Solute transport in magnesium and their uncertainty quantification using density functional theory and green function approach

Predictive control of alloy processing requires an accurate knowledge of the thermodynamic and the kinetic information of the system for microstructure simulation. Phenomena such as solute segregation or growth of precipitates occur regularly during alloy processing which involves transport or diffusion of solutes. We investigate interstitial and vacancy-mediated solute transport in the hexagonal close-packed Mg. We utilize density functional theory calculations to determine the energies of interstitials and solute-vacancy configurations, which inform our diffusion model. The diffusion of light elemental solutes B, C, N, and O is investigated by determining their stable interstitial sites and the interpenetrating network formed by these sites. We employ the elastodiffusion tensor to determine the effect of strains on diffusion and find that B, C, and N diffusivity increases with volumetric crystal expansion, while O diffusivity decreases. The vacancy-mediated solute diffusion requires the jump network of vacancy near and away from the solute but the existing diffusion models oversimplify this jump network, severely affecting the accuracy of the transport coefficients. We identify all the symmetry-unique vacancy jumps in the Mg lattice and use our Green function approach to generate the transport database for 61 solutes. Our predictions of solute diffusion coefficients agree well with the available experimental measurements. We also study drag ratios which quantify the drag of solutes by vacancies, and the ring network topologies elucidate their mechanisms. We develop a Bayesian framework to quantify uncertainties in transport coefficients and use it to study uncertainties in transport coefficients due to approximate treatment of electronic exchange and correlation in DFT computed energies