Scheduling Single-Arm Cluster Tools With an Equipment Front-End Module Subject to Wafer Residency Time Constraints

Semiconductor manufacturing widely employs cluster tools that comprise three critical components: a vacuum module (VM), a loadlock module (LLM), and an equipment front-end module (EFEM). In operating such tools, LLM plays a pivotal role. Additionally, wafer fabrication in VM faces strict wafer residency time constraints, making the coordination among the modules particularly challenging. This paper addresses the cyclic scheduling problem of single-arm cluster tools with EFEM and wafer residency time constraints. It focuses on the tool efficiency in module cooperation and schedulability. We explore cooperative strategies for the robot operating within VM to access loadlocks, elucidating the influence of EFEM and LLM on VM. Based on these strategies, we establish necessary and sufficient conditions for the existence of a feasible periodic schedule, providing a foundational basis for the systematic analysis. For schedulable scenarios, we derive an efficient algorithm to identify the feasible and optimal schedule in terms of tool cycle time. The performance and efficiency of the proposed algorithm are validated through experimental verification. Note to Practitioners—In wafer fabs, single-arm cluster tools with EFEM are widely adopted. Studies that neglect the impact of EFEM on scheduling cluster tools are often inapplicable to real-world scenarios. Wafer residency time constraints are essential for maintaining wafer quality. As VM and EFEM operate in different pressure environments, the cooperation via LLM makes it challenging to meet residency time constraints. To address this problem, this paper analyzes the impact of cooperative strategies on the tool performance, and the influence of EFEM and LLM on VM as well. Under these strategies, we examine the schedulability conditions, helping engineers understand how processing parameters affect scheduling feasibility. If a feasible schedule exists, an efficient algorithm is derived to find an optimal schedule. Experimental results show the efficiency of the proposed algorithm, making it suitable for embedding into cluster tool controllers for efficient implementation. For cases that are not schedulable, with the results obtained, this work can provide guidance of how to redesign the process such that feasibility can be ensured.