應用統計手法於TFT LCD配向膜製程之最佳參數設計

Abstract

隨著3C電子產品(手機、GPS、NB…等)中小尺寸面板的需求急速擴充，導致中小尺寸供應商產能吃緊，迫使產品良率與設備產能利用率必須提升。有鑑於此，如何應用現有生產設備以因應市場需求的快速變化成為多面板廠的主要課題，另一方面此一類客戶產品規格的需求不斷之提升，包含高對比、LC反應速度高、結合Touch panel、電子紙、OLED、LED、太陽能運用，已經成為中小尺寸面板廠最新挑戰。在TFT LCD製造的過程中，液晶分子於玻璃面板上的配向之良率是主要的瓶頸，如何提高此一製程的良率是最直接被要求立即改善項目，本研究提升藉由統計手法將異常現象進行分析，其中包括材料研究（液晶、布毛）、設備穩定性、製程參數最佳化等…深入且系統化的探討。 本論文的研究方向主要在探討液晶與聚亞醯胺(Polyimide；簡稱PI)配向膜摩擦(Rubbing)加工參數對上、下PI配向膜配向不同組成液晶空間之良率損失之影響，以了解PI受配向布壓入量不同之表面型態變化之研究。利用群組腦力激盪方式，找出機台相關參數及液晶種類配合DOE(Design of Experiment)，及統計手法與分析軟體(Minitab & ANOVA)來進行實驗計劃的設計，以4因子(Roller rotation/Stage speed/LC type/Rubbing count)與2水準以確保其交互作用關係。綜合此研究已可明確了其實驗結果與參數關係與重要性。找出Rubbing Process最佳化製程參數，以降低良率損失與提高產品品質為目標。在95%信賴區間下分別依實驗數據中為HLM與Sandy為影響良率最大兩項損失。最後得到兩項結論，影響HLM(H line mura)，考慮單一因子為Rubbing speed，其交互因子為Roller speed*Rubbing time為重要影響因子。另外影響Sandy考慮單一因子為LC type。

Along with the extended application of LCD panels on the products such as mobile phones, GPS and NB etc., the yield rate and the capacity of medium-small sized panels are forced to be improved in order to raise the productivity. The subject of adopting existing facilities to the rapidly increasing LCD market has been discussed among LCD manufacturers. Furthermore, the specification of the LCD panels are asked to upgrade from time to time e.g. better contrast, LC reaction time, Touch screen, E-paper, OLED, LED, solar application, all of these requirements are a great challenge to the medium-small sized panel manufacturers. One of the major problem of the production of TFT LCD is the low yield of alignment of liquid crystal cells. Methods for improving the yield rate are always very important and are treated as one of the critical competitive actions in the industry. In this study the failure modes of the panels were analyzed by using statistics, and the parameters discussed including TFT-LCD, clothes, stability of facilities. Optimization of the process parameters were discussed in detail in this thesis. To investigate the yield loss, the alignment of PI-type rubbing membrane and rubbing processing parameters of liquid crystal were taken into consideration. Moreover, optimization methodologies e.g. DOE(Design of Experiment) were applied for finding suitable combinations of processing parameters. Minitab and ANOVA were used to process the experiment plan. Four factors, namely Roller rotation, Stage speed, LC type, Rubbing count were selected as the parameters. The importance of the processing parameters were proved from the experiment result. It was found that rubbing process parameter is the most significant factor the yield and improving the product quality. Besides, HLM and Sandy are major defects in the experiments. Final findings found that roller speed, rubbing time were the main factor affect the HLM(H line mura). On the other hand, LC type alone was the most significant fact that causes the Sandy defect.