Skip to Main Content
It is widely understood that metrology (MTY) is a critical part of the semiconductor manufacturing process. It accelerates yield enhancement and maintains yield performance in both new and mature processes. In recent years, the semiconductor industry has increased its focus on this significant part of the semiconductor manufacturing process. Often we find papers and publications focused on MTY's technological advancements such as MTY integrated equipment and their cycle time impact. The same can be said about the topic of real time dispatching (RTD). Real time dispatching has been intensively studied since the 1980s and publications still continue to flow. Most publications in the past have focused on order release methods, dynamic dispatching rules, and scheduling decisions because of its major productivity improvement and general ease of implementation. However, unlike this previous industry focus on either topic, this paper concentrates not only on MTY's cycle time impact, but the impact that non-MTY integrated equipment has on primary processing tools dependent on MTY feedback. The paper presents an RTD based solution and methodology used to increase capacity, decrease cycle time, reduce downtime and improve process stability in CMP primary processing equipment influenced by metrology sampling controls. A set of dispatching heuristics were developed and analyzed to examine their impact on various performance and productivity limiters in CMP primary processing equipment induced by metrology sampling requirements. These heuristics included a specific set of dispatching  and scheduling  rules, as well as a method to addressing interactions and tradeoffs among potentially productivity competing factors in CMP. RTD has been used in several and different applications, from sorting a work center's dispatch list to choosing a destination in the automated material handling system (AMHS). In this case, RTD applies two separate heuristic algorithms. The first algo- rithm sorts a dispatch list which is displayed to the operator. The second algorithm actively controls sampling by skipping lots which have been determined to not require metrology.