An engineer staring at a fractured test coupon usually has one question first: does this residue pattern mean the adhesive failed, or does it mean something upstream in the process failed? The answers below work through the questions that come up most often once a failure-mode classification is in hand.
Q: What’s actually different between a cohesive and an adhesive fracture?
A cohesive fracture splits inside the adhesive layer itself, so both substrates come away coated in residue. An adhesive fracture releases at the interface, leaving one or both surfaces largely clean. A third, less commonly discussed outcome — substrate failure — happens when the base material cracks or tears before either the bond or the interface gives way, which in most structural work is the outcome engineers are actually trying to design toward. For a fuller breakdown of how each of these three failure modes is defined and specified, that overview is a useful companion to the FAQ below.
Q: Is a cohesive failure always considered a “pass” on the interface side?
Generally yes, but with a caveat worth stating plainly: a cohesive failure only confirms the interface performed as intended at the load where the part actually broke. If that load is well under the adhesive’s rated bulk strength, the cohesive label is masking a separate problem — usually incomplete cure or an out-of-spec bond-line thickness — rather than proving the joint met its design target.
Q: How do engineers actually classify failure mode in a lab setting rather than guessing from a photo?
Lap shear testing to a recognized standard such as ASTM D1002 is the baseline quantitative method — substrates are bonded, pulled to failure under controlled tension, and the resulting fracture surface is examined under magnification (scanning electron microscopy for very fine bond lines) and reported as a percentage split between cohesive and adhesive area. Peel and T-peel geometries are used instead when the substrate is flexible and stress won’t distribute evenly across a lap shear coupon. The percentage split, not just a binary label, is the number that actually drives a corrective action decision.
Q: If failures trend adhesive, what should change first?
Look at the interface before touching the adhesive formulation. Low surface energy on inert plastics, residual mold-release or machining oil, and skipped or inconsistent plasma or corona treatment are the most common root causes of adhesive-mode failure, and all three are process fixes rather than material swaps. A primer step can also close the gap on substrates that are difficult to treat mechanically. Reaching for a different adhesive chemistry before ruling out these process variables is usually a wasted qualification cycle.
Q: If failures trend cohesive but at a lower-than-expected load, what should change first?
Check cure completeness before concluding the material has reached its real ceiling. For UV and visible-light-curable systems, this means confirming irradiance and total dose reached every point along the bond line — not just the exposed surface — since shadowing from fixtures or nearby components can leave a bond line under-cured even when the top layer looks fully hardened. Bond-line thickness outside the validated process window is the second most common cause: too thick concentrates internal stress, too thin starves the joint of enough bulk material to absorb load.
Q: Does the environment a part sees in service change which failure mode shows up?
Yes, and this is one of the more counterintuitive parts of failure analysis: a joint that passes lap shear testing with a clean cohesive failure at time zero can shift toward adhesive failure months later in the field. Repeated thermal cycling introduces cyclic stress concentrated right at the interface when substrates have mismatched coefficients of thermal expansion — a mechanism examined in more depth in how CTE mismatch drives adhesive bond failure — and moisture or chemical exposure can degrade adhesion at the surface even while the bulk adhesive itself stays intact. This is why a qualification protocol built only on fresh, as-cured coupons misses a real failure mode that shows up only after aging.
Q: What does a complete qualification protocol actually track?
At minimum, three data points per candidate adhesive: failure mode and load as-cured, failure mode and load after thermal cycling representative of the service environment, and failure mode and load after chemical or moisture exposure relevant to the application. A formulation that holds cohesive failure at consistent load across all three has real margin. One that starts cohesive and drifts adhesive after aging needs either a different surface treatment or a different adhesive chemistry before it goes into production — catching that in qualification is considerably cheaper than catching it in the field.
Q: Who should be involved in reviewing ambiguous or mixed-mode results?
Mixed-mode fractures — a coupon showing both cohesive and adhesive regions on the same surface — are common enough that they shouldn’t be treated as inconclusive. Reporting the percentage split and correlating it against process variables (cure dose logs, surface treatment records, batch documentation) usually resolves the ambiguity without additional testing. For substrate combinations where dissimilar coefficients of thermal expansion are already a known risk, cross-referencing failure data against heavy-duty repair bond strength benchmarks can help set a realistic target load before qualification even starts. Engineers building this kind of protocol are welcome to Email Us for guidance on test sequencing specific to their substrate stack.
Reading a fracture surface correctly turns a pass/fail data point into a specific, actionable diagnosis rather than a guess — and Incure’s applications engineers review failure-mode data of exactly this kind on a regular basis. Contact Our Team if you’d like a second read on your own qualification results.
Visit www.incurelab.com for more information.