Understanding Metal Ion Interactions in Solvents Using First-Principles and Machine Learning Interatomic Potentials

The transport of metal ions in liquids with extensive hydrogen (H)-bonding networks such as deep eutectic solvents (DESs) are essential to study as they have applications in areas such as critical mineral recovery and redox flow batteries for grid energy storage. In particular, experiments have found that ion transport in DESs is possible without a counterion, necessitating first-principles understanding of the local charge transfer between metal ion and neat solvent, solvation shells, and resultant ion transport.

As first-principles-based computational modeling of metal ion transport in DESs is limited by simulation size and high viscosity arising from strong H-bonding, we apply machine learning approaches to approximate the potential energy surface (PES) for a metal ion and DES system comprising of divalent nickel and ethaline (a 1:2 molar ratio of choline chloride and ethylene glycol). We describe how reliably benchmarked quantum chemistry calculations with on-the-fly active learning using FLARE: Fast Learning of Atomistic Rare Events [1] generates representative training data which are used to train machine learning interatomic potentials to learn the PES. We apply our computational approach to study the transport of divalent nickel in ethaline.

[1] Vandermause, J., Torrisi, S.B., Batzner, S. et al. On-the-fly active learning of interpretable Bayesian force fields for atomistic rare events. npj Comput Mater 6, 20 (2020).