Abstract: Quantitative structure-activity relationship (QSAR) was used to predict the melting points of two kinds of imidazolium ionic liquids. Using a genetic algorithm a three-parameter model with R2 =0.89 was set up for 16 imidazolium tetrafluoroborates and a four-parameter model with R2 =0.85 was developed for 22 imidazolium hexafluorophosphates. Internal reliability validation was performed using a leave-one-out approach with R2=0.82 for the model of imidazolium tetrafluoroborates and R2 =0.77 for the model of imidazolium hexafluorophosphates. To validate predictive abilities of models an external validation was performed for test sets where the mean absolute deviation were 11.43 and 7.78 K, respectively. These values are better than the values found in the literature. The proposed models can be used to predict melting points of imidazoliumionic liquids.

Key words: QSAR, Imidazoliumionic liquids, Genetic algorithm, Melting point