Solid-State Synthesis and Thermoelectric Properties of Sb-Doped Mg2Si Materials

Solid-State Synthesis and Thermoelectric Propertiesof Sb-Doped Mg2Si Materials

M. IOANNOU,1 G. POLYMERIS,2 E. HATZIKRANIOTIS,2 A. U. KHAN,1

K. M. PARASKEVOPOULOS,2 and TH. KYRATSI1,3

1.—Department of Mechanical and Manufacturing Engineering, University of Cyprus, 1678Nicosia, Cyprus. 2.—Department of Physics, Aristotle University of Thessaloniki, 54124 Thessa-loniki, Greece. 3.—e-mail: [email protected]

Sb-doped magnesium silicide compounds have been prepared through ballmilling and solid-state reaction. Materials produced were near-stoichiometric.The structural modifications have been studied with powder x-ray diffraction.Highly dense pellets of Mg2Si1�xSbx (0 £ x £ 0.04) were fabricated via hotpressing and studied in terms of Seebeck coefficient, electrical and thermalconductivity, and free carrier concentration as a function of Sb concentration.Their thermoelectric performance in the high temperature range is presented,and the maximum value of the dimensionless figure of merit was found to be0.46 at 810 K, for the Mg2Si0.915Sb0.015 member.

Key words: Solid-state reaction, ball milling, antimony doping homogeneity

INTRODUCTION

Silicide compounds seem to be an advantageouschoice for thermoelectrics, not only because of theample availability of their constituent elements innature but also because of their nontoxicity, whichis consistent with the priority for technology that issafe for both the environment and humans. Mg2Si-based compounds were found to have a favorablecombination of physical and chemical propertiesand can form a good basis for development of newefficient thermoelectric devices.1 Mg2Si compoundhas already high ZT, of about 1, and its low densityencourages the development of economical systems.

Regarding synthesis, it is difficult to prepareMg2Si by a melting process due to the large differ-ence in vapor pressures of the constituent elementsand their insolubility. Furthermore, it is difficult tocontrol its composition, mainly due to volatilizationand oxidation of Mg. In general, solid-state tech-niques, such as mechanical alloying or solid-statereactions, are preferable because they are relativelyeasier and better to control, applicable on a largescale, and demand simpler equipment. Mechanicalalloying, which is usually preferred for industrial-scale

applications, is not an effective technique for thismaterial since severe agglomeration during milling isknown to be a major problem. Dry milling is notpractical under various conditions.2 On the otherhand, the wet ball-milling process was found to be auseful tool to fabricate nanocrystalline material usingMg2Si as the starting powder.3

Pure Mg2Si is an indirect-bandgap semiconductor(Eg = 0.77 eV) with less interesting thermoelectricproperties, and for this reason, several either p- orn-type dopants have been proposed. Tani andKido4,5 first reported that Sb can act as a donor forMg2Si. In previous studies, Sb was used as a dopantin either small6 or larger concentrations.7 However,Nolas et al.,7 produced materials with excess Mg,and limited their study to the low temperaturerange, while Jung et al.6 produced material withrather low mobility.

In this work, we study high-mobility near-stoi-chiometric materials, in the high temperaturerange. A combination of ball-milling, solid-statereaction, and hot-pressing techniques is employed,aiming at completion of the reaction of Mg2Si withSb doping and the formation of Mg2Si1�xSbx

(0 £ x £ 0.04) in the form of highly dense pellets.The thermoelectric properties (Seebeck coefficient,electrical conductivity, thermal conductivity, andfigure of merit) are discussed, while the local

(Received July 8, 2012; accepted December 28, 2012;published online February 9, 2013)

Journal of ELECTRONIC MATERIALS, Vol. 42, No. 7, 2013

DOI: 10.1007/s11664-012-2442-6� 2013 TMS

1827

microstructure was examined in comparison withthe local variation in free carriers.

EXPERIMENTAL PROCEDURES

Reagents

Chemicals in this work were used as obtained: (i)magnesium powder [�20 + 100 mesh (= 841 lm to149 lm), 99.8% purity; Alfa Aesar], (ii) silicon pow-der [crystalline, +100 mesh (= 149 lm), 99.9% pur-ity; Alfa Aesar], and (iii) antimony shot [6 mm(0.2 inches) and less, 99.999%]. All manipulationswere carried out under inert gas argon in a dryMBraun glovebox.

Mixing and Milling

The starting materials were mixed inMg2Si1�xSbx (0 £ x £ 0.04) stoichiometry in a stain-less-steel bowl, and ball-milled (Pulverisette 6;Fritsch) using balls of 10 mm diameter. The bowlwas sealed by an O-ring under argon atmosphere.The milling was applied for 60 min and interruptedevery 15 min for 5 min to avoid heating. The speedof the ball-milling process was 300 rpm, and theball-to-material ratio was 23:1. The material wasexamined after the ball-milling process (BMedmaterial).

Heat Treatment

The powders, after milling, were cold pressed intopellets 10 mm in diameter at about 0.5 GPa. Cold-pressing was carried out in order to achieve goodcontact of the grains and enhance the diffusionmechanism in the solid state. Finally, the pelletswere placed inside Mo foil to save them from attackby the hot quartz wall, sealed in silica tubes undervacuum, and heated at temperatures of 400�C and600�C for 1 h. The heating rate was 1�C/min for allsamples. The product of this step was in powderform (SSRed material).

The materials in powder form were, then, sin-tered in order to perform thermoelectric measure-ments. Sintering was carried out using a uniaxialhot-pressing system (HP20) from Thermal Tech-nologies Inc. under argon flow using a graphite dieof 10 mm diameter. The powders were heatedto 860�C and kept at this temperature for 60 minunder pressure of 80 MPa (HPed material). The hot-pressed pellets were highly dense with densitiescorresponding to �98% of the theoretical value forMg2Si1�xSbx. The density of the pellets was esti-mated from mass (m) and volume (V) measurementsusing the equation qMEAS = m/V.

Structural Characterization and ElementalAnalysis

Powder patterns were obtained from all BMed,SSRed, and HPed materials, using a Rigaku Mini-flex powder x-ray diffraction with Ni-filtered Cu Ka

radiation (30 kV, 15 mA) in order to identify thephases and evaluate the purity of the products.

Sample microstructure was examined by a Jeol840A scanning microscope with an energy-disper-sive spectrometer attached (model ISIS 300; Oxford)for energy-dispersive x-ray analysis (EDX). Thebeam spot was 1 lm, the accelerating voltage was20 kV, the beam current was 0.4 nA, the workingdistance was 20 mm, and the counting time was60 s real time. Several measurements were takenon each sample on areas of about 20 lm 9 20 lmand averaged.

Thermoelectric Properties

Seebeck Coefficient and Electrical Conductivity

The measurements were carried out on hot-pres-sed pellets simultaneously using a commercialZEM-3 Seebeck coefficient and electrical resistivitymeasurement system from ULVAC-RIKO. Datawere recorded in the temperature range of roomtemperature to 550�C. The measurements wereperformed under residual pressure of He gas tofacilitate good thermal contact. Estimated uncer-tainties are ±5%.

Thermal Conductivity

A Netzsch LFA-457 system was used to measurethe thermal diffusivity of the hot-pressed pelletswith 10 mm diameter and thickness of �2 mm. Thethermal conductivity was calculated from theexperimental thermal diffusivity, as well as densityvalues and previously reported specific heat capac-ity data of undoped Mg2Si.8 Estimated uncertaintiesare ±5%.

Carrier Concentration

Carrier concentration and sample homogeneitywere estimated from conventional infrared (IR) andl-IR reflectivity measurements. Infrared spectrawere recorded at nearly normal incidence at roomtemperature, with a Bruker 113v Fourier-transformIR (FTIR) spectrometer in the 100 cm�1 to2000 cm�1 range (for conventional IR measure-ments) and with a PerkinElmer spectrometerequipped with an i-series PerkinElmer FTIRmicroscope in the 500 cm�1 to 4000 cm�1 spectralregion (for l-IR). Spectra were recorded at multiplespots on each sample, with an iris of 100 lm diam-eter, to check for sample homogeneity. The reflec-tion coefficient was determined by typical sample in/sample out method with a gold mirror as reference.Measurements were carried out on optically flat,polished hot-pressed pellets.

For selected samples, the carrier concentrationwas estimated by both Hall-effect and IR reflectivityanalysis. Both methods gave similar results interms of carrier concentration, thus all sampleswere further studied using IR reflectivity analysis.The results are presented in Table I.

Ioannou, Polymeris, Hatzikraniotis, Khan, Paraskevopoulos, and Kyratsi1828

RESULTS AND DISCUSSION

Synthesis and Powder x-Ray Diffraction(PXRD)

After Ball Milling

Magnesium, silicon, and antimony powders wereball-milled under argon atmosphere, in order to use

them as pristine material for heat treatments. Ball-milling was limited to 60 min, aiming to avoidMg2Si formation at this step as discussed else-where.9,10

Phase identification after each step was per-formed using PXRD. Figure 1a shows the PXRDpatterns of materials milled for 60 min. It is clearthat only peaks of Mg, Si, and Sb2Mg3 (in smallamount) are identified after ball-milling, whereasthere are no diffraction peaks of the Mg2Si com-pound. Interestingly, the material milled for60 min showed lower intensities for the (101) and(100) diffraction peaks of Mg compared with theintensity of the (002) peak. This behavior has beenalso observed elsewhere,9,10 where preferentialorientation of Mg appeared after 60 min of ball-milling and was attributed to deformation of theMg powder to a flaky shape having the (002) planeparallel to the flake surface.9 The same effect isobserved for Si. The highest peak (220) observed inpure Si (Fig. 1a) is no longer higher in the ball-milled material but has a lower intensity than(111).

Table I. Composition, conduction type, carrier concentration, Seebeck coefficient, electron mobility,and electrical and thermal conductivity at room temperature for Sb-doped Mg2Si (Mg2Si12xSbx;0 £ x £ 0.04) at 300 K

NominalComposition

ConductionType

CarrierConcentration

(cm23)

ElectricalConductivity

(S/cm)

ElectronMobility(cm2/V s)

SeebeckCoefficient

(lV/K)

ThermalConductivity

(W/m K)

Mg2Si n 2.24 9 1018 56.8 158.5 �224.0 11.17Mg2Si0.99Sb0.01 n 1.85 9 1020 1717 58.0 �63.5 7.81Mg2Si0.985Sb0.015 n 2.38 9 1020 2923 76.7 �70.2 7.86Mg2.04Si0.985Sb0.015 n 2.34 9 1020 2583 68.9 �65.2 7.68Mg2.08Si0.985Sb0.015 n 2.45 9 1020 2099 53.5 �63.5 7.98Mg2Si0.98Sb0.02 n 2.48 9 1020 2850 71.8 �62.6 8.53Mg2Si0.97Sb0.03 n 2.78 9 1020 2510 56.5 �52.0 8.55Mg2Si0.96Sb0.04 n 1.85 9 1020 2830 65.7 �51.7 7.13

(b) SSR material

(a) 60min ball milling

Mg2Si

Mg3Sb

2

(112

)

(103

)

(110

)

(102

)

(101

)

(002

)

(100

)

Mg

20 30 40 50 60 70 80

(220

)

(311

)

(111

)

Si

2theta (deg.)

Inte

nsity

Fig. 1. PXRD patterns of 60-min ball-milled material (a) and aftersolid-state reaction at 600�C for 1 h (b). Reference materials areshown for comparison.

0,00 0,01 0,02 0,03 0,04 0,056,350

6,352

6,354

6,356

6,358

6,360

6,362

6,364

Latti

ce P

aram

eter

(A

)

x

Fig. 2. Variation of lattice parameter versus Sb content.

Solid-State Synthesis and Thermoelectric Properties of Sb-Doped Mg2Si Materials 1829

After Heat Treatment

Phase identification in the material after the firststep of heat treatment showed Mg2Si single phase(see Fig. 1b for SSRed material after heating at600�C for 1 h). PXRD patterns after heating(SSRed) as well as after hot-pressing (HPed) wereidentical.

Figure 2 presents the variation of the latticeparameter for Sb-doped Mg2Si. The lattice param-eter increases in proportion to the Sb content. Asthe atomic radius of Sb is larger than that of Si, bysubstituting Sb for Si, the lattice parameter willincrease monotonically, and this is an indication ofsuccessful substitution of Si by Sb.

The heating process also affected the materialsmorphologically, since Mg2Si formation seemed tobe related to the observed disintegration of thepellets into very fine powder. The disintegrationwas observed after the heating process in allexperiments that led to a single Mg2Si phase prod-uct. The reaction of Mg2Si formation has beenrecently reported to follow a self-propagatingmechanism11,12 which is related to the disintegra-tion of the pellets.

Microstructure

SEM and EDS Analysis

In Fig. 3, typical SEM micrographs in thebackscatter mode are shown for the HPed samples.As can be seen, darker areas as well as white spotsappear in a gray background. EDX analysisshowed that the white spots are areas with about6.5% to 7.5% Mg deficiency. The Sb concentra-tion in the dark areas is smaller than in thegray area, resulting in Sb-poorer and Sb-richerphases. The density of the darker areas variesfrom sample to sample, from about 37% for the4% Sb-doped to about 64% for the 1% Sb-dopedsamples.

IR Reflectivity Studies

The main characteristic of the IR spectra is thatthey are dominated by a structureless plasmon,with no clear indication of reststrahlen modes. IRreflectivity, R(x), is expressed through the complexdielectric function e(x) as12

RðxÞ ¼ffiffiffiffiffiffiffiffiffiffi

eðxÞp

� 1ffiffiffiffiffiffiffiffiffiffi

eðxÞp

þ 1

�

�

�

�

�

�

�

�

�

�

2

:

For the spectral analysis we assumed the Drudemodel for the plasmon, which is given by

eðxÞ ¼ e1 �e1x2

P

x2 þ i � cPx;

where e1 is the high-frequency dielectric function,and xP and cP are the plasma frequency anddamping constant, respectively. The plasma fre-quency (xP) is given by

x2PL ¼

n � e2

m�e0e1;

where e0 is the dielectric function of vacuum, m* isthe carrier effective mass, and n is the free carrierconcentration. For Mg2Si, m* is considered equal to0.53m0.7,13 The plasmon frequency was found toincrease with the Sb content, indicating an increasein the free carrier concentration.

Conventional IR reflectivity measurements areusually carried out with an iris of about 2 mm to4 mm in diameter. In l-IR, the iris is reduced to100 lm, which enables mapping of the sample forlocal inhomogeneities. Spectra were recorded atmultiple spots on each sample, as shown in Fig. 4.In the same figure, the local variation of theplasma frequency (xP) for the 1% and 4%Sb-doped samples is also shown as a contour plot.As can be seen, the inhomogeneity in the 1%

Fig. 3. Typical SEM micrographs for 1% Sb-doped (a) and 4% Sb-doped (b) samples.


Sb-doped sample is smaller than in the 4% one;local variation in xP corresponds to local variationin the free carrier concentration. Local inhomoge-

neity results were confirmed from local variationin Sb-richer and Sb-poorer regions, observed inSEM.

-2000

-1000

0

1000Y

( μm

)

X (μm)

1600

1700

1800

1900

2000

2100

2200

(a)

-2000 -1000 0 1000 -4000 -3000 -2000 -1000 0 1000-4000

-3000

-2000

-1000

0

1000

Y (

μm)

X (μm)

1600

1700

1800

1900

2000

2100

2200

(b)

Fig. 4. Optical photograph and contour plots of the plasma frequency (xP) across the sample for the 1% Sb-doped (a) and for the 4% Sb-dopedsample (b).

(a)

300 400 500 600 700 800-260

-240

-220

-200

-180

-160

-140

-120

-100

-80

-60

-40

x=0.000 x=0.010 x=0.015 x=0.020 x=0.030 x=0.040

S (

μ V/K

)

T (K)

(b)

300 400 500 600 700 80010

x=0.000 x=0.010 x=0.015 x=0.020 x=0.030 x=0.040

σ (S

/cm

)

T (K)

(c)

300 400 500 600 700 800 9002

3

4

5

6

7

8

9 x=0.000 x=0.010 x=0.015 y=0.020 x=0.030 x=0.040

κ latti

ce (

W/m

K)

T (K)

(d)

300 400 500 600 700 8000.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

x=0.000 x=0.010 x=0.015 x=0.020 x=0.030 x=0.040

ΖΤ

T (K)

Fig. 5. Temperature dependence of the Seebeck coefficient (a), the electrical conductivity (b), the lattice contribution to the thermal conductivity(c), and ZT (d) for Mg2Si1�xSbx (0 £ x £ 0.04).


Thermoelectric Properties

Table I lists the results of the measurements(mean value of carrier concentration, Seebeck coef-ficient, and electrical and thermal conductivity) at300 K of Mg2Si1�xSbx (0 £ x £ 0.04).

According to Table I, the value of the free carrierconcentration for undoped Mg2Si was found to besignificantly higher than that reported by Junget al.,6 and is closer to that reported by Nolas et al.7

This might result from differences in chemicalpurity and the preparation method used.

The sign of the Seebeck coefficient for all compo-sitions is negative, indicating that electrons are themajor carriers, in good agreement with the litera-ture.6 Room-temperature Seebeck values rangebetween �63 lV/K and �70 lV/K for the Sb-dopedspecimens, in accordance with literature.7 Theelectrical conductivity is quite high for the Sb-dopedsamples, much higher than for pure Mg2Si, and thisis attributed to the significant increase of the freecarrier concentration (due to Sb doping) whichovercompensates the (expected) decrease in carriermobility. In fact, the electron mobility in all (both

pure and doped) samples is much higher than thosereported by Jung et al.6 and slightly higher thatthose reported by Nolas et al.7 The electrical con-ductivity (r) at 300 K increases with Sb doping.Overall, in the composition range of 0 £ x £ 0.04,the member with x = 0.015 shows the highest elec-trical conductivity, having a value of 2923 S/cm.The absolute Seebeck coefficient decreases withincreasing Sb content, due to the electrons that areintroduced as carriers.

Figure 5a shows the Seebeck coefficient of theMg2Si1�xSbx hot-pressed pellets in comparison withthat of undoped Mg2Si. The doped materials havedifferent temperature dependence due to their sig-nificantly higher carrier concentration. The tem-perature dependence of the electrical conductivity ofthe Mg2Si1�xSbx series is shown in Fig. 5b. Theundoped Mg2Si presents semiconducting behavior,in agreement with the intrinsic semiconductingbehavior of Mg2Si discussed elsewhere.5 Intrinsicconduction occurs at temperatures higher than300�C due to the bandgap of 0.77 eV.14 The electri-cal conductivity of the Sb-doped Mg2Si decreases

(a)

300 400 500 600 700 800

-160

-140

-120

-100

-80

-60 x=0.015 2% Mg excess 4% Mg excess

S (

V/K

)

T (K)

(b)

300 400 500 600 700 800100

1000

10000

x=0.015 4% Mg excess 2% Mg excess

(S

/cm

)

T (K)

(c)

300 400 500 600 700 8002

3

4

5

6

7


latti

ce (

W/m

K)

T (K)

(d)

300 400 500 600 700 8000.0

0.1

0.2

0.3

0.4

0.5


T (K)

μ

σ

κ

ΖΤ

Fig. 6. Temperature dependence of the Seebeck coefficient (a), the electrical conductivity (b), the lattice contribution to the thermal conductivity(c), and ZT (d) for Mg2+dSi1�xSbx (x = 0.015).


with temperature, which is typical for materialswith relatively high carrier concentration. In thesecases, the temperature dependence is attributedmainly to the decrease of the mobility, since scat-tering by acoustic phonons is the predominantmechanism.15,16

Figure 5c presents the lattice contribution to thethermal conductivity for Mg2Si1�xSbx (0 £ x £ 0.04)versus temperature. The thermal conductivitydecreased with Sb concentration, mainly due to themass fluctuation that is introduced in the Mg2Silattice. The lattice thermal conductivity was calcu-lated by using the Wiedemann–Franz relation:jL = j � je, where je = LrT with L = 2.45 9 10�8

V2 K�2, using the measured values for thermal (j)and electrical conductivity (r).

The temperature dependence of ZT of Mg2

Si1�xSbx (0 £ x £ 0.04) is shown in Fig. 5d. Thepower factor reaches 25.1 lW/cm K2, and the ZTpresents a maximum of 0.41 at 810 K for themember with nominal composition Mg2Si0.985

Sb0.015. This value is lower than those reported inthe literature,6 also following synthesis with solid-state reaction, mainly due to the lower Seebeckcoefficient.

Mg2Si has 4.7 9 1022 structural units per cm3. Ifthe substitution of Sb for Si only introduces elec-trons, in a rigid-band-type model, the carrier con-centration can be estimated by assuming that oneSb3� substitution for one Si4� generates one elec-tron. Thus, for Sb substituting 4% of Si, the freecarrier concentration is expected to be about6.3 9 1020 cm�3. However, in the case of 4% Sbsubstitution, only 42% is used to increase the carrierconcentration, as shown in Table I. According toWang and Nolas,17 the remaining Sb generatesvacancies at Mg sites in the case of Mg2Si-Mg3Sb2

mix-crystals. The generation of Mg vacancies is inagreement with the white spots (deficiency in Mg)observed in SEM images (Fig. 3). The introductionof the ball-milling step as well as the higher Sbconcentration could result in such vacancies, and asMg vacancies act as acceptors, they could accountfor the discrepancy between the expected and mea-sured values of free carrier concentration.

To compensate the Mg evaporation loss and thusto increase the efficiency of the Mg2Si0.985Sb0.015

member, materials with Mg excess were also stud-ied. The room-temperature values of the thermo-electric properties of Mg2Si0.985Sb0.015 memberprepared with 2% and 4% Mg excess (nominal for-mulas Mg2.04Si0.985Sb0.015 and Mg2.08Si0.985Sb0.015)are presented in Table I. The temperature depen-dence of the Seebeck coefficient and electrical con-ductivity (Fig. 6a, b) present the typical behavior ofmaterials with relatively high carrier concentration.The temperature dependence of the lattice thermalconductivity is the same for all materials, asexpected (Fig. 6c). Overall, the ZT value is higherfor 2% Mg excess, where a value of 0.46 at 810 K is

reached, as shown in Fig. 6d, the highest valuereported in this work.

CONCLUSIONS

Sb-doped Mg2Si (Mg2Si1�xSbx, 0 £ x £ 0.04) com-pounds were prepared by a solid-state reaction andhot pressing, and their thermoelectric and transportproperties were examined. This synthetic routeconsists of two steps: ball-milling aimed at goodmixing and the creation of fine particles to be usedas starting materials, and a heating process totrigger the solid-state reaction. Materials producedwere near-stoichiometric. Free carriers were foundto monotonically increase with the Sb content. Thelocal microstructure was examined and comparedwith the local variation in free carriers. Carrierinhomogeneity was increased at higher Sb content.The local inhomogeneity results from Sb-richer andSb-poorer regions, as observed in SEM.

The electrical conductivity, Seebeck coefficient,and thermal conductivity were strongly affected bythe Sb incorporation. The absolute values of theSeebeck coefficient decreased with increasing Sbcontent, while the electrical conductivity increaseddue to the higher carrier concentration. The ther-mal conductivity decreased with the amount of Sbdue to lattice scattering of phonons. The maximumvalue of the dimensionless figure of merit (ZT) forMg2Si0.915Sb0.015 prepared with 2% Mg excess was0.46 at 810 K.

ACKNOWLEDGEMENTS

The authors thank Dr. E. Pavlidou for the SEM/EDS and Ms. E-C Stefanaki for performing the l-IRmeasurements. This study was supported by theThermoMag Project, which is co-funded by theEuropean Commission in the 7th Framework Pro-gram (Contract NMP4-SL-2011 263207), by theEuropean Space Agency, and by the individualpartner organizations.

REFERENCES

1. E.N. Nikitin, V.G. Bazanov, and V.I. Tarasov, Sov. Phys.Solid State 3, 2648 (1961).

2. M. Riffel and J. Schilz, Scr. Metall. Mater. 32, 1951 (1995).3. M. Ioannou, E. Hatzikraniotis, Ch. Lioutas, Th. Hassapis,

Th. Altantzis, K.M. Paraskevopoulos, and Th. Kyratsi,Powder Technol. 217, 523 (2012).

4. J. Tani and H. Kido, Intermetallics 15, 1202 (2007).5. J. Tani and H. Kido, Phys. B 364, 218 (2005).6. J.-Y. Jung, K.-H. Park, and I.-H. Kim, IOP Conf. Ser.:

Mater. Sci. Eng. 18, 142006 (2011).7. G.S. Nolas, D. Wang, and M. Beekman, Phys. Rev B 76,

235204 (2007).8. M.W. Chase Jr, J. Phys. Chem. Ref. Data 14, 1495 (1985).9. L.Lu,M.O.Lai,andM.L.Hoe,NanoStruct.Mater.10,551 (1998).

10. M. Ioannou, K. Chrissafis, E. Pavlidou, F. Gascoin, and Th.Kyratsi, J. Solid State Chem. 197, 172 (2013).

11. E. Godlewska, K. Mars, R. Mania, and S. Zimowski, Inter-metallics 19, 1983 (2011).

12. F. Gervais and B. Piriou, Phys. Rev. B10, 1642 (1974).


13. M.W. Heller’t and G.C. Danielson, J. Phys. Chem. Solids 23,601 (1962).

14. G. Busch and U. Winkler, Physica 20, 1067 (1954).15. R.J. LaBotz, D.R. Mason, and D.F. O’Kane, J. Electrochem.

Soc. 110, 127 (1963).

16. Y. Noda, H. Kon, Y. Furukawa, N. Otsuka, I.A. Nishida, andK. Masumoto, Mater. Trans. JIM 33, 851 (1992).

17. D. Wang and G.S. Nolas, Advances in Electronic Ceramics,ed. C. Randall, H.-T. Lin, K. Koumoto, and P. Clem (Wes-terville, OH: The American Ceramic Society, 2008), p. 185.


Documents

Solid-State Synthesis and Thermoelectric Properties of Sb-Doped Mg2Si Materials