Multiscale Thermal Transport Across Solid-Solid Interfaces

Abstract

We propose a novel multiscale model in order to better understand thermal transport across solid-solid interfaces in a mesoscale system. While Molecular Dynamics (MD) simulations tend to be very accurate, they are also computationally rather expensive. Continuum simulations such as Symmetric Smoothed Particle Hydrodynamics (SSPH), cannot take temperature discontinuities that may occur across interfaces into account, which can cause erroneous results. As such, we develop a multiscale model in which we run MD simulations over the region containing the interface, while running SSPH simulations over the remainder of the domain. This drastically reduces the number of molecules simulated by MD, reducing computational time, while hopefully still maintaining the accuracy provided by a “pure” MD run. Results from the simulation indicate that when boundary temperatures are specified, the data from the multiscale model is highly similar to the data from the pure MD run. However, when boundary fluxes are specified, the multiscale model tends to predict higher temperatures than does MD. We believe that this may be due to continuum SSPH simulations being unable to take into account phonon scattering with non-periodic boundary conditions.