Skip to Main Content
The benefits of run-to-run control are commonplace and well known: improved process and device performance, increased process tool utilization, improved throughput, reduction of non production (pilot) wafers, etc. Likewise, the benefits of intelligent scheduling and dispatching methods have been shown to reduce cycle time and improve resource utilization. In this paper, we will consider the potential additional benefits that can be obtained through sharing information between a run-to-run control system and a fab dispatch system. Many factors can be taken into consideration when deciding how to dispatch material to process resources-equipment availability, equipment states, time constraints, due dates, bottleneck resources, etc. However, in practice, the decision for lot-to-tool assignment is often based on identifying the highest priority lot and the next available tool. For a process area that has deployed run-to-run control, a number of factors available from the control system can enhance the effectiveness of the dispatch tool. Likewise, the selection of equipment for a given process can affect a controller's efficiency. A controller's state is commonly partitioned by context, such as process tool, product, layer, etc. When this information is used in the dispatch decision, the material can be assigned to a tool with an existing or recently updated partition. The wrong dispatch decision can result in an unnecessary look-ahead run, especially in high-mix environments. This paper will discuss the impact of control-oriented dispatching on factory performance metrics. We describe simulation results where traditional dispatching techniques are compared with a process metrics-based approach, using cycle time and utilization as the key metrics, based on actual production data from a chemical-mechanical planarization process area. We describe a single-resource constrained (process tools only) process and suggest possible enhancements. We illustrate how dispatching can- impact batch processes, process cycle time, and controller performance. Practical considerations for integrating these types of systems, such as scalability constraints, will also be discussed.