Modeling Solvation and Dehydration Mechanisms of Lanthanide Ions Using Machine Learning Potentials

The efficient separation of rare earth elements (REEs) remains a critical challenge for sustainable energy and advanced manufacturing technologies. In this study, we develop machine learning interatomic potentials to model and predict the solvation structure and kinetic behavior of several lanthanide ions in aqueous environments and their dehydration processes with near-quantum accuracy. These ML potentials enable large-scale molecular dynamics simulations revealing binding mechanisms, coordination environments, and free-energy landscapes underlying separation efficiency. Our results provide molecular-level insights that advance understanding and control of REE separation chemistry.