A New Model for Estimation of Activity Coefficient of Water at Infinite Dilution in Ionic Liquids

Estimation of activity coefficient of water at infinite dilution in ionic liquids (ILs) is very important for natural gas dehydration design and application. For this purpose, in this work a new simple quantitative structure-property relationship (QSPR) model with only three parameters was successfully developed to predict this property. To better distinct the effects of anion and cation on the activity coefficient of water in ILs, more than 1000 molecular descriptors were calculated for each of cation and anion of an IL. An efficient approach combining modified particle swarm optimization (MPSO) and multiple linear regression (MLR) was used for selecting the most important molecular descriptors and building the optimal QSPR model. The number of molecular descriptors in final QSPR model was 2 for anions and 1 for cations. The proposed QSPR model resulted in squared correlation coefficient values of 0.959 and 0.965 for the training and test sets, respectively. The obtained results in this paper suggest that by the proposed model, it is possible to obtain a good estimation of activity coefficient of water at infinite dilution in ILs.