Research On Surface Defect Detection System For Copper-covering Of High-density Ultra-thin Flexible Integrated Circuit Packaging Substrate

Flexible Integrated Circuit Substrate(FICS)is a printed circuit board which is etched by copper foil on the surface of polyimide dielectric materials to form circuits.FICS manufacturing process involves three key processes of IC manufacturing,packaging and assembling.It has the characteristics of light weight,small size,high density and flexibility.It has been widely used in aerospace,military,mobile terminal and other fields.With the requirement of miniaturization of electronic products,the size of FICS circuit has reached micron level,and with the development of IC technology to nanometer level,the control of quality and defects in the manufacturing process is becoming more and more stringent.Traditional manual visual inspection method and ordinary CCD camera automatic optical detection system have been unable to meet the need of high density and ultra-thin FICS precision detection.In view of these difficulties,a FICS defect detection system based on ultra-precision micro-imaging is built in this paper.Aiming at the common defects of FICS copper cladding,such as surface oxidation,the high-precision detection algorithm was studied.The main work of this paper is as follows:(1)To meet the demand of high-density and ultra-thin FICS precision detection,a FICS defect detection system based on ultra-precision micro-imaging is built,and an improved Dual CCD combined second-generation system based on metallographic microscope and zoom mirror is introduced.According to the surface defect characteristics of FICS copper cladding,a stable light source system was designed.The image acquisition structure was improved by using the dual CCD system of metallographic microscope and zoom mirror,and the motion platform of the system was improved with the XYZ axis motion module.(2)A method of copper foil surface segmentation based on local histogram constraint is applied.Firstly,the adaptive threshold is calculated by local histogram constraint algorithm,then,the threshold is used for binary segmentation,finally,the image post-processing is processed,and the complete copper-clad surface is obtained by morphological filtering and hole filling.(3)An oxidation defect detection algorithm based on gaussian model is designed.Firstly,the distribution of oxidized pixels in different color spaces is analyzed.Then,gaussian model is established according to the oxidized pixel characteristics of YCbCr color space.Each oxidized pixel is extracted by threshold segmentation of the Gauss model,and the oxidized defect region is obtained by post-processing.(4)The 8-dimensional color features based on RGB,HSV and partitioning strategy is designed,and A three-layer DAG-SVMS classifier to classify the oxidation degree of oxidation defects is built.Firstly,for the detected oxidation region,75% of the central region of the oxidized rectangle region is intercepted,and then its 8-dimensional color feature vector is extracted to input the trained DAG-SVMS classifier for classification and tag feedback.This paper provides a precise defect detection system for the manufacturing process of high-density and ultra-thin flexible integrated circuit packaging substrate,and solves the problem of automatic defect detection on copper clad surface,which has certain theoretical significance and engineering practical value.