Chemical Transferability and Accuracy of Ionic Liquid Simulations with Machine Learning Interatomic Potentials

Ionic liquids (ILs) are an exciting class of electrolytes finding applications in many areas from energy storage to solvents, and have been touted as “designer solvents”, as they can be mixed to precisely tailor the physiochemical properties. As using machine learning interatomic potentials (MLIPs) to simulate ILs is still relatively unexplored, several questions need to be answered to see if MLIPs can be transformative for ILs. Since ILs are often not pure, but are either mixed together or contain additives, we first demonstrate that a MLIP can be trained to be chemically transferable, i.e., can be applied to compositions not directly train on, but with similar chemistries, whilst only being trained on a few compositions. We also investigate the accuracy of MLIPs for a novel IL, which we experimentally synthesis and characterise. Our MLIP trained on ∼200 DFT frames is in reasonable agreement with our experiments and good agreement against DFT, demonstrating the data-efficient power of this method.