2.1. Sample preparation and corrosion test

AA2037 T8 aluminum alloy was manufactured using Hazzellet continuous casting method by Aleris International (Uhrichsville, OH, USA). The aluminum alloy contained Cu (1.48%), Mg (0.45%), Mn% (0.24), Si (0.06), Zn (0.02) and Ti (0.02). The hot bands were cut into pieces of 20 mm × 20 mm × 2 mm. Two types of the samples were prepared. Sample A: The as-received hot band was first cold-rolled to 70% deformation and homogenized at 510°C for 2 h in air. Then it was stretched with 3% deformation and finally aged at 150°C for 7 days. Some cold-rolled samples,, were homogenized at 510°C for 30 s and quenched in ice water, ensuring that premature cooling was minimized. Samples A and B were treated under almost the same condition. There were nano-sized dispersoids in both samples. Sample A by long homogenization time contained a large number of dispersoids, while Sample B by short annealing time only contained very few dispersoids. TEM analysis revealed that the dispersoid composition was Al₂₀Mn₃Cu₂, of about 5 nm in size. Coincidence Doppler broadening (CDB) experiment confirmed that there existed Mn and Cu in T phase ^[24].

Surfaces of both samples were ground and polished, rinsed with alcohol in ultrasonic bath and dried under cold flowing air. 1 M NaOH aqueous solution was prepared from reagent-grade chemicals and deionized water. The samples were immersed in the solution for different durations (0–2 h) at room temperature. After removal from NaOH solution, the corroded samples were thoroughly washed with distilled water and dried with cold flowing air.

2.2. Sample examination

Slow positron beam Doppler broadening measurements were conducted in a vacuum system at 10⁻⁶ Pa at room temperature in Wuhan University ^[26]. A slow positron flux was about 5 million per second for 40 mCi ²²Na radioactive source, a moderator decay rate of 4% per day, and a positron energy spread as narrow as 2.0 eV. Moderation efficiency using solid neon as moderator was about 1%, and monoenergetic beam was in spot size of about Φ5 mm. The positron mean implantation depth Z_m (in nm) can be calculated by

where, ρ_Al = 2.7 g/cm³ is density of aluminum alloy and E is the positron energy (in keV).

A high-purity Ge detector was used to measure the energy spectra of gamma photons. The positron implantation energy ranged from 0 to 30 keV. Each annihilation line spectrum consisted of 5×10⁵ annihilation events, with a counting rate of 1000 cps at 511 keV. In accordance with the previous work ^[27], two standard line-shape parameters S and W were calculated to characterize the sample with a relative accuracy of 0.001. The VEPFIT software was used to fit the result of the S parameter as function of energy E.

SEM (SIRION 200, FEI, the Netherlands) was used to examine surface topography of the corroded samples. AFM examination (SHIMADZU, SPM-9500J30) of samples was carried out in air in the direct contact mode.

The open-circuit potentials (OCP) and potentiodynamic polarization curve measurements were performed using a CHI660C system and a three-electrode electrochemical cell. A Pt flake, a saturated calomel electrode (SCE) and the sample were used as the counter electrode, the reference electrode and the working electrode, respectively. The apparent area of the sample that was exposed to 1 M NaOH solution was 1 cm². The potentiodynamic polarization curve was measured at the scan rate of 5 mV/s from −3.1 V to −0.1 V potential (referred to as open circuit potential) after 5 min initial delay. The corrosion current density and corrosion rate were determined by Tafel extrapolation method. All dissolution experiments were performed at room temperature. The corrosion results were averaged from at least measurements of each sample.

3.1. Slow positron beam measurement

The S parameter as a function of positron incident energy E, i.e. S(E) profiles, for Samples A and B at different dissolution durations in 1 M NaOH solution is shown in Fig. 1. According to Eq.(1), the mean implantation depth of positrons from the surface is given on the top axis. The S parameters are normalized to the bulk value (S = 0.515), which is approached for E >20 keV. All samples have low S values of 0.92–0.94 close to surface for E < 2 keV. For 2 keV<E<8 keV with larger depths, there was a maximum of S value at which the positrons annihilate in low-momentum valence electrons in interfacial defect layer. For E >20 keV, S approaches a constant value since all positrons are essentially implanted into the matrix before being annihilated. A significant decrease in S parameter is found after dissolution for both samples. The broad peak in S(E) indicates that there exists a near-surface defect layer in both samples after 10 min in 1 M NaOH solution. As dissolution time increases, the peaks weaken gradually and then disappear. The results agree with those of the corrosion of water-quenched Al alloy ^[25], but both samples changed little after 10 min dissolution as they further deformed to 3% and aged at 150°C for 7 days, resulting in production of many defects and precipitates induced by deformation and/or peak-ageing.

Fig. 1.Full-screen View

S-parameter as a function of position incident energy and mean depth, for the samples corroded in 1 M NaOH solution for different minutes. The solid lines are fits to the data.

Comparing the S(E) profiles in Figs. 1(a) and 1(b), it can be seen that the virgin Sample A has larger S in interfacial defect layer than the virgin Sample B, and after corrosion, the S(E) profiles of both samples decrease, but the decrease rate of Sample A is slower than that of the Sample B.

To learn about defects types for both samples during early stages of corrosion, W parameters as a function of S parameter are shown in Fig. 2. All the coordinate points (S, W) fall on a common locus and distribute near by a straight line, indicating that only two kinds of positron states contribute to the annihilation ^[28]. Thus, only one type of defect appears in all the measurements. For the original samples, positrons mainly annihilate in interface layer and bulk; while for the corroded samples positrons mainly annihilate in corrosion product in layer and bulk.

Fig. 2.Full-screen View

W parameter as a function of S parameter of the samples corroded in 1 M NaOH solution for different minutes..

To obtain the depth distribution of defects, the VEPFIT program was used to fit the S (E) profiles. The results are plotted as solid lines in Fig. 1. The dissolution rate in NaOH solution was very rapid (which will be discussed in the Section 3.3). The initial defect layer was dissolved in a few minutes. After the treatment, the defect layer was replaced by new surface corrosion layer introduced by the dissolution. Therefore, the simple two-layer model (near-surface corrosion layer and bulk), i.e., assuming no interface layer, was employed for the fitting. The positron diffusion length of bulk Al alloys was fixed at about 200 nm as reported for metals and alloys ^[19,25]. Table 1 shows the fitted results of the interfacial defect layer (or near-surface corrosion layer) S (S_d), positron diffusion length (L_d), and thickness (B_d). The S_d parameter of defect layer decreases with increasing dissolution time in both samples, and the decrease rate of Sample A (from 1.025 to 0.996) is slower than that of Sample B (from 1.022 to 0.985). L_d of both samples increases after dissolution, but does not change too much with dissolution time, indicating a lower defect/void concentration or the formation of near-surface corrosion layer. The layer thickness B_d is 300–350 nm, which is much longer than that of virgin samples (about 50 nm).

Many experiments ^[3-5,25,29] indicated that Cu affects strongly on corrosion of aluminum alloys. The significant decrease in S parameter observed after the corrosion of AA2037 alloy can be explained by invoking two possible causes: copper enrichment within the metal-oxide interface layer and formation of new corrosion layer. The annealed Cu has a low S value of 0.83 due to presence of 3d-electrons. The content of Cu in AA2037 aluminum alloy is only 1.48%. Therefore, Cu enrichment near the metal-oxide interface layer shall induce a decrease of the S parameter. The Cu enrichment increases with dissolution time, and thus the S parameter decreases. In addition, the positron diffusion length increases after NaOH dissolution, indicating the formation of near-surface corrosion layer (i.e., the formation of new oxides or hydroxides). As a result of the pre-existence of many dispersoids, the rate of Cu enrichment or corrosion layer formation was slower in Sample A than in Sample B, and so the S parameter decreased less with increasing dissolution time, which implies that the pre-existence of a number of dispersoids introduced during homogenization shall hinder the corrosion processes. Additional measurements confirmed that a thick oxide film (a few μm) was formed at the surface of AA2037 aluminum alloy samples after immersion in pure water for several months, resulting in a lower near-surface S parameter. Further investigations are in progress and will be reported in the future.

3.2. Morphology characterization

Fig. 3 shows typical SEM surface morphology of Samples A and B after different minutes of exposure to 1 M NaOH solution. Figs. 3(a) and 3(f) are microstructures of the virgin samples, for comparison with the corroded surfaces. When the samples were immersed in NaOH solutions, high effervescence of the solution surrounding the samples was observed due to the formation of gaseous H₂ ^[30,31]. Figs. 3(b) and 3(g) show that the original thin oxide film was dissolved after 1 min immersion by the high OH⁻ concentration in alkaline solution. Both samples mainly consisted of α-Al matrix with tiny precipitates introduced by peak-ageing at 150°C. The precipitates were confirmed to be Cu-containing intermetallics on the alloy surface by corresponding EDS spectrum (not shown). The surface of samples is uniformly thinned by alkaline chemical dissolution, and local corrosion of some grain boundaries or precipitates is facilitated by attack of OH^- from dissolution time of 1 min to 10 min ^[32]. Compared to the surface morphology of Sample A (Figs. 3b–3e), Sample B (Figs. 3g–3j) at the same dissolution time (e.g. 5 min) was corroded more seriously, with more flawed regions at a dissolution time of 10 min. This may be because that, introduced during homogenization, Sample A had more dispersoids, which hindered the corrosion processes.

Fig. 3.Full-screen View

SEM images of Sample A (a–e) and Sample B (f–j) tested in 1 M NaOH solution for different dissolution durations.

Fig. 4 shows AFM images of the corroded surface of the samples. Some cavities on the surface of both samples after immersion for 1 and 2 min are found to be associated with inclusions or large corrosion pits. As shown in Figure 3, Sample B was corroded more seriously and produced more and larger cavities or microvoids than Sample A. The results support that Sample A has a better resistance to corrosion than Sample B due to the presence of more dispersoids, which is in good agreement with Doppler broadening measurements.

Fig. 4.Full-screen View

AFM micrographs of the surface topographies of Samples A after dissolution in 1 M NaOH solution for 1 min (a) and 2 min (b), and of Sample B after dissolution in 1 M NaOH solution for 1 min (c) and 2 min (d).

Doppler broadening measurement (Figs. 1) has a resolution around 10¹–10² nanometer in depth, and can be used to evaluate the thickness of surface layer or interfacial defect layer without cross-sectioning the sample. This resolution cannot be easily achieved by SEM and AFM (Figs. 3 and 4). In characterizing surface layers and near-surface region, slow positron beam technique with such a resolution plays an irreplaceable role. SEM reveals just sample morphology and composition distribution, while the positron beam technique is sensitive to vacancy- type defects and defect’s chemical environment that are closely associated with corrosion reactions. In this sense, the positron beam is an essential complement to SEM.

3.3. Analysis of open-circuit potential and polarization curve

Fig. 5 shows the open-circuit potential as a function of dissolution time for Samples A and B in 1 M NaOH solution. In these transients, at first the potential decreased dramatically to a minimum in less than 20 s. The initial potential decline is attributed to the dissolution of oxide film on the surface, as a result of its high solubility in the alkaline solution. The potential increased rapidly up to 2 min, then the increase became slowly. These are in good agreement with open-circuit corrosion of Al in alkaline solutions ^[33-35]. Based on the mechanism of corrosion of Al in alkaline solutions suggested by Adhikari et al. ^[35,36], the value of the minimum potential of electropolished samples indicated that it was determined by the Nernst potential for oxidation of surface AlH₃. The potential increases quickly first, and then slowly. The former is attributable to the formation of surface hydride, while the latter is due to enrichment of Cu impurities on the surface, which anodically polarizes Al by accelerating the cathodic process. Therefore, we suggest that the corrosion mechanism of Al alloy in alkaline media may be related to the build-up of a protective layer of Al(OH)₃. The transients of Sample A in Fig. 5 are offset to more negative potential than that of Sample B. This may be attributable to surface composition changes or microstructure in sample surface layer due to the presence of more dispersoids, which were removed in the first 2 min of dissolution. The dealloying of Cu-containing intermetallics on the alloy surface may occur in long time immersion. During dealloying, S-phase (Al₂MgCu) ^[37,38] intermetallic undergoes selective dissolution (an increase of the elemental Cu on the alloy surface) in NaOH solution, leading to a slow increase of open-circuit potential value.

Fig. 5.Full-screen View

Comparison of open-circuit potential transients for Samples A and B in 1 M NaOH solution.

Fig. 6 shows the typical potentiodynamic polarization curves of the samples corroded in 1 M NaOH solution. The polarization curves were used to measure the corrosion current density, i_corr at E_corr, by Tafel extrapolation of the polarization curve. It can be seen that both anodic and cathodic polarization curves of Sample A are lower than those of Sample B, indicating that the former is electrochemically less active than the latter. The corrosion rate was calculated using the following equation ^[39]:

Fig. 6Full-screen View

Polarization curves of Sample A and Sample B in 1 M NaOH solution.

where M is atomic weight, n is valence, F is Faraday constant in C·mol⁻¹, i_corr is current density in μA·cm⁻², ν is corrosion rate in g·m⁻²·h⁻¹, ρ is the density in units of g·cm⁻³, and d is corrosion rate in mm·a⁻¹.

The electrochemical parameters of the samples are listed in Table 2. The E_corr of both samples are about the same (1.53 and 1.52 V), indicating that they have similar susceptibility to the corrosion initiation. However, the i_corr (11.0 mA·cm⁻²) and corrosion rate (119 mm·a⁻¹) of Sample A are lower than those of Sample B (16.2 mA·cm⁻² and 176 mm·a⁻¹), revealing that Sample A has a better corrosion resistance than Sample B. The results are similar to the data reported by Refs.[32,40], and are consistent with the results of microstructure analysis and Doppler broadening measurements. In this study, the evaluation of corrosion behavior using the Tafel extrapolation method or the electrochemical technique allows one to determine and understand the difference in tendency to corrode between Samples A and B under the same testing condition.

Corrosion behavior of AA 2037 T8 Al alloys will be further studied with immersion tests in NaCl or acidic aqueous solutions in follow-up studies.