_______________________________________________2

MATHEMATICAL TREATMENT OF
FERROELECTRICS

Physicists usually treat a natural phenomenon using a simple mathematical form:

one is a linear approximation and another is a non-linear expansion theory. Hooke's
law, the stress - strain relation and Ohm's law, the voltage - current relation are two
of the most famous linear laws in physics. These linear relations are extended into
matrix or tensor relations in linear algebra. On the other hand, the Maclaurin or
Taylor series are popularly used to calculate slightly perturbed physical quantities
around an equilibrium state including non-linear effects. In this chapter, we will
consider the tensor representation of physical properties (linear relation) and
phenomenology of ferroelectricity (non-linear relation).

2.1 TENSOR REPRESENTATION OF PHYSICAL

PROPERTIES

(1) Tensor Representation1)

Let us first consider the tensor for electric conductivity. The conductivity is defined
so as to correlate an applied electric field E and the induced current density J as
follows:

J=σE. (2.1)

Since both the electric field and the current density are first rank tensor (that is,
vector) quantities, the conductivity should have a second rank tensor representation
(that is, with two suffixes); this is described as

J1 s 11 s 12 s 13 E1
J2 = s 21 s 22 s 23 E2 (2.2)
J3 s 31 s 32 s 33 E3

or
Ji = Σ σij Ej (2.3)
j
A third-rank tensor is exemplified by piezoelectric coefficients, providing a relation
between the applied field E and the induced strain x:

23
24 Chapter 2

x = d E. (2.4)

Since the E and x are first-rank and second-rank tensors, respectively, the d should
have a third-rank tensor form represented as

xjk = Σ d ijk Ei (2.5)

i
The d tensor is composed of three layers of the symmetrical matrices.

d111 d112 d113

1st layer (i = 1) d121 d122 d123
d131 d132 d133

d211 d212 d213

2nd layer (i = 2) d221 d222 d223 (2.6)
d231 d232 d233

d311 d312 d313

3rd layer (i = 3) d321 d322 d323
d331 d332 d333

Generally speaking, if two physical properties are represented using tensors of p-

rank and q-rank, the quantity which combines the two properties in a linear relation
is also represented by a tensor of (p + q)-rank.

(2) Crystal Symmetry and Tensor Form

A physical property measured along two different directions must be equal if these
two directions are crystallographically equivalent. This consideration sometimes
reduces the number of independent tensor components representing the above
property.

Let us again take electric conductivity as an example of a second-rank tensor. If the

current density J in an (x,y,z) coordinate system is described in an (x',y',z') system
as J', J and J' are related using a unitary matrix# as follows:
J' 1 a11 a12 a13 J1
J' 2 = a21 a22 a23 J2 (2.7)
J' 3 a31 a32 a33 J3

The electric field is transformed in the same way:

Mathematical Treatment of Ferroelectrics 25

E' 1 a11 a12 a13 E1

E' 2 = a21 a22 a23 E2 (2.8)
E' 3 a31 a32 a33 E3

or

E'i = Σ aij Ej (2.9)

Then, we can calculate the corresponding σ' tensor defined by

J’ 1 E’ 1
J’ 2 = σ' E’ 2 , (2.10)
J’ 3 E’ 3

σ' 11 σ' 12 σ' 13 a11 a12 a13

σ' 21 σ' 22 σ' 23 = a21 a22 a23
σ' 31 σ' 32 σ' 33 a31 a32 a33

σ11 σ12 σ13 a11 a21 a31

. σ21 σ22 σ23 a12 a22 a32 (2.11)
σ31 σ32 σ33 a13 a23 a33

# A unitary matrix without an imaginary part has the following relation:

a11 a12 a 31 -1 a 11 a21 a31

a21 a22 a23 = a 12 a22 a32
a31 a32 a33 a 13 a23 a 33

For centro-symmetry, the transformation matrix is written as

-1 0 0
0 -1 0
0 0 -1

and for rotation about a principal axis,

cos q sin q 0
-sin q cos q 0
0 0 1
26 Chapter 2

or

σ' ij = Σ aika jl σkl (2.12)

k,l

When the crystal has a 2-fold axis along the z-axis, the electric conductivity should
have the same tensor form in terms of the transformation:

-1 0 0
0 -1 0
0 0 1 .

From the condition

σ' 11 σ' 12 σ' 13 -1 0 0

σ' 21 σ' 22 σ' 23 = 0 -1 0
σ' 31 σ' 32 σ' 33 0 0 1

σ11 σ12 σ13 -1 0 0

. σ21 σ22 σ23 0 -1 0
σ31 σ32 σ33 0 0 1

σ11 σ12 σ13

= σ21 σ22 σ23 (2.13)
σ31 σ32 σ33

the following equivalencies can be derived:

σ31 = σ13 = σ32 = σ23 = 0

σ11 , σ22 , σ33 ? 0 (2.14)
σ12 = σ21

It is very important to note that most physical constants have a symmetric tensor
form. [The proof involves thermodynamical considerations that are beyond the
scope of this textbook. Refer to Ref. 1).]

For a third -rank tensor such as the piezoelectric tensor, the transformation due to a
change in coordinate system is represented by

d'ijk = Σ aila jmakn d lmn (2.15)

Mathematical Treatment of Ferroelectrics 27

When the crystal has a 4-fold axis along z-axis, for example, the transformation
matrix is given by

0 1 0
-1 0 0
0 0 1

Considering the tensor symmetry with m and n such that d123 = d132 and d213 =
d 231 (each matrix of the ith layer of the d tensor is symmetrical), we can obtain:

d 111 = d 222 = d 112 = d 121 = d 211 = d 221 = d 212 = d 122

= d 331 = d 313 = d 133 = d 332 = d 323 = d 233
= d 312 = d 321 = 0
d 333 ? 0
d 311 = d 322
d 113 = d 131 = d 223 = d 232
d 123 = d 132 = -d 213 = -d 231 (2.16)

Then we get the d tensor as follows:

1st layer
0 0 d131
0 0 d123
d131 d123 0

2nd layer
0 0 -d123
0 0 d131 (2.17)
-d123 d131 0

3rd layer
d311 0 0
0 d311 0
0 0 d333

(3) Reduction of the Tensor (Matrix Notation)

A general third-rank tensor has 33 = 27 independent components. Since dijk is

symmetrical in j and k some of the coefficients can be eliminated, leaving 18
independent d ijk coefficients; this facilitates the use of matrix notation.
28 Chapter 2

So far all the equations have been developed in full tensor notation. But when
calculating actual properties, it is advantageous to reduce the number of suffixes as
much as possible. This is done by defining new symbols, for instance, d21 = d211
and d14 = 2d 123 : The second and third suffixes in the full tensor notation are
replaced by a single suffix 1 to 6 in matrix notation, as follows:

Tensor notation 11 22 33 23,32 31,13 12,21

___________________________________________________________

Matrix notation 1 2 3 4 5 6

In terms of these new symbols the array (2.6) is rewritten as:

d11 (1/2)d 16 (1/2)d 15

(1/2)d 16 d 12 (1/2)d 14
(1/2)d 15 (1/2)d 14 d 13

d21 (1/2)d 26 (1/2)d 25

(1/2)d 26 d 22 (1/2)d 24 (2.18)
(1/2)d 25 (1/2)d 24 d 23

d31 (1/2)d 36 (1/2)d 35

(1/2)d 36 d 32 (1/2)d 34
(1/2)d 35 (1/2)d 34 d 33

The last two suffixes in the tensor notation correspond to those of the strain
components; therefore, for consistency, we make the following change in the
notation for the strain components.

x11 x12 x31 x 1 (1/2)x 6 (1/2)x 5

x12 x22 x23 ---> (1/2)x 6 x2 (1/2)x 4 (2.19)
x31 x23 x33 (1/2)x 5 (1/2)x 4 x3

The reason for the (1/2)s in the substitution (2.19) is due to the cancellation with
(1/2)s in (2.18). Then, we have

xj = Σ d ij Ei (i = 1, 2, 3; j = 1, 2, ..., 6) (2.20)

or
Mathematical Treatment of Ferroelectrics 29

x1 d11 d21 d31 E1

x2 d12 d22 d32 E2
x3 = d13 d23 d33 E3 (2.21)
x4 d14 d24 d34
x5 d15 d25 d35
x6 d16 d26 d36

Concerning the stress components, the (1/2)s are unnecessary.

X 11 X 12 X 31 X1 X6 X5
X 12 X 22 X 23 ---> X6 X2 X4 (2.22)
X 31 X 23 X 33 X5 X4 X3

The marix notation has the advantage of compactness over the tensor notation, and
it makes it easy to display the coefficients on a plane diagram. However, it must be
remembered that in spite of their form, the d ij's do not transform like the
components of a second-rank tensor. An example of a piezoelectric matrix for the
point group 4 is written as

0 0 0 d14 d15 0
0 0 0 d15 - d 14 0 (2.23)
d31 d31 d33 0 0 0

In the solid state theoretical treatment of the phenomenon of piezoelectricity or

electrostriction, the strain xkl is expressed in terms of the electric field Ei or electric
polarization P i as follows:

xkl = Σ d ikl Ei + Σ M ijkl Ei Ej

= Σ g iklP i + Σ QijklP iPj (2.24)

Here, dikl and g ikl are called the piezoelectric coefficients, and Mijkl and Qijkl the
electrostrictive coefficients. Since the E and x are first-rank and second-rank
tensors, respectively, d and M should be third-rank and fourth-rank tensors,
respectively.

Using a similar reduction of the notation for the electrostrictive coefficients Mijkl,
we get the following equation corresponding to Eq. (2.24):
30 Chapter 2

x1 d11 d21 d31 E1

x2 d12 d22 d32 E2
x3 = d13 d23 d33 E3
x4 d14 d24 d34
x5 d15 d25 d35
x6 d16 d26 d36

M 11 M 21 M 31 M 41 M 51 M 61 E 12
M 12 M 22 M 32 M 42 M 52 M 62 E 22
+ M 13 M 23 M 33 M 43 M 53 M 63 E 32 (2.25)
M 14 M 24 M 34 M 44 M 54 M 64 E 2 E3
M 15 M 25 M 35 M 45 M 55 M 65 E 3 E1
M 16 M 26 M 36 M 46 M 56 M 66 E 1 E2

Tables 2.1 and 2.2 summarize the matrices d and M for all crystallographic point
groups.1)

Example Problem 2.1_________________________________________________

Suppose that a shear stress is applied to a square crystal and the crystal is deformed
as illustrated in Fig. 2.1. Calculate the induced strain x5 ( = 2x31).

F
1 deg

F
1
Fig. 2.1 Shear stress and strain configuration.

Solution

Since x5 = 2x31 = tan θ = θ and 1° = π /180 rad., x5 = 0.017.

___________________________________________________________________
Mathematical Treatment of Ferroelectrics 31

Table 2.1 Piezoelectric coefficient matrix.*

32 Chapter 2

Table 2.1 (Continued) Piezoelectric coefficient matrix.*

Mathematical Treatment of Ferroelectrics 33

Table 2.2 Electrostrictive coefficient matrix.*

34 Chapter 2

Table 2.2 (Continued) Electrostrictive coefficient matrix.*

Mathematical Treatment of Ferroelectrics 35

Example Problem 2.2_________________________________________________

For a cube-shaped specimen, tensile stress X and compressive stress -X were

applied simultaneously along the (1 0 1) and (1 0 1) axes, respectively (Fig. 2.2).
When we take the prime -coordinates as illustrated in Fig. 2.2, the stress tensor is
represented as
X 0 0
0 0 0
0 0 -X .

Using the transformation matrix A

cosθ 0 sinθ
0 1 0
-sinθ 0 cosθ ,

calculate A .X .A -1 , and verify that the above-stress is equivalent to a pure shear

stress in the original (non-prime) coordinates.

X
3
3’
1’ -X

2, 2’
1

Fig. 2.2 Application of a pair of stresses X and - X to a cube of material

Solution

Using θ = - 45o , we can obtain the transformed stress representation:

A . X. A-1 = 0 0 X
0 0 0 . (P2.2.1)
X 0 0

The off-diagonal components X13 and X31 have the same magnitude X, and
represent a pure shear stress. Note that a shear stress is equivalent to a combination
36 Chapter 2

of extensional and contractional stresses. Only an extensional stress applied along a

diagonal direction 1' may exhibit an apparently similar diagonal distortion of the
crystal. However, precisely speaking, without the contraction along the 3' direction,
this is not exactly equivalent to the pure shear deformation.
___________________________________________________________________

Example Problem 2.3_________________________________________________

Barium titanate shows a tetragonal crystal symmetry (point group 4mm) at room
temperature. Therefore, its piezoelectric constant matrix is :
0 0 0 0 d15 0
0 0 0 d15 0 0
d31 d31 d33 0 0 0

(a) Calculate the induced strain under an electric field applied along the crystal c
axis.
(b) Calculate the induced strain under an electric field applied along the crystal a
axis.

Solution

x1 0 0 d31 E1
x2 0 0 d31 E2
x3 = 0 0 d33 E3 (P2.3.1)
x4 0 d15 0
x5 d15 0 0
x6 0 0 0

is transformed into

x1 = x2 = d 31 E3
x3 = d 33 E3
x4 = d 15 E2
x5 = d 15 E1 , E6 = 0 (P2.3.2)

(a) When E3 is applied, elongation in the c direction (x3 = d33 E3, d33 > 0) and
contaction in the a and b directions (x1 = x2 = d 31E3 , d 31 < 0) are induced.

(b) When E1 is applied, shear strain x5 (= 2x31) = d15 E1 is induced. Figure 2.3(a)
illustrates a case of d 15 > 0 and x5 >0.
___________________________________________________________________
Mathematical Treatment of Ferroelectrics 37

Polarization
3
2
(a) (b)
1
Fig. 2.3 (a) Orientation of the shear strain in the point group 4mm, and
(b) [111] axis strain in the point group m3m.

Example Problem 2.4_________________________________________________

Lead magnesium niobate exhibits a cubic crystal symmetry (point group m3m) at
room temperature and does not show piezoelectricity. However, large
electrostriction is induced under an applied electric field. The relation between the
strain and the electric field is given by

x1 M 11 M12 M12 0 0 0 E 12
x2 M 12 M11 M12 0 0 0 E 22
x3 = M 12 M12 M11 0 0 0 E 32 (P2.4.1)
x4 0 0 0 M44 0 0 E 2 E3
x5 0 0 0 0 M 44 0 E 3 E1
x6 0 0 0 0 0 M 44 E 1 E2

in a matrix representation. Calculate the induced strain under an electric field

applied along the [111] direction.

Solution

The electric field along the [111] direction , E[111] , is represented as (E[111]/√3,
E[111]/√3, E[111]/√3). Substituting E1 = E2 = E3 = E[111]/√3 into Eq. (P2.2.1), we
obtain
38 Chapter 2

x1 = x2 = x3 = (M 11 + 2M 12 )E[111]2/3 (= x11 = x22 = x33),

x4 = x5 = x6 = M 44 E[111]2/3 (= 2x23 = 2x31 = 2x12). (P2.4.2)

The distortion is illustrated in Fig. 2.3(b). The strain x induced along an arbitrary
direction is given by

x = Σ xij l i lj (P2.4.3)

where li is a direction cosine with respect to the i axis. Therefore, the strain induced
along the [111] direction, x[111]// , is given by

x[111]// = Σ xij (1/√3)(1/√3)

= [x1 + x2 + x3 + 2(x4/2 + x5/2 + x6/2)]/3
= (M 11 + 2M 12 + M 44) E[111]2/3. (P2.4.4)

On the other hand, the strain induced perpendicular to the [111] direction, x[111] ,
is calculated in a similar fashion as

x[111] = (M 11 + 2M 12 - M 44/2)E[111]2/3. (P2.4.5)

Figure 2.11(b) shows the distortion schematically. It is important to note that the
volumetric strain (∆V/V) given by

x[111]// + 2 x[111] = (M 11 + 2 M 12 ) E[111]2 (P2.4.6)

is independent of the applied field direction.

___________________________________________________________________

2.2 PHENOMENOLOGY OF FERROELECTRICITY

(1) Landau Theory of the Phase Transition

A thermodynamical theory explaining the behavior of a ferroelectric crystal can be

obtained by considering the form of the expansion of the free energy as a function
of the polarization P. We assume that the Landau free energy F in one dimension is
represented formally as:

F(P,T) = (1/2)α P2 + (1/4)β P4 + (1/6)γ P6 + ... (2.26)

The coefficients α, β, γ depend, in general, on the temperature. Note that the series
does not contain terms in odd powers of P because the free energy of the crystal will
not change with polarization reversal (P --> -P). The phenomenological
formulation should be applied for the whole temperature range over which the
material is in the paraelectric and ferroelectric states.
Mathematical Treatment of Ferroelectrics 39

The equilibrium polarization in an electric field E satisfies the condition:

(∂F/ ∂P) = E = α P + β P3 + γ P 5 (2.27)

To obtain the ferroelectric state, the coefficient of the P2 term must be negative for
the polarized state to be stable, while in the paraelectric state it must be positive
passing through zero at some temperature T0 (Curie-Weiss temperature):

α = (T - T0 )/ε0 C (2.28)

where C is taken as a positive constant called the Curie-Weiss constant and T0 is

equal to or lower than the actual transition temperature TC (Curie temperature). The
temperature dependence of α is related on a microscopic level to the temperature
dependence of the ionic polarizability coupled with thermal expansion and other
effects of anharmonic lattice interactions. Refer to the discussion in Section 1.2.

Second-order transition

When β is positive, γ is often neglected because nothing special is added by this

term. The polarization for zero applied field is obtained from Eq. (2.27) as

[(T - T0 )/ε0 C] PS + β PS3 = 0 (2.29)

so that either PS = 0 or P S2 = (T0 - T)/β ε0 C.

For T > T0, the unique solution PS = 0 is obtained. For T < T0 the minimum of the
Landau free energy is obtained at:
______________
PS = √(T 0 - T)/(β ε0 C). (2.30)

The phase transition occurs at TC = T0 and the polarization goes continuously to

zero at this temperature; this is called a second-order transition.

The relative permittivity ε is calculated as:

1/ε = ε0/(∂P/∂E) = ε0 (α + 3β P2) (2.31)

Then,
ε = C/(T - T0) (T > T0)
C/[2(T - T0 )] (T < T0) (2.32)

Figure 2.4(a) shows the variations of PS and ε with temperature. It is notable that
the permittivity becomes infinite at the transition temperature. Triglycine sulphate
is an example of a ferroelectric exhibiting the second-order transition.
40 Chapter 2

Permittivity ε Permittivity ε
Spontaneous Inverse Spontaneous
Ps permittivity Ps Inverse
1/ ε permittivity
1/ ε

Tc Tc
Temperature Temperature
(Curie Temp.) (Curie Temp.)
(a) (b)
Fig. 2.4 Phase transitions in a ferroelectric: (a) second-order and (b) first-order.

First-order transition

When β is negative in Eq. (2.26) and γ is taken positive, the transition becomes first
order. The equilibrium condition for E = 0 in Eq. (2.33) leads to either PS = 0 or Eq.
(2.34).

[(Τ − Τ0 )/ε0 C] PS + β PS3 + γ PS5 = 0 (2.33)

_________________
PS2 = [- β + √ β2 - 4γ(T - T0 )/ε0 C] /2γ (2.34)

The transition temperature TC is obtained from the condition that the free energies
of the paraelectric and ferroelectric phases are equal: i.e., F = 0, or:

[(Τ − Τ0 )/ε0 C] + (1/2) β P S2 + (1/3) γ PS4 = 0 (2.35)

Therefore:

TC = T 0 + (3/16)(β2 ε0 C/ γ) (2.36)

Note that the Curie temperature TC is slightly higher than the Curie-Weiss
temperature T0, and that a discrete jump of PS appears at TC. Also, the permittivity
exhibits a finite maximum at TC for a first-order transition [Fig. 2.4(b)]. Barium
titanate is an example of a ferroelectric that undergoes a first-order phase transition.

The free energy curves are plotted for the second- and first-order phase transitions
at various temperatures in Fig. 2.5. In the case of β > 0, the phase transition is not
associated with a latent heat, but with a jump of the specific heat; thus, this is called
a second-order transition. On the other hand, in the case of β < 0, the transition
exhibits a latent heat, and is called a first-order transition, where the permittivity
shows a maximum and a discontinuity of the spontaneous polarization appears at
TC.
Mathematical Treatment of Ferroelectrics 41

Free Energy Free Energy

T>TC T>T1
TC<T<T1
T=TC

T=TC
T<TC

T0<T<TC

P P
T<T0
(a) (b)

Fig. 2.5 Free energy curves plotted for the second- (a) and first-order (b) phase
transitions at various temperatures.

Example Problem 2.5_________________________________________________

Verify the difference between the Curie and Curie-Weiss temperatures as expressed
by:

TC = T 0 + (3/16)(β2 ε0 C/ γ)

for a first-order phase transition, where the Landau free energy is expanded as

F(P,T) = (1/2)α P 2 + (1/4)β P4 + (1/6)γ P6,

α = (T - T0 )/ε0 C.

Hint

The potential minima are obtained from

(∂F/ ∂P) = E = α P + β P3 + γ P5 = 0 . (P2.5.1)

There are generally three minima including P = 0 (F = 0).

At the Curie temperature, the free energy at the non-zero polarization must be equal
to zero (F = 0). Thus we obtain another condition:

F = (1/2)α P2 + (1/4)β P4 + (1/6)γ P6 = 0 . (P2.5.2)

42 Chapter 2

Solution

Equations (P2.5.1) and (P2.5.2) are reduced for non-zero polarizations to

α + β P2 + γ P4 = 0 , (P2.5.3)
α + (1/2) β P2 + (1/3) γ P4 = 0 . (P2.5.4)

Note that Eq. (P2.5.3) is valid for all temperatures below TC, but Eq. (P2.5.4) is
only valid at T = TC. Eliminating the P terms from these two equations, we obtain

α + β (- 4 α/ β) + γ (- 4 α/ β)2 = 0 . (P2.5.5)

Then, taking account of α = (T - T0 )/ε0 C, the Curie temperature is calculated as

TC = T 0 + (3/16)(β2 ε0 C/ γ) . (P2.5.6)
___________________________________________________________________

(2) Phenomenology of Electrostriction

In a ferroelectric whose prototype phase (high temperature paraelectric phase) is

centrosymmetric and non-piezoelectric, the piezoelectric coupling term PX is
omitted and only the electrostrictive coupling term P2X is introduced. The theories
for electrostriction in ferroelectrics were formulated in the 1950s by Devonshire 2)
and Kay.3) Let us assume that the elastic Gibbs energy should be expanded in a
one-dimensional form:

G1 (P,X,T) = (1/2)α P2 + (1/4)β P4 + (1/6)γ P6

- (1/2)s X2 - Q P2 X, (α = (T - T0)/ ε0 C) (2.37)

where P, X, T are the polarization, stress and temperature, respectively, and s and Q
are called the elastic compliance and the electrostrictive coefficient, respectively.
This leads to Eqs. (2.38) and (2.39).

E = (∂G1/ ∂P) = α P + β P3 + γ P5 - 2Q P X (2.38)

x = - (∂G1/∂X) = sX + QP2 (2.39)

Case I: X = 0

When the external stress is zero, the following equations are derived:

E = α P + β P3 + γ P5 (2.40)
x = QP2 (2.41)
1/ε0 ε = α + 3 β P2 + 5 γ P4 (2.42)
Mathematical Treatment of Ferroelectrics 43

If the external electric field is equal to zero (E = 0), two different states are derived;
_______
P = 0 and P = (√ β2- 4αγ - β)/2γ.
2

(i) Paraelectric phase: P S = 0 or P = ε0 ε E (under small E)

Permittivity: ε = C/(T - T0) (Curie -Weiss law) (2.43)

Electrostriction: x = Q ε02 ε2 E2 (2.44)

The previously mentioned electrostrictive coefficient M in Eq. (2.24) is related to

the electrostrictive Q coefficient through

M = Q ε02 ε2 (2.45)
_______
(ii) Ferroelectric phase: P S2= (√ β2- 4αγ - β)/2γ or P = PS + ε0 ε E (under small E)

x = Q(P S + ε0 ε E)2 = QP S2 + 2 ε0 ε QP SE + Q ε02 ε2 E2 (2.46)

where we define the spontaneous strain xS and the piezoelectric constant d as:

Spontaneous strain: xS = QPS2 (2.47)

Piezoelectric constant: d = 2 ε0 ε QP S (2.48)

We see by Eq. (2.48) that piezoelectricity is equivalent to the electrostrictive

phenomenon biased by the spontaneous polarization. The temperature dependences
of the spontaneous strain and the piezoelectric constant are plotted in Fig. 2.6.

Case II: X ? 0

When a hydrostatic pressure p (X = - p) is applied, the inverse permittivity is

changed in proportion to p:

1/ε0 ε = α + 3 β P2 + 5 γ P4 + 2Qp (Ferroelectric state)

α + 2Qp = (T - T0 + 2Qε0 Cp)/(ε0 C) (Paraelectric state) (2.49)

Therefore, the pressure dependence of the Curie -Weiss temperature T0 or the

transition temperature TC is derived as follows:

(∂T0/∂p) = (∂TC/∂p) = - 2Qε0C (2.50)

In general, the ferroelectric Curie temperature is decreased with increasing

hydrostatic pressure (i.e. Qh > 0).
44 Chapter 2

Piezoelectric
Spontaneous constant d
strain x s

Tc (Curie Temp.)
Temperature

Fig. 2.6 Temperature dependence of the spontaneous strain and the piezoelectric
constant.

Example Problem 2.6_________________________________________________

Barium titanate has d33 = 320 x 10-12 C/N, εc (= ε3 ) = 800 and Q33 = 0.11 m4C-2
at room temperature. Estimate the spontaneous polarization PS.

Solution

Let us use the relation:

d 33 = 2 ε0 ε3 Q33P S . (P2.6.1)

PS = d 33/2 ε0 ε3 Q33
= 320x10-12[C/N]/{2 x 8.854x10-12 [F/m] x 800 x 0.11 [m4 C-2]}
= 0.21 [C/m2] (P2.6.2)
___________________________________________________________________

Example Problem 2.7_________________________________________________

In the case of a second-order phase transition, the elastic Gibbs energy is expanded
in a one-dimensional form as follows:

G1 (P,X,T) = (1/2)α P2 + (1/4)β P4

- (1/2)s X2 - Q P2 X, (P2.7.1)

where only the coefficient α is dependent on temperature, α = (T - T0)/ ε0 C. Obtain

the dielectric constant, spontaneous polarization, spontaneous strain and
piezoelectric constant as a function of temperature.
Mathematical Treatment of Ferroelectrics 45

Solution

E = (∂G1/ ∂P) = α P + β P3 - 2Q P X (P2.7.2)

x = - (∂G1/∂X) = sX + QP2 (P2.7.3)

When an external stress is zero, we can deduce the three characteristic equations:

E = α P + β P3 (P2.7.4)
x = QP2 (P2.7.5)
1/ε0 ε = (∂E/∂P) = α + 3 β P2 (P2.7.6)

By setting E = 0 initially, we obtain the following two stable states: PS2 = 0 or

-α/β.

(i) Paraelectric phase -- T > T0 -- PS = 0

1/ε0 ε = α, then ε = C/(T - T 0) (P2.7.7)

___________
(ii) Ferroelectric phase -- T < T0 -- P S = √(T0 - T)/ε0 Cβ (P2.7.8)

1/ε0 ε = α + 3 β P2 = - 2α, then ε = C/2(T0 - T) (P2.7.9)

xS = QPS2 = Q (T 0 - T)/ ε0 Cβ (P2.7.10)

From Eqs. (P2.7.8) and (P2.7.9), the piezoelectric constant is obtained as

d = 2 ε0 ε Q P S
_____
= Q√ε0 C/β (T 0 - T)-1/2 (P2.7.11)
___________________________________________________________________

(3) Converse Effects of Electrostriction

So far we have discussed the electric field induced strains, i.e. piezoelectric strain
(converse piezoelectric effect, x = d E) and electrostriction (electrostrictive effect, x
= M E2 ). Let us consider here the converse effect, that is, the material's response to
an external stress, which is applicable to sensors. The direct piezoelectric effect is
the increase of the spontaneous polarization by an external stress, and expressed as

∆P = d X. (2.51)

On the contrary, since an electrostrictive material does not have a spontaneous

polarization, it does not generate any charge under stress, but does exhibit a change
in permittivity [see Eq. (2.49)]:
46 Chapter 2

∆(1/ ε0 ε) = 2QX (2.52)

This is the converse electrostrictive effect. The converse electrostrictive effect, the
stress dependence of the permittivity, is used in stress sensors.4) A bimorph
structure which subtracts the static capacitances of two dielectric ceramic plates can
provide superior stress sensitivity and temperature stability. The capacitance
changes of the top and bottom plates have opposite signs for uniaxial stress and
same sign for temperature change. The response speed is limited by the capacitance
measuring frequency to about 1 kHz. Unlike piezoelectric sensors, electrostrictive
sensors are effective in the low frequency range, especially DC. The sensor
applications of piezoelectrics will be discussed in Section 7.2 of Chapter 7.

(4) Temperature Dependence of Electrostriction

Several expressions for the electrostrictive coefficient Q have been given so far.
From the data obtained by independent experimental methods such as

1) electric field-induced strain in the paraelectric phase,

2) spontaneous polarization and spontaneous strain (x-ray
diffraction) in the ferroelectric phase,
3) d constants from the field-induced strain in the ferroelectric phase
or from piezoelectric resonance,
4) pressure dependence of permittivity in the paraelectric phase,

Fig. 2.7 Temperature dependence of the electrostrictive constants Q33 and Q13 in
Pb(Mg1/3Nb 2/3)O3.
Mathematical Treatment of Ferroelectrics 47

nearly equal values of Q were obtained. Figure 2.7 shows the temperature
dependence of the electrostrictive coefficients Q33 and Q31 for the complex
perovskite Pb(Mg1/3Nb 2/3)O3 , whose Curie temperature is near 0°C.5) It is seen
that there is no significant anomaly in the electrostrictive coefficient Q through the
temperature range in which the paraelectric to ferroelectric phase transition occurs
and piezoelectricity appears. Q is almost temperature independent.

2.3 PHENOMENOLOGY OF ANTIFERROELECTRICITY

(1) Antiferroelectrics

The previous sections dealt with the case in which the directions of the spontaneous
dipoles are parallel to each other in a crystal (polar crystal). There are cases in
which antiparallel orientation lowers the dipole-dipole interaction energy. Such
crystals are called anti-polar crystals. Figure 2.8 shows the orientation of the
spontaneous electric dipoles in an anti-polar state in comparison with a non-polar
and a polar state. In an anti-polar crystal, where the free energy of an antipolar state
does not differ appreciatively from that of a polar state, the application of an
external electric field or mechanical stress may cause a transition of the dipole
orientation to a parallel state. Such crystals are called antiferroelectrics.

Figure 2.9 shows the relationship between E (applied electric field) and P (induced
polarization) in paraelectric, ferroelectric and antiferroelectric phases. In a
paraelectric phase the P-E relation is linear; in a ferroelectric phase there appears a
hysteresis caused by the transition of the spontaneous polarization between the
positive and negative directions; in an antiferroelectric phase, at low electric field,
the induced polarization is proportional to E, and when E exceeds a certain value
Ecrit , the crystal becomes ferroelectric (electric field induced phase transition), and
the polarization shows hysteresis with respect to E. After removal of the electric
field, the crystal returns to its anti-polar state, and hence, no spontaneous
polarization can be observed as a whole. This is called a double hysteresis curve.

stripe type checker board type

(a) Nonpolar (b) Polar (c) Anti-polar

Fig. 2.8 Schematic arrangement of the spontaneous dipoles in non-polar, polar and
antipolar materials.
48 Chapter 2

Polarization Polarization

Electric field
Electric field Ec

(a) Paraelectric Polarization

(b) Ferroelectric Electric field

Et
(c) Antiferroelectric

Fig. 2.9 Polarization vs. electric field hysteresis curves in paraelectric, ferro-
electric and antiferroelectric materials.

(2) Phenomenology of Antiferroelectrics

We will discuss here the introduction of electrostrictive coupling in Kittel's free

energy expression for antiferroelectrics.6,7) The simplest model for
antiferroelectrics is the "one-dimensional two-sublattice model." It treats the
coordinates as one-dimensional, and a superlattice (twice the unit lattice) is formed
from two neighboring sublattices each having a sublattice polarization Pa and Pb .
The state Pa = Pb represents the ferroelectric phase, while Pa = - Pb , the
antiferroelectric phase. For the electrostrictive effect, ignoring the coupling
between the two sublattices, the strains from the two sublattices are QPa 2 and
QPb 2 , respectively (assuming equal electrostrictive constants Q for both
sublattices). The total strain of the crystal becomes

x = Q(Pa 2 + Pb 2)/2. (2.53)

However, since antiferroelectricity originates from the coupling between the

sublattices, it is appropriate to consider the sublattice coupling also for the
Mathematical Treatment of Ferroelectrics 49

electrostrictive effect. The coupling term for the electrostriction Ω is introduced in

the following form:

G1 = (1/4) α (Pa 2+Pb 2 ) + (1/8) β (Pa4 +Pb 4) + (1/12) γ (Pa6 +Pb 6)

+(1/2) η PaPb - (1/2) χT p 2 +(1/2) Qh (Pa 2 +Pb 2 +2Ω Pa Pb )p (2.54)

in which hydrostatic pressure p is employed, and χT is the isothermal

compressibility, Qh and Ω are the electrostrictive constants. Introducing the
transformations PF = (Pa + Pb )/2 and PA = (Pa - Pb )/2 leads to the following
expression:

G1 = (1/2) α (P F2 + PA 2 ) + (1/4) β (PF 4 + PA 4 + 6P F2 PA 2)

+ (1/6) γ (P F 6+PA 6+15P F4 PA 2+15P F2 PA 4)+(1/2) η (P F 2 - PA 2 )
- (1/2) χ T p 2 + Qh [PF 2 +PA 2 + Ω (PF 2 - PA 2)]p (2.55)

The dielectric and elastic equations of state follow as

∂G1/∂PF = E = PF [α + η + 2 Q (1 + Ω)p + β PF 2 + 3β PA 2 + γ PF 4
+ 10γ P F 2PA 2 + 5γ PA 4] (2.56)
∂G1/∂PA = 0 = PA [α − η + 2 Q (1 - Ω)p + β PA 2 + 3β PF 2 + γ PA 4
+ 10γ P F 2PA 2 + 5γ P F4 ] (2.57)
∂G1/∂p = ∆V/V = - χ T p +Qh (1+Ω)PF + Qh (1 - Ω)PA
2 2 (2.58)

Hence, the induced volume change in the paraelectric phase can be related to the
induced ferroelectric polarization by the following formula:

(∆V/V) ind = Qh (1 + Ω)P F,ind2 (2.59)

Below the phase transition temperature (this temperature for antiferroelectrics is

called Neel temperature) the spontaneous volume strain and the spontaneous
antiferroelectric polarization are related as

(∆V/V)S = Qh (1 - Ω)PA,S 2 (2.60)

Even if the perovskite cystal shows Qh > 0, the spontaneous volume strain can be
positive or negative depending on the value of Ω (Ω < 1 or Ω > 1), that is, if the
inter-sublattice coupling is stronger than the intra-sublattice coupling, a volume
contraction is observed at the Neel point. This is quite different from ferroelectrics,
which always show a volume expansion at the Curie point. Figure 2.10 illustrates
the spontaneous strains in a crystal schemetically for Ω > 0. When Pa and Pb are in
50 Chapter 2

the parallel configuration (ferroelectric phase), the Ω-term acts to increase the strain
xS, when they are in the anti-parallel configuration (antiferroelctric phase), the Ω-
term acts to decrease the strain.

This phenomenological theory explains well the experimental results for the
antiferroelectric perovskite crystal PbZrO3 and others.8) Figure 2.11 shows the
strain in the antiferroelectric ceramic Pb 0.99 Nb 0.02[(Zr0.6Sn 0.4)0.94 Ti0.06 ]0.98O3
as a function of an applied electric field.9) The large change in the strain associated
with the field -induced transition from the antiferroelectric to ferroelectric phase can
be estimated to be

(∆V/V) = Qh (1 + Ω)PF,S2 - Qh (1 - Ω)PA,S2

= 2 Qh Ω PF,S2 (2.61)

Here, we assume that the magnitudes of Pa and Pb do not change drastically

through the phase transition.

(a) Ferroelectric Arrangement

x = Q (1+Ω) (Pa + Pb )2/4

x = Q Pa2 x = Q P b2

(b) Antiferroelectric Arrangement

x = Q (1-Ω) (Pa - Pb )2/4

Fig. 2.10 Intuitive explanation of the sublattice coupling with respect to

electrostriction (for Ω > 0).
Mathematical Treatment of Ferroelectrics 51

Fig. 2.11 Field induced strain in a Pb(Zr,Sn)O3 based antiferroelectric.

CHAPTER ESSENTIALS_________________________________

1. Tensor representation: when two physical properties are represented using

tensors of p-rank and q-rank, the quantity which combines the two properties in
a linear relation is also represented by a tensor of (p + q)-rank.

2. A physical property measured along two different directions must be equal if

these two directions are crystallographically equivalent. This consideration
reduces the number of the independent tensor components representing the
above property.

d ijk' = Σ a il a jm a kl d lmn (a il, a jm, akl : transformation matrices)

l,m,n
d ijk' = d ijk

3. Shear strain: x5 = 2 x31 = 2 φ, taken as positive for smaller angle.

4. Phenomenology:
(1/4) β P 4 > 0 --> second-order phase transition
(1/4) β P 4 < 0 --> first-order phase transition

5. Electrostriction equation:

x = Q PS2 + 2 Q ε0 ε P S E + Q ε0 2 ε2 E2
spontaneous strain piezostriction electrostriction
52 Chapter 2

The electrostrictive Q constant is insensitive to temperature changes.

6. Piezoelectric equations:

x = sE X + d E
P = d X + ε0 εX E

7. In antiferroelectrics, consideration of sublattice couplings is essential to

understanding the stable sublattice polarization configuration and the
significant jump in strain associated with an antiferroelectric - ferroelectric
phase transition induced by an external electric field.
___________________________________________________________________

CHAPTER PROBLEMS

2.1 The room temperature form of quartz belongs to class 32.

(1) Find that the piezoelectric matrix (d ij) is given by

d11 -d11 0 d14 0 0

0 0 0 0 -d14 -2d11
0 0 0 0 0 0
Notice that the piezoelectric tensor must be invariant for a 120° rotation
around the 3-axis and for a 180°rotation around the 1-axis. The
transformation matrices are

-1/2 √3/2 0
-√3/2 -1/2 0
0 0 1
and
1 0 0
0 -1 0 ,
0 0 -1
respectively.

(2) The measured values of the d ij for right-handed quartz are

-2.3 2.3 0 -0.67 0 0 x 10 -12 [C/N]

0 0 0 0 0.67 4.6
0 0 0 0 0 0
Mathematical Treatment of Ferroelectrics 53

(a) When a compressive stress of 1 kgf/cm2 is applied along the 1-axis of a

quartz crystal, find the polarization generated.
(kgf = kilogram force = 9.8 N)

(b) When an electric field of 100 V/cm is applied along the 1-axis, find the
strains generated.

Hint

Let us initially use the following equation for a 180o rotation (with respect
to the a-axis):

d ijk' = Σ a il a jm a kl d lmn (a il, ajm , akl : transformation matrices)

l,m,n

Considering that a11 = 1, a22 = -1, a33 = -1, we obtain the following
relationship:
d111 '=d111 d112 '=- d112 d113 '=- d113
d121 '=- d121 d122 '=d122 d123 '=d123
d131 '=- d131 d132 '=d132 d133 '=d133

d211 '=- d211 d212 '=d212 d213 '=d213

d221 '=d221 d222 '=- d222 d223 '=- d223
d231 '=d231 d232 '=- d232 d233 '=- d233

d311 '=- d311 d312 '=d312 d313 '=d313

d321 '=d321 d322 '=- d322 d323 '=- d323
d331 '=d331 d332 '=- d332 d333 '=- d333

or
d111 0 0
0 d122 d123
0 d123 d133

0 d212 d231
d212 0 0
d231 0 0

0 d312 d331
d312 0 0
d331 0 0
54 Chapter 2

Next, a 120o rotation is considered such that a11 = -1/2, a12 = √3/2, a21 =
-√3/2, a 22 = -1/2, a 33 = 1:

d 111' = Σ a 1l a1m a 1n d lmn

l,m,n

= Σ a11 a 1m a1n d 1mn + Σ a 12 a1m a 1n d 2mn

m,n m,n

= (-1/2) 3 d 111 + (-1/2)(v3/2)2 d 122 + 2(-1/2)(v3/2) 2 d 212

= - (1/8) d 111 - (3/8) d 122 - (3/4) d 212
= d 111

d 122' = Σ a11 a 2m a2n d 1mn + Σ a 12 a2m a 2n d 2mn

m,n m,n
= (-1/2)(-v3/2) 2 d 111 + (-1/2) 3 d 122 + 2(v3/2)(-v3/2)(-1/2) d 212
= - (3/8) d 111 - (1/8) d 122 + (3/4) d 212
= d 122

Continuing the calculations for d123, d212, d231, d312, d331 , we can
obtain all the necessary equations for deriving the final matrix form.

2.2 In the case of a first-order phase transition, the Landau free energy is
expanded as in Example Problem 2.5. Calculate the inverse permittivity in
the vicinity of the Curie temperature, and verify that the slope (?(1/ε)/∂Τ)
just below TC is 8 times larger than the slope just above TC.

Hint

In a first-order phase transition, PS satisfies the following equation in the

temperature range of T < TC :

α + β Ps 2 + γ Ps 4 = 0 .

The permittivity is given by

1/ε0 ε = α + 3 β Ps 2 + 5 γ Ps 4 .

Thus,
1/ε0 ε = α + 3 β Ps 2 + 5 (- α - 3 β Ps 2)
= - 4 α - 2 β Ps 2
_______
Since α = (T - T 0 )/ε0 C, P S2 = (√ β2 - 4αγ - β)/2γ and
Mathematical Treatment of Ferroelectrics 55

(T - T0)/ ε0 C = (3/16)(β 2/ γ) - (TC - T)/ ε0 C,

we can obtain

1/ε0 ε = - 4 α - 2 β Ps 2

= - 4 [(3/16)(β 2/ γ) - (T C - T)/ ε0 C]
______________________________
+ (β / γ) - (β/γ)(√ β 2- 4γ [(3/16)(β2/ γ) - (T C - T)/ε0 C]
2

Considering (TC - T) << 1, obtain the approximation of this equation.

REFERENCES
1) J. F. Nye: Physical Properties of Crystals, Oxford University Press, London, p.123,
p.140 (1972)
2) A. F. Devonshire: Adv. Phys. 3, 85 (1954)
3) H. F. Kay: Rep. Prog. Phys. 43, 230 (1955)
4) K. Uchino, S. Nomura, L. E. Cross, S. J. Jang and R. E. Newnham: Jpn. J. Appl.
Phys. 20, L367 (1981); K. Uchino, S. Nomura, L. E. Cross, S. J. Jang and R. E.
Newham: Jpn. J. Appl. Phys. 20, L367 (1981); K. Uchino: Proc. Study Committee
on Barium Titanate, XXXI-171-1067 (1983)
5) J. Kuwata, K. Uchino and S. Nomura: Jpn. J. Appl. Phys. 19, 2099 (1980)
6) C. Kittel: Phys. Rev. 82, 729 (1951)
7) K. Uchino: Solid State Phys. 17, 371 (1982)
8) K. Uchino, L. E. Cross, R. E. Newnham and S. Nomura: J. Appl. Phys. 52, 1455
(1981)
9) K. Uchino: Jpn. J. Appl. Phys. 24, Suppl. 24-2, 460 (1985)
56 Chapter 2

<< Do You Know ? >>

Cellular Telephone

Portable communication devices such as cordless, portable, and car telephones have
become popular worldwide. Do you know what kind of ferroelectric and dielectric
components are used in a cellular phone?

555-8035

Chip Monolithic Ceramic Capacitors

Microwave Oscillators
Microwave Filters
Chip Monolithic LC Filters
Ceramic Resonators
High Frequency SAW Filters
Ceramic Filters
Piezoelectric Receivers
Piezoelectric Speakers