Adhesive bond failure analysis: can a lab tell why a bonded joint came apart?
Short answer
We can investigate chemical differences between a failed adhesive joint and an unfailed reference to narrow plausible explanations for debonding. We can rank explanations supported by composition evidence, but cannot promise a unique root cause or reconstruct the application site. We need both uncleaned break faces, an unused unfailed control, and adhesive, substrate, application, cure and exposure details.
What are you really asking when an adhesive joint fails?
Did the joint separate at the interface or inside a material?
Bonded joints can join metals, plastics, glass, wood or composites with structural adhesives, sealants, hot melts, pressure-sensitive adhesives or water-based systems. Adhesive failure occurs at an interface; adhesive mainly on one face can direct attention to contamination, preparation or compatibility, without proving which occurred. Cohesive failure occurs within the adhesive and can leave adhesive on both faces, raising questions about mixing, cure, aging or moisture-related degradation. Substrate failure means the joined material fractured, while mixed failure combines locations; neither identifies the cause from appearance alone.
Was the problem the adhesive, the substrate or the application?
For example, a manufacturer investigating a rejected batch or supplier change can ask whether the failed interface differs chemically from a comparable normal joint. Possible contributors include release agents, silicone oils, grease, migrated plasticizers and rust-preventive oils, alongside incorrect mixing, insufficient cure, changed adhesive and heat, moisture, chemicals or sunlight exposure. Our adhesive formulation guide addresses what the adhesive contains; the joint investigation asks how location-specific findings relate to separation.
How would we approach adhesive bond failure analysis?
1. We would review adhesive identity, substrates, application and cure conditions, time before separation, exposure and the number of affected parts, identifying available unused same-batch adhesive and normal joints.
2. We would visually inspect and photograph both break faces, mark sampling locations and record an initial interface, cohesive, substrate or mixed-failure interpretation. We would record which substrate belongs to each face, since adhesive left on a plastic side and a clean metal side directs a different comparison from separation within the adhesive. We would label any declared primer layer separately, because a break within primer raises a different question from separation at the metal surface. Before sending, avoid wiping or touching the faces and package them separately to preserve their individual evidence.
3. We would compare each face, the normal bond line and suitable unused adhesive with FTIR, locating adhesive-family evidence and checking retained cure-related features against a comparable cured reference.
Unused adhesive describes incoming material; comparison with a cured joint must account for reaction and volatile loss before interpreting a missing ingredient as a supplier change, with that distinction retained in the report.
4. We would take controlled surface wipes or extracts for GC-MS, looking for accessible volatile or semivolatile release-related molecules, low-molecular-weight siloxanes, plasticizers, oils and residues against substrate backgrounds.
5. We would use high-resolution LC-MS and LC-MS/MS on suitable extracts to investigate migrated nonvolatile additives that the gas-phase comparison may not reveal.
6. We would use XRF for silicon, zinc or calcium differences, adding ICP-OES on suitably prepared extracts when useful, without naming silicone release agent or zinc stearate from elements alone.
7. We would compare failed, normal and unused-material findings, rank the supported explanations and list the process information needed to test each interpretation.
Why can these methods reveal clues at a failed bond?
FTIR reads chemical-bond absorption patterns, helping distinguish adhesive and substrate families when their signals can be separated. A thin adhesive remnant can be masked by its backing, so comparing both faces avoids assuming that a visually clean surface contains no adhesive. GC-MS separates suitable volatile and semivolatile molecules and compares mass-spectral patterns in mixed surface extracts. High-resolution LC-MS narrows candidate formulas for suitable dissolved nonvolatile compounds, while LC-MS/MS adds fragment evidence without uniquely proving every structure. XRF and ICP-OES add elemental clues; silicon can also belong to fillers, so its presence does not establish silicone contamination. Matched references help distinguish normal joint chemistry from an unusual surface contribution, although sampling may combine adhesive, primer and substrate.
What will the bond-failure report say, and what remains uncertain?
The English PDF would include a conclusion card, location-specific component table, supported estimates where relevant, data appendix and unresolved items, separating observations from ranked explanations. Cure-related findings support comparison with suitable normal material, rather than an absolute degree of cure or the original resin-to-hardener ratio. Whether abrasion, corona, plasma treatment or primer application was performed adequately often remains an indirect inference requiring production records. This study does not provide peel, shear or tensile strength, aging trials or humid-heat resistance results, and it does not assign responsibility or determine legal or insurance outcomes.
Which bonded specimens should you declare before sending?
New unused bonded parts failing at inspection, assembly, storage or testing, with unfailed controls, need declaration and SDS review; parts removed from service are not received from overseas. Water-based adhesives and hot melts need declaration review; reactive polyurethane hot melts need case-by-case review. Declare solvent-based adhesive or uncured A/B components; non-water-based liquids and components containing prohibited hazardous-list substances are not received as supplied, so propose cured specimens for review. For large joints, propose representative cut sections; confirm dimensions first. Use our inquiry page and shipping guide; ship after agreement, payment and an acceptance number.
Which related guides should you read?
Our failure-analysis service explains chemical evidence and its causal limits; Our coating-delamination guide addresses separation in paint systems; Our cured-epoxy guide separates retained network chemistry from original mixing proportions; Our residue guide explains deposits, background controls and preservation.
FAQ
Does adhesive on both faces prove that the bond was adequate?
Adhesive on both faces can indicate cohesive failure, but it does not prove adequate strength or correct processing; mixed failure can also leave adhesive on both sides.
Does a silicone finding identify the cause or its source?
Siloxane evidence can be consistent with release-related contamination, but formulation ingredients and several process materials may share that chemistry, so source and causation need independent support.
Does no reported contaminant rule out a contaminated interface?
A trace interfacial film may be below measurable levels or hidden by adhesive and substrate signals, so a report without a contaminant finding does not establish a clean original interface.