Adhesion Test Failure Modes

A coating system applied to a substrate consists of multiple layers and interfaces. In a typical protective coating system on steel, the stack includes the substrate surface, a surface profile created by blast cleaning, a primer layer, one or more intermediate coats, and a topcoat. Each layer has internal (cohesive) strength, and each interface between layers has a bond (adhesive) strength. The weakest of these determines where failure occurs during a pull-off test.

Failure modes are classified into four principal categories:

Adhesive failure occurs at the interface between two layers — for example, between the primer and the substrate, or between the primer and the intermediate coat. The fracture surface shows clean separation with one material on each side of the break. This type of failure directly indicates the bond strength at that interface.

Cohesive failure occurs within the body of a single layer. The fracture passes through the material rather than along an interface, leaving the same material on both the dolly and the surface. This indicates that the layer’s internal strength is lower than the bond strength at adjacent interfaces.

Substrate failure occurs when the substrate material itself fractures. This is most common on concrete, where the near-surface concrete may be weaker than the coating bond. On steel substrates, true substrate failure is rare; apparent substrate failure usually indicates contamination or degradation at the steel surface.

Glue failure occurs at the adhesive used to bond the dolly to the coating. This means the test has not stressed the coating system to its limit. Glue failures are generally considered invalid because they do not provide information about the coating’s adhesion.

After each pull-off test, the operator examines both the dolly face and the corresponding area on the surface. The fracture surface is compared to the layer structure of the coating system to determine which interfaces or layers are visible. Colour, texture, and layer identification are used to classify the failure.

Most real failures are mixed. A single dolly face may show areas of adhesive failure at one interface, cohesive failure within a layer, and possibly glue failure at the perimeter. Standards require that the operator estimate the percentage of each failure type across the dolly face (typically to the nearest 5% or 10%) and record these alongside the pull-off stress.

A consistent pattern of failure at the same interface across multiple tests indicates a systematic weakness at that location in the coating system. A random distribution of failure modes across tests suggests that the system is near a balance point where multiple planes have similar strength, or that local preparation variability is shifting the weakest plane between tests.

Interpreting failure mode in relation to the pull-off value is essential. A high pull-off stress with cohesive failure in the topcoat means the topcoat’s internal strength was the limiting factor and the interface bonds are stronger than the measured value. A low pull-off stress with adhesive failure at the primer–substrate interface identifies a specific bonding deficiency. A low value with glue failure means the test did not reach the coating system’s capacity and should be repeated with improved preparation.

Where substrate failure dominates (common on concrete), the pull-off value represents the tensile strength of the substrate, not the coating adhesion. In such cases, the coating bond is stronger than the substrate, which may be acceptable or may indicate that the substrate itself needs investigation.

• Material-dependent effects: Multi-layer coating systems have more potential failure planes. Thick coatings may develop internal stresses during curing that weaken specific layers or interfaces. Some coating chemistries produce layers with relatively low cohesive strength compared to their adhesive performance, making cohesive failure the dominant mode. On porous substrates, adhesive can penetrate into the coating layers during dolly bonding, potentially reinforcing the area and masking the true weakest plane.
• Environmental effects: Moisture ingress over time can degrade specific interfaces preferentially, changing the failure mode from cohesive (within a layer) to adhesive (at a now-weakened interface). Temperature affects the ductility of coating layers and can shift failure from one mode to another: cold conditions may embrittle a layer, promoting cohesive fracture, while warm conditions may soften it, promoting interfacial failure under the same stress.
• Instrument/technique-specific effects: Pull rate influences failure mode in some systems. A fast pull may not allow time for stress redistribution, concentrating failure at the weakest point; a slow pull may allow creep in a soft layer, changing where failure initiates. Scoring depth can affect the result: a score that does not fully penetrate to the substrate leaves coating bridges that carry load, while a score that goes too deep into the substrate can initiate substrate failure.
• Operator or setup issues: The most common operator error affecting failure mode classification is misidentification of layers. If the operator does not know the coating system structure (number of coats, colours, primer type), the failure plane can be misassigned. Photography of the fracture surface immediately after the pull provides a record that can be reviewed later. Inadequate lighting or surface contamination (dust, fingerprints on the fracture surface) can obscure the failure pattern.

Failure mode analysis matters in every pull-off adhesion test, but it is particularly critical in acceptance testing for protective coating systems, where the specification may require failure to occur at or above a minimum stress and not at a specific interface. For example, a specification might require ≥5 MPa with no adhesive failure at the primer–substrate interface. A result of 6 MPa with failure at that interface would fail the acceptance criterion even though the numerical value exceeds the threshold.

It is often assumed that partial glue failure on a dolly face can be disregarded when classifying the result. If glue failure is present on any portion of the dolly face, the measured pull-off value may be limited by the glue bond and may not represent the coating system’s true performance. Standards generally require glue failures to be repeated.

Equally problematic is the failure to distinguish between adhesive failure at different interfaces. “Adhesive failure” between the topcoat and the intermediate coat has different implications from “adhesive failure” between the primer and the substrate. Recording “adhesive failure” without specifying the interface is insufficient.

On concrete substrates, it is not uncommon for substrate failure to be interpreted as evidence of excellent coating adhesion. While it does indicate that the coating bond exceeds the substrate’s tensile strength, it can also indicate that the concrete surface is weak (e.g., due to laitance, carbonation, or poor curing), which may itself be a concern.

• Pull-off Adhesion Test Principles
• Surface Preparation and Dolly Bonding
• Adhesion Testing Standards and Reporting

If failure-mode interpretation is leading to questions about whether the whole test setup is robust enough, this guide is the most relevant next step.

• Adhesion Test Setup and Validity if you need to judge whether the wider pull-off workflow is controlled tightly enough for reliable interpretation.