Yield Improvement in Color Filter Manufacturing Using Taguchi Methods and TRIZ’s Substance–Field Analysis

Thin-film transistor liquid crystal display (TFT-LCD) has been widely applied to electronic merchandise in daily lives. All panel makers are trying every possible way to produce high-quality products to fight for market share. However, the particle issue in the clean room often causes scrapped panel and increased costs. In the past, engineers in factory usually carried out quality improvement via the trial-and-error and empirical method, yet the results were rather poor, which often delayed the schedule of mass production of new product. A problem-solving procedure has been proposed in this paper, and we use a TFT-LCD maker as an example to explain the effectiveness of this procedure. The particle and gel defects in the process critical to quality are defined via the historical data analysis of the case company. The statistical analysis is utilized to screen out the factors with a significant impact on defective color filter (CF), Taguchi methods are used to find out optimal parameters combination, and the substance-field analysis in the Teoriya Resheniya Izobretatelskikh Zadatch (Russian) and 76 standard solutions come up with the optimal improvement plan to verify the effectiveness of improvement. Finally, the obtained improvement plan is standardized, and a relevant operating standard is revised. The results indicate that the process yield of CF has been improved, which has led to a hard saving of more than U.S. $3.7 million.