ML-Driven Discovery of Metastable States

Metastable states and their minimum energy pathways (MEPs) are central to understanding transformations and phase stability in complex materials, yet exploring them in a wide pressure-temperature space remains computationally demanding and experimentally challenging. While machine learning (ML) offers acceleration beyond conventional density functional theory (DFT), progress requires benchmarks that capture real-world complexity and enable robust method comparisons. Here, we advance the solid-state nudged elastic band (SSNEB) approach by integrating modern ML with DFT for energy, force, and stress evaluations, achieving a 2-3-fold speedup while converging to the same pathways predicted by first-principles calculations. This framework allows systematic benchmarking, providing both efficiency and reliability in predicting MEPs for diverse solid-state material systems.