Design and Control Using Stochastic Models of Deposition Reactors

The financial feasibility of the creation of a start-up company to sell software developed for the optimization and in-line control of thin film growth in deposition processes was investigated. An analysis of the current marketplace revealed potential for a small start-up company to be competitive with this novel product. The investigation concluded an IRR of 20% for a five year period before possible sale of the company. The kinetic Monte Carlo method was employed as the basis for all simulations in this work. This method retains atomic scale information while enabling simulation of process relevant features such as roughness, growth rate and efficiency. A model predictive controller was designed to reproducibly generate thin films with desired properties under a variety of initial condition disturbances for both single component and multi component systems. The substrate temperature and gas flux were employed as control variables. The control algorithms were investigated using a sensitivity analysis and shown to be robust under a wide range of conditions.