Atmospheric Pitting Corrosion of AA7075-T6 Under Evaporating Droplets THESIS

Transcription

1 Atmospheric Pitting Corrosion of AA7075-T6 Under Evaporating Droplets THESIS Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University By Sean C. Morton Graduate Program in Materials Science and Engineering The Ohio State University 2013 Master's Examination Committee: Professor Gerald Frankel, Advisor Professor Rudolph Buchheit

2 Copyright by Sean C. Morton 2013

3 Abstract Pitting corrosion of polished and etched AA7075-T6 in the presence of 3.5 wt % NaCl electrolyte was investigated both with standard full immersion testing, and with the application of evaporating electrolyte droplets in a fixed humidity environment. Testing included the addition of chromate-, vanadate, and cerous-based corrosion inhibiting salts in concentrations of 3 mm, 0.3 mm and 0.03 mm. A Scanning Kelvin Probe (SKP) was used to investigate the in situ corrosion potential behavior of the electrolyte droplets. In immersion testing, widespread pitting at intermetallic particles was observed in uninhibited solution. Similar but less severe attack was observed with additions of 0.03 mm vanadate and cerous ions. At higher concentration of vanadate, small numbers of very large pits were found, while higher concentrations of cerous ions provided very good inhibition. Chromate provided nearly complete inhibition of corrosion in all concentrations of immersion testing. In all cases of immersion testing, the measured corrosion potential was found to be pinned near to the pitting potential. In droplet testing, corrosion potentials measured by SKP were indicative of a range of behaviors, including metastable pitting, inhibition/passivation, and a form of corrosion attack in which pitting at the edge of the test droplet resulted in the formation of an adjacent, secondary droplet. This secondary droplet effect, observed in uninhibited NaCl droplets as well as with 0.03 mm concentration of both vanadate and cerous ions, was investigated in more detail with SKP. The attack progresses by large-scale separation ii

4 of anodic and cathodic regions under the droplet, with the anodic region initiating near the edge of the droplet. A model is described for the formation of the secondary droplet and associated attack, by deliquescence of water vapor due to an excess of ions from the anode diffusing to the edge of the droplet. iii

5 Dedication To my parents, for everything. iv

6 Acknowledgments I would like to thank my advisor, Dr. Jerry Frankel, for his support, patience, and encouragement throughout my time at OSU. I also wish to acknowledge the financial sponsorship of this project from the Office of the Secretary of Defense and the US Air Force Academy. Finally, I want to thank my friends and colleagues at the Fontana Corrosion Center for all of their help and support. v

7 Vita Deerfield Academy B.S. Materials Science, Mass. Inst. of Tech Test Engineer, A123 Systems, Inc Lead Systems Engineer, Levant Power Corp present...graduate Research Associate, Department of Materials Science, The Ohio State University Publications M. P. Short, S. Morton, S. E. Ferry, R. G. Ballinger. Diffusional Stability of Ferritic Martensitic Steel Composite for Service in Advanced Lead Bismuth Cooled Nuclear Reactors, International Heat Treatment & Surface Engineering, 4: (2010). Fields of Study Major Field: Materials Science and Engineering vi

8 Table of Contents Abstract... ii Dedication... iv Acknowledgments... v Vita... vi List of Tables... viii List of Figures... ix Introduction... 1 Experimental Results Discussion Conclusions and Future Work References vii

9 List of Tables Table 1: ASM composition specification for AA7075.[17]... 8 Table 2: Droplet parameters from SKP results and the calculated approximation of final droplet volume and the factor by which salt concentration increased from the beginning of the experiment viii

10 List of Figures Figure 1: Typical potential profiles over 20 hours of etched AA7075-T6 samples in 3.5 wt % NaCl (top left) and with additions of chromate (top right), vanadate (bottom left), and cerous ions(bottom right) Figure 2: Anodic and cathodic potentiodynamic data on samples immersed in 3.5 wt % NaCl, taken0.5 and 20 hours after immersion. Each of the four scans was run on separate samples Figure 3: Sample surface before (left) and after (right) 20-hour immersion test in 3.5 wt % NaCl Figure 4: Optical images of sample surfaces after 20-hour immersion testing in 3.5 wt % NaCl and chromate concentrations of 3 mm (top left), 0.3 mm (top right), and 0.03 mm (bottom left) Figure 5: Optical images of sample surfaces after 20-hour immersion testing in 3.5 wt % NaCl and vanadate concentrations of 3 mm (top left), 0.3 mm (top right), and 0.03 mm (bottom left). The image at bottom right shows the 0.03 mm vanadate test sample after cleaning Figure 6: Optical images of sample surfaces after 20-hour immersion testing in 3.5 wt % NaCl and cerous ion concentrations of 3 mm (top left), 0.3 mm (top right), and 0.03 mm (bottom left) ix

11 Figure 7: Height profile of sample surface after uninhibited 3.5 wt % NaCl immersion test. Pits exceed 5 μm in depth Figure 8: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.03 mm chromate. Maximum pit depth is less than 3 μm. The sample area shown is the same as pictured in Figure Figure 9: Height profile of sample surface after immersion testing with 3.5 wt % NaCl with 3 mm vanadate. Pit depth exceeds 15 μm. The sample area shown is the same as pictured in Figure Figure 10: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.3 mm vanadate. Large pits are seen up to 15 μm deep. The sample area shown is the same as pictured in Figure Figure 11: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.03 mm vanadate. Pits are narrow and shallow, less than 6 μm deep, but are widespread across the exposed surface. The sample area shown is the same as pictured in Figure Figure 12: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.3 mm cerous ions. Isolated pits are limited to less than 4 μm in depth. The sample area shown is the same as pictured in Figure Figure 13: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.03 mm cerous ions. Pits reach over 10 μm in depth. The sample area shown is the same as pictured in Figure x

12 Figure 14: Potentiodynamic scans in 3.5 wt % NaCl with chromate additions. Anodic inhibition is observed at all concentrations near the corrosion potential, and stronger inhibition at the lowest concentration. All concentrations inhibit cathodic kinetics, with 0.3 mm and 3 mm concentrations performing nearly identically Figure 15: Potentiodynamic scans in 3.5 wt % NaCl and vanadate additions. At 3 mm vanadate, some anodic inhibition is observed, while lower concentrations perform similar to the uninhibited condition. On the cathodic side, all concentrations reduce the oxidation kinetics by an order of magnitude Figure 16: Potentiodynamic scans in 3.5 wt % NaCl and with additions of cerous ions. Slight anodic inhibition is observed by all concentrations slightly above the corrosion potential, but the effect disappears after 50 mv. On the cathodic side, oxidation kinetics are reduced by less than an order of magnitude with the addition of 0.03 mm cerous ions, while the higher concentrations decrease the oxygen reduction rate by over a decade up to 400 mv below the corrosion potential Figure 17: Optical images showing the evolution of a 3.5 wt % NaCl droplet during SKP testing. The initial droplet is shown at top left. Evaporation of the droplet is seen after 30 minutes at top right, and 2 hours at center left. The bottom left image shows the initiation of a secondary droplet after 9 hours. At 12.5 hours, pictured bottom right, the secondary droplet has grown significantly Figure 18: Detailed time progression of the formation of a secondary droplet during SKP testing. The droplet shown is 3.5 wt % NaCl with 0.03 mm vanadate. Corrosion activity is first observed at top left near the edge of the droplet. The attack progresses throughout xi

13 the test, forming a secondary droplet (first seen at center right), which grows during the final 8 hours Figure 19: Measured SKP potential profiles for replicate tests of 3.5 wt % NaCl droplets. All tests resulted in the formation of a secondary droplet Figure 20: SKP potential profiles for droplet tests with 3.5 wt% NaCl and chromate in various concentrations Figure 21: SKP potential profiles for droplet tests with 3.5 wt% NaCl and vanadate in various concentrations. Secondary droplet formation is observed at the 0.03 mm vanadate condition Figure 22: SKP potential profiles for droplet tests with 3.5 wt% NaCl and cerous ions in various concentrations. Secondary droplet formation is observed in condition of 0.03 mm cerous ions Figure 23: Photograph at the conclusion of a 20-hour droplet test with 0.03 mm cerous ions and 3.5 wt % NaCl, prior to gathering linescan data. A secondary droplet is seen at bottom left, with an associated ring of precipitation in the main droplet Figure 24: Shown at top is the SKP linescan profiles for potential and height of a 20-hour droplet test with 0.03 mm cerous ions and 3.5 wt % NaCl. At bottom is an optical image of the surface taken after testing. The arrow indicates the location of the linescan, and corresponding sites are labeled numerically Figure 25: Shown at top is surface seen in Figure 18 after removal of corrosion products. At bottom is topography data, taken with OP, of the boxed area in the top figure xii

14 Figure 26: A droplet at the end of a 20-hour test with 0.03 mm vanadate and 3.5 wt% NaCl, with secondary droplet at the perimeter Figure 27: Shown at top is the SKP linescan profiles for potential and height of a 20-hour droplet test with 0.03 mm vanadate and 3.5 wt % NaCl. At bottom is an optical image of the surface taken after testing. The arrow indicates the location of the linescan, and corresponding sites are labeled numerically Figure 28: Shown at top is the surface seen in Figure 21 after removal of corrosion product. At bottom are OP data of the indicated boxed area Figure 29: A droplet at the end of a 20-hour test with 3.5 wt % NaCl. An irregular area is evident at the edge of the droplet closest to the camera, indicated by an arrow Figure 30: Shown at top is the SKP linescan profiles for potential and height of a 20-hour droplet test with 3.5 wt % NaCl. At bottom is an optical image of the surface taken after testing. The arrow indicates the location of the linescan, and corresponding sites are labeled numerically Figure 31:At left is the surface seen in Figure 24 after the removal of corrosion product. At right is the topography of the indicated boxed area, as measured by OP Figure 32: A ring of salt precipitated around the edge of this droplet during testing with 3.5 wt % NaCl due to a drop in chamber humidity. A secondary droplet that developed during the test is indicated with an arrow Figure 33: The height and potential profiles from an SKP linescan after a 20-hour droplet test with 3.5 wt % NaCl are shown in the top image. At bottom is an optical micrograph xiii

15 of the surface surrounding the secondary droplet after testing. An arrow indicates the location of the linescan, and corresponding sites are labeled numerically Figure 34: Shown at left is the secondary droplet area seen in Figure 27 after removal of corrosion products. At right is OP data showing the topography of the indicated boxed area. The dotted lines indicate the perimeter of the initial and secondary droplets Figure 35: Schematic representation of secondary droplet phenomenon. Severe pitting attack (1) occurs near the edge of the initial droplet where the secondary droplet forms. Many smaller pits also occur throughout the wetted surface area. The dotted arrows indicate ion transport from the anode and cathode sites for hydrolysis xiv

16 Introduction Atmospheric corrosion, as distinct from corrosion occurring on substrates submerged in electrolyte, is corrosion that results from exposure to natural ambient environments.[1] Sources of supporting electrolyte for atmospheric corrosion include humidity, rain, dew, or spray from nearby water. Pollutants and natural chlorides in the atmosphere provide the necessary ions for activation and support of corrosion cells. For example, it has been shown that the presence of the pollutant SO 2 in the atmosphere, in combination with sufficiently high humidity, will cause significant corrosion on carbon steel.[2] Marine and coastal environments are among the more severe environments for atmospheric corrosion, due to the combination of humidity and atmospheric salts.[1] Much work has gone into studying the effects of these conditions by simply exposing samples in the environment for months or years at a time. Ambler and Bain provide one of the first extensive field studies in the literature, exposing steel, brass, and aluminum at a number of sites in Nigeria for a period of one to twelve months, in addition to attempting to measure deposition rates of airborne salts.[3] Rates of corrosion were found to vary widely, with atmospheric salinity and humidity being identified as primary variables in the extent of corrosion observed. In coastal regions, atmospheric deposition of sodium chloride was found to reach as high as 2 g/m 2 per day. Furthermore, tests were performed where samples were only exposed during the day or only during the night, in 1

17 addition to continuously exposed samples. The extent of the day- and night-only corrosion, combined, was found to be less than that of the continuously exposed tests. This result suggests that diurnal changes, likely temperature and/or humidity, play a significant role in initiating and sustaining atmospheric corrosion. Humidity was further identified as a key component of corrosion by comparing the corrosion rates of exposed samples at sites that experienced similar salt deposition rates, but different mean relative humidities. In particular, for two sites with salt deposition rates measured at 22 and 21 mg/m 2 /month and average humidity values of 80% and 50% respectively, corrosion rate at the site with greater humidity was found to exceed that of the arid site by a factor of 18. Clearly humidity plays a very important role in activation of atmospheric corrosion. However, it is not immediately clear whether the critical factor of humidity is simply being above a threshold value for creating absorbed water layers on metallic surfaces, or a more complex function of changing humidity in a wet/dry cycle. Citing a lack of attention in the literature, Duncan and Balance further investigated the extent of airborne salt deposition far from the ocean.[4] While the concentration of atmospheric salt is known to decrease rapidly with distance from the shore, typically within 1 km, there was little data available for sites much further inland. Levels of dry-deposited chloride, measured by exposing sheltered, dry sheets of filter paper in the environment for four weeks at a time, were gathered for sites in New Zealand at distances up to 90 km from the coast. Sites were selected so as to minimize the corrosive effects of atmospheric pollution, and strictly investigate the effects of sea salt. Mild steel and copper samples were also exposed at the chosen locations to measure 2

18 weight loss resulting from atmospheric corrosion. Average chloride deposition rates as far as 50 km from shore were measured to be as much as one tenth of the rates for coastal sites within a kilometer of the ocean. Furthermore, weight loss in exposed samples for sites 50 km inland was as much as 30-40% of that for the coastal. The effects of sea salt are seen to extend far beyond what would nominally be considered a marine environment, and should be taken into account for any study of atmospheric corrosion. It has been noted that investigation of atmospheric corrosion has been performed primarily through environmental exposures in the field rather than laboratory experiments.[5] However, one of the classic experiments in corrosion, the Evans drop, provides insight into the nature of atmospheric corrosion, and has been performed more recently in various forms. The original experiment, published in 1926, was conducted by placing drops of electrolyte on mild steel.[6] The electrolyte drops contain potassium chloride, to promote corrosion of the steel, and a doubly acting ferroxyl indicator, which changes color depending on local ph of the solution. High ph regions, which appear pink due to the indicator, are assumed to be sites of oxygen reduction, while areas that turn blue are of low ph, and associated with the anodic reaction, where metal is oxidized and forms metal salts. Initially, when oxygen is uniformly dissolved through the drop, small areas of anodic and cathodic activity appear throughout the droplet area. However, as the oxygen in solution is depleted through reduction, the cathodic and anodic areas segregate, with oxygen reduction take place at the circumference of the drop where the diffusion path from the air/electrolyte interface to the electrolyte/metal interface is the shortest. Consequently, oxidation of the metal is seen to centralize in the middle of the drop, with 3

19 insoluble products building up in a ring separating the two regions. This experiment, an example of a differential aeration cell, illustrates factors in atmospheric corrosion not present in full immersion conditions, which is the effects of the three-phase air/electrolyte/metal interface as well as variation in electrolyte film thickness.[7] Under immersion testing, such large coalescence of anodic and cathodic regions are typically not observed, as larger volumes of electrolyte prevent significant differences in oxygenation from occurring due to longer diffusion paths and convection in the liquid. In small volumes of electrolyte, convection in the solution plays less of a role.[8] The Evans drop experiment can be extended to investigate corrosion of metals that form passive oxide layers.[9] In the work of Tsutsumi et al., magnesium chloride (MgCl 2 ) drops of varying sizes and concentrations were placed on samples of 304 stainless steel and exposed to controlled humidity environments for up to 100 hours. The role of humidity is to control the equilibrium concentration of MgCl 2 in the drop. Over time, the drop will either evaporate (in dry conditions) or deliquesce (in wet conditions) until the activity of water in the drop is in equilibrium with that of water in the air. In drier conditions, as more water evaporates from the drop, there will be a greater resulting equilibrium concentration of salt. In this way, it was shown that drop with a starting concentration of as low as 0.5 mm MgCl 2 will cause initiation of pits on stainless steel under sufficiently low humidity conditions. This adaptation of the Evans drop experiment provides an excellent basis for lab testing of atmospheric corrosions. In similar work, also on 304 stainless steel, it has been observed that additional wetting of the sample surface may occur around the edges of the electrolyte drop.[10] At 4

20 high humidities, drops of 0.5 M NaCl were shown to induce formation of micron-scale droplets at the periphery of the initial electrolyte drop. It is suggested that these regions start with a layer of absorbed water due to high humidity. Given the very thin nature of the water layer, these areas are preferred sites for the oxygen reduction reaction, supporting corrosion of the substrate under the drop. The cathodic reaction alkalizes the absorbed water layer, and cations are drawn in from the drop to achieve charge balance. The increasing concentration of ions in the water layer in turn causes deliquescence of these areas into visible droplets, as water is pulled from the air to dilute the electrolyte. Maier et al. investigated the effects of MgCl 2 droplets on SS304, and observed that pits forming under droplets tended to grow laterally.[11] Similar work by Li et al. also observed lateral growth of corrosion under droplet of MgCl2 on an Al-0.63Mg- 0.28Si alloy, but attack was in the form of a filiform-like growth that would extend outside the original test droplet.[12] Li proposes a mechanism by which attack edge of the droplet could propagate outwards and cause deliquescence of additional liquid outside the droplet. Li also observed metastable pitting attack in electrolyte droplets with in situ measurement of the corrosion potential with a Kelvin probe, noting a characteristic sharp drop in OCP followed by an immediate relaxation back to the stable baseline potential. This paper investigates the differences in corrosion mechanisms between standard immersion testing and an atmospheric corrosion test similar to the Evans drop experiment and the work of Maier and Li. 5

21 Scanning Kelvin Probe The Kelvin probe (KP) is a non-contact device that allows measurements of the outer potential difference between electrically connected metals.[13] It operates by vibrating a metallic tip vertically above the surface under investigation, which is connected electrically by an external circuit. The surface and tip form a capacitor, which varies in capacitance with the distance between them. The surfaces will become charged due to differences in the outer potential of the two materials (with metals, the potential difference is the difference in their work functions). The vibration of the tip, and resulting variation in capacitance, causes an alternating current to flow in the external circuit. The potential difference between the surface and probe tip can then be determined by measuring the magnitude of the current flow. Kelvin probes were first investigated as tool for corrosion science by Stratmann and Streckel in 1990.[14] The authors showed that for metals corroding under thin electrolyte layers, the potential measured by a KP above the electrolyte is nearly identical to the corrosion potential. Further development of the technique led to the Scanning Kelvin Prove (SKP), adding the capability to maintain constant air gap distance of the tip above a surface that varies in height with time.[13] To do so, an additional AC voltage is applied between the surface and tip through external circuitry. The frequency of the applied voltage must be sufficiently different from the frequency of tip vibration so as to allow independent demodulation of the two signals. The magnitude of the current response from the additional AC voltage is then strictly related to the distance between the surface and tip, and the height of the tip can be adjusted to maintain a constant current 6

22 magnitude response, and thus a constant distance of separation. This capability allows an SKP to measure the potential over electrolyte that is evaporating or condensing, or to scan laterally over areas of varying height. In using an SKP as a reference electrode, the measured potential is first referenced against a known potential.[15] Saturated CuSO 4 on pure copper is commonly used for this, as it maintains a stable potential of +320 mv vs. SHE. Prior to conducting an experiment with the SKP, the potential of Cu/CuSO 4 is measured first, and the potential output of the SKP is calibrated with the controlling software. The SKP has been used to investigate the potential in of steel corroding under Evans drops.[16] Chen and Mansfeld prove the usefulness of the instrument in investigating atmospheric corrosion by measuring a detailed potential map of corrosion under a droplet of NaCl electrolyte. Data from the experiment clearly shows low potential in the center of the droplet, where the anodic reaction is occurring, and higher potential at the edge of the droplet, indicative of the supporting cathodic reaction. AA7075 The 7xxx series of aluminum alloys contain zinc as the primary alloying element along with copper and magnesium.[17] These alloys achieve the highest strength available with aluminum through heat treatment, but become increasingly susceptible to corrosion as a result. The common T6 temper designation indicates that the alloy has been solution heat treated and artificially aged to nucleate and grow secondary phase particles, providing the highest strength available with the alloy. The high strength of the 7

23 7xxx series is particularly useful in aircraft, where it has found widespread use in airframe construction. Table 1: ASM composition specification for AA7075.[17] Of the numerous alloys in the AA7xxx series, AA7075 is the most commonly used and studied, typically in the aforementioned T6 temper. The composition of AA7075 is shown in Table 1. The microstructure of AA7075-T6 temper is complex, but has been thoroughly studied. It contains a wide range of secondary phase particles that have been categorized into three classes.[18, 19] Precipitates are the smallest of these, and are found to nucleate and grow during aging as a result of super saturation of alloying elements after solutionization. The most notable precipitate is the η and ή phases of MgZn 2 and the associated solute-rich clusters known as GP zones. These precipitates are responsible for providing the high strength of the alloy, and are typically several nanometers in size. Other precipitate species identified include Mg 2 Al 3, Al 2 Cu, and Al 32 Zn 49. The second set of particles, known as dispersoids, are found to be distributed evenly throughout the matrix, with particle sizes ranging from tens to hundreds of nanometers in size. Dispersoids are useful in controlling the grain size of the matrix, as well as inhibiting recrystallization. They can be composed of a number of aluminum intermetallics, such as Al 3 Ti, Al 3 Zn, Al 6 Mn, and Al 20 Cu 2 Mn 3. The largest secondary 8

24 particles found in the microstructure of 7075 are known as constituent particles, and may be as large as µm in size. Compositions for constituent particles include the S- phase, Al 2 CuMg, as well as Al 7 Cu 2 Fe, Al 3 Fe and (Al,Cu) 6 (Fe,Cu). These particles are formed during solidification and will often break up and align during working and forming of the metal, though they are not affected by further heat treatment. These particles are easily identified in polished specimens, especially in material that has undergone rolling, as the particles are visible through optical microscopy and can be seen in striations oriented in the rolling direction. After heat treatment of AA7075, the remaining composition by weight of primary alloying elements in the matrix is 3-4% Zn, 2-3% Mn, and 0.5-1% Cu, Table 1. While providing exceedingly high strength, the presence of so many particles of varying composition is directly responsible for significant susceptibility to localized corrosion attack in AA7075. The corrosion potential of the various individual particles has been shown to spread over a range of hundreds of millivolts.[20] As a result, galvanic couples arise between individual second phase particles and the matrix.[21] Matrix dissolution is observed surrounding cathodic particles, in particular those containing iron or manganese. Corrosion of the matrix occurs in a trenching fashion, as material is removed radially about the particle while the cathodic reaction, reduction of oxygen, is catalyzed on the surface of the particle. This attack will tend to proceed so long as the particle maintains an electrically conducting path to the matrix. Conversely, particles such as Al 2 CuMg that are at a lower potential relative to the matrix will anodically dissolve, while the matrix in the area surrounding the particle is host for oxygen 9

25 reduction. These localized corrosion cells are able to form independently at particles across the exposed surface, with trenches and pits commonly observed to be distributed evenly across a sample after lab testing. It has been shown that mechanical polishing of AA7075 will produce a unique surface layer with chemical properties different from those in the bulk.[22] Naturally this is of great importance to studies of corrosion, where mechanical polishing of samples is a nearly universal practice. The exact mechanism by which the layer forms has yet to be described, though it is hypothesized to arise from high shear strains and plastic deformation during polishing. Properties of the surface layer have been investigated electrochemically and with transmission electron microscopy (TEM). High resolution cross-sectional images of polished AA7075 show a homogeneous region at the surface which contains no precipitates, and extending 200 nm into the bulk.[23] This layer is also found to have a different, lower breakdown potential than the underlying bulk material. Thus, any corrosion experiment performed on polished AA7075 where the modified surface layer was not removed will likely not be producing data consistent with the performance of the actual bulk material. The surface layer will more readily corrode due to its lower breakdown potential, and the mechanisms of observed corrosion phenomena on the surface will be different as a result of the lack of secondary particles in the layer. However, this issue is easily remedied by simply removing the surface layer after polishing. The authors report that chemical etching of polished samples, performed by submersion in 1 M NaOH at 60 C for 90 seconds followed by 70% HNO3 for 30 seconds, is effective in completely removing the surface layer and exposing unmodified 10

26 material with bulk properties. This process ensures that data from subsequent testing of the surface will accurately represent properties of the bulk material as used in service. Inhibitors Hexavalent chromium (chromate) has long been used as an effective corrosion inhibitor on various structural metals and alloys.[24] The exact mechanisms of protection are not universally agreed upon, and vary depending on the substrate and environment, but the principals and applications of chromate have been extensively researched. The primary operation of chromate in solution is as a cathodic inhibitor, by reducing out of solution from Cr(VI) to Cr(III) and forming a strong barrier against oxygen reduction. Experiments with electrolyte containing NaCl and K 2 Cr 2 O 7 on pure copper showed rapid reduction of Cr(VI) to hydrated Cr(III), followed by the formation of a strongly adhered monolayer of polymerized Cr(III) oxyhydroxide.[25] This layer greatly suppresses oxygen reduction on the metal surface by preventing absorption of dissolved O 2. The formation of the Cr(III)-based layer is irreversible. After exposure of the surface to a chromate-containing electrolyte, the enhanced passivation properties observed are retained even after the electrolyte is removed and replaced with chromate-free electrolyte. Furthermore, the formation of the layer traps additional Cr(VI) from the solution, but prevents it from reducing. Upon localized removal or penetration of the protective layer and exposure of the metal surface, either by abrasion or aggressive attack, some of the remaining Cr(VI)in the coating is released and reduced onto the newly exposed surface, restoring the integrity of 11

27 the inhibitive coating. The same phenomenon has been observed with chromate on AA2024.[26] Salt spray testing was performed on samples with a 0.2 to 1 µm thick commercial-grade chromate coating, during which the majority of Cr(VI) remaining in the coating was reduced to Cr(III) while remaining a part of the protective layer. This self-healing characteristic of chromate coatings is very desirable in commercial applications, as it gives the protective layers exceedingly long lifetimes even in extreme environments. The cathodic inhibition effects are realized even in very low concentrations of chromate in solution. Significant decreases in oxygen reduction kinetics are observed on AA2024 in the presence of chloride at Cr(VI):Cl - ratios as low as Chromate in solution is also known to have an anodic inhibition effect, retarding the kinetics of metal oxidation from the substrate surface, though this effect is only observed at a significant degree with very high concentration of chromate (Cl - : Cr(VI) 1). However, industrial chromate conversion coatings (CCC) on aluminum have been shown to exhibit much greater anodic inhibition.[27] CCC formation involves use of fluoride to etch away the natural aluminum oxide layer from the metal surface, after which chromate is found to form a Cr(OH) 3 layer. While chromate is an extremely effective corrosion inhibitor, it is also known to be a strong carcinogen.[28] Much work in recent decades has gone into finding inhibitors with lower toxicity to replace chromate, and this search has covered dozens of transition and rare earth metal ions. Two of the more promising of these are ions of cerium and vanadium. 12

28 Cerium, typically added to solution in the form of CeCl 3 salt, has been shown to be effective in reducing the corrosion rate on aluminum alloys.[29] Strong inhibition has been observed in AA7075 in 0.1 M NaCl at a concentration of Ce 3+ of as low as 0.2 mm.[30] Polarization scans of AA7075 in 0.1 M NaCl showed that the addition of 1000 ppm CeCl 3 caused a 160 mv drop in corrosion potential, as well as a large drop in the cathodic limiting current density.[31] These data indicate that cerium acts as a strong cathodic inhibitor, reducing the kinetics of oxygen reduction on AA7075. The inhibiting effect is caused by film formation on the surface, primary in the form of Ce(OH) 3, though CeO 2 and Ce(OH) 4 are also found. The precipitated film has been shown to cover the entire surface of AA2024, though precipitation was strongest on secondary particles typically associated with oxygen reduction sites.[32] There is evidence for the film forming by Ce 3+ first oxidizing to Ce 4+ under alkaline conditions, the Ce 4+ then reducing back to its trivalent state at the metal surface and precipitating asce(oh) 3.[33] Vanadium has also been shown to be an effective inhibitor, particularly in alkaline environments.[34] Iannuzzi, et al. reported cathodic inhibition of AA2024-T3 by vanadate in 0.5 M NaCl solution, with the open circuit potential (OCP) being decreased by as much as 400 mv and over two orders of magnitude decrease in the limiting current density for the oxygen reduction reaction.[35] It is also shown that vanadium species provide a small amount of anodic protection, observed through an increase in measure pitting potentials, regardless of aeration. Speciation of vanadate in solution is very complex, and the present forms vary as a function of ph.[36] It is suggested that inhibition is primarily from monovanadate (V 1 ), which is stable at high ph, while 13

29 decavanadate (V 10 ) was detrimental to corrosion inhibition and formed at low ph. Therefore, vanadate appears to act as an inhibitor in a similar fashion to cerium, where increase in ph due to oxygen reduction causes reduction of vanadium species and subsequent passivation of cathodic surfaces. 14

30 Experimental Sample Preparation Samples measuring 3 cm x 3 cm were cut from AA7075-T6 rolled plate (6 mm thickness) and abraded with SiC paper in ethanol, starting with 120 grit and progressing through 240, 400, 600, and 800 grit. The samples were rinsed in ethanol, then etched in 1 M NaOH at 60 C for 45 seconds and pickled in 70% HNO 3 at 25 C for 10 seconds, then rinsed again with ethanol. Etching was performed to remove the altered surface layer generated by the abrasion process.[22] The samples were stored in a desiccator until use. Upon removal from the desiccator for testing, the samples rinsed with ethanol, DI water, and dried with filtered, pressurized air. Electrolytes The base electrolyte for all testing was 3.5 % NaCl by weight, and was unbuffered in all cases. In addition to uninhibited NaCl tests, inhibitors were added to the electrolyte at a range of concentrations for additional testing. The base salts for the inhibitors were K 2 CrO 4, NaVO 3, and Ce(NO 3 ) 3. These inhibitor salts were added to the base NaCl solution in concentrations of 3 mm, 0.3 mm, or 0.03 mm. Both full immersion and SKP tests were performed for NaCl alone as well as NaCl plus each concentration of the three inhibitors, for a total of 10 test conditions. At least two replicate tests were performed for each test condition. Further replicate tests were performed when data were inconsistent or test conditions were not sustained, such as variation in humidity for the droplet tests. 15

31 Immersion Testing Bulk immersion testing was conducted in an upright cylindrical vessel of polycarbonate, with a diameter of 4 cm, narrowing at the base to a 1.5 cm opening with a nitrile rubber O-ring seal. A lower assembly back plate held samples against the seal during testing. For each test, the vessel was filled with 60 ml of electrolyte, and a saturated calomel electrode (SCE) was positioned within 5 mm directly above the sample. Open circuit potential (OCP) measurements were made by recording the potential vs. SCE of the sample for 20 hours with a measurement period of 3 seconds using a Gamry Reference 600 potentiostat. After 20 hours, the sample was removed and rinsed with DI water and ethanol, dried with filtered air, and returned to the desiccator. Potentiodynamic tests were also performed on samples with the same immersion conditions and equipment. Platinum wire mesh was used as a counter electrode. Anodic and cathodic potentiodynamic tests were performed separately on different samples. After electrolyte was added to the test apparatus, the OCP was monitored for 20 minutes before beginning the scan. Anodic scans were performed from -10 mv vs. OCP to +250 mv vs. OCP. Cathodic scans were performed from +10 mv vs. OCP to -600 mv vs. OCP. Both scans were conducted at a rate of mv/sec. SKP Testing Kelvin probe experiments were performed with an SKP produced by K&M SoftControls (Düsseldorf, Germany) with an actuating sample stage (x- and y-axis 16

32 control) and probe head (z-axis). Controls and measurements were performed with a LabView (National Instruments) interface provided by the SKP supplier. The test chamber of the SKP was first purged with high humidity air by bubbling source air through DI water and into the chamber until the internal humidity was at least 85%. Open beakers of saturated KCl were placed in the chamber to maintain an equilibrium humidity of 85%. The SKP potential was calibrated against a Cu/CuSO 4 electrode composed of a 1 cm diameter copper cup filled with saturated CuSO 4 solution. The potential of this Cu/CuSO 4 electrode is 320 mv SCE. For each test, an electrolyte droplet measuring 6 L in volume was placed on a prepared sample using a glass syringe. The sample was then immediately loaded into the test chamber of the SKP and positioned under the probe tip before beginning data acquisition. The potential and height were measured at a sample rate of 1 Hz for 20 hours, while a digital image of the droplet was captured every minute with a 2 megapixel USB digital microscope (Oasis Scientific, Inc.). Upon completion of the 20-hour test, the probe tip was scanned linearly across the droplet to measure the final height. After visual inspection, additional line scans of potential and height were performed across the droplet over areas of interest. After completion of SKP testing, the sample was removed, washed with DI water and ethanol, dried, and returned to the desiccator. Images of the as-tested samples were taken with an optical microscope. Corrosion products on the samples were then removed by immersion in a solution of 1.5 liters of DI water, 150 ml H 3 PO 4, and 30 g CrO 3 at 80 C for 1 minute. After cleaning, the samples 17

33 were further imaged with optical microscopy, and height profile analysis was performed with a Contour GT-K1 optical profilometer (OP, Veeco Instruments Inc). 18

34 Results Immersion Testing The following results are for samples exposed in bulk solutions as opposed to the small volume droplets used in the SKP experiments. Typical OCP profiles of 20-hour immersion tests for NaCl and all inhibitor concentrations are shown in Figure 1. Figure 1: Typical potential profiles over 20 hours of etched AA7075-T6 samples in 3.5 wt % NaCl (top left) and with additions of chromate (top right), vanadate (bottom left), and cerous ions(bottom right). 19

35 All tests displayed a baseline potential near -720 mv vs. SCE. This indicates that the OCP is pinned near the pitting potential in all conditions owing to the nonpolarizibility of the pitting reaction, and that the addition of inhibitors has no significant affect on the pitting potential. In uninhibited NaCl, the OCP dropped by over 100 mv during testing. This was consistently observed in repeated experiments. The decrease is slow, not typical for the activation of a surface by pitting. Furthermore, the OCP value of-720 mv SCE is considered to be controlled by the pitting potential. Therefore, this drop in potential merits further attention. Potentiodynamic scans were performed on samples soon after immersion and after a long enough time at open circuit that the potential had dropped to the lower value. Separate anodic and cathodic scans were performed on different samples. Figure 2 shows that the OCP values for samples immersed for 0.5 h were close to -720 mv SCE, in agreement with the data shown in Figure 1. Anodic polarization resulted in an immediate rapid increase in current, which supports the notion that the OCP was close to the pitting potential. However, after 20 h of immersion, when the OCP had decreased to the lower stable value, the anodic polarization curve shows a passive region with a pitting potential equal to the OCP/pitting potential measured for samples tested soon after immersion. The cathodic polarization curves indicate that, after 20 h at OCP, the rate of oxygen reduction had decreased significantly, which caused the large decrease in OCP. The solution was continuously exposed to air so the availability of oxygen at the sample surface was unchanged. The drop in OCP during immersion can be attributed to a large increase in the 20

36 diffusion coefficient of dissolved oxygen, likely as a result of gel formation and corrosion products at the sample surface. Figure 2: Anodic and cathodic potentiodynamic data on samples immersed in 3.5 wt % NaCl, taken0.5 and 20 hours after immersion. Each of the four scans was run on separate samples. Figure 1 shows that, with additions of chromate and vanadate, the OCPs remained stable during the entire duration of the test, with very little variation in potential between different concentrations. With cerous ions, the baseline potential was initially higher at greater concentrations, but this difference was not reflected in the OCP of polarization tests, or reliably reproducible. For the two higher concentrations, the potential was initially low and increased over a period of hours. The condition with the lowest concentration of cerous ions experienced a drop in potential during the final hours of the experiment, when the potential decreased by over 100 mv. The data shown in Figure 1 21

37 represent the most severe of these observed fluctuations, but similar potential behavior was observed in replicate tests. This phenomenon may be similar to the drop experienced by the uninhibited NaCl tests. Optical micrographs of the sample surfaces before and after immersion testing in NaCl are shown in Figure 3. Striations of constituent intermetallic particles, from rolling of the plate in production, extend laterally, and scratches from polishing are seen diagonally across the surface. Etching of the sample revealed grain boundaries, which can also been seen in the optical images. Surface attack during exposure to NaCl is observed to occur at the intermetallic particles Figure 3: Sample surface before (left) and after (right) 20-hour immersion test in 3.5 wt % NaCl. Addition of chromate resulted in strong inhibition in all concentrations. Surface images of chromate-inhibited immersion samples are shown in Figure 4. No corrosion is 22

38 observed at concentrations of 3 mm or 0.3 mm chromate, with some isolated pitting occurring at 0.03 mm chromate. Figure 4: Optical images of sample surfaces after 20-hour immersion testing in 3.5 wt % NaCl and chromate concentrations of 3 mm (top left), 0.3 mm (top right), and 0.03 mm (bottom left). 23

39 The presence of vanadate provided superficial inhibition at the two higher concentrations, where the majority of the sample surface was unaffected by corrosion, while significant attack was sustained in localized areas. Surface images following immersion tests are shown in Figure 5. At the lowest concentration of vanadate, 0.03 mm, corrosion attack was similar to the uninhibited case, with widespread pitting attack at intermetallic particles. Figure 5: Optical images of sample surfaces after 20-hour immersion testing in 3.5 wt % NaCl and vanadate concentrations of 3 mm (top left), 0.3 mm (top right), and 0.03 mm (bottom left). The image at bottom right shows the 0.03 mm vanadate test sample after cleaning. 24

40 The addition of cerous ions provided somewhat better inhibition than vanadate. Images of the samples surfaces after exposure are shown in Figure 6. With a concentration of 3 mm cerous ions, no corrosion is observed. At 0.3 mm, small isolated pits are found on the surface. However, at 0.03 mm, a more widespread attack is observed. The extent of attack in this condition is less than that seen with 0.03 mm vanadate. Figure 6: Optical images of sample surfaces after 20-hour immersion testing in 3.5 wt % NaCl and cerous ion concentrations of 3 mm (top left), 0.3 mm (top right), and 0.03 mm (bottom left). 25

41 Topography data from OP of samples with observable corrosion are shown in Figures Pitting from NaCl-only exposure is widespread across the surface, but the pits do not exceed 5 μm in depth. Pits formed under electrolyte with 0.03 mm chromate are smaller, at most 2-3 μm deep. Tests at high concentrations of vanadium, which prevented attack on most of the surface, actually resulted in the formation of deeper pits. The large pits in solutions with 3 mm and 0.3 mm vanadate were found to be 15 μm deep. Similar to the uninhibited condition, the widespread pitting in 0.03 mm vanadate was limited to 5-6 μm in depth. Cerous ions were more effective, with pitting at the 0.3 mm condition limited to under 4 μm in depth. The pitting that was seen with 0.03 mm cerous ions was less widespread but deeper than that with 0.03 mm vanadate, reaching over 10 μm deep. 26

42 Figure 7: Height profile of sample surface after uninhibited 3.5 wt % NaCl immersion test. Pits exceed 5 μm in depth. Figure 8: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.03 mm chromate. Maximum pit depth is less than 3 μm. The sample area shown is the same as pictured in Figure 3. 27

43 Figure 9: Height profile of sample surface after immersion testing with 3.5 wt % NaCl with 3 mm vanadate. Pit depth exceeds 15 μm. The sample area shown is the same as pictured in Figure 4. Figure 10: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.3 mm vanadate. Large pits are seen up to 15 μm deep. The sample area shown is the same as pictured in Figure 4. 28

44 Figure 11: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.03 mm vanadate. Pits are narrow and shallow, less than 6 μm deep, but are widespread across the exposed surface. The sample area shown is the same as pictured in Figure 4. Figure 12: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.3 mm cerous ions. Isolated pits are limited to less than 4 μm in depth. The sample area shown is the same as pictured in Figure 5. 29

45 Figure 13: Height profile of sample surface after immersion testing with 3.5 wt % NaCl and 0.03 mm cerous ions. Pits reach over 10 μm in depth. The sample area shown is the same as pictured in Figure 5. Potentiodynamic Testing Potentiodynamic tests on samples under the same immersion conditions as tested above were performed to investigate the effects of the added inhibitors on the kinetic and critical potentials. The results are shown in Figures Chromate-containing solutions exhibit the strongest anodic inhibition, and the most effective anodic inhibition was observed for solutions containing 0.03 mm chromate. Weaker anodic inhibition was observed near the corrosion potential for the other concentrations and inhibitors. The trend of stronger anodic inhibition at lower concentration of chromate was observed in replicate testing. It is known that chromate can act as an anodic inhibitor, but the present observed correlation is not supported in the literature.[24] However, it is also noted that the conditions of aqueous chromate and etched AA7075-T6 is unique. The majority of 30

46 extant data is for chromate conversion coatings, or aqueous chromate and AA2024. Furthermore, tests with AA7075-T6 have previously not incorporated an etching step following mechanical polishing. Chromate additions resulted in a strong decrease in cathodic kinetics, but the variation for different concentrations of chromate was small. Vanadate showed the weakest cathodic inhibition, with even less difference in inhibition between concentrations. At concentrations of 3 mm and 0.3 mm, cerous ions provided the strongest cathodic inhibition near the corrosion potential, a reduction of close to two orders of magnitude in current density. The lowest concentration of cerous ions was less effective, and all concentrations had significantly decreased effectiveness in the hydrogen evolution region at potentials far below the corrosion potential. Figure 14: Potentiodynamic scans in 3.5 wt % NaCl with chromate additions. Anodic inhibition is observed at all concentrations near the corrosion potential, and stronger inhibition at the lowest concentration. All concentrations inhibit cathodic kinetics, with 0.3 mm and 3 mm concentrations performing nearly identically. 31

47 Figure 15: Potentiodynamic scans in 3.5 wt % NaCl and vanadate additions. At 3 mm vanadate, some anodic inhibition is observed, while lower concentrations perform similar to the uninhibited condition. On the cathodic side, all concentrations reduce the oxidation kinetics by an order of magnitude. Figure 16: Potentiodynamic scans in 3.5 wt % NaCl and with additions of cerous ions. Slight anodic inhibition is observed by all concentrations slightly above the corrosion potential, but the effect disappears after 50 mv. On the cathodic side, oxidation kinetics are reduced by less than an order of magnitude with the addition of 0.03 mm cerous ions, while the higher concentrations decrease the oxygen reduction rate by over a decade up to 400 mv below the corrosion potential. 32

48 Droplet Behavior In droplet tests, evaporation of electrolyte and the resulting decrease in droplet height begins immediately upon exposure of the droplet to the fixed humidity environment of the SKP chamber. A steady drop in potential coincides with the decrease in droplet height, due to the increasing concentration of NaCl in the droplet from evaporation of water. This can be seen in images of the droplet progression in Figure 17. In the third image, 2 h have elapsed, and the droplet has reached its equilibrium height and chloride concentration. During evaporation, only the droplet height changes; no shrinking of the droplet perimeter is seen in any of the tests performed. In tests performed with droplets containing only 3.5 wt % NaCl, or with additions of 0.03 mm concentrations of vanadate or cerous ions, secondary droplets of electrolyte form at the edge of the original electrolyte droplet. A secondary droplet can be seen in the latter two images of Figure 17. A more detailed progression of initiation and growth of a secondary droplet is shown in Figure 18. Pitting initiates at a point close to the edge of the droplet, and the attack is sustained throughout the test, spreading closer to the edge. Eventually, this results in the formation of a separate but connected secondary droplet. These secondary droplets are very small in diameter compared to the original droplet, and the secondary droplet forms such that the two droplets cover a single contiguous wetted surface area. Strong corrosion attack occurs in all cases in the region under the secondary droplet. 33

49 Videos of the droplet experiments shown in Figures 17 and 18 are provided as supplemental material for this document. The interested reader is strongly encouraged to view the videos. Figure 17: Optical images showing the evolution of a 3.5 wt % NaCl droplet during SKP testing. The initial droplet is shown at top left. Evaporation of the droplet is seen after 30 minutes at top right, and 2 hours at center left. The bottom left image shows the initiation of a secondary droplet after 9 hours. At 12.5 hours, pictured bottom right, the secondary droplet has grown significantly. 34

50 Figure 18: Detailed time progression of the formation of a secondary droplet during SKP testing. The droplet shown is 3.5 wt % NaCl with 0.03 mm vanadate. Corrosion activity is first observed at top left near the edge of the droplet. The attack progresses throughout the test, forming a secondary droplet (first seen at center right), which grows during the final 8 hours. 35

51 Scanning Kelvin Probe Testing Potential profiles for droplet tests, as measured by SKP, are shown in Figures All tests show an immediate and rapid drop in potential, as the result of pitting initiation, which is promoted by the evaporation of water from the droplet and increase in chloride concentration. After the initial evaporation, during typically one to two hours, a wide range of behaviors is observed in the measured potential. In 3.5 wt % NaCl, two characteristic behaviors are identified, and are seen in Figure 19. In some tests, the potential is seen to steadily decrease to a stable baseline potential, and remain there for the duration of the test ( 3 and 4 in Figure 19). In other tests the potential stabilized after evaporation, but later experienced a large (>600 mv) and very rapid increase in potential ( 1 and 2 in Figure 19). Following this increase, the potential remained high for the duration of the test. Regardless of the observed potential behavior, secondary droplet formation was observed in all of these tests, in addition to initiation of numerous small pits under the initial droplet. In the potential profile labeled 1 in Figure 19, the upward spike in potential was observed to occur simultaneously with the formation of the secondary droplet. (In the test 2, the secondary droplet formed out of view of the camera and was unable to be observed directly during formation.) 36

52 Figure 19: Measured SKP potential profiles for replicate tests of 3.5 wt % NaCl droplets. All tests resulted in the formation of a secondary droplet. The SKP potential transients for inhibited droplets are shown in Figures Results from droplets inhibited by chromate are shown in Figure 20, where the two higher concentrations displayed rapid decreases followed by slow recovery of the potential. These are characteristic signs of metastable pitting, indicating good protection because of the absence of sustained pitting. The lowest chromate concentration exhibits an initial drop in potential, indicating some pitting in the early stages, but the following rise in potential may indicate passivation as the chromate concentration increases due to evaporation of the droplet. Potential profiles for vanadate-inhibited droplets are shown in Figure 21, and droplets inhibited by cerous ions in Figure 22. The potential behaviors are very similar 37

53 for the two inhibitors, with the higher concentration of both displaying sustained high potential, while the potential for the two lower concentration tests decreases and stays at a low potential through the whole test. For the 0.03 mm concentration of vanadate, the potential goes through a series of relatively rapid changes early in the test, while the droplet is still evaporating. During this period, the SKP needle was observed to be directly above the edge of a large cathodic zone, which is discussed to a greater extent below. The rapid changes in potential are thus hypothesized to be a result of fluctuations in the area of this cathode region, and the changes in measured potential are due to switching between measurements of the active cathode (high potential) and non-active regions (low baseline potential) on the sample surface. The observed potential profiles of inhibited droplets can be categorized into three groups. The two higher concentrations of chromate, 3 mm and 0.3 mm, are unique among the tested inhibitors. The measured potential decreases to a constant baseline potential, as seen in all tests, but the potential profiles contain frequent downward spikes of up to 200 mv. Following these spikes, the potential quickly rises back to the baseline. For the 0.03 mm concentration of chromate, and for both vanadate and cerous ions at 3 mm and 0.3 mm concentrations, a different potential behavior is observed. In all of these tests, the potential decreases initially, stabilizes, and at some point during the test rises to a higher potential. While this behavior is similar to the upward spikes in potential observed in uninhibited NaCl droplets, no corrosion at the edge or exterior of the droplet, associated with the formation of a secondary droplet, is observed. The final category of potential behavior was exhibited by the lowest concentration, 0.03 mm, of both vanadate 38

54 and cerous ion inhibited tests. In these results, the potential profile is similar to that of the NaCl droplets that did not show an upward potential spike during the test. The stable potentials in these two tests are much lower than potentials measured in the other inhibited droplet tests, and are also unique in that they were the only inhibited conditions under which secondary droplets were observed. Figure 20: SKP potential profiles for droplet tests with 3.5 wt% NaCl and chromate in various concentrations. 39

55 Figure 21: SKP potential profiles for droplet tests with 3.5 wt% NaCl and vanadate in various concentrations. Secondary droplet formation is observed at the 0.03 mm vanadate condition. Figure 22: SKP potential profiles for droplet tests with 3.5 wt% NaCl and cerous ions in various concentrations. Secondary droplet formation is observed in condition of 0.03 mm cerous ions. 40

56 Secondary Droplets Further investigation of secondary droplets was performed with the SKP after completion of the 20-hour tests. For tests where a secondary droplet formed, the potential and height of the two droplets were measured by scanning laterally in a single line across the two droplets. The areas under the secondary droplets were investigated with optical microscopy and optical profilometry. These data for four instances of secondary droplet formations are shown in Figures In all of the secondary droplets investigated, line scans of the potential revealed an area of low potential under the secondary droplet, with the adjacent region of the original droplet exhibited an elevated potential relative to the bulk of the droplet. Figure 23: Photograph at the conclusion of a 20-hour droplet test with 0.03 mm cerous ions and 3.5 wt % NaCl, prior to gathering linescan data. A secondary droplet is seen at bottom left, with an associated ring of precipitation in the main droplet. 41

57 The first example shown is from a test with 0.03 mm cerous ions and 3.5 wt % NaCl. A macroscopic view of the droplet at the time of completion is shown in Figure 23. The secondary droplet is clearly seen at the bottom left edge. Also visible is an area of discoloration under the main droplet, which forms an arc around the location of the secondary droplet. This region is seen again in Figure 24 as the area indicated by 3. The potential in the area of the secondary droplet, marked as 1 in Figure 24, is depressed relative to the entire initial droplet, and this area is covered in large amounts of corrosion products. The region between the secondary droplet and the arc of discoloration is at an elevated potential relative to both the secondary droplet and the remaining area of the initial droplet. Scanning away from the secondary droplet, the potential begins decreasing at the discolored region, reaching an intermediate potential far from the secondary droplet. The discolored region appears to be corrosion product, which is supported by EDS analysis showing Al, O, and some Cl. 42

58 Figure 24: Shown at top is the SKP linescan profiles for potential and height of a 20-hour droplet test with 0.03 mm cerous ions and 3.5 wt % NaCl. At bottom is an optical image of the surface taken after testing. The arrow indicates the location of the linescan, and corresponding sites are labeled numerically. 43

59 Figure 25: Shown at top is surface seen in Figure 24 after removal of corrosion products. At bottom is topography data, taken with OP, of the boxed area in the top figure. Pitting of the surface in the regions of the secondary droplet is seen in Figure 25, with pits as deep as 20 μm. Pitting of the surface in the rest of the initial droplet area is seen in the optical micrograph, but the pits are small both in diameter and depth in comparison to the attack at the secondary droplet area. 44

60 These data indicate that the secondary droplet is associated with a large scale corrosion cell, with anodic activity occurring primarily under and near the secondary droplet, as shown by the severe pits and the depressed potential found in the SKP measurement. A wide area surrounding the secondary droplet serves as a cathodic region to support the attack, as seen in the elevated potential of the SKP data. This region is also seen visually by a lack of corrosion product, where the surface is able to support electron transfer. A similar behavior is seen in a secondary droplet formed during a test with 0.03 mm vanadate and 3.5 wt % NaCl, seen in Figure 26. Data from the SKP linescan and OP are seen in Figures 27 and 28. Figure 26: A droplet at the end of a 20-hour test with 0.03 mm vanadate and 3.5 wt% NaCl, with secondary droplet at the perimeter. 45

61 Figure 27: Shown at top is the SKP linescan profiles for potential and height of a 20-hour droplet test with 0.03 mm vanadate and 3.5 wt % NaCl. At bottom is an optical image of the surface taken after testing. The arrow indicates the location of the linescan, and corresponding sites are labeled numerically. 46

62 As with the previous example, the area under the secondary droplet is at a low potential relative to the initial droplet, indicating the site of the anodic half reaction. The region surrounding it, in an arc of surface not covered by corrosion product, is at a high potential, and acts as a large cathode area. Beyond that, where discoloration is notable on the surface, the potential decreases to an intermediate value. In this test there was a region (marked as 2 in Figure 27) between the anodic secondary droplet and cathodic region that was found to have an intermediate potential and was covered with significant amounts of corrosion product. This region may have been initially cathodic, but the deposition of corrosion products by the end of the test mitigated any catalytic activity in the region, resulting in the observed intermediate potential. The region further illustrates the separation of anode and cathode in the secondary droplet phenomenon, as this intermediate area would have incoming migration of metal cations and hydroxide from the two half reaction sites, and is where the most deposition of corrosion product would be expected. 47

63 Figure 28: Shown at top is the surface seen in Figure 27 after removal of corrosion product. At bottom are OP data of the indicated boxed area. From the optical surface and topography images in Figure 28, it can be seen that, while pitting again occurs generally throughout the droplet area, the scale of these pits are insignificant in comparison to the large pits formed under the secondary droplet region. 48

64 Figure 29: A droplet at the end of a 20-hour test with 3.5 wt % NaCl. An irregular area is evident at the edge of the droplet closest to the camera, indicated by an arrow. Data from a droplet test with 3.5 wt % NaCl is shown in Figures In this test, a region at the edge of the droplet developed with similar properties to the previously shown secondary droplets, but did not evolve into a distinct droplet. From the SKP height measurements, this region is seen as a bulge on the otherwise spherically shaped droplet. The surface and potential show the same characteristics of the tests with secondary droplets, with a low potential at the edge of the initial droplet and a circumferential region of high potential and little corrosion product deposition. A single large pit is found near the edge of the droplet where the potential was lowest, and the sample surface immediately surround it, at an intermediate potential, is covered by corrosion product. Beyond that is a ring of clean surface at high potential, acting as the cathode. Further 49

65 away, corrosion products are found on the surface and the potential returns to an intermediate value. Figure 30: Shown at top is the SKP linescan profiles for potential and height of a 20-hour droplet test with 3.5 wt % NaCl. At bottom is an optical image of the surface taken after testing. The arrow indicates the location of the linescan, and corresponding sites are labeled numerically. 50

66 Figure 31:At left is the surface seen in Figure 30 after the removal of corrosion product. At right is the topography of the indicated boxed area, as measured by OP. Shown in Figure 32 is an image from the end of a droplet test with 3.5 wt % NaCl in which the humidity in the SKP test chamber dropped below the desired range, resulting in crystallization of salt around the edge of the droplet. The rim of salt caused the droplet to take the form of disc rather than a sphere. A secondary droplet is observed, with the same common characteristics seen in the previous secondary droplets. These data are shown in Figures

67 Figure 32: A ring of salt precipitated around the edge of this droplet during testing with 3.5 wt % NaCl due to a drop in chamber humidity. A secondary droplet that developed during the test is indicated with an arrow. As was seen in other secondary droplet attacks, the most severe pitting occurred near the intersection of the initial and secondary droplets, the location experiencing a low potential relative to the bulk area of the initial droplet. The surrounding area is cathodic, with a peak in potential where the surface is uncovered by corrosion products. While the pits near the secondary droplet are still the largest found on the sample, they are much smaller than other tests that developed secondary droplets. This is likely due to the evaporation and super saturation of salt caused by the drop in humidity, possibly associated with reduced diffusion kinetics of oxygen.[12] 52

68 Figure 33: The height and potential profiles from an SKP linescan after a 20-hour droplet test with 3.5 wt % NaCl are shown in the top image. At bottom is an optical micrograph of the surface surrounding the secondary droplet after testing. An arrow indicates the location of the linescan, and corresponding sites are labeled numerically. 53

69 Figure 34: Shown at left is the secondary droplet area seen in Figure 33 after removal of corrosion products. At right is OP data showing the topography of the indicated boxed area. The dotted lines indicate the perimeter of the initial and secondary droplets. 54

70 Discussion The clearest contrast between immersion and droplet testing can be seen in the uninhibited 3.5 wt % NaCl tests. Under full immersion, the aluminum surface corrodes by localized pitting at constituent intermetallic particles. These pits are small and widespread across the surface, and there is no indication of interaction between pits to any degree. Conversely, uninhibited NaCl droplet tests formed secondary droplets, under which separation of anode and cathode occurred at a scale much larger than individual constituent particles, promoting pits supported by large regions of oxygen reduction. Similar discrepancies are seen in tests with the cerous inhibitor. At the lowest concentration, the behavior was similar to the uninhibited case. Pits reached 10 μm deep in immersion testing, again localized to intermetallic particles, but grew to 20 μm deep in droplet tests, with large areas of cathode supporting a single region of deep pits. Vanadate-inhibited tests performed similarly at the lowest concentration, with pit depth under secondary droplets double that found in immersion testing. All concentrations of vanadate and the lowest concentration of cerous inhibitors performed worse, in terms of pit depth, than uninhibited NaCl immersion tests. In these tests, similar to the secondary droplet attack, individual pits were able to grow very large, apparently also supported by large cathodic areas. These inhibitors appear to prevent most of the pitting initiation, an observation that is supported by the lack of widespread pitting on the surface, and suggested by the minor anodic inhibition found in potentiodynamic testing. However, once a pit does form under these inhibited conditions, 55

71 the reduction in cathodic kinetics is not sufficient to stem its growth. This is the worst case for inhibitors, where partial inhibition of the anodic and cathodic reactions leads to focusing of the attack at fewer sites. A deeper pit will result in faster perforation or crack initiation. The higher concentrations of cerium inhibitor, 3 mm and 0.3 mm, were shown through cathodic polarization to have a much greater effect on retarding oxygen reduction kinetics, and this effect is realized in the reduction in pit depth found at the higher concentration of cerium inhibited immersion testing. The strong inhibitive effects of chromate, both anodic and cathodic, resulted in very little corrosion in both immersion and droplet testing, providing no useful comparison between corrosion mechanisms in the two test methods. The in situ SKP potential during droplet tests provides a useful metric for comparison between similar tests. Potential characteristics and observed corrosion phenomena were consistent for cerous- and vanadate-inhibited droplet tests, where an elevated potential corresponded to inhibition of attack under the droplet, while a steady low potential was associated with the formation of a secondary droplet. However, in the case of uninhibited NaCl, a large increase in potential during the test was directly related to the formation of a secondary droplet. For the NaCl droplets, the increase in potential at the initiation of a secondary droplet may result from the area under the droplet transitioning from a mixed potential of small local anodes and cathodes to a dominantly large cathodic site. Further investigation into secondary droplet formation in inhibited droplets is needed to identify a consistent explanation for the potential behaviors observed in the presence of inhibitors. Additionally, there is a discrepancy between the 56

72 large change in potential during the SKP droplet test, as much as 500 mv, and the small variation in potential found in the post-exposure line scan, which was typically within a range of 100 mv. It should be noted that between taking these two potential measurements, the SKP chamber was opened to reposition the sample to align the secondary droplet with the axes of the SKP stage. Opening the chamber caused a drop in humidity near the sample, which could have affected the sample-droplet interface by precipitating salts. The observed phenomenon of large-scale anode/cathode separation, with anodic attack at the secondary droplet and supporting cathodic regions under the main droplet, is consistent in uninhibited droplets as well as vanadate and cerium inhibited droplets. A schematic drawing of the observed phenomenon is shown in Figure 35, and the following mechanism for the secondary droplet is proposed. The pitting attack spreads toward the edge of the droplet, and the high concentration of ions from the pits causes diffusion of these ions to the edge of the droplet and out into the absorbed water layer on the sample surface outside the droplet. The presence of ions in external absorbed water layer initiates deliquescence of the secondary droplet. This mechanism has been previous described by Tsuru, et al.[10] Further pitting attack under the secondary droplet accelerates the uptake of atmospheric water vapor. A region surrounding the anode, labeled 3, is the supporting cathodic area. The intermediate area ( 2 ) is covered with corrosion products resulting from hydrolysis of anions from the large pits and reaction with hydroxide from the cathode. The transport of these ions is necessary for completion of the electrochemical circuit created by the large separation of anode and cathode. The 57

73 diffusion of ions from the electrodes to the site of hydrolysis is indicated by the dotted arrows in the figure. Beyond the cathode, in region 4, the supporting cathodic reaction does not occur. The termination of the cathodic region may be affected by a combination of the distance from the anode, which determines the ohmic potential drop, as well as the local height of the droplet, which determines the diffusion length of oxygen from the airelectrolyte interface to the cathode. Alternatively, the lack of cathodic activity in region 4 may be due to inhibition by the layer of deposited corrosion products which are observed. While the cathode is localized to region 3, the boundary between regions 3 and 4 is not necessarily the edge of the cathode, but determined by the local ph at which aluminum hydroxide becomes insoluble. Pits initiate at site 1 underneath the initial droplet, and the area develops into a large local anode. As the pits in this region are large in size and few in number, it is unknown at what surface features these pits initiate. However there is nothing to suggest they differ from the small pits throughout the droplet area, which initiate at secondary particles. It is notable that in droplet tests, the large pits only occur at or very near the edge of the droplet. This suggests that the development of large-scale separation of the anode and cathode is dependent on the droplet morphology. In terms of pit initiation, the capacity for cathodic activity will be greater near the droplet edge, where the electrolyte layer is thinner and the diffusion path of oxygen shorter. Pits in this region would then have access to a large area of supporting cathode with a short ion diffusion path, allowing more metastable pits to grow and become stable. The proximity to the droplet edge will also restrict bulk diffusion of ions around the anode, and may result in the growth of a 58

74 local acidic environment. This effect would be greatest at the edge itself, where there is no direct source of ion transport from the cathode region. The acidic environment would then promote further pit initiation and growth in the region, resulting in the large area of anodic attack observed. Pits at the extreme edge of the droplet provide a source of ions that promote absorption of water from the atmosphere, resulting in the deliquescence of the secondary droplet. This mechanism is similar to the secondary droplets associated with filiform-like attack observed by Li, et al.[12] 59

75 Figure 35: Schematic representation of secondary droplet phenomenon. Severe pitting attack (1) occurs near the edge of the initial droplet where the secondary droplet forms. Many smaller pits also occur throughout the wetted surface area. The dotted arrows indicate ion transport from the anode and cathode sites for hydrolysis. In experiments where secondary droplets develop, there is never more than one secondary droplet associated with one large anodic region. This observation suggests that the isolation of the anode at the edge of the droplet provides cathodic protection to the rest of the droplet, preventing any additional significant attack. 60