SPC在半導體機台微粒數的運用

Abstract

隨著半導體產業競爭越來越激烈，製造業者無不致力於提昇生產能力，而晶圓的良率即是衡量生產能力的一個重要指標。實務上，提升良率可從兩個方向著手，一是管制製程，二是管制機台。統計製程管制(statistical process control, SPC)是製造現場中，幫助提昇產品製程良率的最實用工具之一。傳統上，一般採用SPC中的缺陷點管制圖(c-chart)應用於半導體機台微粒數的管制，藉以管制機台是否有異常，但此管制圖的前提假設就是資料分配要符合Poisson分配。然而，在群聚或不明原因而導致微粒數目分配不符合c-chart的前提下，假警報率往往過高，而使現場人員無法透過微粒數來管制機台。因此，本研究提出結合資料轉換和Neyman分配的方法來構建一個快速方便的管制流程，藉以降低假警報率並使現場人員能快速的監控機台。本文以新竹科學園區某半導體製造公司所提供的實際資料為個案，來說明本研究所提方法的可行性與有效性。

With increasing competition in the semiconductor industry, semiconductor manufacturers are making efforts to increase their productivity. The yield on each wafer is an important index to evaluate productivity. To enhance the yield of IC products, statistical process control (SPC) is the most useful tool in semiconductor manufacturing. The c-chart of SPC has traditionally been used to monitor machine particle counts, thereby controlling the machine condition. However, the clustering phenomenons and unknown factors cause the Possion based c-chart invalid. This study combines data transformation and Neyman distribution to develop a control procedure to monitor machine condition. A case study from a semiconductor company in Taiwan is demonstrated to verify the effectiveness of this proposed method.