Electrochemical Characterization of Galvanic Couples Under Saline Droplets in a Simulated Atmospheric Environment

Galvanic corrosion in atmospheric environments has long been acknowledged in the design and maintenance of aircraft systems, as evidenced by frequent reference in design guidance documents to avoid dissimilar metallic connections wherever possible. Despite this guidance, it is difficult in the design and construction of aircraft structures to completely avoid problematic galvanic materials combinations because of the need for careful management of structure strength-to-weight ratio and the absence of galvanically compatible fastening hardware. While multilayer coating systems can provide long periods of sustained corrosion protection for aircraft components, their inevitable failure raises time of service questions subsequent to coating compromise. In this regard, recent accelerated galvanic atmospheric corrosion tests demonstrated aluminum corrosion rates were most dependent on the cathode material,^1-3  with the coating primarily dictating the corrosion morphology. Extensive knowledge of alloy polarization behavior, both oxidative and reductive pathways, in atmospheric corrosive environments is therefore desirable.

The bimetallic couples of primary interest in this work are those between aluminum alloys, found in aircraft skins and frames, with stainless steel (SS) or titanium alloy fasteners. In particular martensitic stainless steels are commonly used materials for aircraft fasteners because of their high strength, toughness, hardness, and good level of corrosion resistance. Although Ti alloys are also attractive fastener materials because of their corrosion resistance and lower specific gravity, higher cost and lower shear strength have made these alloys less commonplace than their SS counterparts. Despite the role these materials play as fasteners and their relative nobility with respect to aluminum alloys, limited electrochemical cathodic polarization or oxygen reduction reaction (ORR) kinetic data are available for these alloys.^4-10  In the past this could be attributed, in part, to a reliance upon the galvanic series in seawater.^11-12  More recent work presents only a snapshot of stainless steels and titanium alloys that often overlooks relevant atmospheric conditions and their electrochemical behavior when galvanically coupled.^7,13-18 

In the context of atmospheric corrosion cells, polarization behavior itself is nontrivial because of the physical and IR constraints imposed upon the electrode configuration by thin films and droplet electrolytes.^17,19-29  A handful of cell configurations have been developed to examine the electrochemistry within thin films as these electrolytes approach small length scales, altering transport behavior of reaction precursors and complicating the experimental requirements.^30-31  Stratmann, et al., pioneered the use of a scanning Kelvin probe (SKP) to examine surface potential of metals under thin films with an oxygen monitor to determine the ORR rate.^30,32-34  Lyon and coworkers developed a three-electrode cell by embedding the electrodes in an epoxy and a well to submerge the reference electrode.^35-38  Several variants have been adopted by other groups to look at electrochemistry under thin films.^39-41 

A significant quantity of work has gone into developing the understanding of corrosion in sessile droplets on a single alloy surface, often referred to as an Evan’s cell,^{18,26,28,42-49}  while relatively few reports examine electrochemical reactions in a droplet to lead to a better understanding of galvanic atmospheric corrosion. Cathodic polarization of macroscopic 0.1 M NaCl drops on steel and stainless steel by direct immersion of counter and reference electrodes into the droplet was used to demonstrate that cathodic current and corrosion current depended upon the triple edge boundary length between alloy/electrolyte/gas.⁵⁰  Direct immersion of noble metal wires into droplets has also been used to probe droplet electrochemistry.^51-52  Recently, microcapillary salt bridges were exploited to polarize mild steel under microscale droplets to show enhanced oxygen reduction rates and passivation during cathodic and anodic polarization, respectively.³¹  Still, an electrochemical study of galvanic corrosion within droplets has not been explored despite the importance that kinetics play for bimetallic couples and the difference in environments between marine and atmospheric corrosion.

This work investigated galvanic atmospheric corrosion involving titanium and martensitic stainless steel alloys coupled with structural aircraft aluminum alloys UNS A92024⁽¹⁾ and UNS A97075 through droplet polarization scans. Coupon specimens of each alloy were evaluated in an uncoated condition in several bulk electrolytes to identify relevant reaction pathways and quantify the effects of chloride, protons, and oxygen. Microcapillary salt bridges were used to probe the electrochemistry of the droplets, and microelectrodes were used to measure droplet chloride concentrations. Standard corrosion cells were used to interrogate the same alloys in bulk electrolytes for comparison between bulk and droplet electrolyte behavior. Optical images of droplets during polarization and bulk electrolyte polarization scans were utilized to understand unique micro-droplet features that arise during droplet polarization scans.

All experiments utilized 2.5 cm × 2.5 cm or 5 cm × 5 cm coupons of UNS A92024 (AA2024-T3), UNS A97075 (AA7075-T6), UNS S13800 (CRES 13-8 PH1000), UNS S17400 (CRES 17-4 PH1025), UNS R56400 (Ti-6Al-4V), and UNS S46500 (Custom 465 H950^†) cut from panel or bar stock. The surface of each coupon was brought to a 600 grit SiC finish, rinsed with distilled water, cleaned in an ultrasonic bath, and dried with nitrogen just prior to cell assembly. The configuration shown in Figure 1 was used for droplet testing. Coupons were secured to an optical breadboard with plastic fixtures. The micropipettes were held in place with a three-axis micromanipulator. The entire assembly was placed in a temperature- and humidity-controlled environmental chamber with a cable gland for potentiostatic cable pass through. Coupons were introduced to the preset environmental chamber immediately prior to depositing a droplet. Pipettes were used to place droplets of known electrolyte composition and volume on the test alloy or salt crystals were placed on the surface and deliquesced into droplets at high humidity. For the droplet experiments, images of the droplet were taken to determine initial droplet size and at select intervals to understand micro-droplet formation.

Polarization scans within the droplets were performed using micropipettes fabricated in-house. Standard borosilicate glass Pasteur pipettes were manipulated using a Nashirige PC-10^† pipette puller with a two-step protocol to obtain tip diameters between 10 μm and 40 μm. The pipettes were then silanized with N,N-dimethyltrimethylsilylamine at 200°C for 2 h to make the glass hydrophobic. Liquid exchange through the tip opening was controlled by back filling the tip with NaCl saturated agar and the pipette was subsequently filled with saturated NaCl to complete the micro-salt bridge and reduce resistive losses, which were in the range of 50 kΩ to 200 kΩ. The Pt wire counter electrode and a fritted Ag/AgCl reference electrode were inserted into the body of the pulled pipette. The reference electrode was measured against a saturated calomel electrode (SCE) prior to each experiment and all potentials are reported vs. SCE (V_SCE). Polarization scans were preceded by 18 h at open-circuit potential (OCP) and a potentiostatic electrochemical impedance spectroscopy scan. The impedance was measured at OCP from 100 kHz to 0.1 Hz with an amplitude of 0.01 V. The impedance was used to determine the resistance in the system and apply a post scan IR correction to all data. The aluminum alloys were anodically polarized at 0.167 mV/s from OCP −0.02 V to −0.4 V_SCE and −0.6 V_SCE for UNS A92024 and UNS A97075, respectively. The Ti alloy and stainless steels were polarized cathodically at 0.167 mV/s from OCP + 0.02 V to −1.4 V_SCE or −2.0 V_SCE. Diffusion limited onset potential was determined by intersection of the extrapolated linear regions of the activation and diffusion limited current, where the latter was defined by a stagnant current range greater than 0.1 V. Micro-droplet formation, which will be described in the Results and Discussion sections, led to changes in the droplet surface area at very negative potential. The configuration in these cells did not permit accurate measurement of micro-droplet interfacial area between the droplets and the substrate and no action was taken to correct current densities to changes in interfacial area.

A microscale chloride sensor was fashioned from Ag wire by electrochemical means. A notch was mechanically introduced into a 250 μm diameter silver wire with a razor blade. The wire, including the notched region, was immersed in 0.1 M KNO₃, concentric to a Pt wire counter electrode in the shape of a circle at the electrolyte surface. The wire was etched in a two-electrode configuration at 5 V until the current dropped precipitously (~2 min), concurrent with wire breakage and micro-tip formation. The wire was subsequently anodized for 40 s to 120 s in 0.1 M HCl using the same configuration for etching at 0.5 V. The chloride sensor was calibrated against Ag/AgCl (saturated NaCl) reference electrode contained in a micropipette assembly in bulk NaCl electrolytes of various concentrations.

Bulk electrolyte scans were conducted on coupons mounted in Gamry Paracells^† and stabilized for 18 h at OCP to the following electrolytes: 0.6 M NaCl with ambient aeration, 4 M NaCl at ambient aeration, Ar purged 0.6 M NaCl, 0.6 M NaCl + 10 mM HCl, 0.6 M NaCl + 10 mM H₂SO₄, and 0.6 M NaCl + 5 mM HNO₃ + 5 mM H₂SO₄. The aluminum alloys were anodically polarized at 0.167 mV/s from OCP −0.02 V to −0.4 V_SCE and −0.6 V_SCE for UNS A92024 and UNS A97075, respectively. The Ti alloy and stainless steels were polarized cathodically at 0.167 mV/s from OCP + 0.02 V to −1.4 V_SCE. Unless otherwise indicated, the temperature and aeration condition were from the ambient environment.

The droplets for the anode materials were formed in a different manner from the ones used on the cathode materials in that, on the anode materials, the droplets were formed by deliquescing NaCl salt crystals. This process was used because the hydrophilic nature of the Al alloys would cause deposited 0.6 M NaCl droplet to spread significantly over the course of the OCP timeframe and result in a considerable change in surface area and droplet composition before the polarization experiments could be conducted, thereby making droplet contact difficult. The deliquescent approach to droplet formation produced the same electrolyte composition as deposition of more dilute droplets because of evaporation and condensation as the electrolyte sought to establish equilibrium with the relative humidity (RH).

The aluminum alloy polarization scans in Figure 2, measured in droplets at 80% RH and 23.8°C, displayed small Tafel slopes and larger exchange currents compared to the stainless steels and UNS R56400, in agreement with previous reports. The intersection points between anodic and cathodic scans were used to predict the corrosion current and potential for galvanic couples corresponding to those materials.⁵³  The rapid increase in corrosion current in response to small changes in potential for the Al alloys indicated that in galvanic couples, the corrosion rate of these alloys would be under cathodic control, also consistent with previous reports.³  In Figure 2 it can be seen that the predicted galvanic corrosion current for UNS R56400 coupled with the Al alloys was an order of magnitude smaller than the predicted corrosion current for couples with UNS S46500, UNS S13800, and UNS S17800. Importantly, the ~0.15 V corrosion potential difference between UNS A92024 and UNS A97075 also contributes to an order of magnitude increase in the galvanic corrosion current on UNS A97075, as the reduction current on the cathode materials is still increasing in this potential range.

The cathodic polarization curves within droplets on three stainless steels and UNS R56400 are also shown in Figure 2. The polarization curves of the stainless steels overlaid one another closely, while the polarization curve for the Ti alloy, UNS R56400, showed the following significant differences from the stainless steels.

• •

  The OCP for the stainless steels ranged from −0.04 V_SCE to −0.08 V_SCE, while that of UNS R56400 was −0.23 V_SCE.

• •

  The ORR activation region of the polarization curves for the stainless steels exhibited Tafel slopes of 0.06 V/decade to 0.08 V/decade transitioning to diffusion control at approximately −0.25 V_SCE, whereas the UNS R56400 Tafel slope was 0.19 V/decade and did not transition to diffusion control until −0.85 V_SCE.

• •

  Although the cathodic current reached a diffusion controlled limit around 10 μA/cm² to 40 μA/cm² for the stainless steels, the current densities subsequently increased near the galvanic coupling potentials for UNS A92024-T3. The Ti alloy also displayed an order of magnitude increase in cathodic current as the potential was pushed more negative of the current plateau, < −0.5 V_SCE.

The cathodic current increases above the diffusion limit on the tail of the droplet polarization curves was investigated further on UNS S46500 by optical imaging of the droplet behavior during polarization to IR corrected potentials below −1 V_SCE, as shown in Figure 3. At potentials above −0.66 V_SCE, optical images in Figures 3(a) through (c) show no apparent change in the droplet shape or periphery at this resolution. However, as the potential continues to decrease, subsequent images display evidence of droplet expansion around the perimeter of the original droplet. Another contributor to the observed current increase was thinning of the droplets as the electrolyte expanded laterally. Previous reports with thin film electrolytes have suggested that ORR rates increased until the electrolyte film thickness reached 10 μm to 30 μm, at which point oxygen transfer across the air/electrolyte interface becomes slower than oxygen transport through the electrolyte.^32,54 

In order to compare bulk electrolyte and droplet polarization scans, direct experimental measurement of the chloride concentration in the droplets was obtained with a microscale chloride sensor. The sensor, an Ag/AgCl microelectrode, was first calibrated against an Ag/AgCl reference electrode in a microcapillary salt bridge in bulk electrolytes of varying concentration. The natural log relationship shown in Figure 4 was a consequence of the sensor Nernstian response to chloride activity changes relative to saturated NaCl in the reference.^55-56  A calibration curve of the electrolyte concentration against the RH was then obtained by measuring the OCP of the chloride sensor in a 5 μL droplet at different RH. The results are shown in Figure 4, with an image of the chloride sensor (left) and the micropipette tip (right) containing the Ag/AgCl reference electrode probing a NaCl droplet on UNS S46500. The original 5 μL volume of the droplet decreased because of evaporation during equilibration with the ambient RH producing a final salt concentration that was considerably greater than the initial 0.6 M NaCl. The evaporation process on the stainless steels and Ti alloy occurred by pinning of the three phase boundary line, whereby the height of the droplet decreased, as reported previously for aqueous NaCl solutions, resulting in a flattened appearance.⁵⁷  The chloride concentration was nearly linear with respect to RH for the range examined, 72% to 84% RH at 75°F (23.8°C). The chloride concentrations of the droplets were within 4% RH of calculated and experimental equilibrium results for NaCl solutions.

Cathodic polarization scans for 0.5 μL, 2.5 μL, and 5.0 μL droplets with initial concentrations of 0.6 M NaCl are shown in Figure 5. These scans were all collected at 80% RH on UNS S465000. The plot also includes a polarization curve from the same material with a bulk electrolyte configuration containing a comparable 4 M chloride concentration. The three different volumes of electrolyte were shown to behave in a similar fashion, with a well-defined activation-controlled region from 0 V_SCE to −0.3 V_SCE, followed by an apparent diffusion limited region between −0.3 V_SCE and −0.6 V_SCE and finishing with the current increasing by as much as an order of magnitude for the smallest droplet. The final measured current densities and potentials for each droplet scan were dictated by the IR drop, which was greater in the case of larger droplets. One interesting feature detected in the polarization curves using the smallest droplet was a current decrease even as the potential continued to decrease between −0.3 V_SCE and −0.5 V_SCE. This current decrease may be related to phenomena observed in bulk electrolyte solutions at high pH and will be discussed in more detail later.

With respect to the droplet polarization experiments, the bulk electrolyte results exhibited a significantly lower OCP, as shown in Figure 5. The OCP is attributed to passive layer breakdown and is described further in the Discussion section. As a consequence, the mixed or activation-controlled region of the polarization curve in the bulk electrolyte was virtually non-existent because the diffusion limit was rapidly attained. The diffusion limited current for ORR in the droplets was also ~5 times that of the diffusion limited current in the bulk electrolyte. This difference was attributed to the thin profile of the droplet and edge effects that enhanced oxygen availability.

For further deconvolution of the droplet polarization scans and an understanding of pH, chloride concentration, and O₂ concentration, each alloy was polarized in several bulk electrolyte compositions. The electrolytes investigated included:

• •

  0.6 M NaCl.

• •

  Ar purged 0.6 M NaCl to understand the effects of oxygen.

• •

  Saturated NaCl to mimic droplet composition near efflorescence.

• •

  Three different acidified solutions that were used to examine pH and the role of anions. Acidic pH values were of interest because of recent correlations between aerosol acidity and particle size in climates with distinct gaseous pollutant levels.^19,58-60 

The results for the stainless steels and Ti alloy are shown in Figure 6 with additional pH dependent details in Figure 7 and results for UNS A92024-T3 and UNS A97075-T6 in Figure 8.

The baseline curves, in Figure 6, for the stainless steels in 0.6 M NaCl (blue dashed line) were quantitatively similar to one another, while UNS R56400 displayed the same qualitative features. The OCP of UNS S46500, UNS S17S00, and UNS S13800 were between −0.08 V_SCE and −0.11 V_SCE, respectively, while the OCP of UNS R56400 was ~0.21 V. The stainless steels were more kinetically active for the ORR, reaching diffusion limitation (1) near −0.57 V_SCE, with Tafel slopes between 0.16 V/decade and 0.19 V/decade, while that of UNS R56400 was 0.29 V/decade, reaching diffusion limitation near −1 V_SCE. Tafel slope values were taken closer to the diffusion limited region because an inflection point between OCP and the diffusion limited current (2), along with the nonlinear behavior closer to the OCP, which implied most of the activation-controlled region was poorly suited for capture of ORR Tafel slopes. Interestingly, UNS R56400 displayed a decrease in current after reaching the ORR diffusion limited level (1*), suggesting that ORR became kinetically hindered on the Ti alloy as a result of solution or surface chemistry changes. This feature for UNS R56400 is a possible consequence of TiO₂ oxide reduction to the more thermodynamically stable form, Ti₂O₃, at these potentials.⁶¹ 

In the Ar purged polarization scans, OCP dropped by 0.2 V to 0.4 V (3) and the diffusion limited current was shown to drop one order of magnitude (4) or more. Similar reduction reaction currents in the ambient oxygenated and Ar purged curves below −1 V_SCE suggested these Tafel currents, labeled (5) in Figure 6, were a result of water reduction and, thus, unaffected by removal of oxygen. Polarization scans in saturated NaCl solution showed similar features as the Ar purged scan as a result of the decrease in the oxygen solubility from 238 μmol/kg to 69 μmol/kg H₂O as the NaCl concentration increased from 0.6 M to saturation.^62-63  With two of the alloys, UNS S13800 and UNS S17400, bubbling air through the saturated NaCl solution recovered some of the lost current, thereby verifying that oxygen transport limited ORR rates in saturated NaCl solutions. The drop in OCP for both Ar purged and saturated NaCl is related to reduced oxygen availability and passive layer breakdown, which was evident in the latter by crevice formation. These results are in agreement with the droplet and bulk electrolyte polarization curves in 4 M NaCl (Figure 5), which indicate the droplets reduce the diffusion barrier for oxygen transport.

The concentration of protons introduced into the acidified solution shifted the OCP ~0.20 V in the negative direction, which was consistent with passive layer breakdown in acidified solutions. Low pH also increased, by roughly two orders of magnitude, the observed diffusion limited current for the stainless steels as a result of proton reduction reaction contributions. The increased exchange current densities for hydrogen evolution via proton reduction are consistent with this reaction being kinetically facile with respect to ORR. Of the three acids examined, the 10 mM H₂SO₄ displayed a consistently greater diffusion limited current, approximately double that of the other acid electrolytes for UNS S13800, UNS S17400, and UNS S46500, which can be attributed to the diprotic behavior of sulfuric acid. In the case of UNS R56400, the lower pH enhanced the activation region from 0.6 M NaCl, with diffusion limited current that exhibited anion selective behavior. These results, along with the significantly decreased reduction reaction current with respect to the stainless steels, indicate that UNS R56400 does not reach diffusion limitation for the proton reduction reaction, remaining kinetically limited close to −1.4 V_SCE.

Further details of the pH relationship with reduction currents were examined for UNS S13800, as shown in Figure 7. The pH shift from neutral to acidic conditions decreased the OCP monotonically to more negative values with the most significant step occurring from pH 4 to 3. These changes likely represented changes in the passive layer as iron oxide has been shown to be prone to dissolution in acidic environments. The diffusion limited current at pH 7 was the result of ORR, while the diffusion limited values at pH 2 and 3 were attributed to the proton reduction reaction. The increased diffusion limited current densities at these lower pH values was an indication of the large concentration differences between protons and sparingly soluble O₂. However, the reaction kinetics occurring on alloys in solutions with pH values between 3 and 7 are more complex. The apparent drop in the diffusion limited current with decreasing pH, which reaches a minimum around pH 4, is understood as a complex response to the decreasing ORR current and increasing proton reduction current.

The ORR current did not decrease because of a decline in the oxygen concentration, as the minute decrease in O₂ concentration from the acid concentration at pH 5 (10 μM) and 4 (100 μM), respectively, was orders of magnitude lower than the NaCl concentration and therefore insufficient to account for the additional “salting-out” of oxygen that would cause a 50% decrease in ORR diffusion limited current.⁶³  Rather, the pH shift was changing the kinetic behavior of the alloy which was likely a result of changes in thickness and speciation of the metal oxide layer. Close examination of these scans reveal shoulders, highlighted in Figure 7(a), that emerged just below the 1 μA/cm² mark at pH 6 and moved to more negative potentials and higher current densities through pH 4. These shoulders roughly correlate with expected hydrogen evolution reaction (HER) current along with the transition to diffusion limited HER current observed at pH 3 and 2.

The polarization behavior observed in the alkaline solution was more monotonic than behavior in acidic solutions. The most significant difference among scans in Figure 7(b) was the decrease in the OCP as the pH was increased. This behavior was consistent with the Nernstian relationship for ORR at higher pH values.⁶⁴  The diffusion limited current exhibited little change through pH 12, but by pH 13, the reductive current peaked with a 30% drop in apparent diffusion limited current before a further drop of 22% as the potential was scanned down to −1 V_SCE.

The lost ORR activity, also seen in cathodic polarization curves on the titanium alloy, may be attributed to decreased oxygen solubility from the additional ionic strength of ~0.1 M NaOH at pH 13 along with changes in the metal oxide structure. The bulk pH decreased the oxygen solubility by 3.1%, according to modified “salting-out” parameters for the Setchenov equation.⁶³  While this was not sufficient to account for the >50% drop in ORR current, local pH gradients from hydroxide generation at the electrode surface may also contribute to this solubility barrier. For instance, a local increase of 0.5 M NaOH can lower the oxygen concentration by an additional 16%. In this regard, alkaline electrolytes contributed to two general effects on the surface oxides:^65-66 

• •

  Increasing the thickness of surface oxides.

• •

  Hydroxylation of the surface resulting in a more negative surface charge.

The thicker surface oxides have been shown to result from, at least partially, the decreased solubility of metal ions and, for some steels, it had been suggested they form an Fe³⁺-rich outer layer as a result of the preferential dissolution of Fe during anodic polarization. All scans in basic media displayed an increase in reductive current from the onset of water reduction below −1.1 V_SCE.

Anodic polarization scans for UNS A92024 and UNS A97075 in the same electrolytes as were used for the stainless steel and titanium alloys are shown in Figure 8. Comparison between UNS A92024 and UNS A97075 indicate that the latter was generally more active in 0.6 M NaCl, with an OCP value of −0.634 V_SCE for the former and −0.770 V_SCE for the latter. Another characteristic of the UNS A97075 scan was a sharp inflection observed just 20 mV positive of the OCP, which is likely associated with remnants of a more active Zn-rich surface layer.⁶⁷ 

For both Al alloys, deoxygenation pushed the OCP ~0.15 V negative of the breakdown potential, thereby introducing a large passive window. The passive region in UNS A92024 has been associated with dissolution of bulk material around the intermetallic Al₂CuMg S-phase particles and Mg dealloying from the S-phase prior to the onset of intergranular corrosion (IGC) at the breakdown potential.^68-69  UNS A97075 exhibited two breakdown potentials: the first was usually observed in mechanically polished UNS A97075 coupons sitting at OCP for short periods of time and had been attributed to a redistribution of zinc on the a surface of polished coupons and its subsequent dissolution during anodic polarization, and the second was associated with IGC.^67,70 

Saturated NaCl increased the activity of Al alloys with a negative shift in corrosion potential for both alloys. However, in the case of UNS A92024, a small passive window further offsets OCP from the breakdown potential. This distinction between OCP and the IGC breakdown potential was also observed in the deoxygenated 0.6 M NaCl electrolyte, providing a basis for assignment of the passive region to dissolution of the Al₂CuMg S-phase. Attenuation of this passive region was caused by the activation of IGC resulting from increased chloride concentration and partial deactivation of the S-phase particles. Intermetallic deactivation in saturated NaCl was faster than the deoxygenated solution but slower than what was seen in ambient oxygenation of 0.6 M NaCl solution resulting from the presence of oxygen, which drives intermetallic deactivation, albeit at a reduced level with respect to the more dilute NaCl solution.

Conversely, acidification deactivated both alloys by as much as 0.1 V, with less of an impact observed for the UNS A97075. While no anion effect was observed for UNS A97075, acid comparison with UNS A92024 displays a weak trend, with H₂SO₄ having the strongest ennoblement effect.

Experiments in bulk electrolytes were used to understand polarization details in droplets. Droplet experiments on Al alloys exhibited very similar OCP values with their corresponding bulk electrolyte values, occurring within a 10 mV range of one another. However, the corrosion potentials of the aluminum alloys in the droplets were pinned right at the OCP, while in the bulk electrolyte scans the corrosion potentials were shifted ~25 mV positive. These corrosion potential shifts were a consequence of the increased chloride concentration of the droplets, approximately 4 M NaCl at 80% RH. Additionally, droplet polarization scans on UNS A92024 at different RH values (Figure 9) demonstrated that lowering the RH resulted in a decrease in corrosion potential resulting from higher chloride concentrations.

From a galvanic corrosion perspective, the Al alloys displayed small Tafel slopes in both bulk and droplet electrolytes that suggested these alloys would be under cathodic control in a galvanic couple. As a consequence, the bulk electrolyte galvanic couples (0.6 M NaCl) displayed little difference in galvanic corrosion rates (Figure 10) when coupled with stainless steels. Despite a 0.15 V difference in corrosion potential between the two Al alloys, both of their corrosion potentials fell in the diffusion limited regime of the stainless steels, which is relatively stagnant.

In contrast, the droplet galvanic corrosion current in Figure 10(b) displayed a considerable increase over bulk electrolytes in addition to Al alloy dependency. Both of these differences were brought about by the unique polarization behavior of the stainless steels at negative potentials, < 0.6 V_SCE. In contrast to bulk polarization couples, chloride-induced corrosion potential shifts on Al alloys under droplets could cause greater galvanic corrosion as a result of the increasing ORR current in this potential range. The overall effect is captured in Figure 10, which illustrates a 5 to 10 increase in galvanic current under droplets vs. bulk electrolytes and a ~5-fold difference for droplet couples with UNS A92025 and UNS A97075.

Further comparison of cathodic polarization scans in droplets (Figure 2) and bulk electrolytes (Figure 6) for the stainless steels illustrates several key differences.

• •

  Increased ORR kinetics in the activation region with Tafel slopes that shift from 0.16 V/decade to 0.19 V/decade in bulk electrolyte to 0.06 V/decade to 0.08 V/decade in droplets.

• •

  The diffusion limited current increases ~5 fold in droplets of comparable electrolyte composition.

• •

  An order of magnitude increase in current is observed in droplets below −0.6 V_SCE.

The first point captures the importance of the electrode/electrolyte interface on reaction kinetics and a critical difference between marine and atmospheric corrosion. The NaCl concentration of discontinuous electrolytes in atmospheric environments is humidity dependent and can be up to an order of magnitude greater than 0.6 M NaCl.⁷¹  Despite this disparity in concentration, the polarization curves from droplets, which contain 4 M NaCl, appear more similar to the 0.6 M NaCl bulk electrolyte polarization curves than to bulk electrolytes of 4 M NaCl (Figure 5) or saturated NaCl (Figure 6). This is because of the decrease in OCP by ~0.3 V and the order of magnitude drop in diffusion limited current as the chloride concentration increases from 0.6 M NaCl to saturated NaCl in bulk electrolyte (Figures 6[a] and [b]). While the decrease in diffusion limited current is a consequence of the lower oxygen concentration in bulk electrolyte, and does not reflect a change in the material, the drop in OCP is an indication of passive layer breakdown.

Chloride interactions with stainless steel passive layers are complex with much deliberation over the details in the context of marine corrosion. However, limited information about passive layer behavior in atmospheric environments is available.^72-75  From previous studies in bulk electrolytes, the presence of (Mn, Fe, Cr)-S inclusions is well known and they are recognized as pit initiation sites that are sensitive to both chloride and pH.^76-79  As no steps were taken to mitigate S inclusions in this work, these defect sites, along with gasket contact, likely contribute to rapid activation in high chloride bulk electrolytes, as suggested by the initial 75 mV to 90 mV difference in OCP for more concentrated bulk electrolytes in Figure 11 for UNS S17400. Conversely, the OCP of UNS S17400 in 0.6 M NaCl continued to climb despite a series of potential dips indicative of activation-repassivation events. The droplet OCP was initially 100 mV more positive than any of the bulk electrolytes before a rapid 50 mV decay. This was followed by a small activation-repassivation event and progressive ennoblement. This single metastable OCP event in droplets is consistent with a relatively small number of inclusions under droplets vs. bulk electrolytes, based on surface area differences. MnS susceptibility to chemical dissolution at low pH may also explain passive layer breakdown in the acidified bulk solutions (Figures 6 and 7[a]).⁷⁹ 

The enhanced passive layer stability under droplets may be associated with greater availability of gaseous molecules in thin electrolyte films. Although dissolved oxygen is not considered to have an active precursor role for passive layer formation, higher concentrations of oxygen can ennoble the metal surface by lifting its potential to keep it in the passive range.⁸⁰  The slight increase in OCP for droplets (Figures 2 and 5), −0.04 V_SCE to −0.08 V_SCE, with respect to bulk electrolytes (Figure 6), including more dilute 0.6 M NaCl solutions, which have higher oxygen saturation concentrations, at −0.08 V_SCE to −0.11 V_SCE, support this interpretation in the present work. The small volume and high electrolyte/air interfacial area in droplets also enhance CO₂ availability for surface reactions as well. These conditions support greater levels of bicarbonate in neutral solution, which has been shown to inhibit pitting corrosion on stainless steel with delayed onset for pitting and elevated pitting potentials.^81-82  Evidence suggests repassivation occurs by adsorption of bicarbonate to form a surface layer of Fe, Cr(CO₃). These results are also consistent with a previous study of UNS S30400 (Type 304 stainless steel) under MgCl₂ electrolyte thin films in which the transition of metastable pitting events, discerned by 0.25 V drops in OCP, to stable pits was shown to occur at higher chloride concentrations, > 7.2 M Cl⁻, for thinner electrolyte films.⁸³ 

The enhanced ORR Tafel slopes from Figure 2 are thought to represent a change in the oxide layer that may manifest itself as a compositional or thickness change. Previous work has shown the oxide layer to play a crucial role on the observed ORR kinetics of stainless steels with various surface pretreatments.⁷  For example, XPS results of passive films on UNS S31600 (Type 316 stainless steel) indicated that pre-reduced surfaces exhibit greater kinetic activity for ORR than polished surfaces in seawater as a result of increased ratio of Fe²⁺/Fe³⁺ in the oxide layer.⁷  More recent reports on UNS S30400 and UNS S31600 in bulk and thin film electrolytes indicate the presence of Cl⁻ increases the Cr content in the passive film.^84-85  Taken together, these reports lend support for passive layer differences induced by the high chloride and dissolved gaseous species that lead to altered ORR kinetics. In an alternate microcell study of S inclusions on UNS S30400, inclusion dissolution at OCP was shown to result in S adsorption on the passive film and enhance cathodic current in these regions.⁸⁶  Conditioning of surfaces at OCP for experiments in Figure 2 may facilitate this type of S coverage, as reaction products are confined to small electrolyte volumes.

The second feature of the cathodic polarization curves in Figure 2 is the diffusion limited current, represented by a near vertical line between −0.3 V_SCE and −0.5 V_SCE. From Figure 5 the diffusion limited ORR current is shown to be enhanced in droplets by ~5 fold. These results are consistent with polarization curves under thin film electrolytes, which illustrate that oxygen transport is a function of both electrolyte thickness and salt loading. The droplet size examined in this work did not display an ORR diffusion limited size dependence; however, this diffusion limited current is consistent with values reported previously in this potential window.³¹  Close examination of the diffusion limited region in Figures 2 and 5 also reveals current reversals and oscillations not observed in previous reports with small embedded electrodes under thin films. These current fluctuations are attributed to confinement of hydroxide ions generated from ORR during cathodic polarization. Integration of the charge passed from OCP to the onset of the diffusion limited current, 0.1 mC to 1.2 mC, is equivalent to 0.003 M to 0.04 M OH⁻ by the four electron ORR, which dominates in neutral and alkaline conditions. Uniformly dispersed, these concentrations fall short of pH 13, but given these reactions are surface driven and dependent upon precursor availability, it is reasonable to expect radial and longitudinal gradients for near-surface regions, particularly at the perimeter of the droplet, that approach or exceed pH 13. The ORR kinetic decline in the diffusion limited region is attributed to high pH values within the most oxygen accessible regions of the droplet, just as was observed for bulk solution at pH 13 in Figure 7, which lowers oxygen concentration and promotes thicker passive films.

In the final potential regime of the droplet polarization curves, the cathodic current increased by an order of magnitude, giving rise to the enhanced galvanic corrosion currents in Figure 10. This increased current is partly a result of the reduction of the droplet surface by the decreasing potential. As mentioned earlier, the work of Le Bozec and coworkers demonstrated increased ORR kinetics on pre-reduced UNS S31600 was a result of the higher of Fe²⁺/Fe³⁺ ratio in the passive film.⁷  However, in addition to a more ORR active reduced surface, micro-droplets such as those in Figure 3 are formed during cathodic polarization. These micro-droplets have been observed previously in droplet atmospheric corrosion cells and are believed to form in response to hydroxide generation at the edge of the primary droplet which leads to sodium migration and water absorption to ultimately form the micro-droplets.^47,87  The images from Figure 3 suggest a fraction of the additional current is a result of surface reduction, as micro-droplet formation was most noticeable between −0.65 V_SCE and −0.84 V_SCE, which corresponds to images (c) and (d) from Figure 3.

Additionally, one should also consider that ohmic loss is responsive to droplet geometry changes during polarization. Impedance measurements before and after cathodic polarization showed a general increase in resistance at high frequencies, consistent with a thinner droplet and larger contact area. A dynamic impedance may also explain the lack of defined HER currents in Figure 2; however, these results require more accurate understanding of current and ohmic loss distribution to dissect the impact of IR drop from polarization curves.

Although not incorporated into the above discussion, the three general differences observed for bulk and droplet polarization scans on the stainless steels also apply to the case of UNS R56400. Likewise, the polarization features for UNS R56400 under droplets display the three primary features as the stainless steels, albeit at more negative potentials. The kinetic differences observed between Ti and stainless steel fasteners give rise to significant differences in galvanic corrosion rates between stainless steels and UNS R56400 with UNS A92404 and UNS A97075, as shown in Figure 10. Furthermore, as the corrosion potentials of both UNS A92024 and UNS A97075 fall on the activation region of UNS R56400 for both bulk and droplet electrolytes, UNS R56400 displays material-dependent galvanic corrosion rates for bulk and droplet electrolytes.

Comparison of the stainless steels with UNS R56400 in Figure 10 indicates that both the bulk electrolytes and droplets predict significantly higher galvanic corrosion rates for stainless steel-Al alloy couples. A key difference being that the stainless steels did not demonstrate Al alloy dependent galvanic corrosion rates in bulk solution, while UNS R56400 did. This was because the galvanic coupling potential of the Al alloys are in the activation region of UNS R56400 and in the ORR diffusion limited region of the stainless steels. Conversely, under droplets, stainless steels and UNS R56400 impose couple dependent corrosion rates on UNS A92024 and UNS A97075, but for different reasons. UNS R56400 exhibits Al alloy dependent galvanic corrosion rates because it is still in the activation region. The stainless steels induce current increases primarily associated with area expansion.

This work examined galvanic atmospheric corrosion of structural aircraft alloys with droplet and bulk electrolyte polarization scans. Droplets were probed by means of micropipette salt bridges in a simulated atmospheric chamber with controlled humidity and temperature. Bulk electrolytes were evaluated as a basis to understand droplet polarization curves. Corrosion rates were evaluated by overlaying polarization scans of anode and cathode materials. Primary findings are summarized as:

• Cathodic polarization of stainless steels and Ti alloy in droplets exhibit smaller Tafel slopes and the onset of diffusion limitations at more positive potentials in comparison to bulk electrolytes with higher oxygen solubility.

• Droplet galvanic corrosion rates are 5 to 10 fold greater than that observed in bulk electrolytes.

• At galvanic coupling potentials, cathode materials display an increase in current beyond the interpreted ORR diffusion limited current that is primarily attributed to surface area expansion.

• Stainless steel induced galvanic corrosion in bulk electrolyte is relatively insensitive to the aluminum alloy as their corrosion potentials both lie in the ORR diffusion limited region, whereas droplets experience a 2 to 3 fold difference in corrosion rates between UNS A92024-T3 and UNS A97075-T6.

• Galvanic corrosion is under mixed control for couples with UNS R56400 as galvanic potentials lie within the activation region for both bulk and droplet electrolytes.