產品良率對於企業或工廠而言，是衡量其是否具競爭優勢的重要指標。而現行工廠中在面對產品良率的掌控問題時，大多是藉由製程工程師對製程的瞭解，以經驗挑出可能影響良率的製程參數，再以實驗設計或配合實驗批量的生產，來驗證所選擇之參數是否確實為影響良率的主因。但繁複的晶圓製程中會產生大量的製程參數，且參數間又具有相互的影響關係，故以傳統統計方法分析時困難度較高。因此，本研究將以集群分析配合資料視覺，從晶圓廠LPC（lot in-line process control）資料中，找出影響晶圓良率為低的製程參數。本研究之方法首先是將LPC資料轉換成描述製程品質特徵（process quality feature，PQF）的集合，透過PQF指標的集群分析及視覺化，來粹取對良率具影響力的關鍵PQF指標，再利用這些關鍵PQF指標對晶圓批量進行集群分析後，分析相似的晶圓批量良率表現，研究結果驗證了有相似PQF指標的批量，的確呈現了相同良率之表現。

Yield improvement is critical for an IC manufacturing company to remain competitive in the market. High-yield products not only bring more profit to the company but also indicate better manufacturing skills. Current practice toward yield improvement relies on engineers to select candidate manufacturing parameters and use statistical techniques such as design of experiment to verify the hypotheses. This approach is difficult and less effective due to the large number of parameters and the complicated interactions among them. This research employees clustering and visualization technique to help engineers find possible causing factors of low-yield wafers from the lot in-line process control (LPC) data, which are collected from the metrology machine during IC fabrication. We convert the raw LPC data into process quality features (PQF). With the help of clustering & visualization, the key PQF’s are selected. These key PQF’s are then used to cluster the wafer lots. By analysis these low-yield wafer lots that are grouped into the same cluster, i.e., having similar PQF’s, manufacturing parameters causing the low-yield wafers could be identified.

中文摘要………………………………………………………………………………Ⅰ
英文摘要………………………………………………………………………………Ⅱ
誌謝……………………………………………………………………………………Ⅲ
目錄……………………………………………………………………………………Ⅳ
表目錄…………………………………………………………………………………Ⅵ
圖目錄…………………………………………………………………………………Ⅶ
第一章 緒論……………………………………………………………………...…….1
1.1 研究背景…………………………………………………..…………………..1
1.2 研究動機與目的…………………………………………..…………………..2
1.3 研究方法流程……………………………………………..…………………..2
第二章 文獻探討………………………………………………………………...…….4
2.1 現行於分析半導體產品良率之研究……………………..………………….4
2.2集群分析（Cluster Analysis）…………………………..……………….…..5
2.2.1 集群分析之相似度描述………………………….……………………6
2.2.2 集群分析之方法……………………………………………………….8
2.3 視覺探勘之N-維資料視覺化………………………..…………………….11
第三章 LPC資料與製程品質特徵………………………………………………..…15
3.1 LPC製程資料描述…………………………………………………………15
3.2製程品質特徵指標計算…………………………………….………………16
第四章 關鍵PQF指標粹取…………………………………………………………..20
4.1 PQF集群分類………………………………………………………………20
4.2 PQF指標視覺化設計………………………………………..……………..21
4.3 PQF集群分析結果與視覺化探勘………………………………………....21
4.4 以準則判斷式選取重要PQF………………………………………………25
第五章 批量集群分析………………………………………………………………..30
5.1 重要PQF群組之批量群集分類…………………………………………..30
5.2 關鍵良率影響因子驗證…………………………………………………...32
第六章 結論…………………………………………………………………………..36
參考文獻………………………………………………………………………………38
附錄一 重要PQF原始資料誤差盒圖……………………………………………….40
表目錄
表3.1 qf值設定………………………………………………...……………………...18
表4.1 PQF指標之視覺化……………………………………………………………21
表4.2 準則判斷式選取之重要PQF列表……………………………………………25
圖目錄
圖1.1 研究方法流程圖………………………………………….……………………..3
圖2.1 晶圓圖缺陷晶格之特徵………………………………………………………...5
圖2.2 二維平面上點物件之集群……………………………………………………...6
圖2.3 階層式分類法之樹狀結構圖…………………………………………………...9
圖2.4 電腦檔案系統架構以階層式技術視覺化…………………………………….11
圖2.5 金融資料時間序列像素圖…………………………………………………….12
圖2.6 Icon-Based 技術Stick Figures…………………………………………………12
圖2.7 應用Icon-Based技術於繪製PCB廠在製品狀況…………………………….13
圖2.8 以平行座標法繪製PCB主生產計劃與機器閒置成本資料………………....13
圖2.9 圖像式顯示技術……………………………………………………………….14
圖3.1 展開晶圓製造流程的迴圈………………………………………………….…15
圖3.2 LPC資料描述與製程品質特徵轉換示意圖………………………………….16
圖3.3 規格定義區間劃分示意圖…………………………………………………….17
圖4.1 PQF矩陣視覺化…………………………………………………………….….21
圖4.2 PQF集群分類演算結果……………………………………………………….23
圖4.3 具影響良率之重要PQF群組…………………………………………………24
圖4.4 視覺蒐尋於高低良率間具影響之PQF子群…………………………………24
圖4.5不同選取方式之關鍵PQF視覺化圖形………………………………………29
圖5.1以視覺化選取之重要PQF對批量進行集群分類……………………………31
圖5.2 以準則判斷式選取之重要PQF對批量進行集群分類………………………31
圖5.3 關鍵良率影響因子特徵……………………………………………………….33
圖5.4 關鍵良率影響因子原始資料誤差盒圖………………………………………34
圖5.5 良率影響因子原始資料誤差盒圖…………………………………………….35

Brada, D., and A. Shmilovici, “Data Mining for Improving a Cleaning Process in the Semiconductor Industry,” IEEE Transactions on Semiconductor Manufacturing, Vol. 15, No. 1, 91-101,2002.
Dabbas, R. M., and H. N. Chen, “Mining Semiconductor Manufacturing Ddata for Productivity Improvement ─ An Integrated Relational Database Approach,” Computers in Industry, Vol.45, Issue:1, pp.29-44, 2001.
Duda, R. O., P. E. Hart, and D. G. Stork, Pattern Classification, A Wiley-Interscience Publication, John Wiley & Sons, Inc., New York U.S.A., 2001.
Fan, C. M., R. S. Guo, A. Chen, K. C. Hsu, and C. S. Wei, “Data Mining and Fault Diagnosis Based on Wafer Acceptance Test Data and In-line Manufacturing Data,” IEEE International Semiconductor Manufacturing Symposium , pp.171 —174, 2001.
Fayyad, U., “Data Mining and Knowledge Discovery in Databases: Implications for Scientific Databases,” Ninth International Conference on Scientific and Statistical Database Management Proceedings, pp. 2-11, 1997.
Gardner, M., and J. Bieker, “Solving Tough Semiconductor Manufacturing Problems Using Data Mining,” Advanced Semiconductor Manufacturing Conference and Workshop IEEE/SEMI, pp.46 —55, 2000.
Han, J., and M. Kamber, Data Mining : Concepts and Techniques, Morgan Kaufmann Publishers, San Francisco, 2000.
Hansen, C. K., P. Thyregod, “Use of Wafer Maps in Integrated Circuit Manufacturing,” Microelectronics and Reliability, Vol.38, Issue:6-8, pp.1155-1164, 1998.
Kaufman, L., and P. J. Rousseeuw, Finding Groups in Data, A Wiley-Interscience Publication, John Wiley & Sons, Inc., New York U.S.A., 1990.
Keim, D. A., and H. P. Kriegel, “ Visualization Techniques for Mining Large Databases: A Comparison,” IEEE Transactions on Knowledge and Data Engineering, Vol.8, Issue.6, pp.923-938, 1996.
Leivian, R., W. Peterson, and M. Gardner, “CorDex : A Knowledge Discovery Tool,” Proceedings WSOM’97 : Workshop on Self-organizing Maps, Helsinki University of Technology, pp.63-68, 1997.
Robertson, G. G., J. D. Mackinlay, and S. K. Card, “Cone Trees: Animated 3D Visualizations of Hierarchical Information,” Proc. Human Factors in Computing System CHI’91 Conf., New Orleans, L.A., pp.189-194, 1991.
Tufte, E. R., The Visual Display of Quantitative Information, Published by Graphics Press, Cheshire Connecticut U.S.A., 1998.
陳紀宏，「資料視覺化於生產管理報表資料分析」，碩士論文，元智大學工業工程研究所，2001。
宋保全，「以視覺化資料探勘輔助印刷電路板廠之生產管理」，碩士論文，元智大學工業工程研究所，2000。
郭玟琳，「以資料視覺輔助產能規劃的決策」，碩士論文，元智大學工業工程研究所，1999。
范治民，「晶圓允收測試之統計製程管制設計」，博士論文，國立臺灣大學電機工程研究所，2000。
劉亭宜，「GRNN在晶圓製造裡良率模式之建構與分析」，碩士論文，元智大學工業工程研究所，1999。
黃俊英，「多變量分析-第三版」，中國經濟企業研究所，華泰圖書文物公司，台灣台北市，1988。