Nital S. Patel

Adaptive optimization of run-to-run controllers: the EWMA example

This paper presents a recursive scheme for optimizing the gain of an exponentially weighted moving average (EWMA) controller under stability constraints. The objective is to minimize the asymptotic mean square error in the output with minimal a priori information. The algorithm hinges on a simple representation of the optimal EWMA gain. Both step and drift disturbances are considered. It is shown that the gain sequence generated by the algorithm always yields a stable system. Furthermore, this sequence is shown to converge to a suboptimal value. Extensions to the algorithm to the case where there is model uncertainty are also presented. The algorithm is verified via simulation. Data from a manufacturing implementation are presented.

Device dependent control of chemical-mechanical polishing of dielectric films

This paper presents a control scheme for run-to-run control of chemical-mechanical polishing (CMP). The control scheme tracks both device pattern dependent and equipment induced disturbances. The structure of the controller is such that sensitivity to qual (unpatterned blanket oxide) wafer frequency is minimized. Additionally, prethickness variation and metrology delay are accounted for in the design. Results from applying this scheme in volume production are presented.

In situ estimation of blanket polish rates and wafer-to-wafer variation

Presents a scheme for extracting information from interferometry signals off patterned wafer polish during nonendpointed chemical-mechanical polishing (CMP). This enables one to obtain blanket removal rates immediately after a patterned wafer finishes polishing as well as estimate post-polish within lot thickness variation. A nonlinear regression algorithm is presented that enables one to estimate this information from less than a full interferometry trace cycle and with a lower sampling rate compared to peak-valley detection schemes.

Lot allocation and process control in semiconductor manufacturing - a dynamic game approach

This paper proposes a framework to study lot allocation (dispatch) policies for improved process control. Minimizing the sensitivity of overall process performance to process and inventory disturbances is proposed as a performance criterion. The specific policy presented considers process performance in making lot allocation decisions, and introduces soft dedication of machines in case of low-running material. Lastly, this policy is non-idling, and is driven by the contents of a buffer controlled by a daily planning system, or a dynamic scheduling algorithm. It is hoped that this paper will lead to additional studies on the interaction between scheduling/dispatch and semiconductor process control, especially in high-mix environments.