Towards the prediction of infinite dilution activity coefficient (IDAC) of methanol in ionic liquids (ILs) using QSPR-based COSMO descriptors: Considering temperature effect using machine learning

In this study, the ‘Quantitative Structure-Activity/Property Relationship’ (QSAR/QSPR) approach has been applied for the prediction of infinite dilution activity coefficient (IDAC) of Methanol (MeOH) in Ionic Liquids (ILs) using an extensive dataset. A new predictive QSPR model including novel molecular descriptors, called ‘COSMO-RS descriptors’, has been developed for the first time. In this study, the dataset was divided to a training set for the development of models, and a validation set for external validation. According to the obtained results of statistical parameters (R2 = 0.92 and Q2LOO-CV = 0.91), the predictive capability of the developed QSPR model was acceptable for training set. Regarding the external validation, other statistical parameters such as AAD = 0.2034 and RMSE = 0.2926 were also satisfactory for validation set. While the values of IDAC increase or decrease with increasing temperature, the QSPR model based on the van’t Hoff equation takes into account the ‘negative’ and ‘positive’ effects of temperature on the IDAC of MeOH in ILs well, depending on the nature of ILs. It was also shown that the IDAC value in some new ILs, which had not been experimentally studied before, can be predicted using QSPR model. These predicted data can be considered as ‘Pseudo Experimental data’ for future efforts.