A new approach to fabricate the fine redistribution layer (RDL) for 2.5/3-D chiplet integration systems using a damascene planarization patterning process is demonstrated. A smart chemical mechanical polishing/planarization (CMP) process was developed, assisted by deep learning, to enable hybrid polishing of several organic dielectric materials and Cu metal interconnections. A smart system was developed that can predict the polishing rate with an accuracy of about 10% for Cu and photosensitive insulating resin under less than 20 learning conditions. In parallel with this, the CMP of Cu/photosensitive insulating resin that was performed confirmed the formation of wiring patterns.

The Patterned Photosensitive Dielectric Organic Material/Cu Simultaneous Novel CMP Process for Fine Damascene RDL Based on Process Design Assisted by Deep Learning