許多的半導體生產製程,會採用破壞性檢驗的方法收集製程監控的數據,因而花費高額檢驗成本,多次抽樣平均數管制圖(Multiple sampling (平均值)X charts; MS)能有效的降低抽樣數量,改善破壞試驗的檢驗成本,並維持原有修華特平均數管制圖的製程監控成效特性。本研究從統計的觀點設計MS,並評估MS與修華特管制圖的製程監控表現,最後實際應用本研究提出的MS來監控IC封裝的焊線接合製程,以驗證其改善破壞性檢驗樣本數的效益。
In semiconductor processes, many data of process were obtained by the destructive testing methods which will incur high inspection cost. Multiple sampling (average)X charts (MS) can effectively reduce sample size and simultaneously maintain the same performance as Shewhart control charts. In this study, MS were designed based on a statistical viewpoint, and evaluated their monitoring performance in comparison with Shewhart control charts. A real case of wirebonding process of IC packaging was used to demonstrate the effectiveness of proposed MS in reducing the sample size of destructive testing.