虛擬量測(Virtual Metrology,VM)技術利用工件的歷史製程資料和與其相對應的量測資料來建立推估模型，如此可將離線且具延遲特性的品質抽檢改為線上且即時之品質全檢。目前大部分的虛擬量測系統所採用的建模參數組合，是擷取自該製程某時間區段相對應的歷史製程與量測資料，以『資料導向(Data Driven)』分析法挑選出建模參數。
資料導向在決策中仰賴歷史數據的品質、資料集的大小和所在時間區間，挑選出的參數集合好壞常受到以上因素所侷限，例如：不同時間區間的資料變異程度大，最後使得統計分析所得的參數集合會隨著上述限制因素的不同而改變。如此，將無法穩定地選到符合該製程的所有重要參數組合，致預測模型與精度將難以維護。
資料導向所導致的限制使得研究者常需要多次試誤與依賴工程師廠內經驗來提升預測精度；再者，研究經驗中發現，有時候具統計意義之結果，並不能完全代表實際運作之物理現象。綜合以上狀況，將難以說服生產單位接受並使用之。
為解決上述問題，本論文提出一套適用於虛擬量測系統的特徵篩選法。其作法為首先採取基於物理與化學的製程領域原理即所謂模型為基(Model-based)的特徵萃取 方法來挑選主要參數(Primary Parameters)；然後，再採用基於統計與資料探勘等數學工具的資料導向(Data Driven)方法來挑選次要參數(Secondary Parameters)。由範例可得知，本論文所採用的方法，可妥當地挑選出固定的參數組合，且可改善虛擬量測的預測精度。

Virtual Metrology (VM) technology utilizes historical process and corresponding metrology data of workpieces to construct the conjecture model so as to convert off-line sampling inspection with metrology delay into on-line and real-time total inspection. Traditionally, most virtual metrology systems adopt Data Driven Analysis to choose appropriate process indicators by picking a particular process out of a time zone to get the historical process and corresponding measurement data.
Data-Driven Analysis is easily affected and limited by samples qualities, samples sizes and, production time of historical data while conducting decision-making. For example, large variance between data of different time zone would be a factor affecting decision-making. Data from different time zones have different characteristics, thus the selection result would vary over time, which will lead to difficult maintenance of conjecture model and its accuracy.
The limitations of data driven result in continuous tests and errors of the researchers and the dependence of engineers’ fab experiences. In addition, according to past research experiences, the results which meet the statistical meaning was not able to represent their actual operation of the physical phenomena. Due to such reason, it is sometimes infeasible for the manufacturers to accept and adopt this method.
To solve the issues mentioned above, this paper proposes a feature selection architecture which uses model-based theoretical foundation to select primary parameters and adopts the concept of data-driven analysis to find out secondary parameters. The experiment result verifies that the model-based feature selection method is able to select proper combination of fixed parameters and thus enhances VM accuracy.

[1] F.-T. Cheng, H.-C. Huang, and C.-A. Kao, “Developing an Automatic Virtual Metrology System,” IEEE Trans. on Automation Science and Engineering, vol. 9, no. 1, pp. 181-188., Jan. 2012.
[2] F.-T. Cheng, H.-C. Huang, and C.-A. Kao, "Dual-Phase Virtual Metrology Scheme," IEEE Trans. on Semiconductor Manufacturing, vol. 20, no. 4, pp. 566-571, Nov. 2007.
[3] A. Weber, “Virtual Metrology and Your Technology Watch List: Ten Things You Should Know about This Emerging Technology,” Future Fab International, issue 22, section 4, pp. 52-54, Jan. 2007.
[4] F.-T. Cheng, Y.-T. Chen, Y.-C. Su, and D.-L. Zeng, “Evaluating Reliance Level of a Virtual Metrology System,” IEEE Trans. on Semicond. Manuf., vol. 21, no. 1, pp. 92-103, February 2008.
[5] Y.-T. Huang and F.-T. Cheng, “Automatic Data Quality Evaluation for the AVM System,” IEEE Trans. on Semicond. Manuf., vol. 24, no. 3, pp.445-454, August 2011.
[6] S. Lynn , J. Ringwood , E. Ragnoli , S. McLoone and N. MacGearailt "Virtual Metrology for Plasma Etch Using Tool Variables", Proc. IEEE/SEMI Adv. Semicond. Manuf. Conf.(ASMC), pp.143 -148 2009.
[7] B. Lenz, B. Barak, C. Leicht, "Development of smart feature selection for advanced virtual metrology," IEEE/SEMI Advanced Semiconductor Manufacturing Conference(ASMC), pp.145-150, May 2014.
[8] D. Zeng, “Statistical Methods for Enhanced Metrology in Semiconductor/Photovoltaic Manufacturing,” EECS Technical Reports, pp. 41-49, Dec. 2012.
[9] T. Lee and C-O. Kim,” Statistical Comparison of Fault Detection Models for Semiconductor Manufacturing Processes,” IEEE Transactions on Semiconductor Manufacturing, VOL.28, No.1, Feb 2015.
[10] Y. Jang, S. Jeong, S. Park, H-J. Roh, S. Ryu, M. Choi and G-H. Kim, ”Performance of Robust VM Models Based on PCR and Sensitivity Ranking Test for Oxide Etch Process in TF-CCP,” ISSM, PC-P-063, Dec 2014.
[11] A. Emami-Naeini , J. Ebert , R. Kosut , D. de Roover and S. Ghosal "Model-based control for semiconductor and advanced materials processing: An overview", Proc. Amer. Contr. Conf., vol. 5, pp.3902 -3909 2004.
[12] 黃暄恆「適用於全自動虛擬量測系統之智慧型參數篩選機制」，國立成功大學製造資訊與系統研究所碩士論文學位論文，2013.
[13] M. Fukasawa, A. Kawashima, N. Kuboi, H. Takagi, Y. Tanaka, H. Sakayori, K. Oshima, K. Nagahata, and T. Tatsumi, ”Prediction of Fluctuations in Plasma-Wall Interactions Using an Equipment Engineering System,” 2009, Jpn. J. Appl. Phys., 48,08HC01.
[14] S. Park, Y. Janh, S. Jeong, H-J. Roh, S.Ryu and G-H .Kim,” Enhancement of Virtual Metrology Performance for Oxide Etching Process by Using Process Plasma Monitoring,” ISSM, PC-030, Dec 2014.
[15] Y.-T. Huang, H.-C. Huang, F.-T. Cheng, T.-S. Liao, and F.-C. Chang, “Automatic Virtual Metrology System Design and Implementation,” in Proc. 4th IEEE Conference on Automation Science and Engineering, USA, Aug. 2008.
[16] 白木 靖寬、吉田 貞史編著，王建義編譯，薄膜工程學-薄膜工學，第二版，全華科技圖書股份有限公司出版，2006，P4-57~P4-62，譯序。
[17] 王建榮、林慶福、林必窕編譯，半導體平坦化CMP技術，初版，全華科技圖書股份有限公司出版，1999，P1-3~P1-7.