Chemoinformatic Approaches To Predict the Viscosities of Ionic Liquids and Ionic Liquid-Containing Systems

Abstract

Modelling, predicting, and understanding the factors influencing the viscosities of ionic liquids and related mixtures are sequentially checked in this work. The molecular maps of atom-level properties (MOLMAP codification system) is adapted for a straightforward inclusion of ionic liquids and mixtures containing ionic liquids. Random Forest models have been tested in this context and an optimal model was selected. The interpretability of the selected Random Forest model is highlighted with selected structural features that might contribute to identify low viscosities. The constructed model is able to recognize the influence of different structural variables, temperature, and pressure for a correct classification of the different systems. The codification and interpretation systems are highlighted in this work.

Keywords: MOLMAP; Random Forest; chemoinformatics; ionic liquids; viscosity.