27. Experimentation for Equipment Reliability Improvement

Abstract

To be fit for use by a customer, semiconductor process and test equipment must perform reliably under real-world sources of variation in wafer parameters, setup modalities, and environmental conditions. Using design of experiments methodology, an experiment was conducted on the robotic wafer transfer and alignment sub-system of an ESI Model 9000 automated memory repair system. The experiment identified design factors related to a serious reliability problem in the operating environment of a particular customer. The results demonstrated how the wafer handling sub-system could be improved to avoid the effects of environmental factors representing customer usage of the system. This case study describes the experimental design, data analysis, and implementation of the results which achieved a reduction from an average of seven wafer handling assists per 1000 system hours prior to the experiment to zero occurrences after implementation of a design change. Executive Summary Problem In December 1990 a key Electro-Scientific Industries (ESI) customer in Japan informed ESI of their dissatisfaction with the mean-time-between-failure (MTBF) performance of their Model 9000 systems. In the customer's estimation, ESI's competitor provided superior reliability performance; therefore, improvement in MTBF was required if ESI was to remain a vendor of choice for automated memory repair equipment. ESI was given six months to respond with a solution.