Can You Apply Polyurethane Over Epoxy? Adhesion and Recoat Rules

Yes, polyurethane over epoxy works, and it is one of the most common renovation specifications in industrial flooring. It fails only when one of three conditions is skipped.

The epoxy underneath must be sound and still bonded, its surface must be abraded, and it must be free of the waxy blush that forms while epoxy cures. Get those three right and the bond is usually stronger than the concrete beneath it.

This article sets out the recoat adhesion rules, how to detect amine blush, what a pull-off test needs to read, and the cases where no preparation will save the job.

The short answer, and its three conditions

Applying polyurethane over epoxy is mechanically sensible. Polyurethane is more flexible than epoxy, adheres well to a keyed epoxy surface, and adds ultraviolet stability that the epoxy underneath never had.

The conditions are simple to state and easy to skip. The existing film must be bonded, the surface must be mechanically keyed, and the surface must be chemically clean.

Each of those has a test attached, and the tests take less time than the argument that follows when they are omitted.

What none of them can fix is a floor that is already failing. A coating applied over a delaminating film inherits the delamination, and it will lift in the same places within a season.

Why the interface behaves the way it does

Epoxy cures by an addition reaction between resin and hardener, and it cures hard and glassy. Polyurethane cures into a more flexible film with a different modulus.

That difference is an advantage at the interface, because the polyurethane accommodates movement that would crack a second epoxy coat. It is also why the epoxy beneath must be sound: flexibility cannot bridge a void.

The bond itself is mechanical rather than chemical in most cases. The polyurethane wets the keyed profile of the epoxy and locks into it, which is why surface abrasion matters more than any primer.

The modulus difference is the reason polyurethane over epoxy is prescribed over an old epoxy rather than a second epoxy. A flexible film on a rigid one tolerates the movement that a rigid film on a rigid one cannot.

A chemical bond is possible while the epoxy is still within its recoat window, when reactive groups remain on the surface. Miss that window and the bond becomes purely mechanical.

Condition one: the epoxy must be sound

An epoxy that is sound enough to overcoat will resist a blade pushed under the film edge at several points, and it will sound solid under a hammer. Where the film lifts easily it has already failed, and it must come off before anything is ordered.

Tap the floor with a hammer or a chain and listen. A dull note over a hollow sound marks a blister, and blisters do not heal when they are covered.

Look for the classic failures: lifting at doorways and drains, crazing around columns, and white patches where moisture has pushed from below. Each of those is a substrate problem rather than a coating problem.

Where a film is sound across most of the area and failing at 5 per cent of it, cut out the failures and patch them before the full coat. That is normal and it is not a reason to abandon the specification.

Coverage is worth calculating rather than guessing. A 320 micron coat needs roughly 0.45 kilograms per square metre at 1.4 kilograms per litre, so a 1,000 square metre hall needs about 450 kilograms of mixed material before waste.

Clean as you grind. Vacuum the surface the same day, because dust and epoxy residue both stop the new film from wetting the profile you have just created.

The same logic explains why surface abrasion is specified in microns of key rather than in passes of a machine. A crew that makes three light passes has done less than one that makes a single pass at the right depth, and the difference only shows up years later.

Where access is difficult, a hand-held grinder on a polyurethane over epoxy renovation is often the difference between a proper key and a hopeful one. Corners, plinths and drain surrounds are where the coat fails first, and they are the areas a ride-on machine never reaches.

Figure 1: the operator works along a sawn joint and holds the hammer near its head, so the tap lands on a narrow strip.

Condition two: the surface must be keyed

A cured epoxy is glossy and chemically inert. Nothing will bond to it until the surface is opened, and surface abrasion is the only reliable way to do that across a large area.

Diamond grinding with a 30 to 60 grit tool is the normal method, and the surface profile grade it leaves is what the bond depends on. Shot blasting works too and is faster on a large hall, though it can profile soft epoxy unevenly.

The target is a uniformly dull, matte surface with no shiny patches. Shiny patches are unkeyed patches, and they are where the new coat will lift first.

Edge detail and corners need a hand grinder, because the big machine cannot reach them. Those are also the places where coating failure is most visible to a client.

• Grind until the sheen is gone across the whole area.
• Vacuum immediately; dust left overnight re-contaminates the surface.
• Do not acid etch an epoxy; it dulls the surface without keying it.
• Check coverage on hands and knees under a light before priming.

Condition three: no amine blush

Amine blush is a waxy film that forms on epoxy when amine hardeners react with moisture and carbon dioxide in the air. It is invisible, and it is a release agent.

It forms most readily in cool, damp conditions and on slow-curing products. A floor that cured over a humid weekend is a likely candidate, and so is a basement or a chill store.

Water alone will not remove it. A warm detergent wash followed by clean water is the standard treatment, and some specifications require a solvent wipe after that.

The test is simple: pour a cup of water on the surface. If it beads rather than wets out, there is a film on the surface and it is probably blush.

Any epoxy residue left behind, whether blush or a curing compound or a release agent, produces the same result. The new coat bonds to the contamination and the contamination bonds to nothing.

Figure 2: the water sits in separate beads at the centre of the bay and starts to spread where the surface has been abraded.

Test before you commit: the trial patch

A trial patch answers the whole question for the price of a few hours. Prepare one square metre exactly as the full job would be prepared, coat it, cure it, and try to remove it.

Leave the patch for 24 hours, then cut a cross through it with a knife and lever at the corner. If the film resists and tears rather than peeling cleanly, the bond is good.

A trial patch is also where you discover a reaction problem. Some solvent-borne polyurethanes soften a fresh epoxy, and that shows up as wrinkling within an hour rather than at handover.

Where a client has an unknown coating from a previous contractor, the trial patch is not optional. It is the only way to know what is on the floor without a laboratory test.

Pull-off testing and what number passes

A pull-off test measures the force needed to detach a bonded disc, and it is the only quantitative evidence of bond on a renovation. A glued dolly is pulled until something fails.

Read the failure mode before the number. A result of 2 megapascals with the film still attached to the concrete is a pass; the same number with a clean separation at the interface is a fail.

For coating work on concrete we look for at least 1.5 megapascals, with the failure in the substrate. At 1.5 megapascals and above the existing film is sound enough to coat. Below 1.5 megapascals the result describes the slab rather than the film, and a clean break at the epoxy interface overrides the number.

Table 1: Reading a pull-off result on an existing epoxy.

Record the location of every test on a plan before the results are read. A single number without a location cannot be compared with a later survey, and a client asking why one bay failed deserves an answer.

Take at least one test per 500 square metres and one in every area that looks different. A single result on a large floor tells you about one square metre.

When the answer is no

Some incompatible coatings cannot be overcoated whatever the preparation. Moisture-cure polyurethane, oil-modified sealers, waxes and silicone-based products all fall into that group.

A moisture-cure film cures from atmospheric humidity and stays sensitive to solvents. A solvent-borne polyurethane applied over it will wrinkle it, and the damage appears within the hour.

Where the existing system is unknown, test for solvent sensitivity on a small patch: wipe with the solvent in the new product and watch for softening or colour transfer.

The list of incompatible coatings is longer than most contractors expect, and it includes several products sold as floor paints. A moisture-cure urethane is the most common of them on older industrial floors, because it was cheap and it was sold as a floor finish.

Covering a system that has already failed rarely works either. Where an epoxy has blistered from rising moisture, a polyurethane over epoxy over it will blister in the same pattern within a season.

The other clear no is a slab with rising moisture, and a concrete moisture test is the only way to establish it before a polyurethane over epoxy order is placed. If the epoxy is blistering from below, a new coat will blister too, and the fix is a moisture barrier rather than an adhesion promoter.

Table 2: Existing finishes and whether they accept a polyurethane coat.

Where an adhesion promoter earns its place

An adhesion promoter is a thin, low-viscosity resin that wets a difficult surface and provides a bond between two systems that would otherwise struggle.

It is useful on dense, power-trowelled concrete, on surfaces that cannot be fully abraded, and where a primer with a long open time is needed for a large area.

It will not neutralise epoxy residue, dust or blush. A promoter applied over any of those bonds to the contamination, and the failure then occurs one layer higher and costs more to put right.

It also does not raise a pull-off result. The dolly still fails at the same interface, because a promoter changes the surface chemistry rather than the strength of the film above it.

Where a promoter is used, the recoat adhesion result should be checked with the same pull-off test as the rest of the job. An untested promoter is an assumption, not a system.

Intercoat timing and the recoat window

Once the epoxy is prepared, the new coat should follow within the working day. A ground surface attracts dust and moisture, and a floor left overnight needs vacuuming again in the morning.

Between the polyurethane coats, the recoat adhesion rules come from the product datasheet, not from habit. Most epoxy systems allow overcoating between 8 and 24 hours at 20 °C, and the overcoating time window is measured from the end of the previous coat.

Beyond that window the surface must be abraded again. On a large hall this is often the difference between finishing in one visit and returning the next week.

On a renovation the first coat is usually the primer, and the clock starts when it is down rather than when the crew arrives. A polyurethane over epoxy system applied in one visit is a schedule decision made the week before, not on the morning.

Plan the sequence backwards from the window. Primer, body and seal in one continuous operation is achievable on 500 square metres with two crews and impossible with one.

What to write into a renovation specification

A specification that produces a good result names the preparation, the test and the acceptance criterion. Preparation alone is an instruction; a test makes it enforceable.

Where a survey finds incompatible coatings, the clause changes from preparation to removal. That is a different budget and a different programme, and it is better discovered at tender than at handover.

Where the existing film is of unknown origin, require a trial patch and a pull-off test before the main order is released. That single clause prevents most of the failures we are asked to inspect.

Where the film is known and sound, a simpler clause works: abrade to a uniform matte, wash and rinse, verify a water break, then apply within eight hours.

Table 3: A specification clause that can be checked.

Five lines, all of them measurable, and none of them requiring a laboratory. That is the difference between a specification and a wish.

Frequently asked questions

Can you put polyurethane directly over epoxy?

Yes, if the epoxy is sound, abraded to a matte finish and free of amine blush. A small test patch confirms the bond before the main order.

How long should epoxy cure before polyurethane goes on?

Between 24 and 72 hours at 20 °C, counted from the end of the epoxy coat. That is the chemical-bond window. After 72 hours the bond becomes mechanical, and the eight-hour clock in the clause starts when the abrasion is finished.

What happens if you coat over amine blush?

The new film bonds to the blush rather than the epoxy, and it lifts in sheets. Washing with warm detergent and checking a water break prevents it.

What pull-off result is acceptable?

A minimum of 1.5 megapascals on a pull-off test, and the break has to be in the concrete. A high reading that separates cleanly at the interface is still a fail.

Will polyurethane cover a flaking epoxy floor?

No. The existing film must be removed where it has failed. Coating over a delaminating floor transfers the problem upward.

Can polyurethane go over moisture-cure polyurethane?

Generally no. Those films are solvent sensitive and a solvent-borne coat will wrinkle them within the hour.

Do I need a primer between epoxy and polyurethane?

Often not, if the epoxy is freshly abraded. An adhesion promoter helps on dense concrete or where the surface cannot be fully keyed.

Why food plants work with Kaida Paint

Kaida Paint supplies the polyurethane coats and the primers for renovation work, and we will say when a floor should be stripped instead of covered. Send photographs of the existing film and the results of a hammer survey, and we will tell you whether polyurethane over epoxy is the right answer for that slab or whether the money is better spent on removal.