An interdisciplinary study concerning the durability of adhesively elastomer/metal bonded joints in marine environments is reported. The generation of OH⁻ ions at the bondline due to an imposed cathodic current from an external source is suspected to be the predominant cause of failure. A surface analysis study was performed early in this research for the purpose of identifying the cause(s) of failure. Characterization of the rubber and the metal failure surfaces with XPS (X-ray Photoelectron Spectroscopy) showed similar composition on both sides and to that of the bulk degraded primer component of the adhesive. Saponification of the adhesive and the leaching of chlorine (forming HCI that attacks the oxide) are identified as two possible failure mechanisms. The locus of failure is believed to be very close to the . adhesive/oxide interface. The exposure of bulk adhesive free-standing films to different environments showed that the hydroxyl is detrimental in the environmental durability of these specimens judging by the percentage of net mass uptake. An alternate interfacial failure mechanism is also presented where the neutralization of the adhesion-promoting attachment sites (A.S.) at the interface leads to de-adhesion and whereby OH⁻ ions chemically break-up the -COOH-Fe bond forming a non-operative activated complex at the degraded crack tip.

Debond, or loss of adhesion, can exist in two modes. Weakening denotes debond growth which takes place undetected to the naked eye, and is governed by a diffusion-control degradation process that gives a straight line when plotted against the square root of time. Delamination, on the other hand, is a "post weakening" process., Debond rates in this mode are influenced by the applied total strain energy release rate, G_T, and by the environment and can be described by an exponential function in G_T. The effect of shear stress on debond acceleration was determined to be minimal. While compressive stresses seem to be beneficial in slowing the ingression of the bulk hydroxyl into the bondline, no noticeable improvement was detected when an imposed current was used. The use of zinc phosphate-coated steel substrates is shown to improve bond durability significantly at low voltages. Similar trends are observed when silane (γ-aminopropyltrimethoxy) modified primer were used in bonding.

Two approaches are used in order to model debonding: empirical and analytical. Statistical Analysis System (SAS) is used to fit the empirical model which draws heavily on the functional dependencies of debond rates on the accelerating parameters, i.e., temperature, stress, and applied voltage. An Arrhenius relationship is shown to model the effect of temperature very well. Also, voltage effect is correlated with the corresponding current densities which, in turn, were found to obey an exponential relationship with debond rates. SAS fits of the experimental data are shown to model the process accurately and could be utilized for life predictions. Integration of delamination rates in real time is a feasible method to predict durability as well.

A generalized analytical model for debonding is also developed, and it draws on the similarities between this application and corrosion fatigue of metals. The model is based on the conservation of mass of the involved species and is composed of a system of partial equations and their associated boundary conditions. Furthermore, temperature and voltage-dependent diffusion coefficients and reaction rate constants were used. The resulting boundary value problem amounts to a diffusion-chemical reaction mechanism into which a mechano-chemical failure mechanism is incorporated. A simplified version of the full scale analytical model is solved numerically and some interesting conclusions concerning the failure criterion are drawn. The model also simulates the weakening and delamination behavior and allows for temperature and voltage treatment as well. Delay times are also predicted as a function of the applied voltage and temperature. A particularly important conclusion is that the "marching boundary" phenomena seems to account for most of the accelerating influence of applied G.