Design of a film surface roughness-minimizing molecular beam epitaxy process by reduced-order modeling of epitaxial growth

Abstract

Molecular beam epitaxy of germanium was used along with kinetic Monte Carlo simulations to study time-varying processing parameters and their effect on surface morphology. Epitaxial Ge films were deposited on highly oriented Ge(001) substrates, with reflection high-energy electron diffraction as a real-time sensor. The Monte Carlo simulations were used to model the growth process, and physical parameters were determined during growth under time-varying flux. A reduced version of the simulations was generated, enabling the application on an optimization algorithm. Temperature profiles were then computed that minimize surface roughness subject to various experimental constraints. The final roughness after two layers of growth was reduced to 0.32, compared to 0.36 at the maximum growth temperature. The study presented here is an initial demonstration of a general approach that could also be used to optimize properties in other materials and deposition processes.