Learning from metastable grain boundaries

TL;DR

This paper uses a large database of grain boundary structures and physics-inspired descriptors to predict properties and understand atomic environments, advancing data-driven modeling of polycrystal behavior.

Contribution

It introduces a robust, physics-based data-driven approach combining atomistic data and descriptors to predict and analyze grain boundary properties.

Findings

Regression models predict GB properties with only 19 descriptors.

Unsupervised methods reveal atomic environment roles in GB properties.

Method generalizes to nanocrystal environments.

Abstract

Grain boundaries (GBs) govern critical properties of polycrystals. Although significant advancements have been made in characterizing minimum energy GBs, real GBs are seldom found in such states, making it challenging to establish structure-property relationships. This diversity of atomic arrangements in metastable states motivates using data-driven methods to establish these relationships. In this study, we utilize a vast atomistic database (~5000) of minimum energy and metastable states of symmetric tilt copper GBs, combined with physically-motivated local atomic environment (LAE) descriptors (Strain Functional Descriptors, SFDs) to predict GB properties. Our regression models exhibit robust predictive capabilities using only 19 descriptors, generalizing to atomic environments in nanocrystals. A significant highlight of our work is integration of an unsupervised method with SFDs to elucidate LAEs at GBs and their role in determining properties. Our research underscores the role of a physics-based representation of LAEs and efficacy of data-driven methods in establishing GB structure-property relationships.