반도체 제조 공정에서의 비대칭도와 첨도를 고려한 비정규 분포 관리도

Abstract

It is very common to follow non-Gaussian distribution for many quality characteristics, especially electric ones, in semiconductor manufacturing process. Since traditional Shewhart control chart increases type I error for semiconductor electric quality characteristics, this paper develops a modified control chart considering non-Gaussian factors, skewness and kurtosis. The method calculates control limits as Shewhart control chart, shifts control limits based on skewness, and changes tolerance of control limits based on kurtosis. For data of semiconductor electric quality characteristics, the method has shown smaller type I error and better performance than Shewhart control chart and skewness only correction control chart. The method has been applied also to other data which following non-Gaussian distribution.