Using RBF algorithm for Scanning Acoustic Microscopy inspection of flip chip

摘要

With the increasing prevalence of flip-chip technology in high density assembly, more attention has been paid to the microbump defect inspection in the flip chip. However, the traditional techniques have some disadvantages, which makes it difficult to diagnose the chips efficiently. Therefore, new inspection approaches for solder bump are being investigated. In this article, the radial basis function neural network (RBF) combined with the scanning acoustic microscopy (SAM) technology is proposed for inspection of missing microbump defects. The ultrasonic transducer was used to test the flip chip. The solder joints were segmented from the SAM image on the basis of gradient of gray scale, and representative characteristics of micro solder balls were extracted. Then RBF network was used for identification. The results reflect that RBF algorithm with high recognition rate is effective for diagnosis of missing-bump defects.