Peel and cleavage tests on structural adhesive joints typically yield failure loads one to two orders of magnitude below the same adhesive's bulk tensile strength, because both loading modes concentrate stress at a single leading edge of the bondline [S2][S5]. Specifiers who read a peel or cleavage number as "bond strength" are misreading the metric: it is an edge-propagation resistance value, not a design stress [S6].
The two modes are routinely confused because both look like "pull-apart" tests, yet the stress state at the crack tip is geometrically different. Peel applies a flexible arm being lifted at 90° or 180°; cleavage loads one rigid adherend in tension so the bondline opens at a wedge. NPL's 2001 review of adhesion test methods groups both under "joint tests that concentrate stress on an edge" and warns that measured values are geometry-dependent, not material properties [S5].
Stress State: Why Edge Loading Lowers Apparent Strength
Cleavage and peel both create a stress singularity at the bondline leading edge, and the local stress intensity, not the average bond area stress, controls when the crack advances [S5][S2]. 3M's adhesive-joint stress taxonomy states that joints placing stress into cleavage or peel concentrate the load onto the leading edge, and this is the primary driver of premature bond failure in service [S2]. The practical consequence is that doubling the bonded area in a peel-loaded joint does not double the failure load: the resisting cross-section is essentially a line, not the full overlap [S3].
Peel tests are run at controlled angles: 90° peel for edge-lift simulation, 180° peel for full separation, and T-peel for two flexible adherends [S3]. Cleavage, in wedge or tensile-block form, is the standard geometry used in NPL's fracture-mechanics Mode I characterisation of structural adhesives, where the wedge is driven into the bondline to measure the critical strain-energy release rate G_Ic rather than a peak load [S5]. Specifiers who need a material property for design should ask for G_Ic (J/m²) rather than a peak cleavage load in newtons.
Failure Modes: Adhesive, Cohesive, Substrate, and Mixed
Adhesive joint failures are classified into four modes that are independent of whether the test was peel, cleavage, shear, or tensile: adhesive failure (at the interface), cohesive failure (within the adhesive), substrate failure (within the adherend), and mixed mode [S5][S7]. The same adhesive can show 100% cohesive failure in a lap-shear coupon and 100% adhesive failure in a peel coupon, because the leading-edge stress field preferentially seeks the lowest-toughness path [S7].
In the 2020 dynamic-message-sign study by Amatya (ASCE), welded DMS coupons lost up to 13.7% of cleavage strength relative to the baseline, and failed by a mix of weld and rupture modes; the adhesively bonded DMS coupons failed by a different mode, dominated by the bondline, demonstrating that geometry alone can shift the failure locus [S1][S4]. For QA work this is critical: a peel test that ends in substrate tear is reporting the adherend strength, not the adhesive, and the data should be footnoted rather than averaged into an "adhesive peel strength" value [S5].
Peel vs Cleavage vs Shear vs Tensile: A Decision-Criteria Comparison
For specification use, the four common joint tests are not interchangeable. Peel and cleavage both test edge-failure resistance, but peel is the lower-bound test for flexible-laminate construction, while cleavage is the relevant test for rigid adherends loaded in tension at one end. Lap shear and pull-off (per ASTM D4541 and D7234) test area-loaded resistance to sliding or normal separation and report the higher values most specifiers associate with "bond strength" [S8].
Key decision criteria: (1) loading mode, peel/cleavage = edge-loaded, shear/pull-off = area-loaded; (2) typical result magnitude, peel/cleavage failure loads are commonly a small fraction of lap-shear failure loads on the same adhesive; (3) service-relevance filter, choose peel for flexible laminates and dynamic/vibration exposure, choose cleavage for rigid wedge or thermal-expansion loading, choose lap-shear or pull-off for area-loaded structural joints; (4) design-output type, peel/cleavage deliver a load-versus-angle or load-versus-extension curve, while fracture-mechanics cleavage delivers G_Ic for design calculation [S3][S5][S8]. High tensile strength, on this matrix, does not guarantee adequate peel resistance, and the two metrics should be specified independently [S3].
When Peel/Cleavage Numbers Matter, and When They Do Not
Peel and cleavage data are decisive for flexible-substrate laminates, thin-wall assemblies, dissimilar-material stacks (metal-to-composite, aluminium-to-plastic), transportation interior panels under vibration and thermal cycling, and any geometry where the bondline is loaded at a corner or seam rather than across its full face [S3]. They are the wrong test for rigid area-loaded structural bonds, where lap-shear (ASTM D1002 family) or pull-off (ASTM D4541) data govern [S8].
Engineers often see peel strength quoted in N/25 mm or pli and cleavage strength quoted in N or kN, with no failure-mode annotation. Without the failure-mode callout, the number is not comparable across suppliers, because a cohesive-failure value reflects the adhesive's bulk properties while an adhesive-failure value reflects the interface's surface energetics [S5][S7]. The Infinita Lab testing-method overview (2026-09) lists pull-off, lap shear, peel, cross-cut tape and scratch as separate methods with separate ASTM references, reinforcing that a single "adhesive strength" number is not a substitute for a method-specific value [S8].
Standards and Quantification Methods
Standardised peel geometries exist in the ASTM D903 (peel), ASTM D1876 (T-peel) and ASTM D3807 (climbing drum peel) families; pull-off adhesion of coatings to substrates is covered by ASTM D4541 and ASTM D7234 [S8]. For design-calculable fracture-mechanics input, NPL's review recommends Mode I wedge-cleavage testing and reporting the strain-energy release rate G_Ic, which unlike peak load is a geometry-independent material input for finite-element design [S5].
ANSYS ACP (release v251 documentation) implements a peel-stress failure criterion in its adhesive failure model, where the criterion activates when the local peel stress at the bondline exceeds a user-supplied threshold, again confirming that peel is treated as a local edge-stress problem in design software, not as an average stress [S9]. A bonded-joint analysis that omits the peel or cleavage check on the as-built geometry will understate the probability of edge-initiated service failure.
Specifying Peel or Cleavage Data Without Misreading It
A data sheet entry of "peel strength: 80 N/25 mm" tells the engineer only the peak load per unit width on a specific geometry, with a specific adherend, at a specific angle, at a specific test rate. The same adhesive tested at 90° versus 180° versus T-peel will return different numbers, and the failure-mode annotation (cohesive vs adhesive vs substrate) is the gate that determines whether the figure can be used to compare two adhesives [S3][S5].
For new-product qualification, request peel data at the production angle (typically 90° or 180°), the production adherend, the production surface preparation, and the production test rate, plus a reported failure-mode percentage. For design calculation where the joint sees a wedge or corner load, request Mode I fracture toughness G_Ic rather than a peak cleavage load, and feed that value into a finite-element model with an explicit peel-stress criterion [S5][S9]. For a process-engineer audience comparing a new structural adhesive against an incumbent, side-by-side peel and cleavage data on identical substrates is the only fair comparison, and the comparison is incomplete without the failure-mode callout.
Trackable signals to watch: (1) adoption of G_Ic-based design allowables in lieu of peak-load values for peel and cleavage in bonded-joint design guides, and (2) growth of method-specific ASTM peel/cleavage standards covering 3D-printed and composite substrates, where existing flat-coupon standards are not yet validated.
Component reference pages worth checking: industrial adhesive, pressure transmitter, and flow meter.
For related coverage, see 4D 16-Line vs 3D 12-Line Laser Level: Which Plane Count Fits the Job.