Intermetallics 32 (2013) 312e317

Contents lists available at SciVerse ScienceDirect

Intermetallics
j o u r n a l h o m e p a g e : w w w.e l s e v i e r.c o m / l o c a t e / i n t e r m e t

Improving p-type thermoelectric performance of Mg2(Ge,Sn) compounds via solid

solution and Ag doping
Guangyu Jiang a, Luxin Chen a, Jian He b, Hongli Gao a, Zhengliang Du a, Xinbing Zhao a, Terry M. Tritt b,
Tiejun Zhu a, c, *
a

State Key Laboratory of Silicon Materials, Department of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, Peoples Republic of China
Department of Physics and Astronomy, Clemson University, Clemson, SC 29634-0978, USA
c
Cyrus Tang Center for Sensor Materials and Applications, Zhejiang University, Hangzhou, 310027, Peoples Republic of China
b

a r t i c l e i n f
o
Article history:
Received 13 June 2012
Received in revised form
24 July 2012
Accepted 1 August 2012
Available online 9 October 2012
Keywords:
B. Thermoelectric properties
C. Melting
E. Physical properties, miscellaneous
F. Diffraction (electron, neutron and X-ray)
G. Thermoelectric power generation

a b s t r a c t
Mg2X (X Si, Ge, and Sn) compounds have attracted increasing attention owing to their promising
thermoelectric properties and green constituent elements. While most studies to date have been on ntype Mg2(Si,Sn) solid solutions, we herein report Ag-doping study in Mg2(Ge,Sn) solid solutions to
pursue higher p-type performance. Two series of samples, Mg2Ge1 xSn x and Mg2

1. Introduction
The impending energy crisis and the environmental concerns
associated with the use of fossil fuel have been imposed a demand
for sustainable and efcient energy conversion technologies,
including thermoelectricity. Thermoelectric (TE) devices can
directly convert heat from various sources to electricity, or work
reversely as heat pump [1e3]. The energy conversion efciency of
a TE device is largely determined by the performance of the TE
materials, gauged by the dimensionless gure of merit ZT,

ZT
=k

y Ag yGe 0.4Sn 0.6,

were

prepared by melting in evacuated Ta tubes, followed by hot pressing, and the thermoelectric properties
of the solid solutions were investigated from room temperature to 723 K. An n- to p-type crossover in
the Seebeck coefcient was observed for Mg2Ge0.4Sn0.6 and Mg2Ge0.2Sn0.8 at elevated temperatures.
To enhance p-type performance, we doped Ag in Mg2Ge0.4Sn0.6 that exhibited lowest thermal
conductivity among the Mg2(Ge,Sn) solid solutions. It was found that (1) p-type behavior has been
established in Mg2
yAgyGe0.4Sn0.6 (y 0.005, 0.01, 0.02, 0.04) samples in the entire temperature range studied; and (2) the
electrical conductivity increased with increasing Ag content until reaching the solution limit w0.02. As
a result, a dimensionless gure of merit ZT w0.38 is attained for Mg1.98Ge0.4Sn0.6Ag0.02 at 675 K, which is
one of the highest reported values in p-type Mg2X compounds.
2012 Elsevier Ltd. All rights reserved.

a2 sT

where a is the Seebeck coefcient, s the electrical conductivity and

k the thermal conductivity, and T the absolute temperature,
respectively. The state-of-the-art TE materials such as PbTe [4e6],

* Corresponding author. State Key Laboratory of Silicon Materials, Department of

Materials Science and Engineering, Zhejiang University, Hangzhou 310027,
Peoples Republic of China. Tel.: 86 571 87952181; fax: 86 571 87951203.
E-mail address: zhutj@zju.edu.cn (T. Zhu).

0966-9795/$ e see front matter

2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.intermet.2012.08.0 02

GeTeeAgSbTe2 [7] and lled skutterudites [8] have ZT > 1.0 in

the intermediate temperature range (w500 Ke900
K).
While developing higher ZT materials has always been at the
core of thermoelectric study, it is getting more attention
that TE materials should be made from naturally abundant
non-toxic elements. To this end, Mg2X (X Si, Ge and Sn) solid
solutions are promising green TE materials in the intermediate
temperature range.
A typical TE device consists of two legs, each made of n-type or
p-type material. The TE performance and the mechanical
properties of the n-type and p-type materials had better be
similar to maxi- mize the output [3]. In the case of Mg2X-based TE
compounds, the ZT of n-type Mg2X materials are signicantly
higher than that of the p-type counterpart. For example, the ZT of
n-type Sb and Bi doped Mg2(Si,Sn) compounds is w1.0 [9e13],
comparable to those of PbTe and lled skutterudites. On the other
hand, Isoda et al. [14] have reported a ZT w 0.32 for p-type Li
and Ag dual-doped Mg2Si0.25Sn0.75. Ihou-Mouko et al. [15] have
reported a ZT w 0.36 for p-type Ga-doped Mg2Si0.6Ge0.4, which
set the bench mark for high performance p-type Mg2X (X Si, Ge
and Sn) compounds.
The inferior p-type performance in Mg2X compounds arises
largely from that the electron mobility is intrinsically higher than

G. Jiang
G. Jiang
et al.
et /al.Intermetallics
/ Intermetallics
32 (2013)
32 (2013)
312e317
312e317

22

22

the hole mobility, disfavoring p-type conduction [16]. Among pure

binary Mg2X compounds Mg2Sn has highest hole to electron
mobility ratio, making it appropriate for p-type doping. Furthermore, it is established that a high ZT is hard to achieve in any pure
binary Mg2X compounds because of high intrinsic thermal
conductivity, as such, solid solution becomes the routine to attain
higher ZT. To this end, there are mainly two series of Mg2Sn based
solid solutions, Mg2(Si,Sn) and Mg2(Ge,Sn) compounds. Fedorov
et al. early pointed out that despite a relatively high thermal
conductivity Mg2Ge0.4Sn0.6 solid solution has a favorable hole to
electron mobility ratio [17], suggesting that Mg2(Ge,Sn) solid
solutions are more promising for p-type thermoelectricity than
Mg2(Si,Sn) solid solutions [18].
The present work consists of two parts: solid solution and hole
doping. In the rst part, we prepare and characterize Mg2(Ge,Sn)
solid solutions with varying Ge/Sn ratios. The purpose is to
identify the most promising candidate for the subsequent hole
doping study. In literature, Ag and Li have been proved to be
effective p- type dopants for Mg2(Si,Sn) solid solutions [14], while
Ga, In, and Na have been tried for Mg2Ge0.4Sn0.6 [17,19]. In the
second part of this work we conduct Ag doping study in
Mg2Ge0.4Sn0.6, which exhibits the relatively low thermal
conductivity among Mg2(Ge,Sn) solid solutions.
2. Experiment
Mg2Ge1 xSnx (x 0.2, 0.4, 0.5, 0.6, 0.8) and Mg2 yAgyGe0.4Sn0.6
(y 0, 0.005, 0.01, 0.02, 0.04) samples were prepared. Stoichiometric amount of Mg (nominal 3 N), Ge (3 N), Sn (3 N) and Ag (4 N)
powders were weighed and mixed, an excess of 0.5 wt.% Mg was
added to compensate the evaporation loss. The admixture was
sealed in a Ta tube in Ar atmosphere and then was heated at 1273
K for 24 h. The product was pulverized in an agate mortar and
sieved. The ne powder <75 mm was then hot pressed at
873e973 K for 2 h into f 12.6 mm pellets. All hot pressed pellets
had w99% or more of theoretical densities.
The phase purity of samples was checked on a Rigaku D/
MAX02550PC diffractometer with monochromatic Cu ka radiation.
Thermal diffusivity, D, and specic heat, Cp, measurements were
performed on a Netzsch LFA457 Micro Flash Analyzer. The thermal
conductivity k was then calculated by the relation: k rdDCp,
where rd is the density of sample. Then bar shape samples were
cut from the pellets for electrical conductivity, s, and Seebeck
coef- cient, a, measurements on a custom designed apparatus
using a DC four-probe method and differential voltage/temperature
technique, respectively [20]. All TE properties, s, a, and k, were
measured from room temperature to 723 K.
3. Results and discussion
3.1. Undoped Mg2Ge1

xSnx

solid solutions

In this section we present the data taken on undoped

Mg2Ge1 xSnx solid solutions so as to identify a good parent
compound for the subsequent hole doping study. Mg2Ge1 xSnx
(x 0.2, 0.4, 0.5, 0.6 and 0.8) samples were synthesized and the
thermoelectric properties characterized. Fig. 1(a) shows the XRD
patterns of the Mg2Ge1 xSnx (x 0.2, 0.4, 0.5, 0.6, and 0.8) samples.
All peaks of each sample can be indexed to an anti-uorite crystal
structure (space group Fm3m) and are located between those of
binary Mg2Ge and Mg2Sn. The XRD patterns indicate no immiscibility gap, which is consistent with the report by Makarov [21]. A
trace amount of MgO phase was observed in some samples but it
should not affect the following discussion. Fig. 1(b) shows the
lattice parameter, a, increases linearly with the nominal Sn content

Fig. 1. (a) XRD patterns of Mg2Ge1 xSnx samples. Black diamonds denote the MgO
impurity phase. (b) Lattice parameters of Mg2Ge1 xSnx follows the Vegards law (the
straight line).

x. The fairly good linearity is consistent with the Vegards law,

conrming the formation of Mg2GeeMg2Sn solid solutions.
The temperature dependences of s and a for Mg2Ge1 xSnx solid
solutions are presented in Fig. 2(a) and (b), respectively. A characteristic thermal excitation behavior is evident from the
exponential increase of s at elevated temperatures. The ln s
versus 1/T above
500 K is plotted in the inset of Fig. 2(a), the slope is a gauge of the
energy gap. The gap of Mg2Ge1 xSnx systematically decreases with
increasing x: for example, the gap is w0.52 eV for Mg2Ge0.4Sn0.6
and 0.46 eV for Mg2Ge0.2Sn0.8, respectively. We note that the
energy gaps of binary Mg2Ge and Mg2Sn were reported to be 0.55
eV and
0.35 eV, respectively [18,22]. As shown in Fig. 2(b), the Seebeck
coefcients are strongly temperature dependent, roughly following
a 1/T behavior as expected from a semiconductor. All samples are
n- type in the vicinity of room temperature. However, an n-type to
p- type crossover in the Seebeck coefcient takes place at w670 K
for Mg2Ge0.4Sn0.6, and near 600
K for
Mg2Ge0.2Sn0.8. In
Mg2Ge0.6Sn0.4 and Mg2Ge0.5Sn0.5, the sign change does not take
place but the trend of sign change is apparent. All these evidence
indicate that p- type conduction is established at elevated
temperatures in
these samples. Nonetheless, the hole
concentration in undoped Mg2X solid solution is too low to
support a good p-type thermoelectricity [17,19], hence hole
doping is the next step.
Before proceeding to the results of hole doping study, we
inspect the impact of solid solution on the thermal conductivity.
Fig. 3(a) and (b) show the temperature dependences of thermal
conductivity k and (k ke) of Mg2Ge1 xSnx samples, respectively.
We estimated the carrier thermal conductivity, ke, using the
WiedemanneFranz relationship, ke LsT, where L is the Lorenz
number for a degenerate semiconductor and the value is about

Fig. 2. Temperature dependences of electrical conductivity s (a) and Seebeck coef1

cient a (b) for Mg2Ge1 xSn samples. Inset of (a) shows the good (2k T) dependence
x
B
of ln s, indicating a characteristic thermal excitation behavior.

Fig. 4. (a) XRD patterns of Ag doped Mg2

eters of Mg2 yAgyGe0.4Sn0.6 samples.

yAgyGe0.4Sn0.6 samples.

(b) Lattice param-

Fig. 3. Temperature dependences of thermal conductivity k (a) and the (k ke) term (b) for Mg2Ge1 xSnx solid solutions. The term (k ke) is a good approximation of the lattice
thermal conductivity kL near room temperature, where the bipolar thermal conductivity kb only play a minor role. (c) Lattice thermal resistivity plotted against the proportion x at
room temperature.

2.0
10 8 V2 K 2 for Mg2Sn based solid solutions [12].
It should be noted that it is less appropriate to call the (k ke)
term lattice thermal conductivity as the contribution
from bipolar effect is substantial between 500 and 700 K in
this study. The thermal conductivity of all samples rst
decreases and then increases with temperature, due to the onset
of bipolar conduction. The scenario of the bipolar conduction is
also consistent with the n-type to p-type crossover in the
Seebeck coefcient (Fig. 2(b)). Furthermore, the lattice thermal
conductivity kL of a solid solution varies approximately to the
rule [23].

1=kL 1=kL 0 4x1

0

x 1=kL 0:5

1=kL

(1)

where (kL)0.5 is the lattice thermal conductivity when x, the

proportion of the second component in the solid solution, is 0.5
and (kL)0 is the lattice thermal conductivity when x is 0. We note
that the (k ke) term is a good approximation of the lattice
thermal conductivity near room temperature where the bipolar
contribution only plays a minor role. Fig. 3(c) shows the composition
dependence of experimental and calculated lattice thermal resistivities (1/kL) at room temperature according to formula (1). We
use the lattice thermal resistivities of Mg2Ge and Mg2Sn as
references in the calculation [24], respectively. As shown in
Fig. 3(c), the calculated line is in good agreement with our
experimental data, indicating the point defects due to solid
solution are the dominant phonon scatters. Following this line, the
subsequent hole doping is expect to decrease the lattice thermal
conductivity and optimize the electrical properties, and thus
contribute to higher ZT.

Fig. 5. Temperature dependences of electrical conductivity s (a) and Seebeck coefcient a (b) for Mg2 yAgyGe0.4Sn0.6.

3.2. P-type doping for Mg2Ge0.4Sn0.6 solid solutions

Taking into consideration the values and temperature dependences of s, a, and k for Mg2(Ge,Sn) samples, we choose
Mg2Ge0.4Sn0.6 as the parent compound for hole doping study. The
dopant that we choose is Ag in this study. Fig. 4(a) shows the XRD
patterns of Ag-doped Mg2 yAgyGe0.4Sn0.6 (y 0.005, 0.01, 0.02 and
0.04) samples. No secondary phase was detected except for a trace
amount of MgO in some samples. Lattice parameter is plotted in
Fig. 4(b). It rst increases to the top value at y 0.02 and then
decreases with increasing Ag content. We think that y 0.02 is
likely the solubility limit for Ag in Mg2Ge0.4Sn0.6.
Fig. 5(a) and (b) show the temperature dependences of s and
a for Mg2 yAgyGe0.4Sn0.6 samples, respectively. Apparently, Ag is an
effective hole dopant for Mg2(Ge,Sn) solid solutions. First, in
contrast to the undoped samples, the Seebeck coefcients of doped
samples become all positive in the temperature range studied,
indicating that the p-type conduction is established. It is noted
that a
rst increases then
decreases with
increasing
temperature, contrary to
the observed trend in
electrical
conductivity. In addi- tion, the bipolar effect is more pronounced
upon Ag doping. When the minority carriers start to manifest
itself, a drops fast. Second, in contrast to the semiconducting
behavior of the undoped sample, the s of doped samples rst
decreases then increases with increasing temperature, typical of
a degenerate semiconductor. We believe that the decrease of s
below w500 K is caused by electrone phonon coupling while the
increase of s above w500 K by the onset of thermal excitation of
minor carriers. With 0.5 at.% Ag doping, the
s of Mg2 yAg yGe 0.4 Sn 0.6 reaches 20,000 S m 1 at room

Fig. 6. Temperature dependences of thermal conductivity k (a) and the (k ke) term (b)
for Mg2 xGe0.4Sn0.6Agx. At low temperatures, k ke is mainly the lattice thermal
conductivity kL and the bipolar thermal conductivity kb is dominating after the onset of
thermal excitation of minor carriers.

followed by quenching, pulverizing and hot pressing.

Thermoelectric properties of Mg2Ge1 xSnx solid solutions were
evaluated from room temperature to 723 K and Mg2Ge0.4Sn0.6 was
chosen as the parent compound for hole doping by Ag. Ag was
proved to be an
effective acceptor in Mg2Ge0.4Sn0.6, as
corroborated in the signicant increase in the electrical
conductivity and ZT. A maximum gure of merit ZT of
0.38 was achieved at 675 K for Mg1.98Ag0.02Ge0.4Sn0.6, which is one
of the best reported values in p-type Mg2X (X Si, Ge and Sn)
ther- moelectric materials. It was nd that the solubility of
Ag in Mg2Ge0.4Sn0.6 is about 0.02. Over saturated Ag doping in
Mg2Ge0.4Sn0.6 is unfavorable to the improvement of TE properties.
Acknowledgments
The work was supported by the National Natural Science
Foundation of China (50971115, 51061120455 and 51171171),
the Natural
Science Foundation
of
Zhejiang
Province
(Z4090204), the Fundamental Research Funds for the Central
Universities (2011QNA4034), and the National Science Foundation
of the United States (1008073).
References

Fig. 7. Temperature dependences of power factor (a) and gure of merit ZT (b) for
Mg2 yAgyGe0.4Sn0.6 samples.

temperature, w150 times that of the undoped one. With more

Ag doping, s further increases. The maximum s at room
temperature is
27,000 S m 1 for y 0.01 and 0.02, higher than that of Ga-doped
Mg2Si0.4Ge0.6 [15] and those of p-type Mg2Ge0.4Sn0.6 (doped by
Al, In, Na) [19]. However, the s does not show a further increase
with increasing Ag doping: the s actually drops at y 0.04. This
is likely caused by the formation of a secondary phase when the
Ag
content is beyond the solution limit, nonetheless, we do not
observe any secondary phase except MgO by XRD measurements.
Fig. 6(a) shows the temperature dependence of thermal
conductivity k of Mg2

yAgyGe0.4Sn0.6.

The k ke term as a function of

temperature and Ag content is plotted in Fig. 6(b). The (k ke) term

rst decreases with increasing temperature in T 1 relationship,
characteristic of the Umklapp process, then increases with
increasing temperature due to the onset of the bipolar effect.
Concerning the Ag content dependence, the k ke term decreases
roughly with increasing Ag content.
The temperature dependences of the power factor and ZT
are plotted in Fig. 7(a) and (b), respectively. A remarkable
increase in power factor was attained for the Ag doped samples.
The maximum power factor increases with increasing Ag content
before reaching the solution limit y 0.02, and then decreases. The
maximum ZT value of w0.38 was obtained at 675 K for y 0.02.
The maximum ZT for y 0.01 is also above 0.3. Although the ZT
value is still low compared with that of the n-type Mg2(Si,Sn), the
obtained ZT values are the state
of the art for p-type Mg2X (X Si, Ge and Sn) solid
solutions.
4. Conclusions

[1] Snyder GJ, Toberer ES. Complex thermoelectric materials. Nat Mater 2008;7:
105e14.
[2] Tritt TM. Thermoelectric materials e holey and unholey semiconductors.
Science 1999;283:804e5.
[3] Rowe DM, Bhandari CM. Modern thermoelectrics. Englewood Cliffs, NJ: Reston
Publishing Co.; 1983.
[4] Heremans JP, Jovovic V, Toberer ES, Saramat A, Kurosaki K, Charoenphakdee A,
et al. Enhancement of thermoelectric efciency in PbTe by distortion of the
electronic density of states. Science 2008;321:554e7.
[5] Pei YZ, LaLonde AD, Heinz NA, Shi XY, Iwanaga S, Wang H, et al. Stabilizing the
optimal carrier concentration for high thermoelectric efciency. Adv Mater
2011;23:5674e8.
[6] Pei YZ, Shi XY, LaLonde A, Wang H, Chen LD, Snyder GJ. Convergence of
electronic bands for high performance bulk thermoelectrics. Nature 2011;473:
66e9.
[7] Yang SH, Zhu TJ, Sun T, Zhang SN, Zhao XB, He J. Nanostructures in highperformance (GeTe) x(AgSbTe2)100 x thermoelectric materials. Nanotechnology 2008;19:245707.
[8] Shi X, Yang J, Salvador JR, Chi MF, Cho JY, Wang H, et al. Multiplelled skutterudites: high thermoelectric gure of merit through separately
optimizing electrical and thermal transports. J Am Chem Soc 2011;133:
7837e46.
[9] Zhang Q, He J, Zhu TJ, Zhang SN, Zhao XB, Tritt TM. High gures of merit and
natural nanostructures in Mg2Si0.4Sn0.6 based thermoelectric materials. Appl
Phys Lett 2008;93:102109.
[10] Zaitsev VK, Fedorov MI, Gurieva EA, Eremin IS, Konstantinov PP, Samunin AY,
et al. Highly effective Mg2Si1 xSnx thermoelectrics. Phys Rev B 2006;74.
[11] Gao HL, Zhu TJ, Liu XX, Chen LX, Zhao XB. Flux synthesis and thermoelectric
properties of eco-friendly Sb doped Mg2Si 0.5Sn0.5 solid solutions for energy

Mg2Ge1 xSnx and Ag-doped Mg2 yAgyGe0.4Sn0.6 samples are

prepared by melting the starting single elements in sealed Ta
tubes,

harvesting. J Mater Chem 2011;21:5933e7.

[12] Du Z, Zhu TJ, Chen Y, He J, Gao HL, Jiang GJ, et al. Roles of interstitial Mg in
improving thermoelectric properties of Sb-doped Mg2Si0.4Sn0.6 solid solutions. J Mater Chem 2012;22:6838e44.
[13] Liu W, Tan XJ, Yin K, Liu HJ, Tang XF, Shi J, et al. Convergence of conduction
bands as a means of enhancing thermoelectric performance of n-type
Mg2Si1 xSnxSolid solutions. Phys Rev Lett 2012;108:166601.
[14] Isoda Y, Tada S, Nagai T, Fujiu H, Shinohara Y. Thermoelectric properties of ptype Mg2Si0.25Sn0.75 with Li and Ag double doping. J Electron Mater 2010;39:
1531e5.
[15] Ihou-Mouko H, Mercier C, Tobola J, Pont G, Scherrer H. Thermoelectric
properties and electronic structure of p-type Mg2Si and Mg2Si0.6Ge0.4
compounds doped with Ga. J Alloys Compd 2011;509:6503e8.

[16] Winkler U. Die Elektrischen Eigenschaften der Intermetallischen Verbindungen Mg2Si, Mg2Ge, Mg2Sn und Mg2Pb. Helv Phys Acta 1955;28:663e7.
[17] Fedorov MI, Zaitsev VK, Eremin IS, Gurieva EA, Burkov AT, Konstantinov PP,
et al. Transport properties of Mg2X0.4Sn0.6 solid solutions (X Si, Ge) with ptype conductivity. Phys Solid State 2006;48:1486e90.
[18] Zaitsev VK, Fedorov MI, Gurieva EA, Eremin IS, Konstantinov PP, Samunin AY,
Vedernikov MV. Thermoelectrics of n-type with ZT > 1 based on Mg2Sie
Mg2Sn solid solutions. In: Proceedings 24th International conference on
thermoelectrics, IEEE; 2005. p. 189e195.
[19] Fedorov MI, Zaitsev VK, Isachenko GN, Eremin IS, Gurieva EA, Konstantinov
PP, et al. Kinetic properties of p-type Mg2Ge0.4Sn0.6 solid solutions. In:
Proceedings 24th International conference on thermoelectrics, IEEE; 2005. p.
110e113.

[20] Zhang Q, Yin H, Zhao XB, He J, Ji XH, Zhu TJ, et al. Thermoelectric properties of
n-type Mg2Si0.6 ySbySn0.4 compounds. Phys Status Solidi A Appl Mat 2008;
205:1657e61.
[21] Makarov ES, Muntyanu S, Sokolov EB, Slesareva GA. Study of the Mg2Sne
Mg2Ge system. Inorg Mater 1966;2:1830e2.

[22] Mead CA. Photothresholds in Mg2Ge. J Appl Phys 1964;35:2460e2.

[23] Goldsmith HJ. Introduction to thermoelectricity. Berlin: Springer; 2009.
[24] Zaitsev VK, Tkalenko EN, Nikitin EN. Lattice thermal conductivity of Mg2Sie
Mg2Sn, Mg2GeeMg2Sn and Mg2SieMg2Sn solid solutions. Sov Phys Solid State
1969;11:221e4.