Why Is My Epoxy Floor Peeling? Adhesive vs Cohesive Failure

Adhesive failure and cohesive failure are not the same thing

Two terms separate a useful diagnosis from an argument. Adhesive failure is separation at an interface — the coating parts from the surface it was applied to, leaving that surface bare. That interface may be the coating-to-concrete bond line, or it may be between two coats of the same system, in which case it is called intercoat delamination. In both cases the two materials never joined properly, or something in between them prevented the join.

Cohesive failure is different: the material itself tears. It has two forms and they mean opposite things. The coating may split internally, leaving resin on both faces of the break — a sign of an under-cured, mis-mixed or over-thick film, or of a coat applied over a soft, uncured layer. Or the concrete may tear, so that when the peeled coating is turned over, a layer of grey concrete is stuck to its back. That second case is the good news buried in a bad situation: the adhesive bond was stronger than the concrete, and the coating did not fail at all. The slab did.

This distinction governs the money. Adhesive failure at the bond line usually means the whole area prepared the same way is suspect, because the same defect was applied everywhere. Intercoat delamination may leave the primer and base coat serviceable. Cohesive failure of the concrete means substrate reinstatement precedes any coating discussion. Cohesive failure within the film points at mixing, cure or thickness on a particular day, and often maps to a single bay.

Where in the build-up the separation happens

A resin floor is a stack, and each plane in that stack has a characteristic way of failing. The topcoat can wear or chalk. Between topcoat and body coat, and between body coat and primer, sit intercoat planes where delamination occurs when the recoat window was missed or amine blush was left in place. Beneath the primer is the bond line — the interface with the concrete — and beneath that, the concrete itself, which can tear if it is weak, dusting, or covered in laitance.

The bond line carries almost all of the risk. Everything above it is resin joined to resin, chemistry that is designed to be compatible. The bond line is resin joined to a porous mineral surface whose top few millimetres may be the weakest part of the entire slab. Sparco's TDS requires the substrate to be dry, sound, clean and free from oil, grease, loose material and other bond-inhibiting materials, prepared mechanically by ball blasting, milling or diamond grinding, with weak concrete removed and blowholes fully exposed — and it sets a minimum pull-off strength of the prepared substrate of 1.5 N/mm² before the primer goes on.

The primer selection at that plane matters as much as the preparation. An absorbent, open-pored slab drinks a low-viscosity primer and cures with resin locked into the pores; the same product on a dense, closed, power-trowelled surface may sit on top. Sparco Epoxy Bonding Primer #100 is a bonding primer for prepared concrete, with a maximum permissible substrate moisture content of 5%; Sparco SB Prime 107 is a two-part solvent-based epoxy primer whose low viscosity suits penetration into absorbent substrates. Which is appropriate depends on substrate and exposure, and should be confirmed through technical review.

Read the underside of the peeled piece

The most informative object on a failed floor is a lifted fragment of the coating, viewed from below. Nothing else on site tells you as much in as little time. Cut a square through the film with a diamond blade so you get an intact piece rather than a shattered one, lever it up, and turn it over under good light. What is stuck to its back is the diagnosis.

Concrete stuck to it — the concrete failed, not the bond. A clean, glossy, unmarked underside — the bond never formed; the resin cured against something it could not key into. Fine grey or white powder — laitance, the weak cement skin that mechanical preparation exists to remove. A waxy, sheeny film or beading — a curing compound, sealer, or oil contamination. A resin surface on both faces of the break, with no interface at all — cohesive failure within the coating. Blisters or a damp, cupped underside — moisture pressure from beneath.

Where a quantitative figure is needed — for a warranty claim, or to qualify a prepared substrate before recoating — the concrete pull-off method is ASTM D7234, Standard Test Method for Pull-Off Adhesion Strength of Coatings on Concrete Using Portable Pull-Off Adhesion Testers. Coatings thicker than roughly 20 mils must be scored down to the concrete before the dolly is bonded, or the test measures the wrong thing. The test also reports the failure mode, which is exactly the information the underside gives you by eye. Acceptance values come from the specification or the manufacturer's TDS, not from the standard.

What actually breaks the bond

Contamination and unprepared substrate lead the list. Laitance, curing compound, a previous sealer, oil driven into the pores by years of forklift traffic, or rubber and tyre residue all sit between the resin and the concrete. So does an acid-etched or merely swept surface: etching does not reliably remove curing compound, and sweeping removes nothing at all. Mechanical preparation by ball blasting, milling or diamond grinding is what Sparco's TDS calls for, and it does two jobs — it removes the weak layer and it opens a profile for the primer to key into.

Moisture debonds from below. Vapour rising from the ground, or from a slab that has not finished drying, arrives at the underside of an impermeable film and exerts pressure there. Where a hydrostatic head exists — a high water table, a leaking service, ground water at a loading bay after monsoon rain — the pressure can lift a well-bonded film outright. This is the mechanism behind floors that were laid perfectly and separated anyway, and it is why slab moisture testing precedes rather than follows a repair decision.

The remaining causes concentrate on the day of application. A missed recoat window leaves the previous coat overcured, with neither a chemical key nor a mechanical one, and the next coat sits on it rather than joining it. Amine blush — a greasy carbamate bloom formed when a curing epoxy meets moisture and carbon dioxide — must be washed off, not overcoated; in a humid tropical environment it forms readily. Applying when the substrate is within a few degrees of dew point invites condensation on the surface at the moment of application. Choosing a primer that cannot penetrate an absorbent slab, or thermal shock from hot washdown onto a coating not formulated for it, complete the list.

Common mistakes — and the checks that prevent them

The first mistake is recoating over the peeling. A new coat cannot repair a bond line it does not touch; it inherits the old one. The second is treating a hollow-sounding but visually intact area as sound. Tap-testing almost always reveals a debonded zone considerably larger than the peeled one, and the recoat boundary should follow the hollow boundary, not the visible one. The third is changing product after a concrete-cohesive failure, which addresses a bond that never failed.

The fourth is diagnosing without turning the piece over. A photograph of a peeling floor from standing height supports any theory anyone wants to hold, which is why disputes over failed floors last so long. A single lifted fragment, examined from below and photographed, closes most of them. Where the parties disagree, a pull-off programme to ASTM D7234 across failed and apparently sound areas gives a figure and a failure mode rather than an opinion.

Before any recoat is priced, the following should be established on site. None of it requires specialist equipment beyond a diamond blade and a hammer.

• Tap-test well beyond the visible failure and chalk the true hollow boundary
• Cut and lift at least three samples from different areas, and photograph every underside
• Record whether each underside shows concrete, gloss, powder, a waxy film, or torn resin
• Check the slab for a damp-proof membrane and test internal moisture before deciding anything
• Retrieve the original specification, TDS, preparation method and recorded coverage per coat
• Note the ambient and substrate temperature and humidity if a recoat is imminent — application near dew point is a repeat of the original error
• Confirm that the prepared substrate reaches the TDS minimum pull-off strength of 1.5 N/mm² before priming

Can it be patched, or must it come off?

The short answer is that the failure plane decides. A localised cohesive concrete failure — a damaged bay, an impact crater — is a patch: cut back to sound material, reinstate the substrate, re-prime and rebuild. Intercoat delamination over a sound, well-bonded primer may sometimes be corrected by removing the delaminated coat, abrading the coat beneath and reapplying. Adhesive failure at the bond line rarely patches, because the preparation that produced it was applied across the whole floor, and the sound-looking remainder is on the same trajectory.

Where full removal and reinstatement is the outcome, the substrate is rebuilt before the system is rebuilt — commonly with a non-shrink repair mortar such as Sparco Epoxy Mortar, or Sparco Epoxy Thixotropic Compound where sag resistance is needed — and the new system is then applied over a primer selected for the prepared slab. Sparcofloor SL 200, a solvent-free two-component self-smoothing epoxy, is one of the systems commonly used for reinstatement where a seamless finish is required. Suitability depends on traffic, chemical exposure, moisture and downtime.

This is deliberately a summary. The full repair-versus-recoat-versus-replace decision brings in downtime, remaining slab life, moisture mitigation and the cost of removal itself, and it is treated separately in our guide to repairing, recoating or replacing an industrial floor.

Frequently asked questions