Atomistic and mesoscale modeling of dislocation mobility

Abstract

Dislocation is a line defect in crystalline materials, and a microscopic carrier of plastic deformation. Because dislocation has both a localized core and a long-range stress field, linking atomistic and meso scales is often the most challenging step in studying its dynamics. This Thesis presents theories and simulations of dislocations in Si and BCC transition metals, with emphasis on the atomistic-mesoscale coupling. Contributions are made in both methods development and mechanistic understanding of dislocation mobility. For atomistic studies of defects embedded in a mesoscale surrounding, we have given rigorous treatments of two types of boundary effects. A method is derived for quantifying artificial image energies in dislocation simulations with a periodic cell, in which a longstanding conditional convergence problem in lattice summation is resolved. We have also developed a systematic approach based on the linear response theory, which minimizes boundary wave reflections in molecular dynamics simulations without artificial damping. When predictive models are confronted with experiments at the level of mesoscale kinetics, the challenge is to properly incorporate atomistic details into a coarse-grained simulation.

(cont.) We have investigated dislocation core and kink mechanisms and obtained deeper understandings on the shuffle-glide controversy in Si and edge versus screw dislocations in BCC Mo, with some of these breakthroughs related to a better control of artificial boundary effects. The atomistic-mesoscale coupling is then manifested in our formulation of a kinetic Monte Carlo description of dislocation glide in Si at the mesoscale, based on kink mechanisms. As a result, the nature of "weak obstacles" to kink propagation, a long-standing postulate for interpreting low stress dislocation mobility data, is clarified. This model is then generalized to incorporate cross slip for modeling screw dislocation motion in a BCC lattice. Lastly, a physically-motivated procedure is derived for removing the stress singularity in mesoscale dislocation dynamics simulations.