Yang Li,Zhiming Shi,Xingrui Chen,Andrej Atrens

School of Mechanical and Mining Engineering,Centre for Advanced Materials and Processing,The University of Queensland,St Lucia,QLD 4072,Australia

Abstract We measured the anodic hydrogen evolution rates for various applied anodic current densities and estimated the corresponding cathodic hydrogen evolution rates.The estimated cathodic hydrogen evolution rates were less than the measured anodic hydrogen evolution rates,contradicting the enhanced catalytic activity mechanism of Mg corrosion.In addition,this model was contradicted by the measured apparent Mg valence of 1.2±0.1.In contrast,the uni-positive Mg+ mechanism of Mg corrosion was supported by(i)the apparent Mg valence of 1.2±0.1,and(ii)the fact that the measured anodic hydrogen evolution rate increased with increasing weight loss rate.

Keywords: A magnesium;B Galvanostatic;Potentiostatic;Weight loss;Hydrogen evolution;C Negative difference effect.

1.Introduction

1.1.Mg corrosion characteristics

Magnesium corrosion has three important characteristics[1–6]:(i)the rate of hydrogen evolution increases on anodic polarisation,(designated as anodic hydrogen evolution),(ii)the rate of Mg dissolution is greater than expected from the Faraday law that the amount of Mg dissolved is determined by the applied anodic current density,and(iii)the corrosion of Mg is weakly influenced by surface films.The first characteristic is designated the negative difference effect(NDE).The two probable mechanisms of Mg corrosion [1,4,6]are(1)the enhanced catalytic activity mechanism of Mg corrosion and(2)the uni-positive Mg+mechanism of Mg corrosion.

Mg corrosion is the electrochemical splitting of water to produce hydrogen by the following overall corrosion reaction under open circuit conditions at the free corrosion potential[3,4]:

The main cathodic partial reaction is hydrogen evolution[3,4,7]by:

In addition,there could be some cathodic current density associated with a minor cathodic reaction involving magnesium hydride,MgH2as follows [7–11]:

and there is evidence for the presence of MgH2during Mg corrosion [10,11,20].The cathodic reaction,described by Eq.(3)occurs in parallel with the main cathodic partial reaction described by Eq.(2).

Assuming that Eq.(2)is the main cathodic partial reaction,the cathodic current density,icHER,T,of hydrogen evolution can be expressed by

wherei0His the exchange current density,Eis the electrode potential,EHis the equilibrium potential of the cathodic hydrogen evolution reaction andbHis the Tafel slope.Eq.(4)indicates that the cathodic hydrogen evolution reaction rate,icHER,T,decreases rapidly with increasing anodic potential,and is essentially zero when the electrode potential is more positive than the corrosion potential,Ecorr.

Fig.1.The surface film consists of a thin(～6 nm)partially-protective MgO film on the Mg metal surface on top of which is a thicker porous Mg(OH)2 layer,based on [12,13].Corrosion is expected to occur in the breaks in the MgO film,which represent ～0.003 of the surface.

The partly protective nature of the film on corroding Mg is illustrated by Fig.1 [12,13].The steady-state surface film typically consists of a thin(～6 nm)partially-protective MgO layer,on which is a porous Mg(OH)2layer.Corrosion occurs in the gaps in the MgO film,which represented a fraction～0.003 of the surface [13].There is much support for this bi-layer film [12–20].It is expected that the area of the gaps in the surface MgO film increases with anodic polarisation[1,3,4,6,7].

1.2.The enhanced catalytic activity mechanism

The enhanced catalytic activity mechanism of Mg corrosion[21,22]assumes(i)that the anodic partial reaction occurs in one step by the overall anodic partial reaction of Eq.(5):

(ii)that the cathodic partial reaction is hydrogen evolution by Eq.(2),and(iii)that the cathodic current density,icHER,c,is given:

wherei0H,a,the exchange current density,is assumed to increase with increasing anodic polarisation.To compensate for the rapid decrease of the second exponential term in Eq.(6),the enhanced catalytic activity mechanism of Mg corrosion assumed thati0H,aincreases faster with increasing anodic polarisation in order to explain the measured increased hydrogen evolution with anodic polarisation [21,22].

The enhanced catalytic activity mechanism was recently shown[23,24]to be not consistent with experimental evidence for pure Mg.The anodic hydrogen evolution rate,iH2,a,was measured for various values of anodic applied current density.Immediately thereafter the“enhanced”cathodic current density of hydrogen evolution,icHER,c,was estimated from polarisation curves.The“enhanced”cathodic current density of hydrogen evolution,icHER,c,was much smaller(by up to two orders of magnitude)than the measured anodic hydrogen evolution rate,iH2,a,indicating that the enhanced catalytic activity mechanism was contradicted by experimental evidence[23,24].

Thus,recent experimental results for pure Mg [23,24]contradicted the enhanced catalytic activity mechanism [21,22].Nevertheless,it is worthwhile to further experimentally explore this mechanism,because this mechanism has received so much attention over the last years [1–6],and also because a fresh examination leads to considerable new insights.

1.3.Research aims

This research carried out the following series of experiments to experimentally test the two probable mechanisms of Mg corrosion:(1)the enhanced catalytic activity mechanism and(2)the uni-positive Mg+mechanism.The uni-positive Mg+mechanism is summarised in Section 4.4.

?The anodic hydrogen evolution rate on anodic polarisation was measured and compared with the estimated cathodic hydrogen evolution rate at the same potential.The cathodic hydrogen evolution rate was evaluated immediately after the end of the anodic polarisation from cathodic polarisation curves measured immediately after the end of the anodic polarisation.The enhanced catalytic activity mechanism predicts that the cathodic hydrogen evolution rate must equal the anodic hydrogen evolution rate.

?The apparent Mg valence of Mg,V,was evaluated from the electron flux(number of electrons per unit area per unit time)during anodic polarisation divided by the flux of dissolving Mg atoms(number of Mg atoms per unit area per unit time).The apparent Mg valence must be equal to two(V=2.0)in the enhanced catalytic activity mechanism because the anodic partial reaction is assumed to be by Eq.(5),whereas an apparent valence between 1 and 2(1

?The anodic hydrogen evolution rate was compared with the weight loss rate.A direct relationship is expected between these two quantities in the uni-positive Mg+mechanism(see Section 4.4)whereas there should not be any relationship in the enhanced catalytic activity mechanism of Mg corrosion.

2.Experimental methods

2.1.Mg alloys

Pure Mg and the Mg alloy WE43 were used in this research.The pure Mg consisted only of the Mg matrix,with no other phases in the microstructure.The pure Mg exemplified the intrinsic corrosion of Mg.The intrinsic corrosion rate of Mg is 0.3 mm/y in 3.5 wt.% NaCl saturated with Mg(OH)2[25,26].WE43 consists of the Mg matrix with some second phase particles [27,28].The corrosion rate of WE43 in 3.5 wt.% NaCl saturated with Mg(OH)2was 0.23 mm/y[27].The corrosion rate somewhat lower than the intrinsic Mg corrosion rate was attributed to [27](a)no corrosion acceleration by the small second phase particles,and(b)a somewhat more protective surface film.Table 1 provides the chemical composition of the pure Mg and WE43.

Table 1 Chemical composition of the pure Mg and the Mg alloy WE43.

Fig.2.(a)overview of experimental apparatus allowing simultaneous electrochemical and hydrogen evolution measurements using a plug-in specimen.Weight loss measurements were carried out at the end of the experiment.(b)schematic of the assembly of a plug-in specimen.

Fig.2(a)presents an overview of the experimental apparatus [29–32].The plug-in specimen,prepared as indicated in Fig.2(b),was the working electrode.The reference electrode was an Ag/AgCl standard electrode saturated with KCl.A platinum sheet was the counter electrode.This arrangement allowed hydrogen gas collection [33]and simultaneous electrochemical measurements which used the Parstat 2263 and Parstat 2273 electrochemistry systems.The specimens were ground to 4000 SiC grit using ethanol,washed with distilled water and ethanol,dried(in a desiccator for at least a day)and weighed,to give the specimen weight before the galvanostatic experiments,Wb.All experiments were carried out in N2deaerated 3.5 wt.% NaCl saturated with Mg(OH)2.The solution was made with reagent grade chemicals and high-quality deionized water.Experiments were carried out at 23±2 °C.

2.2.Anodic and cathodic hydrogen evolution rates

The first series of experiments were designed to measure the anodic hydrogen evolution rate and to estimate the corresponding cathodic hydrogen evolution rate,in order that there could be made a rigorous comparison between the cathodic hydrogen evolution rate and the anodic hydrogen evolution rate.This allowed a direct test of the enhanced catalytic activity mechanism because the enhanced catalytic activity mechanism predicts that the cathodic hydrogen evolution rate must equal the anodic hydrogen evolution rate.

The anodic hydrogen evolution rate was first measured using galvanostatic experiments.These galvanostatic experiments were carried out with the specimens immersed in N2deaerated 3.5 wt.% NaCl saturated with Mg(OH)2with applied anodic current densities,Iapplied,of 0.01,0.1,0.3,1,3,10,30 and 100 mA cm?2.During each galvanostatic experiment,the potential of the Mg electrode,Eapplied,was recorded.The applied current density caused the measured potential to have an“iR”error because of the applied current density,whereiwas the applied current density,Iapplied,andRwas the resistivityR1between the Luggin probe and the Mg specimen.The evolved hydrogen was also measured in each galvanostatic experiment.The hydrogen evolution rate,NH2,m,was reported in units of mL cm?2h?1.

Immediately thereafter,it was necessary to evaluate the cathodic hydrogen evolution rate at the same potential as for the measurements of the anodic hydrogen evolution rate.The appropriate potentialEc,appliedwas theIappliedRlcorrected potential corresponding toEapplied,whereRlcorresponded to the resistivity between the Luggin probe tip and the metal of the Mg electrode,evaluated as

where(Ef–Ecorr,m)was the potential difference between the Luggin probe tip and the Mg metal,Efwas the measured potential measured at the end of the galvanostatic experiment,measured with the application ofIapplied,andEcorr,mwas the measured corrosion potential measured with no applied current density,measured immediately before measurement of the cathodic polarisation curves.The resistivityRlis usually identified as the resistivity of the solution.However,Rlcorresponds to the resistivity between the Luggin probe tip and the metal of the Mg electrode and may include kinetic and mass transfer contributions.This resistivity is nevertheless the appropriate resistance to use for theiRcorrection of the electrochemical potential.

The cathodic hydrogen evolution rate was evaluated as follows.(i)Immediately after each galvanostatic experiment,a cathodic polarization curve was measured fromEapplied+20 mV toEapplied–600 mV at a scan rate of 10 mV/min.(ii)The cathodic partial hydrogen evolution reaction was extrapolated toEc,appliedto estimate the cathodic hydrogen evolution rate,icHER,c.

2.3.Apparent valence

The apparent valence of Mg corrosion during the galvanostatic experiments was evaluated as follows.The applied current density(mA cm?2)was equivalent to the electron flux,Ne,a(mmol cm?2h?1),associated with the anodic partial reaction,given by [30]:

whereIais the anodic current density,Icis the cathodic current density(evaluated from the polarisation curves)andFis the Faraday.

Immediately after the measurement of the cathodic polarisation curves,the corrosion products were removed by immersion of the specimens in the chromic acid solution of composition of 200 g L?1CrO3+2 g L?1AgNO3at 23±2 °C.The specimen was washed in ethanol and distilled water,dried in a desiccator for at least a day,and weighed,to give the specimen weight after the galvanostatic experiments,Wafrom which the weight loss rate,WL(mg cm?2h?1)was determined.The rate of metal loss,NW(mmol cm?2h?1),was evaluated from [30–32,34]:

The apparent Mg valence of Mg,V,was evaluated from the following:

The apparent Mg valence is the average number of electrons produced by the dissolution of one Mg atom.The key postulate of the enhanced catalytic activity mechanism is thatV=2.0 because of the key assumption that the Mg anodic reaction is fully described by Eq.(5).In contrast,an apparent valence between 1 and 2(1

2.4.Anodic hydrogen evolution rate and weigh loss rate

The anodic hydrogen evolution rate,NH2,m,was compared with the weight loss rate,NW,from Eq.(9).A direct relationship is expected between these two quantities in the unipositive Mg+mechanism.This comparison was facilitated by expressing the hydrogen evolution rate,NH2,m,in the same units of mmol cm?2h?1,for whichNH2,mwas evaluated from[31–34]:

wheretH2is the exposure time(h).

3.Results

Two series of experiments were carried out with each material(pure Mg and WE43).Typical data from the galvanostatic experiments are presented in Figs.3–7.Typical data from the potentiodynamic experiments are presented in Figs.8 and 9.The data from the repeat experiments were similar indicating good reproducibility.Thus,typically,the data for one set of experiments is presented for each Mg alloy as representative.

3.1.Galvanostatic experiments

The galvanostatic experiments measured(i)the anodic hydrogen evolution,and(ii)the potential at which the anodic hydrogen was evolved.

The anodic hydrogen evolution volume versus time for pure Mg for various applied anodic current densities,in the galvanostatic experiments,is presented in Fig.3a.Fig.3b presents an enlarged part of Fig.3a for better display of the data at small applied current densities.Fig.3 shows that the anodic hydrogen volume increased linearly with time.The slope of the plot increased with increasing applied current density.Fig.4 shows similar plots for WE43.The data for WE43 were similar to those of pure Mg.

Fig.5 presents the anodic hydrogen evolution rate versus applied anodic current density for pure Mg and WE43 in the galvanostatic experiments,from Figs.3 and 4.The anodic hydrogen evolution rate increased with increasing applied anodic current density although there was some data scatter for small applied current densities,when there were substantial measurement errors.

The potential at which the anodic hydrogen evolved is presented in Fig.6a for pure Mg for various applied current densities,in the galvanostatic experiments.Fig.6b,an enlarged portion of the plot of Fig.6a,shows that there were some initial potential changes and the potential had stabilized by 3600 s(i.e.by 1 h)for all applied current densities except for 100 mA cm?2for which there was some fluctuation of potential even after 1200 s,attributed to changes in the effective solution resistance produced by the copious hydrogen evolution at the Mg electrode.Fig.7a presents similar data for the measured potential for WE43 for the 2 h experiment,and the detail in Fig.7b shows the initial potential changes and that similarly the potential rapidly stabilized.The data for WE43 were similar to those for pure Mg.The initial potential changes are attributable to the change of the surface film from that formed during specimen preparation to the steady state film [12].The potential thereafter stabilized to a steady state potential.

Fig.3.a.Hydrogen evolution volume versus time for pure Mg in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities,in the galvanostatic experiments.The absolute measurement error for the hydrogen volume was ～2% for each data point for an applied current density of 10 mA/cm2 although the trend indicated a lower relative error between data points.b.Enlarged part of plot of hydrogen evolution volume versus time for pure Mg in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities,in the galvanostatic experiments.

Fig.4.a.Hydrogen evolution volume versus time for WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities,in the galvanostatic experiments.Errors were the same as for Fig.3.b.Enlarged part of plot of hydrogen evolution volume versus time for WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities,in the galvanostatic experiments.

3.2.Potentiodynamic experiments

The corresponding cathodic hydrogen evolution was characterised by the measurement of a potentiodynamic polarisation curve measured immediately after each galvanostatic experiment.Figs.8 and 9 present cathodic polarisation curves for pure Mg and WE43 measured immediately after the galvanostatic experiments.These typical cathodic polarisation curves did not show a systematic variation with the prior galvanostatic applied current density.This was similar to the findings of [23].

Fig.5.Anodic hydrogen evolution rate versus applied current density for pure Mg and WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 in the galvanostatic experiments.The measurement error in the anodic hydrogen evolution rate was ～1% for an applied current density of 10 mA/cm2 and increased with decreasing applied current density to ～20% at an applied current density of 1 mA/cm2.

3.3.Resistance

The cathodic hydrogen evolution rate was evaluated from the cathodic branch of these polarisation curves by the extrapolation of the cathodic polarisation curve to the potential corresponding to the potential measured during the corresponding galvanostatic experiment,as explained in Section 2.2.This extrapolation required the value of the resistivityR1in each case.Fig.10 presents the variation with applied current density ofRlthe resistivity between the Luggin probe tip and the metal of the Mg electrode for pure Mg and WE43 in the galvanostatic experiments.Fig.10(a)provides an overview,that shows a higher resistivity for low applied current densities that decreased with increasing current density up to an applied anodic current density of 1 mA cm?2.These substantial resistivity values are attributed largely to the resistivity of the surface film,with the resistivity of the surface film on WE43 being higher than that of the resistivity for pure Mg.This is consistent with a more protective surface film on WE43,incorporating Y2O3,consistent with the lower corrosion rate of WE43 [27,28],and consistent with the Electrochemical Impedance Spectroscopy data [28].

It is expected that the higher resistivity value at the lowest applied current density(0.01 mA/cm2)has a significant contribution from the intrinsic resistivity of the surface film,and that the resistivity decreases as film becomes degraded as the Mg surface corrodes due to the applied current density.The film free surface area increases,so that the resistivity reflects the solution resistivity.Furthermore,the evolving hydrogen can also influence the measurement.

Fig.10(b)presents the magnified portion of this plot at the higher applied current densities.The resistivityRlfor WE43 increased somewhat with increasing applied current density attributed to the effect of the evolving hydrogen,which effect was more manifest for WE43 attributed to the smaller fraction of film free area.

3.4.Weight loss

Fig.11 presents the weight loss rate versus applied current density for pure Mg and WE43 in the galvanostatic experiments.The weight loss rate also increased with increasing applied anodic current density,although the data was scattered for low applied current densities,when there were substantial measurement errors.

Fig.6.a.Measured potential versus time for pure Mg in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities,in the galvanostatic experiments.b.Enlarged portion of the plot of measured potential versus time for pure Mg in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities.

Fig.7.a.Measured potential versus time for WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities.b.Enlarged portion of the plot of measured potential versus time for WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 for various applied current densities.

Fig.8.Cathodic polarisation curves for pure Mg in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2)measured immediately after the galvanostatic experiments.There was no iR correction made for these curves.

4.Discussion

4.1.Anodic hydrogen evolution

Fig.12 presents the anodic hydrogen evolution rate in comparison with the cathodic hydrogen evolution rate from the cathodic partial reaction,plotted versus applied current density,for pure Mg and WE43 in the galvanostatic experiments,Series 2.The anodic hydrogen evolution rate was determined from the evolved hydrogen during the galvanostatic experiments during which there was applied an anodic current density.The cathodic hydrogen evolution rate from the cathodic partial reaction was evaluated from the cathodic polarisation curves measured immediately after the galvanostatic experiments.Each cathodic polarisation curve was extrapolated to theiR-corrected potential measured during the corresponding galvanostatic experiment,to give the corresponding rate of hydrogen evolved by the cathodic partial reaction,as accelerated by the prior galvanostatic experiment.Each such value of the cathodic hydrogen evolution rate is predicted by the enhanced catalytic activity mechanism to be equal to the measured anodic hydrogen evolution rate.

The data of Fig.12 show that the values of the cathodic hydrogen evolution rate were less than the measured anodic hydrogen evolution rate,indicating that the enhanced catalytic activity mechanism did not provide an adequate explanation for the measured anodic hydrogen evolution rate for pure Mg and WE43 in the galvanostatic experiment Series 2.

Fig.9.Cathodic polarisation curves for WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 measured immediately after the galvanostatic experiments.There was no iR correction made for these curves.

Fig.10.Variation with applied current density of Rl,the restivity between the Luggin probe tip and the metal of the Mg electrode for HP Mg and WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 in the galvanostatic experiments:(a)overview,(b)magnified portion at the higher applied current densities.The error in the resistivity was ～2%.

Similarly,Fig.13 presents the anodic hydrogen evolution rate and the cathodic hydrogen evolution rate from the cathodic partial reaction,plotted versus applied current density,for pure Mg and WE43 in the galvanostatic experiments,Series 3.Just like the data of Fig.12,the data of Fig.13 show that the values of the cathodic hydrogen evolution rates were less than the measured anodic hydrogen evolution rates,indicating that the enhanced catalytic activity mechanism did not provide an adequate explanation for the measured anodic hydrogen evolution rates for pure Mg and WE43 in the galvanostatic experiments,Series 3.

Fig.11.Weight loss rate versus applied current density for pure Mg and WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 in the galvanostatic experiments.The measurement error in the weight loss rate was～1% for an applied current density of 1 mA/cm2 and increased with decreasing applied current density to ～20% at an applied current density of 0.1 mA/cm2.

The conclusion from Figs.12 and 13 was that the experimental data were not consistent with the enhanced catalytic activity mechanism of Mg corrosion.This conclusion was consistent with the conclusion from the previous research of Fajardo,Frankel and co-workers [23,24].

Fig.12.Anodic hydrogen evolution rate(full symbols,left hand axis)and hydrogen evolution rate from the cathodic partial reaction(open symbols,right hand axis),plotted versus applied current density,for pure Mg and WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 in the galvanostatic experiments,Series 2.As per Fig.5,the measurement error in the anodic hydrogen evolution rate was ～1% for an applied current density of 10 mA/cm2 and increased with decreasing applied current density to～20% at an applied current density of 1 mA/cm2.The error in the cathodic hydrogen evolution rate was ～20%.

The values of the cathodic hydrogen evolution rate were evaluated from Tafel extrapolation of the polarisation curves of Figs.8 and 9 using the cathodic part between 70 mV and 150 mV from the free corrosion potential.These polarisation curves had noiRcorrection.TheiRcorrection was relatively small in the region that was used for Tafel extrapolation,between 70 mV and 150 mV from the free corrosion potential.Furthermore,the effect of suchiRcorrection would be to decrease the evaluated values of the cathodic hydrogen evolution rate.This has the effect of strengthening the conclusion that the cathodic hydrogen evolution rates were less than the measured anodic hydrogen evolution rates,indicating that the enhanced catalytic activity mechanism did not provide an adequate explanation for the measured anodic hydrogen evolution rates for pure Mg and WE43.

Fig.13.Anodic hydrogen evolution rate(full symbols,left hand axis)and hydrogen evolution rate from the cathodic partial reaction(open symbols,right hand axis),plotted versus applied current density,for pure Mg and WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 in the galvanostatic experiments,Series 3.Errors were as evaluated for Fig.12.

4.2.“Cathodic”hydrogen evolution

Figs.12 and 13 also do show that there were appreciable“apparent”cathodic hydrogen evolution rates for applied current densities greater than 1 mA/cm2.The“apparent”cathodic hydrogen evolution rates did not decrease to negligible values as expected from Eq.(4).These appreciable“apparent”cathodic hydrogen evolution rates relate to the measured cathodic polarisation curves and were measured as a current density atEc,applied.They represent a cathodic current density,not an actual cathodic hydrogen evolution rate.These values of cathodic current density are attributed to the minor cathodic partial current given by Eq.(3).

This leads to the following conclusion.At the same potentials as measured with an applied anodic current density during the galvanostatic experiment,the cathodic polarisation curves indicated that there were appreciable cathodic current densities for applied current densities greater than 1 mA/cm2.These cathodic current densities could be attributed to a cathodic partial reaction involving magnesium hydride,MgH2.

These modest cathodic current densities in Figs.12 and 13 are attributed to the fact that it is expected that the area of the gaps in the surface MgO film was substantially greater than at the free corrosion potential because of the prior anodic polarization as the free area is expected to increase with anodic polarization [1,3,4,6,7].

4.3.Apparent Mg valence

In addition,Fig.14 provides a plot of independent evaluations of the apparent valence of Mg and of the value of k for the current densities that had the appropriate precision,i.e.low acceptable measurement errors.This indicates that the apparent valence for Mg was between 1.0 and 2.0 as expected by the uni-positive Mg+corrosion mechanism.The average value of the apparent Mg valence was 1.2±0.1.Similarly,Atrens et al.[35]showed that the data of Fajardo and Frankel[23]produced values of the apparent valence of Mg between 1.0 and 2.0,with an average value of 1.3.The average value of k=0.2 in Fig.14 was in good agreement with the average value ofV.

Fig.14.Plot of apparent Mg valence, V,and the value of k,versus applied current density for the applied current densities that produced data with acceptably low measurement errors.The average value of the apparent Mg valence was 1.2±0.1 consistent with expectations of the uni-positive Mg+corrosion mechanism.The error for each data point was ～20%.k=(1-X)/(1 + X), X= NH.m/Ne,a.The average value of k=0.2 was in good agreement with the average value of V.

The enhanced catalytic mechanism is based on the key assumption that the apparent valence of Mg is 2.0,because it is assumed that Eq.(5)is a complete description of the Mg anodic partial reaction.The data of Fig.14 contradict this key assumption of the enhanced catalytic activity mechanism.

4.4.The uni-positive Mg+ corrosion mechanism

The increased anodic hydrogen evolution rate on anodic polarization is predicted by the uni-positive Mg+corrosion mechanism [1–4],so that Song [36]indicated that the production of anodic hydrogen was strong evidence for the uni-positive Mg+corrosion mechanism,particularly the colocation of anodic hydrogen and Mg dissolution,in locations different to those that produced cathodic hydrogen on cathodic polarisation.Fajardo et al.[24]also convincingly showed that the anodic hydrogen was co-located at the locations of Mg dissolution.Recent strong support for the uni-positive Mg+corrosion mechanism has been provided by Gomes et al.[13],Huang et al.[6],Orazem and Tribollet [37],and Ma et al.[40,41].Zhang et al.[38]claimed the electrochemical detection of the univalent Mg cation.

The uni-positive Mg+corrosion mechanism can be described by the following set of reactions under open circuit conditions at the free corrosion potential.The overall anodic partial reaction of Mg dissolution,Eq.(5),occurs in at least two steps because(i)the simultaneous transfer of two electrons is forbidden by quantum mechanics [39],and(ii)recent first principles calculations [40,41]showed that the sequential loss of two electrons by Mg is much easier than the simultaneous loss of two electrons.These two anodic steps are suggested to be:

The anodic partial reaction is balanced by the cathodic partial reaction of hydrogen evolution,written as follows to balance the anodic partial reactions:

The uni-positive Mg+is a reactive,short-lived,intermediate [40,41].A fraction k undergoes an electrochemical anodic reaction by Eq.(13),whilst the complement chemically splits water by following reaction:

Eqs.(12)–(15)sum to give the overall Mg corrosion reaction,Eq.(1).

Thus,the uni-positive Mg+corrosion mechanism provides a simple explanation for the main manifestations of Mg corrosion:(i)anodic polarisation causes the production of anodic hydrogen co-located with the corroding Mg,(ii)the amount of anodic hydrogen is proportional to the amount of corroding Mg,and(iii)the ratio 2/(1 +k)is the ratio of Mg2+produced to the amount expected by application of the Faraday Law,which is greater than 1,becausek <1.

In support of the uni-positive Mg+corrosion mechanism,Fig.14 indicates that the average value of the apparent valence was 1.2±0.1,and in confirmation the average value of k was 0.2.Similarly,Atrens et al.[35]showed that the data of Fajardo and Frankel [23]produced values of the apparent valence of Mg between 1.0 and 2.0,with an average value of 1.3.The uni-positive Mg+corrosion mechanism assumes that the Mg anodic partial reaction occurs in the two steps given by Eqs.(12)and(13)and that some Mg+chemically splits water by the reaction described by Eq.(15).Thus,a valence between 1.0 and 2.0 is at the core of the uni-positive Mg+corrosion mechanism consistent with the data of Fig.14,and a value of k between 0 and 1.

In addition,Fig.15 shows that the data from the present research is as predicted by the uni-positive Mg+corrosion mechanism that the rate of anodic hydrogen evolution is directly related to the rate of Mg corrosion.Fig.15 presents the anodic hydrogen evolution rate,NH,m,plotted versus weight loss rate,NW,for pure Mg and WE43 in the galvanostatic experiments.The anodic hydrogen evolution rate,NH,m,increased with increasing weight loss rate,NW,as expected by the uni-positive Mg+corrosion mechanism.

4.5.Mechanism evaluation

This research experimentally evaluated the enhanced catalytic mechanism for Mg by a study of the corrosion of pure Mg and the Mg alloy WE43.The experimental data in Figs 12 and 13 were not consistent with the enhanced catalytic activity mechanism.This conclusion was consistent with the previous research of Fajardo,Frankel and co-workers[23,24].In addition,the apparent Mg valence of 1.2±0.1 in Fig.14 contradicted the key assumption of the catalytic activity mechanism.In contrast,the experimental data of Fig.14 were consistent with this Mg+corrosion mechanism and further support was provided by the data of Fig.15.

Fig.15.Hydrogen evolution rate, NH,m,plotted versus weight loss rate, NW,for pure Mg and WE43 in N2 deaerated 3.5 wt.% NaCl saturated with Mg(OH)2 in the galvanostatic experiments.As per Fig.9,the measurement error in the anodic hydrogen evolution rate was ～1% for an applied current density of 10 mA/cm2 and increased with decreasing applied current density to ～20% at an applied current density of 1 mA/cm2.As per Fig.10,the measurement error in the weight loss rate was ～1% for an applied current density of 1 mA/cm2 and increased with decreasing applied current density to ～20% at an applied current density of 0.1 mA/cm2.

The key assumptions for the uni-positive Mg+corrosion mechanism are:(i)that there is a reaction intermediate(between metallic Mg and the equilibrium ion Mg2+)that is sufficiently energetic to chemically split water to produce hydrogen,and(ii)that the rate of the anodic dissolution of Mg increases with anodic polarisation.Key assumption(i)was verified by the recent first-principle research of Ma et al.[40,41]that indeed showed that the uni-positive Mg+is sufficiently energetic to chemically split water by reaction Eq.(15).Key assumption(ii)is that the applied anodic potential causes activation polarisation and accelerates Mg corrosion.This is the normal expectation of electrochemistry.

4.6.Additional detail and support

Because the uni-positive Mg+corrosion mechanism predicts that the overall Mg corrosion reaction is given by Eq.(1),it was shown [34]that

whereQappliedis the amount of anodic charge associated with the applied current density,QMgis the charge associated with the measured mass of Mg corroded assuming that the corrosion produced the equilibrium Mg2+ion by the overall anodic partial reaction,Eq.(5),andQH2is the charge associated with measured amount of hydrogen.

The chemically splitting of water by the uni-positive Mg+occurs on the Mg metal surface,not in solution.The unipositive Mg+ion [40,41]is the surface species because the partitioning of charge between Mg and Mg2+is energetically favoured by the following,essentially-instantaneous reaction:

Gas phase studies have indicated that the uni-positive Mg+reacts with water in milliseconds [42–50].There is no expectation that Mg+can exist in an aqueous solution for a time greater than milliseconds.

Strong support for the uni-positive Mg+corrosion mechanism has also been provided by the recent reviews [1,6].Highlights are as follows.Song [36]showed that the locations at which the anodic hydrogen was evolved on a Mg electrode by anodic polarisation were quite distinct from the different locations at which were produced cathodic hydrogen,indicating that the anodic hydrogen was produced by a different mechanism to the cathodic partial reaction of hydrogen evolution.The critical work by Lebouil et al.[51]of the group of Ogle in Paris found that the flux of Mg2+was greater than expected from the Faraday law and corrosion with n=2;they proposed that the extra Mg2+was produced by the splitting of water by metallic Mg(although the data is just as consistent with the splitting of water by Mg+).Furthermore,the mass balance by Atrens et al.[1]indicated that Lebouil et al.[51]measured only 60%of the Mg2+produced,which reinforces their finding that the Mg corrosion occurred with an apparent Mg valence n<2.This mass balance was also applicable to the work of Rossrucker et al.[52],so their work also indicated n<2.Atrens et al.[1]also showed that there was excess hydrogen and therefore corrosion with n<2 in the work of Williams et al.[53],and also in the work of Williams et al.[54].Furthermore,Cain et al.[55]and Cain et al.[56]showed that the experiments data were consistent with Eq.(16)which is consistent with the uni-positive Mg+corrosion mechanism.

4.7.Comparison of WE43 and pure Mg

Fig.5 indicates that the anodic hydrogen evolution rate on WE43 was the same as that on pure Mg within experimental error.This indicated that the hydrogen evolution behaviour on WE43 was not significantly influenced by the second phase particles in the microstructure.This was consistent with the fact that the second phase particles had little influence on the corrosion behaviour of WE43.Similarly,Figs.15,12 and 13 indicate similar values for WE43 and pure Mg of(i)the weight loss rate and(ii)cathodic hydrogen evolution.These results are also consistent with the fact that the second phase particles had little influence on the corrosion behaviour of WE43.This indicates that the experimental data for both WE43 and pure Mg showed that the enhanced catalytic activity mechanism did not provide an adequate explanation for the measured anodic hydrogen evolution rate.

Fig.14 indicates the same apparent Mg valence for both WE43 and pure Mg,in support of the uni-positive Mg+corrosion mechanism.Similarly,Fig.15 indicated the same relationship for WE43 and pure Mg indicating that the results from both alloys supported the uni-positive Mg+corrosion mechanism.

4.8.Limitations

Measurement errors of the anodic hydrogen evolution rate in Fig.5 and the cathodic hydrogen evolution rate in Figs.12 and 13 increased with decreasing applied current density.Nevertheless,the measurement errors were quite small for applied current densities greater than 1 mA/cm2,and under these conditions it was clear that the anodic hydrogen evolution rate was not explained by the enhanced catalytic activity mechanism.A similar consideration for the data of Figs.14 and 15 indicates that there was the compelling agreement with the expectations of the uni-positive Mg+corrosion mechanism.

4.9.Implications

Better understanding of the Mg corrosion mechanism as elucidated in this paper is of importance to the understanding and development of Mg-alloy anodes for Mg-based primary batteries including the Mg-air battery [57–62].The mechanism proposed in this paper helps understand the anodic hydrogen evolution of Mg-based batteries during the discharge process,which also provides an explanation for the decrease of anodic efficiency.An apparent Mg valence of less than 2.0 means that there is less current generated by the Mg anode of the Mg-battery than expected.It is expected that the apparent Mg valence predicts the anodic efficiency in the Mg-air battery.This understanding provides a base for the development of better Mg alloy anodes and better Mg-based batteries.

Implications of a better understanding of the Mg corrosion mechanism for Mg corrosion and corrosion protection were discussed by Huang et al.[6]under the headings:purification,passivity,cathodic protection,coating and environmental modification.Different understanding of Mg corrosion lead to different approaches to corrosion protection of Mg and Mg alloys.

5.Conclusions

This paper measured experimentally the anodic hydrogen evolution rate and estimated the cathodic hydrogen evolution rate for pure Mg and WE43 in N2deaerated 3.5 wt.% NaCl saturated with Mg(OH)2and concluded as follows.

1.The enhanced catalytic activity mechanism was contradicted by the experimental data:(i)the enhanced catalytic activity mechanism predicts that the measured anodic hydrogen evolution rate for pure Mg and WE43 should be equal to the cathodic hydrogen evolution rate,whereas the measured anodic hydrogen evolution rate was significantly greater than the cathodic hydrogen evolution rate,and(ii)the enhanced catalytic activity mechanism predicts that the apparent Mg valence should be equal to 2.0;whereas the experimental data indicated that the apparent Mg valence was equal to 1.2±0.1.

2.The measured anodic hydrogen evolution rate increased with increasing weight loss rate as expected by the unipositive Mg+corrosion mechanism.Further support for this Mg+corrosion mechanism was provided by the apparent Mg valence ofV=1.2±0.1,and the average value of k=0.2.

3.A consideration of the measurement errors indicates that the measurement errors were quite small for applied current densities greater than 1 mA/cm2,and under these conditions it was clear that the anodic hydrogen evolution rate was not explained by the enhanced catalytic activity mechanism;and similarly,that there was the compelling agreement with the expectations of the uni-positive Mg+corrosion mechanism.

Data availability

The data used in this paper is contained within the paper.

Conflict of Interest

There are no conflict of interests.

Acknowledgements

The authors acknowledge the financial support for this research by the Australian Research Council(ARC)through the discovery grant DP170102557.