Structural cluster analysis of chemical reactions in solution.

We introduce a simple and general approach to the problem of clustering structures from atomic trajectories of chemical reactions in solution. By considering distance metrics which are invariant under permutation of identical atoms or molecules, we demonstrate that it is possible to automatically resolve as distinct structural clusters the configurations corresponding to reactants, products, and transition states, even in presence of atom-exchanges and of hundreds of solvent molecules. Our approach strongly simplifies the analysis of large trajectories and it opens the way to the construction of kinetic network models of activated processes in solution employing the available efficient schemes developed for proteins conformational ensembles.

DFT-derived reactive potentials for the simulation of activated processes: the case of CdTe and CdTe:S.

We introduce a new ab initio derived reactive potential for the simulation of CdTe within density functional theory (DFT) and apply it to calculate both static and dynamical properties of a number of systems (bulk solid, defective structures, liquid, surfaces) at finite temperature. In particular, we also consider cases with low sulfur concentration (CdTe:S). The analysis of DFT and classical molecular dynamics (MD) simulations performed with the same protocol leads to stringent performance tests and to a detailed comparison of the two schemes. Metadynamics techniques are used to empower both Car-Parrinello and classical molecular dynamics for the simulation of activated processes. For the latter, we consider surface reconstruction and sulfur diffusion in the bulk. The same procedures are applied using previously proposed force fields for CdTe and CdTeS materials, thus allowing for a detailed comparison of the various schemes.

The Fate of a Zwitterion in Water from ab Initio Molecular Dynamics: Monoethanolamine (MEA)-CO2.

Understanding the fundamental reactions accompanying the capture of carbon dioxide in amine solutions is critical for the design of high-performance solvents and requires an accurate modeling of the solute-solvent interaction. As a first step toward this goal, using ab initio molecular dynamics (Car-Parrinello) simulations, we investigate a zwitterionic carbamate, a species long proposed as intermediate in the formation of a stable carbamate, in a dilute aqueous solution. CO2 release and deprotonation are competitive routes for its dissociation and are both characterized by free-energy barriers of 6-8 kcal/mol. Water molecules play a crucial role in both pathways, resulting in large entropic effects. This is especially true in the case of CO2 release, which is accompanied by a strong reorganization of the solvent beyond the first coordination shell, leading to the formation of a water cage entrapping the solute (hydrophobic effect). Our results contrast with the assumptions of implicit solvent models.

Bridging Static and Dynamical Descriptions of Chemical Reactions: An ab Initio Study of CO2 Interacting with Water Molecules.

Extracting reliable thermochemical parameters from molecular dynamics simulations of chemical reactions, although based on ab initio methods, is generally hampered by difficulties in reproducing the results and controlling the statistical errors. This is a serious drawback with respect to the quantum-chemical description based on potential energy surfaces. This work is an attempt to fill this gap. We apply molecular dynamics, based on density functional theory (DFT) and empowered by path metadynamics (MTD), to simulate the reaction of CO2 with (one, two, and three) water molecules in the gas phase. This study relies on a strategy that ensures a precise control of the accuracy of the reaction coordinates and of the reconstructed free-energy surface on this space, namely, on (i) fully reversible MTD simulations, (ii) a committor probability analysis for the diagnosis of the collective variables, and (iii) a cluster analysis for the characterization of the reconstructed free-energy surfaces. This robust procedure permits a meaningful comparison with more traditional calculations of the potential energy surfaces that we also perform within the same DFT computational scheme. This comparison shows in particular that the reactants and products of systems with only three water molecules can no longer be understood in terms of one structure but must be described as statistical configuration ensembles. Calculations carried out with different prescriptions for the exchange-correlation functionals also allow us to establish their quantitative effect on the activation barriers for the formation and the dissociation of carbonic acid. Their decrease induced by the addition of one water molecule (catalytic effect) is found to be largely independent of the specific functional.