Control Development and Modelling for Enhanced Crystal Growth: Application to Novel New Technologies and Extenstion of Conventional Capabilities

Trustees Of Boston University, Boston

Investigators

Abstract

Characterizing these in terms of the control gain can be misleading. The idea of using feedforward of known effects to assist the feedback control is good, but well known. Micro-level material objectives have been indirectly incorporated in the design of controllers for years. However, the idea of explicitly incorporating the effects of heat flow is a good example of the use of modeling to help determine the controller explicitly. Although adaptive control is proposed as an integral part of both the feedback control and the feedforward control, very little is said about adaptive control per se. I have a few comments on the interpretations of the graphs. The text in Figure 11 states that the oscillation in diameter indicates the need for adaptive control. However, without knowing the measurements that determine both the pull rate and the heater power, and the system dynamics, it is not apparent that adaptive control is required. Certainly, Figure 11 by itself is insufficient. Figure 19 shows oscillations of the boule diameter, which is one indication of nonminimum phase behavior. However, the anomaly near the end of the graph appears to be due to a nonlinear effect (or a time variation). This graph, if the anomalous behavior is not uncommon, would present a good argument for the use of adaptive control. The modeling of the CZ and EMCZ appears to be appropriate. Experimentation to validate the models is reasonable, and I expect that it can be conducted on the available puller. In summary, I think the primary contribution of the proposed effort will be the development of models for the CZ and EMCZ processes. There will be no new contribution to control theory (and the proposal doesn't claim any). Although the proposal states that controllers for these processes will be developed, the analysis of the control issues is weak.