You are on page 1of 6

The effect of Te doping on the electronic structure and thermoelectric

properties of SnSe
Song Chen
a
, Kefeng Cai
a,n
, Wenyu Zhao
b
a
Functional Materials Research Laboratory, Tongji University, 1239 Siping Road, Shanghai 200092, China
b
State Key Lab of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070, China
a r t i c l e i n f o
Article history:
Received 7 May 2012
Received in revised form
20 June 2012
Accepted 29 June 2012
Available online 20 July 2012
Keywords:
SnSe
Doping
Thermoelectric
First-principle theory
a b s t r a c t
SnSe
1x
Te
x
(x0, 0.0625) bulk materials were fabricated by melting Sn, Se and Te powders and then
hot pressing them at various temperatures. The phase compositions of the materials were determined
by X-ray diffraction (XRD) and the crystal lattice parameters were rened by the Rietveld method
performed with DBWS. XRD analysis revealed that the grains in the materials preferentially grew along
the (l 0 0) directions. The structural behavior of SnSe
1x
Te
x
(x0, 0.0625) was calculated using CASTEP
package provided by Materials Studio. We found that the band gap of SnSe reduced from 0.643 to
0.608 eV after Te doping. The calculated results were in good agreement with experimental results. The
electrical conductivity and the Seebeck coefcient of the as-prepared materials were measured from
room temperature to 673 K. The maximum power factor of SnSe is 0.7 mW cm
1
K
2
at 673 K.
& 2012 Elsevier B.V. All rights reserved.
1. Introduction
With the increasing global energy demands and declining fossil
fuel reserves, development of renewable energy alternatives is
extremely urgent. Thermoelectric (TE) devices are solid-state energy
converters that can transform waste heat directly into electricity.
However, until now, TE devices only have niche applications mainly
due to low efciency and high cost. The efciency of such devices
mainly depends on the gure of merit, ZT (a
2
sT/k, where a, s, T,
and k are the Seebeck coefcient, the electrical conductivity, the
absolute temperature, and the thermal conductivity, respectively) of
the materials used. Bi
2
Te
3
-based materials are the best TE materials
around room temperature [1,2]. However, Bi and Te elements are
rare in the earth and their prices are increasing with the develop-
ment of LED industry. Therefore, it is necessary to nd new
TE materials. Sn is an earth-abundant and environment friendly
element, and Se is more abundant than Te in the earth. Like Bi
2
Te
3
,
SnSe also has a layered structure, hence, it may be of interest to
study the TE properties of SnSe.
SnSe has orthorhombic crystal structure with space group
Pmna and lattice parameters: a1.150 nm, b0.415 nm and
c0.444 nm. A unit cell of SnSe contains eight atoms placed in
positions by the scaled co-ordinates (u, 1/4, v) and (1/2, 1/4, 1/2v).
The Sn and Se atoms form double layers made up of two planes of
zigzag SnSe chains perpendicular to the a-axis. Within either
double layer, each atom has three nearest neighbors and two next
nearest neighbors. The layers pile up with a weak van der Waals-
like coupling along the a-axis direction. Bulk SnSe has an indirect
band gap of 0.90 eV and a direct band gap of 1.30 eV [3].
Recently, SnSe has received much attention because it could be
used in many application elds, such as solar cells [4], phase-change
alloys for electronic memory [5], and as a cathodic material in
lithium intercalation batteries, due to the anisotropic character [6].
Several methods have been developed to prepare SnSe, such as
uxing [7], chemical vapor deposition [8], electrodeposition [9],
electron beam irradiation [10], direct vapor transport [11], and one-
pot chemical synthesis [12]. Theoretically, the ab-initio calculations
were used to study the band structure and density of states (DOS) of
SnSe. Makinistian et al. [13] reported that spinorbit does not affect
the band gap in the Brillouin zone of SnSe. The structural behavior of
SnSe under the hydrostatic pressure using a constant pressure ab
initio technique was also studied, and a structural second order
phase transition at 7 GPa was found [14].
In fact, the TE properties of single crystalline SnSe grown by a
direct vapor transport technique have already been studied in
Refs. [15,16]. It is found that the SnSe has very high Seebeck
coefcient that increases with the increasing temperature (from
200 mV/K at 308 K to 1300 mV/K at 573 K) and low electrical
conductivity (0.00055 S/cm at 308 K) [15]. As the pressure
increases, the Seebeck coefcient of the single crystalline SnSe
decreases from 47 mV/K at 0.5 GPa to 24 mV/K at 8 GPa and
S-doping decreases the Seebeck coefcient of the SnSe. And the
values of the band gap decrease with the increasing pressure and
the materials become more conducting [16]. However, the TE
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/physb
Physica B
0921-4526/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.physb.2012.06.041
n
Corresponding author.
E-mail address: kfcai@tongji.edu.cn (K. Cai).
Physica B 407 (2012) 41544159
properties of polycrystalline SnSe and Te doped SnSe have not been
reported, and the relationship between electronic structure and TE
properties of SnSe has also not been studied. In this work, the effect
of Te doping on the electronic structure and TE properties of SnSe
were investigated by both experimentally and theoretically.
2. Experimental
Te powder (Alfa Aesar, 99.99%), Se powder (Alfa Aesar, 99.95%),
and Sn powder (Sinopharm Chemical Reagent, 99.5%) were used
as received. All manipulations were performed in an argon-lled
glove box. The elements, Te:Se:Sn, were placed in a 5-mL BN
powder-lined graphite crucible in the molar ratios of x:1x:1
(x0, 0.0625), respectively. The reactants were sealed in quartz
ampoules under 1/5 atm argon atmosphere and placed in a high-
temperature programmable furnace. The furnace was heated to
1023 K with a rate of 10 K/min, then held at 1023 K for 1 h,
followed by slowly cooling (2 K/min) to room temperature. The
ingots obtained had obviously metallic luster and layer-structure,
which is in agreement with that reported in Ref. [10]. The ingots
were broken and ground into powders under an Ar atmosphere in
glove box, and nally hot pressed into pellet (10 mm in diameter
and 3 mm in thickness) in vacuum (about 10
2
Pa) for 2 h at
different temperatures under 80 MPa.
The phase composition of the samples before and after hot
pressing was determined by XRD (Rigaku DMAX2500VPC) with Cu
K
a
radiation (40 kV, 200 mA). The DBWS suite of programs was used
for Rietveld tting of the powder XRD data. A pseudo-voigt function
with axial divergence was used to describe the peak shape. The
fracture surface of the crystals was observed by eld emission
scanning electron microscopy (FESEM, Quanta 200FEG), equipped
with electron energy dispersive X-ray spectroscopy (EDS, Oxford
7582). Hall effect measurement was carried out at room tempera-
ture using a Hall effect measurement system (HMS 3000, Ecopia)
with a magnetic eld of 0.55 T. The differential scanning calorimeter
(DSC) and thermogravimetric analysis (TG) of the samples were
performed at a heating rate of 10 K/min in a owing nitrogen.
The pellets were cut into rectangles (1022 mm
3
) for s and a
measurements. The measurements were carried out using a home-
made computer control test system from room temperature to
673 K under argon atmosphere. The s measurement was performed
by a steady-state four-probe technique with a square wave current
(10 mA in amplitude). The a value was determined by the slope of
the linear relationship between the thermal electromotive force and
temperature difference (10 K) between the two ends of each
sample. The density of the samples was determined by measuring
the Archimedes method.
3. Computation method
The calculations were performed using CASTEP package pro-
vided by Materials Studio. Our calculations were based on the
density functional theory (DFT) in generalized gradient approx-
imations (GGA) with PerdewBurkeErnzerhof (PBE) exchange-
correlation potential. The electronic structure was calculated by
optimizing all the atoms of the crystal using ultrasoft pseudopo-
tentials for the core electrons. The cut-off energy in plane wave
expansion was 300 eV. The total energy, maximum stress, max-
imum force and maximum displacement were converged to less
than 110
5
eV/atom, 0.05 GPa, 0.3 eV/nm and 0.0001 nm,
respectively. The tolerance in the self-consistent eld (SCF)
calculation was set to 10
6
eV/atom. We also used 233
k-point MonkhorstPack mesh for the bulk. The calculations were
based on the experimental crystal structure data after renement.
We began our calculations with 122 supercell consisting of
16 Sn atoms and 16 Se atoms, and for the Te-doped SnSe,
we replaced one of the Se atoms in the supercell with a Te atom.
4. Results and discussion
Fig. 1(a) and (b) shows the XRDpatterns of the SnSe sample before
and after hot pressing at 575 1C, and Fig. 1(c) shows the XRD pattern
of the Te doped sample hot pressed at 575 1C (the XRD pattern of the
other SnSe samples is almost the same as that shown in Fig. 1(b)).
There is not much difference in the patterns of the SnSe samples
before and after hot pressing. All the XRD peaks can be indexed to the
standard data of SnSe (JCPDS card, 48-1224). This indicates that pure
SnSe was prepared. Compared with the standard data, the intensity of
the (l 0 0) plane peaks of the samples is much increased, which
indicates that the grains in the samples preferentially grew along the
(l 0 0) planes. By comparing Fig. 1(b) with (c), it is known that Te
doping does not introduce any impurity phase into SnSe.
In order to obtain the crystal lattice parameters, Rietveld
renement was performed with DBWS, and the nal calculated,
observed, and residual patterns are shown in Fig. 2. The details of
Fig. 1. XRD patterns of samples SnSe before (a) and after (b) hot pressing at
575 1C, and (c) SnSe
0.9375
Te
0.0625
hot pressed at 575 1C.
Fig. 2. Final Rietveld renement of powder XRD data for the SnSe sample hot
pressed at 575 1C.
S. Chen et al. / Physica B 407 (2012) 41544159 4155
the renement are summarized in Table 1. The t reliability
was converged to give R-wp14.80%, R-p9.87% and S3.56.
The renement result shows that the sample containing a little
amount of impurity: 1.2(4) wt% of Sn (JCPDS card 65-7657), which
is conrmed by DSC analysis result (Supporting material Fig. S1).
FESEM image (Fig. 3(a)) also shows that the samples have
layered structure. This is ascribed to the dominant van der Waals
character of the bonds between adjacent layers [10,11]. EDS
analysis (Fig. 3(b)) reveals that the samples consist of Sn and Se,
and quantitative EDS analysis indicates that the atomic ratio of
Sn/Seo1, which agrees with the renement result. All the data
above indicate that the samples prepared mainly contain Sn
1y
Se
(0oyo1) mixed with a little amount of Sn.
The structure parameters of the SnSe
1x
Te
x
(x0, 0.0625) sam-
ples hot pressed at 575 1C were optimized and listed in Table 2.
Compared with the rened experimental result, the calculated lattice
parameters a, b and c are slightly larger with relative errors of 1.3%,
1.1% and 0.7%, respectively. And the calculated lattice parameters of
the Te-doped sample were also listed in Table 2, with maximum
relative error of 1.6%. Thus, our calculated results are reasonable. Both
the calculated and experimental results indicate that the unit cell size
of SnSe was enlarged after doping Te, because the Te atom was larger
than Se atom.
To better understand the electronic nature of this material, we
have calculated the band structure of SnSe along high symmetry
lines corresponding to a 122 super cell shown in Fig. 4. In
this gure, the G, F, Q, Z and B are (0 0 0), (0 1/2 0), (0 1/2 1/2),
(0 0 1/2) and (1/2 0 0) high symmetry points, respectively. It can
be seen from Fig. 4 that the valence band maximum (VBM)
locates between Z- and G-point, and the conduction band mini-
mum (CBM) between G- and F-point, indicating an indirect band
gap material. It is obvious that the Fermi energy level is on the
VBM and the band gap is 0.643 eV, which is lower than the
experimental result (about 0.9 eV) [3].The low value is caused by
a well-known drawback using standard DFT calculation [17]. The
band structure of SnSe
0.9375
Te
0.0625
has also been calculated (not
shown here), and it reveals that it is also an indirect band gap
material, with band gap of 0.608 eV, which is somewhat smaller
than that of SnSe.
In addition, we also plotted the DOS and the decomposed
partial density of states (PDOS) of each atom for SnSe
0.9375
Te
0.0625
in Fig. 5. We chose the DOS of the SnSe
0.9375
Te
0.0625
as positive
value and that of the SnSe as negative one to clearly show
the doping effect of Te. From the total DOS for the SnSe, the
valence band can generally be divided into lower valence band
from 14.5 to 11.5 eV, middle one from 8.5 to 5.0 eV and
upper one from 5.0 to 0.0 eV (Fig. 5(a)). Moreover, the lower
valence band is mainly contributed to the Se(4s) states, the upper
one is mainly from Se(4p) states, and the middle one is dominated
by Sn(5s) (Fig. 5(b)), which agree with the results reported in
Ref. [14]. In addition, the conduction band from 0.5 to 2.5 eV
is primarily ascribed to the Sn(5p) orbit (Fig. 5(c)). Compared
with the SnSe, the SnSe
0.9375
Te
0.0625
raises the valence band from
11.5 to 10.5 eV on the DOS curve, which is mainly occupied
by Te(5s) with admixture of Sn(5s). As it is far away from
the Fermi energy level, Te(5s) and Sn(5s) interact and form a
strong covalent bond in this region. As the DOS curve of the
SnSe
0.9375
Te
0.0625
is similar to that of the SnSe and in order to
clearly show the peak slightly shifts after Te doping, we added
Table 1
Rened atomic coordinates and thermal parameters at room temperature for the sample SnSe hot pressed at 575 1C.
Space group Pnma 62
Atom Position Atomic coordinates Occu.
Temperature parameter B
iso
(

A
2
)
x y z
Se 4 0.8559(1) 0.2500 0.4836(1) 1 0.022294
Sn 4 0.1180(2) 0.2500 0.1043(5) 0.98 0.025842
Cell dimensions and angles
a1.1500(7) nm b0.4154(2) nm c0.4446 (2) nm abg90
o
U.C. density d6.203 (1) g/cm
3
R-P 9.87%
R-Wp 14.80%
S 3.56
Fig. 3. (a) FESEM image of the fracture surface of the sample SnSe hot pressed at
575 1C and (b) EDS spectrum recorded on the marked area in (a).
S. Chen et al. / Physica B 407 (2012) 41544159 4156
green and blue lines in Fig. 5(a). It should be noticed that the
conduction band peak of the SnSe
0.9375
Te
0.0625
shifts to left
compared with that of the SnSe, but the upper valence bands do
not. So we presumed that the reduction of the band gap is due to
the splitting of the conduction band [18], so-called band gap
renormalization, which results in a gap narrowing [19]. These
ndings are helpful to understand the optical properties of SnSe.
If one of the Se atoms in SnSe is substituted by a Te atom, since
the electronegativity of Te is weaker than that of Se, the energy
level will be broadened to form impurity-induced bands. This is
the result from the interaction between Sn(5s) and Te(5s) orbits
combined with CBM. Furthermore, CBM moves down to the Fermi
level and then the band gap reduces.
To explore the bonding character in SnSe, the charge density
distribution was studied. The counter plot of the valence charge
density on the (1 0 0) plane is shown in Fig. 6. It can be seen from
Fig. 6 that the charge density around Se atom exhibits a directional
distribution toward Sn atom. In addition, the charge density around
Se atom is much higher than that around Sn atom, which indicates
that the SeSn bond has a strong polarization covalent character due
to the hybridization effect between the Sn(5s) and Se(4s) states,
which is also reected in the DOS of SnSe
0.9375
Te
0.0625
. The charge
density around Te atom is much lower than that around Se atom,
which means that the bonding of TeSn atoms has covalent character
due to the hybridization effect between the Te(5s) and Sn(5s) states.
The effective mass of the carrier can be expressed as follows:
m
n

h
2p

2
d
2
E
dk
2
!
1
1
where m
n
is the effective mass of electron, h is the Plancks constant,
E is the energy level and k is the wave vector. We calculated the
effective mass of carrier on the CBM band at G-point along bc plane
(average of GF and ZG) and along the a-axis direction (GB), all
the results are listed in Table 3.
For electron-conduction, the carrier concentration (n) can be
expressed as follows:
n
1
V
Z
1
Ec
f Eg
c
EdE 2
where f E 1=1expEE
F
=kT, and it is FermiDirac distribu-
tion function; V is the supercell volume; g
c
(E) is the DOS function.
Note that we chose the point where E40, DOS40.3 (states/eV) and
is nearest to Fermi-Point as E
c
during the calculation. We calculated
the Seebeck coefcient using the formula as follows [20]:
a
8p
2
k
2
B
3eh
2
m
n
T
p
3n

2=3
3
where k
B
is the Boltzmann constant and e is the electronic charge.
The calculated results are given in Table 3.
Fig. 4. Calculated GGA-PBE band structure of SnSe.
Fig. 5. (a) DOS of SnSe (negative) and SnSe
0.9375
Te
0.0625
(positive), (b) partial DOS
of Se(4s), (4p), Te(5s) (5p), and (c) partial DOS of Sn(5s), (5p) in SnSe
0.9375
Te
0.0625
and SnSe from the GGA calculation. (For interpretation of the references to color in
this gure, the reader is referred to the web version of this article.)
Table 2
The calculated and the experimental lattice parameters of the SnSe
1x
Te
x
(x0,
0.0625) samples hot pressed at 575 1C.
Sample a (nm) b (nm) c (nm) a (deg.) b (deg.) g (deg.)
SnSe
Exp 1.1500 0.8308 0.8892 90 90 90
Cal 1.1650 0.8403 0.8956 90.0 90.0 90.3
Error 1.3% 1.1% 0.7% 0.3%
SnSe
0.9375
Te
0.0625
Exp 1.1559 0.8311 0.8990 90 90 90
Cal 1.1700 0.8446 0.8972 90.0 90.0 90.3
Error 1.2% 1.6% 0.2% 0.3%
S. Chen et al. / Physica B 407 (2012) 41544159 4157
It is seen from Table 3 that all the calculated values agree with
the experimental results. Hall measurement indicates that the
SnSe has higher mobility than the SnSe
0.9375
Te
0.0625
(Table 3). As
the SnSe sample is rich in Se, its carrier concentration is higher
than that of the SnSe single crystals reported in Refs. [15,16], and
also the conduction changes from p-type to n-type.
Fig. 7 shows the temperature dependence of electrical conduc-
tivity and the Seebeck coefcient of the samples prepared. All the
curves have the same change trend: as temperature increases, the
electrical conductivity rapidly increases rst, then decreases a little,
and nally increases again. The electrical conductivity of the SnSe
sample at room temperature is higher than that of the single crystal
SnSe reported in Ref. [15], which should be because the present
sample is nonstoichiometric and mixed with some Sn impurity. The
relative density of all the samples is 95%; therefore, we neglected
the effect of porosity on the TE properties of the samples.
It can be seen from Fig. 7 that when To550 K, at a given
temperature, the absolute Seebeck coefcient of the SnSe samples
increases with the increasing hot pressing temperature, whereas
the electrical conductivities of the samples are close near room
temperature after which these increase rst then decrease with the
increasing hot pressing temperature at intermediate temperatures
(400oTo550 K). Based on the results, we deduce that the hot
pressing process introduced deep level defects in the samples and
the concentration of the defects increased with increasing hot
pressing temperature. The defects increase the scattering factor
which results in the increasing Seebeck coefcient. On the other
hand, the deep level defects can be excited when the temperature is
high enough (T4400 K), leading to increased carrier concentra-
tion. When the carrier concentration is high enough, it will increase
the scattering and hence decrease the mobility. Therefore, at a given
temperature (550 KoT) the changing trend of the electrical con-
ductivity for the samples is not the same as that of the Seebeck
coefcient.
As the measurement temperature increases, the changing
trend of the absolute Seebeck coefcient is consistent with that
of the electrical conductivity of the samples, which is different
from the behavior of the common TE materials. This should be
ascribed to the phonon-drag effect as carrier concentration of the
samples is less than 10
17
cm
3
[21]. The SnSe sample hot pressed
at 575
1
C shows higher power factor at a given temperature. The
maximum value of power factor is 0.7mW cm
1
K
2
at 673 K and
it is close to that of FeSi
2
-based TE material [22].
Combined with the DSC result, it is known that the sT and
aT curves of the samples both changing around 500 K are related
to the melting of Sn impurity in the samples. At about 550600 K,
Table 3
Some physical properties of the samples prepared.
m
n
/m
0
a (mV/K) n
0
s
0
m
0
yz x yz x (10
17
/cm
3
) (S/cm) (cm
2
/Vs)
SnSe
Cal. 4.026 1.06 161.6 44.5 3.5
Exp. 165.2 96.2 2.9 0.23 4.88
SnSe
0.9375
Te
0.0625
Cal. 4.045 168.9 3.3
Exp. 168.3 2.1 0.14 4.21
Fig. 7. Temperature dependence of (a) electrical conductivity and (b) the Seebeck
coefcient of the samples hot pressed at different temperatures.
Fig. 6. Charge density maps for SnSe
1x
Te
x
(x0, 0.0625) on (1 0 0) plane.
S. Chen et al. / Physica B 407 (2012) 41544159 4158
both the electrical conductivity and the absolute Seebeck coef-
cient rapidly increase. If it is because of intrinsic excitation, the
absolute Seebeck coefcient should decrease. In addition, SnSe
has a large band gap (0.9 eV), hence we excluded the possibility
of intrinsic excitation at this temperature range.
To know the effect of the impurity, Sn, on thermal stability of
the samples, recycle measurements were carried out (Fig. S2). The
results indicate that the TE properties of the samples are stable. In
addition, the Seebeck coefcient of the sample hot pressed at
575 1C was measured along the directions perpendicular (?) and
parallel (J) to the hot pressing direction, and the results (Fig. S3)
show that the a
?
is higher than the a
J
, indicating anisotropic
transport behavior of the SnSe due to the anisotropic structure.
By combining the DSC analysis (Fig. S1) and recycle measure-
ment (Fig. S2) results, we think that the strange behavior of a and
s as T550600 K may not have resulted from Sn and/or Se
volatilization. Nariya et al. [15] also found that the absolute
Seebeck coefcient of the SnSe single crystals increased signi-
cantly above 530 K. Hence, we deduce that the electronic
structure of SnSe may change at high temperatures, which needs
to be further studied.
It is seen from Fig. 7 that at a given temperature although the
Te-doped sample has lower electrical conductivity than the
undoped ones (due to the lower carrier concentration and
mobility, see Table 3), its temperature dependence of electrical
conductivity is similar to that of undoped samples; however, its
absolute Seebeck coefcient starts to decrease at rst and then
decreases more signicantly. In general, like S-doping in SnSe
[16], Te-doping also does not improve the electrical transport
properties of SnSe. It may be a good choice to dope at the Sn-site
with proper elements or even better to study other selenides. For
example, more recently, Liu et al. [23] reported that Cu
2
Se has
excellent TE properties with ZT1.5 at 1000 K.
5. Conclusions
In summary, polycrystalline SnSe
1x
Te
x
(x0, 0.0625) samples
were fabricated by melting followed by hot-pressing at various
temperatures and their TE properties were measured from room
temperature to 673 K. The electronic structures of SnSe
1x
Te
x
(x0, 0.0625) were calculated by the rst-principle theory. The
band gap and the band structure near the band gap of SnSe can be
adjusted by doping element Te. The maximum power factor for
SnSe was 0.7 mW cm
1
K
2
at 673 K. Doping Te on Se site does
not improve the power factor of SnSe.
Acknowledgments
This work was supported by the National Natural Science
Foundation of China (50872095), Doctoral Fund of Ministry of
Education of China, the foundation of the State Key Lab of Advanced
Technology for Materials Synthesis and Processing, Wuhan Uni-
versity of Technology.
Appendix A. Supporting information
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.physb.2012.06.041.
References
[1] W.J. Xie, J. He, H.J. Kang, X.F. Tang, et al., Nano Lett. 10 (2010) 3283.
[2] Y.Q Cao, X.B. Zhao, T.J. Zhu, Appl. Phys. Lett. 92 (2008) 143106.
[3] I. Lefebvre, M.A. Szymanski, J. Olivier-Fourcade, J.C. Jumas, Phys. Rev. B 58
(1998) 1896.
[4] C. Guillen, J. Montero, J. Herrero, Phys. Status Solidi A 208 (2011) 679.
[5] K.M. Chung, D. Wamwangi, M. Woda, M Wuttig, W. Bensch, J. Appl. Phys. 103
(2008) 083523.
[6] M.Z. Xue, S.C. Cheng, J. Yao, Z.W. Fu, Acta Phys.-Chim. Sin. 22 (2006) 383.
[7] M.F. de Oliveira, R. Caram, C.S. Kiminami, J. Alloys Compd. 375 (2004) 142.
[8] N.D. Boscher, C.J. Carmalt, R.G. Palgrave, I.P. Parkin, Thin Solid Films 516
(2008) 4750.
[9] M. Bicer, I. Sisman, Appl. Surf. Sci. 257 (2011) 2944.
[10] Z. Li, L.W. Peng, Y.G. Fang, Z.W. Chen, D.Y. Pan, M.H. Wu, Radiat. Phys. Chem.
80 (2011) 1333.
[11] A. Agarwal, S.H. Chaki, D. Lakshminarayana, Mater. Lett. 61 (2007) 5188.
[12] D.D. Vaughn II, S. In, R.E. Schaak, ACS Nano 5 (2011) 8852.
[13] L. Makinistian, E.A. Albanesi, Phys. Status Solidi B 246 (2009) 183.
[14] S. Alptekin, J. Mol. Model. 17 (2011) 2989.
[15] B.B. Nariya, A.K. Dasadia, M.K. Bhayani, A.J. Patel, A.R. Jani, Chalcogenide Lett.
6 (2009) 549.
[16] T.H. Patel, Rajiv Vaidya, S.G. Patel, High Pressure Res. Int. J. 23 (2003) 417.
[17] X.Y. Su, P.P. Si, Q.Y. Hou, X.L. Kong, W. Cheng, Physica B 404 (2009) 1794.
[18] K. Hoang, S.D. Mahanti, M.G. Kanatzidis, Phys. Rev. B 81 (2010) 115106.
[19] J.A. Sans, J.E. Snchez-Royo, A. Segura, G. Tobias, E. Canadell, Phys. Rev. B 79
(2009) 195105.
[20] G.J. Snyder, E.S. Toberer, Nat. Mater. 7 (2008) 105.
[21] D.M. Rowe, Thermoelectrics Handbook Macro to Nano, 2006, pp. 55.
[22] X.B. Zhao, T.J. Zhu, S.H. Hu, B.C. Zhou, Z.T. Wu, J. Alloys Compd. 306 (2000)
303.
[23] H.L. Liu, X. Shi, F.F. Xu, L.L. Zhang, W.Q. Zhang, L.D. Chen, Q. Li, C. Uher, T. Day,
G.J. Snyder, Nat. Mater. (2012), http://dx.doi.org/10.1038/NMAT3273.
S. Chen et al. / Physica B 407 (2012) 41544159 4159

You might also like