Automated procedures for characterizing specific equipment productivity losses with applications in the semiconductor manufacturing industry

This dissertation develops techniques and procedures for characterizing specific productivity losses of complex manufacturing equipment. Cases are taken from the semiconductor industry, and procedures are developed for its particular needs. Based on a survey of improvement paradigms, industry standards, and data technology, it is generally concluded that although ample diagnostic technology exists, standard procedures or techniques do not seem to be universally understood nor applied, especially to the problem of characterizing equipment rate efficiency losses.; To help formalize equipment productivity measurement, techniques are presented for modeling equipment processing sequences as activity-on-node networks of operational elements, where sequence duration and path slack for elements may be determined using ordinary critical path calculations. Given detailed equipment data of sufficient integrity, this procedure of generating network models may be entirely automated.; To help direct specific rate efficiency improvement efforts, two algorithms are presented. First, an elemental criticality algorithm is proposed that uses non-aggregated equipment data to automatically characterize the effects of elemental productivity losses within equipment sequences on the overall sequence duration. A field test of this algorithm successfully exposed losses and suggested specific areas for improvement. However, due to observed data integrity problems, significant manual data conditioning was required in order to obtain results.; Second, a resource-constrained scheduling algorithm is introduced, which builds on techniques developed by Talbot, that automatically measures certain rate efficiency losses due to sub-optimal equipment software and hardware design. Although a field test of this algorithm showed these losses to be insignificant for cluster tools that were dedicated to only one internal process mute, these losses are likely to be significant for cluster tools that simultaneously perform multiple internal process mutes.; In conclusion, the only apparent obstacle to complete automation of the analysis procedures presented is the poor integrity of the automated data collection systems observed. An elemental orientation concept is presented along with recommendations for shared responsibility between equipment owners and equipment vendors for overcoming this obstacle.