A transferable machine-learning framework linking interstice distribution and plastic heterogeneity in metallic glasses

TL;DR

This paper introduces a machine learning framework that predicts atomic plasticity in metallic glasses based on structural features, demonstrating high transferability across different compositions and systems.

Contribution

The study develops a transferable ML-based model linking interstice distributions to plastic heterogeneity in metallic glasses using only quenched structural data.

Findings

The ML model accurately predicts plastic sites in Cu-Zr alloys.

The model generalizes well to other MG systems like Ni62Nb38, Al90Sm10, and Fe80P20.

Predictive power decreases only when shear bands form.

Abstract

When metallic glasses (MGs) are subjected to mechanical loads, the plastic response of atoms is non-uniform. However, the extent and manner in which atomic environment signatures present in the undeformed structure determine this plastic heterogeneity remain elusive. Here, we demonstrate that novel site environment features that characterize interstice distributions around atoms combined with machine learning (ML) can reliably identify plastic sites in several Cu-Zr compositions. Using only quenched structural information as input, the ML-based plastic probability estimates ("quench-in softness" metric) can identify plastic sites that could activate at high strains, losing predictive power only upon the formation of shear bands. Moreover, we reveal that a quench-in softness model trained on a single composition and quenching rate substantially improves upon previous models in generalizing to different compositions and completely different MG systems (Ni62Nb38, Al90Sm10 and Fe80P20). Our work presents a general, data-centric framework that could potentially be used to address the structural origin of any site-specific property in MGs.