Discovering high-strength alloys via physics-transfer learning

Abstract

Predicting the strength of materials requires knowledge over multiple scales, striking a balance between accuracy and efficiency. Peierls stress measures material strength by evaluating dislocation resistance to plastic flow, reliant on elastic lattice responses and the crystal slip energy landscape. Computational challenges due to the non-local and non-equilibrium nature of dislocations prohibit Peierls stress evaluation from state-of-the-art material databases. We propose a physics-transfer learning framework that leverages neural networks trained on force-field simulations to understand crystal plasticity physics, predicting Peierls stress from material parameters derived via first-principles calculations, which are otherwise computationally intractable for direct dislocation modeling. This physics-transfer approach successfully screens strengths of metallic alloys from a limited number of single-point calculations with chemical accuracy. Guided by the prediction, we fabricate high-strength binary alloys previously unexplored via high-throughput ion-beam deposition. The framework solves problems facing the accuracy-performance dilemma by harnessing multi-scale physics in materials sciences.