The pairwise descreening approximation provides a rapid computational algorithm for the evaluation of solute shape effects on electrostatic contributions to solvation energies. In this article we show that solvation models based on this algorithm are useful for predicting free energies of solvation across a wide range of solute functionalities, and we present six new general parametrizations of aqueous free energies of solvation based on this approach. The first new model is based on SM2-type atomic surface tensions, the AM1 model for the solute, and Mulliken charges. The next two new models are based on SM5-type surface tensions, either the AM1 or the PM3 model for the solute, and Mulliken charges. The final three models are based on SM5-type atomic surface tensions and are parametrized using the AM1 or the PM3 model for the solute and CM1 charges. The parametrizations are based on experimental data for a set of 219 neutral solute molecules containing a wide range of organic functional groups and the atom types H, C, N, O, F, P, S, Cl, Br, and I and on data for 42 ions containing the same elements. The average errors relative to experiment are slightly better than previous methods, butmore significantlythe computational cost is reduced for large molecules, and the methods are well suited to using analytic derivatives.

No abstract

We present a new set of geometry-based functional forms for parametrizing effective Coulomb radii and atomic surface tensions of organic solutes in water. In particular, the radii and surface tensions depend in some cases on distances to nearby atoms. Combining the surface tensions with electrostatic effects included in a Fock operator by the generalized Born model enables one to calculate free energies of solvation, and experimental free energies of solvation are used to parametrize the theory for water. Atomic charges are obtained by both the AM1-CM1A and PM3-CM1P class IV charge models, which yield similar results, and hence the same radii and surface tensions are used with both charge models. We considered 215 neutral solutes containing H, C, N, O, F, S, Cl, Br, and I and encompassing a very wide variety of organic functional groups, and we obtained a mean unsigned error in the free energy of hydration of 0.50 kcal/mol using CM1A charges and 0.44 kcal/mol using CM1P charges. The predicted solvation energies for 12 cationic and 22 anionic solutes have mean unsigned deviations from experiment of 4.4 and 4.3 kcal/mol for models based on AM1 and PM3, respectively.

We present eight new parameterizations of the SM5.42R solvation model: in particular we present parameterizations for HF/MIDI!, HF/6-31G*, HF/6-31+G*, HF/cc-pVDZ, AM1, PM3, BPW91/MIDI!, and B3LYP/MIDI!. Two of the new cases are parameterized using the reaction-®eld operator presented previously, and six of the new cases are parameterized with a simpli®ed reaction-®eld operator; results obtained by the two methods are compared for selected examples. For a training set of 2135 data for 275 neutral solutes containing H, C, N, O, F, S, P, Cl, Br, and I in 91 solvents (water and 90 nonaqueous solvents), seven of the eight new parameterizations give mean unsigned errors in the range 0.43±0.46 kcal/mol, and the eighth ± for a basis set containing diuse functions ± gives a mean unsigned error of 0.53 kcal/mol. The mean unsigned error for 49 ionic solutes (containing the same elements) in water is 3.5±3.9 kcal/mol for the Hartree±Fock, Becke±Perdew± Wang-1991 and Becke three-parameter Lee±Yang±Parr cases and 4.1 and 4.0 kcal/mol for parameterized model 3 and Austin model 1, respectively. The methods are tested for sensitivity of solvation free energies to geometry and for predicting partition coecients of carbonates, which were not included in the training set.

We present a new solvation model for predicting free energies of transfer of organic solutes from the gas phase to aqueous and organic solvents. The model is based on class II charges, gas-phase geometries, a generalized Born approximation to the polarization free energy, and SM5-type atomic surface tensions. The initial parametrization of the new model was developed to utilize the MNDO/d Hamiltonian, and we also present parameters for the MNDO, AM1, and PM3 Hamiltonians. These parametrizations are based on reasonably accurate gas-phase geometries for 43 ions and 260 neutral solute molecules composed of H, C, N, O, F, S, Cl, Br, and I and containing a wide variety of functional groups. For aqueous solutions, the parametrization is based on data for 248 of the neutrals and all of the ions. For organic solvents, it is based on 1836 experimental data points for 227 of the neutral solutes in 90 organic solvents. The parametrization based on the MNDO/d Hamiltonian is called SM5.2R/MNDO/d, and it yields a mean unsigned error of 3.8 kcal/mol for the free energy of hydration of ions and a mean unsigned error of 0.38 kcal/mol for the free energy of solvation of neutral solutes. Gas-phase geometries for all solute molecules were calculated at the Hartree−Fock level with a heteroatom-polarized valence-double-ζ basis set (HF/MIDI!), and we confirmed that the average errors increase only about 0.1 kcal/mol if we use the MNDO/d geometries.