Materials Chemistry and Physics 124 (2010) 281–286

Contents lists available at ScienceDirect

Materials Chemistry and Physics

journal homepage: www.elsevier.com/locate/matchemphys

Hysteresis scaling relations in polycrystalline BaTiO3 bulk ceramics

N. Wongdamnern a,∗ , J. Tangsritragul b , A. Ngamjarurojana a , S. Ananta a , Y. Laosiritaworn a , R. Yimnirun c
a
Department of Physics and Materials Science, Faculty of Science, Chiang Mai University, 239 Huay Kaew Road, Muang, Chiang Mai 50200, Thailand
b
Department of Physics, Faculty of Science and Technology, Thammasat University, Pathumthani 12121, Thailand
c
School of Physics, Institute of Science, Suranaree University of Technology, Nakhon Ratchasima 30000, Thailand

a r t i c l e i n f o a b s t r a c t

Article history: The scaling behaviors of the dynamic ferroelectric hysteresis of polycrystalline BaTiO3 bulk ceramics were
Received 1 May 2009 investigated. To enhance the accuracy of prediction, not only sets of the scaling relation of hysteresis area
Received in revised form 1 June 2010 A against frequency f and field amplitude E0 were established, but also were sets of scaling relation of
Accepted 16 June 2010
remnant polarization (Pr ) and coercive field (EC ) against frequency f and field amplitude E0 . All scaling
relations of hysteresis parameters, A, Pr , and EC , displayed obviously the inter-correlation between f- and
Keywords:
Hysteresis  field and above coercive field conditions. By first approximation, the
E0 -exponents in both sub-coercive
scaling relations took a form of A ∝ f −0.39 E01.06 , Pr ∝ f −0.18 E00.47 , and EC ∝ f −0.33 E00.46 above the coercive
Scaling  
BaTiO3 ceramics field condition; whereas the scaling relations in the form of A ∝ f −0.55 E03.40 , Pr ∝ f −0.43 E01.73 , and EC ∝
Polycrystalline
f −0.27 E01.35 were obtained under a sub-coercive field condition. The frequency observation range affected
directly the scaling exponents. Over the same frequency range, the scaling relations obtained in this
study are generally comparable to previously reported scaling behavior in BaTiO3 single crystals, as well
as in other polycrystalline bulk ceramics, suggesting that materials (both single and polycrystalline) with
similar domain switching mechanisms should have comparable dynamic hysteresis and scaling behavior.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction
Many reports have been focused on the scaling behavior of the
dynamic hysteresis in ferromagnetic and ferroelectric materials
because the dynamic hysteresis, i.e., hysteresis area A as a func-
tion of the field amplitude E0 and frequency f has become important
consideration in many applications, whose performance is related
to the signal amplitude and frequency. However, some discrepan-
cies between theoretical and experimental results still exist [1–10].
Many theoretical studies have been focused on the power-law scal-
ing
  ˇ
A ∝ f ˛ E0
(where ˛ and ˇ are exponents that depend on the dimensional-
ity and symmetry of the system) of hysteresis curves in spin and
polarization systems [1–5]. Earlier investigations have reported the
scaling relations in thin-films, bulk ceramics, and liquid crystals of
some ferroelectric materials, such as Pb(Zr,Ti)O3 (or PZT), Bi4 Ti3 O12 ,
SrBi2 Ta2 O9 , PMN-PT film, soft and hard PZT bulk ceramics and
MIP0806 liquid crystal [1,2,4,10–19]. It is interesting to observe
that scaling behavior of different materials derived from theoret-
∗ Corresponding author.
E-mail address: natthapongwongdamnern@hotmail.com (N. Wongdamnern).
ical model, thin-film and bulk ceramics are generally comparable
with slight difference in exponents ˛ and ˇ, suggesting that mate-
rials with similar domain structures should have very comparable
dynamic hysteresis and scaling behavior, as previously proposed
in our investigations [13–15,20]. More interestingly, the scaling
behavior of BaTiO3 single crystals was recently reported by our
group [21]. A direct evidence of the influence of the domain states
contributing to the dynamic hysteresis behavior was indicated in
the study. The scaling exponents ˛ and ˇ in BaTiO3 single crystals
are comparable to those of other more complex materials reported
previously. It could be stated that the dynamic hysteresis (hence the
scaling behavior) is mainly controlled by available domain states
(1) and polarization switching mechanism [21]. It should be noted
that BaTiO3 single crystals are mentioned as single-domain-like
state material [21]. It is, therefore, of interest to investigate the
dynamic hysteresis and scaling behavior in BaTiO3 bulk ceramic
which is polycrystalline material with randomly oriented multi-
domain states, to support the hypothesis that materials with similar
domain states should have very comparable dynamic hysteresis
and scaling behavior. More importantly, all of previous studies
[10–20] have proposed only scaling relations in term of hysteresis
area A which is one of many basic parameters describing hystere-
sis behavior. The other equally important parameters are remnant
polarization Pr and coercive field EC . It is then intuitively expected
that the accuracy and reliability of prediction of the hysteresis

0254-0584/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.matchemphys.2010.06.032
282 N. Wongdamnern et al. / Materials Chemistry and Physics 124 (2010) 281–286

already established in previous investigations [8,13,14], the scaling

behavior can be divided into two distinct regimes; that for the
saturated loops (E0 > EC ) and the other for the minor loops (E0 < EC ).
The same can be applied for the case of the BaTiO3 bulk ceramics,
as seen in an inset of Fig. 1(a) that there are two distinctly different
trends with the applied field amplitude. From Fig. 1(a) inset, it is
reasonable to estimate EC = 3.33 kV cm−1 as a critical turning point
for BaTiO3 bulk ceramics (as opposed to 1.67 kV cm−1 for BaTiO3
single crystals) [21].
Thus, to obtain the scaling relations for the polycrystalline
BaTiO3 bulk ceramics, one can fit the data with Eq. (1) where expo-
nents ˛ and ˇ are determined directly from the experimental data.
By plotting A against f at fixed E0 , one obtains the exponent ˛.
On the other hand, the exponent ˇ can be obtained from plot-
ting A against E0 at fixed f [13,14]. In conjunction with the least
square-fitting method, the methodology outlined above is applied
to obtain the exponents ˛ and ˇ from the saturated loop (E0 > EC )
and the minor loop (E0 < EC ) data. Moreover, the scaling relations
of the other hysteresis parameters: remnant polarization Pr and
coercive field EC , also are investigated in forms of
ˇ
Pr ∝ f ˛ E0 (2)
ˇ
EC ∝ f ˛ E0 (3)

where ˛ and ˇ are exponents that depend on the dimensionality

and symmetry of the system. Based on the same methodology, the
exponents ˛ and ˇ for Pr -scaling relations are extracted by plotting

Fig. 1. P–E hysteresis profiles for polycrystalline BaTiO3 bulk ceramics (a) at various
E0 fixed f = 100 Hz, and (b) at various f fixed E0 = 15 kV cm−1 . Insets show a variations
of A with E0 and f.

behavior can be improved if the scaling relations in terms of Pr and

EC are also observed. Consequently, the scaling relations in forms
of hysteresis area A, remnant polarization Pr and coercive field
EC against frequency f and electric field E0 will be determined and
reported in this study.

2. Experimental procedure

The dynamic hysteresis (P–E) loops of tetragonal BaTiO3 bulk ceramics

(TC = 124.5 ◦ C, εr at 100 kHz = 1143, tan ı at 100 kHz = 0.0477) prepared by a con-
ventional mixed-oxide method with a diameter of 8 mm and thickness of 1.2 mm
were characterized at room temperature (25 ◦ C) using a modified Sawyer–Tower
circuit with f covering from 1 to 100 Hz and E0 from 0 to 15 kV cm−1 . The electric
field was applied to a sample by a high voltage AC amplifier (Trek 610D) with the
input sinusoidal signal from a function generator (HP 3310A). The P–E loops were
recorded by a digital oscilloscope (HP 54645A, 100 MHz). Each loop was obtained
after 20 sampling cycles to average out the noise deformation [13–15].

3. Results and discussion

The P–E hysteresis profiles at various E0 but fixed f (100 Hz),

and at various f but fixed E0 (15.0 kV cm−1 ) are shown in Fig. 1. As
shown in Fig. 1(a), for applied field lower than the coercive field,
the loop does not saturate. With further increase in E0 beyond
EC , the loop area A, remnant polarization (Pr ), and coercive
field (EC ) increase until well-saturated loop is achieved. Similar
Fig. 2. (a) Variation of f-exponent ˛ with E0 , and (b) variation of E0 -exponent ˇ
observations have been reported in ferroelectric single crystals,
with f in sub-coercive field (0.17 ≤ E0 ≤ 3.33 kV cm−1 ) and beyond coercive field
thin-films, and bulk ceramics [4,6,8,10,13,14,21]. The A, Pr , and (3.33 ≤ E0 ≤ 15 kV cm−1 ) conditions. Dotted lines indicate linear relations fitted to
EC decrease with an increase of frequency, as shown in Fig. 1(b). As data in each regime.
 N. Wongdamnern et al. / Materials Chemistry and Physics 124 (2010) 281–286 283

Fig. 3. Scaling relations for polycrystalline BaTiO3 bulk ceramics with different observation f-ranges, (a and b) scaling relations of A against f −0.55 E03.40 and A against
f −0.39 E01.06 for minor loop and saturated loop, respectively, with observed f-range of 1–100 Hz, (c and d) scaling relations of A against f −0.32 E03.68 and A against f −0.25 E00.82
for minor loop and saturated loop, respectively, with observed f-range of 20–100 Hz. The dotted lines represent the fitting.

Pr against f at fixed E0 to obtain exponent ˛ and plotting Pr against
E0 at fixed f to obtain exponent ˇ for both saturated and minor
loops. Similarly, the exponents ˛ and ˇ for EC -scaling relations are
also extracted with the same methodology. Fig. 2(a) and (b) shows a
variation of exponent ˛ with E0 , and a variation of exponent ˇ with
f, respectively, for all A, Pr - and EC -scaling relations. Evidently, the
exponent ˇ of all scaling relations, A, Pr and EC , depends slightly on
the frequency f, while the exponent ˛ depends significantly on the
field amplitude E0 with E0 = EC being the critical turning point. The
variation of E0 -exponents ˇ with f obtained here is similar to that in
BaTiO3 single crystals [21], while the f-exponent ˛ variation with E0
shows a little discrepancy with that of BaTiO3 single crystals [21].
From Fig. 2(a), the linear relations can be obtained for the vari-
ation of f-exponent ˛ with E0 . With E0 < EC (3.33 kV cm−1 ), strong
field-dependent relations ˛ = (−5.93 × 10−1 ) + (1.12 × 10−1 )E0
for A scaling, ˛ = (−5.24 × 10−1 ) + (2.83 × 10−2 )E0 for Pr
scaling, and ˛ = (−3.01 × 10−1 ) + (5.36 × 10−2 )E0 for EC
scaling are obtained, while weaker field-dependent rela-
tions ˛ = (−1.84 × 10−1 ) + (−2.23 × 10−2 )E0 for A scaling,
˛ = (−1.46 × 10−1 ) + (−3.94 × 10−3 )E0 for Pr scaling, and
˛ = (−8.51 × 10−2 ) + (−2.66 × 10−2 )E0 for EC -scaling relations
are attained at E0 > EC . The slopes of the linear relations are notice-
ably different from BaTiO3 single crystals [21]. Furthermore, as
displayed in Fig. 2(b), the E0 -exponents ˇ are slightly dependent on
f for both regimes which can be explained by complex exponential
relations that are not shown here. Notably, E0 -exponents ˇ of
each parameter relation are approaching a certain value at low
frequency. These observations are very similar to those of BaTiO3
single crystals [21]. Nonetheless, further experiments capable of
very frequency region will be needed to arrive at any conclusion.
In addition, a variation of ˛ and ˇ relations in E0 < EC and E0 > EC
regimes may be caused by the difference in polarization switching
mechanisms, since a reversible 180◦ domain switching mainly
controls the dynamic hysteresis in sub-coercive field condition
[13,14,22–29], while an irreversible non-180◦ domain switch-
ing is primarily responsible for above coercive field condition
[13,14,24–30]. More importantly, Fig. 2(a) and (b) clearly reveals
that the f-exponent ˛ is a function of E0 and the E0 -exponent
ˇ is a function of f. The scaling relations can then be written in
  ˇ(f )
an inter-correlation form as A ∝ f ˛(E0 ) E0 for both E0 < EC and
E0 > EC regimes. One can generally predict hysteresis behavior and
in sub-coercive field and above coercive field from the scaling
relations established in Fig. 2.
For a better comparison, approximations of exponent values of
both f-exponent ˛ and E0 -exponent ˇ were obtained. At E0 < EC ,
the exponents ˛ are averaged to a mean value of −0.55 and ˇ of
3.40. As plotted in Fig. 3(a), it is revealed that the minor loop data
can then be fitted reasonably well (R2 ∼ 0.97), within the measured
uncertainty, by
 
A ∝ f −0.55 E03.40 for E0 < EC (4)
Additionally, for the case of E0 > EC , the exponents ˇ = 1.06 and
˛ = −0.39 are attained from approximation. Therefore, the scaling
284 N. Wongdamnern et al. / Materials Chemistry and Physics 124 (2010) 281–286

relation for the saturated loops of the polycrystalline BaTiO3 bulk Table 1
Scaling exponents in different materials. Refer to Eq. (1) for exponents ˛ and ˇ.
ceramics takes the form of
  Systems ˛ ˇ References
A ∝ f −0.39 E01.06 for E0 > EC (5)
Nd-doped Bi4 Ti3 O12 thin-film −0.33 2 [12]
SrBi2 Ta9 O4 thin-film −0.33 3 [10]
which can be fitted well (R2 ∼ 0.96) with the saturated loop data, as Pb(Zr,Ti)O3 thin-film −0.33 1 [31]
plotted in Fig. 3(b). PbTiO3 /polymer composite −1 2 [32]

In the saturated loop case, it is very interesting to observe that Soft PZT bulk ceramic
the E0 -exponent ˇ value for the polycrystalline bulk ceramics is Saturated loops −0.25 1 [13]
Minor loops −0.33 3 [13]
very similar to that for the single crystals [21]. On the other hand,
the f-exponent ˛ is significantly higher, about two times in abso- Hard PZT bulk ceramic
lute value, than that of the BaTiO3 single crystals. The difference Saturated loops −0.28 0.89 [14]
Minor loops −0.43 3.19 [14]
may be results of a near saturation of polarization in the single
crystal under the applied field beyond the coercive field, and of PZT–PZN bulk ceramic (minor loop)
the depolarizing effects in polycrystalline bulk ceramics acting as 0.9PZT–0.1PZN −0.42 3.65 [20]
0.8PZT–0.2PZN −0.36 3.32 [20]
a buffer to polarization-reversal mechanism resulting in slower 0.7PZT–0.3PZN −0.35 3.47 [20]
domain switching [13–15]. 0.6PZT–0.4PZN −0.36 4.03 [20]
For the minor loop scaling relation, the E0 -exponent ˇ (3.40) in 0.5PZT–0.5PZN −0.34 3.68 [20]
the BaTiO3 bulk ceramics is approximately the same as in BaTiO3 Na0.5 Bi0.5 TiO3 ceramic
single crystals [21]. As explained in previous investigations [13,15], Minor loops −0.39 2.63
this is attributed to the fact that the main polarization orientation
{0 0 1}-BaTiO3 single crystals
mechanism in thin-films and in sub-coercive field condition for Saturated loops −0.195 0.95 [21]
bulk ceramics is likely from the reversible 180◦ domain rotation Minor loops (1.667)E0 − 2.804 4.157 [21]
(which also occurs at much faster rate than the irreversible pro- BaTiO3 bulk ceramic (20–100 Hz)
cess), as the irreversible non-180◦ domain switching is normally Saturated loops −0.25 0.82 This work
accompanied with mechanical strain, it occurs at higher E-field than Minor loops −0.32 3.68 This work
reversible 180◦ domain rotation does. Therefore, under low E-fields, BaTiO3 bulk ceramic (1–100 Hz)
one would expect the reversible 180◦ domain rotation to occur first. Saturated loops −0.39 1.06 This work
This explains why the scaling behavior of the bulk ceramics at low Minor loops −0.55 3.40 This work
E-fields is similar to that of thin-films and why the response to E0 Pr -scaling relations of BaTiO3 bulk ceramic (1–100 Hz)
at sub-coercive field occurs at much faster response (ˇ ∼ 2–4) than Saturated loops −0.18 0.47 This work
that at higher field amplitude (ˇ ∼ 1). However, even ˇ = 3.40 can Minor loops −0.43 1.73 This work
be roughly considered as the same as that of BaTiO3 single crys- EC -scaling relations of BaTiO3 bulk ceramic (1–100 Hz)
tal, but these values still have some discrepancy. This may be an Saturated loops −0.33 0.46 This work
influence of many depolarizing effects [13], arisen from domain Minor loops −0.27 1.35 This work
walls, boundaries, space charge, immobile defects, etc., which may
restrain the external field. Consequently, the energy barrier is very
hysteresis parameters: Pr and EC , in the forms of Eqs. (2) and (3)
much higher, which leads to slower polarization-flip kinetics [21].
were also established to provide an additional dimension to the
Therefore, a lower exponent for the E0 term is expected in poly-
hysteresis behavior prediction through scaling relations. As seen in
crystalline BaTiO3 bulk ceramics.
Fig. 1, decaying of Pr and EC with f can be observed. Both parame-
More interestingly, when compared with previous studies in
ters increase to saturation with E0 . As the frequency of the applied
bulk ceramic materials such as commercial soft and hard PZTs,
field increases, it is expected for Pr to decrease because some of
and various PZT–PZN compositions [13,14,20], the exponents ˇ
the domains cannot follow the applied field. Hence, it takes lower
obtained for both regimes are comparable to those of previously
electric field to force the polarizations to become zero, so lower
studied systems. On the contrary, the exponents ˛ obtained in
coercive field is observed [21].
this study are noticeably different from those of previous ones
Similar to the case of hysteresis area A, remnant polarization Pr
for both regimes, as shown in Table 1. The possible cause may
and coercive field EC also show the inter-correlation of f-exponent
be the difference in measuring frequency range. Most of previ-
˛ with E0 and E0 -exponent ˇ with f, as displayed in Fig. 2(a) and
ous investigations observed and established scaling relations in
(b). Through an approximation, the Pr -scaling relations for minor
frequency range of 20–100 Hz, while this study scaling relation
(E0 < EC ) and saturated (E0 > EC ) loops take a form of Pr ∝ f −0.43 E01.73
was performed in frequency range of 1–100 Hz. The different fre-
and Pr ∝ f −0.18 E00.47 , respectively, as plotted in Fig. 4(a) and (b).
quency ranges may cause different exponents ˛. Hence, to confirm
this hypothesis, the hysteresis behavior and scaling relations of the For EC , the scaling relation takes the form of EC ∝ f −0.27 E01.35 for
polycrystalline BaTiO3 bulk ceramic were also examined in fre- the minor loop, while EC ∝ f −0.33 E00.46 is obtained for the saturated
quency range of 20–100 Hz. Based on the same methodology of loop, as plotted in Fig. 4(c) and (d). Clearly, f-exponents ˛ of Pr -
exponent extraction, the exponents ˛ and ˇ in frequency range of and EC -scaling relations for saturated loop in this study, which
20–100 Hz are obtained for both minor and saturated loops. Follow- are −0.18 and −0.33, are also about two times higher in absolute
ingEq. value than those of BaTiO3 single crystal (−0.079 and −0.164) [21].
 (1), the scaling relation
  in minor loop (E0 < EC ) takes the form
of A ∝ f 0.32 E03.68 while A ∝ f −0.25 E00.82 is the form of saturated This is because, as mentioned previously, the depolarizing effects
loop (E0 > EC ), as seen in Fig. 3(c) and (d). As expected, the obtained in polycrystalline bulk ceramics acting as a buffer to polarization-
scaling relations are very similar to those of previous investigations, reversal mechanism resulting in slower domain switching. So the
as summarized in Table 1. Consequently, this strongly suggests that higher absolute values of f-exponents ˛ for both Pr - and EC -scaling
the scaling relations depend significantly on the frequency range of relations of polycrystalline BaTiO3 bulk ceramic are obtained.
observation. Since empirical approximation of the  hysteresis area A  can

In addition to the scaling relations of hysteresis area A against be estimated with (2Pr )(2EC ) [15,34], A ∝ f −0.70 E03.08 and A ∝
frequency f and electric field E0 , the scaling relations of the other f −0.51 E00.93 are available from the Pr - and EC -scaling relations
 N. Wongdamnern et al. / Materials Chemistry and Physics 124 (2010) 281–286
Fig. 4. Pr - and EC -scaling relations for polycrystalline BaTiO3 bulk ceramics with observed f-range of 1–100 Hz, (a and b) scaling relations of Pr against f −0.43 E01.73 and Pr
against f −0.18 E00.47 for minor loop and saturated loops, respectively, (c and d) scaling relations of EC against f −0.27 E01.35 and EC against f −0.33 E00.46 for minor loop and saturated
loops, respectively. The dotted lines represent the fitting.
obtained above for minor and saturated loops, respectively,
through this empirical formula the relations. When compared with
directly obtained scaling relations from experimental data in Eqs.
(4) and (5), the exponents of the empirical relations are lower, yet
comparable, than those obtained directly from the experiment in
both regimes. It is noted that the empirical relation (2Pr )(2EC ) is
clearly based on the square-shape loop. Unlike the BaTiO3 single
crystal case [21] where the above-mentioned relations agreed very
well, this empirical relation is clearly not suitable for estimation
of the hysteresis area of polycrystalline BaTiO3 bulk ceramics and
other materials which do not have square-shape-like loop.
4. Conclusion
The inter-correlated scaling relations in form of A, Pr and
EC against f and E0 for polycrystalline BaTiO3 bulk ceramics are
revealed. Sets of the scaling relation for the polycrystalline ceramics
are established based upon the coercive field. The scaling rela-
  ˇ(f ) ˇ(f )
tions take the form of A ∝ f ˛(E0 ) E0 , Pr ∝ f ˛(E0 ) E0 , and EC ∝
ˇ(f )
f ˛(E0 ) E0 for both E0 < EC and E0 > EC . Through a first approxima-
 
tion, the scaling relations A ∝ f −0.55 E03.40 , Pr ∝ f −0.43 E01.73 , and
 
EC ∝ f −0.27 E01.35 are obtained for E0 < EC and A ∝ f −0.39 E01.06 , Pr ∝
f −0.18 E00.47 , and
EC ∝ f −0.33 E00.46 are obtained for E0 > EC . Over the
same frequency range, the scaling relations obtained in this study
are comparable to those of BaTiO3 single crystals and the other
285
ferroelectric ceramics. Finally, these results further support the
hypothesis that materials with similar domain state and switch-
ing mechanisms should have comparable dynamic hysteresis and
scaling behavior.
Acknowledgments
Financial supports are provided by the Industry/University
Cooperative Research Center (I/UCRC) in HDD Component, the Fac-
ulty of Engineering, Khon Kaen University and National Electronics
and Computer Technology Center, National Science and Technology
Development Agency (NSTDA) and the Thailand Excellent Center
in Physics. N.W. would like to express his gratitude to the Devel-
opment and Promotion of Science and Technology talents project
(DPST) of the Institute for the Promotion of Teaching Science and
Technology (IPST), Faculty of Science, and Graduate School of Chi-
ang Mai University for financial support during his graduate studies.
References
[1] M. Rao, H.R. Krishnamurthy, R. Pandit, Phys. Rev. B 42 (1990) 856.
[2] M. Acharyya, B.K. Chakrabarti, Phys. Rev. B 52 (1995) 6560.
[3] J.-M. Liu, H.L.W. Chan, C.L. Choy, C.K. Ong, Phys. Rev. B 65 (2001) 014416.
[4] J.-M. Liu, H.L.W. Chan, C.L. Choy, Y.Y. Zhu, S.N. Zhu, Z.G. Liu, N.B. Ming, Appl.
Phys. Lett. 79 (2001) 236.
[5] M. Rao, R. Pandit, Phys. Rev. B 43 (1991) 3373.
[6] Q. Jiang, H.N. Yang, G.C. Wang, Phys. Rev. B 52 (1995) 14911.
[7] B. Pan, H. Yu, D. Wu, X.H. Zhou, J.-M. Liu, Appl. Phys. Lett. 83 (2003) 1406.
[8] Y.-H. Kim, J.-J. Kim, Phys. Rev. B 55 (1997) R11933.
286 N. Wongdamnern et al. / Materials Chemistry and Physics 124 (2010) 281–286
[9] J.-H. Park, C.-S. Kim, B.-C. Choi, B.K. Moon, J.H. Jeong, I.W. Kim, Appl. Phys. Lett.
83 (2003) 536.
[10] J.-M. Liu, H.L.W. Chan, C.L. Choy, Mater. Lett. 52 (2002) 213.
[11] J.-M. Liu, B. Pan, H. Yu, S.T. Zhang, J. Phys.: Condens. Matter 16 (2004)
1189.
[12] J.-M. Liu, B. Pan, K.F. Wang, H. Yu, Ceram. Int. 30 (2004) 1471.
[13] R. Yimnirun, Y. Laosiritaworn, S. Wongsaenmai, S. Ananta, Appl. Phys. Lett. 89
(2006) 162901.
[14] R. Yimnirun, R. Wongmaneerung, S. Wongsaenmai, A. Ngamjarurojana, S.
Ananta, Y. Laosiritaworn, Appl. Phys. Lett. 90 (2007) 112908.
[15] R. Yimnirun, R. Wongmaneerung, S. Wongsaenmai, A. Ngamjarurojana, S.
Ananta, Y. Laosiritaworn, Appl. Phys. Lett. 90 (2007) 112906.
[16] R. Yimnirun, N. Wongdamnern, N. Triamnak, M. Unruan, A. Ngamjarurojana, S.
Ananta, Y. Laosiritaworn, J. Appl. Phys. 104 (2008) 104103.
[17] X.F. Chen, X.L. Dong, N.B. Feng, H.C. Nie, F. Cao, G.S. Wang, Y. Gu, H.L. He, Solid
State Commun. 149 (2009) 663.
[18] J. Yang, X.J. Meng, M.R. Shen, C. Gao, J.L. Sun, J.H. Chu, Appl. Phys. Lett. 93 (2008)
092908.
[19] M. Surjya Sarathi Bhattacharyya, A. Rahman, B.K. Mukherjee, A. Chaudhuri,
Yoshizawa, Appl. Phys. Lett. 92 (2008) 122909.
[20] N. Wongdamnern, N. Triamnak, M. Unruan, K. Kanchiang, A. Ngamjarurojana,
S. Ananta, Y. Laosiritaworn, R. Yimnirun, Phys. Lett. A 374 (2010) 391.
[21] N. Wongdamnern, A. Ngamjarurojana, Y. Laosiritaworn, S. Ananta, R. Yimnirun,
J. Appl. Phys. 105 (2009) 044109.
[22] B.-K. Lai, I. Ponomareva, I.I. Naumov, I. Kornev, H. Fu, L. Bellaiche, G.S. Salamo,
Phys. Rev. Lett. 96 (2006) 137602.
[23] D. Ricinschi, M. Okuyama, Ferroelectrics 349 (2007) 111.
[24] T. Tsurumi, T. Sasaki, H. Kakemoto, T. Harigai, S. Wada, Jpn. J. Appl. Phys., Part
1 43 (2004) 7618.
[25] K. Lee, S. Baik, Annu. Rev. Mater. Res. 36 (2006) 81.
[26] R.C. Smith, C.L. Hom, J. Intell. Mater. Syst. Struct. 11 (2000) 62.
[27] C.-T. Sun, A. Achuthan, J. Am. Ceram. Soc. 87 (3) (2004) 395.
[28] X. Zhou, A. Chattopadhyay, J. Appl. Mech. 68 (2001) 270.
[29] A. Kholkin, Ferroelecrrics 238 (2008) 235.
[30] N. Uchida, T. Ikeda, Jpn. J. Appl. Phys. 4 (1965) 867.
[31] J.-M. Liu, H.P. Li, C.K. Ong, L.C. Lim, J. Appl. Phys. 86 (1999) 5198.
[32] B. Pan, Y. Yang, L.C. Yu, J.M. Liu, K. Li, Z.G. Liu, H.L.W. Chan, Mater Sci. Eng. B 99
(2003) 179.
[34] G.L. Yuan, J.-M. Liu, S.T. Zhang, D. Wu, Y.P. Wang, Z.G. Liu, H.L.W. Chan, C.L.
Choy, Appl. Phys. Lett. 84 (2004) 954.