Improving line yield at Fab17

Abstract

Fab 17 was a Digital Equipment Corporation semiconductor manufacturing and development facility. In June of 1998, Intel purchased Fab 17 from Digital. Intel shifted the focus to manufacturing and implemented many Intel operation policies. Intel sold equipment back to Digital and reorganized the manufacturing organization. As a result of Intel's initiatives and increasing demand for Fahl 7 products, Fab 17's profitability was improved. Despite improved profitability, Fab 17 was not meeting line yield performance goals. This thesis examines line yield excursions at Fab 17. Root causes are identified, and recommendations are made. The thesis first analyzes the line-yield loss data at Fahl 7 from a macro perspective. Inexperienced technicians and multitasking are found to be associated with most line yield incidents. The thesis then studies line yield excursions in high leverage functional areas in detail. Two major root causes are identified. First, in the reorganization, many technicians switched to new roles and had to be retrained. The training was rushed. Due to low level of automation, processing wafer at Fab 17 requires experience and proficiency. Fab 17's wafer starts increased in the first two quarters of 1999. Technicians had to rush to process the wafers. This resulted in high stress, which leads to high probability of line yield excursion. The inexperienced technicians are even more prone to line yield excursions under high stress. In addition, Fab 17 lacked a minimum staffing policy. Often an inappropriate number of technicians are present on the manufacturing floor. This necessitates multitasking, which creates higher stress and leads to higher probability of line yield excursions. The thesis recommends operations policies to address these issues and reports the actual implementation and some preliminary results.