Testing standard
ASTM D897 / D2095
ASTM D897, Standard Test Method for Tensile Properties of Adhesive Bonds; ASTM D2095, Standard Test Method for Tensile Strength of Adhesives by Means of Bar and Rod Specimens
Written and technically reviewed by Dak System Inc. engineeringLast reviewed
ASTM D897 and ASTM D2095 pull an adhesive bond apart in straight tension across a butt joint, with no overlap and no offset, so the whole glue line is loaded at once. D897 covers metal-to-metal bonds on the standard shapes it defines; D2095 covers bar- and rod-shaped butt joints in a wider range of adherends. Both describe themselves as comparative — the number ranks adhesives and pre-treatments rather than sizing a joint.
At a glance
- Test type
- Tensile — the specimen is pulled apart
- Published by
- ASTM
- Edition
- D897-08(2024)
- Material
- Adhesives, tapes & bonded joints
- Runs on
- Series 7200 and Series 9000
What the test does
Two adherends are bonded end to end so that the glue line lies square across the axis, and the joint is pulled straight apart. There is no overlap and no offset — the whole bonded area is loaded in tension at once, and the load path runs through the centre of the bond. D897 does this with the standard shaped specimens it defines, for metal-to-metal bonds under defined conditions of pre-treatment, temperature and testing machine speed. D2095 does the same with bar- and rod-shaped butt-jointed specimens, and extends the geometry to adherends other than metal, in similar or dissimilar combinations.
What it measures, and why it matters
Tensile strength of the bond: the load at rupture divided by the bonded area. This is the closest a routine adhesive test comes to loading the glue line uniformly, which makes it the natural companion to lap shear rather than a substitute for it. A single-lap joint rotates as it loads and the ends of its overlap carry peel as well as shear, which is why the result is called apparent; a butt joint has no such rotation to build in.
D897 is explicit that it is primarily comparative. Its declared use is as a discriminator — for variations in surface preparation, for the durability of an adhesive system in a given environment, and for the control of primers and pre-treatments in production. It is a way of finding out whether something has changed, not a stress a designer can carry into a joint of different proportions. D2095 is described in the same terms: relative tensile strength.
The butt joint
Two adherends bonded end to end, so the glue line lies square across the load axis.
- Configuration
- Butt joint, bonded across the full sectionNo overlap and no offset — which is exactly what distinguishes it from a lap-shear specimen.
- D897 adherends
- Metal to metal, on the standard shapes the method defines
- D2095 adherends
- Bar- and rod-shaped, in similar or dissimilar combinationsThe wider adherend scope is the reason D2095 exists alongside D897.
- Bonded area
- Fixed by the methodThe method defines the specimen shape and with it the bonded area. The dimensions are in the purchased text and are not quoted here.
- Surface preparation
- Exactly as specified — degrease, abrade, etch or primeVery often the variable the test is actually comparing.
- Concentricity
- Aligned during bonding, not corrected afterwardsPractice
- Squeeze-out
- Trimmed before testDakAdhesive left proud of the section adds bonded area the calculation does not know about, and under-reports the strength.
- Conditioning
- Standard laboratory atmosphere
D897 names pre-treatment, temperature and testing machine speed as the defined conditions of the test. Change any of them and the result is not comparable with an earlier one.
Test speed
- Rate
- Fixed by the method and reported with the resultThe numeric rate could not be corroborated from two independent published sources and is deliberately not quoted here. Take it from the edition in force.
- Recorded quantity
- Load at rupture
Calculations
σ = P_max / A
- P_max
- load at rupture, N
- A
- bonded area, mm²
Unlike lap shear, this is not called an apparent strength — the butt geometry does not build in the joint rotation that makes a lap result an average over a distribution. It is still comparative, because it remains sensitive to alignment, bond-line thickness and preparation.
How the test runs
- 01Machine or select the adherends to the specified shape.
- 02Prepare both bonding faces with the specified surface treatment.
- 03Bond the two halves concentrically, controlling bond-line thickness.
- 04Cure to the adhesive's schedule with the halves held in alignment.
- 05Trim squeeze-out and measure the bonded area.
- 06Condition in the standard laboratory atmosphere.
- 07Mount through self-aligning fittings at both ends.
- 08Pull at the specified speed to rupture and record the load.
- 09Examine both faces and classify the failure — cohesive, interfacial, or in the adherend.
- 10Divide the rupture load by the measured area and report the failure mode alongside it.
A specimen that fails in the adherend has told you something true about the assembly and nothing about the adhesive. Record it, and do not average it in with bond failures.
What the report has to contain
- Reference to ASTM D897 or D2095, and the edition
- Adhesive identification, batch and cure schedule
- Adherend material, shape and dimensions
- Surface preparation and any primer
- Measured bonded area
- Bond-line thickness where controlled
- Conditioning and test atmosphere
- Testing machine speed
- Tensile strength of each specimen
- FAILURE MODE for every specimen
- Number of specimens, mean and scatter
What the machine must be capable of
A modest frame and a rigorously axial load path. On a bonded section of the size these methods use, a structural adhesive at tens of megapascals fails in the tens of kilonewtons, so a frame in the 30 to 50 kN class covers most work while leaving the load cell in the part of its range where it resolves well.
Alignment is the whole design problem. A butt joint is short, stiff and unforgiving, and any eccentricity converts part of the tension into bending, which lifts one edge of the glue line into peel and fails the joint early. Both methods are run through self-aligning grip fittings that let the specimen find the load line rather than being forced onto it; threaded couplings screwed into fixed adapters do the opposite. Speed is fixed by the method and must be reported with the result. No extensometer is used — the quantity is a rupture load, not a strain.
What goes wrong in practice
Misalignment, which is invisible in the trace and shows only as scatter between nominally identical specimens. Adherend failure rather than bond failure, which is a valid observation but not a bond strength. Squeeze-out left in place, so the effective area exceeds the measured one and the strength is under-reported. Bond lines of uncontrolled thickness, which change the result on their own. And reporting a mean with no failure mode: cohesive rupture through the adhesive says the adhesive set the limit, while a clean interfacial release says the surface preparation did.
Butt tension against the other bond tests
| ASTM D897 / D2095 | ASTM D1002 / ISO 4587 | ASTM D3167 | ASTM D3433 | |
|---|---|---|---|---|
| Loading | Tension across the bond | Shear, with peel at the ends | Peel | Cleavage, Mode I |
| Stress across the bond | Near-uniform, if aligned | Peaks at both ends of the overlap | Concentrated on a moving line | At a crack front |
| Output | Tensile strength | Apparent shear strength | Force per width | Fracture toughness, energy per area |
| Design allowable | No — comparative | No — comparative | No | Closer — an energy quantity |
Only the fracture-mechanics route yields a quantity that transfers to a joint of different geometry. The other three rank adhesives, qualify pre-treatments and monitor a process, which is what they were written for.
Questions we are asked about this test
What is ASTM D897?
It is the ASTM test for the tensile properties of adhesive bonds. Two metal adherends of a standard shape are bonded end to end and pulled straight apart under defined conditions of pre-treatment, temperature and testing machine speed. The rupture load divided by the bonded area is the reported strength.
How does ASTM D2095 differ from D897?
D2095 uses bar- and rod-shaped butt-joined specimens and applies to a wider range of adherend materials, in similar or dissimilar combinations, where D897 is written around metal-to-metal bonds on the shapes it defines. D2095 names D897 as the alternative approach, so the two are companions rather than competitors.
Why use butt tension rather than lap shear?
Because a butt joint loads the whole glue line at once, with no overlap and no offset to make the stress non-uniform. A single-lap specimen rotates as it loads and its overlap ends carry peel as well as shear, which is why its result is called apparent. Butt tension answers a cleaner question, and the two together describe a bond better than either alone.
Can I use the result as a design stress?
No. D897 states that it is primarily comparative, and D2095 speaks of relative tensile strength. The declared use is as a discriminator — for surface preparation, for primers, and for the environmental durability of an adhesive system. A real joint is rarely loaded in pure tension across its bond, and the figure does not transfer to a different geometry.
Why does alignment matter so much on a butt joint?
Because the specimen is short and stiff and has nothing to absorb an off-axis pull. Any eccentricity converts part of the tension into bending, which lifts one edge of the glue line into peel and fails the joint early and inconsistently. Self-aligning fittings at both ends are what keep the load line through the centre of the bond.
What does the failure surface tell me?
What to change. Cohesive failure through the adhesive, with material left on both faces, says the adhesive's own strength set the limit. Interfacial failure, one face coming away clean, points at degreasing, abrasion, etching or priming. Failure in the adherend says the joint outlived the material around it. Same number, three different problems.
What frame size do these tests need?
A modest one. On the bonded sections these methods use, a structural adhesive at tens of megapascals ruptures in the tens of kilonewtons, so a frame in the 30 to 50 kN class covers most work while keeping the load cell in the part of its range where it resolves well.
Do I need an extensometer?
No. The quantity measured is a load at rupture divided by an area — there is no gauge length and no strain to follow. Machine capacity, alignment and a properly sized load cell are what decide the quality of the result.
Running ASTM D897 / D2095 on the Series 7200 and Series 9000
Dak verifies against whichever standard the method names, and where a class applies our frames sit a class tighter than it asks.
| The method asks for | Dak supplies | |
|---|---|---|
| Capacity | On the bonded sections these methods use, a structural adhesive at tens of megapascals ruptures in the tens of kilonewtons, so a 30 to 50 kN frame covers most work while keeping the load cell in the part of its range where it resolves well. | Load cells from 1 kg to 60 ton on the Series 7200, and 0.5 to 100 kN on the Series 9000 |
| Force accuracy | unknown — no force-verification class could be confirmed for either method from a publisher record, and none is asserted here | ISO 7500-1 Class 0.5, verified to ASTM E4, DIN 51221 and BS 1610 |
| Gripping | Self-aligning fittings at both ends, so the butt-joined specimen finds the load line rather than being forced onto it | Wedge, vice-action, pneumatic and hydraulic grips, built to the specimen |
| Environment | Specimens conditioned in the standard laboratory atmosphere; D897 names pre-treatment, temperature and testing machine speed as the defined conditions of the test | 3009 series chambers, −150 °C to +400 °C — temperature only |
This page describes the method as practised. The governing text is the current edition from the issuing body. Tell us what you are testing and we will answer with the machine, the fixture and a quotation.