A novel method to predict the thermal conductivity of nanoporous materials from atomistic simulations

Abstract The low density (smaller than 10% of the bulk density) and the nanostructured porosity of silica aerogels provide their extremely low thermal conductivities but also impact their poor mechanical properties. Atomic scale simulation is the appropriate tool to predict the thermal and mechanical properties of such materials. For such simulations, the interatomic potential should be carefully chosen to ensure result validity but also reasonable computational times. A truncated BKS potential has been used for aerogels as it fairly reproduces the nanostructure. It allows reducing the computational time by a 3000 gain factor on the CPU time per atom per step compared to the original BKS interatomic potential while predicting correctly the mechanical properties. However, when it comes to skeletal thermal conductivity of nanoporous silica, the associated computation times are too large for a representative volume. This is due to the low thermal diffusivity of the material. Here, a new method that takes advantage of the amorphous structure of silica and the diffusive nature of phonon heat transfer at the scale of an aerogel aggregate is proposed. The time dependent temperature profile in the system obtained from Non-Equilibrium Molecular Dynamics simulations is compared to the classical solution of the thermal diffusion equation and an identification procedure is used to determine the thermal conductivity of silica aerogels.