Crack Defect Detection For Flexible Integrated Circuit Package Substrates

Flexible integrated circuit substrate(FICS)has the characteristics of light,thin and flexible,which is suitable for the miniaturization of electronic products in the fields of aerospace,military,and mobile terminals.With the gradual shrinking of the FICS circuit size,its detection accuracy has reached the micron level,and even some areas require nm-level detection accuracy,especially the trace defects on the FICS.If they are not detected,their service life will be reduced.It will directly cause the product to be scrapped.The commonly used manual visual inspection method and visual inspection method cannot meet the requirements due to their low detection accuracy,and have a high rate of missed detection.In contrast,simple microscopic inspections have a low detection rate due to the limitation of the field of view.In view of the above difficulties in FICS detection,this paper builds a double-precision detection system,and designs detection algorithms with different precisions for trace defects.The main work of this paper is as follows:(1)In response to the increasing detection accuracy of FICS,a double-precision detection system that can freely switch between CMOS industrial camera and microscope is designed.After micron-level detection is performed by using CMOS industrial camera,nm-level detection is performed on the gold surface area to ensure detection accuracy.At the same time,the detection speed is improved.(2)Aiming at the useless texture information and random noise of FICS images,this paper introduces a gradient matrix to optimize the existing Directional Relative Total Variation(DRTV)filtering algorithm.Experiments show that this method can FICS useless texture information and random noise are removed.(3)Aiming at the feature of uneven grayscale of mark defects on low-resolution FICS images,a mark defect detection algorithm based on RANSAC is proposed.Experiments show that this method can connect intermittent trace defects under low resolution conditions,and completely mark trace defects.(4)Aiming at the problem of low detection efficiency of mark defects on high-resolution FICS images,this paper proposes a mark defect detection algorithm based on wavelet transform.This method uses Haar wavelet to transform the characteristics of the gold surface,extracts its high-frequency signal for analysis,and finally obtains the location of the defect.It is verified through experiments that this method can distinguish FICS trace defects from other defects under high-resolution conditions,and mark them at the same time.This paper provides a suitable defect detection system for the manufacturing process of flexible integrated circuit packaging substrates.At the same time,it designs corresponding trace defect detection algorithms for different detection accuracy,and verifies the feasibility of the algorithm through practice,which has certain theoretical significance and engineering practical value.