Conventional molecular beam epitaxy (mbe) growth control systems are either operated manually or with simple linear controllers using predefined recipes. We will supplement a standard control system with a neural network based controller. The neural network will be taught by growing gaas/algaas superlattices using the conventional techniques. Subsequently, the neural network will be allowed to take full control of the growth process. Evaluation of the performance of the neural network will be based on a comparison of superlattices grown by conventional and neural techniques. A neural network based mbe growth control system will greatly enhance the quality and repeatability of iii-v film characteristics. This will lower device costs and improve device performance.
Keywords: Neural, Network, Optoelectronic, Control, Epitaxy, Superlattice
