Prediction of thermal decomposition temperatures of binary imidazolium ionic liquid mixtures using improved E-state index descriptors

Ionic liquid (IL) mixtures have been widely used in various fields as new green “design solvents”. However, ILs are often used at high temperatures, which may trigger thermal hazards. The thermal decomposition temperature (T_d) is an important parameter to characterize their thermal hazards. In this work, a quantitative structure-property relationship (QSPR) method is used to develop a model for predicting T_d of binary imidazolium IL mixtures. Twelve kinds of mixing rules are used to improve the original electrotopological state (E-state) index descriptors, which can better describe the interaction of binary IL mixtures as well as the structural characteristics. By using the random forest (RF) method to build prediction models, two models with three descriptors (R² = 0.974) and four descriptors (R² = 0.977) are obtained by comparing their predictive capability. The various validations have demonstrated that those two models have good robustness and predictive capabilities. This work provides two reliable models to predict the T_d of binary imidazolium IL mixtures, which is expected to provide theoretical guidance for the safe use of binary imidazolium IL mixtures.