Abstract

Thermal management is critical for a wide range of technological applications, from microelectronics to renewable energy harvesting and storage. In the design of complex materials for such applications, it is essential to predict their thermal conductivity alongside other functional properties. In this talk, I will illustrate complementary approaches to calculating the thermal conductivity of materials from atomistic simulations, combining lattice dynamics and molecular dynamics. I will demonstrate that the emergence of machine learning models trained on first-principles calculations and the development of a unified theory of heat transport for crystals and glasses allow one to perform, on one side, high-throughput explorations of the thermal conductivity across a broad chemical space, and, on the other side, very accurate and predictive calculations of heat transport for complex systems, such nanostructures, glasses, and nanostructured alloys.