A bond that passed every incoming inspection can still fail on the line six weeks later — and the questions that come up during that investigation tend to repeat across industries far more than the failures themselves do.

Q: A bond failed in the field but passed our original QC check. Where do I actually start?

A: Start with the fracture surface before touching any process records. A visual and low-magnification inspection tells you which of the three recognized failure modes occurred — clean separation at the substrate (interfacial), a split within the adhesive layer itself leaving residue on both sides (cohesive), or breakage of the substrate material with the bond line intact (substrate failure), each covered in more depth in our breakdown of adhesive failure modes — and each mode points investigation toward a different set of process variables. Jumping straight to “the adhesive was bad” before classifying the fracture surface is the most common way an investigation goes in the wrong direction from the start.

Q: How do I tell interfacial failure from cohesive failure without a lab?

A: Look at the residue pattern on both halves of the joint. Interfacial failure leaves one side essentially bare — little to no adhesive residue — indicating the bond to that substrate never really formed, usually from surface contamination or insufficient wetting. Cohesive failure leaves adhesive material on both sides, meaning the bond to the substrate held but the adhesive’s internal strength gave out first, which usually points to under-cure or a mechanical overload beyond what the formulation was rated for. A simple tape-pull or fingernail scrape test on the residue, checking whether it’s present on one surface or both, gets you most of the way to a correct classification without instrumentation.

Q: Is cohesive failure always the “good” outcome everyone says it is?

A: Not automatically. Cohesive failure is preferred over interfacial failure because it proves the bond to the substrate was stronger than the adhesive’s own bulk strength — a real signal that surface preparation worked. But a joint that cohesively fails well below its rated load still indicates a process problem, usually incomplete cure. A cohesive failure at 60% of the expected load is not a success story just because it happened in the “right” mode; it’s evidence the adhesive never reached its designed cross-link density in that specific batch.

Q: We’re seeing the same joint fail at wildly different times in the field — some at three weeks, some at three years. Why?

A: This pattern almost always points to a process variable that drifts rather than a formulation defect, since a true material defect tends to fail consistently. The most common culprits are a UV dose that varies with part position on a fixture (parts at the edge of the cure zone receiving less irradiance than parts in the center), a curing oven with an uneven temperature profile across its width, or surface preparation that depends on an operator’s technique rather than a controlled, verified step. Comparing failed parts against their original process logs — cure zone position, batch date, operator shift — usually surfaces the drifting variable faster than re-testing the adhesive chemistry itself.

Q: What’s the fastest way to rule out an under-cured adhesive before escalating further?

A: A Shore hardness reading on a witness sample cured alongside the same production batch is the quickest non-destructive check — an under-cured sample reads measurably softer than a fully cross-linked one even when both appear visually identical and tack-free to the touch. For UV-cure systems specifically, confirming delivered irradiance and total dose with a radiometer at the actual part location, not just at the lamp head, catches the shadowing and fixture-geometry issues that a lamp-output check alone would miss — a failure pathway closely related to the light guide degradation mechanisms that quietly reduce delivered dose in aging UV curing equipment.

Q: When does an investigation need to escalate beyond a pull test?

A: When the fracture-surface classification and process-log review don’t converge on an obvious cause, differential scanning calorimetry (DSC) can confirm whether an adhesive reached its expected degree of cure by measuring residual exotherm, and FTIR spectroscopy can identify unexpected surface contamination at the molecular level that a visual inspection would miss entirely. These tools are worth the turnaround time on a recurring or high-consequence failure; they’re usually overkill for a one-off return that a hardness check and a process-log review already explain.

Q: Does adding more adhesive volume make a marginal bond design safer?

A: Not reliably, and sometimes the opposite. Excess adhesive squeezed out of a joint under clamping pressure can trap air at the bond line’s center, and a bead sized larger than the joint geometry calls for often cures unevenly through its thickness compared to a bead matched to the actual gap. The more reliable lever is confirming the joint design keeps the load in shear or compression rather than peel — the trade-off explored further in UV glue vs epoxy for heavy-duty repairs — rather than compensating for a peel-prone design with more material.

Email Us with a description of your fracture pattern and process history, and Incure’s applications team can help narrow the investigation before it goes to destructive testing.

A structured investigation — classify the fracture mode first, compare against process records second, escalate to instrumented testing only when the first two steps don’t converge — resolves most recurring adhesive failures faster than re-testing the adhesive formulation itself. Contact Our Team if your production line needs help building that investigation process.

Visit www.incurelab.com for more information.