A computing environment for spatial data analysis in the microelectronics industry

This paper describes a computing environment called S-wafers that is tailored for the analysis of spatial data collected during semiconductor manufacturing processes. At the core of S-wafers lies a new statistical methodology that systematically exploits the basic spatial nature of these data. The S-wafers environment builds on the experience of Lucent Technologies' Microelectronics engineers and Bell Labs researchers who have noticed that patterns in mapped wafer data can provide ''signatures,'' which can be used to help identify and correct process problems. The S-wafers environment provides the means to formalize these observations into implementable strategies by furnishing extensive fools for data visualization; identification of groups of similarly patterned wafers; spatial analysis of designed experiments; assessment of the impact of particle contamination; and much more. In addition to supporting statistical research into problems of process improvement in this area, the S-wafers environment has been used for more than two years by various groups in Microelectronics. Extensions to both the software environment and the underlying statistical methodology continue at a rapid pace.