Quick answer. Bubbling and foaming in a polyurethane adhesive bond line is carbon dioxide, generated when free isocyanate (NCO) groups react with water instead of with polyol or substrate hydroxyls. The water comes from substrate moisture content, ambient humidity condensing on cold panels, or wet fillers and release agents. Control it by holding substrate MC below 8-10% for wood, keeping panel surface temperature at least 3 °C above dew point, and reducing open time so the adhesive is under pressure before the moisture reaction has time to gas off.
Foaming is not a formulation defect in most factory complaints we investigate. It is a process-window defect that shows up as blisters, telegraphed dimples on thin skins, and a bond line that measures 0.9 mm when the spec said 0.25 mm.
This guide covers the chemistry, the measurable limits, and the sequence to diagnose it on a running line.
The CO2 mechanism: what actually gasses
An isocyanate reacts with water in two steps. First, R-NCO + H2O forms an unstable carbamic acid. That acid immediately decomposes into an amine plus CO2 gas.
The amine then reacts with a second isocyanate group to form a substituted urea. Net stoichiometry: one mole of water consumes two moles of NCO and releases one mole of CO2.
The number that matters on the shop floor: 1 gram of water generates roughly 1.24 litres of CO2 at standard conditions. Spread 150 g/m² of a 1K moisture-cure PU across a panel and you only need a few tenths of a gram of excess water per square metre before the gas volume exceeds what the bond line can dissolve or vent.
Two competing reactions run at once. The gelling reaction (NCO + OH, forming urethane) builds strength. The blowing reaction (NCO + H2O, forming urea + CO2) builds gas. Anything that slows gelling relative to blowing — cold substrate, low catalyst, long open time — shifts the balance toward bubbles.
The urea linkages themselves are not the problem; they are stiffer and higher-melting than urethane. The problem is that the gas has nowhere to go once the adhesive skins over.
Moisture sources ranked by how often they are the cause
In field diagnosis, the order below holds for laminating, sandwich panel and footwear lines.
| Source | Typical magnitude | How to confirm | Corrective limit |
|---|---|---|---|
| Substrate moisture content (wood, MDF, plywood, paper honeycomb) | MC 12-18% after monsoon storage vs 6-8% conditioned | Pin-type or capacitance meter per ASTM D4442; oven-dry check on 3 coupons | Wood/MDF 6-10% MC; paper honeycomb ≤8% |
| Condensation on cold metal or stone | Invisible film, 0.1-0.5 g/m² | Surface pyrometer vs psychrometer dew point | Surface temp ≥ dew point + 3 °C |
| Ambient RH during open time | >75% RH at 30 °C accelerates skinning | Datalogger at the coater, not at the office wall | 45-65% RH, 18-28 °C |
| Wet or hygroscopic filler (CaCO3, talc, wood flour) | 0.2-1.5% water by weight as delivered | Karl Fischer titration or 105 °C loss-on-drying | <0.05% water in filler |
| Residual water-based primer, mould release or cleaning solvent | Varies | Flash-off time audit; wipe test | Full flash-off, verified by weight-back |
| Drum headspace humidity in opened 1K PU | Skinning at the drum wall within 24-72 h | Visual crust + viscosity rise | Dry nitrogen blanket, reseal within 15 min |
Reading the defect: bubble morphology tells you the source
Cut a cross-section with a fresh blade and look at the bond line under 10x.
Fine, uniform, spherical voids distributed across the whole glue line usually mean bulk moisture — wet substrate or wet filler. The gas nucleated everywhere at once.
Large, flattened blisters concentrated at panel edges or near the press exit point to condensation and to gas escaping laterally before pressure was applied.
Elongated voids that follow the coating roll direction indicate air entrainment from the applicator, not CO2 at all. Check roll nip, adhesive viscosity at application temperature, and pump cavitation before you blame humidity.
A skinned surface with a soft, gassy core means open time was exceeded: the top surface reacted with airborne moisture, formed a barrier, and trapped the CO2 generated underneath.
Process window: the numbers to hold
These are typical operating ranges for reactive PU laminating and assembly adhesives. Confirm against the technical data sheet for the specific grade you run.
| Parameter | Bubbling risk if… | Working range | Why |
|---|---|---|---|
| Substrate MC (wood-based) | >12% | 6-10% | Free water feeds the blowing reaction |
| Surface temp vs dew point | Below dew point | ≥ +3 °C | Prevents invisible condensate film |
| Ambient RH | >75% | 45-65% | Slows premature skinning |
| Open time before press | Exceeds TDS value | Within 60-70% of stated open time | Gas escapes before skin forms |
| Press pressure | <0.3 MPa on porous stacks | 0.4-0.8 MPa typical for panels | Keeps CO2 dissolved; collapses voids |
| Coat weight | Over-application | 120-180 g/m² for porous, 60-100 g/m² for tight substrates | Thick lines cannot vent gas |
| Press temperature | Cold press <15 °C | 20-70 °C depending on grade | Speeds gelling relative to blowing |
Dew point is the single check most lines skip. A sling psychrometer and a $40 infrared surface thermometer settle the question in 30 seconds, and the pairing is the same principle used in protective coatings practice.
Diagnostic sequence for a running line
Work in this order. Each step is cheap and rules out a whole class of causes.
- Take three substrate coupons from the actual pallet in use, not from the conditioned sample rack, and measure MC per ASTM D4442. Log the value.
- Measure ambient dry-bulb, RH and panel surface temperature at the coater. Calculate dew point. Record the delta.
- Verify adhesive age: check batch date, drum opening date, and whether the drum was blanketed. Skinned 1K PU is already partially ureated.
- Run one panel at 60% of normal open time and one at normal. If the short-open panel is clean, the fault is skinning, not bulk water.
- Cut and inspect both bond lines. Match morphology to the section above.
- If the substrate is a composite or contains recycled fibre, run a 105 °C loss-on-drying on the filler or core.
- Only after those five steps should you request a formulation change.
Most escalations we receive from customers resolve at step 1 or step 2. Formulation changes are the last lever because they cost weeks of requalification.
Formulation levers when the process is already tight
If MC and dew point are in range and bubbles persist, the fixes sit with the adhesive itself.
Lower free-NCO content reduces the pool of isocyanate available for the water side reaction, and also lowers monomeric MDI exposure — relevant because monomeric diisocyanates carry restriction obligations under the EU REACH framework, documented by ECHA, with training requirements for industrial users.
Catalyst rebalancing shifts the gel/blow ratio. Tin-based gelling catalysts accelerate the NCO-OH path; excess tertiary amine pushes the water path. On lines running below 20 °C, a gel-biased catalyst package often eliminates blistering without touching the process.
Moving from 1K moisture-cure to a 2K polyol-cured system removes the dependency on ambient water entirely, at the cost of metering equipment and a defined pot life. For laminating stone, metal skins and low-porosity composites, that trade is usually worth it.
Molecular sieve or reactive water scavenger addition binds trace water in the filler package. It buys margin, not immunity.
Hot-melt PU (reactive HMPUR) sidesteps most of the problem: it develops green strength on cooling before the moisture cure completes, so the bond is mechanically closed before significant CO2 forms.
Substrate-specific notes
Wood and MDF. The plant floor gradient matters. A pallet stored against an exterior wall in a coastal plant can sit 4-5 percentage points higher in MC than the same batch in the middle of the shop. Condition 48 hours minimum before coating.
Aluminium and steel skins in sandwich panels. Metal has thermal mass. A coil unwound from a 12 °C warehouse into a 28 °C, 70% RH shop will sit below dew point for 20-40 minutes and carry an invisible water film the whole time. This is the most common cause of edge blistering in structural insulated panel production.
Paper honeycomb and kraft cores. Hygroscopic and rarely measured. Treat as wood: measure, condition, then coat.
Foam cores (EPS, PU, XPS). Low water content, but closed-cell foam cannot absorb any gas. Every millilitre of CO2 stays in the bond line, so coat weight discipline is stricter here.
Footwear uppers and leather. Leather can carry 12-16% moisture and often arrives with residual finishing agents. Solvent flash-off audits matter more than RH control.
Handling and ventilation while you troubleshoot
Diagnosing bubbling means opening drums, cutting hot bond lines and running short-open trials — all of which raise isocyanate vapour and aerosol exposure at the operator station. Applicable occupational limits and control guidance for diisocyanates are published by OSHA, and respiratory sensitisation data is summarised by NIOSH. Local exhaust at the coater, not general dilution ventilation, is the control that works.
FAQ
Q: Can I just increase press pressure to crush the bubbles?
Higher pressure suppresses void growth and forces CO2 into solution, and moving from 0.3 to 0.6 MPa often clears marginal cases. It does not fix a substrate at 15% MC — the gas reappears as delayed blistering after the panel leaves the press and the bond line relaxes.
Q: Why does bubbling appear only in summer or only on night shift?
Both are dew-point signatures. Summer brings high absolute humidity; night shift often runs with the HVAC set back, so panel surfaces sit closer to dew point. Log RH and surface temperature by shift for one week and the pattern usually resolves itself without any adhesive change.
Q: Is a small amount of foaming acceptable?
For gap-filling assembly work on porous substrates, controlled expansion is designed in and harmless. For structural laminating and thin decorative skins it is not: voids reduce effective bonded area, and a bond line with 15% void fraction loses roughly proportional shear strength while also telegraphing surface dimples on skins under 0.6 mm.
Q: Does the adhesive shelf life affect bubbling?
Yes. An opened drum of 1K moisture-cure PU begins reacting with headspace humidity immediately. Viscosity climbs, free NCO drops, and the surface skins. Reseal within 15 minutes, blanket with dry nitrogen, and treat an opened drum as having a working life measured in days rather than the unopened shelf life on the label.
Q: We changed nothing and bubbling started. Where do I look first?
A new substrate lot or a new filler lot. Ask for the supplier's moisture certificate on the incoming batch, then run your own loss-on-drying. A filler that shifted from 0.03% to 0.8% water will produce visible foam at normal loading levels.
Q: Will switching to hot-melt PU eliminate the problem?
It reduces it substantially because green strength develops on cooling, closing the bond mechanically before the moisture cure runs to completion. You still need substrate MC control, since trapped water will migrate to the bond line during the 24-72 hour post-cure.
Working with a direct manufacturer on this
Bubbling problems get solved faster when the person adjusting the catalyst package is the person who made the batch. As a China-based PU adhesive manufacturer supplying direct rather than through distribution, SPC Adhesives adjusts free-NCO level, catalyst ratio and open time against a customer's actual line conditions — substrate MC range, shop RH, press cycle and coat weight — and ships trial quantities before full-scale qualification.
Custom formulation work starts from the process data collected in the diagnostic sequence above, not from a catalogue selection. Send the MC readings, dew-point delta and a cross-section photo and the reformulation scope is usually clear on the first review.
Browse the reactive polyurethane range at puglue.com PU adhesives, or see the structural panel application notes under SIP and sandwich panel bonding.
