next up previous contents
Next: Blind Channel Deconvolution Up: Control Issues Previous: Photoresist Cure Monitoring

Control of Drop-Ejector

Photo-resist dispensers traditionally apply a generous amount of resist on the wafer and then spin the wafer to reach a uniform desired thickness. With this technique, over 95wasted. The goal of this project is to reduce the waste by using a drop on demand ejection technology to apply the photo-resist.

Due to imprecise manufacturing and aging there is a certain variability in the behavior of the micro-machined/macro drop-ejectors. Our objective is to use feedback control to compensate for this variability. A crucial point is to obtain an adequate model suitable for model based control design. The model we use in our work is a second order non-linear differential equation, where especially the non-linear part is uncertain. The goal is to synthesize a controller that makes the closed loop robust with respect to the uncertainty.

Another issue that is addressed is how to automatically tune the controller. This is of importance, since the application requires a whole array of ejectors, and it would be to expensive to tune each individual controller by hand. The idea used is to feed the ejector with a test-signal, measure the response, estimate the parameters in the model, and to compute the controller parameters on-line. Initial results are presented in Roche and Hansson (1997).


Gokhan PERCIN
Tue Sep 1 11:46:26 PDT 1998