Sei sulla pagina 1di 7

First Principles Study on Intrinsic Vacancies in Cubic and Orthorhombic CaTiO

3
Haksung Lee
1
, Teruyasu Mizoguchi
1;
*
, Takahisa Yamamoto
1;2
and Yuichi Ikuhara
1;2;3
1
Institute of Engineering Innovation, The University of Tokyo, Tokyo 113-8656, Japan
2
Nanostructures Research Laboratory, Japan Fine Ceramics Center, Nagoya 456-8587, Japan
3
WPI, Advanced Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan
The structural relaxations and the formation energies of intrinsic defects in cubic and orthorhombic CaTiO
3
were investigated by a rst
principles projector-augmented wave method. It was found that cations and oxygen vacancies in both phases cause extra levels near the valence
band maximum and the conduction band minimum, respectively, and the Ti-vacancy induced level in orthorhombic CaTiO
3
is closer to the
valence band maximum than that in cubic CaTiO
3
. Among the neutral defect species, including neutral isolated vacancy, partial Schottky, and
full Schottky, it was found that the V
Ca
2
V
O
2
and V
O
0
are the most preferable defect species for orthorhombic CaTiO
3
under reduction and
oxidization conditions, respectively, whereas the V
Ca
2
V
O
2
partial Schottky is always stable in any atmosphere in cubic CaTiO
3
. As
compared to cubic CaTiO
3
, it was found that orthorhombic CaTiO
3
shows higher defect formation energies.
[doi:10.2320/matertrans.MC200813]
(Received December 15, 2008; Accepted March 11, 2009; Published April 15, 2009)
Keywords: calcium titanate, defect energetics, calculation
1. Introduction
CaTiO
3
is an important material for a radioactive waste
and electric materials with doping.
1,2)
Due to its similarity in
the crystal structure, CaTiO
3
is also utilized as a dopant and a
member of the superlattice together with other perovskite
materials, such as BaTiO
3
and SrTiO
3
.
35)
It has been known
that the electric properties and microstructures of those
perovskite materials were strongly dependent on the oxygen
partial pressure, indicating that defects play an essential role
to bring about their properties.
612)
Thus, an understanding of
defect energetics is indispensable for further developing and
applications of these perovskite oxides.
Although the defect energetics of SrTiO
3
and BaTiO
3
has
been studied so far,
1316)
there are only a few reports on the
defect energetics in CaTiO
3
. Recently Mather et al. reported
the energetics of defects, dopant-vacancy association, and the
nanoscale clusters formation in orthorhombic CaTiO
3
based
on a static lattice calculation with empirical pair potentials.
17)
First principles calculations of CaTiO
3
were also performed
by Cockayne et al. and Wang et al., but they were mainly
focused on the dielectric constant, optical properties, and the
surface structures.
18,19)
In this study, the intrinsic defect formation energies in
cubic and orthorhombic CaTiO
3
were calculated by a rst-
principles projector-augmented wave (PAW) method. The
formation energies and atomic relaxations around vacancies
were investigated.
2. Methodology
2.1 First principles calculations of cubic and ortho-
rhombic CaTiO
3
A rst-principles projector augmented wave (PAW)
method within the generalized gradient approximation
(GGA) implemented in VASP code was used to calculate
the defect energetics.
2022)
The wave functions were expand-
ed in a plane-wave basis set with a plane-wave cuto energy
of 400 eV. CaTiO
3
shows three types of polymorphs depend-
ing on temperature: orthorhombic, tetragonal and cubic.
23)
In
this study, defect energetics in cubic and orthorhombic
CaTiO
3
were calculated, because tetragonal phase is a
transient compound in very limited temperature (1523 K <
T < 1622 K), and the cubic and orthorhombic CaTiO
3
are stable over wide temperature range T > 1622 K and
T < 1486 K, respectively.
First, the unitcell calculations were performed to nd
optimized structures for two polymorphs. The primitive cell
of cubic CaTiO
3
was calculated using 10 10 10 k-point
mesh generated by the Monkhors-Pack scheme (35 irredu-
cible k-points). From the energy-volume curve, the lattice
constant and the bulk modulus were estimated to be 3.89 A

and 173 GPa, respectively, which are good agreement with


previously reported experimental value.
18)
For orthorhombic
CaTiO
3
, the structure was also calculated using 10 10 10
k-point mesh (125 irreducible k-points) allowing relaxation
until its residual forces were less than 0.05 eV/A

. Figure 1
shows the optimized atomic structures of the two poly-
morphs. It can be seen that the calculations are satisfactorily
reproduce the experimental structures.
Figure 2 shows the calculated band structures of cubic and
orthorhombic CaTiO
3
. Similar to BaTiO
3
and SrTiO
3
, the
valence-band maximum (VBM) composed of O 2p orbitals
and the conduction-band minimum (CBM) composed of
Ti 3d orbital respectively. In cubic CaTiO
3
, VBM at M is
0.50 eV higher than that at and the indirect gap between M
and is 1.97 eV and the direct band gap at is 2.50 eV.
For orthorhombic, it shows the direct band gap at , 2.54 eV.
The dierence of the direct band gap between cubic and
orthorhombic CaTiO
3
, 0.04 eV, is small. Although a feature
of the band structure is consistent with the previous GGA
report
19)
for cubic CaTiO
3
, the calculated band gap is below
the experimental value, 3.5 eV.
24)
The dierence between
experimental and calculated band gap could underestimate
the formation energies of oxygen vacancies, which will be
discussed later.
*
Corresponding author, E-mail: Mizoguchi@sigma.t.u-tokyo.ac.jp
Materials Transactions, Vol. 50, No. 5 (2009) pp. 977 to 983
Special Issue on Nano-Materials Science for Atomic Scale Modication
#2009 The Japan Institute of Metals
The optimized primitive cells of cubic and orthorhombic
CaTiO
3
were expanded by 3 3 3 and 2 2 2, and the
supercells with 135-atoms and 160-atoms were respectively
obtained. By using these supercells, a vacancy can be located
apart from the neighboring vacancies in the adjacent super-
cells by 11.6 A

for cubic CaTiO


3
and more than 10.7 A

for
orthorhombic CaTiO
3
. To introduce an isolated vacancy of
Ca, Ti, or O, an interior atom was removed from the
supercell. The third nearest neighbor sites from the vacancy
for cubic CaTiO
3
were allowed to relax until their residual
forces were less than 0.05 eV/A

. In orthorhombic CaTiO
3
, it
is dicult to determine the third nearest neighbor exactly.
Therefore, atoms inside a sphere of 3.9 A

, which is slightly
bigger than the relaxation radius of cubic CaTiO
3
, centered at
the defect were relaxed. For the large supercell calculations,
the 2 2 2 k-points mesh generated by the Monkhorst-
Pack scheme was used for the numerical integrations over the
Brillouin zone.
(a) Cubic CaTiO3 (b) Orthorhombic CaTiO3
Pm3m
Pbnm
173 171
3.89 3.90
Exp. Cal.
Bulk modulus(GPa)
a()
Ca
Ti O
Ti
Ca
O1
O2
5.36
5.45
7.62
5.38
5.44
7.64
a
b
c
Exp. Cal.
Ca
x
y
-0.0078
0.0357
-0.0820
0.0433
O1 x
y
0.0736
0.4828
0.0808
0.4789
O2
x
y
z
0.7113
0.2893
0.0375
0.7069
0.2926
0.0424
()
()
()
z 0.25 0.25
z 0.25 0.25
Ti
x
y
0.0
0.5
0.0
0.5
z 0.0 0.0
Fig. 1 The crystal structures of cubic and orthorhombic CaTiO
3
. The calculated and experimental lattice parameters and bulk modulus are
shown in the table, and the optimized atomic coordinates are also shown for the orthorhombic structure. The experimental data were
obtained from Ref. 23).
Y X M R
-5
0
5
E
n
e
r
g
y

/

e
V
Wave Vector
X M R X
-5
0
5
Wave Vector
(a) Cubic CaTiO3 (b) Orthorhombic CaTiO3
E
n
e
r
g
y

/

e
V
Fig. 2 Calculated energy band structure for (a) cubic and (b) orthorhombic CaTiO
3
. The VBM at M and was set at 0 eV for cubic and
orthorhombic, respectively.
978 H. Lee, T. Mizoguchi, T. Yamamoto and Y. Ikuhara
2.2 Defect Formation Energy
The formation energies of vacancies in CaTiO
3
were
calculated by the similar procedure as the previous re-
ports.
15,16)
The formation energy (E
f
) is evaluated as
following equation:
E
f
E
T
defect : q E
T
perfect
fn
Ca

Ca
n
Ti

Ti
n
O

O
g q"
F
E
VBM
1
Here, E
T
(defect:q) and E
T
(perfect) are the total energies of
the supercells containing a defect in a charge state q and that
of the perfect supercell without any defect, respectively. n
Ca
,
n
Ti
and n
O
are the numbers of Ca, Ti, and O atoms removed
from the large supercells to introduce a vacancy.
Ca
,
Ti
,
and
O
are the atomic chemical potentials, and "
F
is the
Fermi energy measured from the VBM. In this study, charge-
neutral (V
Ca
0
, V
Ti
0
, and V
O
0
) and ionized (V
Ca
2
, V
Ti
2
,
V
Ti
4
, and V
O
2
) vacancies were considered.
As can be seen eq. (1), E
VBM
needs to be determined. For
this purpose, it was assumed that the electrostatic potentials
in the supercell without vacancies were similar to those far
from a defect in the supercell with a vacancy. This treatment
can compensate distortion to the band structure around the
band gap, which is caused by a vacancy.
2527)
Since the theoretical band gap was lower than the
experimental value, the energy of the electrons at the donor
state is underestimated. The band gap energy was calculated
from the total energies of the supercell as following equation:
E
g
E
perfect
CBM
E
perfect
VBM
fE
1
T
E
0
T
E
0
T
E
1
T
g 2
where E
q
T
indicates the total energy of a perfect supercell
with the charge state: q. The calculated band gap of cubic
CaTiO
3
and that of orthorhombic CaTiO
3
is 2.07 eV and
2.15 eV smaller than experimental value, respectively. In the
previous report, the formation energy of an oxygen vacancy
was adjusted to be increased the donor level by the band gap
dierence between the calculation and experiment.
28)
Ad-
justing the defect formation energy of V
O
0
is added 2 Eg
and for V
O
1
, 1 Eg is added for calculating the defect
formation energy.
The formation energies also depend on the atomic chemi-
cal potentials. The chemical potentials of the constituent
atoms vary with the equilibrium conditions among their
related phases. Figure 3 shows a schematic phase diagram of
the ternary Ca-Ti-O system, and the vertices of three phase
regions were named AG as for the previous reports.
15,16)
The values of the atomic chemical potential of three elements
were evaluated at those points. Here, it was assumed that
CaTiO
3
was stable at every point, so that the chemical
potentials of the three elements were obtained from the
combinations of the chemical potentials of Ca, CaO, Ti, TiO,
Ti
2
O
3
, TiO
2
and O
2
. The chemical potential of O
2
was
obtained from the supercell calculation at the point
for an isolated O
2
molecule using a large vacuum cell
15 15 15 A

3
. Spin-polarized calculation was performed
for the chemical species because in the previous report spin-
polarized calculations were closer to the experimental
values.
15,16)
Especially, for calculating the chemical potential
of Ti
2
O
3
, antiferromagnetic state was considered because
antiferromagnetism in Ti
2
O
3
has been reported in litera-
ture.
30)
The heats of formation obtained from the calcula-
tions are summarized in Table 1. It is seen that spin-
polarized calculations satisfactorily reproduce the experi-
mental values.
As shown in Fig. 1, orthorhombic CaTiO
3
has two oxygen
sites, O1 and O2. Their formation energies were separately
calculated. As a result, it was found that their formation
energies are very similar and those of V
O1
q(0,1,2)
is slightly
lower compared to those of V
O2
q
. Thus, the defect formation
energies of V
O1
0
and V
O1
q
were employed for calculating the
formation energies of the neutral vacancy and the Schottky
reactions.
3. Results and Discussion
3.1 Atomic and electronic structures of isolated vacan-
cies
Figure 4 shows the one-electron energy levels around the
band gap and the number of electrons occupying these states.
Since the existence of vacancies causes signicant distortion
to the band structures around the band gap, the alignment of
the band structures of the perfect and defective supercells is
necessary. For this purpose, it was assumed that the potentials
in the perfect supercell are similar to those far from a defect
in a defective supercell. The VBM positions of defective
supercells were determined by the following equation:
31,32)
E
VBM
E
perfect
VBM
V
defect
av
V
perfect
av
3
Ti
G
TiO
A
Ti2O3
B
TiO2
C
O
D
CaO
E
Ca
F
Fig. 3 Schematic phase diagram of the ternary system Ca-Ti-O. A hatched
polygon shows the region of CaTiO
3
. Corners of the polygon are indicated
by A-G.
Table 1 Calculated and experimental formation enthalpies for the refer-
ence materials. The experimental data were obtained from Ref. 29).
Formation enthalpies (eV/atom)
Experiment Calculation
CaO F
m3m
3:30 3:03
TiO F
m3m
2:82 2:29
TiO
2
P
42/mmm
3:25 3:03
Ti
2
O
3
R
33C
3:16 2:67
CaTiO
3
P
m3m
3:11
P
bnm
3:17
First Principles Study on Intrinsic Vacancies in Cubic and Orthorhombic CaTiO
3
979
The VBM of perfect cubic and orthorhombic CaTiO
3
was set at 0 eV for each case. In the case of the cation
vacancies (V
Ca
and V
Ti
), the positions of the states in the
band measured from the VBM are displayed in this gure.
In contrast, the positions of the V
O
induced states are
measured from the CBM. As can be seen in Fig. 4, each
vacancy induces an extra level in the band gap. In the case
of V
Ca
and V
Ti
, the extra levels are located near the VBM
and the energy levels become larger with increasing
negative charge for both cubic and orthorhombic CaTiO
3
.
Figures 5(a) and 5(b) shows the contour maps of the
squares of the wave functions for the extra states induced
by V
Ca
or V
Ti
in cubic CaTiO
3
on the {100} plane,
respectively. It can be seen that these states are mainly
composed of 2p orbitals of O ions surrounding the
vacancies. In cubic CaTiO
3
, the wave function of the
V
Ti
0
induced level shows more localized than that of the
V
Ca
0
induced level, indicating that the electrons occupied
in the V
Ti
0
induced level would suer more electronic
repulsions around the vacancy site as compared with the
V
Ca
0
induced level. It could explain that the energy levels
for V
Ti
with negative charges tend to be higher than those
for V
Ca
, as shown in Fig. 4(a). In contrast, the extra levels
by V
O
0
are located close to the CBM and thus this level is
mainly composed of Ti 3d orbitals. Those characteristics in
the defect-induced levels are very similar to the previously
reported results for SrTiO
3
.
15)
Figure 4(b) and Figure 6 show the energy levels of the
vacancy induced states and the corresponding contour maps
for orthorhombic CaTiO
3
. It is interestingly mentioned that
the V
Ti
induced level in orthorhombic CaTiO
3
is closer to the
VBM than that in cubic CaTiO
3
. This can be ascribed to the
fact that the V
Ti
0
induced level and the V
Ca
0
induced level in
orthorhombic CaTiO
3
is more delocalized and localized,
respectively, as compared with that in cubic CaTiO
3
. (Figs. 5
and 6)
The introduction of vacancies induces not only extra levels
in the band gap but also structural relaxations of the
surrounding ions. Table 2(a) lists the calculated interatomic
distances between each vacancy and the rst and second NN
ions before and after structural relaxations in cubic CaTiO
3
.
For all vacancy species, the rst NN ions exhibit outward
relaxations by more than 3%, irrespective of the charge
states. The outward relaxations of the rst NN ions can be
ascribed to that the chemical bonds between the removed ion
and surrounding ions vanishes by introducing the vacancy. In
contrast, the second NN cations of V
Ca
and V
Ti
showed
VBM
0 1- 2- 0 1- 2- 3- 4- 0 1+ 2+
0.0
0.5
1.0
2.8
3.5
E
n
e
r
g
y

/

e
V
VBM
CBM
0.18
0.21
0.32
0.26
0.35
0.39
0.44
0.49
0.09
0.12
0.20
VCa VTi
VO
(a) Cubic CaTiO3
0.0
0.5
1.0
2.8
3.5
E
n
e
r
g
y

/

e
V
CBM
0.16
0.21
0.33
0.18
0.22
0.29 0.41
0.48
0.06 0.09 0.10
0 1- 2- 0 1- 2- 3- 4- 0 1+ 2+
VCa VTi
VO1
(b) Orthorhombic CaTiO3
0.07
0.12 0.11
VO2
0 1+ 2+
Fig. 4 One-electron energy levels for V
Ca
, V
Ti
and V
O
of CaTiO
3
in
various charge states for (a) cubic and (b) orthorhombic CaTiO
3
. Electron
energy levels are given with respect to the VBM in the case of V
Ca
and
V
Ti
, while those from the CBM in the case of V
O
.
Vacancy
Ca
Ti
O
(a)
(c)
(b)
VCa
VTi
VO
Fig. 5 Contour maps of the square of the wave functions of vacancy-
induced levels of (a)V
Ca
0
, (b)V
Ti
0
and (c)V
O
0
on the {100} plane of cubic
CaTiO
3
. The contour lines are drawn from 0.005 to 0.2 with an interval
0.01 in the unit of electrons/A

3
.
980 H. Lee, T. Mizoguchi, T. Yamamoto and Y. Ikuhara
inward relaxations because electrostatic repulsions between
second NN cations and the removed cation are reduced by the
vacancies. V
Ti
in cubic CaTiO
3
induces around 11% inward
relaxations of the second NN Ca while V
Ca
in cubic CaTiO
3
induces less than 2% inward relaxations of Ti. In the case of
V
O
, there are two kinds of atoms, Ca and O which are the
second NN before relaxations. After relaxations, Ca moves
outward and O moves inward due to electrostatic repulsions
and attraction, respectively.
Table 2(b) shows structural relaxations by defects in
orthorhombic CaTiO
3
. Since orthorhombic CaTiO
3
is a
lower symmetry phase than cubic CaTiO
3
, the coordination
number and the distances of surrounding ions are more
complex. Although it is dicult to dene exact nearest
neighbor (NN) atoms, the dierence of the bond length
within 0.1 A

is regarded as the same coordination members


for the rst NN and their average distances were evaluated.
For example, although O2 in Fig. 1 is coordinated by two Ti
ions with 1.961 A

and 1.968 A

bond lengths, they are


considered as the same coordination. The second NN bond
length was calculated by the average of the distance between
a vacancy and the second NN atomic species. For example,
second NN atomic species for V
Ti
are Ca ions, which
locate 3.13 A

, 3.26 A

, 3.33 A

and 3.52 A

apart from V
Ti
. The
(a) (c) (b)
VCa
VTi
VO1
VO2
(d)
Vacancy
Ca
Ti
O
Fig. 6 Contour maps of the square of the wave functions of vacancy-induced levels of (a) V
Ca
0
, (b) V
Ti
0
and (d) V
O2
0
on the (001) plane
and that of (c)V
O1
0
on the (100) plane of orthorhombic CaTiO
3
. The (100) plane was selected only for V
O1
0
to show the Ti site on the
plane. The contour lines are drawn from 0.005 to 0.2 with an interval 0.01 in the unit of electrons/A
3
.
Table 2 The interatomic distance of the neighboring ions from a vacancy (A

) after the structural relaxation in (a) cubic and (b)


orthorhombic CaTiO
3
.
(a) Cubic CaTiO
3
(b) Orthorhombic CaTiO
3
First NN Second NN First NN Second Species
Ca 2.75(O12) 3.37(Ti8) 2.34(O4) 2.62(O4) 3.31(Ti8)
V
Ca
0
2.86(3.83%) 3.32(1:57%) 2.45(4.89%) 2.71(3.57%) 3.26(1:66%)
V
Ca
1
2.86(3.83%) 3.32(1:57%) 2.45(4.71%) 2.72(3.52%) 3.25(1:69%)
V
Ca
2
2.86(3.83%) 3.32(1:57%) 2.44(4.40%) 2.72(3.77%) 3.25(1:88%)
Ti 1.94(O6) 3.37(Ca8) 1.96(O6) 3.31(Ca8)
V
Ti
0
2.01(3.54%) 2.98(11:7%) 2.07(5.34%) 3.19(3:79%)
V
Ti
1
2.01(3.54%) 2.98(11:7%) 2.07(5.27%) 3.19(3:83%)
V
Ti
2
2.01(3.59%) 2.97(11:7%) 2.07(5.45%) 3.18(3:91%)
V
Ti
3
2.01(3.59%) 2.97(11:9%) 2.07(5.64%) 3.17(4:30%)
V
Ti
4
2.01(3.61%) 2.94(12:6%) 2.08(5.89%) 3.16(4:70%)
O 1.94(Ti2) 2.75(O8) O1 1.96(Ti2) 2.37(Ca2)
V
O
0
2.03(4.49%) 2.57(6:25%) 2.12(8.34%) 2.51(6.05%)
V
O
1
2.03(4.62%) 2.56(6:20%) 2.12(8.62%) 2.51(6.07%)
V
O
2
2.04(4.84%) 2.56(6:27%) 2.13(8.73%) 2.51(5.86%)
2.75(Ca4) O2 1.97(Ti2) 2.53(Ca3)
2.92(6.69%) 2.14(8.87%) 2.66(5.06%)
2.92(6.74%) 2.14(9.02%) 2.66(5.01%)
2.92(6.93%) 2.14(9.18%) 2.66(5.01%)
First Principles Study on Intrinsic Vacancies in Cubic and Orthorhombic CaTiO
3
981
average distance becomes 3.31 A

before relaxations and


3.19 A

after relaxations. For V


O
0
in orthorhombic CaTiO
3
,
there are two oxygen vacancy positions (O1 and O2) which
have almost identical surroundings. O1 and O2 were
commonly coordinated by two Ti with 1.957 A

and 1.96 A

bond length, respectively. The rst NN relaxation due to


V
O1
0
, 8.34%, is close to that due to V
O2
0
, 8.87%.
Comparing the rst NN structural relaxations due to V
O1
0
in cubic and orthorhombic CaTiO
3
, the Ti relaxations in
orthorhombic, 8.34%, is approximately twice of that in cubic
CaTiO
3
, 4.49%. In addition, the Ca relaxation around V
Ti
0
in
cubic CaTiO
3
is 11:7%, which is much larger than those in
orthorhombic CaTiO
3
, 3:79%.
3.2 Defect formation energies of neutral isolated vacan-
cies and Schottky reactions
The formation energies of the isolated neutral vacancies
were evaluated. As mentioned above, the formation energy
for the O1 site in orthorhombic CaTiO
3
is slightly lower than
those for the O2 site. Therefore, the O for orthorhombic
CaTiO
3
hereafter corresponds to O1 site. The formation
energy of the isolated vacancies, Ca, Ti and O, were shown in
Fig. 7. It is commonly found that the vacancy formation
energy of Ti is relatively higher than that of Ca in both
phases. This trend is the same as SrTiO
3
and cubic BaTiO
3
,
in which the defect formation energy of V
Ti
0
is also higher
than that of V
A(Sr, Ba)
0
.
15,16)
In order to compare the formation energies in cubic and
orthorhombic CaTiO
3
, the average values of the formation
energies over all equilibrium points were evaluated. The
average values for V
Ti
0
in cubic and orthorhombic CaTiO
3
are 11.8 and 13.1 eV and those for V
Ca
0
in cubic and
orthorhombic CaTiO
3
are 5.00 and 6.33 eV, respectively.
In addition, the average values for V
O
0
in cubic CaTiO
3
,
4.05 eV, are lower than those in orthorhombic CaTiO
3
,
4.47 eV. Namely, it can be concluded that all intrinsic
vacancies are dicult to form in the orthorhombic phase as
compared with those in the cubic phase.
In additional to those isolated vacancies, the following
Schottky reactions were also considered: V
Ca
2
V
O
2
(CaO partial Schottky), V
Ti
2
V
O
2
(TiO partial
Schottky), 2V
Ti
3
3V
O
2
(Ti
2
O
3
partial Schottky),
V
Ti
4
2V
O
2
(TiO
2
partial Schottky), V
Ca
2
V
Ti
4

3V
O
2
(CaTiO
3
full Schottky). In those Schottky reactions,
each vacancy was separately calculated and the association
between vacancies was not considered. The defect formation
energies at each point are plotted in Fig. 7. The calculated
formation energies of the full Schottky, 2.19 eV for cubic
CaTiO
3
and 2.66 eV for orthorhombic CaTiO
3
, are close to
the previously reported value by G. C. Mather et al., of
2.49 eV, which was obtained for orthorhombic CaTiO
3
by
the static lattice calculation.
17)
As compared with the full
Schottky, the CaO partial Schottky defect formation energy
changed with equilibrium point. In the Ti-rich conditions (B,
C, D and E), the CaO partial Schottky reaction for cubic
CaTiO
3
was 1.21 eV, which is smaller than that in Ca-rich
condition (A), 1.44 eV, and the CaO partial Schottky reaction
is the most dominant defect species for the cubic phase in any
equilibrium condition. In the case of orthorhombic CaTiO
3
,
CaO partial Schottky energy is 1.73 eV in Ti-rich (B, C and
D) and 2.11 eV in Ca-rich condition (A), and is the most
stable defect species except for point G. On the other hand, it
is found that both TiO, Ti
2
O
3
and TiO
2
partial Schottky
reactions are always higher formation energy as compared
with other Schottky reactions. This higher formation energy
of the TiO, Ti
2
O
3
and TiO
2
Schottky reactions can be
ascribed to the higher formation energy of the Ti vacancies.
By comparing cubic CaTiO
3
with orthorhombic CaTiO
3
,
it was found that all Schottky reactions in orthorhombic
CaTiO
3
show the higher formation energies due to the higher
formation energies of the isolated vacancies.
4. Summary
First-principles plane-wave-based PAW calculations were
performed for the defect energetic in cubic and orthorhombic
CaTiO
3
.
(1) It was found that cations and oxygen vacancies in
both phases cause extra levels near the valence band
maximum and the conduction band minimum, respec-
tively, and the Ti-vacancy induced level in orthorhom-
bic CaTiO
3
is closer to the valence band maximum than
that in cubic CaTiO
3
.
2
4
6
8
10
12
14
16
18
(a) Cubic CaTiO3
(b) Orthorhombic CaTiO3
Point in Phase Diagram
VCa
VTi
0
0
D
e
f
e
c
t

F
o
r
m
a
t
i
o
n

E
n
e
r
g
y

p
e
r

d
e
f
e
c
t
s
,

E
f

/

e
V
A B C D E F G
0
0
2
4
6
8
10
12
14
16
18
Point in Phase Diagram
A B C D E F G
D
e
f
e
c
t

F
o
r
m
a
t
i
o
n

E
n
e
r
g
y

p
e
r

d
e
f
e
c
t
s
,

E
f

/

e
V
VTi
0
VCa
0
V
Ca
0
V
Ti
0
V
O
0
V
Ca
2-
+V
O
2+
V
Ti
2-
+V
O
2+
2V
Ti
3-
+3V
O
2+
V
Ti
4-
+2V
O
2+
V
Ca
2-
+V
Ti
4-
+3V
O
2+
V
Ca
0
V
Ti
0
V
O
0
V
Ca
2-
+V
O
2+
V
Ti
2-
+V
O
2+
2V
Ti
3-
+3V
O
2+
V
Ti
4-
+2V
O
2+
V
Ca
2-
+V
Ti
4-
+3V
O
2+
Fig. 7 Defect formation energies calculated for (a)cubic and (b)ortho-
rhombic CaTiO
3
at points in schematic phase diagram. As described in
text, O in the orthorhombic case corresponds to the site O1 in Fig. 1.
982 H. Lee, T. Mizoguchi, T. Yamamoto and Y. Ikuhara
(2) Among the neutral vacancy species, the CaO partial
Schottky and V
O
0
are the most preferable for ortho-
rhombic CaTiO
3
under reduction and oxidization
conditions, respectively, whereas the CaO partial
Schottky is always stable in any atmosphere in cubic
CaTiO
3
.
(3) The isolated vacancies as well as the Schottky reactions
were found to be dicult to form in orthorhombic
CaTiO
3
as compared with those in cubic CaTiO
3
.
Acknowledgements
We acknowledge T. Tohei and N. Shibata for helpful
discussions. This study was partially supported by a Grant-in-
Aid for Scientic Research in Priority Area Nano Materials
Science for Atomic Scale Modication 474 and Young
Scientists (B) 20760449, from the Ministry of Education,
Culture, Sports, Science and Technology of Japan.
REFERENCES
1) X. S. Wang, C. N. Xu, H. Yamada, K. Nishikubo and X. G. Zheng:
Adv. Mater. 17 (2005) 12541258.
2) A. E. Ringwood, S. E. Kesson, K. D. Reeve, D. M. Levins and E. J.
Ramm: Radioactive Waste Forms for the future (1988) p. 233.
3) H. N. Lee, H. M. Christen, M. F. Chisholm, C. M. Rouleau and D. H.
Lowndes: Nature 433 (2005) 395399.
4) T. F. Lin and C. T. Hu: J. App. Phys. 67 (1990) 10421047.
5) M. Yamamoto, H. Ohta and K. Koumoto: Appl. Phys. Lett. 90 (2007)
072101.
6) T. Yamamoto and T. Sakuma: J. Am. Ceram. Soc. 77 (1994) 1107
1109.
7) T. Yamamoto and T. Sakuma: Mat. Sci. Forum 491 (1996) 204206.
8) P. R. Rios, T. Yamamoto, T. Kondo and T. Sakuma: Acta Metall. 46
(1998) 16171623.
9) B. K. Lee, S. Y. Chung and S. J. L. Kang: Acta Mater. 48 (2000) 1575
1580.
10) S. Y. Chung, D. K. Yoon and S. J. L. Kang: Acta Mater. 50 (2002)
33613371.
11) S. Y. Choi and S. J. L. Kang: Acta Mater. 52 (2004) 29372943.
12) Y. I. Jung, S. Y. Choi and S. J. L. Kang: Acta Mater. 54 (2006) 2849
2855.
13) G. V. Lewis and C. R. A. Catlow: J. Phys. Chem. Solids 47 (1986)
8997.
14) H. Moriwake: Int. J. Quantum Chem. 99 (2004) 824827.
15) T. Tanaka, K. Matsunaga, Y. Ikuhara and T. Yamamoto: Phys. Rev. B
68 (2003) 205213.
16) H. S. Lee, T. Mizoguchi, T. Yamamoto, S. J. L. Kang and Y. Ikuhara:
Acta Mater. 55 (2007) 65356540.
17) G. C. Mather, M. S. Islam and F. M. Figueiredo: Adv. Func. Mater. 17
(2007) 905912.
18) E. Cockayne and B. P. Burton: Phys. Rev. B 62 (2000) 37353743.
19) Y. X. Wang, M. Arai, T. Sasaki and C. L. Wang: Phys. Rev. B 73
(2006) 035411.
20) G. Kresse: Ph.D. thesis, Technische Universitat, Wien (1993).
21) G. Kresse and J. Furthmuller: Comput. Mater. Sci. 6 (1996) 1550.
22) J. P. Perdew, J. A. Chevary, H. Vosko, K. A. Jackson, M. R. Pederson,
D. J. Singh and C. Fiolhais: Phys. Rev. B 46 (1992) 66716687.
23) R. Ali and M. Yashima: J. Solid State Chem. 178 (2005) 28672872.
24) K. Ueda, H. Yanagi, R. Noshiro, H. Hosono and H. Kawazoe: J. Phys.
Condens. Matter 10 (1998) 36693677.
25) S. B. Zhang, S. H. Wei, A. Zunger and H. Katayama-Yoshida: Phys.
Rev. B 57 (1998) 96429656.
26) D. B. Laks, C. G. Van de Walle, G. F. Blochl and S. T. Pantelides: Phys.
Rev. B 45 (1992) 1096510978.
27) A. Garcia and J. E. Northrup: Phys. Rev. Lett. 74 (1995) 11311134.
28) J. Padilla and D. Vanderbilt: Phys. Rev. B 56 (1997) 16251631.
29) W. G. Mallard and P. J. Linstrom ed.: NIST chemistry webbook,
(2003) NIST standard reference database no. 69. Gaithersbug (MD):
National Institute of Standards and Technology.
30) S. C. Abrahams: Phys. Rev. 130 (1963) 22302237.
31) S. Poykko, M. J. Puska and R. M. Nieminen: Phys. Rev. B 53 (1996)
38133819.
32) T. Mattila and A. Zunger: Phy. Rev. B 58 (1998) 13671373.
First Principles Study on Intrinsic Vacancies in Cubic and Orthorhombic CaTiO
3
983

Potrebbero piacerti anche