近年來不斷有新的技術應用在半導體製程上，在應用方面也使其設計出具有輕薄短小特質的電子性相關產品，並廣泛的為大眾所使用，帶動整個電子資訊工業的蓬勃發展，使半導體成為現代工業極重要的產業之一。在半導體繁複的製程中，晶圓上的晶方由於某些製程問題而導致的功能異常，其發生製程的原因與特定之晶圓允收測試參數（WAT）間有著密不可分的關係與影響，而其關聯性因為半導體製程步驟繁複且參數間存在複雜的交互效應，然而，工程師往往無法在短時間內有效的將其異常的問題找出，而造成良率的損失。因此，本研究目的即針對半導體中晶圓良率值與晶圓允收測試參數間之關聯性，提出一關聯性特徵擷取分析架構與步驟，利用決策樹中CART演算法針對Bin值良率異常的晶圓找出特定WAT參數的影響特徵，並瞭解Bin值表現與WAT參數之間的關係，以提供工程師進行製程監控與良率改善的依據。本研究實際收集新竹某半導體廠的製程資料，共取得50片晶圓的詳細資料進行分析，其實證結果發現有11筆異常資料可以正確找出晶圓良率值與晶圓允收測試參數間之關聯性，且其異常問題分類結果可信度高達100%，準確率為84.6%。透過本研究所提出的關聯性特徵擷取方法可以有效的找出存在於晶圓良率值與WAT參數之間複雜的關聯性問題，並提供工程師進行製程監控與後續之良率改善的依據。

In recent years, new technologies have been designed in semiconductor fabrication and led to design delicate electronic product widely used by public in application purpose. Moreover, new technologies make electronic industry a skyrocket and one of the most important industries. Due to the complexity of semiconductor manufacturing, engineers make efforts to trace causes of chip malfunctions during the process, and the analysis of correlations between WAT parameters and Bin values are one of the promising approaches to find out the root causes.. The relation between them is very complicated due to the complex fabrication and therefore, engineer can not find out the problem in a short time. This research focus on the analysis of relationship between abnormal procedure and WAT parameters. A Decision Tree algorithm is presented to analyze abnormal wafer lots in batch mode to find out valuable information, and then collects relative traits of WAT parameters to realize the relationship between Bin values and WAT parameters. Thereafter, specific WAT parameters that may cause bad Bin values under certain situations can be determined and then engineers can be able to monitor and diagnose following manufacturing process. This research collects 50 samples of wafers for the analysis. The experiment results show the accuracy of presented approach is 84.65.With the extraction method of the relation trait introduced by this research, we can effectively find out the complex relation between yield and WAT parameters and provide some useful information for engineer for further monitoring and yield improvement.

刁建成，「ULSI製程技術」，全華出版社，2000。

李偉傑，「半導體之工程資料與診斷系統」，碩士論文，國立清華大學工業工程與工程管理研究所，1996。

李培瑞，「半導體製程資料挖礦架構、決策樹分類法則及其實證研究」，碩士論文，清華大學工業工程與工程管理研究所，2001。

林演智，「以工程資料為基礎之半導體故障分析系統」，碩士論文，國立清華大學工業工程與工程管理研究所，1998。

林鼎浩，「建構半導體製程資料挖礦架構及其實證研究」，碩士論文，國立清華大學工業工程與工程管理學系研究所，2000。

黃大倫，「半導體製程良率提升專家系統之失效模式與效應&晶圓允收測試異常分析」，碩士論文，清華大學工程與系統科學系研究所，2001。

蔡明正，「半導體製程良率提升專家系統之故障樹建構與分析及專家系統之驗証」，碩士論文，清華大學工程與系統科學系研究所，2001。

蔡智政，「應用CART決策樹與資料視覺技術於低良率晶圓成因探討」，碩士論文，元智大學工業工程與管理學系研究所，2002。
游智翔，「多重bin層良率群集分析之研究」，碩士論文，國立清華大學統計學研究所，2001。

劉博文，「ULSI製程技術」，新文京開發出版社，2003。

盧文彬，「半導體晶圓允收測試之工程資料分析與診斷研究」，碩士論文，國立清華大學工業工程研究所，1997。

簡禎富、林鼎浩、彭誠湧、徐紹鐘，「建構半導體晶圓允收測試資料挖礦架構及其實證研究」，中國工業工程學刊，18(4)，37-48 ，2001。

鍾依芸，「行動電話系統業服務品質滿意度之研究─應用統計分析與決策樹法」，碩士論文，元智大學工業工程與管理學系研究所，2004。

Berry, M. J. and G. Linoff ,“Data Mining Techniques for Marketing ,Sales and Customer Support”, Wiley Publishing, 1997.

Brada, D., and A. Shmilovici,“Data mining for improving a cleaning process in the semiconductor industry,”IEEE Transaction On Semiconductor Manufacturing, Vol.15, No. 1, pp. 91-101, 2002.

Breiman, L., Friedman, J. H., Olshen, R. J., and Stone, C. J., “Classification and Regression Trees,” 1984.

Chih-Min Fan, Ruey-Shan Guo, Chen A., Kuo-Ching Hsu and Chih-Shih Wei,“Data mining and fault diagnosis based on wafer acceptance test data and in-line manufacturing data,”2001 IEEE International Semiconductor Manufacturing Symposium , pp. 171-174, 2001.

Cunningham, S. P., C. J. Spanos, and K. Voros, “Semiconductor yield improvement: Results and best practices,” IEEE Transactions Semiconductor Manufacturing, Vol. 8, pp. 103-109, 1995.

Hastie, T., J. Friedman, and R. Tibshirani,“The elements of statistical learning,”Springer-Verlag, 2001.

Hummel, M., Federl, P., “Analysis of electrical test data using a neural network approach,” IEEE /SEMI Advanced Semiconductor Manufacturing Conference and Workshop 1997, pp. 37-41, 1997.

Kass, G.,“An exploratory technique for investigating large quantities of categorical data,”Applied Statistics, Vol. 29, No. 2, pp. 119.127, 1980.

Koppenhoefer, B., Wuerthner, S., Ludwig , L., Rosenstiel, W., Kuge, H.-H., Sharma, V., “A new modeling methodology combining engineering and statistical modeling methods : a semiconductor manufacturing application,” Massachusetts Institute of Technology, 1996.

Mingers, J., “An Empirical Comparison of Selection Measures for Decision-Tree Induction,” Machine Learning, Vol. 3, pp. 319-342, 1989.

Patton, R., Frank, P. and Clark, R., “Fault Diagnosis in Dynamic System: Theory and Application,” Prentice Hall International, 1989.

Quinlan, J.R., “Induction of Decision Trees,” Machine Learning, Vol. 1, pp. 81-106, 1986.

Rhoads, T.R., Stanley, J.D.,“Multivariate statistical methods for modeling and analysis of wafer probe test data,”IEEE Transaction On Semiconductor Manufacturing, Vol. 15, No 4, pp. 523-530, 2002.
Runyan, W, R., and Bean, K.E., “Semiconductor Integrated Circuit Processing Technology,” Addison-Wesley, 1990.
Safavian, S.R. and Landgrebe, D., “A Survey of Decision Tree Classifier Methodology,” IEEE Transactions on Systems, Man, and Cybernetics, Vol. 21, No. 3, pp 660-674, 1991.
Skinner, K.R., Montgomery, D.C., Runger, G.C., Fowler, J.W., McCarville, D.R., Tsuda, H., Shirai, H., Takagi, O., Take, R.,“Yield analysis and improvement by reducing manufacturing fluctuation noise,” The Ninth International Symposium on Semiconductor Manufacturing, pp. 249-252, 2000.