Corrosivity of haze constituents to pure Mg

Chen Zho, Fuyong Co, Gung-Ling Song,b,?

a Center for Marine Materials Corrosion and Protection, College of Materials, Xiamen University, 422 S. Siming Rd., Xiamen 361005, China

b State Key Laboratory of Physical Chemistry of Solid Surfaces, Xiamen University, 422 S. Siming Rd., Xiamen 361005, China

Abstract The corrosion behavior of pure Magnesium (Mg) in a Mg(OH)2-saturated solution containing different individual constituents of PM2.5 in haze were studied by hydrogen evolution, weight loss and electrochemical experiments. The results indicated that the corrosivity of these constituents to pure Mg decreased in the following order: (NH4)2SO4>Haze-contaminated-solution>NH4NO3>NH4Cl>NaCl≈KCl≈Na2SO4 ≈MgCl2 ≈CaSO4>Mg(OH)2 (basic solution) >Ca(NO3)2. Possible mechanisms behind the different corrosion behaviors of Mg in response to these constituents were also briefly discussed in this paper.? 2020 Published by Elsevier B.V. on behalf of Chongqing University.This is an open access article under the CC BY-NC-ND license. (http://creativecommons.org/licenses/by-nc-nd/4.0/)Peer review under responsibility of Chongqing University

Keywords: Haze (PM2.5); Mg; Corrosion.

1. Introduction

In recent years, a large number of emissions, such as automobile exhaust and industrial pollutants, have been polluting the air seriously. Haze, as a kind of severely polluted atmosphere, has been greatly concerned. It is generally accepted that haze is composed of PM10and PM2.5(particle matters with aerodynamic diameters smaller than 10μm and 2.5μm,respectively) [1,2]. Compared with large particle suspensions in the air, PM2.5has a large specific surface area and high activity, which can keep in motion in the air for a long time,threatening the human health seriously.According to literature[2–6], the main constituents of PM2.5contain inorganic salts,such as water-soluble ions (,,), organic hydrocarbons and microorganisms. Among these constituents,inorganic salt can easily deposit on the metal surfaces to form electrolytes after absorbing water in the air, which may accelerate the corrosion of metals considerably.

Mg alloy could be widely used in aerospace, automotive,electronics and other fields due to their low density and high strength to weight ratio [7,8]. However, their poor corrosion resistance limits their practical applications and many efforts have been made to improve it [9–12]. So far, Mg alloys have been made into some components, such as wheels and bumpers in automobiles and gearboxes and canopy covers in aircrafts, which are serviced in atmospheric environments[13–15]. Therefore, the polluting particles in the atmosphere could cause serious corrosion damage to these Mg components. It is important to study the effects of the constituents of haze on the corrosion behavior of Mg alloys.

There are some reports about the corrosion behavior of Mg alloys in solutions containing one or several constituents of haze. For example, it is well known that the presence of Cl?can destroy the surface film, causing serious localized corrosion of Mg [16,17]. Yang et al. [18] found thatwas less corrosive than Cl?, and the corrosion product films of AZ91D after immersion in Na2SO4and MgSO4solutions were compact, which could effectively prevent the propagation of corrosion of the substrate. Buggio et al. [19] indicated that the presence ofin solution altered the dissociation equilibrium of H2O, which could accelerate the corrosion of Mg by formation of hydride phases. Cui et al. [20] studied the corrosion behavior of AZ31 in NaCl solution containing NH4NO3and found that the presence ofinhibited the precipitation of Mg(OH)2and promoted the reduction ofaccelerating the corrosion of AZ31. Cui et al.[21] studied the effects of different ions in a simulate haze solution containing Na2SO4, NaNO3and NH4Cl on the corrosion behavior of AZ91D. Their results indicated that the presence ofand Cl?initiated the pitting corrosion of AZ91D, but the adsorption ofprotect the specimens by generating a passive film on the surface, while the consumption of OH?by the hydrolysis ofretarded the formation of Mg(OH)2, and therefore accelerated the corrosion process drastically.

As summarized above, the relevant investigations so far have only been focused on the influence of a few individual haze constituents on the corrosion behavior of Mg. Their effects have not been systematically compared. In this study,the reported compositions of PM2.5in haze were summarized according to the research data of haze composition for each city all over the world. Furthermore, the individual influence of the main haze constituents on the corrosion behavior of pure Mg was systematically studied.

2. Experimental methods

2.1. The typical composition of PM2.5 in haze solution

According to [2–6], the main inorganic salts of haze are sulfate ((NH4)2SO4, CaSO4, Na2SO4), chloride (NaCl, KCl,MgCl2,NH4Cl),nitrate(NH4NO3,Ca(NO3)2),carbon(burned coal), SiO2(dust)). They are mainly derived from the automobile exhaust, industrial waste, coal combustion, seawater evaporation, and dust. Among them, only the soluble chemicals were studied as the main constituents of haze in this study, and those insoluble were not considered. The intermediate values of these soluble haze constituents reported in[1–6,22–27] were set as their typical levels in haze.

The concentrations of the haze constituents in water were determined as follows: According to their normal concentrations in haze, the ratios of these constituents were first obtained. Since CaSO4·2H2O has the lowest solubility (2g/L),its normal concentration in water at the solubility limit was set as a benchmark. The normal concentrations of the other constituents in the water were calculated according to their ratios using the solubility concentration of CaSO4·2H2O. This ensured that the other constituents would not exceed their solubility limits, resulting in deposition in solution. Therefore,the composition of haze contaminated water could be generated as shown in Table 1.

It should be noted that in practice, different constituents of haze have different solubility levels and dissolution rates in water. Their normal concentration ratios in water cannot be the same as those in air, and can be influenced by many practical factors. The concentrations of the constituents of haze listed in Table 1 were simply for comparison in this study.

Table 1 Concentrations of constituents in haze-contaminated solution and their corresponding mass percentages [1-6,16-21].

2.2. Materials and specimens

An extruded pure Mg plate was used and the composition was measured by iCAP 7400 optical emission spectroscopy(ICP-OES)from Thermo Scientific as the following(wt%):Al 0.015, Zn 0.006, Mn 0.017, Si 0.018, Fe 0.0028, Cu 0.0028,Ni 0.0004, and Mg balance. In order to reduce the texture of the plate, the Mg plate was heat-treated at 300°C for 3h and cooled in the furnace with the flowing argon as protective atmosphere. The plate was then cut into 1cm×1cm×1cm coupons for immersion test.The electrode specimens for electrochemical tests were joined with copper wires and mounted in epoxy resin with an exposed area of about 1.0 cm2. The surfaces of Mg coupons for immersion test were ground to 2000 grit successively with SiC papers, rinsed with distilled water and dried with cold and warm air. All the experiments were performed at room temperature of 25±1 °C if not specified.

2.3. Solution preparation

The following solutions were prepared:

(i) Basic solution: a Mg(OH)2saturated solution was prepared. This basic solution simulated the liquid film on corroded Mg surface containing saturated Mg(OH)2.

(ii) Haze-constituent-contaminated solutions: individual at their concentrations respectively listed in Table 1 were added into the Mg(OH)2saturated basic solution, and the resulting solutions contained different levels of the haze pollutants. The solutions also included hazecontaminated solution which contained all the constituents at the specified concentrations.

(iii) Equal normal solutions: the same normal concentration(0.0232N) of each individual PM2.5constituent listed in Table 1 was added into the Mg(OH)2saturated basic solution.

These solutions were prepared with distilled water and A.R grade reagents. To eliminate the influence of pH, the pHs of all the solutions were adjusted to 10±0.3 by 0.1M NaOH solution to simulate the surface alkalinity of Mg in atmosphere,as the dissolution of Mg can result in an instantly increased pH of the droplets on the Mg surface to 10.3 [28,29].

Fig. 1. Hydrogen evolution curves of pure Mg immersed in basic Mg(OH)2 saturated solution, haze constituent contaminated solutions [Mg(OH)2+X(pH=10±0.3): X=(NH4)2SO4, NH4NO3, NaCl, KCl, Na2SO4, MgCl2, CaSO4, NH4Cl, Ca(NO3)2], and haze-contaminated solution for 72 h:(a) all the curves, and (b) details of some curves in (a).

2.4. Immersion test

In hydrogen evolution measurement, the evolved hydrogen was collected for three days during immersion. At least three parallel specimens were used in each solution. The hydrogen collection setup was the same as that described in the literature [30]. The funnel over the specimen collected all the hydrogen bubbles from the specimen into a burette which was vertically mounted above the funnel. The burette was initially full of the test solution,which would be replaced by the evolved hydrogen bubbles during the immersion test.Therefore, the hydrogen volume could be measured directly by reading the position of the liquid level.

During weight loss measurement, the specimen was weighed before and after 3 days of immersion. The corrosion rate was calculated in the same way as that described in previous studies [16,31]. The corrosion products formed on Mg surface were removed by 200g/L CrO3+2g/L AgNO3solution after immersion test[16].Typical corrosion morphologies of Mg specimens with and without corrosion products were observed by optical microscopy (Leica DVM6), scanning electron microscopy (TM3000) with an electron acceleration voltage of 15kV. X-ray diffraction (TD-3500) and energy dispersive spectrometer(SU-70)equipped on the SEM were used to analyze the composition of the corrosion products.

2.5. Electrochemical measurements

Before electrochemical measurement,all the Mg electrodes were immersed in the solutions for 1h to stabilize their open circuit potentials (OCPs). The electrochemical impedance spectroscopy (EIS) spectra and polarization curves were measured with standard three-electrode electrolyte system. The reference electrode was Ag/AgCl/ Sat. KCl and the counter electrode is a Pt mesh.

In EIS measurement, the frequency range was 100kHz –30 mHz, and the sinusoid potential perturbation had an AC amplitude of 5mV. The polarization resistance, RP, was obtained as the resistance at the lowest frequency after deduction of the solution resistance.After EIS measurement,potentiodynamic polarization was carried out from ?0.3V to +0.5V vs.the OCP at scanning rate 1mV/s. The corrosion current density, icorr, from the polarization curve was obtained by Tafel fitting as detailed described in [16].

3. Results

3.1. Corrosion in haze constituent contaminated solutions

Fig. 1 shows the hydrogen evolution curves of pure Mg during 72 h of immersion in basic solution saturated Mg(OH)2, haze-constituent-contaminated solutions and hazecontaminated solution. The hydrogen evolution rate of pure Mg in the (NH4)2SO4contaminated solution was substantially higher than those in the other constituent-contaminated solutions, even much higher than that in the solution contaminated by all the haze pollutants. The hydrogen evolution rate in Ca(NO3)2solution was the lowest of all, which was even lower than that of Mg in the basic Mg(OH)2solution. Except for in (NH4)2SO4solution and haze contaminated solution,the hydrogen evolution rates of specimens in other solutions were relatively low and their differences were small as shown in Fig. 1(b). Table 2 lists the corresponding corrosion rates calculated from weight loss of Mg specimens. Except for the NH4NO3solution, the corrosivity of other constituents revealed by weight loss is consistent with that revealed by hydrogen evolution rates.In the NH4NO3contaminated solution,the corrosion rate calculated from weight loss is much higher than that from hydrogen evolution measurement.According to the corrosion rates, the PM2.5constituents could be classified into the following 4 groups: 1) ammonium salts ((NH4)2SO4,NH4NO3, NH4Cl), 2) chlorides (NaCl, KCl, MgCl2),3) sulfates (CaSO4, Na2SO4), and 4) Ca(NO3)2. Fig. 1 and Table 2 both show that the corrosion rates of Mg in the ammonium salts containing solutions are in a relatively high level, and those in the chloride salt containing solutions are closed to those in the sulfate solutions. Therefore, the corrosivity of these constituents to pure Mg could be arranged in the following order: (NH4)2SO4>Haze-contaminated-solution>NH4NO3>NH4Cl, chloride≈ sulfate, Mg(OH)2(basic solution)>Ca(NO3)2.

Fig. 2. EIS Nyquist plots of pure Mg in basic Mg(OH)2 saturated solution, haze constituent contaminated solutions [Mg(OH)2+X (pH=10±0.3):X=(NH4)2SO4, NH4NO3, KCl, MgCl2, CaSO4, NH4Cl, Ca(NO3)2], and haze-contaminated solution: (a) all the curves, and (b) details of some curves in (a).

Table 2 Corrosion rates calculated from weight loss of Mg specimens immersed in basic Mg(OH)2 saturated solution, haze constituent contaminated solutions [Mg(OH)2+X (pH=10±0.3): X=(NH4)2SO4, NH4NO3, NaCl, KCl,Na2SO4, MgCl2, CaSO4, NH4Cl, Ca(NO3)2], and haze-contaminated solution for 72h.

Fig. 2 shows the EISs of pure Mg in basic solution saturated Mg(OH)2,haze- constituent-contaminated solutions and haze-contaminated solution. It is noted that the EISs of Mg in the NaCl and Na2SO4contaminated solutions were not shown, because they were similar to those in the KCl and CaSO4contaminated solutions. All the EIS spectra of the Mg specimens in the basic Mg(OH)2saturated solution and the haze-constituent-contaminated solutions have two capacitive arcs:one at high frequencies and the other at low frequencies,as illustrated by the schematic Nyquist plots in Fig. 3(a).The corresponding equivalent circuit is shown in Fig. 3(b),where Rsis the solution resistance, Rfis the film resistance,Rais the pseudo resistance for the electrochemical reaction at the film/Mg interface, Cfis the surface film capacitance,and Cais the pseudo capacitance for the electrochemical reaction at film/Mg interface [32]. The polarization resistance Rp, is the resistance at the lowest frequency after deduction of the solution resistance, which is approximately equal to the sum of Rfand Ra. The polarization resistance Rpof each individual constituent or haze contaminated solution was calculated as shown in Table 3. In (NH4)2SO4solution, the Rpvalue was the smallest, which was followed by that in the haze contaminated solution. The Rpvalues of the specimens in MgCl2and in Mg(OH)2were larger than those in the other constituents and haze contaminated solutions. The influences of most constituents on Mg corrosion indicated by RPvalues were basically consistent with those indicated by hydrogen evolution rates and weight loss data (except for MgCl2).

Fig. 3. (a) The schematic theoretical Nyquist plots. HF=high frequency range; LF=low frequency range; (b) Equivalent circuit model [26].

Fig. 4. (a) Polarization curves of pure Mg in basic Mg(OH)2 saturated solution, haze constituent contaminated solutions [Mg(OH)2+X (pH=10±0.3):X=(NH4)2SO4, NH4NO3, KCl, MgCl2, CaSO4, NH4Cl, Ca(NO3)2], and haze-contaminated solution, and (b) curve-fitted corrosion current densities icorrs.

Table 3 Curve-fitted parameters of the EISs and polarization curves of Mg specimens in basic Mg(OH)2 saturated solution and haze constituent contaminated solutions [Mg(OH)2+X (pH=10±0.3): X=(NH4)2SO4, NH4NO3, KCl, MgCl2,CaSO4, NH4Cl, Ca(NO3)2], and haze-contaminated solution.

Fig. 4(a) shows the potentiodynamic polarization curves of pure Mg in basic solution saturated Mg(OH)2, hazeconstituent-contaminated solutions and haze-contaminated solution. The polarization curves of NaCl and Na2SO4were not shown, as they were similar to those in KCl and CaSO4.The corrosion current density, icorr,was obtained from polarization curves through Tafel fitting. The icorrof pure Mg in (NH4)2SO4was the highest, followed by that in the haze contaminated solution and NH4NO3solution as shown in Fig. 4(b). The magnified image in Fig. 4(b) shows that the icorrs of the specimens in the other constituents contaminated solutions were low,in which the icorrs of Ca(NO3)2and MgCl2were the lowest two. The influences of haze constituents indicated by the polarization curves and by hydrogen evolution rates as well as weight loss data were generally consistent,except for the NH4NO3and MgCl2solutions.

Fig. 5. Surface morphologies of pure Mg after 14 days immersion in haze constituent contaminated solutions (pH=10±0.3): (a) Mg(OH)2+ (NH4)2SO4, (b)Mg(OH)2+NH4NO3, (c) Mg(OH)2+NaCl, (d) Mg(OH)2+KCl, (e) Mg(OH)2+Na2SO4, (f) Mg(OH)2+MgCl2, (g) Mg(OH)2+CaSO4, (h) Mg(OH)2+NH4Cl,and (i) Mg(OH)2+Ca(NO3)2.

In summary, according to the hydrogen evolution curves(Fig. 1), weight loss data (Table 2) and the electrochemical data (Figs. 2 and 4), the corrosion rates of pure Mg in basic solution saturated Mg(OH)2,haze-constituentcontaminated solutions and haze-contaminated solution could be ordered as the following: (NH4)2SO4>Haze-contaminatedsolution>NH4NO3>NH4Cl>NaCl≈KCl≈Na2SO4≈MgCl2≈CaSO4>Mg(OH)2(basic solution)>Ca(NO3)2.

Fig. 5 shows the corrosion morphologies of pure Mg after 14 days of immersion in haze-constituent-contaminated solutions. The surfaces of the corroded Mg were covered with various thick cracked corrosion products. The corrosion product layer in the (NH4)2SO4contaminated solution was obviously thicker and more seriously cracked than those in the other solutions. In CaSO4solution, the corrosion products were needle-like and elliptical, while those in Ca(NO3)2solution were cubic.

Fig. 6 shows the corresponding optical corrosion morphologies of pure Mg after removal of corrosion products.There are evident corrosion pits on the surfaces of all the Mg specimens. In the ammonium and chloride containing solutions, some circular corrosion cavities were distributed over the whole surface. In the sulfate containing solution,the corrosion pits were distributed in the strips. While in the Ca(NO3)2solution, the corrosion pits were relative small and discrete.

Table 4 Chemical composition of the corrosion products on the surfaces of Mg specimens after 14 days of immersion in CaSO4 and Ca(NO3)2 contaminated solutions (pH=10±0.3) in wt% determined by EDS.

Fig. 7 shows the corresponding SEM images of the corroded morphologies,which is consistent with Fig.6.The pure Mg in the ammonium containing solutions suffered the most serious corrosion with deep cavities on the Mg surface. The corrosion of Mg in the chloride and Na2SO4containing solutions was relatively milder with shallower cavities, while in CaSO4solution, there are corrosion pits on the local small area and the corrosion of most area was slight. The specimen in the Ca(NO3)2solution suffered the lightest corrosion with some small and shallow pits along the grinding scratches.

Horns! so that was what he promised me! Let someone find the plum-seller at once and bring him to me! Let his nose and ears be cut off! Let him be flayed53 alive, or burnt at a slow fire and his ashes scattered54 to the winds! Oh, I shall die of shame and despair! Her women ran at the sound of her screams, and tried to wrench13 off the horns, but it was of no use, and they only gave her a violent headache

Table 4 shows the chemical composition in wt% determined by EDS of the corrosion products on the surfaces of Mg specimens after 14 days of immersion in CaSO4and Ca(NO3)2contaminated solutions (pH=10±0.3). The corrosion products on the Mg surface in both solutions mainly contained three elements: Mg, Ca and O, which were identified as Ca(OH)2and Mg(OH)2by X-ray diffraction(Fig. 8). In the Ca(NO3)2solution, N element was not detected in the corrosion products, indicating the nitrate might not be involved in the corrosion reaction or all the nitrogen containing corrosion products were soluble or evaporable.

Fig. 6. The optical corrosion morphologies of Mg without corrosion products after immersion in the haze constituent contaminated solutions (pH=10±0.3):(a) Mg(OH)2+ (NH4)2SO4, (b) Mg(OH)2+NH4NO3, (c) Mg(OH)2+NaCl, (d) Mg(OH)2+KCl, (e) Mg(OH)2+Na2SO4, (f) Mg(OH)2+MgCl2, (g)Mg(OH)2+CaSO4, (h) Mg(OH)2+NH4Cl, and (i) Mg(OH)2+Ca(NO3)2.

Fig. 7. Typical corrosion morphologies of Mg after removal of corrosion products after 14 days of immersion in haze constituent contaminated solutions(pH=10±0.3): (a) Mg(OH)2+ (NH4)2SO4, (b) Mg(OH)2+NH4NO3, (c) Mg(OH)2+NaCl, (d) Mg(OH)2+KCl, (e) Mg(OH)2+Na2SO4, (f) Mg(OH)2+MgCl2,(g) Mg(OH)2+CaSO4, (h) Mg(OH)2+NH4Cl, and (i) Mg(OH)2+Ca(NO3)2.

Fig. 8. The X-ray diffraction pattern of the corrosion products of Mg specimens immersed in (a) CaSO4 and (b) Ca(NO3)2 solutions.

Fig. 9. Hydrogen evolution curves of pure Mg immersed in basic Mg(OH)2 saturated solution and that with equal normal concentration of individual haze constituents [Mg(OH)2+X (pH=10±0.3): X=(NH4)2SO4, NH4NO3, NaCl, KCl, Na2SO4, MgCl2, CaSO4, NH4Cl, Ca(NO3)2] for 72 h: (a) all the curves,and (b) details of some curves in (a).

3.2. Corrosion in solutions containing the same equal normal concentration of individual haze constituents

In order to further understand the influence of each haze constituent on the corrosion of Mg, the corrosion behavior was investigated in basic solution saturated Mg(OH)2and equal normal solutions with individual haze constituents at 0.0232 normal concentration, which is the solubility of CaSO4·2H2O in water.

Fig. 9 shows the hydrogen evolution curves of pure Mg in basic solution saturated Mg(OH)2and equal normal solutions with individual haze constituents for 72 h. The Mg specimens in the ammonium containing equal normal solutions had the highest hydrogen evolution rates, followed by those in the chloride and sulfate containing equal normal solutions.The hydrogen evolution rates of Mg specimens in the chloride containing equal normal solutions (KCl, NaCl and MgCl2)were close, but that in Na2SO4solution was higher than that in CaSO4containing solution. However, the hydrogen evolution rate of the Mg specimen in Ca(NO3)2was the lowest,which was the same with that in Fig. 1. Table 5 lists the corresponding corrosion rates calculated from weight loss data of pure Mg. The influence of each individual haze constituent on the corrosion of Mg indicated by hydrogen evolution rate is basically consistent with that indicated by weight loss data,except in the NH4NO3containing solution, in which the corrosion rate is substantially higher.

Fig. 10 shows the EIS spectra of pure Mg in basic solution saturated Mg(OH)2and equal normal solutions. All the impedance spectra were consisted of two capacitive arcs.The polarization resistance Rpvalues of pure Mg in the equal normal solutions with different individual haze constituents are listed in Table 6. The influence of PM2.5constituents(except for MgCl2) on Mg corrosion indicated by EIS results was basically consistent with that indicated by the weight loss data. In the ammonium ((NH4)2SO4, NH4NO3, and NH4Cl)containing equal normal solutions, the Rpvalues of the Mg specimens were the smallest,followed by those in the chloride(KCl and NaCl) and sulfates (Na2SO4and CaSO4) containing solutions. However, the Rpvalue of Mg specimen was the largest in the MgCl2containing equal normal solution.

Fig. 10. EIS Nyquist plots of pure Mg in basic Mg(OH)2 saturated solution and that with equal normal concentration of individual haze constituents[Mg(OH)2+X (pH=10±0.3): X=(NH4)2SO4, NH4NO3, KCl, MgCl2, CaSO4, NH4Cl, Ca(NO3)2]: (a) all the curves, and (b) details of some curves in (a).

Table 5 Corrosion rates calculated from weight loss of pure Mg immersed in basic Mg(OH)2 saturated solution and that with equal normal concentration of individual haze constituents [Mg(OH)2+X (pH=10±0.3): X=(NH4)2SO4,NH4NO3, NaCl, KCl, Na2SO4, MgCl2, CaSO4, NH4Cl, Ca(NO3)2] for 72h.

Fig. 11(a) shows the potentiodynamic polarization curves of pure Mg in basic solution saturated Mg(OH)2and equal normal solutions with different individual haze constituents.The corrosion current densities icorrs in the (NH4)2SO4and NH4Cl equal normal solutions as shown in Fig. 11(b) were significantly higher than those in other solutions.Whereas,the corrosion current densities icorrs in the MgCl2and Ca(NO3)2equal normal solutions were the lowest.

Table 6 Curve-fitted parameters of the EISs and polarization curves of Mg specimens in basic Mg(OH)2 saturated solution and that with equal normal concentration of individual haze constituents [Mg(OH)2+X (pH=10±0.3):X=(NH4)2SO4, NH4NO3, KCl, MgCl2, CaSO4, NH4Cl, Ca(NO3)2].

In summary, the corrosivity of the individual haze constituents at the same normal concentration (0.0232N)to Mg could be ordered as: (NH4)2SO4>NH4Cl>NH4NO3>KCl≈MgCl2≈NaCl>Na2SO4>CaSO4>Mg(OH)2(basic solution)>Ca(NO3)2, which is slightly different from that in the haze constituent contaminated solutions.

Fig. 12 shows the corrosion morphologies of Mg without corrosion products after 14 days immersion in the basic Mg(OH)2saturated solution and that with equal normal concentration of individual haze constituents solutions. The Mg suffered serious corrosion in the ammonium containing solutions. While in chloride and sulfate containing solutions,the corrosion of Mg was mild with some pits. The corrosion morphologies of Mg in the CaSO4and Ca(NO3)2solutions were similar to those shown in Fig. 7(g) and (i). In the basic Mg(OH)2saturated solution, there were only shallow pits distributed discretely on the Mg surface.

Fig. 11. (a) Polarization curves of pure Mg in basic Mg(OH)2 saturated solution and that with equal normal concentration of individual haze constituents[Mg(OH)2+X (pH=10±0.3): X=(NH4)2SO4, NH4NO3, KCl, MgCl2, CaSO4, NH4Cl, Ca(NO3)2], and (b) curve-fitted corrosion current densities icorrs.

Fig. 12. Typical corrosion morphologies of Mg after removal of corrosion products after 14 days of immersion in basic Mg(OH)2 saturated solution and that with equal normal concentration of individual haze constituents (pH=10±0.3): (a) Mg(OH)2+ (NH4)2SO4, (b) Mg(OH)2+NH4NO3, (c) Mg(OH)2+NaCl,(d) Mg(OH)2+KCl, (e) Mg(OH)2+Na2SO4, (f) Mg(OH)2+MgCl2, (g) Mg(OH)2+CaSO4, (h) Mg(OH)2+NH4Cl, (i) Mg(OH)2+Ca(NO3)2, and in (j) basic Mg(OH)2 saturated solution.

4. Discussion

4.1. The overall corrosivity sequence of haze constituents to Mg

In order to better understand the influence and mechanism of the haze constituents on the corrosion of Mg, the experiments were repeated in the equal normal solutions which contained the same normal concentration of the haze constituents.All the haze constituents were compared at the same normal concentration for their corrosivity. According to hydrogen evolution curves (see Fig. 9) and weight loss data(Table 5), the corrosivity of these constituents could be classified into 4 groups in a decreasing sequence: ammonium>chloride> sulfate> nitrate.

4.2. , Cl?and

The hydrogen evolution curves, weight loss rates and electrochemical data all indicate that the most aggressive ion in PM2.5isand it is far more detrimental than the other constituents in haze on the Mg components. Cao et al.[33]found that the acceleration effect ofon Mg in alkaline solution was related to the dissolution of the inner MgO film of Mg corrosion products byIt is well accepted that in an aqueous solution there is a somewhat protective inner MgO film and a porous outer Mg(OH)2layer on Mg surface [34,35]. Thecould penetrate into the surface film and dissolved the MgO film, providing paths for the solution to reach the substrate and accelerating the corrosion of Mg substantially. MgO could be dissolved more rapidly than Mg(OH)2in acontaining solution [33]. The corrosion morphology with serious localized corrosion in Figs. 6 and 7 is consistent with the above mechanism. Sinceis mainly present in the form of (NH4)2SO4and NH4NO3in the haze simulated solution, the content of NH4Cl is so small(only 6mmol/L) that the hydrogen evolution rate of Mg in NH4Cl solution is low (Fig. 1).

In a chloride containing solution, such as NaCl or KCl, as the Cl?has a small radius and a large negativity [16,31], it could deteriorate the surface film by penetrating through the film and corroded the substrate. Compared with accelerated dissolution by, the inner MgO penetrated or dissolved by chlorides could be more localized and hence at a relatively lower rate. Nevertheless, the corrosion of Mg overall was still accelerated and suffered serious localized corrosion in the chloride solutions.

In sulfate solutions, such as Na2SO4and CaSO4, the corrosion of Mg was uniform and mild which could be attributed to the less corrosivity ofto pure Mg than that of Cl?because of its larger radius [36–38]. However, this characteristic seems to be contradictory to the current research that the corrosion rate of Mg in (NH4)2SO4was higher than that in NH4Cl at the same normal concentration. Although(NH4)2SO4and NH4Cl had the same concentration of NH4+,the hydrolysis ofin the NH4Cl and (NH4)2SO4solutions could be different when the pHs of the solutions were adjusted using NaOH. As a result, the concentration of the free NH4+in (NH4)2SO4could be larger than that in NH4Cl in the equal normal solutions (see Appendix A). Therefore,the corrosion rate of Mg in the equal normal(NH4)2SO4solution was higher than that in the equal normal NH4Cl solution due to the higher concentration ofin the former solution.

4.3. Mg2+, Ca2+ and

The corrosion rates of Mg estimated from hydrogen evolution (Fig. 9) and weight loss data (Table 5) in Cl?containing solutions (NaCl, KCl and MgCl2) at the same concentration were nearly equal. Their corrosion morphologies were also similar (see Fig. 12). However, the electrochemical data, such as icorrand RP, indicate that the corrosion rate of Mg in the MgCl2was the lowest among all of the solutions.This could be mainly attributed to that the presence of Mg2+in the solution might accelerate the formation of a Mg(OH)2corrosion product film on the Mg surface instantly when the Mg specimen was immersed in the solution initially. The deposited Mg(OH)2film was porous and relatively thick with a rough surface and could trap more hydrogen bubbles than the surface film formed on Mg in the NaCl solution. The trapped H2bubbles could to some extent isolate the corrosion of Mg from the outside environment. However, corrosion could still proceed within the local area beneath the hydrogen bubbles, but the corresponding electrochemical signals could become weaker in measurement. After the system reached its steady state, the hydrogen generated under the bubbles could still come out and be collected, while new bubbles could be trapped by the porous thick rough Mg(OH)2film and replace the previous bubbles. As a result, the measured icorrvalues was smaller and RPvalues higher than in the MgCl2solution.

In Ca2+containing solutions, such as CaSO4, the corrosion of Mg was mild as indicated by the hydrogen evolution rates in Figs. 1 and 9, and by its corroded morphology in Figs. 6 and 7. This is mainly attributed to that a layer of Ca(OH)2film formed on the surface of pure Mg after immersion in the solutions(Table 3 and Fig.8).The corrosion products on Mg surface in both solutions mainly contained Mg,Ca and O, which were identified as Ca(OH)2and Mg(OH)2by X-ray diffraction (Fig. 8). This layer had some protectiveness for Mg matrix in absence of aggressive ions, such asand Cl?.

The corrosion rate of pure Mg in Ca(NO3)2solution is the lowest among all the solutions (Figs. 1 and 9), which could be attributed to: (i) the protective Ca(OH)2film formed on the Mg surface, and (ii) the passivity of the surface film enhanced by[21]. However, when thewas present in thecontaining solution, the protectiveness offor Mg could turn to a damage effect because of the synergistic effect betweenandaccording to reference[20], in whichis believed to have a buffering effect,continuously producing H+to promote the cathodic reductionthat could additionally play the same role as the reduction of H2. Therefore, in the current study, the corrosion rates estimated from hydrogen evolution (Fig. 9) for Mg immersed in the NH4NO3containing solutions underestimated the real corrosion rates, which were substantially lower than those from the weight loss data (Table 5).

4.4. Implication of corrosivity of haze constituents

The current study systematically investigated the effect of haze constituents on the corrosion of pure Mg. The haze constituents and their tyipcal concentrations in air summarized in this paper will be an important database for applications of Mg alloys in industries, since most Mg components could be used in atmospheric conditions. Furthermore, this investigation indicates that the damage of haze on Mg is mainly caused by the.Therefore,the strategy to improve the corrosion resistance of Mg in haze polluted atmosphere should mainly be focused on the reduction ofeffect. However, the corrosion behavior and mechanism of Mg incontaining environment is still equivocal and need more systematic investigations. Moreover, haze polluted marine atmosphere is an important service environment that Mg alloys have to be exposed. It must be even more corrosive. There will be many interesting fundamental understandings regarding the synergistic effect between haze pollutant and marine factors, deserving intensive investigations in future.

5. Concluding remarks

The corrosivity of the haze constituents to pure Mg can be ordered as: (NH4)2SO4> Haze-contaminated-solution>NH4NO3>NH4Cl >NaCl≈KCl≈Na2SO4≈MgCl2≈CaSO4>Mg(OH)2(basic solution) >Ca(NO3)2, which can be classified into 3 groups: ammonium> chloride≈sulfate> nitrate.Some unexpected phenomena and results, such as the corrosion rate of Mg in (NH4)2SO4higher than that in NH4Cl,the lowest corrosion rate in Ca(NO3)2at the same normal concentration, suggest that current knowledge about the corrosion performance of Mg obtained from conventional solutions cannot be directly applied under the complicated conditions. More systematic studies are needed in order to enable the uses of Mg in these kinds of industrial service environments.

Acknowledgments

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

Appendix A

The following solutions were prepared: 2N NaOH(500mL), 0.0232N (NH4)2SO4(1L), 0.0232N NH4Cl (1L),0.1N (NH4)2SO4(1L), 0.1N NH4Cl (1L). All reagents and distilled water were A.R. grade, and all experiments were performed at room temperature.

The pH of (NH4)2SO4and NH4Cl solutions at the same normal concentration of 0.0232N and 0.1N were adjusted to 10 with 2N NaOH, respectively. The initial pH values of the(NH4)2SO4and NH4Cl solutions were measured using a pH meter. The volumes of the required NaOH for pH adjustment were measured using burette and recorded in Table A1.

Table A1 shows that the initial pH values of the(NH4)2SO4solution is a bit higher than that of the NH4Cl solution at the same normal concentrations 0.0232N and 0.1N. The NaOH volumes added in the (NH4)2SO4solutions to adjust pH to 10 were obviously less than those required in the NH4Cl solutions.At 0.0232 normal concentration,the initial difference of pH between the (NH4)2SO4solution (pH=5.74) and NH4Cl solution (pH=5.58) was only about 10?7mol/L, which negligible.

The added NaOH not only led to increased pHs for the solutions, but also combined withto NH3·H2O. Since more NaOH was needed in pH adjustment for NH4Cl than(NH4)2SO4, the excess NaOH added in the NH4Cl solution was mainly consumed by the combination of freeand OH?.As a result,the freeconcentration in NH4Cl after pH adjustment could be lower than that in (NH4)2SO4.

Table A1 Initial pH values of (NH4)2SO4 and NH4Cl solutions at the same normal concentrations of 0.0232N and 0.1N and the volumes of 2N NaOH required to adjust the initial pHs to 10.