Electrochemical Behavior of Al–Al 9 Co 2 Alloys in Sulfuric Acid

: Al–Co alloys of various Co contents (2–20 wt.% Co) were fabricated by vacuum arc melting (VAM) with the scope to investigate the inﬂuence of cobalt on the microstructure and corrosion resistance of Al in 1 M H 2 SO 4 . The obtained microstructures were directional, consisting of Al 9 Co 2 platelets (grown to coarse acicular plates as the Co content increased) uniformly dispersed in an Al-matrix. Alloying Al with Co did not decrease the rate of uniform corrosion of Al but it considerably increased its passivation ability. Moreover, all Al–Co alloys displayed lower uniform corrosion rate and notably higher passivation ability than market leading Al-alloys. The underlying mechanisms during anodic polarization in 1 M H 2 SO 4 were identiﬁed and correlated with the microstructure. High Co content alloys (7–20 wt.% Co) presented superior passivation ability in 1 M H 2 SO 4 as compared to the low Co content alloys.


Introduction
During the last decades, rapidly solidified (RS) or intensively cooled Al-alloys have attracted significant attention owing to improved properties, such as mechanical properties and corrosion resistance. More specifically, many works have reported that RS of Al-alloys has led to refined microstructures as compared to the conventionally cast counterparts [1][2][3], supersaturation of Al with transition metals (TM), alkaline earth metals or metalloids [1,2,[4][5][6][7], formation of intermetallic compound (IC) phases (their morphology depending on the cooling rate and the content of the solute element in the alloys [1,2,4]), and formation of semi-crystalline or amorphous phases [8,9].
Although the processing and physical metallurgy of RS bulk Al-alloys have extensively been studied, their corrosion behavior is much less investigated. RS of Al-alloys is reported to lead to improved corrosion resistance, mainly due to microstructure refinement [10][11][12][13] (in terms of grain size [12] and IC dispersion [10][11][12]), microstructure homogeneity [11], reduction of the precipitate content [10,14], and supersaturation of Al with elements like Ti, Mn, Cr, Ni, Cu, Zr, and Nb [11,14,15]. However, the RS-due changes in the electrochemical characteristics of the constituent phases may negatively affect the corrosion resistance, as in the cases of supersaturation of Al with Zn [14,16], Mg and Fe [14].
Established knowledge on rapid solidification mainly addresses cooling rates higher than 10 4 K/s. Intensive cooling rates, as those employed in vacuum arc melting (10-150 K/s) have also been shown to lead to refined structures and non-equilibrium phenomena, such as formation of supersaturated phases, non-equilibrium transformations, and suppression of equilibrium phase diagram regimes [17,18].

Microstructural Analysis
The microstructures of the produced Al-Co alloys are presented in Figure 1. A detailed microstructural analysis in context with the solidification mechanisms is reported elsewhere [20]. All alloys are composed of αAl and Al 9 Co 2 , as has been identified by EDX and XRD analysis [20]. The volume fractions of the Al 9 Co 2 phase in Al-2Co, Al-5Co, Al-7Co, Al-10Co, Al-15Co, and Al-20Co (wt.%), have been determined (by image analysis) as 19 ± 2, 30 ± 2, 36 ± 3, 41 ± 1, 50 ± 1, and 63 ± 1 vol.%, respectively. Despite the negligible solid solubility limit of Co in Al under equilibrium conditions [45], notably higher concentrations of Co have been detected within the Al-matrix that increase with Co content. They range from max 0.49 wt.% in Al-5Co to max 5.51 wt.% in Al-20Co [20]. The supersaturation of the Al-matrix with Co has been caused by the fast solidification that stretched out the solid solubility limit of Co in Al.  Although all the compositions are hypereutectic [45], primary Al 9 Co 2 is hardly discerned in the low Co (2, 5, 7, 10 wt. % Co) microstructures. The almost entirely eutectic morphologies suggest the suppression of the pre-eutectic stage due to rapid cooling. Figure 1a-d manifests that fine plates of Al 9 Co 2 are distributed in the Al-matrix in a uniform and directional pattern, the outcome of preferential heat extraction [40]. In the microstructures of Al-(2-10) wt.% Co, relatively coarse stripes of bright contrast are observed to delimit the boundaries of elongated colonies. These colonies include fine dispersions of Al 9 Co 2 eutectic platelets. In the case of Al-2 wt.% Co, this dispersion is very fine, dense and uniform. This morphology is considered to be the outcome of the high cooling rates, in the following way: At the beginning of solidification, an Al 9 Co 2 stripe is formed. Due to the high cooling rate, there is not enough time for further crystal growth, since atomic diffusion to the surface of the existing crystal is constrained. To continue their growth, the crystals must develop along new directions that are easy for an effective atom capture. These preferential directions can be screw dislocations, the density of which increases with cooling rate [46,47]. Amongst the examined alloys, Al-2 wt.% Co has the highest solidification rate, owing to the lowest content of Co.
As far as Al-15 wt.% Co is concerned, planar Al 9 Co 2 is discerned along with the eutectic Al 9 Co 2 /Al microconstituent (Figure 1e). In Figure 1b-e, it is observed that the Al 9 Co 2 eutectic stripes are not continuous but they comprise small building units with a directional alignment. Such directional but interrupted growth has been observed in various types of arc-melted in-situ composite materials [46,48]. It has been attributed to the different cooling rates. When solidification starts, the cooling rate is high enough to cause the formation of a continuous Al 9 Co 2 stripe. With the progress of solidification, the cooling rate decreases and the remaining liquid around the stripe becomes impoverished in Co. Such impoverishment interrupts the mode of continuous growth resulting in a brick-like pattern.
The predominant microstructural feature of Al-20 wt.% Co is that of coarse Al 9 Co 2 plates with acicular acmes (Figure 1f). This morphology suggests an extensive presence of primary Al 9 Co 2 , which is attributed to the much lower thermal conductivity of Al 9 Co 2 as compared to Al (Al 9 Co 2 , as a complex metallic alloy, has very low conductivity [49]). Hence, the cooling rate of Al 9 Co 2 is much lower than that of Al, allowing the crystallization and development of coarse pre-eutectic Al 9 Co 2 . The above considerations can justify the observed increase in the amount of coarse acicular Al 9 Co 2 plates and the corresponding decrease in the fraction of fine Al 9 Co 2 eutectics, with Co content increasing.  Figure 2a,b, respectively. The voltammograms of two leading commercial Al-alloys (Al2024-T3, Al7075-T6), widely used in the aviation and military industries, are included in Figure 2c for comparison. Table 1 lists critical electrochemical values drawn from the polarization curves. Table 2 lists the data extracted from the Tafel extrapolation on the cathodic polarization curves. Figure 3 shows the microstructural state of the alloys after cyclic polarization in 1 M H 2 SO 4 (surface and cross-section views).
Corros. Mater. Degrad. 2020, 1 FOR PEER REVIEW 6 At a first glance, Figure 2 and Tables 1 and 2 show that the differences in the electrochemical behavior of the Al-Co alloys are small. Steep changes in the current density trends define distinct stages on the forward anodic portions of the polarization curves, noted in Figure 2b. However, on a closer observation, the alloys are classified into two groups on the basis of their anodic polarization behavior, namely: Al-(2-5) wt.% Co: Stage 1 is due to the active corrosion of the (Al) phase, since Al is more active than Al9Co2 [35]. Electrolyte penetration through the defects of the Al-surface film (existing after 4 h of immersion under open-circuit [50]) has taken place during stage 1. The defects can be the Al9Co2/(Al) interfaces and/or (Al)-grain boundaries where AlFeSi intermetallics reside [40]. Hydrogen evolution during the preceding cathodic polarization may also have caused damage to the surface film existing at Ecorr [51,52]. More specifically, stage 1 is governed by ionic migration across the pre-existent surface film [53].
Passive-like stage 2 follows, extending for over 1000 mV (actually, no depassivation has taken place during the forward anodic scan). In the case of the Al-matrix, passivation reportedly occurs by film growth at the metal/film interface, with direct ejection of Al 3+ to the film-electrolyte interface [53]. The relatively high values of current density (i > 0.1 mA/cm 2 ) suggest pseudopassivation. The pseudopassivation of Al is justified by the formation of a film that is soluble in the outer layers [54].
The positive hysteresis loop through stage 2 ( Figure 2a) indicates the efficient protectivity of the surface film (although the film on the Al-matrix is soluble at the oxide/electrolyte interface, it has a barrier effect at the metal/oxide interface [54].) However, the hysteresis turns to negative, sectioning the forward anodic scan at potentials just a couple of decades lower than the critical passivation potential (Ecp). As a consequence, Ea/c tr is less noble than Ecorr by a few decades of mVs, indicating a slightly less noble surface at Ea/c tr (upon reverse polarization) than that at Ecorr (upon forward polarization).      In 75% of all the tested specimens (7-20 wt.% Co), the reverse polarization curves presented multiple anodic-to-cathodic transition potentials ("cathodic loop"), usually forming "spikes", and positive hysteresis through essentially the whole reverse scan, as illustrated in Figure 4. These multiple (2-3, commonly 3) anodic-to-cathodic transition potentials correspond to the intersection points of the anodic and cathodic Evans lines and they are often observed in the reverse polarization scans in the absence of pitting [55]. Their origin is attributed to the shift of the return anodic line towards smaller current densities as compared to the forward ones due to the formation of passive films during the preceding forward anodic scanning. Hence, during reverse scanning, the unchanged At a first glance, Figure 2 and Tables 1 and 2 show that the differences in the electrochemical behavior of the Al-Co alloys are small. Steep changes in the current density trends define distinct stages on the forward anodic portions of the polarization curves, noted in Figure 2b. However, on a closer observation, the alloys are classified into two groups on the basis of their anodic polarization behavior, namely: Al-(2-5) wt.% Co: Stage 1 is due to the active corrosion of the (Al) phase, since Al is more active than Al 9 Co 2 [35]. Electrolyte penetration through the defects of the Al-surface film (existing after 4 h of immersion under open-circuit [50]) has taken place during stage 1. The defects can be the Al 9 Co 2 /(Al) interfaces and/or (Al)-grain boundaries where AlFeSi intermetallics reside [40]. Hydrogen evolution during the preceding cathodic polarization may also have caused damage to the surface film existing at E corr [51,52]. More specifically, stage 1 is governed by ionic migration across the pre-existent surface film [53].
Passive-like stage 2 follows, extending for over 1000 mV (actually, no depassivation has taken place during the forward anodic scan). In the case of the Al-matrix, passivation reportedly occurs by film growth at the metal/film interface, with direct ejection of Al 3+ to the film-electrolyte interface [53]. The relatively high values of current density (i > 0.1 mA/cm 2 ) suggest pseudopassivation. The pseudopassivation of Al is justified by the formation of a film that is soluble in the outer layers [54].
The positive hysteresis loop through stage 2 ( Figure 2a) indicates the efficient protectivity of the surface film (although the film on the Al-matrix is soluble at the oxide/electrolyte interface, it has a barrier effect at the metal/oxide interface [54].) However, the hysteresis turns to negative, sectioning the forward anodic scan at potentials just a couple of decades lower than the critical passivation potential (E cp ). As a consequence, E a/c tr is less noble than E corr by a few decades of mVs, indicating a slightly less noble surface at E a/c tr (upon reverse polarization) than that at E corr (upon forward polarization).
Al-(7-20) wt.% Co: Stage 1 corresponds to the selective dissolution of (Al), seen in Figure 3b-d. Stage 1 is followed by a passive-like stage (stage 2) extending for a potential range (E b1 − E cp1 ) of 385 to 432 mV. Unlike the low Co compositions, an increase in current density follows (stage 3), eventually leading to passive-like stage 4. The high slope of the E vs. log(i) curve during stage 3 along with the positive hysteresis upon reverse polarization through stage 3 ( Figure 2a) suggest that this stage is owing to the occurrence of an oxidation process rather than a localized degradation process (the processes causing stages 3 and 4 are attempted to be clarified by chronoamperometry testing in 3.2.2). In the 25% of all the tested specimens (7-20 wt.% Co), hysteresis has remained positive through the whole reverse anodic scan resulting in an anodic-to-cathodic transition potential that is nobler than the corrosion potential by several hundred mVs ( Figure 2a). The latter indicates that the surface at E a/c tr (return scan) is nobler than the surface at E corr (forward scan).
In 75% of all the tested specimens (7-20 wt.% Co), the reverse polarization curves presented multiple anodic-to-cathodic transition potentials ("cathodic loop"), usually forming "spikes", and positive hysteresis through essentially the whole reverse scan, as illustrated in Figure 4. These multiple (2-3, commonly 3) anodic-to-cathodic transition potentials correspond to the intersection points of the anodic and cathodic Evans lines and they are often observed in the reverse polarization scans in the absence of pitting [55]. Their origin is attributed to the shift of the return anodic line towards smaller current densities as compared to the forward ones due to the formation of passive films during the preceding forward anodic scanning. Hence, during reverse scanning, the unchanged cathodic line intersects the reverse anodic line at more than one intersection points [55]. Therefore, a non-susceptibility to localized corrosion is suggested. Table 1 includes the values of the uppermost and lowermost anodic-to-cathodic potentials, since these are considered the most stable ones [55]. A third of the Al-(2-5) wt.% Co tested specimens and a quarter of the CP-Al tested specimens also presented a cathodic loop. Usually, one of the two E a/c tr potentials corresponded to the formation of a short and relatively blunt protrusion (a "nose" rather than a "spike"), as the one shown in Figure 4 (Al-5wt.% Co).
hydrous aluminum sulfate macromolecules to active sites on cathodic zones [51] (the adsorption of sulfates, also occurring on anodic sites, will be discussed in 3.2.2.). The effective protectivity of the surface film suggested by cyclic polarization is manifested by the good surface state of the Al-Co alloys in the SEM micrographs of Figure 3 (both surface view and cross-section view). Figure 3a-d also shows that the lower the Co content, and hence, the finer the Al9Co2 particles, the better defined the "cellular" morphology of the surface film. At 5 wt.% Co (Figure 3a), the film "cells" present a surface morphology resembling the well-defined hexagonal cells of the films on pure Al [53]. As the Co content increases, the cell diameter becomes smaller and is confined between the plates of Al9Co2. A closer observation of Figure 3d reveals a similar "cellular" morphology on the Al9Co2 plates. Figure 5 illustrates the chronoamperometry plots of a low Co-containing alloy (Al-5 wt.% Co) and a high Co-containing alloy (Al-20 wt.% Co) at potentials corresponding to different corrosion states (active (stage 1), passive-like (stage 2), transition (stage 3), and passive-like (stage 4)). In the case of polarization at a potential in stage 1 (−360 mV vs. Ag/AgCl), the initial rapid increase in the current density is followed by a gradual decrease in the rate of current density increase. Eventually, the current density is stabilized. The above trend signifies an active dissolution process gradually weakening due to the gradual reduction of the available active surface and/or surface deposition of corrosion products.

Chronoamperometry
Al-5 wt.% Co: The current density vs. time curves of Al-5 wt.% Co at two potentials in stage 2 (a potential in the low range of "passive" potentials (0 mV vs. Ag/AgCl) and a potential in the high range of "passive" potentials (600 mV vs. Ag/AgCl)) present trends of passive-like behavior in agreement with the potentiodynamic polarization findings. More specifically, the initial steep increase in the current density (notably steeper at 600 mV) reflects fast oxidation. The subsequent current decrease (faster at 600 mV) corresponds to the growth of surface films. Eventually, current stabilization is attained signifying the stabilization of the surface films in terms of thickness and extent (after ~95 min in the case of polarization at 0 mV vs. Ag/AgCl and ~70 min in the case of polarization at 600 mV vs. Ag/AgCl). However, the high values and the oscillations of the current density, especially at 600 mV vs. Ag/AgCl, imply unstable surface products and/or metastable pitting. Metastable pitting of Al at defective sites of the oxide film in Al immersed in pure H2SO4 under open circuit has been observed by Mercier et al. [50]. The surface film on Al exposed to aqueous H2SO4 dissolves in the outer zones (film/solution interface) that are composed of Al2O3 hydrates and show that the reverse cathodic currents are slightly lower than the forward cathodic currents, indicating a cathodic inhibition mechanism reportedly based on the adsorption of hydrous aluminum sulfate macromolecules to active sites on cathodic zones [51] (the adsorption of sulfates, also occurring on anodic sites, will be discussed in Section 3.2.2).
The effective protectivity of the surface film suggested by cyclic polarization is manifested by the good surface state of the Al-Co alloys in the SEM micrographs of Figure 3 (both surface view and cross-section view). Figure 3a-d also shows that the lower the Co content, and hence, the finer the Al 9 Co 2 particles, the better defined the "cellular" morphology of the surface film. At 5 wt.% Co (Figure 3a), the film "cells" present a surface morphology resembling the well-defined hexagonal cells of the films on pure Al [53]. As the Co content increases, the cell diameter becomes smaller and is confined between the plates of Al 9 Co 2 . A closer observation of Figure 3d reveals a similar "cellular" morphology on the Al 9 Co 2 plates. Figure 5 illustrates the chronoamperometry plots of a low Co-containing alloy (Al-5 wt.% Co) and a high Co-containing alloy (Al-20 wt.% Co) at potentials corresponding to different corrosion states (active (stage 1), passive-like (stage 2), transition (stage 3), and passive-like (stage 4)). In the case of polarization at a potential in stage 1 (−360 mV vs. Ag/AgCl), the initial rapid increase in the current density is followed by a gradual decrease in the rate of current density increase. Eventually, the current density is stabilized. The above trend signifies an active dissolution process gradually weakening due to the gradual reduction of the available active surface and/or surface deposition of corrosion products.

Chronoamperometry
Al-5 wt.% Co: The current density vs. time curves of Al-5 wt.% Co at two potentials in stage 2 (a potential in the low range of "passive" potentials (0 mV vs. Ag/AgCl) and a potential in the high range of "passive" potentials (600 mV vs. Ag/AgCl)) present trends of passive-like behavior in agreement with the potentiodynamic polarization findings. More specifically, the initial steep increase in the current density (notably steeper at 600 mV) reflects fast oxidation. The subsequent current decrease (faster at 600 mV) corresponds to the growth of surface films. Eventually, current stabilization is attained signifying the stabilization of the surface films in terms of thickness and extent (after~95 min in the case of polarization at 0 mV vs. Ag/AgCl and~70 min in the case of polarization at 600 mV vs. Ag/AgCl). However, the high values and the oscillations of the current density, especially at 600 mV vs. Ag/AgCl, imply unstable surface products and/or metastable pitting. Metastable pitting of Al at defective sites of the oxide film in Al immersed in pure H 2 SO 4 under open circuit has been observed by Mercier et al. [50]. The surface film on Al exposed to aqueous H 2 SO 4 dissolves in the outer zones (film/solution interface) that are composed of Al 2 O 3 hydrates and transitory complexes of SO 4 2− , whereas it reforms in the inner zones that are composed of barrier anhydrous alumina [50,54]. Hence, a dynamic equilibrium is established through the simultaneous formation and dissolution of the film. The outer surface film layers are responsible for the high currents recorded during chronoamperometry ( Figure 5) and potentiodynamic polarization through stages 2 and 4 (Figures 2 and 4). The anhydrous inner layer provides the protection shown by the positive hysteresis loop through the passive stages in Figures 2 and 4.
Corros. Mater. Degrad. 2020, 1 FOR PEER REVIEW 11 transitory complexes of SO4 2− , whereas it reforms in the inner zones that are composed of barrier anhydrous alumina [50,54]. Hence, a dynamic equilibrium is established through the simultaneous formation and dissolution of the film. The outer surface film layers are responsible for the high currents recorded during chronoamperometry ( Figure 5) and potentiodynamic polarization through stages 2 and 4 (Figures 2 and 4). The anhydrous inner layer provides the protection shown by the positive hysteresis loop through the passive stages in Figures 2 and 4. Through the whole range of passive-like stage 2, the Al-matrix of Al-5 wt.% Co appears oxidized (Figure 6a, spectrum 31 and Figure 6b, spectrum 29). In the case of polarization at 0 mV vs. Ag/AgCl, the Al9Co2 phase has remained free of any oxidation signs (Figure 6a, spectrum 25); hence, it is shown that oxidation has taken place only on the surface of the Al-matrix. Instead, in the case of polarization at 600 mV vs. Ag/AgCl, oxygen is detected on the surface of the Al9Co2 particles ( Figure 6b, spectrum 26). Therefore, as far as the low Co Al-alloys are concerned (2, 5 wt.% Co), it is drawn that passivation in the low range of anodic potentials is exclusively caused by the formation of an Al2O3-based film on the Al-matrix; at the high range of anodic potentials, passivation of Al9Co2 also takes place. Through the whole range of passive-like stage 2, the Al-matrix of Al-5 wt.% Co appears oxidized (Figure 6a, spectrum 31 and Figure 6b, spectrum 29). In the case of polarization at 0 mV vs. Ag/AgCl, the Al 9 Co 2 phase has remained free of any oxidation signs (Figure 6a, spectrum 25); hence, it is shown that oxidation has taken place only on the surface of the Al-matrix. Instead, in the case of polarization at 600 mV vs. Ag/AgCl, oxygen is detected on the surface of the Al 9 Co 2 particles (Figure 6b, spectrum 26). Therefore, as far as the low Co Al-alloys are concerned (2, 5 wt.% Co), it is drawn that passivation in the low range of anodic potentials is exclusively caused by the formation of an Al 2 O 3 -based film on the Al-matrix; at the high range of anodic potentials, passivation of Al 9 Co 2 also takes place.
Al-20 wt.% Co: The polarization behavior of Al-20 wt.% Co at 0 mV vs. Ag/AgCl (stage 2) and 600 mV vs. Ag/AgCl (stage 4) presents passive-like trends in agreement with the potentiodynamic polarization results. The shapes of the curves are very similar to the shapes of the respective curves of Al-5 wt.% Co. The fluctuations and high values of current density also suggest pseudopassivity and/or metastable pits. The current density values, especially in the stabilization parts, are close to the respective values of Al-5 wt.% Co. The similar curve shapes and current density values indicate similar corrosion phenomena for both the low and high Co Al-alloys. Indeed, the microstructural observations in Figure 7 confirm the above claim. Figure 7a reveals the formation of alumina-based films on the Al-phase of Al-20 wt.% Co polarized at 0 mV vs. Ag/AgCl (spectrum 53), as in the case of Al-5 wt.% Co. However, unlike Al-5 wt.% Co, the EDX spectra of the matrix surface of Al-20 wt.% Co include Co. The Al 9 Co 2 phase has remained free of any oxidation signs (Figure 7a, spectrum 51). Therefore, it is deduced that, as in the case of the low Co Al-alloys polarized in the low range of passive potentials, stage 2 in the high Co Al-alloys has been caused by the formation of alumina-based films (containing some Co) on the Al-phase.  Figure 7 confirm the above claim. Figure 7a reveals the formation of alumina-based films on the Al-phase of Al-20 wt.% Co polarized at 0 mV vs. Ag/AgCl (spectrum 53), as in the case of Al-5 wt.% Co. However, unlike Al-5 wt.% Co, the EDX spectra of the matrix surface of Al-20 wt.% Co include Co. The Al9Co2 phase has remained free of any oxidation signs (Figure 7a, spectrum 51). Therefore, it is deduced that, as in the case of the low Co Al-alloys polarized in the low range of passive potentials, stage 2 in the high Co Al-alloys has been caused by the formation of aluminabased films (containing some Co) on the Al-phase.
Regarding passive-like stage 4, Figure 7b shows that, besides passivation of the (Al) matrix, passivation of Al9Co2 has also occurred ( Figure 7b, EDX elemental mapping).  Regarding passive-like stage 4, Figure 7b shows that, besides passivation of the (Al) matrix, passivation of Al 9 Co 2 has also occurred ( Figure 7b, EDX elemental mapping).
The shape of the current density vs. time curve (Al-20 wt.% Co) at a potential in stage 3 (280 mV vs. Ag/AgCl) enhances the claim of an oxidation process proposed in 3.2.1. Comparing the shapes of the curves at passive-like stages 2 and 4 with the curve at stage 3, it is evident that the current initially rises at intermediate values between those of active stage 1 and those of passive-like stages 2 and 4. Then, the current density does not drop as in stages 2 and 4 but it continues to rise at a slow rate that gradually decreases with time elapsing (up to~25 min). A stabilization trend is eventually attained. Taking into account the presence of oxygen on the surface of Al 9 Co 2 in stage 4 (Figure 7b-EDX mapping) and its respective absence in stage 2 (Figure 7a-spectrum 51), it is drawn that intermediate stage 3 is mainly attributed to the oxidation of Al 9 Co 2 . Its distinctiveness, unlike the case of the low Co Al-alloys, is most likely due to the high surface area of the Al 9 Co 2 plates.
Often, minor amounts or traces of sulfur have been detected on the surface of the oxidized matrix after chronoamperometry at passive potentials (Figure 6b-spectrum 29, Figure 7a-spectrum 53), as well as after cyclic polarization (Figure 8-spectrum 6). Adsorption of SO 4 2− on the hydrous alumina can block active sites of the surface film [50]. Sulfates can rapidly displace aquo groups (Al-OH 2 + ) or hydroxo (A-OH 2 + ) groups on hydrous Al 2 O 3 [56]. This blocking action does not, however, ensure inhibition of localized corrosion. The latter depends on the defects present in the structure of the surface film [50] that can be numerous [52]. This film is a porous-type anodic film, with a barrier layer at the metal/film interface and numerous imperfections [54]. Sulfur is also detected on the surface of oxidized Al 9 Co 2 ( Figure 8-spectrum 5).
Corros. Mater. Degrad. 2020, 1 FOR PEER REVIEW 14 To summarize, both potentiodynamic polarization and chronoamperometry experiments showed that the anodic polarization of Al-Co in 1 M H2SO4 at 25 °C proceeds via the stages: (1) Active dissolution of the Al-matrix by ionic conduction across the preexistent surface film; (2) pseudopassivation of the Al-matrix by film formation at the metal/oxide interface and film dissolution at the film/electrolyte interface; (3) oxidation of the surface of the Al9Co2 particles; and (4) extension of pseudo-passivation to the surface of the Al9Co2 particles. Stage 3 is distinct only in the case of the high-Co alloys (7-20 wt.% Co).

Polarization Performance in 1 M H2SO4 Versus Co Content
Tables 1 and 2 show that neither the corrosion potential values nor the corrosion current density values as a function of the Co content present any meaningful trends. Regarding Ecorr, on closer observation, one could claim that the high Co Al-alloys present nobler corrosion potentials as compared to the low Co Al-alloys (compare the Ecorr of Al-2 wt.% Co with the Ecorr of Al-20 wt.% Co in Table 1). This trend can be justified by the increased Co dissolution in the Al-matrix at the high Co contents, as aforementioned in 3.1 (the corrosion potential of electrolytic Co in 1 M H2SO4 has been measured by potentiodynamic polarization as (−305 ± 11) mV vs. Ag/AgCl, whereas the corrosion potential of CP-Al in 1 M H2SO4 has been measured as (−463 ± 8) mV vs. Ag/AgCl.) Anyway, this ennoblement is not conclusive, as the variations in the mean Ecorr values are small and often within experimental error.
Regarding icorr, the insignificant variations (within experimental error) in the icorr values of the various Al-Co compositions can be justified by the following competing contributions: As the Co To summarize, both potentiodynamic polarization and chronoamperometry experiments showed that the anodic polarization of Al-Co in 1 M H 2 SO 4 at 25 • C proceeds via the stages: (1) Active dissolution of the Al-matrix by ionic conduction across the preexistent surface film; (2) pseudo-passivation of the Al-matrix by film formation at the metal/oxide interface and film dissolution at the film/electrolyte interface; (3) oxidation of the surface of the Al 9 Co 2 particles; and (4) extension of pseudo-passivation to the surface of the Al 9 Co 2 particles. Stage 3 is distinct only in the case of the high-Co alloys (7-20 wt.% Co).

Polarization Performance in 1 M H 2 SO 4 Versus Co Content
Tables 1 and 2 show that neither the corrosion potential values nor the corrosion current density values as a function of the Co content present any meaningful trends. Regarding E corr , on closer observation, one could claim that the high Co Al-alloys present nobler corrosion potentials as compared to the low Co Al-alloys (compare the E corr of Al-2 wt.% Co with the E corr of Al-20 wt.% Co in Table 1). This trend can be justified by the increased Co dissolution in the Al-matrix at the high Co contents, as aforementioned in 3.1 (the corrosion potential of electrolytic Co in 1 M H 2 SO 4 has been measured by potentiodynamic polarization as (−305 ± 11) mV vs. Ag/AgCl, whereas the corrosion potential of CP-Al in 1 M H 2 SO 4 has been measured as (−463 ± 8) mV vs. Ag/AgCl.) Anyway, this ennoblement is not conclusive, as the variations in the mean E corr values are small and often within experimental error.
Regarding i corr , the insignificant variations (within experimental error) in the i corr values of the various Al-Co compositions can be justified by the following competing contributions: As the Co content increases, a decrease in i corr would be expected due to the decrease in the number of interfaces per unit area (coarser microstructure) and the increase in the surface area of the noble CMA/intermetallic compound (Al 9 Co 2 ). A 9 Co 2 is a CMA with a structural complexity between B2-AlCo and QC-Al 13 Co 4 . At its surface, A 9 Co 2 presents a pseudogap in the electron density of the states across the Fermi energy [57], which is typical for CMAs, and it is considered responsible for their low surface energy, and, consequently, their high corrosion resistance. On the other hand, as the Co content increases, an increase in i corr would be expected due to the decrease in the anode (Al)-to-cathode (Al 9 Co 2 ) ratio and the increase in the size of the cathodic Al 9 Co 2 particles.
However, a more consistent trend is observed in the current density values of pseudopassive stage 2: Figures 2, 4 and 5, as well as Table 1, show that, generally, as the Co content increases, the passive current (i p1 ) decreases (especially, as far as the 0-10 wt.% Co compositions are concerned). Although the variations are within experimental error (only in the range 10-20 wt.% Co), the decreasing trend with Co increasing is consistent. In addition, the i p1 values of the low-Co alloys appear clearly higher than the i p1 values of the high-Co alloys, classifying the studied compositions in two groups, in compatibility with the observations in 3.2.1: Al-(0-5) wt.% Co and Al- (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) wt.% Co. The decreasing trend of stage 2-current values with Co content can be associated with the increase in the Co dissolved in the Al matrix, causing an increasing participation of Co in the alumina surface film. Indeed, minor amounts of Co have been detected in the surface film of the (Al) phase (Figure 7, spectrum 53, Figure 8, spectrum 6). Here, it should be noted that in order to ensure the validity of the point EDX qualitative analysis, all measurements were performed at distances greater than 2-3 µm from the interfacial boundaries, abiding by the principle that the spatial resolution of EDX is 1-5 µm 3 for low atomic number elements and 0.2-1 µm 3 for high atomic number elements in the voltage range of 15-25 kV [58]. It has been stated that the presence of oxidized species of TMs in the alumina-based surface layer improve the corrosion resistance of Al-TM alloys, aluminides, and Al-based CMAs in chloride solutions and sulfuric acid solutions by blocking the aggressive-ion entry sites in the passive layers [59][60][61] or stabilizing the passive layers [62]. Recently, Esquivel and Gupta [63] summarized the proposed mechanisms for the contribution of TMs in the passivation ability of Al-TM alloys prepared by non-equilibrium processing techniques. Here, it should be noted that the similarity in the i p1 values of the high-Co alloys (10-20 wt.% Co), may be justified by the following reasoning: The increase in i p1 (stage 2) due to the decrease in the passive surface area, i.e., Al matrix, can to some extent counterbalance the decrease in i p1 due to the increase in the dissolved (in the matrix) Co content.
On the other hand, Figures 2 and 4 and Table 1 show similar values of the current in passive-like stage 4 (i p2 , 7-20 wt.% Co) and the current in the upper part of passive-like stage 2 (2-5 wt.% Co), regardless of the Co content. Moreover, despite the comparatively high Co content in the Al-matrix, Al-20 wt.% Co presents only slightly higher current densities than Al-5 wt.% Co during chronoamperometry at 600 mV vs. Ag/AgCl ( Figure 5). The most likely reason for these discrepancies is that surface films after polarization at 600 mV vs. Ag/AgCl (passive-like stage 4) are heavily hydrated, thus highly conductive [35]. As such, their conductivity is rather governed by the extent of the adsorbed water molecules than the compositional differences of the anhydrous parts of the surface complex compounds.
Another data group with a consistent trend is that of E cp1 − E corr (Table 1). A general trend of reduction with Co content along with small STDEV values are observed. This decreasing trend can be attributed to the increasing instability of the Al-matrix as the Co content increases due to the increasing amount of Co dissolved in Al (supersaturated state) and, consequently, the increasing ease of Al-matrix oxidation/passivation. A moderately higher resistance to localized corrosion is suggested for the Al- (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) wt.% Co as compared to the Al- (2)(3)(4)(5) wt.% Co, on the basis of positive hysteresis through the whole anodic scan (25% of tested specimens against 0% of tested specimens), cathodic looping (75% of tested specimens against 33% of tested specimens), and positive hysteresis turning to negative a couple decades of mV lower than E cp1 (0% of tested specimens against 25% of tested specimens). This higher resistance to localized corrosion can be attributed to the mediating effect that the increased Co dissolution at the high Co contents has on the galvanic coupling between Al matrix and Al 9 Co 2 , at the Al/Al 9 Co 2 interface.
The relatively low resistance to localized corrosion of the Al-(2-5) wt.% Co alloys in comparison with the Al- (7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) wt.% Co alloys could also be correlated with the greater occurrence of interfacial defects in the low-Co compositions. A close observation of Figure 3a-d reveals some microcracks and inhomogeneities on the surface structure. These defects are located at the Al/Al 9 Co 2 interphase. Their directional alignment in the case of Al-5 wt.% Co, where they are more abundant, suggests their formation along the boundaries of the long and coarse bright stripes of Al 9 Co 2 (Figure 1b). According to Kharitonov et al. [64], the Volta potential difference between the intermetallic particles and the alloy matrix along with the interphase boundary zone that constitutes a potential gradient zone (similar to the sensitization zone) are responsible for localized microgalvanic corrosion, as well as accumulation of strain and microstructural mismatch in a multi-phase alloy. The accumulation of strain along these long, coarse Al 9 Co 2 stripes is high even in the as-solidified state, as the stripes have grown along screw dislocations [46,47]; the density of screw dislocations is higher at higher cooling rates, namely at lower Al 9 Co 2 contents. In the case of the low Al 9 Co 2 compositions (Al- (2)(3)(4)(5) wt.% Co), not only is the Volta potential difference between Al and Al 9 Co 2 relatively high owing to the reduced dissolution of Co into the Al matrix, but the long, coarse Al 9 Co 2 stripes constitute large cathodic zones leading to locally high cathode-to-anode ratios. The above considerations are compatible with Zhang et al. [65], who observed localized dissolution in parallel to anodic Al-oxide formation in Al-alloys containing cathodic intermetallic particles and anodic intermetallic particles.
The high resistance of the Al-Co alloys to 1 M H 2 SO 4 also becomes evident when comparing their voltammograms with the voltammograms of market-leading Al-alloys. Figure 2c and Table 2 show that the Al-Co alloys present better performance than Al2024-T3 and Al7075-T6 in terms of lower corrosion current density values and lower passive current densities. It should be noted that the lower passive current densities of the Al-Co alloys with respect to those of CP-Al, Al2024-T3, and Al7075-T6, imply more stable passive films, a promising characteristic regarding the effectiveness of a potential anodization process. Figure 9 illustrates representative Raman spectra obtained from various spots of the surface of Al-15 wt.% Co polarized at potentials in passive-like stage 2 (0 mV vs. Ag/AgCl), stage 2-to-stage 4 transition stage 3 (280 mV vs. Ag/AgCl), and passive-like stage 4 (600 mV vs. Ag/AgCl). All Raman spectra present bands (clearly to hardly discerned) characterizing alumina hydrates, Co-oxide-based structures, and sulfate/bisulfate ions. The differentiation between the various forms of hydrous alumina, as well as the various products of cobalt oxidation, is a difficult task, due to the overlapping of bands, the significant presence of amorphous phases/dispersed species, and the metallic character of the intermetallic compounds as substrates to thin films. Nevertheless, an interpretation of the main features of the spectra is further attempted. It should be noted that the assignment of low-intensity peaks abundantly found in the spectra of Figure 9 can be speculative, since there is a risk of a baseline artefact to be misinterpreted as an active Raman band. For this reason, only the most prominent features of the received spectra are herein discussed. Furthermore, Figure 10 includes the baselines of the spectra in magnification (×8 along the y-axis) to ensure the correct assignments of low-intensity peaks. of the spectra in magnification (×8 along the y-axis) to ensure the correct assignments of low-intensity peaks.

Nature of Surface Films in 1 M H 2 SO 4
The spectra received from the surface after chronoamperometry at 0 mV vs. Ag/AgCl are classified into two groups: Type "Al-0 mV" obtained from spots on the Al-matrix surface and Type "Al9Co2-0 mV" obtained from the surface of the Al9Co2 phase. Although the humpy morphologies of the base lines, exhibited in Figures 9 and 10, indicate amorphicity, intensive or weak Raman bands characteristic of vibrations in crystalline phases stand out.  Corros. Mater. Degrad. 2020, 1 FOR PEER REVIEW 17 Spectrum "Al-0 mV": A hump extending from ~565 cm −1 to ~705 cm −1 forms the baseline of two broad bands peaking at 652 cm −1 and 604 cm −1 (Figure 9). The broad band extending from ~565 cm −1 to ~630 cm −1 peaks at 604 cm −1 , a wavenumber corresponding to the F2g mode of CoOOH vibrations [66]. The broad band extending from ~630 to ~705 cm −1 peaks at 652 cm −1 , which is the highest peak in the spectrum. It is assigned to the γ(ΟΗ) vibrations (B1g symmetry) of diaspore {α-AlOOH} [67], whereas it is accompanied by a "shoulder" peak at 662 cm −1 , ascribed to the Ag symmetry/γ (ΟΗ) vibrations of diaspore [68]. High and sharp peaks at 986 cm −1 and 614 cm −1 are assigned to the v1 (SO4 2− ) symmetric stretching vibration mode and the ν4 (SO4 2− ) bending and antisymmetric stretching vibration mode, respectively [69][70][71]. In Figure 9, another hump is seen to extend from ~435 cm −1 to ~490 cm −1 . Figure 10a shows minor peaks standing out from this hump at 456 cm −1 (v2 (SO4 2− ) symmetric bending vibrations [70], 467 cm −1 (Al-O vibrations in the diaspore lattice [68]) and 471 cm −1 (Eg mode of CoO [66,72]). Several minor peaks have also been identified throughout the spectrum; they are caused by vibrations in the lattices of alumina hydrates (diaspore, boehmite (γ-AlOOH), bayerite {α-Al(OH)3}, gibbsite {γ-Al(OH)3} [68]), and Co-oxide-based compounds (CoO, Co3O4, CoOOH [66,72]), the exact assignments of which, are shown in Figure 10a.  Figure 10. The baselines of the spectra of Figure 9 magnified along the y-axis. Spectra from the surface of the (a,e) Al-matrix after chronoamperometry at 0 mV and 600 mV vs. Ag/AgCl, respectively; (b-d) Al 9 Co 2 phase after chronoamperometry at 0 mV, 280 mV and 600 mV vs. Ag/AgCl, respectively. The spectra received from the surface after chronoamperometry at 0 mV vs. Ag/AgCl are classified into two groups: Type "Al-0 mV" obtained from spots on the Al-matrix surface and Type "Al 9 Co 2 -0 mV" obtained from the surface of the Al 9 Co 2 phase. Although the humpy morphologies of the base lines, exhibited in Figures 9 and 10, indicate amorphicity, intensive or weak Raman bands characteristic of vibrations in crystalline phases stand out.
Spectrum "Al-0 mV": A hump extending from~565 cm −1 to~705 cm −1 forms the baseline of two broad bands peaking at 652 cm −1 and 604 cm −1 (Figure 9). The broad band extending from~565 cm −1 tõ 630 cm −1 peaks at 604 cm −1 , a wavenumber corresponding to the F 2g mode of CoOOH vibrations [66]. The broad band extending from~630 to~705 cm −1 peaks at 652 cm −1 , which is the highest peak in the spectrum. It is assigned to the γ(OH) vibrations (B 1g symmetry) of diaspore {α-AlOOH} [67], whereas it is accompanied by a "shoulder" peak at 662 cm −1 , ascribed to the A g symmetry/γ (OH) vibrations of diaspore [68]. High and sharp peaks at 986 cm −1 and 614 cm −1 are assigned to the v 1 (SO 4 2− ) symmetric stretching vibration mode and the ν 4 (SO 4 2− ) bending and antisymmetric stretching vibration mode, respectively [69][70][71]. In Figure 9, another hump is seen to extend from~435 cm −1 to~490 cm −1 . Figure 10a shows minor peaks standing out from this hump at 456 cm −1 (v 2 (SO 4 2− ) symmetric bending vibrations [70], 467 cm −1 (Al-O vibrations in the diaspore lattice [68]) and 471 cm −1 (E g mode of CoO [66,72] [66,72]), the exact assignments of which, are shown in Figure 10a. Spectrum "Al 9 Co 2 -0 mV": Peaks are very weak hardly protruding from their background. Few weak peaks (noted in Figure 10b) could be due to vibrations in the alumina hydrate lattices, respectively [68]; however, their low intensities do not allow for reliable conclusions. The only distinguishable pattern is a broad band that extends from~926 cm −1 to~994 cm −1 and peaks at 979 cm −1 (wavenumber ascribed to the v 1 (SO 4 2− ) vibrations). The above observations enhance the indications reported in 3.2.2 that stage 2 in the Al-Co alloys of high Co content (7-20 wt.% Co) is owing to the passivation of the Al-matrix (without the participation of Al 9 Co 2 in the passivation process). Spectrum "Al 9 Co 2 -280 mV": A representative spectrum received from the surface of primary Al 9 Co 2 stripes after chronoamperometry at 280 mV vs. Ag/AgCl is included in Figure 9. Spectrum "Al 9 Co 2 -280 mV" is characterized by several very broad bands of low intensity and jagged morphology, the latter clearly seen in Figure 10c. More analytically, a broad hump extends from~520 to~700 cm −1 , which is centered around 602 cm −1 (the characteristic wavenumber of the F 2g mode of CoOOH vibrations). This broad band is typical of a dispersed Co-oxide phase [73]. Weak peaks are discerned to protrude from this hump; they are assigned to the Al-O bonding (<500 cm −1 ), γ(OH) translation modes (<900 cm −1 ), and δ(OH) deformation modes of alumina hydrates (>900 cm −1 ) [68]; their exact assignments are noted in Figure 10c. The high degree of amorphicity (indicated by the humpy and jagged morphology, as well as the very weak peaks of the spectrum in Figure 9) is in accordance with the evidences of the voltammograms in Figures 2 and 4, and the current density vs. time curve at 280 mV vs. Ag/AgCl in Figure 5, of a transition stage governed by oxidation processes on the Al 9 Co 2 surface. The main products of these oxidation processes (after chronoamperometry for 2 h) are dispersed Co-O species and barrier amorphous Al-O films, probably precursor phases to the phases formed during stage 4. Regarding the claim for barrier Al-oxide films, these are postulated to be "substrates" to the thin alumina hydrate layers detected. The bi-layered structure of the passive film of Al in contact with an aqueous environment, namely an amorphous Al 2 O 3 barrier layer beneath a permeable hydrous Al 2 O 3 layer, is long acknowledged [74].
Two main types of spectra have been obtained from various spots of the surface after chronoamperometry at 600 mV vs. Ag/AgCl. Spectrum type "Al 9 Co 2 -600 mV", received from the surface of an Al 9 Co 2 primary stripe, and spectrum type "Al-600 mV", received from the surface of the Al-matrix. Spectrum "Al 9 Co 2 -600 mV" contains major peaks assigned to characteristic vibration modes of the lattices of CoOOH (F 2g : 604 cm −1 ), Co 3 O 4 (F 2g : 622 cm −1 ), and diaspore (γ(OH), B 1g : 650 cm −1 ). Other accountable peaks, more clearly discerned in Figure 10d, are ascribed to lattice vibrations of alumina hydrates (diaspore A g symmetry/γ(OH) at 663 cm −1 -shoulder to the B 1g band-, bayerite/γ(OH) at 899 cm −1 , bayerite/Al-O at 434 cm −1 , gibbsite/Al-O at 410 cm −1 , gibbsite/γ(OH) at 751 cm −1 , gibbsite/δ(OH) at 895 cm −1 , boehmite/γ(OH) at 670 cm −1 ) [20,68]. Some minor peaks, more clearly discerned in Figure 10d, are considered to be due to vibrations in the lattices of CoO (A 1g symmetry at 674 cm −1 , F 2g symmetry at 507 cm −1 , E g symmetry at 470 cm −1 ), CoOOH (807 cm −1 ), and Co(OH) 2 (A 1g symmetry at 424 cm −1 ) [66,72]. Amorphicity is indicated by the hump that forms the base of the intensive bands peaking at 604 cm −1 , 622 cm −1 , and 650 cm −1 , whilst it extends from~545 to~700 cm −1 , as seen in Figure 9. The well-formed bands of high intensity, as compared to the "humpy" morphology of the spectrum at 280 mV vs. Ag/AgCl, suggest that the oxidation having occurred in transitional stage 3 has led to the formation of surface films on the Al 9 Co 2 phase during potentiodynamic polarization through stage 4. The hump that forms the base of the band peaking at 604 cm −1 enhances the postulation that dispersed Co-O species are precursors to Co(OH) 2 , CoOOH, Co 3 O 4 , and CoO. Figure 9 shows that Spectrum "Al-600 mV" includes three wide bands that give the impression of an overall humped morphology with numerous protruding small peaks forming a "jagged" shape: A band peaking at 510 cm −1 (F 2g (CoO)), a band peaking at 584 cm −1 (B 1g /γ(OH) in diaspore), 605 cm All Raman spectra in Figure 9 present sharp or wide bands of notable intensity peaking at 975-987 cm −1 . As aforementioned, these bands are assigned to the symmetric stretching vibrations of the SO 4 2− ion. A question arises whether the identified sulfate-due bands are due to vibrations in the electrolyte remaining in the pores of the alloys or they are due to bonding between surface metal and sulfate ions. The presence of residual H 2 SO 4 solution is supported by the following considerations: The variety in the half-width of the v 1 (SO 4 2− ) band, observed in Figure 9, has been attributed to the variety in the H 3 O + concentrations [69], suggesting the presence of aqueous H 2 SO 4 . Myhre et al. [69] showed that an increase in H 2 SO 4 concentration leads to a significant decrease in the intensity and an increase in the half-width of the v 1 (SO 4 2− ) band. On the other hand, the presence of sulfate salts is supported by the following considerations: The max 12 cm −1 shift observed in the v 1 (SO 4 2− ) band constitutes evidence of bonding between the adsorbed sulfate and surface atoms [70,76]. According to literature, the Raman position of the v 1 mode of a solid sulfate shifts to a high wavenumber compared to the aqueous solution [76,77]. According to Omori [78], "every atom of the SO 4 molecule in solution vibrates freely, but in a crystal, the atom vibrates under additional influences from the cations, causing variations in the wavenumber of the absorption band". For example, the v 1 (SO 4 2− ) band of aqueous H 2 SO 4 of various concentrations at room temperature is reported to center around 980 cm −1 [69], whereas the v 1 (SO 4 2− ) band of Al 2 (SO 4 ) 3 ·xH 2 O has been reported to peak at 991 cm −1 [79]. As such, for hydrated sulfates, the ν 1 symmetric stretching-due peak would shift to lower vibrational wavenumbers with increasing hydration [80]. Additionally, in solid salts, the vibrational wavenumber of each sulfate-due Raman band has been shown to decrease as the cation radius increases and, consequently, the cation's polarizing power decreases [76,77]. Even in the case of sulfate salt presence, the ν 1 mode can exhibit a linewidth increasing from the anhydrous to hydrated and then to aqueous solution, which can be due to the increase in the structure disordering from the solid to the liquid state [76]. Splitting of the SO 4 2− -due Raman bands (as in spectra Al 9 Co 2 -0 mV (ν 1 ), Al 9 Co 2 -600 mV (ν 4 ), Al-600 mV (ν 1 )) may also be related to the metal cation radius in the solid sulfate salt [77]. Hence, although the possibility for the simultaneous presence of residual sulfuric acid solution or sulfuric salt solution cannot be excluded, the wide v 1 (SO 4 2− ) Raman bands in spectra Al 9 Co 2 -0 mV, Al 9 Co 2 -280 mV, Al 9 Co 2 -600 mV, and Al-600 mV could also be due to sulfate salts, in hydrated forms. In fact, splitting of the SO 4 2− -due Raman bands favor the latter consideration. The v 1 (SO 4 2− ) band in spectrum Al-0 mV, presenting markedly higher sharpness and intensity and located at a higher wavenumber compared to the corresponding bands of the other spectra in Figure 9, could be due to the presence of a solid sulfate salt, most likely Al 2 (SO 4 ) 3 ·xH 2 O. The different intensities of the v 1 (SO 4 2− ) band are owing to the different concentrations of the SO 4 2− ions in the analyzed spots [81]. Weak peaks uniquely ascribed to HSO 4 − at (878-887) cm −1 (asymmetric stretching vibrations) are also observed [69]. The considerable shift in the frequency of this band is attributed to the variation of the concentration of HSO 4 - [69]. Two peaks at 593-594 cm −1 and 422-425 cm −1 arise from fundamental vibrations of both SO 4 2− and HSO 4 − ions [69].
It should be emphasized that the studied system is complex and multicomponent; therefore, significant overlapping of bands is an additional reason for the variations in the band characteristics (intensities, peak shifts, width, etc.) In conclusion, Raman spectra confirmed the findings reported in 3.2.1 and 3.2.2, namely: (a) The nature of the surface films is complex, consisting of mixtures of crystalline and amorphous phases. Amorphicity is associated with dispersed Co-oxide species, alumina underlayers to alumina hydrates, and intensively hydrated surface films at high passive potentials.
(b) Al 9 Co 2 has remained unoxidized at the low potentials of the passive stage (stage 2), for the low-Co compositions (2-5 wt.% Co), or during the first passive stage (stage 2) for the high-Co compositions (7-20 wt.% Co).
(c) Passive films on the Al-matrix contain Co-oxide based species due to the oxidation of Co dissolved in Al. (It is well established that Al alloyed with TMs forms Al 2 O 3 -based passivating layers in various electrolytes; these layers may incorporate oxidized species of the TM. Such systems include: CMAs (Al-Cu-Fe-Cr [82], Al-Cr-Fe [62,83] and Al supersaturated in TMs (W [59], Mo [59], and Ta [59,84]).
(d) Mixtures of alumina hydrates and Co-oxides (Co-oxide dispersed species, CoO, Co 3 O 4 , CoOOH) compose the films on Al 9 Co 2 in compatibility with works on the passivation of Fe-aluminides in H 2 SO 4 [60,85].
(e) Sulphate salts in hydrated forms may also be included in the surface films.

Some Comments on the Objectives of This Work
Regarding the primary objective of this work, which was to identify low enough Co additions that will satisfy the threefold: Ductility-corrosion resistance-cost of raw materials, this work produced the following data: The high Co alloys (7-20 wt.% Co) led to greater reductions of passive current, in comparison with the low Co alloys (2-5 wt.% Co), namely: During the first pseudopassive stage, a reduction of 3.5-5.0 times (as compared to pure Al) was attained for the 7-20 wt.% Co range, whereas a reduction of 1.4-2.0 times was attained for the 2-5 wt.% Co range. On the other hand, the differences in the electrochemical behaviors of the 2-5 wt.% Co and 7-20 wt.% Co compositions concerning the resistances to localized corrosion and uniform corrosion were not significant enough. Taking also into account that: (a) High Co alloys are more brittle than low Co alloys due to the high amounts of Co-aluminides [35], (b) addition of 7 wt.% Co to Al has led to higher localized and uniform corrosion resistances in 3.5 wt% NaCl, as compared to richer Co compositions [20], (c) Co additions raise the cost of the raw materials, and (d) low and high Co alloys have herein shown comparable resistances to localized corrosion and uniform corrosion in 1 M H 2 SO 4 , the present work tends to support that alloying Al with 7 wt.% Co could be a quite satisfactory option. However, safer conclusions will be drawn when the present study will be complemented with an investigation on the effect of Co on the wear resistance of Al, planned to be published in the immediate future.
As aforementioned in the Introduction, the main subject of this effort is to examine the electrochemical response of a wide compositional range of Al/Al 9 Co 2 alloys to aqueous H 2 SO 4 , from the viewpoint of corrosion, as H 2 SO 4 is a key substance in the chemical industry, an important agent in many industrial sectors, like mineral processing, oil refining, wastewater processing, as well as the main constituent of acid rain. For this purpose, a common commercial concentration of diluted H 2 SO 4 was selected, which is also frequently used in electrochemical studies of Al (1 M) [86]. A wide potential range was employed including cathodic polarization, anodic polarization, and reverse polarization to gain as much information as possible (in a systematic manner) on the electrochemical performance of the Al-Al 9 Co 2 alloys (a new combination of Al-intermetallic compound). A further objective was to evaluate whether off-equilibrium alloying Al with Co can improve the passivation ability of Al. As such, the present work can set the foundations for a future detailed investigation on the selection of the parameters of an anodizing treatment (e.g., electrolyte concentration, composition and temperature, current/potential values, time, pore sealing conditions, etc.), towards an optimum response to aggressive environments. Within the above framework, it should be emphasized that although the sulfuric acid concentration of 1 M has been employed for anodizing [87], the present effort should not be used to draw conclusions on the anodization procedure in practice.

•
Al-Co alloys of various Co contents in the 2-20 wt.% Co range were produced by vacuum arc melting. The alloys were composed of Al 9 Co 2 particles uniformly distributed in an Al matrix in a directional pattern. As the Co content of the alloy increased, the volume fraction and coarseness of Al 9 Co 2 increased. • Both potentiodynamic polarization and chronoamperometry experiments showed that the anodic polarization of Al-Co in 1 M H 2 SO 4 at 25 • C proceeded via the stages: (1) Active dissolution of the Al-matrix by ionic conduction across the imperfections of the preexistent surface film; (2) pseudo-passivation of the Al-matrix by film formation at the metal/oxide interface and film dissolution at the film/electrolyte interface; (3) oxidation of the surface of the Al 9 Co 2 particles; and (4) extension of pseudo-passivation to the surface of the Al 9 Co 2 particles. Stage 3 was distinct only in the case of the high-Co alloys (7-20 wt.% Co). • Different passivation behaviors were observed for the 2-5 wt.% Co and 7-20 wt.% Co alloys in 1 M H 2 SO 4 . The low-Co alloys presented one stage passivation governed by the passivation of the (Al) matrix; this stage included passivation of Al 9 Co 2 at high overpotentials. The high-Co alloys presented passivation of two distinct stages. The first stage was governed by the passivation of the (Al) matrix. The second stage was governed by the passivation of Al 9 Co 2 . A transitional active stage separated the two passive stages; it corresponded to the oxidation of Al 9 Co 2 .

•
Alloying Al with Co did not decrease the rate of uniform corrosion. Nevertheless, increasing the Co content led to a decrease in the passivation current density (during the first stage) in 1 M H 2 SO 4 , due to the increasing participation of Co in the surface layer of the (Al) matrix. Even low percentages of Co led to accountable decreases in the passive current density of Al (Al-2 wt.% Co: 1.4 times, Al-5 wt.% Co: 2 times, Al-7 wt.% Co: 3.5 times). • All Al-Co compositions displayed greater passivation ability in 1 M H 2 SO 4 than commercially pure Al, Al7075-T6, and Al2024-T3, in terms of passive current density and initiation of passivation at lower potentials with reference to the corrosion potential. • All Al-Co compositions showed high resistance to localized corrosion. High Co alloys (Co ≥ 7 wt. %) did not show any susceptibility to localized corrosion.
• Taking into account that: (a) High Co alloys are more brittle than low Co alloys due to the high amounts of Co-aluminides, (b) addition of 7 wt.% Co to Al has previously led to higher localized and uniform corrosion resistances in 3.5 wt% NaCl, as compared to richer Co compositions, (c) Co additions raise the cost of raw materials, and (d) low and high Co containing alloys have herein shown comparable resistances to localized corrosion and uniform corrosion in 1 M H 2 SO 4 , the present work tends to support that alloying Al with 7 wt.% Co could be a quite suitable option, towards the threefold: Ductility-corrosion resistance-cost of raw materials. A planned publication on the effect of the Co content on the tribological performance of Al will further elucidate this matter. Funding: This research received no external funding.