運用品質管制圖在半導體機台資料校正與分析

Using Quality Control Diagram for Semiconductor Machine Data Correction and Analysis

張蓓琪(PEI-CHI CHANG)

指導教授 ： 陳榮靜

朝陽科技大學/資訊學院/資訊管理系/碩士(2016年)

摘要

在IC晶圓半導體製造過程中，品質的改善與機台良率的提升已成為半導體工廠獲利的主要指標。技術的進步固然重要，掃描式電子顯微鏡工作站(SEM)為主要晶圓檢測站點，重要關卡其確保產品之良率維持在應有之水準更為重要。晶圓製造之步驟通常都超過百道，而每一道製作過程都需要仔細監控。可能因其中某一道程序出了偏差而浪費昂貴的原料。球面像差校正不準確會讓機台良率不正確，而直接影響良率品質。在現今製程越做越小的情況來看，維持高良率相當的重要。現階段，是以人工進行手寫數據資料校正工作，長時間的檢測，會增加檢測人員的誤判。依但造成誤判，良率不正確會讓損失提高，增加製造成本。為了確保球面像差校正準確，資料數據精準，晶圓製造過程減少損失，穩定良率。本論文針對SX機型(某機型代碼)，自動化帶出參數平均值 ，再利用2013年、2014年、2015年資料做分析，用誤差百分比公式與標準差公式算出校正值。並利用品質管制七大手法的管制圖帶入常態分配圖，減少誤差。用管制圖辨別電子束產生之二次電子的球面像差是否有機台異常之現象，有效減少人工方式做紀錄的誤判，清楚知道校正Statistical Process Control簡稱SPC統計製程控制安全範圍(±)是否有超出異常值，有效改善失控機台校正，增加機台良率品質。

關鍵字

球面像差 ； 誤差百分比 ； 標準差 ； 信賴區間 ； 管制圖

並列摘要

IC wafer in semiconductor manufacturing processes, quality improvement and upgrading quality have become the main indicator of the semiconductor factory profit. The advance technologies are important but scanning electron microscopy workstation (SEM) is still as the main wafer important points of inspection sites. To ensure that the yield of their products is maintained at the proper level the wafer fabrication is worse than one hundred steps. Every production process requires careful monitoring, which may be due to a deviation of a channel program and wasting expensive material. If the correction data of spherical aberration quality inaccurate or incorrect, it will be a direct impact on quality. There are about correction in a machine in a year. The manufacture processes are getting very small, the quality becomes quite important. Good wafer quality is direct impact on the of operations production line. However, the current pressing is carried out by handwritten data. Prolonged testing, it will increase the false detection of staff, and caused a miscarriage of justice. Then the worst quality increase lead to increasing manufacturing costs. To ensure accurate spherical aberration correction, accurate information and data, wafer manufacturing process to reduce losses, and stable quality In this thesis, we focus on SX model to automatically bring out the parameters mean, We use 2013, 2014, 2015 data that about to do the analysis. Then use the percentage error formula and formula to calculate standard deviation correction values and to use statistical methods to establish the graph, to reduce errors. The control charts is correct to judgment the secondary electrons generated from the electron beam spherical aberration to detect abnormal organic (SPC + -). Cutis effectively to reduce the artificial way to do record misjudgment and to, well aware of the correction SPC safe range (+ -) If there is beyond exception, effectively improve the machine calibration work, the system can increase yield quality of machine.

並列關鍵字

spherical aberration ； percentage error ； standard deviation ； confidence intervals ； control charts

參考文獻

[2] 陳復生，半導體廠SEM作業員視覺疲勞之評估改善碩士論文，2005年清華大學數位人因工程研究室

[14] 陳鴻基李有仁，全面品管於軟體開發之應用，資訊管理研究，第四卷，第一期，2002年1月

[18] 李東穎王復德陳瑛瑛，應用管制圖偵測群突發事件，國立陽明大學 醫學系，感染控制雜誌，第十七卷第四期

[24] Beane, Rachel Using the Scanning Electron Microscope for Discovery Based Learning in Undergraduate Courses, Journal of Geoscience Education, vol 52 #3 ( 2004)pp250-253

[25] Czochralski J. Ein neues Verfahren zur Messung der Kristallisationsgeschwindigheit der Metalle. Festschrift fur physikalische Chemise. (1918) pp219–221