(394l) A Data Science Framework for the Analysis of Ion Transport Mechanisms in Ionic Liquids

Ionic liquids are increasingly investigated for electrochemical energy storage as an alternative to flammable organic electrolytes due to their high stability and tunable chemical structures [1]. However, the strong intermolecular interactions which stabilize these molecules produce highly correlated ionic environments that complicate ionic liquid electrolyte design. The transport models commonly used to model ion transport in liquid electrolytes have led to a framework for maximizing conductivity by minimizing viscosity in ionic liquids [2]. Yet, recent studies show that many ionic liquids deviate from hydrodynamic behavior and instead suggest the presence of ion-hopping transport mechanisms in ionic liquids [3, 4].

To overcome the limitations of theory in understanding ionic liquid design, this work builds upon classical theory frameworks by using machine learning to identify structural and energetic motifs important for describing ion transport. We merge databases of experimentally measured properties and simulated molecular features for 218 ionic liquids. We then use these molecular descriptors in a machine learning model to correct hydrodynamic transport model conductivity predictions and gain insight to the physical origin of non-hydrodynamic ion transport behavior in ILs. Further, this work provides an in-depth analysis of the capacity for various ionic liquid information (structure, energetics, theory, and other properties) to predict ionic liquid ion transport. Intriguingly, we find that individual ion molecular features can predict some materials properties, such as conductivity, while failing to predict other properties that rely on larger length scale events, such as viscous dissipation.

Overall, this new framework provides a new avenue for identifying ionic liquid candidates with desirable properties while learning insights into the material property’s physical origin.

1. Baskin, I., A. Epshtein, and Y. Ein-Eli, Benchmarking machine learning methods for modeling physical properties of ionic liquids. Journal of Molecular Liquids, 2022. 351.
2. Nordness, O. and J.F. Brennecke, Ion Dissociation in Ionic Liquids and Ionic Liquid Solutions. Chemical Reviews, 2020. 120(23): p. 12873-12902.
3. Cashen, R.K., et al., Bridging Database and Experimental Analysis to Reveal Super-hydrodynamic Conductivity Scaling Regimes in Ionic Liquids. J Phys Chem B, 2022. 126(32): p. 6039-6051.