A round robin test aiming at measuring the high-temperature thermoelectric properties was carried out by a group of European (mainly French) laboratories (labs). Polycrystalline skutterudite Co0.97Ni0.03Sb3 was characterized by Seebeck coefficient (8 labs), electrical resistivity (9 labs), thermal diffusivity (6 labs), mass volume density (6 labs), and specific heat (6 labs) measurements. These data were statistically processed to determine the uncertainty on all these measured quantities as a function of temperature and combined to obtain an overall uncertainty on the thermal conductivity (product of thermal diffusivity by density and by specific heat) and on the thermoelectric figure of merit ZT. An increase with temperature of all these uncertainties is observed, in agreement with growing difficulties to measure these quantities when temperature increases. The uncertainties on the electrical resistivity and thermal diffusivity are most likely dominated by the uncertainty on the sample dimensions. The temperature-averaged (300-700 K) relative standard uncertainties at the confidence level of 68% amount to 6%, 8%, 11%, and 19% for the Seebeck coefficient, electrical resistivity, thermal conductivity, and figure of merit ZT, respectively. Thermal conductivity measurements appear as the least accurate. The moderate value of the temperature-averaged relative expanded (confidence level of 95%) uncertainty of 17% on the mean of ZT is essential in establishing Co0.97Ni0.03Sb3 as a high temperature standard n-type thermoelectric material.

We theoretically investigate the thermoelectric properties of sintered SiGe alloys, compare them with new and previous experimental measurements, and evaluate their potential for further improvement. The theoretical approach is validated by extensive comparison of predicted bulk mobility, thermopower, and thermal conductivity, for varying Ge and doping concentrations, in the 300–1000K temperature range. The effect of grain boundaries is then included for Si0.8Ge0.2 sintered nanopowders and used to predict optimized values of the thermoelectric figure of merit at different grain sizes. Our calculations suggest that further optimization of current state of the art n-type (p-type) material would be feasible, possibly leading to ∼5% (4%) ZT enhancement at 1000 K and 16% (6%) at room temperature. Even larger enhancements should be possible if the phonon scattering probability of the grain boundaries could be increased beyond its present value.

No abstract