Development of a computer control system with self-tuning growth rate control for the Czochralski crystal growth process.

摘要

This dissertation is a presentation of the development of an advanced computer control system for the Czochralski crystal growth system in the crystal growth laboratory at the Center for Research and Education of Optics and Lasers, University of Central Florida.; Czochralski crystal growth process is very sensitive to disturbances in nature. Frequent human interference is required in order to achieve a successful growth run and to obtain a crystal of high quality. A computer control system provides crystal growers a much more friendly environment than the commercial digital controller for monitoring the process. It also allows the growers to be able to analyze the process in detail according to the collected data after the growth runs, and thereafter improve their process strategies. Automation is increased in the computer controlled growth system. Several crystals have been grown with the new system.; Computerization makes it possible to apply modern control theories to the control of the crystal growth. To date much is still unknown about the dynamics of crystal growth. On the other hand, it is known that growth dynamics vary greatly with growth conditions. Therefore adaptive control is desirable. The self-tuning control (STC) is implemented in the system for the control of growth rate which determines crystal uniformity. The self-tuning controller applies the generalized minimum variance (GMV) strategy with the recursive-least-squares identification algorithm (RLS) as the on-line parameter estimator. Successful growth runs have been achieved while growth rate is under the control of the GMV STC.; This is the first time that self-tuning control has been implemented in an actual crystal growth system. The results are promising. Comparing the crystals yielded by the experimental runs, we see that the crystals grown under the control of the GMV STC have fewer striations then those grown under PI control. This indicates significant improvement, although to this end, we cannot conclude that the crystals grown with the advanced controller will all be superior to those grown with conventional controller. It shows that further research and development of adaptive control in crystal growth process is worthwhile. Actually we have made the software in such a way that any new discrete control strategy can now be easily implemented and tested for the crystal growth process in this system.