Share Email Print

Proceedings Paper

Improved estimates of the range of errors on photomasks using measured values of skewness and kurtosis
Author(s): Henry Chris Hamaker
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Statistical process control (SPC) techniques often use six times the standard deviation sigma to estimate the range of errors within a process. Two assumptions are inherent in this choice of metric for the range: (1) the normal distribution adequately describes the errors, and (2) the fraction of errors falling within plus or minus 3 sigma, about 99.73%, is sufficiently large that we may consider the fraction occurring outside this range to be negligible. In state-of-the-art photomasks, however, the assumption of normality frequently breaks down, and consequently plus or minus 3 sigma is not a good estimate of the range of errors. In this study, we show that improved estimates for the effective maximum error Em, which is defined as the value for which 99.73% of all errors fall within plus or minus Em of the mean mu, may be obtained by quantifying the deviation from normality of the error distributions using the skewness and kurtosis of the error sampling. Data are presented indicating that in laser reticle- writing tools, Em less than or equal to 3 sigma. We also extend this technique for estimating the range of errors to specifications that are usually described by mu plus 3 sigma. The implications for SPC are examined.

Paper Details

Date Published: 8 December 1995
PDF: 10 pages
Proc. SPIE 2621, 15th Annual BACUS Symposium on Photomask Technology and Management, (8 December 1995); doi: 10.1117/12.228171
Show Author Affiliations
Henry Chris Hamaker, Etec Systems, Inc. (United States)

Published in SPIE Proceedings Vol. 2621:
15th Annual BACUS Symposium on Photomask Technology and Management
Gilbert V. Shelden; James N. Wiley, Editor(s)

© SPIE. Terms of Use
Back to Top