Nick J. Mosey
Assistant Professor and
Graduate Coordinator
Department of Chemistry
Queen's University
90 Bader Lane
Kingston, ON, Canada K7L 3N6

email: nick.mosey@chem.queensu.ca
phone: 1-613-533-2650
fax: 1-613-533-6669

Temporal QM/MM

MOLECULAR DYNAMICS (MD) is a powerful tool for study simulating the atomic-level evolution of chemical systems. In an MD simulation, the nuclei are treated at classical particles that are propagated according to Newton's equations of motion. In MD simulations, the forces acting on the nuclei are typically obtained through either of two methods: force-fields (FFs) or quantum chemical (QC) calculations. FFs are empirical relationships between a molecule's potential energy and geometry. These models are computationally-inexpensive, but cannot generally describe the changes in bonding that occur during reactions. Meanwhile, QC methods can describe the changes in electronic structure that occur when bonds form and break during reactions. Unfortunately, the computational expense of QC calculations limits QC-based MD simulations to studying sub-nanosecond time scales, which is too short to observe most reactions without special sampling techniques, thereby limiting the potential of using MD as a tool for identifying reactions.

Temporal QM/MM Temporal QM/MM: Using FFs when a system is close to a potential energy minimum, and switching to QC methods only when the system approaches the transition state provides access to longer MD simulations and the ability to model changes in bonding.

TO OVERCOME the limited time scales accessible in MD simulations of reactions, we are developing a hybrid technique that employs forces derived from both FF and QC methods. Specifically, FFs are used to describe the system near minima on the potential energy surface, and QC methods are used to model the system when it undergoes transitions between these minima. Specifically, the system is modeled with a FF and its behaviour is monitored for the onset of reactions. When the onset of a reactive event is detected, the description of the system is switched smoothly from FF to QC models, and when the reaction is complete, the description of the system reverts to an FF. Since reactive events occur very infrequently and complete very quickly, this approach promises to significantly reduce the use of QC methods in MD simulations of reactions.

IN PRACTICE, this method promises to permit MD simulations on the microsecond timescales accessible with FFs, while retaining the ability to describe chemical reactions. This method is analogous to existing technique called QM/MM that spatially divide a chemical system into a region that is describe with QC methods and another described with FFs as a way to study large systems. Since the method we are developing divides the use of QC and FF methods in time, we call it temporal QM/MM.