A POLE-PLACEMENT ADAPTIVE-CONTROL OF A RAPID THERMAL PROCESSOR

Rapid thermal processing (RTP) techniques are often applied in microel ectronic technology. The main characteristics of RTP techniques lie in single wafer treatments, fast thermal kinetics at low thermal cost an d reduced contamination risks; moreover, many types of reaction can be realized with one single wafer treatment. Problems of measurement and control of space-time temperature profiles have not yet been solved i n a satisfactory way; in particular, classical control algorithms (suc h as PID control) can give bad dynamics to the system if nominal tempe rature profiles must be modified. Modelling of the thermal behaviour o f the process leads to very complex mathematical models. These are the reasons why adaptive control techniques are chosen. A CARIMA model is an interesting input-output model of the thermal processor, whose par ameters can be estimated on-line, using an identification scheme which includes supervisory actions. The adaptive control algorithm is a ful l-state feedback pole-placement law with an adaptive observer. A parti al-state reference model was included in the control law algorithm. Th is type of control law can be implemented even for non-minimum-phase s ystems, time-varying dynamics, a wrong assumption about the time delay of the system, and when disturbances on the load exist. Experimental results are presented which show a better performance when using an ad aptive control strategy.