Powder Coating Peeling and Adhesion Failure: How to Diagnose It
Peeling is the defect that ends supplier relationships. A coating that fades or orange-peels is a complaint; a coating that lifts off in sheets is a rejected consignment. It is also the defect most often blamed on the wrong party, because the visible failure appears in the coating while the cause almost always sits upstream of it.
This guide covers adhesion failure specifically: peeling, flaking, delamination and poor cross-hatch results. For the full defect map see the complete diagnostic guide.
What you are looking at
Adhesion failure has a distinctive signature. The coating separates as a continuous film rather than crumbling, and it usually lifts from an edge, a hole, a weld or an impact point rather than from the middle of a flat panel. Peel a piece away and the back of it is smooth and glossy, carrying a clean impression of the metal it came off.
The single most useful diagnostic is to look at the metal underneath, not at the coating.
| What the exposed metal looks like | What it tells you |
|---|---|
| Bright, clean, unoxidised | The coating never bonded. Pre-treatment failed, or the surface was contaminated after rinsing. |
| Rusted, with corrosion spreading sideways under the intact film | Corrosion undercutting. The bond was adequate initially and moisture reached the interface through a breach or through the film itself. |
| Grey, powdery residue on the back of the peeled film | An exhausted or poorly rinsed phosphate bath. The conversion layer formed but was never properly bonded to the substrate. |
| Oily sheen or fingerprints visible | Handling contamination between pre-treatment and coating. |
Measuring it properly
Adhesion is not a matter of opinion. ASTM D3359 defines a cross-hatch test: cut a lattice through the coating to the substrate, apply and remove a specified tape, and classify the result from 5B (no removal) to 0B (more than 65 per cent removed). ISO 2409 uses an inverted scale where GT-0 is the best result.
Oracle Industries tests every batch to this method and reports 5B / GT-0 across all four chemistries. If your incoming material tests below 4B on a properly prepared panel, that is a material problem. If it tests 5B on a lab panel and fails on your production parts, the difference is in your line, not in the powder.
That distinction is worth establishing before any dispute, which is why keeping a coated test panel from each batch is worth the two minutes it costs.
The four causes, in order of frequency
1. Pre-treatment failure
This accounts for the clear majority of adhesion failures. Powder coating has no solvent and no mechanical key of its own; it relies entirely on a chemically converted surface to bond to. If the conversion coating did not form, nothing else in the process can compensate.
The usual specific causes are a degreasing bath run past its titration limit, a phosphate bath below temperature or concentration, insufficient dwell time in any stage, or a final rinse using hard or recycled water that leaves salts behind. Check bath concentrations and final rinse conductivity before you look anywhere else. The pre-treatment guide covers each stage and what it does.
2. Contamination after pre-treatment
A correctly pre-treated part can be ruined between the rinse and the booth. Bare hands, oily conveyor hooks, airborne mist from an adjacent machining operation, or a compressed air line carrying oil will all deposit a film that the powder cannot bond through.
The fix is procedural rather than chemical: gloves after the rinse, a covered transfer path, and a properly maintained oil and moisture separator on the compressor. Oil in the air line is a common cause and often goes unsuspected for months because it produces both fisheyes and adhesion failure and looks like two unrelated problems.
3. Under-cure
An under-cured film has not fully crosslinked, so it has neither its designed mechanical properties nor its designed adhesion. It often looks correct. The diagnostic is a solvent rub: a cotton swab soaked in methyl ethyl ketone, rubbed with firm pressure across the surface. A properly cured thermoset resists roughly fifty double rubs. An under-cured one softens, dulls or transfers colour to the swab within ten.
Cure is counted from when the metal reaches temperature, not from when the part enters the oven. A thick casting entering a 190 degree oven may not reach 180 degrees on the metal for eight or ten minutes, and if the schedule allows fifteen minutes total, that part received five minutes of actual cure. Log a thermal profile on the heaviest part in the load. Oracle specifies 180 to 200 degrees Celsius metal temperature held for 10 to 15 minutes across all four chemistries.
4. Excessive film thickness
An over-thick film shrinks more as it cures and carries more internal stress. On sharp edges and complex geometry that stress can exceed the adhesive bond. If your dry film thickness is running above 100 microns where the specification is 70 to 80, reduce it before changing anything else.
Edges peel first, and that is a design problem
Molten powder pulls back from a sharp edge through surface tension, so an edge always ends up the thinnest point on a part. It is also where handling damage concentrates and where corrosion starts. Breaking sharp edges to a small radius before coating does more for edge performance than any change to the powder, and it costs nothing once it is in the fabrication drawing.
This is the single most under-used improvement available to fabricators. If your rejects cluster on edges and corners, the answer is in the press shop.
A working sequence
- Keep a failed part. Photograph the peel and the exposed metal before anything is cleaned.
- Look at the metal underneath and classify it using the table above.
- Cross-hatch test a retained panel from the same batch. If it passes at 5B, the material is sound.
- Solvent-rub the failed part to rule cure in or out.
- Titrate the pre-treatment baths and measure final rinse conductivity.
- Check dry film thickness at several points, including an edge.
- Only then change a process parameter, and change one at a time.
Working in this order takes an hour. Changing gun settings first and pre-treatment last, which is the common instinct, typically costs a shift.
Where Oracle can help
If you are seeing adhesion failure on Oracle material, send photographs of the peel and the exposed substrate along with your pre-treatment schedule, film thickness readings and oven profile. Most cases resolve on the pre-treatment data alone. Get in touch and we will work through it with you, including when the answer is that the material is not right for the application.