Neural networks provide an efficient, general interpolation method for nonlinear functions of several variables. This paper describes the use of feed-forward neural networks to model global properties of potential energy surfaces from information available at a limited number of configurations. As an initial demonstration of the method, several fits are made to data derived from an empirical potential model of CO adsorbed on Ni(111). The data are error-free and geometries are selected from uniform grids of two and three dimensions. The neural network model predicts the potential to within a few hundredths of a kcal/mole at arbitrary geometries. The accuracy and efficiency of the neural network in practical calculations are demonstrated in quantum transition state theory rate calculations for surface diffusion of CO/Ni(111) using a Monte Carlo/path integral method. The network model is much faster to evaluate than the original potential from which it is derived. As a more complex test of the method, the interaction potential of H2 with the Si(100)-2×1 surface is determined as a function of 12 degrees of freedom from energies calculated with the local density functional method at 750 geometries. The training examples are not uniformly spaced and they depend weakly on variables not included in the fit. The neural net model predicts the potential at geometries outside the training set with a mean absolute deviation of 2.1 kcal/mole.

Tautomerization of amino acids between the neutral form (NF) and the zwitterionic form (ZW) in water has been extensively studied, often using glycine as a model to understand this fundamental process. In spite of many advanced studies, the tautomerization reaction remains poorly understood because of the intrinsic complexities of the system, including multiple accessible reaction pathways, charge transfer, and variations of solvation structure. To establish an accurate model that can be used for molecular dynamics simulations, a ReaxFF reactive force field has been developed for glycine. A training set for the ReaxFF hydrocarbon potential was augmented with several glycine conformers and glycine-water complexes. The force field parameters were optimized to reproduce the quantum mechanically derived energies of the species in the training set. The optimized potential could accurately describe the properties of gas-phase glycine. It was applied to investigate the effect of solvation on the conformational distribution of glycine. Molecular dynamics simulations indicated significant differences in the dominant conformers in the gas phase and in water. This suggests that the tautomerization of glycine occurs through a conformational isomerization followed by the proton transfer event. The direct reaction mechanism of the NF→ZW proton transfer reaction in water, as well as mechanisms mediated by one or two water molecules, were investigated using molecular dynamics simulations. The results suggest that the proton transfer reaction is most likely mediated by a single water molecule. The ReaxFF potential developed in this work provides an accurate description of proton transfer in glycine and thus provides a useful methodology for simulating proton transfer reactions in organic molecules in the aqueous environment.

Adsorption of water on the Si(100)-(2×1) surface has been investigated using density functional theory and cluster models of the surface. The reaction pathway and geometries of the product, the transition state and a molecular precursor state are described. There is no energy barrier to dissociative chemisorption. Adsorbed H and OH fragments are most stable when bonded to the same surface dimer with the hydroxyl oriented away from the surface dimer bond. The orbital and electrostatic interactions that determine the adsorbate and transition state geometries are analyzed. Surface distortion (dimer buckling) is a recurring theme in this analysis. Interactions of adsorbed molecular fragments with each other and with dangling bonds have significant effects, modifying the adsorbate geometry and leading to adsorbate islanding. Calculated vibrational frequencies of adsorbed H2O on Si(100)-(2×1) are discussed. The theoretical results are consistent with most available experimental results, and provide a microscopic description of the interactions that account for the observations.