Materials genomics of thin film strain relaxation by misfit dislocations

Abstract

We summarize the development and implementation of a “process simulator” for modeling thin film strain relaxation by injection of misfit dislocations. The process simulator, initially developed for GexSi1−x/Si(100) lattice-mismatched epitaxy, integrates elasticity and dislocation theory with experimental measurements of kinetic parameters describing dislocation nucleation, propagation, and interactions. This enables predictive simulation of the development of misfit dislocation arrays during growth and thermal annealing sequences. Further, in the spirit of the materials genome initiative, we show how once a relatively complete description is built for one materials system, extension to a related system may be implemented using a greatly reduced data set. We illustrate this concept by translation of the simulator for GexSi1−x/Si(100) epitaxy into predictive simulation for the GexSi1−x/Si(110) system (which has quite different dislocation microstructure and kinetics) using greatly reduced data sets for the latter system and incorporating data refinement methods to extract unknown kinetic parameters. This sets the platform for extension of these methods to a broader set of strained layer systems.