A Computational Framework for Long-Term Atomistic Analysis of Solute Diffusion in Nanomaterials

Diffusive Molecular Dynamics (DMD) is a class of recently developed computational methods for the simulation of long-term mass transport with a full atomic fidelity. Its basic idea is to couple a discrete kinetic model for the evolution of mass transport process with a non-equilibrium thermodynamics model that governs lattice deformation and supplies the requisite driving forces for kinetics. Compared to previous atomistic models, e.g., accelerated Molecular Dynamics and on-the-fly kinetic Monte Carlo, DMD allows the use of larger time-step sizes and hence has a larger simulation time window for mass transport problems. This dissertation focuses on the development, assessment and application of a DMD computational framework for the long-term, three-dimensional, deformation-diffusion coupled analysis of solute mass transport in nanomaterials. First, a computational framework is presented, which consists mainly of: (1) a computational model for interstitial solute diffusion, which couples a nonlinear optimization problem with a first-order nonlinear ordinary differential equation; (2) two numerical methods, i.e., mean field approximation and subcycling time integration, for accelerating DMD simulations; and (3) a high-performance computational solver, which is parallelized based on Message Passing Interface (MPI) and the PETSc/TAO library for large-scale simulations. Next, the computational framework is validated and assessed in two groups of numerical experiments that simulate hydrogen mass transport in palladium. Specifically, the framework is validated against a classical lattice random walk model. Its capability to capture the atomic details in nanomaterials over a long diffusive time scale is also demonstrated. In these experiments, the effects of the proposed numerical methods on solution accuracy and computation time are assessed quantitatively. Finally, the computational framework is employed to investigate the long-term hydrogen absorption into palladium nanoparticles with different sizes and shapes. Several significant findings are shown, including the propagation of an atomistically sharp phase boundary, the dynamics of solute-induced lattice deformation and stacking faults, and the effect of lattice crystallinity on absorption rate. Specifically, the two-way interaction between phase boundary propagation and stacking fault dynamics is noteworthy. The effects of particle size and shape on both hydrogen absorption and lattice deformation are also discussed in detail.