Physics-Guided Descriptors for Prediction of Structural Polymorphs

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 BiFeO 3 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 BiFeO 3 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.

Prediction of the hardest BiFeO3 from first-principles calculations

“…Besides, we calculated and analyzed the elastic properties of the predicted structures with lower energy and the structures reported recently. 19,47 The results are shown in Table SIII and SIV (ESI†). Based on the predicted c ij 's, the corresponding bulk modulus and shear modulus can be calculated in terms of the Voigt-Reuss-Hill method.…”

Pressure-Induced Phase Diagram and Electronic Structure Evolves during the Insulator–Metal Transition of Bulk BiFeO3

“…Polar mode is used to distinguish between polar and nonpolar phases. Where possible, octahedron tilting modes are represented by the traditional Glazer representation. − In Figure a, we demonstrated seven new low-energy structures through a USPEX search. In Figure b, the main structural distortion modes are shown, with the relative movement of the Bi atoms represented by black arrows.…”