Cycle Time Reduction Through Preventive Maintenance De-clustering

Cycle-time (CT) is one of the key performance metrics for silicon wafer fabrication factories (fabs). The largest component of CT is the time lots spend in queue waiting for processing. A primary cause of queue-time is the variability of toolset availability (V_A). One component of V_A occurs when usage-based preventative maintenance (PM) events are clustered in time, i.e., unevenly spaced. This portion of V_A may be significantly reduced by intelligent PM scheduling to de-cluster PMs. In this paper, we present a case study on how eManufacturing is applied to enable PM spacing. The PM schedules are generated by a MILP-based (mixed-integer linear programming) optimizer. The fab automation systems visually display the solver results to technicians. An advanced simulator-optimizer system has been developed to test this eManufacturing approach. It demonstrated ~20% reduction in V_A and ~30% reduction in CT for a thin film toolset.