Multiscale Modeling of Grain Boundaries in Magnesium Alloys for Improved Strength and Ductility

Abstract

This dissertation investigates grain boundaries (GBs) in magnesium (Mg) alloys, specifically the effects of solute segregation thermodynamics, GB deformation mechanisms, and migration kinetics on mechanical properties. Mg alloys are attractive for lightweight structural applications in the automotive industry to make energy-efficient vehicles. Some obstacles to their widespread commercial usage stem from their hexagonal-close packed structure, leading to low strength and poor room-temperature formability. Tuning GB properties in Mg alloys can mitigate these issues. This requires a comprehensive understanding of solute effects on GB properties, to fulfill the goal of building accurate predictive alloy design models. In this dissertation, we couple atomistic simulations with statistical and machine learning approaches to systematically study GB properties in Mg alloys. At each step, our atomistic studies are informed by experiments, higher length-scale simulation methods, or classical models for the estimation of average GB behavior that can be useful for realistic predictive alloy design.

This dissertation presents work on GB thermodynamics, deformation behavior, and kinetics, in that order. First, we use molecular dynamics (MD) simulations to sample accurate finite temperature segregation free energies for Y at a few symmetric tilt GB (STGB) sites. A physics-informed surrogate model is constructed to predict the segregation energetics over many GB sites and obtain segregation free energy spectra for a range of temperatures. Using a spectral segregation model, we estimate high- temperature segregated concentrations for Y, which are validated with experimental measurements for micro-scale grains. In the second part, we use realistic solute-segregated structures obtained from the first study to explore the activation of non-basal slip modes at GBs, which can improve ductility and formability. MD simulations for < c>-tilt axis STGBs under compression show twin nucleation at GBs followed by distinct deformation pathways for different Mg alloys. Mg STGBs segregated with Y reveal a new mechanism for pyramidal I < c+a> slip formation from the twin nuclei formed at GBs. On the other hand, Mg-Al STGBs show twin growth and propagation from the initial nuclei at GBs to form {10-11} compression twins. A stochastic mesoscale model is employed to quantitatively show that Y reduces the critical twin nucleation stresses at GBs and consequently also reduces the critical stress for formation of pyramidal I < c+a> slip. In the third and fourth section, we consider GB kinetics in pure Mg and Mg alloys for optimizing gain texture evolution during thermomechanical processing for better mechanical performance. We first perform a survey GB migration kinetics under an applied driving force using MD for a large dataset of STGB structures with [1-100], [1-210], and [0001] tilt axes. GB mobilities are found to have a large anisotropy for different STGBs, but a detailed consideration of GB atomic structure features reveals important correlations that can be exploited to construct predictive models. A simple machine learning model helps us identify important physically-relevant GB structural features and evaluate their utility for robust predictions of GB mobility. General findings from MD simulations are also incorporated into a classical solute drag model to compare the impact of solute segregation thermodynamics versus solute diffusion on GB migration. Lastly, we use semi-empirical models to make estimates of co-segregation energetics in Mg-Ca-Zn and use an extended solute drag model for multicomponent alloys to determine solute drag effects in the ternary alloy.