How to Prepare a Surface So a Bond Holds

Why this matters

Surface prep is where bonded and coated work is won or lost, and it is the step that gets compressed when the job is running late. The damage does not show up that day. It shows up at week three as a bracket on the floor or a coating peeling in sheets, and by then nobody can tell whether the product was wrong or the prep was. What follows is a sequence, and the sequence itself is the content: every step is positioned so the step before it cannot undo it. Move two of them and you get a joint that looks identical, passes every check you can do on site, and fails anyway.

Before any abrasive touches a coated surface

Two hazards live in this work and both are in the dust, not the adhesive.

• If the coating is on a pre-1978 residential structure or a child-occupied facility, assume lead until tested. Renovation, repair and painting that disturbs painted surfaces in those buildings is covered by EPA's Renovation, Repair and Painting rule at 40 CFR Part 745, Subpart E, which requires a certified firm, a certified renovator on site, containment of the work area, and specified cleanup and verification. Sanding first and asking later is the violation. A recognized lead test kit or a certified inspector settles it before the sander comes out.
• If you are grinding or cutting concrete, masonry, tile, mortar or engineered stone, that dust is respirable crystalline silica. Use the tool's integrated water feed or its dust collection with a HEPA-filtered vacuum, and a respirator selected and fit tested under 29 CFR 1910.134. The silica standard itself has two homes and a field-service shop can fall under either: 29 CFR 1910.1053 in general industry, 29 CFR 1926.1153 in construction. Dry-cutting masonry indoors with no control is the single worst version of this step.

Solvent wiping brings the third: vapors need ventilation and no ignition source in the room, and a standing pilot on a nearby water heater or furnace counts as an ignition source. Shut it down or move the work.

Step 1: Name the substrate before you decide anything

You cannot select a prep method for a material you have not identified. Bare steel, galvanized steel, aluminum, stainless, a powder-coated panel, a painted panel, ABS and polypropylene all take different treatment, and three of those look similar under shop lighting.

The cheap discriminators: a magnet separates steel and most cast iron from aluminum, stainless of the common austenitic grades, brass and copper. A drop of water tells you high energy from low energy (a continuous film versus beads). On plastics, a molded recycling or resin code inside the part is usually the fastest honest answer, and if there is none, treat an unmarked flexible white or translucent plastic as a polyolefin until proven otherwise, because that is the assumption that fails safe.

What you lose by skipping it: you prep for the wrong material. Abrading anodized aluminum through to bare metal changes the surface you are bonding to. Solvent-wiping polycarbonate or acrylic with the wrong solvent crazes it, and the crazing is a network of stress cracks that the bond will later follow.

Step 2: Degrease before you abrade, not after

Every surface arrives with something on it: mold release on new plastic, drawing lubricant on new metal, hand oils, silicone from a previous repair, dust bound in a film of moisture. Remove it with a solvent or cleaner the substrate tolerates, applied to a clean cloth rather than poured on the part, wiping in one direction and turning to a fresh face of the cloth on every pass. Wiping back and forth with one face redistributes the contamination you just picked up.

This is the step the SSPC-SP 1 solvent-cleaning standard describes, and it comes first for a physical reason. Abrading a contaminated surface drives the contamination into the profile you are creating. After that, no amount of later cleaning gets it out, because you have just built it a set of pockets to hide in. Silicone is the worst offender: a surface that once had a silicone product on it will repel adhesive and coating for a long time, and the visible film is not the problem, the migrated residue is.

What you lose by skipping it or moving it later: the bond fails adhesively at a clean-looking substrate face, and every subsequent repair on that spot fails the same way.

Step 3: Abrade to create profile and fresh surface

Abrasion does two things: it removes weak boundary layers such as oxide, scale, and chalked coating, and it multiplies the real contact area so the adhesive has something to key into.

• On steel and aluminum for structural bonding, a general working range is roughly 80 to 120 grit. Finer than that leaves too little profile, coarser leaves peaks that the adhesive cannot fill.
• On plastics, stay finer, in the region of 180 to 220 grit, and use light pressure. Heat from aggressive sanding smears thermoplastics and creates a soft skin that comes off with the adhesive.
• Uniform dulling of the whole bond area is the target. A shiny patch inside a sanded area is a spot the abrasive never reached, and that spot is a starting point for a peel.

Where a coating system specifies an anchor profile, that is a measured value in the coating manufacturer's data sheet, expressed as a depth, and it is matched to a named blast or power-tool standard rather than to a hand-sanding grit. Read theirs; do not carry a number over from another product.

What you lose by skipping it: the adhesive bonds to the oxide layer or the old coating rather than to the part, and the joint fails at whatever that layer's own strength was.

Step 4: Remove the abrasion debris with a second clean wipe

Sanding leaves loose particles and, on metal, a fine dust that is partly the substrate. Bonding over it means bonding to loose powder. Vacuum first if the debris is heavy, then repeat the one-direction solvent wipe with fresh cloth faces, and let the solvent flash off completely before anything else touches the surface. A solvent still evaporating out of the profile will blow bubbles into the bondline or the coating.

Do not blow the surface off with shop air unless the line has a working oil and water separator you have verified. Compressed air off a lubricated compressor is a very efficient way to deposit a thin film of oil on a surface you just spent ten minutes cleaning.

Do not touch the prepared area with bare hands after this point. Skin oil is a contaminant at exactly the scale that matters.

Step 5: Treat what abrasion cannot fix

Two cases need chemistry, not mechanics.

Low-surface-energy plastics (polyethylene, polypropylene, acetal, and anything that beads water after cleaning) do not become bondable through sanding. They need flame treatment, plasma or corona treatment, or a chemical adhesion promoter sold for that resin family. Sanding alone leaves you a rough surface that still will not wet.

Inactive metals for anaerobic chemistries (stainless, plated, anodized, and most non-ferrous) cure slowly or incompletely without the primer the adhesive manufacturer specifies. This is a cure problem, not a cleanliness problem, and it survives perfect prep.

Step 6: Prove the surface with a water break test

Flick or mist clean water across the prepared area and watch it for about 30 seconds. A continuous unbroken film means a clean, high-energy surface. Beading, or a film that pulls back into islands, means residual contamination or a substrate whose energy is too low for the adhesive to wet, and you return to step 2 or step 5 depending on which it is. This test is standardized as a method (ASTM F22), which is worth knowing because it means the result is repeatable rather than a matter of opinion.

The test only reads valid on a surface that is otherwise dry. Dry it fully before proceeding, since trapped water is its own bondline defect.

Step 7: Bond inside the open window

A prepared surface starts degrading immediately. Freshly abraded aluminum grows a new oxide film in minutes. Flame and plasma treatment decay over hours to days depending on the resin. Any exposed surface collects airborne contamination continuously.

Follow the adhesive or promoter manufacturer's stated open time. Where none is stated, treat a freshly prepared metal surface as good for minutes rather than hours, and prep in the order you will bond rather than prepping everything and then starting to glue.

What it looks like when the order is violated

A shop bonds eight identical mounting pads, same substrate, same adhesive, same day. One tech works the run in the order above. The other, working faster, sands first and then wipes once with solvent at the end, on three of the eight, reasoning that one thorough wipe after sanding covers both jobs.

At the ninety-day mark, three of the eight pads are off. They are the same three, all sanded before degreasing. That is 3 of 8, 37.5 percent of the run, and 3 of 3, the whole of the reversed subgroup. The break faces are the tell: all three show adhesive on one side only and clean substrate on the other, which is adhesive failure at the interface, not the adhesive tearing. If the adhesive itself had been the weak point, the material would be split down the middle with residue on both faces.

The labor math, in unbilled hours against unbilled hours: the correct sequence adds one extra wipe per pad, call it 0.1 hour across the run of eight. The three recovery visits ran about 1.0 hour each including drive, so 3.0 hours. That is a 30-to-1 ratio, and both figures are the same currency, absorbed shop time on work already paid for. It would take an implausibly slow prep step to change that answer.

One caution on reading this result: those three pads were also the ones done fastest overall, so the comparison is not perfectly clean between prep order alone and general haste. The break-face evidence is what makes the call, not the count.

Verifying prep on work you did not watch

You often inherit a surface someone else prepared, or you need to confirm your own before a coating goes over it and hides everything.

• Tape-pull a coated surface. Press a strip of firm adhesive tape onto the cured coating, smooth it, and pull it back sharply at a steep angle. Coating on the tape means the bond to the substrate is weaker than the tape's bond to the coating, which is a failing result whatever the appearance.
• Look at the edges of the prepared zone. A prep boundary that is a sharp line means someone masked and sanded to it. A feathered boundary means they worked outward and probably covered the whole bond area. A prepared zone smaller than the bond footprint is a guaranteed edge failure.
• Check for a witness on the scrap. If the tech kept an offcut prepped alongside the part, water-break it now. Same surface, same treatment, and it is destructive to test without harming the job.
• Ask what solvent was used and on what plastic. Crazing from an incompatible solvent can take days to become visible and will not appear in a photo taken on the day.

References

• 40 CFR Part 745, Subpart E (EPA Renovation, Repair and Painting rule), for the certification, containment and cleanup requirements that apply before disturbing paint in pre-1978 target housing and child-occupied facilities
• 29 CFR 1910.1053 (general industry) and 29 CFR 1926.1153 (construction), respirable crystalline silica, and 29 CFR 1910.134 for respirator selection and fit testing
• SSPC-SP 1, Solvent Cleaning, now published by AMPP, for the solvent-cleaning method referenced in step 2
• ASTM F22, water-break test method, for the surface-cleanliness check in step 6
• See related: What an Adhesive Needs in Order to Bond; Common Construction Adhesives Reference