Atomistic and Machine Learning Studies of Solute Segregation in Metastable Grain Boundaries

Grain boundary (GB) segregation plays a decisive role in determining the mechanical, thermal, electrical,and corrosion properties of polycrystalline materials. While equilibrium grain boundaries have beenextensively studied, metastable grain boundaries—formed during rapid solidification, severe plasticdeformation, or non-equilibrium processing—remain less understood. These metastable interfaces exhibitcomplex atomic structures and local energy landscapes that strongly influence solute segregation behavior.This study integrates atomistic simulations and machine learning (ML) techniques to investigate solutesegregation in metastable grain boundaries. Atomistic modeling based on molecular dynamics (MD)and Monte Carlo (MC) simulations is used to compute segregation energies, local structural motifs, anddiffusion characteristics. Subsequently, supervised ML models are trained on atomistic datasets to predictsegregation tendencies across diverse boundary types and chemical compositions. The results revealthat metastable boundaries possess heterogeneous segregation sites with wide energy distributions, oftenleading to enhanced segregation compared to equilibrium configurations. ML models demonstrate highpredictive accuracy and significantly reduce computational costs by identifying dominant structuraldescriptors influencing segregation energy. The study highlights the synergistic potential of atomisticmodeling and data-driven approaches for accelerating alloy design and tailoring grain boundary chemistry.