Effects of Surface Preparation on Long-Term Durability of Composite Adhesive Bonds

Jason D. Bardis, Keith T. Kedward

Department of Mechanical and Environmental Engineering

University of California Santa Barbara, Santa Barbara, California 93106

The evaluation of adherend surface preparation techniques on the long-term durability of composite bonded joints is addressed. Several potential factors are evaluated, concentrating on the effects of peel plies, release fabrics, release films, and grit blasting on the strength, durability, failure modes of adhesively bonded composites, as well as paste vs. film adhesives, all with materials and configurations typical of commercial and general aviation aircraft.

Initial work concentrated on an evaluation of the floating roller peel test (ASTM D3167), in an attempt to extract quantitative data as a quality control test method (Fig. 1a). This test method, designed for the peeling of a thin ductile metal sample from a thicker one, proved to not be conducive to the brittle behavior of a thin composite laminate. Subsequently, a unique form of the double cantilever beam (DCB) test was developed and used for a sequence of test evaluations of carbon fiber adherends bonded with 2-part epoxy paste adhesive (Fig. 1b). This DCB specimen incorporates features of the ASTM D3433, D3762, and D5528 samples. The DCB research results are intended to aid in the interpretation of a modified version of the wedge test where the usual aluminum adherends are substituted with composite adherends (Fig. 1c). In this wedge test, the samples are forced partially apart to initiate a crack in the bondline, the specimen is placed in a chamber at elevated humidity (to promote crack growth and simulate service conditions), and the crack length is measured over time.

Preliminary evaluations of an alternate version of the wedge test [1], where the wedge is driven under position control rather than left in place, have been conducted. This travelling wedge test method involves using a custom tabletop test rig to force a razor blade into a bonded joint at an extremely slow velocity (quasi-static condition) and optically measuring crack length. A series of travelling wedge specimens and DCB specimens cut from the same bonded panels were tested and compared. The results showed similar average critical strain energy release rate (G_Ic) values, but the DCB test results were grouped more tightly than these initial travelling wedge tests’. Modifications to the travelling wedge test, to better suit it to perform more reliably with the materials in this study, are underway—the razor blade has been replaced with a 1/8" thick stainless steel wedge, and the tests are being conducted in an Instron screw-driven test machine.

Examination of the crack front shape, for both test methods’ specimens, is in progress, using X-ray photography. A Zinc Iodide solution injected into the crack before exposing the film provides clear images of the crack front, aiding in determining the accuracy of measuring the crack tip location by visual inspection from the side of the sample. Photos showed that the crack fronts are relatively straight, with little deviation from one side to the other or from one side to the center of the specimen, confirming the validity of crack measurement by viewing one side of the specimen during testing.

DCB test load-deflection curve data (Fig. 2) was combined with visual crack position measurements to compute G_Ic values of the joints. The critical strain energy release rate calculations for DCB tests were made according to the "modified beam theory method" [2-4]. For the travelling wedge tests, G_Ic was computed based on fracture mechanics, as a function of the wedge thickness and the crack length [1].

Fig. 2 Sample DCB Test Load-Deflection Curves

Curing adherends with a coated nylon release fabric on the surface tends to deposit release agents such as silicone onto the adherend surface [5, 6]. Designed to facilitate removal, these contaminants also inhibit proper chemical bonding of the adhesive to the adherend. DCB tests have revealed results of this phenomenon in the form of interfacial failures and intermittent crack propagation, with reduced loads and crack opening displacements. This fracture behavior resulted in significantly lower G_Ic than equivalent joints with bonds made to adherend surfaces cured against PTFE vacuum bag material. The bond failures of surfaces cured on PTFE always failed cohesively, indicating a strong chemical bond between the adhesive and adherend.

In DCB tests, grit-blasted bonded joints tended to have higher G_Ic values, initial failure loads, and maximum opening displacement than their non-blasted counterparts. However, the mode of failure (interfacial or cohesive) was unchanged by the blasting.

Since it has been shown that the coated nylon release fabrics inhibit proper bonding, peel plies more typical of those used on commercial airliners are being used on future autoclave-cured specimens. Subsequent samples are being cured using various versions of a polyester peel ply (unmodified, mechanically calendered to reduce porosity, and coated with a siloxane chemical finish).

2. Johnson, W. Steven, Butkus, Lawrence M., and Valentin, Rodolfo V., "Applications of Fracture Mechanics to the Durability of Bonded Composite Joints," DOT/FAA/AR-97/56 final report, May 1998.