Statistical Approaches to Yield Forecast and Yield Mining of IC Manufacturing

• Main Authors: KU I CHIA, 顧宜佳
• Other Authors: HORNG-SHING LU
• Format: Others
• Language: en_US
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/46091666420537123301

碩士 === 國立交通大學 === 統計所 === 91 === The uncontrolled factors in manufacturing cause variations in the yields of wafers in the IC industry. This thesis contains two sections. In the first part, we discuss the problem of yield forecast and decide the number of wafers to be picked from the wafer bank to meet the order of a customer. For this problem, we study the statistical methods of outlier detection, selection of training sets from historical data, and resampling methods. Empirical studies based on the production data in one IC company are conducted to evaluate the performance. In the second part, we aim to find the malfunction equipments and time periods during the manufacturing process that result in low yields at the end. We investigate the statistical methods of outlier detection, clustering analysis, and classification techniques for this problem. Empirical and simulation studies are used to evaluate the performance. Gaussian mixtures are also discussed and studied as well.