Thermoelectric transport effects beyond single parabolic band and acoustic phonon scattering

Abstract

Thermoelectric materials have been extensively studied for applications in solid-state power generation and cooling. Progress has been made over the past decade in multiple materials systems, hence, it becomes increasingly valuable to be able to analytically model the transport behavior to optimize materials and compare different systems. The well-known effective mass modeling approach is often used to fit the data to the form expected for a single, parabolic band, with charge carrier scattering dominated by acoustic phonons, i.e., deformation potential scattering. However, many high-performance thermoelectric materials benefit from having multiple bands (multi-valley) and many have non-parabolic bands or complex scattering. Understanding how these effects alter properties from that given by the effective mass model provides rational strategies for new materials. In this review, we discuss how this can be done in three scenarios. The first is how to evaluate the influence of point defects on charge carrier mobilities, as well as thermal conductivity. Established methods are available for considering additional scattering mechanisms for phonons and electrons. We focus on how to determine the parameters used in modeling that require the least amount of fitting. We discuss the thermoelectric transport in two different types of materials: lead chalcogenides and half-Heuslers. The second scenario involves systems with multiple sets of conduction or valence bands, which are not necessarily aligned. We discuss different conditions in hypothetical materials systems by considering quality factors for each set of bands. We then demonstrate how the lessons learned are reflected in real thermoelectric materials systems. The third scenario has resonant dopants, and lead chalcogenides have become model systems. These dopants create a distortion in the density of states; inherently, the parabolic dispersion assumption can no longer be used. It is possible nonetheless, to quantitatively undrestand thermoelectric transport properties, providing insights on how to best utilize resonant dopants. Finally, we provide an outlook, identifying limitations and challenges to solve in order to model, and better yet, predict the thermoelectric performance of different materials.