Review of the atmospheric corrosion of magnesium alloys

1. Introduction

Magnesium (Mg) alloys are the lightest of the structure alloys, and are receiving increasing attention because of their low density and adequate mechanical properties, including good machinability, conductivity, and damping [1,2]. They are considered for a wide range of uses in the aerospace and automotive industries, and for biodegradable implants [[3], [4], [5], [6], [7], [8], [9], [10], [11], [12]]. However, their range of applications is more restricted than that of aluminum alloys due to their poorer corrosion behavior [[13], [14], [15], [16]]. Common corrosion protection methods for Mg alloys include microstructure control [17,18], composition modification [[19], [20], [21]], and surface treatments and coatings [[22], [23], [24], [25], [26], [27]]. The corrosion behavior of a Mg alloy depends on the environment, the alloy composition and production details [28,29]. A better understanding of these factors will lead to better Mg alloys.

Many researches in the past few decades focused on Mg corrosion in aqueous solutions, especially in chloride containing solutions for automobile applications, and in synthetic body fluids for biomedical applications. The influence of chemical composition, microstructure, surface condition and solution compositions was intensively investigated and is summarized in recent reviews [[30], [31], [32]].

In contrast, there have been fewer investigations into the atmospheric corrosion of Mg alloys, even though atmospheric corrosion is important for most Mg components in practical service for automobile, aerospace, military and telecommunication applications [[33], [34], [35], [36], [37], [38], [39]]. This is partly because atmospheric corrosion of Mg alloys is less severe than aqueous corrosion, and partly because there are greater experimental difficulties in understanding the electrochemistry involved in atmospheric corrosion. New insights [[40], [41], [42]] have indicated that atmospheric environmental factors have a much more complicated influence on the corrosion behavior, and their influence is difficult to predict based on the corrosion behavior in aqueous solution [43,44]. The important atmospheric parameters include: temperature, relative humidity, time of wetness, atmospheric composition, the presence of industrial pollutants, and the presence of inorganic salts. The successful use of Mg components requires understanding the influence of these factors. Therefore, this critical review provides a framework, on which to build the further development of Mg alloys that are particularly suited for atmospheric service.

This review summarizes the influences of metallurgical and environmental factors on Mg atmospheric corrosion. This review is expected to deepen the current understanding and lay a foundation for exploration of possible future breakthroughs.

2. Characteristics of Mg atmospheric corrosion

2.1. Mg corrosion mechanism

Mg alloys show good resistance in dry air and little or no corrosion occurs. Atmospheric corrosion depends on the availability of water, and generally proceeds by means of electrochemical reactions under an aqueous layer or in droplets on the Mg surface [13,45]. The electrolyte allows ionic electrical paths and permits galvanic corrosion cells. Secondary phases (especially impurity rich phases) act as efficient cathodes and accelerate the corrosion of the Mg matrix, and degrade partially protective surface films. Gases such as NO_x, SO₂ and CO₂ are dissolved in the aqueous surface layer and are transported to the Mg alloy surface to influence the corrosion processes. This causes the corrosion rate to be greater in industrial environments than in rural atmospheres. In rural atmospheres, the corrosion rate of Mg alloys is comparable or lower than that of mild steel [46]. This is attributed to the fact that the surface films formed on the surface of Mg alloys in mild atmospheric conditions are much more protective that those typically formed under immersion conditions, and moreover are more protective than the film formed on the surface of steel [47].

Under a thin aqueous layer, the anodic and cathodic reactions involved in the corrosion of Mg alloys are similar to those in immersion corrosion, except for the non-negligible contribution of oxygen reduction, which normally can be ignored for immersion conditions. The overall anodic reactions is:

Mg(s) → Mg²⁺(aq)+ 2e^- (1)

The overall cathodic reaction is either of the following:

2H₂O(aq)+2e^- → H₂(g)+2OH^-(aq) (water reduction) (2)

O₂(g)+2H₂O(aq)+4e^- → 4OH^-(aq) (oxygen reduction) (3)

The overall corrosion reaction is:

Mg + 2H₂O → Mg(OH)₂ + H₂ (water reduction) (4a)

or

2Mg + O₂ +H₂O → 2Mg(OH)₂ (oxygen reduction) (4b)

Investigations on the contribution of oxygen reduction to the corrosion of Mg are rare, and this is an important scientific topic still to be studied. Also, the anodic dissolution of Mg (1) involves detailed electrochemical steps, which result in“anodic hydrogen evolution”and the negative difference effect [[48], [49], [50]]. Another interesting question is whether the involvement of oxygen in atmospheric corrosion changes the anodic dissolution mechanism.

The corrosion mechanisms of Mg alloys in atmospheric and immersion environments are similar, but the atmospheric corrosion rate is significantly lower. This can be evidenced by a comparison of the corrosion rates of pure Mg and some commercial Mg alloys in atmosphere and 3 wt.% NaCl solution (Table 1). Moreover, pure Mg and Mg alloys in atmosphere and solution usually suffered localized corrosion [39,51,52], but the size or depth of the pits formed in atmosphere are considerably smaller.

2.2. Surface films

An oxide film forms rapidly over the Mg surface when it exposed to the air due to the large reactivity of Mg [53]. This oxide film protects the Mg substrate because it is dense and stable. However, water molecules in wet air can trigger hydration of the oxide and lead to corrosion. The film formed on a scratched Mg surface after exposure to humid air (∼65% RH, 30 °C) exhibited a bilayer structure with an apparently dense outer layer (mainly Mg(OH)₂) and a cellular inner layer (mainly MgO) [54]. The structure and the formation mechanism of this surface film are affected by the chemical composition of the Mg alloy, the constituents of the atmosphere, the temperature, and the humidity [55]. The protectiveness of this surface film is the main determinant of the atmospheric corrosion behavior. Therefore, understanding Mg atmospheric corrosion requires understanding the factors which influence the film on their surface.

In fact, the atmospheric corrosion of a Mg alloy is also affected by micro-galvanic couples. Apart from the surface film, many metallurgical and environmental factors under atmospheric conditions can lead to invalidation of these micro-galvanic couples. Hence, the atmospheric corrosion is usually slower than the immersion corrosion for a Mg alloy.

3. Metallurgical factors

Atmospheric corrosion behavior is influenced by alloying elements, second phases or other intermetallic particles and their distribution, grain size and grain orientation. However, compared with aqueous corrosion, the understanding of the influence of these factors on Mg atmospheric corrosion is relatively incomplete. Only a few alloying elements have been studied.

3.1. Alloying elements

3.1.1. Aluminum (Al)

In aqueous solutions containing chlorides, alloying with Al decreases the corrosion rate [14,56,57]. However, in some cases the corrosion rate increased with increasing Al content (e.g., Liu et al. [32]), which can be attributed to the confounding influence of the impurity Fe. The improvements have been attributed to the barrier effect of the β-phase network inhibiting corrosion, and to the improved protectiveness of the surface film on the matrix caused by Al dissolved in the Mg-matrix solid solution [15,58]. This increased protectiveness is attributed to Al oxide accumulating at the interface between the Mg substrate and the surface film [59,60], increasing the protectiveness of the surface film, which increases the resistance to local film breakdown, and consequently decreases the corrosion rate [61].

The atmospheric corrosion rate is decreased with increasing Al concentration [62,63]. For example, the corrosion rate of AZ61 was 20% lower than that of AZ31 in a continuous condensation atmosphere, attributed to the higher concentration of Al in AZ61 [64]. Similarly, Jönsson and Persson [62] found that AZ91D had a corrosion rate lower than that AM50 for indoor exposure, further supporting the decrease of corrosion rate with increasing Al content. Merino et al. [63] also reported that in salt fog, the corrosion rate decreased with increasing Al concentration: Mg > AZ31 > AZ91D > AZ80, as shown in Fig. 1(a). The better performance of AZ80 was attributed to the higher Al content (13%) in the AZ80 dendrites compared with the lower amount of Al (8%) in the α-phase in the interdendritic areas of AZ91D due to its higher solidification rate [14]. Similarly Feliu et al. [29] found a decreased corrosion rate with increasing Al concentration as shown in Fig. 1(b), which was caused by a higher percentage of Mg carbonate on the surface of the Mg-Al alloy. These findings indicated that the Al concentration and the Al distribution both play an important role in the corrosion of Mg alloys [65]. The effect of Al on decreasing the atmospheric corrosion rate derives from the Al-enriched surface film which is more protective than the MgO and Mg(OH)₂ films [52]. The beneficial effect of Al alloying on corrosion of Mg alloys has also been found in real atmospheric environments [28,34,35].

3.1.2. Rare earth elements (RE)

Alloying with rare earth elements (RE) also decreased corrosion rates [66,67] of AZ-series and AM-series Mg alloys in aqueous chloride containing solutions, attributed to two factors [[68], [69], [70]]：1) the formation of RE-containing intermetallic phases reduces the amount of the β-phase and impurity (e.g. Fe) containing particles; and 2) the presence of RE alloying in the Mg-matrix improved the protectiveness of the corrosion product surface film. A small amount of RE decreases the corrosion rate, but a larger concentration increases the corrosion rate because of the microgalvanic effect of a larger amount of RE-containing second phase particles [69,71].

In atmospheric environments, a considerable decrease in the corrosion rate occurs for Mg alloys containing RE elements [66,72]. Arrabal et al. [72] found that alloying with Nd or Gd decreased the corrosion rate of AM50. The Nd or Gd reduced the volume fraction of Al-Mn intermetallics as a result of the formation of Al-Mn-RE and Al₂RE intermetallic phases. The micro-galvanic corrosion became weaker because the Al-Mn-RE and Al₂RE intermetallic phases are less efficient cathodes. Similarly, Arrabal et al. [66] found that alloying with Nd reduced the corrosion rate of AZ91 by 70% under real atmospheric corrosion conditions. The Nd alloying refined the β-phase and caused the formation of a more protective corrosion product layer. These studies thus indicate that RE elements decrease the atmospheric corrosion rate caused by: (i) the decreased micro-galvanic effect between the α-Mg matrix and second phases; (ii) formation of a beta phase network; and (iii) a more protective surface film.

3.1.3. Other alloying elements

The alloying elements, Zn, Ca and Sr, dissolved in Mg matrix also have a positive effect on the atmospheric corrosion performance. Zn dissolved in the Mg improved the corrosion performance of Mg alloys by making the surface film more compact [73]. Alloying with calcium decreased the corrosion rate. Compared with Zn, Ca is relatively difficult to dissolve in the Mg matrix, but tends to form the intermetallic compound (Mg₂Al)₂Ca that is nobler than the Mg matrix, and which increases the corrosion rate [74,75]. Sr has been found to decrease the Mg corrosion rate [76].

The amount of alloying element must be limited to below the solubility limit, as a larger amount results in the formation of intermetallic phases which increase the corrosion rate by causing micro-galvanic corrosion [66,77]. However, the micro-galvanic effect is usually less serious than that in aqueous solutions as it is suppressed by the limited electrolyte conductivity and by the corrosion products [43,44].

3.2. Microstructure

3.2.1. Influence of the matrix phase

Grain refinement improves the mechanical properties [[78], [79], [80]], and the corrosion performance [[81], [82], [83], [84], [85]]. Grain size can influence the electrochemistry of Mg alloys [86,87], and there are a number of publications on the role of grain size on Mg atmospheric corrosion [28,88,89]. Liao et al. [89] reported that the corrosion rate of AZ31B had an inversely linear relationship with d^1/2 (grain size) in cyclic neutral-salt spray test. In marine and urban environments, the corrosion rate of the AZ31B with fine grains was lower to that of the AZ31B with coarser grains, especially in urban environments. Li et al. [90] attributed the lower corrosion rate of extruded AM60 to the smaller grain size and more homogeneous microstructure compared with the ingot AM60. Grain refinement by extrusion resulted in a uniformly distributed more-protective corrosion film. Moreover, grain refinement may also reduce the possibility of surface film cracking, because of the lower residual stress in the surface film [88,91,92].

3.2.2. Intermetallic phases

Intermetallic particles significantly increase the corrosion rate of a Mg alloy. β-phase (Mg₁₇Al₁₂), Mg₂Zn and β-Mn are the intermetallic phases often present in Mg-Al, Mg-Zn and Mg-Mn alloys, respectively. The β-phase precipitates along grain boundaries, has a lower corrosion rate, but is cathodic to the Mg matrix and consequently increases the corrosion rate of Mg-Al alloy due to its micro-galvanic acceleration of the corrosion of the α-Mg matrix [60]. The galvanic effect of the β-phase is more localized under atmospheric conditions than in a bulk solution [93]. Li et al. [90] found that in an industrial environment, the corrosion rates of ingot and extruded AM60 samples were 1.4 and 1.1 g/m²/y, respectively. The higher corrosion rate of the ingot AM60 was attributed to the higher volume fraction of β-phase particles in the alloy, resulting in more galvanic acceleration. Furthermore, they [90] found that the atmospheric corrosion rate of AM60 was four times lower than that of die-cast AZ91 because of the lower amount of the β-phase. However, in contrast Yang et al. [94] found that the finer grains and the continuous net-like β-phase in a high pressure die-cast (HPDC) specimen decreased the corrosion rate. Merino et al. [63] found that the more continuously distributed finer β-phase in AZ80 was a more efficient corrosion barrier than the coarse β-phase in AZ91D Mg, so that the AZ80 Mg had a lower corrosion rate. Therefore, similar to the corrosion of Mg alloy in a bulk solution [32], the β-phase plays a dual role in Mg atmospheric corrosion as proposed by Song et al [93,95]. For immersion corrosion, the continuous β-phase network with fine grains acts as a barrier to restrain further corrosion; otherwise, the beta-phase acts as a micro-galvanic cathode to increase the corrosion rate if it occurs as coarse particles in the Mg matrix [28]. A schematic illustration of the dual role of the β-phase in Mg-Al alloys is presented in Fig. 2 [96].

However, an interesting corrosion phenomenon has also been reported under atmospheric conditions [42]. The β-phase was preferentially corroded, which does not occur in a neutral bulk solution. This suggests that the existing knowledge regarding the effect of Mg alloy composition and microstructure of corrosion gained from bulk solutions cannot always be simply transplanted to atmospheric corrosion.

The Al-Mn intermetallic is also cathodic to the Mg matrix, and hence accelerates the corrosion of Mg alloys [72,97,98]. However, a recent study [89] found that the precipitated Al₆Mn/Al₈Mn₅ particles had an insignificant impact on the atmospheric corrosion of AZ31B, because of their small quantity. In real atmospheric environments, the corrosion rates of AMX602 series with different microstructures increased in the order: AMX602-S < AMX602-C < AMX602-E [28] (S,C,E respectively represent the manufacturing processes: S - spinning water atomization, C - gravity casting, E - hot extrusion), indicating that the atmospheric corrosion performance of AMX602 series alloys was principally dependent on the grain size and the distribution of Al₂Ca intermetallic particles in the alloy. When the grain size was small and the intermetallic phase was finely dispersed in the matrix, the intermetallic mainly acted as a corrosion barrier; but the intermetallic particles became micro-galvanic cathodes and promoted corrosion when they were coarse and dispersed in the Mg substrate, which was consistent with the results reported by Jönsson et al. [35].

Since the atmospheric corrosion of Mg alloys can be affected by the alloy composition and microstructure simultaneously, it is important to know which factor dominates the corrosion behavior. Merino et al. [63] found that the corrosion in salt fog was more significantly influenced by the Al distribution and the β-phase morphology than the Al-Mn particles, which is in good agreement with Jönsson et al. [35]. Nevertheless, Al-Mn particles do cause some corrosion acceleration for Mg-Al alloys in NaCl solution (3.5 wt. %) [14].

So far, limited investigations have indicated that, except for some unexpected phenomena in some cases, such as the dissolution of the β-phase, the influence of Mg microstructure on atmospheric corrosion is similar to that for immersion corrosion, and correspondingly there has been no new influencing mechanism. However, the Mg atmospheric corrosion rate is lower than the immersion corrosion rate.

4. Environment factors

The environment parameters include relative humidity (RH), temperature, gaseous composition, electrolyte composition, deposited salt species and dust particles on the Mg alloy surface [[99], [100], [101]].

4.1. Relative humidity (RH)

Relative humidity (RH) plays the most important role among all the environmental factors. Mg can be very corrosion resistant in dry air at ambient temperature [13,45]. However, when RH increases and reaches a certain level, the corrosion of Mg will change from a relatively slow chemical process to a significantly fast electrochemical reaction. The thickness of the liquid film on a metal surface increases with the RH of the ambient air as shown in Table 2 [102]. At 95% RH the water film can be more than 16 monolayers and can exhibits characteristics close to those of bulk water [103]. Furthermore, hygroscopic deposits also cause the formation of a water film. For example, an aqueous film occurs on the surface at RH > 76% in the presence of NaCl [45]. Therefore, the corrosion can occur in atmospheric environments with RH < 100%.

In principle, this phenomenon also applies to Mg alloys. However, the hygroscopicity of the oxide and hydroxide film on Mg alloys could be different from that on the other engineering metals. There is a possibility that the thickness of the adsorbed liquid water layers could be different. Unfortunately, there has been no investigation on the liquid film for Mg alloys under atmospheric conditions.

A higher relative humidity produces a thicker aqueous film, which may lead to the generation of a less-protective corrosion product layer. However, there have been few papers on the influence of relative humidity on the atmospheric corrosion of Mg alloys. Nevertheless, it is generally accepted that the corrosion rate of Mg alloys increases with RH. For example, a small increase in RH led to increased levels of surface tarnish for Mg alloys [41]. LeBozec et al. [38] showed that the corrosion rates of AZ91D and AM50 both increased with increasing RH from 75% to 95% at 25 °C and 35 °C, which has been summarized in Fig. 3. A Fourier transformed infrared spectroscopy (FTIR) analysis indicated that the corrosion products formed at 95% consisted mainly of hydromagnesite as has been found in nature [105], while at 75% RH magnesite was the principal corrosion product in the form of nesquehonite [36]. This was consistent with Arrabal et al. [106] who reported that there was no obviously visible appearance change in an atmosphere with 80% RH.

4.2. Temperature

In a natural atmosphere, the local metal temperature can be different to the average air temperature because of heat transfer (e.g. radiation, conduction, and convection) between the metal surface and the air [38].

In general, temperature increases the kinetics of chemical reactions, so that the corrosion rate would be expected to increase with increasing temperature. In contrast, if the speed of corrosion is limited by a compact surface oxide film, the speed of corrosion would be expected to decrease with temperature because, according to Cabrera-Mott mechanism, at low temperatures the thickness of a surface oxide film increases with temperature [107]. Thus it is not surprising that the influence of temperature on the corrosion rate is different for different metallic materials [[108], [109], [110]]. For example, Blücher and Svensson [109] found that the corrosion rate of Al increased with increasing temperature as the relevant reactions are thermally activated. In contrast, Niklasson et al. [100] claimed that the corrosion rate of lead induced by acetic acid vapor decreased with increasing temperature (in the range from 4 °C to 22 °C), which was attributed to the decreasing adsorption of acetic acid molecules on the lead surface with increasing temperature. A similar negative correlation with temperature (in the same range) was observed for the corrosion of zinc induced by SO₂ at a high relative humidity by Svensson and Johansson [111], which was ascribed to the formation of zinc hydroxyl sulfate at room temperature.

With respect to the influence of temperature on the atmospheric corrosion of Mg alloys, most investigations have been performed at room temperature or above. There are few investigations below room temperature, especially close to or below 0 °C [41,99]. Song et al. found that, in a laboratory environment chamber without any contaminants, increasing temperature accelerated the surface tarnishing for Mg alloys. Lebozec et al. [38] found an increase in the average corrosion rate by approximately 30% induced by a temperature increase from 25 °C to 35 °C for all Mg alloys studied as shown in Fig. 3, attributed to increasing kinetics with increasing temperature. Similarly, Merino et al. [63], under a salt fog condition, found an increasing corrosion rate caused by increasing temperature from 25 °C to 35 °C, which was greater than that caused by increasing the chloride concentration from 2 wt.% to 5 wt.%. Furthermore, the temperature effect was alloy dependent. The temperature effect was relatively insignificant for AZ80 and AZ91D compared with that on pure Mg and on AZ31, which can be attributed to the better film on the alloys with higher Al content.

The Mg atmospheric corrosion at low temperatures is also of concern for the automotive industry, because external surfaces of automotive components are exposed to NaCl containing deicing salt in winter. The presence of chlorides can significantly accelerate surface tarnishing [41]. Esmaily et al. [110] found that the corrosion rate of AM50 in an atmospheric environment containing NaCl increased with increasing temperature in the temperatures range from -4 to 22 °C. However, for 99.97 wt.% purity Mg there was little increase of corrosion rate with temperature. This is contradictory to Merino et al. [63] that the temperature effect was inhibited by Al alloying in Mg.

4.3. Chloride

Chloride is one of the most corrosive species, which can induce breakdown of the Mg surface oxide film, resulting in rapid Mg corrosion [13,66,101,112,113]. This has been widely investigated in aqueous solution [51,[113], [114], [115], [116], [117], [118]]. The immersion experiments of Dhanapal [119] and Altun [120] lead to the following conclusions:

1 The corrosion rate increased with increasing Cl^- concentration for Cl^- concentrations less than 1 M.

2 For higher concentrations, the corrosion rate increased only slightly.

3 The corrosion potential usually shifted negatively with increasing Cl^- concentration.

The increasing corrosion rate with increasing concentration of Cl^- was attributed to Cl^- ions penetrating and breaking down the somewhat protective corrosion product film. The partially protective layer formed on the AM50 was entirely penetrated by Cl-, which significantly increased the corrosion rate [121]. The more negative corrosion potential could be attributed to the greater amount of actively corroding surface [114].

Thus atmospheric corrosion is similar to aqueous corrosion [42,44] in the detrimental effect of chlorides. Merino et al. [63] found that in a salt fog environment the corrosion rate of AZ31, AZ80 and AZ91D increased with increasing concentration of Cl^-, attributed to increased breakdown of the surface oxide film at higher chloride concentrations. The detrimental effect of the chloride concentration for AZ91D was more pronounced than for AZ80. However, these results do not correspond well to outdoor exposures. The corrosion rate was accelerated too much, about an order of magnitude higher than the typical corrosion results in atmospheric exposures, which might lead to a corrosion mechanism different from that in natural atmospheric exposure [122]. The well-controlled laboratory tests [36,38] of Lebozec et al. [38] found that the atmospheric corrosion rates of both AZ91D and AM50 exhibited a linear dependence on the amount of NaCl deposition on the exposed surface regardless of the exposure conditions as shown in Fig. 4(a), which was explained by the increase in conductivity of the thin NaCl electrolyte film with increasing chloride concentration. This is analogous to the results reported by Zhou et al. [123], who found that the corrosion rate of AZ91D with 120 μg/cm² NaCl deposition on the surface was obviously higher than that with 50 μg/cm² NaCl deposition as indicated in Fig. 4(b). The corrosion rate decreased with exposure time, because the initial corrosion product inhibited the transfer of the corrosive medium to the surface and thereby delayed further corrosion. Jönsson et al. [36] investigated the atmospheric corrosion of AZ91D and found that soluble chloride ions formed a layer of electrolyte solution on the metal surface, promoting the dissolution of AZ91D and forming poorly corrosion resistant products. At the same time, the surface electrolyte layer provided a highly conductive medium for the spatially separated anodes and cathodes present on the surface; and the primary corrosion products mainly consisted of Mg carbonate.

The detrimental effect of chloride ions has also been observed on the field-exposure corrosion behavior of Mg alloys. Liao et al. [89] found that the corrosion rate of AZ31B in an marine atmosphere was much higher than in an urban region, which was attributed to the larger amount of chlorides and higher RH in the marine environment. This is consistent with observation by Jönsson et al. [35]. Cui et al. [33] found that corrosion rate of AZ31 evidently fluctuated with the variation of chloride deposition rate caused by the changes of the environmental condition in a marine atmosphere.

4.4. Oxygen

The gaseous composition has a strong effect on the atmospheric corrosion of Mg alloys by affecting the electrochemical processes and the formation of droplets [124]. Normal air is mainly composed of N₂, O₂, CO₂ and some rare gases. N₂ and rare gases do not significantly affect the corrosion of Mg alloys due to their inertness [125]. O₂ has a significant contribution to cathodic depolarization and thus differentiates atmospheric corrosion from immersion corrosion.

O₂ is present in atmospheric corrosion in a dissolved form in the surface water film. O₂ does not significantly contribute to corrosion of Mg in distilled water or saline solutions, because hydrogen evolution is the main cathodic reaction in this case [32]. For example, Bedjoudi et al. [126], found that the cathodic curve for Mg in a NaCl solution saturated with Mg(OH)₂ was the same with aeration and deaeration. For Na₂SO₄ solutions, Braril and Pébère [127] also found that the dissolved oxygen had no influence on the cathodic reaction, which indicated that the cathodic process was not influenced by oxygen diffusion, and consequently oxygen had no impact on the corrosion of Mg. However, the oxygen in theory can participate in the cathodic reaction, which could accelerate the atmospheric corrosion rate, although the oxygen accelerated corrosion have not been frequently observed experimentally. Some special environments with positive potentials and good aeration have indicated some contribution of oxygen to Mg atmospheric corrosion [128].

4.5. Carbon dioxide

A number of studies have examined the influence of CO₂ on the atmospheric corrosion. The concentration of CO₂ in the natural ambient air is 330 ppm. Dissolution in water forms carbonic acid, so that CO₂ can reduce the pH of the electrolyte on a metal surface and can transform Mg hydroxide into Mg hydroxy-carbonates on the Mg alloy surface [13].

The development of the air-formed film and its transformation on Mg in humid air can be summarized as follows [129,130]. (i) In the absence of CO₂, the original air-formed films mainly consists of a MgO/Mg(OH)₂ layer. Mg(OH)₂ can be dissolved in the water film on the Mg surface by Eq. (5). MgO is more soluble than Mg(OH)₂ and dissolves by Eq. (6) [131]. Dissolution of MgO and Mg(OH)₂ decreases the thickness of the initial oxide film. (ii) In CO₂-containing air, the surface electrolyte film becomes acidified by the dissolution of CO₂ by Eqs. (7) and (8), which enhances the dissolution of the surface MgO/Mg(OH)₂ layer [130].

Mg(OH)₂ ⇄ (s) Mg²⁺(aq) + 2OH^-(aq) (5)

MgO + H₂O ⇄ Mg²⁺(aq) + 2OH^-(aq) (6)

CO₂+H₂O → HCO₃^- + H⁺ (7)

HCO₃^- +OH^- → CO₃^2-+H₂O (8)

In an atmosphere with a normal concentration of CO₂ and a low relative humidity, no continuous electrolyte layer is present on the Mg surface, and the Mg(OH)₂ (brucite) directly react with CO₂ to form Mg carbonate:

Mg(OH)₂(surface) + CO₂(g) → MgCO₃(ads) + H₂O (9)

The generation of Mg carbonate inhibits Mg corrosion, so that CO₂ decreases the Mg corrosion rate. This was shown by the well-controlled atmospheric exposure experiments [130] in which the corrosion rate (68 μg/cm²) of Mg in an atmosphere without CO₂ was higher than that (15 μg/cm²) in presence of CO₂, and moreover the corrosion rate in the presence of CO₂ decreased with exposure time, which suggested that the CO₂-containing air caused the formation of more-protective corrosion products that reduced the corrosion rate [132], which is in consistence with the results reported by Esmaily et al. [110] as shown in Fig. 5. Moreover, in the presence of CO₂ the corrosion was uniform, while it was localized in the absence of CO₂. The main corrosion product in the presence of CO₂ was a relatively thick layer of Mg hydroxy carbonate [130]. Likewise, the atmospheric corrosion rate of Mg alloys contaminated by NaCl can be decreased by CO₂. The corrosion rates of the Mg alloy exposed in a CO₂-free atmosphere containing NaCl were about three times greater than that in the absence of the NaCl [132]. For AZ91, the inhibiting effect of CO₂ was attributed to synergic effect of decreased pH in the surface electrolyte and the formation of a relatively protective carbonate-containing film. The decreased pH could lower the solubility of alumina and thus stabilized the Al-containing surface film [129]. A analogous study by Jönsson et al. [36] showed that at 95% relative humidity the main corrosion products were composed of hydromagnesite (Mg₅(CO₃)₄(OH)₂·4H₂O) and brucite (Mg(OH₂)).

The effect of carbonates can also be verified by the corrosion behavior of Mg in solution, in which the presence of CO₂ is mostly detrimental [127,133], because it is relatively more difficult in solution to the deposit Mg oxide/hydroxide and carbonate surface films. Qu et al. [133] found that the corrosion rate of AZ31B in a solution saturated with CO₂ was much higher than in the solution without CO₂ due to the increased conductivity and acidity of the solution.

4.6. Sulfur dioxide

Some of the trace atmospheric gases with a concentration of less than 10 ppm may nevertheless have an important effect on Mg atmospheric corrosion. For example, SO₂, a product of the combustion of sulfur-containing fossil fuels, has been considered important for Mg atmospheric environment [90]. SO₂ can accelerate the atmospheric corrosion of Mg alloys by combining with water to form a strongly corrosive H₂SO₃/H₂SO₄ electrolyte, and producing soluble Mg sulphite or sulphate [90,94]. The reactions of SO₂ can be expressed by the following equations [134]:

SO₂(g)+H₂O → HSO₃^-(aq)+H⁺(aq) (12)

HSO₃^-(aq)+OH^- → SO₃^2-+H₂O (13)

Mg(OH)₂(surface) + SO₂(g) → MgSO₃(ads) + H₂O (14)

Laboratory and field tests have shown that the Mg atmospheric corrosion rate increases with increasing SO₂ concentration [94,135], as shown in Fig. 6. Yang et al. [94] reported that the corrosion rate of AZ91D exposed in a polluted environment was approximately two times higher than those in unpolluted regions. Esmaily et al. [134] found that the corrosion of AZ91D increased in well-controlled laboratory exposure tests in the presence of SO₂ at only ppb levels. The main corrosion products were MgSO₃·6H₂O. An outdoor exposure study [90] showed that in the atmospheric environment polluted severely by SO₂ the dominated corrosion products were MgSO₃·6H2O and MgSO₄·6H₂O. In some cases, Mg₂Al₂(SO₄)₅·39H₂O could also be detected in the field [94].

4.7. Nitrogen oxides

NO and NO₂ mainly stem from high-temperature combustion in power plants and vehicles. Both NO and NO₂ can be readily dissolve in an aqueous film, and catalytically oxidized to form nitric acid which can corrode Mg to producing soluble nitrate compounds according to the following reactions [90]:

4NO+3O₂+2H₂O → 4HNO₃ (15)

3NO₂+H₂O → 2HNO₃+NO (16)

HNO₃+Mg+6H₂O → Mg(NO₃)₂·6H₂O+H₂ (17)

The literature [134,136] indicates that the presence of NO₂ strongly increases the Mg corrosion rate in the presence of SO₂. The corrosion products are primarily Mg sulphate, and remain unchanged even if NO₂ is present. Detailed studies on the mechanism by which NO and NO₂ accelerate Mg corrosion are currently lacking.

4.8. Dust particles

Dust particles commonly present in the atmosphere include the products of natural weathering of rocks and soil, emission of combustion engines and other artificial sources. The dust particles can absorb moisture and conglomerate on metal surfaces and accelerate the corrosion.

Dust particles can accelerate Mg corrosion by decreasing the critical relative humidity for the formation of a liquid water film. The size of dust particles influences the pit size on Mg alloys [90]. Chen et al. [124] found that dust particles stimulate the formation of micro-droplets. The hydrophilicity decreases in the following order: salt particle > dust particle > AlMn phase > other phases (e.g. primary α-Mg and eutectic Mg phase). In a marine environment, the air borne particles are mainly soluble particles, such as sea salt [13] containing chlorides that accelerate Mg corrosion.

4.9. Other factors

The ultraviolent (UV) radiation has usually been used to accelerate the degradation of organic coatings in order to rapidly evaluate their protection performance [137]. The influence on the corrosion behavior of metals has not been widely reported [138,139], and there is no investigation on the effect of UV radiation on the atmospheric corrosion of Mg alloy. The influence of stress on the atmospheric corrosion of Mg has been widely studied under the topic of “stress corrosion cracking of Mg” [[140], [141], [142]], which has a totally different behavior and mechanism. It is not included in this review.

5. Corrosion products

Laboratory investigations [130] indicated that the corrosion products of Mg exposed to humid air, in the presence of CO₂, consisted mainly of hydrated Mg hydroxy carbonate Mg₂(OH)₂CO₃·3H₂O, whereas the dominated corrosion product in the absence of CO₂ was crystalline Mg(OH)₂ (brucite). Pitting can occur with the less protective Mg hydroxide film. The relatively thicker carbonated film was more protective, and thus reduced the Mg corrosion [130]. In the presence of NaCl and 350 ppm CO₂, the corrosion products on AZ91D were a hydrated Mg hydroxyl carbonate, hydromagnesite (Mg₅(CO₃)₄(OH)₂·4H₂O) [36]. A similar finding was reported by Lindström et al. [129]. Feliu et al [29] found that Mg-Al alloys had a larger amounts of Mg carbonates on the surface, and thus a lower corrosion rate in humid air [29] possibly attributed to the formation of layered double hydroxides (LDHs).

Different outdoor atmospheric environments produce different corrosion products. Jönsson et al. [35], in field studies of AZ91D and AM50 exposed in urban and rural regions in Sweden and a marine field in France, found that (Mg₅(CO₃)₄(OH)₂·4H₂O) was the primary corrosion product at all sites, and the corrosion rate increased approximately linearly with exposure time, which suggested that the corrosion products were not protective [35]. Liao et al. [28] proposed that the increasing corrosion rate with exposure time was partially due to the increased size and/or number of cracks in the corrosion products, which led to easy penetration of rainwater and pollutants through the layer to the Mg alloy matrix and thereby accelerated the corrosion. However, Cui et al. [33] reported that, in the tropical marine atmospheric environment at Xisha Islands, the primary corrosion products were Mg₅(CO₃)₄(OH)₂·xH₂O, and slightly slowed the corrosion rate by restraining both the anodic and cathodic processes. The results were supported by Wang et al. [143]. Cui et al. [33] found that AZ31 exposed to marine atmospheric environment for 24 months had the best barrier effect against further corrosion attacks. In a polluted environment, sulphite, sulphate, and nitrate corrosion products can be formed on the Mg surface [90]. Mg carbonate and ((Mg_0.833Al_0.167)(OH)₂(CO₃)_0.083·0.75H₂O) was also detected on the surfaces of ten different Mg alloys after 3 years of field-exposure [28]. Yang et al. [94] suggested that when the Al ion concentration in the surface electrolyte increased and reached a threshold value, Mg₂Al₂(SO₄)₅·39H₂O formed on the surface of AZ91D.

Jönsson et al. [35] found that Mg(OH)₂ was absent in the field-exposed samples. However, in some recent studies, Mg(OH)₂ was the main corrosion product found on Mg alloy surface in field tests [94,144]. One possible reason for the deviation could be the presence of hydromagnesite that facilitated the formation of brucite.

These studies indicate that the corrosion product layer has a partially beneficial effect on the corrosion performance of Mg alloys, and the protectiveness of the corrosion products depends on chemical composition, structure, adherence, solubility, compactness, atmospheric environment and exposure time. However, the mechanism of the dependence of the protectiveness on those atmospheric environment parameters is still unclear, which needs further investigation.

6. Tests of atmospheric corrosion

6.1. Field exposure

Field exposure experiments provide actual information on the atmospheric corrosion behavior, which is the most important method for atmospheric corrosion studies. Cui et al. [33] found that the corrosion of AZ31 in a marine atmosphere increased with the chloride deposition rate. The corrosion products formed on the specimens weathered for 24 months suppressed further corrosion attacks. Liao et al. [89] found that the corrosion rates of AZ31 in marine environments were higher than in urban because of the presence of far higher amounts of airborne sea salt, which is not found in indoor testing. However, field exposures are limited and suffer from complicated influences of the environmental parameters, variable corrosion conditions, time-consuming experiment durations and the difficulty of control and bad reproducibility. Hence, acceleration tests under controlled conditions in the laboratory are widely conducted to evaluate Mg corrosion in a relatively short period of time. In laboratory tests, the accuracy can be controlled and each factor can be studied separately, which is essential to the understanding of corrosion mechanisms.

6.2. Salt spray

The salt spray test is a common test, which somewhat simulates marine atmospheric corrosion and greatly accelerates the corrosion rate. The most widely used salt spray test follows ASTM B-117. According to ASTM B-117 [145], samples are exposed to a neutral 3.5% NaCl solution at 35 °C. However, the test results do not correspond well to actual Mg atmospheric corrosion [146]. Hence, some different wet-dry cyclic salt spray tests have been developed, to better reproduce the key characteristics of the atmospheric environment, wherein RH and temperature are the most important factors. For example, one typical wet-dry cyclic testing standard is Volkswagen PV1210 [147] and consists of the following in each cycle: firstly 4 h salt spray fog at 35 °C, then 4 h at 50% RH and 23 °C, and finally 16 h at 100% RH and 40 °C. Although these cyclic tests appear to be more reasonable compared to a constant salt spray test, there is still an issue with correlation of the test results with actual Mg corrosion behavior in atmospheric environments.

6.3. Laboratory atmospheric test

Apart from the overall atmospheric environment controlled in standard tests, some parameters critically influence atmospheric corrosion and are carefully controlled in the lab.

6.3.1. Thin electrolyte layer

Wet atmospheric corrosion in nature is a result of electrochemical processes of a metal under a thin electrolyte layer on the metal surface. The thin electrolyte layer can be a continuous film or discontinuous droplets. There have been numerous studies on the corrosion of metals under a thin electrolyte investigated by electrochemical methods since the 1950s [148,149]. The big challenge of applying electrochemical techniques in such a thin electrolyte film is the placement of a reference electrode and/or counter electrode in such a thin liquid film/droplet and the continuity of the liquid film between the metal, reference and counter electrodes. Also, the ohmic drop between the working and reference electrodes during electrochemical measurements is another problem.

The Volta potential measurement using a Scanning Kelvin Probe (SKP) is an powerful technique, which can be correlated to the corrosion potential of the metal underneath a thin liquid film, but the oscillation of the scanning Kelvin probe tip may transport more oxygen and carbon dioxide to the metal surface [150,151]. Zhang et al. [151] investigated the corrosion behavior of Mg under thin electrolyte using a device to sense the thin electrolyte film. Liu et al. [152] investigated the corrosion behavior of AM60 with rare earth (RE) addition under a thin electrolyte [151], in which the thickness of the thin electrolyte layer was also carefully measured. These liquid film thickness detection and control methods are helpful in understanding the atmospheric corrosion behavior of Mg alloys.

6.3.2. Relative humidity (RH)

Relative humidity (RH) is an important parameter influencing the atmospheric corrosion as it affects the liquid film thickness in the atmospheric environment and the aggressiveness of the medium under a dry atmospheric condition. If Mg alloy surfaces are contaminated by NaCl (typically 14, 70, 140 μg/cm²), their corrosion will be accelerated in RH controlled atmospheric environments [146]. The RH can be regulated using saturated salt solutions [153]. The dependence of relative humidity in air on salt solution in a closed chamber is listed in Table 3:

Using this technique, LeBozec et al [146] found that the corrosion rate of AZ91D and AM50 exposed in increased when the RH inceased from 75% to 95%.

6.3.3. Temperature

Temperature control at temperatures above 0℃ is relatively easy and Mg corrosion for these conditions is relatively well studied [41]. However, there is significant application of Mg alloys used in the exterior of cars below 0℃ during winter. Esmaily et al [110] designed a corrosion system as shown in Fig. 7 which used ethylene glycol as antifreeze to perform corrosion test at sub-zero temperatures and revealed that the corrosion behavior of AM50 at -4℃ was quite different with that at room temperature (see section 4.2). Furthermore, Esmaily et al [45] indicated that with minor modification, the set-up can be used in a wider range from -20 to 60℃.

6.3.4. Laboratory atmosphere

More realistic atmospheres in the laboratory can be simulated using complicated setups. For example, to control the laboratory atmosphere [130], a wide range of trace gases such as SO₂, NO₂, H₂S and O₃ can be added in the test chamber using N₂ as a carrier.

7. Measurement of corrosion rate

Many techniques have been used to study atmospheric corrosion. They include traditional gravimetric measurements, electrochemical methods and surface analyses.

7.1. Gravimetric measurements

Reaction (4) indicates that the corrosion of Mg leads to a weight gain if all the corrosion products stay on the surface after corrosion, and a weight loss after the corrosion products are removed. However, in aqueous corrosion not all the corrosion products stay on the specimen surface [51].

The experimental errors introduced in the gravimetric measurement can result from the loss of corrosion products before weight gain detection or the incomplete or over removal of corrosion products during weight loss detection. A solution containing chromium trioxide and silver chloride has been often used to dissolve the corrosion products without attacking the un-corroded metallic Mg at room temperature [[154], [155], [156]].

The gravimetric method has been widely used to measure Mg atmospheric corrosion, especially in field exposures and salt spray tests. For example, Jönsson et al. [35] measured the mass loss of AZ91D and AM50 exposed to three different field-exposure locations and revealed that the weight loss of the Mg alloys was linear with time, and that the highest corrosion rate was measured in the marine environment.

The advantages of the method is that the method is reliable and direct [35,62]. However, limitations include deviation from the real loss mass caused by the detachment of loose corrosion products during the exposure and excessive or incomplete elimination after the pickling [44].

7.2. Electrochemical measurement

Electrochemical measurement is an effective approach to investigate the corrosion behavior of Mg alloys.

7.2.1. Polarization curves

The typical polarization curve of a metal consists of cathodic and anodic polarization branches. This curve allows determination of the free corrosion potential and the corrosion rate. However, it should be noted that the corrosion rates extrapolated from the polarization curve is questionable for Mg or Mg alloys due to anodic hydrogen evolution [15,157].

Due to the difficulty in forming an electrolyte cell on the metal surface for an atmospheric condition, direct measurements of polarization curves of Mg alloys have not been widely reported. There are only some polarization curves indirectly performed after atmospheric corrosion. For example, Cui et al. [33] measured polarization curves of AZ31 after field-exposure from 1 to 24 months, which indicated that the current densities of the anodic branch decreased with exposure time, and the current densities of the cathodic branch had also a decreasing trend after the field-exposure.

There also some polarization curves directly measured under thin electrolyte layer to simulate the atmospheric conditions for Mg alloys as shown in Fig. 8 [151,152]. The corrosion behavior of pure Mg under thin electrolyte layers has been studied by means of cathodic polarization using a setup as shown in Fig. 8 [144]. The results show that the cathodic process of pure Mg under thin electrolyte was dominated by hydrogen evolution. The cathodic process was inhibited slightly while the anodic process was retarded significantly as the thin electrolyte layer thickness increased. Liu et al. [152] also used the setup to study the corrosion behavior of AM60 Mg alloys containing Ce or La under thin electrolyte layers and found that corrosion resistance was enhanced by decreasing the thin electrolyte layer thickness. So far, this setup has produced some meaningful results for Mg under thin electrolyte layer. However, due to the surface tension of the liquid layer, the thickness must be very carefully measured and analyzed

7.2.2. Electrochemical impedance spectroscopy (EIS)

EIS can be utilized to study the corrosion mechanisms and corrosion properties of Mg alloys. For example, Cao et al. [51] used the EIS to investigate the development of the corrosion product film on Mg in 3.5 wt.% NaCl solution saturated with Mg(OH)₂. The corrosion rate of Mg alloy estimated from EIS is not accurate, because the Stern-Geary relationship is not strictly followed due to the involvement of anodic hydrogen evolution process or the negative difference effect in the corrosion of Mg [65].

The use of EIS to study Mg atmospheric corrosion is rare for the same difficulties as for the polarization curve measurements under atmospheric corrosion conditions, but indirect EIS measurements Mg atmospheric corrosion are quite popular [33,64,158]. An in-situ EIS technique to study the atmospheric corrosion of other metals can be introduced to study Mg atmospheric corrosion as shown in Fig. 9 [159].

7.3. Surface analytical methods

Under atmospheric conditions where electrochemical techniques are difficult to employ to investigate corrosion reactions, surface analytical methods are important. Many such techniques have been used in investigating Mg atmospheric corrosion.

7.3.1. Corrosion morphology

The corrosion morphology with or without corrosion products is important to understanding the Mg atmospheric corrosion behavior. Current techniques include confocal microscopy (CM), scanning electron microscopy (SEM) and transmission electron microscope (TEM).

Confocal microscopy (CM), a development of optical microscopy (OM), uses a laser beam and two pinhole apertures to obtain a three-dimensional image of the surface of a sample. Jönsson et al. [62] used this technique to study the corrosion damage for AZ91D exposed in an atmospheric environment for two weeks. SEM and TEM are also used to characterize the microstructure and surface topography of the surface after atmospheric corrosion [1]. SEM and TEM have better resolution and depth of field than that OM (including CM). TEM is useful for determining the crystal structure, crystallographic orientation and chemical composition of corrosion layers.

7.3.2. Corrosion products

For a valid indoor accelerating test, the corrosion products formed on the specimens have to be similar to those on the field exposure specimens. Various techniques are used to analyze the corrosion products, such as X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), Fourier Transform Infrared Spectroscopy (FTIR), auger electron spectroscopy (AES), energy dispersive spectrometer (EDS) and scanning Kelvin probe-force microscopy (SKPFM).

XRD allows identification and determination of the relative quantity of phases in corrosion products.

XPS provides detailed information on both elemental and chemical composition of the surface. Cui et al. [33] found the main corrosion products on AZ31 after weathering in a marine atmosphere were composed of Mg, C, O and Al by XPS analysis and the concentration of CO₃^2- in the outer layer was $\widetilde{2}$6%, and was higher than that of the inner layer ($\widetilde{1}$6%).

The Fourier Transform Infrared Spectroscopy (FTIR) has been used in identifying Mg atmospheric corrosion products under ambient conditions such as in-situ atmospheric corrosion. The FTIR spectroscopy of AZ91D exposed in an industrial environment for 12 months [94] showed that the dominated corrosion products consist of Mg(OH)₂, MgCO₃, and Mg₂Al₂(SO₄)₅·39H₂O for both the ingot and high pressure die-cast samples.

AES can also provide both the elemental and chemical composition from the surface layer of Mg. Moreover, AES has a higher lateral resolution than XPS. Jönsson et al [34] measured the composition of the outermost surface layers of AZ91D Mg alloy using AES, and found that in the different phases there was an Al enrichment in the oxide layer and that this enrichment decreased in the order of β phase > eutectic α-/β phase > α phase.

EDS is often combined with SEM to investigate the element composition of the corrosion product. EDS can analyze any interested area (point, line or mapping).

SKPFM measures the potential difference between the probe and the surface of sample. This method determines the surface Volta potential of different phases in an Mg alloy. Arrabal et al. [66] measured the surface potential using SKPFM and found that Al-Mn or Al-Mn-Nd inclusions were more cathodic than the surrounding α-Mg matrix.

8. Concluding remarks

As a green engineering material, Mg alloys have been found in a variety of innovative applications. However, their further industrial uses are limited due to their unsatisfactory corrosion properties. Therefore, corrosion resistance is currently a critical issue for Mg alloys. In the recent decades, while a large number of investigations have been concentrated on the corrosion of Mg alloys in bulk solutions, their atmospheric corrosion, including the corrosion behavior, mechanisms, influencing factors, and methods of detecting the corrosion properties has been increasingly concerned by more and more engineers, because Mg alloys are rarely used in solutions directly and atmospheric corrosion is in fact more relevant to the damage that Mg alloys may suffer in their service conditions.

This review of Mg atmospheric corrosion indicates the following:

1 Atmospheric corrosion is induced by the presence of a thin aqueous layer on the metal surface. Depending on the relative humidity and amounts of water-soluble species, the thin aqueous films or droplets can be considerably different in thickness, and thus differently affecting the atmospheric corrosion of Mg alloys.

2 The main factors that can mechanistically influence the atmospheric corrosion of Mg alloys are microgalvanic acceleration between different phase, corrosion protection by a continuous second phase distribution, the protection of corrosion products, and the degradation of protective layers by aggressive species such as chloride ions. However, due to the poor conductivity and small thickness of the electrolyte layer under atmospheric conditions, the influence of composition and microstructure of the substrate alloy on the atmospheric corrosion performance can be substantially different from that in bulk solutions.

3 The Mg atmospheric corrosion rate typically increases as a function of temperature, relative humidity, and concentrations of aggressive species, such as chloride and sulfate ions.

4 O_2, SO₂ and NO₂ accelerate atmospheric corrosion rate, whereas the corrosion rate is decreased by CO₂. Mg carbonate is a primary corrosion product formed on the surfaces, which has better protectiveness than the Mg hydroxide. The sulphite, sulphate and nitrate corrosion products formed in a polluted environment are less protectiveness.

5 The dust particles can accelerate the corrosion process of Mg through declining the local critical relative humidity on the dust-contaminated surface

6 The traditional gravimetric can evaluate effectively the corrosion behavior of Mg alloys effectively. Some modern material analyses, such as the CM, XPS, FTIR and SKPFM are very useful in understanding the corrosion mechanism. More advanced techniques are needed in the study of atmospheric corrosion of Mg alloys.

9. Outlook and prospective

The above review indicates that there are a few critical issues regarding the atmospheric corrosion of Mg alloys not properly addressed yet. A deep insight into the corrosion behavior of Mg, such as the negative difference effect, under atmospheric conditions will be the first important research topic in the corrosion science of Mg alloys.

Secondly, since atmospheric factors, such as the temperature, relative humidity, gas composition and pollutants in air, can significantly affect the corrosion of Mg alloys, a mechanistic study on the roles of these changing parameters in atmospheric corrosion will be a scientifically interesting topic. Identification of a dominating factor for atmospheric corrosion of Mg (e.g., the effect of oxygen on Mg corrosion which is normally overlooked under immersion conditions) could be a starting point in the research field.

On top of the influences of the substrate and environment on the atmospheric corrosion of Mg alloys, the detailed chemical composition and microstructure of the corrosion product layers formed on the Mg alloy surfaces under atmospheric conditions could be the third critical research topic. Clarification of the surface films, as well as their roles in atmospheric corrosion, will build a bridge between the substrate alloy and the environment, which could be a key to the comprehensive understanding of Mg atmospheric corrosion.

Atmospheric corrosion is a special electrochemical process, in which traditional electrochemical techniques have difficulty being used to characterize the corrosion damage. The lack of effective methods that can reliably and accurately obtain the electrochemical information from corroding Mg surface under atmospheric conditions is currently one of the crucial obstacles in atmospheric corrosion studies. Developing innovative techniques will lay a foundation for gaining insight into the atmospheric corrosion of Mg. This should be another interesting and hot topic in future work.

Based on the above, the following research topics may be prioritized for the atmospheric corrosion of Mg and its alloys in the future:

1 The role of oxygen in the cathodic reaction on Mg under atmospheric conditions.

2 The anodic dissolution behavior and negative difference effect during the atmospheric corrosion of Mg.

3 The influence of atmospheric parameters and alloy factors on the atmospheric corrosion of Mg.

4 The composition and microstructure of corrosion products and surface films formed on Mg alloys under atmospheric conditions, their dependence on alloy and environmental factors, and their roles in the corrosion.

5 The development of innovative electrochemical theories and techniques for atmospheric corrosion.

Acknowledgements

This research was supported by the National Natural Science Foundation of China (No. 51731008) and the National Environmental Corrosion Platform of China.

The authors have declared that no competing interests exist.

---