Use of GAMESS/COSMO program in support of COSMO-SAC model applications in phase equilibrium prediction calculations

Thermodynamic models based on conductor-like screening models (COSMO) offer viable alternatives to existing group-contribution methods for the prediction of phase equilibria. Normally a COSMO-based model requires input of the distribution of screening charges on the molecular surface, aka. the sigma profile, determined from a specific quantum chemistry program and settings. For example, the COSMO-SAC model requires input of DMol³ generated sigma profiles. In this paper, we investigate the proper settings for an open-source quantum chemistry package GAMESS in order to generate sigma profiles to be used directly in the COSMO-SAC model. The phase behaviors (VLE and VLLE) of 45 binary mixtures from 10 commonly used solvents and the solubilities of 4 complex drug compounds in these solvents calculated from DMol³ and GAMESS generated sigma profiles are compared. While noticeable fine structure differences are observed in the individual sigma profiles for the same chemical compound generated from the two packages, it is found that the accuracy in the VLE/VLLE and solubility predictions from the two packages are comparable. Based on the systems we studied here, the open-source GAMESS/COSMO program with proper program settings could be used as an alternative sigma profile generation source in support of COSMO-SAC model applications in phase equilibrium prediction calculations.