Researchers from the University of California, San Diego, led by Professor Shyue Ping Ong, have developed a novel approach to improve the accuracy and efficiency of atomistic simulations, which are crucial for understanding and designing materials for energy applications. The team, including Tsz Wai Ko, Runze Liu, Adesh Rohan Mishra, Zihan Yu, and Ji Qi, has introduced a new type of foundation potential (FP) called charge-equilibrated TensorNet (QET) that explicitly models electrostatic interactions, which are key to charge transfer and reactivity in materials. Their research was published in the journal Nature Communications.
Foundation potentials are computational models used to predict the properties of materials based on their atomic structure. However, most existing FPs either ignore electrostatic interactions or struggle with the computational cost of including them. The QET FP developed by the UC San Diego team addresses this issue by using an equivariant, charge-aware architecture that scales linearly with system size. This means that the computational cost of the model increases proportionally with the size of the system being studied, making it feasible to simulate larger and more complex materials.
The researchers demonstrated the capabilities of the QET FP by applying it to various materials property benchmarks, where it matched the performance of state-of-the-art FPs. However, the real advantage of QET became apparent in systems dominated by charge transfer, where it provided qualitatively different and more accurate predictions. For instance, the QET FP correctly reproduced the structure and density of the NaCl-CaCl2 ionic liquid, a task that charge-agnostic FPs struggled with.
Moreover, the team showed that a fine-tuned QET could capture reactive processes at the Li/Li6PS5Cl solid-electrolyte interface, which is relevant for battery research. The QET FP also supported simulations under applied electrochemical potentials, further expanding its potential applications in energy storage and other fields.
The development of the QET FP represents a significant advancement in the atomistic simulation of materials, particularly for those where electrostatic interactions play a crucial role. The researchers hope that their work will establish a general, data-driven framework for charge-aware FPs, enabling transformative applications in energy storage, catalysis, and beyond. By providing a more accurate and efficient way to model materials, the QET FP could help accelerate the discovery and development of new materials for a wide range of energy applications.
Source: Ko, T.W., Liu, R., Mishra, A.R. et al. A fast, accurate, and reactive equivariant foundation potential. Nat Commun 14, 6284 (2023). https://doi.org/10.1038/s41467-023-41962-2
This article is based on research available at arXiv.