Application of new Apriori algorithm MDNC to TFT-LCD array manufacturing yield improvement

With the trend and demand of larger panels and higher resolution in Thin Film Transistor-Liquid Crystal Display (TFT-LCD) panels, yield improvement has become the key factor in TFT-LCD manufacturing. This paper presents a successful and effective data mining methodology for TFT-LCD manufacturing yield improvement. We have modified an Apriori algorithm called Multi-Dimension Non-Continuous (MDNC) by eliminating the limitations imposed by traditional pattern matching of continuous data, to mine the association rules in the cross-day discrete manufacturing data and find out some valuable information. The results show how to effectively locate any machine with low yield and vastly improve it in TFT-LCD large panel manufacturing yield-rate, thereby reducing the manufacturing cycle time, the frequency of holding lot by adaptation of MDNC.