- 學位論文
薄膜電晶體液晶顯示器產線中破片檢測及欠陷分析
The fracture glass examination and defect analysis in the TFT-LCD production line
摘要
近十年來，液晶顯示器的迅速發展，幾乎取代了傳統的陰極射線市場，光電產業的總值也超過了半導體產業，也是目前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流程，包含了陣列工程、彩色濾光片工程、面板工程及模組工程，透過流程的介紹，來說明在產線中檢測的重要性及方法。 目前彩色濾光片工程中所使用到的產線中檢查機，共分為影像缺陷檢查機、線寬檢查機、巨觀檢查機及破片檢查機等四種機台，透過產線中破片檢查機台來檢出面板製造過程中所會產生的欠陷，並將欠陷所造成的原因及設備來分門別類，如此一來，可以提供工程人員來解決良率問題，提高公司的效益。 未來產線中的檢查設備的發展重點，主要在於自動化及知識管理，來提高產線中檢查設備的檢查效率，並且提昇生產良率及減少生產成本，另未來檢查設備的發展重點，主要在於檢查機台大型化、量測精度準確化及稼動率的提高，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。 最後未來研究方向，主要在於知識化管理系統及自動化破片影像分類系統兩方面。在面板製造過程中，事實上造成破片、裂痕及非破片的欠陷檢出的原因非常多。如此一來，就會影響產線中欠陷分析的準確度。為了解決未見的破片的原因及造成非破片的欠陷檢出問題，面板廠需要建立一套知識管理系統來加強產線中欠陷的分析及判定;另外來產線中欠陷分類，可藉由整合個種數位影像影像處理技巧、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，進行自動即時分類。
並列摘要
Over the past decade, LCD has been developing so swiftly that almost replace CRT in display market. The optoelectronics industry is the star industry fully supported by the government and the total value of it has exceed that of semiconductor industry. Introduction of the manufacturing process of TFT LCD will be cited to illustrate the method and importance of the panel inspection which includes array process, color filter process, cell process and module process. In color filter process there are four types of inspectors, including automatic optical inspection machine, Micro measure machine, Macro machine and fragment inspector. Fragment inspectors are used to detect the defect of the panel caused in the manufacturing process and classify the defect by reasons and equipments. Therefore, this inspection will help engineers to improve the yield of production and the benefit of corporation. To improve the accuracy of the equipment, the yield of the productions and the cost of the productions, the study of the inspector equipments will focus on automation and knowledge management. Furthermore, the inspection equipment will be designed to adapt to large size glass substrate in three ways: equipment enlargement, measurement accuracy and utilization improvement. At last, discussions on future study focus on knowledge management and fragment classify automation. Actually there are various reasons causing fragments, split, and non-fragment defects in the process. Consequently the accuracy of defect analysis will be influenced. To solve unseen fragment defects and non-fragment defects, panel factory need to build a knowledge management system to enhance the defect analyzing and determining by immediately automatic classifying defect images integrating digital imaging processing techniques, texture abstraction, data mining and neural network.