臺灣博碩士論文加值系統

全球半導體製造業目前正在快速發展。在半導體製造過程中，晶圓上的缺陷往往是由機器參數設置、製程表或環境因素等造成的。在資料科學時代，資料品質在機器/深度學習模型的訓練階段起著重要作用。為了在機器學習的背景下獲取專家知識，手動資料標註是監督學習中一項繁瑣且耗時的任務。然而，人工標註的專家在長時間工作後容易分心或疲勞，導致誤判、誤標等。
本論文研究晶圓缺陷圖的模式識別，其主要目標是用有限的手動標記數據訓練卷積神經網絡（CNN）模型，使訓練的模型能夠自動標註。隨後，根據卷積神經網絡的能力，對錯誤自動標註的樣本圖進行標籤修正。因此，在保留所有專家標註的晶圓圖的同時，可以確保晶圓缺陷模式有一樣好的分類表現。

The global semiconductor manufacturing industry is currently undergoing rapid development. In the semiconductor manufacturing process, defects on the wafer are often caused by machine parameter settings, process schedules, or environmental factors, etc. In the era of data science, data quality plays an important role in the training phase of machine/deep learning models. To capture expert knowledge in the context of machine learning, manual data labeling is a tedious and time-consuming task in supervised learning. However, a large number of domain experts who manually annotate lines are easily distracted or fatigued after long hours of work, resulting in misjudgment, mislabeling, etc.
This thesis studies pattern recognition of wafer defect maps, the main goal of which is to train a convolutional neural network (CNN) model with limited manually labeled data so that the trained model is capable of performing pseudo labeling. Subsequently, according to the ability of convolutional neural network, label rectification is performed on the sample maps of mistaken automatic annotations. As a result, while all expert annotated wafer maps are preserved, the equally high classification performance of wafer defect patterns can be assured.

摘　要 i
ABSTRACT ii
誌謝 iv
List of Content v
List of Tables vii
List of Figures ix
Chapter 1 Introduction 1
1.1 Background 1
1.2 Motivation 2
1.3 Description of the Objectives 3
1.4 Organization of the Thesis 4
Chapter 2 Literature Review 5
2.1 Semiconductor Manufacturing 5
2.1.1 Oxidation 6
2.1.2 Lithography 6
2.1.3 Etching 7
2.1.4 Ion Implantation 7
2.1.5 Sputtering 7
2.2 Machine Learning 8
2.2.1 Supervised Learning 9
2.2.2 Unsupervised Learning 11
2.2.3 Semi-supervised Learning 12
2.3 Deep Learning 12
2.4 Related Work 13
Chapter 3 Proposed Methodology 15
3.1 The Flow Chart of Proposed Methods 15
3.2 Methodology 16
3.2.1 Convolutional Neural Network 16
3.2.2 VGGNet 19
3.3 Label Rectification 21
3.4 Confusion Matrix 21
Chapter 4 The Experimental Results of Proposed Method 24
4.1 Feasibility testing - CIFAR-10 dataset 24
4.2 Dataset of Wafer Defect Pattern 29
4.3 Patterns of softmax value changes over epochs 31
4.4 Select the number of epochs for Label Rectification 58
4.5 With and without Label Rectification 63
4.6 Improvement Ratio 64
4.7 Softmax of Pseudo Labeled wafer maps 68
Chapter 5 Conclusions and Future Research 71
References 72
Appendix 75

Freund, Y., and Schapire, R. E., 1997, “A decision-theoretic generalization of on-line learning and an application to boosting,” Journal of computer and system sciences, vol. 55 (1), pp. 119-139.

Fang, Z., and Yang, Y., Li, W., Chen, Y., Ma, L., and Du, Q., 2022, “Confident Learning-Based Domain Adaptation for Hyperspectral Image Classification,” IEEE Transactions on Geoscience and Remote Sensing, vol. 60.

He, K., Zhang, X., Ren, S., and Sun, J., 2016, “Deep residual learning for image recognition,” In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770- 778.

Hu, H., He, C., and Li, P., 2021, “Semi-supervised Wafer Map Pattern Recognition using Domain-Specific Data Augmentation and Contrastive Learning,” 2021 IEEE International Test Conference (ITC).

Hwang, J. Y., and Kuo, W., 2007, “Model-based clustering for integrated circuit yield enhancement,” Eur. J. Oper. Res., vol. 178 (1), pp. 143-153.

Kong, Y., and Ni, D., 2020, “A Semi-Supervised and Incremental Modeling Framework for Wafer Map Classification,” IEEE Transactions on Semiconductor Manufacturing, vol. 33 (1).

Krizhevsky, A., Sutskever, I., and Hinton, G. E., 2017, “Imagenet classification with deep convolutional neural networks,” Communications of the ACM, vol. 60 (6), pp. 84-90.
Kyeong, K., and Kim, H., 2018, “Classification of Mixed-Type Defect Patterns in Wafer Bin Maps Using Convolutional Neural Networks,” IEEE Transactions on Semiconductor Manufacturing, vol. 31 (3), pp. 395-402.

Kim, Y., Cho, D., and Lee, J., 2020, “Wafer Map Classifier using Deep Learning for Detecting Out-of-Distribution Failure Patterns,” 2020 IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits (IPFA).

LeCun, Y., Bottou, L., Bengio, Y., and Haffner, P., 1998, “Gradient-based learning applied to document recognition,” Proceedings of the IEEE, vol. 86 (11), pp. 2278-2324.

Li, K., Jiang, X., Chen, L., Wang, S., Huang, A., Chen, J. and Liang, H., 2022, “Wafer Defect Pattern Labeling and Recognition Using Semi-Supervised Learning,” IEEE Transactions on Semiconductor Manufacturing, vol. 35 (2), pp. 291-299

MO, J., Gan, Y., and Yuan, H., 2021. “Weighted Pseudo Labeled Data and Mutual Learning for Semi-Supervised Classification,” IEEE Access, vol. 9, pp. 36522-36534.

Mascarenhas, S., and Agarwal, M., 2021, “A comparison between VGG16, VGG19 and ResNet50 architecture frameworks for Image Classification,” 2021 IEEE International Conference on Disruptive Technologies for Multi-Disciplinary Research and Applications (CENTCON).

Saqlain, M., and Jargalsaikhan, B., 2019, “A Voting Ensemble Classifier for Wafer Map Defect Patterns Identification in Semiconductor Manufacturing,” IEEE Transactions on Semiconductor Manufacturing, vol. 32 (2), pp. 171-182

Xu, X., and Zhang, L., 2021, “Rectifying Pseudo Label By Mutual Disagreement Learning For Unsupervised Domain Adaptation Person Re-Identification,” 2021 IEEE International Conference on Multimedia and Expo (ICME).

Yu, J., and Lu, X., 2016, “Wafer Map Defect Detection and Recognition Using Joint Local and Nonlocal Linear Discriminant Analysis,” IEEE Transactions on Semiconductor Manufacturing, vol. 29 (1), pp. 33-43.

Zhang, T., Ma, Y., and Li, H., 2021, “Analysis of Semi-Supervised Algorithms in Natural Language Processing,” 2021 International Conference on Electronic Information Technology and Smart Agriculture (ICEITSA).