R S Court, TWI
Extended abstract of paper presented at 6th European Adhesion Conference and Adhesion '02, 10-13 September 2002, Glasgow, Scotland
A new test method to measure the shear properties of adhesive materials has been developed, which attempts to avoid some of the problems with test procedures currently in use. Details are presented of a new adhesive shear sample based on a torsion test using a tube bonded to a flat plate. The new tube-plate test sample has several advantages over other test methods: the adhesive bondline thickness is accurately controlled; there are no alignment issues; a tube gives a uniform strain distribution around the circumference of the sample; pin loading on to the tube-plate eliminates the possibility of the sample slipping during clamping or loading.
Experimental details are presented of shear tests using the new tube-plate sample with a two-part acrylic adhesive, which has shear failure strains in excess of 100%. The effects of bondline thickness were investigated and the results show a thickness effect, with thinner adhesive layers being up to twice as strong and with 40% higher failure strains than thick layers. An analytical model is presented which gives insight into why thin adhesive layers are stronger than thick layers.
Understanding the shear properties of adhesives is important for both the final mechanical performance of an adhesive joint and for use in the joint design process. In order to obtain the required shear properties, test methods such as the thick adherend shear test (TAST, ISO 11003-2) and the 'napkin ring' (ASTM E 229-70) have been developed and were reviewed by Dean et al.  But these suffer from problems of either non-uniform shear stress states, as with the TAST, or complicated and difficult procedures for preparing samples, as with the 'napkin ring' and other torsion tests using aligned tubes.
To try and avoid some of the problems with existing shear samples, a new test sample was developed, using some of the features from the ASTM and ISO standards. A sample based on a tube was selected, since this gives a uniform distribution of strain around the circumference of the tube. Several iterations of specimen design were needed to obtain the final design, which uses an aluminium tube bonded to an aluminium plate.
The adhesive used was a two-part acrylic, QuickBond 5002 from Permabond. The shear test sample was made from aluminium alloy tube and plate.
2.2 Tube-plate shear test sample and test procedure
Shear test specimens were made as shown schematically in Figure 1
. A small step was machined on one end of the tube. The depth of this step controlled the adhesive thickness, as shown in Figure 2
. The surfaces of the tube and base plate were prepared for bonding by abrading with ScotchBrite TM
and degreasing with a Lotoxane (petroleum distillate) wipe. A 1mm wide strip of single-sided pressure sensitive adhesive with polytetrafluoro-ethylene (PTFE) backing layer was located on the machined step of the tube to prevent this part from being bonded to the base plate, Figures 1 and 2
Adhesive was applied to the stepped region of the tube using a cartridge and gun system with a static mixer. The base plate was placed onto the adhesive, and held in place for one hour. The adhesive fillet formed was checked for uniformity and any excess adhesive removed before it had cured. After the adhesive had cured, the adhesive thickness, ta , was calculated from Equation 1. The dimensions for ta , Lb and Ls are shown in Figure 2.
Fixtures for clip gauge extensometers were bonded to the tube and plate. Samples were tested in torsion with pin-loading on the tube using a tension-torsion machine manufactured by ESH/MTS. The base plate was prevented from rotating by loading pins acting on its edges ( Figure 1). Using pins bearing on parts of the sample eliminates the possibility of the sample slipping during clamping or loading. No axial tension or compression load was applied to the tube. The applied torque, time and the displacement at the two clip gauges were measured. Data for torque and displacements were collected every second. The torsion load was increased at a rate to cause failure of the sample within 2 to 3 minutes. At the same time as the sample was being loaded, a video recording of the edge of the adhesive layer was made using a CCD video camera and VHS video recorder.
The torque readings were converted into shear stress at the radial mid-point of the adhesive layer using Equation 2. The geometry and symbols used in the shear strain calculation are shown in Figure 3
, and the shear strain was calculated using Equation 3.
The results from the shear tests on four adhesive bondline thicknesses are shown in Figure 4. These results showed that different thickness bondlines exhibit the same initial stiffness of around 100MPa. However, the strength and failure strain of the adhesive depended on its thickness, with thin bondlines having a higher strength and failure strain than thick bondlines.
A key point from the investigation is the result that thin adhesive layers are stronger than thick layers. An explanation for this can be developed using the observation from the video images of the test, (Court 
) which showed whitening and cracks developing in the adhesive layer at an angle of 45° to the axis of the tube. Harte and Fleck 
developed an analysis based on the cracks effectively fragmenting the adhesive into a series of beams, which then rotate under the applied shear stress. Based on this analysis, a relationship was developed by Court 
between the length, l
, of the adhesive beams, their thickness, tc
, and the applied shear stress τxy
, as shown in Equation 4.
A comparison of predictions from Equation 4 and experimental results is given in Table I, which shows the correct predicted trend of thinner layers being stronger, and gives good agreement between experimental and theoretical values for joint strengths.
Table I. Comparison of predicted and experimental shear failure stresses for acrylic adhesive.
|Adhesive thickness (mm)||Beam length,|
|Beam thickness, tc (mm)||Predicted shear fail stress (Nmm-2)||Expt shear fail stress (Nmm-2)|
||0.15 ± 0.05
||0.04 ± 0.02
- A new shear tube-plate test sample has been developed. This sample is simple to assemble, has no alignment problems and gives good control of adhesive layer thickness.
- This type of sample would significantly reduce the cost of specimen manufacture, in order to produce shear data on adhesives, providing a torsion test machine is available.
- A thickness effect was found with thin (0.1mm) adhesive layers being up to 100% stronger and with 40% higher failure strain than thick (1.1mm) layers.
- An analytical model was developed, explaining the adhesive layer thickness effect.
- Dean G, Duncan B, Adams R, Thomas R, Vaughn L, Third European Conference on Adhesion - Euradh '96, Cambridge, UK. (1996) 113.
- Court R S, PhD thesis, 'The long-term durability of adhesive joints', Cambridge, 2001.
- Harte A-M, Fleck N, Eur J Mech A/Solids 19 (2000) 259.
7. List of symbols
δc = estimated displacement at centre of adhesive layer (mm)
δg = measured displacement at clip gauge extensometer (mm)
γ = shear strain
τ, τxy or τmax = shear stress, applied shear stress or shear failure stress (Nmm-2)
ψ = angle of shear cracks in adhesive layer, taken as 45° in Equation 4
l = length of adhesive beam in cracked shear sample (mm)
Lb or Ls = length of bonded tube sample or length of stepped part of tube (mm)
Q = torque (Nmm-1)
rc or rg = radius at centre of adhesive layer or radius at clip gauge (mm)
ri or ro = inner or outer radius (mm)
ta or tc = thickness of adhesive layer or of adhesive beam in cracked shear sample (mm)