New sequences of Dual-Armed cluster tools with long cleaning times and periods and reducing wafer completion interval variability

Abstract

In semiconductor manufacturing, the schedule of chamber cleaning plays a crucial role in preventing failures in wafer quality and the wafer completion interval. This thesis centers around scheduling a dual-arm cluster tool featuring a parallel wafer flow pattern with K-periodic chamber cleaning. Specifically, it investigates a scenario where the previous All Backward Cleaning(ABC) sequence fails to achieve optimality and the variability of wafer completion interval times. Consequently, a modification of the ABC sequence that divides the cleaning into a different work cycle is proposed. The Half-ABC sequence is proposed for handling much longer cleaning time than processing time. This modification involves dividing the cleaning chambers into two distinct work cycles. We utilize the Timed Event Graph(TEG) and define the cycle time by calculating the circuit ratio that exists in circuits in TEG. The Half-ABC sequence demonstrates superior performance in minimizing the cycle time and achieving optimality in a larger area. At the same time, a similar approach is proposed to handle the variability of wafer completion interval times by dividing the cleaning chambers into d number of work cycles and spreading the divided cleaning operations across a range of k work cycles. The performance of these proposed sequences was evaluated by comparing the standard deviation of the wafer completion interval. The experimental results revealed that the proposed Spread One-to-One ABC sequence exhibited the smallest variability in wafer completion interval.