Dispersion and repulsion contributions to the solvation free energy: Comparison of quantum mechanical and classical approaches in the polarizable continuum model

Abstract

We report a systematic comparison of the dispersion and repulsion contributions to the free energy of solvation determined using quantum mechanical self-consistent reaction field (QM-SCRF) and classical methods. In particular, QM-SCRF computations have been performed using the dispersion and repulsion expressions developed in the framework of the integral equation formalism of the polarizable continuum model, whereas classical methods involve both empirical pairwise potential and surface-dependent approaches. Calculations have been performed for a series of aliphatic and aromatic compounds containing prototypical functional groups in four solvents: water, octanol, chloroform, and carbon tetrachloride. The analysis is focused on the dependence of the dispersion and repulsion components on the level of theory used in QM-SCRF computations, the contribution of those terms in different solvents, and the magnitude of the coupling between electrostatic and dispersion-repulsion components. Finally, comparison is made between the dispersion-repulsion contributions obtained from QM-SCRF calculations and the results determined from classical approaches.

How accurate are continuum solvation models for drug-like molecules?

“…This is four times more than in the original investigation, 4 kJ/mol, but such a low uncertainty is highly suspicious, considering that the uncertainty in the experimental results is 8-29 kJ/mol [35,39,58,94]. In fact, the main reason for the large difference is that we have used different experimental data: The experimental estimates for the nine ions that are included in both investigations differ by up to 33 kJ/mol.…”

“…In fact, the five PCM methods give positive solvation energies for these two molecules, whereas all the other methods give negative solvation energies. The difference can be traced entirely to the non-polar terms and may be connected to the use of pair-potentials for the dispersion and repulsion terms [94] or to the use of the van der Waals surface (which contains many small cavities inside large molecules), rather than the solvent-accessible surface, for the cavitation term. If these two molecules are removed from the analysis, the average range decrease to 70 kJ/mol, but this is still four times larger than for the small organic molecules.…”

Dependence of the wittig reaction mechanism on the environment and on the substituents at the aldehyde group and at the phosphonium ylide

“…The positive cavitation contributions turned out to be larger indeed for the reactants than for the TS. Usually, there is a close balance between cavitation and dispersion, whereas this does not occur in those critical points, because dispersion and repulsion terms (determined in the code using classical expressions with self‐consistency between solute and solvent limited to the electrostatic component 47) are both fairly conserved, in contrast to cavitation.…”

Performance of the IEF-MST solvation continuum model in the SAMPL2 blind test prediction of hydration and tautomerization free energies

“…The dielectric response of the polarizable medium is characterized by a single value of permittivity, which is taken to be equal to the bulk value of pure solvent, and DG ele thus rather accounts for the bulk-electrostatic interaction between solute and solvent, neglecting then in principle the contribution of specific solute-solvent interactions. The non-electrostatic component includes a priori a variety of physical contributions, such as the energy cost associated with the creation of the solute cavity, and exchange-repulsion and dispersion interactions between solute and solvent, but in practice it also accounts for any deviation introduced upon assumption of the bulk-electrostatic interaction for DG ele : Finally, though both DG ele and DG nÀele should a priori be treated self-consistently [6][7][8][9], in practice most QM-SCRF models limit self-consistency to DG ele ; while DG nÀele depends on nuclear coordinates using either a single term for nonelectrostatic contributions or by computing separately cavitation and repulsion-dispersion components [10]. Keeping in mind the preceding considerations, the computational protocol optimized for each QM-SCRF model, which also involves a careful parametrization against experimental data, must be kept for an accurate prediction of DG sol [11,12].…”