Multiscale diffusion Monte Carlo simulation of epitaxial growth (Q2508892)

The paper describes an accelerated kinetic Monte Carlo (KMC) simulation algorithm for molecular beam epitaxial growth in the step flow regime. In the algorithm, an ensemble of off-lattice random walkers is used to simulate a diffusion field. It is considered epitaxial growth on 1-D surface in absence of direct interaction between atoms. Four possible events are considered, namely: (i) hopping of adatoms on the surface, (ii) attachment of adatoms to the edge, (iii) detachment of atoms from the edge, and (iv) flux of atoms to the surface. First, the classic KMC algorithm for epitaxial growth is considered, and analyzed its computational complexity. Then, the authors develop accelerated algorithms from the idea that the detailed hopping of adatoms far away from the edges are not important to the overall behavior of the system, especially in the considered model, where the adatoms are coupled to each other only through the movements of the edges. The requirement that adatoms far away from the step edges take larger jumps with rates proportional to the inverse square of the step size allows to greatly accelerate KMC simulation of epitaxial growth, while retaining the accuracy of physical results, including stochastic fluctuations in the considered region. The accelerated algorithms and complexity analysis are verified by comparing the physical results and performance those algorithms with continuum models and classical KMC simulation. The proposed algorithms are orders of magnitude faster than the classical one and superior in terms of computational complexity.