J. Li et al. / Journal of Alloys and Compounds 741 (2018) 1148e1152
1151
clear that kl decreases significantly with increasing Sb content. The
lattice thermal conductivity for Mg2Si0.985Sb0.015 drops to values
around 1.0 Wmꢀ1Kꢀ1 at temperatures higher than 700 K, which is
one of the lowest values for Mg2Si systems. Substituting Sb atoms on
Si sites in Mg2Si induces point defects that can scatter phonons
effectively due to the large mass/size difference between Sb and Si
atoms. Higher Sb content means stronger phonon scattering thus
lower lattice thermal conductivity. Moreover, the high pressure
synthesis method employed in current work promotes the forma-
tion of interstitial Mg atoms [37], which can also scatter phonons
and further reduce the lattice thermal conductivity.
Fig. 4 demonstrates the temperature dependent ZT of HPS Sb-
doped Mg2Si samples based on the thermoelectric properties
measurements. Representative results of ambient-pressure syn-
thesized Sb-doped Mg2Si from literature are also presented for
comparison reason [20,24,36]. For the optimal Mg2Si0.985Sb0.015
sample, ZT increases with elevating temperature and reaches
0.94 at 873 K, with an average ZT as high as 0.49 for the whole
temperature range. As a comparison, the highest ZT values are less
than 0.7 for those ambient-pressure synthesized samples, with a
much lower average ZT values as well [20,24,36]. The better ther-
moelectric performance of our HPS samples originates from the
enhanced power factor and suppressed (lattice) thermal conduc-
tivity benefitted from high pressure synthesis. During the synthe-
sis, the sealed reaction environment ensured by high pressure
prohibits the volatilization or oxidation of Mg, which are usually
suffered by ambient-pressure synthesis methods. Moreover, high
pressure effectively promotes the formation of interstitial Mg. Both
substitutional Sb and interstitial Mg can contribute electrons to
increase the carrier concentration and scatter phonons to decrease
the lattice thermal conductivity, leading to the superior thermo-
electric performance. High pressure synthesis thus offers an alter-
native approach to synthesizing Mg2Si-based thermoelectric
materials. Note the temperature and pressure conditions for syn-
thesizing Mg2Si can readily be realized in industrial production
[41], which is advantageous for commercial applications of ther-
moelectric materials.
Fig. 4. ZT of Mg2Si1-xSbx (x ¼ 0, 0.005, 0.01, 0.015) samples as
a function of
temperature.
substantially lower than those attained in ambient-pressure syn-
thesized samples [20,24,36], which can be attributed to the higher
carrier concentration in the HPS sample. As shown in Fig. 3b, all the
samples show negative Seebeck coefficients indicating n-type
conduction. The undoped Mg2Si sample shows an early onset of
intrinsic conduction occurring around 650 K due to its low carrier
concentration. The Seebeck coefficient of Sb-doped samples shows
a roughly linear dependence on temperature, a typical behavior of
heavily doped semiconductors, and the absolute value of S de-
creases with increasing Sb content. The trend of the Seebeck co-
efficient with respect to temperature and Sb content is consistent
with that of the electrical resistivity. Compared with the samples
synthesized under ambient-pressure with similar composition
[20,24,36,38], the HPS Mg2Si0.985Sb0.015 sample shows a slightly
lower Seebeck coefficient (64
its higher carrier concentration.
m
VKꢀ1) at room temperature due to
Fig. 3c shows the temperature dependent power factor
(PF ¼ S2=
r
) of Mg2Si1-xSbx samples. PF for undoped Mg2Si is rela-
tively small with a peak value of 1400
m
Wmꢀ1Kꢀ2 reached around
4. Conclusions
400 K. After Sb doping, the power factor is greatly enhanced, and
the maximal PF of 3300
m
Wmꢀ1Kꢀ2 is achieved for the optimal
Highly densified single-phase Mg2Si1-xSbx samples were suc-
cessfully synthesized with high pressure synthesis followed by
spark plasma sintering. HPS can effectively eliminate the oxidation
or volatilization of Mg, and promote the formation of interstitial
Mg. Consequently, the thermoelectric performance of our HPS
samples is significantly improved compared with those ambient-
pressure synthesized samples. The optimal Mg2Si0.985Sb0.015 sam-
Mg2Si0.985Sb0.015 at 773 K. This value is higher than those achieved
in ambient-pressure synthesized Sb-doped Mg2Si samples since
the decrease of electrical resistivity is much more significant than
the degradation of Seebeck coefficient in our HPS sample pos-
sessing higher electron concentration. Moreover, the power factor
of Mg2Si0.985Sb0.015 maintains a relatively high value (>3000
m
Wmꢀ1Kꢀ2) when temperature is higher than 650 K, which is
ple shows substantially enhanced power factor (3300 m
Wmꢀ1K2 at
beneficial for the overall thermoelectric performance.
773 K) and greatly suppressed lattice thermal conductivity (ca. 1.0
Wmꢀ1Kꢀ1 in the high temperature region), leading to a ZT value as
high as 0.94 achieved at 873 K. Through implementing the strate-
gies of nanostructuring as well as alloying Si with Ge and Sn, further
enhancement of thermoelectric performance of Mg2Si-based
thermoelectric materials is anticipated with HPS method.
The temperature dependent thermal conductivity of Sb-doped
Mg2Si is shown in Fig. 3d. The thermal conductivity for undoped
Mg2Si is qualitatively consistent with the results reported previ-
ously [36,39]. After Sb doping, the thermal conductivity decreases
mildly but does not show a markedly dependence on Sb content,
especially at high temperature. The total thermal conductivity k is a
sum of the lattice thermal conductivity kl and the electronic
contribution ke. For our HPS samples with increasing Sb content, the
rapid increase in ke offsets the decrease in kl, resulting in the thermal
conductivity almost the same across different samples. ke is gener-
Acknowledgements
This work was supported by the Natural Science Foundation of
China (51525205, 51421091, 51402254, 51722209, and 51332005),
and the Key Basic Research Project of Hebei (14961013D).
allyestimated with the Wiedemann-Franz law (ke ¼ LT=
the Lorenz number) from the measured electrical resistivity. L (in the
unit of 10ꢀ8 W ꢀ2) can be calculated from the experimentally
determined Seebeck coefficient (in the unit of
VKꢀ1) through
L ¼ 1:5 þ expðꢀjSj=116Þ [40]. kl can then be assessed by subtracting
ke from . The inset to Fig. 3d presents kl of Mg2Si1-xSbx samples. It is
r, where L is
U
K
References
m
k