An Accelerated Approach to Learn the Solute-Solute Interaction Spectra in Grain Boundaries

Recent progress in understanding the thermodynamic stabilization of nanocrystalline alloys via grain boundary solute segregation has revealed the need for spectral information which captures the full distribution of grain boundary environments. Our recently developed accelerated framework, which predicts segregation energy spectra from the local atomic environments at the grain boundary, has made spectral information accessible even with ab-initio methods. However, this framework has only been applied in the dilute limit, and does not yet account for the spectral nature of solute-solute interactions. In this talk, we review our recent development of an atomistic, physically motivated method to measure the full spectrum of grain boundary solute interactions, and discuss a case study on the results of this method for a general polycrystal of Al(Mg). Additionally, we present a modified version of our accelerated framework, which uses the Smooth Overlap of Atomic Positions (SOAP) to construct descriptors for specific nearest neighbor pairs, and demonstrate its use for multiple binary alloys, including Al(Mg), Pt(Au), and Ag(Cu), using existing interatomic potentials.