Daniil Poletaev and Artem Oganov, researchers from the Skolkovo Institute of Science and Technology, have made significant strides in the field of crystal structure prediction (CSP), particularly in systems with lightweight atoms like superconducting hydrides. Their work, published in Nature Communications, addresses the challenge of accurately predicting crystal structures at finite temperatures while accounting for quantum anharmonic effects.
The researchers integrated machine-learned interatomic potentials (MLIPs) with the stochastic self-consistent harmonic approximation (SSCHA) to enable evolutionary CSP on the quantum anharmonic free-energy landscape. This approach allows for the prediction of crystal structures under conditions where quantum nuclear motion and anharmonicity play a dominant role.
Using LaH10 at 150 GPa and 300 K as a test case, Poletaev and Oganov compared two methods for SSCHA-based CSP. The first method involved using lightweight active-learning MLIPs (AL-MLIPs) trained on-the-fly from scratch. The second method utilized foundation models or universal MLIPs (uMLIPs) from the Matbench project. The researchers found that AL-MLIPs correctly predicted the experimentally known cubic Fm3m phase as the most stable polymorph at 150 GPa but required corrections within the thermodynamic perturbation theory for consistent results. The uMLIP Mattersim-5m allowed for SSCHA-based CSP without the need for per-structure training and achieved correct structure ranking near the global minimum, though fine-tuning may be needed for higher accuracy.
The study demonstrated that including quantum anharmonicity simplifies the free-energy landscape and is essential for accurate stability rankings. This is particularly important for high-temperature phases that could be missed in classical 0 K CSP. The proposed approach extends the reach of CSP to systems where quantum nuclear motion and anharmonicity dominate, offering valuable insights for the energy sector, particularly in the development of superconducting materials and high-pressure research.
This research was published in Nature Communications, providing a robust framework for future studies in crystal structure prediction under complex conditions. The practical applications for the energy sector include the design and discovery of new materials with desirable properties, such as high-temperature superconductors, which could revolutionize energy transmission and storage technologies.
Source: Nature Communications
This article is based on research available at arXiv.