Gauge capability assessment for high-yield manufacturing processes with truncated distribution

Abstract

In this article, an alternative measurement capability assessment based on a nonparametric method is proposed for high-yield manufacturing processes. This method is easy to use and the results have shown that the nonparametric method is robust with respect to measurement errors that follow a truncated distribution. A truncated distribution is defined as any distribution that has finite upper and lower limits. We compare the nonparametric gauge repeatability and reproducibility (GRR) method with the analysis of variance (ANOVA) method and find that the nonparametric method provides a more accurate assessment of the gauge capability in most cases. In general, the ANOVA method tends to give conservative estimates. Consequently, a more precise measurement system may be proposed if commonly used GRR methods are adopted, even though this may be unnecessary. The nonparametric method is also applied to an actual manufacturing gauge system in a semiconductor industry to illustrate its potential usefulness.