Cathodic Dealloying of α-Phase Al-Zn in Slightly Alkaline Chloride Electrolyte and Its Consequence for Corrosion Resistance

Cathodic dealloying of Al-Zn phases may be a significant, albeit poorly understood, phenomenon occurring during field corrosion of Zn-Al and Zn-Al-Mg coatings. The reactivity of α-phase Al-Zn (Zn-68 wt% Al, Al5.2 Zn) has been investigated as a function of potential in pH = 10.1, 30 mM NaCl. The dealloying of Al during cathodic polarization (cathodic dealloying) led to the formation of a Zn(0) enriched surface layer. The dealloyed layer did not affect the rate of cathodic Al dissolution or the open circuit corrosion rate. The dissolution of Zn from the Zn(0) enriched layer showed an onset potential below the critical potential (Ec) where Al and Zn simultaneously dissolve. The cathodic dissolution of Zn was observed in O2 saturated electrolyte which correlated with an order of magnitude decrease in the Al dissolution rate. © The Author(s) 2018. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any medium, provided the original work is properly cited. [DOI: 10.1149/2.0581807jes]

Alloys of Zn, Al and Mg are frequently used as coatings for galvanized steel due to their enhanced corrosion resistance in a wide variety of conditions. This enhanced resistance has been confirmed in atmospheric field exposures, accelerated atmospheric corrosion tests, and in immersion in chloride containing electrolytes.  The role of microstructure on the corrosion resistance was highlighted by Prosek et al. 11 They produced Zn-Al and Zn-Al-Mg alloys with identical chemical composition but differing microstructure by changing the cooling rate when they were casted. They concluded that the finer microstructure of Zn-5 wt% Al alloy showed better corrosion resistance under atmospheric exposure conditions. The origin of this effect and the interactions between the different phases however has yet to be revealed. Indeed, the commercial Zn-Al and Zn-Al-Mg alloy coatings present a complex multiphase structure. For example, Zn-5 wt% Al alloy (Galfan, denoted as Zn-5Al) has a eutectic structure consisting of a dendritic η-phase of Zn interspersed within the lamella of Zn rich Al phase (β-phase of Al). [24][25][26][27] The Zn-55 wt% Al-1.6 wt% Si, (Galvalume, denoted as Zn-55Al) is composed of a dendritic α-phase of Al, and a Zn rich interdendritic phase. [28][29][30][31][32] The Zn-Al-Mg alloys have a dendritic η-phase of Zn, a Zn-MgZn 2 binary eutectic, and a Zn-Al-MgZn 2 ternary eutectic phase. [1][2][3][4][5][6][7][8]23 In order to understand and predict the corrosion of complex Zn-Al and Zn-Al-Mg alloys, it is important to characterize the electrochemical stability of the individual phases. The MgZn 2 intermetallic phase and the α-phase of Al represent the extremes of the phases present in the coating alloys in terms of Mg content and Al content respectively. In previous work, the dealloying mechanism of MgZn 2 intermetallic phase was investigated in slightly alkaline chloride containing electrolyte. 33 It was concluded that a Mg-depleted Zn metallic/oxide layer inhibits the dissolution of Mg and Zn resulting in improved corrosion resistance. This work extends the previous work to a consideration of a pure Al-Zn α-phase system. The α-phase of Al (0∼32 wt% Zn) is representative of the high Al content phases of the Zn-Al and Zn-Al-Mg system. The α-phase of Al is present in the ternary eutectic phase of Zn-Al-Mg where it may be in intimate contact with the η-phase of Zn and the MgZn 2 phase. 34,35 The ternary eutectic has been shown to be the most reactive phase, preferentially corroding leaving the Zn rich phase unattacked. 8,23,35 Other phases such as the β-phase of Al may dissociate to form the η-phase of Zn and the α-phase of Al at room temperature. 36,37 A particularity of Al and Al alloys is their tendency to undergo dissolution in the presence of cathodic currents, commonly referred to as cathodic dissolution. [38][39][40][41][42][43][44][45] This is usually thought of as a two-step process, involving the dissolution of the oxide film by reaction with hydroxide generated by the cathodic reaction and the * Electrochemical Society Student Member.
* * Electrochemical Society Member. z E-mail: kevin.ogle@chimie-paristech.fr subsequent oxidation of the exposed Al metal. The net reaction, with either water or O 2 as the oxidizing agent, may be written as: Al + 4H 2 O + e − → Al(OH) 4 − + 2H 2 [1] Al Reactions 1 and 2 predict a 1:1 stoichiometry of electrons to dissolved Al ions (v e /v Al ) during cathodic polarization. In fact, a value of v e /v Al = 1.83 ± 0.35 was obtained where the excess electrons were attributed to the diffusion of hydroxide away from the interface in the electrochemical flow cell. 42 The stoichiometry was obtained for Al metal, Al 2 Cu pure phase and various Al alloys (2000 and 6000 series). [40][41][42] Cathodic Al dissolution has also been observed during the polarization of Zn-5Al and Zn-(3∼4) wt% Al-(3∼4) wt% Mg. 7 During corrosion, the α-phase of Al may be polarized cathodically when coupled to a less noble phase such as MgZn 2 . Therefore, it is reasonable to think that cathodic Al dissolution (cathodic dealloying) may be important under field conditions. This could play an important role in corrosion resistance: the Zn-Al-Mg and Zn-Al alloy coatings are known to form Zn-Al layered double hydroxides (LDH) which have been associated with improved corrosion resistance. 1,15 Cathodic Al dissolution may serve as a reservoir of Al 3+ release into solution even when the Al rich alloy is the cathode in a galvanic couple.
In this work, the anodic and cathodic polarization of the α-phase of Al is investigated with an emphasis on cathodic dealloying. To this end, atomic emission spectroelectrochemistry (AESEC) is used to measure the dissolution rates of Al and Zn simultaneously with the electrical current. This permits a direct measurement of cathodic Al dissolution which cannot be obtained by consideration of the conventional polarization curve since the anodic dissolution of Al is masked by the intense cathodic current. The effect on corrosion of the dealloyed metallic Zn enriched (Zn(0)) layer formed during cathodic dealloying is investigated. Other phenomena brought to light in this work are the cathodic dissolution of Zn and the passivation of the Al film in the presence of oxygen. The pH = 10.1 was chosen because, during corrosion, the interfacial pH of the Zn-Al-Mg alloy becomes alkaline due to the cathodic reaction. 17,23,46 Experimental Materials.-Al metal (99.99%) from Goodfellow and a α-phase of Al (Zn-68 wt% Al) were used in this work. The latter was produced and characterized by the Department of Metals and Corrosion Engineering, University of Chemistry and Technology, Prague. The chemical composition of α-phase of Al was chosen as 68 wt% Al because it showed a single α-phase with only a trace of η-phase of Zn. In this work this phase will be denoted as Al 5.2 Zn based on its molar composition. Al 5.2 Zn was produced from pure metals, heated in a C335 ceramic crucible in a muffle furnace, filtered through a frit and cast into a 50 mm diameter metal mold. Al 5.2 Zn was treated by water quenching after 24 hours in a 400 • C muffle furnace. The chemical composition of the Al 5.2 Zn was 67.6 wt% Al and 32.4 wt% Zn obtained by atomic absorption spectroscopy (AAS). The sample and electrolytes were prepared as in our previous publication. 33 30 mM NaCl, pH = 10.1 electrolyte was prepared from analytical grade reagent and deionized water (18 M cm) obtained via a Milipore system either deaerated by Ar or saturated by O 2 . All the experiments presented herein showed reproducible results from the measurements repeated at least three times.
The AESEC technique.-AESEC was used for analyzing the elemental dissolution kinetics of the specimens. The AESEC technique has been described in detail in previous publications. 33,[47][48][49] The specimen of interest was brought into contact with a flowing electrolyte in a small volume (≈ 0.2 cm 3 ) three-electrode electrochemical flow cell. A saturated calomel electrode, SCE and a Pt foil were used as reference and counter electrodes, respectively. In this work, the time resolved concentration of Zn was determined from the emission intensity of the plasma monitored at 213.86 nm wavelength on the polychromator and Al at 167.08 nm on the monochromator using standard inductively coupled plasma atomic emission spectrometry (ICP-AES) calibration techniques.
A Gamry Reference 600 potentiostat was used to control and measure the electrochemical potential (E) and current density (j e ). A number of different electrochemical experiments were performed including measurements of the spontaneous dissolution at the open circuit potential (dissolution profile), linear sweep voltammetry (LSV) using a scan rate of 0.5 mV s −1 , and constant potential (E ap ) chronoamperometry (CA). Specific experimental descriptions will be given in the text.
Surface characterization.-The material surface before and after electrochemical experiments was characterized by scanning electron microscopy (SEM) using a Zeiss Leo 1530 microscope with field emission gun (FEG) source at 15 keV, 15 mm working distance. Raman spectroscopy was performed to identify oxide species on the sample surface using a Reinshaw Invia confocal Raman microscope with excitation by a Co diode pumped solid state (DPSS) using a green laser (532 nm). The acquisition and evaluation of Raman spectra were conducted by using Wire software.
Data analysis.-The data analysis was described in the previous publication concerning MgZn 2 dealloying 33 and elsewhere. [47][48][49] The basic principle of AESEC is to simultaneously measure the rates of elemental dissolution, v M (M = Al and Zn in this work), the electrochemical current density, j e and the potential, E. The dissolution rates may be expressed as equivalent elemental current densities, j M (j M = zFv M where F is the Faraday constant and z is the oxidation state of each element) to facilitate comparison with j e . For the determination of the cathodic dealloying stoichiometry, it is convenient to present the cathodic current as a rate of electron transfer, v e = −j e /nF, for j e < 0.
During dealloying or selective leaching, the least noble element (Al) will dissolve leaving behind a surface layer enriched in the more noble element (Zn). The elemental dissolution rates may be used to construct a mass balance so that the surface excess of an element, Q M(0) , may be determined. This permits a real time, indirect analysis of the growth of a surface enriched (dealloying) layer during the selective dissolution of one or more elements.
A mass/charge balance may be constructed by comparing j e with the sum of the dissolution rates, j (= j Al + j Zn ). The faradaic yield of dissolution, η, may be determined as the ratio of η = j /j e * . For η = 1, it is often possible to calculate the rate of either the cathodic reaction, j c (η > 1), or the formation of undissolved corrosion products, j ox for (η < 1). Note that the comparison of transients in j M and j e requires the convolution of the latter with the residence time distribution in the flow cell. The convoluted form of j e is designated as j e * .  dissolution rate, j Al . The transpassive breakdown was somewhat more intense in the presence of O 2 . It is obvious from Fig. 2 that a conventional Tafel extrapolation of j e * to E j = 0 does not in any way predict the corrosion rate for this system as j Al is significantly higher than j e * throughout the potential sweep in cathodic branch. The dashed lines in Fig. 2 represent a mixed potential analysis of the Al dissolution rate. The vertical lines designate the E j = 0 and the horizontal lines indicate j Al observed at E j = 0 . It is seen that both E j = 0 and j Al values are in excellent agreement with the E oc and j s Al values observed in the dissolution profiles ( Fig. 1). The interesting point is that O 2 saturation markedly increased the cathodic intensity ( Fig. 2), however the potential change exactly compensated this increase such that the open circuit j Al value was apparently independent of O 2 saturation. This is reasonable if we consider that the anodic reaction may be due to a passivation reaction such as and is, by and large, independent of potential. In this interpretation, O 2 saturation would shift the cathodic branch of the Evans diagram, corresponding to Eqs. 1 or 2, to higher potentials but would have little effect on the anodic reaction.

The potential dependence of Al 5.2 Zn dissolution: overview.-
The open circuit dissolution rates of Al 5.2 Zn in Ar deaerated and O 2 saturated electrolytes are shown in Fig. 3. A steady state Al dissolution rate was measured at j s Al = 36.3 ± 2.3 μA cm −2 (Ar deaeration) and 27.3 ± 0.4 μA cm −2 (O 2 saturation) respectively. The error bar represents the standard deviation of five different measurements.
The steady state open circuit potential in Ar deaerated electrolyte (E oc = −1.31 V) was 200 mV lower than that in O 2 saturated electrolyte (E oc = −1.11 V). In both cases, j Zn was below the detection limit. The positive shift of the potential with respect to pure Al ( Fig.  1) may be attributed to the presence of Zn. In both cases, the O 2 saturation caused a further anodic shift in the potential by 150 to 200 mV and, in this case, a diminution of the Al dissolution rate.
The AESEC polarization curve of Al 5.2 Zn in an Ar deaerated electrolyte (Fig. 4) may be divided into three potential domains based on the rate of Zn dissolution in Fig. 4: Zn is the onset potential for Zn dissolution and E c is the onset potential of spontaneous Al/Zn dissolution. The cathodic dissolution of Al was evident during the cathodic branch (domain I) as was observed for pure Al in Fig. 2. Throughout the cathodic branch, j Zn remained below the detection limit. By mass balance, it follows that a Zn(0) enriched layer was building up on the surface via a "cathodic dealloying" mechanism. Zn dissolution became active in domain II. The peak maximum at E = −1.05 V was due to the dissolution of Zn(0) built up on the surface during the cathodic Al dissolution phase as will be demonstrated later. In the same potential range, j Al reached its lowest level corresponding approximately to the passive dissolution rate of pure Al metal (Fig. 2). In domain III, simultaneous dissolution of Al and Zn occurred at E c = −0.88 V approaching congruent dissolution (j Al /j Zn = 5.8 ± 0.1 and still increasing at the end of the polarization, as compared to a bulk value j Al /j Zn = 7.8) and close to a 100% faradaic yield (η = j /j e * = 1.09 ± 0.12 with j = j Al + j Zn ). The AESEC polarization curve of Al 5.2 Zn (Fig. 4) can be used to predict the spontaneous, open circuit dissolution rate and potential (Fig. 3). From Fig. 4, E j = 0 = −1.28 V as compared to E oc = −1.31 V in Fig. 3. Further, j Al = 37 μA cm −2 at E j = 0 in Fig. 4 was reasonably identical to j s Al in Fig. 3. The stoichiometric relationship between cathodic current and Al  Table I. The values are similar to what was measured for pure Al, the Al 2 Cu pure phase, and various Al alloys in previous work 42   or 2) due to the diffusion of OH − away from the surface. A kinetic model taking into account mass transport has been presented. 42 For Al metal, the value of v e * /v Al was identical within experimental error for the Ar deaerated and the O 2 saturated electrolytes (Table I), despite the significantly more intense cathodic current in the latter, demonstrating that it is indeed the cathodic current that drives Al dissolution.
For Al 5.2 Zn, the v e * /v Al values were slightly higher than those of pure Al probably due to the presence of metallic Zn making Al less available at the surface. Further, for the Al 5.2 Zn phase, the v e * /v Al value in O 2 saturated solution was slightly higher (2.3) than that in Ar deaerated electrolyte (1.6). Note that the initially high v e * /v Al was due to the high current associated with the imposition of the starting potential for the sweep and was not taken into account.  Fig. 6. Note that the sample was exposed at the open circuit for 10 s prior to the potential sweep, sufficient time for the flow cell to fill with electrolyte but short enough to minimize the buildup of residual metallic Zn. Following a short rise period, a stable j e * ≈ j Al of approximately 5 μA cm −2 was obtained, essentially independent of the potential sweep. This is consistent with passive behavior as observed for pure Al metal in this potential range in Fig. 2. Throughout this potential range, j Zn was below the detection limit indicative of Al selective dissolution. At E = −0.88 V, Al and Zn dissolved simultaneously showing a slight Zn dissolution peak at E = −0.85 V. Note that the onset of Zn dissolution was shifted by +260 mV in the positive direction with respect to Fig. 4.

Anodic reactivity of Al
The anodic LSV is less perturbed by residual metallic Zn formed by the cathodic dealloying mechanism. Therefore, E = −0.88 V may be considered as approaching the critical, onset potential (E c ) for si-   Fig. 7 showing a similar result to the LSV in Fig. 4. For E < −1.34 V, j e * and j Al increased simultaneously indicating that once again, j Al was coupled to the cathodic reaction. Cathodic dealloying occurred as no Zn dissolution was observed during the potential sweep. Interestingly, for −1.34 V < E < E oc , the total current was anodic showing a maximum j e * of 5.4 μA cm −2 , with j Al at approximately the same value. This is consistent with the potential transient of Fig. 3 where E oc was initially around −1.2 V and then dropped to approximately −1.31 V at steady state.
Further insight into the cathodic dealloying reaction may be gained by measuring the response of the system (chronoamperometry, CA) to a series of applied potentials, E ap . Two types of experiment were performed; (1) E ap was applied for 1000 s, and then the relaxation of the system to open circuit was recorded. This permits a verification of the cathodic dissolution mechanism free from the effect of sweep rate or surface conditioning and observation of the effect of the Zn(0) enriched dealloyed layer on the j s Al ; (2) E ap was applied for 1000 s and then the potential was stepped to a more positive value to dissolve the Zn from the Zn(0) enriched dealloyed layer. Fig. 8 shows the CA experiments at various applied potentials as indicated. In all cases, j Zn (not shown) was below the detection limit, demonstrating a mechanism of cathodic dealloying. The average values of j Al and j e * at the steady state are summarized in Table II. The v e * /v Al values measured during a constant cathodic potential pulse were 1.86 ± 0.06 at E ap = −1.60 V and 1.00 ± 0.04 at E ap = −1.40 V in rather good agreement with those measured from the polarization experiment of Fig. 5 summarized in Table I. For all applied potentials, after release of E ap , j Al returned to j s Al decaying with the time constant of the flow cell and E returned to E oc . This result demonstrates unambiguously that the Zn(0) enriched dealloyed layer had no significant effect on the selective dissolution of Al. Further, the dissolution transients give no indication that either Zn oxide or Al oxide was present on the alloy surface following E ap .
To corroborate the formation of the Zn(0) enriched layer during the cathodic potential domain, a Zn(0) enriched layer was formed at a constant cathodic potential (E ap = −1.60 V) during 1000 s. Then the potential was instantly switched to an anodic value at which the Zn(0), formed by cathodic dealloying would dissolve, but below the critical potential of Zn dissolution from the bulk alloy (E c = −0.88 V from Figs. 4 and 6). Fig. 9A gives an example of the j Zn transients during an anodic step (E ap = −1.00 V) either with or without a preceding 1000 s at E ap = −1.60 V. Without the cathodic potential step, j Zn was always below the detection limit, as in Fig. 8 demonstrating that Zn in the Al 5.2 Zn matrix did not dissolve at these potentials. Following the cathodic potential step, however, a large peak of Zn dissolution was observed, clearly linked to the enrichment of Zn(0) during cathodic dealloying.
This attribution is also consistent with a mass balance: integration of j Al during the cathodic dealloying step yields 420 nmol Al cm −2 , as compared with 63 nmol Zn cm −2 obtained from the j Zn transient during the anodic step. This yields a stoichiometry of Al/Zn = 6.7, somewhat higher than the bulk composition of Al/Zn = 5.2. The incomplete dissolution of the excess Zn(0) in the dealloyed layer is not surprising as the quantity of Zn dissolved depends strongly on potential. At E ap = −1.10 V (Fig. 9B), the rate of Zn(0) dissolution was decreased by a factor of 5 (integration of peak gives 6.4 nmol Zn cm −2 ), and at E ap = −1.20 V (Fig. 9C) no Zn dissolution was observed.
SEM images following each potential pulse are shown in Fig. 10. At E ap = −1.60 V, a rough surface was observed caused by highly selective dissolution of Al. Lighter spots are observed after E ap = −1.60 V probably due to undissolved Al(OH) 3 . At E ap = −1.40 V, Zn oxide/hydroxide species could be seen at a grain boundary. These were identified as ZnO by Raman spectroscopy with characteristic Raman frequency of 345, 442 and 563 cm −1 . 50,51 The nature of the crystals was investigated at higher resolution in the middle right of Fig. 10. Grain boundary corrosion of Zn-Al alloys was observed by Devillers et al. 52 by polarization experiments and SEM characterization in pH = 9∼11 electrolytes. They concluded that the cathodic reaction occurred at the tip of the corroding grain boundary was due to hydrogen evolution At the more positive potentials, E ap = −1.20 V, −1.10 V (not shown) and −1.00 V, the surface was not remarkably altered as compared to the surface after the humidity chamber. This may be attributed to less pronounced selective Al dissolution at these potentials.    (Fig. 11A) as previously described with a stoichiometric ratio slightly higher than observed in the Ar deaerated electrolyte ( Fig. 5 and Table I). These results were qualitatively and quantitatively similar to the case of the deaerated electrolyte.
From Fig. 11A, the cathodic dissolution of Zn was observed over a 200 mV range (−1.46 V < E < −1.20 V) and correlated with a substantial drop in j Al by approximately a factor of 30. There was also a strong enhancement of the cathodic reaction in this potential range. The cathodic dissolution and its inhibitive effect on Al dissolution was confirmed by the cathodic LSV (Fig. 11B) and the CA at E ap = −1.40 V (Fig. 12). The E c Zn/Zn−OH is the onset potential of cathodic Zn dissolution. Note that for the cathodic LSV starting at E oc (Fig. 11B) the potential range of Zn dissolution was shifted to more negative potential (−1.60 V < E < −1.40 V) as compared to that in Fig. 11A, no doubt due to the slower increase in interfacial pH associated with the cathodic potential sweep. The anodic LSV with O 2 saturation, shown in Fig. 11C, demonstrates that O 2 has no discernable effect on anodic dissolution of Al 5.2 Zn with E c = −0.88 V for both electrolytes. For the CA experiment at E ap = −1.40 V (Fig.  12), the cathodic current (j e * ) with O 2 saturation was 6 times more intense than with the Ar deaeration.
Insight into the mechanism of cathodic Zn dissolution may be gained from the SEM micrographs (Fig. 13) of the surface obtained after the CA experiment at E ap = −1.40 V. With Ar deaeration (Fig. 13A), crystals of ZnO were clearly observed at what appears to be grain boundaries as in Fig. 10. In the O 2 saturated condition  (Fig. 13B), however, these crystals were not observed. These results support the hypothesis that Zn oxidation occurs at E = −1.40 V especially around the grain boundaries. The cathodic dissolution of Zn occurs in the presence of saturation O 2 , because the ZnO formed will dissolve directly by complexation with OH − generated by O 2 reduction. The overall reaction would be as follows: ZnO where the hydroxide generated by the cathodic reaction drives Eq. 7.

Discussion
As stated in the introduction, the α-phase Al 5.2 Zn, investigated here, and the MgZn 2 investigated previously, 33 are representative of the extremes of Al and Mg content in the Zn-Al-Mg coating. Both phases were demonstrated to undergo dealloying reactions when polarized to cathodic potentials. In both cases, this results in the formation of a Zn(0) enriched layer. For MgZn 2 , the Zn(0) enriched layer significantly reduced Mg dissolution. 33 However, the cathodic dealloying of α-phase Al 5.2 Zn was not restrained by this layer despite the fact that the quantity of Zn(0) in the enriched layer was approximately 9 times thicker for Al 5.2 Zn than for MgZn 2 intermetallic. (A mass balance in Ar deaerated solution from Fig. 12, predicts 155 nm of Zn(0) as compared to 17 nm from a similar experiment in Ref. 33, assuming a uniformly distributed, homogenous metallic Zn(0) layer.) The consequence of this is that the MgZn 2 phase undergoes a type of passivity at open circuit as the Zn(0) layer forms and progressively hinders further dissolution. Further, the presence of Mg maintains the potential below E c Zn , preventing the dissolution of the Zn(0) film. The α-phase Al 5.2 Zn however remains active during cathodic polarization due to the high Al dissolution rate. In practice, this would suggest that the high %Al phase remains anodic to the high %Mg phase despite the fact that Mg is the least noble element in the Zn-Al-Mg coating. This may be understood if it is considered that the dealloyed Zn(0) layer is no doubt porous, and the cathodic reaction within the confined porosity would intensify the pH increase. The precipitation of Mg(OH) 2 in the porosity would most likely block the contact between the electrolyte and the underlying MgZn 2 phase. No similar effect would be predicted for the α-phase Al-Zn pure phase, as the solubility of Al(III) oxides and hydroxides increases with pH due to the formation of Al(OH) 4 − . A schematic dealloying model of α-phase Al 5.2 Zn is proposed in Fig. 14. The cathodic deallyoing of Al occurs in potential domain I forming a Zn(0) enriched layer. In the O 2 saturated electrolyte, Al dissolution is inhibited in domain I'. In domain II, E c Zn < E < E c , Zn in the Zn(0) rich film dissolves independently of the Zn in the matrix. Finally, in domain III, E > E c , both Al and Zn dissolve simultaneously.
Pickering et al. 53 classified dealloying reactions based on whether or not dissolution was selective at the critical potential. For Type I dealloying, the critical potential corresponds only to the onset of dissolution of the least noble element. For Type II dealloying, both elements dissolve congruently at the critical potential. The polarization curve of MgZn 2 intermetallic 33 showed a clear Type II behavior, as does the anodic LSV (Fig. 6) of Al 5.2 Zn, in that j Al increased simultaneously with j Zn . However, the onset of cathodic dealloying of Al precludes application of the Pickering classification in that Al dissolution increases in the cathodic direction and Zn dissolution in the anodic direction. For example, in the potential domain II of the cathodic to anodic LSV (Fig. 4), the enriched Zn(0) (more noble) is dissolved selectively with respect to Al (least noble). In the domain III, however, dissolution is congruent.
The nature of the cathodic reaction in the presence of saturated O 2 is not well understood. Saturated O 2 showed no effect on the cathodic current below the onset of cathodic Zn dissolution (E < E c Zn/Zn−OH , domain I) irrespective of the sweep direction (Figs. 11A and 11B), suggesting that only water reduction occurred in this potential domain. This is surprising in that O 2 reduction was clearly observed on pure Al (Fig. 2) and pure Zn. 33 In the potential domain I' (Figs. 11A and 11B), the cathodic reaction increased markedly. Simultaneously, j Al decreased by an order of magnitude while j Zn rose above the background level indicating cathodic Zn dissolution. The enhanced cathodic reaction suggests that O 2 reduction begins at this potential. The increase in j Zn was described as due to the oxidation of the Zn followed by its dissolution as hydroxide complexes (Zn(OH) 3 − and Zn(OH) 4 2− ). The decrease in j Al is more difficult to explain. It is possible that the higher redox potential associated with saturation O 2 might favor a more compact passive film less susceptible to reaction with hydroxide. In any case, the low dissolution rate of Zn and the absence of Zn based corrosion products would argue against an inhibitive mechanism induced by Zn corrosion products on Al dissolution. Dafydd et al. 54 observed that Zn metal, Zn-0.1Al and Zn-4.3Al showed clear 4e − (n = 4) pathway O 2 reduction reaction at pH = 9.6 whereas for Zn-55Al, an apparent n = 2.8 was determined. They attributed the low n value to O 2 reduction occurring only on the Zn rich zones of the Zn-55Al alloy. Cathodic Al dissolution of the α-phase of Al-Zn in the Zn-55Al could also contribute to an apparently low n  value since the real cathodic current would be more intense than the measured cathodic current.
As an aid in predicting the coupling between the different phases in the technical coatings, a galvanic series was constructed for the alloys and metals investigated in this study. Table III summarizes the potential values obtained for Al, Zn, and Mg metals, Zn-5Al, Zn-55Al commercial alloy coatings, the MgZn 2 intermetallic, and the α-phase Al 5.2 Zn in Ar deaerated and O 2 saturated electrolyte. For Zn, Mg metals and the MgZn 2 intermetallic, these values were obtained from data of Ref. 33. For the commercial alloys of Zn-5Al and Zn-55Al, the potential values were obtained in the same experimental condition. Based on the E oc data, the α-phase Al 5.2 Zn is cathodic to pure Al and anodic to both Zn-Al alloy coatings, pure Zn and MgZn 2 . The more noble potential of the MgZn 2 is due to the formation of the Zn(0) film by selective dissolution of Mg.
The E c Zn values are nearly identical for Al 5.2 Zn, Zn-55Al and Zn-5Al, which may indicate that the onset potential of Zn(0) enriched layer dissolution is determined independently of Al content. Likewise, the E j = 0 values of Al 5.2 Zn in both Ar pure and O 2 saturation were very similar to those of Zn-55Al and Zn-5Al coatings. This suggests that the electrochemical behavior of the alloys is determined by the Al 5.2 Zn phase contained in these alloys. On the other hand, the spontaneous dissolution of Zn-5Al is controlled by the Zn of the Zn rich phase. Preferential dissolution of the ternary eutectic phase of Zn-Al-Mg alloys can be attributed to the more negative E oc of Al 5.2 Zn as compared to that of the other phases present in the alloy coating.
This hypothesis is consistent with field exposure data of the coating alloys. Higher reactivity of the α-phase of Al-Zn was evidenced by the less noble potential observed by Ramus Moreira et al. 29 during atmospheric corrosion tests of the Zn-55Al alloy coating in six different exposure sites. The result showed that the Zn rich interdendritic phase of Zn-55Al was composed of one other Zn rich phase and the other Al rich phase (α-phase of Al-Zn). A preferential corrosion of Al rich phase in the interdendritic region was observed by SEM and X-ray energy dispersive spectroscopy (SEM-EDS) after five years of atmospheric exposure demonstrating that the significance of α-phase of Al-Zn phase for the corrosion of the Zn-Al alloys.

Conclusions
1) The α-phase Al 5.2 Zn, and pure Al undergo a mechanism of cathodic dealloying due to the selective dissolution of Al during the application of a cathodic current. The stoichiometry of the reaction (v e * /v Al ) was between 1 and 2 consistent with Eqs. 1 and 2.
2) Cathodic dealloying led to the formation of a Zn(0) enriched layer. This layer did not inhibit further Al dissolution. This is very different from the previously studied MgZn 2 , in which the dealloyed film of Zn(0) did yield an inhibitive effect on Mg dissolution. 3) The Zn(0) layer dissolved at a potential, E c Zn , below the critical, onset potential, E c of Zn in the Al 5.2 Zn matrix. 4) Al and Zn in the α-phase Al 5.2 Zn dissolved simultaneously for E > E c = −0.88 V with almost 100% faradaic yield. The E c was not affected by the precedent potential sweep or O 2 saturation in the electrolyte. 5) The cathodic dissolution of Zn was observed in O 2 saturated electrolyte for E > E c Zn/Zn−OH . This was attributed to the enhanced cathodic reaction due to the increased O 2 concentration resulting in the dissolution of ZnO/Zn(OH) 2 by complexation with OH − . 6) The cathodic dissolution of Zn correlated with a simultaneous cathodic current increase and an order of magnitude decrease in the Al dissolution rate. The latter probably reflects a passivation of the Al due to reaction with O 2 . 7) The α-phase Al-Zn present in Zn-Al and Zn-Al-Mg alloy coatings maybe polarized anodically or cathodically depending on conditions. Cathodic Al dissolution observed in this work can be important to predict corrosion behavior of these alloys because the Al rich phase may supply Al 3+ into solution contributing the formation of the corrosion resistant Zn-Al LDH.