Long time scale simulations of solids undergoing atomic and spin transitions

by Prof. Hannes Jónsson (Háskoli Íslands)

Monday 23 Apr 2012, 15:15 → 16:00 Europe/Stockholm

FA 31

Atomic rearrangements in solids, for example migration of defects, diffusion of atoms or chemical reactions, typically involve overcoming an energy barrier and the time interval between such events is many orders of magnitude longer than the time scale of atomic vibrations. Computer simulations based on direct solution of the classical equation of motion often cannot cover long enough time intervals. A different simulation approach is needed to simulate long time scale evolution in solids. For complex systems where the atomic structure and mechanism of transitions is unknown, such long time scale simulations are particularly relevant. The adaptive kinetic Monte Carlo (AKMC) approach will be described in this presentation. It is based on the two step WKE procedure, which involves the identification of an optimal transition state and subsequent dynamical trajectories started at the transition state. For each state of the system, possible transition mechanisms are explored and transition rates determining, followed by a kinetic Monte Carlo algorithm to pick a transition and advance the clock. The key problem is the identification of a good transition state for each state of the system without bias towards a particular mechanism or final state. Within the harmonic approximation to transition state theory this requires finding all relevant saddle points on the potential energy rim surrounding the energy well corresponding to the current state. More generally, a dividing surface corresponding to maximum free energy needs to be identified. We have developed methods for finding such transition state surfaces and implemented them in software for distributed computing. Application to defects and diffusion in and on the surface of solids as well as spin transitions in nano-clusters will be presented.