ASTM C633

What the test does

A coating is thermally sprayed onto the flat end face of a cylindrical substrate fixture. That coated face is then bonded with an adhesive to the matching face of a second, uncoated loading fixture, and the completed assembly — substrate, coating, adhesive, loading fixture — is pulled apart in tension normal to the plane of the coating. The force at separation, divided by the coated area, is the reported strength. The coating is loaded in the one direction it is rarely loaded in service, straight off its substrate, because that is the direction in which a sprayed deposit is weakest and the only one in which a single number describes the bond.

What it measures, and why it matters

Adhesion or cohesion, and which of the two you have measured is decided by where the assembly separated, not by the operator. If the break runs along the coating-to-substrate interface, the figure is adhesion: how well the deposit is keyed to the prepared surface. If it runs within the deposit itself, the figure is cohesion: how well the sprayed particles have bonded to one another. Both are useful, and they point at different corrections — a low adhesion figure sends you back to grit blasting, preheat and stand-off, while a low cohesion figure sends you back to particle temperature and velocity. Because the number is a routine condition of approval for a coating and for the shop applying it, the failure mode is not a footnote to the result; it is half of it.

The coated assembly

Two cylinders, one coated on its end face, bonded together and pulled apart along their common axis.

Substrate fixture

Cylinder, coated on one flat end face

Loading fixture

Matching uncoated cylinder, bonded to the coated face

Diameter

Of the order of 25 mmPracticeVendor fixtures and laboratory descriptions put the standard specimen at about 25 mm; the exact figure is in the purchased text of the method.

Minimum coating thickness

Greater than 0.38 mm (0.015 in.)Below this the bonding adhesive wicks through the coating porosity to the substrate, and the test then measures the glue.

Coating processes covered

Combustion flame, plasma arc, two-wire arc, high-velocity oxygen fuel, detonation

Coating and substrate materials

Ceramic and metal coatings on metal or ceramic substrates

Test temperature

Normally ambientPracticeThe method does not fix a test temperature; it notes that very low temperatures may suit fundamental investigation.

Bonding agent

A high-strength adhesive, chosen to out-perform the coatingPracticeIt is a ceiling on the whole method. If the glue lets go first, you have a lower bound, not a result.

Bonding-agent control

A set of uncoated fixtures bonded alongside the coated onesBonded at essentially the same time, so the adhesive's own strength is measured on the same batch and the ceiling is a known number rather than an assumption.

The bonded area is measured, not assumed. Every reported strength is a load divided by that area, so an over-run of adhesive beyond the coated face quietly inflates the denominator.

Rate of loading

Rate

Constant crosshead travel between 0.013 and 0.021 mm/s (0.030 to 0.050 in./min)Applied as an increasing tensile load until rupture. The assembly is drawn steadily apart rather than snapped.

Machine requirement

To Practices E4, with permissible variation under 1 %

Load path

Self-aligning devices that permit no eccentric load or bending momentThis is a clause of the method, not a refinement of it.

Recorded quantity

Maximum load applied before ruptureThere is no curve to interpret — the assembly holds until it does not.

Calculations

How the test runs

1. 01Prepare the substrate fixture face — clean and roughen as the coating process requires.
2. 02Apply the thermal spray coating to that face, thicker than the method's minimum.
3. 03Prepare and clean the mating face of the uncoated loading fixture.
4. 04Bond the two together with the chosen adhesive and cure it to its schedule, holding the fixtures parallel and aligned throughout.
5. 05Bond a set of uncoated fixtures at the same time, to measure the bonding agent's own strength.
6. 06Measure the bonded area.
7. 07Mount the assembly through self-aligning couplings at both ends.
8. 08Pull in tension normal to the coating plane, at 0.013 to 0.021 mm/s of crosshead travel, to rupture.
9. 09Record the maximum load applied.
10. 10Examine both faces and classify the failure — at the coating-to-substrate interface, within the coating, in the bonding agent, or a combination; on a multilayer coating, note any internal adhesive failure between layers.
11. 11Divide peak force by the measured area, and report the failure mode with the number.

A specimen that fails in the bonding agent has not measured the coating. Report it as a lower bound and repeat with a stronger adhesive.

What the report has to contain

• Reference to ASTM C633 and the edition
• Coating material or designation, the spray technique and the equipment used
• Spray parameters
• Final coating thickness, and whether the surface is finished or as-sprayed
• Substrate material
• Description of the surface preparation of the substrate
• Bonding agent, and the bonding procedure where it departs from the maker's instructions
• Number of specimens sprayed and number tested
• Adhesion or cohesion strength of each specimen
• Average, maximum and minimum strength
• DESCRIPTION OF THE FAILURE — at the coating-to-substrate interface, in the coating, in the bonding agent, or a combination, plus any internal adhesive failure on a multilayer coating

What the machine must be capable of

Modest capacity, excellent alignment. On a 25 mm face the bonded area is around 500 mm², so a well-bonded coating carrying tens of megapascals fails somewhere in the tens of kilonewtons, and a frame in the 50 kN class covers the work with headroom. Poorly bonded or deliberately weak coatings fail at a small fraction of that, so force resolution at the bottom of the range matters as much as capacity.

Alignment is not a refinement here, it is a clause: the apparatus must apply the load through self-aligning devices that do not permit an eccentric load or a bending moment on the specimen. The assembly is short, stiff and brittle, and any eccentricity in the load path puts bending across the coating plane, which peels one edge first and depresses the result. Self-aligning couplings at both ends, rather than fixed threaded adapters, are what keep the pull axial. The frame is verified to Practices E4, and the load is applied at a constant rate of crosshead travel between 0.013 and 0.021 mm/s (0.030 to 0.050 in./min) until rupture — the assembly is drawn steadily apart, not snapped.

What goes wrong in practice

Failure of the bonding adhesive instead of the coating, which yields a lower bound rather than a result and must be reported as such. Adhesive penetration through a coating applied below the thickness limit. Misalignment, which reads as an unexplained spread between nominally identical specimens. Contamination of the coated face before bonding. And reporting a mean strength with no statement of where each specimen broke, which leaves the number uninterpretable.

Pulling normal to a bonded plane — four methods

All four are limited by whatever glue holds the fixture on. In every case a failure in that glue is a lower bound on the property, not a measurement of it.

Questions we are asked about this test