A simple group contribution correlation for modeling the surface tension of pure ionic liquids

Abstract

A simple but general atomic contribution model was developed to predict the surface tension of pure ionic liquids (ILs) at atmospheric pressure as a function of temperature and the number of atoms in the IL (anionic carbon, cationic carbon, oxygen, nitrogen, etc.). The correlation was developed with 1246 data points of 61 ILs, many of which contained different anions and cations. In order to find the constants of the proposed functionality relation through simulated annealing (SA) algorithm, the database was randomly divided into two subsets of training and testing data with the ratios of 70 and 30%, respectively. A thorough statistical assessment indicated that the proposed correlation with the relative mean square error (RMSE) of 2.232, average absolute relative deviation percent (AARD %) of 4.16, and correlation coefficient (R2) of 0.9190 is very robust and has an acceptable error trend. In an attempt to validate the generality of the suggested model, 15 ILs not included in the optimization process were fed to the developed correlation. Although some of these ILs included cations and anions not employed in developing the correlation, the predicted values were acceptably in agreement with corresponding experimental data. A generalized and user friendly correlation with accessible input variables and adequate error trends was established for modeling the surface tension of pure ILs.