A corona roller is one of the few thermal spray applications where the coating is judged by an electrical test rather than by how it looks. A roll can leave the grinder with a perfect surface finish, measure full thickness everywhere, and still fail at 8 kV on the hipot set. Understanding why is the difference between a coating that runs for years and one that is reworked twice a year.
These are the failure modes we see most often on rollers returned from the field, and the controls that prevent each one.
Failure mode 1: dielectric breakdown through porosity
A plasma sprayed alumina layer is built from flattened droplets. Between them are sub-micron gaps that connect, in places, from the bond coat to the surface. Under a high-frequency corona field, air in those channels ionises and the channel becomes a conducting path. The result is a pinhole puncture, usually invisible from outside, surrounded by a burned crater.
Prevention: seal the coating. A silicone or epoxy sealer drawn into the porosity by vacuum or warm dip closes the through-thickness paths. Thickness alone does not solve this - a 2 mm porous coating breaks down before a well-sealed 0.8 mm one.
Failure mode 2: edge and end flashover
At the roller ends, the spray angle becomes oblique and the coating thins, sometimes to half the nominal thickness over the last 20 mm. The electric field concentrates at the roller edge, and the thin zone flashes over. This is the most common cause of a roller that tests good in the middle and fails at the ends.
Prevention: chamfer or radius the substrate edge before spraying, program the gun to over-travel past the end face, and map thickness at 50 mm intervals from the end rather than only in the centre.
Failure mode 3: mud-cracking from thermal mismatch
Alumina has a thermal expansion coefficient of about 8 × 10-6/K; steel is around 11.5 × 10-6/K. On cooling from spray temperature the ceramic is left in tension, and once the tensile stress exceeds the coating strength it relieves itself as a network of cracks - the "mud-crack" pattern that is normal in thin layers but dangerous in thick ones, because the cracks can run through the full thickness.
Prevention: keep the coating within its design window (0.5-1.8 mm for our standard corona build, 2.0 mm maximum), use a NiAl or NiCrAl bond coat, apply a graded ceramic-metal transition, and control preheat and cooling rate instead of quenching the roll after spraying.
Failure mode 4: contamination and grit embedment
Alumina grit embedded in the substrate during blasting, or oil residue in a porous cast iron roll, volatilises in the plasma plume. The gas expands under the coating and produces a blister or a localised unbonded area. It often survives grinding and fails weeks later in service.
Prevention: degrease and outgas the part before blasting, use clean angular grit, blast at a controlled pressure, and blow off with dry oil-free air immediately before spraying. On castings, preheat to drive out trapped oil.
Failure mode 5: grinding damage
Diamond grinding generates heat. Too aggressive an infeed burns the surface, and the burnt layer contains microcracks that become breakdown paths under voltage. The coating then fails while the thickness and finish still look acceptable.
Prevention: low infeed with generous coolant, a dressed diamond wheel, and a final polish pass to Ra 0.20. A dyed penetrant check after grinding catches the damage before the roll ships.
Failure mode 6: sealer degradation in ozone
Corona discharge produces ozone continuously. Some epoxy sealers embrittle and craze within a year, reopening the porosity that was sealed at dispatch. The roller then fails progressively, starting at the highest-field region.
Prevention: specify an ozone-resistant silicone sealer, and include re-sealing in the maintenance schedule - typically every 12-24 months depending on duty.
Inspection checklist we run before shipment
| Check | Method | Acceptance |
|---|---|---|
| Coating thickness | Eddy current or magnetic, 12 points along the face plus both ends | Within drawing tolerance, ends not below 80% of nominal |
| Dielectric strength | AC hipot, 50 Hz or high frequency per customer spec | No breakdown at 20 kV, or at the customer-specified test level |
| Surface cracks | Dye penetrant after grinding | No through-thickness indications |
| Bond strength | ASTM C633 on a witness coupon sprayed in the same batch | Above 65 MPa |
| Roughness | Profilometer | Ra 0.20 standard |
| Runout | Dial indicator on centres | Per drawing, typically under 0.02 mm |
Frequently asked questions
How long should a ceramic corona roller coating last?
In two-shift film treatment duty, five to eight years is typical when the roller is correctly sealed and re-sealed on schedule. Failures inside two years almost always trace back to porosity, edge thinning or grinding damage.
Can a failed roller be repaired instead of replaced?
Yes. The coating is stripped, the substrate inspected and re-ground, and the roll recoated. A steel roller body can normally be recoated several times as long as the wall thickness and balance remain within specification.
Does a thicker coating give higher breakdown voltage?
Only if it is sealed. Thickness raises the intrinsic breakdown voltage, but an unsealed 2 mm coating can fail below a sealed 0.8 mm one because the limiting path is the connected porosity, not the bulk material.
What test voltage should be specified?
Match the test to the working voltage with a safety margin - commonly 20 kV for rollers running at 10-20 kV. We test to the customer's specification and report the result per roller.
