半導體產業中，薄膜製程為晶圓加工的基礎，依據鍍上薄膜材質的不同，可擁有不同的性能，然而此階段含有大量複雜的交互作用，會影響晶圓最終之性能及良率，因此為確保最終產品之品質，需定期執行抽樣檢測，以獲取薄膜之實際厚度，不過此方法不僅無法獲取及時的薄膜厚度亦無法針對該膜厚來進行即時的機台參數調整。因此本研究目標在於建立虛擬量測系統並綜合大數據分析及人工智慧等技術，為個案公司減少量測成本，並達到100%全檢以及提升產品良率之目標，同時本研究亦處理了個案公司所遇到之四大問題：類別變數處理、機台特徵篩選、機台短期趨勢變動及模型超參數優化。模型在實際上線後，部分類別資料會出現未曾看過之情形，導致模型無法進行準確的預測，且在眾多機台特徵中並非所有皆為關鍵特徵，同時機台狀況會隨時間出現變化，因此本研究不僅針對關鍵特徵進行分析，亦結合了超參數優化演算法、文字嵌入以及個案公司同仁之實務經驗及領域知識，建構客製化虛擬量測系統架構以處理上述所提到的四大問題，藉由此架構，可以幫助個案公司獲取即時的膜厚資訊，以確保其產品品質，同時幫助決策者進行機台參數調整，並及時發現異常之情形。

In the semiconductor industry, thin film processes are fundamental to wafer manufacturing. Depending on the material deposited, different performances can be achieved. However, this stage involves many complex interactions that can affect the final performance and yield of the wafer. Therefore, to ensure the quality of the final product, regular sampling inspections are conducted to obtain the actual thickness of the thin films. However, this method not only fails to provide real-time thickness data but also cannot enable immediate adjustments to machine parameters based on the thickness measurements. The goal of this study is to establish a virtual metrology system that integrates big data analysis and artificial intelligence technologies to reduce measurement costs, achieve 100% full inspection, and improve product yield for the case company. Additionally, this study addresses four major issues faced by the case company: handling categorical variables, selecting critical machine features, managing short-term machine trend changes, and optimizing model hyperparameters. When the model is deployed in practice, some categorical data may not have been encountered before, leading to inaccurate predictions. Not all of the numerous machine features are critical, and machine conditions can change over time. Therefore, this study not only analyzes key features but also incorporates hyperparameter optimization algorithms, word embedding techniques, and practical experience and domain knowledge from the case company to construct a customized virtual metrology system architecture to address the aforementioned four issues. This architecture helps the case company obtain real-time film thickness information, ensuring product quality while assisting decision-makers in adjusting machine parameters and detecting anomalies in a timely manner.

Bergstra, J., & Bengio, Y. (2012). Random search for hyper-parameter optimization. Journal of Machine Learning Research, 13(2).
Bergstra, J., Bardenet, R., Bengio, Y., & Kégl, B. (2011). Algorithms for hyper-parameter optimization. Advances in Neural Information Processing Systems, 24.
Bischl, B., Binder, M., Lang, M., Pielok, T., Richter, J., Coors, S., ... & Lindauer, M. (2023). Hyperparameter optimization: Foundations, algorithms, best practices, and open challenges. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 13(2), e1484.
Chang, K. H., Tsai, C. C., Wang, C. H., Chen, C. J., & Lin, C. M. (2021). Optimizing the energy efficiency of chiller systems in the semiconductor industry through big data analytics and an empirical study. Journal of Manufacturing Systems, 60, 652-661.
Chang, K. H., Sun, Y. J., Lai, C. A., Chen, L. D., Wang, C. H., Chen, C. J., & Lin, C. M. (2022). Big data analytics energy-saving strategies for air compressors in the semiconductor industry–an empirical study. International Journal of Production Research, 60(6), 1782-1794.
Chen, C. H., Zhao, W. D., Pang, T., & Lin, Y. Z. (2020). Virtual metrology of semiconductor PVD process based on combination of tree-based ensemble model. Isa Transactions, 103, 192-202.
Chen, T., & Guestrin, C. (2016, August). Xgboost: A scalable tree boosting system. In Proceedings of The 22nd Acm Sigkdd International Conference On Knowledge Discovery And Data Mining , 785-794.
Chou, P. H., Wu, M. J., & Chen, K. K. (2010). Integrating support vector machine and genetic algorithm to implement dynamic wafer quality prediction system. Expert Systems with Applications, 37(6), 4413-4424.
Dreyfus, P. A., Psarommatis, F., May, G., & Kiritsis, D. (2022). Virtual metrology as an approach for product quality estimation in Industry 4.0: A systematic review and integrative conceptual framework. International Journal of Production Research, 60(2), 742-765.
Fan, S. K. S., Hsu, C. Y., Tsai, D. M., He, F., & Cheng, C. C. (2020). Data-driven approach for fault detection and diagnostic in semiconductor manufacturing. IEEE Transactions on Automation Science and Engineering, 17(4), 1925-1936.
Feng, Z., Guan, N., Lv, M., Liu, W., Deng, Q., Liu, X., & Yi, W. (2020). Efficient drone hijacking detection using two-step GA-XGBoost. Journal of Systems Architecture, 103, 101694.
Feurer, M., Springenberg, J., & Hutter, F. (2015, February). Initializing bayesian hyperparameter optimization via meta-learning. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 29, No. 1).
Guyon, I., Weston, J., Barnhill, S., & Vapnik, V. (2002). Gene selection for cancer classification using support vector machines. Machine learning, 46, 389-422.
Guo, J., Yang, L., Bie, R., Yu, J., Gao, Y., Shen, Y., & Kos, A. (2019). An XGBoost-based physical fitness evaluation model using advanced feature selection and Bayesian hyper-parameter optimization for wearable running monitoring. Computer Networks, 151, 166-180.
Jones, D. R. (2001). A taxonomy of global optimization methods based on response surfaces. Journal of Global Optimization, 21, 345-383.
Kuhn, M., & Johnson, K. (2013). Applied predictive modeling (Vol. 26, p. 13). New York: Springer. doi:10.1007/978-1-4614-6849-3.
Lesch, V., Züfle, M., Bauer, A., Iffländer, L., Krupitzer, C., & Kounev, S. (2023). A literature review of IoT and CPS—What they are, and what they are not. Journal of Systems and Software, 200, 111631.
Li, Z., Duan, M., Xiao, B., & Yang, S. (2022). A novel anomaly detection method for digital twin data using deconvolution operation with attention mechanism. IEEE Transactions on Industrial Informatics, 19(5), 7278-7286.
Limanond, S., Si, J., & Tsakalis, K. (1998). Monitoring and control of semiconductor manufacturing processes. IEEE Control Systems Magazine, 18(6), 46-58.
Liu, D., Du, Y., Chai, W., Lu, C., & Cong, M. (2022). Digital twin and data-driven quality prediction of complex die-casting manufacturing. IEEE Transactions on Industrial Informatics, 18(11), 8119-8128.
Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A. V., & Gulin, A. (2018). CatBoost: Unbiased boosting with categorical features. Advances in Neural Information Processing Systems, 31.
Psarommatis, F., Sousa, J., Mendonça, J. P., & Kiritsis, D. (2022). Zero-defect manufacturing the approach for higher manufacturing sustainability in the era of industry 4.0: a position paper. International Journal of Production Research, 60(1), 73-91.
Zhang, C., Chen, D., Tao, F., & Liu, A. (2019). Data driven smart customization. Procedia CIRP, 81, 564-569.