(711D) Predicting Properties of Ionic Liquid Mixtures Using  Molecular Simulations and Machine Learning

Ionic liquids are composed exclusively of an organic cation and anion that can be either organic or inorganic in nature. Highly articulated nature of the ions implies a delocalization of either the positive or negative charge, resulting in melting points of these compounds below ambient conditions. Extremely low vapor pressures and ability to target properties suitable for a given application by carefully selecting the ion combination are the primary drivers for research on the application of ionic liquids in CO₂ capture, gas separations, batteries, solar cells, and reactions, etc. While designing an ionic liquid, one is confronted with the vast chemical space for selecting cations and anions; it is estimated that there are more than a billion ionic liquids. Considering that almost all applications of ionic liquids are likely to involve mixtures of ionic liquids or ionic liquid-organic solvents, the number of such possible combinations would be in trillions. High throughput screening using machine learning and understanding molecular phenomena with atomistics simulations play an important role to navigate the vastness of the available chemical space. In this presentation, challenges and opportunities for predicting properties of ionic liquid mixtures from atomistic simulations will be highlighted. We will also discuss our efforts at employing machine learning in rapidly estimating properties of ionic liquid-ionic liquid mixtures and gaps that must be bridged for property predictions.