The application of experimental design to the analysis of semiconductor manufacturing lines

Abstract

Flexible, interactive software is used to apply a two-level fractional factorial design with ten factors to the discrete event simulation of a semiconductor manufacturing line. The sensitivity of cycle time to the number of tools and the operator-to-tool ratio at five tool groups that tend to have large queues preceding them is analyzed. The analysis confirms the proposed model for cycle time and illustrates the efficiency of using design of experiments for discrete event simulation. A review of ancillary output data from each run corroborates the results of the experimental analysis, namely that tool effects predominate. It reveals that the main effects for the tools are ordered according to utilization.