The Recoat That Peeled Off in Sheets

Why this matters

A screen and recoat is the cheapest service a hardwood shop sells and one of the easiest to get wrong, because everything that decides whether it holds happened before the crew arrived. The new coat can only bond to what is already on the floor. If that surface carries years of polish or wax, the recoat goes down clean, dries hard, looks perfect at the walkthrough, and peels within weeks. This case reconstructs one of those failures from the job file and shows where the record was blank.

The callback

Three and a half weeks after the shop screened and recoated about 220 square feet of red oak in a kitchen and breakfast nook, the homeowner sent photos. In front of the sink, at the refrigerator and through the doorway to the dining room, the new finish was lifting at the edges and coming off in clear, flexible sheets the size of a hand. Away from the traffic paths the floor looked fine.

The crew lead who took the call started where the evidence was, at a peel edge.

What the peel says

He slid a razor blade under a lifting edge and pulled a strip. It came off cleanly as one clear film, and the surface underneath was the old finish, intact, dull where the screen had abraded it. No bare wood showed anywhere.

That locates the failure precisely. There are two places a floor finish can let go, and they mean different things:

• Between the new coat and the old finish (intercoat failure). The new coat never bonded to the surface it was put on. The cause is on that surface: a contaminant, or abrasion too light to give the new coat something to grip.
• Between the old finish and the wood. The whole finish system is lifting, which points at the wood or the original finish: moisture from below, an incompatible original finish, or a finish applied over a contaminated sanding.

Clean sheets over intact, scuffed old finish is the first kind. The peel pattern added to it. Failure concentrated in the traffic paths, with the rest of the floor holding, is what an intercoat bond that is weak everywhere looks like once foot traffic, grit and chair legs start working it. The weak bond is floor-wide; the traffic is where it shows first.

So the question became: what was on that old finish?

Reading the job file backwards

The crew lead pulled the file and read it from the failure back to the first visit.

The crew's job sheet. "Screened with 120-grit screen, vacuumed, tacked with dry microfiber, one coat waterborne polyurethane, 2 hours between steps, room closed." All of that is a normal recoat. What the sheet did not have was a line for an adhesion test, because the form had no such field.

The estimate sheet. "Oak, site-finished, about 15 years old. Finish in fair condition, worn at traffic areas, no bare wood. Recommend screen and recoat." No question about how the floor had been cleaned or maintained. No test patch recorded.

The homeowner's answer to the question nobody asked. The crew lead called and asked what she had been using on the floor. For about six years she had applied a "floor restorer" product every couple of months, a liquid that went on with a mop and left the floor shiny again. Its label described it as a polish that fills scratches.

That was the cause, sitting in the one place the record never looked. Products sold to "restore" or "renew" the shine on a finished floor generally leave an acrylic polish film on top of the finish, and repeated applications build it up layer on layer. Here is the stack on this floor and the plane that let go:

 new coat, waterborne poly   =======================
 the plane that let go  -->  - - - - - - - - - - - -
 polish, top scuffed by the  \/\/\/\/\/\/\/\/\/\/\/\
   120-grit screen           .......................
 polish below, six years of  .......................
   restorer, untouched       .......................
 original finish, 15 years   -----------------------
 red oak                     | | | | | | | | | | | |

A screening pad scuffs the top of that film; it does not reliably cut through it. The new coat bonded to polish, and polish is designed to be a sacrificial, removable layer with weak adhesion to anything put on top of it. Wax and oil-soap residues and silicone-containing cleaners cause the same failure through the same path.

Proving it before re-scoping the job

The crew lead did not want to sell a full sand on the strength of a phone call, so he tested on site.

The tape test on a non-peeled area. In the nook, where the new coat was still intact, he scored a small crosshatch through the new film with a razor, pressed a strip of strong tape over it, and pulled it off sharply. The new finish came up with the tape in squares. A sound intercoat bond leaves the crosshatch nearly intact.

A test of the old surface. In a closet corner that had been recoated but saw no traffic, he lifted a section of new film, then wiped the exposed old surface with a cloth dampened in the polish maker's recommended remover. The cloth came back cloudy, and the surface beneath went from glossy to dull. That is polish coming off; cured polyurethane does not dissolve that way.

The tests confirmed what the file reconstruction suggested: polish on top of the original finish, never removed, never tested for.

Why the fix is a full sand, not a second recoat

Three repairs were possible on paper. Only one would hold.

Recoat again after a heavier screening. Rejected. Six years of polish is not evenly distributed; it is thickest where the homeowner mopped most and thinnest under furniture, and screening cannot tell which is which. A second recoat would bond in some places and fail in others, and there would be no way to know which before it peeled.

Strip the polish, then recoat. Polish removers exist, and on a floor with one or two applications this can work. After six years, the remover would have to dissolve polish that had worked into the scratches and worn spots of the old finish, and any residue left there would sit under the new coat. Water-based removers on a 15-year-old floor with worn spots also put water at bare-wood edges. Too much uncertainty to warranty.

Sand to bare wood and refinish. Accepted. Sanding removes the polish, the old finish and the failed recoat together and leaves a surface whose condition the crew controls. It is a larger job than the recoat, and the shop said so plainly: the recoat was the wrong service for this floor, and the shop should have found that out before selling it.

The homeowner agreed. The shop credited its recoat against the refinish on its own warranty ticket.

The refinish

Before sanding, check the house age and the floor's surroundings. The house was built in 1994, after the 1978 line, so the lead-safe rules for pre-1978 homes did not apply; on an older house the shop would have worked under 'The Lead-Safe Floor Work SOP' before a sander touched the floor.

Dust. Sanding oak finish and wood produces fine wood dust, and hardwood dust is a respiratory hazard and a recognized carcinogen. The crew ran the sander with its dust collection connected and the bag emptied before it was half full, sealed the doorways with poly, covered the HVAC returns, and wore respirators under the shop's respiratory program. Sanding dust from finishes is also combustible: the bag contents went into a metal container outside, never left in a pile or in the van.

Sanding. The crew followed the grit sequence in 'Sanding Machines and Grit Sequences' for a finished oak floor, starting coarse enough to cut through the polish and finish in one pass without chasing it, then working up through the sequence, with the edges matched to the field.

Finishing. Waterborne polyurethane, three coats, per the finish maker's recoat windows, with the light abrasion between coats the maker specifies. The finish maker's instructions also give ventilation and temperature limits: windows open with fans exhausting outward, and no pilot lights or space heaters in the work area for a solvent-bearing product. Applicator pads and rags from an oil-modified product can self-heat and ignite; these were waterborne, but the crew's standing practice is to lay all rags flat outdoors to dry or put them in a water-filled closed metal can.

How it was confirmed

At the walkthrough the floor was judged standing, in normal light, which is how finished work is judged. The crew lead ran the crosshatch tape test again on a closet corner after the finish maker's cure time: the crosshatch stayed down. He returned at eight weeks, after the traffic paths had seen two months of use: no lifting at the sink, the refrigerator or the doorway.

He also gave the homeowner a written care sheet naming the cleaner the finish maker recommends and stating that shine restorers, polishes, wax and oil soaps go on the list of products not to use, because they would make the next recoat impossible too.

What the shop changed

Two forms changed, each matching the step that had been blank.

The estimate sheet in 'The Flooring Site Assessment SOP' now asks, for any recoat: what has been used to clean or shine the floor in the last several years, including any product called a restorer, polish or wax; and whether a recoat has been done before. Any yes to a polish, wax or oil-soap product moves the job to a test patch before it can be sold as a recoat.

And the crew's recoat sheet now has a required adhesion line from 'How to Screen and Recoat a Worn Floor Finish': the test area, the date, and the tape-test result, recorded before the full floor is coated. That is the test-patch-and-adhesion check NWFA guidance, in substance, calls for before recoating a floor whose maintenance history is unknown.

References

• Finish manufacturer's product data sheet: compatibility, recoat windows, ventilation and cure times.
• National Wood Flooring Association technical guidance on screen and recoat and finish adhesion, current edition.
• See related: How to Screen and Recoat a Worn Floor Finish; The Flooring Site Assessment SOP; Hardwood Floor Finishes for Refinishers.
• See related: Sanding Machines and Grit Sequences; The Lead-Safe Floor Work SOP.