半導體生產設備的問題,往往因原廠設備設計的瓶頸所造成。在設備異常時設備端無足夠的回饋機制亦無良好的人機介面可提供警報、記錄、查詢;在製程端,無精準的生產參數模型做預警監控,導致無法立即發現生產參數已偏移基準線,最後造成晶片大量報廢損失。因此有必要建立半導體設備監控系統,以可即時掌控設備正常狀況,並讓處理機況作業更加便捷,且有效降低單位生產成本、穩定產品品質與增進企業競爭力。 本研究以8吋晶圓廠為個案研究,藉由日常管理中設備異常率的統計發現失效警報率過多。經過分析機台監控系統,可在設備端建立回饋機制來監控防堵電磁閥瞬間失效。在製程端建置失誤偵測與分類防護網,可在設備端增加額外感測器再與資訊端建立相關的監控參數模型讓防護網作預警監控,藉由兩者的導入確認導入前後所達成的效果,並透過使用者上線成效滿意度評量,來驗證建立後之改善績效。 本研究結果證實,設備端人機監控的改善與製程過程中提昇設備參數的訊號及失誤偵測與分類防護網參數模型建立兩者的導入與運用,確實能即時監控設備問題,改善產品良率與設備工程師工作的壓力,明顯提升生產效率與縮短偵錯時間效果。可提供日後建廠規畫參考,以提升機台運轉的穩定度。
The problemes of Semiconductor manufacturing equipment are often due to bottlenecks in original equipment design, As equipment malfunctioned , there ware no sufficient feedback mechanism nor good Humman Machine Interface to provided,etc In process ends,there were no accurate-parametric models to make early warning surveillance, and thus lead production parameters to be offset baseline. Finally,it resulting in a large number of scrapped wafers loss. Therefore, it is nessary to establish a semiconductor device monitoring system to instantly control device status, and let fault detection task easier to reduce unit production costs, improve product quality and enhance the competitiveness of enterprises. In this study, the 8-inch wafer fab equipment were taken as a case. The daily management statistics revealed high failure rate of equipment malfunction. After analyzing the process of machine monitoring systems, a feedback mechanism has to be built to monitor the instantaneous failure of the solenoid valve, In process ends, (Fault Detection and Classification; FDC) fence should be built Bessdes, extra sensor in the equipment should be re-established to monitor parameters related to (Information technology :IT) end models,and thus allow for early warning monitoring The performance before and often establishment were compared and satisfaction of on-line users wereassessed to verify the effect of improvement. The results of this study confirmed that the improvement off process interface and establishment of the protective net may solve real-time monitoring equipment problems and improve product yield and redace equipment engineers working pressure.It significantly improve production efficiency and shorten debug time and provides a reference for future construction planning to enhance the stability of the machine operation.