Controllable Preparation And Property Optimization Of Mg₂(Si,Sn) Based Thermoelectric Materials

Mg2(Si,Sn) based thermoelectric (TE) material is being paid more and more attention for the low cost, non-toxicity and good TE performance. However, the oxidation and volatilization of Mg and the big discrepancy between the melting temperatures of the raw materials make the preparation difficult. And many investigations were inhibited for this reason. We employed two new preparation methods-B2O3 flux method and Ta-tube method. Which is effective to suppress the oxidation and volatilization of Mg, and hence control the composition. Using the methods, we systematically investigated the key factors that impact the electrical performance. Samples with low thermal conductivity were prepared using isoelectronic substitution. The effect of cooling rates on the phase constitution and TE performance was investigated. The main results are summarized as follows:(1) It was found that insufficient Mg excess results in Mg vacancy in doped Mg2(Si, Sn) solid solution. Mg2Sio.59-χxSbχn0.41(χ=0,0.005,0.0075,0.01) and Mg2Sio.58Sn0.42-χBiχ(0≤χ≤0.015) samples were prepared by melting in Ta-tube method. The Mg excess is 1.5%. Hall measurements show that the carrier concentration is obviously smaller than the calculated one assuming that all the carriers come from the dopants, indicative of Mg vacancy. Which significantly reduces the electron concentration, mobility and electrical conductivity. A ZTmax of 0.65 at 700 K was obtained for x=0.015 in Mg2Sio.58Sno.42-χBiχ(0≤χ≤0.015) samples.(2) It was found that interstitial Mg obviously enhances the carrier concentration and TE properties of Sb-doped Mg2(Si, Sn) solid solutions. Mg2(i+χ)Sio.3gSno.6Sbo.o2 (0.05≤χ≤0.12) samples were prepared by B2O3 flux method. The presence of interstitial Mg was corroborated by X-ray powder diffraction, X-ray photoelectron spectroscopy, Hall coefficient measurement, and compositional analysis. The electrical conductivity, Seebeck coefficient, and thermal conductivity of Mg2(i+ x)Sio.38Sno.6Sbo.o2 (0.05≤χ≤0.12) as a function of Mg excess were studied between 300 K and 730 K. The electrical properties havebeen analyzed in the framework of a single parabolic band model to gain more insight on the roles of Mg interstitials. We found that increasing Mg excess content increased the carrier concentration, electronic effective mass, and electrical conductivity, while it decreased the Seebeck coefficient and led to a non-monotonic change in the lattice thermal conductivity. As a result, a maximum ZT-0.85 was attained at 700 K for Mg excess x= 0.1. a-60% enhancement compared to that of the sample x= 0.05.(3) It was found that the lattice thermal conductivity of Mgo(Si, Sn) can be effectively reduced by isoelectronically replacing Si4"(Sn4") with Ga5- and Sb3'. Mg2Si0.5Sn0.5-xGaSb(0≤χ≤0.15) and Mg2Si0.8Sn0.2-xlnSb (0≤χ≤0.15) samples were prepared. The measurements show that the lattice thermal conductivity is proportional to1/T, indicating that the U process dominates the phonon transport. With increasing GaSb content, the lattice thermal conductivity decreases.(4)The lattice thermal conductivity of quaternary Mg2Si0.333Ge0.333Sno.333 solid solution is lower than that of Mg2(Si, Sn) ternary solid solution. We prepared Sb-doped Mg2Si0.333Ge0.333Sn0.333 solid solutions using Ta-tube method. And the effect of Mg excess on the TE property of Mg2Si0.333Ge0.333Sn0.333 solid solution was investigated. It was found that the lattice thermal conductivity is about 30%lower than that of Mg2Si0.5Sn0.5 ternary solid solution. It increases with increasing Mg excess. Which is in good agreement with the calculated one using Abeles model.(5) It was found that with decreasing cooling rates, the content of second phase increases and TE performance decreases. The effect of cooling rates on phase constitution, microstructure and TE property of Mg2SixSn1-χis investigated. The cooling rates include liquid nitrogen cooling and furnace cooling. XRD patterns show that it consists of two solid solution phase. The fraction of the second phase for furnace-cooled samples is obviously smaller than the quenched ones. The cooling rates have no obvious effect on the lattice thermal conductivity.