How Adhesive Bonding Chemistry and Curing Decide Whether a Waterproof Bag Holds

The break tells you what to fix: three failure loci

Every bonded joint has three places it can come apart, and the adhesive bonding chemistry discussion is only relevant to one of them. Adhesive failure means the bond let go at the interface: one face comes away clean and the adhesive stays on the other side. Cohesive failure means the adhesive layer itself split: both faces carry adhesive residue. Substrate failure means the fabric, film or foam tore before the bond did. On a production bag the third outcome is what you actually want, because it means the joint is stronger than the materials it joins.

The reason this matters commercially is that each locus points at a different fix and the fixes differ by orders of magnitude in cost. An interfacial failure is fixed by cleaning, abrading, priming, corona or flame treatment, or by raising the surface energy of the film before bonding, all of which cost cents or nothing. A cohesive failure is fixed by changing the cure, the mix ratio or the formulation, which costs process time. A substrate failure is fixed by changing the material, which costs real money. Guess wrong and you pay for the expensive fix while the cheap one was sitting unused.

The diagnostic is simple enough to run at a returns desk: look at both faces of the failure. If one is clean and bare, the adhesive never bonded to it, and no strength upgrade will help because the interface was never the load path. If both faces are uniformly coated with adhesive, the adhesive was fine and the cure was not, or the layer was too thick, or it was loaded before it had developed strength.

• Clean substrate face, adhesive all on the other side: interfacial failure. Fix the surface, not the glue.
• Adhesive residue on both faces: cohesive failure. Fix the cure, the layer thickness or the grade.
• Fabric or film torn, adhesive still attached to both: substrate failure. The joint won; upgrade the material.
• Mixed patterns across one seam: uneven pressure or uneven surface treatment, which is a process control problem.

Surface energy in dynes: the number nobody puts on a tech pack

Adhesion is fundamentally a wetting problem. An adhesive can only bond to a surface it can spread across, and it will only spread across a surface whose surface energy is higher than the surface tension of the liquid adhesive. Surface energy is reported in dynes per centimetre, or equivalently millinewtons per metre, and the practical thresholds are well established: untreated polyethylene situs around 30 dynes and is famously unbondable without treatment, treated polypropylene runs 38 to 44, polyester film around 42 to 46, and properly treated TPU in the mid forties to low fifties.

The field test uses dyne pens or dyne solutions: a liquid of known surface tension is brushed onto the surface, and if the film stays continuous for two seconds it wets, which means the surface energy is at or above that value. The rule of thumb that follows is widely used and worth memorising: the surface energy should exceed the surface tension of the adhesive by roughly 10 dynes for a reliable bond. Below that margin you get a joint that passes on the day and creeps apart under load and humidity.

Two entries on that table deserve emphasis because they cause the majority of real failures. The second row is the invisible one: a fabric delivered with a durable water repellent finish, a slip agent that has bloomed to the surface of a film during storage, or mould release residue carried over from a component, all push the effective surface energy down into a range where the adhesive appears to wet during application and then lets go under humidity. The third row carries a timing trap that catches buyers out: corona and flame treatments decay. Atreated roll bonded three weeks later may have lost ten dynes, and the supplier will report the treatment was performed correctly, which is true and irrelevant.

Surface preparation: four routes and what each one removes

Preparation is where bond reliability is won, and the four routes are not interchangeable because they remove different things. Solvent wiping removes oils, release agents and some processing aids, and it is the minimum that should ever happen. Mechanical abrasion removes weak surface layers and oxide, and critically it increases the effective bonding area. Corona and flame treatment raise surface energy by oxidising the surface, which is the only practical route for polyolefins. Chemical primer applies a coupling layer that bonds to both sides, and it is the answer when two incompatible materials must be joined.

• Solvent wipe: removes oils and release agents. Fast and cheap, but re-contamination is immediate if the operator touches the surface afterwards.
• Abrasion: removes weak boundary layers and adds mechanical key. Watch the dust; abraded dust left in place becomes a release layer of its own.
• Corona or flame: raises surface energy by oxidation. Effective but time-limited, and the decay accelerates with temperature.
• Primer or coupling agent: the only answer for genuinely low-energy substrates. Adds a process step and a cure window of its own.

The failure pattern that shows up again and again is a factory that abrades and wipes but does not treat, paired with a film that has been stored long enough for slip agent to bloom. Abrasion removes the bloom locally and the bond is fine for a week, then the remaining agent continues migrating into the interface and the joint creeps. The correct sequence for a difficult film is wipe, abrade, wipe again, treat, and bond inside the treatment window, with the whole sequence written into the process sheet rather than left to operator habit.

There is also a human factor that no specification fixes on its own. A surface prepared and then handled with bare hands picks up skin oils within seconds, and skin oil is enough to drop the local surface energy below the wetting threshold. Gloves, and a rule that prepared surfaces are bonded within a stated time, cost nothing and remove a whole class of sporadic failure that otherwise looks random.

Two-component polyurethane: mix ratio, pot life and the Monday problem

Most structural bonding on waterproof bags uses a two-component polyurethane: an isocyanate-bearing component and a polyol-bearing component that react after mixing to build molecular weight and crosslink density. The ratio is not a suggestion. Deviation in either direction alters the stoichiometry of the reaction, and the consequences are specific rather than vague: too much isocyanate leaves unreacted monomer that later reacts with atmospheric moisture and generates carbon dioxide, producing bubbles and a foamed, weak bond line; too little leaves the system under-crosslinked, soft, and permanently sensitive to heat and hydrolysis. A ratio error of five per cent is enough to move a joint from substrate failure to cohesive failure.

Pot life is the second control variable and the one that produces the most inconsistent batches. Once the two components are mixed, viscosity rises as the reaction proceeds, and beyond the stated pot life the material no longer wets the surface properly. Wetting failure looks exactly like surface contamination failure, so a batch bonded with expired material gets misdiagnosed as a preparation problem, the factory increases its cleaning, and the problem persists. Materialmixed on a Friday afternoon and topped up on Monday morning is a documented and entirely avoidable source of this.

The three controls that actually work are unglamorous: weigh both components rather than measuring by volume, record the mix time on the batch, and discard rather than top up. None of them require laboratory equipment. All of them are more predictive of field bond performance than any adhesion test run on a fresh coupon.

Moisture cure, heat cure and two-part: three mechanisms, three windows

Curing mechanism determines what the production line has to control. A moisture-curing polyurethane reacts with water vapour from the air, which means it needs humidity, it cures from the outside in, and it is slow: full cure through a thick section can take days. A heat-cured system reacts when raised to temperature, cures uniformly, and reaches full strength in minutes, but it needs a fixture and an oven and it applies thermal load to the fabric. A two-part system cures on mixing, needs neither humidity nor heat, and is the most controllable of the three, at the cost of pot life discipline.

• Moisture cure: no equipment, excellent gap filling, tolerates flexible substrates. Slow, humidity dependent, and it can foam if applied too thick.
• Heat cure: fast, uniform, high final strength, easily automated. Requires fixturing, applies heat to heat-sensitive films, and needs a recorded temperature profile.
• Two-part: predictable and equipment-light, cures at room temperature. Strict on ratio and pot life, and unforgiving of poor mixing.

The choice is usually made for the wrong reason, which is line speed, when it should be made on substrate sensitivity. A heat-cured adhesive on a thin TPU film is asking for distortion: the film softens well below the temperature the adhesive needs, and the result is a strong bond on a distorted panel. A moisture-curing adhesive used in a dry winter factory without humidity control simply does not cure on schedule, and the operator compensates by stacking parts, which then block. The compatibility question is addressed in more depth in our review of cold and heat resistance in waterproof materials.

Where a bag will be made by a combination of welding and bonding, the two processes have to be sequenced rather than interleaved. Welding first and bonding second risks exposing a cured adhesive line to weld heat; bonding first and welding second risks the solvent or moisture in the adhesive interfering with the weld. The construction routes and their sequencing are set out in our comparison of stitched, welded and bonded construction.

Temperature, pressure and time: three interlocking limits

A curing schedule is not three independent settings that can each be optimised. Raise the temperature and you shorten the time, but you also shorten the open time and risk substrate distortion. Raise the pressure and you thin the bond line, which usually raises shear strength and lowers peel strength, and excessive pressure starves the joint of adhesive entirely. Extend the time and you improve conversion, but past full cure you gain nothing and pay in throughput. The window is the region where all three sit together, and the width of that window is what a buyer should be asking about, not the nominal settings.

Bond-line thickness deserves its own paragraph because it is the most commonly uncontrolled variable in bag bonding and it moves both strength directions at once. A very thin bond line is stiff and brittle: it concentrates stress at the edges of the joint and performs badly in peel. A thick bond line has more compliant material to absorb strain and performs better in peel, but it carries load less efficiently in shear and takes longer to cure through. For most structural bag joints the useful target is modest and controlled, in the region of a few tenths of a millimetre, achieved with a defined pressure rather than by eye.

• Ask for the curing window as a range, not a set point. A supplier who gives one temperature and one time has not characterised the process.
• Set bond-line thickness with a shim, a stop or a controlled pressure, and verify it on a cut section rather than assuming it.
• Require that parts not be loaded, folded or packed until the stated handling strength has been reached, and name that time on the process sheet.
• Record the actual temperature at the bond line, not the oven set point. On thick or multi-layer stacks the two differ substantially.

The third bullet is where a surprising amount of field failure originates. A joint that has reached handling strength can be moved; a joint that has not will creep permanently if it is loaded, and it will creep in a way that looks like a material defect rather than a process one. The distinction between handling strength and full strength is routinely collapsed on production lines, and the seam that looked fine at packing opens at the customer.

Peel, shear and tensile rank adhesives in different orders

This is the point most specifications get wrong, and it is worth stating as a rule: the adhesive that wins a lap shear test is frequently not the adhesive that wins a peel test, and a bag joint rarely sees the loading that either test applies in isolation. Shear loads the whole bonded area roughly uniformly, so it rewards a stiff, strong, thin bond line. Peel concentrates the entire load at a moving line at the front of the crack, so it rewards a tough, compliant, thicker bond line. Tensile pulls normal to the plane, is highly sensitive to misalignment, and is the least informative of the three for a flexible product.

The practical consequence is that a buyer should name the loading mode the joint will see and test that mode. A seam tape is peeled, so it should be specified by peel strength. A stiffener panel is sheared, so it should be specified by shear. A strap attachment sees fatigue, and no single static number predicts it at all, which is why strap attachments are validated by cyclic load testing of the kind described in our article on load stress testing for straps and handles.

A useful diagnostic rule follows from the table: if a joint performs dramatically worse in peel than in shear, the adhesive is too brittle or the bond line is too thin, and the fix is a tougher grade or a controlled increase in thickness. If it performs worse in shear than in peel, the bond line is probably too thick or insufficiently cured, and the fix is pressure and cure rather than a new adhesive.

When to bond and when to weld: a decision with real cost consequences

Bonding and welding are not competing versions of the same operation. Welding fuses two compatible thermoplastics into one continuous mass, so there is no interface left to fail, no consumable, no cure time and no mixing discipline. Bonding joins dissimilar materials, works on thermosets and coated fabrics that cannot be welded, and distributes stress over a wide area rather than concentrating it at a seam line. Each is the right answer in a defined set of cases, and choosing bonding where welding was available is a permanent cost and reliability penalty.

The rule that usually holds: if both substrates are weldable thermoplastics of the same family, weld. If one substrate is a thermoset, a metal, a foam, a coated woven that will not fuse, or a different polymer family, bond. If the joint will be loaded in peel at an exposed edge, consider a welded primary seal with a bonded secondary reinforcement, because that combination is how the most reliable waterproof constructions are actually built.

• Welding removes the interface entirely, which removes the surface energy problem, the mix ratio problem and the pot life problem in one step.
• Bonding is the only option for dissimilar materials, for hardware attachment and for repairing or reinforcing a finished panel.
• Bonding adds consumable cost, labour time, cure time and a process window that must be controlled and recorded.
• A bonded joint on a weldable material is usually a sign that the material selection, not the process, should be revisited.

The welding side of this comparison, including the parameter windows for radio frequency and hot air routes, is covered in our article on RF and hot air welding in production, and the alternative thermal routes are compared in our piece on heat sealing versus ultrasonic welding.

Open time, tack time and green strength on a real line

Three timings govern whether a bonded joint survives its first hour, and they are almost never written down. Open time is how long the coated surface stays bondable after application. Tack time, or assembly time, is the window within which the two faces must be brought together for full strength. Green strength is how much load the joint can carry before it has cured. A line that applies adhesive across twenty panels and then assembles them all at the end has quietly exceeded the open time on the first panel, and the resulting joints range from good to non-existent across a single batch.

This is the mechanism behind the failure pattern that looks like randomness: a batch where most units are fine and a scattering fail early, with no correlation to operator, material lot or machine. Randomness of that shape is nearly always a timing distribution rather than a material property, and it is fixed by batching the work to the open time rather than by changing anything about the adhesive.

The fix is procedural and cheap. Define the maximum number of panels coated before assembly begins, define the pressure and dwell at assembly, and define the minimum time before the assembly can be moved. Three numbers on a process sheet remove a failure mode that no amount of incoming material testing will catch, because the material is perfectly good.

The adhesive families used on waterproof bags

Five chemistry families cover essentially all bonding on a waterproof bag, and each has a clear domain. Choosing within the domain is straightforward; choosing across domains is where programmes get into trouble, usually because an adhesive that worked on one component was carried over to a different material without re-validation.

The last row is worth highlighting because heat-activated tapes are the most under-specified item in this category. A tape is a complete adhesive system supplied in dry form, which removes mixing and pot life entirely, but it substitutes a thermal process window that is just as demanding: too little heat and the adhesive does not wet, too much and it wicks into the fabric or degrades. The routes and their verification are covered in our guide to seam taping and sealing methods.

Compliance is a live constraint on the third and fourth rows. Solvent-based systems carry volatile organic compound and restricted substance obligations that vary by market, and adhesive selection has to clear the same regulatory screen as the fabric. The relevant obligations are summarised in our article on REACH and CPSIA compliance testing, and material declarations are discussed in our piece on chemical transparency in manufacturing.

Testing a bonded joint so the number predicts the field

A peel or shear figure measured on a fresh coupon predicts almost nothing about year two, because it samples one point in a process distribution and one moment in a cure schedule. Three additions turn it into a prediction, and all three are cheap: test after conditioning, test at the extremes of the process window, and test the failure locus rather than just the force.

• Condition before testing. A joint tested dry at room temperature is not the joint the customer will use. Water immersion, humidity ageing and thermal cycling all attack the interface specifically.
• Test at the edges of the stated window, not the centre. A process that only works at its nominal set point will drift in production.
• Record where the sample failed, not only the force. A high number with interfacial failure is a warning, not a pass.
• Sample across the shift and across the batch. Timing drift produces a distribution, and a single coupon cannot show it.

Conditioning is the one that adds the most information per dollar, because the dominant degradation path for a bonded joint on a bag is hydrolytic and thermal rather than mechanical. Water attacks the interface, and it attacks it preferentially: a joint that loses half its strength after humidity ageing and none after thermal ageing is telling you the interface is the weak link, which points straight back at surface preparation. The accelerated protocols and how to interpret them are set out in our article on accelerated ageing tests for durability prediction, and the humidity and temperature chambers themselves are described in our piece on environmental testing of waterproof bags.

Independent laboratories such as SATRA run peel, shear and conditioned bond testing routinely, and the standard methods are published by ASTM International with parallel procedures in the ISO series. Cite the method, the conditioning, the rate and the failure locus with every number, because a peel strength without those four is not a specification.

The eight bonded-joint failures and the line that prevents each

Almost every bonded joint failure seen in the field resolves to one of eight causes, and each has a specific prevention that belongs on the tech pack. Written this way the adhesive stops being a brand name and becomes a controlled process.

• Untreated low-energy surface: require a minimum dyne level measured at the bonding station, immediately before assembly.
• Contamination from finish, slip agent or release agent: require a wipe-and-abrade step and a dyne check after it, not before.
• Treatment decay: require that treated material be bonded within a stated number of hours, and record the treatment date on the roll.
• Mix ratio error: require both components to be weighed, with the ratio and batch time recorded per mix.
• Expired pot life: require mixed material to be discarded rather than topped up, and log the discard.
• Wrong cure conditions: require a recorded temperature at the bond line and a stated time to handling strength.
• Starved or over-thick bond line: require a controlled pressure and a bond-line thickness target verified on a cut section.
• Wrong test for the loading mode: name peel, shear or fatigue to match how the joint is actually loaded in service.

The first three are surface problems, and they account for the majority of real failures, which is the whole thesis of this article in one line. The fourth through seventh are process problems that no material change will fix. Only the eighth is a design problem, and it is the one most often left to chance.

Writing the adhesive specification: the lines that get quoted

A bonded joint is specified the same way any other controlled process is specified: name the chemistry, the surface condition, the process window and the acceptance test, and require records. Twelve lines cover it, and a factory can quote accurately against them, which is the point — an adhesive that is specified only by brand invites substitution, and substitution is where reproducibility is lost.

• Chemistry family and grade, named, with substitution requiring written approval and re-validation.
• Surface energy requirement in dynes, measured at the bonding station immediately before assembly.
• Surface preparation sequence: wipe, abrade, treat, and the maximum time between preparation and bonding.
• Mix ratio by weight for two-part systems, with a tolerance and a per-mix record.
• Pot life and a discard rule, stated explicitly rather than left to operator judgement.
• Cure mechanism and window: temperature at the bond line, time, and time to handling strength.
• Pressure and bond-line thickness target, verified on a cut section at a stated frequency.
• Acceptance test: method, loading mode, rate, conditioning, minimum value, and the required failure locus.
• Open time and the maximum number of parts coated before assembly.
• Prohibited handling: no bare-hand contact with prepared surfaces, and gloves required at the station.
• Traceability: adhesive lot linked to the production batch, and the mix record retained.
• Regulatory clearance for the target markets, with a declaration on file.

The eighth line is the one that converts a test from a formality into a control, because it forces the failure locus to be recorded. A batch that produces high numbers with interfacial failure is a batch that will fail in the field, and without that line it passes inspection looking identical to a good batch.

The economics are favourable, because almost every line on that list costs process discipline rather than money. Surface preparation, weighing, timing and recording are free; the alternative is a warranty line. The broader cost picture is set out in our breakdown of custom waterproof bag cost, and the defect categories these lines prevent are catalogued in our article on common waterproof bag defects and how to prevent them.

If you want a bonded construction developed against a specific duty, send the substrate specification, the loading mode and the target life, and the adhesive system can be selected and validated against those rather than chosen from a catalogue. You can see how a programme moves from first enquiry through sampling into bulk production, and every style we produce starts at 500 pieces minimum, with samples in 6–10 working days and bulk in 35–50 days, quoted FOB Xiamen.

Frequently Asked Questions

Q1. What is the difference between adhesive failure and cohesive failure?

Adhesive failure means the bond let go at the interface, leaving one face clean. Cohesive failure means the adhesive layer itself split, leaving residue on both faces. Adhesive failure is a surface problem; cohesive failure is a cure or formulation problem.

Q2. Why does my bonded seam let go even though I upgraded the adhesive?

Because the failure was almost certainly at the interface, which means the surface was never properly prepared or its surface energy was too low. A stronger adhesive on an unprepared surface fails in exactly the same place.

Q3. What dyne level do I need for a reliable bond?

As a working rule, the substrate surface energy should exceed the surface tension of the adhesive by roughly ten dynes per centimetre. Most polyurethane adhesives want a substrate in the low to mid forties at minimum.

Q4. How do I measure surface energy on a fabric or film?

With dyne pens or dyne solutions. A liquid of known surface tension is applied and must remain as a continuous film for about two seconds. If it beads up, the surface energy is below that value and will not wet reliably.

Q5. Why does corona or flame treatment stop working over time?

The treatment oxidises the surface and that oxidised layer decays, faster at higher temperature. Material treated and then bonded three weeks later may have lost ten dynes, which is why a bonding window must be specified.

Q6. What happens if I get the two-component mix ratio wrong?

Excess isocyanate reacts with atmospheric moisture and produces carbon dioxide, foaming and weakening the bond. Excess polyol leaves the system under-crosslinked, soft and heat sensitive. A five per cent error is enough to change the failure mode.

Q7. What is pot life and why does it matter so much?

Pot life is the working window after mixing before viscosity rises too far for proper wetting. Material used past pot life produces joints that look like contamination failures, so the problem gets misdiagnosed and the fix gets missed.

Q8. Should I use a moisture-cure or a heat-cure adhesive?

Moisture cure needs no equipment and fills gaps well but is slow and humidity dependent. Heat cure is fast, uniform and strong but applies thermal load to the substrate. Two-part is the most controllable if you can discipline the mixing.

Q9. Why does my adhesive bond foam up after application?

Usually isocyanate excess reacting with moisture, either from the air or from the substrate. Check the mix ratio by weight, then check substrate moisture content and ambient humidity at the bonding station.

Q10. What is the ideal bond-line thickness?

A few tenths of a millimetre for most structural bag joints. Thinner lines raise shear strength but become brittle in peel; thicker lines improve peel but carry shear less efficiently and cure more slowly.

Q11. Why does one adhesive win a shear test and lose a peel test?

Shear loads the whole area uniformly and rewards stiff, thin bond lines. Peel concentrates load at a crack front and rewards tough, compliant ones. Test the mode the joint will actually see in service.

Q12. When should I weld instead of bond?

Whenever both substrates are weldable thermoplastics of the same family. Welding removes the interface entirely, which removes the surface energy, mixing and pot life problems in one step and lowers unit cost.

Q13. What does green strength mean and why does it cause failures?

Green strength is the load a joint can carry before it is fully cured. Parts moved or packed before reaching it creep permanently, and the resulting deformation looks like a material defect rather than a process one.

Q14. How do I stop sporadic bond failures that look random?

Look for a timing distribution rather than a material cause. Open time, tack time and treatment decay all produce a spread of results across a batch with no correlation to operator or lot.

Q15. How should I condition samples before bond testing?

Water immersion, humidity ageing and thermal cycling all attack the interface preferentially. A joint that loses strength after humidity ageing but not thermal ageing is telling you the interface is the weak link.

Q16. Does a high peel number guarantee a durable joint?

No. You must also record the failure locus. A high number achieved with interfacial failure is a warning rather than a pass, because the interface will degrade faster than the bulk adhesive.

Q17. What should I write on the tech pack for a bonded joint?

Chemistry and grade, minimum dyne level at the station, the preparation sequence and bonding window, mix ratio by weight with a record, pot life and discard rule, cure window, pressure and bond-line thickness, and the acceptance test with a required failure locus.

People Also Ask

What causes a bonded seam to fail on a waterproof bag?

Most often the surface, not the adhesive. Around eight in ten field bond failures are interfacial, which means poor preparation, contamination or surface energy that was too low to wet.

What dyne level is needed for adhesive bonding?

Generally the substrate should exceed the adhesive surface tension by about ten dynes. Clean polyester sits near 42 to 46; treated TPU runs 44 to 52.

Is cohesive failure good or bad?

Cohesive failure is usually the better outcome. It means the adhesive bonded to both surfaces and the limit was the adhesive itself, so the surface preparation was adequate.

Why does two-part polyurethane need weighing?

Because the stoichiometry of the reaction is fixed. A five per cent ratio error moves the joint from substrate failure to cohesive failure and can cause foaming or permanent softness.

Should I test bonded joints in peel or shear?

Test whichever mode the joint sees in service. Seam tapes are peeled, stiffener panels are sheared, and strap attachments need cyclic fatigue testing rather than a static number.

Does a stronger adhesive fix a failing bond?

Rarely. If the break is at the interface, the fix is surface preparation, treatment or contamination control, and a stronger grade will fail in the same place.