控擋片回收製程之改善方案的選擇

Abstract

控擋片回收製程之改善方案的選擇 研究生：簡嘉賢 指導教授：巫木誠 博士 國立交通大學工業工程與管理研究所 摘要 晶圓廠製造過程中，幾乎所有機台都會使用控擋片作為監測、維護機台環境的輔助工具。控擋片可藉由清洗回收、降級和研磨回收以重複使用，來減少全新控片的投入量。而目前研究中多為控擋片降級法則之改善，以降低全新控片投入量，該種研究屬於營運法則之改善，甚少探討有關清洗回收之體質改善的問題。 本研究將探討控擋片清洗回收製程改善時，應該優先選擇那些控擋片的清洗回收製程作改善和改善多少，可以讓全新控片投入量減少最多，以使改善成本發揮最大的效益。本研究透過基因演算法和邊際分配法之協助，可以針對不同的改善預算，找到近似最佳解的改善方案。在比較搜尋時間方面，邊際分配法遠較基因演算法來得快。 關鍵詞：控片、擋片、清洗回收、基因演算法、邊際分配法

lection of Alternatives for Improving Recycle Processes of C/D Wafers Student：Chia-Shien Chien Advisor：Dr. Muh-Cherng Wu Institute of Industrial Engineering National Chiao Tung University ABSTRACT Control/dummy (C/D) wafers are used to measure the process quality in semiconductor manufacturing. C/D wafers are reusable by recycle cleaning and can be downgraded to the downstream processes. Previous studies on C/D wafers aimed to reduce the use of C/D wafers by making appropriate downgrade decisions. Yet, the yield improvement in the recycle process C/D wafers is seldom explored. This research formulates a decision problem and proposes solution methods for the selecting the yield improvement alternatives in the C/D recycle processes. The decision problem is to determine the yield improvement targets for each recycle process in order to minimize the use of C/D wafers, under a given budget for yield improvement. The two solution methods involve a genetic algorithm and a marginal allocation algorithm. The two methods yield same solutions but the marginal allocation method is good in requiring much less computation. key word：control wafer, dummy wafer, recycle cleaning, genetic algorithm, marginal allocation