X-ray analysis reveals the internal morphology of solder joints, solder voids, interface condition and crack location. It is the first-line method for failure analysis of hidden joints such as BGA and QFN. It shows morphology, but pinpointing the failure interface requires cross-sectioning alongside.
What X-ray can show
Solder volume: the number and proportion of voids inside the joint; a high void ratio is an important driver of cracking.
Crack location and width: wider cracks appear as linear dark regions in a two-dimensional image.
Pad-to-solder bonding: whether there are non-wetted or poorly wetted areas.
Solder ball morphology: whether there is tilt, collapse or offset.
Multi-layer structures: CT or multi-angle imaging allows layer-by-layer observation and resolves projection overlap.
Limitations of X-ray
Conventional X-ray is a two-dimensional projection, so multi-layer package structures (such as the bottom pad and top balls of a QFN) overlap and interfere with judgement.
Micro-cracks narrower than the resolution may not be visible.
It cannot determine the nature and depth of a crack — that is a question for cross-sectioning.
CT (computed tomography) increases both cost and lead time.
Choosing between 2D X-ray and CT
2D X-ray: low cost and fast, suited to void-ratio statistics on BGA and screening of obvious cracks.
Angled imaging: separates some overlapping structures and adds observation at specific angles.
CT (industrial CT): three-dimensional reconstruction with layer-by-layer views of hidden structures, suited to complex packages and failure localisation.
How to choose: package complexity, how suspect the failure location is, and how deep the information needs to go.
Frequently asked questions
Can X-ray show the cause of failure directly?
No. X-ray provides morphological evidence — you can see voids and cracks, but why they exist requires cross-sectioning and composition analysis. A complete analysis usually runs X-ray (localisation), then cross-sectioning (interface), then EDS (composition).
What void ratio counts as out of specification?
Industry practice usually takes 25% as the acceptable upper limit (the IPC-A-610 requirement for most devices), but high-reliability applications such as automotive and military are stricter; the customer standard governs. The measurement method and sampling rules must be specified, because different calculation methods give very different results.
Service notes
Our laboratory holds both CMA and CNAS accreditation, covers about 10,000 m2 and runs more than 500 sets of equipment. All testing is carried out in our own laboratories — nothing is subcontracted. Plant visits and on-site witnessed testing are welcome.
Scheduling follows the actual project pipeline. You can state your preferred lead time when you submit samples; we will confirm a feasible plan against current capacity and keep you updated at each milestone. Once the report is issued we send the electronic copy; corrective-action advice can be arranged separately.
