Adhesive Failure Mode Chart: What Each Fracture Points To

To decide which pattern a failed joint shows, check the fracture pattern, both surfaces and cure records against ISO 10365:2022 failure pattern designations. The chart below maps each pattern to candidate causes and the next check that separates them; it selects tests to run and does not, by itself, confirm a cause.

• ISO 10365:2022 (edition 2) designates the main failure patterns of bonded assemblies and applies to all mechanical tests on them (ISO, 2022).

• The 2022 edition added stress whitening failure (SWCF) and debonding due to bondline corrosion (COR) (ISO 10365:2022, foreword).

• ASTM D5573-99(2019) classifies failure modes in bonded fiber-reinforced-plastic joints by visual observation without magnification (ASTM, 2019).

• Cohesive failure indicates adequate bonding, with adhesive strength and joint design as the limit (TWI, accessed 2026).

This adhesive failure mode chart covers structural and semi-structural bonded joints that failed in a mechanical test or in service, with both halves available. It does not cover pressure-sensitive tape peel grading, sealant joints judged by their own standards, or cases where only one half of the joint was kept.

When do you use an adhesive failure mode chart?

An adhesive failure mode chart is used when a bonded joint has failed in a test or in service and you need to decide which causes to investigate first. Quality engineers, adhesive application engineers and technical service staff use it after a lap shear or peel test gives low or scattered values, after a field return, or when a supplier and a customer disagree about whether the adhesive or the surface is at fault.

The chart reads the fracture surface, which is the first evidence available and the cheapest to record. ISO 10365:2022 "specifies the designations for the main types of failure pattern of bonded assemblies and illustrates, through diagrams, their respective appearances", and it "applies to all mechanical tests performed on a bonded assembly, regardless of the nature of the adherends and adhesive" (ISO 10365:2022, clause 1).

The chart does not replace analysis. A failure pattern narrows the candidate causes to a short list, and each item on that list is a hypothesis to test against records and measurements. ASTM D5573-99(2019) makes the same distinction for FRP joints: it "does not address the acceptability of any specific failure mode" (ASTM, 2019). Use the chart early in a complaint, before cure records, surface preparation logs and retained samples are pulled, so that you know which of those records to pull first. The editable Word decision chart is at the end of this page.

What do you need to read an adhesive failure mode?

To read an adhesive failure mode you need both halves of the failed joint, the test or service conditions, and the records of how the joint was made. Without both halves, a bare-looking surface cannot be read.

Collect these inputs before you classify the joint:

1. Both fracture surfaces, kept clean and dry, photographed under the same light with a scale.

2. The test method and conditions, or the service history: load type, temperature, humidity, time in service and any exposure to water, salt or chemicals.

3. The failure load and the expected value for the same test from the qualification report or the technical data sheet.

4. Adhesive lot numbers, mix ratio records for two-part systems, cure time and temperature.

5. Surface preparation records: method, time between preparation and bonding, and the procedure revision.

6. Bond-line thickness as designed and as measured on the failed joint.

7. A retained good joint or coupon from the same period, if one exists.

ASTM D5573-99(2019) classifies each mode "based solely on a visual observation of the failure surface without the aid of a microscope or other means to magnify the surface" (ASTM, 2019). Start with that naked-eye reading, then use a hand lens or microscope where the chart says the naked eye can mislead, such as a surface that looks bare but carries a thin adhesive film.

Record the area share of each pattern as a percentage of the bonded area, as the mixed-failure figures in ASTM D5573-99(2019) do (ASTM, 2019).

Adhesive failure mode chart: fracture pattern, candidate causes and next checks

The adhesive failure mode chart works as a sequence of numbered decision steps followed by one row per pattern. Run the steps on each half of the failed joint:

1. Is adhesive present on both fracture surfaces? If yes, go to step 4. If one surface looks bare, go to step 2.

2. Under magnification, does the bare-looking surface carry a thin, continuous adhesive film? If yes, record thin-layer cohesive failure. Then go to step 5. If no, go to step 3.

3. Are there corrosion products at the interface, spreading from the edges? If yes, record debonding due to bondline corrosion (COR). If no, record adhesive (interfacial) failure (AF). Then go to step 5.

4. Is the adhesive whitened along the fracture? If yes, record failure with stress whitening (SWCF). If no, record cohesive failure (CF). Then go to step 5.

5. Is the adherend itself torn, delaminated or broken? If yes, record substrate failure (SF) for that area. Either way, go to step 6.

6. If more than one pattern appears, record mixed failure with an approximate percentage after each designation. Then go to step 7.

7. Compare the failure load with the expected value for the same test, then take the next check from the pattern's row.

The table below decides which cause hypotheses to test first; the candidate causes are hypotheses, not conclusions.

The key row is adhesive (interfacial) failure: a clean adherend on one side points first to the surface and its preparation, not to the adhesive formulation.

What are the main adhesive failure modes in ISO 10365?

ISO 10365:2022 designates nine main failure patterns in its Table 1: substrate failure (SF), cohesive substrate failure (CSF), delamination failure (DF), cohesion failure (CF), special cohesion failure (SCF), failure with stress whitening of adhesive (SWCF), adhesion failure (AF), adhesion and cohesion failure with peel (ACFP) and debonding due to bondline corrosion (COR) (ISO 10365:2022, Table 1). ISO 10365:2022 is the second edition, published in January 2022, and it cancels and replaces ISO 10365:1992 (ISO, 2022).

The foreword of ISO 10365:2022 lists the main changes from the first edition: "failure with stress whitening of adhesive (SWCF) has been added" and "debonding due to bondline corrosion (COR) has been added" (ISO 10365:2022, foreword). Where more than one pattern occurs, "an approximate percentage for each type of failure pattern shall be given after each designation", and an oscillating failure is marked with "OSC", as in AF (50 %) + CF (50 %), OSC (ISO 10365:2022, clause 4). The standard shows each appearance as a diagram, so the chart describes appearances in words from NPL and ASTM D5573.

The definitions behind the main patterns are published elsewhere. The NPL Good Practice Guide No. 28 defines adhesive failure as "Failure of an adhesive bond, such that separation appears to be at the adhesive/adherend interface", cohesive failure as failure "within the body of the adhesive", and adherend failure as "Failure of a joint in the body of the adherend" (NPL, 2023). For bonded fiber-reinforced plastics, ASTM D5573-99(2019) uses finer classes: adhesive, cohesive, thin-layer cohesive, fiber-tear, light-fiber-tear, stock-break and mixed failure (ASTM, 2019). The fiber-tear and stock-break classes are forms of substrate failure in a composite adherend.

What does adhesive (interfacial) failure point to?

Adhesive (interfacial) failure points first to the surface: contamination, low surface energy, surface preparation that departed from the procedure, or water that reached the interface after bonding. The bond to that surface was weaker than the adhesive itself.

Contamination is the first hypothesis to test. The European Aluminium Automotive Manual lists contaminants including "lubricants, oxides, dust, and dirt" and states that surface contamination "will reduce bond strength by interfering with the ability of the adhesive to form a proper bond" (European Aluminium, 2015). The NPL guide adds that after surface preparation the adherends "must not be exposed to physical handling or uncontrolled atmospheric environments to prevent surface contamination prior to bonding" (NPL, 2023).

Low surface energy is the second hypothesis on plastics. The same manual notes that "polymers and other organic materials have low surface energy while metals and ceramics have high surface energy" (European Aluminium, 2015), so an untreated or under-treated plastic surface may not wet. Surface preparation guidance for metals and plastics is set out in ISO 17212:2012 and, for wrought metals, in ASTM D2651-01(2024).

For a joint that failed after service, test water at the interface: water "may also displace the adhesive at the interface and cause true interfacial failure" (European Aluminium, 2015). Surface analysis of the bare side, such as XPS, which NPL describes as surface-sensitive and used on fracture surfaces (NPL, 2023), can show what sits on it.

What does cohesive failure in the adhesive point to?

Cohesive failure in the adhesive points first to the adhesive and the joint design, not to the surface, because the adhesive layer itself gave way. TWI states that "Cohesive failure indicates that bonding between the adhesive and the substrate is adequate and the limiting factor is the strength of the adhesive and the design of the joint" (TWI, accessed 2026).

Cohesive failure at the expected load is the intended result on many metal joints. The European Aluminium manual reasons that because adhesive properties are lower than those of aluminium, "it is reasonable to assume that an adhesive bond on aluminium can (and should) fail internally in the adhesive itself" (European Aluminium, 2015). Compare the failure load with the expected value before treating cohesive failure as a defect.

Cohesive failure at a low load points to the adhesive's condition in the joint. Test these hypotheses in order: under-cure (cure time and temperature records, then degree of cure; NPL notes that differential scanning calorimetry "can be used to measure Tg and degree of cure" (NPL, 2023)); wrong mix ratio in a two-part system (dispense records and metering checks); bond-line thickness outside the adhesive's range; and voids in the bond line. A vendor troubleshooting guide lists cure state, mix ratio and bond-line thickness as the first places to look after cohesive failure (Incure, 2026).

For joints aged in humid service, also test water uptake, which may lower the glass transition temperature of the adhesive and its load-bearing capacity (European Aluminium, 2015).

How do you read substrate and mixed adhesive failure modes?

Substrate failure means the adherend broke before the bond did, and mixed failure means two or more patterns share the bonded area; read substrate failure as a statement about the adherend and mixed failure as a map of local variation. Neither one points to a single cause.

Substrate failure in a metal joint usually means the adherend yielded or broke. The NPL guide notes that in a single-lap joint "the overlap length L should be sufficient to ensure adhesive failure occurs before the adherend yields" (NPL, 2023), so substrate failure in a test can point to test geometry or adherend thickness rather than to the bond. In fiber-reinforced plastics, ASTM D5573-99(2019) separates fiber-tear, light-fiber-tear and stock-break failures (ASTM, 2019); these show that the laminate surface or the laminate itself was the weakest part.

Mixed failure needs the area share of each pattern and its location. Record, for example, interfacial failure at the edges and cohesive failure in the middle, with each as a percentage of the bonded area, as ASTM D5573 figures report mixed modes (ASTM, 2019). Location tells you more than the shares do. Interfacial zones at the edges of a joint aged in salt water point to edge ingress, since corrosive undermining "can easily propagate from the plain edges" on plain test specimens (European Aluminium, 2015). Interfacial patches in the middle point to local contamination or uneven preparation.

What is the difference between cohesive and adhesive failure modes?

The difference between cohesive and adhesive failure modes is where the crack ran: cohesive failure runs through the adhesive and leaves adhesive on both surfaces, while adhesive failure runs along the interface and leaves one surface bare. That difference sends the investigation in opposite directions, toward the adhesive and joint design for cohesive failure and toward the surface for adhesive failure.

The boundary between them is the thin-layer case. ASTM D5573-99(2019) lists thin-layer cohesive failure as its own class (ASTM, 2019), and a surface with a very thin adhesive film can look bare to the naked eye. If that joint is called adhesive failure, the team may chase contamination when the weak zone is in the adhesive close to the interface, for example in a primer layer or an interphase that cured differently. Check a bare-looking surface under magnification, or wipe a small area with a solvent that the adhesive's safety data sheet (SDS) and your site's chemical handling rules allow, before you classify it.

The terms also differ by source. NPL defines adhesive failure by where separation "appears to be" (NPL, 2023), which admits that the naked-eye call can be wrong. Some vendor guides call adhesive failure "interfacial" and describe residue on the opposite surface (Incure, 2026). Use the ISO 10365:2022 designations in reports, and add the descriptive name in brackets so readers who use other terms can follow.

Worked example of the adhesive failure mode chart (illustrative)

The worked example of the adhesive failure mode chart uses an invented case to show how the steps and rows lead to a test plan; it describes no real product, customer or result. A line bonding powder-coated steel brackets to a glass-fiber reinforced polyester panel with a two-part epoxy reports low lap shear values on one shift (illustrative).

The table below shows what the chart decides for the five failed coupons in the illustrative case.

Steps 1 to 3 send coupons 1 to 3 to the surface, not to the epoxy, because the steel side is bare under magnification (illustrative). Coupon 5 shows the bond to the panel was not the weak point. The team therefore tests contamination of the coated steel first, pulls the preparation log for the shift, and keeps cure and mix ratio as open hypotheses until the dispense records are checked.

A public case shows why the plan matters. In a laboratory study of an acrylic foam tape that detached from ABS housings, the analysts found a difference in the adhesive's composition and in the stretching marks left after detachment, yet their first conclusion was "Unfortunately, there is no irrefutable proof" (C2MI, 2025). The case is a tape bond, outside this chart's scope; it is cited only for how far surface evidence reaches. The pattern chose the checks; the checks did not close every hypothesis.

Where the adhesive failure mode chart does not apply

The adhesive failure mode chart does not apply when only one half of the joint was kept, when the fracture surfaces were cleaned, handled or exposed after failure, or when the failure load and test conditions are unknown. In each case the surface no longer records the failure reliably.

It also does not judge acceptability. ASTM D5573-99(2019) "does not address the acceptability of any specific failure mode" (ASTM, 2019), and ISO 10365:2022 is scoped to designations and their appearances (ISO, 2022). Whether a joint with 20 % interfacial failure is acceptable is decided by the product specification or the customer's qualification plan, not by this chart (illustrative share).

Visual reading has limits. FTIR, often used on adhesive surfaces, can miss surface detail: in the C2MI study its penetration depth of "between < 1 µm and 10 µm approximately" was noted as possibly too high to see the finest surface composition (C2MI, 2025). A contaminant layer thinner than that may need XPS or another surface-sensitive method.

Single-lap test values carry their own limit. ASTM D4896-01(2024) warns that using single-lap strength values "as allowable design-stress values for structural joints could lead to product failure, property damage, and human injury" (ASTM, 2024). The failure pattern from such a test describes the test joint, not the design strength of the part.

Lubinpla's root cause analysis task records the fracture pattern as an observation, fixes each candidate cause with its prediction before the evidence is opened, and judges each one against the records.

Key takeaways

• Read both halves of the failed joint before you classify it.

• Interfacial failure points to the surface; cohesive failure points to the adhesive and joint design.

• Check bare-looking surfaces under magnification for a thin adhesive film.

• Record mixed failure with area shares and locations.

• Treat every candidate cause as a hypothesis until a record or test settles it.