Observation of metal-insulator transition in hollandite vanadate, K 2V8O16

Article abstract of DOI:10.1143/JPSJ.75.073801

We have synthesized hollandite vanadate, K₂V₈O ₁₆, by a solid-state reaction under 4 GPa at 1473 K for one hour, and investigated its structural and electromagnetic properties. We found a metal-insulator transition with a two-step jump of resistivity of about three orders at around 170 K. The magnetic susceptibility is reduced to a small value at the transition, suggesting the formation of V⁴⁺-V⁴⁺ singlet pairs and V³⁺-V³⁺ pairs in the low-temperature insulator phase. The transition is of first order accompanied by a structural change from a tetragonal to a monoclinic structure. The low-temperature phase has a superlattice of √2a × √2a × 2c, where a and c denote the parameters of the primitive cell for the simple tetragonal hollandite structure, suggesting a charge ordering of V⁴⁺ and V³⁺. We construct a possible charge order model from the obtained results, in which two kinds of double-chain formed by V³⁺ and V⁴⁺ chains and by single V⁴⁺ chains order in a manner that gives a superlattice of √2 × √2 in the a-b plane. The V⁴⁺-V⁴⁺ and V³⁺-V³⁺ pairs are formed in each chain along the c-axis, resulting in a duplication of the c-axis. 2006 The Physical Society of Japan.

Full text of DOI:10.1143/JPSJ.75.073801

Journal of the Physical Society of Japan
Vol. 75, No. 7, July, 2006, 073801
#2006 The Physical Society of Japan
LETTERS
Observation of Metal–Insulator Transition in Hollandite Vanadate, K2V8O16
ꢀ
Masahiko ISOBE , Shigenori KOISHI, Naomi KOUNO, Jun-Ichi YAMAURA,
1
1
Touru YAMAUCHI, Hiroaki UEDA, Hirotada GOTOU , Takehiko YAGI and Yutaka UEDA
Materials Design and Characterization Laboratory, Institute for Solid State Physics, University of Tokyo,
5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8581
1Division of New Materials Science, Institute for Solid State Physics, University of Tokyo,
5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8581
(Received April 10, 2006; accepted May 8, 2006; published June 26, 2006)
We have synthesized hollandite vanadate, K2V8O16, by a solid-state reaction under 4 GPa at 1473 K
for one hour, and investigated its structural and electromagnetic properties. We found a metal–insulator
transition with a two-step jump of resistivity of about three orders at around 170 K. The magnetic
susceptibility is reduced to a small value at the transition, suggesting the formation of V4þ–V4þ singlet
pairs and V3þ–V3þ pairs in the low-temperature insulator phase. The transition is of first order
accompanied by a structural change from a tetragonal to a monoclinic structure. The low-temperature
pffiffi
pffiffi
phase has a superlattice of 2a ꢁ 2a ꢁ 2c, where a and c denote the parameters of the primitive cell
for the simple tetragonal hollandite structure, suggesting a charge ordering of V4þ and V3þ. We construct
a possible charge order model from the obtained results, in which two kinds of double-chain formed by
pffiffi pffiffi
V3þ and V4þ chains and by single V4þ chains order in a manner that gives a superlattice of 2 ꢁ 2 in
the a–b plane. The V4þ–V4þ and V3þ–V3þ pairs are formed in each chain along the c-axis, resulting in
a duplication of the c-axis.
KEYWORDS: K2V8O16, metal–insulator transition, magnetic susceptibility, structural transition
DOI: 10.1143/JPSJ.75.073801
Since 1949 when a metal–insulator (MI) transition was and 8 GPa for A ¼ Na and 8 K and 7 GPa for A ¼ Ag.19,20)
first observed in V2O3 by Foex,1) vanadium oxides have The obtained P–T phase diagrams are similar to those of
¨
been a central material in the study of strongly correlated other superconducting systems, f-21,22) and ꢁ-electron23)
electrons. Many compounds exist in a binary V–O system, systems, except for the magnetic ordered phases adjacent to
many of which show MI transitions as a function of tem- the superconducting phases in the f- and ꢁ-electron systems.
perature.2,3) The MI transitions in mixed-valent vanadium
We have been continuously engaged in the development
oxides such as V4O7 are accompanied by charge separation of new vanadium oxides. Our next target is the development
and charge order.4,5) In their low-temperature insulator of hollandite vanadium oxides because a series of isostruc-
phases, spin singlet V4þ–V4þ pairs are formed, resulting tural compounds will enable us to systematically investigate
in a large reduction in magnetic susceptibility at the MI their structural and physical properties. ‘‘Hollandite’’ is one
transitions.4,5) In the extension of our study on vanadium of the manganese minerals. Its structure was determined in
oxides to ternary systems, two (MI and YU) of the authors 1950.24) Hollandite-type compounds have the general chemi-
reported ternary AV2O5 (A ¼ Li, Na, Cs, Mg, and Ca) which cal formula AxM8O16 (x ꢂ 2). Its structure consists of a
are insulators and quantum spin systems.6–8) The monovalent tubular M8O16 network and A-cations at the tunnel sites
V4þ-compounds, MgV2O5 and CaV2O5, are spin singlets in of the M8O16 framework, as shown in Fig. 1. The M8O16
the ground state, which originates from the geometry of the framework is constructed of double-chains, and each double-
structure, that is, the trellis lattice with two leg ladders.9,10) chain is formed by the sharing of the edges of the MO6
On the other hand, mixed-valent compounds (A ¼ Li,7,11) octahedra. The M8O16 network has rectangular tubes sur-
Na,6) Cs7)) with V4þ/V5þ ¼ 1 are low-dimensional magnets. rounded by four double-chains and A-cations occupy the
In particularly, the discovery6) of a spin-Peierls-like tran- sites within each rectangular tube.
sition in NaV2O5 triggered an explosion of worldwide
Some hollandite vanadium oxides with A ¼ K,25,26) Ba,27)
research on NaV2O5. It is now established that the transition Bi,28–30) Pb31) andd Tl26) have been reported. Among them,
in NaV2O5 is a charge-order transition accompanied by only the A ¼ Bi compound, in which the MI transition was
the formation of a spin gap.12,13) The research on NaV2O5 observed, has been studied in detail.28–30) The compounds
reached a peak at the discovery of the ‘‘devil’s flower’’-type with A ¼ K and Tl were synthesized under the high pressure
pressure–temperature (P–T) phase diagram.15) Another series of about 3 GPa. The first report on K2V8O16 was by Okada
of ternary vanadium oxides is the ꢀ-vanadium bronzes et al. in 1978.25) They reported a semiconductor–semi-
AxV2O5 (A ¼ Li, Na, Ag, Ca, Sr, Pb). We found MI conductor-like transition and an anomaly in the magnetic
transitions as a function of temperature in ꢀ-A0:33V2O5 susceptibility possibly caused by an antiferromagnetic tran-
(A ¼ Li, Na, Ag, Ca, Sr).16–18) Interestingly we discovered sition.16) There has been no report on K2V8O16 since then.
pressure-induced superconductivity in ꢀ-A0:33V2O5 (A ¼ Na, We successfully obtained K2V8O16 by high-pressure syn-
Ag).19,20) The charge-ordered insulator phases are suppressed thesis and observed an MI transition at 170 K. In this letter,
under hydrostatic pressure, and the superconducting phases we report the structural and electromagnetic properties of
adjacent to the charge-ordered phases appear at around 9 K K2V8O16.
Powder samples of K2V8O16 were prepared by a solid-
state reaction of KVO3, V2O3 and V2O5 under 4 GPa at
ꢀE-mail: isobe@issp.u-tokyo.ac.jp
073801-1

J. Phys. Soc. Jpn., Vol. 75, No. 7
LETTERS
M. ISOBE et al.
(a)
(b)
K+
VO6
c
a
a
VO6 double-chain
Fig. 1. (a) Schematic crystal structure of tetragonal K2V8O6 viewed from
c-direction and (b) schematic illustration of double-chain made up of
corner- and edge-sharing VO6 octahedra (gray polyhedra). The V8O16
framework is formed by the sharing of the edges of VO6 octahedra with
the double-chains, and Kþ ions (filled circles) occupy the tunnel sites
surrounded by the double-chains.
1473 K for one hour. Starting reagents were ground thor-
oughly and the mixture was wrapped in platinum foil and
put into a pressure cell. KVO3 was obtained by a reaction of
K2CO3 and V2O5. V2O3 was prepared by reducing of V2O5
under a H2 gas flow at 1173 K. A cubic-type apparatus was
used for the high-pressure synthesis. Sample identification
and the determination of the lattice parameters were per-
formed using powder X-ray diffraction. Very small single
crystals were found in the sintered sample. The typical size
of a crystal was 0:1 ꢁ 0:1 ꢁ 0:4 mm3. The structure analysis
was performed by X-ray diffraction of the obtained single
crystal using an imaging plate Weissenberg camera. The
electrical resistivity was measured along the c-axis by a
conventional four-probe method using a single crystal. The
magnetic susceptibility of the powdered sample was meas-
Fig. 2. (a) Temperature dependence of resistivity measured along c-axis
in single crystal K2V8O6. (b) Temperature dependence of magnetic
susceptibility for powder K2V8O6. The solid line shows the magnetic
susceptibility derived by subtracting the Curie term in the low-temper-
ature region from the raw data.
ured using a superconducting quantum interference device subtraction of the Curie tail from the raw data, the magnetic
(SQUID) magnetometer. susceptibility is almost constant below 150 K, as shown in
The powder X-ray diffraction pattern of the obtained Fig. 2(b). Both curves of resistivity and magnetic suscepti-
K2V8O16 shows a typical tetragonal hollandite structure bility show significant hysteresis at the transition, which
at room temperature. The crystal symmetry determined by indicates a first-order transition. The X-ray diffraction
single-crystal X-ray diffraction is the space group of I4/m pattern at 100 K below TMI can be indexed to a pseudo-
˚
and the lattice parameters are a ¼ 9:9455ð1Þ A and c ¼ orthorhombic structure, or more precisely, a monoclinic
2:9116ð1Þ A at room temperature, in agreement with the structure with ꢀ ꢃ 90:07ꢄ. It was difficult, however, to
˚
previous report.26) The composition of K ions was deter- determine the crystal symmetry of the low-temperature
mined from single-crystal X-ray diffraction to be almost phase because the diffraction spots were split into some
stoichiometric. Temperature dependences of electrical re- spots across the MI transition, which may be due to the first-
sistivity and magnetic susceptibility are shown in Figs. 2(a) order transition. The temperature dependences of the lattice
and 2(b), respectively. We first found a MI transition with constants obtained from powder X-ray diffraction are shown
a two-step jump of resistivity at about three orders at around in Fig. 3. The X-ray diffraction pattern shows the coex-
170 K (TMI). The resistivity behavior above TMI is that of istence of the two phases through the transition, indicating
a typical metal, which is in disagreement with the previous a first-order transition. Below TMI, the a–b plane in the
report.25) Such a discrepancy might be ascribed to the quality tetragonal structure shrinks and the c-axis extends, resulting
of the crystals. The MI transition is accompanied by a in an increase of volume of about 0.14%. Furthermore, from
gradual reduction in magnetic susceptibility from 175 to a single-crystal X-ray photograph using an imaging plate
150 K. The two-step transition observed in the electrical Weissenberg camera, we observed superreflections associ-
pffiffi
pffiffi
resistivity was not clearly seen in the magnetic susceptibility ated with a supercell of 2a ꢁ 2a ꢁ 2c, where a and c
measured using the powdered sample. The magnetic sus- denote the primitive cell for the simple tetragonal hollandite
ceptibility showed an increase at low temperatures, which structure. We also observed the splitting of diffraction spots
obeys the Curie law. The obtained Curie constant corre- through the transition in the X-ray diffraction pattern, which
sponds to about 4.5% impurity with S ¼ 1=2. After the implied a two-step transition from tetragonal to monoclinic
073801-2

J. Phys. Soc. Jpn., Vol. 75, No. 7
LETTERS
M. ISOBE et al.
V3+-V3+ pair
V4+-V4+ pair
a
2a
a
2a
c
V3+
V4+
A
B
A
B
Fig. 4. Model of charge order in low-temperature phase of K2V8O6.
Where two kinds of rectangular tube formed by two kinds of double-
chain, A and B, are ordered in a checkerboard pattern, giving a super-
pffiffiffi pffiffiffi
lattice of 2 ꢁ 2 in the a–b plane. The double-chains A and B are made
up of V3þO6 and V4þO6 octahedral chains and by only V4þO6 octahedral
chains, respectively. In each chain, V4þ–V4þ and V3þ–V3þ pairs are
formed along the c-axis, resulting in a duplication of the c-axis.
each chain along the c-axis, resulting in a duplication of the
c-axis.
In summary, we have successfully synthesized hollandite
vanadate, K2V8O16, by a solid-state reaction under 4 GPa
at 1473 K for 1 h, and measured its electrical resistivity,
magnetic susceptibility and structure as functions of temper-
Fig. 3. Temperature dependences of lattice constants for K2V8O6. The
crystal structure changes from a tetragonal structure in the high-
temperature phase to a monoclinic structure in the low-temperature
phase. The monoclinic b-axis corresponds to the tetragonal c-axis.
and then from monoclinic to monoclinic upon cooling. ature. We found a MI transition with a two-step jump of
However, it was difficult to determine the precise structure resistivity of about three orders at around 170 K. The tran-
of the intermediate- and low-temperature phases because the sition is accompanied by a reduction in magnetic suscepti-
crystal was split into diffrent domains across the transition. bility to a small value, suggesting the formation of V4þ–V4þ
The origin of the two-step transition is still unknown.
Many binary vanadium oxides show MI transitions insulator phase. The transition is of first order and accom-
spin-singlet pairs and V3þ–V3þ pairs in the low-temperature
accompanied by the reduction in magnetic susceptibility, panied by a structural change from a tetragonal to a mono-
32,33)
as mentioned above. In some oxides, such as V2O3
and clinic structure. The low-temperature phase has a super-
pffiffi
pffiffi
V8O15,34) antiferromagnetic transitions occur simultaneously lattice of 2a ꢁ 2a ꢁ 2c, where a and c denote the
MI transitions, whereas in other oxides, such as VnO2nꢅ1 parameters of the simple tetragonal cell, suggesting a charge
(n ¼ 4; 5; 6),35,36) MI transitions are followed by antiferro- ordering of V4þ and V3þ. We propose a possible charge-
magnetic transitions below TMI. Furthermore, the MI tran- order model, in which two kinds of rectangular tube are
sitions in VnO2nꢅ1 are accompanied by the charge separation ordered in a checkerboard pattern, giving a superlattice of
pffiffiffi pffiffi
4,5)
and charge ordering of V4þ and V3þ
.
The reduction in
2 ꢁ 2 in the a–b plane. The two kinds of rectangular
magnetic susceptibility at the MI transitions in VnO2nꢅ1 tube consist of two kinds of double-chain formed by V3þ and
originates from the formation of V4þ–V4þ spin-singlet V4þ chains and by only V4þ chains. V4þ–V4þ and V3þ–V3þ
pairs and V3þ–V3þ pairs. We cannot confirm whether the pairs are formed in each chain along the c-axis. resulting in
reduction in magnetic susceptibility at the MI transition in a duplication of the c-axis.
K2V8O16 is due to an antiferromagnetic transition or to the
formation of V4þ–V4þ spin-singlet pairs and V3þ–V3þ pairs;
nevertheless, we consider the latter to be plausible because
Acknowledgements
The authors thank Dr. K. Ohgushi, F. Sakai and Y. Kiuchi
of the rather small magnetic susceptibility, about 5 ꢁ 10ꢅ4 for valuable discussions. This study is partly supported
(emu/V-mol), of the insulator phase. The superlattice of by a Grant-in-Aid for Scientific Research on Priority Area
pffiffiffi
pffiffi
2a ꢁ 2a ꢁ 2c observed in the low-temperature phase of (No. 16076203) from the Ministry of Education, Culture,
K2V8O16 suggests charge ordering. Taking these results into Sports, Science and Technology, Japan.
consideration, we propose a possible charge-ordered model,
as shown in Fig. 4. In this model, there are two kinds of
1) M. Foex: C. R. Acad. Sci. Paris B 223 (1946) 1126.
¨
double-chain: one (A) is formed by V3þO6 and V4þO6
chains and the other (B) is formed by only V4þO6 chains.
Four A and B double-chains form two kinds of rectangular
tube, respectively. These tubes are ordered in the manner
2) H. Okinaka, K. Kosuge and S. Kachi: Trans. JIM 12 (1971) 44.
3) S. Kachi, K. Kosuge and H. Okinaka: J. Solid State Chem. 6 (1973)
258.
4) M. Marezio, D. B. MaWhan, P. D. Dernier and J. P. Remeika: Phys.
Rev. Lett. 28 (1972) 1390.
shown in Fig. 4. This model satisfies the ratio of V4þ/V3þ
¼
pffiffiffi pffiffi
5) Y. Ueda, Y. Kitaoka, H. Yasuoka, K. Kosuge and S. Kachi: J. Phys.
Soc. Jpn. 47 (1979) 333.
6) M. Isobe and Y. Ueda: J. Phys. Soc. Jpn. 65 (1996) 1178.
3=1 and forms the superlattice of 2 ꢁ 2 in the a–b plane.
Furthermore the V4þ–V4þ and V3þ–V3þ pairs are formed in
073801-3

J. Phys. Soc. Jpn., Vol. 75, No. 7
LETTERS
M. ISOBE et al.
7) M. Isobe and Y. Ueda: J. Phys. Soc. Jpn. 65 (1996) 3142.
8) Y. Ueda: Chem. Mater. 10 (1998) 2653.
22) S. S. Saxena, P. Agarwal, K. Ahilan. F. M. Grosche, R. K. W.
Haselwimmer, M. J. Steiner, E. Pugh, I. R. Walker, S. R. Julian,
P. Monthoux, G. G. Lonzarich, A. Huxley, I. Sheikin, D. Braithwaite
and J. Flouquet: Nature (London) 406 (2000) 587.
9) H. Iwase, M. Isobe, Y. Ueda and H. Yasuoka: J. Phys. Soc. Jpn. 65
(1996) 2397.
10) M. Isobe, Y. Ueda, K. Takizawa and T. Goto: J. Phys. Soc. Jpn. 67
(1998) 755.
23) D. Jerome and H. J. Schulz: Adv. Phys. 51 (2002) 293.
24) A. Bystrom and A. M. Bystrom: Acta Crystallogr. 3 (1950) 146.
¨
¨
11) N. Fujiwara, H. Yasuoka, M. Isobe, Y. Ueda and S. Maegawa: Phys.
Rev. B 55 (1997) R11945.
25) H. Okada, N. Kinomura, S. Kume and M. Koizumi: Mater. Res. Bull.
13 (1978) 1047.
12) H. Seo and H. Fukuyama: J. Phys. Soc. Jpn. 67 (1998) 2602.
13) T. Ohama, H. Yasuoka, M. Isobe and Y. Ueda: Phys. Rev. B 59 (1999)
3299.
26) W. Abriel, F. Rau and K. J. Range: Mater. Res. Bull. 14 (1979) 1463.
27) Y. Kanke, E. Takayama-Muromachi, K. Kato and K. Kosuda: J. Solid
State Chem. 115 (1995) 88.
14) K. Ohwada, Y. Fujii, N. Takesue, M. Isobe, Y. Ueda, H. Nakao,
Y. Wakabayashi, Y. Murakami, K. Ito, Y. Amemiya, H. Fujihisa,
K. Aoki, T. Shobu, Y. Noda and N. Ikeda: Phys. Rev. Lett. 87 (2001)
086402.
28) H. Kato, T. Waki, M. Kato, K. Yoshimura and K. Kosuge: J. Phys.
Soc. Jpn. 70 (2001) 325.
29) T. Waki, Y. Morimoto, H. Kato, M. Kato and K. Yoshimura: Physica
B 329–333 (2003) 938.
15) K. Ohwada, Y. Fujii, H. Nakao, Y. Murakami, M. Isobe and Y. Ueda:
Mod. Phys. Lett. B 20 (2006) 199.
30) T. Waki, H. Kato, M. Kato and K. Yoshimura: J. Phys. Soc. Jpn. 73
(2004) 275.
16) H. Yamada and Y. Ueda: J. Phys. Soc. Jpn. 68 (1999) 2735.
17) Y. Ueda: Proc. Int. Workshop ‘‘Magnetic Excitations in Strongly
Correlated Electrons, J. Phys. Soc. Jpn. 69 (2000) Suppl. B, p. 149.
18) M. Isobe and Y. Ueda: Mol. Cryst. Liq. Cryst. 341 (2000) 1075.
19) T. Yamauchi, Y. Ueda and N. Mori: Phys. Rev. Lett. 89 (2002)
057002.
31) F. Abraham and O. Mentre: J. Solid State Chem. 109 (1994) 127.
32) R. M. Moon: J. Appl. Phys. 41 (1970) 883.
33) K. Andres: Phys. Rev. B 2 (1970) 3768.
34) S. Kachi, Y. Ueda and K. Kosuge: Ferrites (1980) 131.
35) H. Okinaka, K. Nagasawa, K. Kosuge, Y. Bando, S. Kachi and
T. Takada: J. Phys. Soc. Jpn. 28 (1970) 798.
20) T. Yamauchi, M. Isobe and Y. Ueda: Soild State Sci. 7 (2005) 874.
21) D. Jaccard, K. Behnia and J. Sierro: Phys. Lett. A 163 (1992) 475.
36) K. Kosuge, K. Okinaka and S. Kachi: IEEE Trans. Magn. 8 (1972)
581.
073801-4