Detection of Faulty Semiconductor Wafers using Dynamic Growing Self Organizing Map

• Main Authors: Russ, G., Kruse, R., Karim, M.A., Hsu, A.L., Halgamuge, S., Smith, A.J., Islam, A.
• Format: Conference Proceeding
• Language: English
• Subjects: Clustering algorithms; Data mining; Data visualization; Fault detection; Informatics; Manufacturing processes; Neurons; Organizing; Semiconductor device manufacture; Time to market
• ISSN: 2159-3442; 2159-3450
• DOI: 10.1109/TENCON.2005.301056

Solving product yield and quality problems in a manufacturing process is becoming increasingly more difficult. There are various types of failures and their causes have complex multi-factor interrelationships. Semiconductor manufacturing is very complex due to the large number of processes, diverse equipment set and nonlinear process flows. Its manufacturing database comprises of hundreds of process control, process step and wafer probe data. This huge volume of data coupled with quicker time to market expectations is making finding and resolving problems quickly an overwhelming task. In this study, a methodology developed using dynamic growing self-organizing map (GSOM) to detect the faulty products in a wafer manufacturing process. As part of the methodology, a clustering quality measure was developed to evaluate the performance of the algorithm in separating good and faulty products. Results show that the algorithm was able to separate good and faulty products from the raw data. Even though this work has focused mainly on clustering good and faulty products, the technique can be extended to model the failure causes of the lower yielding products.