Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

CHINESE JOURNAL OF PHYSICS VOL. 42, NO.

5 OCTOBER 2004

Study of the Defect Structure Effect on Electrical Conductivity in


Nonstoichiometric Lithium Tantalate

N. Masaif,1, 2 S. Elhamd,1 S. Jebbari,1 A. Jennane,1 and F. Bennani3


1
Département de Physique, Faculté des Sciences et Techniques, LMM, BP 577 Settat, Morocco
2
Département de Physique, Faculté des Sciences, LPNA, BP 133 Kénitra, Morocco
3
Département de Physique, Faculté des Sciences, LPVC, BP 133 Kénitra, Morocco
(Received July 10, 2003)
An analysis of ceramics specimens of LiTaO3 indicates that there is a modification of the
conductivity with different nonstoichiometric compositions. To understand the mechanism
responsible in this phenomenon, we have proposed a theoretical description of the defect
structure in LiTaO3 on the basis of the lithium and tantalate vacancy models combined
with a ferroelectric phase transition theory. A comparative study between the calculated
and measured values of the conductivities shows that the lithium vacancy model is the
best model that has been suggested for describing the defect structure in nonstoichiometric
lithium tantalate.
PACS numbers: 72.15Eb, 72.20Dp

I. INTRODUCTION

Lithium tantalate has been the subject of many theoretical and practical investiga-
tions due to their properties in electro-optics, electo-acoustics, and non-linear optics. This
material, which shows a ferroelectric behaviour, is well known to be a narrow-range nonsto-
ichiometric compound. In LiTaO3 , the solid solubility range extends from about 46 to 50.4
mol % of Li2 O at room temperature [1]. The structure of ferroelectric LiTaO3 belongs to
the space group R3c and can be considered as a superstructure of the α-Al2 O3 corundum
structure with Li+ and Ta5 cations along the c-axis [2].
The presence of nonstoichiometry in the form of lattice vacancies profoundly alters
the physical properties, such as the electrical conductivity. In this work, we analyse the
electrical conductivity as a function of temperature for some lithium tantalate samples with
different compositions. Dielectric measurements have shown that there is a considerable
effect of nonstoichiometry on the phase transition temperature in ferroelectric solid solu-
tions LiTaO3 and LiNbO3 . So these compounds are characterized, in order to clarify the
relationships between the defect structure and the optical properties.
Several studies have reported on the changes in electrical conductivity of the lithium
tantalate solid solutions [3–6]. By complex impedance spectroscopy, Huanosta et al. [4]
studied the variation of the conductivity as a function of the stoichiometry in Li1−5x Ta1+x O3
compounds.
In a previous paper [7], the successive phase transitions corresponding to the various
compositions in the LiTaO3 system have been re-examined. In brief, the ceramic speci-

http://PSROC.phys.ntu.edu.tw/cjp 649 c 2004 THE PHYSICAL SOCIETY


OF THE REPUBLIC OF CHINA
650 STUDY OF THE DEFECT STRUCTURE EFFECT ON . . . VOL. 42

mens of LiTaO3 with different compositions were studied by impedance spectroscopy in the
1 Hz-1 MHz frequency range and the 500–1200 K temperature range. The temperature
dependence of the dielectric permittivity of these specimens was measured, in order to de-
termine the temperature of the phase transition. Thus, the ionic conductivity is determined
for all compositions from Cole-Cole diagrams. The analysis of the ceramic samples, which
were prepared from Li2 O and Ta2 O5 , indicates that there is a modification of the ionic
conductivity of LiTaO3 with different compositions.
To understand which defect or defects are involved in this phenomenon, we have
proposed a theoretical description of the defect structure in LiTaO3 on the basis of a set
of vacancy models combined with a ferroelectric phase transition theory. The theoretical
models which predict such a composition dependent conductivity either assume oxygen and
lithium vacancies, that an excess of niobium ions might occupy the lithium sites, or that
there is a new defect structure corresponding to the Nb-site vacancy type. The purpose of
this work is to investigate the defect possibly responsible for these phenomena and try to
explain their mechanism of conduction.
LiTaO3 is isomorphous with LiNbO3 ; several models of defects have been proposed for
LiNbO3 . Fay et al. [8] describe a ceramic sample assumed to contain oxygen and lithium
vacancies by the oxygen vacancy model that has the formula [Li1−2x 2x ][Nb][O3−x x ],
where [.] represents the sublattice and denotes the vacancies. This model presents a
certain anomaly in describing the density variation as a function of the composition. In
order to correct this anomaly, Lerner et al. [9] have predicted that an excess of niobium
ions might occupy the lithium sites. This is the lithium vacancy model, which is given by
the formula [Li1−5x Nb−x 4x ][Nb]O3 , where the lithium vacancies are introduced taking
into account the charge compensation. However, Peterson and Carnevale [10] observed a
new defect structure corresponding to the Nb-site vacancy type using the NMR technique.
Their niobium vacancy model is arranged as follows: [Li1−5x Nb5x ][Nb1−4x 4x ]O3 . Contrary
to the result specified by Abrahams and Marsh [11], Donnerberg et al. [12] proved that
the niobium vacancy model is unfavorable, by comparing the energetic aspect of the defect
structure models. A comparative study between the densities of these models allowed Iyi
et al. [13] to reject the oxygen vacancy model. In accord with this result, we have only
used the lithium and tantalate vacancy models to describe the defect structure effect on
the conductivity.
A theory of ferroelectric phase transition in the crystal LiNbO3 has been proposed
to understand and predict the properties of this crystal [14]. In this theory, the solution of
the dynamic problem of the crystal planes system exhibits the existence of a “soft mode”
at the ferroelectric transition. These recent studies provide a useful understanding of the
phase transition dynamic behavior and the defect structure of LiNbO3 .
The approach of this theory has been applied to the defect structure analysis of the
lithium niobium single crystals [15]. It allows for a comparison between the calculated values
and the experimental data of the Curie temperature for the three vacancy models. The
present paper is dedicated to the experimental and analytical study of the ionic conductivity
of the nonstoichiometric LiTaO3 . A comparative study between the calculated values and
those measured is presented, taking into account the temperature and composition effects
VOL. 42 N. MASAIF, S. ELHAMD, et al. 651

on the conductivity.

II. EXPERIMENTAL RESULTS

The experimental technique and the preparation of the solid solutions have been
described elsewhere [7]. By the conventional mixed oxide method, the nonstoichiometric
LiTaO3 samples were prepared from Li2 CO3 and Ta2 O5 . The powders were isostatically
pressed at 2500 bars, to give pellets of 13 mm in diameter and 1 mm in thickness, and were
sintered at 1500 K for 4h with a heating rate of 100 K/h. The ceramics were characterised
by X-ray diffraction, and the microstructure of these samples was examined using scanning
electron microscopy (SEM). The chemical composition of the samples was obtained using
inductively coupled plasma-atomic emission spectroscopy (ICP-AES).

TABLE I: Chemical formula obtained by the analysis, and the proposed formula according to the
(b) and (c) models with different Li/Ta rations.
Experimental formula Proposed formula
Li/Ta
(b) (c)
1 LiTaO3 [Li1.0 ][Ta1.0 ]O3 [Li1.0 ][Ta1.0 ]O3
0.976 Li0.98 Ta1.005 O3 [Li0.98 Ta0.02 ][Ta0.985 0.015 ]O3 [Li0.98 Ta0.005 0.015 ][Ta]O3
0.95 Li0.958 Ta1.008 O3 [Li0.958 Ta0.042 ][Ta0.966 0.034 ]O3 [Li0.958 Ta0.008 0.034 ][Ta]O3
0.88 Li0.90 Ta1.02 O3 [Li0.90 Ta0.10 ][Ta0.92 0.08 ]O3 [Li0.90 Ta0.02 0.08 ][Ta]O3

The formulae obtained, which was estimated to be about 0.8% for Li and 0.1% for Ta,
are reported in Table I. We denote by (b) and (c) the tantalate and lithium vacancy models.
Among the electrical measurements investigated, we were especially interested in finding
the evolution of the conductivity σ by complex impedance spectroscopy. The isothermal
measurements were carried out between 600 and 1200 K, with temperature steps of 50 or 100
K, except in the Tc range where the steps were reduced to 20 K. The range of measuring
frequencies was 1 Hz–1 MHz for the ceramic samples. A Pt(Rh 10%)/Pt thermocouple
located near the sample was connected to a Keithley 2000 electrometer to measure the
temperatures; reported temperatures are estimated as being accurate to ±2K.
The ionic conductivity was determined for all compositions from Cole-Cole diagrams,
and the evolution vs temperature was studied as a function of the Li/Ta ratio [7]. The mea-
sured conductivities for nonstoichiometric LiTaO3 with different compositions are shown
in Fig. 1 in the form of Arrhenuis plots for a range of temperature from 830 to 1250 K.
This figure illustrates the logarithmetical conductivity at different compositions as a func-
tion of the reciprocal temperature. In pure LiTaO3 , the conductivity ranged from 1.1 10−6
Ω−1 cm−1 to 31.1 10−6 Ω−1 cm−1 for a temperature varying from 873 K to 1073 K. The con-
ductivity represented in Fig. 1 increased with increasing composition. The same analysis
technique based on an Arrhenuis plot was also applied to estimate the activation energy
(∆E) of the materials (Table II). The data obtained were interpreted in the framework of
652 STUDY OF THE DEFECT STRUCTURE EFFECT ON . . . VOL. 42

σ Ω

&#
3
FIG. 1: Evolution of log(σ) as a function of 10 /T in the 830–1250 K range for pure LiTaO3 with
different compositions x. The full circles, squares, triangles, and crosses indicate the experimental
data for the different Li/Ta ratios, and the solid lines are the linear fits to the data.

a new theory of the defect structure in LiTaO3 proposed by Safaryan [14].

TABLE II: Values of the activation energy and log(σ0 ) of the LiTaO3 for the different Li/Ta ratios
studied.

Li/Ta Activation energy (eV) log(σ0 ) (Ω−1 cm−1 )


1 1.16 0.88
0.98 1.1.4 1.1.8
0.95 0.91 0.11
0.88 0.86 0.19

III. THEORETICAL APPROACH

In this work, we suppose that the ceramic samples of LiTaO3 consist of a single crystal
on the basis of their structure. Such an assumption is based on the experimental fact that
the ferroelectric phase transition occurred in the ceramic samples which are formed by the
parallel planes along the polar “c” axis. In Fig. 2, distances between planes (Li, Ta, and O
at T =0 K) are denoted as follows: RO-O (a= 2.30Å), RLi-O (R20 = 0.601 Å), RTa-O (R10 =
0.954 Å), RLi-Ta (R12 = a − R10 − R20 ) [16].
Safaryan’s new approach on the ferroelectric transition in the crystal LiNbO3 is newly
tested in the LiTaO3 compound in order to discuss the main role of the nonstoichiometric
VOL. 42 N. MASAIF, S. ELHAMD, et al. 653

+
!

*;

FIG. 2: Different planes in an elementary cell of crystal LiTaO3 .

compositions. We avoid the detail of the theory of ferroelectric transition in the crystal
LiTaO3 , which is similar to that of reference [14], and only report the useful expressions in
the appendix. At 0 K, the soft mode frequency, ω 2 , is proportional to the Curie temperature.
Substituting ω 2 by expression (A.10) we obtain the following relation which allows for the
calculation of the Curie temperature of the defect structure:
  ∗ 
T∗ M1∗ + M2∗ + M0∗ M1 M2 M0 P1 P2
= . (1)
T M1 + M2 + M0 M1 M2 M0∗ ∗ ∗ P1 P2∗
The element X represents the exact stoichiometric composition and X∗ is the nonstoichio-
metric composition.
The electrical conductivity can be described by the law of Arrhenuis,

σ = σ0 exp (−∆E/kB T ) , (2)

where σ0 is the pre-exponential factor, ∆E the activation energy, kB the Boltzmann con-
stant, and T is the temperature (K). Note that the presence of nonstoichiometry in the form
of lattice vacancies profoundly alters the electrical conductivity, consequently we must re-
define the law of Arrhenuis to take into account of this effect.
We introduce the conductivity, σ ∗ , to explain the role of the defect structure in
nonstoichiometric lithium tantalate compound such that:

σ ∗ = σ0∗ exp (−∆E ∗ /kB T ∗ ) . (3)

Since the expression for T ∗ is determinate at the Curie temperature point, this relation
seems correct for calculating the conductivity about T ∗ .
Taking into account the experimental results of σ0 and ∆E for each composition x,
we have used two possible models (the tantalate and lithium vacancy models) to determine
the conductivities according to Eq. (3).
We represent these models, described previously, in the following condensed form:
[Liα1 Taα2 α3 ] [Taβ1 β2 ][O3 ], and K ∗ = gK with K = M , q, and g = α, β. Here α and β
allow the two models to be identified as follows:
654 STUDY OF THE DEFECT STRUCTURE EFFECT ON . . . VOL. 42

i) The tantalate vacancy model corresponds to α1 = 1 − 5x, α2 = 5x, β1 = 1 − 4x,


β2 = 4x, and K2∗ = α1 K2 + α2 K1 , K1∗ = β1 K1 , K0∗ = K0 .

ii) The lithium vacancy model corresponds to α1 = 1 − 5x, α2 = x, α3 = 4x, β1 = 1,


and K2∗ = α1 K2 + α2 K1 , K1∗ = β1 K1 , K0∗ = K0 .

In this representation α = β = 0 signified that ions and vacancies are absent in these non-
stoichiometric models. In order to provide an adequate description of the defect structure
in nonstoichiometric lithium tantalate, we have analytically performed the calculations of
the Curie temperature as a function of the composition x, which we illustrate in Table III.
Estimates of T ∗ for LiTaO3 have been obtained using the following values of the charges
and masses of the ions: q0 = 2, q1 = 5, q2 = 1, M0 = 48, M1 =180.95, and M2 = 6.94.

TABLE III: The Curie temperature T ∗ as a function of the nonstoichiometric composition x. (b)
and (c) represent the vacancy models.

(b) (c)
P1∗ (1 − 17.88x)P1 P1
1 + 78.01x 1 + 10.86x
P2∗ P2 P2
(1 − 4x)(1 + 20x)(1 + 125.38x) 1 + 21.07x
(1 − 17.88x)(1 + 0.62x)(1 + 20x) 1 + 0.62x
T∗ T T
(1 + 78.01x) 1 + 10.86x

IV. RESULTS

In principle, one could test these models by comparing the experimental results with
the calculated values of the conductivities from this theoretical approach. To explain the
defect structure effect on the conductivity, we have applied the results of the procedure to
a discussion of the interpretation of the experimental data.
By a linear fit of the experimental values and the calculation of the slope of the log σ
curve in the form of the Arrhenuis plot (Fig. 1), we obtained σ0 and ∆E for each composi-
tion x. The estimated activation energy with different Li/Ta ratios from Eq. (2) is reported
in Table II. The calculated and experimental values of σ ∗ for various nonstoichiometric
compositions are illustrated in Figs. 3.1–3.4. For each composition x, we note that the con-
ductivity increases with increasing temperature. Comparing the measured conductivity for
nonstoichiometric LiTaO3 with the two vacancy models (the lithium and tantalate vacancy
models), the results demonstrate clearly that the plot of the lithium vacancy model conduc-
tivity is close to that measured. Hence, we suggest that the lithium vacancies contribute
to the conductivity of this material.
To describe the behaviour of the activation energy, we have fitted the experimental
values of log(σ) for each composition x (Table II). The activation energy dependence on
VOL. 42 N. MASAIF, S. ELHAMD, et al. 655

B H

E
+
E
+

+
&Dσ EDΩ
+
&Dσ EDΩ

L
L

" +
, B D= E " +
, B DK= E

(1) (2)

B H, >>
E

E
+

+
+
&Dσ EDΩ

+
&Dσ EDΩ
L

(3) (4)

FIG. 3: Logarithmical conductivity of LiTaO3 solid solutions as a function of the reciprocal tem-
perature with different compositions x: (1) Li/Ta=1, (2) Li/Ta=0.98, (3) Li/Ta=0.95, and (4)
Li/Ta=0.88. The full circles, squares, triangles, and crosses are the experimental data and the solid
lines represent the linear fits to the data. The dashed and dotted lines indicate the linear fits of the
(b) and (c) models, respectively.

the composition x is reported in Fig. 4. In this case, the curve indicates an abrupt fall
in the activation energy. The composition dependence of the conductivity for different
temperatures is shown in Fig 5. The increase in the conductivity can be interpreted in
terms of competition between Li+ and Ta5+ ions.
656 STUDY OF THE DEFECT STRUCTURE EFFECT ON . . . VOL. 42

",

,
∆Ε D ME
,*

,,

, ?*

, ?,

, >*

, >,

, )*

, ,,, , ,,* ,, , ,, * ,, ,
2

FIG. 4: Activation energy as a function of the composition x.


E
+
+
&Dσ EDΩ
L

K= ,*"K= ,,,K=
?* K= ?,?K= >),K=

FIG. 5: Variation in the logarithmical conductivity versus the composition x for different values of
temperatures.

V. DISCUSSION

The influence of the nonstoichiometric composition on the ionic conductivity is quan-


titatively analyzed on the basis of a simplified structural description of lithium tantalate.
In these analyses the estimation of the conductivities of the lithium and tantalate vacancy
models shows that the lithium model approaches the measured values well. It must be
noted that the calculated and experimental conductivities have the same activation energy.
It is clear that the comparison between the vacancy models and the experimental values
of the conductivities allows for a prediction of the conduction mechanism. The results in
VOL. 42 N. MASAIF, S. ELHAMD, et al. 657

Figs. 3.1–3.4 illustrate that the conductivity is due to the lithium vacancies. So, the lithium
vacancy model is the best for describing the conductivity in nonstoichiometric LiTaO3 solid
solutions. The different lines show that no anomaly, such as a slope change, is manifested
at the Curie temperature. Thus, the conduction mechanism seems to be the same in both
the ferroelectric and paraelectric phases.
The effect of the lithium defect is predominant over of that of the tantalate vacancies
for the conductivity of LiTaO3 . However, the number of Li vacancies connected with the
composition Li/Ta ratio is mainly responsible for the increase found in the conductivity.
Fig. 4 illustrates the activation energy at different compositions. According to the results
obtained on the conductivity, the activation energy is not affected by the temperature but
undergoes an abrupt decrease from 1.14 eV to 0.86 eV when the composition is increased
above x=0.004 (R=0.98). It seems that this phenomena is due to the increase in the mass
difference between Li and Ta cations (the mass effect).
Fig. 5 shows the plots of the conductivity as a function of the composition x for
various temperatures. In this figure, we note that the conductivity rapidly increased in the
low composition region. The increase of the conductivity with increasing composition x
results from an increasing Ta density and a decreasing Li density. We have therefore used
this theory for a consistent study of vacancy defects in nonstoichiometric lithium tantalate
using the available vacancy models.

VI. CONCLUSION

We have presented the conductivity as a function of the temperature and composition


in nonstoichiometric LiTaO3 ceramics. In this simple system, the theory of the ferroelec-
tric phase transition gives a good quantitative description of the experimental results. A
comparative study, between the vacancy models and the experimental values of the conduc-
tivities, shows that the lithium vacancy model is the defect model which can be proposed
for describing the defect structure in nonstoichiometric lithium tantalate.

APPENDIX

The energy of the electrostatic interaction of two electrically charged lines is,
2l
W = e2 δi δj b2 ln , (A.1)
Rij

where δi = Nlbi qi with b, l, Rij , qi , and Ni being the cell parameter, the line length, the
distance between lines, the charge of ions, and their number respectively.
We deduce the full energy of interaction of the planes such that:

e2 qi qj 2l Bij
Uij = − ln + 2 , (A.2)
b Rij Rij
658 STUDY OF THE DEFECT STRUCTURE EFFECT ON . . . VOL. 42

us vs ξs

≡ 3O 2- ≡ Ta5+ ≡ Li+

FIG. 6: Displacement of Li+ , Ta5+ and 3O2− ions in a linear lattice.

e2 q q 0 )2 with R0 as the equilibrium distance between planes and n is the


where Bij = bni j (Rij ij
index of non-electrostatic repulsion for interacting planes.
We can expand the potential energy in terms of a plane’s displacement from equi-
librium for the positions, xij = Rij − Rij 0 . 13
The energy, developed up to the second order,
is
!
e2 qi qi 1 2l e2 qi qj n x2ij
Uij = − ln 0 + 0 )2 2 . (A.3)
b n Rij b(Rij

e2 q i q j
The coefficient Cij = b(R0ij )2
n describes the interaction between the Li and O planes (as
well as for Ta and O).
However, the interaction between the Li and Ta will deduct from the total energy:

e2 qi qj 2l Bij
Uij − ln 0 − . (A.4)
b Rij Rij

Finally, for the LiTaO3 crystal the Cij coefficients are

q0 q2 e2
CLi-O ≡ C20 = −3 2 n,
bR20
q0 q1 e2
CTa-O ≡ C10 = −3 2 n, (A.5)
bR10
q1 q2 e2
CTi-Ta ≡ C21 = 2 n,
bR21

where q1 , q2 , and q0 are, respectively, the electric charges of the Ta5+ , Li+ , and O2− ions.
The value of the constant n is achieved at n = 1 for interacting planes.
To solve the dynamic problem, we reduced the structure of the charged planes to a
vibrating linear lattice system (Fig. 6). The displacements of the three ions are indicated by
vs (Li+ ), us (Nb5+ ), and ξs (O2− ). Then the system is described by the differential equations

M1 üs = C10 (ξs+1 − us ) + C21 (vs − us ) ,


M2 v̈s = C21 (us − vs ) + C20 (ξs − us ) , (A.6)
M0 ξ̈s = C20 (vs − ξs ) + C10 (vs−1 − ξs ) ,
VOL. 42 N. MASAIF, S. ELHAMD, et al. 659

where M1 , M2 , and M0 are the masses of elements Ta, Li, and 3O, respectively. The choice
of solutions in the form of plane waves,

gs = gei(ωt+kas ) with g = u, v, or ξ , (A.7)

leads to a system of linear equations. This system has a nontrivial solution. We put k = 0
in the determinant equation (∆ = 0) in order to obtain the fundamental frequencies of the
optical branches. Then we obtain the following equation,

ω 4 − Bω 2 + D = 0 ,

where
     
1 1 1 1 1 1
B = C10 + + C21 + + C20 + ,
M1 M0 M1 M2 M2 M0
(A.8)
M1 + M2 + M0
D= (C10 C20 + C10 C21 + C21 C20 ) .
M1 M2 M0
The two optical modes are given by
2 1 2
1/2
ω12 = 12 B ± 4B −D . (A.9)

For small parameter ((4D/B2)  1) we deduce

D ne2 M1 + M2 + M0 P1
ω22 = = , ω12 = B − ω22 ,
B b M1 M2 M0 P2
with
2 2 2
P1 = 3q0 R12 − q1 R20 − q2 R10 ,
(R20 R21 )2 
1 1

(R10 R21 )2

1 1

(R10 R20 )2

1 1

P2 = + + + − + .
q2 M1 M0 q1 M2 M0 q0 M1 M2
(A.10)

References

[1] M. Paul, M. Tabuchi, and A. R. West, Chem. Mater. 9, 3206 (1997).


[2] S. C. Abrahams and L. J. Bernstein, Phys. Chem. Solids, 28, 1685 (1967).
[3] Y. Fujino, H. Tsuya, and K. Sugibachi, Ferroelectrics 2, 113 (1971).
[4] A. Huanosta and A. R. West, J. App. Phys. 61, 5386 (1987).
[5] D. C. Sinclair and A. R. West, Phys. Rev. B 39, 13486 (1989).
[6] I. Tomeno and S. Matsumura, Phys. Rev. B 38, 606 (1988).
[7] F. Bennani and E. Husson, J. of European Cearamic Society 21, 847 (2001).
[8] H. Fay, W. J. Alford, and H. M. Dess, Appl. Phys. Lett. 12, 89 (1968)
[9] P. Lerner, C. Legras, and J. P. Dumas, J. Cryst. Growth, 3/4, 231 (1968).
[10] G. E. Peterson and A. Carnevale, J. Chem. Phys. 56, 4848 (1972).
660 STUDY OF THE DEFECT STRUCTURE EFFECT ON . . . VOL. 42

[11] C. S. Abrahams and P. Marsh, Acta Crystallogr. Sect. B 42, 61 (1986).


[12] H. Donnerberg, S. M. Tomlinson, C. R. A. Catlow, and O. F. Schirmer, Phys. Rev. B 40, 909
(1989).
[13] N. Iyi, K. Kitamura, F. Izumi, J. K. Yamamoto, T. Hayashi, H. Asano, and S. Kimura, J. of
Solid State Chem. 101, 340 (1992).
[14] F. P. Safaryan, Physics Letters, A 191, 255 (1999).
[15] F. P. Safaryan, R. S. Feigelson, and A. M. Petrosyan, J. Appl. Phys. 85, 8079 (1999).
[16] M. E. Lines and A. M. Glass, Principles and Application of ferroelectrics and related materials,
(Clarendon Press, Oxford, 1977).

You might also like