Author： Besse P.
Publisher： Taylor & Francis Ltd
ISSN： 0361-0918
Source： Communications in Statistics: Simulation and Computation, Vol.35, Iss.2, 2006-04, pp. : 479-496
Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.
Abstract
This article presents a reliable method for highlighting a defective stage within a manufacturing process when the existence of a failure is only known at the end of the process. It was developed in the context of integrated circuit manufacturing, where low costs and high yields are indispensable if the manufacturer is to be competitive. Change detection methods were used to point out the defective stage. Two methods were compared and the best chosen. Thanks to this approach, it was possible to solve some yield problems for which the engineers' investigations were far from the real cause of failure. However, there is a strong requirement to assess the reliability of the suspicions cast on the incriminated stage, otherwise engineers could be made to do useless work and time could be wasted looking into events that are not the true cause of failure. Two complementary tools were implemented for this reliability assessment and their efficiency is illustrated by several examples.
Related content
Access to resources
Recommend
Bootstrapping Sequential Change-Point Tests
Sequential Analysis, Vol. 27, Iss. 3, 2008-07 ,pp. :