Modeling of pure compounds surface tension using QSPR

A theoretical method for predicting the surface tension of pure liquid compounds at 25 °C from their molecular structure is presented. A back propagation artificial neural network algorithm was used to select the appropriate functional groups and investigate their contribution to the surface tension property. The networks were used to probe the functional groups and determine the ones that have significant contribution to the overall surface tension property and arrive at the set of groups that can best represent the surface tension for about 560 substances. The 46 functional groups arrived at can predict the surface tension of pure compounds from the knowledge of the molecular structure alone with a correlation coefficient of 0.99 and an AAD of 0.69 dyne/cm. The results are further compared with other methods in the literature.