On Neural Networks for Grain Boundary Dynamics

Nearly all structural and functional materials are polycrystals; they are composed of differently oriented crystalline grains that are joined at grain boundaries (GBs). Such interfaces play a critical role in controlling many engineering and functional properties. Understanding GB physics is, therefore, a key aspect of materials discovery and design efforts. Owing to their local atomic environments, GBs provide preferential sites for alloying elements to occupy. Indeed, the direct manipulation of GB chemical states has been found to influence a host of materials properties and processes, such as cohesion and fracture resistance, transport, electrochemical response, electrical conductivity, and processability during advanced manufacturing techniques. Of particular interest is the impact of GB solute segregation on boundary dynamics, as it influences microstructure formation and evolution pathways during processing treatments or under operating environments. For example, GB segregation has been used to mitigate grain coarsening and thermally stabilize nanograined structures. While GB solute segregation has been the subject of active research, most existing studies focus on the thermodynamics of GB segregation and the kinetic role (i.e., dynamic solute drag) remains unexplored. GB solute drag results when segregated alloying elements exert a resistive force on migrating GBs hindering their motion. The challenge here is that solute drag depends on several properties, such as alloy thermodynamics (e.g., heat of mixing), and dynamic processes including solute diffusion and boundary migration; solute drag is a hypersurface. In this talk, we unravel GB dynamic solute drag through theoretical analysis, mesoscale modeling, and machine learning studies. We establish design maps relating drag effects to relevant bulk alloy and GB properties. We find that solute drag is dominant in immiscible alloys with far-from-dilute compositions in agreement with experimental observations of GB segregation in metallic alloys. Further, asymmetric GB segregation is found to greatly influence solute drag values. On the whole, our work provides future avenues to employ GB segregation engineering to control GB dynamical processes.