真空時延系統之穩健控制

Abstract

真空系統的壓力控制在半導體製程的品質上有決定性的影響。以金屬有機物化學氣相沉積為例(Metal Organic Chemical Vapor Decomposition ,MOCVD)， 其腔體內壓力的穩定度都是以PID (proportional-integral-derivative)控制器來控制，但此方法對於時間延遲、非線性、不確定性較高的製程，能夠修正的有限，且易受干擾。故本論文建立一個簡易真空平台，透過系統鑑別，了解實驗模型之動態特性，並透過理想真空模型之數學推導，得到一線性理論系統模型，驗證實驗模型之準確性，再以具有雙互質分解干擾觀測器(Doubly coprime disturbance observer, DCFDOB)之兩階段H∞設計方法設計穩健控制器，來處理不確定性、干擾等問題，並針對於高不確定之性真空時延系統做穩定度分析，且與傳統之Mu-synthesis、Loopshaping做比較，其模擬和實驗之結果顯示DCFDOB兩階段H∞設計方法及所設計之穩健控制器能夠更快穩定且準確控制壓力的變化。

Pressure control of vacuum system is the most crucial thing for the quality of semiconductor process. Take MOCVD (Metal Organic Chemical Vapor Decomposition) for example. The chamber pressure of MOCVD is controlled by PID, which is limited to fix the problem like time delay, nonlinear, and uncertainty. Moreover, disturbance has a strong effect on semiconductor process. The main theme of this thesis is to investigate above-mentioned phenomena and fix them by the developed method. First, the linear model of the vacuum-chamber dynamic characteristics was identified experimentally at various operating points; then compared the identified model with the linear, theoretical model which is derived from ideal gas law. Furthermore, we used a cascaded H∞ design method for DCFDOB to design observer and controller, and then compare the stability and performance of the closed-loop system with those used Mu-synthesis and Loop shaping design method. The simulation and experiment result shows that cascaded H∞ design method with DCFDOB can perform faster response with better disturbance rejection capability than others.