Minimising idle times in cluster tools in the semiconductor industry

Abstract

This paper is dedicated to the problem of sequencing handler's moves in cluster tools in the semiconductor industry. A suboptimal sequence of the handler's moves usually leads to long idle times in the cluster tool's chambers and hence longer makespan and lower throughput. Therefore, an effective algorithm that minimises idle times in the cluster tool's chambers is required. An algorithm called MinBIT (minimising bottleneck idle time) is presented as a new method for sequencing the handler's moves. The MinBIT algorithm gives priority to the bottleneck stage, and thus leads to a reduction in unnecessary idle times and hence a shorter makespan. We compared the MinBIT algorithm with other algorithms. Our algorithm achieved best performance in 98% of the cases. In the other 2% of the cases it achieved second or third best performance. The computational results show that the MinBIT algorithm can improve the performance of a cluster tool by up to 10%. A comparison between the MinBIT solutions and the optimal solutions found by using branch and bound, shows that the deviation between them ranged between 0% and 0.22%, which indicates that MinBIT gives optimal or near optimal results.