Atomistic Simulations of Nanoscale Transport Phenomena

Molecular simulations have, for many years, been used to investigate transport phenomena, with a goal of validating theories of transport and predicting material properties. Although Equilibrium Molecular Dynamics (EMD) simulations can be utilized for these purposes, they are somewhat limited due to noise issues. Previous Non-Equilibrium MD (NEMD) methods have improved the quality of transport coecients obtained from simulation results, but their application is still limited to homogeneous materials. In this dissertation, I present two algorithms, the “Non-Isotropic Velocity Scaling” (NIVS) and “Velocity Shearing and Scaling” (VSS) approaches that improve the applicability of the NEMD methods. The implementation of these algorithms in MD simulations will be discussed in detail. Applications of these methods in homogeneous fluid and solid systems as well as complex metal-liquid interfaces with surface capping agents follows, along with simulation results for thermal transport and interfacial shearing. Our results demonstrate that these algorithms can provide reasonable estimates of transport coefficients as well as retaining desirable features of previous NEMD methods, while fixing defects in these previous methods. Furthermore, with the VSS algorithm, an efficient evaluation of shear viscosity over a wide range of temperatures can be obtained with a single simulation. We expect these methods can be used for more advanced applications in complex systems such as coupled transport properties.