No abstract

The approach to the extension of the optimized potentials for liquid simulations all-atom (OPLS-AA) force field, which was tested for pyridine and the diazenes in the first paper of this series, has been extended to pyrrole, pyrazole, imidazole, furan, isoxazole, and oxazole. Standard OPLS Lennard-Jones parameters are used for the nonbonded interactions in conjunction with partial charges obtained from fitting to RHF/6-31G* electrostatic potential surfaces. The harmonic bond stretching and angle bending terms are adopted mostly from the AMBER force field, although the addition of two new atom types was required. The resultant force field was tested by computing the gas-phase structures of the heterocycles, heterocycle-water interaction energies, properties of the pure liquids, and the relative free energies of hydration for pyrrole and imidazole.

The formation and stability of self-assembled monolayers (SAMs) of aryl sulfmates and, for the first time, aryl sulfonates are described. The ways in which the molecules interact with the surface and the stability of the resulting SAMs were characterized by surface enhanced Raman (SER) spectroscopy. Aryl sulfinate monolayers can be reversibly oxidized to sulfonate monolayers, but the sulfonate is readily displaced by sulfinate in solution.The relative adsorptivities of aryl sulfur species were found to be ArSC>3_ < < ArSC>2-< ArS-. Through a novel application of perturbation theory, in which the adsorbate-surface Coulombic and charge transfer interactions and the change in the solvation free energy of the adsorbate are taken into account, we have been able to explain this trend. The differences in the adsorptivities of the anions studied here are primarily attributable to differences in the adsorbate-surface charge transfer interactions. These were evaluated by calculating the adsorbate HOMO energies at the ab initio HF/3-21G(d) level. Higher adsorbate HOMO energies are correlated with higher adsorptivities, consistent with established trends in the adsorptivities of soft, basic anions on metal electrodes. Statistical perturbation theory was used to calculate the relative free energies of hydration of the three anions. The sulfonate is most strongly solvated, followed by the sulfinate and thiolate. Thus, the trend in solvation energies is consistent with and probably reinforces the trend in the adsorbate-surface charge transfer interactions. The combination of computational methods used here may prove generally useful for predicting the relative adsorptivities of molecules and ions on metal surfaces.

The intermolecular potential functions that are commonly used for the simulation of biomolecular systems in programs such as AMBER, BOSS, CHARMM, and GROMOS are pairwise-additive; polarization effects are only represented in an average sense through the use of partial charges that yield gas-phase dipole moments enhanced 10-20% over experimental values.' This approach has been remarkably successful for simulations of a wide range of systems including pure water,1 2 aqueous solutions of neutral organic molecules,3 and even C1~(H20)" and Na+(H20)" clusters.4 However, without polarization, energetic errors become increasingly unacceptable for stronger ion-molecule interactions.5 Furthermore, it was noted by Kumpf and Dougherty that alkali cationinteractions are too weak by ca. 50% with the OPLS model of benzene and Aqvist's ion parameters.6 The problem also occurs for ammonium ion-benzene interactions and can be remedied by explicit consideration of polarization, as recently reported by Caldwell and Kollman.7 Though the need for polarization with such strong interactions is not unexpected, we document here that polarization is also important for polar molecules in lowdielectric media, e.g., hydrocarbon solvents or hydrophobic regions in proteins.

Monte Carlo statistical mechanics simulations were used to compute absolute free energies of solvation in chloroform for 16 organic molecules. The intermolecular interactions were described by classical potential functions consisting of Coulomb and Lennard-Jones interactions. The partial charges for the solutes were derived from fitting to the electrostatic potential surfaces of ab initio 6-31G* wavefunctions. First, free energy perturbation (FEP) calculations yielded relative free energies of solvation. These were converted to absolute quantities through perturbations to the reference molecule, methane, which was annihilated. The average error in the FEP-computed free energies of solvation is 0·8 kcal mol . Combination of the predictions of free energies of solvation in water and chloroform yields partition coefficients, log P, with an average error of 0·3-0·4 log unit.