Quick answer. Most PU adhesive bond failures trace to three root causes: substrate moisture content outside 8-12% for wood or missing surface treatment on plastics/metals below 38 dyne/cm surface energy, no primer on non-porous or oxide-forming substrates, and insufficient open humidity (below 40% RH) starving the NCO-moisture cure reaction. Adhesive-failure mode (clean substrate, adhesive lifts intact) points to surface chemistry; cohesive failure (adhesive splits internally) points to under-cure or contamination. Fix the substrate and cure environment before reformulating the adhesive.
A one-component moisture-cure PU adhesive cures by reacting free NCO groups in the prepolymer with atmospheric or substrate-bound water, generating urea linkages and releasing CO2 as a byproduct. This reaction is diffusion-limited: cure starts at the exposed surface and works inward, roughly 2-4 mm per 24 hours at 23°C/50% RH for a typical 3500-4500 mPa·s prepolymer. Below 30% RH, full-depth cure in a 5 mm glueline can take over 7 days instead of 48-72 hours, and joints loaded before that point show cohesive splitting under peel testing rather than clean substrate release.
We formulate and ship PU prepolymers and primers directly out of our Zhejiang plant, so the failure patterns below come from claims data on wood lamination, footwear upper-to-sole bonding, and sandwich panel lines, not from a lab data sheet.
Adhesive Failure vs Cohesive Failure vs Substrate Failure: Reading the Break
ASTM D903 (180° peel) and ASTM D1002 (lap shear) both classify break mode as the first diagnostic step. Under a loupe, an adhesive-failure surface shows a mirror-clean substrate with no polymer residue — the bond never chemically or mechanically keyed into the surface at all.
Cohesive failure shows adhesive residue split roughly evenly on both sides, meaning the bond formed but the cured polymer itself was weaker than the joint load — usually under-cure, moisture starvation, or an NCO:OH ratio that drifted during prepolymer synthesis.
Substrate failure (the target outcome for structural joints) shows the wood, foam, or laminate itself tearing before the adhesive line lets go. On birch plywood lamination we require substrate failure at ≥70% of the bonded area per ASTM D905 shear block testing before releasing a lot for shipment.
| Break Mode | What You See | Most Likely Root Cause | Corrective Action |
|---|---|---|---|
| Adhesive failure | Clean substrate, no residue, adhesive lifts as an intact film | Surface energy <38 dyne/cm, oil/mold-release residue, no primer on PP/PE/anodized aluminum | Dyne-pen test surface; corona or flame treat plastics; apply silane-based primer, 15-30 min flash-off |
| Cohesive failure (soft, tacky core) | Adhesive splits, uncured or rubbery interior | RH <40% during cure, glueline >5mm without moisture-injection assist, NCO:OH ratio off-spec | Humidify workspace to 50-65% RH or use two-part accelerated system; cap glueline thickness per TDS |
| Cohesive failure (brittle, chalky) | Adhesive splits, hard and friable | Substrate moisture content >15% causing excess CO2 foaming and micro-voiding | Kiln-dry wood to 8-12% MC before bonding; moisture-meter check at press, not at intake |
| Substrate failure | Wood/foam tears, adhesive line intact | Bond performing as designed | None — this is the acceptance criterion |
Substrate Moisture: The Variable Everyone Skips at the Press
Wood moisture content is checked at intake and then ignored until the bond fails weeks later. Kiln-dried lumber re-absorbs ambient moisture in humid coastal warehouses (common at export ports) at roughly 0.5-1.5% MC per week without climate control.
Below 6% MC, there is not enough bound water at the interface to initiate cure at a useful rate, and the adhesive stays tacky at the glueline for days. Above 15% MC, the moisture-cure reaction runs too fast and generates excess CO2, creating micro-voids that show up as a chalky, brittle cohesive failure under peel load.
We recommend a pin-type moisture meter reading taken at the press, not at the receiving dock, for every production shift — not a spot check once a week.
Primer Selection: Match Chemistry to Substrate, Not to Habit
Primer choice is where most cross-substrate lines lose the most bonded area, because one primer gets used across plastics, metals, and composites out of convenience.
Non-Porous Plastics (PP, PE, PTFE)
These substrates sit at 29-31 dyne/cm untreated surface energy, well below the ~38 dyne/cm threshold PU adhesives need to wet out. Corona or flame treatment raises surface energy for roughly 48-72 hours before it decays back toward baseline, so bonding must happen inside that window — not the next shift.
Metals (Aluminum, Galvanized Steel)
Native oxide layers on aluminum are inconsistent and can be several microns thick depending on alloy and storage humidity. A silane-based primer with a 3-aminopropyltriethoxysilane coupling agent bridges the oxide layer to the PU backbone; skip it and you get adhesive failure at the metal interface even when peel values look acceptable on day one, because the oxide layer itself lets go under long-term humidity cycling.
EVA and Rubber Soling (Footwear)
EVA foam soling needs a halogenation or UV-ozone surface treatment before priming — PU primer alone does not etch the EVA surface enough for mechanical interlock, which is why footwear delamination complaints cluster on EVA midsoles more than on rubber outsoles.
| Substrate | Typical Surface Energy (Untreated) | Required Treatment | Primer Flash-Off Time |
|---|---|---|---|
| Polypropylene (PP) | 29-31 dyne/cm | Corona or flame treatment, then primer | 15-20 min at 23°C |
| Anodized Aluminum | Variable, oxide-dependent | Silane coupling agent primer | 10-15 min at 23°C |
| EVA Foam | 32-34 dyne/cm | Halogenation/UV-ozone, then primer | 20-30 min at 23°C |
| Solid Wood / Plywood | Porous, N/A | None required at 8-12% MC | N/A |
Our PU primer line is formulated substrate-by-substrate rather than as one universal product, because a single silane loading that works on aluminum under-treats PP.
Open Time, Press Time, and the Peel Test That Catches Both
Open time — the window between adhesive application and substrate mating — runs 3-8 minutes for most PU hot-melt and moisture-cure grades at 23°C, and compresses fast above 30°C. Mating substrates after the open time window closes gives you a partially skinned film that looks wet but has already lost its ability to flow into surface texture, which reads as adhesive failure on peel testing even though the correct adhesive and primer were used.
ASTM D903 peel testing at 7 days post-application (full cure) is the only reliable acceptance gate — testing at 24 hours on a moisture-cure system will show artificially low peel strength that has nothing to do with the final bond.
Isocyanate Handling and Regulatory Exposure Limits
PU prepolymers contain residual free monomeric diisocyanate (typically <0.1% for MDI-based systems, disclosed on the SDS). OSHA sets permissible exposure limits for MDI and TDI vapor and requires respiratory protection and ventilation controls where spray or heated application generates aerosol, per OSHA's isocyanates guidance.
Under EU REACH, ECHA's restriction on diisocyanates (entry 74, Annex XVII) requires documented training for industrial and professional users above the 0.1% w/w concentration threshold, with the transitional deadline that took effect in phases through 2023-2025 — see ECHA's diisocyanates restriction page for the current compliance dates applicable to your import market.
Sourcing Direct: Why Manufacturer Support Catches Failures Faster Than Distributor Purchasing
Buying PU adhesive through a distributor means the technical file — NCO%, viscosity curve, recommended primer pairing — sits with the formulator, not the reseller, and troubleshooting a bond failure means routing through a middleman with no visibility into the batch chemistry.
We ship prepolymer, primer, and application data sheets as a matched set out of our own plant, with MOQ starting at 1 ton for standard grades and custom NCO:OH ratio adjustments available at 3-5 ton MOQ with a 15-20 day lead time. When a customer's peel data comes back low, our lab can pull the exact production batch record rather than asking a distributor to relay a claim.
See our PU adhesive product range for grade-by-grade viscosity and open-time specifications by application (lamination, footwear, panel).
FAQ
Q: My PU adhesive stays tacky for days and never fully cures — what's wrong?
Check ambient RH first. Moisture-cure PU needs 40-65% RH to cure at a normal rate; below 30% RH, a 5mm glueline can take over a week instead of 48-72 hours. Low substrate moisture content (below 6% MC on wood) causes the same symptom.
Q: Do I need a primer on plastic if the adhesive datasheet doesn't mention one?
Test surface energy with a dyne pen before assuming you don't. Untreated PP and PE sit around 29-31 dyne/cm, well under the ~38 dyne/cm PU adhesives need — a generic datasheet often assumes corona-treated stock that your incoming material may not be.
Q: Peel test passed at 24 hours but the joint failed in the field weeks later — why?
24-hour peel data on a moisture-cure system measures partial cure, not final bond strength. Test at 7 days per ASTM D903, and re-check humidity cycling resistance separately if the application sees repeated wet/dry exposure.
Q: Can we switch primers between our aluminum and EVA foam lines to simplify inventory?
Not without losing bond strength on one side. Aluminum needs a silane coupling agent primer to bridge the oxide layer; EVA needs surface halogenation or UV-ozone treatment before the same primer chemistry will key in. One universal primer under-treats at least one substrate.
Q: How do we tell if a returned joint failed from our process or from an adhesive batch defect?
Read the break mode first — adhesive failure with a clean substrate usually points to surface prep or open-time timing on your line; cohesive failure with soft, uncured adhesive points to cure conditions or a batch NCO:OH ratio issue, which we can trace against the production lot record when you supply the batch number.