| A method has been developed to realistically model diffusion and accumulation of sparse defects within large systems over long time scales. Starting from an atomistic regime, regions containing low barrier processes are identified as defective and are classified using a graph representation. Transitions available to these defects are determined on the fly using Dimer, ARTn or RAT searches and are assigned to the appropriate defect in a process database. Rediscovered defects are mapped against the stored form and the saved processes are applied in a symmetrically and rotationally invariant fashion. Once the permutations for each defect have been explored, the system is evolved very rapidly in a purely object KMC regime - returning to atomistic level detail only when new configurations are discovered, for example when 2 defects come within interaction distance. Problems are demonstrated including vacancy cluster aggregation in Cu, where over 95% of the (104 step) KMC simulation occurs within the object KMC framework. |
|