成功大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	張尹甄 Chang, Yin-Chen
論文名稱：	考慮電信檢測次序的影響進行晶圓片缺陷型態之分群與分類 The Defect Pattern Clustering and Classification of Wafer Bin Map with Masked Effect in Circuit Probe Test
指導教授：	鄭順林 Jeng, Shuen-Lin
共同指導教授:	夏啟峻 Hsia, Chi-Chun
學位類別：	碩士 Master
系所名稱：	管理學院 - 數據科學研究所 Institute of Data Science
論文出版年：	2020
畢業學年度：	108
語文別：	英文
論文頁數：	73
中文關鍵詞：	電信檢測、晶圓圖像、分群、分類、影像辨識、影像遮蓋、卷積神經網路
外文關鍵詞：	Circuit Probe Test, Wafer Bin Map, Clustering, Classification, Image Recognition, Image Masking, Convolution Neural Network
相關次數：	點閱：93 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

隨著晶圓片的構造越來越複雜，晶圓片在製造的過程中常常會出現不同的缺陷情形。半導體公司為了維持競爭力，所以很重視製程中的缺陷偵測與良率的提升，也因此晶圓片的製程中會穿插一些檢測的步驟來檢視晶圓片生產的情形。其中，電信檢測(circuit probe test) 是一種常見的檢測方式，主要是針對每一個晶粒去檢查他們的各項功能是否正常。電信檢測的輸出結果稱為晶圓圖(wafer bin map)，工程師不但能透過晶圓圖得知晶圓片的缺陷情形，也能夠透過缺陷的晶粒在晶圓圖中所組成的特殊形態來判斷這些缺陷晶粒是由製程中的哪些部分造成的，並進一步去改良製程、提升良率。

目前為止，已有許多關於晶圓圖上的缺陷型態的研究。一部份的研究是採用傳統統計方法，他們需要較多的資料前處理步驟來達到特徵擷取的目的；而另一部份的的研究則是採用深度學習的方法，他們往往需要較多已標註好的資料去訓練模型並進行預測。現實情況來說，晶圓片都需要進行人工標記來獲得標註好的資料，故要得到大量的標註資料是相對困難的。除此之外，在電信檢測中所產生的遮蓋效應也是在分析上的一個很重要的問題之一，但過去的研究卻較少去探討。

在我們的研究中，我們使用了少量的標註資料去訓練深度學習模型，並從訓練好的模型中去直接擷取晶圓圖的特徵。我們希望能夠用這些擷取出來的特徵來對晶圓圖上的缺陷型態進行適當的分群，並利用其他產品中曾經出現過的缺陷型態來進一步地將新出現的缺陷型態從原有的缺陷型態中分開。在方法驗證的部分，我們使用模擬資料來對晶圓片缺陷型態的分群進行驗證；並使用模擬資料及一些修改過後的真實資料來對新出現缺陷型態的偵測進行驗證。在實驗過程中，我們也發現我們的方法相對其他傳統的統計方法來說，所受到遮蓋效應的影響會較小。

As wafer fabrication processes become more and more complicated, new process failures increase, which makes defect detection and yield enhancement become crucial issues to maintain competitive advantages in semiconductor companies. A wafer bin map (WBM) is the result of a circuit probe test (CP test) on a wafer after the completion of the manufacturing process. The specific defect patterns on WBMs provide crucial information for engineers to trace the failure causes in the complicated manufacturing process.

Many research for WBM image recognition were done using statistical and deep learning methods. The statistical methods are often followed by additional image transformation process. As for the deep learning method, they often focus on dealing the classification task with many labelled data, which is difficult to apply on real analysis because real data has no label unless the labels are marked artificially. There are also common problems in WBM image recognition that are seldom discussed by previous studies, for example, the crucial masking problem caused by the mechanism of CP test.

In our study, we use few labels to train deep learning models and extract features from the models without additional transformations on WBMs. We then use the features to cluster the WBMs. We aim to give proper clustering for the typical defect patterns on WBM. If there is a new defect pattern in the current WBM product, we also use the deep learning methods to separate the new pattern from the typical defect patterns by using the help of untypical defect patterns from other WBM products. The evaluation of our clustering procedure is conducted with simulation data and the evaluation of our new pattern detection is conducted with simulation data and some modified real data. Furthermore, we have investigated about the effect of the masking problem caused by the mechanism of CP test.

摘要i
Abstract ii
誌謝iii
Table of Contents iv
List of Tables vi
List of Figures viii
Chapter 1. Introduction 1
1.1. Background and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2. Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3. The Flow Chart of Our Study . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Chapter 2. Literature Review 7
2.1. Defect Patterns on WBM . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2. Statistical Feature Engineering on Image Recognition . . . . . . . . . . . . 10
2.3. Deep Learning Methods on WBM Defect Pattern Detection . . . . . . . . . 10
2.4. Summary of Previous Studies . . . . . . . . . . . . . . . . . . . . . . . . . 11
Chapter 3. Methodology 14
3.1. Radon Transform and RadonRelated
Methods . . . . . . . . . . . . . . . . 14
3.1.1. Radon Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.1.2. RadonRelated
Methods . . . . . . . . . . . . . . . . . . . . . . . 15
3.2. Convolutional Neural Networks (CNN) . . . . . . . . . . . . . . . . . . . 18
3.2.1. Introduction of Convolutional Neural Networks . . . . . . . . . . . 18
3.2.2. The CNN Architecture of Convolutional Neural Networks . . . . . 19
3.3. Kmedoids
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.4. Proposed Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.4.1. Typical Defect Pattern Detection on WBMs . . . . . . . . . . . . . 23
3.4.2. New Defect Pattern Detection on WBMs . . . . . . . . . . . . . . . 26
Chapter 4. Experiment 27
4.1. Dataset and Data Preprocessing . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1.1. Simulation Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1.2. Real Dataset . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.2. Evaluation of our Models in Classification Task . . . . . . . . . . . . . . . 34
4.2.1. Overall Performance . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.2.2. Performance without Mixed Noise . . . . . . . . . . . . . . . . . . 36
4.2.3. Performance under Different Level of Random Noises . . . . . . . 38
4.3. Evaluation of the Models in Clustering Procedure . . . . . . . . . . . . . . 42
4.3.1. Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.3.2. Evaluation Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.3.3. Baseline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.3.4. Clustering Result . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.4. New Defect Pattern Detection . . . . . . . . . . . . . . . . . . . . . . . . 51
4.4.1. Dataset Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.4.2. Classification Result . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.5. Execution Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Chapter 5. Conclusion 62
5.1. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.2. Discussion and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . 63
References 64
Appendix A. Confusion Matrix of New Pattern Prediction 67
                                    

[1] Elie Aljalbout, Vladimir Golkov, Yawar Siddiqui, Maximilian Strobel, and Daniel Cremers. Clustering with deep learning: Taxonomy and new methods. arXiv preprint arXiv:1801.07648, 2018.
[2] Pierre Baldi and Peter Sadowski. The dropout learning algorithm. Artificial intelligence, 210:78–122, 2014.
[3] Cheng-Wei Chang, Tsung-Ming Chao, Jorng-Tzong Horng, Chien-Feng Lu, and Rong-Hwei Yeh. Development pattern recognition model for the classification of circuit probe wafer maps on semiconductors. IEEE Transactions on Components, Packaging and Manufacturing Technology, 2(12):2089–2097, 2012.
[4] Chen-Fu Chien, Shao-Chung Hsu, and Ying-Jen Chen. A system for online detection and classification of wafer bin map defect patterns for manufacturing intelligence. International Journal of Production Research, 51(8):2324–2338, 2013.
[5] Keystone Microtech Coproration. Introduction of probe card / substrate. http://yieldwerx.com/conductingyieldanalysissemiconductormanufacturing/.
[6] Mengying Fan, Qin Wang, and Ben van der Waal. Wafer defect patterns recognition based on optics and multilabel classification. 2016 IEEE Advanced Information Management, Communicates, Electronic and Automation Control Conference (IMCEC), pages 912–915, 2016.
[7] Chia-Yu Hsu. Clustering ensemble for identifying defective wafer bin map in semiconductor manufacturing. Mathematical Problems in Engineering, 2015, 2015.
[8] Ming-Jie Hsu. The order effect of wafer defect detection in the clustering of wafer bin maps. Master’s Thesis of National Cheng Kung University, Department of Statistics, 2017.
[9] Shao-Chung Hsu and Chen-Fu Chien. Hybrid data mining approach for pattern extraction from wafer bin map to improve yield in semiconductor manufacturing. International Journal of Production Economics, 107(1):88–103, 2007.
[10] Tsutomu Ishida, Izumi Nitta, Daisuke Fukuda, and Yuzi Kanazawa. Deep learningbased wafermap failure pattern recognition framework. 20th International Symposium on Quality Electronic Design (ISQED), pages 291–297, 2019.
[11] Leonard Kaufman and Peter J Rousseeuw. Finding groups in data: an introduction to cluster analysis, volume 344. John Wiley & Sons, 2009.
[12] Keras. Github keras. https://github.com/kerasteam/keras.
[13] Kiryong Kyeong and Heeyoung Kim. Classification of mixedtype defect patterns in wafer bin maps using convolutional neural networks. IEEE Transactions on Semiconductor Manufacturing, 31(3):395–402, 2018.
[14] Kudrov Maksim, Bukharov Kirill, Zakharov Eduard, Grishin Nikita, Bazzaev Aleksandr, Lozhkina Arina, Semenkin Vladislav, Makhotkin Daniil, and Krivoshein Nikolay. Classification of wafer maps defect based on deep learning methods with small amount of data. 2019 International Conference on Engineering and Telecommunication (EnT), pages 1–5, 2019.
[15] Nwayyin Najat Mohammed and Adnan Mohsin Abdulazeez. Evaluation of partitioning around medoids algorithm with various distances on microarray data. 2017 IEEE International Conference on Internet of Things Things) and IEEE Green Computing and Communications (GreenCom) and IEEE Cyber, Physical and Social Computing (CPSCom) and IEEE Smart Data (SmartData), pages 1011–1016, 2017.
[16] Gordon E Moore et al. Cramming more components onto integrated circuits, 1965.
[17] Takeshi Nakazawa and Deepak V Kulkarni. Wafer map defect pattern classification and image retrieval using convolutional neural network. IEEE Transactions on Semiconductor Manufacturing, 31(2):309–314, 2018.
[18] Takeshi Nakazawa and Deepak V Kulkarni. Anomaly detection and segmentation for wafer defect patterns using deep convolutional encoder–decoder neural network architectures in semiconductor manufacturing. IEEE Transactions on Semiconductor Manufacturing, 32(2):250–256, 2019.
[19] Melanie Po-Leen Ooi, Eric Kwang Joo Joo, Ye Chow Kuang, Serge Demidenko, Lindsay Kleeman, and Chris Wei Keong Chan. Getting more from the semiconductor test: Data mining with defectcluster extraction. IEEE Transactions on Instrumentation and Measurement, 60(10):3300–3317, 2011.
[20] Johann Radon. 1.1 über die bestimmung von funktionen durch ihre integralwerte längs gewisser mannigfaltigkeiten. Classic papers in modern diagnostic radiology, 5:21, 2005.
[21] Zongli Shen and Jianbo Yu. Wafer map defect recognition based on deep transfer learning. 2019 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM), pages 1568–1572, 2019.
[22] Salvatore Tabbone, Laurent Wendling, and JP Salmon. A new shape descriptor defined on the radon transform. Computer Vision and Image Understanding, 102(1):42–51, 2006.
[23] Ghalia Tello, Omar Y Al-Jarrah, Paul D Yoo, Yousof Al-Hammadi, Sami Muhaidat, and Uihyoung Lee. Deepstructured machine learning model for the recognition of mixeddefect patterns in semiconductor fabrication processes. IEEE Transactions on Semiconductor Manufacturing, 31(2):315–322, 2018.
[24] Tzu-Yi Tseng. The masked effect of defect detection from wafer bin maps. Master’s Thesis of National Cheng Kung University, Department of Statistics, 2017.
[25] Junliang Wang, Zhengliang Yang, Jie Zhang, Qihua Zhang, and Wei-Ting Kary Chien. Adabalgan: An improved generative adversarial network with imbalanced learning for wafer defective pattern recognition. IEEE Transactions on Semiconductor Manufacturing, 32(3):310–319, 2019.
[26] Rui Wang and Nan Chen. Wafer map defect pattern recognition using rotationinvariant features. IEEE Transactions on Semiconductor Manufacturing, 32(4):596–604, 2019.
[27] Rui Wang and Nan Chen. Defect pattern recognition on wafers using convolutional neural networks. Quality and Reliability Engineering International, 36(4):1245–1257, 2020.
[28] Wikipedia. K-medoids. https://en.wikipedia.org/wiki/Kmedoids.
[29] Ming-Ju Wu, Jyh-Shing R Jang, and Jui-Long Chen. Wafer map failure pattern recognition and similarity ranking for largescale data sets. IEEE Transactions on Semiconductor Manufacturing, 28(1):1–12, 2015.
[30] yieldWerx. Conducting yield analysis for semiconductor manufacturing. http://yieldwerx.com/conductingyieldanalysissemiconductormanufacturing/.
[31] Jianbo Yu and Xiaolei Lu. Wafer map defect detection and recognition using joint local and nonlocal linear discriminant analysis. IEEE Transactions on Semiconductor Manufacturing, 29(1):33–43, 2015.
[32] Naigong Yu, Qiao Xu, and Honglu Wang. Wafer defect pattern recognition and analysis based on convolutional neural network. IEEE Transactions on Semiconductor Manufacturing, 32(4):566–573, 2019.

校內：2025-07-23公開
校外：不公開電子論文尚未授權公開，紙本請查館藏目錄

簡易檢索 / 詳目顯示

相關論文