(115a) Guiding the Design of Ionic Liquids with Machine Learning

Ionic liquids can be considered as highly concentrated electrolytes which are composed of an organic cation and organic/inorganic anion. The ions are generally asymmetric in nature rendering many as liquids at ambient conditions. Ability to tune the properties of ionic liquids by systematically altering the identity of the cations, anions or substituents on the ions to design novel ionic liquids with desired properties is probably the most important factor for the explosion in the ionic liquids research. Over the years, a large number of ionic liquids have been synthesized and characterized in terms of their thermophysical and phase-equilibria properties. The availability of ionic liquid properties provides exciting opportunities to employ machine learning-based techniques to not only develop structure-property relationships, but also accelerate the discovery of pure, binary and higher order ionic liquids. In this presentation, application of machine learning methods to identify ionic liquids with high conductivity and high electrochemical window will be described. Additionally, it will be demonstrated how the high-dimensional feature space of ionic liquids can be compressed into latent space to generate ionic liquid families not yet synthesized. Finally, challenges and opportunities in using machine learning/data science approaches in the field of ionic liquids will be discussed.