Mechanisms of thermal conduction at grain boundaries and in nanocrystalline materials by computational and machine learning techniques

Abstract

Lattice thermal conductivity is an important property that determines the heat insulation and dissipation of materials. Recently, to reduce the lattice thermal conductivity, nanocrystalline materials have been extensively fabricated in the field of thermoelectrics by utilizing the thermal resistance of grain boundaries. However, the suppression mechanism of thermal conduction in nanocrystalline materials remains to be elucidated. For example, the correlation between grain boundary structure and thermal conductivity, the decrease in intragrain thermal conductivity in nanocrystals, and the effect of grain shape on thermal conductivity are still unclear. This paper introduces two cases that have clarified these issues from a microscopic point of view, using atomic-level computational methods and machine learning techniques such as structure descriptors.