Scheduling cluster tools using filtered beam search and recipe comparison

Cluster tools have gained a lot of importance in today's semiconductor manufacturing. A cluster tool basically consists of several processing chambers in a mainframe, several load locks to insert wafer lots and a robot arm to move them. This means that these tools are able to work on more than one lot at the same time. Since the lot combination processed together can have an influence on the cycle times of these lots, scheduling is needed to ensure that the overall cycle times are kept low. In a previous work, we presented a method based on filtered beam search using slowdown factors as evaluation methods. Here, we present another evaluation method based on recipe comparison that produces even better results. We also show results of a beam width parameter study.