Experimental and statistical study of three adherence tests for an epoxy-amine/aluminum alloy system: Pull-Off, Single Lap Joint and Three-Point Bending tests

The mechanical resistance of a bonded joint depends on the adhesive interaction onto the substrate and the mechanical properties of the adhesive itself. Many existing tests can be useful for measuring the adherence or evaluating mechanical adhesive response. All these tests do not provide the same information: in particular, adherence measurements can be split into initiation tests and propagation ones. In this paper, three adherence tests have been considered for the evaluation of the fracture initiation between a poly-epoxide adhesive (a mixture of pure epoxy and amine) and an aluminum surface (AA 2024-T3), namely the Pull-Off, Single Lap Joint (SLJ) and Three-Point Bending tests. Various surface preparation protocols before bonding have been tested and optimized for aluminum substrates, including mechanical and chemical surface treatments, followed by the application of an appropriate primer before bonding. This study paves the way for the future development of adhesive systems as it provides reliable surface preparation protocols for aluminum surfaces and gives an insight into the choice of an adequate adherence test dedicated to high-performance adhesives. The load at break (FMax), the experimental error, the failure mode and statistical studies according to the Weibull model and Principal Component Analysis (PCA) were studied on each surface preparation configuration. It has been shown that the application of a primer, especially a sol-gel product increases the load at break and provides more reliable results. Then, this paper shows that the two tests can quantify the failure initiation and distinguish the different surface preparation efficiency, are the Single Lap Joint test (mode II or mode I + II) and the Three-Point Bending test (mode I), with an increase of the results reliability with the latter one. The Pull-Off test (mode I) is useful as a routine checking, and particularly interesting because its response does not depend on the substrate thickness, even though it cannot highlight the difference between all surface preparations.