Toward molecular modeling of thiophene distribution between the ionic liquid and hydrocarbon phases: Effect of hydrocarbon structure

Abstract

In the present study, Quantitative Structure-Properties Relationship (QSPR) methodology has been employed to propose predictive and descriptive models for the prediction of thiophene distribution between the ionic liquid (IL) and hydrocarbon-rich phases in ternary systems containing IL, thiophene, and hydrocarbon solvent. By comprehensive literature survey, three different datasets (the first dataset:[C2MIM][EtSO4] (1) – thiophene (2) – six different hydrocarbon solvents (3), the second dataset: [C8MIM][BF4] (1) – thiophene (2) – eight different hydrocarbon solvents (3), and the third dataset: [C8MIM][NTF2] (1) – thiophene (2) – eight different hydrocarbon solvents(3)) with total 179 data points were found for model development. After categorizing the data points to the train and test data in each data set, QSPR models were constructed for each data set using genetic algorithm multiple linear regression (GA-MLR). Acceptable values of the statistical parameters such as root mean square error (RMSE), absolute relative deviation (ARD), and average absolute relative deviation (AARD) confirmed the validity of the developed model for each data set. The validity of models was also confirmed by external validation. It was found that the polarity number (Pol) of hydrocarbon solvent, as a topological descriptor which relates to the steric properties of molecular structures, has a remarkable effect on the thiophene distribution between the IL and hydrocarbon-rich phases.