(656h) Accelerated Discovery of Novel Ionic Liquid Cations Using a Continuous Latent Space Representation of Chemical Space

Ionic Liquids (ILs) have generated tremendous interest in the research community over the years due to several unique and desirable properties compared to conventional solvents. However, its slow transport properties and high material costs have limited its commercial application development. Due to the vastness of the ionic liquid chemical space, narrowing down the potential candidates using experiment or atomistic simulation with faster dynamics, high stability, and low material cost has proven to be difficult. Machine learning (ML) algorithms, on the other hand, could serve as an alternative tool for accelerated material discovery with desirable properties. This paper employs a generative-based deep machine learning method to discover new cations in the ionic liquid chemical space, thereby increasing the number of potential cation candidates for battery applications. According to our findings, some of the newly discovered cations belong to known cation families, whereas most of the cyclic/aromatic cations occupy a new region of chemical space. To assess their suitability as potential electrolytes, we calculate the electrochemical stability of these cations using DFT calculations. Few of the cations have very high electrochemical stability, making them ideal for battery applications.