Productivity Improvement From Using Machine Buffers in Dual-Gripper Cluster Tools

Abstract

Cluster tools (also referred to as robotic cells) are extensively used in semiconductor wafer fabrication. We consider the problem of scheduling operations in an -machine cluster tool that produces identical parts (wafers). Each machine is equipped with a unit-capacity input buffer and a unit-capacity output buffer. The machines and buffers are served by a dual-gripper robot. Each wafer is processed on each of the machines, and the objective is to find a cyclic sequence of robot moves that minimizes the long-run average time to produce a part or, equivalently, maximizes the throughput.We first obtain a tight upper bound on the optimal throughput and then use this bound to obtain an asymptotically optimal sequence under conditions that are common in practice. Next, we quantify the improvement in productivity that can be realized from the use of unit-capacity input and output buffers at the machines. Finally, we illustrate our analysis on cluster tools with realistic parameters, based on our work with a Dallas-based semiconductor equipment manufacturer.Note to Practitioners-Semiconductor manufacturers often seek methods to increase the productivity of their cluster tools. This paper considers the option of adding single-unit-capacity input and output buffers at each processing chamber of a cluster tool with a dual gripper robot. We demonstrate that this is an effective approach for tools that are transport-bound, i.e., those in which the robot is the bottleneck resource. We also present a sequence of robot movements that provides the optimum throughput for such a tool. The use of these internal buffers is not beneficial for tools that are process-bound, i.e., those in which a processing chamber is the bottleneck.

Bi-objective cyclic scheduling for single hoist with processing-time-window constraints considering buffer-setting

“…And the hoist capacity constraints can be formulated as constraints (8)- (11). In particular, inequality (8) guarantees that the start time of move i in a cycle must be after the hoist traveling time. And inequality (9) guarantees that all the leaving time in a cycle cannot exceed the CT, T. Constaints (10) and (11) indicate that a hoist only transports one part at the same time.…”

“…In addition, other more complicated processing lines are extensively applied to the industry practice, for example, those with several identical tanks, a re-entrant tank, multiple hoists or considering buffers. [4][5][6][7][8] In this article, we deal with HSP in the lines considering buffer-setting.…”

“…Drobouchevitch et al 33 considered a cell in which each machine has a one-unit input buffer and a one-unit output buffer and concluded that few throughput advantages existed in the machine with input buffers over the one with input and output buffers. Geismar et al 8 quantified that the setting of unit-capacity input and output buffers at the machines could improve the productivity and provided an asymptotically optimal cyclic solution.…”

Noncyclic Scheduling of Cluster Tools With a Branch and Bound Algorithm

“…There have also been many studies on scheduling hoists, robotic cells, cluster tools, and multicluster tools without a Petri net [15], [35]- [40]. Che et al [35], [36] have proposed an efficient B&B algorithm for cyclic scheduling of a hoist by analyzing some properties and solving a linear programming (LP) problem for the lower bound.…”

Time optimal supervisory control of discrete event systems

“…[25,46]. Another common approach to improve steady-state, as well as transient behaviors of the cluster tools was achieved by structural analysis [30,54,58,72]. Such an approach was highly dependent on the analyzed cluster tool and its properties and became quite complicated for tools which were complex.…”