Synthesis, structure, and thermoelectric properties of the new antimonide sulfide MoSb2S

Article abstract of DOI:10.1002/1099-0682(200203)2002:3<591::AID-EJIC591>3.0.CO;2-R

The structure of a new ternary compound, MoSb₂S, was identified by single-crystal X-ray diffraction. The title compound was synthesized by heating the elements in stoichiometric ratios in sealed silica tubes at 600-800 °C using small amounts of

Full text of DOI:10.1002/1099-0682(200203)2002:3<591::AID-EJIC591>3.0.CO;2-R

FULL PAPER
Synthesis, Structure, and Thermoelectric Properties of the New Antimonide
Sulfide MoSb2S
Chi-Shen Lee[a] and Holger Kleinke*[a]
Keywords: Antimony / Sulfur / Materials science / Chalcogens / Molybdenum
The structure of a new ternary compound, MoSb2S, was
identified by single-crystal X-ray diffraction. The title com- contains CdI2-related layers (MoSbS) and Sb columns,
pound was synthesized by heating the elements in stoi- namely a zigzag (cis-trans) chain and an equidistant ladder,
95.093(2)°, Z = 16, R1/wR2 = 0.035/0.059]. Its crystal structure
chiometric ratios in sealed silica tubes at 600−800 °C using both of which are located between the MoSbS layers. The
small amounts of I2 as a mineralizator. The compound crystal- results of thermoelectric measurements and electronic struc-
lizes in a new structure type in the monoclinic space group
C2/m [a = 36.157(3) A, b = 6.3823(5) A, c = 6.5379(5) A, β =
ture calculations indicate metallic properties.
˚
˚
˚
Introduction
Results and Discussion
Structure
Thermoelectric materials exhibit a unique property,
namely the conversion of heat into electrical energy and
vice versa. Such materials, which are usually small-gap
semiconductors, have been utilized in applications like refri-
gerators and power generators. Commercial materials in-
clude semiconductors based on Bi2Te3, Si0.7Ge0.3, and
PbTe.[1] Extensive research activities were carried out with
investigations of new thermoelectric materials, including the
filled skutterudites LnM4Sb12 (Ln ϭ lanthanoid; M ϭ Fe,
Co, Ni, ...) and most recently the ternary telluride
CsBi4Te6.[2Ϫ5] Good thermoelectric materials have a high
figure-of-merit that combines a high Seebeck coefficient,
low thermal conductivity, and high electrical conductivity.
To approach the goal of preparing new thermoelectrics,
heavy main group metalloid elements such as antimony, sel-
enium or tellurium may be used as the common element.
Our group has focused on the syntheses and characteriza-
tions of new materials in the ternary MϪSbϪQ systems
(M ϭ transition metal atoms; Q ϭ S, Se, Te). During the
course of this study, a new ternary compound, MoSb2Se,
was identified.[6] The structure of MoSb2Se can be treated
as the combination of a CdI2-related layer structure
(MoSbSe layer) with interstitial Sb chains binding the Sb
atoms of MoSbSe layers. A similar structure is realized in
the MoϪSbϪS system; the structure of MoSb2S exhibits
topologically equivalent MoSbS layers, yet significant dif-
ferences in the Sb atom substructure. With this contribu-
tion, the synthesis, characterization and electronic structure
studies of MoSb2S are presented.
MoSb2S crystallizes in the space group C2/m, with six
independent positions for Sb, three for Mo, and three for S
atoms. The perspective view of the MoSb2S crystal struc-
ture along [010] is illustrated in Figure 1. In general, the
structure of MoSb2S contains a CdI2-related layer of the
composition MoSbS and two types of columnar Sb atom
substructures, which bind the Sb atoms of the two sur-
rounding MoSbS layers. Pseudo-octahedral voids remain
present in the interlayer Sb atom substructure.
[a]
Department of Chemistry, University of Waterloo
Figure 1. The three-dimensional structure of MoSb2S projected
along [010]; dark circles: Mo; small gray circles: S; large circles: Sb,
with dark circles representing interlayer and bright circles intra-
layer atoms
Waterloo, ON, N2L 3G1, Canada,
Fax: (internat.) ϩ 1-519/746-0435
E-mail: kleinke@uwaterloo.ca
Eur. J. Inorg. Chem. 2002, 591Ϫ596
WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002 1434Ϫ1948/02/0303Ϫ0591 $ 17.50ϩ.50/0
591

C.-S. Lee, H. Kleinke
FULL PAPER
˚
The Mo atoms are sandwiched between the approxim- and long (3.54 A) SbϪSb distances alternate along [010].
ately closed packed Sb and S atom layers, forming MoSbS This type of cis-trans chain is also found in the structure of
layers that are reminiscent of the MoTe2 layers of the mono- MoSb2Se (Figure 2b). Furthermore, each Sb atom of the
[7]
clinic β-MoTe2
structure. Each Mo is coordinated by zigzag chain binds an Sb atom of the MoSbS layer (d ϭ
˚
three Sb and three S atoms in a distorted octahedral coor- 2.84 A). Thus, each Sb atom of the zigzag chain takes part
dination. The averaged MoϪSb and MoϪS distances are in three SbϪSb interactions described best as single bonds.
˚
2.83 and 2.38 A, respectively. This may be compared to the
The second type is an Sb ladder at x close to 1/2. Each
corresponding interatomic distances in Mo3Sb7 (dMoϪSb
2.77 and 2.85 A) and MoS2 (dMoϪS ϭ 2.41 A).
ϭ
ladder contains two parallel Sb atom columns with equi-
distant SbϪSb contacts of 3.19 A parallel to the b axis.
[8]
[9]
˚
˚
˚
The Mo atoms in MoSb2S are connected to form zigzag These two columns are interconnected to form the ladder
˚
chains along the b axis, as is the case for the β-MoTe2 struc- by SbϪSb bonds of 2.85 A, and to the Sb atoms of the
˚
˚
ture. The shortest MoϪMo distances are 2.76Ϫ2.77 A, MoSbS layer by a second SbϪSb single bond of 2.84 A.
shorter than that of the topologically equivalent chains of This particular Sb atom substructure is not observed in the
˚
˚
MoSb2Se (2.81 A) and β-MoTe2 (2.89Ϫ2.90 A). The MoSb2Se structure, but the Sb atom ladder compares well
˚
MoϪMo bonds in elemental molybdenum (cubic body- to that found in FeSb2 with SbϪSb distances of 3.20 A in
˚
˚
centered packing, W type) are 8 ϫ 2.73 and 6 ϫ 3.15 A per the columns and 2.89 A for the bonds connecting the col-
Mo atom. Thus, the MoϪMo interactions in MoSb2S have umns to the ladder.[15] Figure 3 illustrates the different co-
significant bonding character.
lumnar Sb atom substructures of MoSb2S, namely the cis-
The SbϪSb distances within the MoSbS layer are be- trans chain and the ladder.
˚
tween 3.18 and 3.33 A. While these distances are larger than
˚
a typical SbϪSb single bond (2.80Ϫ2.90 A), they are
shorter than the weak bond in elemental antimony (d ϭ
˚
3.35 A). Interactions like these are typically weakly bond-
ing.[10] The averaged SϪS and SϪSb distances are 3.19 and
˚
3.64 A, respectively. These presumably nonbonding con-
tacts are much longer than the sums of the covalent radii
[11]
˚
(rS ϭ 1.02, rSb ϭ 1.40 A).
A detailed discussion of the
bonding character of these interactions will be given in the
electronic structure section.
Chains comprising Sb atoms are running along the b axis
between the MoSbS layers. Two types of interlayer SbϪSb
chains exist in the structure of MoSb2S. The first type
(located at x close to 1/4 and 3/4 as shown in Figure 2a) is
Figure 3. The columnar Sb atom substructure of MoSb2S: a) zig-
zag chain; b) ladder
˚
SbϪSb bonds of 3.0Ϫ3.2 A are usually identified as half
a zigzag chain with SbϪSb bonds of 2.84 (parallel to b) and bonds.[16Ϫ21] Instead of forming three full SbϪSb bonds
˚
2.88 A (approximately parallel to c), which are both in the like the Sb atoms of the zigzag chain, each Sb atom of the
range of the SbϪSb single bond lengths, as comparisons ladder forms two full and two half SbϪSb bonds. This
[12]
˚
with KSb (2.83 and 2.85 A),
A),
cyclo-Sb55Ϫ (2.81Ϫ2.91 bonding situation resembles those of the FeSb2 structure
[13]
3Ϫ
[14]
˚
˚
˚
and Sb11 (2.76Ϫ2.85 A) reveal.
Short (2.84 A) and of the Li2Sb structure. In the latter, two different linear
Sb atom chains occur; one comprises equidistant SbϪSb
˚
distances of 3.26 A, and the second one has alternating
[22]
˚
short and long distances of 2.97 and 3.56 A.
The question remains, why two different columnar Sb
atom substructures occur in MoSb2S, but not in MoSb2Se.
In general, the presence of one full bond is preferred over
two half bonds (Peierls distortion). However, the Li2Sb
structure proves that the energetic differences may be very
small, since both alternatives exist in this same structure Ϫ
as they do in MoSb2S. Also, the differences between
MoSb2S (one linear equidistant, thus an undistorted chain
is present) and MoSb2Se are reminiscent of those between
the high- and low-temperature form of LaSiAs3.[23] In this
arsenide, an almost square-planar, i.e. undistorted, As layer
˚
(with intermediate AsϪAs bonds of 3.01 and 3.05 A) is
present in the high-temperature modification, resembling
the MoSb2S case. This As layer is transformed into zigzag
chains in the low-temperature modification (thus resem-
Figure 2. The pseudo two-dimensional Sb atom network of a)
bling the MoSb2Se case), with short and long AsϪAs dis-
MoSb2S and b) MoSb2Se; solid lines: short SbϪSb contacts
˚
˚
˚
(2.8Ϫ2.9 A); dash lines: long SbϪSb contact (3.1Ϫ3.4 A)
tances of alternating 2.56 and 3.52 A along [100] and 2.76
592
Eur. J. Inorg. Chem. 2002, 591Ϫ596

Synthesis, Structure, and Thermoelectric Properties of the New Antimonide Sulfide MoSb2S
FULL PAPER
˚
and 3.29 A along [010]. In analogy, one can consider
MoSb2Se as the distorted variant of MoSb2S.
However, even if the differences are subtle, why is there
no undistorted Sb chain in MoSb2Se, but one in MoSb2S?
We speculate that this is due to the matrix effect imposed
by the larger chalcogen atom Se, which would make the half
bonds longer than in MoSb2S as reflected in the longer b
˚
˚
axis of MoSb2Se, namely 3.25 A compared to 3.19 A.
The formal charges of each atom in MoSb2S may be as-
signed qualitatively based on the ZintlϪKlemm con-
cept.[24,25] Among these elements, the electronegativity of
sulfur is the largest on any electronegativity scale, and the
distances between two nearest S atoms fall in the range of
˚
3.2Ϫ3.5 A, which are normal van der Waals distances.
Then, the S atom can be treated as S2Ϫ ions. The short
Figure 4. Band structure of MoSb2S; the dashed line denotes the
MoϪMo contacts in the Mo zigzag chain suggest a d2 con- Fermi energy; the interlayer Sb contributions are emphasized
figuration, that is Mo4ϩ, which is the preferred valence-elec-
tron configuration for the transition metal atom zigzag
chain. The analogous assignments for MoS2 and MoTe2 re- nounced local minimum in the densities of states at the
sult in (Mo4ϩ)(S2Ϫ)2 and (Mo4ϩ)(Te2Ϫ)2, respectively. The Fermi level.
formal charge of each Sb atom is assigned based on the (8
The interatomic interactions in MoSb2S were analyzed
Ϫ N) rule, with N ϭ number of full SbϪSb bonds. There- by integrating the area of the COHP curves, resulting in
fore, the Sb atom on the MoSbS layer is Sb2Ϫ (one full ICOHP values. This is a similar procedure to obtaining the
bond, thus N ϭ 1, ergo 8 Ϫ N ϭ 7 valence electrons), and widely used Mulliken overlap populations (MOPs)[27] from
the Sb atom of the zigzag chain is Sb0 (three full bonds, the Crystal Orbital Overlap Population (COOP) curves[28]
N ϭ 3). The Sb atom from the ladder carries the same calculated with the Extended Hückel method.[29,30] In con-
charge (i.e. 0), since it comprises two full and two half trast to the MOPs, bonding character is reflected in nega-
bonds. We therefore arrive at an electron-precise formula, tive ICOHPs. MoϪMo COHP curves (solid line in Fig-
that is Mo4ϩSb0Sb2ϪS2Ϫ, which in general implies that ure 5b) are nearly optimized at the Fermi level EF
semiconducting properties are possible. In this case, how- (ICOHP ϭ Ϫ1.16 eV per bond). A small gap at EF separ-
ever, semiconducting properties are unlikely due to the pres- ates weak antibonding from strong antibonding MoϪMo
ence of other intermediate SbϪSb interactions (3.18Ϫ3.20 states, which suggests that the MoϪMo bonding will be
˚
A) within the MoSbS layer. These increase N of the intra- destabilized by small changes in the valence-electron con-
layer Sb atom, and thus decrease the formal charge to (2 Ϫ centration. The COHP curves of the MoϪS (dotted) and
x)Ϫ with x Ͻ 1, resulting in the more accurate formulation MoϪSb (dashed) interactions also exhibit a nearly maxim-
Mo4ϩSb0Sb(2Ϫx)ϪS2Ϫ(eϪ)x that hints towards metallic ized bonding interaction at the Fermi level (ICOHP:
properties.
MoϪSb ϭ Ϫ1.69; MoϪS ϭ Ϫ2.93 eV per bond). All three
kinds of interactions shown in Figure 5b comprise zero con-
tributions directly at the Fermi level.
Electronic Structure
Theoretical band calculations using the LMTO method
The SbϪSb interactions are illustrated in Figure 5c for
were carried out to study the electronic structure of two kinds of SbϪSb contacts. The first COHP curve (solid
˚
MoSb2S. The band structure of MoSb2S is illustrated in line, I) represents the shorter SbϪSb contacts Ͻ 2.9 A. Al-
Figure 4. Bands cross the Fermi level along each direction, most no antibonding states are filled, the interactions are
which strongly suggests three-dimensional metallic prop- almost optimized at the Fermi level with an averaged
erty. The different orbital contributions can be evaluated by ICOHP value of Ϫ1.66 eV per bond. The second COHP
the fat band analyses.[26] The results indicate that the orbital curve (dotted line, II) represents SbϪSb contacts within the
˚
contributions near the Fermi level stem mostly from Mo MoSbS layer (3.18 Ͻ dSbϪSb Ͻ 3.33 A). These interactions
2
dx Ϫy2 and Sb p orbitals from the interlayer Sb atoms (Sb1, exhibit antibonding characters at the Fermi level. The aver-
Sb5 and Sb6, as emphasized by the fat band representa- age ICOHP value for these SbϪSb bonds is Ϫ0.31, indicat-
tions). The contributions from the S atom are near zero at ing weaker yet still bonding character. The longer bond of
˚
the Fermi level. Thus, the MoϪMo and SbϪSb interactions the equidistant ladder of 319 A (curve III, dashed line) was
are likely responsible for the metallic behavior of MoSb2S.
considered as a half bond, which is reflected in an interme-
Density of states (DOS) curves for MoSb2S and the par- diate ICOHP value of Ϫ0.63 eV per bond. These ICOHPs
tial projections onto the Mo and Sb atoms are illustrated may be compared to those of the two different bonds in
in Figure 5. The s block of sulfur occurs in the energy win- elemental antimony, which are Ϫ1.63 for the short one of
˚
dow of Ϫ13 to Ϫ15 eV, and the broad peak between 2.91 and Ϫ0.23 eV for the longer one (3.35 A). For the
[31]
˚
Ϫ13 eV and the Fermi level at 0 eV mainly comprises S p, SbϪSb bonds of 2.91 and 3.10 A in Mo3Sb7,
we calcu-
Sb p and Mo d orbitals. There is no band gap, but a pro- lated similar ICOHPs of Ϫ0.79 and Ϫ0.40, respectively.
Eur. J. Inorg. Chem. 2002, 591Ϫ596
593

C.-S. Lee, H. Kleinke
FULL PAPER
Figure 5. DOS and COHP curves for the MoSb2S; the horizontal dashed line denotes the Fermi energy: a) total DOS (solid line), PDOS
from Mo (dashed), Sb (dotted); b) COHP curves for MoϪMo (solid line), MoϪSb (dotted), MoϪS (dashed) interactions; c) SbϪSb
˚
˚
˚
contacts (solid line: I, 2.83 Ͻ dSbϪSb Ͻ 2.88 A; dotted line: II, intralayer 3.18 Ͻ dSbϪSb Ͻ 3.33 A, dashed line: III, interlayer 3.19 A)
˚
Conclusion
Correspondingly, the half bond (3.11 A) in Zr11Sb18 ex-
hibits an ICOHP of Ϫ0.49, while the full bonds in KSb
have larger ICOHPs of Ϫ2.29 and Ϫ1.87 eV per bond.[32]
In summary, the synthesis and characterization of a new
ternary compound, MoSb2S, were presented. MoSb2S crys-
tallizes in a new three-dimensional structure type exhibiting
CdI2-related layers (MoSbS) and Sb columns (cis-trans
chains and equidistant ladders). The structure comprises
strong heteronuclear MoϪS and MoϪSb bonds, and addi-
tional MoϪMo and SbϪSb bonds. The Sb atom substruc-
ture of MoSb2S may be considered to be a partly undis-
torted variant of MoSb2Se. Both antimonide chalcogenides
can be formally derived from MoSbS layers topologically
equivalent to the β-MoTe2 layers, with additional interlayer
Sb atoms. The (formal) intercalation of anionic Sb atoms
actually enlarges the octahedral voids between the sheets,
thereby rendering an additional intercalation of cations
more facile.
Physical Properties
The temperature-dependent Seebeck coefficient measure-
ment is shown in Figure 6. The Seebeck coefficient drops
from Ϫ17 µV/K at 300 K to Ϫ30 µV/K at 600 K. The nega-
tive sign indicates that electrons are the main charge car-
riers. Typical magnitudes of Seebeck coefficients at 300 K
range from a few µV/K for most metals to 10Ϫ20 µV/K
for some transition metal elements.[1] The observed Seebeck
coefficient of MoSb2S is at the upper end for metallic mat-
erials.
MoSb2S exhibits metallic behavior as predicted by elec-
tron counting, electronic structure calculations and con-
firmed by physical property measurements (Seebeck coeffi-
cients). The results from the electronic structure calcula-
tions show the smallest number of states directly at the
Fermi level; thus, changing the valence-electron concentra-
tion is likely to further increase the number of itinerant val-
ence electrons, a disadvantage when looking for thermoelec-
tric materials.
Experimental Section
Synthesis: All starting materials were stored in an Argon-filled
glove box. The starting materials were molybdenum (Aldrich,
99.9%), antimony powder (Aldrich, 99.8%, melting point 903 K,
Figure 6. Temperature-dependent Seebeck coefficients of MoSb2S
594
Eur. J. Inorg. Chem. 2002, 591Ϫ596

Synthesis, Structure, and Thermoelectric Properties of the New Antimonide Sulfide MoSb2S
FULL PAPER
99.5%), sulfur (Aldrich, 99.98%), Mo3Sb7 and Sb2S3. Mo3Sb7 was
synthesized by starting from the stoichiometric ratio of Mo/Sb ϭ
3:7 and heating at 973 K for 3 d in an evacuated fused silica tube,
whereas Sb2S3 was prepared at 773 K within 3 d in an evacuated
fused silica tube. Initial reactions were carried out by preparing
mixtures in stoichiometric ratios of the elements with a small
Table 1. Crystal data and conditions of data collection for MoSb2S
Refined composition
Instrument
MoSb2S
Smart APEX CCD
371.5
Molecular mass
Crystal size [mm]
Space group, Z
0.16 ϫ 0.04 ϫ 0.01
C2/m, 16
36.157(3)
6.3823(5)
6.5379(5)
95.093(2)
1502.8(2)
34.99°
amount of I2 as the transport reagent, in a silica tube, sealed under
˚
a [A]
Ϫ3
vacuum (Ͻ 10
mbar), and heated at 1073 K for 3 d. Black,
˚
b [A]
needle-shaped crystals of MoSb2S and mixtures of Sb and binary
phases (Mo3Sb7, MoS2) were observed from these reactions. Pure
MoSb2S can be synthesized either by heating stoichiometric
amounts of the elements at 873 K for 3 d or annealing binary com-
pounds Mo3Sb7 and Sb2S3 in a 1:1 ratio at 973 K for 2 d. The
purity of MoSb2S was checked by powder X-ray diffraction and
elemental analyses (EDS, see below). The second reaction route
(Mo3Sb7 ϩ Sb2S3) yielded MoSb2S and elemental antimony, that
crystallized on different sides of the reaction container. Higher re-
action temperatures (Ͼ 1123 K) for both reaction routes led to the
disproportionation of MoSb2S into MoS2, Mo and Sb. MoSb2S is
stable in air for at least two weeks.
˚
c [A]
β [°]
V [A ]
3
˚
2θmax
Reflections collected
Independent, observed reflections (Rint
dcalcd. [g/cm3]
12316
3506, 1553 (0.043)
6.568
17.86
0.82Ϫ1.0
298(2)
0.035, 0.059
0.80
2.54/Ϫ2.53
)
Absorption coeff. [mmϪ1
Transmission range
Temperature [K]
]
R1[a], wR2 [I Ͼ 2σ(I)]
Goodness-of-fit on F2
3
˚
∆ρ(max/min) [e/A ]
[a]
2
2 2
2 2
R1 ϭ Σ||Fo| Ϫ |Fc||/Σ|Fo|; wR2 ϭ [Σ[w(Fo Ϫ Fc ) ]/Σ[w(Fo ) ]]1/2
.
Structure Determination: Needle-shaped crystals for single-crystal
X-ray diffraction experiments were selected under an optical micro-
scope and then mounted on glass fibers. Single-crystal X-ray dif-
fraction data were collected with the use of graphite-monochromat-
˚
ized Mo-Kα radiation (λ ϭ 0.71073 A) at 298(2) K with a Bruker
˚
APEX CCD diffractometer. The crystal-to-detector distance was
4.550 cm. Data were collected by a scan of 0.3° in ω groups of 606
frames each at ϕ settings of 0°, 60°, 120°. The exposure time was
120 s/frame. The 2θ values varied between 3° and 70°. The data
were corrected for Lorentz and polarization effects. Absorption
corrections were based on fitting a function to the empirical trans-
mission surface as sampled by multiple equivalent measure-
ments.[33] Diffraction peaks obtained from all frames of reciprocal
space images were used to determine the unit cell parameters. Ini-
tially, the data revealed a primitive unit cell similar to the one of
MoSb2Se and additional weak reflections, which suggested a su-
percell. The unit cell parameters of MoSb2S indicated monoclinic
symmetry, as later proven by a comparison of symmetry-equivalent
reflections (Rint ϭ 0.043). The systematic absences revealed C-cent-
ering. Therefore, the space group C2/m was chosen for subsequent
structural analyses. The structure model was obtained by direct
methods and refined by full-matrix least-squares refinement based
on F2 with the SHELXTL5.12 package.[34] Table 1 summarizes the
crystallographic data of MoSb2S. Table 2 lists selected interatomic
distances. Further details of the crystal-structure investigations may
be obtained from the Fachinformationszentrum Karlsruhe, 76344
Eggenstein-Leopoldshafen, Germany, on quoting the depository
number CSD-412092. All Mo and S atoms exhibit reasonable ther-
mal displacement parameters. However, the Sb5 and Sb6 atoms
show 2Ϫ3 times higher thermal displacement parameters along
each main direction, which might suggest an even higher order of
supercell or lower symmetry. However, lowering the symmetry, for
example by refining in C2 or Cm, did not result in any improve-
ments. Also, we found no evidence for the existence of a higher su-
perstructure.
Table 2. Selected bond lengths [A] for MoSb2S
Sb1
Sb2
Sb1
Sb1
Sb3
Sb3
Sb4
Sb5
Mo1
Mo2
Mo1
Mo2
Mo3
Sb3
Sb3
Sb4
2.839(1)
2.878(3)
2.844(2)
3.3294(7)
3.1912(2)
2.841(2)
2.840(2)
2.823(2)
2.826(2)
2.844(2)
2.824(2)
3.1846(9)
3.1977(9)
3.3283(7)
Sb5
Sb5
Sb6
2.853(4)
3.1912(3)
Mo1
Mo2
Mo3
S2
2.7650(9)
3.1912(3)
2.382(4)
2.390(5)
3.188(1)
3.194(1)
2.7596(9)
2.382(4)
2.389(5)
2.381(4)
2.388(5)
2.380(4)
3.1912(3)
3.188(3)
3.195(3)
S3
Mo2
Mo2
Mo2
Mo3
S1
S2
S3
S1
S2
S3
S2
Sb3
Sb4
Mo3
Sb6
Mo2
Mo3
2.842(2)
2.823(2)
2.846(2)
S1
S2
S2
Thermopower Measurements: Seebeck coefficients were determined
on a cold-pressed bar by a commercial thermopower measurement
apparatus (MMR Technologies) in the temperature range of 300 to
600 K, using Constantan as an internal standard. Silver paint
(AMI DODUCO Technology) was used to create electric contacts.
Electronic Structure Calculations: The tight-binding LMTO (Linear
Muffin Tin Orbitals)[35Ϫ37] band calculations were carried out in
order to understand the electronic structure of MoSb2S. The integ-
ration in k space was performed by an improved tetrahedron
method[38] on a grid of 16 ϫ 16 ϫ 8 irreducible k points of the first
Elemental Analysis: Energy dispersive spectroscopy (EDS, LEO
1050) was performed on the crystalline and powdered samples. The
Mo/S ratio could not be determined due to the severe overlap of Brillouin zone (i.e. 554 of 2048 k points). We analyzed the elec-
the peaks of Mo (Lα ϭ 2.293 eV) and S (Kα ϭ 2.307 eV). No im- tronic structure by extracting information from the band structure,
purities of other elements (like iodine, silicon or oxygen) were de- Densities of states (DOS), and Crystal Orbital Hamiltonian Popu-
tected.
lation curves (COHP).[39]
Eur. J. Inorg. Chem. 2002, 591Ϫ596
595

C.-S. Lee, H. Kleinke
H. Kleinke, Eur. J. Inorg. Chem. 1998, 1369.
H. Kleinke, Chem. Commun. (Cambridge) 1998, 2219.
M. J. Ferguson, R. W. Hushagen, A. Mar, J. Alloys Compd.
1997, 249, 191.
H. Kleinke, Chem. Soc. Rev. 2000, 29, 411.
G. Papoian, R. Hoffmann, J. Am. Chem. Soc. 2001, 123, 6600.
G. A. Papoian, R. Hoffmann, Angew. Chem. Int. Ed. 2000,
39, 2409.
W. Müller, Z. Naturforsch., Teil B 1977, 32, 357.
H. Hayakawa, S. Ono, J. Less-Common Met. 1988, 144, 177.
W. Klemm, Z. Anorg. Allg. Chem. 1941, 247, 1.
E. Zintl, Angew. Chem. 1939, 1, 52.
O. Jepsen, O. K. Andersen, Z. Phys. 1995, 97, 25.
R. S. Mulliken, J. Chem. Phys. 1955, 23, 2343.
T. Hughbanks, R. Hoffmann, J. Am. Chem. Soc. 1983, 105,
3528.
R. Hoffmann, J. Chem. Phys. 1963, 39, 1397.
M.-H. Whangbo, R. Hoffmann, J. Am. Chem. Soc. 1978, 100,
6093.
E. Dashjav, A. Szczepenowska, H. Kleinke, J. Mater. Chem.,
in press.
I. Elder, C.-S. Lee, H. Kleinke, Inorg. Chem., in press.
SAINT Version 4, Siemens Analytical X-ray Instruments Inc.,
Madison, WI., 1995.
FULL PAPER
[16]
[17]
[18]
Acknowledgments
Financial support from the Canadian Foundation for Innovation,
the Ontario Innovation Trust, the Materials Manufacturing Onta-
rio, the Province of Ontario, and the National Science and Engin-
eering Research Council of Canada is appreciated.
[19]
[20]
[21]
[1]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
D. M. Rowe, CRC Handbook of Thermoelectrics, CRC Press,
Boca Raton, FL, 1995.
D.-Y. Chung, T. Hogan, P. Brazis, M. Rocci-Lane, C. Kannew-
[2]
urf, M. Bastea, C. Uher, M. G. Kanatzidis, Science
(Washington, D. C.) 2000, 287, 1024.
T. M. Tritt, M. G. Kanatzidis, G. D. Mahan, H. B. J. Lyon, in
[3]
Materials Research Society Symposium Proceeding, Materials
Research Society, Warrendale, PA, 1999, vol. 545.
[4]
[29]
[30]
B. C. Sales, D. Mandrus, R. K. Williams, Science 1996, 272,
1325.
[5]
F. J. DiSalvo, Science 1999, 285, 703.
[6]
[31]
H. Kleinke, Chem. Commun. (Cambridge) 2000, 1941.
[7]
B. E. Brown, Acta Crystallogr. 1966, 20, 268.
[8]
[32]
[33]
A. Brown, Nature 1965, 206, 502.
[9]
K. D. Bronsema, J. L. De Boer, F. Jellinek, Z. Anorg. Allg.
Chem. 1986, 540, 15.
P. Villars, L. D. Calvert, Pearson’s Handbook of Crystallo-
[10]
[34]
SHELXTL Version 5.12, Reference Manual, Siemens Analyt-
ical X-ray Systems, Inc, Madison, WI, 1996, 1995.
U. van Barth, L. Hedin, J. Phys. C 1971, 4, 2064.
O. K. Andersen, Phys. Rev. B 1975, 12, 3060.
H. L. Skriver, The LMTO Method, Springer, Berlin, 1984.
P. E. Blöchl, O. Jepsen, O. K. Andersen, Phys. Rev. B: Condens.
Matter 1994, 49, 16223.
graphic Data for Intermetallic Phases, 2nd ed., ASM Interna-
tional, Metals Park, OH, 1991.
L. Pauling, The Nature of the Chemical Bond, 3rd ed., Cornell
[35]
[36]
[37]
[38]
[11]
University, Ithaca, NY, 1948.
[12]
W. Hönle, H.-G. von Schnering, Z. Kristallogr. 1981, 155, 307.
[13]
N. Korber, F. Richter, Angew. Chem. Int. Ed. Engl. 1997, 36,
1512.
U. Bolle, W. Tremel, J. Chem. Soc., Chem. Commun. 1992, 91.
H. Holseth, A. Kjekshus, Acta Chem. Scand. 1969, 23, 3043.
[39]
R. Dronskowski, P. E. Blöchl, J. Phys. Chem. 1993, 97, 8617.
Received September 13, 2001
[14]
[15]
[I01360]
596
Eur. J. Inorg. Chem. 2002, 591Ϫ596