Aerospace structures often involve dissimilar materials to optimize structural performance and cost. These materials can then lead to the formation of galvanic couples when moisture is present. Specifically, noble metal fasteners (such as SS316) are often used in aluminum alloy load-bearing structures, which can lead to accelerated, localized corrosion attack of the aluminum alloy due to the cathodic current supplied by the SS316 fastener. This localized attack is difficult to predict, and tests are often expensive, so modeling of these galvanic couples could be of great utility. The work reported here focuses on the galvanic coupling between fasteners installed in a panel test assembly, and the resultant corrosion damage down the fastener holes. This arrangement is a common assembly geometry in aerospace applications. A specific sol-gel coating was applied to the fasteners, to determine its effectiveness on mitigating galvanic corrosion; bare fasteners were also tested, to investigate a worst-case scenario. Geometric constraints in the model were made to match those of an experimental test panel, which was exposed to ASTM B117 salt fog for 504 h. The electrochemical boundary conditions were generated in solutions appropriate to the material and environment to which it would be exposed. Anodic charge passed during exposure was calculated from image analyses of the corrosion damage in the experimental test, and the results were compared with the model. The Laplacian-based model provides a very good first approximation for predicting the damage within the fastener hole. Validation was provided by both experimental results generated in this study as well as comparison to results in the literature that used similar, but not identical, conditions.

Aluminum alloys are commonly used in aerospace structures, with AA7075-T6 (UNS A97075(1)) specifically used in the fuselage and frames due to its high strength: weight.1  Due to microstructural constraints, however, AA7075-T6 cannot be welded, and components must be joined together mechanically.2  Commonly, more noble fasteners are the solution, as they supply adequate mechanical properties. If both the fastener and panel are painted with a coating which has no defects, the assembly would experience no corrosion. Problems arise when defects in the coating are present, either developed during service or from installation, which allow pathways for an aqueous solution to interact with the metals. This intrusion of solution can initiate the onset of galvanic corrosion. Repairs to aircraft sometimes lead to the installation of bare, noble metal fasteners,3  which exacerbates the galvanic corrosion. Dry installation (i.e., without the inclusion of sealant in the hole) represents a worst-case scenario for this panel/fastener design with respect to corrosion susceptibility, but is known to happen.

Experimental measurements of localized corrosion damage can be time-consuming in terms of sample construction, testing duration, cross-sectional metallography, and post-test optical profilometry. Having a validated computational model allows focusing of experiments in the most important parts of parameter space (explored more in Part 2). Therefore, finite element modeling has become a common approach to study the effects of galvanic coupling, through the utilization of both the Laplace equation,4-12  and the Nernst-Planck equation.4,13-16  A thorough discussion on both approaches was recently published by Liu and Kelly,17  however, this work will only focus on a Laplacian-based model.

Much modeling research has been done on the galvanic coupling of dissimilar metals in simple geometries.4-,6,9,14,18  However, less modeling research has been done on more complex geometries, such as panel/fastener assemblies,8,11-12,19-20  which are more common in real-life situations and are known to trap water in crevices between dissimilar metals.21-22 

The complex geometry of structural components in the aircraft increases the difficulty in both experimental measurement and modeling of the galvanic couple. The fastener shaft and hole form an occluded cell with the panel, which leads to local changes in chemistry and thus electrochemical behavior. The occluded geometry also requires complex ohmic drop considerations. Previous experimental work has focused on the external surface corrosion of a scribed fastener/panel assembly,3,23-24  however, fastener holes are known to be the location of most crack initiation, due to the stress concentration that occurs there. Corrosion damage, which is known to serve as an initiation site for fatigue cracks,25-28  can severely impact the fatigue life of the aircraft if it occurs in a fastener hole, where it is very difficult to detect without disassembly.28-29 

The present work aims to develop a validated modeling strategy for galvanic corrosion in complex geometries, particularly down a fastener hole. The main goals were to quantitatively evaluate damage within the fastener holes of corroded panels, including the effects of applying a sol-gel coating to the fastener to mitigate that damage, and then compare results to model predictions of damage down fastener holes. Specifically, the computational portion of this research evaluated the galvanic corrosion current within fastener holes containing bare noble metal fasteners. The model was also used to create calculations that could be compared to measurements of the galvanic current during corrosion via zero resistance ammeters, as reported by Feng, et al.,23,30  Wang, et al.,31  and Boerstler.24 

In order to evaluate the corrosion in the complex geometry described above, an accelerated galvanic corrosion test plate designed by Matzdorf, et al.,3  was used. Referred to in this document as the “NAVAIR test panel,” it consisted of cathodic fasteners and hardware assembled in a coated aluminum plate. Scribes were placed on the surface of the coated aluminum plate to simulate defects in the coating.

To mitigate the galvanic coupling happening inside of the fastener hole, a specific sol-gel coating was applied to the cathodic fasteners. The rationale for coating the cathode is based on simple electrochemical theory. The conservation of electric charge requires that the total anodic current (TAC) (Ia) must equal the total cathodic current (Ic). Utilizing this relation, one can see that the cathode surface area (Ac) to anode surface area (Aa) ratio becomes extremely important when trying to decrease the amount of anodic current density in a system (see Equation [1]): where ia is the anodic current density and ic is the cathodic current density. If the cathodic surface area decreases, then the anodic dissolution current density must decrease (for a constant anode surface area). Conversely, if the anodic surface area decreases, then the anodic dissolution current density will increase. Such considerations are important when assessing coating schemes in galvanic couples. If the anode is coated, any area exposed by defects will experience a very high current density, making the local dissolution worse than if there were no coating at all. Conversely, if the cathode is coated, any defect in the coating will only increase the dissolution of the anode by an amount proportional to the exposed area, and the galvanic attack will never be worse than if no coating were applied. If perfect coatings existed, there would be no difference between coating the anode vs. coating the cathode. As this is not the case, coating the cathode is the more robust means of minimizing the effects of a galvanic couple. Any surface treatment on the cathode would act in the same way as a coating, limiting the amount of cathodic current available to galvanically couple.

formula

The work reported here involved both experimental and computational methods.

Accelerated Testing

An accelerated corrosion test panel configuration designed by NAVAIR3  was utilized (see Figure 1). This panel has been used in previous literature to study the effects of galvanic coupling between dissimilar fasteners assembled in a panel.3,19,23,31-32  Around each of the lower two fastener holes, two scribes were made through the coating in the aluminum plate in order to assess the effects of coating defects on the corrosion morphology. The panels were 3 in × 6 in × 0.25 in, with four fasteners on one half and four fasteners on the other half; in the current research, half-panels (3 in × 3 in × 0.25 in) were utilized. 10-32 threaded SS316 (UNS S31600) and 10-32 threaded Ti-6Al-4V (UNS R56400) fasteners were offset from each other in two horizontal rows. SS316 washers with an outer diameter of 0.75 in were placed under the SS316 fasteners and Ti-6Al-4V washers with an outer diameter of 0.5 in were placed under the Ti-6Al-4V fasteners. All bolts were torqued to 100 in oz. Further details of the NAVAIR plates can be found elsewhere.3,23,32 

FIGURE 1.

AA7075-T6 NAVAIR galvanic test panel assembly with bare SS316 and Ti-6Al-4V fasteners; numbers represent different simulated real-life scenarios.

FIGURE 1.

AA7075-T6 NAVAIR galvanic test panel assembly with bare SS316 and Ti-6Al-4V fasteners; numbers represent different simulated real-life scenarios.

Close modal

These panels were constructed from an AA7075-T651 plate (6 mm), with two SS316 fasteners and two Ti-6Al-4V fasteners installed.3,19,23,31  AA7075-T6 and SS316 are both commonly used alloys in aircraft design, for structure and fasteners, respectively. Ti-6Al-4V fasteners are more noble than SS316 fasteners, and they are generally much more expensive. Both fasteners were assembled in the same panel, in order to make testing most efficient. The combination of this panel design and alloys is consistent with literature.

For the current work, the AA7075-T6 panel was coated front and back with a Surtec 650V pretreatment, and a 44GN072 chromate-containing primer (MIL-PRF 85582). Along the front and edges of the panel, a 03W127A topcoat (MIL-PRF 85285) was applied. In the first panel, all fasteners were uncoated (i.e., bare) and were thus dry installed. This represented a worst-case scenario for a repair to aircraft, making it most susceptible to corrosion. It is important to note that the interior of the holes in the present work were also intentionally not coated.

Overall, there were four “real life” scenarios that the experimental panel attempts to emulate (Figure 1).

  • I.

    A section of an aircraft with bare Ti-6Al-4V fasteners, dry installed in a panel with no coating defects (best-case for minimal galvanic corrosion)

  • II.

    A section of an aircraft with bare SS316 fasteners, dry installed in a panel with no coating defects (better case)

  • III.

    A section of an aircraft with bare Ti-6Al-4V fasteners, dry installed in a panel which contains a surface defect modeled by a scribe in the coating (bad case)

  • IV.

    A section of an aircraft with bare SS316 fasteners, dry installed in a panel which contains a surface defect modeled by a scribe in the coating (worst case)

A second NAVAIR panel was tested, however, both types of fasteners were spray coated with a protective sol-gel coating, rather than being assembled bare. These fasteners were still dry installed. The sol-gel applied was approximately 5 μm to 10 μm thick. The goal of applying this sol-gel was to limit cathodic current availability and thus limit the anodic dissolution of the AA7075-T6 panel. This sol-gel coated fastener panel was made to be compared with the four bare fastener scenarios above, to determine the effect of the sol-gel. Information regarding the sol-gel formulation can be obtained from Luna Innovations, Inc.(2) An additional test was done in which all of the fasteners were nylon in order to assess the contribution of crevice corrosion to the damage in the fastener holes.

The panels were exposed to ASTM B11733  continuous salt spray for 504 h (21 d). This testing environment is very common for evaluating the effects of corrosion damage within a galvanic couple.3,18,22-23,31-32,34  Although this test is known to be much more aggressive than most service conditions, and cannot predict specimen lifetimes,34  it has become a standard and results from it can be gathered and compared with other data from the literature. This test consists of a continuous vertical spray of 5% NaCl salt water, which creates a fog over the samples mounted at a 15° tilt, while the temperature and humidity remain constant.33 

Once the plates were removed from the testing environment, they were chemically stripped of the primer and coating, and the hardware was removed. Metallographic analysis was done by cutting the plate cross-sectionally through the holes. The cross-sectioned samples containing the exposed holes were polished to a 3-µm mirror finish and imaged using a Hirox RH 8800 optical microscope.

After the samples were imaged, a MATLAB algorithm developed specifically for the depth analysis of cross-sectional corrosion damage19  was used to quantify the results. The algorithm first converts the optical image into a binary image and extracts the outline of the sample based on a gray tolerance factor. Next, the algorithm plots the corroded surface of the sample as a function of distance. The profile of corrosion damage was used to calculate the approximate total volume of corrosion damage within the hole, by assuming that the damage exposed in the four cross-sections made was the same throughout the entire hole.

Finite Element Modeling

The modeling software used was COMSOL Multiphysics© (version 5.3a). A secondary current distribution model was used to calculate the total interface current on each surface in the NAVAIR panel. The three main pillars of this type of modeling are the assumptions used, the boundary conditions implemented, and the geometry built.

There were four main assumptions in the model which were made intentionally to reduce the computational power. First, the system was assumed to be in steady-state conditions over the entire testing period. This assumption has been made in finite element models before.4,6,10,12,18,35  The current distribution was calculated in the model and was converted to anodic charge in order to compare with experimental data. Rather than multiplying the current by the total amount of time which the experimental samples were exposed to the testing environment (504 h in our case), a 4-d “initiation” period for corrosion was accounted for. That initiation period came from an approximation from previous data found in literature, where steady-state corrosion appeared after 4 d in B117, as measured by a zero-resistance ammeter.23-24,30-31  As both the experimental panels and testing environment in this literature were similar to the current study, the same initiation time was used. Including this initiation period of corrosion helped make the “steady-state” assumption in the model more accurate.

The second assumption made was that migration of charged ions was the main mass transport mechanism, and therefore diffusion and convection could be ignored. In utilizing this assumption, the more general Nernst-Planck equation becomes the Laplace equation, which only takes into account the migration of ions. This simplification has been shown to give a good approximation of current densities in galvanic couples, and significantly decreases the computational power/time required.8,10,12-13,18,36 

The presence of a perfectly insulating coating system on the panel, with zero flux, was the third assumption in the model. This assumption meant that actively corroding AA7075-T6 was only considered to occur within the fastener holes and in the scribes on the surface. Note that the model cannot account for undercutting of the coating because of this assumption. At short exposure times, this assumption is valid, as the cathodic current will take the path of least resistance which is generally toward the initially bare AA7075-T6. However, at long exposure times, the aggressive solution is known to undercut surface coating and cause corrosion outside of the scribes.

The solution in the model was selected to represent the continuous salt spray environment to which the experimental panels were subjected. Therefore, the fourth assumption was that the water layer thickness in B117 was 4,000 μm. Multiple water layer thicknesses were considered, but previous work determined that a thickness of 4,000 μm most accurately represented a constant salt spray environment.37-38  Recent literature has also determined that all water layers above the natural convection boundary layer of 800 μm behave as bulk immersion, with minimal change in the total current.36  This indicates that although the water layer thickness in B117 testing has never experimentally been confirmed, as long as it is greater than 800 μm, the value input into the model does not change the overall results significantly. Therefore, a 4,000 μm water layer thickness was used in the model, along with a solution conductivity of 6 s/m.19  Investigations of experimentally measuring the water layer thickness in B117 will further be discussed in Part 2.

The boundary conditions input into the model were experimentally-derived via potentiodynamic scans of bare AA7075-T651, SS316, and Ti-6Al-4V. The experimental set up utilized a 250 mL glass corrosion flat cell with a 1 cm2 working electrode area and a 5 mL Luggin well. Platinum mesh was used as the counter electrode. Quiescent curves were generated in bulk 5 wt% NaCl at 35°C (as used in B117 testing),33  without any addition of gas to the solution. Deaerated curves were generated in the same environmental conditions as above, with research grade nitrogen bubbled in the solution at about 50 cc/min. The scan rate for the quiescent tests was 0.1 mV/s, while the scan rate for the deaerated tests was 5 mV/s. For all tests, the starting potential was held for 1 h, before scanning. Once the open-circuit potential was reached, another 1 h hold was initiated, before continuing to the end of the scan. The solution was not agitated through stirring for any of the tests described above.

The AA7075-T6 quiescent polarization curve was generated after the sample was held at an initial potential of –0.3 VSCE to activate localized corrosion on the panel, and then was scanned downwards in potential until reaching –1.2 VSCE. The potential range of the scan was aimed to encompass both the cathodic and anodic reactions on the AA7075-T6 surface. In this test, the solution at the surface of the AA7075-T6 was assumed to best represent the aggressive solution that develops in the occluded region of the hole during the B117 testing.

Deaerated cathodic SS316 and Ti-6Al-4V curves were generated, to represent the change in chemistry down the fastener hole. These curves were input into the model as boundary conditions for the occluded fastener surfaces. Quiescent cathodic SS316 and Ti-6Al-4V curves were generated to represent the surface of the fasteners. These curves were input into the model as boundary conditions for all fastener surfaces outside of the occluded region. The anodic behavior of these two alloys was not of particular interest in the current research and therefore was not tested.

The geometry in the model was constructed in the modeling software with the dimensions specified to match that of the experimental panels. The average gaps between the washers, fasteners, and panel were determined by metallographic analyses of an identically prepared plate assembly that had not been exposed to the corrosion chamber (see Figure 2). To represent the “dry installation” of the fasteners, the surfaces of the fastener holes were considered to be bare aluminum.

FIGURE 2.

(a) Cross-sectional analysis of assembled NAVAIR panel to assess gaps in the system and (b) gaps input into computational geometry.

FIGURE 2.

(a) Cross-sectional analysis of assembled NAVAIR panel to assess gaps in the system and (b) gaps input into computational geometry.

Close modal

The model was used to calculate the TAC, and both the net current and net current density at all positions where the aluminum metal was bare. The net current results were quantitatively related to the experimental data set by converting the current to charge using Faraday’s law for alloys, assuming a 408 h period of active localized corrosion. False-color net current density plots were used to qualitatively see the current distributions, and were compared with experimental data.

In all cases, the total anodic and cathodic currents of the entire panel were compared to verify that the conservation of charge was being observed. Throughout all of the calculations, the percent difference between those values never rose above 0.7%. The error tolerance level, for the current in the electrolyte, was set to 0.1%. The model was extended to several sets of experimental data found in the literature.23-24,30-31  These studies also used the NAVAIR panel, albeit with some differences in the geometry and required boundary conditions.

B117 Exposure Results

Upon removal of the panels after 504 h exposure to the test chamber, corrosion damage was easily visible on the NAVAIR panel with the bare fasteners. After the surface coating was stripped off and the hardware was removed, further corrosion damage became visible (Figures 3[a]and [b]). The majority of the damage appeared to be in the scribes. It is important to note that the surface surrounding the top left hole (which contained a bare Ti-6Al-4V fastener) received local damage where the coating blistered, likely due to a defect during installation. In the top right hole (which contained a bare SS316 fastener), there was no visible damage because the coating there did not blister. When comparing the scribes surrounding the two different fasteners, it can be seen that the scribes closest to the SS316 fastener (including those portions of the scribes on the Ti-6Al-4V fastener hole) suffered more severe damage. The damage in these scribes appeared to be localized at the tips, outside of the washers. We denoted these as corrosion “bulbs.”

FIGURE 3.

Cross sections of fastener holes in AA7075-T6 plate (a) after 504 h exposure to ASTM B117; (b) surface of panel after hardware and coatings were stripped off, dashed boxes represent each the location of each respective cross section; (c) bare Ti-6Al-4V fastener hole without scribes; (d) bare SS316 fastener hole without scribes; (e) bare Ti-6Al-4V fastener hole with scribes; and (f) bare SS316 fastener hole with scribes.

FIGURE 3.

Cross sections of fastener holes in AA7075-T6 plate (a) after 504 h exposure to ASTM B117; (b) surface of panel after hardware and coatings were stripped off, dashed boxes represent each the location of each respective cross section; (c) bare Ti-6Al-4V fastener hole without scribes; (d) bare SS316 fastener hole without scribes; (e) bare Ti-6Al-4V fastener hole with scribes; and (f) bare SS316 fastener hole with scribes.

Close modal

Down-hole analysis was conducted through metallographic cross-sectioning of the panels. Large amounts of intergranular corrosion (IGC) was seen down the bare fastener holes (Figures 3[c] through [f]). The top-right hole (which contained a bare SS316 fastener but no intentional coating defect) had deep pits and one IGC fissure reaching about 2 mm in length (Figure 4). This fastener hole showed no surface appearance of damage. This indicated that although the surface coating did not appear to be damaged, electrolyte was still able to enter the fastener hole and form an aggressive occluded cell.

FIGURE 4.

Micrograph of 2 mm fissure observed in AA7075-T6 hole with scribes containing bare SS316 fastener after exposure to B117 for 504 h; taken with (a) optical microscopy and (b) concentric backscatter (CBS) scanning electron microscopy.

FIGURE 4.

Micrograph of 2 mm fissure observed in AA7075-T6 hole with scribes containing bare SS316 fastener after exposure to B117 for 504 h; taken with (a) optical microscopy and (b) concentric backscatter (CBS) scanning electron microscopy.

Close modal

The top-left hole (which contained a bare Ti-6Al-4 V fastener and no intentional coating defect) appeared to suffer mechanical damage, as can be seen by the systematic hemispheres of damage (Figure 3[c]). It is unknown whether this mechanical damage helped initiate the surface coating failure. In all of the data presented in this research, mechanical damage was accounted for and filtered out. Significant exfoliation of the AA7075-T6 plate was observed on the surface under the Ti-6Al-4V washer (Figure 5). This type of corrosion is known to appear on AA7075-T6 in particular.39-40  The cross-sectioned images of the bare fastener holes were input into an algorithm to quantify the corrosion damage. The amount of anodic coulombic charge that must have been present for the given mass loss was calculated. The results showed that the damage down all of the bare fastener holes was nearly equivalent, independent of fastener type (Ti-6Al-4V or SS316) or surface defects (scribes or no scribes). This equivalence was in spite of the fact that that the SS316 fastener produced more cathodic current than the Ti-6Al-4V fastener.

FIGURE 5.

Exfoliation of AA7075-T6 surface observed near hole with no scribes containing bare Ti-6Al-4V after exposure to B117 for 504 h; taken with (a) CBS scanning electron microscopy, (b) secondary electron (SE) scanning electron microscopy, and (c) optical microscopy.

FIGURE 5.

Exfoliation of AA7075-T6 surface observed near hole with no scribes containing bare Ti-6Al-4V after exposure to B117 for 504 h; taken with (a) CBS scanning electron microscopy, (b) secondary electron (SE) scanning electron microscopy, and (c) optical microscopy.

Close modal

The surface damage on the NAVAIR panel containing sol-gel coated fasteners, after 504 h exposure to the test chamber, was much more limited than that of the bare fasteners panel. Small amounts of discoloration in the scribes were observed, but the surfaces with no scribe contained no corrosion attack (Figures 6[a] and [b]). The panel containing sol-gel coated fasteners was cross-sectioned with the same process as the bare fastener panel, in order to observe damage inside of the fastener holes. It was seen that there was no corrosion damage down either of the fastener holes with no intentional scribe defects (Figures 6[c] and [d]). However, there was a small amount of IGC observed down both of the fastener holes with surface scribed defects (Figures 6[e] and [f]). This damage down the holes was independent of the fastener type but was dependent on the surface defect.

FIGURE 6.

Cross-sections of sol-gel coated fastener holes in AA7075-T6 plate (a) after 504 h exposure to ASTM B117; (b) surface of panel after hardware and coatings were stripped off, dashed boxes represent each the location of each respective cross section; (c) sol-gel coated Ti-6Al-4V fastener hole without scribes; (d) sol-gel coated SS316 fastener hole without scribes; (e) sol-gel coated Ti-6Al-4V fastener hole with scribes; and (f) sol-gel coated SS316 fastener hole with scribes.

FIGURE 6.

Cross-sections of sol-gel coated fastener holes in AA7075-T6 plate (a) after 504 h exposure to ASTM B117; (b) surface of panel after hardware and coatings were stripped off, dashed boxes represent each the location of each respective cross section; (c) sol-gel coated Ti-6Al-4V fastener hole without scribes; (d) sol-gel coated SS316 fastener hole without scribes; (e) sol-gel coated Ti-6Al-4V fastener hole with scribes; and (f) sol-gel coated SS316 fastener hole with scribes.

Close modal

Modeled Secondary Current Distribution Results

The model was assembled with the assumptions, boundary conditions, and geometry described above. Figure 7 shows all of the polarization curves which were used in the present model, and the respective surfaces in the model to which they were assigned. False-color current density plots were generated for a qualitative comparison to the surface appearance of the NAVAIR panels (Figure 8). Only the active AA7075-T6 surfaces are shown on the plot for convenience, although all fasteners have negative cathodic current in the system. The false color plot showed the peak current density on the tip of the scribes outside of the washers.

FIGURE 7.

(a) Polarization curves for quiescent (solid lines) and deaerated (dashed lines) cathodic and anodic alloys; surfaces in blue represent (b) bare AA7075-T6 (all blue surfaces used the activated AA7075 polarization curve as a boundary condition); (c) Ti-6Al-4V fasteners and washers; and (d) SS316 fasteners and washers. Note: for both (c) and (d), the shafts used the deaerated polarization curves as boundary conditions, while the washers and head of the fasteners used the quiescent polarization curves as boundary conditions.

FIGURE 7.

(a) Polarization curves for quiescent (solid lines) and deaerated (dashed lines) cathodic and anodic alloys; surfaces in blue represent (b) bare AA7075-T6 (all blue surfaces used the activated AA7075 polarization curve as a boundary condition); (c) Ti-6Al-4V fasteners and washers; and (d) SS316 fasteners and washers. Note: for both (c) and (d), the shafts used the deaerated polarization curves as boundary conditions, while the washers and head of the fasteners used the quiescent polarization curves as boundary conditions.

Close modal
FIGURE 8.

False-color current density plot of exposed AA7075-T6 substrate under 4,000 μm water layer thickness with active Ti-6Al-4V fastener and active SS316 fastener installed; solid circles highlight peak current density at tip of scribes, qualitatively agreeing with experimental data.

FIGURE 8.

False-color current density plot of exposed AA7075-T6 substrate under 4,000 μm water layer thickness with active Ti-6Al-4V fastener and active SS316 fastener installed; solid circles highlight peak current density at tip of scribes, qualitatively agreeing with experimental data.

Close modal

To quantify the damage, the current density was integrated over the bare AA7075-T6 surface area in order to obtain the average current. This summation was done for each bare AA7075-T6 fastener hole individually. These results were then converted to columbic charge considering a 4-d initiation period, i.e., it was assumed that after 4 d, the corrosion initiated and immediately propagated at steady-state for the remaining 17 d. The length of this initiation period came from literature with similar experimental test parameters as our own.23-24,30-31 

Investigations of Damage Down Ti-6Al-4V Fastener Hole

To follow up on the experimental results observed, three tests were conducted to determine why the holes containing bare Ti-6Al-4V fasteners were experiencing severe corrosion. As it is known that Ti-6Al-4V creates a less strong galvanic couple than SS316,23,31-32  these results were intriguing. An experimental test was conducted to investigate the possibility of crevice corrosion in the NAVAIR panel. Two modeling tests were conducted to evaluate the level of interactions between the SS316 and Ti-6Al-4V fasteners.

The goal of the experimental test was to determine whether the damage observed in the hole containing Ti-6Al-4V fasteners was due to crevice corrosion of the AA7075-T6 in the occluded cell, independent of the nature of the fastener. Nylon fasteners and hardware of the same dimensions as the SS316 and Ti-6Al-4V fasteners and hardware were installed in the AA7075-T6 NAVAIR panel, providing similar occluded regions but without any galvanic interactions. The test panel was exposed to the same test conditions as described above (504 h in B117). Metallographic analysis showed a complete lack of corrosion damage (see Figure 9), demonstrating that crevice corrosion was not contributing damage to this test assembly.

FIGURE 9.

(a) Nylon fasteners assembled in AA7075-T6 NAVAIR plate after 504 h exposure to ASTM B117; (b) hardware and coatings stripped off of panel; and (c) no corrosion visible in cross section of fastener hole.

FIGURE 9.

(a) Nylon fasteners assembled in AA7075-T6 NAVAIR plate after 504 h exposure to ASTM B117; (b) hardware and coatings stripped off of panel; and (c) no corrosion visible in cross section of fastener hole.

Close modal

The first modeling test was conducted to determine the level of cathodic current which the Ti-6Al-4V fastener was contributing toward the galvanic couple. Ti-6Al-4V boundaries were set to have zero flux while the geometry and remaining boundary conditions stayed the same. In this case, the only galvanic current that could occur would be due to the SS316 fasteners. The model showed that the scribes surrounding the inert Ti-6Al-4V fastener and the fastener hole experienced nearly the same amount of current density as before being inert. This supported the proposal that the SS316 fasteners must be communicating with the Ti-6Al-4V holes via the water layer, and that the contribution of the Ti-6Al-4V fasteners was almost negligible (Figure 10). Interactions between fasteners in the NAVAIR configuration have been observed in previous work.3,19,32  However, this finding shows that the Ti-6Al-4V fasteners are not contributing a significant amount to the corrosion of the aluminum, but are just hiding the true extent of damage caused by the SS316 fasteners down the fastener holes.

FIGURE 10.

Total current down the fastener holes with Ti-6Al-4V surfaces active vs. nonactive.

FIGURE 10.

Total current down the fastener holes with Ti-6Al-4V surfaces active vs. nonactive.

Close modal

The second modeling test was conducted to verify the theory of the SS316 fastener galvanically coupling with bare AA7075-T6 surrounding the Ti-6Al-4V fastener. Both of the fasteners had active boundary conditions, as well as the AA7075-T6 surfaces. The water layer thickness was changed from 4,000 μm to 800 μm, while all other test parameters remained the same. A water layer thickness of 800 μm is still considered “bulk” electrolyte (as it is at the threshold of the boundary layer thickness),36  however, the migration of ions should feel increased resistance from the 4,000 μm to 800 μm water layer thickness. The results showed an asymmetry in the scribes surrounding the Ti-6Al-4V fastener (see Figure 11). This indicates that the SS316 fastener is contributing the majority of the galvanic current surrounding the Ti-6Al-4V fastener, however, with a thinner water layer, not all surfaces of the bare AA7075-T6 are as easy to couple with.

FIGURE 11.

False-color current density plot of exposed AA7075-T6 substrate under 800 μm water layer thickness with active Ti-6Al-4V fastener and active SS316 fastener installed; dashed circles highlight asymmetry in current distribution on the scribes, indicating interaction between the scribe and SS316 fastener.

FIGURE 11.

False-color current density plot of exposed AA7075-T6 substrate under 800 μm water layer thickness with active Ti-6Al-4V fastener and active SS316 fastener installed; dashed circles highlight asymmetry in current distribution on the scribes, indicating interaction between the scribe and SS316 fastener.

Close modal

From a structural perspective, damage down fastener holes is very important because of the increase to the inherent stress intensification of the hole itself by the corrosion damage. Damage on the order of hundreds of micrometers, as was observed in our samples, could serve as crack initiators if high stress was applied that had a component perpendicular to the corrosion damage or if bending stresses existed. The surface appearance was not a reliable predictor of the corrosion damage down the fastener holes. However, it was seen that with the application of sol-gel on the noble fasteners, the severe corrosion damage down the fastener holes was greatly mitigated. The modeled results agreed qualitatively with the damage distribution on the surface, and quantitatively with the total damage down the fastener holes. The model was able to predict the complex interaction between the SS316 fastener and the bare AA7075-T6 panel.

Comparison of Damage Down Holes for Bare vs. Sol-Gel Coated Fasteners in the NAVAIR Plate

Previous work using similar NAVAIR plates has shown that the SS316 fasteners produce nearly double the cathodic current of the Ti-6Al-4V fasteners.23,31-32  This is known to be due to the slow cathodic kinetics on the Ti-6Al-4V surface, although its potential is higher relative to SS316. Feng, et al.,23,32  thus concluded that SS316 fasteners will therefore cause more severe corrosion attack. In theory, an isolated SS316 fastener would cause corrosion attack on the panel which is twice as severe as an isolated Ti-6Al-4V fastener, as indicated by the total current measurements. However, close proximity of fasteners relates to a more realistic situation, which is represented by the NAVAIR panel design. The research presented here indicates that when fasteners are put in this close proximity, the damage distribution can no longer be predicted by the measured currents on the individual fasteners. It was seen that, although the current from the SS316 fastener was nearly double that of the Ti-6Al-4V fastener (as known from literature), the damage down the fastener holes was all of the same order of magnitude. The measured cathodic current in this configuration is a poor proxy for damage within the fastener hole, which is of more structural importance. It should also be noted that crevice corrosion does not appear to play any role in the damage within the holes as shown by the experiment with the nylon fasteners (Figure 9). The SS316 fastener’s ability to couple with the AA7075-T6 panel, as a function of distance, is investigated further in Part 2.

Figure 12 compares the overall damage down the sol-gel fastener holes to the bare fastener holes, through anodic charge comparisons. Because the holes with the sol-gel fasteners suffered visible corrosion only on the scribed holes, the damage in the nonscribed holes was not reported on the bar chart and is approximately zero. The error bars represent the minimum and maximum damage seen down the cross section of the holes. Due to the nonuniform corrosion down the fastener holes (see Figure 3), the errors bars could become very large.

FIGURE 12.

Comparison of damage down bare vs. sol-gel fastener holes after 504 h exposure to ASTM B117.

FIGURE 12.

Comparison of damage down bare vs. sol-gel fastener holes after 504 h exposure to ASTM B117.

Close modal

The sol-gel coating reduced the total anodic charge by a factor of approximately 84% after 504 h of exposure to the aggressive B117 testing. A statistical analysis of the damage was also performed for the two fastener systems. The maximum depth of corrosion damage down the bare fastener holes ranged from about 280 μm to 370 μm, whereas the maximum depth of corrosion damage down the sol-gel coated fasteners holes ranged from about 180 μm to 200 μm. The area corroded however, differed by one order of magnitude, with the bare fasteners having approximately 0.21 mm2 and the sol-gel fasteners having approximately 0.069 mm2 worth of damage. These results show that at least up to 504 h in the accelerated salt spray chamber, the sol-gel coating limited cathodic current, thereby stifling the galvanic coupling which causes the corrosion initiation and propagation.

Comparison Between Model and Experimental Results

The false-color current density plots generated in the model agreed qualitatively with observations of the experimental test panels, where severe bulb-shaped corrosion was seen at the ends of the scribes (see Figure 8). The model demonstrates that the end of the scribe is most accessible for the cathodic current and is thus an area of high current density. The current density decreases as one moves closer to the washer. Details regarding the current distribution are discussed in Part 2.

The results from the model were compared with down-hole experimental data gathered from metallographic analysis(3) (see Figure 13). The experimental data of 504 h of exposure compared well with the computational data for the bare fasteners, supporting the assertion that the use of the simplified model (i.e., one based on the Laplace Equation) works well when the boundary conditions are accurate, representing the electrochemical kinetics in the expected occluded solution. As observed experimentally, the computed damage in the four holes with the bare fasteners was similar. The differences in the computational results between Ti-6Al-4V and SS316 is within the error of the cross-sectional methodology.

FIGURE 13.

Comparison of modeling vs. experimental data of charge down bare fastener holes after 504 h exposure to B117; solid bars represent experimental data while the striped bars represent computational data.

FIGURE 13.

Comparison of modeling vs. experimental data of charge down bare fastener holes after 504 h exposure to B117; solid bars represent experimental data while the striped bars represent computational data.

Close modal

To further demonstrate the utility of this modeling approach, the geometry was adapted to represent the experimental procedure described in other work. The boundary conditions remained the same as the testing above, however the geometries were changed slightly, along with the arrangement of fasteners. One paper tested two bare SS316 and two bare Ti-6Al-4V fasteners in a single NAVAIR panel,31  similar to the configuration described above, whereas a different paper and a dissertation tested four bare SS316 fasteners.24,30  One other paper tested both of the experimental scenarios above, with both SS316 and Ti-6Al-4V fasteners.23  One of the largest differences in the experimental set up was the size of the SS316 washers. The ones used in the literature were much smaller than the ones used in this research, which would affect the total amount of cathodic current available due to the decrease in the cathodic surface area.

The model was run with the new considerations described above and the current density was integrated over all four fasteners, in order to compare with the experimental data taken with a zero-resistance ammeter. The results are plotted against each other in Figure 14. The model is seen to give similar results to the experimental data from literature, during steady-state corrosion. Note that the data from Boerstler24  in Figure 14(b) used the NAVAIR plate without scribes, and therefore has a longer corrosion initiation period, however, steady-state corrosion is similar to those with scribes. The quality of the agreement between the computations and data generated independently is strong evidence of the validity of the modeling approach.

FIGURE 14.

Comparison of galvanic current in the system calculated with the model vs. calculated in literature with ZRA: (a) 2 SS316 fasteners and 2 Ti-6Al-4V fasteners installed in AA7075-T6 panel; and (b) 4 SS316 fasteners installed in AA7075-T6 panel. Note: steady-state corrosion begins in both experimental scenarios around 4 d.23-24,30-31 

FIGURE 14.

Comparison of galvanic current in the system calculated with the model vs. calculated in literature with ZRA: (a) 2 SS316 fasteners and 2 Ti-6Al-4V fasteners installed in AA7075-T6 panel; and (b) 4 SS316 fasteners installed in AA7075-T6 panel. Note: steady-state corrosion begins in both experimental scenarios around 4 d.23-24,30-31 

Close modal

Although the results produced by the model compared well with experimental data, both in-house and from literature, the approach does have limitations that should be appreciated. The model assumes a perfect coating on the exterior surfaces, and thus as constituted cannot be used to assess the effects of different coatings or surface treatments. Experimentally, clear distinctions have been seen in the literature between coatings containing chromium and nonchromium coatings (Figure 14). The steady-state assumption in the model requires that either the initiation period be known or the test is long enough that the initiation time is small in comparison. Finally, the electrochemical kinetics used are critical to the results of the model. In the present work, the anodic activation of the AA7075-T6 was used to better capture the active localized corrosion kinetics. Future work will be done on determining whether the corrosion down the fastener hole is more sensitive to changes in the cathodic or anodic boundary conditions.

  • This work shows the importance of investigating the corrosion damage within fastener holes of galvanic panels. A lack of corrosion damage on the panel surface does not correlate to a lack of damage within the hole. All studies using the NAVAIR galvanic panel or similar designs to assess the performance of coatings should assess damage within the holes.

  • Damage down bare and sol-gel coated fastener holes was independent of fastener type due to interactions between the SS316 fasteners and all bare AA7075-T6 surfaces throughout the panel. FEM calculations showed this independence and interaction as well. Crevice corrosion was ruled out as the cause of the damage.

  • Application of a specific sol-gel formulation to cathodic fasteners greatly reduced corrosion both down fastener hole and on the surface of the panel by reducing the cathodic current available to drive the anodic dissolution.

  • Damage down fastener holes was reduced by 84% after 504 h of exposure to B117 with use of sol-gel coated fasteners as opposed to bare fasteners.

  • The finite element model developed was shown to give a good approximation of both the charge down fastener holes, and of total current of fasteners found in literature at times before severe coating degradation. These results indicated that the simplifying assumptions made in the model did not change the system significantly.

(1)

UNS numbers are listed in Metals & Alloys in the Unified Numbering System, published by the Society of Automotive Engineers (SAE International) and cosponsored by ASTM International.

Trade name.

(3)

Mass loss for samples of this size (6 in × 3 in × 0.25 in = 12 g) due to localized corrosion is a poor metric as the maximum damage loss observed would represent a mass loss of 0.08%.

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of Office of Naval Research. This work is supported by the Office of Naval Research under Contract No. N68335-16-C-0121 (Mr. William Nickerson) through a subcontract with Luna Innovations, Inc.

1.
P.
Rambabu
,
N.
Eswara Prasad
,
V.V.
Kutumbara
,
R.J.H.
Wanhill
,
“Aluminum Alloys for Aerospace Applications,”
in
Aerosp. Mater. Mater. Technol.
,
eds.
N.
Eswara Prasad
,
R.J.H.
Wanhill
(
Singapore
:
Springer
,
2018
),
p
.
29
52
.
2.
“Aluminum 7075–T6; 7075-T651,”
ASM Aerospace Specification Metals Inc
(
n.d.
), http://asm.matweb.com/search/SpecificMaterial.asp?bassnum=ma7075t6 (
June
11
,
2018
).
3.
C.A.
Matzdorf
,
W.C.
Nickerson
,
B.C.
Rincon Tronconis
,
G.S.
Frankel
,
L.
Li
,
R.G.
Buchheit
,
Corrosion
69
,
12
(
2013
):
p
.
1240
1246
.
4.
D.
Mizuno
,
R.G.
Kelly
,
Corrosion
69
,
7
(
2013
):
p
.
681
692
.
5.
C.
Liu
,
R.G.
Kelly
,
Electrochem. Soc. Interface
23
,
4
(
2014
):
p
.
47
51
.
6.
M.
Turner
,
E.
Gutierrez-Miravete
,
“Modeling Galvanic Corrosion,”
COMSOL Conf.
,
held October 9-11
(
Boston, MA: COMSOL Multiphysics
,
2013
),
p
.
1
4
.
7.
C.
Liu
,
P.
Khullar
,
R.G.
Kelly
,
J. Electrochem. Soc.
166
,
6
(
2019
):
p
.
153
161
.
8.
C.
Liu
,
V.N.
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, paper no. 5579
(
Houston, TX
:
NACE International
,
2015
).
9.
A.D.
King
,
J.S.
Lee
,
J.R.
Scully
,
J. Electrochem. Soc.
163
,
7
(
2016
):
p
.
342
356
.
10.
P.
Kramer
,
T.
Curtin
,
M.
Merrill
,
M.
Kim
,
F.
Friedersdorf
,
R.
Adey
,
“Atmospheric Corrosion Measurements to Improve Understanding of Galvanic Corrosion of Aircraft,”
CORROSION
2018
,
paper no. 10968
(
Houston, TX
:
NACE
,
2018
),
p
.
1
11
.
11.
D.
Snihirova
,
D.
Höche
,
S.
Lamaka
,
Z.
Mir
,
T.
Hack
,
M.L.
Zheludkevich
,
Corros. Sci.
157
(
2019
):
p
.
70
78
.
12.
R.
Adey
,
A.
Peratta
,
J.
Baynham
,
T.
Curtin
,
“Predicting Cumulative Galvanic Corrosion Damage in Aircraft Structures,”
AVT-303 Research Workshop on Corrosion Management
,
held December 10-12
(
Athens, Greece
:
North Atl. Treaty Organ. Sci. Technol. Organ.
,
2018
),
p
.
1
14
.
13.
N.
Murer
,
R.
Oltra
,
B.
Vuillemin
,
O.
Néel
,
Corros. Sci.
52
,
1
(
2010
):
p
.
130
139
.
14.
T.M.
Amorim
,
C.
Allély
,
J.P.
Caire
,
“Modelling Coating Lifetime: First Practical Application for Coating Design,”
COMSOL Conf.
,
held November 4-6
(
Hannover, Germany
:
COMSOL Multiphysics
,
2008
).
15.
N.
Murer
,
N.A.
Missert
,
R.G.
Buchheit
,
J. Electrochem. Soc.
159
,
6
(
2012
):
p
.
265
276
.
16.
O.
Dolgikh
,
H.
Simillion
,
S.V.
Lamaka
,
A.C.
Bastos
,
H.B.
Xue
,
M.G.
Taryba
,
A.R.
Oliveira
,
C.
Allely
,
B.
Van Den Bossche
,
K.
Van Den Bergh
,
J.
De Strycker
,
J.
Deconinck
,
Mater. Corros.
70
(
2018
):
p
.
780
792
.
17.
C.
Liu
,
R.G.
Kelly
,
Corrosion
75
,
11
(
2019
):
p.
1285
1299
.
18.
C.
Liu
,
G.W.
Kubacki
,
R.G.
Kelly
,
Corrosion
75
,
5
(
2019
):
p
.
465
473
.
19.
C.
Tattersall
,
“Modeling and Analysis of Geometrically Complex Corrosion”
(
Ph.D. diss.
,
University of Virginia
,
2018
).
20.
R.
Jones
,
M.
Lo
,
D.
Peng
,
A.
Bowler
,
M.
Dorman
,
M.
Janardhana
,
N.S.
Iyyer
,
Meccanica
50
,
2
(
2015
):
p
.
517
532
.
21.
V.
Rafla
,
A.
Davenport
,
J.R.
Scully
,
Corrosion
71
,
11
(
2015
):
p
.
1300
1303
.
22.
A.
Moran
,
J.
Jennings
,
H.
Nee
,
S.
Pearson
,
B.
Clark
,
R.S.
Lillard
,
Corrosion
75
,
5
(
2019
):
p
.
484
498
.
23.
Z.
Feng
,
G.S.
Frankel
,
C.A.
Matzdorf
,
J. Electrochem. Soc.
161
,
1
(
2014
):
p
.
42
49
.
24.
J.T.
Boerstler
,
“Corrosion Degradation of Coated Aluminum Alloy Systems through Galvanic Interactions”
(
Ph.D. diss.
,
The Ohio State University
,
2018
).
25.
J.T.
Burns
,
J.M.
Larsen
,
R.P.
Gangloff
,
Fatigue Fract. Eng. Mater. Struct.
34
,
10
(
2011
):
p
.
745
773
.
26.
T.J.
Harrison
,
B.R.
Crawford
,
G.
Clark
,
M.
Brandt
,
Adv. Mater. Res.
891
(
2014
):
p
.
242
247
.
27.
T.J.
Harrison
,
B.R.
Crawford
,
M.
Brandt
,
G.
Clark
,
Comput. Mater. Sci.
84
(
2014
):
p
.
74
82
.
28.
G.K.
Cole
,
G.
Clark
,
P.K.
Sharp
,
“The Implications of Corrosion with Respect to Aircraft Structural Integrity,”
DSTO Aeronautical and Maritime Research Laboratory
,
DSTO-RR-0102
,
1997
.
29.
M.
Lo
,
R.
Jones
,
A.
Bowler
,
M.
Dorman
,
D.
Edwards
,
Fatigue Fract. Eng. Mater. Struct.
40
,
10
(
2017
):
p
.
1664
1675
.
30.
Z.
Feng
,
J.
Boerstler
,
G.S.
Frankel
,
C.A.
Matzdorf
,
Corrosion
71
,
6
(
2015
):
p
.
771
783
.
31.
S.
Wang
,
J.
Boerstler
,
G.S.
Frankel
,
Corrosion
74
,
5
(
2018
):
p
.
499
508
.
32.
Z.
Feng
,
G.S.
Frankel
,
Corrosion
70
,
1
(
2014
):
p
.
95
106
.
33.
ASTM B117
,
“Standard Practice for Operating Salt Spray (Fog) Apparatus”
(
West Conshohocken, PA
:
ASTM International
,
1994
).
34.
Z.
Feng
,
G.S.
Frankel
,
W.H.
Abbott
,
C.A.
Matzdorf
,
Corrosion
72
,
3
(
2016
):
p
.
342
355
.
35.
Z.
Haque
,
B.A.
Clark
,
R.S.
Lillard
,
Corrosion
74
,
8
(
2018
):
p
.
903
913
.
36.
C.
Liu
,
J.
Srinivasan
,
R.G.
Kelly
,
J. Electrochem. Soc.
164
,
13
(
2017
):
p
.
845
855
.
37.
C.
Liu
,
“Assessment of the Utility of Laplace’s Equation for Modeling of Localized Corrosion”
(
Ph.D. diss.
,
University of Virginia
,
2018
).
38.
L.M.
Blohm
,
“Galvanic Coupling of AA5XXX-H116 and CDA 706 Utilizing Laboratory, Outdoor, and Accelerated Testing”
(
Ph.D. diss.
,
University of Virginia
,
2018
).
39.
A.S.
El-amoush
,
Mater. Chem. Phys.
126
,
2
(
2011
):
p
.
607
613
.
40.
F.
Andreatta
,
H.
Terryn
,
J.H.W.
De Wit
,
Electrochim. Acta
49
,
17-18
(
2004
):
p
.
2851
2862
.