Physics-guided descriptors for prediction of structural polymorphs

TL;DR

This paper introduces a physics-guided machine learning approach combined with DFT calculations to efficiently predict low-energy polymorphs in materials, validated on BiFeO3, discovering new stable phases.

Contribution

The method uniquely integrates structural distortion modes as descriptors in ML models, enabling systematic exploration and prediction of low-energy polymorphs.

Findings

Successfully rediscovered known metastable phases of BiFeO3.

Identified 21 new low-energy polymorphs.

Validated the approach's efficiency and accuracy.

Abstract

We develop a method combining machine learning (ML) and density functional theory (DFT) to predict low-energy polymorphs by introducing physics-guided descriptors based on structural distortion modes. We systematically generate crystal structures utilizing the distortion modes and compute their energies with single-point DFT calculations. We then train a ML model to identify low-energy configurations on the material's high-dimensional potential energy surface. Here, we use BiFeO3 as a case study and explore its phase space by tuning the amplitudes of linear combinations of a finite set of distinct distortion modes. Our procedure is validated by rediscovering several known metastable phases of BiFeO3 with complex crystal structures, and its efficiency is proved by identifying 21 new low-energy polymorphs. This approach proposes a new avenue toward accelerating the prediction of low-energy polymorphs in solid-state materials.