Modeling surface tension of ionic liquids by chemical structure-intelligence based models

Abstract

Surface tension is one of the most significant physicochemical characteristics in process design, industrial applications of heterogeneous systems, and scientific applications. To estimate the surface tension of ionic liquids, acquiring a reliable and accurate model is crucial since experimental measurements are costly and time-consuming. In this regard, the main objective of this study is to estimate the surface tension of ionic liquids (ILs) by utilizing two mathematical models based on the radial basis function (RBF) and least square-support vector machine (LSSVM), coupled with two optimization algorithms namely firefly algorithm (FFA) and differential evolution (DE). In this work, a huge experimental dataset including 1042 data points from 69 ILs was utilized. The dataset consists of surface tension and temperature over a range of 18.5–70.3 mN/m and 268.29–532.4 K, respectively, where the pressure is constant (0.101 MPa). The input parameters were chosen to be chemical structure and temperature, whereas the surface tension was the output parameter. Sensitivity analysis, statistical, and graphical error were used in order to evaluate the performance and accuracy of the proposed models. The results showed that the LSSVM-FFA model accurately predicts the surface tension of ILs with the percentage of average absolute relative deviation of 1.8440% and determination coefficient of 0.9828. The sensitivity analysis showed that surface tension is affected by some substrates such as NH2 and SO2 more dominantly compared to other substructures. Furthermore, statistical and graphical error analyses of the models developed in this study along with those obtained from the literature demonstrate that the LSSVM-FFA model significantly outperforms all the existing models in terms of accuracy and range of validity. Finally, conclusions drawn from understanding the impact of temperature on the surface tension using the proposed models reveals a declining surface tension with increasing temperature owing to the effect of intermolecular forces. The outcome of this study can help not only circumvent the challenges of predicting surface tension of ILs, but also establish modern and reliable predictive approaches for an extensive dataset over a wide range of temperatures.