No abstract

Kinetic Monte Carlo (KMC) methods have the potential to extend the accessible timescales of off-lattice atomistic simulations beyond the limits of molecular dynamics by making use of transition state theory and parallelization. However, it is a challenge to identify a complete catalog of events accessible to an off-lattice system in order to accurately calculate the residence time for KMC. Here we describe possible approaches to some of the key steps needed to address this problem. These include methods to compare and distinguish individual kinetic events, to deterministically search an energy landscape, and to define local atomic environments. When applied to the ground state  ∑5(2 1 0) grain boundary in copper, these methods achieve a converged residence time, accounting for the full set of kinetically relevant events for this off-lattice system, with calculable uncertainty.

Abstract

The solute diffusion of tungsten at low concentrations in chromium has been investigated both by experiments and computational methods. From finite-source diffusion experiments measured with an Electron Probe Micro Analyzer at temperatures from 1526−1676 K, it was found that the diffusivity of tungsten in chromium follows the Arrhenius relationship D = 0 exp(-Q/RT), where the activation energy was found to be Q = 386 ± 33 kJ/mol. Diffusion of tungsten in chromium was investigated computationally with both the activation-relaxation technique (ART) and molecular dynamics (MD) using a hybrid potential. From ART, the effective diffusion activation energy was determined to be = 315 ± 20 kJ/mol based on a multi-frequency model for a monovacancy mechanism. From MD, the square displacement of tungsten was analyzed at temperatures between 1200 and 1700 K, and the diffusion activation energy was determined to be = 310 ± 18 kJ/mol. In spite of possible complications arising due to experimental compositions away from the dilute limit, the agreement between experiments and simulations falls within the calculated uncertainties, supporting a monovacancy mechanism for diffusion of tungsten in chromium.

We have implemented a modified kinetic Monte Carlo method which achieves a demonstrated convergence in residence time to study grain boundary kinetics of the Σ5 (210) GB in copper at temperatures in the range 200–1173 K. We have observed two regimes of kinetic behavior: a high temperature regime in which low energy dynamics of the disordered GB dominate and a low temperature regime in which high energy events required to escape from the ground state configuration dominate. Whereas our high temperature simulations result in kinetic parameters that are consistent with values from molecular dynamics (MD) simulations, the kinetic parameters from our low temperature simulations (for which simulations reached cumulative real times of approximately 50 d) are quite different and align more closely with experimental literature results. We attribute this difference to the kinetic improbability that MD would access the relevant kinetic events at low temperatures. The methods used in this work can be extended to simulate other properties, grain boundaries, and materials under experimentally relevant conditions.