Self-repairing oxides to protect zinc: Review, discussion and prospects

Self-repairing oxides to protect zinc: Review, discussion and prospects

Corrosion Science 69 (2013) 11–22 Contents lists available at SciVerse ScienceDirect Corrosion Science journal homepage: www.elsevier.com/locate/cor...

1MB Sizes 0 Downloads 8 Views

Corrosion Science 69 (2013) 11–22

Contents lists available at SciVerse ScienceDirect

Corrosion Science journal homepage: www.elsevier.com/locate/corsci

Review

Self-repairing oxides to protect zinc: Review, discussion and prospects S. Thomas a,b,⇑, N. Birbilis b, M.S. Venkatraman a, I.S. Cole a a b

CSIRO Material Science and Engineering, Clayton, 3169 Victoria, Australia ARC Centre of Excellence for Design in Light Metals, Department of Materials Engineering, Monash University, Clayton, 3800 Victoria, Australia

a r t i c l e

i n f o

Article history: Received 9 March 2012 Accepted 4 January 2013 Available online 16 January 2013 Keywords: C. Corrosion A. Zinc C. Passivation A. Oxide C. Self-repair

a b s t r a c t Zinc is protected in the short term by chromate conversion coatings. The Cr (VI) based surface layer has the unique ability to self-repair which is attributable to the aqueous properties of its constituent ions. The thermodynamic feasibility of achieving chromium-like protection for the short term using other candidates is reviewed. Long-term protection of zinc is naturally affected by its own corrosion products (patina), which form a multilayered structure. The role of this patina in protecting the underlying metal is also reviewed, and processes within the patina, including the interaction between its various layers, are elaborated and discussed. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction A chromium-based coating is conventionally used to prevent the white staining (corrosion) of zinc/galvanised steel prior to its industrial application [1]. This coating is particularly effective, since the chromium ions within it can exist in two oxidation states Cr (III) and Cr (VI), the Cr (III) state provides a barrier effect, and the Cr (VI) state a ‘self-repairing’ effect [2]. The thick surface layers formed during the application of Cr (VI) based conversion coatings, leach mobile Cr (VI) aqueous species, which undergo reduction at the coating defects to form the passivating Cr (III) oxide (Cr2O3) [3]. However, exposure to hexavalent chromium (Cr (VI)), a known carcinogen [4–7] is now prohibited due to its toxicity, mandating more viable ‘green’ alternatives. Although a number of different corrosion inhibition strategies have been applied to zinc, including the use of molybdates [8,9], rare earths [10–13], phosphates [14– 16] and organic systems [17–21], the ability to self-repair has not been realised using non-chromium-based inorganic inhibitor systems. In the present work, the pH-potential behaviour of Cr (VI)–Cr (III) in aqueous media has been reviewed, with the aim of understanding the properties which would need to be replicated, using other ions to develop an alternative self-repairing system to protect zinc. The protective capacity of the surface coatings (either chromium, phosphate or organic materials) applied onto zinc decreases with time, and eventually, after up to 10 years, such coatings disappear from the surface [1]. The long-term protection of zinc, after ⇑ Corresponding author at: CSIRO Material Science and Engineering, Clayton, 3169 Victoria, Australia. Tel.: +61 420424024. E-mail address: [email protected] (S. Thomas). 0010-938X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.corsci.2013.01.011

the disappearance of such surface coatings, is then achieved by zinc’s own corrosion products (patina) that are formed during its atmospheric corrosion. It was observed that 37% of the corroded zinc was retained within this patina, in the form of different zinc compounds after 10 years of exposure to a marine atmosphere [1]. The barrier properties of the patina also improved with time as the polarisation resistance (Rp) increased up to 80 times compared to that of an unexposed reference surface after 10 years [1]. Concomitantly, these oxides were noted to have decreased the corrosion rate of zinc over a 10 year period (around 0.46 lm year1), when compared to that of pure zinc exposed for 1 year (around 0.67 lm year1), at the same exposure site [1,22]. The morphology and composition of the zinc compounds which constitute the patina usually vary depending on the nature of the atmospheric environment (marine, urban, rural and industrial environments) [23]. Reactive metals such as Fe, Cr, Ni, Al, Ti, Pb and Zn generally undergo direct oxidation on exposure to air, forming thin oxides on their surfaces, which effectively self-protect these metals in environments where these oxides are insoluble [24]. The Pilling–Bedworth ratio [25] of ZnO is 1.58 [26], implying that the oxide formed on zinc during its direct reaction with oxygen in air could serve to passivate the surface. However, during the atmospheric corrosion of zinc, direct exposure of the metal to either a thin film of moisture or to aerosols disrupts this initially formed oxide [27] and results in the gradual development of the zinc patina. This means that we need to focus on improving the barrier properties and ‘self-repair’ capabilities of this patina. The literature on zinc corrosion can be classified into three main bodies depending on the mode of exposure of the metal to the electrolyte:

12

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

i. Thin film exposure wherein zinc is exposed to a thin film of moisture [28–32]. ii. Droplet exposure corresponding to zinc exposed to aerosols of various sizes and compositions [33–36]. iii. Bulk exposure pertaining to experiments carried out in bulk electrochemical cells [37,38]. The electrode kinetics, being a function of the surface properties of the metal, are independent of the mode of exposure. However, the rate of oxygen diffusion to the metal surface and hence the oxygen concentrations within these systems will vary significantly depending on the mode of electrolyte exposure. In the case of thin film exposure, since the electrolyte layer is thin, oxygen diffusion to the metal surface will be faster than in the case of bulk exposure to electrolyte [39]; in droplet exposure, the oxygen distribution within the system is non-uniform, since the electrolyte forms a hemispherical layer upon the metal surface. A lower volume of electrolyte (corresponding to thin film or droplet exposure) could also result in increased local concentrations of dissolved metal ions, as opposed to bulk solutions, which could favour precipitation of corrosion products. In the current work, all three bodies of literature have been collectively reviewed to gain insights into the corrosion processes occurring within the zinc patina. In bulk solutions, a well defined region of zinc passivity exists only under alkaline conditions [37,38]. The zinc patinas formed during field exposures has a different composition in comparison with the passive oxides developed in laboratory experiments, and the role played by this patina in protecting the metal is still not clear. The patina is porous [33], providing pathways for the corrosive environment, and hence requiring repair strategies to ensure long-term protection of zinc. Repair of porous oxide films in bulk solutions was studied as early as 1949 by Hoar for iron [40], who reported that oxide repair occurs when certain anions (OH,  CO2 3 , NO2 ) are in sufficient supply (within pores), to seal the pores before local acidity attains a critical value. However, such insights into repairing oxides have not been extended to develop strategies to repair the porous zinc patina. The composition and morphology of the zinc-based patina has been widely studied [33,41–44], along with its chemical stability and barrier properties [45–47], but their impact on chemical/electrochemical processes within the patina itself is an aspect which has not been clarified. The present work reviews existing literature to identify strategies for building self-repairing oxides on zinc in the future. The review looks to narrow down the problem and find some key aspects which will require research to make the development of selfrepairing oxides possible. The current work includes:

1b) [48–50]. Below pH 8, Zn2+ and ZnOH+ are the most stable phases predicted to form by thermodynamics, whereas above pH 2 12.5, zinc hydroxycomplexes such as ZnðOHÞ are 3 and ZnðOHÞ4 + more stable. Metal hydrolysis generating Zn(OH) is an important solution reaction, in the near neutral pH range, which undermines zinc passivity [38] and promotes pit propagation [51]. Thus, as a basic strategy to improve the barrier properties of a surface layer, its pH stability needs to be extended/enhanced using inhibitors. The thermodynamic diagrams do not consider the influence of kinetics within the system, nor the effect of different anions present when zinc is exposed to atmospheric conditions. However, they serve as a basic guide to understand the corrosion and passivation of zinc in a given pH range. In the presence of anions like Cl, 2 SO2 4 , and CO3 , basic zinc salts (BZS) like zinc hydroxychloride, zinc hydroxysulphate and zinc hydroxycarbonate are formed and could potentially protect zinc [34]. Mixed oxides like gordiate (sodium zinc chlorohydroxysulphate), which have different pH stability in comparision to other BZS, may also form, but their role in protecting zinc is not completely understood. As zinc’s own oxides are not effective in protecting it across a wide pH range, short-term protection systems (like chromate conversion coatings) are conventionally used to protect zinc. The next section discusses some of the key aqueous properties of constituent ions of the ‘self-repairing’ chromium system, aiming to identify a broad basis to select alternative candidates which could mimic chromium’s beneficial functionalities. 2.1. Towards replicating chromium-based protection Chromate based conversion coating generates a hydrophobic and irreversibly adsorbed chromium (III) oxide (Cr2O3) layer on metals such as aluminium [52]. This oxide has a wide pH stability in the pH range 4–14 (Figs. 2a and 2b) and thus could serve as a stable surface barrier on zinc. The solubility of Cr3+ ions across the entire pH range is shown in Fig. 2b. This indicates that above pH 4, chromium oxides have low solubility, close to 108 mol L1 in the pH range 6–10, in comparison with that of the zinc oxides (Fig. 1b). This implies that the chromium oxides could serve as a stable insoluble barrier on the surface. Conversely, the Cr (VI) oxyanions adsorbed onto the Cr (III) oxide are highly soluble and mobile [52,53]. This makes them readily leach and migrate to cathodically active sites, where they are reduced to form passivating chromium (III) oxides. The Pourbaix diagram of chromium shows that the line for the redox reaction

(i) A review of the pH-potential behaviour of chromium in aqueous media, to identify some key properties which could serve as a rationale to design non-chromium-based protection systems. We focus on building self-repairing oxides for the short-term protection of zinc and also suggest criteria for selecting inhibitor systems to protect zinc. (ii) Corrosion related processes taking place through the zinc patina to identify key problems which, if understood better, could be used to build strategies to enhance self-repair of the zinc patina (towards long-term protection). (iii) Some prospects to enhance self-repair of the zinc patina, based on the literature survey. 2. Self-repairing systems for short-term protection of zinc Thermodynamic diagrams depicting the zinc–water system 2 (without consideration of anions such as Cl, SO2 4 and CO3 ) show that the zinc oxides are stable in the pH range of 8–12 (Figs. 1a and

Fig. 1a. Fractions of different Zn2+ species formed as a function of pH for a zinc– water system as predicted by thermodynamic calculations using the software Hydra-Medusa [48]. The Zn2+ ion concentration is 0.0001 M in the calculation.

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

Fig. 1b. Solubility of Zn2+ as a function of pH as predicted by thermodynamic calculations using the software Hydra-Medusa [48]. The Zn2+ ion concentration is 0.0001 M in the calculation.

13

(Cr (VI) to Cr (III)) is located in between the oxygen reduction reaction (ORR) line and the hydrogen evolution reaction (HER) line (region of water stability) [50]. This implies that in the presence of chromates, there is an additional cathodic reaction (Cr (VI) to Cr (III)) in the system, taking place along with the HER and ORR. This particular aspect was verified on different electrodes namely copper, glassy carbon and platinum [54]. In the presence of chromates in the solution, cyclic voltammograms of the electrodes showed a peak between +0.25 and 0.6 V (vs. Ag/AgCl), indicating the reduction of soluble CrO2 4 ions to insoluble Cr2O3. After the culmination of this reduction reaction, the cathodic currents decreased, indicating that the Cr2O3 monolayer formed inhibits cathodic reactions. This clarified that Cr (VI) is a ‘site-directed’ inhibitor, which migrates to active cathodic sites, where it gets reduced to an inhibiting Cr (III) oxide layer [54]. The Cr (III) oxide and the Cr (VI) ions thus act together to form a self-repairing barrier on the metal. In order to mimic the chromium-based system, any other alternative inhibiting system must (i) provide a supply of aqueous ions, which can self-reduce to passivating oxides at defects of the surface layer (like the Cr (VI) to Cr (III) reduction); (ii) form surface layers with wide pH stability (like Cr2O3) and (iii) readily leach highly soluble and mobile ions (like Cr (VI) oxyanions) into the solution. Some prospects towards replacing chromium, specifically targeting the above three functionalities, are reviewed in the following section. 2.2. Chromium-free protection: review and prospects

Fig. 2a. Fractions of different Cr3+ based species formed as a function of pH for a chromium–water system. The Cr3+ ion concentration is 0.0001 M in the calculation. The thermodynamic calculations were performed using the software HydraMedusa [48].

Fig. 2b. Solubility of Cr3+ ions as a function of pH for a chromium–water aqueous system. The Cr3+ ion concentration is 0.0001 M in the calculation. The thermodynamic calculations were performed using the software Hydra-Medusa [48].

2.2.1. Reduction to passivating oxides McCafferty [55] calculated the free energy change, for the interaction of 19 different oxyanions with iron and aluminium. These oxyanions were those of the transition metals in the fourth through sixth periods of the periodic table. The standard free energy change DG° (per mole of oxyanion) for the reaction between the metal and the oxyanion was calculated, with the reduction in the oxyanions (example CrO2 4 ) to their oxides (Cr2O3) taken as the cathodic half-cell reaction. For iron, the oxyanions, MnO 4, 2 2 2 NiO4 , RuO 4 , RuO4 and RhO4 were found to produce more nega2 tive values of DG° (per mole of oxyanion) than CrO2 4 . OsO4 and 2 2 IrO4 , produced values of DG° similar to CrO4 , whereas oxyanions   3   2 2 such as MoO2 4 , TcO4 , NbO3 , VO4 , TaO3 , ReO4 , ReO4 and WO4 were found to produce less negative values of DG° than CrO2 4 . Some of these oxyanions could thus serve to mimic the ‘self-repairing’ effect of the chromates. However, they need to be tested in the manner prescribed by Clark and McCreery [54], to identify whether they indeed transform to passivating oxides, on reduction, which could inhibit cathodic reactions on their surface. In the past, molybdate ðMoO2 4 Þ based coatings have been extensively investigated as a protective barrier on zinc [56–58]. However, in general, they have not demonstrated a capacity to ‘self-repair’ like the chromate based surface layers. This is probably due to the fact that molybdates are weaker oxidising agents than the chromates [8], and also, that the species formed from molybdate reduction tend to redissolve in solution [56]. The nature of the surface oxides, formed during molybdenum conversion coatings, varies depending on the process parameters and the predominant states of Mo in these surface oxides are (IV), (V) and (VI) [8,57,58]. The barrier properties of the molybdate-based layers have been improved upon by the synergistic addition of phosphates [58]. However, in this case, the MoO2 oxyanions were 4 found to only inhibit attack by chlorides and not undergo cathodic reduction to form passivating oxides like the chromates [58]. 2.2.2. Developing surface layers with wide pH stability A chromium-free surface layer must be stable across a wide range of pH to be resistant against dissolution like the Cr2O3 layer.

14

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

Usually, inhibitors previously tested on zinc do not form surface layers which have wide pH stability as found for Cr2O3. Such inhibitors can provide effective protection only within specific pH domains. For example, rare earth based inhibitors have been used as cationic inhibitors to protect zinc [10–13]. Rare earths are also used in zinc alloys and surface coatings [59,60]. These inhibitor systems have been postulated to form surface layers, which inhibit the ORR on zinc. Aramaki [11] proposed that addition of certain rare earth ions (Ce3+, La3+) in solution formed a thick protective film, which resulted in the shift of the cathodic reaction from the ORR to an alternate, proton (H+) consuming, reaction. However, those studies were conducted in neutral solutions, and since the cerium oxides/hydroxides are thermodynamically predicted to dissolve below pH 10 (Fig. 3) [61], the growth and protection mechanisms invoked by these oxides on zinc are still not clear. Cerium 2+ species, such as CeðOHÞþ are also predicted to form 2 and Ce(OH) when the pH is below 8 (Fig. 3) [61,62], these species could also influence the corrosion kinetics of zinc. Arenas and de Damborenea [63] reported that the local alkalinity, generated by the ORR, was critical to attain the required pH to precipitate cerium oxides/ hydroxides. Therefore, rare earth ions precipitate oxides under alkaline conditions, especially near to the ORR sites (where the pH is greater than 10). However, these oxides do not have wide pH stability and thus cannot protect zinc exposed to acidic or neutral solutions. Similarly, zinc–magnesium coatings are known to be more corrosion resistant than their pure zinc counterparts [64,65]. The presence of Mg2+ ions in the electrolyte also decreased zinc corrosion [66]. This better performance in the presence of Mg2+ ions is attributable to the formation of Mg(OH)2 in alkaline regions. Mg(OH)2 possibly forms a layer which restricts oxygen diffusion and also buffers the local pH (close to 10.2), making the zinc oxides less prone to alkaline dissolution. Those works also reported that the presence of Mg forms MgCO3, which removes carbonates from the corrosive medium, and correspondingly limits the transformation of zinc hydroxychloride (simonkolleite) to zinc hydroxycarbonate (hydrozincite). On the premise that zinc hydroxychloride is a more protective corrosion product than zinc hydroxycarbonate, those works concluded that the presence of Mg serves to lower the corrosion of zinc. The pH stability of a surface layer is thus an important criterion in designing surface layers on zinc. Cationic inhibitors, such as cerium and magnesium, form inhibiting surface oxides under alkaline

conditions, which could provide inhibition against the ORR, and also mitigate the alkaline corrosion of zinc. However, unlike Cr2O3, they cannot serve as effective barrier on a metal across a wide range of pH. Anionic inhibitors supply soluble anions which react with zinc ions, released during metal dissolution, to precipitate protective zinc compounds. Phosphates have been found to inhibit the anodic processes on zinc by the formation of Zn3(PO4)2, whereas silicates were found to inhibit both the anodic and cathodic processes; the anodic inhibition being due to formation of ZnSi2O5 [67,68]. The solubility of zinc in a phosphate solution across the pH range is shown in Figs. 4a and 4b. This indicates that formation of zinc phosphates could extend the pH stability of the surface layer down to pH 6, due to the formation of Zn3(PO4)2. However, the solubilities of the phosphates are still 1000 times higher than that of the Cr (III) oxide (inferred from Fig. 2b), implying that phosphate layers tend to dissolve and are less stable compared to chromium oxides. A solely cationic inhibitor or anionic inhibitor will not form an effective surface barrier on zinc. This is because of the wide range of pH (ranging from 3 to 8) manifested within zinc corrosion cells [35]. Synergistic inhibitors, which include compounds of form XY, with ions of X (cations) forming inhibiting layers at cathodic sites, and ions of Y (anions) doing the same at anodic sites, could serve to further decrease the corrosion of zinc. The pH stability of the surface layers produced by X and Y together must be able to mimic the pH stability of the Cr (III) oxide. Although synergistic inhibitors have been extensively researched upon zinc [69–73], the pH stability of such synergistically inhibiting surface layers has not been previously studied. Aramaki [69–73] reported improvement of self-healing capabilities of a inhibiting film on using a number of different synergistic inhibitors, usually cerium oxides, modified with sodium phosphate [69], sodium silicate [70], calcium and magnesium nitrate [71], zinc nitrate and sodium phosphate [72], and sodium hexadecanoate [73]. A similar sort of approach has been carried out upon the aluminium alloy AA2024, using rare earths with dibutyl and diphenyl phosphates [74]. The degree of inhibition achieved in that case was close to that offered by chromium-based layers. However, the surface layers formed using these inhibitors have been tested mostly for anodic and cathodic inhibition, and the overall pH stability of these synergistic surface layers is still not known. As an example, the effect on the pH stability of a surface layer comprising both Ce3+ and PO3 ions can 4 be appreciated by considering Figs. 3, 4a and 4b. The presence of

Fig. 3. Fractions of different Ce3+ species formed as a function of pH for a cerium– water system. The Ce3+ ion concentration is 0.0001 M in the calculation. The thermodynamic calculations were performed using the software Hydra-Medusa [48].

Fig. 4a. Fractions of different Zn2+ species, formed in the presence of PO3 4 ions, as a function of pH. The Zn2+ and PO3 concentrations are both 0.0001 M in the 4 calculation. The thermodynamic calculations were performed using the software Hydra-Medusa [48].

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

2+ Fig. 4b. Solubilities of Zn2+ and PO3 and 4 ions together as a function of pH. The Zn PO3 concentrations are both 0.0001 M in the calculation. The thermodynamic 4 calculations were performed using the software Hydra-Medusa [48].

PO3 ions extends the oxide stability down to pH 6 (Fig. 4b), 4 whereas the addition of Ce3+ ions extend the stability of the oxides to about pH 13.5 (Ce(OH)3 – Fig. 3).Thus, these two ions together extend the stability of zinc surface layers from pH 8–12 (without any inhibitors) to pH 6–13.5 by their synergistic combination. The pH stability of the surface layer could also be improved using other elements, such as niobium, tantalum, zirconium or titanium. The Pourbaix diagrams of these elements show that their oxides have wide pH stability [50]. Niobium and tantalum are in theory passive throughout their E–pH diagram; the passive regimes of zirconium and titanium extend from pH 4 to 13. Kendig and Buchheit reviewed the use of these elements as coatings for aluminium [52]. 2.2.3. Leachability of inhibiting ions The surface layers formed on metals intrinsically contain defects, pores or even holidays. These defects serve as active sites, forming local electrochemical cells upon exposure to aqueous media. Quick migration of the inhibiting ions, from the surface layer towards these active sites, is thus necessary to restrict localised attack of the metal. For chromate based surface layers, the highly soluble CrO2 4 ions leach from within the surface layer and migrate to the active sites (where they get reduced) to achieve this functionality. Zhao et al. [75] used an artificial scratch cell to study the leachability of chromate ions from the coating to the bare metal surface. Those studies demonstrated that the chromates indeed leach from within the coating to active sites on the bare metal. This consequently results in a decrease in the corrosion potential (Ecorr) and increase in Rp of the bare metal. Similar experiments were carried out by Guan and Buchheit [76] to test vanadate conversion coating layers upon aluminium. However, in the context of testing inhibiting systems on zinc, artificial scratch experiments have not been carried out, and thus, information on the leachability and transport of inhibiting ions from surface layers formed on zinc is limited. Importantly, the impact of certain anions found within 2 the zinc patina itself, namely Cl, SO2 and OH, in terms 4 , CO3 of leachability and migration may be addressed in future research. 3. Towards enhancing self-repair of the zinc patina Zinc forms a number of corrosion products, namely zinc oxide, zinc hydroxide, zinc hydroxychloride, zinc hydroxysulphate, zinc chlorohydroxysulphate [33], boyelite, zinc chlorate hydrate, sodium zinc carbonate hydrate, sodium zinc chloride hydrate and

15

gordaite during exposure to marine environments [34]. Therefore, the zinc patina formed in such conditions serves as a reservoir of 2 + 2+ Zn2+, Cl, SO2 and OH ions. The role played by 4 , CO3 , Na , Mg these ions in the corrosion of the underlying metal is still not well understood, especially for the scenario of rewetting, where the whole system (zinc and zinc patina) is exposed to fresh electrolyte. Previous research examining the barrier properties of the zinc patina, revealed that sulphate based zinc compounds (gordaite, zinc chlorohydroxysulphate) served as more protective barriers than chloride based zinc compounds (zinc hydroxychloride, zinc chloride) [35,45]. However, a number of studies have found that zinc hydroxychloride is a protective corrosion product [64,65,77], reportedly because its crystals have a platelet-like morphology and form more compact surface layers than other BZS. The evolution of the zinc patina on a polished zinc specimen, exposed to a single seawater droplet, is shown in Fig. 5 (adapted from [33]). Initially, a thin oxide layer was detected on the surface, having many voids at the metal–oxide boundary (Fig. 5a). However, after 30 min, localised attack of the metal was observed to take place along its grain boundaries (Fig. 5b). After 6 h exposure, a multilayered zinc patina was detected upon the surface (Fig. 5c). Fig. 5d shows an enlarged version of Fig. 5a, revealing that the thin oxide formed on zinc initially consists of two layers: an inner barrier oxide layer and an outer porous precipitated layer. This oxide formed thus has a duplex structure, as previously reported [37]. The lateral view of the zinc patina, formed after 6 h exposure to a seawater droplet, is shown in Fig. 6a. The cross-section of the patina is displayed in Fig. 6b. The zinc patina transforms into a multilayered structure with time (Figs. 5c, e and 6b). The lowermost layer is the barrier oxide, which is about 10–20 nm thick. A precipitated layer consisting of two sub-layers forms upon this barrier oxide. The inner layer 1 is compact and dense, whereas layer 2 is porous (Fig. 6b). The patina, when seen in lateral view (Fig. 6a), shows a platelet-like morphology interspersed with a large number of pores. However, beneath this layer, there is a compact layer with fewer pores present (Fig. 6b). Thus, the protective capacity of the patina may not be due to the morphology of the surface products, as has been widely reported [64,65], but due to the dense layers formed beneath the surface. The natural densification (repair) of certain layers in this patina is a key aspect which requires further study, as optimising this process could serve to enhance the overall barrier properties, and ‘self-repair’ capabilities of the zinc patina. The role of the barrier oxide and the precipitated porous layer in the context of zinc corrosion has been reviewed in the following sections. 3.1. Barrier oxide 3.1.1. Formation and morphology Passive films form on the surfaces of most reactive metals and serve to self-protect the metal from corrosion [37,78]. The general view is that passive films have a bilayer structure, consisting of an inner defective barrier oxide (the primary passive film), and an outer precipitated oxide, which forms by the hydrolysis of cations ejected through the barrier oxide [37]. The details of the mechanistic processes involved in the formation of ZnO on zinc under atmospheric conditions can be found elsewhere [78]. Zinc usually undergoes passivation during anodic polarisation in alkaline solutions, forming a barrier oxide on its surface. However, the specific mechanism by which this barrier oxide forms upon zinc in aqueous solutions is a topic of some debate as reviewed in [79]. This oxide has been reported to form by several different models, the most prominent ones being: (i) the dissolution– precipitation model; (ii) the adsorption model and (iii) the nucleation and growth model. In the dissolution–precipitation model,

16

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

Time

(a)15 min

(d)

Platinum Coating

Initial thin oxide

(b) 30 min

Growing oxide, with voids and defects (e)

(c) 6 hour

Multiple layers BZS/oxides.

Densification Fig. 5. Oxide surfaces and cross sections at the centre of the drop after 15 min (5a), 30 min (5b), and 6 h (5c). The labels 1, 2, and 3 refer to the three layers of the zinc patina: 1 is below the original metal surface; 2 is immediately above the original surface; 3 is a loose top layer (figure adapted from [33]). A platinum layer of 1–2 lm is deposited over the substrate prior to Focussed Ion Beam (FIB) analysis.

the formation of passive ZnO or Zn(OH)2 was proposed to be due to precipitation, which was induced by the super-saturation of ZnðOHÞ2 ions in the electrolyte close to the metal surface (1) 4 and (2). Platinum  Zn þ 4OH ! ZnðOHÞ2 4 þ 2e

ð1Þ

 ZnðOHÞ2 4 ! ZnO þ H2 O þ 2OH

ð2Þ

The adsorption model postulated that certain adsorbed (ads) species, like ZnOH(ads), reject protons at some critical potential to form a passive oxide layer (3).

ZnOHðadsÞ ! ZnO þ Hþ þ e

ð3Þ

Layer 2- Porous

Layer 1- Dense and compact Barrier Oxide

In the nucleation and growth model, passivation was proposed to result from the direct reaction of Zn with hydroxyl groups (4).

Zn þ 2OH ! ZnO þ H2 O þ 2e

ð4Þ

Fig. 6a. Lateral view of the zinc patina formed under a 1 ll seawater droplet after 6 h exposure.

Fig. 6b. Cross sectional view of the zinc patina formed under a 1 ll seawater droplet after 6 h exposure. A platinum layer is deposited on the substrate prior to FIB analysis.

The zinc oxides formed in alkaline solutions, during anodic polarisation have different morphologies, depending on the reaction mechanisms by which they form. A type of zinc oxide (known as Type I oxide) is usually formed by precipitation reactions ((1) and (2)). Type II zinc oxide forms by solid-state reactions (reactions (3) or (4)), between the metal and Type I oxide. The local pH is a critical factor for forming this Type II oxide, which passivates zinc [80]. Type I oxide serves as a barrier to the transport of OH ions transported to the metal surface, decreasing the interfacial pH and allowing formation of the Type II oxide. Mokaddem et al. [81] reported that three different types of oxides were formed during anodic polarisation of zinc in alkaline solutions, with a Type III oxide formed in small quantities at certain electrode potentials being responsible for the passivation of zinc. In solutions with neutral pH, Zn oxidises to ZnO in the potential range 1.3 VSCE to 1.1 VSCE [82]. However, during potentiodynamic tests, the

17

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

formation of these oxides is usually detected during cathodic polarisation, unlike in alkaline solutions, where the passive oxide formation is detected during anodic polarisation. The morphology of the surface oxides formed on zinc in alkaline solutions is sensitive to local pH [83]. Correspondingly, the Ecorr of zinc in alkaline sulphate solutions showed significant deviation from thermodynamic predictions in the pH range 12–13 [83]. It was proposed that below pH 12 and above pH 13, the surface oxides formed on zinc were homogenous and compact, whilst between pH 12 and 13, the surface oxides formed were inhomogeneous and had a porous morphology. Between pH 12 and 13, zinc corrosion potentials were more noble than at other pH values, due to the formation of zinc hydroxyl precipitates. These precipitates mean that an ion transfer mechanism prevails across the precipitated oxide-solution interface, in contrast to the electron transfer mechanism, which was prevalent when a compact oxide was formed on the surface [83]. The potentiodynamic polarisation curves of zinc in unbuffered 0.1 M chloride solutions with pH 1, 7, 11, 12 and 13 are displayed in Fig. 7 (adapted from [38]). Steady state current densities, corresponding to passivity, are seen only at pH 12 and 13, when the electrode potentials are above 1.15 VSCE. At pH 11, where ZnO is thermodynamically predicted to be stable on the surface, and pseudo-passive behaviour is observed (as opposed to the steady passive behaviour observed for the pH 12 solution). This indicates that with decrease in pH, the surface oxides formed on zinc might be more porous, due to local acidification or chloride induced dissolution processes. The absence of steady state passive currents at pH 11 implies that dissolution through the surface oxides under these conditions is spontaneous, due to the local solution chemistry. However, at pH 12, an impressed electric field (anodic polarisation) is required to break down the barrier oxide (and cause pitting).Thus, the formation of the barrier oxide might be a function of electrode potentials, but its morphology (porous or compact) and susceptibility to break down depends on the local solution chemistry (pH, chloride concentration, etc.).

3.1.2. Corrosion through the barrier oxide For an electrode comprising of the metal and its oxide, the Ecorr settles either in the vicinity of the electrode potential of oxide formation (with the system behaving as an ionic electrode) or the flat band potential of the oxide (if the system is behaving as an electronic electrode) [84]. Thus, for zinc which is in contact with its n-type semiconducting oxide (ZnO), the Ecorr should be around the oxidation potential of Zn to ZnO (1.3 to 1.1 VSCE) [82] or close to the flat band potential of ZnO (around 0.9 VSCE) [85]. The Ecorr values of zinc as a function of pH are plotted on the Pourbaix diagram of zinc in Fig. 8. Under alkaline conditions, the Ecorr of zinc is around 1.3 VSCE, whereas in the pH range 7–11, the Ecorr is in a potential window between the flatband potential of ZnO and the Zn to ZnO oxidation potential. A possible explanation for this behaviour is that ZnO, which is sandwiched between Zn and the electrolyte, is porous, and there are both oxide-electrolyte and metal-electrolyte interfaces in this electrode system, causing the Ecorr to settle in this window. The anodic polarisation curves at pH 12 and 13 (Fig. 7) show that the onset passivation takes place around 1.15 VSCE, with steady passive currents manifest on further polarisation. Macdonald et al. [37] characterised this passive state of zinc and attributed it to the formation of zinc oxide, driven by point-defect reactions. Passivity breakdown was linked to the condensation of metal vacancies at the metal–oxide interface [37]. However, steady state passive currents are not observed at lower pH (Fig. 7), implying that corrosion within the barrier oxide (if present) occurs either: (i) by point-defect reactions within the oxide (which could form or destroy the oxide); (ii) through pores in the oxide or (iii) by both mechanisms. In order to estimate the mode of corrosion through the barrier oxide, its morphology in different solutions need to be ascertained. Cathodic reactions such as the oxygen reduction reaction (ORR) are postulated to occur on the surface of semi-conducting oxides growing on metals [39]. For steel, Stratmann and Streckel [39] reported that Fe2+ species, created within the FeOOH lattice, increase

Pseudo-passive behaviour

Steady state passive behaviour

Fig. 7. Polarisation curves of zinc in 0.1 M NaCl solutions at pH 1, 7, 11, 12 and 13. Steady state passive currents are not observed in pH 11 solution, where ZnO is expected to form a stable layer, based on thermodynamic predictions (adapted from [38]).

18

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

ZnOH+ 2Zn(OH)4 Zn(OH)3

Zn2+

ZnO Zn

A: pH-potential domain for forming passivating Type II/ Type III oxides

Fig. 8. Ecorr of zinc at different pH values plotted on the E–pH diagram for zinc. ‘A’ depicts the regime for passivation of zinc, where a protective Type II/III oxide is expected to form [38,81]. The E–pH diagram was generated using the software Hydra-Medusa [48] (adapted from [38]).

the conductivity of the n-type semiconducting oxide on steel, and also the density of electrons at the oxide/electrolyte interface. This causes a higher rate of ORR on steel, as the ORR takes place on the rust, which has a much higher surface area than that of the metal. In the case of zinc, its oxide (ZnO) is an n-type semiconductor having a band gap of around 3.4 eV [86], and electrical conductivity ranging from 1015 S m1 to 105 S m1. ORR on oxide covered zinc has been reported by a number of researchers [87–90] and is understood to be limited by a 2 electron process producing H2O2. ZnO itself is also capable of undergoing cathodic reduction itself (5) [87].

ZnO þ 2Hþ þ 2e ! Zn þ H2 O

ð5Þ

Goux et al. [91] used a rotating disc electrode to study the ORR on bare platinum and ZnO thin films electrodeposited on platinum. In the absence of Zn2+ ions in solution, the Tafel slope of ORR on ZnO is 139 mV dec1, which is close to that of ORR on Pt (133 mV dec1). The reaction rate, in this case, is limited by the kinetics of the first electron transfer reaction (6):

O2 þ H2 O þ e ! ðHO2 Þads þ OH

ð6Þ

2+

In the presence of Zn ions in solution, the current density decreased significantly (Tafel slope 282 mV dec1) due to the electrode being blocked by the intermediates of the ZnO formation reaction. The exchange current densities (i0) of the ORR as a two electron process, on Zn/ZnO electrode under alkaline conditions (reaction (7)) were close to 109 A cm2 [87]. 





O2 þ H2 O þ 2e ! ðHO2 Þ þ OH

ð7Þ

Flitt and Schweinsberg [82] surveyed the i0 values of the different cathodic reactions on zinc. For water reduction, the i0 varied from 109 to 1012 A cm2; for ORR as a 4 electron process, the i0 varied from 1036 to 1042 A cm2; and for ORR as a 2 electron process, the i0 varied from 1015 to 1025 A cm2. The oxides formed on zinc thus could support the ORR at substantial rates, and hence, the impact of the ORR on zinc oxides needs to be accurately estimated to evaluate its impact on localised corrosion and passivation. A porous oxide model of corrosion was developed to understand the impact of the ORR taking place on oxides on the overall corrosion of the metal [92]. This model also predicts the electrochemical polarisation induced by the galvanic coupling between the metal (anode) and its own oxide (cathode). The different pro-

cesses modelled are depicted in Fig. 9; the ORR occurs on both the metal and oxide surface. Some important parameters which dictate the extent of ORR on the oxide surface were proposed as follows: (i) electrical conductivity of the oxide; (ii) i0 of the ORR on the oxide surface; (iii) specific contact resistivity and (iv) contact potential difference between the metal and its oxide. The Ecorr and corrosion current are influenced by the above parameters, along with physical properties of the oxide (porosity and thickness). The electrical conductivity of the pore solution also plays an important role in the corrosion through porous oxides. In the scenario where the oxide is compact and homogenous, with corrosion being driven through the oxide by a point-defect mechanism, it is difficult to envisage the influence of the ORR on processes within the oxide. In this case, the oxide would be an ionic/electronic conductor located upon the metal surface, with ORR occurring at the oxide/solution interface, and controlling the electric field within the compact oxide. A point-defect mechanism for ORR taking place on passive zinc was hypothesised by Pilbath and Sziraki [87]. However, there is still a distinct lack of knowledge about such non-Faradaic processes. Under open circuit conditions or anodic polarisation tests where point defects characterise the passive state of zinc, the impact of the ORR on the oxide surface and its subsequent effect on the point-defect chemistry within the barrier oxide need to be carefully considered. The ORR releases hydroxyl ions into solution; these ions could either neutralise localised acidity or combine with zinc ions to precipitate different zinc compounds. The location of the ORR sites within zinc corrosion cells thus plays a key role in the precipitation of zinc-based corrosion products. Self-healing at the cut-edge of galvanised steel has been associated with cathodic inhibition (with steel as the cathode), induced by zinc’s corrosion products [93]. Interaction of zinc ions produced at the anodic sites with hydroxyl ions produced at the cathodic sites (by ORR) is responsible for the precipitation of these protective zinc corrosion products [93]. Zinc oxides themselves support ORR and hence release OH ions, however the impact of these processes on the local solution chemistry and subsequently on the dissolution/repair of passive oxides is still not clear. OH ions released by the ORR may have a strong influence on localised corrosion; a low concentration of OH ions could adjust the local pH to favour partial hydrolysis of metal ions, thereby inducing localised acidification, which could promote pit propagation [51]. Optimum supply of OH ions could result in complete metal hydrolysis and precipitation of Zn(OH)2 [94], which could repair the oxide defects. Higher supply of OH ions would cause alkaline dissolution of zinc by the formation of zinc hydroxycomplexes [38]. 3.2. Precipitated layers – their role in protecting zinc The precipitated layers which form upon the barrier oxide are shown in Figs. 5e and 6b. These layers predominantly comprise of BZS, which precipitate out from the solution, due to local super-saturation of zinc ions (released into solution by metal dissolution). The BZS usually form by the reaction of precipitated zinc hydroxide (Zn(OH)2) with different anions [41]. The precipitated layers are usually located upon the barrier oxide. However, if the barrier oxide is porous, the BZS could be present in the pores of the barrier oxide. It is extremely difficult to view pores within the barrier oxide and also to analyse whether BZS precipitates within the oxide pores (Fig. 5e). A more advanced analytical technique is required to give a clear picture of the processes taking place within the barrier oxide. In conventional electrochemical tests like potentiodynamic polarisation or electrochemical impedance spectroscopy, zinc is usually exposed to bulk solutions for short periods of time. In this case, the metal ions released in solution would still be in dilute

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

19

Fig. 9. Depiction of the phenomena modelled in the porous oxide model of corrosion [92]. MDR represents metal dissolution reaction, WR represents water reduction. The cathodic reactions have been modelled to occur both on the oxide as well as the metal surface. The key parameters governing the extent of ORR on the oxide is also shown.

concentrations due to the large volume of electrolyte used. Hence, precipitated layers may not readily form during such experiments, and correspondingly, the implications of the precipitated layer on local corrosion processes cannot be accurately estimated from such experiments. Metal dissolution rates recorded from thin film and droplet exposure are usually reported in terms of mass-loss [29], Rp [30] and volume loss [36]. This data provide clear correlations between the measured corrosion rates and precipitated corrosion products. However, they do not give an ample understanding of the mechanistic role played by the precipitated layers in protecting the metal. It is evident that the precipitated layer serves as a diffusion barrier on the metal; however, its role as an electrical/electrochemical element within the local electrochemical cells is a key aspect which may be addressed in future research. The BZS in the patina potentially impact localised corrosion in both their precipitated state (solid) and their aqueous state (solution). In the precipitated state, they could form salt films upon passive oxides, which could support localised corrosion [95,96]. Formation of a salt film upon a pit (or upon pores in the barrier oxide) would result in the ionic concentrations in the pit solution reaching its saturation (in equilibrium with the precipitated salt). Thus, the pit growth rate (or metal dissolution through the oxide layers) will be under mass-transfer control, depending on the diffusion of ions through this salt layer [97]. The precipitated layers may also behave as an electrical/electrochemical barrier, as they may induce potential drops within local electrochemical cells. If the electrical resistance offered by the precipitated layer is large, it may dominate the interfacial impedance and hence polarisation resistance [98]. It is still not clear whether the BZS (‘zinc rusts’) also support cathodic reactions like the ORR on their surfaces. If these layers support the ORR, zinc corrosion would increase with time as the BZS continues to grow upon the metal. In an aqueous state, the constituent ions of the BZS control the IR drops in local corrosion cells. The IR drops depend on the electrical conductivity of the solution, which in turn depends on the mobility, concentration and solubility of the different ions in the solution. Thus, in pores replete with salt solutions, if the solution

conductivity decreases (due to release of ions with lower mobility or by precipitation of salts), the IR drop would correspondingly increase. This would lower the corrosion cell potential, and hence reduce metal dissolution. For zinc, the role played by the precipitated layer in the protecting the metal is still not clear. The densification observed in the lowermost layers of the patina (Fig. 6b) could be due to any of the above impacts of the precipitated layer. However, it is difficult to apply any specific framework to build repair strategies for the zinc patina. Most inhibitors usually form a protective layer of precipitated compounds upon the barrier oxide. The Atomic Emission Spectroscopy (AES) depth profile of the chromate conversion coating deposited on zinc showed that the actual coating is a mixture of chromium and zinc oxides [99]. The concentration of chromium was found to decrease with depth, and zinc oxide was the main compound detected at the interface between the coating and the zinc substrate. Similarly, characterisation of rare earth metal (hydr) oxide films on zinc using in situ ellipsometry showed that the rare earth metal (hydr) oxide, deposits and nucleate efficiently only upon native (hydr) oxide covered metal surfaces [100]. Aramaki [11] reported that the inhibition achieved using cerium based inhibitors was due to the hydroxylated Ce3+ ions reacting with Zn(OH)2, which initially forms on zinc during its reaction with water. These examples indicate that inhibiting compounds are mainly formed in the precipitated layer, and hence the role of these precipitated layers on processes within the underlying barrier oxide needs to be clarified and related to their inhibiting mechanisms. 4. Prospects of self-repair The literature review has discussed certain problems pertaining to zinc corrosion: (i) Determination of the morphology of the barrier oxide (if present) in non-passivating conditions; this would clarify the mode of corrosion through the barrier oxide (whether

20

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

it occurs through pores or some other solid-state mechanism involving point defects). (ii) Identification of impact of the ORR on the local solution chemistry in zinc-based systems, with the ORR on zinc oxides being an important case. (iii) Impact of the multilayered zinc patina on processes in the underlying barrier oxide and zinc. The corrosion products in a precipitated or aqueous state could impact processes within the local corrosion cells. Densification seen within the precipitated layer could be due to various processes occurring within the precipitated layer, but this particular aspect needs further scrutiny. (iv) Interaction of inhibitors with the barrier oxide and the precipitated layer; the former is usually present underneath the inhibiting compounds, whereas the latter usually forms mixed phases with inhibiting compounds. A better understanding of these aspects could serve to build strategies to enhance self-repair of the zinc patina. Although natural densification of the precipitated layers takes place as indicated in Figs. 5c, e and 6b this particular process will need to be optimised/enhanced. In bulk solutions, passivation of zinc by the formation of Type II/III oxides takes place in alkaline solutions at potentials around 1.15 VSCE (Fig. 8) [38,81]. The true repair of the surface oxide occurs when conditions to form these specific oxides are achieved. The Ecorr values of zinc in non-alkaline solutions or after prolonged exposures are significantly more anodic to 1.15 VSCE. In the presence of artificially produced BZS on the surface, the Ecorr of zinc steadily increase with time, settling around 0.85 VSCE after 1 h of exposure [101] and around 0.55 VSCE after 40 h of exposure [45]. The steady shift to more anodic Ecorr could be due to increase in cathode reaction rates on the surface oxides. Strategies to lower the Ecorr of exposed zinc samples to the protective potential range (around 1.15 VSCE) may be investigated using some specific approaches like (i) reduction in the cathodic reactant supply to the surface (ii) adjustment of the cell potential drop by formation of precipitated layers and (iii) optimisation of the local alkalinity utilising OH ions released by the ORR. Some strategies to enhance self-repair of the patina have been discussed; the approach is mainly focussed on the impact of ORR on the zinc oxides and, depending on the nature of the oxide layer, the strategies could vary. 4.1. Improving effectiveness of the barrier oxide The Pourbaix diagram of zinc [50] reveals that the pH-potential region of ZnO stability extends across the region of water stability. Therefore, both the cathodic reactions (HER and ORR) are thermodynamically feasible upon ZnO. As discussed in the previous sections, kinetic experiments carried out by a number of researchers reveal that the ORR is actively supported on the ZnO surface. Thus, the cathodic current delivering capacity [102] of the local corrosion cell could increase with the formation of ZnO, which could correspondingly increase the metal dissolution rates. In the presence of pores in the surface oxide, since the ORR rates are different on the oxide and metal surface, oxygen concentration gradients could be established between the oxide/solution interface and the metal/ solution interface. These oxygen gradients could induce a potential drop between these two interfaces, which could drive localised corrosion. An approach to lower the ORR rate on ZnO is to reduce the electrochemical activity of ZnO. This strategy was tried as attempted as 1981, wherein zinc was predipped in solutions of cobalt chloride (CoCl2) and nickel acetate to decrease the electrochemical activity of the surface oxides prior to cathodic polarisation tests [103]. Those studies hypothesised that cobalt or nickel atoms served as

electron traps within zinc oxides hence reducing the rate of cathodic reactions. In the case of zinc–magnesium coatings, the introduction of MgO into the zinc oxide has been found to lower the ORR rate on the surface oxides. An increase in Mg content causes the flat band potentials of the surface oxides to shift to more negative values, thereby lowering the surface concentration of electrons, which correspondingly lowers the ORR rate upon the oxides [104,105]. The topic of decreasing the n-type conductivity of ZnO has been reviewed by Janotti and Van de Walle in [106]; lowering the conductivity of zinc oxides could result in a lowered ORR rate and hence could serve as a rationale to develop more corrosion resistant oxides. i0 of the ORR on the oxide surface, specific contact resistivity and contact potential difference between the metal and its oxide are other parameters which could be potentially tuned to control the extent of ORR upon the oxide surface (Fig. 9) [92]. Eliminating the electrolyte contact with ZnO by having an insulating barrier at the electrolyte-oxide interface is another approach to lower the ORR rate upon the oxide surface. An example of this approach is the chromium oxide layer which forms upon ZnO during the chromate conversion coating process. The chromium oxide monolayer is known to decrease the electron transfer rate to outer sphere redox systems that do not require adsorption to the electrode surface namely ferrocene and RuðNH3 Þþ3 6 [54], indicating that the adsorbed Cr (III) film is a good insulating barrier against electron transfer. Hence, the Cr (III) layer formed over the zinc oxides insulates it from the electrolyte, curbing its electrochemical activity. 4.2. Enhancing utilisation of precipitated layers Optimising/enhancing the processes which cause densification, as seen in Figs. 5c, e and 6b, is an approach to promote repair of the precipitated layer within the zinc patina. In the case of the cutedge of galvanised steel, the OH ions which are released into the corrosion cell by the ORR plays a key role in the formation of the precipitated layer [93]. Similarly, OH ions released by ORR on zinc oxide could be utilised efficiently for densification of the precipitated layer. A local increase in alkalinity could result in the dissolution of zinc oxides into certain zinc hydroxycomplexes such as ZnðOHÞ 3 and ZnðOHÞ2 4 . Therefore, certain elements have to be added to zinc, either by alloying or as inhibitors, to precipitate corrosion products which remain stable under alkaline conditions and also to ensure that conditions within the patina do not become highly alkaline. An example of this approach is again the zinc–magnesium coating system, which forms Mg(OH)2 which is insoluble even under alkaline conditions. The formation of this oxide further buffers the pH, rendering non-alkaline conditions, promoting stability of the zinc oxides. Mg2+ ions also favour densification of the patina, as they readily react with carbonates and sulphates during short exposure to form protective/insoluble compounds. They also form layered double hydroxides (LDH) after long exposure. The low electron density and compact structure of the LDH further improve the barrier properties of the precipitated layer [107]. As discussed in Section 2 of this study, ions need to be added into the system to improve its overall pH stability. In the case discussed above, a single cation (Mg2+) has been used to achieve this functionality, and this aspect could be further improved by using certain anions in addition to these cations (as discussed in Section 2). Incorporation of certain ions which could reduce the electrical conductivity of the corrosive electrolyte could serve to increase the IR drop in the local corrosion cells. This could lower the corrosion cell potential and decrease the metal dissolution rates. In the case of fresh exposure to the electrolyte, the precipitated layers and the barrier oxide tend to dissolve, and hence would release

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

some of their constituent ions into the local corrosion cell. These ions impact the IR drop within these cells, and strategies to lower corrosion by controlling such IR drops may also be investigated. 5. Conclusions The short-term protection of zinc is usually achieved using chromate conversion coatings. The chromium-based surface layer formed during this process consists of chromium ions which can exist in dual oxidation states, Cr (III) and Cr (VI). Cr (III) forms Cr2O3 which serves as a protective barrier on the metal, whereas Cr (VI) exists as soluble CrO2 4 , and can self-repair defects in the surface layer. In order to mimic these key functionalities of the chromium-based system, an alternate protective system must (i) be able to form a surface layer with wide pH stability (like Cr2O3) (ii) possess soluble ions which easily leach to defects (like CrO2 4 ) and can (iii) undergo reduction to form passivating oxides at those defects (like the Cr (VI) to Cr (III) reduction). The thermodynamic feasibility of some these beneficial aqueous properties and methods to test them have been elucidated in the current work. The long-term protection of zinc is achieved by its own corrosion products formed during atmospheric corrosion. On droplet exposure, which mimics atmospheric corrosion of zinc by deposition of marine aerosols, the zinc patina gradually grows. It eventually forms a triple layered structure, with a thin barrier oxide and two kinds of precipitated layers. The outer precipitated layer is usually porous, whereas the inner layer naturally undergoes densification with time and becomes extremely compact. The corrosion mechanism through the barrier oxide in non-passivating conditions is still not clearly understood. The processes which naturally densify this inner precipitated layer have not been identified, and also the impact of the precipitated layer (as both an electrical element and a diffusion barrier) on processes in the underlying barrier oxide and metal dissolution has not been clarified in past research. This knowledge could serve to develop strategies to improve the barrier properties of the zinc patina, and also enhance its ‘self-repair’ capabilities. On exposure to bulk solutions, passivation (by formation of a Type II/III oxide) of zinc occurs in alkaline solutions at an electrode potential around 1.15 VSCE. The Ecorr values of zinc in non-alkaline solutions or after long-term exposure are usually more noble than this potential. Lowering the Ecorr of exposed zinc to this potential could favour formation of this passivating Type II/III zinc oxide, and strategies to achieve this may be investigated. The ORR has been reported to occur on the surface of the passive films on zinc (mainly ZnO). Past research has concentrated on lowering the ORR on ZnO by incorporating certain elements such as Co, Ni and Mg into the oxides. These elements have been proposed to lower the n-type conductivity of ZnO and hence curb its electrochemical activity. Some other parameters which could be tuned to control the extent of the ORR on the oxide (in a galvanic couple with the metal as anode) have been identified: (i) i0 of the ORR on the oxide surface, (ii) specific contact resistivity and (iii) Contact potential difference between the metal and its oxide. The ORR releases hydroxyl ions into solution. Local alkalinity could cause either partial or complete hydrolysis of zinc ions, which could either dissolve or repair the zinc oxides. Optimising the ORR upon the surface oxides so that OH ions are released to facilitate oxide repair rather than oxide dissolution is a potential approach towards building protective oxides on zinc. Acknowledgements Dr. Katherine Nairn is gratefully acknowledged for her help in preparing this manuscript. The authors acknowledge the financial

21

support from the CSIRO and the Australian Research Council and also wish to thank Dr. B. Emmanuel and Dr. A.E. Hughes for their helpful comments. Microscopy was performed in part at the Melbourne Centre for Nanofabrication (MCN), which is the Victorian node of the Australian National Fabrication Facility, an initiative partly funded by the Commonwealth of Australia and the Victorian Government. In particular Manoj Sridhar (MCN) is thanked for his technical assistance in performing FIB-SEM.

References [1] D. Lindstrom, I. Odnevall, Environ. Monit. Assess. 173 (2011) 139–153. [2] G.D. Wilcox, J.A. Wharton, Trans. IMF 75B (1997) 140–142. [3] X. Zhang, C. van den Bos, W.G. Sloof, A. Hovestad, H. Terryn, J.H.W. de Wit, Surf. Coat. Technol. 199 (2005) 92–104. [4] International Agency for Research on Cancer (IARC), Chromium, Nickel and Welding. Monographs on the Evaluations of Carcinogenic Risk to Humans, vol. 49, 1990, pp. 1–208. [5] A. Leonard, R.R. Lauwerys, Mutat. Res. 76 (1980) 227–239. [6] S. Langard, Am. J. Ind. Med. 17 (1990) 189–215. [7] L.S. Levy, S. Vanitt, Carcinogenesis 7 (1986) 831–835. [8] G.D. Wilcox, D.E.T. Walker, Int. Met. Finish. 86 (2008) 251–259. [9] O.R. Mattos, A.A.O. Magalhaes, I.C.P. Margarit, J. Electroanal. Chem. 572 (2004) 433–440. [10] B.R.W. Hinton, L. Wilson, Corros. Sci. 29 (1989) 967–&. [11] K. Aramaki, Corros. Sci. 43 (2001) 2201–2215. [12] M.A. Arenas, J. de Damborenea, Corros. Sci. 48 (2006) 3196–3207. [13] K. Aramaki, Corros. Sci. 43 (2001) 1573–1588. [14] A. Jardy, R. Rosset, R.J. Wiart, J. Appl. Electrochem. 14 (1984) 537–545. [15] G.H. Nancollas, Corrosion 39 (1983) 77–82. [16] W.J. Vanooij, A. Sabata, Surf. Coat. Technol. 39 (1989) 667–674. [17] S. Manov, A.M. Lamazouere, L. Aries, Corros. Sci. 42 (2000) 1235–1248. [18] K. Aramaki, Corros. Sci. 43 (2001) 1985–2000. [19] S. Biallozor, J. Dobryszycki, Corros. Sci. 43 (2001) 1309–1319. [20] K. Aramaki, Corros. Sci. 49 (2007) 1963–1980. [21] A. Pilbath, L. Nyikos, I. Bertoti, E. Kalman, Corros. Sci. 50 (2008) 3314–3321. [22] W. He, I. Odnevall Wallinder, C. Leygraf, Runoff rates of zinc—a four year field and laboratory study, in: H.E. Townsend (Ed.), Outdoor Atmospheric Corrosion, ASTM STP 1421, 2002, p. 216. [23] X.G. Zhang, Corrosion and Electrochemistry of Zinc, Plenum Press, New York, 1996. pp. 157–181. [24] D.D. Macdonald, Pure Appl. Chem. 71 (6) (1999) 951–978. [25] N.B. Pilling, R.E. Bedworth, J. Inst. Met 29 (1923) 529–582. [26] S.A. Bradford, ASM Metals Handbook, 1990, p. 64. [27] I.S. Cole, N.S. Azmat, A. Kanta, M. Venkatraman, Int. Mater. Rev. 54 (2009) 117–133. [28] D.C. Stiles, E.O. Edney, Corrosion 45 (1989) 896–901. [29] Q. Qu, L. Li, W. Bai, C. Yan, C. Cao, Corros. Sci. 47 (2005) 2832–2840. [30] Y.Y. Chen, S.C. Chung, H.C. Shih, Corros. Sci. 48 (2006) 3547–3564. [31] P. Qiu, D. Persson, C. Leygraf, J. Electrochem. Soc. 156 (2009) C81–C86. [32] T. Falk, J.–E. Svensson, L.-G. Johansson, J. Electrochem. Soc. 145 (1998) 2993– 2999. [33] I.S. Cole, T.H. Muster, D. Lau, N. Wright, N.S. Azmat, J. Electrochem. Soc. 157 (2010) C213–C222. [34] I.S. Cole, T.H. Muster, S.A. Furman, N. Wright, A. Bradbury, J. Electrochem. Soc. 155 (2008) C244–C255. [35] N.S. Azmat, K.D. Ralston, B.C. Muddle, I.S. Cole, Corros. Sci. 53 (2011) 1604– 1615. [36] N.S. Azmat, K.D. Ralston, B.C. Muddle, I.S. Cole, Corros. Sci. 53 (2011) 3534– 3541. [37] D.D. Macdonald, K.M. Ismail, E.J. Sikora, J. Electrochem. Soc. 145 (1998) 3141– 3149. [38] S. Thomas, N. Birbilis, M.S. Venkatraman, I.S. Cole, Corrosion 68 (2012) 015009. [39] M. Stratmann, H. Streckel, Corros. Sci. 30 (1990) 697–714. [40] T.P.T. Hoar, Faraday Soc. 45 (1949) 683–693. [41] T.E. Graedel, J. Electrochem. Soc. 136 (1989) C193–C203. [42] I. Odnevall, C. Leygraf, Atmos. Corros. 1239 (1995) 215. [43] I. Odnevall, C. Leygraf, in: 12th Scandinavian Corrosion Congress & Eurocorr ‘92, vol. I, 1992, p. 189. [44] M. Morcillo, D. de la Fuente, J.G. Castano, Corros. Sci. 49 (2007) 1420–1436. [45] V. Ligier, M. Wery, J.Y. Hihn, J. Faucheu, M. Tachez, Corros. Sci. 41 (1999) 1139–1164. [46] J. Morales, F. Diaz, J. Hernandez-Borges, S. Gonzalez, Corros. Sci. 48 (2006) 361–371. [47] R. Ramanauskas, P. Quintana, P. Bartolo-Perez, L. Diaz-Ballote, Corrosion 56 (2000) 588–597. [48] http://www.kemi.kth.se/medusa/ (accessed 20.10.11). [49] B. Beverskog, I. Puigdomenech, Corros. Sci. 39 (1997) 107–114. [50] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, second ed., National Association of Corrosion Engineers, Houston, 1974.

22

S. Thomas et al. / Corrosion Science 69 (2013) 11–22

[51] J.R. Galvele, J. Electrochem. Soc. 123 (4) (1976) 464–474. [52] M.W. Kendig, R.G. Buchheit, Corrosion 59 (2003) 379–400. [53] E. Akiyama, A.J. Markworth, J.K. McCoy, G.S. Frankel, L. Xia, R.L. McCreery, J. Electrochem. Soc. 150 (2003) B83–B91. [54] W.J. Clark, R.L. McCreery, J. Electrochem. Soc. 149 (2002) B379–B386. [55] E. McCafferty, J. Appl. Electrochem. 40 (2010) 543–549. [56] G.M. Treacy, G.D. Wilcox, M.O.W. Richardson, J. Appl. Electrochem. 29 (1999) 647–654. [57] C.G. Silva, I.C.P. Margarit-Mattos, O.R. Mattos, H. Perrot, B. Tribollet, V. Vivier, Corros. Sci. 51 (2009) 151–158. [58] Chen-Yang Tsai, Jen-Shou Liu, Pei-Li Chen, Chao-Sung Lin, Corros. Sci. 52 (2010) 3385–3393. [59] F. Rosalbino, E. Angelini, D. Maccio, A. Saccone, S. Delfino, Electrochim. Acta 54 (2009) 1204–1209. [60] F. Pedraza, Y. Hamlaoui, C. Remazeilles, M. Bordes, L. Tifouti, Corros. Sci. 52 (2010) 1020–1025. [61] S.A. Hayes, P. Yu, T.J. O’Keefe, M.J. O’Keefe, J.O.J. Stoffer, J. Electrochem. Soc. 149 (2002) C623–C630. [62] F.H. Scholes, A.E. Hughes, S.G. Hardin, P. Lynch, P.R. Miller, Chem. Mater. 19 (2007) 2321–2328. [63] M.A. Arenas, J.J. de Damborenea, Electrochim. Acta 48 (2003) 3693–3698. [64] N.C. Hosking, M.A. Strom, P.H. Shipway, C.D. Rudd, Corros. Sci. 49 (2007) 3669–3695. [65] P. Volovitch, C. Allely, K. Ogle, Corros. Sci. 51 (2009) 1251–1262. [66] T. Prosek, D. Thierry, C. Taxen, J. Maixner, Corros. Sci. 49 (2007) 2676–2693. [67] S.M. Powell, H.N. McMurray, D.A. Worsley, Corrosion 55 (1999) 1040–1051. [68] K. Aramaki, Corros. Sci. 43 (2001) 591–604. [69] K. Aramaki, Corros. Sci. 44 (2002) 2621–2634. [70] K. Aramaki, Corros. Sci. 44 (2002) 871–886. [71] K. Aramaki, Corros. Sci. 45 (2003) 2361–2376. [72] K. Aramaki, Corros. Sci. 45 (2003) 1085–1101. [73] K. Aramaki, Corros. Sci. 48 (2006) 3298–3308. [74] T.A. Markley, M. Forsyth, A.E. Hughes, Electrochim. Acta 52 (2007) 4024– 4031. [75] J. Zhao, G. Frankel, R.L. McCreery, J. Electrochem. Soc. 145 (1998) 2258–2264. [76] H. Guan, R.G. Buchheit, Corrosion 60 (2004) 284–296. [77] W. Miao, I.S. Cole, A.K. Neufeld, S.A. Furman, J. Electrochem. Soc. 154 (2007) C7–C15. [78] G.A. Alimenti, M.E. Gschaider, J.C. Bazan, M.L. Ferreira, J. Colloid Interf. Sci. 276 (2004) 24–38.

[79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107]

L.M. Baugh, A. Higginson, Electrochim. Acta 30 (1985) 1163–1172. R.W. Powers, M.W. Breiter, J. Electrochem. Soc. 116 (1969) 719–&. M. Mokaddem, P. Volovitch, K. Ogle, Electrochim. Acta 55 (2010) 7867–7875. H.J. Flitt, D.P. Schweinsberg, Corros. Sci. 52 (2010) 1905–1914. S. Bonk, M. Wicinski, A.M. Hassel, M. Stratmann, Electrochem. Commun. 6 (2004) 800–804. N. Sato, Corros. Sci. 42 (2000) 1957–1973. R. Hausbrand, M. Stratmann, M. Rohwerder, J. Electrochem. Soc. 155 (2008) C369–C371. X.G. Zhang, Corrosion and Electrochemistry of Zinc, Plenum Press, New York, 1996. pp. 93–124. Zs. Pilbath, L. Sziraki, Electrochim. Acta 53 (2008) 3218–3230. K.G. Boto, L.F.G. Williams, J. Electroanal. Chem. 77 (1977) 1–20. H. Dafydd, D.A. Worsley, H.N. McMurray, Corros. Sci. 47 (2005) 3006–3018. A.P. Yadav, A. Nishikata, T. Tsuru, J. Electroanal. Chem. 585 (2005) 142–149. A. Goux, T. Pauporte, D. Lincot, Electrochim. Acta 51 (2006) 3168–3172. M.S. Venkatraman, I.S. Cole, B. Emmanuel, Electrochim. Acta 56 (2011) 8192– 8203. K. Ogle, S. Morel, D. Jacquet, J. Electrochem. Soc. 153 (2006) B1–B5. J.R. Galvele, Corros. Sci. 21 (1981) 551–579. P.C. Pistorius, G.T. Burstein, Philos. Trans. Roy. Soc. Lond. A 341 (1992) 531– 559. N.J. Laycock, R.C. Newman, Corros. Sci. 39 (1997) 1771–1790. G.S. Frankel, J. Electrochem. Soc. 145 (1998) 2186–2189. D.D. Macdonald, Electrochim. Acta 56 (2011) 1761–1772. X. Zhang, W.G. Sloof, A. Hovestad, E.P.M. Van Westing, H. Terryn, J.H.W. de Wit, Surf. Coat. Technol. 197 (2005) 168–176. S. Bohm, R. Greef, H.N. McMurray, S.M. Powell, D.A. Worsley, J. Electrochem. Soc. 147 (9) (2000) 3286–3293. T.H. Muster, I.S. Cole, Corros. Sci. 46 (2004) 2319–2335. Z.Y. Chen, F. Cui, R.G. Kelly, J. Electrochem. Soc. 155 (7) (2008) C360–C368. H. Leidheiser Jr., I. Suzuki, J. Electrochem. Soc. 128 (1991) 242–249. R. Hausbrand, M. Stratmann, M. Rohwerder, Corros. Sci. 51 (2009) 2107– 2114. T. Prosek, A. Nazarov, U. Bexell, D. Thierry, J. Serak, Corros. Sci. 58 (2008) 2216–2231. A. Janotti, C.G. Van De Walle, Rep. Prog. Phys. (2009) 126501. P. Volovitch, T.N. Vu, C. Allely, A. Abdel Aal, K. Ogle, Corros. Sci. 53 (2011) 2437–2445.