The Role of Chromate Conversion Coating in the Filiform Corrosion of Coated Aluminum Alloys

The experiments were carried out using commercially coated rolled AA6016 T4 aluminum alloy (composition by weight: 0.94% Si, 0.32% Fe, 0.07% Cu, 0.043% Mn, 0.44% Mg, remainder Al). The samples were pretreated by a conventional CCC (Alodine 1200) process, followed by a full paint covering (20 μm thick electrodeposited coating and a 100 μm thick top coating). Chomating was applied after degreasing and etching the aluminum, using Alodine 1200 (15 g L⁻¹, pH 1.85, 30°C). The thickness of the CCC was approximately 1 μm, which corresponds to approximately 1 g m⁻². A final coating of transparent polyurethane, 10 μm thick was applied to samples destined for XANES measurements. Some samples were treated by a chromium-free conversion coating process (e.g., tricationic phosphate and a commercial titanium-zirconium treatment) before being painted. The phosphate coating was applied using a commercial zinc phosphate solution at 50°C, to obtain a phosphate layer of 2.6 g m⁻². Some samples were treated with Ti-Zr in order to compare how different pretreatments inhibit filiform corrosion. This treatment was carried out at the premises of a chemical supplier. Two scratches forming an X were made to simulate a defect. The total scribe length was 7 cm and the scribe width was 0.5 mm. Filiform corrosion was initiated by contamination with sodium chloride followed by an exposure at a relative humidity of 85% and at two temperatures, 25 and 35°C.¹⁶ Three replicates of each pretreatment were tested. Samples of AA6016 covered by Alodine without any organic coating were used for Raman investigations. The CCC was carefully removed by fine grinding on a part of the sample. The corrosion was initiated by placing a droplet of 0.1 N HCl onto the bare metal, which was then exposed to humid air (95% relative humidity, RH) at room temperature (20°C).

XANES is characterized by two regions. Below a characteristic energy, the absorption edge, relatively few photons are absorbed. As the energy increases above the absorption edge, incident X-ray is strongly absorbed. The absorption edge corresponds to the excitation of a core electron from a bound state level to the continuum vacuum. The position of the edge shifts to higher energy as the valence of the Cr species increases.^{13 14} A distinct pre-edge peak is observed for Cr(VI) in tetrahedral coordination which is not observed for Cr(III) species. The XANES spectrum is a linear combination of spectra of Cr(III) and Cr(VI) components, and thus, the ratio of the intensity above the edge to the intensity below the edge provides an estimate of the Cr(VI) concentration.

Spectra were obtained near the Cr K absorption edge (5990 eV). The experiments were performed in air on beam X 26A (beam diam 15-35 μm) at the National Synchrotron Light Source of Brookhaven National Laboratory.

Micro-Raman spectra were obtained with a LABRAM spectrometer from ISA-Jobin Yvon. The red line at 632.8 nm of a He-Ne laser with a constant power output of 9.7 mW was used. Neutral filters with optical densities of 1 or 2 were used when the product was unstable under the laser light. The laser power was chosen at a value that did not modify the spectra due to thermal effects. The area investigated was limited to 1.2 μm² when using an 80 ULWD objective lens. A confocal microscope was used with hole diam fixed at 120 μm in order to separate contributions coming from the top and the bottom of the analyzed object. The resolution in depth was approximately 20 μm when the product was transparent to the excitation radiation.

Figure 1 shows the area of underfilm corrosion following 6 weeks of exposure at 85% RH as a function of temperature for the different chemical pretreatments: chromating, phosphating, and Ti-Zr treatment. The length of the longest filament is given on top of the bars. The rate of propagation of filiform corrosion was lower for chromating than it was for the two other chemical pretreatments. Similar results were found independently of the temperature in the range from 25 to 35°C. This is consistent with a study performed by Wernick and co-workers,¹⁷ which showed that chromate films have a lifetime after salt spray exposure that is twice that of chromate-phosphate and phosphate films.

Zoom In Zoom Out Reset image size

Figure 2 shows a XANES spectrum for uncorroded AA6016 aluminum alloy covered with Alodine 1200. The spectrum was normalized with respect to the edge intensity. Thus, the intensity of the pre-edge peak at 5990 eV relative to the edge intensity indicates the fraction of Cr(VI) in the CCC. There was about 19% of Cr(VI) in the AA6016 samples treated with Alodine 1200. This agrees well with previous studies, which have found Cr(VI) content between 15 to 27%.⁸ Similar results to those shown in Fig. 2 were found below a 10 μm organic coating. The ratio between Cr(VI) to the total chromium in this case was 19%, which lies within the previous range, although the total signal was lower than that shown in Fig. 2. Filiform corrosion was then initiated according to the experimental procedure described above. Figure 3 is a laser scan micrograph of a filament formed on chromated AA6016 covered with a 10 μm polyurethane coating after 6 weeks exposure at 85% RH and 25°C. Figure 4 shows XANES maps at characteristic energies for Cr(VI) (Fig. 4a) and total Cr (Fig. 4b) over the filament shown in Fig. 3. The total amount of chromium decreased in regions close to the scratch and at some locations in the filament. Cr(VI) was more homogeneously depleted in the entire filaments, although some regions showed a higher decrease in the amount of Cr(VI). Both Cr(VI) and total Cr decreased to a great extent in the scratch (see Fig. 4c and d). This is shown more clearly in Fig. 5, which shows the ratio of Cr(VI) to total Cr in the filament (Fig. 5a) and in the scratch (Fig. 5b). Cr(VI) accumulated in the scratch close to the opening of the filament, whereas Cr(VI) was mainly depleted at the opening of the filament and at some locations along the filament. Hence, Fig. 5 shows that the change of Cr(VI) content was higher than the change in the total chromium content at specific locations in the filament. It is believed that these locations correspond to former heads of filament, which are known to act as pits.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Comparing Fig. 4 and 5 shows that the amount of Cr(VI) decreased in the filament while the total amount of Cr remained almost unchanged. This indicates that Cr(VI) was reduced to Cr(III) in the CCC. Filiform corrosion is a dynamic process, and the active head moves continuously as long as the humidity is sufficiently high. Cr(VI) was not depleted at all (or only to a very small degree) at the current location of the head, which suggests that the corrosion inhibition of the anodic head was not completed when the measurements were performed. Cr(VI) was depleted and the total amount of Cr had decreased at the opening of the filament. Thus, it seems that Cr(VI) is preferentially leached at the opening of the filament. The chromate leached from the CCC then accumulated in the scratch, close to the opening of the filament. The location at which chromate accumulated was then the location of an early pitting attack. At these locations, an accumulation of aluminum corrosion products was observed. It should be noted that the areas of highest concentration of Cr(VI) correspond to the center of the pits (Fig. 4). The results clearly show that chromate is released from the CCC and migrates to the nearby uncoated scratch region. This agrees well with recent results obtained by Raman microspectroscopy.¹⁰

Raman microspectroscopy is well adapted for the study of Cr(VI) compounds and this technique has been widely used to provide information on the structure of Cr(VI) species.^{18 19 20 21} Figure 6 shows the Raman spectrum of AA6016 treated with Alodine 1200. The spectrum has peaks of various widths, at ∼240, 500, 855, 1670, 2083, and 2130 cm⁻¹. The peak at 855 cm⁻¹ arose from the presence of Cr(VI), from the symmetric stretching mode of Cr(VI)-O. The spectrum of CCC was similar to that of chromate in aqueous solution, which has a band at 840 cm⁻¹.^{21 22} Previous studies on CCC using Raman spectroscopy have shown that this band is characteristic of chromate vibration in a polymeric mixed oxide Cr(III)-Cr(VI) that is present in Alodine CCCs.²³ The broad bands centered at ∼500 and ∼240 cm⁻¹ arose from the symmetric stretch and from the bending of Cr-O-Cr, respectively, suggesting that a dimeric species may be present.²⁴ The bands at 2083 and 2130 cm⁻¹ arose from CN vibration in FeCN from the treatment bath, while HOH bending appears around 1670 cm⁻¹.²³

Zoom In Zoom Out Reset image size

Raman microspectroscopy was performed directly on the corrosion products after a droplet of aggressive solution (HCl, 0.1 N) was deposited inside the scratched area (the scratch penetrated down to metallic aluminum). Two cases were investigated: (i) the droplet of HCl extended onto the Alodine coating and the droplet was not in contact with Alodine. HCl was carefully deposited onto the samples, and was left for 10 min in air. The samples were thereafter exposed to humid air (95% RH) at room temperature (20°C) for various times before Raman spectra were recorded.

In the first case, Alodine dissolved rapidly during the first hour of exposure to humid air. Spectrum a in Fig. 7 shows the Raman spectrum from such samples. It has a chromate peak at 850 cm⁻¹, which is the same position as for Alodine, showing that chromate was present in the scratch. After three days in humid air, the corrosion layer became thicker due to the formation of a corrosion product, probably aluminum hydroxide gel that contained carbonate. The carbonate peaks at 1065 and 1100 cm⁻¹ and the Al-O peak at 550-600 cm⁻¹ support this conclusion (spectra b and c in Fig. 7).⁷ The relative intensity of the Cr(VI) peak in spectra from inside (spectrum b, near the metal surface) and outside the corrosion products (spectrum c, near the external surface) shows that the concentration of chromate was higher close to the metallic part. Furthermore, the widths and the positions of the bands were different than those of the peak recorded from Alodine (855 cm⁻¹), as shown in the enlargement of this part of the spectra. The peak arising from inside the corroded layer can be deconvoluted into two peaks, one centerd at 840 cm⁻¹ corresponding to free chromate and another at 870 cm⁻¹ corresponding to the an aluminum/chromate compound during the corrosion process.¹⁹ The peak in the spectrum recorded outside the corroded layer was symmetric and centred at 860 cm⁻¹, close to the value of peak recorded from Alodine.

Zoom In Zoom Out Reset image size

In the second case, in which there was no contact between HCl and Alodine 1200, pitting corrosion of aluminum resulted in the formation of an aluminum hydroxide containing carbonate, after 1 day of exposure. Raman bands attributed to the vibrations of Al-O in a tetrahedral environment and the vibrations of carbonate were clearly observed at 550-600 and 1065-1100 cm⁻¹, respectively⁷ (see spectrum a, Fig. 8). The pit was located a few tens of micrometers from the CCC. It should be stressed that, an acidic salt layer of aluminum chloride and its hydrolysis products was formed during the first steps of pitting process.^{5 25} The initial salt layer incorporated carbonates during the growth of the pit, and was transformed into an aluminum hydroxide-containing carbonate. A mechanism of the propagation of filiform corrosion is detailed elsewhere.⁷ Chromate appeared in the corroded area after only three days of exposure to humid air. Note that Raman investigations were not performed at the exact center of the pit because the signal was disturbed by fluorescence coming from the thick layer of corrosion products. All measurements were recorded between the CCC and the center of the pit. The broad peak at approximately 850 cm⁻¹ in the Raman spectrum from corrosion products close to the metal surface (spectrum b, Fig. 8) arose from chromate. The Raman peak was shifted to ∼900 cm⁻¹ when the Raman laser was focused at the outermost layer of corrosion products (spectrum c, Fig. 8). These observations show that chromate species were released from the Alodine film and migrated towards corroded areas, where they reacted with aluminum corrosion products near the pitting place. It has been suggested that the increased concentration of chromate around a pit results in the formation of a Cr mixed oxide and/or an Al/Cr mixed oxide.¹⁰ The latter is more likely to form near the pit, because the formation of a Cr(VI)/Cr(III) mixed oxide is inhibited in the conditions in the pit.¹⁹

Zoom In Zoom Out Reset image size

Several explanations have been proposed for shifts of peak position of chromate toward higher wavenumbers. A shift has been observed when increasing chromium content and dehydratation of the chromate layer on chromium oxide supported by alumina, and this shift has been attributed to the formation of a polymeric chromate compound.²⁰ A shift also occurs when passing from chromate (840 cm⁻¹) to dichromate (910 cm⁻¹) in aqueous solution.¹⁸ The equilibria of Cr(VI) species in solution depend on the pH: (dichromate) and (bichromate) exist in acid media, whereas is the sole species at pH higher than 8. This leads us to expect that will predominate at the interface between the metal and the corrosion products, since the environment here is basic.