Prediction of Melting Temperature of bis(trifluoromethylsulfonyl) Imide Anion Based Ionic Liquids Using QSPR Approach

In this study, the melting temperature of ionic liquids based on bis(trifluoromethylsulfonyl)imide anion using experimental data were modeled using the QSPR approach. The structures of ionic liquids were drawn by Avogadro software and their 3D structure were optimized by Merck molecular force field. Descriptors were calculated using Dragon and Padel software. After reducing the number of descriptors, the ionic liquids were divided into training and test set employing Kenard- Stone algorithm. For choosing the best set of descriptors, a genetic algorithm with Friedman's lack of fitness was used. Then, the numbers of optimized variables were determined by the multiple linear regression MLR method and the linear equation between variables was obtained. Linear and nonlinear models were built by means of Multiple Linear Regressions (MLR), MultiLayer Perceptron (MLP) neural networks, and using genetic algorithm. R2, AARD values were 0.787, 0.043 for linear model and 0.785, 0.0428, for nonlinear model. Both models showed the same accuracy. The applicability domain of the proposed models was also determined. Also using the obtained descriptors, the effect of molecular structure on the physical properties such as melting temperature was addressed.