Characterization of 2024 T3 Aluminum Alloy after Rare Earth Desmutting

The effect of CeCl 3 and CeCl 3 + H 2 O 2 additions to a nitric acid solution for desmutting of 2024 T3 aluminum alloy has been investigated using scanning electron microscopy and energy dispersive X-ray analysis, potentiodynamic polarization and X-ray photoelectron spectroscopy. No S-, θ -phase and Al-Cu-Fe-Mn-(Si) particles were detected on the alloy surface after desmutting with added CeCl 3 . A similar result was attained with addition of H 2 O 2 , although with slower removal of the particles. In contrast, θ -phase, Al-Cu-Fe-Mn-(Si) and dealloyed S-phase particles remained following desmutting in nitric acid alone. The absence of the particles on the alloy surfaces following desmutting in the presence of CeCl 3 was also conﬁrmed by potentiodynamic anodizing of the alloy in H 2 SO 4 solution.

Surface pre-treatment of aluminum alloys is important for improvement in the properties and the anticorrosion performance of anodic films and conversion coatings used in the aerospace industry. 1 The main purpose of the pre-treatment is to prepare a reproducible, chemically active surface for application of the subsequent treatment. The pre-treatment of the alloys involves several steps, including degreasing, acid or alkaline etching and desmutting.
Enhanced mechanical properties of the alloys are achieved by addition of alloying elements that form second phase particles. 2,3 The presence of cathodically active intermetallics on the alloy surface sustains cathodic reactions that cause susceptibility to corrosion due to microgalvanic coupling between the intermetallics and the aluminum matrix and the associated alkiline corrosion. Porous anodic films formed on Al-Cu alloys are less regular than films produced on high purity aluminum due to the incorporation of copper into the film and oxygen evolution during anodizing. 4 Furthermore, the anodic layer formed on the intermetallic particles has an altered morphology and contains various defects, thus providing reduced corrosion protection compared with the aluminum matrix. [5][6][7] For the last two decades, rare earth metals (cerium, neodymium and lanthanum) have been investigated in surface treatments for enhancing the corrosion resistance of anodized aluminum alloys. [8][9][10] The treatments can reduce the number of residual cathodic intermetallic particles on the alloy surface and hence enable formation of an anodic film that contains fewer defects. In this regard, relatively little information about cerium-containing desmutting solutions has been published. 8,[11][12][13] Hughes et al. 8,11 proposed a high etch rate desmutting solution containing 1.28 M HNO 3 + 0.5 M H 2 SO 4 + 0.04 M F − + 0.05 M ((NH) 4 Ce(SO 4 ) 4 · 2 H 2 O) with / without oxidants (K 2 S 2 O 8 , H 2 O 2 ) for 2024 T3 aluminum alloy. The addition of fluoride ions and cerium (IV) sulfate resulted in removal of the S-phase particles from the surface, partial dissolution of Al-Cu-Fe-Mn intermetallics with the loss of Fe, Al, Mn and re-deposition of copper oxide products with dimensions of 200 nm.
Subsequently, Hughes et al. 12 and Kimpton et al. 13 investigated a lower etch rate cerium-containing desmutter (0.5 M H 2 SO 4 + 0.05 M (NH) 4 Ce(SO 4 ) 4 · 2 H 2 O) for pre-treatment of 2024 T3 and 7075 T6 aluminum alloys, which was shown to improve the performance of subsequently applied cerium conversion coating.
In the present work, alkaline etching followed by desmutting in nitric acid with CeCl 3 and CeCl 3 / H 2 O 2 additions are investigated for the removal of intermetallics from the surface of an 2024 T3 alloy. The effect of the additions on the anodizing behavior and the corrosion protection of the alloy are also investigated.  3 , the pH of the base solution was in the region of −0.14, and it is unlikely that the low levels of CeCl 3 and H 2 O 2 additions resulted in significant differences in pH values. The etching and desmutting processes were followed by rinsing in deionized water and drying in a cool air stream.
Prior to potentiodynamic polarization tests and anodizing, the individual specimens were examined by scanning electron microscopy (SEM) and energy dispersive X-ray analysis (EDX), using a Carl Zeiss Ultra 55 instrument at accelerating voltages of 1.5 and 15 keV, respectively, in order to determine their chemical compositions and surface morphologies. EDX analysis was performed at an accelerating voltage of 15 keV. Intermetallic particles on alloy before etching and desmutting were observed by SEM using block specimens prepared using ultramicrotomy with a diamond knife using a Leica Ultracut Instrument.
Potentiodynamic polarization tests were carried out in naturally aerated 3.5 wt% NaCl solution at ambient temperature using a threeelectrode cell with a Solartron SI 1287 potentiostat. The alloy specimens, with an exposed area of 1 cm 2 , a platinum electrode and a saturated calomel electrode (SCE) were used as the working electrode, the counter electrode and the reference electrode, respectively. The potential was varied between −0.03 V vs. the open circuit potential (OCP) and −0.2 V vs. the SCE for anodic polarization, and between 0.03 V vs. OCP and −1.8 V vs. SCE for cathodic polarization. A scan rate of 0.16 mV/s was employed.
Potentiodynamic anodizing was undertaken in stirred 0.46 M H 2 SO 4 electrolyte at ambient temperature using a similar threeelectrode cell with the potential scan performed from 0 V vs. OCP to 14 V vs. SCE at 16.66 mV/s, with an exposed specimen area of 1 cm 2 .
The surface composition was investigated using a Kratos Axis Ultra DLD XPS spectrometer with a monochromated Al K α source (1486.6 eV). The operating pressure was less than 2 × 10 −8 Pa and the analysis area was 700 × 300 μm 2 . The C1s hydrocarbon peak (285.0 eV) was used for calibration of the binding energies. Individual specimens of area of approximately 0.25 cm 2 were mounted on a stainless steel holder using double sided adhesive tape. Aluminum oxide thicknesses were determined from the ratio of the Al 3+ /Al 0 intensities and calculated using the following equation: 8,12 where is the photoemission angle (angle of detection), λ Al is the mean free path of electrons with kinetic energy corresponding to the Al2p photoelectron line, I Al 3+ and I Al 0 are the intensities of the oxidized and metal components obtained after fitting of the Al2p peak, and I 0 Al 0 and I 0 Al 3+ are the intensities obtained from the pure aluminum. The variation in values of thickness is considered real and confidence in the values arises from repeated measurements for each condition. All data processing was carried out using CasaXPS version 2.3.17 (Casa Software, Teignmouth, UK). For each surface treatment, 10 specimens were produced under nominally identical conditions and the analysis presented was performed on at least three different specimens produced under nominally identical conditions for reproducibility; the average behavior is presented.

Results and Discussion
Surface characteristics.-Initially, the morphology and composition of the intermetallic particles (IMCs) in the 2024 T3 aluminum alloy were characterized by SEM/EDX. The distributions of the particles on the ultramicrotomed block specimen of the as-received alloy are shown in Fig. 1. Four types of intermetallic particle were present. The micrographs in Fig. 1 reveal round-shaped Al-Cu-Mg (S-phase) particles (labeled 1 and 2), Al-Cu (θ-phase) particles (labeled 3), Al-Mn dispersoids (labeled 4), and the irregularly-shaped Al-Cu-Fe-Mn-(Si) IMCs (labeled 5-10). The respective particles have typical sizes in the ranges of 1-5 μm, 0.5-1 μm and 2-8 μm. These particles have been characterized in many previous studies. [14][15][16][17][18][19] Table I shows the elemental atomic concentrations detected by EDX analysis. The Mg/Cu ratio helps to distinguish the θ and S-phases. The ratio for respective particles are in the region of 0.22-0.27 and 0.78-0.90. The magnesium detected at locations of the θ-phase particles originates from the underlying matrix. A cluster Al-Cu-Fe-Mn-(Si) particle is shown in Fig. 1b. In some of the Al-Cu-Fe-Mn-(Si) particles no silicon was detected (labeled 5 and 6) and the internal parts of the particles are enriched with Fe or Fe and Mn (labeled 7, 8 and 9, 10 respectively). The differing brightness of the internal and external parts of Al-Cu-Fe-Mn-(Si) particles in the backscattered electron image suggests a non-uniform distribution of elements within the particles. EDX analysis identified an enrichment of iron and manganese in the internal part of the particles ( Table I) that was possibly generated during the early stages of solidification. Small particles with different concentrations of Fe and Mn within larger Al-Cu-Fe-Mn-(Si) intermetallics were observed in earlier work. [20][21][22] The formation of relatively small size, individual Al-Cu-Fe-Mn-(Si)containing particles grouped in clusters can be explained by the low solubility of Fe and Mn in the aluminum matrix.
After etching in sodium hydroxide solution and desmutting for 30 or 60 s in nitric acid, the surface revealed a scalloped appearance with removed, partly dissolved and whole intermetallic particles present in cavities with trenching around them (Figs. 2a and 2b). Fig. 2b also reveals deposition of copper oxide particles. Intermetallic particles other than S-phase appeared to be unaffected by etching, according to EDX analysis. In contrast, after desmutting for 30 and 60 s, residues of S-phase revealed Al, Mg and Cu, with Mg decreasing to negligible levels after 60 s due to the progressive de-alloying. The alloy matrix around intermetallics is dissolved resulting in undermining and loss of smaller particles. However, larger intermetallic particles remain for short times of desmutting.
In contrast to desmutting in nitric acid alone, no S, θ-phases and Al-Cu-Fe-Mn-(Si) particles were observed on the surface after desmutting 30 and 60 s with addition of CeCl 3 (Figs. 2c and 2d). The surface revealed scalloping of similar shape and size as the intermetallic particles, which had been removed by dissolution and detachment. Only a small amount of copper (1.6-2 at.%) was left on the surface compared with specimens treated in nitric acid only (16-25 at.%). A longer time of desmutting resulted in dissolution of aluminum at the periphery of the Al-Cu-Fe-Mn-(Si) particles (Fig. 2d).
The micrographs of specimens desmutted in the solution with CeCl 3 and H 2 O 2 additions for 3, 6 and 9 min show features on the surface similar to those resulting from treatment with added CeCl 3 only (Figs. 3a-3c). After 3 min, the surface has a scalloped appearance with partly removed and dissolved Al-Cu-Mg, Al-Cu-Fe-Mn and Al-Mn particles and precipitated copper oxide particles (Fig. 3a). It is difficult to distinguish the size of the individual copper oxide particles due to their agglomeration. After 6 min, more S-phase particles were removed and no Al-Cu-Fe-Mn-(Si)-containing particles were detected (Fig. 3b). Some partly dissolved Al-Cu-Mg particles were covered by copper oxide particles. After 9 min, no intermetallic particles were observed (Fig. 3c). According to EDX analysis, the copper concentration decreased significantly with time, with concentrations of 18-32, 1.9-2.5 and 1.7-1.9 at.%, respectively. More detailed measurements of the Cu species are presented later using XPS analysis. The slower removal of intermetallic particles in the presence of H 2 O 2 is possibly  due to a local increase in the pH that slows the dissolution of oxide covering the particles.
Optical micrographs show that in the presence of CeCl 3 , grain boundary attack is significantly reduced. Figs. 4a and 4b compare specimens after desmutting in the absence and in the presence of CeCl 3 respectively; arrows indicate the grain boundaries that were attacked. Clearly, the number of the grain boundaries attacked and the extent of the attack was higher for the specimens desmutted in the absence of CeCl 3 .
Additional desmutting experiments were undertaken in order to distinguish the effects from Ce 3+ and Cl − ions using two solutions with different sources of Ce 3+ and Cl − . Fig. 5a shows that S-phase  (Fig. 5b). In the absence of cerium, de-alloyed S-phase remained on the surface following desmutting in nitric acid alone and also with addition of NaCl. In contrast, residual S-phase was not detected when CeCl 3 or Ce(NO 3 ) 3 were added, suggesting that the cerium accelerates the removal of the residues irrespective of the presence of Cl − ions. In contrast, Cl − ions appear to accelerate the removal of Al-Cu-Fe-Mn-(Si) particles, which were not detected following desmutting in solutions containing NaCl or CeCl 3 . The anodic dissolution is facilitated by the microgalvanic coupling between the particles and the matrix; it is assisted by Cl − ions that penetrate / attack the alumina film. [23][24][25] Al-Kharafi and Badawy found a significant decrease of the corrosion resistance of Cu-containing aluminum alloys with increase of the chloride ion concentration (>35 mM) in nitric acid solution. 26,27 The presence of copper in the alloys increases the attack by Cl − ions due to the formation of oxyhalide complexes that may cause the natural passivity to decrease.  (Table II). Carbon is present as a contaminant from the environment. As shown below, the oxide films on the alloy surface were 4 nm thick. Since the depth of the XPS analysis was about 5 to 10 nm, elements present in the substrate may be detected. No significant change in the element concentrations was observed with increasing time of treatment in either HNO 3 or HNO 3 + CeCl 3 solutions. A low concentration of chlorine was detected on the surface of the alloy treated in the CeCl 3 -containing solution. The concentrations of Cl − ions after desmutting for 30 and 60 s with addition of CeCl 3 were 0.5 and 1.5 at.% respectively, and 0.6, 0.9 and 1.5 at.% after desmutting for 3,6, and 9 min, respectively with additions of CeCl 3 and H 2 O 2 (Table II).
Due to the Al2p and Cu3p peaks overlapping, the real concentrations of elements can only be obtained by peak fitting. Figs. 7a and 7b display the results of peak fitting of the Al2p high resolution spectra. The following four components were found: i) two metallic Al peaks (Al2p 3/2 and Al2p 1/2 ) at 71.9 and 72.3 eV, respectively, ii) oxidized    metal (Al 3+ ) AlOx2p 3/2 and AlOx2p 1/2 at 74.3 and 74.7 eV, respectively, and iii) Cu3p 3/2 and Cu3p 1/2 at 75.2 and 77.5 eV, respectively (Fig. 7). The alloy surface was covered with aluminum oxide after removing the smut layer formed during etching in sodium hydroxide solution. The thickness of the oxide layer was determined using eq.1. After desmutting in nitric acid only for 30 and 60 s, the film thicknesses were 4.4 and 4.0 (±0.1) nm, respectively, while in the presence of CeCl 3 the respective values were 4.1 and 3.6 (±0.1) nm. The thickness of the aluminum oxide layer formed after desmutting in HNO 3 is in good agreement with the results presented in the literature. 28 The thicknesses of the aluminum oxide layer after desmutting in HNO 3 + CeCl 3 + H 2 O 2 solution for 3, 6, 9 min were 4.8, 3.9, 3.6 (±0.1) nm, respectively. The findings suggest a small reduction in thickness with increased desmutting time for all solutions.
Fitting of the Cu2p 3/2 peaks was undertaken to determine the oxidation state of copper present on the alloy surface after desmutting (Figs. 8a, 8b). The peaks revealed the presence of Cu 0 (metal) and Cu + (Cu 2 O) species with binding energies of 932.6 and 932.2 eV, respectively. Their presence was also identified using the Auger electron peaks (Cu LMM). 29,30 The amount of Cu 2 O relative to Cu 0 was greater after desmutting with added CeCl 3 and CeCl 3 + H 2 O 2 than with nitric acid alone (Table III).
Corrosion behavior.-Potentiodynamic polarization curves for the variously pre-treated specimens were obtained after OCP mea- surements for 15 min (Figs. 9a, 9b). The corrosion potentials (E corr ) are slightly higher for samples treated in HNO 3 + CeCl 3 and HNO 3 + CeCl 3 + H 2 O 2 (−0.56 and −0.57 V vs. SCE, respectively), than for nitric acid only (−0.6 V vs. SCE), probably due to a decreased anodic activity. Comparison of the polarization curves for the specimens reveals a reduced current density with the cerium-containing solution at small cathodic polarizations. In this region, oxygen reduction occurs under activation control mainly on large intermetallic particles. Thus, a reduced number of the particles results in a reduced  cathodic current. As the cathodic polarization increases, the diffusion limited current density is reached, which is similar for all specimens, with hydrogen evolution also occuring below −1.1 V (SCE). Since the removal of intermetallics is slower in the presence of hydrogen peroxide, the reduction of the cathodic current for small polarization is greater with increasing desmutting time.
Potentiodynamic anodizing.-The potential-current density responses during potentiodynamic anodizing of the alloy in 0.46 M H 2 SO 4 electrolyte at ambient temperature with a potential scan from 0 V vs. OCP to 14 V vs. SCE are presented in Figs. 10a, 10b. Two peaks occur after desmutting for 30 and 60 s in nitric acid only: the first, lying close to the OCP, is associated with oxidation of S-phase particles; 5 the second, with a maximum at 4.8 V vs. SCE, is associated with the oxidation of θ-phase (Al-Cu) and Al-Cu-Fe-Mn-(Si) particles (Fig. 10a). 5 With addition of CeCl 3 and CeCl 3 / H 2 O 2 to the desmutting solution, these two peaks were absent (Figs. 10a and 10b respectively). Intermetallic particles in the alloy oxidize at higher or lower potentials than the aluminum matrix during potentiodynamic anodizing in sulfuric acid electrolyte. 5,31,32 The absence of the peak at 0 and 4.8 V confirms the removal of intermetallic particles in the cerium-containing solutions.

Summary
This study is focused on removal of intermetallic particles from the surface of 2024 T3 aluminum alloy using CeCl 3 additions to a nitric acid desmutting solution. No S-, θor Al-Cu-Fe-Mn-(Si) particles were observed on the surface after desmutting for 30 or 60 s in nitric acid with added CeCl 3 . In contrast, only S-phase particles were significantly affected by the CeCl 3 -free solution. Addition of both CeCl 3 and H 2 O 2 to the nitric acid achieved a similar result to addition of CeCl 3 , but resulted in slower removal of the particles. Potentiodynamic anodizing in 0.46 M H 2 SO 4 solution confirmed the improved surface condition produced by the CeCl 3 -containing solutions, with the absence of peaks associated with oxidation of intermetallic particles.