The continuum solvation model COSMO and its extension beyond the dielectric approximation (COSMO-RS) have been carefully parametrized in order to optimally reproduce 642 data points for a variety of properties, i.e., ∆G of hydration, vapor pressure, and the partition coefficients for octanol/water, benzene/water, hexane/ water, and diethyl ether/water. Two hundred seventeen small to medium sized neutral molecules, covering most of the chemical functionality of the elements H, C, N, O, and Cl, have been considered. An overall accuracy of 0.4 (rms) kcal/mol for chemical potential differences, corresponding to a factor of 2 in the equilibrium constants under consideration, has been achieved. This was using only a single radius and one dispersion constant per element and a total number of eight COSMO-RS inherent parameters. Most of these parameters were close to their theoretical estimate. The optimized cavity radii agreed well with the widely accepted rule of 120% of van der Waals radii. The whole parametrization was based upon density functional calculations using DMol/COSMO. As a result of this sound parametrization, we are now able to calculate almost any chemical equilibrium in liquid/liquid and vapor/liquid systems up to an accuracy of a factor 2 without the need of any additional experimental data for solutes or solvents. This opens a wide range of applications in physical chemistry and chemical engineering.

No abstract

No abstract

Continuum solvation models have proven to yield very valuable information about solvation effects, if cavities close to the van der Waals surface of the solutes are used for the calculation of the screening charges. Unfortunately, such cavity size implies that a small but significant portion of the solute electron density reaches out of the cavity. This outlying charge causes serious problems in the context of the dielectric continuum treatment of the solvent. The present paper presents a critical consideration of the origin, the magnitude, and the different strategies for treatment of this problem. Finally, a novel ansatz using an additional, outer cavity is presented which provides rather accurate correction of the corresponding error.

No abstract