Failure Analysis & Troubleshooting
When a bond fails, the quickest explanation is rarely the right one. We work the interface, the cure and the process together until the evidence points at a single root cause — then hand back the control points that stop it recurring. Yield comes back through engineering data, not guesswork.
Failure index
Common failure modes — and what they usually turn out to be
Twelve failures we are asked about most, written the way engineers describe them rather than the way a report classifies them. Each one gives the causes we find most often and how we confirm which one it actually is. They sit in the four layers we work through — interface, cure, build, and the load the product sees — and which layers we open depends on what your evidence points at.
Adhesion — the interface
Did the joint ever bond properly in the first place? We read the fracture surface, the substrate surface, and whatever was sitting on it before the adhesive arrived — release agents, coatings, contamination, and whether the surface treatment was still doing anything by the time the parts reached the dispenser.
What we usually find
- A clean peel means the joint let go at the interface, not inside the adhesive — so this is an adhesion problem, not a strength problem.
- The substrate has low surface energy, so the adhesive never properly wetted it.
- A mould release agent, plasticiser or anti-static coating has migrated to the surface.
- A surface treatment step exists, but its effect decays before the parts reach the dispenser.
How we confirm it
- Read the fracture surface first: interfacial failure tells us which side of the joint to investigate.
- Wetting behaviour on incoming parts, and contamination screening on the failed surface.
- Treated vs untreated coupons from the same lot, bonded side by side.
What we usually find
- Initial strength comes from wetting and mechanical interlock; durability depends on whether the interface survives moisture, heat and sustained load. The two are not the same property.
- The bond relied on a surface treatment that was never chemically anchored, so moisture displaced it over time.
- Low-molecular-weight species migrated out of the substrate and accumulated at the interface.
- The adhesive was cured enough to hold on day one, but never fully cross-linked.
How we confirm it
- Aged units against retained samples from the same build — the comparison is where the evidence is.
- Whether the failure mode shifts from cohesive to interfacial after aging.
- Cure state on the retained parts, to separate an aging problem from an under-cure problem.
What we usually find
- The process is stable. The incoming surface is not.
- Different mould release or lubricant practice at the second supplier.
- The same resin grade, but a different filler or additive package underneath it.
- Different storage and handling, so parts arrive at the line with a different surface age.
How we confirm it
- Side-by-side surface and wetting checks on parts from both sources.
- Incoming material documentation compared line by line, not by trade name.
- A controlled build in which the part source is the only thing that changes.
Cure
Did the adhesive reach the state it was supposed to? Dose and shadow against the assembled geometry, mix and mechanism against the joint it has to fill, and whether the cure that was qualified is the cure the line is still delivering today.
What we usually find
- Light-cure chemistry cures only where light reaches. If part of the bond line is shadowed, that part is still liquid — whatever the timer says.
- The joint geometry shadows the fillet once the parts are closed.
- One substrate blocks or absorbs the wavelength the adhesive needs.
- Lamp position is correct on the fixture drawing but not in the fixture that is actually running.
How we confirm it
- Map where light actually lands on the assembled part, not on an open bead.
- Check substrate transmission at the lamp's wavelength.
- Test whether the material has a secondary cure mechanism able to reach the shadowed volume.
What we usually find
- A tacky skin over a solid bulk is usually oxygen inhibition at the exposed surface — the joint itself may be fine.
- The exposed fillet receives a lower dose than the bulk of the bond line.
- In two-part systems: mix ratio, mixer condition, or material past its shelf life.
- The cure mechanism chosen was never intended for an exposed surface.
How we confirm it
- Test surface tack and bulk cure separately — they are different questions.
- Compare the exposed geometry against what the cure mechanism can actually do.
- Review material dating, storage and mixer handling.
What we usually find
- Something upstream is drifting while the set points stay the same.
- Lamp output falls as the source ages, so identical settings deliver less energy than they did at qualification.
- Ambient temperature and humidity move with shift and season, and several chemistries are sensitive to both.
- Open time between dispense and cure varies with operator pace and line stoppages.
How we confirm it
- Log what the process actually did across several batches, rather than what it was set to do.
- Correlate the drift against the batches that failed.
- Re-run with the suspected variable held fixed, and see whether the variation disappears.
Dispense & fixturing
Was the joint built the way it was designed? Bead pattern, dispensed volume, bond line gap, clamp sequence and open time — measured on real assembled parts rather than read off the programme.
What we usually find
- The air is almost always introduced before cure, not generated by it.
- Air entrained in the cartridge, the feed line or the static mixer.
- The bead pattern traps air as the parts close.
- The joint gives displaced air nowhere to escape.
How we confirm it
- Section assembled parts and look at where the voids sit relative to the bead.
- Compare bead pattern against closing direction and clamp sequence.
- Re-run with a pattern and vent path that let air out, and check whether the voids follow.
What we usually find
- Bond line thickness is set by the parts and the fixture, not by the dispenser.
- No bond line control feature in the joint, so clamp force decides the gap.
- Part flatness varies within its own drawing tolerance.
- The clamp sequence tilts the part before the adhesive is captured.
How we confirm it
- Measure the gap at several locations on assembled parts, not on the drawing.
- Check incoming part flatness against the assumption the joint was designed on.
- Test whether a stop feature or a changed clamp sequence removes the variation.
What we usually find
- Dispensed volume and actual gap are not matched consistently — one is fixed, the other moves.
- Volume was set for a nominal gap the parts do not hold in practice.
- Viscosity shifts as material temperature changes through the shift.
- The bead sits close enough to an edge that it escapes under clamp.
How we confirm it
- Weigh dispensed shots to see the true volume variation rather than the programmed one.
- Track material temperature at the nozzle across a shift.
- Re-run with bead placement moved in from the edge.
Environment & reliability
Does the joint survive what the product actually sees? Expansion mismatch across dissimilar materials, impact and peel loading, moisture ingress and thermal aging.
What we usually find
- Thermal expansion mismatch between two dissimilar substrates, carried by a bond line too rigid or too thin to absorb the movement.
- A stiff, high-strength adhesive chosen for its pull figure, joining materials that move at different rates.
- A bond line thin enough that all the strain concentrates at the interface.
- Sharp corners and stiff features acting as stress risers.
How we confirm it
- Work out the expansion mismatch for the actual material pair and joint geometry.
- Look at where the cracks initiate, which usually names the mechanism on its own.
- Test whether a more compliant chemistry or a controlled bond line thickness removes them.
What we usually find
- The two tests load the joint in completely different ways. Static pull loads it slowly and evenly; a drop loads it fast, in peel and cleavage, at whichever edge is weakest.
- An adhesive with high tensile strength but low toughness.
- An unsupported edge that gives a peel somewhere to start.
- A joint stiff enough to transmit the shock rather than absorb it.
How we confirm it
- Find where the failure starts in the dropped units, not just that it failed.
- Compare failure modes between the two tests — they usually disagree, and that disagreement is the answer.
- Screen candidate chemistries by impact behaviour instead of static strength alone.
What we usually find
- Moisture is reaching the interface and displacing the bond there.
- The interface bond is physical rather than chemical, so water can get underneath it.
- The adhesive itself takes up moisture and softens.
- An exposed edge gives water a direct path to the interface.
How we confirm it
- Aged and unaged samples from the same build, tested the same way.
- Whether the failure mode shifts from cohesive to interfacial after aging — that shift is the fingerprint.
- Test whether a coupling treatment or a sealed edge holds the strength.
Why there are no numbers on this page. Dose, gap, temperature and strength targets are meaningless without your substrates, your joint and your line. We publish the mechanisms so you can recognise your own failure here — and we work out the numbers against your parts, not against a general-purpose table. Some of the checking above we run ourselves, some runs at your line and some at a partner lab; the report states which, so nothing in it arrives unattributed.
What you receive
Evidence you can act on
Final Root Cause Report
The failure mode, the evidence behind it and the mechanism — written so design, process and quality can all act from the same document instead of three different theories.
Corrective action set
The specific parameter, surface or fixture changes to make, each with the reasoning attached — so your team can adapt them to the line instead of following them blind.
Production control points
The short list of things to hold and monitor, so the same failure does not return when volume, season or operator changes.
Engagement
How an investigation runs
Describe the failure
Tell us what failed, when, and what changed just before it started. We come back with an initial failure-mode assessment.
Start hereSample & process review
We examine failed parts, retained good parts and the conditions the line actually ran at — then define the hypothesis.
Controlled re-run & DOE
We isolate the suspected variable in a designed run, so the cause is proven rather than argued.
Corrective actions & handover
You receive the root cause report, the corrective action set and the control points to hold in production.
FAQ
Frequently Asked Questions
Failed parts if you still have them, parts from the same lot that passed, and the conditions the line actually ran at — dispense settings, cure setup, fixture, and what changed shortly before the failure appeared. The good samples matter as much as the failed ones, because the comparison between them is where the evidence comes from.
We give a first failure-mode assessment from your description and photographs, before any parts move. It narrows the likely mechanisms and — just as importantly — tells you what to preserve, so nothing gets reworked or discarded before it can be examined.
Yes. The investigation is vendor-neutral, and most of what we examine is the interface, the cure and the process rather than the brand on the cartridge. If the material does turn out to be the wrong fit for the joint, we say so and explain the reasoning — including when the answer is that your current material is fine.
You get the corrective actions and the production control points to hold. If the fix needs proving on real parts before it goes back on the line, that runs as a pilot build; if it needs a different joint design or a different material, it moves upstream into process design or material substitution.
Let's Find the Root Cause Together
Tell us what failed, what changed, and what you have already tried. We help you untangle the interactions between material, equipment and process — then define the preventive measures and the mass-production control points.
Vendor-neutral · Mechanism-level · Documented