Hole-doping and pressure effects on the metal-insulator transition in single crystals of Y1-xCaxTiO3 (0.37 ≤ x ≤ 0.41)

Article abstract of DOI:10.1143/JPSJ.72.3182

The measurements of electrical resistivity, magnetic susceptibility, specific heat, thermopower and powder x-ray diffraction under pressure have been performed on single crystals of Y_1-xCa_xTiO₃ (0.37 ≤ x ≤ 0.41) to investi

Full text of DOI:10.1143/JPSJ.72.3182

Journal of the Physical Society of Japan
Vol. 72, No. 12, December, 2003, pp. 3182–3188
#2003 The Physical Society of Japan
Hole-doping and Pressure Effects on the Metal–Insulator Transition
in Single Crystals of Y1ÀxCaxTiO3 (0:37 ꢀ x ꢀ 0:41)
ꢀ
Masami TSUBOTA, Fumitoshi IGA , Toshihiro NAKANO, Kazuyuki UCHIHIRA, Souichiro KURA,
1
2
Masaki TAKEMURA, Yoshio BANDO , Kazunori UMEO, Toshiro TAKABATAKE, Eiji NISHIBORI ,
2
2
3
3
Masaki TAKATA , Makoto SAKATA , Kenichi KATO and Yasuo OHISHI
Department of Quantum Matter, ADSM, Hiroshima University, Higashi-Hiroshima, Hiroshima 739-8530
1Department of Electrical Engineering, Kure National College of Technology, Kure, Hiroshima 737-8506
2Department of Applied Physics, Nagoya University, Nagoya 464-8603
3The Japan Synchrotron Radiation Research Institute (JASRI), SPring-8, Hyogo 679-5198
(Received June 12, 2003)
The measurements of electrical resistivity, magnetic susceptibility, specific heat, thermopower and
powder x-ray diffraction under pressure have been performed on single crystals of Y1ꢁxCaxTiO3
(0:37 ꢂ x ꢂ 0:41) to investigate the hole-doping effect and pressure effect on the metal–insulator (MI)
transition. In the low-temperature metallic state, the Ti moment is much reduced from S ¼ 1=2 spin
moment in the high-temperature insulating state. Upon applying pressure on the sample with x ¼ 0:37 at
temperatures below 200 K, the volume fraction of the orthorhombic phase exceeds that of the monoclinic
phase, which is associated with the insulator to metal transition. The good correspondence between the
x–T and P–T phase diagrams in the vicinity of the MI transition shows that the effect of hole doping by
1% Ca substitution is equivalent to the effect of application of pressure of 0.41 GPa. The phase separation
temperature agrees well with the MI transition temperature, and both temperatures increase linearly with
increasing x or pressure. These results indicate that the MI transition in Y1ꢁxCaxTiO3 is driven by
percolation of the low-temperature orthorhombic domains.
KEYWORDS: Mott insulator, metal–insulator transition, Y1Àx Cax TiO3, phase separation, percolation, crystal
growth, pressure effect
DOI: 10.1143/JPSJ.72.3182
YTiO3 is a d1-system, which is well known as a Mott–
Hubbard insulator with the large U ꢅ 1 eV.3) The ferromag-
In the field of transition-metal oxides, the metal–insulator netic state14–17) with a Curie temperature TC ꢅ 30 K is
(MI) transition induced by strong interactions between d- atypical for a Mott insulator because most Mott insulators
electrons has been extensively investigated for last two are antiferromagnetic as in LaTiO3 and V2O3.5,6) In Ca
decades.1–6) Various interesting phenomena, such as high-Tc substituted alloys Y1ꢁxCaxTiO3, the number of 3d electrons
superconductivity in cuprates and colossal magnetoresist- per Ti atom can be controled from 1 to 0 with increasing Ca
ance in manganites, have been discovered in the vicinity of content x from 0 to 1, and thereby hole carriers are doped at
the MI transition. Electron–electron interactions play the the Ti site. The previous studies of Y1ꢁxCaxTiO3 by using
central role in these phenomena. To understand the electron polycrystalline samples revealed that a ferromagnetic-to-
correlation effects on the MI transition, the Mott–Hubbard paramagnetic transition occurs at x ꢅ 0:15, while the system
model with the ratio U=t is often used, where U is the intra- remains insulating up to x ꢅ 0:4.3,4,18–23) Above x ꢅ 0:4, the
atomic Coulomb energy of a d-electron orbital and t is a system changes into a metallic state. For x ¼ 0:37{0:39, a
transfer integral between neighboring d-atoms.2) In the case temperature induced transition from the insulating state to
of U ꢃ t, the system is insulating due to the strong repulsion the metallic state occurs upon decreasing tempera-
of electrons, whereas in the case of U ꢄ t, the system ture.3,4,18–22) At the critical concentration xcr ¼ 0:39, the
becomes metallic. The orbital degree of freedom in addition electrical resistivity ꢀðTÞ drops by two order of magnitude at
to the charge and spin degrees of freedom plays an important 150 K.3,4,19–22) Upon approaching to xcr from the metallic
role in determining the electronic and magnetic properties of region, both the electronic specific-heat coefficient and Pauli
these materials. Among perovskite-type 3d transition-metal paramagnetic susceptibility are enhanced significantly.4,18)
oxides, the eg-electron system such as manganites, in which This enhancement was taken as an experimental evidence
t2g orbitals are fully occupied and eg orbitals are partially for the divergence in the carrier mass. Thus, the MI
filled, exhibits a large Jahn–Teller distortion in the octahe- transition in Y1ꢁxCaxTiO3 has been considered as a typical
dron due to a strong electron–lattice coupling. The resultant example of a Mott–Hubbard type transition.2)
1. Introduction
orbital ordering have attracted much attention.7) On the other
More recently, temperature dependent structural trans-
hand, the t2g-electron system such as titanates, in which t2g formations have been found in the vicinity of the MI
orbitals are partially filled and eg orbitals are unoccupied, transition. The high-resolution powder x-ray diffraction
also attracts a lot of attention. Despite the weak electron– experiments on samples with x ¼ 0:37 and 0.39 have
lattice coupling compared to the eg-system, orbital ordering revealed a phase transition from orthorhombic (space group
occurs in certain t2g-electron systems like in YVO3 and Pbnm) into monoclinic (P21=n) occurring at T ꢅ 230 K.24)
YTiO3.8–13)
For the monoclinic phase, the angle between the a and c axes
gradually increases to 90.15ꢆ.24) On further cooling below
100 K and 150 K for x ¼ 0:37 and 0.39, respectively, a low-
ꢀE-mail: iga@hiroshima-u.ac.jp
3182

J. Phys. Soc. Jpn., Vol. 72, No. 12, December, 2003
M. TSUBOTA et al.
3183
temperature orthorhombic phase (Pbnm, which will be decreased to 0 by the heat treatment at 400ꢆC for 4 h and
referred as LTO) appears and coexists with the monoclinic changed no more by the treatment in longer periods.
phase.24) For x ¼ 0:41, the LTO phase coexists with the However, heating above 400ꢆC resulted in the gradual
monoclinic phase even at room temperature.24) The volume formation of Y2Ti2O7. Based on this result, all samples were
fraction of the LTO phase increases with decreasing annealed at 400ꢆC for 4 h in air to recover the oxygen
temperature and saturates to the value of 25, 80 and 90% stoichiomerty.
for x ¼ 0:37, 0.39 and 0.41, respectively, at 10 K.24) Since
The single-crystal nature was confirmed by Laue photos,
the MI transition associated with such a phase separation is powder x-ray diffraction analysis and electron-probe micro-
beyond the conventional theory based on the Mott–Hubbard analysis (EPMA). The Laue pattern indicated that the crystal
model, further studies on the MI transition of Y1ꢁxCaxTiO3 structure is the orthorhombic GdFeO3-type (space group:
are necessary.
Pbnm) at room temperature. EPMA showed that there is no
From this view point, we have undertaken the investiga- deviation in Ca concentration x from the nominal value
tion of transport, magnetic and thermal properties of well within the absolute resolution of 1%. For the following
characterised single crystals of Y1ꢁxCaxTiO3 for 0 ꢂ x ꢂ measurements of physical properties, we chose homoge-
0:41. We have also studied pressure effects on both the neous parts where the spetial homogeneity in x is better than
resistive transition and structural transformation in the single 0.5%.
crystals of x ¼ 0:37 and 0.39. Application of pressure may
The electrical resistivity was measured by a conventional
increase the magnitude of the transfer integral t and thus four-probe method. The magnetic susceptibility and specific
decrease the ratio U=t. This provides an alternative method heat were measured using a commercial SQUID magnetom-
to control the MI transition, which is otherwise induced by eter (MPMS, Quantum Design) and a physical property
hole-doping. In fact, application of pressure of 1.7 GPa on a measurement system (PPMS, Quantum Design), respective-
polycrystalline sample with x ¼ 0:4 changed the semi- ly. For thermoelectric power measurement, a differential
conducting behavior in the resistivity to the metallic method was used. The sample was suspended between two
one.4,26,27) This large pressure effect indicated that the ratio electrically isolated copper blocks. Powder x-ray diffraction
U=t is very sensitive to the change in the volume. We will experiment at ambient pressure was carried out at BL02B2,
present the x–T and P–T phase diagrams of Y1ꢁxCaxTiO3, SPring-8.28) The electrical resistivity under pressures up to
based on the transport, magnetic and structural measure- 2 GPa was measured using a piston cylinder type cell with
ments on single crystal samples. By comparing these phase Daphne oil as the pressure transmitting medium. Pressure
diagrams, we discuss what is the driving force of the MI was determined from the known pressure dependence of the
transition in Y1ꢁxCaxTiO3.
electrical resistance of a manganin wire. Powder x-ray
diffraction experiment under pressure was carried out using
a diamond anvil cell (DAC) at BL10XU, SPring-8. Pressure
2. Experimental
Single crystals of Y1ꢁxCaxTiO3 (x ¼ 0, 0.1, 0.15, 0.2, 0.3, was generated in the DAC through a 1 : 1 mixture of
0.37, 0.38, 0.39, 0.41) were prepared by the following methanol/ethanol as pressure transmitting medium. During
methods. At the first step, the mixture of CaCO3 and TiO2 cooling and heating, the pressure was kept constant by the
was heated in an alumina crucible in the air at 1000ꢆC for fine control of the gas pressure loaded to the membrane
36 h. Thereby, the following reaction occurs,
which holds the gasket in the DAC. The absolute pressure
was determined from the shift of the fluorescence line of a
small piece of ruby enclosed in the DAC with the sample.
CaCO3 þ TiO2 ! CaTiO3 þ CO2 " :
Thus obtained CaTiO3 was mixed with Y2O3 and Ti2O3 The wavelength of an incident x-ray was selected to be
ꢀ
powders and stuffed into a rubber tube. The sample was 0.735 A not to overlap the large background due to the K-
pressed into a rod of 8 mm in diameter and 100 mm in edge scattering from yttrium in the sample.
length. The rod was inserted into a boron nitride tube and
3. Experimental Results
heated in a vacuum at about 1000ꢆC for 45 min. Thereby the
following reaction takes place,
3.1 Hole-doping effect
With the increase of x from 0 to 0.41 in Y1ꢁxCaxTiO3, the
electrical resistivity ꢀ at 300 K is decreased by four orders of
magnitude, as can be seen in Table I. Figure 1(a) presents the
temperature and x dependences of ꢀ for samples showing the
ð1 ꢁ xÞðY2O3 þ Ti2O3Þ þ xCaTiO3
y
! Y1ꢁxCaxTiO3ꢁy
þ
O2 " :
2
Powder x-ray diffraction analysis of the sintered rod showed MI transition. The electrical current direction was parallel to
the presence of a few % of raw materials and Y2Ti2O7. The the orthorhombic a axis. To examine the anisotropy of ꢀðTÞ,
rods were used to grow single crystals by a floating zone we have measured ꢀðTÞ for current directions I k a, I k b and
method being free from contamination from a crucible. In an I k c. As a result, no systematic differences were observed.
ellipsoidal mirror furnace with four halogen lamps, a molten
The high-temperature semiconducting behavior of ꢀðTÞ is
zone was scanned by a speed of 10 mm/h. The atmosphere described by the Arrhenius’ equation; ꢀðTÞ ¼ ꢀ1 exp ðEg=
was a mixture of Ar 50% and H2 50% which was flowed at a 2kBTÞ, where ꢀ1, Eg and kB are a coefficient, an energy gap
rate by 5 ‘/min. Typical size of a single crystal was of 8 mm and the Boltzmann constant, respectively. The magnitudes
diameter and 80 mm length. Thermogravimetric analysis of of Eg are listed in Table I. On cooling below ꢅ200 K, ꢀðTÞ
the as-grown crystal indicated the oxygen deficiency y in for x ¼ 0:38 and 0.39 shows a sharp peak which is followed
Y
1ꢁxCaxTiO3ꢁy to be 0.02–0.05. The analysis also showed by a drastic decrease. The thermal hysteresis is characteristic
that for both bulk and powdered samples the value of y was of a first order transition, which is coupled with lattice

3184
J. Phys. Soc. Jpn., Vol. 72, No. 12, December, 2003
M. TSUBOTA et al.
Table I. The parameters to describe the electrical resistivity ꢀðTÞ of
1ꢁxCaxTiO3. At low temperatures, ꢀðTÞ obeys the T2 law, ꢀðTÞ ¼ ꢀ0 þ
AT2. Above Tꢀ, ꢀðTÞ obeys the activation law with the energy gap Eg.
105
103
101
10-1
10-3
Y
x = 0.30
0.37
15
10
5
ꢀ (300 K)
(ꢁ cm)
ꢀ0
A
Tꢀ
Eg
x
(ꢁ cm)
(ꢁ cm/K2)
(K)
(K)
0.37
0.39
0.41
0.38
8
0.0
3:2 ꢈ 101
8:7 ꢈ 100
1:7 ꢈ 100
3:5 ꢈ 10ꢁ1
3:6 ꢈ 10ꢁ2
2:9 ꢈ 10ꢁ2
1:3 ꢈ 10ꢁ2
6:1 ꢈ 10ꢁ3
—
—
—
—
—
—
—
0
0
4
12
0.10
0.20
0.30
0.37
0.38
0.39
0.41
—
T
2 ( 103 K2 )
0.38
—
—
—
—
—
—
—
—
2:45 ꢈ 101
2:64 ꢈ 10ꢁ3
6:87 ꢈ 10ꢁ4
8:46 ꢈ 10ꢁ5
1:19 ꢈ 10ꢁ3
2:74 ꢈ 10ꢁ7
1:02 ꢈ 10ꢁ8
7:35 ꢈ 10ꢁ9
85
3600
2560
1600
—
0.39
145
165
—
0.41
I ll a (a)
10-5
1.5
x = 0.30
degree of freedom.3) With increasing x, the hysteresis width
gradually decreases. As for the MI transition temperature,
Tꢀ, we take the temperature where the data of ꢀðTÞ on
heating has the maximum. Thus determined Tꢀ’s are 85, 145
and 165 K for x ¼ 0:37, 0.38 and 0.39, respectively. The
ratio ꢀðTꢀÞ=ꢀð4:2KÞ for x ¼ 0:39 is ꢅ104, which is larger by
two orders of magnitude tha that reported for polycrystalline
samples.3,4,19–22) For x ¼ 0:41, by contrast, ꢀðTÞ shows a
metallic behavior without hysteresis in the whole temper-
ature region. At low temperatures, ꢀðTÞ for all samples
obeys the T2-law, ꢀ ¼ ꢀ0 þ AT2 as was observed in
polycrystalline samples.3,22,23) The residual resistivity ꢀ0
decreases from ꢅ20 ꢁ cm for x ¼ 0:37 to 10ꢁ4 ꢁ cm for
x ꢇ 0:39.
B = 1T ll a
0.37
0.38
1.0
0.5
0.39
0.41
(b)
0
80
60
40
20
0
0.41
0.39
x = 0.37
(c)
300
In Fig. 1(b), we show the temperature dependences of
magnetic susceptibility ꢁðTÞ of selected samples under a
magnetic field of 1 T applied parallel to the a axis. No
anisotropy in ꢁðTÞ was observed, for all paramagnetic
samples with x in the range 0:3 ꢂ x ꢂ 0:41. ꢁðTÞ for T >
200 K obeys the Curie–Weiss law, ꢁðTÞ ¼ Ch=ðT ꢁ ꢂph Þ,
where Ch is the Curie constant and ꢂph is the Weiss
temperature. The parameters obtained by a least squares fit
0
100
200
T ( K )
Fig. 1. Temperature dependences of (a) electrical resistivity, (b) magnetic
susceptibility and (c) volume fraction of LTO in Y1ꢁxCaxTiO3. Dashed
and solid lines show heating and cooling processes, respectively. The
inset shows the normalized resistivity ðꢀ ꢁ ꢀ0Þ=ꢀMAX vs T2, where ꢀ0 is
the residual resistivity and ꢀMAX is the maximum value below 300 K.
Normalized data for x ¼ 0:38, 0.39 and 0.41, are multiplied by 10000,
10000 and 100, respectively. Data of the volume fraction of the LTO
phase are taken from ref. 24.
are listed in Table II. The effective magnetic moment ꢃeff
was evaluated from Ch; ꢃeff
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼
3kBCh=NA, where NA is the
Avogadro number. In order to describe the data of ꢁðTÞ
below 50 K for 0:37 ꢂ x ꢂ 0:41, the inclusion of the
temperature independent term was necessary, i.e., ꢁðTÞ ¼ x ¼ 0:37, no peak appears in ꢁðTÞ, even though ꢀðTÞ has a
ꢁ‘0 þ C‘=ðT ꢁ ꢂp‘ Þ, where ꢁ0 is a constant term including broad peak at Tꢀ ¼ 100 K. For x ¼ 0:38 and 0.39, a broad
both diamagnetic and Pauli paramagnetic contributions. For peak appears in ꢁðTÞ at Tꢁ ¼ 170 K and 180 K, respectively.
Table II. The parameters to describe the magnetic susceptibility ꢁðTÞ of Y1ꢁxCaxTiO3. Tꢁ is the peak temperature. ꢃs is the average spin moment per Ti
atom evaluated by the localized spin model. ꢁðTÞ for T < 50 K obeys a modified Curie–Weiss law, ꢁðTÞ ¼ ꢁ‘0 þ C‘=ðT ꢁ ꢂp‘ Þ. At temperatures above Tꢁ,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ꢀðTÞ obeys a Curie–Weiss law, ꢁðTÞ ¼ Ch=ðT ꢁ ꢂph Þ. Each effective spin moment per Ti atom ꢃeff is calculated from ꢃeff
¼
3kBC=NA.
x
Tꢁ
ꢁ0‘
C‘
ꢂp‘
ꢃ‘eff
Ch
ꢂph
ꢃheff
ꢃs
ꢃheff=ꢃs
(K)
(10ꢁ4 emu/mol)
(emuK/mol)
(K)
(ꢃB)
(emuK/mol)
(K)
(ꢃB)
(ꢃB)
2 < T < 50 K
Tꢁ < T < 350 K
0.0
—
—
—
—
—
—
—
—
—
—
0.375
0.316
0.298
0.257
0.239
0.236
0.229
—
29.8
1.73
1.59
1.54
1.43
1.38
1.37
1.35
—
1.73
1.64
1.55
1.45
1.37
1.36
1.35
1.33
1.00
0.97
0.99
0.97
1.01
1.01
1.00
—
0.10
0.20
0.30
0.37
0.38
0.39
0.41
22.0
7.5
—
—
—
—
—
—
—
—
—
—
ꢁ1:7
ꢁ10:8
ꢁ38:0
ꢁ49:4
—
—
8.48
7.46
4.25
3.78
0.0414
0.0162
0.00566
0.00512
ꢁ2:46
ꢁ2:39
ꢁ2:55
ꢁ2:46
0.18
0.070
0.025
0.023
150
175
250

J. Phys. Soc. Jpn., Vol. 72, No. 12, December, 2003
M. TSUBOTA et al.
3185
100
101
100
Y
0.615Ca0.385TiO3
Y0.61Ca0.39TiO3
(a)
80
60
40
20
0
I ll a
30
20
10
0
10-1
10-2
10-3
10-4
P = 0 GPa
0.16
0.47
0.57
0.75
0.96
0
50
100
T 2 ( K2
)
101
100
10-1
10-2
10-3
0
-20
-40
(b)
0
100
200
T ( K )
300
Fig. 3. Temperature dependences of electrical resistivity in
Y0:61Ca0:39TiO3 under various constant pressures up to 1 GPa. Dashed
and solid lines show heating and cooling processes, respectively. With
increasing pressure above 150 K, ꢀðTÞ rapidly decreases and shows a
metallic behavior, while ꢀðTÞ is almost independent of pressure below
150 K.
0
100
200
300
T ( K )
Fig. 2. Temperature dependences of (a) specific heat Cp, (b) thermopower
S and resistivity ꢀ for Y0:615Ca0:385TiO3 on heating process. CpðTÞ shows
three anomalies as indicated by arrows. The inset shows Cp=T as a
function of T2.
ison, we have chosen single crystalline samples with x ¼
0:37 and 0.39 for the resistivity measurements under
pressure. Here, it should be noted that the samples were
The values of Tꢁ are almost same as those of Tꢀ.
Figure 1(c) represents the temperature dependences of the cut from a different rod from that used for the measurements
volume fraction of the LTO phase estimated from powder x- of ꢀðTÞ and ꢁðTÞ under ambient pressure.
ray diffraction analysis.24) We define the phase separation
Figure 3 represents the results of ꢀðTÞ under various
temperature TPS as the temperature in which the volume pressures for x ¼ 0:39. The sharp transition at Tꢀ ¼ 165 K
fraction of LTO is 10% of its saturated value.
under ambient pressure is changed to a gradual transition at a
We measured the specific heat CpðTÞ and thermopower low pressure of 0.16 GPa. Here, no maximum is observed,
SðTÞ for the sample with x ¼ 0:385 in which the drastic MI and for convenience we take the transition temperature
transition was observed at Tꢀ ¼ 147 K. Figure 2 shows the TꢀðPÞ as the inflection point in the ꢀðTÞ curve. Upon
temperature dependences of Cp, S and ꢀ. At about 240 K, increasing pressure up to 0.96 GPa, Tꢀ increases and the
CpðTÞ has a small jump which is associated with the value of ꢀ ðT ¼ 280 KÞ decreases by one order of magnitude.
structural transition from the orthorhombic phase to the It is noteworthy that in the metallic state below 150 K, ꢀðTÞ
monoclinic phase.24) CpðTÞ shows a small peak at 165 K, does not depend on pressure, and obeys the T2-law below
which agrees with the onset temperature of the phase 70 K. The overall temperature dependence at P ¼ 0:96 GPa
separation as determined by powder x-ray diffraction.24) At resembles that for x ¼ 0:41 at P ¼ 0 GPa shown in Fig. 1(a).
around 150 K where the MI transition takes place, the CpðTÞ
For the sample of x ¼ 0:37, we show in Fig. 4 the
curve bends downward without showing a marked peak. The temperature dependences of ꢀðTÞ and volume fraction of the
Cp=T vs T2 plots shown in the inset of Fig. 2(a) allow us to LTO phase under various pressures. For the resistivity
estimate the electronic specific-heat coefficient ꢄ ¼ measurements, we used a clamp-type piston–cylinder pres-
17:4 mJ/molK2, being comparable with the value reported sure cell, in which the pressure decreased by about 0.1 GPa
for polycrystalline sample for x ¼ 0:4.4)
on cooling due to the solidification of the pressure trans-
In Fig. 2(b), SðTÞ turns off very sharply at 165 K and mitting medium. The correction of the pressure value was
changes sign from negative to positive on cooling. The made by using the known pressure dependence of resistance
drastic change in both CðTÞ and SðTÞ at TPS ¼ 165 K in a manganin wire.25) In Fig. 4(a), the pressure value
confirms the bulk nature of the phase separation, which sets estimated at Tꢀ in the heating process is noted. Upon
in just before the resistivity reaches the maximum on increasing pressure, Tꢀ increases steadily, while the residual
cooling.
resistivity ꢀ0 decreases by three orders of magnitude in the
narrow range 0:44 ꢂ P ꢂ 0:58 GPa. It should be noted that
the sharp transition in ꢀðTÞ at 0.84 GPa is similar to that of
3.2 Pressure effect
The previous study of pressure effect on the MI transition x ¼ 0:39 at P ¼ 0 (see Fig. 3). The width of hysteresis
in polycrystalline samples of Y1ꢁxCaxTiO3 (x ¼ 0:375, 0.39 decreases and vanishes at P ¼ 1:2 GPa. At higher pressure,
and 0.40) showed that application of pressure of about 1 GPa the system behaves metallic in the whole temperature
changes the system into a metallic state.4,26,27) For compar- region, and ꢀðTÞ obeys the T2-law below 70 K.

3186
J. Phys. Soc. Jpn., Vol. 72, No. 12, December, 2003
M. TSUBOTA et al.
x ¼ 0:41 just above xhc, in fact, a metallic state is stabilized
even at 290 K.
102
Y
0.63Ca0.37TiO3
P = 0 GPa
We pay our attention to the metallic phase at low
temperature, where ꢀðTÞ follows the T2 law. From the values
of the residual resistivity ꢀ0 and coefficient A listed in
Table I, one notice that both A and ꢀ0 rapidly decrease by
four orders of magnitude between x ¼ 0:37 and 0.38. In
order to consider the reason for the large drop of ꢀ0, we
recall that the metallic LTO phase appears at 100 K and
150 K for x ¼ 0:37 and 0.39, respectively.24) For x ¼ 0:41,
however, the volume fraction of the LTO phase is 60% at
room temperature.24) The volume fraction of the LTO phase
increases on cooling and saturates to 20, 80 and 90% for
x ¼ 0:37, 0.39 and 0.41, respectively.24) By comparing the
results of ꢀðTÞ and the volume fraction for x ¼ 0:39 in Figs.
1(a) and 1(c), we find that the ꢀðTÞ starts to drop when the
LTO phase appears. Similarly sharp drop in ꢀðTÞ for x ¼
0:37 occurs under pressures at P ¼ 0:58 and 0.84 GPa, as the
LTO phase develops. When the volume fraction of the LTO
phase is saturated, then ꢀðTÞ obeys the T2 law.
I ll a
101
100
0.26
0.44
0.58
10-1
10-2
10-3
0.84
1.2
1.5
1.8
(a)
100
80
60
40
20
0
1.9
1.7
0.8
0.6
In certain manganites and cuprates, phase separations into
metallic and insulating domains are known to be responsible
for the transition from an insulating state to a metallic
state.29) Monte Carlo simulations of electrical resistivity of
such systems using two or three dimensional clusters
indicated that the resistivity behaves metallic when the
volume fraction of metallic phase amounts to 20–30%.30–32)
In the present system Y1ꢁxCaxTiO3, the percolation of the
metallic LTO phase may lead to the T2 behavior in ꢀðTÞ.
We discuss the hole-doping effect on ꢁðTÞ by analyzing
the parameters listed in Table II. At high temperatures,
where the whole sample is in a single phase, ꢁðTÞ obeys a
Curie–Weiss law, ꢁðTÞ ¼ Ch=ðT ꢁ ꢂph Þ. By using the Curie
P = 0 GPa
(b)
0
100
200
300
T ( K )
Fig. 4. Temperature dependences of (a) electrical resistivity and (b)
volume fraction of the LTO phase in Y0:63Ca0:37TiO3 under pressures up
to ꢅ2 GPa. Dashed and solid lines show heating and cooling processes,
respectively.
constant Ch, the effective magnetic moment ꢃeff per Ti atom
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
The volume fraction of the LTO phase under pressure was was evaluated as ꢃehff
¼
3kBCh=NA. Based on the assump-
estimated by powder x-ray diffraction experiments with the tion of the valence states Y31þꢁxCax2þ(Ti13þꢁxTi4xþ)O3, on the
method described in ref. 24. At P ¼ 0, the LTO phase other hand, the average spin moment ꢃs per Ti atom is
pffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
appears below 120 K, and the volume fraction increases on calculated as ꢃs ¼ 1 ꢁ xꢈ2 SðS þ 1Þ, where S ¼ 1=2 for
cooling and saturates to a value of 25% of the whole sample. the Ti3þ ion. From ꢃehff and ꢃs, the ratio ꢃheff=ꢃs was
We define the phase separation temperature TPS by the same evaluated and listed in Table II. The ratio is nearly 1 in the
way as at ambient pressure. With increasing pressure up to whole range. This fact confirms that x holes per formula unit
0.8 GPa, both TPS and the volume fraction increase. At P ¼ is really doped by the Ca substitution in the insulating region
1:7 GPa, TPS eventually exceeds room temperature and the (0 ꢂ x ꢂ 0:39) In the metallic region x ꢇ 0:4, the doped hole
LTO phase is stabilized in the whole volume.
x correspondes to the carrier number, n ¼ 1 ꢁ x, which was
first pointed out from the x dependence of the Hall
coefficient.4)
4. Discussion
4.1 Hole-doping effect
At temperatures below 50 K, the samples for x ꢇ 0:37 are
First, we discuss the hole-doping effect on ꢀðTÞ of separated into the monoclinic and the LTO phases. When the
Y
1ꢁxCaxTiO3. A simple picture for the MI transition based volume fraction of the LTO is saturated, ꢁðTÞ obeys a
on the Mott–Hubbard model is as following. If x holes per modified Curie–Weiss low, ꢁðTÞ ¼ ꢁ‘0 þ C‘=ðT ꢁ ꢂp‘ Þ. In
formula unit are doped, the number of 3d electrons per Ti Table II, we represent the parameters ꢁ‘0, C‘, ꢂp‘ and ꢃe‘ff. It
ion n would decrease as n ¼ 1 ꢁ x. At a small x, say 0.05 as should be noted that ꢃ‘eff=ꢃehff is much less than 1, and the
was observed in La1ꢁxSrxTiO3,18) the holes produced in the values are 0.17, 0.069 and 0.025 for x ¼ 0:37, 0.38 and 0.39,
lower Hubbard band would become mobile, then the respectively. This significant decrease in ꢃ‘eff=ꢃehff can be
insulating state changes to a metallic state. However, the understood if the localized moments in the insulating phase
transition in Y1ꢁxCaxTiO3 occurs at a large value x ꢅ 0:39, above Tꢀ are reduced in the monoclinic phase below Tꢀ. On
as shown in Fig. 1. For the insulating phase, the energy gap going from x ¼ 0:37 to 0.41, the value of ꢁ‘0 is decreased to
Eg estimated from ꢀðTÞ is listed in Table I. The magnitude of less than half. This x dependence of ꢁ‘0 can be smoothly
Eg as a function of x decreases linearly for 0:37 ꢂ x ꢂ 0:39, connected to the previous result for x ꢇ 0:4.3,18) This drop in
and the linear extrapolation of EgðxÞ to larger x region leads ꢁ‘0 for x ꢇ 0:4 indicates the reduction in the electron
to a critical concentration xhc ¼ 0:405 where Eg vanishes. At correlation when the system changes into the metallic phase.

J. Phys. Soc. Jpn., Vol. 72, No. 12, December, 2003
M. TSUBOTA et al.
3187
transition may remain until x is decreased to x ¼ 0:33 at
ambient pressure. This critical concentration is much larger
than xc ¼ 0:05 for a typical Mott transition system
La1ꢁxSrxTiO3.18)
P ( GPa ) for x = 0.39
0
0.5
1.0
300
250
200
150
100
One possible reason for the large difference in the critical
concentration between 0.33 in Y1ꢁxCaxTiO3 and 0.05 in
La1ꢁxSrxTiO3 is the difference in the charge gap in the host
compounds. Another reason may be the orbital ordering in
the former. It is known that the mother material LaTiO3 for
La1ꢁxSrxTiO3 is an antiferromagnetic insulator without
orbital ordering, then a metallic state is easily realized by
hole-doping at a low level.35) In contrast, the mother material
YTiO3 for Y1ꢁxCaxTiO3 is a ferromagnetic insulator with
orbital ordering, then the insulating state might be stable as
long as the orbital order remains.10–13) If this is the case, the
orbital ordered state may be retained up to x ¼ 0:33. To
observe the orbital ordered state in the present system,
resonant x-ray scattering experiments have been performed.
The results will be reported elsewhere.
HTO
Mono
LTO + Mono
1.5 2.0
0
0.5
1.0
P ( GPa ) for x = 0.37
0.37 0.38 0.39 0.40 0.41
x in Y1-xCaxTiO3
5. Summary
Fig. 5. Pressure dependences of TꢀðPÞ and TPSðPÞ for x ¼ 0:37 and 0.39,
and x dependences of Tꢀ at ambient pressure for Y1ꢁxCaxTiO3. Open (
In order to study the mechanism of the MI transition in
)
Y
1ꢁxCaxTiO3 at x ꢅ 0:4, we compared the hole-doping
and closed ( ) circles show TꢀðPÞ and TPSðPÞ for x ¼ 0:37, respectively.
Open squares ( ) represent TꢀðPÞ for x ¼ 0:39, and the zero point of
TꢀðPÞ for x ¼ 0:39 is shifted by 0.82 GPa higher than that of TꢀðPÞ for
x ¼ 0:37. Crosses (ꢈ) show x dependence of Tꢀ at ambient pressure in
Y1ꢁxCaxTiO3. Closed triangles ( ) represent the temperature of the
phase transition from the high temperature orthorhombic phase (HTO) to
the monoclinic one.
effect with the pressure effect on the MI transition. By using
single crystalline sample for 0:3 ꢂ x ꢂ 0:41, we measured
electrical resistivity, magnetic susceptibility, specific heat
and thermopower and powder x-ray diffraction under
pressure. The significant decrease in magnetic susceptibility
associated with the MI transition indicates that the localized
3d electrons at the Ti site in the insulating phase are
4.2 Comparison between hole-doping effect and pressure delocalized in the two-phase state. Two characteristic
effect
temperatures, MI transition temperature Tꢀ and phase
In this section, we compare the hole-doping effect with separation temperature TPS were determined based on the
the pressure effect on both the MI transition and phase results of electrical resistivity and x-ray diffraction, respec-
separation in Y1ꢁxCaxTiO3. In Fig. 5, we represent four sets tively. The data were used to draw the P–T and x–T phase
of data of characteristic temperatures; the pressure depend- diagrams. It is found that TPS agrees well with Tꢀ, and ꢀðTÞ
ence of Tꢀ and TPS for x ¼ 0:37, the pressure dependence of starts to decrease when the metallic LTO phase appears as a
Tꢀ for x ¼ 0:39, and x dependence of Tꢀ for 0:37 ꢂ x ꢂ 0:41 result of phase separation. The P–T and x–T phase diagrams
at ambient pressure. Here, the origin at P ¼ 0 for x ¼ 0:39 is indicate that both hole-doping and application of pressure
shifted upward by 0.82 GPa with respect to that for x ¼ 0:37 stabilize the metallic state. The effect of hole doping by 1%
so that the curve of TꢀðPÞ for x ¼ 0:39 lies on the curves of Ca substitution on the MI transition is equivalent to the
TꢀðPÞ and TPSðPÞ for x ¼ 0:37. This agreement of the two effect of application of pressure of 0.41 GPa. From these
sets of data implies that the effect of hole-doping of Ca 2% results, we conclude that the MI transition in Y1ꢁxCaxTiO3 is
on the MI transition corresponds to that of application of not a simple Mott–Hubbard type but is caused by the
pressure of 0.82 GPa. This correspondence was used to percolation of the metallic LTO domains. Further structural
normalize the horizontal scale x in order to plot the data of study of this system is necessary to understand why the LTO
TꢀðxÞ for 0:37 ꢂ x ꢂ 0:41 together in Fig. 5. Now, all the phase is metallic while the high-temperature orthorhombic
data of TꢀðPÞ, TPSðPÞ and TꢀðxÞ agree well. Thus, both hole- phase with the same space group Pbnm is insulating.
doping and application of pressure have equivalent effects
Acknowledgements
on both the MI transition and phase separation in
Y
1ꢁxCaxTiO3. Similar correspondence was reported in
We thank Y. Shibata for electron-probe microanalysis
manganite performed at N-BARD and H. Matsuhata and H. Nakao for
another titanate LaTiO3þꢅ and in
a
La1ꢁxSrxMnO3, where both hole-doping and application of fruitful discussions. This work was supported by the COE
pressure stabilize the metallic state as a common result of Research (13CE2002), the Basic Scientific Research
the decrease in the ratio U=t.33,34)
(No.10440110) in a Grant-in-Aid from the MEXT. The
Powder x-ray diffraction analysis has shown that the MI synchrotron x-ray diffraction was done at the BL02B2 and
transition in Y1ꢁxCaxTiO3 is a result of percolation of the BL10XU in SPring-8 under Proposal No. 2001B2004-LD,
metallic LTO domains. In Fig. 5, the linear extrapolation of 2001B0229, 2000B0053, 2001A0172 and 2001A0174.
the data of TꢀðxÞ and TPSðxÞ to smaller x region leads to the
lower critical concentration xc ꢅ 0:33 where TꢀðxÞ or TPSðxÞ
vanishes. This fact suggests that the phase separation and MI

3188
J. Phys. Soc. Jpn., Vol. 72, No. 12, December, 2003
M. TSUBOTA et al.
Tokura: Phys. Rev. B 48 (1993) 7636.
19) F. Iga, T. Nishiguchi and Y. Nishihara: Physica B 206&207 (1995)
859.
1) N. F. Mott: Metal–Insulator Transitions (Taylor & Francis, London,
1985) 2nd Ed.
2) M. Imada, A. Fujimori and Y. Tokura: Rev. Mod. Phys. 70 (1998)
1039.
20) F. Iga, Y. Nishihara, J. Sakurai and M. Ishikawa: Physica B 237&238
(1997) 14.
3) Y. Taguchi, Y. Tokura, T. Arima and F. Inaba: Phys. Rev. B 48 (1993)
511.
21) F. Iga, N. Shirakawa, K. Murata, Y. Nishihara, T. Ishii, G. Oomi and
Y. Uwatoko: Bull. Electrotech. Lab. 59 (1995) 459.
22) F. Iga, T. Naka, T. Matsumoto, N. Shirakawa, K. Murata and Y.
Nishihara: Physica B 223&224 (1996) 526.
4) Y. Tokura, Y. Taguchi, Y. Moritomo, K. Kumagai, T. Suzuki and Y.
Iye: Phys. Rev. B 48 (1993) 14063.
ˆ
5) Y. Taguchi, T. Okuda, M. Ohashi, C. Murayama, N. Mori, Y. Iye and
23) F. Iga, H. Fukuchi, I. H. Inoue and Y. Nishihara: Proc. 9th Int. Conf.
Ternary and Multinary Compounds, Yokohama, 1993, Jpn. J. Appl.
Phys. 32 (1993) Suppl. 32-3, p. 332.
Y. Tokura: Phys. Rev. B 59 (1999) 7917.
6) S. A. Carter, T. F. Rosenbaum, J. M. Honig and J. Spalek: Phys. Rev.
Lett. 67 (1991) 3440.
24) K. Kato, E. Nishibori, M. Takata, M. Sakata, T. Nakano, K. Uchihira,
M. Tsubota, F. Iga and T. Takabatake: J. Phys. Soc. Jpn. 71 (2002)
2082.
7) Y. Murakami, J. P. Hill, D. Gibbs, M. Blume, I. Koyama, M. Tanaka,
H. Kawata, T. Arima, Y. Tokura, K. Hirota and Y. Endoh: Phys. Rev.
Lett. 81 (1998) 582.
25) E. S. Itskevich: Cryogenics (1964) 365.
26) T. Naka, T. Matsumoto, A. Matsushita, F. Iga and Y. Nishihara:
Physica B 304 (2001) 27.
8) M. Noguchi, A. Nakazawa, S. Oka, T. Arima, Y. Wakabayashi, H.
Nakao and Y. Murakami: Phys. Rev. B 62 (2000) R9271.
9) T. Mizokawa, D. I. Khomskii and G. A. Sawatzky: Phys. Rev. B 60
(1999) 7309.
27) T. Naka, T. Matsumoto, F. Iga and Y. Nishihara: Rev. High Pressure
Sci. Technol. 7 (1998) 608.
10) H. Ichikawa, J. Akimitsu, M. Nishi and K. Kakurai: Physica B
281&282 (2000) 482.
28) E. Nishibori, M. Takata, K. Kato, M. Sakata, Y. Kubota, S. Aoyagi, Y.
Kuroiwa, M. Yamakata and N. Ikeda: J. Phys. Chem. Solids 62 (2001)
2095.
11) J. Akimitsu, H. Ichikawa, N. Eguchi, T. Miyano, M. Nishi and K.
Kakurai: J. Phys. Soc. Jpn. 70 (2001) 3475.
29) H.-D. Jostarndt, U. Walter, J. Harnischmacher, J. Kalenborn, A.
Severing and E. Holland-Moritz: Phys. Rev. B 46 (1992) 14872.
30) J. Burgy and E. Dagotto and M. Mayr: Phys. Rev. B 67 (2003) 014410.
´
31) M. Mayr, A. Moreo, J. A. Verges, J. Arispe, A. Feguin and E. Dagotto:
Phys. Rev. Lett. 86 (2001) 135.
12) M. Itoh, M. Tsuchiya, H. Tanaka and K. Motoya: J. Phys. Soc. Jpn. 68
(1999) 2783.
13) H. Nakao, Y. Wakabayashi, T. Kiyama, Y. Murakami, M. v.
Zimmermann, J. P. Hill, D. Gibbs, S. Ishihara, Y. Taguchi and Y.
Tokura: Phys. Rev. B 66 (2002) 184419.
32) K. Yoshida: Phys. Rev. B 60 (1999) 9325.
33) A. Urushibara, Y. Moritomo, T. Arima, A. Asamitsu, G. Kido and Y.
Tokura: Phys. Rev. B 51 (1995) 14103.
14) J. E. Greedan: J. Less-Common Met. 111 (1985) 335.
15) J. D. Garrett, J. E. Greedan and D. A. MacLean: Mater. Res. Bull. 16
(1981) 145.
34) Y. Moritomo, A. Asamitsu and Y. Tokura: Phys. Rev. B 51 (1995)
16491.
16) J. P. Goral, J. E. Greedan and Maclean: J. Solid State Chem. 43 (1982)
244.
35) B. Keimer, D. Casa, A. Ivanov, J. W. Lynn, M. v. Zimmermann, J. P.
Hill, D. Gibbs, Y. Taguchi and Y. Tokura: Phys. Rev. Lett. 85 (2000)
3946.
17) M. Tsubota, F. Iga, T. Takabatake, N. Kikugawa, T. Suzuki, I. Oguro,
H. Kawanaka and H. Bando: Physica B 281&282 (2000) 622.
18) K. Kumagai, T. Suzuki, Y. Taguchi, Y. Okada, Y. Fujishima and Y.