ANN ARBOR, MI
In a new review published in the Journal of Energy Chemistry, Toyota Research Institute of North America (TRINA) researchers Dr. Qian Chen, Dr. Siwen Wang, and Dr. Chen Ling present a groundbreaking perspective on how machine learning is revolutionizing the simulation of all-solid-state batteries (ASSBs). These batteries promise improved safety, higher energy density, and enhanced cycle life, but their development is slowed by the complexity of modeling chemical and interfacial phenomena at realistic scales.
Traditional methods such as density functional theory (DFT) and classical force fields have long been used to simulate atomic interactions in battery materials. However, these approaches struggle to capture the disorder, dynamics, and large system sizes involved in real ASSB environments. The authors argue that machine learning interatomic potentials (MLIPs) overcome these limitations by offering near-DFT accuracy with much greater efficiency. This enables long-timescale, large-scale, and high-throughput simulations—unlocking new insight into ion transport and interface evolution.
The work provides a comparative analysis of the three modeling strategies and highlight how MLIPs are uniquely suited to meet the demands of modern battery research. Their work lays the foundation for more predictive and scalable simulation frameworks, offering a powerful toolset for accelerating the discovery and optimization of solid-state battery materials
Please see the complete work published in Journal of Energy Chemistry
