Comprehensive models for density prediction of ionic liquid plus molecular solvent mixtures at different temperatures

To present comprehensive models for density prediction of the binary mixtures of ionic liquids (ILs), the density data for 193 sets of ionic liquid + molecular solvents were collected from the literature and were analysed by Jouyban-Acree and Jouyban-Acree-van't Hoff and their combined version with Abraham solvation parameters. To check the capability of the trained models for density prediction, the 'leave many out' cross-validation method was used and the obtained overall % ARD were 3.7, 8.7, 4.1 and 8.5 for Jouyban-Acree, Jouyban-Acree-van't Hoff, Jouyban-Acree-Abraham and Jouyban-Acree-van't Hoff-Abraham models, respectively. It was concluded that all predictive models have the excellent accuracy for density prediction of the studied mixtures but the Jouyban-Acree and Jouyban-Acree-Abraham with lower %ARD are better than other models. However, the Jouyban-Acree-van't Hoff-Abraham models may be preferred due to the lack of necessity of density data for components 1 and 2.