Abstract

The rational design of ionic liquids (ILs) is often hindered when promising candidates, such as carboxy-functionalized imidazolium chlorides, exhibit properties like extreme viscosity that preclude direct experimental measurement. In this study, we synthesized a series of these ILs and addressed this “experimental gap” with a combined computational strategy. For the few liquids accessible to measurement, we obtained density, viscosity, and conductivity data. For the majority, we turned to atomistic modeling and machine learning. Symmetry-adapted perturbation theory (SAPT2) energy decomposition uncovered the dominance of electrostatic interactions in governing viscosity, an insight obscured by total binding energies from DFT. In addition, a recently developed machine learning model, named IonIL-IM-D1, predicted the density of [C2COOHeim][Cl] with an error of less than 1% upon validation, though experimental verification for the solid candidates was not possible. This predictive framework was extended to propose and evaluate new IL candidates, offering a complementary strategy for exploring macroscopic behavior when direct experimental measurements are not feasible.

IPC Classification

Keywords