Adaptive control of a semiconductor processing plasma

P.C. Russell*, G.R. Jones* and P.J.G. Lisboa

*Dept. Elec. Eng., University of Liverpool

Plasma emissions in the chromatic range are modelled using neural and statistical response surface models. This supports an expert system at the centre of a real-time supervisory control system which monitors the emissions using chromatic sensors and corrects for any deviations from the optimal operating conditions. Closed loop control of fine adjustments in the processing variables was achieved in simulation.

This project was funded by EPSRC and CHELL Instruments.

Illustration of the plasma processing plant and response surfaces.

pH control in a sulphonation loop reactor

P.J.G. Lisboa

Real-time control of the pH concentration in a re-circulating batch reactor is achieved by mapping the desired control response to disturbances, onto a neural network. This results in a hybrid on-line system comprising analytical plant models and three term controllers, as well as neural components. Closed-loop control was successfully tested on Unilever’s pilot plant at Port Sunlight.

EPSRC CASE studentship in collaboration with Unilever Research Port Sunlight Laboratories.

Measured titration curve showing a sever non-linearity around the desired operating point at pH 10.5.