Performance Evaluation of Serial Photolithography Clusters: Queueing Models, Throughput and Workload Sequencing

Abstract

For clustered configuration of a photolithography toolset, operating under a scheduling policy inducing serial processing, measures of system performance are deduced. Queueing models demonstrate that, due to the parallelism inherent in the system configuration, the normalized cycle time behavior is different than that of the standard single server queue. Cluster throughput is evaluated based on measures of the frequency and magnitude of events common in manufacturing operation. It is shown that the maximum throughput of a serial photolithography cluster tool is not influenced by the order in which two classes of lots with different wafer processing speeds are processed