Phase transitions and electrical transport in the mixed-valence V 2+/V3+ oxide BaV10O15

Article abstract of DOI:10.1016/j.jssc.2003.10.025

BaV₁₀O₁₅ can be regarded as a Ba-doped V ₂O₃ in which the Ba²⁺ ions substitute in the O^2- close-packed layers. The Ba²⁺ ions order within these layers and direct the occupation of the octahedral sites by V²⁺ and V³⁺ ions resulting in a structure with subtle differences from that of V₂O₃ which can be described in Cmca at room temperature. Magnetic susceptibility data show evidence for two phase transitions at 135 and 40K. The higher temperature transition at 135K is shown to be structural in origin to another orthorhombic form, Pbca. The structural transition temperature, T_s, decreases with decreasing V²⁺ content to a minimum value of 105K. Crystallographic and DSC data support a first-order transition driven by partial bond formation which results in a 7% reduction in the distance between two of the five crystallographically distinct V atoms. There is no conclusive crystallographic evidence for V ²⁺/V³⁺ charge ordering in either the Cmca or Pbca forms. Electrical conductivity data show semiconducting behavior above T_s which can be fitted to a small polaron hopping model for the most reduced samples (T_s=135K). The same sample shows a sharp but not discontinuous decrease in conductivity below T_s, consistent with carrier removal due to bond formation. More oxidized materials with T _s=105K show a more subtle anomaly. Evidence for correlated (Efros-Shklovskii) variable-range hopping at low temperatures is seen in the T_s=105K sample from analysis using a Hill-Zabrodski (logdE vs. logT) plot. Thermopower data on the T_s=135K material show an anomalously small value of S～+1μV/K at room temperature which increases to >+200μV/K upon cooling to 90K. Plots of dS/dT show evidence for the T _s=135K phase transition. These results are not consistent with a simple one carrier model for small polaron hopping assuming that the V ²⁺ ions are the carriers, which would predict S～-100μV/K, but seem to demand a two carrier model with n～p at room temperature for which n-type carriers are trapped as a result of bond formation at the phase transition as temperature is lowered. The lower temperature phase transition near 40K is magnetic in origin and will be discussed in a subsequent publication.

Full text of DOI:10.1016/j.jssc.2003.10.025

ARTICLE IN PRESS
Journal of Solid State Chemistry 177 (2004) 1098–1110
Phase transitions and electrical transport in the mixed-valence
2
V /V oxide BaV O
+
3+
10 15
ꢀ
C.A. Bridges and J.E. Greedan
Department of Chemistry and Brockhouse Institute for Materials Research McMaster University, 1280 Main St. w., Hamilton, Ont., Canada L8S 4M1
Received 21 July 2003; received in revised form 8 October 2003; accepted 14 October 2003
Abstract
2
+
2ꢁ
2+
BaV10
order within these layers and direct the occupation of the octahedral sites by V and V ions resulting in a structure with subtle
differences from that of V which can be described in Cmca at room temperature. Magnetic susceptibility data show evidence for
two phase transitions at 135 and 40 K. The higher temperature transition at 135 K is shown to be structural in origin to another
O15 can be regarded as a Ba-doped V O in which the Ba ions substitute in the O close-packed layers. The Ba ions
2 3
2
+
3+
2 3
O
2
+
orthorhombic form, Pbca. The structural transition temperature, Ts; decreases with decreasing V content to a minimum value of
05 K. Crystallographic and DSC data support a first-order transition driven by partial bond formation which results in a 7%
reduction in the distance between two of the five crystallographically distinct V atoms. There is no conclusive crystallographic
1
2
+
3+
evidence for V /V charge ordering in either the Cmca or Pbca forms. Electrical conductivity data show semiconducting behavior
above Ts which can be fitted to a small polaron hopping model for the most reduced samples (Ts ¼ 135 K). The same sample shows
a sharp but not discontinuous decrease in conductivity below Ts; consistent with carrier removal due to bond formation. More
oxidized materials with Ts ¼ 105 K show a more subtle anomaly. Evidence for correlated (Efros-Shklovskii) variable-range hopping
at low temperatures is seen in the Ts ¼ 105 K sample from analysis using a Hill-Zabrodski (log dE vs. log T) plot. Thermopower
data on the Ts ¼ 135 K material show an anomalously small value of SB þ 1 mV/K at room temperature which increases to
4
+200 mV/K upon cooling to 90 K. Plots of dS=dT show evidence for the Ts ¼ 135 K phase transition. These results are not
2+
consistent with a simple one carrier model for small polaron hopping assuming that the V
ions are the carriers, which would
predict SB ꢁ 100 mV/K, but seem to demand a two carrier model with nBp at room temperature for which n-type carriers are
trapped as a result of bond formation at the phase transition as temperature is lowered. The lower temperature phase transition near
40 K is magnetic in origin and will be discussed in a subsequent publication.
r 2003 Elsevier Inc. All rights reserved.
2+
3+
Keywords: V /V mixed valence oxide; First order crystallographic phase transition; Partial V–V bond formation; Electrical conductivity;
Magnetic susceptibility; Thermoelectric power; Electron trapping; Small polaron hopping; Variable range hopping
1
. Introduction
simplicity, V O exhibits a rich variety of properties
2 3
including a spectacular metal/insulator transition below
B160 K associated with a first-order crystallographic
Transition metal oxides which exhibit strong electro-
%
phase transition from R3c to I2=a [1,2]. The monoclinic
nic correlation have attracted extraordinary attention
over the past few decades. One material which has been
central to experimental and theoretical studies of
correlated systems is V O for which a literature of
form is insulating and antiferromagnetic [3] and is often
regarded as a canonical Mott-Hubbard insulator [4–6],
though recent results indicate a more complexpicture
[7]. The material shows a remarkable sensitivity to
applied pressure and doping with small amounts of
2
3
more than 3000 publications exists. This material
crystallizes in the well-known corundum structure,
which consists of a hexagonal close-packed sublattice
3
+
adjacent ions, especially Cr [8,9].
of oxide ions in which V occupies 2 of the octahedral
3
+
As vanadium exhibits a wide range of stable oxidation
states in oxides, from 2+ to 5+, a large number of
mixed-valence vanadium oxides exist, many of which
3
sites. Surprisingly, given its apparent crystallographic
4
+
ꢀ
show phenomena of current interest. NaV O , a V
5
/
material, shows both charge ordering and spin gap
Corresponding author. Fax: +905-522-2773.
E-mail address: greedan@mcmail.cis.mcmaster.ca (J.E. Greedan).
2
5
+
V
0022-4596/$ - see front matter r 2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.jssc.2003.10.025

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1099
1
8
formation at low temperatures [10]. The binary oxides,
V O and V O show both metal/insulator transitions
and charge (V /V ) ordering [11,12].
occupying of the sites in layers of composition BaO7
which alternate with O layers [27]. The stacking of the
4
7
5
9
+
8
3
4+
layers is modified from strict hcp as in V O to
2
3
Other vanadium oxides with related structures and
chemistry are worthy of interest but have been less well
studied. A number of materials with structures derived
from magneto-plumbite have been reported with com-
position MV O (M ¼ Na; K; Sr; Pb) and hollandite-
alternating hcp and cubic close-packing, ccp, Fig. 1a.
C
H
BaO7
O8
6
11
based phases, Ba V O and BaV
O , for example,
x
8
16
10ꢁx 17
[
13–18]. NaV O is the best studied member of this
6 11
series and shows a range of interesting properties
including crystallographic phase transitions and metallic
weak ferromagnetism [19–21]. A partial form of
magnetic frustration is also known, as one set of V sites
BaO7
O8
C
(
lattice [22]. All of the above materials involve mixed-
out of three) in the crystal structure forms a Kagome
H
C
4
+
3+
valence V /V , both of which are common oxidation
states for vanadium oxides.
BaO7
(a)
Considerably less is known about mixed valence
vanadium oxides involving the rare +2 oxidation state.
To our knowledge only three such ternary oxides are
2
+
3+
known, each with the V /V combination: the spinel
AlV O with average oxidation state +2.5 [23],
MV O with average oxidation state +2.62 [24,25],
2
4
1
3
18
and MV O [26,27] with average oxidation state
1
0
15
+
the latter two can be described in terms of close-packed
2.80, where M ¼ Ba or Sr. The crystal structures of
2
+
layers of Ba and O with vanadium ions occupying
2ꢁ
%
octahedral sites, with different symmetries, R3 for
MV O and Cmca for MV O . The study of the
18
1
3
10 15
physical properties of these materials is in the very early
stages. AlV O shows a structural phase transition from
2
4
%
Fd3m to R3m below 700 K and accompanying small
anomalies in both resistivity and susceptibility. The
phase transition is ascribed to a very weak charge
(b)
+
2.5ꢁd +2.5+3d
ordering described as V
23]. Both BaV O and SrV O are described as n-
/V
, where dB0:1
[
type semiconductors but the resistivity values are in the
1
3
18
13 18
mO cm range and the Seebeck coefficients range from
ꢁ
10 to ꢁ50 mV/K. The magnetic susceptibilities are
significantly smaller than expected based on a local
moment model for the V ions. A coherent understanding
of the properties of these materials is not yet available.
Turning to the other series, MV O , more specifi-
1
0
15
cally BaV O , which is the subject of this report, a
15
1
0
more detailed description of its crystal structure and the
close relationship to that of V O is in order. BaV O
10 15
2
3
can be considered to result from ‘‘doping’’ of V O with
2
3
2
Ba . One obvious result of such doping is the
+
3
+
2+
ions to V , a rare
reduction of 20% of the V
oxidation state for this element in oxides as already
mentioned. As well, due to the negligible difference in
(
c)
2
+
2ꢁ
Fig. 1. (a) Stacking of the BaO
7
and O
8
close-packed layers normal to
˚
effective radii, 1.42 A for both Ba
and O in VIII-
the c-axis in an alternating hcp and ccp sequence. (b) Occupation of the
V octahedral sites within a close-packed layer. (c) The V sublattice in
two layers showing the presence of V10 clusters and the condition for
geometric magnetic frustration, i.e., edge sharing tetrahedra.
fold coordination [28], the barium ions substitute in the
closed-packed oxide sublattice. Due to electrostatic
2
+
forces, the Ba ions order within a close-packed layer

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1100
2
+
Electrostatic repulsion, due the presence of Ba in the
close-packed layers, directs the occupation of the
were measured using an optical pyrometer. For most of
the runs the sample was heated to the reaction
temperature over 7 h, held for 17 h and quenched by
turning off the RF generator. This approach generally
resulted in small grained polycrystalline samples. To
encourage the growth of crystals a variant was used in
which the crucible was cooled over 6 h from the reaction
temperature to 1400 C, followed by quenching. This
approach resulted in crystals of dimension 0.1–0.5 mm
which were suitable for single crystal X-ray diffraction.
3
+
2+
octahedral sites by the V and V ions, as shown in
Fig. 1b. The possibility of charge ordering on the
octahedral sites arises due the occurrence two-oxidation
states of vanadium, an issue which will be addressed in
the following section. As mentioned, a very slight degree
of charge ordering has been claimed for the spinel
AlV O [23]. The other remarkable feature of this
structure type is the topology of the M sublattice shown
in Fig. 1c, which is seen to consist of interconnected M10
clusters of edge-shared tetrahedra and thus provides a
necessary condition for geometric magnetic frustration.
Evidence for frustrated magnetism has been demon-
strated for BaCr O [27]. Thus, BaV O represents a
material with the potential for highly correlated
electronic behavior, charge ordering and geometric
magnetic frustration. In this work a remarkable
structural phase transition will be investigated by means
of a detailed crystallographic study, electrical conduc-
tivity and thermopower measurements. Subsequent
papers will describe the effects of chemical substitutions
and unusual magnetic properties associated with the
geometric frustration.
ꢂ
2
4
2.2. Chemical analysis
The samples were analyzed by gravimetric oxidation
1
0
15
10 15
using a Netzsch STA409 to measure the weight gain in
air. As well, the Ba:V ratio was measured by Inductively
Coupled Plasma Mass Spectrometry (ICPMS) on a
Perkin-Elmer Elan 6100. The samples were dissolved in
aqua regia at room temperature overnight, and diluted
5
by a factor of 10 . A calibration curve, created from
standard barium and vanadium solutions, was used to
determine the concentrations of these species for the
analysis.
2.3. X-ray powder diffraction
2
. Experimental
˚
A Guinier-Ha
˚
gg camera (CuKa l ¼ 1:54056 A) was
1
used to monitor phase purity and determine unit cell
constants. For the latter purpose an internal standard of
high purity Si powder was added.
2
.1. Synthesis
Previous reports from this laboratory indicated that
both SrV O and BaV O could be prepared using
the reaction Scheme (I) below at 1350 C for 24 h [27]:
1
0
15
10 15
ꢂ
2.4. X-ray single crystal diffraction
BaCO3 þ 5V2O3 þ H2-BaV10O15 þ H2O þ CO2
ðIÞ
Data were obtained on a Siemens/Bruker P4 four-
circle diffractometer using a rotating anode Mo source
and a 1 K CCD area detector (Bruker). SMART and
SAINT software packages were used for data collection
and processing, respectively [29]. The crystal was ground
into a roughly spherical shape of dimension B0.05 mm.
Absorption and l=2 corrections were performed using
SADABS [30]. SHELXL97 was utilized for structure
solution [31]. In some experiments the crystal was cooled
and the temperature was controlled by an Oxford
Cryostream Cooler. The sample temperature was
monitored by a thermocouple placed B2 mm from the
crystal.
Recent attempts using this approach resulted in
substantially pure samples but in all cases varying
amounts of VO were always present. Refiring under the
x
same conditions led to the diminution of the amount of
BaV O in the product, raising the possibility that this
1
0
15
ꢂ
compound is metastable at 1350 C. This led to another
approach, Scheme (II) below, using much higher
ꢂ
temperatures, 1550–1600 C for 17 h in inert (Ar) gas
atmosphere:
Ba3V2O8 þ 10VO þ 9V2O3-3BaV10O15
Ba V O was prepared in air at 950 C by reacting
ðIIÞ
ꢂ
3
2
8
BaCO and V O for 24 h. Hydrogen reduction of V O
3
2
5
2
5
ꢂ
at 700 C for 24 h yielded V O , the oxygen content of
2
2.5. Powder neutron diffraction
3
which was monitored by thermal gravimetric analysis
TGA). VO was synthesized by arc melting V metal and
V O , and with TGA analysis of the oxygen content.
The starting materials in (2) were sealed in a molybde-
num crucible by welding under an atmosphere of
ultrapure argon gas. Radio-frequency induction heating
was used to attain the indicated temperatures, which
(
Data were collected at the C2 diffractometer operated
by the Neutron Program for Materials Research of the
National Research Council of Canada, Chalk River,
2
3
˚
Canada, at wavelengths, l ¼ 1:329 and 2.379 A. Sam-
ples were contained in closed cycle refrigerator with
temperature control of 70.1 K.

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1101
2
.6. DC magnetic susceptibility
During the course of this study several samples of
nominal composition BaV O were prepared and
1
0
15
A Quantum Design MPMS S.Q.U.I.D. magnetometer
characterized. Fig. 2a shows the inverse susceptibility
behavior for a typical sample over the temperature
range 2–350 K. There are several obvious features. First,
note that the Curie–Weiss law appears to hold for
T4140 K. Fitting this regime yields the constants
was utilized to collect susceptibility data over the
temperature range 2–350 K at various applied fields.
Data were taken in both the field-cooled (FC) and zero-
field-cooled (ZFC) modes.
C ¼ 11:6ð1Þ emu-K/mol and y ¼ ꢁ1156ð15Þ K. The
c
2
.7. Electrical resistivity measurements
Curie constant can be compared with the spin-only
2
+
3+
value 2C(V =1.87)+8C(V =1.00)=11.74 emu-K/
mol, giving good evidence for the presence of the rare
DC resistivity measurements were performed on
2
+
dense, well sintered samples cut into the shape of
elongated bars. A standard four-probe geometry was
used and contacts were made using silver epoxy to silver
wire potential and current leads. Either a Quantum
Design PPMS or an Oxford Maglab measurement
system was used. As the resistivity of the samples
changed by more than 5 orders of magnitude over the
temperature range investigated, currents between 0.01
and 100 mA were employed in various temperature
regimes to give ohmic behavior and acceptable signal
to noise. In all cases the average potential for both
forward and reverse bias was used to compute the
resistivity.
V
. The very large, negative Weiss temperature
indicates strong antiferromagnetic spin correlations
which is significant in view of the frustrated topology
of the V sublattice. There is also evidence for complex
behavior for To135 K, as at least three features can be
seen: a continuous increase in the susceptibility (de-
crease in the inverse susceptibility) just below 135 K
without a FC–ZFC divergence, an abrupt FC–ZFC
1
1
1
1
30
20
10
00
ZFC
FC
BaV O
15
10
9
0
0
0
0
0
0
0
0
2
.8. Thermoelectric power
8
7
6
5
4
3
2
C= 11.6(1) emu-K/mol
2+
3+
C(s.o.)=2C(V )+8C(V ) = 11.74 emu-k/mol
Measurements of the Seebeck coefficient were mea-
sured using an apparatus constructed locally and
described elsewhere [32]. Again, samples were in the
form of bars cut from dense, well-sintered discs and
contacts were made to the measuring heads using silver
paste. Data could be collected, normally, over the range
θ= - 1156(15)K
0
50
100
150
200
250
300
350
3
5–350 K using a closed-cycle refrigerator.
(a)
Temperature (K)
0
0
0
0
0
0
0
.0162
.0153
.0144
.0135
.0126
.0117
.0108
2
.9. Thermal analysis
Differential scanning calorimetry studies were carried
out using a TA Instruments DSC 2910. Samples of
0–30 mg mass were sealed in thin aluminum discs for
Various BaV O
15
Preparations
10
1
measurement. An empty aluminum disc was used as the
reference. Temperatures are measured with Chromel
and Alumel thermocouple wires attached directly under
the sample and reference pans. The system was
calibrated using indium. Sample cooling was attained
using a liquid nitrogen cooling accessory within the
range 110–155 K.
0.0099
0.0090
0
.0081
.0072
0
0
50
100
150
200
250
300
350
(b)
Temperature (K)
3
. Results and discussion
Fig. 2. (a) (TOP) Inverse magnetic susceptibility for BaV10
O
15
showing evidence for two-phase transitions below B135 and B40 K
and a Curie–Weiss fit with the parameters shown. (b) (BOTTOM)
3
.1. Magnetic susceptibility
Susceptibilities of several samples of nominal composition, BaV10O
15
In previous work the magnetic susceptibility was not
reported for BaV O due to problems of small
15
impurities and non-reproducibility of the results [27].
showing the variation of the high temperature phase transition, Ts;
between the limits of 105 K (oxidized samples) to 135 K (nearly
stoichiometric samples).
1
0

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1102
divergence below 50 K and a weak maximum at B25 K
in the ZFC data. Fig. 2b shows similar data, ZFC mode
only, for several samples of varying composition. It is
clear that all of the transitions are fairly robust with the
onset for the highest temperature feature falling within
an envelope between 105 and 135 K. It is the high
temperature transition which will be the subject of this
paper. Analytical data in the form of oxidative weight
gain (theoretical=19.85%) show a value of 19.7% for
the samples with a transition at 135 K and 18.2% for
that at 105 K, indicating that the decrease in transition
temperature is apparently correlated with a decrease in
3.3. Crystal structure near the high temperature
transition
3.3.1. Powder neutron diffraction
Powder neutron diffraction data were obtained for a
sample showing a TsB125 K at several temperatures
near the phase transition in the heating mode. Figs. 4
and 5 show the thermal evolution of a selected part of
the diffraction pattern, indicating a sharp phase transi-
tion. Fig. 5 displays the thermal evolution of the refined
unit cell constants. Note that all three cell constants
change sharply at the transition with a and c increasing
with decreasing temperature while b decreases. The
2
+
V
content (decrease in weight gain). ICPMS data give
˚
˚
a Ba:V ratio of 0.0978(4) for the sample with Ts ¼ 135 K
magnitudes, [Da ¼ þ0:026ð2Þ A, Dc ¼ þ0:033ð2Þ A,
˚
and 0.0973(4) for the T ¼ 105 K material, i.e., evidence
Db ¼ ꢁ0:052ð2Þ A], are such that the total volume
s
for a small Ba deficiency which does not vary
significantly as a function of Ts:
change is very small. The transition temperature was
estimated by locating minima in plots of the temperature
derivative for each cell constant and the mean value for
this sample is Ts ¼ 124:2ð3Þ K.
3
.2. Differential scanning calorimetry (DSC)
3
.3.2. Single crystal X-ray diffraction
The thermodynamics of the high temperature transi-
Full data sets were collected at 130 and 100 K on a
single crystal isolated from a preparation which showed
a bulk T B125 K. The reported temperature is that
measured at a thermocouple about 2 mm from the
sample. The 130 K data were consistent with the Cmca
space group found at room temperature in the previous
study [27]. The details are shown in Table 1a–d. The
three occupied V sites, V1, V2 and V3, have multi-
plicities in the ratios 8:16:16, respectively, which does
tion was investigated using DSC as seen in Fig. 3 on a
sample with a transition near 135 K. Clearly, first-order
behavior is seen in the form of a latent heat and thermal
hysteresis. The transition onset, Ts; was measured at
133.3 K upon heating and 130.6 K with cooling. The
transition enthalpy was estimated by integrating the
s
peak area. Two samples with Ts ¼ 133:9 and 133.3 K
gave DH
associated entropy, DS
¼ 791 and 978 J/mol, respectively. The
TRANS
¼ DH
=T ; is 5.9 and
s
2+ 3+
TRANS
TRANS
permit the possibility of V /V charge ordering. The
expected V–O distances (sum of the Shannon radii) for a
7
6
.3 J/mol-K for the two samples with an average of
.6 J/mol-K. This value is comparable to the ex-
two configuration problem,
2+
charge ordered material are V –O=2.15 A and V
3+
˚
–
O=2.00 A [28]. The weighted, average V–O distance is
pected value for
a
˚
DSTRANS ¼ R ln 2 ¼ 5:6 J/mol-K. The above observa-
tions suggest, strongly, a first-order crystallographic
phase transition near 135 K in BaV O .
˚
.03 A and the corresponding Bond Valence Sum (BVS)
2
1
0
15
7
6
6
5
5
4
4
3
3
2
2
1
1
000
500
000
500
000
500
000
500
000
500
000
500
000
-
-
-
-
-
-
-
2
3
4
5
6
7
8
5
4
3
2
1
0
-1
Cooling
1
63.4K
29.8K
24.6K
23.8K
21.2K
01.7K
1
1
1
1
1
Heating
73.8K
5
00
0
54
56
58
60
62
64
1
15 120 125 130 135 140 145 150 155
Temperature (K)
2θ
Fig. 3. Differential scanning calorimeter data for
BaV10O15 in the range of the structural phase transition.
a
sample of
Fig. 4. Temperature evolution of a portion of the neutron diffraction
pattern for a sample of BaV10 15 with TsB125 K.
O

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1103
1
1.655
for the 100 K data yielded agreement indices equivalent
to those for Cmca at 130 K, Table 2a. As a result of the
lowered symmetry, two of the V sites, V2 and V3, each
split into eight-fold sites, denoted V2 and V2B and V3
and V3B in Table 2b. This splitting arises due to the
aforementioned loss of the mirror plane. Given the
11.650
11.645
11.640
apparent correlation of the onset temperature, T ; with
s
2
+
the V
content, it is reasonable to ask if charge
11.635
ordering plays any role in the phase transition.
Examination of Table 2c and d shows immediately that
the verdict is negative, the average V–O distances and
the BVS values remain nearly unchanged.
11.630
11.625
What, then, is the origin of the transition? The
contents of Table 1c and 2c, aided by Fig. 6, indicate the
changes in interatomic distances which result from the
phase transition. A careful comparison of the V–V
distances shows the following remarkable feature.
Before the phase transition the V2–V3 distances alter-
9
.95
.94
.93
.92
.91
.90
.89
.450
.445
.440
.435
.430
.425
.420
.415
.410
9
9
˚
nate along a zig–zag path between 2.73 and 2.99 A, the
9
shorter separation links the V10 clusters and the longer
corresponds to an edge of the cluster. At 100 K one
intercluster distance (V3–V2B) shrinks markedly to
9
˚ ˚
2.53 A (from 2.73 A) while the intracluster (V3–V2B)
9
˚
˚
distance expands to 3.09 A (from 2.99 A). The former is
one of the shortest V–V separations known in vanadium
oxide chemistry. The relative V–V bond shrinkage
involved, 7.3%, is also remarkable and comparable in
magnitude to that seen in the well-known case of the
9
9
9
phase transition in VO in which V–V bond formation
2
9
occurs, for which the reduction is 7.7% [35,36]. This
may be taken as evidence for V–V bond formation, or at
least bond strengthening, for this fraction of the V–V
bonds in BaV O . The increase in intracluster V3–V2B
9
9
1
0
15
9
bond length is only by 3.3%. Note that the other pair of
inter- and intracluster V–V bonds, V2 and V3B, show
the opposite trend but much less dramatically. The
9
˚
9
intercluster bond lengthens slightly (by 2.5%) to 2.80 A
and the intracluster bond decreases slightly (by 1.0%) to
˚
9
2.96 A. The overall enthalpy change is relatively small,
40
60
80
100
120
140
160
180
B0.8–1.0 kJ/mol, in comparison with VO , 4.69 kJ/mol
T (K)
2
[
of the V–V bonds in BaV O are involved in the
37]. This is not unreasonable given that only a fraction
Fig. 5. Temperature dependence of the unit cell constants for the
sample of Fig. 4. The transition temperature is estimated to be
Ts ¼ 124:2ð3Þ K. See text.
1
0
15
shrinkage and that other V—V bond distances actually
increase as a result of the phase transition. It is worth
noting that the crystallographic phase transition in
BaV O , in spite of the similarity in crystal structure,
1
0
15
is 2.8 using the program VaList [33,34]. Examination of
Table 1c and d indicates a random occupation of the V
sites as had been observed at room temperature in the
previous study [27].
The 100 K data showed a large number of weak
reflections of the type h þ k þ l ¼ 2n þ 1 which violate
the C-centering condition. The pattern of systematic
absences was consistent with space group Pbca, a
maximal, isomorphic subgroup of Cmca. Note that the
mirror plane normal to a is also lost. A solution in Pbca
has little in common with that in V O , wherein the V–V
2 3
distances increase as the symmetry is lowered from
rhombohedral to monoclinic [2].
Further details of the crystal structure experiment
can be obtained from the Fachinformationzentrum
Karlsruhe, 76344 Eggenstein-Leopoldhafen, Germany
(FAX: (49) 7247-808-666; email: crysdata@fiz-karlsru-
he.de) on quoting the depository numbers (CSD) 413301
for BaV O at 100 K and 413302 for BaV O
10 15
1
0
15
at 130 K.

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1
104
Table 1
a) BaV10
(
O15 Details for the single crystal X-ray diffraction experiment at 130 K
Cmca
Space group
Refinement
˚
a (A)
11.6015(9)
9.9186(7)
9.3940(7)
1081.0(1)
130
R1 (I42sðIÞ)
wR2 (I42sðIÞ)
R1 (all)
0.0214
0.0528
0.0229
0.0534
1.244
˚
b (A)
˚
c (A)
3
V (A )
˚
wR2 (all)
Goodness of fit
Temperature (K)
Radiation type
Color
Mo Ka
Black
No. of reflections
No. of parameters
Extinction
1340
68
Indexranges
ꢁ19php18
ꢁ
16pkp11
15plp12
0.0054(2)
0.000
ꢁ
Mean shift/su
2
2
11
12
(
b) Final atomic and isotropic displacement parameters in Cmca at 130 K. T ¼ expðh a ꢀ U þ ? þ 2hka ꢀ b ꢀ U þ ?Þ
2
˚
Site
x
y
z
Ui=Ue(A )
Ba1
V1
V2
V3
O1
O2
O3
O4
O5
O6
0
0.5
0.5
0
0.00468(7)
0.00320(9)
0.00667(8)
0.00464(7)
0.0066(3)
0.0063(2)
0.0046(3)
0.0041(2)
0.0045(3)
0.0054(4)
0.67623(4)
0.40976(3)
0.67201(3)
0.5
0.13922(4)
0.13548(3)
0.10991(3)
0
0.37097(3)
ꢁ0.24606(3)
ꢁ0.2541(2)
0.1185(1)
0.25
0.2472(2)
0.3257(2)
0.5887(1)
0.3429(2)
0.5
0.0022(1)
0.25
0.6298(1)
0.5
0.5
0.2446(1)
0.2477(2)
0
˚
(c) Selected bond distances (A) for BaV10
O15 in Cmca at 130 K
Bond length
Bond length
Bond length
(
(
(
(
(
(
(
(
(
a)
b)
c)
d)
e)
f)
Ba–O5 ( Â 2)
Ba–O2 ( Â 4)
Ba–O1 ( Â 2)
Ba–O4 ( Â 4)
Mean
2.836(2)
2.860(2)
2.949(2)
2.966(1)
2.903
V1–O5
1.965(2)
2.000(1)
2.010(2)
2.183(1)
2.028
V3–V3
V3–V3
V2–V3
V2–V3
V1–V3
V2–V3
V2–V2
V1–V2
V1–V2
2.5819(6)
2.6336(6)
2.7317(5)
2.8410(5)
2.9592(4)
2.9872(5)
2.9939(7)
3.0377(5)
3.1028(5)
V1–O4 ( Â 2)
V1–O2 ( Â 2)
V1–O6
Mean
g)
h)
i)
V2–O5
V2–O3
V2–O2
V2–O4
V2–O1
V2–O6
Mean
1.947(1)
1.9551(9)
2.028(2)
2.049(1)
2.063(2)
2.1591(3)
2.034
V3–O2
V3–O1
V3–O2
V3–O3
V3–O4
V3–O4
Mean
1.986(2)
1.9964(3)
2.012(2)
2.014(2)
2.088(1)
2.091(1)
2.031
(d) Selected bond valence sums for BaV10
Atom
O15 in Cmca at 130 K ( in v.u.)
Atom
ꢀ
Bond valence
Bond valence
Ba
V1
V2
V3
2.266
2.760
2.724
2.710
O1
O2
O3
O4
1.975
2.130
2.043
1.841
1.888
1.866
Mean bond valence
V=2.726
O=1.971
O5
O6
Note. Superscripts (a)–(i) on the V–V distances allow comparison with corresponding distances in the low temperature, Pbca phase (Table 2c).
Entries in bold font show the largest changes resulting from the phase transition.
3
.4. Electrical transport: conductivity
expect consequences for the transport properties. Fig. 7
shows a simple Arrhenius plot of the temperature
dependence of the conductivity for samples with Ts ¼
135 and 105 K. First, both samples are clearly semi-
conducting or insulating within the temperature interval
If V–V bond formation, involving at least some of
the bonds in BaV O , is indeed the driving force
for the crystallographic phase transition, one would
1
0
15

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1105
Table 2
a) BaV10O15 details of the single crystal X-ray diffraction experiment at 100 K
(
Space group
Pbca
Refinement
˚
a (A)
11.6155(9)
9.8735(7)
9.4178(7)
1080.1(1)
100
R1 (I42sðIÞ)
wR2 (I42sðIÞ)
R1 (all)
wR2 (all)
Goodness of fit
0.0227
0.0536
0.0278
0.0600
1.200
˚
b (A)
˚
c (A)
3
V (A )
˚
Temperature (K)
Radiation type
Colour
Mo Ka
Black
No. of reflections
No. of parameters
Extinction
2556
122
0.0029(1)
0.000
Indexranges
ꢁ18php19
ꢁ
ꢁ
16pkp11
12plp15
Mean shift/su
2
2
11
12
(
b) Final atomic and displacement parameters in pbca at 100 K. T ¼ expðh a ꢀ U þz ? þ 2hka ꢀ b ꢀ U þ ?)
2
˚
Site
x
y
Ui=Ue (A )
Ba
V1
V2
V3
V2B
V3B
O1
O2
O2B
O3
O4
O4B
O5
0
0.5
0
0.00323(5)
0.00201(7)
0.00246(7)
0.00226(7)
0.00274(7)
0.00219(7)
0.0038(3)
0.0036(2)
0.0033(2)
0.0033(2)
0.0032(2)
0.0031(2)
0.0033(2)
0.0038(4)
0.50407(3)
0.37552(3)
ꢁ0.24143(3)
0.63553(3)
0.24878(3)
ꢁ0.2566(2)
0.1141(1)
0.8793(1)
0.2512(1)
0.6332(1)
0.3725(1)
0.5017(1)
0.5
0.67522(4)
0.41157(3)
0.67945(3)
0.09703(4)
0.82959(3)
0.5039(2)
0.2505(2)
0.2545(2)
0.3279(2)
0.5884(2)
0.9116(2)
0.3426(2)
0.5
0.14202(4)
0.13676(3)
0.11246(3)
0.63208(3)
0.60593(3)
0.0023(1)
0.0064(2)
0.5040(2)
0.2481(1)
0.2451(2)
0.7444(2)
0.2438(2)
0
O6
˚
(c) Selected bond distances (A) for BaV10
O15 in Pbca at 100 K
Bond length
Bond length
Bond length
(
(
(
(
a)
b)
c)
c)
(
Ba–O2 ( Â 2)
Ba–O5 ( Â 2)
Ba–O2B ( Â 2)
Ba–O4B ( Â 2)
Ba–O1 ( Â 2)
Ba–O4 ( Â 2)
Mean
2.798(2)
2.870(2)
2.877(2)
2.958(2)
2.981(2)
2.986(2)
2.912
V1–O5
V1–O4
V1–O4B
V1–O2B
V1–O2
V1–O6
Mean
1.973(2)
1.981(2)
2.000(2)
2.002(2)
2.016(2)
2.1873(4)
2.027
V3B–V3
V3B–V3
V3B–V2
V3–V2B
2.5378(5)
2.6561(5)
2.7997(5)
2.5334(5)
2.7120(5)
2.9567(5)
2.9696(5)
2.9851(5)
2.9636(5)
3.0869(5)
3.0217(6)
3.0014(5)
3.0929(5)
3.0958(5)
d)
V3B–V2B
(
d)
e)
(
V3–V2
V3–V1
(
e)
f)
V3B–V1
V3B–V2
V3–V2B
V2B–V2
(
(
(
V2–O5
V2–O3
V2–O4B
V2–O2B
V2–O1
V2–O6
Mean
1.905(2)
1.967(2)
2.019(2)
2.041(2)
2.078(2)
2.1241(4)
2.022
V3–O3
V3–O2
V3–O2B
V3–O1
V3–O4B
V3–O4
Mean
1.971(2)
1.980(2)
2.009(2)
2.028(2)
2.092(2)
2.119(2)
2.033
f)
g)
(
h)
(
V1–V2
h)
V2B–V1
(
i)
V2–V1
V2B–O3
V2B–O5
V2B–O1
V2B–O2
V2B–O4
V2B–O6
Mean
1.905(2)
1.969(2)
2.014(2)
2.015(2)
2.118(2)
2.2233(4)
2.041
V3B–O2B
V3B–O1
V3B–O2
V3B–O3
V3B–O4B
V3B–O4
Mean
1.994(2)
2.002(2)
2.010(2)
2.052(2)
2.103(2)
2.104(2)
2.044
(d) Selected bond valence sums for BaV10
Atom
O15 in Pbca at 100 K (in v.u.)
Atom
Bond valence
Bond valence
Ba
V1
V2
V3
V2B
V3B
2.245
2.774
2.812
2.708
2.724
2.621
O1
O2
O2B
O3
O4
O4B
O5
O6
1.956
2.164
2.119
1.741
1.893
1.821
2.099
1.864
Mean bond valence
V ¼ 2:728
O ¼ 1:966
Note. Superscripts (a)–(i) on the V–V distances allow comparison with corresponding distances in the high temperature, Cmca phase (Table 1c).
Entries in bold font show the largest changes resulting from the phase transition.

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1106
2
.96
.99 3.04 3.04 2.99
.99
2.96
2.96
2.96
3
2
1
0
1
2
3
2
2
2.99 3.04 3.04 2.99
2.99
2
.73
2.73
2.73
2.73
2
.96
.99 3.04 3.04 2.99
.99
2.96
2
2
-
-
-
2
.73
2.73
2
.99 3.04 3.04 2. 9. 9
.99
.96
2.96
2.96
2.96
2
2.99 3.04 3.04 2.99
2.99
2
-4
(a)
b
c
2
4
6
8
10
12
3
-1
2
.99
2.97
2.99
2.97
a
10 /T (K )
2
2
.96 3.00 3.09 3.09
2.96 3.00 3.09 3.09
3.02
Fig. 8. Fit of the conductivity of BaV10
Polaronic Model. The derived activation energy is 0.120(1) eV.
O
15, Ts ¼ 135 K, to the Mott
3
.02
.80
2.53
V1
2.80
2.53
2
.97
2.99
V3
V3B
3.5. Samples with Ts ¼ 135 K
3
.09 3.09 3.00 2.96
There is a clear and dramatic break at T ¼ 136ð1Þ K,
essentially T : For T4T the data can be fit, roughly, to
a simple Arrhenius law, whereas, this simple approach
V2B
V2
3.02
s
s
2
.53
2.80
V1
V3
2.97
2
.96 3.00 3.09 3.09
.02
.99
2.97
2.99
V3B
V2
clearly fails for ToT : Taking the T4T data first, a
s
s
2
2.96 3.00 3.09 3.09
number of possible transport models have been pro-
posed for narrow band materials, such as transition
metal oxides, which are consistent with Arrhenius-like
behavior and involve versions of activated processes
falling under the heading of small polaronic conduction
[38]. In most models an equation such as (1) below is
used
3
3.02
V2B
(b)
Fig. 6. Comparison of V–V distances in one V layer between the high
temperature form, Cmca (a, top), and the low temperature form, Pbca
(
b, bottom). See text for discussion and Tables 1c and 2c.
2
1
s ¼ ATꢁ ½expðꢁW=kTÞꢃ;
n
ð1Þ
where s is the conductivity, W the activation energy for
Ts = 105K
2
electron hopping, A ¼ ½cð1 ꢁ cÞe u0=kR] where c is the
0
concentration of lower-valent polaron states, u0 is the
characteristic hopping frequency, R is the average
hop distance and n ¼ 1 if the hop occurs adiabatically.
For non-adiabatic hopping, considered to be less
-1
-2
-3
-4
-5
1
36(1)K
3
2
likely [39], n ¼ : In Fig. 8 the data are fitted according
to (1) with n ¼ 1 and it is noted that this is superior
to the simple Arrhenius approach (n ¼ 0 in the
prefactor) as the linearity range is extended down to
Ts = 135K
1
36 from 180 K. [Results of the fit to the Arrhenius
expression gives Ea ¼ 0:096ð1Þ eV.] Constants derived
from the fitting process are W ¼ 0:120ð1Þ eV and
0
.000
0.005
0.010
0.015
-1
0.020
0.025
0.030
ꢁ
1
ꢁ1
-1
A ¼ 10:32ð1Þ O cm K. As a consistency check one
T (K )
can solve for u from the A value assuming c ¼ 0:2 (the
0
2
+
˚
Fig. 7. Temperature dependence of the conductivity of BaV10
samples with Ts ¼ 105 and 135 K.
O
15
V
concentration) and R ¼ 2:87 A (average V–V
12 ꢁ1
distance), which yields u0 ¼ 2:93ð3Þ Â 10 s . This is
in the lower range of optical phonon frequencies, which
places this material near the borderline between
adiabatic and non-adiabatic hopping. Attempts to
studied, B30–300 K, and the conductivity varies over a
5
range of B10 . The conductivity of the Ts ¼ 105 K
sample is always greater than that of the T ¼ 135 K
analyze the data for ToT for these samples was much
s
s
sample.
less successful. One might expect a gradual transition to

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1107
a variable-range hopping (VRH) mechanism as the
-3.0
3
5K
temperature is lowered. Unfortunately, our data extend
only to 90 K due to the large resistance of the sample
below this temperature. Attempts to analyze the
available data using the Hill-Zabrodskii approach to
be described in the next section, where the quantity
log dE vs. log T is plotted (dE is the gradient of an
Arrhenius plot), did not yield a linear region. It must
be concluded that the VRH regime, if present at all,
occurs at lower temperatures. Nonetheless, the sharp
decrease in the conductivity just below Ts is consistent
with a corresponding decrease in the carrier concentra-
tion that is qualitatively compatible with V–V bond
formation, which would constitute a carrier trapping
mechanism.
100K
-
-
-
3.1
3.2
3.3
7
0K
- 0.54(4)
T
-3.4
1
.6
1.8
2.0
2.2
log T
3
.6. Samples with T ¼ 105 K
Fig. 9. Analysis of the low temperature conductivity data for
BaV10
15, Ts ¼ 105 K using the Hill-Zabrodskii method [41]. Agree-
ment with the ES [40] model, T ; is clear.
s
O
ꢁ
1=2
In this case there is no clear anomaly near T ¼ 105 K.
This is not unreasonable as the magnitude of the
susceptibility jump is much less for this sample which
may imply a lesser degree of V–V bond formation and
correspondingly weaker carrier trapping for this sample.
Unfortunately, it was not possible to obtain single
crystals of this sample which precluded a detailed study
of the variation of V–V distances with temperature. The
high temperature data (180–290 K) were fitted either to a
simple Arrhenius law or to Eq. (1), the Mott polaronic
model, yielding Ea ¼ 0:089ð1Þ eV and W ¼ 0:108ð1Þ eV,
respectively. These values are very similar to each other
and to the values found for the Ts ¼ 135 K sample
and analysis by plotting log ðDEÞ vs. log ðTÞ which
allows extraction of the exponent from the slope of any
linear region according to
log dE ¼ A þ n log T:
ð4Þ
Data from 65 to 35 K, analyzed according to Eq. (4),
are shown in Fig. 9. A linear regime is found giving a
slope, n ¼ ꢁ0:54ð4Þ: This seems to be evidence for the
ES model which implies a correlation or Coulomb gap
at the Fermi level for BaV10O15.
The temperature range for which ES behavior is
observed is similar to other materials reported by
Hill and Zabrodskii, which consisted of amorphous
semiconductors. ES behavior has also been found up to
(
0.120 eV), suggesting that there is little difference in the
conduction mechanism in this thermal regime between
samples at the extreme ranges of composition. Due to
the relatively higher conductivity of this sample, it was
possible to extend the measurements to 35 K and to
search for evidence of VRH. For purposes of this
analysis the conductivity can be expressed in a general
form by the following equation:
30 K in other correlated oxide materials, such as
Sr Y0:5Ca0:5Co O [43]. The characteristic temperature
2
2
7
3
from T ¼ T in (2) is 2 Â 10 K. This is a somewhat
0
ES
higher value than normally seen; for example, TES for
ꢁ1=2
2
Sr Y0.5Ca Co O is 3 Â 10 K. Nonetheless, the T
2
0.5
2
7
law is clearly observed and there are few other tenable
explanations. Note, finally, that in the Hill-Zabrodskii
plot a prominent anomaly is seen near 105 K, the Ts
value for this sample, which was not evident in a
standard Arrhenius plot.
s ¼ s0 expðꢁT0=TÞꢁ
n
:
ð2Þ
Two models for VRH in three-dimensional systems
are known, the original Mott model which predicts a
ꢁ
1=4
T
temperature dependence, and the Efros-Shklovs-
kii (ES) model which includes the effects of electron
ꢁ
1=2
3.7. Thermopower data for T ¼ 105 K and 135 K
correlation resulting in a T
law [40,41]. It is
s
samples
notoriously difficult to distinguish between these power
laws with conventional data analysis and, indeed, fits of
the low temperature data to either the Mott or ES
models for this sample appeared to be of equal quality.
Hill and Zabrodskii, independently, have suggested a
more rigorous test which involves the extraction of an
effective activation energy, dE; from the data as below
Thermopower data on these samples are shown in
Fig. 10a. Two facts are immediately obvious. First, the
thermopower is positive over the entire temperature
range, indicating that holes are the predominant carriers
in BaV O . Secondly, the data show a very strong
10 15
temperature dependence with near zero values down to
250 K, which increase rapidly as the temperature is
lowered further. Careful examination indicates an
[
42]
dE ¼ ꢁ1=T dðln sÞ=dð1=TÞ
ð3Þ

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1108
T = 105K
TS = 105K sample
TS = 135K sample
s
3
2
2
2
1
1
20
80
40
00
60
20
0
T = 135K
s
T = 135K
TS = 135K sample
s
-1
-2
-3
-4
8
0
0
0
4
60
120
180
240
300
360
50
100
150
200
(a)
T (K)
T (K)
6
Fig. 11. Temperature derivative of the Seebeck coefficient for the
samples in Fig. 10a.
BaV O
15
1
0
5
4
3
2
1
0
T = 135K
s
range from +0.4 to +1.2 mV/K for samples with Ts
values from 135 to 105 K, respectively, as seen from
Fig. 11. As the thermopower increases to the range
200–300 mV/K at 50 K, this simple approach is clearly
wrong. On the Heikes model or in fact any more
sophisticated approach that takes electron or spin
correlation into account [45] a value of pD0:5 would
be needed. This suggests a two carrier model with both
holes and electrons with details such that at higher
temperatures the two contributions are nearly equiva-
lent while at lower temperatures the holes dominate.
That is, the phase transition must somehow trap
electrons, predominantly.
2
40 250 260 270 280 290 300 310 320 330
T (K)
(b)
Fig. 10. (a) (TOP) Temperature dependence of the Seebeck coefficient
for BaV10
15 samples with Ts ¼ 105 and 135 K. (b) (BOTTOM)
Thermopower for BaV10
O
At this stage it seems that comparison with Cr-doped
V O would be useful. As already emphasized,
BaV O may be regarded as a heavily doped version
O
15 with Ts ¼ 135 K near room temperature.
2
3
The different symbols represent different runs for the same sample.
1
0
15
of V O , which is in fact equivalent, electronically, to
3
2
3+
0% Cr , i.e., Cr0.4V1.6O . Unfortunately, the max-
2
imum doping range which has been well studied for
3
inflection point near T ; seen most clearly in the plots of
s
dS=dT for the 135 K samples, Fig. 11. Though the
anomaly is less well defined for the 105 K material, its
presence in both samples is further evidence for a
dramatic decrease in carrier concentration below Ts:
A quantitative understanding is more difficult to
realize. The conductivity data seem to support a small
polaron or electron hopping model for the high
temperature regime, at least. On this basis one would
expect a very different set of thermopower data. First,
the simplest small polaron model involves the Heikes
formula given below [44]:
V O extends at most to B6% Cr. Nonetheless, the
2
3
available, albeit scant, results for high Cr doping levels
indicate positive room temperature Seebeck coefficients
of a few mV/K which rise rapidly as the temperature
decreases [46], i.e., parallel to the behavior of BaV O
15
1
0
although there is no phase transition in Cr-doped V O .
2
3
A possible model for the electronic structure of
BaV O involves as a starting point something
analogous to Cr-doped V O , i.e., an electron doped
1
0
15
2
3
Mott-Hubbard insulator. The high electron doping
+
2
3
–d ) would introduce
levels (20%
V
a
signi-
ficant aperiodic perturbation, i.e., Anderson localiza-
tion, and significant mobility edges to the bands. The
Fermi level would be placed such as to mimic the
situation for an intrinsic semiconductor and roughly
equivalent densities of states for holes and electrons. As
the temperature is lowered through the phase transition,
S ¼ ꢁðk=eÞ ln½ð1 ꢁ pÞ=pꢃ;
where p is the ratio of carriers to available sites. Taking
p ¼ ½V ꢃ ¼ 0:2; the expected S ¼ ꢁ119 mV/K which is
temperature independent. While the Seebeck coefficient
is constant over the range B270–B330 K the values
ð5Þ
2
þ

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1109
states from the nominal conduction band are removed
Acknowledgments
due to V–V bond formation and electronic carriers are
removed permitting hole conduction to predominate at
low temperatures.
We thank Professors R.W. Datars, J. Britten, and
M.A. White for helpful discussions. C.A.B. was sup-
ported by an OSSTF award from the government of
Ontario and J.E.G. by the Natural Sciences and Engi-
neering Research Council of Canada through a Research
Grant. We thank the following for assistance with
various procedures and the collection of data: J.D.
Garrett and H.F. Gibbs of the Brockhouse Institute for
Materials Research, high temperature synthesis and ther-
mal analysis, respectively; Ian Swainson of the Neutron
Program for Materials Research of the NRC, the neutron
diffraction experiments; Pamela Collins, ICPMS data;
and J.F. Britten, single crystal X-ray diffraction.
3
.8. Summary and conclusions
2
+
BaV O is a highly correlated, mixed-valence (V
+
/
) vanadium oxide with a crystal structure and
1
0 15
3
V
composition very similar to that of the well-studied
V O . It exhibits a remarkable crystallographic phase
2
3
transition below Ts: Ts varies from 135 to 105 K as the
2
+
sample becomes more oxidized, i.e., as the V content is
diminished. DSC studies indicate that the transition is
largely first-order showing a latent heat and hysteresis
with a value for DStransBR ln 2 which is consistent with a
two configuration problem. Evidence from X-ray and
neutron diffraction shows a change in space group
symmetry from Cmca above Ts to Pbca below Ts and
discontinuous changes in cell constants. There is no
convincing evidence for charge ordering from the crystal-
lographic data. Examination of V–V distances in the high
and low temperature forms discloses that for one pair of
V atoms involved in an intercluster link, a 7% decrease in
interatomic distance occurs suggesting partial bond
formation as the driving force for the transition. Thus,
while the crystal structure of BaV O is more related to
References
[
1] M. Foe
F.J. Morin, Phys. Rev. Lett. 3 (1959) 34.
[2] P.D. Dernier, M. Marezio, Phys. Rev. B 2 (1970) 3771.
¨
x, C. R. Acad. Sci. 223 (1946) 1126;
[
[
[
[
[
3] R.M. Moon, Phys. Rev. Lett. 8 (1970) 527.
4] W.F. Brinkman, T.M. Rice, Phys. Rev. B 2 (1970) 4302.
5] J.B. Goodenough, Prog. Solid State Chem. 5 (1972) 145.
6] J. Spalek, A. Datta, J.M. Honig, Phys. Rev. Lett. 59 (1987) 728.
7] J.-H. Park, L.J. Tjeng, A. Tanaka, J.W. Allen, C.T. Chen,
P. Metcalf, J.M. Honig, F.M.F. de Groot, G.A. Sawatzky, Phys.
Rev. B 61 (2000) 11506.
1
0 15
[
[
8] D.B. McWhan, T.M. Rice, J.P. Remeika, Phys. Rev. Lett. 23
(
V O the origin of the phase transition more closely
2
3
1969) 1384.
9] H. Kuwamoto, J.M. Honig, J. Appel, Phys. Rev. B 22 (1980)
2626.
resembles that in rutile structure VO and the related
2
Magneli phases, V O and V O . Transport property
9
4
7
5
measurements show that all samples of the compound are
semiconducting at all temperatures studied. The
[10] H. Seo, H. Fukuyama, J. Phys. Soc. Japan 67 (1998) 2602;
T. Ohama, H. Yasuoka, M. Isobe, Y. Ueda, Phys. Rev. B 59
(
1999) 3299;
M.V. Mostovoy, D.I. Khomskii, Solid State Commun. 113 (1999)
59.
T ¼ 135 K sample shows a pronounced decrease in
s
conductivity below the phase transition, while this effect
is more subtle for the Ts ¼ 105 K material. Analysis of
the high temperature conductivity data for both samples
suggests a polaronic hopping model, and thermopower
data require a quite complexsituation. The room
temperature Seebeck coefficient is anomalously small
and positive, SB þ 1 mV/K, and increases sharply with
decreasing temperature to values near +200mV/K with a
clear anomaly at Ts: This observation is consistent with a
two carrier model with essentially equal densities of states
and mobilities for holes and electrons at high tempera-
tures but for which electron states are removed due to V–
V bond formation below Ts: It is worth noting that the
1
[11] M. Marezio, P.D. Dernier, D.B. McWhan, S. Kachi, J. Solid State
Chem. 11 (1974) 301.
[
[
12] J.-L. Hodeau, M. Marezio, J. Solid State Chem. 23 (1978) 253.
13] Y. Kanke, K. Kato, E. Takayama-Muromachi, Y. Matsui,
J. Solid State Chem. 89 (1990) 130.
[
[
14] Y. Kanke, Phys. Rev. B 60 (1999) 3764.
15] Y. Hata, Y. Kanke, E. Kita, H. Suzuki, G. Kido, J. Appl. Phys.
85 (1999) 4768.
[
[
[
[
16] O. Mentre, Y. Kanke, A.-C. Dhausy, P. Conflant, Y. Hata,
E. Kita, Phys. Rev. B 64 (2001) 174404/1-11
17] Y. Kanke, E. Takayama-Muromachi, K. Kato, K. Kosuda,
J. Solid State Chem. 113 (1994) 125.
18] Y. Kanke, E. Takayama-Muromachi, K. Kato, K. Kosuda,
J. Solid State Chem. 115 (1995) 88.
2
+
3+
19] Y. Kanke, F. Izumi, Y. Morii, E. Akiba, S. Funahashi, K. Kato,
M. Isobe, E. Takayama-Muromachi, Y. Uchida, J. Solid State
Chem. 112 (1994) 429.
related V /V
SrV O , show n-type conductivity over the same
vanadium oxides, BaV O
and
1
3
18
13 18
temperature range, indicating a profound difference in
electronic structure [25]. Finally, analysis of the low
temperature conductivity data for the Ts ¼ 105 K sample
of BaV O using the Hill–Zabrodskii method shows
clear evidence for correlated, Efros-Shklovskii, variable-
range hopping at low temperatures, strong evidence for
the importance of correlation in this partially disordered,
transition metal oxide.
[
20] A. Akiba, H. Yamada, R. Matsuo, Y. Kanke, T. Haeiwa, E. Kita,
J. Phys. Soc. Japan 67 (1998) 1303.
[21] Y. Uchida, Y. Kanke, E. Takayama-Muromachi, K. Kato,
J. Phys. Soc. Japan 60 (1991) 2530.
1
0 15
[22] Y. Uchida, Y. Onoda, Y. Kanke, J. Magnetism, Magn. Mater.
26–230 (2001) 446.
2
[
23] K. Matsuno, T. Katsufuji, S. Mori, Y. Moritomo, A. Machida,
E. Nishibori, M. Takata, M. Sakata, N. Yamamoto, H. Takagi,
J. Phys. Soc. Japan 70 (2001) 1456.

ARTICLE IN PRESS
C.A. Bridges, J.E. Greedan / Journal of Solid State Chemistry 177 (2004) 1098–1110
1110
[
[
[
24] K. Iwasaki, H. Takizawa, K. Uheda, T. Endo, M. Shimada,
J. Solid State Chem. 158 (2001) 61.
[37] G.V. Chandrashekhar, H.L.C. Barros, J.M. Honig, Mater. Res.
Bull. 8 (1973) 369.
[38] T. Holstein, Ann. Phys. 8 (1959) 343;
25] K. Iwasaki, H. Takizawa, H. Yamame, S. Kubota, J. Takahashi,
K. Uheda, T. Endo, Mater. Res. Bull. 38 (2003) 141.
26] D. Chales de Beaulieu, H. Muller-Buschbaum, Z. Anorg. Allgem.
Chem. 472 (1981) 33.
A. Ghosh, J. Chem. Phys. 102 (1994) 1385.
[39] D. Emin, Phys. Rev. B 48 (1993) 13691.
[40] N.F. Mott, Conduction in Non-Crystalline Materials, Oxford
University Press, New York, 1987, pp. 20–29.
[41] A.L. Efros, B.I. Shklovskii, J. Phys. C: Solid State Phys. 8 (1975)
L49.
[
[
[
27] G. Liu, J.E. Greedan, J. Solid State Chem. 122 (1996) 416.
28] R.D. Shannon, Acta. Crystallogr. A 32 (1976) 751.
29] SAINT release 4.05. Siemens Energy and Automation Inc.
Madison WI, 1996.
[42] R.M. Hill, Phys. Stat. Sol. A 35 (1976) K29;
A.G. Zabrodski, Sov. Phys. Semicond. 11 (1977) 345.
[43] K. Yamaura, D.P. Young, R.J. Cava, Phys. Rev. B 63 (2001)
64401.
[
30] G.M. Sheldrick, SADABS: Siemens Area Detector Absor-
ption Correction Software, University of Go
996.
31] G.M. Sheldrick, SHELXL97, Program for the Refinement of
Crystal Structures, University of Gottingen, Germany, 1997.
32] G. Amow, N.P. Raju, J.E. Greedan, J. Solid State Chem. 155
2000) 177.
¨
ttingen, Germany,
1
[
[
[44] R.R. Heikes, J.R.W. Ure, thermoelectricity: science and engineer-
ing, in: p.M. Chaikin (Ed.), An Introduction to Thermopower for
those who Might want to use it to Study Organic Conductors
and Superconductors, Organic Superconductivity, Interscience
Publishers, New York, 1961;
¨
(
[
[
[
[
33] I.D. Brown, D. Altermatt, Acta. Crystallogr. B 41 (1985) 244.
34] A.S. Wills, VaLIst for GSAS, 1998.
V.Z. Kresin, W.A. Little Plenum Press, New York, 1990.
[45] P.M. Chaikin, P. Beni, Phys. Rev. B 13 (1976) 647.
[46] A.P.B. Sinha, G.V. Chandrashekhar, J.M. Honig, J. Solid State
Chem. 12 (1975) 402.
35] G. Andersson, Acta. Chem. Scand. 10 (1956) 623.
36] D.B. McWhan, J.P. Remeika, M. Marezio, P.D. Dernier, Phys.
Rev. B 10 (1974) 490.