Thermodynamic property prediction through the group contribution methods has been improving. However, the approaches considered in the past present not only limitations on the physical basis but often have restrictions as to certain substances, such as isomers. A new group contribution method is proposed that uses AIM theory, which is based on computational chemistry and quantum mechanics, to overcome these limitations by treating each molecule individually. An evaluation of this method as applied to the Knox model is proposed and analyzed for Vapor Liquid Equilibrium (VLE) of mixtures with the help of nine global parameters that are obtained by correlation.

This method is able to calculate with accuracy VLE for many systems. Both binary and ternary mixtures have been evaluated and have shown that the model can predict the behavior of the systems for several types of mixtures. The model has proved to work well with systems that have presented trouble in the past, such as isomers or polar mixtures, giving very small errors.