Thin Film Growth Simulation Using Cellular Automata State Space, and Neural Nets Methods

摘要

The objective of this research was to demonstrate simulation of multi-species thin film growth that is fast, displays on a desktop computer, and utilizes a unique combination of cellular automata, state space, and neural nets. Software to accomplish this objective was refined by developing computer code that implements a physical model of film growth. Neural nets were into rated into the software to reduce the time needed for computing the atom positions, and generation of exemplars needed or training and testing the neural nets was accomplished via visual basic code. Display of the simulation using a desktop computer (Macintosh Power PC) was achieved. Functionalities for live rotation and modification the film are included, as well as the capability to record the simulation as a computer movie. Study of vacancy behavior using the simulation of about 30, believed to be a result of scaling issues in used model. This software can be used to simulate film growth for molecular beam epitaxy and pulsed laser deposition processes. Physical and chemical model studies and process refinement studies can be accomplished using this software.