When modeling epitaxial growth, the properties of interest (e.g. the growth morphology) develop over time scales of the order of seconds and involve several thousands of atoms, while the ruling microscopic processes operate in the length and time domains of 0.1 - 1 nm, and femto- to pico-seconds. Even with a further increase in computer power, such time and length scales will not be accessable to simulations using ab initio molecular dynamics. We are thus challenged by the need to bridge this gap in length and time scales by many orders of magnitude using multiscale modeling techniques. Here, kinetic Monte Carlo (kMC) simulations offer an efficient and accurate simulation tool. While earlier kMC simulations employing empirically derived effective parameters lacked predictive power, we show that the inherent reliability of the ab initio approach can be extended to experimentally relevant situations, provided that the rates entering in a kMC simulation are directly derived from DFT calculations.
Using the results of DFT calculations  for the atomic-scale processes of growth on the b2 (2×4)-reconstructed GaAs(001) surface, including adsorption, desorption , surface diffusion  and nucleation , we show how the molecular beam epitaxy of GaAs can be modeled in atomistic detail. The main results of these investigations are:
 M. Bockstedte, A. Kley, J. Neugebauer, and M. Scheffler, Comput. Phys. Commun. 107, 187 (1997); http://www.fhi-berlin.mpg.de/th/fhimd
 C. G. Morgan, P. Kratzer, and M. Scheffler, Phys. Rev. Lett., 82, 4886 (1999).
 A. Kley, P. Ruggerone, and M. Scheffler, Phys. Rev. Lett. 79, 5278, (1997).
 P. Kratzer, C. G. Morgan, and M. Scheffler, Phys. Rev. B 59, 15246 (1999).
 P. Kratzer, and M. Scheffler, to be published