薄膜電晶體液晶顯示器線中欠陷檢查之分析與分類

Analysis and Classification of In-line Defect Inspection for Thin Film Transistor Liquid Crystal Display

李銘周(Ming-Jiu Lee)

指導教授 ： 劉益宏

中原大學/工學院/機械工程研究所/碩士(2008年)

https://doi.org/10.6840/cycu200800013

摘要

近十年來，液晶顯示器的迅速發展，幾乎取代傳統陰極射線管的應用市場，光電產業的總值也超過半導體產業，已成為政府積極推動的明星產業。此份技術報告將會介紹液晶面板製作的流程，包括陣列工程、面板工程及模組工程。透過流程的介紹，來說明線中檢測的重要性及方法。
目前陣列工程中所使用到的線中檢查機台，共分為圖案缺陷檢查機、圖案缺陷觀察機、斜光目視檢查機及吋法檢查機等四種機台，透過線中檢查機台來檢查出面板製造過程中所會產生的欠陷，並且將欠陷按照所發生的工程來分門別類，並且分析其欠陷發生的原因，如此一來，便能提供工程人員解決良率問題，提高公司效益。
未來線中檢查站的發展重點，主要在於自動化及知識管理，來提高線中檢查站的檢查效率，並且提昇生產良率及減少生產成本；另外未來檢查機台的發展重點，主要在於檢查機台大型化、量測精度準確化及稼動效率提高，來符合未來大型玻璃基板的稼動與量測，這是未來量測機台的設計趨勢。
最後探討未來研究方向，主要在於知識管理系統及自動化欠陷分類系統兩方面。在面板製造過程中，事實上仍有多種欠陷從未見過。如此一來，就會影響線中欠陷分析的準確度。為了解決未見欠陷的問題，面板廠需要建立一套知識管理系統來加強線中欠陷的分析及判定；另外未來線中欠陷分類，可藉由整合各種數位影像處理技巧、統計紋理特徵抽取、資料探勘及類神經網路辨識等方法，針對檢查機台所拍攝的欠陷影像，進行自動即時分類，也就是自動化欠陷分類系統。

關鍵字

線中檢測 ； 欠陷分類 ； 液晶顯示器

並列摘要

Over the last decade, thin-film-transistor liquid-crystal-display (TFT-LCD), as display, has grown rapidly, and has almost replaced the market of cathode ray tube (CRT) monitor. The total value of photoelectric industry has exceeded the one of semiconductor. It has been a glaring industry that the government promotes actively. The manufacturing process of TFT-LCD, including array process, cell process and module assembly process, will be introduced in this technical report. Through the introduction, the importance of inline defect inspection in array process will be pointed out and detailed.
In current practice, there are several equipments used for in-line inspection: automatic optical inspection (AOI) machine, micro-review machine, macro/micro machine, and critical-dimension overlap machine. These machines are used for inspecting various kinds of defects in different processes of the manufacturing. The defects are classified and analyzed according to the corresponding engineering. Therefore, the task of in-line inspection is able to enhance the yield rate and increase the business benefit.
To achieve the goals of inspection efficiency, yield enhancement, and the production-cost reduction, the future works for inline inspection will be focused on automation and knowledge management. As for the inspection equipment, the key issues will be focused on several points, including how to enlarge the size of equipment, how to improve the accuracy of the measurement, and how to lift the efficiency. These are adapt to the large-size LCD panels, and are the future trends for the design of inspection equipment.
In the final we will discuss the future works including the knowledge management and the automatic defect classification. In TFT-LCD manufacturing, there are some unknown defects actually. Due to unknown defects, a high and satisfactory defect classification rate can not be obtained. For solving this problem, it is necessary to develop a knowledge management system to improve the defect classification accuracy. In addition, several techniques, such as image processing, statistical texture extraction, data mining and neural network-based classification, can be used and integrated to achieve the goal of real-time defect recognition. The above-mentioned refers to the automatic defect-classification system.

並列關鍵字

defect classification ； Liquid Crystal Display ； Inline Inspection

參考文獻

[15] 林思賢，“ 應用視覺及資料探勘技術於TFT-LCD陣列電路工程線中瑕疵辨識之研究-源/汲電極光罩瑕疵自動分類系統開發，” 中原大學機械工程研究所，2006。

[19] 方育斌，“ LCD背光模組之光學最佳化設計，” 成功大學工程科學研究所，2004。

[1] 顧鴻壽，光電液晶平面顯示器-技術基礎及運用，新文京開發出版股份有限公司，2004年3月。

[2] 顏丹青，“ 次世代液晶顯示器用彩色濾光片元件之平坦化製程技術開發與擬真度評價指標系統，” 清華大學動力機械研究所，2006年。

[3] 戴亞翔，TFT-LCD面板的驅動與設計，五南圖書出版公司，2006年4月。