臺灣博碩士論文加值系統

半導體製程改善在過去幾十年來的快速發展，許多現代科技得以快速研發並導入市場，半導體產業也因此被視為現代科技產業的指標產業；而半導體製程之良率，可說是反應此產業中最關鍵的企業獲利指標。簡單來說，良率可以定義成產出良品佔所有投入生產總數的百分比；而良率管理乃是指半導體製造過程中，對所產生的海量資料予以整合分析，以達對良率的改善與預測等，整個相關過程之總稱。而在晶圓製程中，離子佈植的正確程度，則是影響良率好壞的重要關鍵。
本論文利用倒傳遞類神經網路(BP-ANN)所具有之學習、容錯等優點，對離子佈植機(Ion Implanter)佈植的正確程度，建立了一套有效的預測模式。我們首先將離子佈植機機台各功能模組實際量測的電性參數值，經由統計分析與迴歸分析，作資料的前置處理，篩選出可作為倒傳遞類神經網路預測模式的輸入變數，以有效控制類神經網路建構的複雜度與準確度。接著我們利用倒傳遞類神經網路，對離子佈植機佈植結果判定的指標變數，建立準確的預測模式。根據此指標變數值可判定離子佈植機佈植的結果屬於：正常(Normal)，不良(Bad)，或輕微(Slightly)三者之一。本論文對於有否作資料前置處理，對佈植結果預測所造成的誤差，提出具體的比較分析；發現若未進行資料前置處理，將造成預測結果的嚴重誤差。經測試驗證本論文所提出之預測模式的準確度，在經過資料前置處理後，其預測值與實際值之間的誤差可接近0.4%。經由類神經網路的敏感度分析，本論文更進一步找出影響離子佈植機佈植結果的重要電性參數。

In the past few decades the rapid development of semiconductor manufacturing process improvements has led to the speedy growth of many modern industries and imports them into market. Semiconductor industry has therefore been regarded as an indicator of modern industries of science and technology. The semiconductor process yield is one of the most crucial indicators to the corporate profits of this industry. The yield can be defined as a percentage of success of the total number of all production. The yield management but rather refers to the general term for the entire relevant process to the integrated analysis of the mass data produced by the semiconductor process, and to the achievements of yield improvements and predictions. While in the wafer manufacturing process, the correct degree of ion implantation, it is the key to good or bad influence of the yield.
Using the back-propagation artificial neural network (BP-ANN) with its advantages of learning and fault tolerance, we build an effective prediction model to forecast the implantation correct level of the Ion Implanter. We first let the actual measured electrical parameter values of various functional modules of Ion Implanter to go through statistical analysis and regression analysis for data preprocessing. The screened result of the data preprocessing can be used as the input variables to the BP-ANN model to effectively control the complexity and accuracy of the constructed prediction model. Then we build a BP-ANN model to accurately predict the value of the indicator variable of which the implantation correct level of the Ion Implanter is measured. According to the value of this indicator variable the implant results of the Ion Implanter can be categorized as ‘Normal’ or ‘Bad’ or ‘Slightly’. A comparative study of whether to do data preprocessing for the accuracy of the proposed BP-ANN model has been performed in this thesis. We observed that if not to do data pre-processing, it will cause a serious error of prediction. Test result verifies the accuracy of the proposed prediction model, the error between the predictive value and the actual value may be less than 0.4%, after data preprocessing. Through a sensitivity analysis of the neural network, we further identify those electrical parameters that are the most important influence factors on the ion implantation result.

摘要 i
ABSTRACT ii
誌謝辭 ii
目錄 v
表目錄 vi
圖目錄 vii
第一章 緒論 1
第一節 研究背景與動機 1
第二節 研究目的 2
第三節 研究流程 2
第二章 文獻探討 5
第一節 半導體製程 5
第二節 離子佈植機 7
第三節 半導體製程參數 12
第三章 研究方法 17
第一節 迴歸分析 31
第二節 類神經網路 17
第三節 類神經網路建構流程 23
第四章 實例研究 30
第一節 輸入變數與輸出變數 36
第二節 統計分析與迴歸分析 38
第三節 建立倒傳遞類神經網路預測模型 41
第四節 小結 48
第五章 結論與建議 48
第一節 結論與貢獻 52
第二節 建議 53
參考文獻 54

莊達人(2000)。「VLSI 製造技術」。台北市：高立圖書有限公司。

葉怡成(2000)。「類神經網路模式應用與實作」。台北市：儒林圖書公司。

林瑞山(2004)。類神經網路於預測晶圓測試良率之應用。未出版之碩士論文，國立成功大學工學院工程管理在職專班，台南市。

蕭宏(2002)。「半導體製程技術導論」。台北市：學銘圖書總代理。

陳逸新(2009)。半導體研發設計階段以WAT 參數建構系統化黃金晶方抽樣分析模型。未出版之碩士論文，國立清華大學工業工程與工程管理研究所，新竹市。

陳志安(2005)。「應用線性迴歸模型」。台北市：華泰文化事業股份有限公司。

李友錚(2007)。「作業管理」。台北市：前程文化事業股份有限公司。

張家銘(2007)。以 CART 決策樹方法建構晶圓良率值與晶圓允收參數間之關聯性模式。未出版之碩士論文，國立清華大學工業工程與工程管理研究所，新竹市。

吳喜成(2011)。40keV硼離子佈植對於氧化鋅薄膜之影響研究。未出版之碩士論文，國立台北科技大學機電整合研究所，台北市。

Badiru, A.B. and Sieger, D.B(1998). Neural network as a simulation metamodel in economic analysis of risky projects. European Journal of Operation Research,(105),130-142

Chaveesuk, R. and Smith, A. E(1999). A neural network metamodel approach to capital investment decision analysis. IEEE Neural Networks International Joint
Conference ,(6),3844-3849.

Freeman, J. A. and Skapura, D. M(1992).Neural Networks Algorithms, Applications ,and Programming Techniques. Addison-Wesley Publishing Company.New York.


Stapper, C. H(1985). The effects of wafer to wafer density variations on integrated circuit defect and fault density. IBM Journal of Research development ,(29), 87-97.

Vellido, A., Lisboa, P. J.G. and Vaughan, J( 1999). Neural Networks in Business: A Survey of Applications .Expert Systems with Applications ,(17), 51-70.

Walczak S., Cerpa N( 1999). Heuristic principles for the design of artificial neural networks. Information and Software Technology,(41), 107-117.

Williams, R. J. and Zipser, D(1989) .A learning algorithm for continually running fully recurrent neural networks.Neural Computatiol,(1), 270-280.

Hecht-Nielsen,R(1989). Theory of the backpropagation neural network. International Joint Conference on Neural Networks. Washington, DC.

Utans, J. Moody, J. Rehfuss, S. and Siegelmann, H(1995) .Input variable selection for neural networks application to predicting the U.S. business cycle. Proceedings of the IEEE/IAFE on Computational Intelligence for Financial Engineering. New York.

Zurada, J. M. Malinowski, A. and Cloete, I(1994).Sensitivity analysis for minimization of input data dimension for feedforward neural network. IEEE International Symposium on Circuits and Systems 6. London.

Hashem, S(1992). Sensitivity analysis for feedforward artificial neural networks with differentiable activation functions. Proceedings of the International Joint Conference on Neural Networks. Baltimore, (7),419-424. Baltimore.

Douglas C.Montgomery(2006).Introduction To Linear Regression Analysis.New York：Wiley.

Dagli,C.et al.,1994.Enhancing the Performance of Neural Network Models for the Wire Bonding Process.Intelligent Engireering Systems through Artificial Neural Networks, (4),1041-1047.

Geman, S. Bienenstock, E.and Doursat, R.(1992).Neural Networks and the Bias/Variance Dilemma,Neural Computation, Neural Computation,(4),1-58.

Hush,D.R.and Horne,B.G.(1993).Progress in Supervised Neural Networks：What’s New since Lippmann,IEEE Signal Processing Magazine.8-39.

Lee,Y.H.and Kim,S.(1993).Neural Network Applications for Scheduling Jobs on Parallel Machines,Computers and Industrial Engineering,(25),227-230.

Williams,R.J.and Zipser,D.(1997).A Learning Algorithm for Continually Running Fully Recurrent Neural Networks,Neural Computation,(1),271-279.

Wang,S.(1998).An Insight into the Standard Back-propagation Neural Network Model for Regression Analysis,International Journal of Management Science,(26),133-140.

Vaithyanathan,S.and Ignizio,J.P.(1992).A Stochastic Neural Network for Resource Constrained Scheduling,Computers and Operations Research,(19),241-254.

Montgomery,D.C.(1996).Introduction to Statistical Quality Control.WILEY.New York.

Liu,D.and Dong,J.(1996).Dispatching Rule Selection Using Artificial Neural Networks for Dynamic Planning and Scheduling,Journal of Intelligent Manufacturing,(7),243-250.