Synthesis, crystal structure, and thermoelectric properties of clathrates in the Sn-In-As-I system

Article abstract of DOI:10.1002/zaac.201100287

Clathrates of the general formula Sn_{24-x- γ}In_xAs_22-yI₈ have been synthesized with x_max = 10.7. They crystallize in the clathrate-I type of crystal structure, space group Pmβar{3}ν, with the cubic unit cell parameter ranging from a = 11.088 (x = 0) to a = 11.343 (x = 10.7) A. The crystal structure refined for five compositions with different indium contents revealed that despite the linear dependence of the unit cell parameter upon x the details of the crystal structure change twice upon varying the indium contents. These changes occur without altering the space group and show up in positional disorder of atoms within the framework and the concomitant change of the coordination of part of tin atoms from 3+2 to 3+3, which is additionally confirmed by the Moessbauer spectroscopy data. In general, these changes reflect the transition from the Sn₂₄P _19.3I₈ structure type for high indium contents to the Sn_20.5As₂₂I₈ structure type for low indium contents. It is shown that the Sn_24-x-γIn _xAs_22-yI₈ clathrates have n-type conductivity with high values of the Seebeck coefficient ranging from -322 to -594 γVA·K^-1 at room temperature. They display an abnormally low thermal conductivity at 293 K, from 0.36 to 0.57 WA·m^-1A·K^-1. The calculated thermoelectric figure-of-merit for Sn_24-x-γIn _xAs_22-yI₈ has its room-temperature maximum at 0.041 for x = 6.5. Copyright

Full text of DOI:10.1002/zaac.201100287

ARTICLE
DOI: 10.1002/zaac.201100287
Synthesis, Crystal Structure, and Thermoelectric Properties of Clathrates in
the Sn-In-As-I System
Evgeny A. Kelm,[a,b] Andrei V. Olenev,[c] Mikhail A. Bykov,[a] Alexey V. Sobolev,[a]
Igor A. Presniakov,[a] Vladimir A. Kulbachinskii,[d] Vladimir G. Kytin,[d] and
Andrei V. Shevelkov*[a]
Keywords: Clathrates; Tin; Thermal conductivity; Thermoelectric materials
Abstract. Clathrates of the general formula Sn24–x–δInxAs22–yI8 have of tin atoms from 3+2 to 3+3, which is additionally confirmed by
been synthesized with xmax = 10.7. They crystallize in the clathrate-I the Mössbauer spectroscopy data. In general, these changes reflect the
¯
type of crystal structure, space group Pm3n, with the cubic unit cell
transition from the Sn24P19.3I8 structure type for high indium contents
to the Sn20.5As22I8 structure type for low indium contents. It is shown
parameter ranging from a = 11.088 (x = 0) to a = 11.343 (x = 10.7) Å.
The crystal structure refined for five compositions with different in- that the Sn24–x–δInxAs22–yI8 clathrates have n-type conductivity with
dium contents revealed that despite the linear dependence of the unit
cell parameter upon x the details of the crystal structure change twice
upon varying the indium contents. These changes occur without alter-
ing the space group and show up in positional disorder of atoms within
the framework and the concomitant change of the coordination of part
high values of the Seebeck coefficient ranging from –322 to –594
μV·K–1 at room temperature. They display an abnormally low thermal
conductivity at 293 K, from 0.36 to 0.57 W·m–1·K–1. The calculated
thermoelectric figure-of-merit for Sn24–x–δInxAs22–yI8 has its room-
temperature maximum at 0.041 for x = 6.5.
erature, which are labeled with the Latin numbers I–IV, VIII,
and IX according to the classification of Jeffrey;[1] recently,
the new type, clathrate-X, was discovered.[7] Current resurrec-
tion of interest into these clathrates is attributed to their pro-
spective thermoelectric properties.[8–14] Good thermoelectric
material should exhibit good transport of charge carrier and
poor transport of heat. Such a combination of properties is very
rare; moreover, these properties are coupled and their indepen-
dent optimization is almost impossible.[15] Intermetallic clath-
rates show spatial separation of host and guest substructures,
which ensures effective charge carrier transport through the
covalently-bonded framework and simultaneous scattering of
heat-carrying phonons by guest atoms that rattle inside the
oversized cages of the framework.[16,17] Thus, they show good
prospects for almost independent optimization of the charge
carriers and heat transport.
Introduction
Clathrates are peculiar inclusion compounds that feature se-
questering of guest atoms inside the polyhedral cages of the
host framework.[1,2] Apart from clathrates–gas hydrates,[3,4]
there exists a vast family of inorganic/intermetallic clathrates,
the frameworks of which are built mainly of the group 14 ele-
ments whereas large either electropositive or electronegative
atoms fill the polyhedral cages compensating for the charge
of the framework. Depending on the framework charge, the
intermetallic clathrates are subdivided into frequently observed
anionic clathrates and more rare cationic ones.[5,6] Currently,
six types of intermetallic clathrates are documented in the lit-
* Prof. Dr. A. Shevelkov
Fax: +7-495-939-4788
E-Mail: shev@inorg.chem.msu.ru
[a] Chemistry Department
Recently, we reported the crystal growth and thermoelectric
properties of two cationic clathrates Sn17.8In4.0As21.7I8 and
Sn15.1In8.0As21.5I8,[18] which belong to the clathrate-I structure
type.[19] It was shown that these compounds demonstrate n-
type conductivity with high absolute values of the Seebeck
coefficient, up to 600 μV·K–1 and at the same time possess
an abnormally low thermal conductivity, not exceeding
0.4 W·m–1·K–1 at room temperature. In this paper, we report
on the family of clathrates in the Sn-In-As-I system with
Sn24–x–δInxAs22–yI8 with xmax = 10.7. We present and discuss
the methods of their synthesis and crystal growth, peculiarities
of their local and extended crystal structure, and their thermo-
Lomonosov Moscow State University
Leninskie Gory 1–3
Moscow 119991, Russia
[b] Present address: Institute of Microelectronics Nanotechnologies
Russian Academy of Sciences
Moscow, Russia
[c] “SineTheta“ Ltd.
Scientific Park of Lomonosov Moscow State University
Moscow, Russia
[d] Faculty of Physics
Lomonosov Moscow State University
Leninskie Gory 1–2
Moscow, Russia
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/zaac.201100287 or from the author. electric properties as a function of indium contents.
Z. Anorg. Allg. Chem. 2011, 637, 2059–2067
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
2059

A. V. Shevelkov et al.
ARTICLE
Results and Discussion
Synthesis and Characterization
Initially, it was assumed that the target compositions form
upon substituting indium for tin in the ternary clathrate
[20]
Sn20.5As22I8
and that, similar to the parent compound, they
adhere to the Zintl electron counting scheme, which relates the
number of vacancies and substituting atoms (indium in this
case), keeping the whole compound electrically balanced.[5]
The starting compositions were selected in accordance with
this assumption for the general formula Sn24–x–δInxAs22–y 8
I
with x varying from 1 to 14 with the step of 0.2–1 and the
parameters δ and y linked to x to maintain electroneutrality.
Two different routes were followed for the synthesis of the
target clathrates. According to the first scheme, the target com-
positions were synthesized from the mixture of stoichiometric
amounts of elements (Sn, In, and As) and SnI4. However, an-
nealing at different temperatures, from 723 to 973 K always
resulted in an incomplete reaction, with tin and indium arse-
nides being the admixtures. It should be noted that tin and
indium are liquids at the temperature of reaction, and solid
products, such as their arsenides, form on the surface of the
melt, preventing the equilibrium achievement in the system
even after the second annealing that followed regrinding. In
the second method of synthesis, we used pre-synthesized SnAs
and InAs as precursors for accelerating the achievement of
equilibrium. In this case, the reaction involves only ground
and tightly mixed solid components and the gas phase. By this
Figure 1. Dependence of the unit cell parameter on the indium content.
Solid line shows linear fit up to x = 10.7.
SnAs with the concomitant increase of the unit cell parameter
of the remaining clathrate phase (Figure 2). The TG study of
the sample with the same starting composition performed un-
der argon flow shows that the mass loss starts at 623(1) K,
whereas heating the sample to 623 K in air leads to no weight
change. This suggests that if pressure is not maintained the
decomposition due to liberation of volatile products starts at
623 K, otherwise the compound is stable until 845.5 K.
Crystal Structure Determination
method we were able to obtain samples of Sn24–x–δInxAs22–y 8
with the content of admixtures (unreacted SnAs) not greater
I
Several crystal structures with different indium content were
than 2% by mass, as revealed by the Rietveld refinement, with refined but only five of them were selected for further analysis
the optimal annealing temperature ranging from 803 to 823 based on the quality of the data. They demonstrate crystal
K. Similarly, the optimal synthetic conditions for Sn20.5As22I8 structure solutions with R1 Ͻ 0.035, which ensures the accu-
resulted in almost 1 mass% of SnAs as a sole admixture.[20]
racy of the atomic parameters suitable for the discussion of the
The X-ray analysis of the samples obtained after the second fine structure details, and show a sufficient spread of the cubic
annealing showed that the target clathrate phase formed for all unit cell parameter, 11.186 Å Ͻ a Ͻ 11.348 Å, which corre-
values of x. However, the inspection of the diffractograms sponds to the indium content varying from x = 4.1 to x = 9.5
showed that for the samples with x Ͼ 10.7 the admixture con- (Table 1).
tent starts to grow significantly, with InAs becoming the main
The analysis of the systematic extinction conditions con-
side phase. The cubic unit cell parameter was calculated for firmed that all five crystal structures retain the space group
the samples with x Յ 10.5 and for x = 14 for comparison; its Pm-3n typical for the type-I clathrates.[5,6] The refinement of
dependence upon the indium content is shown in Figure 1. the crystal structures was based on the crystallographic data
There is some scatter, but generally the dependence is linear for Sn24P19.3I8.[21] At the first step, the guest atomic positions
until the maximal indium content is reached at x = 10.7. It (2a and 6d) were set as iodine atoms, the 16i and 6c positions
should be noted that we observed extra reflections on the X- were set as arsenic atoms, and the 24kЈ and 24kЈЈ positions
ray patterns of the samples with x = 1 and 2, which were inde- were partially filled by tin atoms with their occupancy sum
xed by doubling the unit cell parameter and choosing the face- fixed as unity. The subsequent refinement showed the presence
centered cubic space group, similar to Sn20.5As22I8 (see Figure of one more electron density peak near two 24-fold positions.
S1 of Supporting Information). Therefore, the clathrates retain This was set as the third partially filled position of tin (24kЈЈЈ)
the structure of Sn20.5As22I8 if indium content is small.
and introduced into refinement, after which no more chemi-
DSC and TG investigations were performed for the sample cally meaningful peaks were found. At the next step, the coor-
with x = 6. The DSC measurements indicate that the reversible dination of three positions deliberately set as occupied by tin
decomposition occurs at 845.5(5) K. To analyze the products was considered. Three coordination modes were observed.
of the decomposition, the pre-synthesized sample of the same Two of them are relevant to the structure with x = 9.5 only,
composition was annealed at 860 K for 48 hours and then where two atoms have a distorted tetrahedral environment of
quenched into icy water. The X-ray pattern of the product re- three arsenic and one metal atoms, while the third atom has a
vealed that upon decomposition the starting clathrate liberates distorted 3+3 environment composed of one metal and two
2060
www.zaac.wiley-vch.de
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2011, 2059–2067

Clathrates in the Sn-In-As-I System
Figure 2. XRD pattern for the clathrate sample with x = 6 before (right) and after (left) annealing at 860 K. The positions of the two highest
peaks of SnAs are marked.
Table 1. Crystal structure data collection and refinement parameters for Sn17.5(3)In4.1(1)
ᮀ
2.4As21.97(5)
ᮀ
0.03I8 (I), Sn17.0(3)In4.80(9)
ᮀ2.2As21.95(5)
ᮀ
0.05I8 (II), Sn17.1(3)In5.0(1)
ᮀ
1.9As21.76(6)
ᮀ
0.24I8 (III), Sn15.8(2)In6.9(1)
ᮀ
1.3As21.62(6)
ᮀ
0.38I8 (IV), Sn14.0(2)In9.5(1)
ᮀ
0.5As21.42(5) 0.58I8 (V).
ᮀ
Crystal number
Space group
Data collection temperature /K
Cell parameter, a /Å
V /Å3
I
II
III IV
V
¯
Pm3n (#223)
293(2)
11.1680(10)
1392.9(2)
11.1819(8)
1398.1(2)
11.1970(10)
1403.8(2)
1
11.2590(10)
1427.2(2)
11.3480(10)
1461.4(2)
Z
Density (calcd.) /g·cm–3
6.210
26.738
6.210
26.690
6.209
6.169
6.131
25.871
μ /mm–1
26.595
26.270
Data collection range /deg.
Reflections collected
Independent reflections
Parameters refined
R1 [Fo Ͼ 4σFo]
2.58Ͻθ Ͻ32.94
2733
500 [Rint = 0.060] 467 [Rint = 0.076] 397 [Rint = 0.053]
2.58ϽθϽ31.93
2544
2.57ϽθϽ29.99
2155
2.56ϽθϽ29.94
2266
403 [Rint = 0.062] 411 [Rint = 0.064]
2.54ϽθϽ29.94
2238
42
42
33
33
31
0.0285
0.0523
1.09 and –1.19
1.142
0.0286
0.0601
1.01 and –1.38
0.955
0.0325
0.0693
1.41 and –1.49
1.109
0.0299
0.0557
0.69 and –1.05
1.155
0.0263
0.0639
0.72 and –1.15
1.030
wR2 (all data)
Largest diff. peak and hole /e·Å–3
Goodness-of-fit
arsenic atoms in the first coordination sphere and three more fold positions and of the 6c position were left varied with the
distant metal atoms. The latter is of the same type as observed sole constraint that relates the total number of vacancies and
for one sort of tin atoms in Sn24P19.3X8 (X = Br, I).[22] In all the occupancy of the indium position such that the electroneu-
other crystal structures, the distorted tetrahedral coordination trality of the crystal structure is maintained according to the
remains, but the 3+3 coordination is replaced with the 3+2 Zintl counting scheme. Finally, the occupancy factors for the
environment made of three arsenic atoms and two distant io- guest positions were checked and found to be equal to unity,
dine atoms occupying the 6d position, which is similar to the and no sign of positional disorder was found. Using such con-
environment of one sort of the tin atoms in the crystal structure ditions, all five structures were refined anisotropically to get
of Sn19.3Cu4.7As22I8.[23] Based on the typical coordination, the compositions and atomic parameters listed in Table 1 and
these atoms were set as tin in all refinements. The remaining Table 2. It should be noted that the refined compositions were
two atoms having very similar coordination were differentiated in good agreement with those obtained by EDX analysis.[18]
according to different M–As bond length (M = Sn, In). As
Also, a crystal was selected from the sample with x = 2.0.
indium has a slightly greater atomic radius than tin,[24] we as- The analysis of the crystal data and systematic extinction con-
sumed that tin occupies the position with shorter M–As bonds; ditions showed the doubled F-centered cubic cell with the unit
the observed difference varied from 0.01 to 0.07 Å for five cell parameter a = 22.249(2) Å and possible space groups F23
structures. After distributing tin and indium atoms over three and Fm3. The refinement was tried in both space groups and
24-fold positions we took notice of a possible vacancy forma- revealed a severe disorder of the framework, similar to the
tion. It is documented in the literature that vacancies form in situation observed for the crystal structure of Sn20.5As22I8.[20]
[20]
the 24-fold positions in the crystal structure of Sn20.5As22I8
Similar to the latter, no better crystal structure solution than
but in the 6c position in the crystal structure of with R = 0.17 was found. Consequently, correct distribution of
Sn24P19.3I8.[21,22] Therefore, the occupancy factors of all 24- vacancies and indium and tin atoms over several atomic sites
Z. Anorg. Allg. Chem. 2011, 2059–2067
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2061

A. V. Shevelkov et al.
ARTICLE
Table 2. Atomic parameters for the crystal structures I–V. W denotes Wykoff position; U is always U(equivalent) in Å2; unless otherwise stated,
Sof = 1 and x/a = 0.
Atom
I
II
III
IV
V
I(1) at 0,0,0
w = 2a
w = 2a
w = 2a
w = 2a
w = 2a
U = 0.0094(2)
U = 0.0118(3)
w = 6d
U = 0.0190(3)
w = 24k
U = 0.0150(4)
w = 6d
U = 0.0225(3)
w = 24k
U = 0.0132(3)
w = 6d
U = 0.0216(3)
w = 24k
U = 0.0124(3)
w = 6d
U = 0.0213(3)
w = 24k
I(2) at 0,¼,½ w = 6d
U = 0.0168(2)
w = 24k
Sn(1)
y/b = 0.3074(4)
z/c = 0.1265(5)
U = 0.0122(4)
Sof = 0.63(1)
w = 24k
y/b = 0.3094(6)
z/c = 0.1253(5)
U = 0.0141(5)
Sof = 0.61(1)
w = 24k
y/b = 0.3099(8)
z/c = 0.1236(9)
U = 0.0191(9)
Sof = 0.63(1)
w = 24k
y/b = 0.3096(5)
z/c = 0.1217(6)
U = 0.0162(7)
Sof = 0.61(2)
w = 24k
y/b = 0.3100(7)
z/c = 0.1207(3)
U = 0.011(1)
Sof = 0.50(1)
w = 24k
In(1)
y/b = 0.3294(12) z/c =
0.123(2)
y/b = 0.321(2)
y/b = 0.321(3)
y/b = 0.319(1)
y/b = 0.312(1)
U = 0.014(2)
Sof = 0.17(1)
z/c = 0.127(3)
U = 0.032(5)
Sof = 0.20(1)
w = 24k
z/c = 0.131(3)
U = 0.033(5)
Sof = 0.21(2)
w = 24k
z/c = 0.129(1)
U = 0.032(3)
Sof = 0.29(2)
w = 24k
z/c = 0.1201(8)
U = 0.025(3)
Sof = 0.40(1)
w = 24k
Sn(2)
As(1)
w = 24k
y/b = 0.3269(4)
z/c = 0.0482(6)
U = 0.023(1)
Sof = 0.101(2)
w = 48l
y/b = 0.3264(6)
z/c = 0.049(1)
U = 0.024(2)
Sof = 0.094(4)
w = 48l
y/b = 0.3258(8)
z/c = 0.049(2)
U = 0.023(3)
Sof = 0.081(4)
w = 16i
y/b = 0.326(1)
z/c = 0.050(2)
U = 0.021(4)
Sof = 0.047(4)
w = 16i
y/b = 0.351(2)
z/c = 0.148(2)
U = 0.015(3)
Sof = 0.08(1)
w = 16i
x/a = 0.185(3)
x/a = 0.197(1)
x/a = y/b = z/c =
0.18822(6)
x/a = y/b = z/c =
0.18842(4)
x/a = y/b = z/c =
0.18831(4)
y/b = 0.180(2)
z/c = 0.198(1)
U = 0.012(1)
Sof = 0.3333
w = 24k
y/b = 0.184(2)
z/c = 0.183(3)
U = 0.016(2)
Sof = 0.3333
w = 24k
U = 0.0222(3)
U = 0.0168(2)
U = 0.0099(2)
As(2)
w = 12g
w = 12g
w = 6c
x/a = 0.986(1)
y/b = 0.5
z/c = 0.2682(3)
U = 0.020(2)
Sof = 0.249(1)
x/a = 0.987(2)
y/b = 0.5
z/c = 0.2671(7)
U = 0.024(3)
Sof = 0.248(2)
y/b = ½
y/b = ½
y/b = ½, z/c = ¼
U = 0.0104(4)
Sof = 0.903(5)
z/c = 0.2341(9)
U = 0.028(2)
Sof = 0.480(4)
z/c = 0.238(1)
U = 0.019(2)
Sof = 0.468(4)
was not determined. The crystal data are given in Supporting
Information (Tables S1 and S2).
Crystal Structure Description and Evolution
The clathrates in the Sn-In-As-I system have the crystal
structure of the clathrate-I type. Two guest atoms, I1 and I2,
sit in the centers of the polyhedral cages formed by 20 and
24 framework atoms, respectively (Figure 3). The 20-vertex
dodecahedra and 24-vertex tetrakaidecahedra alternate in the
crystal structure in the 1:3 ratio. The framework is composed
of tin, indium, and arsenic atoms. Its structure is slightly dif-
Figure 3. Crystal structure of Sn14.0In9.5As21.42I8 (left). Large grey, io-
dine; medium dark, metals; small light grey, arsenic. Polyhedral pres-
ferent from that observed for the ideal type-I clathrates in that entation of the clathrate-I structure (right). Polyhedra around I1, dark
grey; polyhedra around I2, light grey. Unit cell is shown with black
solid lines.
the 24-fold position is split into three closely lying positions
that alternate. As a consequence, not all framework atoms dis-
play tetrahedral coordination typical for all clathrate-type
structures.[5] Particularly, the Sn2 atoms have either 3+2 or splits into three and not two, and the vacancies are observed
3+3 coordination, depending on the indium content.
both in the pnicogen and metal substructures. Upon lowering
The dependence of the unit cell parameter upon the indium the indium content, the coordination of the Sn2 atom changes
content is linear as shown in Figure 4. However, on going from to 3+2 (Figure 5). At the same time, the As2 atom shifts from
the maximal (x = 9.5) to minimal (x = 4.1) indium content the its ideal 6c position towards the 12g position; however, the
structural details change twice. For x = 9.5, the Sn2 atom dis- vacancies remain present in both metal and pnicogen substruc-
plays the 3+3 coordination (Figure 5), resembling the atomic tures. Such a structural variation has no analogs in the clathrate
coordination in the crystal structure of Sn24P19.3X8 structures, although
a similar 3+2 coordination of tin
(X = Br, I)[21,22] with just two exceptions: the 24-fold position was observed previously in the crystal structure of
2062
www.zaac.wiley-vch.de
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2011, 2059–2067

Clathrates in the Sn-In-As-I System
Sn19.3Cu4.7As22I8.[23] However, the latter exhibits no vacancies
at all. For the lowest indium content (x = 4.1 and 4.7), the shift
of the As2 atom is even greater, it now occupies the 24k posi-
tion; besides, the As1 atom is displaced from its ideal 16i posi-
tion towards the general position 48l. Also, the analysis of
the occupancy factors for the arsenic atoms indicates that the
concentration of vacancies in the As2 position is low for x =
4.1 and 4.7 and falls within the accuracy of its determination
(Figure 4). The trends in selected interatomic distances that
accompany the changes in the fine structural details are de-
picted in Figure 6. Finally, we note that, despite the different
degree of disorder, the equivalent atomic displacement param-
eters for the I2 guest atom are nearly twice those for the I1
guest atom (Table 2), which is observed for the majority of the
type-I clathrate compounds.[5,8]
Figure 5. Coordination of the Sn2 atom in the crystal structures of
clathrates with different indium content: x = 9.5 (left) and x = 6.9
(right).
Figure 4. Dependence of the unit cell parameter on the indium content,
x, from the single crystal data; the linear fit is shown with the solid
line (top). Plot of the arsenic contents, n, vs. the indium contents, x,
from the single crystal data (bottom).
Figure 6. Plot of the host-guest distances in the crystal structures vs.
indium content, x. Lines are drawn to guide an eye.
Mössbauer Study
admixture of tin oxide derivatives on the surface, as was pre-
The existence of different types of coordination of tin atoms viously observed for other tin clathrates.[25] The second
was confirmed by 119Sn Mössbauer spectroscopy. Figure 7 impurity component with the hyperfine parameters
shows the spectra collected at 293 K for two samples of the δ = 3.25(4) mm·s–1 and Δ = 0.69 mm·s–1 clearly shows that it
composition Sn17.8In4As21.7I8
and Sn13.3In10.7As21.3I8. Both arises from the Sn2+ compound, which nature remains unclear.
spectra are best fitted as the superposition of major non-mag- In general, it should be noted that the contents of the impurity
netic doublets with minor impurity doublets, the hyperfine pa- phases is always overestimated in this type of fits,[25] and we
rameters of which are summarized in Table 3. Obviously, the believe that the total quantity of the impurities is negligibly
component with the zero isomer shift (δ) corresponds to the low.
Z. Anorg. Allg. Chem. 2011, 2059–2067
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2063

A. V. Shevelkov et al.
ARTICLE
Sn13.3In10.7As21.3I8 merely reflects greater distortion of the tin
tetrahedral environment in this compound. The principal dif-
ference in the two spectra is the unmatched hyperfine param-
eters for the Sn2 component (Table 3), which reflects the dif-
ference in the coordination of the respective tin atoms, 3+2 in
Sn17.8In4As21.7I8 and 3+3 in Sn13.3In10.7As21.3I8. The limited
amount of the literature data makes the comparison difficult;
however, it should be noted that the value of the quadrupole
splitting for the latter compound (Δ = 0.55 mm·s–1) resembles
that observed for Sn24P19.3I8 and Sn14In10P21.2I8 (Δ = 0.44 and
0.38 mm·s–1 respectively). In the latter compounds, tin atoms
,
have the analogous 3+3 environment with the phosphorus
atoms in place of the arsenic ones.[26] Figure 5 shows that the
environment of the Sn2 atom is much less symmetric in
Sn17.8In4As21.7 8 than in Sn13.3In10.7As21.3I8 that probably
I
leads to higher quadrupole splitting. Concluding, the 119Sn
Mössbauer spectra confirm the existence of two types of tin
atom coordination in each investigated clathrate, give evidence
for the structural transformation with changing indium content,
and hint for their surface oxidation.
Thermoelectric Properties
Our preliminary study[18] showed that the crystals of
Sn24–x–δInxAs22–yI8 with x = 4 and 8 behave as n-type semi-
conductors, display high room-temperature values of the See-
beck coefficient, S = |375–600| μV·K–1, and possess anoma-
lously low thermal conductivity, k Յ 0.4 W·m–1·K–1, typical
for amorphous solids. In this work, we measured thermoelec-
tric properties of the samples with x = 3, 6.5, and 9.5. We
summarize and discuss the results of all measurements below.
Figure 8 shows the results obtained with the sample of the
composition Sn14.0In9.5As21.5I8. The electrical resistivity drops
with the increasing temperature, and the log(ρ) = f(T–1) plot
shows two linear regions, with the activation energy of 26 and
43 meV for low and high temperatures, respectively, and with
a kink at 130(1) K. The Seebeck coefficient is negative for
the entire temperature range, indicating that electrons are the
dominant charge carriers. Its absolute value rapidly increases
upon increasing temperature in the range of 77–130 K, leading
to S ≈ –450 μV·K–1 at 130 K. At temperatures 130–300 K, the
Seebeck coefficient remains nearly constant. This dependence
of S(T) correlates with the change of the activation energy of
resistivity near 130 K. Sn14.0In9.5As21.5I8 possesses very low
thermal conductivity. Its value decreases from 1.22 W·m–1·K–1
at 77 K to 0.42 W·m–1·K–1 at room temperature, being roughly
proportional to T–1, which is typical of crystalline solids[27]
and was observed for various clathrates.[28] The total thermal
conductivity is a sum of the lattice and electronic parts. The
Figure 7. 119Sn Mössbauer spectra of Sn13.3In10.7As21.3 8
I (top) and
Sn17.8In4As21.7I8 (bottom). Black areas correspond to Sn(1), gray areas
correspond to Sn(2); impurity contributions are shown with solid lines.
Table 3. Hyperfine parameters of the 119Sn Mössbauer spectra.
Composition
Position
δ /mm·c–1 Δ /mm·c–1 A /%
Sn(1)
Sn(2)
2.05(1)
2.42(3)
0.75(1)
0.96(1)
0.51(2)
0.69(2)
72.8(8)
15.3(1)
6.2(2)
Sn17.8In4As21.7 8
I
Sn4+-imp. –0.01(2)
Sn2+-imp. 3.25(4)
5.7(8)
Sn(1)
Sn(2)
2.148(4)
2.14(1)
1.27(1)
0.55(5)
79(4)
14(4)
7.5(2)
Sn13.3In10.7As21.3 8
I
Sn4+-imp. –0.005(4) 0.47(2)
The remaining Sn1 and Sn2 subspectra (Table 3) correspond latter can be estimated using the Wiedemann–Franz relation,
to the tin atoms constituting the clathrate. The hyperfine pa- κe = σL0T, where L0 is the ideal Lorentz number L0
rameters pertinent to the major Sn1 component (δ = 2.05 and 2.45·10 W·Ohm·K–2, T the absolute temperature, and σ the
2.148 mm·s–1 and 0.75 and 1.27 mm·s–1 for electrical conductivity.[29] Given the room-temperature electri-
Sn17.8In4As21.7 8
and Sn13.3In10.7As21.3I8, respectively) are cal conductivity of 183 S·m–1, the electronic part of the ther-
close to those reported for the tin atom surrounded by one tin mal conductivity was calculated as 1.3 10–3 W·m–1·K–1 at
atom and three arsenic atoms in Sn19.1Cu4.9As22I8.[23]
293 K, proving that the total thermal conductivity of
slightly higher quadrupole splitting for this component in Sn14.6In9As21.5 8 is almost entirely phononic. The dimen-
=
Δ
=
I
A
I
2064
www.zaac.wiley-vch.de
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2011, 2059–2067

Clathrates in the Sn-In-As-I System
Figure 8. Thermoelectric properties of Sn14.6In9As21.5I8: (top left) log-plot of resistivity vs. reciprocal temperature, with two linear regions
marked with their Ea values; (top right) temperature dependence of the Seebeck coefficient; (bottom left) temperature dependence of the thermal
conductivity; (bottom right) temperature dependence of the thermoelectric figure-of-merit.
sionless thermoelectric figure-of-merit calculated as ZT = to 2.5 W·m–1·K–1.[5] Just a few exceptions with lower values,
[22]
around 0.4 W·m–1·K–1, are known for Sn24P19.3I6Br2
and
σS2T/κ increases with the temperature and reaches the value
of 0.028 at 293 K. Compared to various thermoelectric materi-
als with higher values of ZT, the Sn24–x–δInxAs22–yI8 clathrates
possess lower electrical conductivity by at least one order of
magnitude.[30]
Sn20.5As22I8.[20] In the latter case, however, the low thermal
conductivity is associated with the severe disorder in the crys-
tal structure and is coupled to a very low electrical conductiv-
ity, altogether leading to ZT of about 10–5. Contrastingly,
Sn24P19.3I6Br2 displays the figure-of-merit of about 0.02,
which is only slightly lower than the values observed in this
work. The low thermal conductivity of Sn24P19.3I6Br2 arises
not only from the rattling of the guest atoms but also due to
the mass alteration. The bromine and iodine guests alternate
randomly at the centers of the polyhedral cavities, leading to
shorter phonon free path, to which the lattice thermal conduc-
tivity is proportional above the characteristic Debye tempera-
ture.[29] Sn24–x–δInxAs22–yI8 displays a less disordered crystal
structure compared to Sn20.5As22I8 but has a higher unit cell
mass, which leads to a combination of the lower thermal con-
ductivity and higher figure-of-merit compared to both
Sn24P19.3I6Br2 and Sn20.5As22I8.
Figure 9 shows the variation of the room-temperature ther-
moelectric figure-of-merit ZT for Sn24–x–δInxAs22–yI8 with vari-
able indium contents, x. The dependence has a cupola-like
shape, with the maximum observed for x = 6.5. The
thermoelectric figure-of-merit at room temperature for
Sn24–x–δInxAs22–yI8 is the highest of all cationic clathrates at
room temperature but is almost one order of magnitude lower
than for the best anionic clathrates. For instance, the anionic
clathrate Eu8Ga16–xGe30+x has ZT of 0.35 at 300 K.[33] How-
ever, interpolation of the experimentally observed ZT vs. T de-
pendence gives ZT ≈ 0.4 at 600 K, indicating a great potential
of Sn24–x–δInxAs22–yI8 to be a base for the creation of thermo-
electric materials for mid-temperature applications.
All other samples show similar trends in their thermoelectric
properties, except that the feature around 130 K is not ob-
served for the samples with x other than 9.5. The room tem-
perature values are given in Table 4. Evidently, the most inter-
esting property of the Sn24–x–δInxAs22–yI8 clathrates is their ab-
normally low thermal conductivity. This property arises from
the localized motion (“rattling”) of the guest atoms inside the
oversized cavities of the framework, leading to either scat-
tering or low group velocity of heat-carrying phonons.[31,32] At
the first approximation, this is independent of the charge car-
rier transport through the framework; consequently, the lattice
thermal conductivity can be optimized by tuning the host-guest
mismatch and by increasing the atomic mass of the atoms com-
posing the clathrate.
It is known from the literature that the room-temperature
thermal conductivity of intermetallic clathrates ranges from 0.6
Table 4.
Room-temperature
thermoelectric
parameters
for
Sn24–x–δInxAs22–yI8 at different x.
x
σ /S·m–1 Ea /meV S /μV·K–1
κ /W·m–1·K–1
ZT
3.0 461
4.0 263
6.5 279
8.0 135
9.5 183
33
33
37
51
43
–322
–375
–427
–594
–468
0.57
0.36
0.36
0.39
0.42
0.024
0.030
0.041
0.036
0.028
Z. Anorg. Allg. Chem. 2011, 2059–2067
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2065

A. V. Shevelkov et al.
ARTICLE
Sample Characterization: The samples were characterized by means
of the X-ray powder and EDXS analyses. The X-ray analysis was per-
formed by using a Stoe STADI-P diffractometer, Cu-Kα1 radiation, λ
= 1.540598 Å; the unit-cell parameters were calculated from least-
squares fits using germanium (cubic, a = 5.6576 Å) as an internal stan-
dard, utilizing the program package Stoe WinXPOW. Rietveld analysis
was performed using Jana2000 programs.[34] For the EDXS analysis,
single crystals were glued with a carbon adhesive tape to an aluminum
holder such that the crystal surface was parallel to the holder plane.
The composition of the samples was determined using the EDX mode
of the scanning electron microscope JEOL JSM-6490LV (W cathode,
accelerating voltage 30kV) equipped with the analytical attachment
INCAx-sight. To quantify the relative content of elements in the mix-
ture, spectral K-and l-lines were used. In all cases, the data were col-
lected at 7–9 points and then averaged. The results were normalized
to eight iodine atoms. The estimated standard deviations slightly ex-
ceeded 5 mol-% mainly because the examined crystal surfaces were
not strictly parallel to the holder plane.
Figure 9. Dependence of ZT upon the indium contents x in Sn24–x–δ
InxAs22–yI8 at 293 K.
-
Crystal Growth: Large crystals for thermoelectric measurements were
grown using the chemical transport reactions. A two-zone scheme was
employed. In a typical experiment, the pre-synthesized sample of the
clathrate was mixed with a small amount (about 5 mg) of tin tetraio-
dide and loaded into a silica ampoule. The ampoule was sealed under
vacuum and placed into a gradient furnace such that the load was kept
at 820 K and the empty zone was heated to 835 K. After keeping at
such conditions for 36 hours, the temperature profile was changed in
6 hours to maintain the load temperature at 840 K while the crystal
growth zone temperature was set as 825 K. After 12 days, crystals of
the target clathrates having irregular shape with the maximal linear
dimension of 6 mm were grown. The crystal composition was mea-
sured by two different techniques. First, the smaller crystals were
crushed and subjected to X-ray diffraction analysis, from which the
cubic unit cell parameter was refined and related to the linear function
a = f(x), where x is the indium content, obtained from the powder
diffraction data. Then, another crystal from the same bunch was exam-
ined by the EDXS analysis. Afterwards, the results of the composition
determination were compared and analyzed for convergence.
Conclusions
The Sn24–x–δInxAs22–yI8 clathrates have been prepared with
varying indium contents limited by x = 10.7. They all crys-
tallize in the cubic unit cell with the clathrate-I structure type.
The crystal structure determined for five different composi-
tions shows the peculiar structural evolution despite the linear
dependence of the cubic unit cell upon the indium contents. In
particular, upon increasing the indium contents the degree of
disorder lowers and at high x values the coordination of a part
of tin atom changes from 3+2 to 3+3, which is confirmed by
the 119Sn Mössbauer study. The thermoelectric studies show
that these clathrates possess an abnormally low thermal con-
ductivity, which is almost entirely phononic. The room-tem-
perature thermal conductivity values of 0.36–0.57 W·m–1·K–1
are typical for amorphous solids and can be explained by a
combination of the localized motion of the guest atoms inside
the oversized polyhedral cages and a substantial disorder
within the clathrate framework. The maximal thermoelectric
figure-of-merit is 0.041 at room temperature and is interp-
olated to reach 0.4 at 600 K. This indicates that the
Sn24–x–δInxAs22–yI8 clathrates provide a solid based for the de-
velopment of thermoelectric materials for mid-temperature ap-
plication.
Thermal Analysis: DSC measurements were carried out with a pre-
calibrated DSC 204F1 Phoenix (Netzsch) calorimeter and the TG study
was performed using an STA 449c Jupiter (Netzsch) analyzer. About
10 mg of investigated sample were placed in an aluminum crucible.
For the DSC analysis, the samples were pressed and closed inside the
crucible and then twice heated up to 873 K and cooled in a dry argon
atmosphere at a ramp rate varying from 5 to 10 K·min–1 for different
measurements. For the TG analysis, the heating was performed with
the rate of 10 K·min–1 in dry argon flow to 1273 K or in air to 623 K.
The Proteus Thermal Analysis (Netzsch) program was used for the
data processing.
Experimental Section
Synthesis: The samples in the Sn-In-As-I system were prepared by
means of a two-step ampoule synthesis. Metallic tin and indium (both
Chempur, 99.99%), tin(IV) iodide (Alfa, 99.998%), and pieces of ar-
Single Crystal X-ray Diffraction Experiments: Suitable single crys-
tals of different composition were mounted on a goniometer head of a
senic (Chempur, 99.999%) were used as received. First, tin and indium CAD-4 (Nonius) diffractometer. The unit cell parameters were refined
arsenides (SnAs and InAs) were synthesized by annealing a stoichio- on the basis of 24 well-centered reflections in the angular range 11°Ͻ
metric mixture of initial components at 800 K for 7 days. Then the
θ Ͻ 14° and agreed well with those found from the powder data. The
appropriate stoichiometric amounts of tin and indium arsenides, tin(IV) data were collected at room temperature and corrected for polarization
iodide, and arsenic were mixed together; the mixture was sealed in a and Lorentz effects. A semiempirical absorption correction was ap-
silica tube under vacuum and then annealed at 803–823 K for five
days. The annealed samples appeared as black air- and moisture-stable their χ angles close to 90°. SHELXL-97 program was used for the
homogeneous powders. For X-ray diffraction crystal structure analysis,
structure refinement.[35] Data collection and refinement parameters are
single crystals were selected from the reaction product after annealing given in Table 1, important atomic parameters are summarized in
plied to the data on the basis of azimuthal scans of 9 reflections having
the pre-synthesized samples with a small amount of SnI4 at 823 K for
seven days.
Table 2, and a list of anisotropic ADP’s is given in Table S1 of Sup-
porting Information. Further details of the crystal structure determi-
2066
www.zaac.wiley-vch.de
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
Z. Anorg. Allg. Chem. 2011, 2059–2067

Clathrates in the Sn-In-As-I System
[8] M. Christensen, S. Johnsen, B. B. Iversen, Dalton Trans. 2010,
39, 978–992.
[9] M. Beekman, G. S. Nolas, J. Mater. Chem. 2008, 18, 842–851.
[10] J. V. Zaikina, T. Mori, K. A. Kovnir, D. Teschner, A. Senyshyn,
U. Schwarz, Yu. Grin, A. V. Shevelkov, Chem. Eur. J. 2010, 16,
12582–12589.
[11] U. Aydemir, C. Candolfi, H. Borrmann, M. Baitinger, A. Ormeci,
W. Carrillo-Cabrera, C. Chubilleau, B. Lenoir, A. Dauscher, N.
Oeschler, F. Steglich, Yu. Grin, Dalton Trans. 2010, 39, 1078–
1088.
nation may be obtained from Fachinformationszentrum Karlsruhe,
76344 Eggenstein-Leopoldshafen, Germany, on quoting the depository
numbers CSD-423636–423640.
Mössbauer Spectroscopy: The 119Sn spectrum was recorded by using
a conventional constant-acceleration Mössbauer spectrometer. The
measurements were performed at 300 K with a Ca119mSnO3 source
maintained at room temperature. Isomer chemical shifts were refer-
enced to a BaSnO3 absorber at 300 K.
[12] N. Tsujii, J. H. Roudebush, A. Zevalkink, C. A. Cox-Uvarov, G. J.
Snyder, S. M. Kauzlarich, J. Solid State Chem. 2011, 184, 1293–
1303.
[13] T. Tanaka, T. Onimaru, K. Suekuni, S. Mano, H. Fukuoka, S.
Yamanaka, T. Takabatake, Phys. Rev. B 2010, 81, 165110.
[14] M. Beekman, C. P. Sebastian, Yu. Grin, G. S. Nolas, J. Electron.
Mater. 2009, 38, 1136–1141.
[15] G. S. Nolas, M. Beekman, J. Gryko, G. A. Lamberton Jr., T. M.
Tritt, Appl. Phys. Lett. 2003, 82, 910–912.
[16] G. A. Slack, in: CRC Handbook of Thermoelectrics (Ed.: D. M.
Rowe), CRC Press, 1995.
[17] G. S. Nolas, J. L. Cohn, G. A. Slack, S. B. Schjuman, Appl. Phys.
Lett. 1998, 73, 178–180.
[18] A. V. Shevelkov, E. A. Kelm, A. V. Olenev, V. G. Kytin, V. A.
Kulbachinskii, Z. V. Lavrukhina, Fizika I Tekhnika Poluprovodni-
kov (in Russian) 2011, 45, 1454–1458.
Property Measurements: The selected large crystals were polished
with a diamond grinder to achieve a shape of a parallelepiped with
approximate dimensions of 4 ϫ 1 ϫ 1 mm. The simultaneous mea-
surements of temperature dependence of the Seebeck coefficient, ther-
mal conductivity, and electrical resistivity were performed using an
original experimental setup including a low temperature cryostat. Mea-
surements of all quantities were performed by the four-probe method.
The sample chamber was evacuated up to 10–5 mbar. Temperature was
controlled by the main heater and measured by a calibrated carbon
resistance thermometer TSU-2. Another heater was used to apply a
small temperature gradient (0–1 K) to the sample. A temperature gradi-
ent was determined as the difference between the temperatures at two
points on a sample spaced about 2.5 mm apart. The potential and tem-
perature difference required to calculate the Seebeck coefficient was
measured at the same points. At each temperature the heat flow used
to measure the thermal conductivity was determined from the tempera-
ture gradient measured on the reference sample with a well known
thermal conductivity connected in series with a sample under investi-
gation during measurements. The temperature gradient in the reference
sample was determined with a differential thermocouple, as in the in-
vestigated sample. The described experimental setup was tested on
Bi2Te3 crystals with well-known thermal conductivity and Seebeck co-
efficient, which showed that the maximum experimental error did not
exceed 5%.
[19] J. S. Kasper, P. Hagenmuller, M. Pouchard, Science 1965, 150,
1713–1714.
[20] J. V. Zaikina, K. A. Kovnir, A. V. Sobolev, I. A. Presniakov, Yu.
Prots, M. Baitinger, W. Schnelle, A. V. Olenev, O. I. Lebedev, G.
Van Tendeloo, Yu. Grin, A. V. Shevelkov, Chem. Eur. J. 2007, 13,
5090–5099.
[21] M. M. Shatruk, K. A. Kovnir, A. V. Shevelkov, I. A. Presniakov,
B. A. Popovkin, Inorg. Chem. 1999, 38, 3455–3457.
[22] J. V. Zaikina, W. Schnelle, K. A. Kovnir, A. V. Olenev, Yu. Grin,
A. V. Shevelkov, Solid State Sci. 2007, 9, 664–671.
[23] K. A. Kovnir, A. V. Sobolev, I. A. Presniakov, O. I. Lebedev, G.
Van Tendeloo, W. Schnelle, Yu. Grin, A. V. Shevelkov, Inorg.
Chem. 2005, 44, 8786–8793.
Supporting Information (see footnote on the first page of this article):
i) XRD pattern for the sample with x = 2 with superstructure reflec-
tions; ii) crystal data for the crystal with x = 2; iii) a full list of U(aniso)
for the structures I–V.
[24] D. M. Mingos, Essential Trends in Inorganic Chemistry. Oxford
University Press, Oxford, 1998.
[25] A. Kaltzoglou, T. F. Fässler, M. Christensen, S. Johnsen, B. Iv-
ersen, I. Presniakov, A. Sobolev, A. V. Shevelkov, J. Mater. Chem.
2008, 18, 5630–5637.
[26] M. M. Shatruk, K. A. Kovnir, M. Lindsjö, I. A. Presniakov, L. A.
Kloo, A. V. Shevelkov, J. Solid State Chem. 2001, 161, 233–242.
[27] C. Kittel, Introduction to Solid State Physics. 7th Ed. John
Wiley & Sons, New York, 1995.
[28] G. S. Nolas, B. C. Chakoumakos, B. Mahieu, G. J. Long, T. J. R.
Weakley, Chem. Mater. 2000, 12, 1947–1953.
Acknowledgement
This work was supported in part by the Russian Foundation for Basic
Research, grant # 10-03-00277.
[29] M. A. White, Properties of Materials. Oxford University Press,
Oxford, 1999.
[30] A. V. Shevelkov, Uspekhi Khim. 2008, 77, 3–21; Russ. Chem. Rev.
2008, 77, 1–19.
[31] S. Paschen, V. Pacheco, A. Bentien, A. Sanchez, W. Carrillo-Cab-
rera, M. Baenitz, B. B. Iversen, Yu. Grin, F. Steglich, Physica B
2003, 328, 39–43.
[32] M. Christensen, A. B. Abrahamsen, N. B. Christensen, F. Juranyi,
N. H. Andersen, K. Lefmann, J. Andreasson, C. R. H. Bahl, B. B.
Iversen, Nat. Mater. 2008, 7, 811–815.
[33] A. Bentien, V. Pacheco, S. Paschen, Yu. Grin, F. Steglich, Phys.
Rev. B 2005, 71, 165206.
References
[1] G. A. Jeffrey, in: Inclusion compounds (Eds.: J. L. Atwood,
J. E. D. Davies, D. MacNichol), Academic Press, London, 1983
p. 135.
[2] A. F. Wells, Structural Inorganic Chemistry. 5th Ed. Clarendon
Press, Oxford, 1986, Chapter 15.
[3] A. V. Kurnosov, V. Yu. Komarov, V. I. Voronin, A. E. Teplykh, A.
Yu. Manakov, Angew. Chem. Int. Ed. 2004, 43, 2922–2924.
[4] S. Alavi, K. Udachin, J. A. Ripmeester, Chem. Eur. J. 2010, 16,
1017–1025.
[34] V. Petricˇek, M. Dusek, Jana2000, Structure Determination Soft-
ware Programs, Institute of Physics, Praha, Czech Republic, 2000.
[35] G. M. Sheldrick, Acta Crystallogr., Sect. A 2008, 64, 112–122.
[5] A. V. Shevelkov, K. Kovnir, Struct. Bonding (Berlin) 2011, 139,
97–144.
[6] S. Bobev, S. C. Sevov, J. Solid State Chem. 2000, 153, 92–105.
[7] M. A. Kirsanova, A. V. Olenev, A. M. Abakumov, M. A. Bykov,
A. V. Shevelkov, Angew. Chem. Int. Ed. 2011, 50, 2371–2374.
Received: June 29, 2011
Published Online: November 8, 2011
Z. Anorg. Allg. Chem. 2011, 2059–2067
© 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.zaac.wiley-vch.de
2067