Thermally stable thermoelectric Zn4 Sb3 by zone-melting synthesis

Article abstract of DOI:10.1063/1.2916705

The thermal stability of thermoelectric Zn4 Sb3 has been investigated on samples produced by a new zone-melting technique, as well as by the conventional quench method. The multitemperature synchrotron powder diffraction data reveal that while conventionally synthesized, Zn4 Sb3 samples have almost 40% degradation in the first heating cycle at 625 K, samples prepared by zone melting only have 3% degradation. Repeated thermal cycling induces additional degradation of the quenched sample of up to 58%, compared to ～9% degradation in the zone-melted sample. Thus, zone-melting produces Zn4 Sb3 samples that are significantly more thermally stable, which make them promising for commercial implementation.

Full text of DOI:10.1063/1.2916705

Thermally stable thermoelectric Zn4Sb3 by zone-melting synthesis
B. L. Pedersen and B. B. Iversen
Citation: Appl. Phys. Lett. 92, 161907 (2008); doi: 10.1063/1.2916705
View online: http://dx.doi.org/10.1063/1.2916705
View Table of Contents: http://apl.aip.org/resource/1/APPLAB/v92/i16
Published by the American Institute of Physics.
Related Articles
High-temperature thermoelectric properties of the double-perovskite ruthenium oxide (Sr1−xLax)2ErRuO6
J. Appl. Phys. 112, 073714 (2012)
Thermoelectric properties of n-type Mn3−xCrxSi4Al2 in air
J. Appl. Phys. 112, 073713 (2012)
Thermoelectric effect in a graphene sheet connected to ferromagnetic leads
J. Appl. Phys. 112, 073712 (2012)
Thermoelectric properties of p-type Bi0.5Sb1.5Te2.7Se0.3 fabricated by high pressure sintering method
J. Appl. Phys. 112, 073708 (2012)
Dimensional crossover of charge density wave and thermoelectric properties in CeTe2−xSbx single crystals
Appl. Phys. Lett. 101, 143901 (2012)
Additional information on Appl. Phys. Lett.
Journal Homepage: http://apl.aip.org/
Journal Information: http://apl.aip.org/about/about_the_journal
Top downloads: http://apl.aip.org/features/most_downloaded
Information for Authors: http://apl.aip.org/authors
Downloaded 11 Oct 2012 to 171.67.34.69. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

APPLIED PHYSICS LETTERS 92, 161907 ͑2008͒
Thermally stable thermoelectric Zn4Sb3 by zone-melting synthesis
B. L. Pedersen and B. B. Iversena͒
Centre for Energy Materials, Department of Chemistry and Interdisciplinary Nanoscience Center,
University of Aarhus, 140 Langelandsgade, DK-8000 Aarhus, Denmark
͑Received 30 January 2008; accepted 5 April 2008; published online 25 April 2008͒
The thermal stability of thermoelectric Zn4Sb3 has been investigated on samples produced by a new
zone-melting technique, as well as by the conventional quench method. The multitemperature
synchrotron powder diffraction data reveal that while conventionally synthesized, Zn4Sb3 samples
have almost 40% degradation in the first heating cycle at 625 K, samples prepared by zone melting
only have 3% degradation. Repeated thermal cycling induces additional degradation of the
quenched sample of up to 58%, compared to ϳ9% degradation in the zone-melted sample. Thus,
zone-melting produces Zn4Sb3 samples that are significantly more thermally stable, which make
them promising for commercial implementation. © 2008 American Institute of Physics.
͓DOI: 10.1063/1.2916705͔
The ␤ phase of the p-type semiconductor Zn4Sb3 has
excellent thermoelectric properties in the intermediate tem-
perature range ͑473–673 K͒.1 The thermoelectric figure of
merit is defined as ZT=S2T/␳␬, where S is the thermopower,
␳ is the electrical resistivity, ␬ is the thermal conductivity,
and T is the absolute temperature. The outstanding thermo-
electric performance1,2 originates from the complex crystal
structure including three highly disordered interstitial Zn
sites, which effectively lower the thermal conductivity while
preserving the electrical conductivity.3,4 Although Zn4Sb3 is
now known to exist in at least four crystalline phases, only
the ␤ phase, stable from 263 to 765 K, is a high performing
thermoelectric material.1,2,5 When the temperature is lowered
from ambient, the crystal structure goes through two phase
by a new zone-melting technique.13 A large batch of Zn4Sb3
was prepared from stoichiometric ratios of 99.99% Zn shots
and 99.5% antimony powder in a quartz ampoule. The
vacuum sealed ampoule was horizontally placed in a furnace
and heated with a 400 K/h ramp to 973 K under continuous
rotation for 2 h before quenching in ice water. This resulted
in a polycrystalline, rod shaped sample. Part of the quench
synthesized product was used as a feeding rod in the subse-
quent zone-melting process. The feeding rod was placed on
top of a nonstoichiometric rod ͑Zn52Sb48͒ in a quartz am-
poule. Below the nonstoichiometric rod, a small piece of
feeding rod was placed as support. The zone melting was
performed in an induction furnace in a 100 psi He atmo-
sphere, choosing the interface between the lower feeding rod,
and the nonstoichiometric rod as the starting point. After
establishing a stable melt zone, the ampoule was moved
through the heating zone under continuous rotation
͑ϳ8 rpm͒ with a speed of approximately 1 mm/h.
transitions, ␤→␣ and ␣→␣ , before eventually reaching an
Ј
ordered triclinic ␣ -phase at ϳ230 K.6 In the ␣ phase, the
Ј
Ј
Zn interstitials become part of the crystalline network.7–9
In order to become commercially applicable, it is essen-
tial for a thermoelectric material to be stable at the operating
temperature as well as during thermal cycling. Zn4Sb3 was
originally reported to be stable under argon up to 673 K, and
up to 523 K in dynamic vacuum.1,2 When held at tempera-
tures higher than 523 K, the compound partly decomposed
but the degree of decomposition was not quantitatively
determined.1,2 The originally reported thermal stability has
been questioned by several authors who found that Zn4Sb3
decomposes at temperatures below the expected stability
range, forming ZnSb and elemental Zn as decomposition
products.6,10 These decomposition products cause a signifi-
cant degradation of the thermoelectric properties and the
main challenge therefore was to stabilize Zn4Sb3 at T
Ͼ200 °C.6,10 Doping of Zn4Sb3 has been suggested as a pos-
sible way to improve the stability11 but although it was
shown that doping can change the decomposition pattern
relative to pure undoped samples, no immediate improve-
ment of thermal stability was achieved.12
Multitemperature, high-resolution powder diffraction
data were recorded using the large Debye–Scherrer camera at
beam line BL02B2, SPring8, Japan.14 To ensure a homoge-
neous grain size, an ethyl alcohol floating technique was ap-
plied before transferring the sample to a 0.2 mm capillary.
The data were measured at ␭=0.422 36͑1͒ Å with the wave-
length determined from refinement on a CeO2 standard, and
the exposure time was set to 5 min. Data sets were recorded
at 300, 400, 500, 505, 510, 515, 520, 525, 530, 535, 540,
575, and 625 K. After the first heating, the sample was
cycled between 300 and 625 K three times to investigate the
long term stability. All the data were Rietveld refined using
the program FULLPROF.15 Data above 2␪=40° were found to
be very weak for all samples and they were omitted in the
refinements. A pseudo-Voigt function with seven parameters
͑three Gaussian, two Lorentzian, and two asymmetry͒ was
found to give a good description of the peak shapes; the
background was fitted using linear interpolation between ap-
proximately 60 points. Selected crystallographic details are
listed in Table I for data sets recorded at 300 K before and
after heating. Full details of the multitemperature refine-
ments are given in the supporting information ͑Tables S1–
S4͒ for both zone-melted and quenched sample.
Here, we investigate the thermal stability of two com-
pletely phase-pure undoped Zn4Sb3 samples; one prepared
by the powder synthesis quench method2 and one prepared
a͒
Iversen et al. have shown that the zone-melting tech-
nique leads to samples with ZTϳ1.3 at 673 K, which is
Author to whom correspondence should be addressed. Tel. :ϩ45
89423969. Fax: ϩ45 86196199. Electronic mail: bo@chem.au.dk.
0003-6951/2008/92͑16͒/161907/3/$23.00
92, 161907-1
© 2008 American Institute of Physics
Downloaded 11 Oct 2012 to 171.67.34.69. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

161907-2
B. L. Pedersen and B. B. Iversen
Appl. Phys. Lett. 92, 161907 ͑2008͒
TABLE I. Crystallographic details and refinement residuals for the zone
melted ͑ZM͒ and the quenched ͑Q͒ samples. Start refers to refinement of the
dataset recorded before heating and after refers to refinement of the dataset
recorded at 300 K after three heating cycles.
ZM start
Q start
ZM after
Q after
T ͑K͒
300
5
300
5
300
5
300
5
Texpt ͑min͒
No. of data points
No. of reflections
No. of parameters
RF ͑%͒
7513
4682
88
7507
4863
90
7506
4613
99
7507
4682
116
2.05
2.12
2.72
3.72
4.11
1.34
1.84
2.39
3.15
2.91
3.19
4.09
4.82
6.59
14.10
2.83
2.71
4.06
6.16
12.50
RBragg ͑%͒
Rp ͑%͒
Rwp ͑%͒
2
␹
a=b ͑Å͒
c ͑Å͒
12.2182͑1͒ 12.2257͑1͒ 12.2338͑1͒ 12.2383͑2͒
12.4106͑1͒ 12.4162͑1͒ 12.4271͑1͒ 12.4256͑2͒
X
0.220͑4͒
0.019͑4͒
1.11͑2͒
1.27͑3͒
2.19͑5͒
1
0.241͑4͒
0.023͑4͒
1.26͑2͒
1.25͑3͒
2.11͑5͒
1
0.053͑3͒
0.010͑4͒
1.13͑3͒
1.13͑3͒
2.26͑6͒
1
0.218͑4͒
0.0001͑0͒
1.01͑5͒
0.94͑6͒
2.11͑11͒
1
Y
Biso Sb1
Biso Sb2
Biso Zn
Occupancy Sb1
Occupancy Sb2
Occupancy Zn1
Occupancy Zn2
Occupancy Zn3
Occupancy Zn4
Volume ͑Å3͒
wt % ZnSb
wt % ZnO
wt % Sb
FIG. 1. Powder patterns of Zn4Sb3 prepared by zone melting ͑a͒ and by
quench synthesis ͑b͒ recorded at 300 K before heating and after three heat-
ing cycles. The vertical lines correspond to the position of Bragg peaks of
͑from the top͒ Zn4Sb3, ZnSb, ZnO, and Sb.
1
1
1
1
0.921͑5͒
0.063͑2͒
0.055͑4͒
0.052͑1͒
1604.5͑3͒
0
0.905͑3͒
0.052͑1͒
0.047͑2͒
0.032͑1͒
1607.2͑3͒
0
0.954͑4͒
0.059͑3͒
0.022͑2͒
0.045͑2͒
1610.7͑1͒
4.8͑2͒
0.944͑7͒
0.065͑4͒
0.034͑5͒
0.045͑4͒
1611.7͑4͒
4.0͑2͒
sample only exhibits small additional impurity peaks and it is
clear that the thermal degradation is much more limited for
this sample. Figure 2͑a͒ plots the weight fraction of Zn4Sb3,
determined by Rietveld refinement, in the first heating cycle
͑300–625 K͒ of the two samples. It is quite surprising to
observe that the quenched sample starts to decompose even
below 500 K. At 500 K, the sample has already lost
ϳ4 wt. % of the initial Zn4Sb3 phase, and decomposition
phases ZnSb, ZnO, and Sb are observed. Further heating
accelerates the degradation of Zn4Sb3 and at 625 K, only
ϳ60 wt. % of the original Zn4Sb3 phase is left. Simulta-
neously, the amount of elemental Sb linearly increases and
reaches almost 20 wt. % at 625 K ͓Fig. 2͑b͔͒. A large quan-
tity of ZnO forms between 575 and 625 K, where it increases
from ϳ1 wt. % at 575 K to ϳ10 wt. % at 625 K, which
corroborates the idea of Zn disappearing from the structure
and transforming into ZnO during heating in air.12 Recent
tracer diffusion studies have shown that Zn is highly mobile
with properties approaching ionic conductors such as AgI.17
The extreme mobility of Zn in Zn4Sb3 may also be the origin
of the extreme sensitivity of the thermoelectric properties to
the sample compaction method.18
0
0
2.5͑4͒
1.57͑8͒
21.1͑4͒
32.3͑2͒
0
0
comparable to the highest values reported for samples
prepared by the conventional quench method.13 Figure 1
shows powder synchrotron x-ray diffraction patterns re-
corded at 300 K on the two samples. Refinement against the
structural model of Snyder et al.3 reveals that both samples
are completely phase pure. Even with an intense third
generation synchrotron source, which routinely allows the
detection of crystalline phases down to a 0.3% level,16 no
impurity peaks are observed. The structural data for the
two samples are comparable to each other and to the values
given by Snyder et al.3 The lattice constants are a
=12.2257͑1͒ and 12.2182͑1͒ Å, and c=12.4162͑1͒ and
12.4106͑1͒ Å for quenched and zone-melted sample, respec-
tively, compared to a=12.2282͑3͒ Å and c=12.4067͑4͒ Å
reported by Snyder et al.3 from single crystal data. When
comparing the structural parameters, it is seen that the zone-
melted sample has a significantly higher Zn content than the
quench sample ͑see supporting material͒ but further analysis
is needed to fully understand the atomic scale differences
between the samples.
Figure 1 shows the synchrotron powder patterns at
300 K after three heating cycles up to 625 K ͑i.e., the
thermal history is 300→625→300→625→300→625
→300 K͒. The difference in the thermal behavior of the two
samples is striking. The sample synthesized by the conven-
tional quench method has a large number of additional high-
intensity peaks generated from decomposition phases, iden-
tified by Rietveld refinement as ZnSb, ZnO, and elemental
Sb ͑details are given in Table I͒. In contrast, the zone-melted
The zone-melted sample displays a very different behav-
ior with essentially no weight loss up to 520 K ͓Fig. 2͑a͔͒.
Between 520 and 525 K, there is a small weight loss of
ϳ3 wt. % Zn4Sb3, simultaneous with the formation of ZnO
and elemental Sb. This is consistent with Zn leaving the
sample and forming ZnO in an atmosphere of air. Thus,
Zn4Sb3 appears to have a subtle phase transition at ϳ523 K.
In contrast to the quenched sample, further heating to 625 K
does not induce additional decomposition of the zone-melted
sample and at 625 K, there is still ϳ97 wt. % of the original
phase left. Thus, for the zone-melted sample slight decom-
position is an isolated event at ϳ523 K.
Figure 2͑c͒ shows the weight fraction of the remaining
Zn4Sb3 phase determined at 300 K after each of the three
Downloaded 11 Oct 2012 to 171.67.34.69. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions

161907-3
B. L. Pedersen and B. B. Iversen
Appl. Phys. Lett. 92, 161907 ͑2008͒
FIG. 2. ͑Color online͒ ͑a͒ Weight frac-
tion of Zn4Sb3 as a function of tem-
perature in the first heating cycle. ͑b͒
Weight fraction of impurity phases
ZnSb, ZnO, and Sb as a function of
temperature in the first heating cycle.
͑c͒ Weight fraction of Zn4Sb3 as a
function of heating cycles. ͑d͒ Weight
fraction of impurity phases ZnSb,
ZnO, and Sb as a function of heating
cycles. Zn4Sb3 prepared by quench
synthesis has open circles and the zone
melting sample has open squares.
Blue, black, and red squares corre-
spond to ZnSb, ZnO, and Sb in the
zone-melted sample, and blue, black,
and red circles correspond to ZnSb,
ZnO, and Sb in the quenched sample,
respectively.
heating cycles. Clearly, the sample produced by quench syn-
thesis has a serious stability problem. After the first heating
cycle, only ϳ60 wt. % remains and repeated heating causes
further degradation leaving only ϳ42 wt. % of the original
Zn4Sb3 phase after three cycles. The zone-melted sample is
much more stable than the quenched sample, also during
thermal cycling ͓Fig. 2͑d͔͒. Although each heating/cooling
cycle induces a small decomposition of about 3% at 523 K,
more than 91 wt. % of the target phase is still remaining after
three cycles. Iversen et al. observed that zone-melted
samples reach constant thermoelectric properties after re-
peated thermal cycling.13 This suggests that if cycling was
continued, the zone-melted phase would have reached a
stable plateau of degradation. The limited degradation of the
zone-melted sample is an enormous improvement in stability
compared to samples prepared by the quench method. The
present experiments were performed on powders and it is
conceivable that compacted samples would be more stable
during thermal cycling. However, it is unlikely that this
would improve stability properties of the quenched sample
enough to meet commercial requirements.
In summary, the thermal stability of two Zn4Sb3
samples, prepared by different synthesis methods, was inves-
tigated by multitemperature synchrotron powder diffraction.
ZnSb, ZnO, and elemental Sb have been identified as decom-
position products, and the degree of decomposition was
quantified by Rietveld refinement. The stability of Zn4Sb3 is
enormously improved by the zone-melting synthesis tech-
nique. Thermal cycling leaves only ϳ42 wt. % of the origi-
nal phase in case of the quenched sample compared with
ϳ91 wt. % in the zone-melted sample. The high temperature
decomposition of Zn4Sb3 is found to be extremely complex
and challenging studies remain to understand its detailed be-
havior. We conclude that Zn4Sb3 prepared by zone melting is
significantly more stable than Zn4Sb3 synthesized by the
conventional quench method and indeed, zone-melted
Zn4Sb3 appears to be a promising candidate for commercial
implementation in thermoelectric generators operating at in-
termediate temperatures.
The authors gratefully acknowledge the beam time ob-
tained at beam line BL02B2 at SPring8, Japan, and Dr. Eiji
Nishibori and Shinobu Aoyagi are thanked for the assistance
during measurements. The work was supported by DAN-
SCATT and GRUNDFOS A/S.
1T. Caillat, J.-P. Fleurial, and A. Borshchevsky, J. Phys. Chem. Solids 58,
1119 ͑1996͒.
2T. Caillat, A. Borshchevsky, and J.-P. Fleurial, U.S. Patent 6,942,728 ͑13
September 2005͒.
3G. J. Snyder, M. Christensen, E. Nishibori, T. Caillat, and B. B. Iversen,
Nat. Mater. 3, 458 ͑2004͒.
4F. Cargnoni, E. Nishibori, P. Rabiller, L. Bertini, G. J. Snyder, M. Chris-
tensen, C. Gatti, and B. B. Iversen, Chem.-Eur. J. 10, 3861 ͑2004͒.
5Y. Mozharivskyj, A. O. Pecharsky, S. Bud’ko, and G. J. Miller, Chem.
Mater. 16, 1580 ͑2004͒.
6Y. Mozharivskyj, Y. Janssen, J. L. Harringa, A. Kracher, A. O. Tsokol, and
G. J. Miller, Chem. Mater. 18, 822 ͑2006͒.
7A. S. Mikhaylushkin, J. Nylen, and U. Häussermann, Chem.-Eur. J. 11,
4912 ͑2005͒.
8J. Nylén, M. Andersson, S. Lidin, and U. Häussermann, J. Am. Chem.
Soc. 126, 16306 ͑2004͒.
9J. Nylen, S. Lidin, M. Andersson, B. B. Iversen, H. X. Liu, N. Newman,
and U. Häussermann, Chem. Mater. 19, 834 ͑2007͒.
10L. T. Zhang, M. Tsutsui, K. Ito, and M. Yamaguchi, J. Alloys Compd.
358, 252 ͑2003͒.
11B. L. Pedersen, H. Birkedal, E. Nishibori, A. Bentien, M. Sakata, M.
Nygren, P. T. Frederiksen, and B. B. Iversen, Chem. Mater. 19, 6304
͑2007͒.
12B. L. Pedersen, H. Birkedal, P. T. Frederiksen, and B. B. Iversen, Proceed-
ings of the 25th International Conference on Thermoelectrics, 2006 ͑un-
published͒, p. 520.
13B. B. Iversen, B. Lundtoft, M. Christensen, and D. Platzek, WO Patent
No. 128,467 ͑7 December 2006͒.
14E. Nishibori, M. Takata, K. Kato, M. Sakata, Y. Kubota, S. Aoyagi, Y.
Kuroiwa, M. Yamakata, and N. Ikeda, J. Phys. Chem. Solids 12, 2095
͑2001͒.
15J. Rodríguez-Carvajal, FULLPROF 2000 ͑2001͒.
16S. Johnsen, A. Bentien, G. K. H. Madsen, M. Nygren, and B. B. Iversen,
Chem. Mater. 18, 4633 ͑2006͒.
17E. Chalfin, H. X. Lu, and R. Dieckmann, Solid State Ionics 178, 447
͑2007͒.
18B. L. Pedersen, H. Birkedal, M. Nygren, P. T. Frederiksen, and B. B.
Iversen, Appl. Phys. Lett. 89, 242108 ͑2006͒.
Downloaded 11 Oct 2012 to 171.67.34.69. Redistribution subject to AIP license or copyright; see http://apl.aip.org/about/rights_and_permissions