Characterization of Solder Defects in Package-on-Package with AXI Systems for Inspection Quality Improvement (Part 2)

Estimated reading time: 7 minutes

Our experience tells us 2DX and/or CT are good non-destructive method to verify BGA and POP defects, especially for HIP/Open. However 2DX and CT are rather considered manual testing tools. It is desirable to have an AXI machine with 2DX and pCT (planar CT for large board) capabilities. Recently we work with AXI 5 team, and test several packages which have challenging to achieve good results on the AXI systems, the POP being one example.

Computed tomography (CT) inspection technology has long been successfully applied to conduct failure analysis projects of PCB assemblies. In conventional CT for failure analysis, effective sample rotation of 360 degrees is necessary to acquire x-ray projected images and reconstruct 3 dimensional volumes with great structural details. However, this CT procedure usually takes too long time to routinely employ at production line environment. Therefore, 2D AXI inspection is more common so far on the plant floors.

AXI 5 with planar computer tomography (pCT) acquires a stack of slices which is typically 100 slices, but it can be more whenever necessary. The distance between 2 slices is topically 20µm, but it can be less for higher resolution. From this stack of slices, the 3D reconstruction is done in real time on the machine PC while scanning. AXI 5 is especially good for inspecting 3D packages, like POP which require 3D-CT volumetric inspection for quality assurance. AXI 5 employ planar CT as well as other new approaches like limited angle CT, etc., to shorten the acquisition and reconstruction speed. With planar CT approach, the speed comes down to 16s for 35x35mm² area FOV at 20µm voxel resolution. The system also has selective programmable inspection for any component with planar CT to catch defect in wide area, and operator independent pass/fail judgments including CT results/images which is very attractive feature for SMT lines.

Figure 1 shows AXI 5 images for POP and HIP. Shape difference between good solder and bad solder pins for POP and BGA can easily be seen, and the speed with image qualities are comparable to normal 2DX and CT. Same to the familiar AXI systems, AXI 5 detects defects based on Algorithm Threshold, however some measurement items of AXI 5 Algorithm can be direct measurements of absolute volume and volume ratio because of it has 3D measurement capability.

Experiments on AXI 5

With AXI 5, each solder ball can be framed to inspect under one or more measurement window. Two volume measurement algorithms were applied: (1) Absolute Volume of the ball; (2) Upper and Lower Dome Volumes for their ratio calculation (Volume Ratio). The volume measurements are done using voxel element, a benefit of pCT images: (1) Use reconstructed pCT images of a real solder ball, its volume can be directly measured in voxel units; (2) Any volume deviation from standard value indicate the defects like open, less wetting (HIP), as well as potential bridging of adjacent pins.

The ideal solder ball is perfect sphere. If ideal sphere, the solder volumes of Upper and Lower domes will be identical and their Volume Ratio (Vld/Vud) = 1. But there is almost no perfect ideal solder ball during the SMT process. There are several factors that influence the ball shape to deviate from a perfect sphere: gravity; asymmetric contact angles due to different pads at the low and upper contacts in terms of pad sizes and material compositions.

In this project with AXI 5, we focus the top layer of POP because it was identified the most challenging area during previous AXI tests. Based on the 2DX images, 27 defective pins were confirmed. AXI 5 automatically focused the BGA balls in POP three layers, and qualities of solder balls are quantify by using volume algorithms: (1) Measure absolute volume for open / less wetting; (2) Comparing solder volumes composing spherical dome sections sandwiched between the two facing pads. The volume dome section is 30% for the Upper and Lower ball.

The total number of pins for top layer POP is 216. The Volume Ratio threshold is set as ≥1.25. The average Volume Ratio (Vld/Vud = 1.13) with maximum (Vld/Vud =1.45) and minimum (Vld/Vud =0.97). Figure 2 shows the Volume Ratio measurement data for the top layer of POP from AXI 5. There is separation between good pins and bad pins with the 216 pins of Volume Ratio measurement data.

Figure 2: AXI 5 Volume Ratio Measurement Data for the Top Layer of POP, Volume Ratio Threshold ≥ 1.25.

With the threshold as ≥1.25, total fail pins is 30, in which 4 pins are false call, and 1 pin is defect escaped based on the 2DX as the adopted reference. Table 1 is the comparison between 2DX and AXI 5 for these 5 pins discrepancy. When we look at details for these pins with X-ray measurement data, three pins are considered as on the borderline of adopted threshold in earlier study.

Table 1: Comparison between 2DX and AXI 5 for top layer of POP.

Page 1 of 2

• Next Page >