機台的健康狀況如果可以用單一指標來顯示之﹐不但有助於製程設備工程師對製程的掌握﹐更可以用來作為機台維修的根據﹐故如何整合機台即時狀況資料來描述機臺的健康狀況﹐是目前為最新的研究趨勢。此課題的主要研究挑戰是如何從大量機台參數數據中選取最具代表性的統計量來反應真正機台健康狀況。本論文首先利用Hidden Extrapolation Diagnosis來分析出不良的晶圓(Wafer), 進而由Modified Western Electric Rules 和Score Mechanism來建構一套有系統的策略進而制定出一個整合機台的健康指標。最後﹐文中針對多變量t分配資料做更深入的探討。由實證研究可發現在小樣本且在不同自由度(Degree of Freedom)下所決定的臨界值(Critical Point)﹐確實可顯著地改善誤用多變量常態分配所造成晶圓誤判之False Alarm Rate。

Equipment Health Index indicates the equipment condition with a score, which not only helps manufacturing engineers grasp the current manufacturing circumstances but also provides suggestion for the schedules of preventive maintenance. This thesis is to design a generic algorithm, the novel combination of methods, to portray scores for the health condition of equipments. Besides, the use of this algorithm to better characterize processes and provide opportunities for real-time adoption may improve cycle times and lower engineering costs by reducing the number of processing runs necessary to tune and retune a given process. In addition, this approach reduces equipment downtime by providing real-time diagnosis of tool health and process condition and proactive, scheduled maintenance.

Bibliography

1.行政院國家科學委員會自然科學發展處學門與資源規劃報告:統計學,九十四年十月修訂
2.Chen, A., “Real-Time Equipment Health Evaluation and Dynamic Preventive Maintenance”, ISSM2000, pp375- 378, September, 26-28, TOKYO, JAPAN
3.Weisberg, S. Applied Linear Regression, 2nd ed. New York: Wiley, 1985
4.RencherR, A. C. Methods of Multivariate Analysis, 2nd ed. New York: Wiley, 2002.
5.Montgomery, D.C., Design and Analysis of Experiments, 6th ed. New York: Wiley, 2005.
6.Western Electric. Statistical Quality Control Handbook. Western Electric Company: New York, 1958.
7.Chang, C.J., “Health Index of Semiconductor Equipment,” Working Paper.
8.Jensen, D.R., “Closure of multivariate t and related distributions,” Statistics and probability Letters 20, 307-312, 1994.
9.Huang, D.Y. and Tseng, S.T., “A decision procedure for determining the number of components in principal component analysis,” J. Statistic. Planning Inference, vol. 30, pp. 63–71, 1992.
10.Pearson, K, “On non-skew frequency surfaces,” Biometrika 15, 231, 1923.
11.Nadarajah, S. and Kotz, S. “Sampling Distributions associated with the multivariate t-distribution,” Statistica Neerlandica, 59(2), 214-234, 2005