Publishing Venue
The IP.com Prior Art Database
Abstract
Disclosed is a method to reduce the manual processes, and thus the potential for handling defects, for inspecting whole wafers. The novel method uses facial recognition software that is trained to identify the definition of a properly processed whole wafer image versus a defective wafer image.
Page 01 of 1
Automated Whole Wafer Imaging Inspection
Inspection techniques for whole wafers for gross defects are slow processes, requiring the manual handling of wafers resulting in unwanted handling defects. An improved
method to quickly and automatically inspect whole wafers for gross defects is needed.
Facial recognition software can be trained to identify the definition of a properly processed whole wafer image versus a defective wafer image.
A system and method includes using facial recognition software to train the software to inspect whole wafers to recognize the components and features of a good wafer image at particular operations. The wafer can be imaged at multiple angles, multiple rotations, and multiple light sources/frequencies to enhance the inspection's capability of identifying a good wafer versus a defective one and identify any mixed product errors. This allows the system to be automated without any manual intervention required once the system has been trained, thus improving the speed of the inspection, and significantly reducing handling defects.
1