The control of particulate contamination is integral to many semiconductor manufacturing processes. In these processes, particle contamination levels are measured and compared with performance standards and statistically analyzed to determine the process state. Such particle data often exhibit a high relative standard deviation. This impacts the ability to make appropriate decisions about the state of process cleanliness. Further compounding the problem is that traditional statistical tools (e.g., normality assumption, Shewhart control charts) may perform poorly when applied to particle data. This paper presents an overview of many key issues surrounding the application and misapplication of statistical methods to particle data.

This content is only available as a PDF.
You do not currently have access to this content.