Work

SSCHA-based evolutionary crystal structure prediction at finite temperatures with account for quantum nuclear motion

Year: 2026

Type: article

Abstract: Reliable crystal structure prediction (CSP) at finite temperatures, including quantum anharmonic effects, remains challenging but is particularly important for systems containing light atoms such as superconducting hydrides. Here, we integrate machine-learned interatomic potentials (MLIPs) with the stochastic self-consistent harmonic approximation (SSCHA) to enable evolutionary CSP on the quantum anharmonic free-energy landscape. Using LaH 10 as a test case, we compare three approaches: (i) active-learning MLIPs (AL-MLIPs) trained on the fly, (ii) universal MLIPs (uMLIPs), and (iii) temperature-dependent effective potentials (TDEPs) trained on SSCHA ensemble data. AL-MLIPs correctly predict the cubic Fm $$\bar{3}$$ 3 ¯ m phase but require thermodynamic perturbation theory corrections for consistent results. The foundation uMLIP Mattersim-5m enables SSCHA-based CSP without per-structure training, but fine-tuning is required for higher accuracy. Temperature-dependent effective potentials trained on SSCHA data enable accurate and very efficient CSP for large unit cells on the quantum anharmonic free-energy landscape. We demonstrate that quantum anharmonicity simplifies the free-energy landscape and is essential for the correct energy ranking of LaH 10 structures.

Source: npj Computational Materials

DOI: 10.1038/s41524-026-02333-1

Authors: Daniil Poletaev, Artem R. Oganov

Institution: Skolkovo Institute of Science and Technology

Language: English

Topic: Machine Learning in Materials Science

Subfield: Materials Chemistry

Field: Materials Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $3,490

Funder: Russian Science Foundation