Molecular Dynamics Simulations of Heat Conduction in Nanostructures: Effect of Heat Bath

TL;DR

This study systematically examines how different heat bath models in molecular dynamics simulations affect heat conduction results in nanostructures, highlighting the importance of proper heat bath selection for accurate thermal property predictions.

Contribution

It provides guidelines on choosing heat bath parameters and models to ensure realistic simulation of heat conduction in nanostructures, especially for thermal rectification studies.

Findings

Multiple Nosé-Hoover layers reduce boundary temperature jumps.

Langevin heat bath with a single layer suffices for linear temperature profiles.

Langevin bath yields results consistent with experiments across parameter ranges.

Abstract

We investigate systematically the impacts of heat bath used in molecular dynamics simulations on heat conduction in nanostructures exemplified by Silicon Nanowires (SiNWs) and Silicon/Germanium nano junction. It is found that multiple layers of Nos\'e-Hoover heat bath are required to reduce the temperature jump at the boundary, while only a single layer of Langevin heat bath is sufficient to generate a linear temperature profile with small boundary temperature jump. Moreover, an intermediate value of heat bath parameter is recommended for both Nos\'e-Hoover and Langevin heat bath in order to achieve correct temperature profile and thermal conductivity in homogeneous materials. Furthermore, the thermal rectification ratio in Si/Ge thermal diode depends on the choice of Nos\'e-Hoover heat bath parameter remarkably, which may lead to non-physical results. In contrast, Langevin heat bath is recommended because it can produce consistent results with experiment in large heat bath parameter range.