Integrated Continuum Dielectric Approaches to treat Molecular Polarizability and the Condensed Phase: Refractive Index and Implicit Solvation.

Abstract

The idea of using a dielectric continuum inside a molecule to accurately model molecular polarizability is extended to include a larger spectrum of bioorganic molecules and the condensed phase. Atomic polarization radii and an internal dielectric (εin) were fitted to reproduce ab initio B3LYP/aug-cc-pVTZ polarizability tensors taken from a data set of 707 molecules. The average unsigned error on the isotropic polarizability and anisotropy are 2.6% and 5.2%, respectively. It is shown that usual Poisson−Boltzmann contact radii and a low internal dielectric are not appropriate and require major revision. To account for the anisotropy of polarizability, the internal dielectric (εin) constant needs to be larger than 6.0. Reinterpreting the theoretical link between εin and the experimental refractive index (n), this study shows, with a set of 23 organic molecules spanning the entire range of n, that even with εin = 24 the obtained refractive indices can correlate well with experiment (slope of 1.00, intercept of 0.05, and R = 0.95). The novel methodology used here to calculate a macroscopic-like refractive index shows that the application of the EPIC parametrization to condensed phase leads to suitable behavior. Although the primary goal in developing EPIC was to include polarizability in explicit solvent calculations, we also extend the model to work with implicit solvent. This requires the use of a 3-zone smooth dielectric function to transition from the polarization dielectric inside the molecules to the dielectric continuum of the solvent. The parametrization and validation of this model are performed against 485 experimental free energies of hydration. Using 8 solvent cavity atomic radii and a single surface tension an average unsigned error of 1.1 kal/mol and a correlation coefficient of 0.9 are obtained, validating the use of the EPIC model in the condensed phase.