Effect of iron-containing intermetallic particles on film structure and corrosion resistance of anodized AA2099 alloy

1College of Materials Science and Engineering, Chongqing University of Technology, Chongqing 400054, People’s Republic of China 2Corrosion and Protection Centre, School of Materials, The University of Manchester, Manchester M13 9PL, United Kingdom 3Science and Technology on Power Beam Processes Laboratory, AVIC Manufacturing Technology Institute, Beijing 100024, People’s Republic of China 4Chongqing University of Education, Chongqing 400067, People’s Republic of China

Iron is one of the main impurity elements present in aluminum (Al) alloys, and it is present predominantly in the form of intermetallic particles (IMPs) due to its very low solid solubility in Al matrix. 1 Most of the iron-containing IMPs, such as Al 6 (Fe, Mn), Al 3 Fe, α-Al (Fe, Mn, Si) and Al 7 Cu 2 Fe, are found in final Al alloy products because they are stable to thermomechanical treatments. 2 Usually, iron-containing IMPs are cathodic with respect to Al matrix, driving anodic dissolution of surrounding Al matrix and consequently initiating localized corrosion. A number of studies have demonstrated that iron-containing IMPs are detrimental to corrosion resistance of the high strength Al alloys. 1,[3][4][5][6] Anodizing is one of common surface treatments employed to improve corrosion resistance of Al alloys. As Al alloys are anodized, alloying elements, in forms of solid solutions and/or IMPs, may participate in the anodizing process, leading to modification of the resultant anodic film in terms of morphology and composition. The effect of alloying elements in solid solutions is relatively limited because they are minor elements compared with Al matrix and distributed uniformly. Therefore, the anodic film formed on the matrix of Al alloys is, to a large extent, similar to that formed on pure Al. Anodizing of 2xxx Al alloys, for instance, leads to extra lateral porosity in the side walls of the normal vertical pores. 7,8,10 This is associated with enrichment and subsequent anodizing of copper, as well as generation/rupture of oxygen gas bubbles due to anodizing of copper, at the film/alloy interface. [7][8][9][10] Anodizing of IMPs, however, can lead to significant modification of the anodic film because IMPs are usually heavily alloyed and heterogeneously distributed. The study on anodizing behavior of IMPs in practical Al alloys is a rather challenging task and only limited literature is available in this field. [11][12][13][14] Moon et al., 11,12 using confocal scanning laser microscopy (CSLM), found two different types of imperfections in the anodic film formed on an AA5020 Al alloy and attributed them to anodizing of Al-Mg and Al-Fe-Mg particles. Curioni et al. 13 and Saenz de Miera et al., 14 with a combination of electrochemical measurements and electronoptical observations, successfully separated the phenomena related to anodizing of IMPs in AA2024-T3 and AA7075-T6 Al alloys from those associated with anodizing of the alloy matrix.
Recently, researchers have become interested in the correlation between anodizing of IMPs and corrosion resistance of the anodic z E-mail: myl@cqut.edu.cn; huagnweijiu@cqut.edu.cn films. [15][16][17] Zhang et al.,15 with the assistance of atomic force microscope (AFM) and electrochemical impedance spectroscopy (EIS), demonstrated that IMPs in 6060 Al alloy could cause localized dissolution during anodizing and, consequently reduced film growth rate and corrosion resistance. Veys-Renaux et al. 16 investigated the structure and electrochemical behavior of the anodic film formed on different Al alloys (1050, 7175 and 2618) using scanning electron microscope (SEM) and EIS. It was found that the anodic films contained defects related to the size and the composition of the intermetallic phases precipitated within the Al matrix, and the electrochemical resistances of both the porous and the barrier layers were reduced in the presence of defects. Zhu et al. 17 investigated the influence of Si content and Si particle morphology on the corrosion protection of anodic oxide layers on Al-Si alloys, revealing that Si particles made the oxide layer locally thinner and more defective in the eutectic region, thereby increasing the ease of substrate corrosion attack.
AA2099 alloy-a representative third generation aluminumlithium (Al-Li) alloy-is increasingly used on commercial aircrafts such as Airbus A380 and Boeing B777-X. 18,19 There are two types of iron-containing IMPs in AA2099 alloy, namely low-coppercontaining Al-Fe-Mn-Cu particles (LCCP, Cu wt% = 4.0 ± 0.6) and high-copper-containing Al-Fe-Mn-Cu particles (HCCP, Cu wt% = 22.8 ± 4.5). 20 When exposed to 3.5% NaCl solution, both LCCPs and HCCPs can result in superficial pits on the alloy, and the high level of lithium in HCCPs renders them electrochemically more active than LCCPs. 6 During anodizing in tartaric-sulfuric acid solution, LCCPs are anodized at approximately one-half of the rate of the alloy substrate, leaving particle remains in the anodic film; HCCPs dissolve rapidly, resulting in cavity defects in the anodic film. 21 Further work on the corrosion behavior of anodized AA2099-T8 alloy exposed to neutral salt spray test indicates that degradation of the anodic film is localized and related to cavities arising from anodizing of HCCPs. 22 The evident link between HCCP-introduced cavities and localized corrosion of the anodized alloy promotes the end users of the alloy to evaluate the degree of such effect and to seek possible strategies. With this background, it becomes urgent for both the end users and researchers to know how HCCPs are transformed into cavities in the anodic film and how such defects affect the corrosion resistance of the anodic film.
In this work, a cast, homogenized and solution treated AA2099 alloy was firstly potentiodynamically polarized to separate the phenomena related to anodizing of iron-containing IMPs from those associated with anodizing of the alloy matrix. The alloy was then anodized at characteristic potential(s) determined in the previous step to highlight the anodizing behavior of IMPs. Finally, the effect of IMPs on the structure and corrosion resistance of the anodic film was investigated under normal anodizing conditions. The selected alloy conditions and tempers exclude the effect of precipitates on film structure and corrosion resistance, allowing only the effect of insoluble IMPs on structure and corrosion resistance of the anodic film being investigated. It is found that HCCPs dissolve preferentially at ∼ 0 V (SCE) and the corrosion susceptibility of the anodized alloy is related to the location of dissolved HCCPs in the anodic film.

Experimental
Specimens of 10 × 10 × 3 mm were prepared from an AA2099 alloy ingot, with the nominal compositions listed in Table I. The ingot was first homogenized (510 • C × 12 h + 530 • C × 36 h) and then solution treated (isothermally treated at 540 • C for 1 h and then quenched in ice water). The specimens after solution treatment were ground sequentially with 400, 600, 800, 1200 and 2500 grit silicon carbide papers and then polished using 1 μm diamond paste. In order to accurately correlate specific IMPs with characteristic film structures and corrosion events, the areas of interest on the polished specimens were carefully marked with indentations and then the compositions of the IMPs within the marked area were determined using energy dispersive X-ray (EDX). IMPs were exposed to the electron beam for as short a time as possible (< 60 s) to reduce the effect of carbon contamination. 23 The specimens subjected to anodizing treatment were masked with epoxy to expose an area of 1cm 2 . The epoxy was allowed to cure in air for 24 h. Anodizing was performed in a mixed aqueous solution containing 0.46 M sulfuric acid and 0.53 M tartaric acid (TSA) according to a patent. 24 The addition of tartaric acid has been found to effectively improve the corrosion resistance of the anodic film. [25][26][27][28][29][30] For potentiodynamic anodizing, the specimen was polarized from open circuit potential (OCP) to 10 V vs. saturated calomel electrode (SCE) at a sweep rate of 0.1 V/min, at 22, 37, or 42 • C, respectively. For potentiostatic anodizing, the specimen was anodized at a constant voltage of 14 V vs. the counter electrode, at 37 • C, for 1500 s. A large pure aluminum sheet and the sample were set as cathode and anode, respectively, with the sample surface facing the pure aluminum sheet cathode. A KR50003-500 V/3 DC power supply was used, with the current density-time responses recorded.
Selected specimens after anodizing for 1500 s were immersed in 3.5% NaCl solution for up to 84 h at 20 • C to introduce corrosion. After the immersion test, the specimens were cleaned with deionized water and then dried in a cold air stream. In order to reveal the morphology of the alloy substrate immediately beneath the anodic film, the anodic film on selected specimens was removed by immersing the specimens in a mixed solution containing 20 g/L CrO 3 and 35 mL/L H 3 PO 4 at 60 • C for 1.5 h. 31,32 The morphology and compositions of IMPs in the alloy were examined using a Zeiss Sigma HD scanning electron microscope, fitted with EDX facilities, operated at an accelerating voltage of 20 kV. The morphology of the anodic film before and after the corrosion test was examined using the same microscope, but operated at a reduced accelerating voltage of 2 kV to avoid charging effects. Cross sections of the anodic film before and after the immersion test were obtained either through mechanical bending or by ultramicrotomy (Leica Ultracut) using a diamond knife. 33 For mechanical bending, selected specimens with a notch on their back side were immersed in liquid nitrogen for 10 min and then bent using two pliers, with the film side opposite to the bending direction. Ultramicrotomy was also employed for sequential cutting to generate suitable surfaces from the specimen for tomography. 34 Volumetric SEM images from a set of sequentially ultramicrotomed serial surfaces were acquired at accelerating voltage of 1.5 kV. Successively acquired 2D images were aligned to generate 3D volumetric reconstructions. Figure 1 shows backscattered electron micrograph of a mechanically polished alloy specimen, showing the distribution of IMPs (with size varying from several to tens micrometers). Twenty IMPs/spectra were selected at random, with their compositions analyzed by EDX before the anodizing and corrosion test (Table II). The particles are classified based on the Cu/Fe ratios since the copper and iron levels in the alloy matrix are low due to their low solubility in aluminum, and the characteristic X-rays of copper and iron are mainly generated from the particles. The particles mainly fall into two groups, i.e. particles with Cu/Fe ratios between 0.3-0.4 (i.e. LCCPs) and particles with Cu/Fe ratios between 2.2-2.8 (i.e. HCCPs). An exception is found for Particle 8, which corresponds to a multiphase particle and has a Cu/Fe ratio of 1.4; as shown later, it behaved similarly as other particles with Cu/Fe ratios between 2.2-2.8. According to previous work, 6 HCCPs also contain high level of lithium, which could not be detected by the EDX employed in this work. Figure 2 shows current density-voltage responses recorded during potentiodynamic polarization of the alloy from OCP to 10 V (SCE) at different temperatures. The current densities generally increased with increasing voltages and temperatures. Particularly, a current density surge appeared between −0.25 V (SCE) and 0.25 V (SCE). The current density surge became stronger and was slightly shifted to lower voltages with increasing of temperature. According to literature, 13,14 such current density surges suggest preferential anodizing of IMPs relative to alloy matrix. Thus, a selected specimen was potentiostatically polarized at 0 V (SCE) for 300 s in order to correlate the current density surge to specific type of IMPs in the alloy. Figure 3a shows a backscattered electron micrograph of HCCPs before anodizing; EDX spectra were taken from two locations of the particle with different brightness (Spectra 8 and 9). After anodizing at 0 V (SCE), at 22 • C, for 300 s, the particle was transformed into nanoparticles of 50-200 nm diameters, despite of the composition variation (Figs. 3b, 3c and 3d). EDX analysis of the nanoparticles shows trace amounts of iron (0.4 wt%) and manganese (0.5 wt%), some oxygen (1.6 wt%) and high level of copper (70 wt%), suggesting dealloying of the HCCPs during anodizing. Figs. 3e and 3f show the morphology of another HCCP after anodizing at 0 V (SCE), at 37 • C, for 300 s. Under the selected conditions, a surface cavity (as indicated by the arrow in Fig. 3e)  nanoparticles were revealed more clearly, probably due to reduced residual corrosion products. This suggests that increasing temperature accelerated the dealloying and dissolution process of HCCPs. The suggestion is also supported by the relatively stronger current density surge and its shift to lower voltage end at higher temperatures. Figure 4 compares the morphology of a multiphase particle consisting of LCCP (Particle 6) and HCCP (Particle 11) before (Fig. 4a) and after (Fig. 4b) anodizing at 0 V (SCE), at 37 • C, for 300 s. Note that in Fig. 4a the HCCP parts are relatively brighter than the LCCP parts, in the backscattered SEM image, due to much higher level of copper contents in HCCPs than LCCPs. However, in Fig. 4b the HCCP parts are revealed darker compared with the LCCP parts after anodizing under the selected conditions, confirming preferential dissolution of HCCPs.

Results
The alloy was then anodized at 14 V, at 37 • C, for 1500 to generate a relatively thick anodic film. The recorded current density-time response curve (Fig. 5) agrees with that of the AA2099-T8 alloy anodized under the same conditions, 21 suggesting that the selected alloy conditions (casted, homogenized and then solution treated) does not evidently change the general anodizing behavior of the alloy. Figs. 6a and 6b show secondary electron micrographs of the alloy after anodizing, in two typical areas. With the determination of IMP compositions in these regions before anodizing (Table II), it is deduced that HCCPs on the alloy surface in these regions were transformed into surface cavities (Figs. 6a and 6c) while an anodic film was formed on LCCPs (Fig. 6d). Similar phenomena were observed in other regions and the results were not included to avoid repetition. Figure 7 shows the cross sections of the alloy after anodizing, revealing a partially anodized particle (Fig. 7a) and a cavity (Fig. 7b) surrounded by an anodic film of ∼4.6 μm thickness. As expected, the partially anodized particle corresponds to LCCP (75.0% Al-14.8% Fe-5.4% Mn-4.8% Cu) and the anodic film formed from the particle is indicated by the arrow in Fig. 7a. The cavity is about 5.4 μm long and 1.6 μm thick. Nanoparticles (∼100 nm in diameter), similar to those shown in Figs. 3d and 3f, are revealed on the bottom of the cavity (see inset of Fig. 7b). It is suggested that the cavity was caused by dissolution of a HCCP as a consequence of anodizing. Due to rapid dissolution of the HCCP and consequent formation of the cavity, a sunken region developed in the alloy substrate below the dissolved particle, with its profile indicated by the dashed curve in Fig. 7b. Figure 8 shows 3D volumetric reconstruction of the anodized alloy. The cavities in the anodic film, arising from dissolved HCCPs, are evident in Fig. 8a. The anodic film grew into the alloy substrate in the sunken regions and is revealed as hemispherical features in Fig. 8b. The relatively large curvatures of the hemispherical features suggest sudden change of film growth direction in these regions. As discussed later, additional defects might be introduced in the anodic, deteriorating corrosion resistance of the anodic film in these regions.
In order to investigate the effect of anodized IMPs on corrosion resistance of the anodized alloy, the specimen after anodizing at 14 V (SCE), at 37 • C, for 1500 was immersed in 3.5% NaCl solution for up to 84 h. Figure 9 shows open circuit potential-time curves of the bare and anodized AA2099 Al alloy recorded during the immersing test. The OCP of the bare alloy increased from −1.05 V to −0.59 V in the first several seconds, then it decreased slightly to −0.63 V in the next 4 hours; thereafter, it dropped twice before finally stabilizing at −0.82 V. It is believed that the OCP-time response of the bare alloy suggests passivation, initiation of localized corrosion and propagation of localized corrosion during the immersion test. The OCP of the anodized alloy fluctuated mildly between −0.63 and −0.67 during the immersion process. Compared with the bare alloy, there is no sharp increase of OCP at the commencement of immersion for the anodized alloy because a porous anodic film has already existed on the alloy surface. The relatively stable OCP of the anodized alloy suggests that the anodic film delays the corrosion process of the anodized alloy. Figure 10 shows secondary electron micrographs of the anodized alloy after immersion for 24 h. Petal-like cracks varying from several to tens micrometers are revealed on the specimen surface (Figs. 10a  and 10b). Scrutiny of the specimen surface indicates that the petal-like cracks are absent in the surface regions containing dissolved HCCPs (Fig. 10c) or anodized LCCPs (Fig. 10d, the part as labeled). This suggests that the petal-like cracks are related to structures which are within or below the anodic film. Figure 11a shows the same area as shown in Fig. 10d, however, after removal of the anodic film, revealing remained LCCPs and sunken regions left by HCCPs. A circular region (as indicated by the dashed-line arrow in Fig. 11a) is revealed in the periphery of the sunken region. Figure 11b shows the framed area K at increased magnification, revealing different morphology in the bulk surface and the circular regions, with the boundary being schematically indicated by the dashed line. The regular scallop feature on the bulk surface corresponds to the pore base and is generated by normal growth of the anodic film. The presence of such feature suggests integrity of the film and intact of the alloy substrate in this region after the immersion test. The irregular feature in the circular region, which is also found in the sunken region, suggests that the electrolyte penetrated the anodic film and reached the alloy substrate in this region, resulting in localized corrosion of the anodized alloy. Figure 11c shows the framed area L at increased magnification, revealing regular scallop feature in the surrounding regions of the LCCP. This indicates that the partially anodized LCCP did not lead to localized corrosion of the anodized alloy under the selected conditions. Another region after film removal is shown in Fig. 11d. Note that all attacked sites (as suggested by the circular region and indicated by solid-line arrows) are associated with sunken regions (i.e. dissolved HCCPs). However, the circular region is not found in the periphery of the sunken regions indicated by dashed-line arrows, suggesting that not all dissolved HCCPs led to localized corrosion of the anodized alloy after immersion in 3.5% NaCl solution for 24 h.
With the consideration that the petal-like cracks are associated with localized corrosion of the anodized alloy, a cross section of the anodic film containing petal-like cracks was prepared using ultramicrotomy (Fig. 12a). The anodic film in the cracked area detached from the alloy substrate. Additionally, a cavity and a remained particle were detected below the detached anodic film, in the central region of the petal-like cracks. EDX analysis of the remained particle gives 74.0% Al, 1.9% Fe, 0.6%Mn, 2.8% Cu, 1.9% Zn, 0.7% Si, 0.5% S, 13.5% O and 4.1% C, confirming that it belongs to a partially anodized LCCP. The co-existence of the cavity and remained particle suggests that the cavity is at the film/alloy interface because the remained particle should be attached to the alloy substrate. Little anodic film is found at the side wall of the cavity, suggesting that the cavity was formed just before the termination of anodizing. Scrutiny of the attacked region reveals corroded alloy substrate below the detached anodic film and different film morphology above the corroded alloy substrate (Fig. 12b). It is believed that corrosion products arising from the corroded alloy substrate modified the surrounding anodic film, probably through a diffusion process. Figure 12c shows the cross section of another film region, again, revealing cracked anodic film, cavity at the film/alloy interface, corroded alloy substrate below the cracked anodic film and modified anodic film. There are no evident petal-like cracks revealed on the surface of the anodic film in this region, suggesting that the corrosion event shown in Fig. 11c is still in earlier stages compared with that shown in Fig. 11a. Figs. 11d and 11e show the cross sections of the anodic film region containing cavities several hundred nanometers above the film/alloy interface. Neither corrosion attack of the alloy substrate nor modified anodic film is revealed in these regions. Thus, it is suggested that the location of the cavities (or dissolved HCCPs) affects the corrosion behavior of the anodized alloy. Figure 11f shows the cross section of the anodic film region containing a partially anodized LCCP. As expected, no evident corrosion phenomenon is revealed in the alloy substrate below the partially anodized LCCP.

Discussion
It is demonstrated that HCCPs dissolved preferentially at ∼0 V (SCE) in tartaric-sulfuric acid solution at 22, 37 and 42 • C (Fig. 2).  The potential is comparable to that of S (Al 2 CuMg) phase in AA2024-T3 and AA7075-T6 alloys anodized under similar conditions. 14 The preferential dissolution of HCCPs proceeded through a dealloying mechanism, leading to formation of copper-rich nanoparticles of 50∼200 nm diameters. Theoretically, the remained copper-rich nanoparticles might be anodized when the anodizing voltage is above 3 V if they are electrochemically connected to the alloy substrate. 13 However, a continuous and stable anodic film is unlikely to be formed on copper-rich nanoparticles because large amount of oxygen gas will be generated during anodizing of copper. 25,[34][35][36][37] Thus, it is deduced that HCCPs would dissolve, completely and rapidly, during normal anodizing upon exposure to the electrolyte and electric filed.
Plan view of the alloy after normal anodizing and subsequent immersion in 3.5% NaCl solutions for 24 h reveals petal-like cracks in the anodic film (Figs. 10a, 10b and 10d). After film removal, a sunken region is shown in the alloy substrate beneath the petal-like cracks (Fig. 11a), suggesting that the petal-like cracks are related to dissolved HCCPs. The corrosion attack of the alloy substrate in the sunken and its surrounding regions (Fig. 11b) indicates that the dissolved HCCP is responsible for the penetration of the electrolyte through the anodic film in the localized region. It is noted that there were always sunken regions in the corrosion sites (Sites 1 in Fig. 11a and Sites 2 and 3 in Fig. 11d); while no evident corrosion phenomena were detected at the sunken regions indicated by dashed-line arrows in Fig. 11d. Such observation indicates that the dissolution of HCCP is the necessary but not sufficient condition for the development of localized corrosion in the anodized alloy under the selected conditions. Figure 11d indicates that the corroded sunken regions are not necessarily larger or deeper than the intact sunken regions, suggesting that the size of the HCCP is not the decisive factor for initiating  localized corrosion. Remained LCCP is only occasionally observed at the corrosion sites (Figs. 11d and 12a), suggesting that co-existence of HCCP and LCCP is not the decisive factor either. Cross-sectional view of the alloy after anodizing for 1500 s and subsequent immersion in 3.5% NaCl solution for 24 h indicates that dissolved HCCPs (or cavities) at the film/alloy interface (Figs. 12a and 12c) are more likely to cause localized corrosion than those away from the film/alloy interface (Figs. 12d and 12e). Therefore, it is deduced that the location of the dissolved HCCPs (or cavities) plays a critical role in initiating localized corrosion. As suggested in Fig. 8b, there should be a sudden change of film growth direction in the sunken regions, probably from the normal direction of the bulk surface to the normal direction of the cavity surface. Such localized film growth might introduce defects in the anodic film, deteriorating the corrosion resistance of the anodic film. However, full understanding of the corrosion mechanism is still not clear and will be the subject of future work.
Remained LCCPs are frequently observed at the film/alloy interface because of the relatively smaller anodizing rate of LCCPs relative Figure 11. Secondary electron micrographs of solution-treated A2099 alloy after anodizing for 1500 s, subsequent immersion in 3.5% NaCl solution for 24 h and removal of the anodic film: (a) the same area as shown in Fig. 10c after film removal; (b) the framed area K at increased magnification; (c) the framed area L at increased magnification; and (d) another area containing both corroded and intact sunken regions.  to Al matrix. As a consequence, local film thickness should be slightly reduced at the sites of remained LCCPs. However, no evident corrosion phenomena were detected at the sites of partially oxidized LCCPs under the selected conditions (Figs. 11c and 12f). This indicates that LCCPs have less detrimental effect than HCCPs on the corrosion susceptibility of the anodized alloy. Therefore, it might be interesting for metallurgists to develop a procedure to transform HCCPs into LC-CPs if it is technically or economically not possible to further reduce iron-containing IMPs in the alloy.

Conclusions
1) High-copper-containing Al-Fe-Mn-Cu particles (HCCPs) dissolved preferentially at ∼ 0 V (SCE) in tartaric-sulfuric acid solution through dealloying, forming copper-rich nanoparticles of 50-200 nm diameters at the sites of the dissolved particles; such electrochemical activity of HCCPs explains the dissolution of them and formation of cavity defects of micrometer dimensions in the anodic film under normal anodizing conditions. 2) Not all dissolved HCCPs led to localized corrosion of the anodized alloy after immersion in 3.5% NaCl solution for 24 h; the corrosion resistance of the anodic film is associated with the location of dissolved HCCPs, with those at the film/alloy inference being more likely to cause localized corrosion. 3) Although a thinner and more flawed anodic film would be formed on low-copper-containing Al-Fe-Mn-Cu particles (LCCPs) com-pared with the bulk alloy substrate, it seems that LCCPs have much less detrimental effect on corrosion resistance of the anodic film. Therefore, it is possible to improve the corrosion resistance of AA2099 Al-Li alloy by transforming HCCPs into LCCPs.