Effect of Laminar Flow on the Corrosion Activity of AA6061-T6 in Seawater

The electrochemical behaviour and surface changes on AA6061-T6 alloy exposed to Caribbean seawater from the Cozumel Channel for 30 days under laminar flow (0.1 ms (cid:1) 1 ) were studied, these contrasting then with stationary conditions. Open circuit potential monitoring and electrochemical current fluctuations, considered as electrochemical noise (EN), were employed as two nondestructive methods. The calculated corrosion current, based on Rn, was one order higher in laminar flow. The fluctuations of current were transformed in the frequency domain. Their power spectral density (PSD) plots were obtained in order to gain information concerning the dynamic of the spontaneous release of energy during the corrosion process. The value of the exponent β in PSD graphs suggested that the localised corrosion on AA6061-T6 surface occurs as a persistent stationary process, which dynamic is controlled by oxygen diffusion. The changes in the morphology and elemental composition of the formed layers revealed that the localised attacks occurred in the vicinity of intermetallic particles rich in Fe and Cu, which act as cathodes.


Introduction
Aluminium alloy (AA) 6061-T6 is popular as a nonferrous material for structures in seawater [1] and is characterised by various properties, such as strength-toweight ratio, extrudability-particularly for the manufacture of profiles with complex geometry, low thermal expansion coefficient, good wear resistance and corrosion resistance [2]. The alloy presents good corrosion resistance in many environments having neutral pH, because of the formation of protective amorphous aluminium oxide film on its surface of approximately 2-3 nm thickness and is insoluble in water [3,4]. The addition of alloying elements to aluminium increases its mechanical properties [5]; however, the precipitated intermetallic particles (IMPs) have a harmful effect on the corrosion resistance of the Al alloys [4,[6][7][8][9][10]. IMPs present electrochemical behaviour different from the alloy matrix, and they may be classified in two types: cathodic and anodic [7,[9][10][11][12]. The electrochemically active anodic particles are rich in Mg, Si and Al, with Mg preferential dissolution, leaving a cavity in the oxide layer, while the cathodic particles are rich in Fe, Si and Cr, acting as preferential sites for oxygen reduction.
The localised corrosion becomes greater in the presence of aggressive ions, such as chlorides [4,[13][14][15], and corrosion pits initiate in oxide film sites, weakened by chloride attack. Moreover, the heterogeneity of the surface could result in favourable nucleation sites.
Seawater is a complex electrolyte of different ions, with a high salinity (3.5%, density 1.023 g/cm 3 at 25°C), which causes damage to metals in a short time [16,17]. The principal parameters that affect the corrosion behaviour of metals immersed in this electrolyte are oxygen content, dissolved mineral salts, pH, temperature, specific contaminants and flow velocity [16,18].
The characterisation of the corrosion process requires electrochemical nondestructive techniques, and the preferred methods are those that do not apply external polarisation. The monitoring of the open circuit potential (φ corr , free corrosion potential) or corrosion current is one of these and is easy to handle. Fluctuations may be interpreted as electrochemical noise (EN), which is useful for the purposes of corrosion mechanism characterisation. Electrochemical noise is presented by random fluctuations of corrosion potential or current, typically with frequencies below 10 Hz and low amplitude [19]. This technique can provide information concerning the nature of the corrosion process and the rate thereof. The main sources of EN observed in corrosion systems are attributed to microscopic and macroscopic events [19][20][21]. EN measurements can be performed under corrosion potential or any constant potential/current, depending on the research objective, to analyse the corrosion mechanism and obtain the corrosion rate [20].
EN measurements can be analysed transforming the data in the frequency domain by fast Fourier transform (FFT) to obtain power spectral density (PSD) [22,23]. PSD plots display a slope, β exponent, which enables the differentiation between series with fractional Gaussian noise (fGn, β from À1 to 1) and fractional Brownian motion (fBm, β from 1 to 3). The fGn is a stationary process, and the fBm is nonstationary [24,25]. The β exponent is a parameter correlated with the strength of persistence in a process [26]. In our previous studies, EN technique was carried out to characterise the first stages of corrosion in stationary seawater of copper [27], aluminium [28] and aluminium alloys [29], as well as the initial stages of AZ31B Mg alloy in simulated body fluid [30].
The object of this study is to investigate the electrochemical behaviour and surface changes on AA6061-T6 alloy, exposed to Caribbean seawater (Cozumel Channel) under laminar flow, contrasting these with stationary flow. Two nondestructive electrochemical methods were used to test the corrosion resistance of the alloy. The corrosion current and φ corr were considered as EN and transformed in the frequency domain, in order to gain information on the dynamics of the spontaneous release of energy during the corrosion process. X-ray photoelectron spectroscopy (XPS) measurement was employed to analyse the composition of the formed corrosion layers, as well as SEM-EDS surface analysis. To the best of own knowledge, there is still no study on the initial stages of localised corrosion of AA6061-T6 alloy in laminar flow conditions.
Samples of 1 cm 2 were embedded in Epofix resin as working electrodes for electrochemical tests. In addition, samples of 1 cm 2 were cleaned with ethanol and immersed in 100 mL of seawater (triplicated) under flow conditions, for 0, 5, 15 and 30 days, respectively. Before exposure, the specimens were ground with silicon carbide paper to 4000 grit, using distilled water as lubricant. They were washed with distillated water and dried in air.

Laminar flow of seawater
The speed range found in the Cozumel Channel indicates that the ocean current speed oscillates between 0.1 and 1 m s À1 . The tubing employed to connect the peristaltic pump (LAMBDA Laboratory Instrument) with the electrochemical cell has a diameter of 6 mm, and using the lower limit of the flow speed (0.1 m s À1 ), the calculated Reynolds number was 622.97, which corresponds to laminar hydrodynamic flow [31]. The dimensionless N Re characterises the diffusion-dependent system and is governed by the ratio of the inertial forces acting on the fluid to the viscous forces: where ρ is the fluid density (seawater, 997.8 kg m À3 ), u is the flow speed (0.1 m s À1 ), d is the tubing diameter (6 mm) and μ is the seawater viscosity (0.961 Â 10 À3 N s m À2 ). All experiments were carried out at 21°C.

Surface analysis
AA6061-T6 samples were characterised before and after exposure in seawater by SEM-EDS (Philips-XL and ESEM-JEOL JSM-7600F), in order to observe the microstructure and elemental changes on their surface. Corrosion products formed on AA6061-T6 sample surfaces were identified through X-ray photoelectron spectroscopy (XPS, K-alpha, Thermo Scientific, Waltham, MA, USA). In addition, corrosion products were removed in accordance with ASTM G1-90 standard [32], and the alloy surfaces were reexamined.

Electrochemical tests
A multiport electrochemical cell kit (Gamry Instruments, 1 L total volume) and a Gamry PCI4G750-52103 potentiostat were used for all electrochemical measurements. The experimental setup employed was according to ASTM G199-09 standard [33]: two identical AA6061-T6 working electrodes (WEs) connected to a zero resistance ammeter (ZRA) and a saturated calomel electrode (SCE, Hg 2+ /Hg 2 Cl 2 = 0.244 V) as reference electrode. The electrodes were immersed in an electrochemical cell with seawater, employed as test electrolyte. All tests were carried out in laminar and stationary flow conditions up to 30 days. EN data were collected at different times during 3 h, initial, 1, 5, 15 and 30 days in OCP, with a sampling frequency of 10 Hz. The obtained data of current and potential fluctuations were plotted vs. time. The current fluctuations were considered as EN and processed in the frequency domain by fast FFT to graph PSD. The PSD as a function of low frequencies was analysed on bi-logarithmic scale (10 À3 -1 Hz) of power per unit frequency (A 2 /Hz) vs. frequency (Hz). These procedures allowed the fitting of a straight line and obtention of the β slope. The β value characterises the corrosion mechanism on AA6061-T6 surface. The processing of data was realised with Electrochemical Signal Analyser V.7.0.1 software (Gamry Instruments, Philadelphia, PA, USA). All measurements were checked in triplicate.

Energy spectral density
The energy spectral density expresses how the energy of a time series is dispersed with a frequency; it can reflect the change in system dynamic. For a signal X(t) the energy EE is [34] For signals with a finite total energy, an equivalent expression for the energy is expressed as And using the Fourier transform of the signal: where f is the frequency in Hz. The integral on the right-hand side of Eq. (3) is the energy of the signal, and the integrandx f ð Þ j j 2 describes the energy per unit frequency contained in the signal (ESD). Figure 1 shows SEM images of the as-received specimen of 6061-T6 aluminium alloy surface. Small precipitates (labelled as A, B, C) may be seen (Figure 1b). In accordance with EDS, presented in Table 1, these precipitates correspond to particles rich in Fe and could be considered as elements of the following phases: Al 3 Fe, Al▬Si▬Mn▬Fe and α-Al(Fe, Mn, Cu) [9,10,35]. The elemental composition of the surface layer is indicated by zone D. Figure 2a presents the SEM image of the film formed on the surface of the sample after 5 days of immersion in seawater from the Cozumel Channel under laminar flow. It can be clearly seen that the passive layer begins to break down around the alloying elements (Figure 2b). Probably, this layer is less protective in the vicinity of the intermetallic particles, causing the formation of local electrochemical cells with the Al matrix. EDS analysis, Table 2, confirmed the presence of particles mentioned above. According to EDS ( Table 2), zone A is characterised by its higher content of Al, O and Cl, which could relate to corrosion products. According to previous studies, at pH > 8.5, Al(H 2 O) 6 3+ cations appear, while in the range of pH 4.5-8.5, Al(OH) 3 predominates [36,37]. In chloride solutions,  aluminium metal ionises rapidly to the Al 3+ ion, which also hydrolyses very rapidly (owing to the negative potential value) [38]. Both of these Al cations can react with chloride ions and form AlCl 3 soluble in water (31.77 wt%) [39]; this is converted later to a relatively stable species of basic aluminium chloride (AlCl 3 •H 2 O), transformed slowly to Al(OH) 3 and finally to Al 2 O 3 •H 2 O, an important corrosion product for the repassivation process of the aluminium surface [38].

Surface analysis
Based on EDS analysis, in zone B in addition to oxygen and aluminium, elements of calcium and carbon were present, both possibly as a part of a CaCO 3 precipitate, originating from seawater. Meanwhile, the layer in zone C maintained a composition similar to that of the alloy, which indicates that the corrosion process was still beginning on the surfaces of similar areas.
After removal of the layer of corrosion products (Figure 2b), area damaged by pitting and cracking was observed on the alloy surface. However, some precipitates remained on the surface of AA6061-T6, which according to EDS ( Table 2) correspond to cathodic particles rich in Fe, α-Al(Fe,Mn,Cu)Si, which promoted the preferential dissolution of the aluminium matrix (local alkalisation) [10].  (Figure 3a and c) and stationary flow (Figure 3b and d).
In these micrographs, the products formed on the surfaces of the alloy (Figure 3a and b) can be seen and compared, whose EDS analysis is summarised in Table 3. In laminar flow (Figure 3a), the segregation of particles rich in Cu (particles A) with the presence of O and Al could be considered as the phases of Al 2 Cu, AlMgSiCu (Q-phase) and Al 7 Cu 2 Fe; both relatively resistant to corrosion, because they are nobler than the aluminium matrix [4,12]. Q-phase acts as a cathode and does not contribute to intergranular corrosion when it is not connected to any other Cu particle. Otherwise, the Q-phase as cathode promotes the development of intergranular corrosion, initiated in the presence of chloride ions (NaCl). In the corrosion layer, particles without Cu appeared (particles B), rich in Fe, Al and Mg, which could be attributed to the following phases, Al▬Si▬Mn▬Fe and Al▬Mg▬Si, reported for aluminium alloy series 6xxx [35]. Zone C of the layer formed under stationary flow (Figure 3b) presented a similar composition of the alloy (Table 3), however, with the oxide layer of Al 2 O 3 •H 2 O on the alloy surface 6061-T6, as a transformation product of basic aluminium (AlCl 3 •H 2 O) in the presence of NaCl [38]. This layer is part of the entire surface (Figure 3a), since the three zones (A, B and C) have a high oxygen content ( Table 3) [4]. Figure 3c corresponds to the SEM image of the aluminium alloy 6061-T6 exposed in seawater at 30 days with flow, after removal of the layer of corrosion products. An area on the surface damaged by cracking and fissures can be observed, while the surface exposed to seawater without flow (Figure 3d) shows less damage, with pitting holes of several diameters. In contrast, on the surface exposed to laminar flow for this same time period, the pits are not clearly observable, but localised corrosion damage appeared in the form of cracks and fissures. This effect is due to the seawater flow, which accelerates the corrosion process, causing detachment of the destroyed passive layer and the appearance of new "fresh" areas, where the oxygen diffuses and is the oxidising agent in the cathodic corrosion reaction. It is also observed that on both surfaces (Figure 3c and d), some particles remained (named as D and F), and according to EDS (Table 3), they correspond to cathodic rich in Fe [7], reported in the reference sample as α-Al(Fe,Mn,Cu)Si [35]; these intermetallic particles promoted the preferential dissolution of the aluminium matrix [40,41]. Figure 4 compares the φ corr fluctuations of AA6061-T6 specimens exposed for different periods of times (0, 5 and 30 days) under laminar flow conditions (Figure 4a) and without flow (Figure 4b). The average values are summarised in Table 4. The trends in the changes, towards more or less negative values, are a response to the transformations that have occurred on the surface of the aluminium alloy with the advance of the corrosion process. These are in the morphology, elemental composition of the layers formed, as well as the type of localised corrosion attack discussed previously. It can be seen ( Table 4) that the initial values of φ corr are relatively similar, being 30 mV nobler in respect of the exposed surface in laminar flow. At 5 days, which implies the initial destruction of the passive layer (Figure 2a), the φ corr is 100 mV more negative with laminar flow than under  Table 3.  (Table 4). However, when the product layer was removed (Figure 3c and d), the SEM images revealed a greater localised attack on the aluminium surface under laminar flow (Figure 3c), while in the absence of flow, this attack has been less aggressive, presenting shallow pits. Free corrosion potential (φ corr ) tendency towards more or less negative values indicates periods of corrosion activation or repassivation of the surface, both facts related to the characteristics of the layers formed [4,42].

Surface characterisation by XPS
In order to identify the composition of the corrosion product layer created on the alloy surface after 30 days of exposure in laminar flow, XPS analysis was carried out on the specimen immersed, taking into account that aluminium corrosion products were not provided by XRD analysis as crystalline phases and are possibly amorphous. Figure 5 shows the full XPS spectrum of the corrosion products formed on the aluminium alloy surface of 6061-T6. The XPS spectrum revealed signals of Mg, Na, O, Cl, C, Si and Al, which accord with EDS analysis. The high-resolution peak for Al2p, situated at 74. 38 Figure 6 shows the current oscillations, and it can be seen that at the beginning of the experiment (0 days), the current density of AA6061-T6 surface immersed in   Table 4.

Average of free corrosion potential (φ corr ) values for AA6061-T6 in Caribbean seawater under laminar and stationary flows at different times.
laminar flow of seawater is higher in ≈50 μA cm À2 than that current in stationary flow. This suggested that the corrosion of the aluminium alloy 6061-T6 surface in laminar flow initiates faster, when the oxide layer on the alloy begins to break down. However, at the end of the experiment (30 days), the current value diminished suddenly, compared with the initial values. However, in stationary flow the current shifted to one order higher values than those in laminar flow, suggesting an acceleration of the corrosion process at that period of time [44]. Conversely, the current oscillations in stationary conditions (Figure 6b) presented slow variations, while for flow conditions (Figure 6a), intense current fluctuations were acquired with greater amplitude, which suggest greater corrosion [45]. The observed current oscillations registered in (μA cm À2 ) (Figure 6) correspond to the variation of the free corrosion potential (φ corr ) values in several mV (Figure 4). The corrosion current density was calculated from the value of polarisation resistance R p (on the assumption that R p is equivalent to polarisation resistance  noise R n ) obtained by ECN tests, according to the Stern-Geary equation (Eq. (5)). R n is calculated by dividing the standard deviation of potential by standard deviation of current (the potential noise can be modelled as the action of the current noise on the metal-solution impedance): where R p is the polarisation resistance and b a and b c are the Tafel coefficients. In this research, the B value employed was 0.26 V, taking common values for b a and b c of aluminium alloys [46,47]. The calculated values ( Table 5) show that the corrosion current of AA6061-T6 increases with the time of exposure, being more than one order higher when the alloy is exposed under laminar flow of seawater, while in stationary conditions, it maintains almost similar values up to 30 days.
Subsequently, Faraday's law was applied to calculate the corrosion rate (CR, Table 6) in the following form: where E w stands for the equivalent mass of AA6061-T6, I corr is the corrosion density (A cm À2 ), ρ is the metal density, K is a constant (3272 mm/A cm year) and A is the exposed specimen area (1 cm 2 ) [48].
The corrosion rate values presented in Table 6 indicate that under laminar flow, the values varied 0.08 and 0.20 mm per year, while in stationary conditions, they were between 0.012 and 0.162 mm per year.
With the statistical data obtained from the corrosion current, the pitting index (PI) [19] was calculated in order to reveal AA606-T6 susceptibility to localised corrosion for the laminar and stationary: where σ i is the standard deviation and i rms the main square root of current noise. Values of PI above 0.1 may indicate localised corrosion [20,49]. The pitting indexes are shown in Table 7. Thus, the calculated PI value suggests that at the end of the experiment (30 days), for both flow cases, AA606-T6 showed pitting corrosion, approximately four times higher in flow conditions, reaching PI = 0.96. These facts agree with the SEM images (Figure 3) comparing the corrosion attacks on AA6061-T6 exposed to both flow conditions.
The current fluctuations, considered as EN, were transformed into the frequency domain to estimate PSD slopes (β exponent). Figure 7 compares the PSD plots in bi-logarithmic scale, corresponding to AA6061-T6 surfaces after 30 days of exposure in seawater under laminar and stationary flows. In each case, the β exponent decreases as the frequency increases, and this fact could be associated with the advance of the localised corrosion attacks on the alloy surface [34]. At 30 days, β values are similar in laminar and stationary flows (1.0 and 0.94, respectively) and may be attributed to the fractional Gaussian noise (fGn), associated with a persistent process [26]. This type of noise (fGn) is considered also as a stationary process [24]. Figure 8 shows the spontaneous energy E during the corrosion process. At the beginning, after 5 days the energy was of an order of magnitude higher in laminar flow (1.1 Â 10 À4 ). This fact is consistent with the SEM image presented in Figure 2, which showed the accelerated corrosion process causing severe damage to the alloy surface exposed to laminar flow.
However, at the end of the experiment (30 days), the energy diminished in magnitude, being very similar in the order of 10 À7 , for both flows, probably on account of the formation of layers of corrosion products with different  2. The calculated values showed that the corrosion current (icorr) of AA6061-T6 increases with the time of exposure more than one order higher when the alloy is exposed under laminar flow of seawater (7.12-18.08 μA cm À2 ), while in stationary conditions, it maintains almost similar values (17.51-19.32 μA cm À2 ).
With the statistical data obtained from the corrosion current, the calculated pitting index (PI) revealed that AA606-T6 is four times more susceptible to localised corrosion in seawater under laminar flow (PI = 0.96), compared to that in stationary conditions (PI = 0.22). The estimated PSD slopes (β exponent) of the current fluctuations transformed into the frequency domain revealed that in laminar and stationary flows (β = 1.0 and 0.94, respectively), the electrochemical corrosion may be attributed to the fractional Gaussian noise (fGn), associated with a persistent stationary process.
3. The spontaneous energy release in the initial stages is one order higher in laminar flow (ΔE = 1.1 Â 10 À4 ). However, at the end of the experiment (30 days), the energy diminished in magnitude, being very similar in an order of 10 À7 , for both flows, probably on account of the formation of layers of corrosion products with different characteristic, which act as a physical barrier on the alloy surfaces, slowing down the corrosion attack. 4. The observed effect of the laminar seawater flow on the AA6061-T6 corrosion process should be considered as a consequence of the facilitated diffusion of the oxygen at the metal-seawater interface, resulting in specific transformation in the composition of the formed corrosion layers. The presented surface SEM-EDS and XPS analysis agree positively with the results obtained with both nondestructive electrochemical methods.