Rapid Capture, Analysis, and Storage of Integrated Circuit Images for Defect Analysis

Abstract

An automated rapid capture, analysis and storage system for identifying chip defects is essential to ensure Intel's "Copy Exactly" methodology with low labor cost and no human subjectivity. In this system, the software will identify where the defects are with a standardized definition of defects. Constraints include the 80% defect detection rate, and Windows XP compatibility. There should be no defects that can be observed by human eye in the product shipments. Three design concepts were generated and two of them are combined to form the final design. The final design consists of a line scan cameras and two LED arrays with different wavelengths. The software design is discussed under detection and compression algorithms. Based on the software testing done, the software can identify on average 96% of the defects and compress the image up to 22 times over the current bitmap format.