多特性多量測點資料管制流程之構建

Abstract

隨著製造技術進步以及顧客對產品愈趨嚴格的品質要求，管制產品時所需量測的資料特性及數量也跟著複雜起來。以晶圓廠為例，在管制這類儀器成本高、製造過程繁複的高科技產品時，對產品作多點或多特性的量測已成為業界普遍的管制方式，因此，單靠一般傳統管制圖已無法達到確切的管制功能。 本研究之主要目的即是藉由主成份分析及多變量管制圖來瞭解多個品質特性彼此之間以及多個量測點間的關係，建構出一套適用於管制多特性多量測點資料之管制流程，並期望在管制圖出現異常點時，能夠迅速有效地回溯失控情形發生的相關訊息。本研究除了以虛擬之案例說明回溯失控點之步驟外，並以新竹科學園區某積體電路製造廠所提供之實際資料說明如何應用所發展之管制流程，此管制流程除了能有效地簡化管制工作外，亦有助於工程人員對產品整體品質的判讀，同時提高管制效率及管制結果的可靠性。

With the improvement of manufacturing technologies and severe demand of customers, high quality assurance has become much more important nowadays. For instance, quality control on multiple readings form multiple quality characteristics has been a common problem to control wafer quality for a wafer fabrication. The process of controlling such kind of measurements is usually very complicated and the traditional Shewart control charts such as x-bar & R charts are no longer suitable. This main objective of this study is to construct a novel procedure to control multiple readings from multiple quality characteristics by using principal component analysis (PCA) and multivariate control chart. The proposed procedure allows the analyst to trace back information related to the out-of-control sample points quickly and effectively. Engineers can accurately judge the overall quality of products. The efficiency and reliability of the proposed control procedure can improve the quality control work for the manufacturing company, especially for the high-tech industry. Three virtual cases and one real case provided by an integrated circuits manufacturer in Taiwan demonstrate the application of the proposed procedure. 1.1研究動機與目的 1 1.2研究架構與流程 2 第二章 文獻探討 4 2.1與積體電路相關文獻探討 4 2.2主成份分析法原理簡介 9 2.3 Hotelling T2多變量管制圖簡介 10 第三章 多特性多量測點資料管制流程之構建 13 3.1研究步驟 13 3.2以虛擬資料試行失控點之回溯 16 第四章 實例說明 38 4.1晶圓允收測試簡介 38 4.2測試特性說明 38 4.3研究流程 39 4.4與業界現行管制方式之比較 44 第五章 結論 46 參考文獻 48