Thermoelectric Performance Predictions Combining Experiments with Multi-Band Modelling

Abstract

The search for high-performance thermoelectric (TE) materials requires accurate property predictions and the development of analytical models to mimic the temperature dependent charge and heat transport in semiconductors. This is a non-trivial task as most thermoelectric materials have complex electronic band structures with multiple bands contributing to charge transport. In this work, it is shown that using a combination of experiments and a recently developed multi-band modelling technique, it is possible to accurately predict the temperature and doping dependent properties of TE materials. The steps involved are experimental data collection, model parameter generation, and data interpolation. The methodology is elaborated using the example of Mg₂Si_0.3Sn_0.7 which is a high-performance, low-cost thermoelectric material. 3-D maps of power factor and thermoelectric figure of merit ($zT$) are generated as a function of temperature and doping concentration. Model validation is carried out for a randomly prepared composition which yields maximum deviations of ±10% in the $zT$ and the power factor plots. The results highlight the potential of this tool for rapid screening of high-performance compositions.