The breakdown of a Mil-spec coating system at fasteners in aluminum alloys after fatigue and environmental exposure was studied using electrochemical impedance spectroscopy. In addition, a statistical analysis of the galvanic/crevice corrosion in the fastener counter sink is presented. To study coating breakdown, specimens of aluminum alloys 2024-T6 and 7075-T3 with fasteners were coated using an epoxy primer and polyurethane top and studied as a function of mechanical fatiguing and exposure to temperature cycling, UV, and salt fog. Using the porous penetration model, the effects of each stage of exposure on the coating integrity at the fastener were tabulated across a population of specimens. It was found that fatiguing in the elastic region resulted in a sharp decrease in the pore resistance (Rpore). Environmental exposure also decreased Rpore but to a lesser extent. To study the combined galvanic/crevice separate specimens were exposed in immersion and salt fog tests. The resulting corrosion occurred in the form of small corrosion events (pits, intergranular corrosion, and cracks) on the countersink surfaces. The location and magnitude of damage was investigated by cross-sectional analysis of the countersink. The extent and distribution of corrosion events were found to differ between aluminum alloys 2024 and 7075. The observed corrosion trends were explained through the combined effect of galvanic coupling and crevice solution chemistry changes.

With many aircraft having been in service for over 20 y and expected to remain in service for another 25 y, corrosion problems present significant challenges to maintenance and upkeep.1  Being the largest component of aircraft exposed to the environment, the aircraft skin presents unique corrosion prevention challenges, especially at fasteners (typically cadmium-plated steel), and around lap joint regions. During this lifetime, aircraft are exposed to a multitude of aggressive environments which can induce corrosion at these susceptible locations. The first step in the corrosion at fasteners is the breakdown of the coating. This breakdown results from a combination of environmental effects such as mechanical fatigue, humidity, salt spray, UV light, and temperature variations.2  Once solution has penetrated the coating at the fastener, the underlying 2xxx or 7xxx series aluminum alloys are susceptible to galvanic corrosion with the cadmium-coated steel fastener. Further, the gap at the countersink will also create an environment favorable to crevice corrosion.3  From an engineering perspective, corrosion at the fastener/skin interface has been reported to shorten lifetimes not only due to combined galvanic and crevice corrosion effects, but also corrosion fatigue.4-5  Thus, to extend service lifetimes we must find ways to mitigate coating failure during fatigue at fasteners and the ensuing galvanic/crevice corrosion that occurs at these locations.

Few investigations have studied the combined mechanical/environmental effect on coating breakdown and most of those in the literature are on coated steel systems. None that we could find studied fatigue or coated aluminum alloy systems. With respect to steel systems, Touzain studied the effect of visco-elastic stress state (VE), both compression and tension, on a variety of epoxy coatings in both artificial and natural seawater.6-8  In one of those studies, two epoxy paints, one solvent base and the other nonsolvent base, were immersed in 3 wt% NaCl over a four month period.6-7  In these tests, electrochemical impedance spectroscopy data (EIS) were recorded as a function of time during simultaneous immersion and stress. While the low frequency impedance for the unstressed specimens (both coating systems) remained high after 4 months of immersion, it decreased for specimens under compression and tension, by as much as 3 orders of magnitude. In addition, for specimens under tension the coating capacitance increased proportional to the stress whereas it decreased for specimens under compression. It was further observed that this capacitance effect was different for solvent base vs. nonsolvent base coating and this was rationalized via a “thermo-activated” function of stress.7  In a later study, however, Nguyen and Touzain found that visco-elastic stress improved the properties of an epoxy coating. During compression and tension (±7 MPa & ±9 MPa), specimens were tested for water uptake as a function of temperature (30°C to 60°C) also using EIS.9  It was found that both tension and compression decreased water uptake, diffusion, and permeability in epoxy coatings. They proposed that visco-elastic stress improved the barrier properties of the coating in a way that should “delay the corrosion process.”

In a similar study, Hong investigated the effect of elastic stress on water uptake in two-layer coating system.10  In that study, mild steel panels were coated with a zinc-rich epoxy primer followed by a polyurethane top coat. Once again, EIS was used to characterize water uptake and its diffusion during static tensile and compressive loading as a function of time. In general, it was found that the low-frequency impedance decreased with time while the coating capacitance increased with time resulting indicating an increase in water uptake. Finally, Ranade and Tan reported on the initiation and propagation of coating defects in epoxy-coated mild steel specimens during loading and unloading to simulate pipeline conditions.11  In those tests, specimens were strained to 1%, 2.5%, and 3.5%. After removing the strain, specimen degradation was evaluated using EIS. It was found that higher strains were associated with lower low-frequency impedance values. Post-testing SEM analysis found that loading resulted in shear band formation in the coating and, with increasing strain, these bands formed cracks.

In comparison with the lack of studies on the influence of fatigue on coating failure, the corrosion mechanisms of aerospace aluminum alloys are well studied. However, significant knowledge on the interactions of coupled materials at the fastener and localized corrosion is still lacking. Several attempts have been made to characterize the combined galvanic and crevice corrosion in aluminum systems. Payer and Young studied the long-term atmospheric corrosion of plates of aluminum alloy (AA) 2024 with steel and cadmium-plated fasteners.12  Their study included a teardown of the fastener assembly and comprehensive analysis of the corrosion damage modes through cross-sectional analysis of the countersink regions of fastener assemblies. Significant intergranular corrosion was observed with evidence of a transition over time to stress-corrosion cracking. The authors found that fastener material and time of exposure had a large effect on the depth of corrosion events. While the data presented by Payer and Young showed the stages of damage evolution in the countersink, it was missing critical aspects of the extent and location of damage within the fastener countersink. In addition, Young and Payer did not attempt to quantify the total number of corrosion events and no tabulation of their location in the countersink or bore hole. This information would provide a more complete understanding of the corrosion of this system. The Young study was also limited by the choice of alloy. AA2024 is normally cathodic to the cadmium-plated fasteners in marine solutions while other alloys, such as AA7075, may not be. Also, the stochastic nature of these corrosion events, combined with the few specimens analyzed, leaves many questions about the applicability of these data to modeling efforts and to conclusions about the galvanic action of fastener materials.

More recently, Rafla, et al., through an operando x-ray computed tomography technique, studied a galvanic couple between AA7050-T7451 plates and SS Type 304 pins exposed to 4.0 M NaCl or 2.0 M MgCl2.13  Their technique allowed for the precise tabulation of crevice fissures and the tracking of these corrosion events through time. The found that the fissures were intragranular and did not follow bands of intermetallic particles that they also had identified. In their aluminum samples, corrosion occurred extensively over short periods of time (96 h). The authors concluded that corrosion occurs intragranularly in this system and is supported by cathodic currents from the fastener and replated copper on the aluminum surface. As it relates to this study, they also tabulated damage area as a function of depth from the surface. For the specimen exposed to 4 M NaCl, damage accumulation was bimodal with most of the damage being near the surface. In comparison, for the specimen exposed to 2.0 M MgCl2, damage accumulated predominantly at deeper depths.

Mizuno and Kelly developed a mathematical model for the combined crevice/galvanic corrosion at an aluminum stainless steel fastener along with experimental verification by quantifying the depth of IGC attack in a couple.14-15  They used the Laplace equation and finite element methods to determine potential distributions and galvanic current densities. These studies, along with others such as the galvanic interaction measurements by Feng and Frankel,16  can serve to indicate galvanic action from fasteners on bulk surfaces but fail to account for the extent of galvanic or crevice action within restricted fastener-skin gaps. These approaches are further complicated on passive metals like aluminum by noncontinuous current density distributions. Liu, et al., tried to account for crevice conditions by modeling a galvanic couple along with predicted changes in aeration in the gap space.17  Results indicated a corrosive-to-passive transition in an aerated region on the anodic crevice surface of AA7050 where the corrosive region was defined by potentials above the pitting potential of the aluminum alloy. Liu’s study effectively demonstrates how the galvanic action of the skin-fastener system can extend into the crevice gap as well as how the local chemistry changes within this gap might affect galvanic currents and the likelihood of corrosive attack. However, the study represents only one alloy-fastener material combination and no experimental support was provided. It also did not account for the chemistry changes in the crevice beyond deoxygenation such as metal dissolution and the establishment of more acidic or basic conditions.

The present study aims to supplement efforts to characterize the corrosion mechanisms at fasteners in aluminum aircraft skin. Specifically, EIS was used to gain an understanding of the onset of coating breakdown at fasteners due to combined mechanical and environmental effects. In addition to characterizing coating failure, we also report on the combined galvanic/crevice corrosion that occurs subsequent to coating breakdown. Through exposure to seawater immersion and subsequent cross-sectional analysis of exposed fastener assemblies of two of the most common aerospace aluminum alloys, 2024 and 7075, we characterize the number, type, and location of corrosion events. The results offer clues to the complex interaction of galvanic coupling and crevice chemistry effects in both alloy systems.

Specimen Preparation

Aerospace fastener assembly prototypes were designed to study coating failure around the countersink region and to evaluate the subsequent galvanic/crevice corrosion damage. The specimens were 0.313 cm thick-rolled aluminum plates of alloy 2024-T3 and 7075-T6. The nominal composition of the aluminum plates is given in Table 1. Etching to reveal the grain structure of the as-received alloys was performed with Keller’s reagent and is shown in Figure 1. No significant difference was found between the longitudinal (rolling) direction and the transverse direction (perpendicular to the rolling direction). Both alloys displayed a pronounced elongation in the longitudinal and long transverse directions with grains flattened to a height of around 10 μm to 20 μm with an average length of 100 μm though there was a large variance of grain size. The microstructure appeared similar between the alloys with the exception of the presence of small dull gray intermetallics in AA7075 and copper-colored intermetallics in AA2024. These were likely Zn-Mg-Cu particles in AA7075 and Al-Cu particles in AA2024.18-19 

Table 1.

Elemental Composition of As-Received Plates of Aluminum Alloys 2024-T33 and 7075-T6 in wt%

Elemental Composition of As-Received Plates of Aluminum Alloys 2024-T33 and 7075-T6 in wt%
Elemental Composition of As-Received Plates of Aluminum Alloys 2024-T33 and 7075-T6 in wt%
FIGURE 1.

Image of an etched area of 1/8 in. thick rolled aluminum 2024 (left) and 7075 (right) showing grains elongated in the rolling direction and flattened in the transverse direction. Both alloys exhibited similar grain shapes and sizes and no difference was found in the shape of grains between the L and LT directions. T is the transverse direction, L is the longitudinal (rolling) direction, and LT is the long transverse direction.

FIGURE 1.

Image of an etched area of 1/8 in. thick rolled aluminum 2024 (left) and 7075 (right) showing grains elongated in the rolling direction and flattened in the transverse direction. Both alloys exhibited similar grain shapes and sizes and no difference was found in the shape of grains between the L and LT directions. T is the transverse direction, L is the longitudinal (rolling) direction, and LT is the long transverse direction.

Close modal

These aluminum sheets were used to fabricate two different types of specimens: test panels and dog-bone specimens, as seen in Figures 2(a) and (b). The dog-bone specimens were similar to “Specimen 6” in ASTM E8/E8M (Figure 15[b]) with holes and countersinks added for fasteners. The holes were spaced 2.5 cm on center and the countersink bevel was angled at 100° and at a depth to allow the head of the fastener to lie flush with the exposed outer surface of the plate. Four dog-bone specimens for each alloy were studied in this configuration. In addition to this configuration, specimens were designed to mimic “off-normal” conditions in the machining or installation processes and included countersinks that were machined with a change in the countersink angle or had channels machined between the fastener and the countersink region of the plate. With respect to the test panels, the specimens measured 15 cm × 2.5 cm and contained four holes that were also spaced 2.5 cm on center. As with the dog-bone specimens, specimens were designed to mimic “off-normal” conditions in the machining or installation processes were fabricated. Prior to coating, in some cases, the area around a fastener head was taped off to achieve either a 1-to-1 or 3-to-1 plate-to-fastener area (anode to cathode) ratio (Figure 3). Specimens were prepared with either cadmium-coated steel fasteners or bare-steel fasteners. Altogether, 12 plate specimens of each alloy were produced.

FIGURE 2.

(a) Immersion and salt spray exposure test panel schematic. (b) Dog-bone specimen for environmental and fatigue testing of coating integrity.

FIGURE 2.

(a) Immersion and salt spray exposure test panel schematic. (b) Dog-bone specimen for environmental and fatigue testing of coating integrity.

Close modal
FIGURE 3.

(Left) Image of a crosshatch fastener assembly in AA2024 mid-exposure in seawater immersion conditions. White corrosion product and bubbling from within the countersink gap is visible. (Right) 4:1 plate: fastener exposure specimen after 7 d of immersion in artificial seawater with corrosion products removed.

FIGURE 3.

(Left) Image of a crosshatch fastener assembly in AA2024 mid-exposure in seawater immersion conditions. White corrosion product and bubbling from within the countersink gap is visible. (Right) 4:1 plate: fastener exposure specimen after 7 d of immersion in artificial seawater with corrosion products removed.

Close modal

For both the dog-bone and test panels, the specimens were first ground with 600 grit SiC paper then degreased by cleaning with DI water, acetone, ethanol, and DI water (in that order). The fasteners used were HI-LOK pin and collar fasteners which met military specifications: HL19PB6-2 (pins) and HL77-6 (collars). In this fastener, the head and pin were made from cadmium-coated steel, and the collar was made of an aluminum alloy. The fasteners were tightened to a torque-off of 20 in to 30 in lb and the specimens were cleaned again in the same manner as they were degreased after polishing.

Coating System

The coating system used here was a Navy aircraft system that is also used by the Air Force and consisted of three layers: Top Coat: MIL-PRF-85285E class H, a solvent-based polyurethane, Primer: MIL-PRF-23377J Type I class C2, a solvent-based epoxy containing strontium chromate, and Surface Treatment: MIL-C-5541, a nonchromated chemical conversion treatment for Al substrates.20-21  On the day of the coating application, the MIL-C-5541surface treatment PreKote was applied after cleaning the specimen with 2-butanone solvent (MEK). This surface treatment was applied three times in accordance with the instructions that included scrubbing, drying, and rinsing. Following the PreKote applications, the MIL-PRF-23377K epoxy primer was applied. After curing, this was followed by the MIL-PRF-85285E polyurethane topcoat. A spray gun was used to apply the coatings. Once the coating had cured (either primer or top coat) to ensure the proper thickness according to military specifications (primer: 0.75±0.2 mil, topcoat 2.2±0.2 mil), the thickness was measure using an eddy current base gauge. The final system thickness was also measured and this was nominally 3 mils. To ensure proper adhesion to the substrate, coating was applied to separate test plates of aluminum and a pull-off adhesion test verified adhesion exceeding the limits of the test device. This was taken as proof of proper adhesion. Prior to testing, a silicone was thickly applied on and around the collars on the backside of the specimen to ensure no solution penetration was occurring at this location.

Electrochemical Testing

The polarization responses of the materials were recorded in solutions meant to simulate conditions experienced in bulk seawater solution and crevice conditions within the countersink crevice. As such, polarization tests included aerated ASTM substitute ocean water,22  deaerated ocean water (deaerated by bubbling Argon gas for 24 h prior to immersion), and simulated crevice solution as predicted by thermodynamic electrolyte simulation software (OLI Systems, Inc) using metal salts (described below in Results - Simulated Crevice Solutions).

Polarization responses were recorded for AA2024-T6, AA7075-T3, cadmium, and 4140 steel specimens after they achieved a stable open circuit potential, usually about 1 h, although AA2024 specimens often did not achieve relative stability over longer time frames. Polarization experiments were performed using a scan rate of 0.167 mV/s in a 1 L cell with a calomel reference electrode and platinum mesh counter electrode. Aluminum specimens were made from cut bars of alloy while steel specimens were made by removing the cadmium coating from fastener pins (with concentrated nitric acid) and mounting in epoxy. Lollipop-shaped specimens (no mounting) of aluminum were used in crevice solution experiments to avoid localized corrosion of the specimen edges.

Exposure Testing

Flat plate test panels of alloy 2024 and 7075 were immersed in ASTM substitute ocean water for 1 week to 6 weeks in separate 10 in diameter crystallizing dishes. Air was circulated in the immersion dish with a small fish pump to maintain saturated oxygen. After the immersion period, specimens were immediately removed from solution, cleaned with a nylon brush, rinsed, dried, and then examined for corrosion damage. In addition to full immersion exposures, test panels of each aluminum alloy were exposed to salt spray corrosion testing according to ASTM B117 for 1 week.

The sequence of tests for the dog-bone specimens is presented in Table 2. Prior to testing, baseline EIS data were collected after 1 h of open circuit potential (OCP) measurement in 5% NaCl. These EIS tests were performed on each fastener individually using a small O-ring cell that isolated each fastener for separate EIS measurements. This cell had a diameter of 1.5 cm and contained a Silver/Silver Chloride (Ag/AgCl) reference electrode and a stainless-steel or graphite counter electrode. The EIS scans were performed at the open circuit potential, from 100,000 Hz to 0.01 Hz with an AC voltage of 10 mV, collecting 10 data points per decade.

Table 2.

Test Sequence for Fatigue and Environmental Exposure of Dog-Bone Specimens

Test Sequence for Fatigue and Environmental Exposure of Dog-Bone Specimens
Test Sequence for Fatigue and Environmental Exposure of Dog-Bone Specimens

The specimen was then fatigued while exposed to a 5% NaCl solution at a peak stress of 11.7 ksi, R ratio of 0.05, and a frequency of 5 Hz for a total of 10,000 cycles. Following the fatigue cycle, a second set of EIS data were collected. Following this second EIS test, the specimens were placed into a Cincinnati Sub Zero freeze/thaw chamber cycling between −54°C and 30°C (tg for epoxides are typically on the order of 75°C) over a 4-d time period. The cycle started with ramping down to −54°C over the course of 1 h (step 1), and holding at that constant temperature for 23 h (step 2). After this hold, the temperature in the chamber was ramped to 30°C over the course of 1 h (step 3), and was held at that temperature for 23 h (step 4). This cycle was repeated a second time for a total exposure of 4 d. Following this freeze/thaw exposure, the specimens were moved to a UV chamber where they were exposed for a week. In these tests, specimens were held at a temperature of 60°C, an irradiance of 0.89 W/m2 and a humidity of 100%. The specimens were placed in the chamber oriented such that the head of the fastener countersunk into the specimen was facing the UV lamps and the entire gauge section was exposed. From there, the specimens were placed into a Q-Fog ASTM B117 salt fog chamber for 2 weeks.23  After the specimens finish the 2-week salt fog exposure, they were evaluated one final time using EIS.

Coating Integrity at Fasteners

Figure 4 presents the Bode magnitude and phase plots for a single cadmium plated fastener in AA2024 as a function of exposure. Similar data were observed for fasteners in AA7075. The data labeled “initial” were taken in 0.1 M NaCl just prior to fatiguing the specimen while the data labeled “post fatigue 1” were taken just after. After this EIS measurement, the specimen was exposed to temperature, UV, and salt fog, as described in Table 2. Following that exposure another EIS data set was taken, and is shown in Figure 4 as “post environment.” Finally, following this EIS test, the specimen was subject to a second fatigue cycle followed by another EIS data set, “post fatigue 2.” As seen in this figure, there is a clear deterioration of the coating integrity at this fastener after each exposure step with the most pronounced deterioration occurring after the fatigue steps. The data are best modeled by a two-time constant equivalent circuit and the model in Figure 5 was used to fit the data where: Rct is the charge transfer resistance, Cdl is the double layer capacitance, Rpore is the geometric resistance associated with water penetration in the coating, CC is the coating capacitance, and RS is the geometric resistance associated with the bulk solution between the working and reference electrode. A fit of this model to the data in Figure 4, labeled “initial,” is presented in Figure 6.

FIGURE 4.

Bode magnitude (a) and phase (b) plots for EIS of an area on AA2024 specimens around coated fasteners performed at different steps in fatigue and environmental exposure testing.

FIGURE 4.

Bode magnitude (a) and phase (b) plots for EIS of an area on AA2024 specimens around coated fasteners performed at different steps in fatigue and environmental exposure testing.

Close modal
FIGURE 5.

Equivalent circuit used to model the EIS results from coatings around the fastener heads.

FIGURE 5.

Equivalent circuit used to model the EIS results from coatings around the fastener heads.

Close modal
FIGURE 6.

Plot of fitted data for Bode phase and magnitude using the equivalent circuit model of Figure 5.

FIGURE 6.

Plot of fitted data for Bode phase and magnitude using the equivalent circuit model of Figure 5.

Close modal

For comparison with the fastener data, the EIS data from a nonfastener area on the AA2024 tensile bar is presented in Figure 7. For clarity, only the “initial” and “post fatigue 2” data are presented. As seen in this figure, there is only a single time constant and the low frequency impedance for the nonfastener area is orders of magnitude higher than the fastener. While there is a clear degradation of the coating with fatigue and environmental exposure, it is not nearly as dramatic as that associated with the fastener. As such, we will focus on the effects of fatigue and environment on Rpore at the fastener as an indication of the onset of coating breakdown and the subsequent galvanic/crevice corrosion. The nonfastener data and the values of Rct, Cdl, and CC, will be discussed in later publications.

FIGURE 7.

Bode magnitude and phase plots from a nonfastener area before testing (initial) and after a full cycle of environmental exposure plus a second round of fatigue testing (post fatigue 2).

FIGURE 7.

Bode magnitude and phase plots from a nonfastener area before testing (initial) and after a full cycle of environmental exposure plus a second round of fatigue testing (post fatigue 2).

Close modal

Figure 8 presents values of Rpore for three different fasteners in AA2024 as a function of exposure. Similar data were observed for fasteners in AA7075. The most notable characteristic in this plot is the variation in the initial Rpore values, the highest being 6.7 × 106 Ω·cm2 and the lowest 1.5 × 104 Ω·cm2. It can be further seen that while the impedance at the fastener with the lowest Rpore (fastener 3) remains relatively unchanged as a function of exposure, the impedance at the fastener with the highest Rpore value (fastener 1) changes dramatically reaching a minimum after the environmental exposure period. Further, it can be seen that the second fatigue cycle had no additional influence on the value of Rpore for this fastener. To best illustrate the trends in the data as a function of exposure, we have created a series of cumulative distribution function (cdf) plots of Rpore from all fasteners in AA2024 and AA7075 in Figures 9(a) and (b). These plots were generated by taking the data from all of the fasteners at a given step, sorting them from min (first) to max (last) and assigning each value of Rpore a probability (cdf) based on its position in the sorted data (cdf = position/total). The cdf for a given Rpore is the probability of finding a value in the data set that is less than or equal to that value. For example, the probability of finding an “initial” value of Rpore in a fastener in AA2024 (Figure 9[a]) that is less than or equal to 5.1 (approximately 1.25 × 105 Ω·cm2) is 0.49, e.g., 49% of this data set is less than or equal to 1.25 × 105 Ω·cm2. As can be seen in Figure 9(a), for fasteners in AA2024, there is a wide distribution of Rpore values prior to testing (initial). With exposure, the values of Rpore become lower as does the variability and, thus, the specimen standard deviation decreases as well. This is another way of saying that Rpore is approaching the same value for all fasteners as described by Figure 8. In comparison, for fasteners in AA7075, the variability in the initial values is much less. Correspondingly, there is much less change in the average value of Rpore from “initial” to “post environment.” It should be noted that, for the AA2024 specimen, the EIS data from one of the specimens had only one time constant.

FIGURE 8.

Comparison of Rpore for all three fasteners on a single dog-bone specimen at each step in the environmental exposure sequence.

FIGURE 8.

Comparison of Rpore for all three fasteners on a single dog-bone specimen at each step in the environmental exposure sequence.

Close modal
FIGURE 9.

Cumulative distribution plot of Rpore values from fasteners in AA2024 specimens (a) and AA7075 specimens (b).

FIGURE 9.

Cumulative distribution plot of Rpore values from fasteners in AA2024 specimens (a) and AA7075 specimens (b).

Close modal

Electrochemical Testing

The polarization responses of AA2024, AA7075, cadmium, and 4140 steel in aerated ASTM artificial seawater are shown in Figure 10. These curves simulate the current response of the materials in bulk solution. To represent the initial conditions of oxygen depletion in the crevice, polarization curves in deaerated seawater are shown in Figure 11. These data show a clear difference as compared to aerated conditions, especially for the aluminum alloys. Current densities in the cathodic portion of the steel and cadmium curve are noticeably decreased as compared to aerated conditions, but their OCPs are not largely different. As expected, AA2024 and AA7075 show a decrease in OCP when compared to aerated seawater and a large passive region is visible. The galvanic tendencies of the material pairings (AA2024-Cd, AA2024-steel, AA7075-Cd, AA7075-steel) can be predicted by the intersection points of the more polarization curves. AA2024 having the most positive OCP, is expected to be cathodic when coupled to either cadmium or steel. AA7075 is anodic to steel but has nearly the same open circuit potential as cadmium. On this basis, cadmium is unlikely to afford as much protection for AA7075 as it will for AA2024.

FIGURE 10.

Polarization curves of aluminum alloys 2024 and 7075 and fastener materials Steel 4140 and cadmium, in ASTM artificial seawater with ambient aeration.

FIGURE 10.

Polarization curves of aluminum alloys 2024 and 7075 and fastener materials Steel 4140 and cadmium, in ASTM artificial seawater with ambient aeration.

Close modal
FIGURE 11.

Polarization curves of aluminum alloys 2024 and 7075 and fastener materials, Steel 4140 and cadmium in deaerated ASTM artificial seawater.

FIGURE 11.

Polarization curves of aluminum alloys 2024 and 7075 and fastener materials, Steel 4140 and cadmium in deaerated ASTM artificial seawater.

Close modal

Simulated Crevice Solutions

Restricted transport in the crevice between the fastener and countersink wall leads to oxygen depletion, an increase in metal cations and concomitant increase in Cl and a decrease in pH. The result is a more aggressive chemistry in the crevice than in the bulk solution and it is reasonable to assume that this solution may be different for each alloy or, if the solutions are similar, that their electrochemical response is different. The predicted pH and speciation diagram calculated as a function of metal chloride concentration is presented in Figure 12 for AA2024 (OLI Stream Analyzer). The plot assumes stoichiometric dissolution and the data are presented as a function of molality of AA2024. That is, for AA2024 any given concentration, the solution contains metal chloride based on Al(III), Cu(II), Mg(II), Mn(II), Fe(II), and Cr(III) in the same stoichiometric ratio as they appear in alloy 2024. As seen in this figure, Al(OH)3 and AlOHCl2 precipitate from solution early on though they do not exist in large concentrations. At a concentration of approximately 3.5 m, the solution becomes saturated in Al3+ and, as a result, AlCl3 hexahydrate begins to precipitate from solution. It was found that the diagram for AA7075 was nearly identical (not shown).

FIGURE 12.

Thermodynamic electrolyte simulation of a crevice solution showing the amount of precipitates due to increased stoichiometric dissolution of alloy 2024. The solubility limit of Al cations can be observed at the molality origin of AlCl3 precipitation (approximately 4 m).

FIGURE 12.

Thermodynamic electrolyte simulation of a crevice solution showing the amount of precipitates due to increased stoichiometric dissolution of alloy 2024. The solubility limit of Al cations can be observed at the molality origin of AlCl3 precipitation (approximately 4 m).

Close modal

Simulated crevice solutions for AA2024 and AA7075 (3.5 m) were created using the following salts: AlCl3*6H2O, CuCl2*2H2O, MgCl2*6H2O, ZnCl2, and then deaerated prior to testing with Argon to simulate oxygen depletion. Polarization responses for specimens in this simulated crevice solution are shown in Figure 13. In these solutions, the aluminum alloys show active behavior with corrosion current densities nearly three orders of magnitude greater than in aerated seawater conditions. Cathodic activity on all metals, likely the hydrogen evolution reaction, is also much greater in simulated crevice solutions.

FIGURE 13.

Polarization curves of aluminum alloys 2024 and 7075 and fastener materials, Steel 4140 and cadmium, in a simulated crevice environment made using metal salts

FIGURE 13.

Polarization curves of aluminum alloys 2024 and 7075 and fastener materials, Steel 4140 and cadmium, in a simulated crevice environment made using metal salts

Close modal

Seawater Immersion Tests

The cadmium coating of fasteners is quickly lost for both AA2024 and AA7075 test panel assemblies in about two weeks of exposure, though this progression was noticeably faster for AA7075 panels. This quick degradation of the cadmium coating indicates that long-term corrosion study should concentrate mostly on the effect of bare steel fasteners, assuming the cadmium coating has been depleted.

Images from fastener assemblies of AA2024 and AA7075 after immersion testing in ASTM ocean water are shown in Figures 14(a) and (b). With respect to immersion tests, general corrosion in the machined countersink surface was nearly nonexistent thus requiring cross-sectional analysis to identify small localized corrosion events on the countersink surface. Cadmium-coated fasteners served well to protect the aluminum plates from any severe general corrosion at all exposure lengths. However, even bare-steel fasteners did not induce large amounts of general corrosion damage on the aluminum plates.

FIGURE 14.

View of corrosion damage on alloys 2024 and 7075 after 7 d of immersion (a) and (b) and seven days of salt fog exposure (c and d). Discoloration within the countersink is residual cadmium.

FIGURE 14.

View of corrosion damage on alloys 2024 and 7075 after 7 d of immersion (a) and (b) and seven days of salt fog exposure (c and d). Discoloration within the countersink is residual cadmium.

Close modal

Analysis of corrosion damage within the countersink after seawater immersion was performed by successive cross-sectioning and polishing of specimens to create data (images) from slices of a single countersink specimen. These slices were prepared by removing the fasteners and collars from each countersink area and then halving the area to reveal the cross-section profile of the countersink. A diagram showing how a fastener countersink was sectioned and examined using optical metallography is presented in Figure 15 where 0 refers to the mouth of the fastener/plate and 2000 refers to the bottom of the countersink where it intersects with the borehole. The cross-section specimens were then mounted in epoxy and successively ground with 400 grit to 1200 grit SiC paper. Then, specimens were polished with 1 μm diamond paste until a mirror-like finish was achieved. Each polished cross-section slice revealed a 2D profile of the countersink where corrosion damage in the form of pits or cracks (fissures) at the crevice-exposed surface of the countersink could be observed and tallied. These intergranular corrosion fissures and pits were both classified as corrosion events. The location of events within the crevice region were noted for their distance down the length of the crevice by designating the top surface as the origin. Although small, corrosion events were easily identifiable with a highly polished specimen. The total number of corrosion events found among a sampling of 112 cross-section slices (56 for each alloy) is shown in Figure 16 for AA2024 and in Figure 17 for AA7075 as a function of the distance down the crevice surface. The two alloys display distinct distributions of corrosion damage in their countersink crevices. No damage was found in the bore-hole region in either alloy.

FIGURE 15.

Diagram showing how a fastener countersink was sectioned and examined using optical metallography where 0 refers to the mouth of the fastener/plate and 2000 refers to the bottom of the countersink where it intersects with the borehole.

FIGURE 15.

Diagram showing how a fastener countersink was sectioned and examined using optical metallography where 0 refers to the mouth of the fastener/plate and 2000 refers to the bottom of the countersink where it intersects with the borehole.

Close modal
FIGURE 16.

Histogram of corrosion event locations on the countersink wall for aluminum 2024. Events were recorded for a total of 56 cross sections from 7 different countersinks. A distribution profile is superimposed on the histogram to demonstrate a trend of damage occurring at distinct distances down the crevice length.

FIGURE 16.

Histogram of corrosion event locations on the countersink wall for aluminum 2024. Events were recorded for a total of 56 cross sections from 7 different countersinks. A distribution profile is superimposed on the histogram to demonstrate a trend of damage occurring at distinct distances down the crevice length.

Close modal
FIGURE 17.

Histogram of corrosion event locations on the countersink crevice surface of aluminum 7075. Events were recorded for a total of 56 cross sections from 7 different countersinks. A distribution profile is superimposed on the histogram to demonstrate a trend of damage occurring at distinct distances down the crevice length.

FIGURE 17.

Histogram of corrosion event locations on the countersink crevice surface of aluminum 7075. Events were recorded for a total of 56 cross sections from 7 different countersinks. A distribution profile is superimposed on the histogram to demonstrate a trend of damage occurring at distinct distances down the crevice length.

Close modal

Salt-Spray Exposure Tests

Images of specimens after salt fog exposure are shown in Figure 14(c) and (d). Salt fog was a far more corrosive environment than seawater immersion. Specimens showed large amounts of general corrosion on their exposed top surface, which was not observed on immersion testing. Corrosion of the countersink surface, as revealed by cross-sectional analysis, was also much greater in salt fog exposure for both cadmium-coated and bare-steel fasteners. In salt-spray experiments, general corrosion was substantially greater on countersinks with steel fasteners compared to cadmium-coated fasteners. Immersion test specimens were inconclusive regarding differences between fastener material.

Coating Failure

In theory, the impedance of a nondefective polymer coating contains no in-phase or real component and is characterized by a purely capacitive response. In practice, the coating capacitance is typically associated with a parallel resistor representing the intrinsic resistivity of the coating (e.g., a single-time constant parallel RC circuit). As the coating degrades, the development of ionic conduction occurs. This pathway for electrolyte to the metal interface can be thought of as a network of pores in the coating though they are seldom macrosopic defects that can be observed optically or in an SEM. Rather, they are a network of many pathways that may result in regions of chemical bond breaking at the polymer chain level. These new ionic pathways manifest themselves as a second time constant in the EIS response. This time constant is in parallel with the coating capacitance and consists of a resistance representing the cumulative effect of all of the coating pores (Rpore) in series with a simple Randle’s circuit for the electrolyte that has now reached the metal surface, a polarization resistance in parallel with a double-layer capacitance. This idea serves as the basis for the porous penetration model developed by Kendig and Scully and presented in the equivalent circuit in Figure 5.24  The initial EIS data collected at the fasteners for both the AA2024 and AA7075 specimens (Figure 6) were distinctly characteristic of this porous penetration model, the time constant associated with Rpore in the frequency range of 10 Hz to 1,000 Hz being clearly evident. This was in sharp contrast to the EIS data for a nonfastener area. In that case, the data had only one time constant and the low frequency impedance values were orders of magnitude higher. The implication of this result is that the coating in this area was allowing water penetration at the fastener to some degree prior to any testing, fatigue, or environmental chamber. There may be several reasons for this. One might speculate that there were “macroscopic defects” as a result of poor coating coverage, however, when we tested off-set fasteners where the countersink was drilled at an angle less than the normal 100°, the values for Rpore were two orders of magnitude lower than on normal fasteners, as seen in Figure 18. Here, Rpore is approximately 400 Ω·cm2 at the breakpoint frequency of 1,200 Hz. Therefore, if defect fasteners possessed macroscopic defects due to poor coating coverage, we would expect any similar defects on normal fasteners to also have this low Rpore value. As they possess a much greater Rpore, we do not believe that macroscopic defects are responsible for this behavior. In addition, in comparison with the AA2024 tensile specimens, the initial Rpore values from AA7075 were typically lower. As the coatings on AA2024 and AA7075 nominally had the same thickness and this trend persisted across multiple dog-bone assemblies, there is no reason to believe it is an issue related to coating application as this would tend to average out. Rather, we believe this is related to a lack of wetting at coating/metal interface in the region of the fastener. A free-standing film that spans the fastener-aluminum interface (as may be the case in AA7075) would be in tension, whereas, if the coating wetted the gap (as may be the case in AA2024), the gap would nominally be “filled” and close to the same stress state across the interface. If films in tension have greater water uptake as discussed above, a free-standing film (AA7075) would necessarily have lower Rpore values.

FIGURE 18.

Bode magnitude and phase data for an off-set fastener in ASTM artificial ocean water without fatigue or environmental exposure. The countersink was drilled 10º off from the normal angle while the main through hole for the fasteners was drilled normal. This resulted in a fastener that was noticeably raised above the surface at a skewed angle.

FIGURE 18.

Bode magnitude and phase data for an off-set fastener in ASTM artificial ocean water without fatigue or environmental exposure. The countersink was drilled 10º off from the normal angle while the main through hole for the fasteners was drilled normal. This resulted in a fastener that was noticeably raised above the surface at a skewed angle.

Close modal

The effect of fatigue and environmental chamber exposure on Rpore is clear, Rpore decreases with fatigue and increasing exposure. While the initial value of individual fasteners varied (Figures 9[a] and [b]), the population of Rpore values showed a decrease in the cdf = 0.5 with each step in the experimental procedure. As it relates to the onset of galvanic corrosion between the aluminum sheet and the fastener, the coating offers little delay in solution ingress and its onset in the presence of fatigue cycling. Even moderate fatigue cycling resulted in Rpore decreasing by more than a factor of 10. For a constant coating thickness, a drop in Rpore by a factor of 10 is equivalent to an increase in “wetted” area by the same amount.

Corrosion Damage Within the Countersink

Corrosion in salt-spray tests was significant, however, it differed greatly in magnitude and appearance from the results in immersion testing and previous atmospheric testing.12  For this reason, it was assumed that immersion testing more accurately represents the corrosion expected from field applications and focus in this study was restricted to such results.

Corrosion within a countersink appeared mostly as intergranular corrosion (corrosion fissures) and preferential grain dissolution. This is clear from the etched surface of AA2024 in Figure 19, where the corrosion fissure pathway precisely follows the grain boundaries, with occasional specific grain dissolution. This type of corrosion was typical of most corrosion events observed through cross sections of the countersink in both AA2024 and AA7075. Often, a vast network of intergranular corrosion and grain dissolution was visible, extending from the countersink surface into the aluminum sheet, sometimes as far as 2 mm. The elongation of grains that results from cold working of the alloys into a thin sheet is responsible for the significant horizontal spread of intergranular corrosion fissures while the heat treatment, meant to provide precipitation strengthening, creates susceptible intergranular regions. Similar IGC and grain dissolution has been observed by Payer and Young for atmospheric corrosion of the same type of fastener assembly.12  In that study, the greatest fissure length observed was 0.6 mm after 24 months of exposure. Additional examples of corrosion events found in the countersink surface are shown in Figure 20.

FIGURE 19.

(Left) Intergranular corrosion network extending from the countersink surface of an aluminum 2024 plate. (Right) The same network after etching to reveal the grain structure. Intergranular corrosion cracks follow the grain boundaries with larger corrosion pits extending into the grains themselves.

FIGURE 19.

(Left) Intergranular corrosion network extending from the countersink surface of an aluminum 2024 plate. (Right) The same network after etching to reveal the grain structure. Intergranular corrosion cracks follow the grain boundaries with larger corrosion pits extending into the grains themselves.

Close modal
FIGURE 20.

Examples of typical corrosion events found in the crevice region of an aluminum 2024 countersink during cross-section analysis. Pits and cracks on the surface of the countersink were measured for their distance within the countersink-fastener crevice with the measurement origin at the horizontal top surface.

FIGURE 20.

Examples of typical corrosion events found in the crevice region of an aluminum 2024 countersink during cross-section analysis. Pits and cracks on the surface of the countersink were measured for their distance within the countersink-fastener crevice with the measurement origin at the horizontal top surface.

Close modal

In the absence of a galvanic couple, such as the steel fastener in this study, the mechanism of intergranular corrosion in Al-Cu-Mg alloys involves precipitation of θ and S phases at the grain boundaries which are anodic to the grain matrix.25  The AA2024 alloy used in this study likely presents S phase precipitates at the grain boundary owing to the ratio of Cu/Mg being in the range of 4 wt% to 1.5 wt%.26  Intergranular corrosion in AA2024 can be present without any pitting corrosion, starting out as grain boundary attack at the surface and propagating before any observable pitting corrosion.25  Corrosion events on the countersink surface were often seen to coincide with the presence of a constituent particle at, or immediately near, the exposed surface, as can be seen by the darker particle at the angled surface in Figure 19. These particles (Al-Cu-Mn-Fe) are commonly associated with pitting in AA2024, and may act as local cathodes and accelerating dissolution of the surrounding matrix.27  The mechanisms for intergranular corrosion of AA7075 in the literature are similar to those reported for AA2024. For AA7075, Maitra and English proposed that crack propagation to owe to anodic dissolution of grain boundaries.28  In Al-Mg-Zn alloys, the contents of grain boundaries has been determined to be enriched in Mg, Zn, and Cu.29  In the present study, cathodic particles were not observed at the initiation sites of corrosion pits in AA7075 as they were with AA2024.

Knight found that fissure length in AA2024 and AA7050 was limited in depth and they proposed that adjacent sites stifled one another as they competed for cathodic current from neighboring intermetallic particles.30  For systems where the alloy is coupled to a cathode (e.g., steel fastener), such as Young, Rafla, and the present study, corrosion fissure length is much longer. Rafla proposed that the presence of the cathodic fastener promoted reduction rate and capacity allowing greater depths as corrosion fissures did not follow obvious clusters of intermetallic particles.13 

Mechanisms of Damage

The number and distribution of corrosion events in the countersink is clearly different for AA2024 as compared to AA7075. For AA2024, there were fewer events and they were more uniformly distributed along the countersink interior (Figure 16). In comparison, for AA7075 the number of events was much greater and their distribution appeared to be a function of distance from the surface (Figure 17). A probability distribution was overlaid onto these histograms for a visual representation of the distribution of corrosion events. For AA2024, this distribution takes the appearance of a log-normal probability density function. For AA7075, the damage is greater at the boldly exposed surface and near the mouth of the crevice, as compared to AA2024. This may be expected as Ecorr for AA2024 is actually more noble than the fastener (Figure 10). In addition to this difference between the alloys, for AA7075 there is a second increase in damage accumulation at a distance of 0.8 mm to 1 mm into the crevice. In comparison, Rafla found for alloy 7050/SS316 assembly that the damage distribution was different for 4 M NaCl exposure, as compared to 2 M CaCl2. For 4 M NaCl solution, damage accumulation was bimodal with most of the damage being near the surface. In comparison, for the specimen exposed to 2.0 M MgCl2, damage accumulated predominantly at deeper depths and little surface damage was reported.

Using Figures 16 and 21 as a reference, we propose that corrosion within the countersink of AA2024 occurs in the following manner.

  • 1.

    At the mouth of the crevice, the solution is not yet deaerated. The coupled potential is then determined by the aerated polarization curves (region 1) where the OCP of AA2024 is more positive than the fastener (cadmium or steel). Thus, at the couple potential in region 1 AA2024 is the cathode and the steel fastener is the anode (Figure 21). This is consistent with the low level of pitting corrosion observed in the probability density plot in Figure 16.

  • 2.

    At some distance into the crevice, the restricted transport results in oxygen depletion. Assuming hydrogen reduction within the crevice does not play a role in the cathodic reaction, reduction occurs on the boldly exposed steel (in aerated seawater) and the couple potential would be in the passive region of the polarization curve for the bulk solution (Figure 21). However, as time progresses, the restricted mass transport would also result in a buildup of Al+++ (and corresponding Cl) in this region. This occluded chemistry decreases the pitting potential of AA2024 and stable pits and cracks begin to form. This is supported by the observation in Figure 16 of an increase in pitting and cracking in region 2.

  • 3.

    Deep within the crevice in region 3 the IR drop becomes more significant, influenced even more by a buildup of corrosion product around region 2. The result is to lower the potential of AA2024 in this region. The magnitude of damage in this region will depend on the concentration of Al+++ and Cl, however, one could easily envision that this region tends toward the OCP of AA2024 in deaerated seawater, resulting in a decrease in anodic currents.

FIGURE 21.

Polarization curves for AA2024 and steel overlaid to show the galvanic responses expected at different regions in the countersink crevice.

FIGURE 21.

Polarization curves for AA2024 and steel overlaid to show the galvanic responses expected at different regions in the countersink crevice.

Close modal

In summary, the reason for a spike in the number of corrosion events in region 2, at short distances from the crevice, may owe to a balance between chemistry changes within the crevice and IR drop between the separated half-cell reactions. Essentially, the middle region experiences aggressive chemistry changes and anodic events can still be supported by the bulk cathodic reactions due to the relatively lower IR (as compared to deeper distance). Thus, in the AA2024 system, its more positive OCP in the bulk solution only serves to protect it for a short distance into the crevice.

With respect to aluminum alloy 7075, using Figures 17 and 22 as a reference, we propose that corrosion within the countersinks occurs in the following manner.

  • 1.

    At the mouth of the crevice (aerated conditions, region 1 in Figure 22), the OCP of AA7075 is more negative than that of the steel fastener. As a result, the couple potential is more positive than the pitting potential of AA7075 and corrosion events in this region are likely whether an aggressive solution is formed or not. This is consistent with the large number of events observed in this region in Figure 17.

  • 2.

    At short distances from the mouth, corrosion events are still likely, independent of transport considerations, but as the IR drop increases their likelihood decreases. This is consistent with the decrease in events, as seen in region 1b in Figure 17.

  • 3.

    As was the case for AA2024, at some distance into the crevice, the restricted transport results in oxygen depletion and an increase in Al+++ and Cl concentration. As a result, the pitting potential decreases and the number of events peaks. This is consistent with the increase in observation in region 2 for AA7075 in Figure 17. Comparing region 2 for AA2024 and AA7075 in Figures 16 and 17, corrosion events in alloy 7075 outnumber events seen on AA2024 in this region by 2 to 1 or more. This can be explained by the differences in the polarization curves for the alloys in the crevice simulant. As seen in Figure 11, the pitting potential and OCP for AA7075 in the simulant are lower than that of AA2024 by as much as 100 mV in each case indicating a greater susceptibility to damage for the same concentration.

  • 4.

    As was the case for AA2024, deep within the crevice in region 3 the IR drop becomes more significant. The result is a lower potential of AA7075 and, correspondingly fewer damage events in this region, as seen in Figure 17.

FIGURE 22.

Polarization curves for AA7075 and steel overlaid to show the galvanic responses expected at different regions in the countersink crevice.

FIGURE 22.

Polarization curves for AA7075 and steel overlaid to show the galvanic responses expected at different regions in the countersink crevice.

Close modal

In AA2024 crevices, the buildup of an aggressive chemistry is necessary to explain the occurrence of corrosion events. Otherwise, the AA2024 would be expected to be protected by the coupling to the fastener materials. For AA7075, galvanic effects alone could explain the corrosion damage profile, but it is likely that chemistry changes in the crevice also contributed to accelerate corrosion.

This analysis fails to indicate the progress or extent of corrosion beyond the number of events observed. As shown in Payer and Young, corrosion events such as pits or intergranular corrosion on the countersink surface of AA2024 increase with time.12  Their data indicates that intact cadmium-plated fasteners (CdPtd specimens in that study) effectively decrease the number of events and possibly delay the onset of the transition to SCC as compared to specimens where the cadmium coating was damage (CdSteel). In the current study, no time effect was observed on the corrosion extent, although our immersion times were significantly lower. Therefore, while the number of corrosion events are higher for alloy 7075, we are uncertain if this correlates with a transition to more significant corrosion damage modes.

In this investigation, the breakdown of an epoxy base coating system (MIL-PRF-85285E/MIL-PRF-23377J) on aluminum alloys 2024-T3 and 7075-T6 and the subsequent corrosion at cadmium-plated steel fasteners (HL19PB6-2, HL77-6) was investigated. Coating integrity at the fasteners was investigated after subjecting dog-bone specimens to combined fatigue, salt fog, temperature, and UV exposures. From the results of these tests, the following was concluded.

  • The porous penetration model is adequate to describe the coating system breakdown at the fastener, where Rpore is the electrolyte path resistance through the coating.

  • The effect of fatigue and environmental chamber exposure on Rpore was clear, Rpore decreased with fatigue and increasing exposure.

  • Analysis of the total population of Rpore results across all specimens produced a distinct cumulative distribution at each stage of exposure and shows that relatively few fatigue cycles and environmental exposure resulted in failure of the coating around fasteners.

The combined galvanic/crevice corrosion at the fastener was investigated in a series of immersion tests on coated specimens with intentional defects. Following immersion, damage location along the countersink wall was quantified using optical metallography and serial cross sections through the countersink. From the results of these tests, the following conclusions were made.

  • Corrosion in the countersink was mostly in the form of intergranular corrosion or intragranular pits at the surface of the aluminum countersink. The majority of corrosion events were small (<100 μm) with only a few observed to have spread significantly into the matrix of the aluminum sheet.

  • Comparing AA2024 and AA7075, corrosion events in alloy 7075 outnumbered events seen on AA2024 in this by 2 to 1 or more.

  • The distribution of corrosion events along the countersink was explained by considering galvanic interactions with the fastener materials as well as the development of an aggressive crevice chemistry in the gap between the fastener and alloy. Specifically, at short distances from the crevice, corrosion damage was attributed to a balance between chemistry changes within the crevice and IR drop between the separated half-cell reactions. Essentially, the middle region experiences aggressive chemistry changes and anodic events can still be supported by the bulk cathodic reactions due to the relatively lower IR (as compared to deeper distance).

This work was supported by the DoD Technical Corrosion Collaboration sponsored by the U.S. Department of Defense Office of Corrosion Policy and Oversight through the U.S. Air Force Academy Grant No. FA7000-14-2-20016. The authors thank Daniel J. Dunmire, Richard Hays (CPO), and Gregory Shoales (CAStLE) for their continued support. The authors also thank Chad Hunter and the Air Force Research Labs for support and guidance.

1.
National Research Council
,
Aging of U.S. Air Force Aircraft
(
Washington, D.C.
:
National Academies Press
,
1997
).
2.
C.R.
Hegedus
,
S.J.
Spadafora
,
A.T.
Eng
,
D.F.
Pulley
,
D.J.
Hirst
,
“Aerospace and Aircraft Coatings, MNL12239M Paint and Coating Testing Manual: 15th Edition of the Gardner-Sward Handbook”
(
West Conshohocken, PA
:
ASTM International
,
2012
),
p
.
739
750
.
3.
K.S.
Ferrer
,
R.G.
Kelly
,
Corrosion
58
,
5
(
2002
):
p
.
452
459
.
4.
R.C.
Kinzie
,
Aircr. Eng. Aerosp. Technol.
75
,
2
(
2003
):
p
.
123
129
.
5.
S.
Russo
,
P.K.
Sharp
,
R.
Dhamari
,
T.B.
Mills
,
B.R.W.
Hinton
,
G.
Clark
,
K.
Shankar
,
Fatigue Fract. Eng. Mater. Struct.
32
,
6
(
2009
):
p
.
464
472
.
6.
N.
Fredj
,
S.
Cohendoz
,
X.
Feaugas
,
S.
Touzain
,
Prog. Org. Coatings
63
,
3
(
2008
):
p
.
316
322
.
7.
N.
Fredj
,
S.
Cohendoz
,
X.
Feaugas
,
S.
Touzain
,
Prog. Org. Coatings
72
,
3
(
2011
):
p
.
260
268
.
8.
N.
Fredj
,
S.
Cohendoz
,
X.
Feaugas
,
S.
Touzain
,
Prog. Org. Coatings
74
,
2
(
2012
):
p
.
391
399
.
9.
D.
Nguyen Dang
,
B.
Peraudeau
,
S.
Cohendoz
,
S.
Mallarino
,
X.
Feaugas
,
S.
Touzain
,
Electrochim. Acta
124
(
2014
):
p
.
80
89
.
10.
H.Q.
Yang
,
Q.
Zhang
,
S.S.
Tu
,
Y.
Wang
,
Y.M.
Li
,
Y.
Huang
,
Prog. Org. Coatings
99
(
2016
):
p
.
61
71
.
11.
S.
Ranade
,
M.
Forsyth
,
M.Y.J.
Tan
,
Prog. Org. Coatings
110
(
2017
):
p
.
62
77
.
12.
P.S.
Young
,
J.H.
Payer
,
Corrosion
71
,
10
(
2015
):
p
.
1278
1293
.
13.
V.N.
Rafla
,
A.D.
King
,
S.
Glanvill
,
A.
Davenport
,
J.R.
Scully
,
Corrosion
74
,
1
(
2018
):
p
.
5
23
.
14.
D.
Mizuno
,
R.G.
Kelly
,
Corrosion
69
,
6
(
2013
):
p
.
580
592
.
15.
D.
Mizuno
,
R.G.
Kelly
,
Corrosion
69
,
7
(
2013
):
p
.
681
692
.
16.
Z.
Feng
,
G.S.
Frankel
,
Corrosion
70
,
1
(
2014
):
p
.
95
106
.
17.
C.
Liu
,
V.
Rafla
,
J.R.
Scully
,
R.G.
Kelly
,
“Mathematical Modeling of Potential and Current Distributions for Atmospheric Corrosion of Galvanic Coupling in Airframe Components,”
CORROSION 2015, NACE-2015-5579
(
Houston, TX
:
NACE International
,
2015
).
18.
N.
Birbilis
,
R.G.
Buchheit
,
J. Electrochem. Soc.
152
,
4
(
2005
):
p
.
B140
.
19.
A.E.
Hughes
,
N.
Birbilis
,
J.M.C.
Mol
,
S.J.
Garcia
,
X.
Zhou
,
G.E.
Thompson
,
“High Strength Al-Alloys: Microstructure, Corrosion and Principles of Protection,”
in
Recent Trends in Processing and Degradation of Aluminium Alloys
(
London, United Kingdom
:
InTech
,
2011
).
20.
Navair
,
“Cleaning and Corrosion Control Volume V Consumable Materials and Equipment for Avionics,”
2009
.
21.
W.
Nickerson
,
C.
Matzdorf
,
“Non-Chromate Aluminum Pretreatments,”
NAVAIR
,
2012
.
22.
ASTM D1141
,
“Standard for the Preparation of Substitute Ocean Water”
(West Conshohocken, PA: ASTM International, 1999).
23.
ASTM B117-18
,
“Standard Practice for Operating Salt Spray (Fog) Apparatus”
(West Conshohocken, PA: ASTM International, 2004).
24.
M.
Kendig
,
J.
Scully
,
Corrosion
46
,
1
(
1990
):
p.
22
29
.
25.
C.
Luo
,
X.
Zhou
,
G.E.
Thompson
,
A.E.
Hughes
,
Corros. Sci.
61
(
2012
):
p
.
35
44
.
26.
Y.C.
Lin
,
Y.C.
Xia
,
Y.Q.
Jiang
,
H.M.
Zhou
,
L.T.
Li
,
Mater. Sci. Eng. A
565
(
2013
):
p
.
420
429
.
27.
G.S.
Chen
,
M.
Gao
,
R.P.
Wei
,
Corrosion
52
,
1
(
1996
):
p
.
8
15
.
28.
S.
Maitra
,
G.C.
English
,
Metall. Trans. A
12
,
3
(
1981
):
p
.
535
541
.
29.
R.G.
Song
,
M.K.
Tseng
,
B.J.
Zhang
,
J.
Liu
,
Z.H.
Jin
,
K.S.
Shin
,
Acta Mater.
44
,
8
(
1996
):
p
.
3241
3248
.
30.
S.P.
Knight
,
M.
Salagaras
,
A.M.
Wythe
,
F.
De Carlo
,
A.J.
Davenport
,
A.R.
Trueman
,
Corros. Sci.
52
,
12
(
2010
):
p
.
3855
3860
.