Void Measurement by X-Ray: Grey Scale and Acceptance Criteria
A void is a bubble of gas trapped inside a solder joint, and X-ray inspection is the standard way of finding and sizing it without destroying the board. The measurement looks simple: an image of the joint is thresholded, the dark areas are identified as voids and their total area is divided by the joint area. In practice every step in that chain carries an assumption, and the assumptions decide whether two machines will report the same figure for the same joint.
What Void Measurement Reports
The primary output is void percentage, the ratio of void area to joint area as seen in the image. Secondary outputs are the number of voids, the size of the largest single void and its position within the joint. Different acceptance criteria use different combinations of these.
The figure is a two-dimensional projection of a three-dimensional defect, which is the first and largest of the assumptions behind it. A spherical void near the top of a joint projects as a large dark area, while the same void deep inside a ball grid array joint may be hidden entirely by the solder above it. Every reporting method therefore has to state the geometry it assumes, and a criterion transferred between systems without that statement is not reproducible.
X-Ray Inspection Basics and Resolution
A void is detected because solder attenuates X-rays more than the gas inside the bubble, so the detector sees more intensity through the void and the image appears brighter at that point. The contrast between the two depends on the X-ray energy, the thickness of the joint and the detector characteristics.
Resolution sets the smallest void that can be detected reliably. A system with a spot size of a few micrometres resolves voids of a few tens of micrometres; a system with a larger spot blurs them into the surrounding solder. Where the acceptance criterion is written in terms of the largest permitted void, the resolution must be fine enough to measure that size with confidence. The comparison between the two main families of technique is covered in the discussion of X-ray versus AOI inspection.
Grey Scale Threshold and Void Detection
The grey scale threshold is the intensity value below which a pixel is treated as void. It is usually set as a percentage of the difference between the solder intensity and the background intensity, and it is the single largest source of disagreement between systems.
A threshold set too low includes the darker regions at the edges of the joint and overstates the void area; one set too high misses small voids near the limit of detection. The threshold should be established once for a given joint type and then held, recorded with every measurement, and re-established whenever the machine, the detector or the joint geometry changes.
Slice Height and the Projection Problem
Computed tomography systems can reconstruct the joint in three dimensions and measure voids at a chosen slice, which removes most of the projection ambiguity. Two-dimensional systems cannot, so they report the projected area of all voids along the beam path.
For a ball grid array joint the projection problem is worst, because the solder ball, the pad and the package substrate all overlap in the image. The practical compromise is to measure at a plane just above the board surface, where the solder is at its thickest and the surrounding material contributes least, and to state that plane in the report.
Calculating Void Percentage
The joint area used as the denominator has to be defined. Some methods use the nominal pad area, some the area of the solder ball at the slice, and some the convex hull of the joint as imaged. Each gives a different percentage for the same void, so a specification written as a percentage without a stated denominator is not reproducible.
The most defensible approach is to define the denominator from the design rather than from the image, so that the same joint on two boards has the same reference area. Where the criterion comes from a customer standard, the standard normally states the method and the area to be used.
Acceptance Criteria and Their Basis
Void criteria are usually expressed as a maximum total void area, sometimes with an additional limit on any single void. Values around 25 percent total are common for ball grid array joints, with tighter figures for thermal pads and for high-reliability applications.
The basis of the criterion matters more than the number. Voids reduce the cross-section that carries current and heat, and they concentrate stress at their edges, but joints with substantial voiding often pass thermal cycling without failing. Where a criterion is applied strictly, the failures it produces are usually cosmetic rather than functional, and the causes of voiding in BGA joints deserve more attention than the threshold itself.
Comparing Systems and Repeatability
Two machines will not agree exactly on the same joint unless they are matched in energy, resolution and threshold. Before a criterion is transferred from a customer’s reference system to a production system, the two should be compared on a set of joints with known void content.
Repeatability is the more important property. A system that reads slightly high but repeats within a fraction of a percent is more useful than one that is accurate on average and varies by several percent between measurements. Measuring the same joint ten times and reporting the spread is a simple check that should be part of the automated X-ray inspection setup.
Limits of Two-Dimensional Measurement
Even with careful thresholds, a two-dimensional image cannot report the depth of a void. This matters when a void is open to the surface of the joint or when it extends into the pad interface, because those conditions affect mechanical reliability differently from a void enclosed inside the solder.
Where the distinction matters, the answer comes from sectioning rather than from imaging. The microsection methods used for joint inspection show the void in profile and its relation to the interfaces. Using X-ray for screening and sectioning for investigation is the practical division of labour, and it keeps the slow destructive method for the joints that actually need it.
Records and Reporting
The report for a lot should carry the machine identification, the energy and magnification used, the threshold setting, the slice height, the denominator method and the distribution of void percentages rather than only the worst result. Those fields make the data comparable between shifts and between sites.
Trending the mean void percentage is more useful than reacting to individual readings. A shift in the mean points to a change in the paste, the profile or the component surface finish, and it is usually visible in the trend before any single joint approaches the criterion.
FAQ
What void percentage is acceptable in a BGA joint? A common criterion is 25 percent of the joint area, with tighter limits for thermal pads and high-reliability products. The number is less important than the method used to measure it.
Why do two X-ray systems disagree? Usually because of different grey scale thresholds, energy settings or slice heights. Before transferring a criterion, the systems should be compared on joints with known void content.
Can X-ray tell how deep a void is? Not in a two-dimensional image. Laminography or computed tomography can, and sectioning shows the void in profile together with its position relative to the pad interface.
