Journal of the Korean Physical Society, Vol. 51, No. 2, August 2007, pp.

878881
Origin of Ultrahigh Dielectric Constants for Barium Titanate Nanoparticles
S. Wada,

T. Hoshina, K. Takizawa, M. Ohishi, H. Yasuno, H. Kakemoto and T. Tsurumi

Department of Metallurgy and Ceramics Science, Tokyo Institute of Technology,
2-12-1 Ookayama, Meguro, Tokyo 152-8552, Japan
C. Moriyoshi and Y. Kuroiwa
Hiroshima University, 2-3-1 Kagamiyama, Higashihiroshima, Hiroshima 739-8526, Japan
(Received 9 November 2006, in nal form 6 April 2007)
Barium titanate (BaTiO
3
) nanoparticles with various particle sizes from 17 to 1,000 nm were
prepared by using the 2-step thermal decomposition method of barium titanyl oxalate under various
degree of vacuum. Various characterizations revealed that these particles were impurity-free, defect-
free, dense BaTiO
3
nanoparticles. When the degree of vacuum was high (pressure of 150 Pa at 650

C), the dielectric constant of BaTiO

3
particles with a size of around 60 nm exhibited a maximum
of around 15,000. On the other hand, when the degree of vacuum was low (pressure of 400 Pa at 650

C), no dielectric maximum was observed. To explain this size dependence, we precisely investigated
a particle structure by using synchrotron radiation. As a result, the particles were always composed
of two layers, i.e., a surface cubic layer and a bulk tetragonal layer, and the thickness of the
surface cubic layer decreased with increasing degree of vacuum during the preparation of BaTiO
3
nanoparticles. Thus, we conrmed that the surface structure was an important factor in determining
the dielectric properties of BaTiO
3
nanoparticles.
PACS numbers: 77.22.Ch, 77.84.Dy, 77.84.Fa
Keywords: Size eect, Barium titanate, Nanoparticles, Dielectric properties
I. INTRODUCTION
Ferroelectric BaTiO
3
ne particles have been used as
raw materials for electronic devices, such as multilayered
ceramic capacitors (MLCC). With the miniaturization of
electronic devices, the down-sized MLCC have been de-
veloped. As a result, the thicknesses of the dielectric
layers in MLCC are expected to become less than 0.3
0.5 m within 10 years. Consequently, the particle size
of the BaTiO
3
raw materials will decrease to a few ten
nm. However, in ferroelectric ne particles, the ferro-
electricity is known to decreases with decreasing particle
and grain sizes and to disappear below certain critical
sizes; this is called the size eect in ferroelectrics [1
8]. Therefore, the size eect in the BaTiO
3
is one of the
most important phenomena for industry and science.
To date, some researchers have investigated the size ef-
fect of BaTiO
3
using particles [3,4,6]. Recently, Wada et
al. reported a unique particle-size dependence in which a
dielectric maximum of 15,000 was observed at 60 nm [9].
Moreover, Wada et al. also reported a size dependence
for the relative Slater mode frequency, and they found
that at the size (60 nm) with the minimum Slater mode
frequency, a dielectric maximum of 15,000 was clearly

E-mail: swada@ceram.titech.ac.jp; Fax: +81-3-5734-2514

observed [10, 11]. This result suggested that a high di-
electric constant of 15,000 should originate from a change
in the optic phonon behavior in the THz region by down-
sizing the BaTiO
3
particles. However, it should be noted
that this particle size of 60 nm is not a size-induced
phase transition size (20 30 nm) between cubic and
tetragonal. Thus, to explain the origin of the size eect,
other factors must be considered in addition to the size-
induced phase transition size. In the above paper, Wada
et al. also found that all of the BaTiO
3
nanoparticles
were always composed of two parts, i.e., (a) a surface
cubic layer and (b) a bulk tetragonal layer [10,11]. This
suggested that the surface layer might be related to the
dielectric maximum and the minimum Slater mode fre-
quency. Thus, clarifying the relationship between the
surface layer thickness and the size dependence of di-
electric constants is very important.
In this study, to investigate the above relationship,
BaTiO
3
nanoparticles with dierent surface layer thick-
nesses were prepared using a 2-step thermal decompo-
sition method of barium titanyl oxalate under various
degrees of vacuum. Their surface layer thicknesses were
determined using a high energy X-ray diraction (XRD)
method with synchrotron radiation and a Rietveld anal-
ysis. Moreover, the dielectric constants were measured
at 20

C as functions of the particle size.
-878-
Origin of Ultrahigh Dielectric Constants for Barium S. Wada et al. -879-
Fig. 1. Temperature dependence of the partial pressure
at the 2nd step when using the Ba
2
Ti
2
O
5
CO
3
intermediate
particles of 20 g.
II. EXPERIMENTAL AND DISCUSSION
The 2-step thermal decomposition method was de-
veloped to prepare impurity-free and nm-sized BaTiO
3
particles [12]. In this method, at the 1
st
step, BaTiO
(C
2
O
4
)4H
2
O particles with a size of ca. 200 m are
decomposed into Ba
2
Ti
2
O
5
CO
3
intermediate nanopar-
ticles at 500

C in atmosphere while at the 2nd step,
Ba
2
Ti
2
O
5
CO
3
intermediate nanoparticles are decom-
posed into BaTiO
3
nanoparticles at 650

C under vac-
uum. At the 2nd step, the volume of the reaction cham-
ber is xed; thus, the degree of vacuum in the reaction
chamber depends on only the number of Ba
2
Ti
2
O
5
CO
3
intermediate nanoparticles as the starting material. In
this study, three amounts of Ba
2
Ti
2
O
5
CO
3
intermediate
nanoparticles, i.e., 20, 35 and 50 g, were loaded in the
reaction chamber.
Figure 1 shows the temperature dependence of the par-
tial pressure at the 2nd step using the Ba
2
Ti
2
O
5
CO
3
in-
termediate particles of 20 g. In Figure 1, desorption of
surface adsorbed H
2
O is clearly observed from 100 to 400

C while desorption of CO
2
by thermal decomposition of
the intermediates is observed from 600 to 800

C. The
maximum partial pressure of CO
2
gas was around 150 Pa
at 650

C. A similar measurement was performed using
the Ba
2
Ti
2
O
5
CO
3
intermediate particles of 35 and 50
g. For the intermediate particles of 35 g, the maximum
partial pressure of the CO
2
gas was around 200 Pa at
650

C while for the intermediate particles of 50 g, the
maximum partial pressure of the CO
2
gas was around
400 Pa at 650

C. Each amount of the Ba
2
Ti
2
O
5
CO
3
intermediates was calcined at a temperature from 630 to
1,000

C for 1 hr under vacuum at dierent CO
2
partial
pressures from 150 to 400 Pa. As a result, three sets
of BaTiO
3
nanoparticles with sizes from 17 to 1,000 nm
were prepared.
These BaTiO
3
particles were characterized using vari-
ous methods. The crystal structure was investigated us-
ing a powder X-ray diractometer (XRD) (RINT2000,
Fig. 2. Particle size dependence of the dielectric constants
for three sets of the BaTiO
3
particles.
Rigaku, Cu-k, 50 kV, 30 mA). The average particle
sizes and crystallite sizes were estimated using a trans-
mission electron microscope (TEM) (CM300, Philips,
300 kV) and XRD. The impurity in the particles was
analyzed using a Fourier transform infrared spectrome-
ter (FT-IR) (SYSTEM 2000 FT-IR, Perkin Elmer) and
by dierential thermal analysis with thermogravimetry
(TG-DTA) (TG-DTA2000, Mac Science). The absolute
density of the BaTiO
3
particles was measured using a py-
cnometer, and the relative density was calculated using
the theoretical density estimated from the lattice param-
eters obtained by using the XRD measurements. The
Ba/Ti atomic ratios for the BaTiO
3
particles were de-
termined by using an X-ray uorescence analysis. The
particle sizes increased with increasing temperature at
the 2nd step, and the particle size from the TEM ob-
servation was consistent with the crystallite size from
the XRD measurement, which revealed that one parti-
cle was a single crystal particle. The relative densities
were almost over 98 %, and there were no impurities in
lattice. The Ba/Ti atomic ratio was always 1.00. In
this study, three sets of BaTiO
3
particles with dier-
ent particle sizes from 17 to 1,000 nm were prepared,
and on the basis of the above characterizations, there
was no dierences among these particles. Therefore, the
BaTiO
3
particles prepared by using the 2-step thermal
decomposition method were defect-free and impurity-free
nanoparticles, despite the amount of the intermediate
particles at the 2nd step. The dielectric constants of
these particles were measured using slurries and a pow-
der dielectric measurement method [13]. The slurries of
the BaTiO
3
particles obtained in this study were pre-
pared using ball milling with propylene carbonate. The
dielectric constants of these slurries were measured at
10.0 vol%, 20.00

C and 20 MHz. As a reference, com-
mercial BT-05 particles (BaTiO
3
particles of 500 nm,
Sakai Chemical Co.) were used. The dielectric constants
of these BaTiO
3
particles were estimated by using a -
nite element method (FEM) calculation with the mea-
sured dielectric constants of BaTiO
3
slurries. Figure 2
-880- Journal of the Korean Physical Society, Vol. 51, No. 2, August 2007
Fig. 3. Result of a Rietveld tting using a two-layer model
for the BaTiO
3
nanoparticles of 140 nm.
shows the particle size dependence of dielectric constants
for the three sets of BaTiO
3
particles. In Figure 2, for
the BaTiO
3
particles prepared using the Ba
2
Ti
2
O
5
CO
3
intermediate particles of 20 g at the 2nd step, a dielectric
maximum of around 15,000 is clearly observed at a size
of 60 nm. This result is consistent with a previous report
[9]. For the BaTiO
3
particles prepared using intermedi-
ate paticles of 35 g at the 2nd step, a dielectric maximum
of around 5,000 is clearly observed at a size of 140 nm.
On the other hand, BaTiO
3
particles prepared using in-
termediate particles of 50 g at the 2nd step, there is no
dielectric maximum. It should be noted that for these
three sets of the BaTiO
3
nanoparticles, although there
was no dierence in the above characterizations, the par-
ticle size dependence of the dielectric constants was com-
pletely dierent. Therefore, other factors responsible for
these dierences must be investigated. For this purpose,
it is very important to clarify the relationship between
the dielectric constant and the particle structure (two
layer model with a surface cubic layer and a bulk tetrag-
onal layer).
To determine the particle structure, we measured that
the high- intensity XRD patterns of the above BaTiO
3
nanoparticles by using the synchrotron XRD technique.
First of all, prior to the Rietveld analysis, crystal sym-
metry must be determined. Thus, the temperature
dependences of the FWHM for the broad (111) and
(200) planes were carefully measured and compared to
determine the crystal symmetry at room temperature.
As a result, for the BaTiO
3
particles prepared using
Ba
2
Ti
2
O
5
CO
3
intermediate particles of 20 g at the 2nd
step, the phase transition size, d
PT
, of the BaTiO
3
par-
ticles, which is the size of ferroelectric phase transition
from tetragonal to cubic at room temperature, was re-
vealed to be between 17 and 30 nm [11]. On the other
hand, for the BaTiO
3
particles prepared by using inter-
mediate particles of 35 and 50 g at the 2nd step, the
d
PT
existed between 25 and 40 nm. These results reveal
that the d
PT
is almost constant at around 20 30 nm,
despite the amount of Ba
2
Ti
2
O
5
CO
3
intermediate par-
Fig. 4. Relationship between the thickness of the surface
cubic layer and the amount of the Ba
2
Ti
2
O
5
CO
3
intermediate
at the 2nd step.
ticles at the 2nd step. Moreover, it should be noted that
the dPT is always much smaller than the particle size for
the dielectric maximum.
On the basis of the above result, the particle structure
was determined using a Rietveld tting. In this study,
for the BaTiO
3
particles with sizes over the d
PT
, a two
phase model was always applied [14], as shown in Figure
3. As a result, for all of the BaTiO
3
particles obtained
in this study, the volume ratio of tetragonal to cubic was
determined, and for a comparison using BaTiO
3
particles
with the same particle size, the volume ratio of tetragonal
to cubic was found to signicantly increase with decreas-
ing amount of the Ba
2
Ti
2
O
5
CO
3
intermediate particles
at the 2nd step. Moreover, the thickness of the surface
cubic layer was calculated using two values, i.e., (a) the
particle size and (b) the volume ratio of tetragonal to cu-
bic. As a result, for BaTiO
3
particles prepared using the
Ba
2
Ti
2
O
5
CO
3
intermediate particles of 20 g at the 2nd
step, the thickness of the surface cubic layer was around
1 2 nm while for the BaTiO
3
particles prepared using
intermediate particles of 35 g at the 2nd step, the thick-
ness of the surface cubic layer was around 3 5 nm. For
the BaTiO
3
particles prepared using intermediate parti-
cles of 50 g at the 2nd step, the thickness of the surface
cubic layer was around 8 10 nm. Therefore, the thick-
ness of the surface cubic layer decreased with decreasing
amount of Ba
2
Ti
2
O
5
CO
3
intermediate particles at the
2nd step, as shown in Figure 4.
The above discussion reveals that there is a relation-
ship between the thickness of the surface cubic layer and
the dielectric constants; i.e., the dielectric maximum in-
creased and the particle size for the dielectric maximum
decreased with decreasing thickness of the surface cu-
bic layer. This suggests that for the dielectric constant,
the surface layer is a very important factor. Thus, the
origin of ultrahigh dielectric constant can be strongly re-
lated to the surface layer, i.e., to the particle structure
for the nanoparticles. If this relation is claried, the
structure of the surface cubic layer must be investigated.
Origin of Ultrahigh Dielectric Constants for Barium S. Wada et al. -881-
Moreover, the relationship between the thickness of the
surface layer and the preparation condition (the amount
of Ba
2
Ti
2
O
5
CO
3
intermediate particles at the 2nd step)
must be investigated. For these purposes, continuous
study is required.
III. CONCLUSIONS
In this study, three sets of the BaTiO
3
nanoparticles
with various particle sizes from 17 to 1,000 nm were pre-
pared by using a 2-step thermal decomposition method
with barium titanyl oxalate under various degrees of vac-
uum. For this purpose, the amount of Ba
2
Ti
2
O
5
CO
3
intermediate particles at the 2nd step of the 2-step ther-
mal decomposition method was changed from 20 to 50
g. As a result, the partial pressure of the CO
2
gas at the
decomposition of the Ba
2
Ti
2
O
5
CO
3
intermediate parti-
cles changed from 150 to 400 Pa. For the above three
sets of BaTiO
3
nanoparticles, various characterizations
were performed, but there was no dierences between
these BaTiO
3
nanoparticles. Moreover, these character-
izations revealed that these particles were impurity-free,
defect-free, dense BaTiO
3
nanoparticles. The powder di-
electric measurement claried that the size dependence
of the dielectric properties was signicantly depended on
the amount of the Ba
2
Ti
2
O
5
CO
3
intermediate particles.
When the amounts of Ba
2
Ti
2
O
5
CO
3
intermediate parti-
cles was 20 and 35 g, the dielectric maximum of BaTiO
3
nanoparticles with sizes of around 60 and 140 nm exhib-
ited a maximum of around 15,000 and 5,000, respectively.
On the other hand, when the amount of Ba
2
Ti
2
O
5
CO
3
intermediate particles was 50 g, no dielectric maximum
was observed. To investigate the factors responsible for
these dielectric behaviors, we investigated precisely par-
ticle structure by using synchrotron radiation. As a re-
sult, the thickness of the surface cubic layer decreased
with decreasing amount of Ba
2
Ti
2
O
5
CO
3
intermediate
particles at the 2nd step. Thus, we conrmed that the
surface structure was a factor responsible for the dielec-
tric properties of the BaTiO
3
nanoparticles. However,
there are still many questions about the surface struc-
ture, the factor for the surface thickness, and the rela-
tionship between the surface thickness and the dielectric
properties. To clarify these questions, we will perform
continuous study will be performed in the future.
ACKNOWLEDGMENTS
We would like to thank Mr. M. Nishido of Fuji Tita-
nium Co., Ltd. for preparing high purity barium titanyl
oxalates and Mr. K. Abe of Sakai Chemical Industry Co.,
Ltd. for providing high purity BT-05 powders. We also
would like to thank Dr. K. Kato and Dr. M. Takata of
SPring-8, JASRI for helpful discussion of the high inten-
sity XRD measurement using the synchrotron radiation
technique. The experiment at SPring-8 was carried out
under Program No. 2005A0691 and 2005B0724. This
study was partially supported by a Grant-in-Aid for Sci-
entic Research (15360341) from the Ministry of Edu-
cation, Science, Sports and Culture, Japan and by the
Ookura Kazuchika Memorial foundation.
REFERENCES
[1] K. Kinoshita and A. Yamaji, J. Appl. Phys. 45, 371
(1976).
[2] G. Arlt, D. Hennings and G. De With, J. Appl. Phys.
58, 1619 (1985).
[3] K. Ishikawa, K. Yoshikawa and N. Okada, Phys. Rev. B
37, 5852 (1988).
[4] K. Uchino, E. Sadanaga and T. Hirose, J. Am. Ceram.
Soc. 72, 1555 (1989).
[5] M. H. Frey and D. A. Payne, Phys. Rev. B 54, 3158
(1996).
[6] S. Wada, T. Suzuki and T. Noma, J. Ceram. Soc. Jpn.
104, 383 (1996).
[7] D. McCauley, R. E. Newnham and C. A. Randall , J.
Am. Ceram. Soc. 81, 979 (1998).
[8] M. H. Frey, Z. Xu, P. Han and D. A. Payne, Ferroelectrics
206, 337 (1998).
[9] S. Wada, H. Yasuno, T. Hoshina, S.-M. Nam, H. Kake-
moto and T. Tsurumi, Jpn. J. Appl. Phys. 42, 6188
(2003).
[10] S. Wada, H. Yasuno, T. Hoshina, H. Kakemoto, Y.
Kameshima, T. Tsurumi and T. Shimada, J. Electro-
ceram, in press (2006).
[11] S. Wada, T. Hoshina, K. Takizawa, A. Yazawa, M.
Ohishi, H. Yasuno, H. Kakemoto and T. Tsurumi, In-
ter. J. Appl. Ceram. Tech. (2006), in press.
[12] S. Wada, M. Narahara, T. Hoshina, H. Kakemoto and
T. Tsurumi, J. Mater. Sci. 38, 2655 (2003).
[13] S. Wada, T. Hoshina, H. Yasuno, S.-M. Nam, H. Kake-
moto and T. Tsurumi, Key Eng. Mater. 248, 19 (2003).
[14] S. Aoyagi, Y. Kuroiwa, A. Sawada, I. Yamashita and T.
Atake, J. Phys. Soc. Jpn. 71, 1218 (2002).