Abhijit Chatterjee1, Blas P. Uberuaga2, and Arthur F. Voter1. (1) Theoretical Division, Los Alamos National Laboratory, T-12 MS B268, Los Alamos, NM 87545, (2) Material Science and Technology Division, Los Alamos National Laboratory, Los Alamos, NM 87545
The temperature accelerated dynamics (TAD) method is a powerful tool for studying rare event dynamics in materials and provides significant computational speed-up compared to molecular dynamics simulations. TAD simulations can easily access large time scales, such as milliseconds to seconds. However, modeling of large length scales with TAD is challenging. Kinetic Monte Carlo (KMC) is an ideal tool for studying dynamics and steady state of noise-controlled phenomena at large length and time scales. However, KMC requires a list of atomic processes, i.e., a process catalog, for system evolution, which are typically not available a priori. As a result, several assumptions are introduced into the process catalog about possible atomic mechanisms. In addition, often atoms are assumed to reside on a lattice. This leads to unrealistic kinetics and thermodynamic behavior in multicomponent and strained material systems.
To overcome these shortcomings of the two methods, we introduce a TAD-based KMC method that generates an on-the-fly process catalog with a prescribed confidence level. The process catalog provides rate information regarding the possible escape pathways from visited potential energy basins to neighboring basins. For the first time confidence measures associated with the TAD process catalogs are derived, which guarantees that the process catalog contains all the relevant escape pathways for the confidence level chosen. The system evolution is studied by employing the KMC method. Computational requirement of the TAD-based KMC method is discussed. In addition, the local TAD-based KMC is discussed, which exploits past visit to similar local atomic environments and provides additional computational speed-up while retaining accuracy of molecular dynamics. Finally, domain decomposition strategies based on the TAD-based KMC method, that can access even larger length and time scales, are discussed. The application of TAD-based KMC method to metal alloys and oxide materials for electronic, catalytic and energy applications is discussed.