Multiscale Genome Modeling for Predicting the Thermal Conductivity of Silicon Carbide Ceramics

Abstract

Silicon carbide (SiC) ceramics have been widely used in industry due to its high thermal conductivity. Understanding the relations between the microstructure and the thermal conductivity of SiC ceramics is critical for improving the efficiency of heat removal in heat sink applications. In this paper, a multiscale model is proposed to predict the thermal conductivity of SiC ceramics by bridging atomistic simulations and continuum model via a materials genome model. Interatomic potentials are developed using ab initio calculations to achieve more accurate molecular dynamics (MD) simulations. Interfacial thermal conductivities with various additive compositions are predicted by nonequilibrium MD simulations. A homogenized materials genome model with the calculated interfacial thermal properties is used in a continuum model to predict the effective thermal conductivity of SiC ceramics. The effects of grain size, additive compositions, and temperature are also studied. The good agreement found between prediction results and experimental measurements validates the capabilities of the proposed multiscale genome model in understanding and improving the thermal transport characteristics of SiC ceramics.