LAUSR

Abstract Transferable force fields allow the prediction of physical properties for substances with scarce or absent experimental data. For substances with a comprehensive experimental database, however, transferable force fields produce results with higher errors than desired, because the transferable force fields are designed to represent a compromise in correlating the properties of many substances. For applications in chemical engineering, where requirements for accurate vapor pressure correlations and predictions exist, the results from transferable force fields are sometimes insufficient. We individualize the transferable anisotropic Mie force field (TAMie) for 38 substances of various chemical families, by introducing a correction parameter that scales all van der Waals energy parameters e i of the considered substance. We find markedly reduced deviations, mainly in the description of vapor pressures while the errors in liquid density do not change significantly. For polar species, the improvement in vapor pressure is typically a factor four in absolute average deviation to experimental data, when compared to the original TAMie force field. The improved description of pure substances enables more reliable predictions of phase equilibria in binary mixtures. The concept of individualizing substances that are well-characterized by experimental data within a transferable force field is appealing, because the ability to predict mixtures with substances that are not covered by experimental data is preserved.