Run-by-run estimation-based control of meniscus coating

This paper describes the application of process control techniques to the meniscus coating process. Meniscus coating may be used to deposit polymers used in integrated circuits and electronics packages. Three control schemes are presented and each employs Kalman filtering to identify process coefficients. The three schemes differ according to the process model and amount of process information that is available. The dominating process parameters are applicator speed and material viscosity which affect the output of film thickness. The first scheme addresses the case where the least amount of information is known; viscosity is not measured. Identification and control is applied to a meniscus coater that does not have hardware to sense viscosity. In the second control scheme, viscosity sensors are used. However, limitations of the process model and the nature of the process produce controller lag in the presence of a shift in the process. This lag is undesirable in a manufacturing setting, therefore the third scheme is introduced which improves upon the second by reducing lag. The third scheme employs curves fitted to a history of process data to reinitialize the coefficient estimates when there is a shift in the process.