Finding Unprecedentedly Low-Thermal-Conductivity Half-Heusler Semiconductors via High-Throughput Materials Modeling

Abstract

Experimentally determining the lattice thermal conductivity of materials with very high or low values is expensive and time consuming. An efficient computational approach using machine-learning techniques finds a much larger range of conductivity than expected for an impressive number of half-Heusler compounds and also offers a way to rapidly evaluate other classes of materials.