 38, 39, 78,
appears to be a repulsion between them. This and 80-86.
repulsion results in the formation of a very
narrow metastable region which suddenly dis-
appears when an electric field is applied, creat- 4.2.5 Ferroelectric Materials
ing Barkhausen pulses. There are now many hundred materials that
The 90° walls are relatively difficult to exhibit ferroelectric or antiferroelectric proper-
move and usually require an applied electric ties. In Section 4.2.2, the properties of some
field greater than a certain threshold value. In representative ferroelectrics were discussed. In
general, the motion involves the nucleation and this section, we will not list all the ferroelectric
growth of new domains with the polarization in materials but summarize some important prop-
the direction of the applied field. Thus, making erties of some ferroelectrics with the potential
the domains with polarization perpendicular to for practical applications in Table 4-2.
the applied field change to polarization along The most commonly used ferroelectrics have
the field direction may sometimes cause the the Perovskite structure, with the chemical
crystal to change shape. The process for a com- formula AB0 3 . The PZT system is a good
plete polarization in the direction of the applied example, consisting mainly of the alloy of
field for a multiaxial ferroelectric could be quite PbTi0 3 and PbZr0 3 . The general formula of the
complicated. Let us take a simple case as an PZT system is PZ^T,, so the ferroelectric
example. For a single-domain crystal with the properties can be tailored by varying the com-
polarization along the c-axis in the tetragonal position of Zr and Ti, that is, by varying the
phase, if an electric field is applied along the a- value of x.
axis (perpendicular to the c-axis), then a large It is not always convenient to keep ferro-
number of needlelike new domains nucleate at electric devices in an oven or a cryostat in order
the surface and are polarized in the direction to maintain the required properties. Now, mate-
of the applied field, growing in the form of rials can be modified so that certain values of
wedges.81 The formation of new domains con- the required properties can be achieved at room
tinues until the whole crystal is polarized in the temperature. The control of properties is possi-
same direction of the applied field. These newly ble by varying the composition in solid solution
formed domains originate at the surface of the or by incorporating suitable additives.87 One
crystal specimen (usually at the electrode good example is a ceramic formed by a solid
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 247

Table 4-2 General transition properties of some ferroelectric crystals.

Spontaneous
Curie Polarization Crystal Structure
Chemical Temperature, Ps(HCmT2) at
(Abbreviation) Formula Tc (°Q [T(°Q] Above Tc Below Tc

Barium Titanate BaTi0 3 120 26.0 [23] Cubic Tetragonal

Lead Titanate PbTi0 3 490 50.0 [23] Cubic Tetragonal

Potassium Niobate KNb0 3 435 30.0 [250] Cubic Tetragonal

Potassium KH 2 P0 4 -150 4.8 [-177] Tetragonal Orthohombic

Dihydrogen Phosphate
(KDP)

Triglycine Sulfate (NH2CH2COOH)3* 49 2.8 [20] Monoclinic Monoclinic

(TGS) H 2 S0 4 (Centrosymm.) (Noncentrosymm.)

Potassium-Sodium
Tartrate-Tetrahydrate KNaC4H406 •4H2Q 24 0.25 [5] Orthorhombic Monoclinic
(Rochelle salt) (Centrosymm.) (Noncentrosymm.)

Antimony Sulfo-iodide SbSI 22 25.0 [0] Orthorhombic Orthorhombic

(Centrosymm.) (Noncentrosymm.)

Guanidinium Aluminium C(NH2)3A1(S04)2*6H20 None 0.35 [23] Trogonal

Sulfate Hexahydrate
(GASH)

solution of PbTi0 3 in BaTi0 3 , in which the 500 r

— SrTi03 PbTi03
transition temperature depends on the con- 400
- -BaZr03
centration of Pb in BaTi0 3 . In this case, all 300 ---BaSn03 -f
concentrations are possible and they are ferro-
5*200
electric. An increase in Pb concentration causes
~o100
the transition temperature Tc to increase from
120°C to 490°C. Similarly, the substitution of 0
zirconium for titanium in BaTi0 3 lowers the -100
value of Tc, as shown in Figure 4-29. Iron, -200
100 80 60 40 20 0 20 40 60 80 100
strontium, and calcium also tend to lower the ABO, Mole % of ABO, BaTiO, Mole % of PbTiO, PbTiO,
value of Tc. For details about this subject, see
references 39, 88, and 89. Figure 4-29 The effect of various additives in BaTi0 3 on
Ferroelectrics in ceramic forms are com- the Curie temperature Tc.
monly used for practical applications because
they have the advantage of being a great deal
easier to prepare than their single crystal coun- together with any required additives, into
terparts. Also, in many cases their ferroelectric powder. This reacted powder is usually ball-
properties approach quite closely those of the milled to increase homogeneity and particle
single crystal. As mentioned above, a wide reactivity. This may be processed with or
range of ceramic compositions and additives without binder. Then the material is extruded or
can be used to adjust the properties of the mate- pressed into some shape and fired at a higher
rial for different applications. Ceramics are temperature (e.g., 1300°C), at which sintering
generally prepared by grinding the material, occurs.44 This fabrication procedure produces a
248 Dielectric Phenomena in Solids

complicated polycrystalline material that has ceramics are possible. This is why the hystere-
pores, some unavoidable impurities, and many sis loops are generally rounded, the sponta-
crystallites of various sizes. Some properties of neous and the remanent polarizations are lower,
a ceramic may differ from those of the corre- and the coercive fields are larger; this is also
sponding single crystal material because the why the aging effect is enhanced. As knowl-
crystallites in the ceramic are randomly ori- edge of ferroelectric ceramics progresses, new
ented and some of them are under stresses even effects and new fabrication techniques, as well
when no external force is applied; also, there as new materials, will certainly be discovered.
exist grain boundaries and porosity. It is worth mentioning that ferroelectric thin
The ferroelectric properties of ceramics films are very important for practical applica-
depend greatly on the grain (crystallite) size. In tions. For certain applications, ferroelectrics in
a ceramic with small grain sizes, a significant thin-film form are preferred. In general, the
portion of the material volume is affected by dielectric and ferroelectric properties are thick-
grain boundaries. The important difference ness dependent if the thickness of the films is
between grain boundaries and free surfaces is below a certain value. For example, BaTi0 3
that in high-density ceramics, the material close films prepared by electron-beam evaporation or
to the grain boundaries is elastically clamped sputtering techniques exhibit normal dielectric
by the neighboring grains at temperatures well and ferroelectric behavior if the film thick-
below the sintering temperature. On cooling nesses are larger than l^um. Ferroelectric
to the ferroelectric phase, large mechanical behavior becomes unstable for film thicknesses
stresses may be developed by the anisotropic below 0.1/xm.90 But some investigators91 have
spontaneous strain, and these stresses may in reported that vacuum-evaporated BaTi0 3 films
turn affect the domain dynamics and configu- show well defined dielectric anomalies near
ration. If the grains are too small, the polariza- 120°C for films with thicknesses down to
tion may be completely clamped, preventing 0.023 )Um. For BaTi0 3 films prepared by flash
the action of domain reversal when the applied evaporation, the properties remain approxi-
field is turned to the opposite direction. As the mately similar to those of thick films down to
size of grains increases, domain reversal a thickness of 0.1 ^urn.92 The discrepancy can be
becomes possible, but the peak dielectric con- attributed to different techniques of film fabri-
stant decreases, due mainly to the inhomoge- cation, which result in various structural and
neous distribution of the mechanical stresses chemical defects. Defects aside, the interface
and electric fields. If the polarization in the between the film and the electrical contacts, the
neighboring grains is not parallel, then the width of the depletion regions inside the film,
nonzero polarization at the boundaries creates and electronic conductivity also play important
depolarizing fields, tending to compensate free roles in the thickness dependence of the dielec-
charges at the electrodes. tric and ferroelectric properties of thin films.
The grain sizes of ceramics can be controlled Although several film fabrication techniques
to a certain degree during the fabrication pro- (chemical vapor deposition, single or multiple
cessing by adjusting the sintering temperature target sputtering, multiple target ion beam,
and time, the composition of the material, and pulsed deposition techniques, etc.) have been
the additives. Most ceramics consist of grains put forward,58,93 none can be considered very
with a range of grain sizes, and their crystallo- satisfactory. However, the required degree of
graphic axes are randomly oriented. It can be perfection varies according to the specific
imagined that the poling of a ceramic under an application; thus, the technique for film fabri-
external electric field involves microscopic cation should be chosen accordingly. For
processes much more complicated than those example, for ferroelectric films used in small
for single crystals. Because of the grain bound- capacitors, we would want the films to have a
aries, neither dielectric nor elastic relaxations high dielectric constant, low dielectric loss, and
under an alternating field within the bulk of a minimal variation of these parameters with
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 249

temperature, applied field, and frequency. But for practical applications. The most commonly
for films used in optical devices, we would used effects are the piezoelectric, pyroelectric,
want the films to have a high optical quality. and electro-optic effects, which are discussed,
The trend of using ferroelectric films is ever respectively, in Piezoelectric Phenomena and
increasing. The development of new techniques Pyroelectric Phenomena in this chapter and
and the improvement of the existing techniques Modulation of Light in Chapter 3. In this
have to a great extent been motivated by section, we shall briefly describe some applica-
various technical applications. It is certain that tions related directly to the special features of
new experimental and theoretical results in this ferroelectrics: mainly, spontaneous polariza-
area will be available in the near future. tion, nonlinearity, and hysteresis behavior.

4.2.6 Applications of Ferroelectrics Capacitors

Based on the classification of crystals given in Capacitors are important elements in electrical
Figure 4-1 the following statements can be and electronic circuits, performing various
made: functions that include blocking, coupling and
decoupling, AC-DC separation, filtering, power
• All ferroelectrics possess piezoelectric and
factor correction, energy storage, etc. For a
pyroelectric effects.
capacitor with the dielectric material of dielec-
• All pyroelectrics possess piezoelectric tric constant er and thickness d between two
effects, but not all are ferroelectric. parallel metallic plates of area A, the capaci-
• Piezoelectrics may have only piezoelectric tance is given by
effects, both piezoelectric and pyroelectric
A
effects, or all three effects: ferroelectric, C = ere0 — (4-51)
d
piezoelectric, and pyroelectric.
If the working voltage is V, the average
Depending on the crystal structure, sponta-
working field F is V/d, which must be lower
neous polarization in a poled ferroelectric
than the breakdown strength of the dielectric
crystal can be varied to produce different
material Fb by factor k. The volume efficiency
effects by the application of an external force,
of the capacitor is defined as
such as a mechanical stress, an electric field, or
a change in temperature. Figure 4-30 illustrates C ere0
(4-52)
these effects. Each of these effects can be used

•Piezoelectric Effect (Due to change in polarization P)

Poled
•Piezocalorie Effect (Due to change in entropy S)
Mechanical Stress Ferroelectric
•Piezo-Optic Effect (Due to change in strain ft—elastic displacement
Specimen
resulting in change in refractive index)

•Converse Piezoelectric Effect (Due to change in strain h)

Poled
•Electrocalorie Effect (Due to change in entropy S)
Electric Field Ferroelectric
•Electro-Optic Effect (Due to change in polarization and hence in
Specimen
susceptibility)

Poled [•Pyroelectric Effect (Due to change in polarization P)

Temperature Change - Ferroelectric •Thermoelectric Effect (Due to change in strain h)
L
Specimen »Heat Capacity Effect (Due to dQ = SdT)

Figure 4-30 The effects of a mechanical stress, an electric field, or a temperature change on the behavior of a poled
ferroelectric specimen.
250 Dielectric Phenomena in Solids

The maximum permissible energy density

stored in the capacitor is thus
rcc-lf (4-55)
From Equation 4-51 we obtain

The value of £rF2b is a figure of merit for TCC = TJL +—- —

dielectrics. er dT AdT d dT (4-56)
All dielectric materials have a finite value for = TCD+aL
DC resistivity. Thus, a charged capacitor will where TCD is the temperature coefficient of the
discharge gradually through its own resistance dielectric constant and aL is the linear expan-
following sion coefficient. So the change in capacitance
G(0 = aexp(-;/T) (4-54) due to a change in temperature is caused by the
change of the capacitor's dimension, which is
where Q0 and Q(t) are, respectively, the initial usually not very significant, and by the change
charge and the charge remaining in the capaci- of the dielectric constant, which is significant,
tor at time t\ r is the dielectric relaxation time particularly at temperatures near the transition
of the material, which is equal to ere0p\ and p point.
is the resistivity of the material. A capacitor dis- Ferroelectric ceramics are commonly used
sipates its stored energy through DC leakage for capacitors. The Electronic Industries Asso-
resistance and, more importantly, the dielectric ciation (EIA) in the United States has grouped
losses under an AC electric field. These account ceramic capacitors into three classes:
for the dissipation factor or the loss tangent
tan<5 (see section 2.2.3 in Chapter 2). It should
Class I: The materials have a low dielectric
also be noted that the leads and the electrodes
constant (ranging from 15 to 500), low
always introduce an inductance L in the circuit.
dielectric losses (tan<5 < 0.003), a low TCC,
This is why a capacitor has a resonance fre-
and a low rate of aging of the capacitance
quency which may be very high, depending on
value. The working temperature range is
the value of L. For general applications, it is
-55°C to +85°C.
important to keep the value of the unwanted
stray inductance L as small as possible. Class II: The materials have medium to large
However, from Equations 4-52 and 4-53, dielectric constants (ranging from 500 to
both the volume efficiency and the maximum 20,000), with general properties primarily
energy stored in a capacitor increase propor- similar to BaTi0 3 ceramics. The dependence
tionally with increasing dielectric constant. of temperature, electric field, and frequency
Thus, ferroelectric materials with high dielec- is stronger than for Class I, and the TCC
tric constants are obviously ideal materials for value is higher.
use in miniature capacitors. But the capacitance Class III: The materials consist of a conduc-
of the capacitors made of ferroelectric materi- tive phase near the electrodes that reduces
als is dependent on temperature, applied elec- the effective thickness of the dielectric mate-
tric field, and frequency. Usually, it is not so rial. General properties are similar to those
difficult to tolerate a small change in electric of Class II, but their working voltage is
field and frequency, but the range of tempera- lower. The EIA has also introduced a scheme
ture variation is large under certain environ- for specifying the variability of the capaci-
mental conditions, and also the dielectric tance with temperature in the range of prac-
constant is strongly dependent on temperature, tical interest for high-permittivity capacitors.
particularly at temperatures near the transition Table 4-3 lists the EIA codes. For example,
point. So for high-permittivity capacitors, it is a capacitor of EIA code X7E implies that its
necessary to choose a material with a small capacitance may change by no more than
temperature coefficient of the capacitance ±4.7% in the range of temperatures from
(TCC), which is -55°C to +125°C.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 251

Table 4-3 The El A codes for the specification of high-permittivity capacitors.

EIA Code Range of Temperature Variation (°C) EIA Code Change of Capacitance (%)

XI -55 to +125 D -3.3 to +3.3

X5 -55 to +85 E -4.7 to +4.7
Y5 -30 to +85 F -7.5 to +7.5
Z5 +10 to +85 P -10.0 to + 10.0
R -15.0 to +15.0
S -22.0 to +22.0
T -33.0 to +22.0
U -56.0 to +22.0
V -82.0 to +22.0

Dielectric materials with dielectric constants log8r = Vfl l o g £ r / + V/2 l o g £ r 2 + V/3 l o g £ r 3 + . . •

larger than 1000 are mainly ferroelectric mate-
rials, and they are very sensitive to temperature, = I>^n (4-57)
i=\
electric field, and frequency. Obviously, the
development of techniques for modifying the where eri and vfi are, respectively, the dielectric
materials to improve stability and to retain constant and the fraction of the total volume of
the desirable high dielectric constant feature is the mixture of the /th constituent.
very important. In fact, this work has been In general, it is not possible to obtain a com-
carried out continuously for about 50 years. pletely flattened er - T curve in a homogeneous
In general, the dielectric constant is meas- ceramic having BaTi0 3 as a major constituent.
ured with a small AC signal. A bias of a large However, in the region below the peak, at tem-
DC field will cause the value to decrease, but peratures below Tc, the value of the dielectric
the field coefficient of the dielectric constant is constant does not change much with tempera-
normally small. Frequencies in the range of ture. This is due partly to the further change
0-100 MHz do not affect the dielectric constant in crystal structure to rhombohedral and then
to a great extent. The difference in dielectric to orthorhombic, and partly to the transition
constant between power and radio frequencies changing from first order to second order.95
is only 5-10%, and for most ceramics based on Some ceramics with different compositions,
BaTi0 3 and its isomorphs, the dielectric con- which would provide usefully high permittiv-
stant does not vary greatly with frequency until ity-temperature characteristics, are given here
it reaches the gigahertz range. The major as examples. Many similar mixtures have
concern is the temperature dependence of the already been developed.58'95'96
dielectric constant. To reduce the temperature
BaTi0 3 andBi 2 (Sn0 3 )3 (£ r =1000at25°C)
coefficient of the dielectric constant (TCD) is BaTi0 3 and MgSn0 3 (£r = 2000 at 25°C)
thus the goal that many researchers have BaTi0 3 , SrTi0 3 , and CaTi0 3 (sr = 2000 at 25°C)
attempted to reach. High-permittivity ceramic BaTi0 3 and MgF2 (£r = 3500 at 25°C)
capacitors are based mainly on ferroelectric BaTi0 3 , CaSn0 3 , and CaO (£r = 3500 at 25°C)
BaTi0 3 and CaZr0 3 (er = 5000 at 25°C)
BaTi0 3 , its isomorphs, and their solid solutions.
BaTi0 3 and CaSn0 3 (£r = 8000 at 25°C)
According to Lichtenecker's rule,94 the TCD of
a ceramic formed by the mixture of several Furthermore, the locally inhomogeneous
components will be equal to the volume microstructure tends to broaden and flatten the
average of the TCDs of its constituents, if there £r — T curve. This tendency is enhanced if the
is no chemical reaction between constituents growth of grain sizes during sintering can be
leading to the formation of new compounds. sufficiently inhibited.
Following this rule, the dielectric constant of There is a large variety of ceramic capaci-
the mixture can be predicted approximately by tors, including thin-film capacitors, thick-film
Lichtenecker's empirical formula capacitors, and plate capacitors (based on the
252 Dielectric Phenomena in Solids

thickness of the dielectric material), single- represent the points of thermal equilibrium for
layer discrete capacitors in disk or wafer, or ferroelectric materials. The material itself can
cylindrical tube form, or multilayer capacitors stabilize its own temperature, such as point b
based on the construction shape. The design near the transition temperature Tc where non-
depends primarily on the requirements of the linearity is high.99-101 The elements based on the
capacitors for certain specific applications. For changes in the dielectric constant, and hence
more details about capacitors, see references the capacitance with the change of the applied
58, 59, and 95-98. AC field, can be used to replace varactors as
circuit elements in frequency modulation, ther-
mostat control, etc.
Thermo-Autostabilization Nonlinear
Dielectric Elements (TANDEL)
When a ferroelectric specimen is switched on High-Energy Electrical Pulse Generators
to the ferroelectric phase with an alternating Ferroelectric ceramics possess a special feature
field, heat will be produced in the material each of high spontaneous polarization and high
cycle, part of which is due to normal dielectric dielectric constant, the latter being typically
losses and part to hysteresis losses. As a result, several hundred to several thousand, much
the power dissipation WA in the material larger than for organic solid insulators (usually
increases with temperature, reaches a peak below 10) and for inorganic solid insulators
at the transition temperature Tc, and then (below 20). Ferroelectric ceramics have already
decreases rapidly for T> Tc, as shown in Figure been used widely for energy storage capaci-
4-31. The value of WA increases also with tors.102 Ferroelectric ceramics also have a large
increasing applied AC field. There is heat WB piezoelectric constant, implying that the elec-
lost by the material through heat conduction tric field produced per unit mechanical stress
and convection to the surrounding medium, could be very large. For example, a compres-
which is proportional to temperature. Thus, at sion stress of about 0.05 Gpa would generate
the intersection points (such as a, b, c), when an open-circuit field of about S-lSkVcm"1.44
the WA-T curve intersects with the WB-T The direction of the applied stress for such
curve, the heat produced in the material is a piezoelectrically generated high voltage is
equal to the heat lost from the material to its parallel to the poling direction, and the depol-
surrounding medium. So the points a, b, c ing process is based on the linear piezoelectric
effect. However, exploring piezoelectrically
generated electric power beyond the linear limit
T
of the linear piezoelectric effect, Neilson103
was the first to suggest that the remanent (or
spontaneous) polarization of a poled ferroelec-
tric material could be released by a shock wave
of sufficiently high mechanical stress to cause
the transition from the ferroelectric phase
to an antiferroelectric one, thus resulting
yyy \ — WA(F) in the generation of high-energy voltage or
/ / UJ WA(F') current pulses. Since then, several investiga-
' ' WA{F") tors62'63104105 have studied further this method
for the generation of high-energy power sup-
Temperature Tc
plies actuated by a shock wave of mechanical
compression through a poled ferroelectric
Figure 4-31 The variation of the power dissipated in the
ferroelectric specimen WA and the power lost by the spec-
ceramic material.
imen to its surroundings WB with temperature at the applied The basic principle is simple. It is important
AC field F>F>F'. to choose a material with its ferroelectric phase
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 253

very close to the boundary between the reduced to zero and then removed from the
ferroelectric phase and the antiferroelectric specimen. Thus, the poled specimen has a
phase. The phase diagram of the PZT system remanent polarization Pr with a bound charge
shown in Figure 4-19 shows that the PZT 95/5 Qb on the electrodes to balance the polarization
material with x = 0.05 in the composition charge in the specimen, thus keeping the whole
PbZri_xTix03 should be suitable for this appli- system neutral, as shown in Figure 4-32(a). If
cation. This material is in the ferroelectric a shock wave of compressive stress is now
phase (F) with the rhombohedral structure applied to the specimen in the direction per-
between 50°C and 220°C. It becomes antifer- pendicular to the direction of the polarization,
roelectric (AF) with an orthorhombic structure then the original ferroelectric phase will change
at temperature below 50°C. There is a phase to an antiferroelectric phase, in which Pr —> 0,
boundary between the F and AF phases that is and the bound charge Qb on the electrodes will
temperature dependent. However, external become free charge Q, ready to flow out of the
electrical field tends to broaden the ferroelec- system, as shown in Figure 4-32(b).
tric phase, implying that the F-AF phase Based on this principle, we can generate
boundary tends to move toward the left side. high-energy electric pulses by a mechanical
This means that external electric field may shock wave. Let us consider a simple mechan-
convert the original antiferroelectric phase to a ical-electrical energy conversion system, as
ferroelectric phase. A mechanical (compres- shown in Figure 4-33(a), with three sides xa, y09
sion) stress tends to extend the antiferroelectric and z0- The shock wave of the compressive
phase region, implying that the F-AF phase stress travels in the z direction, perpendicular to
boundary tends to move to the right side. This the poling direction along the x-axis, with the
means that a mechanical shock wave may velocity u. In this case, the operation is in
convert the original ferroelectric phase to an the normal mode. It takes r = zDlu second for the
antiferroelectric phase. wavefront to travel across the specimen of
A poled specimen is a specimen that has been width Zo- So, when the wave front has traveled
polarized by a sufficiently high electric field F for time t, the portion of specimen up to the
until the spontaneous polarization reaches the dashed line in Figure 4-33(a) has already been
saturation value Pp (see Figure 4-4). Usually, depoled. The equivalent circuit of the system is
the polarizing field is kept across the specimen shown in Figure 4-33(b). The system can be
for a period of time to ensure that all of the considered a current generator with an internal
spontaneously polarized domains are aligned in capacitance Ca, resistance R0, and inductance
the direction of the field before the field is L0. The value of L0 depends on the amount of

-Qb = -PrA (Bound Charge) -Q = -Qb (Free Charge)

Electrode _ _ _\_
,11e e i i e
i efe i e
11 ^ °™^r A |TTTTTT]
* * * * * *
* * * * * *
* * * * * *
1******1
+ + + + + +
/
+Qb = +PrA (Bound Charge)
+Q=+Qb (Free Charge)
(a)
(b)
Figure 4-32 (a) Poled ferroelectric ceramic specimen with a bound charge Qb on the metallic electrodes, (b) After the
impact of a mechanical shock wave, the original ferroelectric phase changes to an antiferroelectric phase, and the bound
charge on the electrodes is released and becomes a free charge.
254 Dielectric Phenomena in Solids

The current i(t) is thus

i{t) = dQ(t)/dt = PrA/z 0<t<r
(4-63)
(a) -0 t>T

The variation of V and i with time is shown in

Figure 4-34 (solid-line curves). However, the
ideal case never occurs experimentally because
there is always leakage current flowing through
X the specimen. So Q(t) should include a loss
factor, and thus Q(t) should be expressed as
Q{t) = PrA(t/T)/(1 -at) 0< t< r
(b) _n_(T>/ ~ L I i i l L1! = PrA(l-az) t>T
(4-64)
1 i l l < ( \ The variation of V and i will change accord-
ingly. The dashed-line curves shown in Figure
Figure 4-33 (a) Schematic diagram of the mechanical-
electrical energy conversion system and (b) the equivalent
4-34 represent V and / for a nonideal case.
circuit. We carried out a simple experiment to
demonstrate how this type of high-energy pulse
generator would perform. We used PZT with
leakage current flowing through the specimen. the composition of PbZr0.975Tio.o2503 plus 1 wt%
The external load impedance may consist of Nb 2 0 5 , because this material has the boundary
capacitance CL alone, CL in parallel with resist- between the F and AF phases at about room
ance RL, or CL in parallel with inductance LL. temperature. The phase diagram of this is
However, it is important to have CL always con- shown in Figure 4-35(a). Its general properties
nected to the circuit to protect the system in case
are as follows63:
breakdown occurs during the phase transition.
For a simple case, we can assume R0 = • Density = 7.6 g cm"3
oo, L0 = °°, zL consists of CL only and the per- • Dielectric constant = 1600 (high field)
mittivity of the specimen remains the same just
• Loss factor tan 8 = 3 x 10~2
after depoling. Then we have
• Curie temperature Tc = 215°C
C0=ey0z0/x0 0<t<r (4-58)
The charge released for t = 0 to t = r is
Q(t) = q(t) in the system + qL(t) in the load
(4-59)
and the voltage between the electrodes is
V(t) = q(t)/C0=qL(t)/CL
(4-60)
= GW/(C + Q )
Since
Q(t) = PrA(t/r) = Pry0z0 (t/r) 0 < t < r
= PrA = Pry0z0 t>%
(4-61)
0 x Time t
SO
Figure 4-34 V and i as functions of time. Solid-line
V(t) = PrA(t/r)/(C0 + CL) 0<t<z curves are for the ideal case and the dashed line curves for
(4-62)
= PrA/(C0 + CL) t>T a nonideal case.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 255

Figure 4-35 (a) Phase diagram of the PbZiv xTix03 + 1 Wt% Nb 2 0 5 system, (b) Short-circuit current-time curve.
(c) Voltage-time curve.

• Electromechanical coupling factor kp = 0.1 is 107kV, which again is very close to the
• Piezoelectric coefficient d33 = 1 x 10~12 calculated value of lllkV.
coulomb/newton It should be noted that most ceramics are not
good insulators. They consist of many pores and
The shock wave of the compressive stress was other defects, and generally they have a low
produced by an explosive system consisting of resistivity and low breakdown strength. During
an initiating detonator, a plane wave generator, depoling, a high voltage would be developed
and a spacer to control the magnitude of the in the depoled section, which may enhance
stress. The stress used for this investigation filamentary carrier injection from the electrodes,
was about 3Gpa, the shock wave velocity leading to local heating in certain regions. All of
was 0.385 cmjUs-1, and the width of the wave these factors may operate simultaneously to
was much longer than T. This compressive limit and affect the performance of the system.
stress was sufficient to cause all remanent
polarization (bound charge) to be released.
Some typical results are briefly described as
Memories
follows.
The bistable polarization of ferroelectrics can
be used for binary memory systems in the same
Short-Circuit Current way as the bistable magnetization of ferromag-
We used specimens with x0 = 0.5 cm, y0 = netics. The memory is nonvolatile and does not
28 cm, and zQ = 1 cm, so T=z0lu = 2.6 /us. Since require a holding voltage. The idea of using
Pr = 32.3 \xCcm"2, the total charge on the elec- ferroelectric bistable polarization for memories
trode is Qb = Pry0zo = 904.4 jUC. For this system, was first suggested by Anderson.106
C0 = sr£0y0ZoX0~l = 8.4 x 10"9F. Thus, the charge The basic principle is simple. A ferroelectric
Q(t) = Qbt^'1 and the current i = dQIdt = matrix store is shown schematically in Figure
Qbr~l for 0 < t < T, and i = 0 for t > T. The result 4-36 (a). The row and column electrodes are on
is shown in Figure 4-35(b). The experimental opposite surfaces of the ferroelectric specimen.
value of 337 A for / is close to the calculated Thus, there are a number of square portions of
value of 904.4/2.6 = 348A. the specimen having electrodes on both sur-
faces; each portion is one cell of the memory.
Voltage across Load Impedance Consisting The cell may have two remanent polarization
of Q and RL states (+Pr or -P r ). To write can be achieved by
For this experiment we used specimens with applying an electric field pulse Fx to the row
x0 = 4 cm, y0 = 4 cm, and z0=l cm. In this case, electrode and a pulse Fy to the column elec-
C0 = 1.5 x 10"10F, we used CL = 9.75 x 10"10F trode, both additively to produce a polarization
and RL = 39 kQ. The result is shown in Figure of full magnitude 2Pr in the cell, and so it is
4-35(c). The experimental value of V across RL dressed. To read, similar pulses Fx and Fy (but
256 Dielectric Phenomena in Solids

(a)

^ t .1 - c
1 1 2

f*
Fyl

Writing Reading
IFL+IFL

(c) (d)

Figure 4-36 (a) Electrode arrangement for a ferroelectric matrix store, (b) electrical circuit for the application of voltages
(or electric field) pulses to the electrodes, (c) the polarities of electric field pulses for Writing and for Reading, and (d) the
total polarization 2Pr attained by simultaneous application of Fx and Fy pulses additively.

with the polarities opposite to those for writing) roelectric stores are similar to ferrite stores, but
are applied to the electrodes to depole the polar- the former have some major shortcomings,
ized cell, so the stored charge 2Pr will be mainly the low switching speed, the decay of
released generating a current or a voltage pulse the stored signal with repeated uses due to the
across a registered resistor. inherent fatigue behavior, and the aging effect.
Figure 4-36 (b) through (d) illustrates this Many suggestions have been put forward
writing and reading process. It can be seen that to deal with these shortcomings.107"110 For
the cells receiving a single pulse Fx or Fy are example, the combination of thin ferroelectric
hardly affected; the two pulses must be coinci- film technology with silicon-based memory
dent for both the writing and the reading. It technology offers the potential of combining
should be noted that for a nondestructive nonvolatility with the fast read and write char-
readout, arrangements must be made to restore acteristics of dynamic random access memory
the original state of a cell after reading it. Fer- (DRAM) and the small cell size of the
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 257

memory.108110 For certain applications, such movement actuated by an electric field (acting
as for satellites, the radiation hardness of like a motor). There are two principal mecha-
ferroelectric materials is a very attractive nisms for piezoelectricity. Based on the first
feature.108109 Depending on the requirements of mechanism, the dipole moments may just
the particular application, ferroelectric memo- mutually cancel each other in the material
ries have their own merit and deserve further under the unstrained condition. Piezoelectricity
development. By comparison, however, it ap- may occur if the crystal has no center of sym-
pears that in some respects ferroelectric memo- metry, and in this case the relation between the
ries cannot compete with their well established electric field and the mechanical strain is linear
magnetic and semiconductor counterparts. As in the first approximation. This linear relation
a result, attention has been shifted to optical is sometimes referred to as the linear piezo-
memories. For some excellent literature on electric effect. However, based on the second
optical memories, see references.98111"114 mechanism, the dipole moment components
may remain, but they add to a resulting moment
along a polar axis of the unit cell, so the occur-
4.3 Piezoelectric Phenomena
rence of piezoelectricity is accompanied by
Crystals formed by polar molecules with a non- pyroelectricity, involving spontaneous polar-
centrosymmetic structure (i.e., without a center ization. Thus, for ferroelectric piezoelectricity,
of symmetry) will exhibit a piezoelectric effect. the variation of the mechanical strain with the
This implies that a mechanical stress applied to applied electric field follows the change of
the crystal specimen will create an overall polarization in the hysteresis loop, as shown in
polarization and hence a voltage across it. The Figure 4-37. During the poling process, there is
reverse of the stress direction will cause the a small expansion of the material along the
reverse of the polarity of the polarization and poling direction and a contraction in directions
hence the voltage. The piezoelectric effect is (lateral directions) perpendicular to it. So the
convertible. This means that an applied electric strain along the poling direction is positive. In
field will create a mechanical strain, expansion, Figure 4-37, the solid line traces the poling
or contraction, depending on the direction of strain from a virgin state 0 to saturation state
the field. C and remanent state D. The lateral strain is
In general, an applied electric field always negative. Both the poling and the lateral
causes mechanical distortion in the geometric strain-field relations form a butterfly-shaped
shape of the material, because matter is con- hysteresis. Figure 4-37 shows the basic differ-
structed of charged nuclei surrounded by a ences among electrostriction, linear piezoelec-
compensating electron cloud. The polarization tricity, and ferroelectric piezoelectricity.
induced by the applied field will cause changes In general, ferroelectric piezoelectrics, such
in charge distribution and hence mechanical as ceramics, have advantages for use in trans-
distortion. The strain resulting from the ducers because they have a large piezoelectric
mechanical distortion is proportional to the coefficient, especially near the transition tem-
square of the field, and this phenomenon, called perature, and a high dielectric constant that
electrostriction, is not inversive. This means allows the electro-mechanical coupling factor
that a mechanical stress acting on mass points to approach unity. In comparison with non-
cannot induce dipole moments from the neutral ferroelectric materials, ferroelectrics may have
state of the material. Electrostriction occurs in a high coupling constant, but they also have
all materials, although its effect, for most prac- high dielectric losses.
tical cases, is extremely or negligibly small.
The direct effect of the piezoelectricity is the 4.3.1 Phenomenological Approach to
generation of electric polarization by a mechan- Piezoelectric Effects
ical stress (acting like a generator), while the Thermodynamically reversive interactions
converse or inverse effect is the mechanical occur among the electrical, mechanical, and
258 Dielectric Phenomena in Solids

Lateral Direction
0,

Strain Strain

of C

r 4.f
Electrostriction Ideal Piezoelectricity
Poling Direction

Ferroelectric
Piezoelectricity
tJ
Figure 4-37 Schematic illustration of the variation of mechanical strain with electric field for electrostriction, ideal piezo-
electricity, and ferroelectric piezoelectricity in poling and lateral directions.

thermal properties, as illustrated in Figure 4-38. Thus, the displacement induced by the stress
The lines joining pairs of circles indicate that a can be expressed as
small change in one of the variables causes a
D = dX (4-68)
corresponding change in the others. The three
direct relations are as follows. where d is one of the piezoelectric constants (or
The relation between electric field F and dis- coefficients) whose unit is coulombs/newton.
placement D defines the permittivity e In the converse piezoelectric effect, we can
also express the relation between electric field
D = sF (4-65) and the strain as
The relation between mechanical stress X and (4-69)
h = dF
strain h defines the elastic compliance or elastic
constant y (v = 1/F, where Y is the modulus of In this case, the unit for d is meters/volt. In fact,
elasticity or the elastic stiffness coefficients). for both the piezoelectric and the converse
piezoelectric effects, d is identical, so
h = yX (4-66)
The relation between temperature T and d- — - — (4-70)
~ X~ F
entropy S defines the heat capacity Q
In terms of the electric field produced by a
S = QT~ (4-67) mechanical stress, the relation becomes
The lines joining the circles at different corners F = gX (4-71)
define the coupling effects. For ordinary solids,
In this case, g is another one of the piezoelec-
the properties follow Equations 4-65 through 4-
tric constants. Its unit is [volts/meter]/
67. But for piezoelectric materials, additional
[newtons/meter2]. It can be expressed as
terms are required. For example, a mechanical
stress causes not only a strain, but also electric d_ d
(4-72)
polarization, even at a constant temperature.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 259

ELECTRICAL

( Field \
F
\

/ /w
/ /£•
/ ">
/ '**
/ "E
/ *w / »—
/ O / CD
> / CL

/
// ,*§
*** .p / X^"^X \ $ \
/ -^ i
' /Displacement]
/ \ \&W \
// / \ \ CL \
^y •© /
&/ \\
\
D
/
// \W
o
\ "2- \*
/ /
f/ i7 / A^ \ <<i
) /
<S/f
/
Sy \\
\ \\
\ <&
w
\
\
<S>

/ \
/ \
/ \
/. \
/ \
/ Strain Entropy \
h S \
\
3*^ \
Etfetf Their ^a/
p\ezoca\ori<> £*P<)ansfo
n Tempe-
Stress rature
X Thermoelastic Effects T

MECHANICAL THERMAL

Figure 4-38 The relations between the mechanical, electrical, and thermal properties for crystals with a noncentrosym-
metric structure (without a center of symmetry).

There are two other piezoelectric constants,

namely e and/, which are only occasionally \dXJFT
used. Their relation is
8=
X = -eF (4-73) ~{dx)DT
(4-75)
F = -fh (4-74)
dh ) F T
Rigorous analysis and discussion of these
piezoelectric constants (or coefficients) has \dh)DJ
been reported by Mason115 and Jaffe and Berlin-
court.53 These four piezoelectric constants can Similarly, the piezoelectric constants for the
be defined in partial derivatives as converse piezoelectric effects can be written as
260 Dielectric Phenomena in Solids

resulting from the direct piezoelectric effect, or

2 _ Mechanical energy from conversion
Total input electrical energy
(4-83)
(4 ?6)
, (3X< " resulting from the converse piezoelectric effect.
Since the electro-mechanical conversion is
always incomplete, k is always less than unity.
Typical values of k are 0.1 for quartz, 0.4 for
barium titanate, 0.7 for PZT, and close to 0.9
Based on the thermodynamic argument, d = d\ for Rochelle salt.
g = g\e = e\mdf=f. In general, piezoelectric properties are
The interactions between the electrical and dependent on orientational direction, so they
the elastic variables describe the piezoelectric must be described in terms of tensors. A con-
effects. The equations of state relating the elec- venient way to specify the directional proper-
trical and the elastic variables can be written in ties is to use subscripts that define the direction
general form as and orientation, as illustrated in Figure 4-39.
The subscript 3 refers to the polar axis (or
D = dX + exF (4-77) poling axis); 1 and 2 refer to arbitrarily chosen
F
h = y X + dF (4-78) orthogonal axes perpendicular to 3. Subscripts
4,5, and 6 refer the shear planes of the mechan-
where the superscripts denote the parameters ical stress and strain perpendicular to axes 1, 2,
held constant, so ex denotes the permittivity at and 3, respectively. For example, subscript 4
constant stress X and yF denotes the elastic indicates the change of angle of the stress or the
compliance at constant electric field F. For the strain between the two initially orthogonal axes
ferroelectric piezoelectricity, the properties are 2 and 3 in the shear plane (subscript 4) normal
nonlinear, so the equations of state should be to axis 1. Similar meaning is applied to sub-
written in differential form as scripts 5 and 6. Stress and strain due to shear-
8D = d8X + ex8F (4-79) ing action are referred to as the shear stress and
shear strain.
8h = yF8X + d8F (4-80)
Piezoelectric coefficients are usually indi-
The piezoelectric coefficients are inter- cated with two subscripts denoting the direction
dependent. For example, of the properties. The first subscript refers to the
direction of the electric field F (or the dis-
d x
— = £X placement D). The second subscript refers to
! (4-81) the direction of the mechanical stress X (or the
strain h). For example,

d3l
To specify the overall strength of the piezo-
electric effect, it is convenient to use the t\
electro-mechanical coupling factor k, which Stress or strain direction normal to the field direction
Field (or displacement) direction
can be considered the direct way to measure the
ability of piezoelectric materials to convert one
form of energy to another. This coupling factor ^Stress or strain direction parallel to the field direction
k is defined as
Field (or displacement) direction
2 _ Electrical energy from conversion
Total input mechanical energy In most cases, it can be assumed that the plane
(4-82) perpendicular to the polar axis is isotropic. This
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 261

Figure 4-39 Subscript symbols for the notation of the directional properties for poled piezoelectric materials.

implies that an electric field parallel to polar components being negligible, so one subscript
axis 3 interacts in the same way, with the is sufficient. For example, £33 = £3, e31 = eu
mechanical stress either along axis 1 or along £32 = £2, etc. indicating that £3 is the permittiv-
axis 2. Thus, d31 = d32, di5 = d24, yn = y23, etc. ity along the polar axis 3, and ex and £2 are the
It should be noted that shear can only occur permittivities along axes 1 and 2, respectively.
when an electric field is applied at right angle Following the above conventions, Equation
to the polar axis, so there is only one shear 4-81 becomes
piezoelectric coefficient di5 (since dX5 = d24).
dtj _ (dhj/dFl)x,T
There are also piezoelectric coefficients corre-
8ij
sponding to hydrostatic stress. In this case, dh = (4-84)
^33 + d3i + d32 = d33 + 2d3X. Similar notations are
used for other piezoelectric coefficients, such as v dFi )X,T

£33, g3u gis, gh, etc.

Elastic properties can also be specified in For example,
terms cf elastic compliance y, such as yn, yl2, ^33
= £3X3=£3*
J13, J335 J44, J66» to denote the interaction of a #33
strain and a stress under the condition of con-
d3\
stant field. Thus y should be written as yfi, yf2, = e3i=e'
gsi
yk yL yu, yL (usually for F = 0). Each y
relates to the application of a single stress Thus, Equation 4-79, due to the direct piezo-
in one direction; the other ys are kept fixed so electric effect, becomes
that there is no lateral restraint. Stress and strain
8I\= dl58X3 + £xidFx
are interchangeable, so yi3 = y31. The permittiv-
ity e should also be a tensor, but for most cases, 8D2 = dl58X4 + ex28F2 (4-85)
x
the field has only one direction, oblique 5A = d3l(8X, + 8X2) + d338X3 + e 38F3
262 Dielectric Phenomena in Solids

Equation 4-80, due to the converse piezoelec- fR: The resonant frequency at which Xs = 0
tric effect, becomes (XL = Xc)\ in this case, XL - Xc increases with
frequency for frequencies below fR
Sh = y?Mx + y?28X2 + y[38X3 + d3lSF3
fA: The antiresonant frequency at which Xs = 0
8h2 = y[l8X2 + ynSXl + y[38X3 + d32SF3
(XL = Xc), but in this case, XL - Xc decreases
8h3=y[3(8Xl+8X2) + y3F38X3 + d338F3 with the frequency for frequencies around ^
(4-86)
8h4 = y448X4 + di58F2 fp: The frequency at which the resistive
8th = yb8X5 + dl58F{ component Rs is a maximum
8K = yL8X6 fs: The frequency at which the series arm of the
upper portion of Figure 4-40(b) has zero
4.3.2 Piezoelectric Parameters and reactance (Xm = 0)
Their Measurements
It can be seen from the equivalent circuit and
The best-known crystal showing a piezoelectric
its frequency response in Figure 4-40 that the
effect is quartz (Si0 2 ). Lord Kelvin115 was the
piezoelectric specimen stores energy mechani-
first to develop a model to explain this phe-
cally in one part of the cycle and feeds it back
nomenon. For a quantitative analysis of the
into the electrical portion in another part of the
piezoelectric effect, it is necessary to know first
cycle. The specimen operates in sequence as an
the piezoelectric and dielectric properties of the
electro-mechanical transducer.
material. In the previous section, it was shown
It can be shown from the equivalent circuit
that piezoelectric properties are primarily
Figure 4-40(b) and (c) that
governed by the electro-mechanical coupling
coefficient k\ the piezoelectric coefficients,
mainly d and g\ the elastic compliance y; and /* = iK^LmCm (4-87)
the dielectric constant er. These parameters are
generally dependent on the crystal structure and 1 \cm + ce (4-88)
IA=:
the direction of the crystal axes, so it is impor- 2K \ LmCmCe
tant to know these parameters. The convenient The ratio of fA/fR is related to the electro-
way to determine these parameters is to use an mechanical coupling factor by the following
equivalent circuit which consists of two por- equation119
tions: one involving an analog between
mechanical and electrical quantities repre- Ai =1+ . (4-89)
senting mechanical impedance, and the other fRJ ' n2U-k2
representing the electrical impedance.4,116"118 Thus, from Equations 4-87 through 4-89, we
The analog between mechanical and electrical obtain
quantities follows.
Cm JA ~ IR
(4-90)
Mechanical Quantity Electrical Quantity
ce + cm fi
Elastic compliance: y Capacitance: Cm Equation 4-90 gives good approximations,
Mechanical mass: m Inductance: Lm provided that the quality factor Q of the speci-
Viscous damping: T Resistance: Rm
men is sufficiently high. Q is the ratio of the
First, we measure the resonance frequencies of energy stored to the energy dissipated per half-
the piezoelectric specimen under investigation cycle, which is equal to (tan<5)-1. For most
with a sinusoidally varying electric field, as dielectric materials, tan 8 is very small. For
shown in Figure 4-40(a). By comparing the example, when tan 8 <0.01, Q > 100. In Figure
phase of the AC voltage with that of the AC 4-40(b), we have already assumed Q > 100,
current, it is easy to determine the frequencies so only the capacitance Ce is included in the
at which Xs = XL - Xc > 0 or Xs < 0 or Xs = 0. electrical portion, the leakage resistance being
The various frequencies are defined as follows: ignored.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 263

Piezoelectric
Specimen

Ammeter
c^-^TOP - j | — AAM-

AC Voltage
Source
(a) (c)

r-^JW- -AAM-
Mass
Compliance \
Damping

Capacitance«

Ce

(b) (d)

Figure 4-40 (a) A piezoelectric specimen under an alternating electric field, (b) equivalent circuit of (a) near the funda-
mental resonances (the top branch of the circuit represents the mechanical portion and the bottom branch represents the
electrical portion), (c) the equivalent series circuit of the impedance of (b), and (d) frequency response near the fundamental
resonances.

Once the parameter k has been determined, Thus, we can write

the other parameters can be determined easily.
The mechanical energy stored in the specimen (4-93)
per unit volume is given by Ue ye

The relations between fR and fA and the elastic

Um (4-91) compliance y are given by
2 2 y

and the electrical energy stored per unit volume 1

/« = (4-94)
is 2l{pyFT

Ue=\\D\\F\=\eF> (4-92) h = 2l{pyDf2 (4-95)

264 Dielectric Phenomena in Solids

where p is the density of the specimen material some special features of some piezoelectric
and € is the length of the specimen. The super- materials.
script F in yF means y measured at constant Quartz is a commonly used piezoelectric
field, signifying that the specimen is short- crystal, but crystals of the water-soluble type
circuited, while the superscript D refers to y at have a much larger electromechanical coupling
constant displacement, signifying that the spec- factor k. For example, ammonium dihydrogen
imen is open-circuited. The permittivity £ can phosphate (ADP) has k =* 0.30. Still higher
be determined from the capacitance measured coupling can be achieved with ferroelectrics
at a frequency well below resonance. Thus, such as PZT (a PbTi0 3 and PbZr0 3 mixture).
from Equations 4-93 through 4-95, we can The PZT specimen, after being poled with a
determine v, d, and g, since g = die. However, sufficiently strong electric field, exhibits high
the piezoelectric parameters depend on the geo- values of kp and d depending on the com-
metric shape and dimensions of the specimen. position, as shown in Figure 4-41. This PZT
The geometry commonly used is a thin disc system has been studied by several investiga-
of diameter dia and thickness t, which is much tors.44120~124 Aliovalent substituents would
smaller than dia. Metallic electrodes are modify the properties of ceramics with a Per-
deposited on both surfaces of the specimen, and ovskite structure. In order to achieve a given set
the specimen is poled in a direction perpendi- of properties, PZT-based ceramics may require
cular to the surfaces. In this case, the planar composition of more than one type of additive.
(radial) electromechanical coupling factor kp One system that has been widely used for
is of a radial mode, excited through the piezoelectric devices is PZT doped with lan-
piezoelectric effect across the thickness of the thanum to form a new PLZT system. Such
disc. kp is related to the resonant frequencies^, systems usually contain 3-12mol.% La and
and/ 5 , defined by the following relation 5-30mol.% Ti. For the system with the com-
position of La: Zr: Ti=7:60:40, the values of
fp-fs kp and d33 have reached 0.72 and 7.1 x 10"10
= F\j0Juv (4-96)
l-k2n fs rn/V.125,126 Such high values of kp and d33 have
where J0 and J\ are the Bessel functions of the been attributed to the compositions located
orders of zero and one, respectively; and v is within the boundary region between the
Poisson's ratio. The other piezoelectric param- rhombohedral and tetragonal phases.127-129
eters d and g can be determined in the same PLZT may also contain vacancies on B (as
manner as for the thick specimens, but the
expressions for the thin disc are more
0.40 160
complex.96 For more details about piezoelectric I Rhombohedral/1
r 1 i T
Tetragonal
1

parameters, see references 44, 59, 96, 120, 0.35 140

121. 0.30 - 120
kp
0.25 /'\ 100
/ "33
4.3.3 Piezoelectric Materials 0.20
/ / \\ 1 80
The most important piezoelectric materials are 0.15 \ J 60
the ceramics consisting of crystallites of Per- I '' " ' ' \\ A 1\ 40
0.10 vA
ovskite structure, such as PZT and PLZT; the * \\ 1
0.05 w\
S \ 20
synthetic polymers, such as PVDF; and the
I I I I
A
r\\ i i i
1
1
ceramic-polymer composites. Table 4-4 lists PbZr0 3 Mole % PbTi0 3
the typical values of some important para- Increasing PbTi0 3 content - •
meters of some commonly used piezoelectric
materials. All ferroelectric materials are
Figure 4-41 The planar electromechanical coupling
piezoelectric, but not all piezoelectric materials factor kp and piezoelectric coefficient d as functions of the
are ferroelectric. We will describe briefly composition of the PZT ceramics.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 265

Table 4-4 Selected properties of some piezoelectric materials: d = piezoelectric constant, k - electro-mechanical
coupling factor, and p = density.

Material (Abbreviation) Chemical Formula d (cm Y-1) P f g cm 3)

Silicon Dioxide (Quartz) Si0 2 2.25 x lO"10 0.099 2.65

Potassium-Sodium Tartrate-Tetrahydrate KNaC 4 H 4 0 6 -4H 2 0 5.30 x 10"9 0.78 1.77

(Rochelle Salt)

Barium Titanate BaTi0 3 1.90 xl(T 8 0.38 5.70

Ammonium Dihydrogen Phosphate (ADP) AH 2 P0 4 2.46 x 10"9 0.29 1.80

9
Potassium Dihydrogen Phosphate (KDP) KH 2 P0 4 1.07 x 10" 0.12 2.31

Triglycine Sulfate (TGS) (NH2CH2COOH)3*H2S04 5.00 x 10"9 — 1.69

9
Potassium Niobate or Sodium Niobate KNb0 3 or NaNb0 3 4.90 x 10~ 0.42 4.45

Lithium Niobate LiNbO, 0.85 x 10~10 0.035 4.64

Lithium Tantalate (LT) LiTa0 3 3.00 x 10"10 0.10 7.46

Lead Titanate PbTiO, 7.40 x lO"10 — 7.12

8
Lead Zirconate Titanate (PZT) Pb(Zr1_JCTiJ03 2.34 xlO" 0.66 7.70

Polyvinyl Chloride (PVC) see Table 5-2 0.70 x lO"10 — 1.40

10
Polyvinyl Fluoride (PVF) see Table 5-2 1.00 x 10~ 0.03 —
Polyvinylidene Fluoride (PVDF or Kynar) see Table 5-2 4.00 x 10"10 0.12 1.78

well as on A) sites, providing a favorable way

15
to change their polar states under the influence
of applied fields.
Polymer films have the following advantages:
o 10
they are flexible and tough, they can be made
very thin (<10 jjm) and large in area, they can be
shaped into any geometric forms, they have a
low mechanical impedance, etc. Polymers,
particularly PVDF, have been used widely for
various transducers and sensors. For PVDF, the
piezoelectric coefficient d and the pyroelectric 50 100
coefficient p are closely related to the poling Poling Field (MVrrr 1 )
parameters. They increase linearly with increas-
ing poling field Fp, poling temperature Tp9 and Figure 4-42 The piezoelectric coefficient dn of PVDF at
poling time tp, as shown in Figure 4-42. The data room temperature as a function of poling field and poling
are from Murayama et al.130,131 temperature for the fixed poling time of 30 minutes.
Apart from PVDF, other polymers, such as
polyvinyl fluoride (PVF), polyvinyl chloride trie coefficients when poled at high tempera-
(PVC), PTFE, etc., also exhibit piezoelectric tures. It should be noted that the properties of
effects. PVC is a noncrystalline polymer but polymers depend strongly on the preparation
has relatively high piezoelectric and pyroelec- techniques and conditions.132
266 Dielectric Phenomena in Solids

To produce materials with optimal properties Force

for a special application, we may choose a suit-
able ceramic-polymer composite. In general, a
composite refers to a mixture in which ferro-
electric ceramic particles are dispersed in a
polymer matrix. The properties are largely
determined by the choice of components, their
Electrodes
relative composition, and the manner in which
they are interconnected. The composites made
by mixing PZT ceramic, PVDF polymer, and
fluorinated rubber have a good piezoelectric
coefficient, depending on the composition.133
Ceramic-polymer composites have been devel-
oped mainly to search for a suitable combina-
tion that can be adapted for piezoelectric Force
or pyroelectric devices to be used in stringent
environments, such as sonar devices in water, Figure 4-43 Schematic diagram illustrating the principle
sensors for medical diagnostics, etc. of a gas igniter.

4.3.4 Applications of Piezoelectrics gradually would leak away through the leakage
Both the direct and converse piezoelectric across the electrode, instead of building up to
effects can be used for many practical applica- generate a high voltage across the spark gap.
tions, such as high-voltage generators, cigarette This is why igniters are of the momentary
lighters, gas igniters, etc.; transducers for high- impact type rather than the slow squeeze type.
intensity ultrasounds employed in ultrasonic Since most igniters have two pieces of poled
cleaning, ultrasonic therapy, medical diagnosis, ceramic elements connected back to back, we
delay lines, etc.; resonators for filters and oscil- need consider only one for our analysis. Based
lators; microphones; various sensors; etc. The on Equation 4-71, the voltage developed across
following sections use four simple and typical the ceramic element when an impact force is
examples to demonstrate the importance of applied is given by
piezoelectric effects.
V = F£ = g33lX = g33£^- (4-97)
A
Gas Igniters where y is the force and € and A are, respec-
A schematic diagram of a poled piezoelectric tively, the length and the area of the ceramic
ceramic igniter is shown in Figure 4-43, which element. Ignoring all possible losses in the
is self-explanatory. The principle is simple. operation, the energy that may be dissipated in
When a force is applied to the poled ceramic, a the spark is
voltage will be generated between electrodes.
It is usual to use two pieces of poled ceramic W = -e33-(g33£^-) = -d33g633 —
elements connected back to back in order to 2 £V AJ 2 A
(4-98)
double the charges released for the spark. On
releasing the force, a voltage of about the same If two elements are connected back to back (in
magnitude, but of the opposite sign, will be fact, they are connected in parallel), the total
produced, yielding another spark across the gas energy available for the spark is two times that
gap. If the force were applied too slowly, no shown in Equation 4-98. PZT and PLZT are
spark would occur because the charges released commonly used for gas igniters.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 267

Delay Lines Piezoelectric Positioners and Actuators

The transmission of a microwave signal can be Piezoelectric positioners and actuators have
delayed by the following steps: been used widely in areas related to precision
position controlling, vibration damping, relays,
1. The microwave signal is guided through a
phonograph pickup, pressure sensing, etc.135"137
coupling network to a piezoelectric trans-
The basic principle is simple. A thin piezoelec-
ducer, which converts it to an acoustic wave.
tric plate with metallic electrodes deposited
2. This acoustic wave then propagates in a on both surfaces can be used as a mechanical
dielectric medium and reflects when it positioner or relay when the required move-
reaches the end. ments are very small (e.g., a few micrometers
3. This reflected acoustic wave will be con- or less).
verted back to the original microwave signal Figure 4-45 shows schematically a poled rec-
by the transducer, as shown in Figure 4-44. tangular piezoelectric plate with the following
The major advantage of this type of delay line dimensions: length a, width b, and thickness c,
is the use of the transducer to convert the which are in the x, v, and z directions, respec-
high-speed microwave signal to a low-speed tively. When a DC voltage V is applied across
acoustic wave, and of the dielectric medium to the thickness in the z direction (which is also
produce the time delay. Since the speed of an the poled direction), then the plate will expand
acoustic wave is about 105 times slower than an or contract in the z direction, and contract or
electromagnetic wave, this type of delay line expand in the x and y directions, depending on
can be made much smaller in size and weight the polarity of the voltage with respect to the
than a conventional one.134 However, for the poled direction. The magnitude of such move-
design of such delay lines, it is important to ments can be easily calculated. From Equation
consider the following factors: 4-70, we can write

• The dielectric medium must have a low

acoustic attenuation for the acoustic wave to U a
propagate.
• The transducer must have a high electro-
mechanical coupling factor k and a high lon-
gitudinal mode acoustic wave velocity for its
use at microwave frequencies.
/ t
• The acoustic impedance matching between
the dielectric medium and the transducer must
t£
be good, and the insertion loss must be low. (a)
V>0
d Acoustic
Wave
Microwave
Input

H h-Aa/2
(b)
Microwave
Output Figure 4-45 Schematic illustration of the deformation of
a poled piezoelectric plate under an electric field: (a) the
Figure 4-44 The structure of a microwave delay line; a: original shape of the plate at V = 0 and (b) the deformation
electrode A, b: electrode B, c: transducer, and d: dielectric at V > 0 and V < 0. (The deformation is exaggerated for
medium. clarity.)
268 Dielectric Phenomena in Solids

_ h _ Ac/c _ Ac
^
(4-99) ^1
33
~~F~ V/c ~ V~

or
Ac = d~V
4P ^3
31-
This indicates that the change in thickness (i.e., V=0
the movement) depends on the piezoelectric
coefficient d33 and the applied voltage V, but is
independent of the thickness.
For movements in the direction normal to the
poled direction (for example, in the x direction),
we have V>0(+)
_ h _ Aa/a _ ( c \ Aa
Figure 4-46 Schematic illustration of the bending of a
V/c a) V (4-100) bimorph cantilever-piezoelectric actuator. (The bending
magnitude is highly exaggerated for clarity.)
Aa = \-\d3lV
<c,

It can be seen that in this case, a can be made

much larger than c if a » c. For a very small
adjustment in the position, we may use a small A»-f(frl*,v (4-101)
and more precise movement in the direction
In this case, AH is proportional to (a/cf. This
parallel to the poled direction. But for a larger
is why bending actuators are widely used for
movement, such as for relays, sideways move-
the applications requiring large movements,
ment in the direction normal to the poled direc-
because they provide the largest movement
tion may be preferable.
(deflections) for a given applied voltage. It
Elastic compliance can be greatly increased
should be noted that bending bimorphs are not
by using long, thin piezoelectric strips or plates
suitable for some applications because they
and mounting them as cantilevers. Bending
have high mechanical inertia and cannot
such a strip or plate causes one half to stretch
produce significant forces.
and the other half to compress, so there can be
no net electrical response. However, if there are
two such strips or plates with an intervening Piezoelectric Transformers
electrode and electrodes on outer surfaces to A transformer can be formed simply by a
form a bimorph, as shown in Figure 4-46, then piezoelectric plate with electrodes deposited on
the bending will generate a voltage between the half of its two flat surfaces and an electrode on
outer electrodes. By the converse piezoelectric the edge of the plate, as shown in Figure 4-
effect, a bimorph will bend when a voltage is 47(a). One half of the plate is poled in the direc-
applied between the outer electrodes. The tion perpendicular to the surfaces, and the other
half is poled in the direction parallel to the
bending is governed by d' = —— (see Equa-
surfaces. The operation requires the application
tion 4-76). The applied field will also produce
of an alternating voltage, like the conventional
uniform strains along the length of the bimorph
transformer, at a frequency that can excite a
so that it will be in the form of a circular arc. length-mode resonance. In the first half of the
Detailed analysis of the behavior of the plate, the input electrical energy through the
bimorph under static conditions has been done input terminal is converted into mechanical
by several investigators.138139 The reflection at energy, which results in oscillatory vibration.
the free end of the bimorph cantilever is given This mechanical energy is then converted back
by to electrical energy and picked up at the output
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 269

(a)
Stress
Distribution
. i
Y///X////&////zzm
T
Supporting
Point
Halfwave
X Full
Wave
Supporting
Points
(b)

Figure 4-47 (a) Schematic illustration of a piezoelectric transformer and (b) the half- and full-wave resonances.

terminal. It is important that the length of the Based on the equivalent circuit shown in
plate be equal to a number of half-wavelengths Figure 4-40, when the resonance occurs, the
of the acoustic wave to allow standing waves reactance due to Lm and Cm falls to zero and the
to form. In short, the mechanical oscillations in overall impedance becomes very small if Rm is
the first part are transmitted at resonance to the small. So the piezoelectric material can be used
second part, which reconverts the mechanical for wave filters to allow only a selected fre-
energy into electrical energy, resulting in the quency band close to the resonance frequency
generation of a higher voltage. The resonant to pass and to stop waves of other frequencies.
frequency is determined by the acoustic veloc- The efficiency of such wave filters depends on
ity v in the piezoelectric material. If the total the value of the electro-mechanical coupling
length of the transformer is 2€, then for the factor k.
transformer length equal to one full wave-
length, the full wave resonant frequency is
4.4 Pyroelectric Phenomena
f = y£ (4-102)
Of 20 noncentrosymmetric crystal classes, 10
For the transformer length equal to a half- contain a unique polar axis under the unstrained
wavelength, the half-wave resonant frequency is condition. This implies that such crystals are
already spontaneously polarized in a certain
temperature range. Thus, they exhibit both
piezoelectric and pyroelectric effects. For the
Figure 4-47(b) shows that half-wave and the latter, the crystal will develop free electric
full-wave resonance. Since the operation of the charges at the surfaces when uniformly heated
transformer is based mainly on the resonance because the polarization inside the crystal is
conditions, the location of the supporting points changed by the heat. It should be noted that if
in the installation of such transformers is very a crystal specimen is heated nonuniformly (for
important. In general, supporting points must example, with a temperature gradient along the
be located at places where the oscillatory vibra- specimen), then the temperature gradient will
tion is zero, as shown in Figure 4-47(b). also develop a mechanical stress. So in this
270 Dielectric Phenomena in Solids

case, the specimen may have mixed piezoelec- circuited. However, a change in temperature
tric and pyroelectric effects. However, under a will cause a change in the resultant dipole
uniform heating condition (i.e., a hydrostatic moment of all dipoles and hence a change in
condition), the resultant piezoelectric effect the spontaneous polarization. The surface
vanishes. charges will redistribute themselves to com-
Since pyroelectricity results mainly from the pensate for the new internal polarization, thus
temperature dependence of the spontaneous producing a flow of charges when the specimen
polarization of the polar materials, it occurs in is short-circuited.
ferroelectric materials whether they are single- Since
domain crystals or poled ceramics. A change in
temperature will cause a corresponding change D = e0F+Pi0XA
(4-105)
in polarization and hence a change in the com- = £0F + (induced + *s )
pensating charges on the metallic electrodes where Pinduced and Ps are, respectively, the field-
deposited on its surfaces. This change will induced polarization and the spontaneous
produce a current in an external circuit. If the polarization. Pinduced is much smaller than Ps and
pyroelectric specimen is not connected to an less temperature dependent. So px can be
external circuit and kept perfectly insulated expressed as
from its surroundings, the surface charges on
the electrodes released due to a change in tem- x dPs
P = (4-106)
perature will flow inside the material through dT X,F
the intrinsic electrical conductivity of the mate-
The pyroelectric coefficient under the condition
rial a, but it may take e/a time for the released
of constant mechanical stress implies that the
charges to be neutralized.
crystal is not clamped, that is, it is free to
expand or contract thermally. Based on the rela-
4.4.1 Phenomenological Approach to tions between F, D, X, and T shown in Figure
Pyroelectric Effects 4-38, we can write
The interactions between electrical and thermal dh = yFJdX + dTdF+aFdT
variables are the direct concern of this section.
Elasticity A Thermal
A change in a crystal's temperature can produce | expansion (4-107)
a number of electrical effects, which are Converse
referred to as pyroelectricity. These effects Piezoelectricity
depend upon the thermal, mechanical, and elec-
trical constraints on the crystal. The relation dD = dT(dX) + eXTdF+pxdT
between D and T defines the pyroelectric coef-
ficient (see Figure 4-38, the dashed line joining Direct | Pyroelectricity
circle D and circle T) or Piezoelectricity
Permittivity
dD = pxdT (4-108)
dD (4-104)
P = dT where v, d, £, a, and p are, respectively, elastic
compliance, piezoelectric coefficient, permit-
The unit of p is coulomb m_2K_1. tivity, thermal expansion coefficient, and pyro-
Pyroelectricity occurs only when the mate- electric coefficient. These equations should be
rial exhibits spontaneous polarization. If the written in tensor form, like those for piezoelec-
temperature of the specimen remains constant tric effects, with subscripts to denote the direc-
for a sufficient length of time, the charges accu- tion and rotation of the vectors. For simplicity,
mulated on the surfaces will compensate for the we will ignore the complicated mathematical
internal polarization, so there will be no charge expressions and emphasize the physical con-
flow between the surfaces when they are short- cepts of the equations.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 271

Under the constraint condition of constant It should be noted that, theoretically, all fer-
electrical field (i.e., dF = 0), from Equations roelectric materials are pyroelectric, but for
4-107 and 4-108, we obtain applications only those with the second-order
transition at the Curie point are usable. From
dh the Ps - T characteristics shown in Figure 4-9
dX = F,T
y
•dT
(4-109) it can be seen that the pyroelectric coefficient
dD is high just below the Curie point for the case
, - $ « with a second-order transition because dPJdT
is large. For example, TGS exhibits a second-
Under the condition of constant strain, dh = 0. order transition and has a high pyroelectric
This implies that the crystal is rigidly clamped, coefficient just below the Curie point. The
so expansion or contraction is not possible. materials exhibiting a steep first-order transi-
Thus, from Equation 4-109, we obtain tion at the Curie point are not suitable for
T F pyroelectric applications because they exhibit
dD d a
a thermal hysteresis and it is difficult in most
dT h y applications to maintain a sufficiently constant
T F (4-110)
= ph +d d
„F,T
temperature environment (see also Ferroelec-
tric Phenomena in Chapter 4).
The electrocaloric effect is the converse of
Equation 4-110 indicates that the total pyro- the pyroelectric effect. The change of entropy
electric coefficient at constant stress px consists can be expressed as
of two components:
• The pyroelectric coefficient at constant strain dS = aFdX + pxdF+—^— dT (4-111)
TyX,F
ph due to the primary pyroelectric effect
Under the condition of constant entropy, which
• The secondary pyroelectric effect, which
is an adiabatic process, dS = 0 and constant
may occur when the crystal is free to react
stress, dX = 0, then the electrocaloric coefficient
As shown in Figure 4-1, all pyroelectric mate- cf can be written as
rials are piezoelectric, but not all piezoelectric
materials are pyroelectric, because the latter d£ -TyXFpx (4-112)
requires spontaneous polarization for its occur- dF
rence. For example, a-quartz possesses three The unit of q* is KmV"1.
twofold axes, so a stress applied along any of
these axes would cause a piezoelectric effect.
But with heating, the material will expand 4.4.2 Pyroelectric Parameters and
equally along each of these axes. The angle Their Measurements
between two adjacent axes is 120 degrees, so The major parameter is the pyroelectric coeffi-
the pyroelectric effects tend to cancel each cient px since the electrocaloric coefficient is
other. Hence, a-quartz is not pyroelectric. Sim- usually very small and scarcely used for the
ilarly, zinc blend ZnS (zinc sulfide) possesses evaluation of pyroelectric effects. Several
four threefold axes and hence is piezoelectric methods can be used for measuring px. Here,
and not pyroelectric. On the other hand, the we describe only two simple ones.
hexagonal zinc sulfide or cadmium sulfide with One is the static method. In this method, the
a wurtzite-type structure possesses one sixfold pyroelectric specimen is heated in a constant
axis, so it is piezoelectric and also pyroelectric. temperature chamber in which the temperature
In general, pyroelectric materials also exhibit can be varied. Figure 4-48 shows the arrange-
the piezoelectric effect under hydrostatic pres- ment for measuring the charges developed on
sure, and pyroelectric materials have secondary the pyroelectric specimen. These charges are
pyroelectric and piezoelectric effects. then compensated by static charges induced on
272 Dielectric Phenomena in Solids

Pyroelectric Specimen electric materials: one is ferroelectric, and the

other is not ferroelectric but possesses sponta-
neous polarization, which may not be reversed
when the electric field is reversed. For poled
Electrometer
specimens, the remanent polarization is tem-
perature dependent and falls to zero as the
Curie temperature is approached. This Ps - T
relation below Tc is the cause of the pyroelec-
Adjustable
tric effect. However, ferroelectric materials
Temperature Chamber
with the second-order transition at Tc generally
Figure 4-48 Schematic diagram illustrating the static
have higher pyroelectric coefficients than non-
method for the measurement of pyroelectric coefficient p. ferroelectric ones, particularly at temperatures
For the dynamic method, capacitor C is replaced by a resis- just below Tc.
tor/?. Table 4-5 lists the important parameters of
some commonly used pyroelectric materials.
capacitor C, which can be measured simply by The selection of pyroelectric materials for a
an electrometer. Based on Equation 4-104, px special application should not be based solely
can be evaluated by on the figure of merit (see Equation 4-138);
other properties, such as chemical stability,
Qi-Qi
(4-113) mechanical strength, power handling ability,
PA =
A(T2-TX) and fabrication processes, should also be con-
where Q2 and Qi a r e t n e charges measured at sidered. For example, triglycine sulfate (TGS)
temperatures T2 and Tu respectively, and A is has a high figure of merit and is therefore
the surface area of the specimen. a commonly used material for low-power
The other method is the dynamic method. In detection applications, but, it has its inherent
this method, we can use the same arrangement shortcomings. It cannot be subjected to high
shown in Figure 4-48 but with capacitor C heat—even sunlight may make it hot enough to
replaced by a resistor R. In this case, the spec- depole. It is also mechanically weak and hygro-
imen is heated at a uniform rate of temperature scopic. However, the incorporation of alanine
rise, and it generates a current flowing through dopants into TGS greatly reduces its dielectric
R measured also by an electrometer. We can loss and dielectric constant, improving its
write performance as a radiation detector.140141
Anyhow, this material is not suitable for
dQ AdD dDdT A xdT devices to be used in an environment involving
••-=. = A— = A = Apx —
dt dt dT dt dt high vacuum or high humidity. Other materials
Thus, with better mechanical strength and chemical
stability than TGS, such as LiTa0 3 and SBN,
(4-114) have already been used widely for pyroelectric
A(dT/dt) devices. They are also insensitive to humidity.
where i is the current and dT/dt is the rate of Lead zirconate (PZ)-based ceramics, such
as PZT and PLZT, have a high value of
temperature rise.
pyroelectric coefficient and also of dielec-
tric constant. It has been found that PZ
4.4.3 Pyroelectric and Thermally doped with Fe, Nb, Ti, and U, such
Sensitive Materials as Pbi.o2(Zr0.58Feo.2oNb0.2oTio.o2)o.994 U0.006O3
Similar to piezoelectric materials, the most (PZFNTU), gives a higher pyroelectric coeffi-
important pyroelectric materials are ceramics, cient and a lower dielectric constant.142
synthetic polymers, and ceramic-polymer com- As mentioned earlier, pyroelectric polymers
posites. In general, there are two types of pyro- are now widely used for many applications
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 273

Table 4-5 Selected properties of some pyroelectric materials: p = pyroelectric coefficient at constant stress and
electric field, and cp = heat capacity.

Temperature
Material (Abbreviation) Chemical Formula p (C cmr2K-1) cp (J cm^KT1) Range (K)

Triglycine sulfate (TGS) (NH2CH2COOH)3 •H 2 S0 4 3.0 x 10~8 1.70 273-321

Deuterated Triglycine Sulfate (DTGS) (ND2CD2COOD3)«D2S04 3.0 x 10-8 2.40 243-334

Lithium Tantalate (LT) LiTa0 3 1.9 x 10'8 3.19 273-891

Strontium Barium Niobate (SBN) Sr0.6Bao.4Nb206 8.5 x 10"8 2.34 248-303

8
Lead Zirconate Titanate (PZT) Pb(Zr0.52Tio.48)03 5.5 x 10" 2.60 298-523

Lanthanum Modified PZT (PLZT) Composition 7.6 x 10~8 2.57 —

Ratio: La/Zr/Ti =
6/80/20

Barium Titanate BaTi0 3 3.3 x 10-8 — 293-320

Polyvinyl Chloride (PVC) see Table 5-2 0.4 x 10~9 2.40 —

9
Polyvinyl Fluoride (PVF) see Table 5-2 1.8 x 10~ 2.40 —
Polyvinylidene Fluoride (PVDF or Kynar) see Table 5-2 4.0 x 10"9 2.40

because of their ease of preparation in any controllers (e.g., thermostats, thermisters),

shape and size. For example, polyvinyl fluoride current-limiting devices, etc.
(PVF) and polyvinylidene fluoride (PVDF) can
be made extremely thin (<2/jm). Although they
have a low figure of merit, a low pyroelectric NTC Materials
coefficient, and also a relatively high dielectric The most commonly used NTC materials are
loss, they have a low heat conductivity and a based on solid solutions of metallic oxides with
low permittivity, facilitating both thermal and a spinal structure, such as Fe 3 0 4 - ZnCr 2 0 4 and
electrical coupling between neighboring ele- Fe 3 0 4 - MgCr 2 0 4 , and also those based on
ments. Their mechanical strength, thermal sta- Mn304 with partial replacement of Mn with M,
bility, and chemical stability are also good. Co, or Cu.96 On the basis of electrical conduc-
Apart from pyroelectric materials, there are tion in semiconductors due to transport of
some materials whose resistivity is very sensi- charge carriers in the bands or hopping between
tive to temperature based on other mechanisms localized states in the band gap, the tempera-
and not pyroelectric effects. These materials are ture dependence of the resistivity follows the
usually called negative temperature coefficient relation
(NTC) materials, whose resistivity increases
with decreasing temperature, or positive tem-
perature coefficient (PTC) materials, whose p^ c (D = A e x p ^ J (4-115)
resistivity increases with increasing tempera-
ture. Obviously, there are numerous applica-
tions of such materials with a high temperature where k is the Boltzmann constant and A and B
coefficient of resistivity (TCR) in temperature are constant. A is related to the resistivity at
indicators (e.g., thermometers), temperature T —> °°, and B is the activation energy for the
274 Dielectric Phenomena in Solids

carrier movement. The negative TCR is material increases with increasing temperature
and reaches a peak at T> Tc, with the magnitude
1 dpurc B
&NTC — (4-116) of several orders higher than that at Tc, as shown
PNTC dT 'kT2
in Figure 4-49(a). This phenomenon is not
observed in material with a single crystal struc-
PTC Materials ture. It occurs only in polycrystalline ceramics.
The behavior of the DC electrical resistivity of Heywang144 has proposed that this phenomenon
ferroelectric ceramics is anomalous in the neigh- is caused by the Schottky barrier at the grain
borhood of the Curie temperature Tc. A typical boundaries in polycrystalline structures.
example is BaTi0 3 ceramic doped with 0.05 x The energy band diagram near the grain
10"2mol of Sn3+ or Nb3+. When the trivalent boundary is shown in Figure 4-49(b). In the
donor Sn3+ or Nb3+ substitutes for Ba2+, the extra paraelectric phase, the band bending near the
positive charge in the material is compensated boundary is determined by the number of inter-
by an electron in the conduction band. The face states Nint at the boundary. Similar to the
material behaves like a semiconductor, but Schottky barrier at the metal-semiconductor
electron mobility is also changed at the Curie junction with the interface states at the
temperature.143 As a result, the resistivity of the metal-semiconductor interface, there is a

100 150 200 250 300

25 rmin Temperature (°C)

(a)

Conduction Band

© e © © ©® : \ @
© © © © ©
Acceptor States
*d Fermi Level

Grain Boundary

(b)

Figure 4-49 (a) The resistivity p as a function of temperature for BaTi03 based PTC material and (b) the energy band
diagram near a grain boundary of a polycrystalline material.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 275

depletion region xd and a barrier height 0. In the following sections, we discuss some
Assuming that all donors are ionized, it can be typical cases as examples to illustrate the basic
shown that the barrier height is given by principles of pyroelectric devices.
q2NDx2d
(4-117)
2ere0 Pyroelectric Radiation Detectors
where ND is the concentration of donors. The Radiation detectors are mainly used for the
number of interface states at the boundary that detection of infrared radiation. A poled pyro-
accept electrons from the donors, acting as electric element is typically a thin plate of
acceptor-like traps, must be Nint = Nj>xd. The rectangular shape with metallic electrodes
barrier height depends on the size of grains deposited on both surfaces. It is important that
because it affects the energy levels of the inter- the element be very thin, so that it heats up
face states.145 The resistivity of the ceramic quickly and uniformly, and that at least one of
material is determined by the magnitude of 0. the electrodes is a good absorber of radiation,
Thus, we can write so that it can convert radiation energy to heat.
Figure 4-50(a) shows schematically the basic
pPTCooQXp\- — (4-118) arrangement for the detection of infrared radi-
ation. Under a short-circuit condition, there is
For temperatures T > Tc, the dielectric constant a current flow due to the pyroelectricity, which
decreases with temperature following the is given by
Curie-Weiss relation (see Equation 4-1). So,
from Equations 4-1, 4-117, and 4-118, we • pA (4-121)
dt
obtain
where p and A are, respectively, the pyroelec-
p p r c =A'exp (4-119) tric coefficient and the area of the element. In
the following discussion, the superscript x ofpx
where A' and B' are constants related to doping is not shown for simplicity. The pyroelectric
concentration and structure of the material. It element behaves like a current generator with
can be seen that for T > Tc the resistivity an internal resistance Rt and an internal capac-
increases with increasing temperature. The itance Q. The equivalent circuit to Figure 4-
TCR for this case is thus 50(a) is shown in Figure 4-50(b). The detector
also includes an amplifier represented by a
1 dppTc &_Tc_
CCpjc ' 2 (4-120) shunt resistor Rd, a shunt capacitor Cd, and a
PPTC dT k T
voltmeter V.
It is assumed that all incident infrared radia-
4.4.4 Applications of Pyroelectrics tion energy is absorbed by the element in time
Pyroelectric materials respond to changes in dt and rapidly distributed throughout the whole
temperature either by continuous heating or by volume of the element, so thermal diffusion can
the absorption of sinusoidally modulated radi- be ignored. The total power W incident to the
ation. Pyroelectric materials may be used for element may be written as
thermal detectors and calorimeters, but they do
W = W0 + Wiexp(jo>t) (4-122)
not respond to a temporally steady temperature
or to radiation intensity. To obtain a response where W0 in the input power from the sur-
from a stationary temperature or radiation, it is rounding environment corresponding to tem-
necessary to make the heat or radiation period- perature T0, Wt&xp (jcot) is the incident infrared
ically interrupted by a rotating chopper placed power, and co is its frequency. When the
between the heat or radiation source and the element temperature is increased from T0 to T,
pyroelectric element. Obviously, there are part of the absorbed power will be lost to the
many applications of pyroelectric phenomena. surroundings by re-radiation, conduction, or
276 Dielectric Phenomena in Solids

Incident Radiation

Measuring
System
Pyroelectrlc
Axis
(a)

Pyroelectric Measuring
Element System
(b)

Figure 4-50 (a) Experimental arrangement for infrared radiation detection and (b) the equivalent circuit of the radiation
detector shown in (a).

convection at a rate of G per unit temperature Using the boundary condition (when t = 0, T =
rise in the element due to incident radiation. T0 and when t- tu T= Tx), the solution of Equa-
Thus, we can write the energy balance equation tion 4-124 yields
as
T-T0=(Tl-T0)exp(-—) (4-125)
7iWiexp(jcot)-G(T-T0) = H— (4-123)
at where rT is the thermal relaxation time, which is
where rj is the fraction of the incident power
that heats the element; H is the heat capacity of rT=^ (4-126)
the element, which is equal to pcAL in which p KJ

is the density; c is the specific heat; and A and Assuming that the temperature of the element
L are, respectively, the area and the thickness varies sinusoidally following the incident radi-
of the element. If the radiation is switched on ation frequency CD, then the solution of Equa-
at t = 0 and switched off at t = tu then when tion 4-123 gives the continuous response of the
t = tu Wt = 0, Equation 4-123 becomes pyroelectric element98146 as follows:

H— + G(T-To) = 0 (4-124) T-T0= W(G2 + co2H2 y1'1 exp [j(o)t + 0)]

dt (4-127)
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 277

where 6 is the phase angle between the radia- From Equation 4-127, T- T0 is independent of
tion and the temperature oscillation, which is frequency for frequencies co « ?>* and
given by decreases with increasing frequency for
0 = tm-l(o)H/G) (4-128) (0» X~j.
From Equations 4-132 and 4-135, rv is maximal
Based on the equivalent circuit shown in Figure
at the frequency co = (TTTE)~l/2. Both rt and
4-50(b), the total resistance RT and the capaci-
rv increase with increasing frequency for w <
tance CT are
XjX and decrease with increasing frequency
RiRi for co> fj.
RT (4-129)
Ri + Rd
The maximum value of rv occurs at co =
m
(4-130) (TTTE) and is given by
So the element-generated current ip due to
the variation of temperature can be written W> = - J ^ - (4-136)
G(T £ + T r )
as For high frequencies, c&x\ » 1, ofrj » 1 and
dT „ dV V Ct > Cd, so Equation. 4-135 can be simplified
iPp = ypA — = CT — H
dt dt RT to
= r\WiPA\CO
— l[l + coWY11 exp [j(cot + 0)] . r\pA
coHQ
(4-131) (4-137)
_JtP_
The useful criteria for pyroelectric detectors Acocp£
are the current responsivity rt and the voltage
responsivity rv, which are defined as follows: where cp = pc. Since the sensitivity of the radi-
ation detector increases with increasing rv, the
.P^VQn -(l +. ™2^2\-i/2
fl)zTf)- (4-132) p
Wt quantity — : in Equation 4-137 may be used
Cpt,
From Equations 4-129 and 4-130, we can easily as a figure of merit for the pyroelectric radia-
find the total admittance YT of the circuit. The tion detector Fv, which is
voltage due to the change in temperature across
Rd is given by Fv=- (4-138)
c„e
V= This is mainly a function of the material prop-
erties. In general, we would like the material to
= 7iwiPART(—Vi+coh2Tyll2{\+co2T2Ey112 have a high pyroelectric coefficient and a low
dielectric constant.
(4-133)
However, all signal detectors encounter the
where TE is the electrical relaxation time, which effect of noise, which is always present in any
is detector circuit. So, the sensitivity of any detec-
TE=RTCT (4-134) tor is determined by the level of noise in the
amplifier output signal. Therefore, the signal-to-
Thus, the voltage responsivity can be expressed noise ratio must be made as large as possible.
as For pyroelectric detectors, the principal sources
V rjpARTco of noise are Johnson noise, thermal fluctuations,
rv = 2 l/2 2 2 /2 and amplifier noise. The noise level can also be
W, G(l + coh T) (l + co T E?
(4-135) described by the so-called noise equivalent
power (NEP), which is defined as
Assuming that the circuit parameters are inde-
pendent of frequency, we can see the following AVN
NEP = (4-139)
features: rv(Aff/2
278 Dielectric Phenomena in Solids

where AVN is a signal voltage equal to the noise nection of this compensating element and the
voltage, produced by a necessary amount of active element is back to back, so the response
power input to the detector, and Af is the band- of the compensating element to the ambient
width of the amplifier. It is often convenient to temperature fluctuation will cancel that of the
describe the minimum detective power of a active element. In general, such compensation
detector by the so-called detectivity (D) which is essential for many applications. The basic
is defined as principle is the same as that described in the
previous section. The output voltage of the
D =-
1 (4-140) system is to switch on the alarm system.
NEP

Pyroelectric Thermometry
Pyroelectric Burglar Alarm Systems In contrast to radiation detectors, pyroelectric
A blackbody at 300 K would radiate infrared thermometers receive energy through conduc-
rays of about 9.8 jim in wavelength. Human or tion or convection. As a result, noise due to
animal bodies whose temperature is around thermal fluctuations is low. The principle is
310 K are expected to generate infrared radia- simple.148,149 Based on the equivalent circuit
tion of about lOjUm in wavelength. Thus, a shown in Figure 4-50, Equation 4-131 can be
moving intruder produces variable infrared rewritten as
radiation which, when reaching a pyroelectric
AdT ^ dV V
detector, will switch on the alarm system. pA— = CT — + —
y
(4-141)
Figure 4-51 shows the arrangement of the pyro- dt dt RT
electric elements and the alarm system. Since This equation indicates that pyroelectric ther-
pyroelectric materials are also piezoelectric, mometers can be used to measure the rate of
they would produce electric charges due to temperature changes or the steps of steady
external mechanical stresses caused by the temperature changes.
expansion or contraction when the ambient tem-
perature of the surroundings rises or falls. These
Measuring the Rate of Temperature Changes
stress-induced charges may interfere with the
In this case, the electrical relaxation time
response of the active pyroelectric element to
must be small (i.e., TE = RTCT « 1), so the first
radiation. To prevent this interference, we need
term on the right side of Equation 4-141 can be
a dummy compensating element similar to the
neglected. Thus, we have
active element, but which does not respond to
infrared radiation because the electrode facing dT (4-142)
the radiation is a reflecting electrode. The con- dt pART

Incident Radiation FET

T T T T

Compensating Detecting
Element Element

Figure 4-51 Schematic diagram of a pyroelectric burglar system alarm consisting of an active detecting element, a com-
pensating (dummy) element, and an alarm unit.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 279

The thermometer directly measures the rate of

temperature changes by measuring the voltage
across the pyroelectric element.
Win =
C2P(\ + (D-2TT2)(£2+£'2)(OL ©
(4-147)

y/in becomes a maximum for onT = 1. Using the

Measuring Steady Temperature Changes average temperature change given by Equation
In this case, the electrical relaxation time 4-127, AT=T-T0 = Wt/cpA, we have
must be large (i.e., rE = RTCT » 1), so the
p2AT
second term on the right side of Equation 4-141 Vin = (4-148)
2(£2+£-2)cP
can be neglected. Thus, we have
In practice, £ » e1. So, we may consider p2/cp£
pA_dT_ dV_
(4-143) as the figure of merit for energy conversion of
CT dt ' dt
pyroelectric materials. For typical values of
Integrating Equation 4-143 with the initial con- rT = 1 s and Wt = lOOmWcm"2 and the temper-
dition, V(0) = V0 and T(0) = T0, we obtain ature change AT=T-T0 = 50°C, the internal
conversion efficiency is \\fin — 10~3. It has been
T-T0=(V-V0) (4-144) found that p2/cp£ does not vary much for most
pA pyroelectric materials.98150 So it is unlikely that
The thermometer measures the steady temper- the energy conversion efficiency would greatly
ature changes in steps by measuring V at T and exceed 10"3 under practical conditions, based
V0 at T0. on pyroelectric effects.
For more about the limitations of this method
and the factors involved in temperature meas-
urements, see references 146-149. References
1. J. Valasek, Phys. Rev., 77, 475 (1921).
2. J. Valasek, Phys. Rev., 79, 478 (1922).
Pyroelectric Energy Conversion 3. J. Valasek, Phys. Rev., 24., 560 (1924).
Thermal energy input to a pyroelectric element 4. W.G. Cady, Piezoelectricity, (Dover, New
York, 1964).
will cause changes in the spontaneous polariza- 5. J. Curie and P. Curie, Compt. Rend., 97,
tion and hence generate electric charges on the 294 (1880) and 95, 1137 (1881).
electrodes. If a load is connected between the 6. Bunget and M. Popesco, Physics of Solid
electrodes, a current will flow through the load, Dielectrics, (Elsevier, Amsterdam 1984),
and electrical energy will be consumed there. p. 383.
7. M. Born and K. Huang, Dynamical Theory
The energy conversion efficiency is defined as of Crystal Lattices, (Clarendon, Oxford,
1954).
Electrical energy output
¥ = 8. J.F. Nye, Physical Properties of Crystals,
Thermal energy input (Oxford University Press, Oxford, 1960).
= iEV 9. C.S. Smith, "Macroscopic Symmetry and
(4-145) Properties of Crystals," in Solid State
Physics, Edited by F. Seitz and D. Turn-
= Wtrtrv bull, Vol. 6 (Academic Press, New York,
From Equations 4-132 and 4-135, y/ can be 1958), p. 175.
10. K.R. Brain, Proc. Phys. S o c , 36, 81
written as (1924).
WiP2RT{onT)2 11. E.L. Kern and S.M. Skinner, J. Appl.
V- (4-146) Polym. Sci., 6, 404 (1962).
C pL (l + 0)2T2T)(l + G)2T2Ey/2
2 2
12. S.N. Levine, J. Appl. Polym. Sci., 9, 3351
When Rt « Rd, Q » Cd, Q = eAIL and a = (1965).
13. G.E. Hauver, J. Appl. Phys., 36, 2113
LIARi = coe\ we obtain the internal energy con- (1965).
version efficiency of the pyroelectric element 14. E. Fukada, M. Date, and N. Hirai, Nature,
as 277, 1079 (1966).
280 Dielectric Phenomena in Solids

15. H. Kawai, Jpn. J. Appl. Phys., 8, 975 44. B. Jaffe, W.R. Cook, and H. Jaffe, Piezo-
(1969). electric Ceramics, (Academic Press, New
16. C.B. Sawyer and C.H. Tower, Phys., Rev., York, 1971).
35, 269 (1930). 45. H.M. Barnett, J. Appl. Phys., 33, 1606
17. W.J. Merz, Phys. Rev., 76, 1221 (1949) (1962).
18. W.J. Merz, Phys. Rev., 91, 513 (1953). 46. C. Michel, J.M. Moreau, G.D. Achenbach,
19. E.J. Huibregtse and D.R. Young, Phys. R. Gerson, and W.J. James, Solid State
Rev., 103, 1705 (1956). Commun., 7, 865 (1969).
20. A.F. Davenshire, Phil. Mag., 42, 1065 47. E. Sawaguchi, J. Phys. Soc. Japan, 8, 615
(1951). (1953).
21. H.H. Weider, Phys. Rev., 99, 1161 (1955), 48. S. Roberts, J. Am. Ceram. Soc, 33, 63
and J. Appl. Phys., 26, 1479 (1955). (1956), and Phys. Rev., 83, 1078 (1951).
22. W.J. Merz, Phys. Rev., 95, 690 (1954). 49. G. Shirane, E. Sawaguchi, and Y. Takagi,
23. R. Landauer, D.R. Young, and M.E. Phys. Rev., 84, 476(1951).
Drougard, J. Appl. Phys., 27, 752 (1956). 50. E. Sawaguchi and K. Kittaka, J. Phys. Soc.
24. H.H. Weider, J. Appl. Phys., 28, 367 Japan, 7, 336 (1952).
(1957). 51. R. Redin, G. Marks, and C. Antoniak, J.
25. D.S. Campbell, J. Electronics and Control, Appl. Phys., 34, 600 (1963).
3, 330 (1957). 52. G. Jonker, J. Am. Ceram. Soc, 55, 57
26. W. J. Merz, J. Appl. Phys., 27, 938 (1956). (1972).
27. V. Janovec, Czechoslov. J. Phys., 8, 3 53. H. Jaffe and D. Berlincourt, Proc. IEEE.,
(1956). 53, 1372 (1965).
28. W.J. Merz and J.R. Anderson, Bell Lab. 54. K. Hardti, Ferroelectrics, 12, 9 (1976).
Record, 33, 335 (1955). 55. L. Benguigui, Solid State Commun., 11,
29. J.R. Anderson, G.W. Brady, W.J. Merz, and 825 (1972).
J.P. Rameika, J. Appl. Phys., 26, 1387 56. K. Okazaki, Ferroelectrics, 35, 173 (1981).
(1955). 57. P. Thatcher, Appl. Opt., 16, 3210 (1977).
30. E. Fatuzzo and W.J. Merz, Ferroelectricity, 58. R.C. Buchanan (Ed.), Ceramic Materials
(North-Holland, Amsterdam, 1967). for Electronics, (Marcel Dekker, New
31. A.G. Chynoweth, Phys. Rev., 110, 1316 York, 1986).
(1958). 59. L.M. Levinson, Electronic Ceramics,
32. R.C. Miller, Phys. Rev., Ill, 736 (1958). (Marcel Dekker, New York, 1988).
33. V.M. Rudyak, A.Y. Kudzin, and T.V. 60. G.H. Hacetling and C.E. Land, J. Am.
Panchenko, Sov. Phys.—Solid State, 14, Ceram. Soc, 54, 1 (1971).
2112(1973). 61. R.H. Dungan, H.M. Barnett, and A.H.
34. G.F. Bacon and R.S. Pease, Proc. Royal Stark, J. Am. Ceram. Soc, 45, 226 (1962).
Soc. London, A220, 397 (1953) and A230, 62. Y.L. Wang, W.Z. Yuan, G.H. He, S.W. Lin,
359 (1955). Y.H. Ling, C.F. Qu, and B.G. Wang, Fer-
35. M. Tokunaga, Ferroelec tries, 1, 195 roelectrics, 49, 169 (1983).
(1970). 63. Y.L. Wang and K.C. Kao, in Conference
36. G. Busch, J. Helv. Phys. Acta, 11, 269 Record of the 2000 IEEE International
(1938). Symposium on Electrical Insulation
37. A. Von Arx and W. Bantle, J. Helv. Phys. (IEEE-DEIS, New York, 2000) pp. 58-
Acta, 17, 298 (1944). 61.
38. W. Kanzig, "Ferroelectrics and Antiferro- 64. WM. Prest, Jr. and D.J. Luca, J. Appl.
electrics," in Solid State Physics edited by Phys., 46, 4136 (1975).
F. Seitz and D. Turnbull, Vol. 4 (Academic 65. R. Hasegawa, Y Takahashi, Y. Chatani,
Press, New York 1957) pp. 1-197. and H. Tadokoro, Polym. J., 3, 600 (1972).
39. F. Jone and G. Shirane, Ferroelectric Crys- 66. M.A. Bachmann and J.B. Lando, Macro-
tals, (MacMillan, New York, 1962). molecules, 74,40(1981).
40. H.M. Barkla and D.M. Finlayson, Phil. 67. A.J. Lovinger, Science, 220, 1115 (1983).
Mag., 44, 109 (1953). 68. J. Scheinbeim, C. Nakafukus, B.A.
41. VL. Hablutzel, J. Helv. Phys. Acta, 12,489 Newman, and K.D. Pae, J. Appl. Phys., 50,
(1939). 4399 (1979).
42. A.N. Holden and W.P. Mason, Phys. Rev., 69. A.J. Lovinger, in Developments in Crys-
57, 54 (1940). talline Physics, Vol. 1, edited by D.C.
43. S. Hoshino, T. Mitsui, F. Jona, and R. Bassett (Applied Sciences, London, 1982)
Repinsky, Phys. Rev., 107, 1255 (1957). Chapter 5.
 Ferroelectrics, Piezoelectrics, and Pyroelectrics 281

70. T. Takahashi, M. Date, and E. Fukada, 94. K. Lichtenecker, Phys. Z. 10, 1005 (1909)
Appl. Phys. Lett., 37, 791 (1980). and 27, 115(1926).
71. A.J. Lovinger, G.T. Davis, T. Furukawa, 95. K.W. Plessner and R. West "High Permit-
and M.G. Broadhurst, Macromolecules, tivity Ceramics for Capacitors," in
75, 323 and 329 (1982). Progress in Dielectrics, Vol. 2, edited by
72. J.G. Bergman, Jr., J.H. McFee, and G.R. J.B. Birks and J.H. Schulman (Heywood,
Crane, Appl, Phys. Lett., 18, 203 (1971). London, 1960) pp. 165-192.
73. A.F. Devonshire, Quarterly Supplement of 96. A.J. Moulson and J.M. Herbert Electroce-
Phil. Mag., Advances in Physics, 3, 85 ramics: Materials, Properties and Appli-
(1954). cations, (Chapman and Hall, London,
74. E.J. Huibregtse and D.R. Young, Phys. 1997).
Rev., 103, 1705 (1956). 97. J.M. Herbert, Ceramic Dielectrics and
75. D. Meyerhofer, Technical Report 121, Capacitors, (Gordon and Breach, London,
Laboratory for Insulation Research, Mass- 1985).
achusetts Institute of Technology, 1957; 98. M.E. Lines and A.M. Glass, Principles
also Phys. Rev., 112, 413 (1958). and Applications of Ferroelectrics and
76. C. Kittel, Phys. Rev., 82, 729 (1951). Related Materials, (Clarendon, Oxford,
77. L.E. Cross and B.J. Nicholson, Phil. Mag., 1977).
46, 453 (1955). 99. S. Nomura and S. Sawada, J. Phys. Soc.
78. P.W. Forsbergh, "Piezoelectricity, Elec- Japan 6, 36 (1951); also ibid., 10, 108
trostriction and Ferroelectricity" in (1955).
Handbuch der Physik, Vol. 17, (Springer- 100. A. Glanc, Z. Malek, J. Mastner, M. Novak,
Verlag, Berlin, 1956) pp. 264-392. and J. Strajblova, J. Appl. Phys. 35, 1870
79. G.J. Goldsmith and J.G. White, J. Chem. and 1875 (1964).
Phys., 31, 1175(1959). 101. J. Fousek, J. Appl. Phys. 36, 588 (1968).
80. H.L. Stadler and P.J. Zachmanidis, J. Appl. 102. R. Koontz, G. Blookhina, S. Gold and
Phys., 34, 3255 (1963). A. Krasnykh, Annual Report of 1998
81. E.A. Little, Phys. Rev., 98, 978 (1955). Conference On Electrical Insulation and
82. J.C. Burfoot, Ferroelectrics: An Introduc- Dielectric Phenomena (IEEE Publica-tion
tion to the Physical Principles, (Van Nos- 98CH36257-1, New York) pp. 23-26.
trand, Princeton, 1967). 103. F.W. Neilson, Bull. Am. Phys. Soc, 2, 302
83. J.A. Hooton and W.J. Merz, Phys. Rev., 98, (1957).
409 (1955). 104. PC. Lysne and C M . Percival, J. Appl.
84. R.C. Miller and A. Savage, Phys. Rev., Phys., 46, 1519 (1975).
112, 755 (1958); Phys. Rev. Lett., 2, 294 105. W. Mock and [Link], J. Appl. Phys., 49,
(1959); also J. Appl. Phys. 30, 808 (1959). 5846(1978).
85. R.C. Miller and G. Weinreich, Phys. Rev., 106. J.R. Anderson, Electrical Engineering, 71,
117, 1460 (1960). 916 (1952).
86. M. Hayashi, J. Phys. Soc. Japan, 33, 617 107. J.F. Scott and C.A. Araujo, Science, 246,
and 739 (1972). 1400 (1989).
87. M. McQuarrie, Bull. Am. Ceram. Soc. 34, 108. J.F. Scott, C.A. Araujo, H.B. Meadows,
170 (1955). L.D. McMillan, and A. Schawabkeh, J.
88. G. Shirane and K. Sukuzi, J. Phys. Soc. Appl. Phys., 66, 1444 (1989).
Japan 7, 333 (1952). 109. J.F. Scott, C.A. Araujo, and L.D, McMil-
89. A. Glanc, V. Dvorak, V Janovec, E. lan, Ferroelectrics, 116, 107 (1991).
Rechziggel and V. Janouser, Phys. Lett. 7, 110. RK. Larson, R. Cuppens, and C.A.C.N.
106 (1966). Spielings, Ferroelectrics, 128, 265 (1992).
90. A.E. Feuersanger, in Thin Film Dielectrics, 111. L.K. Anderson, Ferroelectrics, 7, 55
edited by F. Vratny (Electrochemical (1974); also 3, 69 (1972).
Society, New York, 1969), p. 209. 112. S.Y Wu, W.J. Takel, and M.H. Francombe,
91. Y Y Tomashpolski, M.A. Sevostianov, Appl. Phys. Lett. 22, 26 (1973).
M.V. Pentagova, L.A. Sorokina, and Y.N. 113. S.A. Keneman, A. Miller, and G.W. Taylor,
Venevstsev, Ferroelectrics 7, 257 (1974). Ferroelectrics, 3, 131 (1972).
92. J.R. Slack and J.C. Burfoot, J. Phys. C. 114. P.P. Peercy and C. Land, IEEE Trans. Elec-
Solid State, 4, 898 (1971). tron Devices, ED-28, 756 (1981).
93. J.L. Vossen and W. Kern (Eds.) Thin Film 115. W.P. Mason, Piezoelectric Crystals and
Processes, (Academic Press, New York, their Applications in Ultrasonics, (Van
1978). Nostrand, New York, 1950).
282 Dielectric Phenomena in Solids

116. K.S. Van Dyke, Proc. IRE., 16,742 (1928). 134. D. Liufu and K.C. Kao, J. Vac. Sci.
117. H.D. Megan, Ferroelectricy in Crystals, Technol., A16, 2360 (1998).
(Methuen, London, 1957). 135. K. Uchino, Piezoelectric Actuators and
118. D. Berlincourt, D. Curran, and H. Jaffee, in Ultrasonic Motors, (Kluwer Academic,
Physical Acoustics, edited by W.P. Mason Boston, 1996).
(Academic Press, New York, 1964). 136. K. Uchino, Ferroelectrics, 91, 281 (1989).
119. A.R. Von Hippel, Molecular Science and 137. K. Uchino, "Ceramic Actuators: Principles
Molecular Engineering, (John Wiley, New and Applications," (MRS Bulletin, April
York, 1959), p. 300. 1993), p. 42.
120. J.M. Herbert, Ferroelectric Transducers 138. J.G. Smith and W.S. Choi, IEEE Trans.
and Sensors, (Gordon Breach, London, Ultrason., Ferroelect, Freq. Contr. 38, 256
1982). (1991).
121. J. Van Randeraat and R. Setterinoton 139. Q.M. Wang, X.H. Du, B.M. Xu, and L.E.
(Eds.), Piezoelectric Ceramics, (Milliard, Cross, IEEE Trans. Ultrason., Ferroelect.,
London, 1974). Freq. Contr. 46, 638 (1999).
122. G. Shirane, K. Suzuki, and A. Takada, J. 140. PJ. Lock, Appl. Phys. Lett. 19, 390 (1971).
Phys. Soc. Japan 7, 12 (1952). 141. E.T. Keve, K.L. Bye, R.W. Whipps, and
123. G. Shirane and K. Suzuki, J. Phys. Soc. A.D. Annis, Ferroelectrics, 3, 39 (1971).
Japan 7, 333 (1952). 142. R.W. Whatmore, Rep. Prog. Phys. 49,1335
124. B. Jaffe, R. Roth, and J. Marzullo, J. Appl. (1988).
Phys., 25, 809 (1954). 143. C.N. Berglund and W.S. Baer, Phys. Rev.
125. G. Haertling and C. Land, J. Am. Ceram. 157, 358 (1967).
Soc, 54, 1 (1971). 144. H. Heywang, J. Mater, Sci. 6, 1214 (1971).
126. S. Liu, S. Pai, and J. Kyonka, Ferro- 145. N. Hirose and H. Sasaki, J. Am. Ceram.
electrics, 22, 689 (1978). Soc. 54, 320 (1971).
127. K. Hardti, Ferroelectrics, 72, 9 (1976). 146. S.B. Lang, Sourcebook of Pyroelectricity,
128. K. Okazaki, Ferroelectrics, 35,173 (1981). (Gordon and Breach, London, 1974).
129. M. Multani, S. Gokarn, R. Vijayaraghavan, 147. S.B. Lang and F. Steckel, Rev. Sci. Instru.
and V Polkar, Ferroelectrics, 37, 652 36, 1817 (1965).
(1981). 148. S.B. Lang, S.A. Shaw, L.H. Rice, and K.D.
130. N. Murayama, T. Oikawa, T. Katto, and Timmerhaus, Rev. Sci. Instru. 40, 274
K. Nakamura, J. Polym. Sci. Polym. Phys. (1969).
Ed. 13, 1033 (1975). 149. S.B. Lang, "Pyroelectricity: Its Phenome-
131. N. Murayama and Hazhizume, J. Polym. nology and Its Application to Temperature
Sci. Polym. Phys. Ed. 14, 989 (1976). Measurements" in Heat Transfer,
132. G. Pfister, M. Abkowtz, and R.G. Crystal, (Elsevier, Amsterdam, 1970), Section MT
J. Appl. Phys. 44, 2064 (1973). 1.6.
133. T. Furukawa, K. Ishida, and E. Fukada, J. 150. J.D. Zook and S.T. Liu, Ferroelectrics, 11,
Appl. Phys. 50, 4904 (1979). 371 (1976).