A label rarely peels off because of “bad glue”. In most cases the substrate’s surface energy is too low (below 36 dyne/cm), its texture cuts the real contact area, its curvature applies permanent stress, or the application never gave the adhesive time to wet out. All four causes are diagnosed by looking at where the bond failed.
The symptom is always the same: the label holds at the end of the line, then lifts at one corner three days later, or drops off after an oven cycle. The instinct is to ask for a stronger adhesive. That is rarely the right answer and sometimes the worst one, because a more aggressive adhesive applied to a substrate it cannot wet produces exactly the same result.
So this article starts where a diagnosis should start: at the failure surface. Then it explains the four physical causes behind it, with the figures you need to tell them apart.
Start here: where did the bond fail?
The most useful question is not “why did it peel?” but “where did the separation happen?”. Peel a failed label off, look at both faces, and match what you see against this table.
| What you observe | Interpretation | Likely cause | First correction |
|---|---|---|---|
| The label comes away clean, no residue on the substrate | Failure at the adhesive / substrate interface | Wetting failure: surface energy too low, contaminated surface, or cold application | LSE adhesive, degrease, control the application temperature |
| The adhesive stays on the substrate and separates from the facestock | Failure at the adhesive / facestock interface | Insufficient anchorage of the adhesive to the facestock, often a missing film surface treatment | Change the adhesive construction, not the adhesive family |
| The adhesive splits in two, residue on both sides | Cohesive failure of the adhesive | Mechanical overload, service temperature exceeded, or creep under permanent stress | Adhesive with higher shear resistance |
| Lifting at the corners or edges only | Permanent peel stress | Curvature, facestock too rigid, label too large, a ridge crossed | Soft facestock, radiused corners, reduced size |
| Bubbles or blisters appearing after a few days | Substrate outgassing or trapped air | Unstabilised plastic or coating, applied without a squeegee | Air-egress adhesive or micro-channels, progressive roller application |
| Holds correctly, then fails after an oven cycle | Loss of adhesion at temperature, or migration | Service temperature exceeded, plasticiser migration from a flexible PVC | High-temperature adhesive, migration barrier |
| Holds in-house, fails on site | Real conditions differ from test conditions | Humidity, UV, solvents, chemical cleaning | Requalify against the real service cycle |
Everything below explains the four causes this table points to.
What is surface energy, and why does it block adhesion?
Surface energy measures a material’s ability to attract the molecules of a liquid, in this case the adhesive. It is expressed in dyne/cm (or mN/m, the same value).
A pressure-sensitive adhesive is not a glue that cures: it is a very viscous liquid that has to flow out across the substrate and penetrate its micro-roughness. This is called wetting. If the substrate’s surface energy is lower than the adhesive’s, the adhesive behaves like a water droplet on a non-stick pan: it beads up instead of spreading.
A pressure-sensitive adhesive has to flow out to bond. On a low surface energy substrate such as polypropylene it only wets partially: the label holds on application and lifts a few days later.
The reference threshold: 36 dyne/cm. Below it, the material is described as low surface energy (LSE) and is hard to bond. Above it, adhesion becomes far more predictable.
Surface energy of common substrates
| Substrate | Indicative surface energy (dyne/cm) | Class | Practical consequence |
|---|---|---|---|
| PTFE | ≈ 18 | Extreme LSE | Not bondable without surface treatment |
| Silicone | ≈ 24 | Extreme LSE | Not bondable; also contaminates neighbouring surfaces |
| PVF | ≈ 28 | LSE | LSE adhesive and adhesion promoter both required |
| Polypropylene (PP) | ≈ 29 | LSE | Cause number one of label failures on plastic parts |
| Polyethylene (PE) | ≈ 31 | LSE | Corona or flame treatment recommended |
| EVA | ≈ 33 | LSE | Dedicated LSE adhesive |
| Acetal (POM), polystyrene | ≈ 36 | Borderline | Risk zone: always test |
| ABS, rigid PVC | ≈ 35 – 42 | Variable | Strongly dependent on formulation and plasticisers |
| Polyamide (PA), PET, polycarbonate | ≈ 38 – 46 | MSE / HSE | Adhesion generally reliable |
| Powder and epoxy coatings | Very wide range | Unpredictable | To be treated as a separate case, see below |
| Glass | ≈ 200 – 300 | HSE | Excellent, provided it is clean |
| Aluminium | ≈ 840 | HSE | Excellent, but oxidation and oils change everything |
Sources: LSE / MSE / HSE classification and polyolefin values based on 3M technical documentation and adhesive converter selection guides. Orders of magnitude for clean, untreated materials.
A word of caution on this table. A single plastic family varies with formulation, release agents, fillers and ageing. The table is there to direct a diagnosis, not to validate a specification: only a measurement on the actual part, with dyne test inks or a contact angle reading, is authoritative.
If your substrate is PP, PE, silicone or a powder coating, the default hypothesis is not “the glue is bad”, it is “the adhesive is not wetting this substrate”. You change the adhesive family, not the supplier.
Why do powder coatings make labels peel off?
This is the most common case in sheet metal work, and the most often misdiagnosed, because four mechanisms stack up.
1. Surface energy is unpredictable. Unlike a thermoplastic, a powder coating has no characteristic surface energy value: it depends on the resin, the fillers and the additives. Two batches of the same reference can behave differently.
2. Waxes and slip additives form a weak boundary layer. A powder formulation contains waxes (polyethylene, polypropylene, sometimes PTFE) used as gloss inhibitors and for scratch resistance, along with silicone flow and slip additives. These additives are deliberately incompatible with the resin and have a low surface tension: that is precisely how they work. During cure they migrate to the surface and form a layer a few molecules thick, invisible and with no change in gloss, on which the adhesive cannot build adhesion. Work on polyester powders is unambiguous: this weak boundary layer degrades adhesion regardless of the powder system used.
3. Migration continues after application. When these additives are not crosslinked into the resin network, they keep migrating — this time into the adhesive of the label just applied. The label can therefore bond correctly at first, then be progressively poisoned by its own substrate. This is exactly the symptom described at the top of this article: a correct bond at the end of the line, a failure three days later.
4. Texture cuts the real contact area. Structured, hammered or orange-peel finishes offer the adhesive only the peaks of their relief. The effective contact area can drop to a fraction of the apparent area.
What to do: test on a part that has been through the real cure cycle, never on a clean test panel. Choose an acrylic adhesive designed for low surface energy, and increase the adhesive thickness or move to a foam construction to bridge the relief. Note that an alcohol wipe is often not enough on powder: the film is smoother and more solvent-resistant than a liquid paint, so the solvent does not bite into the surface. A light abrasive is sometimes more effective than a solvent. An adhesion promoter remains the last resort.
And one question to ask upstream: ask the coater whether the powder contains waxes or slip additives, and what the finish is. A matt powder is very often a wax-loaded powder, and therefore a risk substrate, identifiable before the first trial
Roughness and texture: is the adhesive really touching the surface?
Adhesion is proportional to the area actually in contact, not to the area of the label. On a rough, grainy or ribbed substrate, a thin adhesive does not reach into the valleys: it bridges the peaks and leaves voids.
The correction logic is simple: the more irregular the substrate, the thicker and softer the adhesive has to be.
- A thin transfer adhesive, a few tens of microns, assumes a smooth substrate: glass, polished metal, glossy moulded plastic.
- On a textured surface you need a thicker, more flowable adhesive, or a foam construction able to conform to the relief.
- On a porous surface such as timber, uncoated board or concrete the opposite applies: the adhesive can be absorbed, and you have to trade off penetration against holding power.
The same reasoning applies to release agent traces, cutting oils, dust and condensation: these are parasitic layers that physically separate the adhesive from the substrate. Degreasing with isopropyl alcohol and a lint-free cloth, dried before application, resolves a surprising number of “adhesive problems”.
Curvature and small diameters: why does the label lift at the corners?
When a label is applied to a curved surface, the facestock tries to return to its flat shape. That elastic memory applies permanent peel stress to the edges: 24 hours a day, for the whole life of the product. Failure therefore always starts at a corner or an edge, never at the centre.
This flagging effect gets worse with a small diameter, a rigid or thick facestock, a label that is large relative to the radius, square corners, and any ridge or seam crossed by the label.
The facestock’s elastic memory works against the adhesive around the clock. Failure starts at a corner, which is why radiusing the corners is the cheapest correction available — and free at the die-cutting stage.
Fixing flagging on a curved surface
- Soften the facestock. A PE film or a soft paper follows a curve that PET refuses. This is the most powerful lever and often the least expensive.
- Radius the corners. A radius removes the peel initiation point. Free at the die-cutting stage.
- Reduce the dimension across the curvature, and keep the label edge clear of any ridge, rib or seam. A few millimetres are enough.
- Choose an adhesive with high shear resistance, able to hold under continuous stress, not merely one with high initial tack.
- Move to a self-laminating or wrap-around format when the diameter is very small, as on cables, syringes and tubes: the clear tail closes onto itself rather than onto the substrate.
What if the problem is not the adhesive, but the application?
A pressure-sensitive adhesive needs three things that production regularly forgets: pressure, time, and a correct temperature.
How much pressure does an adhesive need?
Contact is not made by gravity. Technical datasheets for high-performance acrylic adhesives recommend an application pressure of the order of 100 kPa, about 15 psi, measured at the adhesive — a roller or a press platen, not a finger.
How long before the adhesive reaches full strength?
At room temperature, an acrylic adhesive reaches roughly 50 % of its final strength after 20 minutes, 90 % after 24 hours and 100 % after 72 hours. A pull test carried out five minutes after application therefore does not measure the product’s performance, it measures its initial tack. Never test before 24 hours.
At what temperature can a label be applied?
The ideal range is 21 to 38 °C, that is 70 to 100 °F. Below 10 °C (50 °F) the adhesive becomes too firm to wet out and application is not recommended, unless the product is specifically formulated for cold surfaces. The good news: once the assembly has been made correctly, low-temperature performance remains satisfactory. Another useful lever is a short moderate heat cycle, one hour at 65 °C (150 °F), for instance, which brings the bond to its final performance immediately. That matters when the part goes straight into service off the line.
Five habits that fix most cases
- Identify the substrate before choosing the adhesive: exact material, treatment, coating, surface condition. A generic plastic reference is not enough.
- Clean and let it dry. Isopropyl alcohol, lint-free cloth, complete drying.
- Apply real pressure with a roller or a platen, and allow 24 to 72 hours of wet-out before any loading or any test.
- Match the facestock to the geometry, not just the adhesive: flexibility, thickness, radiused corners, size.
- Test on the actual part, through the actual cycle, not on a clean test panel at 22 °C.
Frequently asked questions
How do I know whether my plastic is LSE?
A dyne pen gives an answer in seconds: if the ink pulls back into droplets instead of staying as a continuous film for two seconds, the surface energy is below the ink’s rating. By default, treat PP, PE, EVA, silicone, PTFE and powder coatings as LSE.
Does a stronger adhesive solve a surface energy problem?
No. “Stronger” usually means a higher adhesion value measured on steel, therefore on a high surface energy substrate. On polypropylene, a very high-performance adhesive that does not wet the surface delivers a poorer result than a dedicated LSE adhesive with lower figures on paper.
Is corona treatment permanent?
No, it decays. The effect peaks immediately after treatment and falls away over days to weeks depending on the material and storage conditions. A treated substrate has to be labelled promptly, otherwise the treatment has to be renewed.
How long should I wait before testing adhesion?
24 hours minimum, 72 hours for a measurement representative of final performance. An immediate test measures tack, not adhesion.
Why does my label hold in-house and fail at my customer’s site?
Because the conditions differ: application temperature, substrate cleanliness, application pressure, but also the service environment — chemical cleaning, condensation, UV, vibration. Qualification has to reproduce the real cycle, not workshop conditions.
See our labelling solutions for product information.
Alsace Techniques Étiquetage has been converting adhesive and non-adhesive materials since 1989 in Val-de-Moder, Alsace, France: flexographic, digital and screen printing, precision die-cutting and laminating, up to 610 mm web width, with more than 700 material references in stock and an ISO 7 classified area.