Quantitative Atomistic Simulations of Reactive and Non-Reactive Processes

Abstract

The interpretation of physico-chemical observables in terms of atomic motions is one of the primary objectives of atomistic simulations. Trajectories from a molecular simulation contain much valuable information about the relationship between motion of the atoms and physical observables related to them, provided that the interactions used to generate the trajectories are of sufficiently high quality. On the other hand, many experimental observables are averages over a large number of physical realizations of the system. Thus, a statistically large number of trajectories needs to be generated and analyzed in order to provide a meaningful basis for comparison with and interpretation of experiments. The preferred computational approach which allows such extensive averaging while retaining the quantitative aspects of the intermolecular interactions are accurate force field-based molecular dynamics simulations. This contribution provides an overview of our group's current technological improvements in force field technology and its application to fundamental physico-chemical questions.Keywords: Computational spectroscopy · Force fields · Molecular dynamics simulations · Multipoles · Reaction dynamicsIn the physical sciences simulations provide a third approach -next to experiment and theory -to characterize and understand a wide range of phenomena. For chemistry and chemical physics in particular, atomistic simulations are paramount in providing insights into the energetics and dynamics of complex systems, such as proteins, chemical reactivity, solvation dynamics or spectroscopy.The use of atomistic simulations in reaction dynamics has been established for almost 50 years when molecular dynamics (MD) simulations were used to better understand the H 2 + H reaction dynamics. [1] Since then, MD simulations have become an important complement to investigate complex systems at atomic resolution. The major driving forces behind this development are i) the increase in computational power, ii) the improvement of algorithms and energy functions and iii) the more direct interaction between experiment and simulation. It is the latter that will eventually let this field mature to a degree which allows quantitative and ideally predictive work to be carried out which must be the ultimate goal of atomistic simulations.In the present contribution I will summarize our efforts to item ii) above. More specifically, our group is concerned with improved and widely applicable energy functions for spectroscopic and thermodynamic applications and the development of computational methods to follow chemical reactions in the gas phase and in solution. Quantitative Intermolecular InteractionsWithin the framework of atomistic simulations, intermolecular interactions are described by an empirical energy function ('force field') which is based on a ball-andspring model for the bonded interactions (such as bonds, valence angles, dihedrals) and Coulomb and van der Waals interactions for the nonbonded interactions. [2] This is in contrast t...