Increased Filiform Corrosion Resistance Utilizing a Zirconium-Based Conversion Coating on an Al-Zn-Mg-Cu (AA7075-T6) Alloy as well as Selected Surface Treatments

Filiform corrosion (FFC) is a form of localized atmospheric corrosion that originates from coating defects and occurs as a thread-like formation underneath effected coatings as a result of exposure to high humidity.^1-6  FFC has been shown to cause aesthetic damage to painted aluminum alloys and is a continuing concern for the automotive industry that must be accounted for in the development of new coating technologies.⁷  Although FFC is often superficial in nature, from a structural integrity standpoint, it is unacceptable to consumers. The search for alternative solutions to replace the highly effective,⁸  but toxic,⁹  chromate conversion coatings (CCCs) is ongoing.¹⁰  Phosphate conversion coatings have been the most widely used in recent years but require costly waste-disposal procedures and energy intensive conversion processes.^11-12  Zirconium-based conversion coating pretreatments have been studied extensively and shown to be one of the most promising alternatives.^13-16  However, a knowledge gap exists regarding optimization of such conversion coatings in terms of prior surface preparation and the effect this has on corrosion performance of coated AA7075-T6 (UNS A97075⁽¹⁾) aluminum alloys in atmospheric conditions. As such, the FFC performance of a zirconium-based conversion coating on AA7075-T6 aluminum alloys underneath a model organic coating will be discussed herein as a function of surface treatment.

As such, the presence of the copper additions accelerates the Al corrosion process where the cathodic hydrogen evolution reaction (HER) and oxygen reduction reaction (ORR) are enhanced at the copper-rich sites. This amplifies the local pH increase which in turn promotes the growth of the zirconium oxide layer.¹⁸  Such an enhancement has been demonstrated for hot dip galvanized steel (zinc), cold-rolled steel, and aluminum alloy substrates¹⁹  where, in all cases, the addition of a small amount of copper to the conversion bath is a typical ingredient.^12,18,23-24  In the case of AA7075-T6, copper is already present in the Al matrix and in the form of IMPs. Cerezo, et al., reported that chemical etching of the substrate prior to immersion in the conversion bath can be used to release “functional species,” i.e., copper via dealloying of embedded IMPs that dissolve and subsequently redistribute elements such as Cu and Fe on the surface, starting at remaining IMP sites.^3,23,25-26  This results in an additional source of surface copper that can grow in size during the conversion treatment process. Prior to the conversion coating pretreatment process, surface treatments are often to remove organic contaminants and aged oxide films.^23,27  A typical surface treatment process can involve chemical cleaning, etching, and/or desmutting.²⁷  Cerezo, et al., showed that Al surfaces enriched with copper prior to conversion produced a thicker, more uniform Zr layer relative to a copper-free AA1050 (UNS A91050) alloy that relied on only the copper added to the conversion bath.²³  In this way, the surface treatment parameters can be used to control the final conversion coating thickness and homogeneity.

It should be noted that while the presence of IMPs have been shown to be beneficial to the promotion of the Zr film build, they are also known to be a driver for FFC initiation and propagation, where noble second-phase particles microgalvanically couple to the active matrix.^28-30  However, it has been demonstrated that their removal from the surface can only delay the onset of FFC and not prevent it because the FFC filaments penetrate the substrate to a depth that is sufficient to uncover fresh IMPs.^5,31 

The aims of the present study are:

1. To assess the effect of an alkaline rinse surface treatment procedure, with and without a subsequent acid deoxidation step on the performance of a Zr-based pretreatment for the protection of AA7075-T6 automotive Al alloys against FFC. A comparison of FFC filament propagation rates will be made against a control specimen with no surface treatment.

2. To elucidate the mechanisms by which FFC is inhibited as a result of the surface treatments. Surface characterization techniques will be used to assess the composition and thickness of the Zr-based pretreatments and anodic and cathodic corrosion kinetics will be monitored.

3. A further aim is to determine the effect of the aforementioned surface treatments on IMP density and copper enrichment. Any correlation between such phenomena will be identified in terms of the structure of the conversion coating and resistance to FFC.

Aluminum alloy AA7075-T6 was used throughout this study and was prepared using the three surface treatment variations listed in Table 1. All specimens were first cleaned with a solvent on as-received material and the control specimens had no further treatment. Specimens prepared with Surface Treatment A were alkaline cleaned.³²  Specimens with Surface Treatment B were prepared with an additional nitric acid/sulfuric acid deoxidation treatment step. The term “Surface Treatment” refers to the surface preparation processes listed in Table 1. The term “bare” refers to specimens where no subsequent Zr-based conversion coating pretreatment was applied. “Pretreated” refers to specimens that have been given a Zr-based conversion coating pretreatment³³  after the relevant Surface Treatment. A poly-vinyl butyral (PVB) topcoat was used only in the FFC experiments. All other experiments were conducted with no topcoat. In all cases, copper additions were present in the conversion coating bath.

Scanning electron microscope (SEM) imaging was performed on all specimens with no pretreament and no topcoat and images were collected using a concentric-ring backscatter (CBS) detector, spot size 5 and 20 kV at 200× magnification. Six separate images were collected for each specimen and ImageJ^† analysis software was utilized to estimate the average IMP size and the number of IMPs per area. IMPs were identified using the energy dispersive spectroscopy (EDS) capabilities of a Quanta 650^† SEM and Oxford Instrument’s AZtec^† software. A Helios UC G4^† Dual Beam SEM was used for secondary electron imaging and EDS mapping at 1 kV electron beam voltage (again, using Aztec^† software) of pretreated specimens with no PVB topcoat.

X-ray photoelectron spectra were collected on a PHI-Versaprobe^† (III) x-ray photoelectron spectrometer (XPS) using an Al anode achromatic x-ray source operating at 100 W (20 kV) in large area slot mode (300 μm × 800 μm). Samples were sputter-depth profiled by rastering a 20 kV Ar₂₅₀₀⁺ beam with an approximately 2 mm × 2 mm spot size over a 3 mm × 3 mm area. The spectra were collected from an area of approximately 100 μm in diameter at the center of the sputtered region. Depth profiling was performed every 0.5 min for a sputter time of up to 50 min. A sputter depth of 2.4 nm per min was estimated based on measurements made on SiO₂. This was verified by oxide measurements made using SEM imaging of a focused ion beam (FIB)-prepared cross section (images not included here). After each sputter, a survey spectrum was taken with a pass energy of 120 eV and 0.1 eV step size. Aluminum peak areas were adjusted to remove any contribution from Cu3p by subtraction of a fraction of the observed Cu2p peak area at each point from the Al2p peak area.²⁷  Area values for Cu2p and Cu3p peaks, obtained from spectra of sputter-cleaned, pure copper collected using identical settings as those used for the alloy samples, were used to obtain a scaling fraction. High-resolution spectra were also obtained for Al2p, O1s, Zr3d, F1s, and Cu2p3 peaks. These data were collected using a pass energy of 26 eV and 0.1 eV step size.

Filiform experiments were conducted in a manner described in previous publications.^34-35  Specimens of size 5 cm × 5 cm were prepared by placing a tape of 50 μm thickness to two opposite edges. An ethanol-based 15.5% PVB coating was bar cast onto the sample using the tape as a height guide creating a “trough” for the coating. Samples were air dried and a final film thickness of 30±5 μm was measured. The height-guide remained in place throughout experiments to provide edge protection. Coated samples were scribed using a scalpel to create a 10 mm scribe that penetrated through to the bulk metal. A micropipette was used to inject 100 μL of 5.4 M HCl aqueous electrolyte solution along each scratch. Samples were then placed in a chamber and held at a constant humidity of 82% and 20°C for up to 10 weeks and photographed periodically. The FFC filament lengths were measured using ImageJ^† analysis software. Data were taken from three scribes on each specimen; the five longest filaments were measured and the average was taken.

Individual anodic and cathodic polarization scans were conducted in ambient 0.6 M aqueous NaCl solution adjusted to pH 2 to simulate the conditions in the FFC experiments.¹  Scans commenced after a 2 h period at the open-circuit potential (OCP). A test area of 1 cm² was exposed to solution in each case. A scan rate of 1 mV/s was used, anodic polarization scanning was performed from −0.05 V_OCP to −0.5 V_SCE and cathodic polarization scanning was performed from 0.05 V_OCP to −2 V_SCE. Experiments were conducted on bare and pretreated specimens where no PVB topcoat was present in either case.

Figure 1 shows optical images of representative scribed defects through a PVB coating for both bare (upper images) and pretreated (lower images) specimens after a time of 28 d from the introduction of 5.4 M HCl aqueous electrolyte solution. All of the bare specimens are shown to offer no resistance to FFC. This is also the case for the control specimen when pretreated with a Zr-based conversion coating, which is shown to be very similar in appearance to the bare control specimen at 28 d. In contrast, the propagation of the filaments that formed on pretreated specimens prepared with Surface Treatments A and B is shown to be greatly hindered.

Anodic and cathodic potentiodynamic sweeps were conducted, in separate experiments, after a hold time of 2 h at OCP in 0.6 M aqueous NaCl electrolyte solution, adjusted to pH 2. The results of cathodic potentiodynamic sweeps are given in Figure 4 for (a) bare and (b) pretreated specimens, respectively. The data show that the cathodic kinetics for bare specimens is very similar. In contrast, the pretreated specimen prepared with Surface Treatment A is shown to have cathodic kinetics one order of magnitude lower than the other pretreated specimens. Similarly, very little difference in anodic kinetics can be observed between the three bare specimens (Figure 5[a]), while the pretreated specimen prepared with Surface Treatment A is shown to have anodic kinetics two orders of magnitude lower than the other pretreated specimens (Figure 5[b]). Moreover, the pitting potential is raised from approximately −0.66 V_SCE to −0.61 V_SCE.

SEM imaging was used to examine the surfaces of bare AA7075-T6 specimens, prepared with the control treatment and Surface Treatments A and B. Representative SEM-CBS micrographs of typical surfaces for all surface treatments are shown at (a) 400× and (b) 200× magnification in Figure 6 and on inspection, no obvious differences in surface morphology were observed. A summary of the average IMP density and sizes as measured using image analysis is given in Figure 6(c). Very similar values were found for each specimen. EDS analysis (not presented here) revealed that the most common particles were found to be Al-Fe, Al-Cu-Fe, Al-Cu, and Al-Cu-Fe-Mg-Si; similar findings have been well documented elsewhere.^36-39  The data imply that Surface Treatments A and B have no effect on IMP size or density on exposed surfaces relative to the control specimen based on the magnification used here.

Figure 7 gives secondary electron SEM micrographs at high magnification (up to 50,000×) of pretreated specimens. The pretreated control specimen (Figures 7[a] through [c]) was shown to have a high density of large Cu-rich islands or particles of approximately 0.5 μm in diameter (Figure 7[a]). An example of the EDS data is given in Figure 8, where a composition of Al 57.3 at%, Zr 12.3 at%, O 11.8 at%, Mg 7.2 at%, Cu 5.1 at%, Zn 3.6 at%, and F 2.2 at% was found for this region. A slight enhancement of Zr and O signal can be observed on the copper particles. The nonuniform dark regions visible in Figure 7(b) were found to be Mg-rich. In contrast, pretreated specimens prepared with Surface Treatments A and B were free from this Mg-rich layer at the surface and in both cases a uniform layer of densely-packed Cu-rich particles of approximately 100 nm in diameter can be observed (Figures 7[d] through [i]).

XPS analysis was used to study the composition and thickness of the oxide film of bare and pretreated AA7075-T6 specimens prepared with Surface Treatments A and B and the control. Oxide thicknesses were taken as the sputter time (converted to nm) for the O1s signal to decrease in intensity to half of the initial signal.⁴⁰  The XPS sputter profile data presented in Figure 9 show total elemental contributions, independent of possible mixed chemical states, for the three bare specimens. The oxide thickness of the AA7075-T6 specimens, prepared with the control and Surface Treatments A and B, were approximated to be 19 nm, 60 nm, and 4 nm, respectively, based on an assumed equivalent sputter rate of 2.4 nm/min (summarized in Table 3). Analysis of Cu2p3 and Zn2p3 signal for the control specimen (Figure 9[a]) reveals very limited elemental contribution of Cu or Zn alloying components throughout the oxide and into the bulk material. In contrast, the thick oxide “smut” layer produced by the alkaline cleaning step used in Surface Treatment A is shown to be rich in Zn and, to a slightly lesser extent, Cu shown by the enhanced Zn2p3 and Cu2p3 signals detected in this region (Figure 9[b]). The subsequent acid deoxidation step in Surface Treatment B is shown to greatly reduce the oxide thickness and completely remove any sign of Zn enrichment (Figure 9[c]). However, the elevated Cu2p3 signal in this oxide region suggests that the acid deoxidation step causes Cu enrichment.

Photoelectron spectroscopy data for AA7075-T6 specimens pretreated with a Zr-based conversion coating are given in Figures 10(a) through (c). The XPS sputter profiles show total elemental contributions for Al⁰, O1s, F1s, and Zr3d peaks. In each case, enhanced Zr3d and F1s signals were observed within the oxide region. It was possible to measure the thicknesses of a Zr-rich outer layer in addition to the total oxide thickness. This was based on a model described previously where the outer layer thickness is measured as the sputter depth at the point where the Zr signal intensity has decreased to half of its maximum value, multiplied by the sputter rate.^24,41  The total oxide thickness comprises both the Zr pretreatment oxide and the pre-existing oxide formed during the surface treatment (where applicable). Table 3 summarizes the total oxide thicknesses and outer layer thicknesses for all specimens. The total oxide thickness for the control specimen was approximately 43 nm with the Zr pretreatment (outer) layer measured as 14 nm. In contrast, the total oxide thickness for the pretreated specimen prepared with Surface Treatment A is three times that of the control, measured at 148 nm. The majority of this layer (144 nm) is made up of the Zr outer layer conversion coating as a result of the pretreatment. The specimen prepared with Surface Treatment B is shown to have the thinnest total oxide of 17 nm, however, like Surface Treatment A, the majority (15 nm) is shown to be rich in Zr.

In all cases, very low signal contributions from Zn2p3 were observed (data not given). This is in contrast to the specimens with no Zr-based conversion coating, and suggests that such enrichment phases are completely removed during the pretreatment procedure. This is likely considering the amphoteric nature of zinc⁴²  and the elevated local pH conditions during the pretreatment process.

Figures 10(d) through (f) give XPS sputter profile data for Cu2p3 and Mg1s from the pretreated control, and pretreated specimens prepared with Surface Treatment A and Surface Treatment B, respectively. The dashed lines show (1) the interface between the Zr-rich upper layer and the oxide inner layer and (2) the interface between the oxide inner layer and the bulk metal (marked as the “total oxide thickness”). The pretreated control specimen and the pretreated specimen prepared with Surface Treatment A are shown to have enhanced Mg1s signal within the oxide region. For the control specimen, the Mg1s signal appears to be minimal in the outer, Zr-containing portion and enhanced within the inner oxide region closest to the substrate. This result supports the notion that the dark regions observed in the SEM data given in Figures 7(a) through (c) are Mg-rich. In contrast, the Mg1s detected on the specimen prepared with Surface Treatment A is shown to be in the Zr-rich outer layer. No Mg1s signal was detected for the specimen prepared with Surface Treatment B.

In each case, some enhancement of Cu2p3 signal can be observed within the Zr-rich outer layer of the oxide. For the specimen prepared with Surface Treatment B, the signal is more than double that of the control and Surface Treatment A. Figure 11 gives an exploded-view schematic diagram breaking down each layer within the total oxide for each specimen.

Representative examples of high-resolution spectra are presented in Figures 12(a) through (e) for Al2p, O1s, Zr3d, F1s, and Cu2p3, respectively. These were taken from XPS high-resolution measurements made on a specimen prepared with Surface Treatment B with a subsequent Z-based conversion coating. Contributions from Al⁰ from the base metal were detected close to the interface within the Zr-rich zones for all pretreated specimens (Figures 10[a] through [c]). Previous studies have suggested that this indicates the presence of an interphase composed of Zr and Al oxide/hydroxides.²⁴  However, it is also possible that surface roughness of the samples and a possible asymmetric ion-sputtered region which broadens the interfacial region cause this behavior.

The specimen prepared with only the alkaline cleaning step (Surface Treatment A) followed by a Zr-based conversion coating pretreatment is shown to provide superior resistance to FFC for automotive AA7075-T6 as indicated in Figures 1 through 3 and summarized in Table 2. A reduction of ∼62% in the FFC filament propagation rate relative to the nonpretreated control specimen was demonstrated. Furthermore, anodic and cathodic kinetics for this specimen are shown to be at least an order of magnitude lower than the control and Surface Treatment B (pretreated specimens).

Previous reports on the characterization of Zr-based pretreatments have demonstrated a bi-layer structure where the inner layer is formed from Al oxides and the outer layer (i.e., near the surface) contains Zr, O, and F compounds.²⁴  This demonstrates that both ZrO₂ and ZrF₄ were present in the outer layer, where the latter has been shown in recent publications to be beneficial for long-term corrosion protection, where fluorine was found to leach out of the coating, causing the transformation to ZrO₂·xH₂O_(s) in the remaining oxide layer.^43-45 

The results presented here demonstrate that the ZrO₂-rich conversion coating produced on the specimen prepared with Surface Treatment A makes up 97% of the total oxide thickness with the remaining 3% made up of an Al-rich oxide at the interface with the substrate. The total thickness of this oxide is three times that of the pretreated control specimen where contributions from the Zr oxide portion make up only 33% of the total oxide. The alkaline cleaning step in Surface Treatment A seems to produce a surface that promotes a robust Zr-based conversion coating that likely limits electron transfer across the oxide to the outer layer where ORR and HER occur. The rate of anodic dissolution is also shown to be lowered (Figure 4[b]), suggesting that such a thick oxide may be capable of covering the IMPs present on the surface.¹⁶  This ultimately provides better resistance to underfilm FFC enabled by anodic undercutting at the filament tip (head), enabled by the cathodic reaction on the surrounding matrix.^46-47  It is suggested that the IMP initiation sites under the coating are not completely altered but the global cathodic rates in support of anodic undercutting are.

The poor resistance to FFC of the pretreated control specimen can be attributed to a relatively thick Al oxide inner layer that is rich in Mg. Such oxides have been shown to reduce topcoat adhesion, particularly in 5xxx series Al alloys which have high Mg content.⁴⁸  It is thought that the incorporation of magnesium into the native film acts to destabilize the oxide enabling accelerated transport of solute ions to the metal interface.⁴⁸  This is combined with an enrichment of coarse Cu-rich clusters of approximately 0.5 μm in diameter that will likely promote nonuniform ZrO₂ film-build^13,21-23  and be particularly detrimental to the advancement of FFC filaments.²⁸  Cerezo, et al., previously demonstrated such Cu “islands” on AA6014 (UNS A96014)¹⁹  and suggested that copper deposition can only occur in regions where the oxide has completely dissolved. For the current study, this suggests that the native oxide dissolution was nonuniform and insufficient.

Notably, the pretreated specimen prepared with the additional acid deoxidation step (Surface Treatment B) also showed an appreciable resistance to FFC, where a reduction of 46% was approximated relative to the bare control specimen. This is compared to only a 16% reduction for the pretreated control specimen. Both pretreated specimens prepared with Surface Treatments A and B are shown to extend the time required for FFC filaments to initiate by 24 h. This delay in the onset of FFC is highly desirable in-service because it provides an enhanced window of opportunity for suppression of initiation where any penetrating electrolytes might dry out before FFC can initiate. In contrast to the control, both nonpretreated specimens prepared with Surface Treatments A and B were shown to possess an enrichment of copper at the surface (Figure 9). The presence of these cathodic elements at the surface have been shown to promote the conversion coating build process.^12-13,18,23  It should be noted that the faster cathodic kinetics observed for bare specimens prepared with Surface Treatments A and B (Figure 4[a]) is likely a result of this surface copper enrichment. Interestingly, the Zr-based conversion coating formed on the specimen prepared with Surface Treatment B is shown to be only 17 nm in thickness, one ninth the thickness of Surface Treatment A and approximately one third of the control (pretreated specimens). It appears that the acid deoxidation step promotes the dissolution and subsequent redistribution of copper in addition to the removal of the thick Zn-rich smut layer (produced during the alkaline cleaning stage). The samples prepared with Surface Treatment B are shown to have a homogenous layer of Cu-rich particles and a Cu2p3 signal double that of the other two pretreated specimens within the oxide. The Zr oxide portion of the pretreated specimen prepared with Surface Treatment B accounts for 88% of the total thickness. The relatively high resistance to FFC demonstrated for this specimen, and the contrast in overall film thickness with the pretreated specimen prepared with Surface Treatments A, demonstrates the importance of a continuous Zr film, perhaps in suppression of global cathodic rates that support anodic undercutting. Furthermore, it is likely that minimal inner Al oxide layers found on both specimens that received a surface treatment prior to conversion coating, and not on the control, is an indicator of good topcoat adhesion in light of the improved FFC resistance.¹⁹  Ultimately, the enhanced thickening of the ZrO₂ layer observed on the pretreated specimen prepared with Surface Treatment A provides superior protection against FFC as an extra dense blocking layer and affecting both anodic and cathodic processes supporting FFC attack.

The performance of a Zr-based pretreatment conversion coating for the protection of AA7075-T6 Al alloys was assessed as a function of surface treatment. The following conclusions can be made:

• Specimens with a Zr-based conversion coating that were prepared with an alkaline cleaning step (Surface Treatment A) and the subsequent acid deoxidation step (Surface Treatment B) both provided superior resistance to filiform corrosion (FFC) underneath a PVB topcoat when compared to a control specimen. A ranking order was established where A > B > control. The FFC propagation rate was reduced by 62% and 46% by pretreated specimens prepared with Surface Treatments A and B, respectively.

• Pretreated specimens prepared with Surface Treatments A and B were shown to delay the onset of FFC by 24 h.

• A bi-layer structure was observed in all cases where a ZrO₂ outer layer was combined with an Al oxide inner layer. It was found that maximizing the ZrO₂ outer layer (i.e., a 97% contribution on Surface Treatment A) and minimizing the inner layer correlated with improved resistance to FFC.

• The study demonstrated that nanoscale copper enrichment present on the surface (produced during the surface treatment process) prior to conversion coating generation resulted in a superior conversion coating and, ultimately, superior resistance to FFC. In the absence of a prior enrichment, deposition of conversion-bath copper was dependent on the complete dissolution of the native oxide and resulted in the formation of large copper islands that would likely produce a heterogeneous layer due to excess film-build at those sites and promotion of FFC filament propagation.

• Both surface treatments were shown to have no effect on the surface intermetallic particle (IMP) density and size and FFC filament propagation rates for bare specimens were found to be similar to that of the control, despite a large variation in oxide thickness.

This study was based on work supported by the following individuals from PPG, 4325 Rosanna Dr. Allison Park, PA: K. Olson, P. Votruba-Drzal, and J. Martin. The authors gratefully acknowledge the National Science Foundation under NSF #162601 MRI acquisition of an x-ray photoelectron spectrometer for chemical mapping of evolving surfaces: a regional instrument for research and teaching. The authors would also like to thank undergraduate students James Davis, Gregory Vavoso, Samuel Ong, and Conor Moran.