Effect of heat treatment on the mechanical and biocorrosion behaviour of two Mg-Zn-Ca alloys

N.Pulido-González,P.Hidalgo-Manrique,S.García-Rodríguez,B.Torres,J.Rams

Departamento de Matemática Aplicada,Ciencia e Ingeniería de Materiales y Tecnología Electrónica,ESCET,Universidad Rey Juan Carlos,C/Tulipán s/n,Móstoles,Madrid,28933,Spain

Abstract The effect of heat treatment on the mechanical and biocorrosion behaviour of the Mg-1wt.% Zn-1wt.% Ca (ZX11) and Mg-3wt.%Zn-0.4wt.% Ca (ZX30) alloys was evaluated.For this purpose,three-point bending tests as well as electrochemical and immersion tests in Hank’s solution were performed on both alloys in four different thermal conditions:as-cast,solution-treated,peak-aged and over-aged.Microstructural examinations revealed that the as-cast ZX11 and ZX30 alloys exhibit a microstructure composed of α-Mg grains separated by large Mg2Ca and Ca2Mg6Zn3 particles and by large Ca2Mg6Zn3 particles,respectively.During solution treatment,the Ca2Mg6Zn3 precipitates at the grain boundaries (GBs) are fully dissolved in the ZX11 alloy,but mainly redistributed to form a more connected configuration in the ZX30 alloy,showing a poor age-hardening response.Consequently,after solution-treatment,galvanic corrosion and corrosion rate decreases in the former,but increases in the latter.The peak-aged condition displays the highest corrosion rate for both alloys,maybe due to an excessive number density of fine Ca2Mg6Zn3 particles acting as cathodic sites.However,the over-aged condition shows the lowest corrosion rate for the ZX11 alloy and a very similar one to that of the as-cast sample for the ZX30 alloy.The ZX11 alloy shows generally better biocorrosion behaviour than the ZX30 alloy to its lower content in the Ca2Mg6Zn3 phase and thus reduced galvanic corrosion.The Mg2Ca phase present in the studied ZX11 alloy has been proved to exhibit an increased corrosion potential,which has been related to an observed enrichment with Zn.

Keywords: Magnesium alloys;Biodegradable implants;Heat treatment;Microstructure;Mechanical properties;Corrosion.

1.Introduction

Metallic materials play an important role as implant materials for hard tissue replacement.Due to their high strength,durability and biocompatibility,metallic alloys such as stainless steel (316LSS) or surgical grade 5 titanium (Ti-6Al-4V)are commonly used as load-bearing implants [1].However,bone fixation devices made of these alloys must be removed by a second surgical procedure after the healing procedure.Moreover,derived from the high stiffness of these materials compared to that of cortical bone,hardware made of these alloys often results in post-surgical problems like implant breakage or bone loss around the implant [2].For this reason,a new domain of research on metallic implants focuses on biodegradable materials,which can dissolve in the biological environment after a certain length of functional use,while being replaced by natural bone tissue.

Biodegradable Mg alloys emerge as promising alternatives for biomedical applications,including orthopaedic implants[3-7].As an essential element for human metabolism,Mg shows excellent biocompatibility.In fact,owing to its important role in the formation of biological apatite,Mg may actually have stimulatory effects on the growth of new bone tissue.Moreover,Mg is a lightweight metal that exhibits low stiffness,actually close to that of the bone.However,the fast degradation rate of Mg in physiological environment accompanied by large amounts of evolved hydrogen gas,which may be detrimental to the healing process,has limited the use of Mg as an orthopaedic material so far.A rapid loss of mechanical integrity before the new bone tissue can properly heal is associated with this process.

Alloying is a very effective method to increase the corrosion resistance as well as mechanical properties of Mg.However,for biomedical applications,alloying elements must be chosen with careful consideration of the toxicity and other possible adverse effects on the body of the degradation products [8].As an example,accumulation of Al in the body has been identified as a potential cause of hepatoxicity and is possibly linked to Alzheimer’s disease.This prevents the use of some commercially available alloys such as AZ91 and AZ31 for medical implants.In reality,there is a very small number of elements that can be tolerated in the human body and retard the biocorrosion of Mg at the same time.Alloying elements such as Ca and Zn have been extensively studied in order to achieve Mg alloys with better biocompatibility and slower corrosion rate since they could be released from bone fixation devices at levels consistent with what can be absorbed from human’s diet [9,10].Calcium can improve both corrosion resistance in solutions with a high concentration of chloride,like the human body fluid and blood plasma,and mechanical properties of Mg alloys.Moreover,as a major component of human bone,the release of Ca2+ions as a degradation product could benefit bone healing.Zinc is one of the most abundant nutritionally essential elements in human body.It is beneficial in increasing the tolerance limits of impurities and thus improves corrosion resistance.Also,Zn can improve castability and strength.In fact,it was found to be next to Al in strengthening effectiveness as alloying element in Mg.

Mg-Zn-Ca alloys have recently attracted much attention as candidates to be used for orthopaedic applications due to their gradual biodegradation in physiological environment,their superior biocompatibility and their good age-hardening response compared with the binary Mg-Ca and Mg-Zn alloys[9,10].Several studies [11-16] proved that the noticeable improvement of mechanical properties of these alloys after heat treatment,which comprises solution treatment,quenching and ageing,is attributable to a uniform precipitation of very fine secondary phases (mainly Ca2Mg6Zn3) within theα-Mg matrix phase.However,very few studies [17,18] focused on the influence of heat treatment on the biocorrosion behaviour of Mg-Ca-Zn alloys,despite the known role of microstructure in corrosion properties.Indeed,most of the studies devoted to the biocorrosion behaviour of these alloys [19-23] focused only on the effect of the Zn/Ca atomic ratio after solidifica tion.

In the present investigation,the age-hardening response of the Mg-1wt.% Zn-1wt.% Ca (ZX11) and Mg-3wt.% Zn-0.4wt.% Ca (ZX30) alloys is evaluated.In addition,threepoint bending tests as well as electrochemical and immersion tests in Hank’s solution were performed on both alloys in the as-cast,solution-treated,peak-aged and over-aged conditions.The main objective of this study was to assess the effect of heat treatment on the mechanical and biocorrosion perfor-mance of Mg-Zn-Ca alloys,which was rationalised in terms of second phase content and distribution.

2.Materials and experimental procedure

2.1.Preparation of Mg alloys

The materials investigated are the ZX11 and ZX30 alloys,with a Zn/Ca atomic ratio of 0.50 and 4.75,respectively.The alloys were supplied in the as-cast condition by the Helmholtz Zentrum Geesthacht (Germany).The details of the casting process can be found elsewhere [24].The chemical composition of the alloys,analysed by spark emission spectroscopy,is shown in Table 1.

Table 1Chemical composition (wt.%) and Zn/Ca atomic ratio of the studied ZX11 and ZX30 alloys.

Table 2Chemical composition of Hank’s solution (g/L).

2.2.Heat treatments

Sections of the ZX11 and the ZX30 alloy cast ingots were solution treated at 450 °C for 24 h in an electric furnace and immediately quenched in water in order to preserve the microstructure.Afterwards,the samples were aged in an oil bath at 180 °C.The ageing response of both alloys was monitored by Vickers microhardness tests with a load of 0.1kg.Vickers hardness was determined from 10 individual indentations made on each specimen.

2.3.Microstructural characterisation

Microstructural examinations were performed by Scanning Electron Microscopy (SEM) and Transmission Electron Microscopy (TEM).SEM was conducted in a Hitachi S-3400N microscope and TEM in a JEOL JEM 2100 microscope,both equipped with an energy dispersive spectrometer (EDS).In preparation for SEM,samples were manually ground and polished down to 1μm with progressively fine SiC emery papers and diamond pastes and then chemically etched in a Nital solution (2vol.% nitric acid and 98vol.% ethanol).For TEM examinations,thin foils,prepared by mechanical grinding down to approximately 50μm,were punched into discs of 3mm in diameter and then ion milled to perforation using a Gatan precision ion polishing system.

2.4.Three-point bending tests

Samples of dimensions 2.5×1×30mm3were tested under three-point bending conditions.Before the tests,one of the surfaces of the bending test samples was ground and polished down to 1μm and then etched in a Nital solution.Three-point bending tests were carried out at room temperature and a constant displacement rate of 0.5mm·min-1using a Deben bending machine in the above-mentioned SEM.Thetests were performed to failure,but they were interrupted at different displacements in order to examine the crack progression in the samples.At least two tests were performed for each condition.

2.5.Electrochemical tests

Electrochemical tests were carried out using an Autolab PGSTAT302N potentiostat-galvanostat,provided with the Nova 2.1 software.The tests were conducted at room temperature using a standard three-electrode cell configuration consisting of a silver/silver chloride(Ag/AgCl,KCl 3M)electrode as a reference,a graphite rod as the counter electrode and the ZX11 and ZX30 alloy samples,with an exposed area of 0.78 cm2,as the working electrode.Hank’s Balanced Salt Solution (Hank’s solution),whose composition is shown in Table 2,was used as electrolyte.The samples were ground with SiC emery papers up to 1200 grade before the tests.

Table 3Electrochemical parameters of the ZX11 and the ZX30 alloy samples in Hank’s solution obtained from the anodic-cathodic polarisation tests.

Anodic-Cathodic Polarisation measurements were performed after 1h of immersion.The curves were obtained at a scanning rate of 1mV/s,polarising the samples from-400mV to 800mV relative to their Open Circuit Potential(OCP).Before the tests,samples were immersed in the Hank’s solution until a steady value for their OCP was reached.Three tests were performed for each sample.

Linear polarisation tests were performed in order to obtain the polarisation resistance (Rp) of the samples.The scanning rate was 1mV/s for all the samples and the applied potential ranged±10mV from their OCP.Rpvalues were acquired after different immersion times ranging from 0 to 168h.Three tests were performed for each sample.

2.6.Immersion tests

In order to estimate the degradation or corrosion rates,immersion tests were performed in Hank’s solution saturated with hydrogen gas at pH 7.4 and 37 °C.Three different specimens,with dimensions of 30×20×2.5mm3,were used for each condition of both alloys.They were ground with SiC emery papers up to 1200 grade and afterwards immersed in the thermostatic bath for up to 336h.When the samples are immersed in the solution,Mg is oxidised by water,what generates hydrogen (H2) and magnesium hydroxide(Mg(OH)2).The reaction is as follows:

The hydrogen bubbles generated by the soaked samples were collected with a funnel into a burette [25],the degradation rate of magnesium being quantified via the amount of evolved hydrogen over time.However,this reaction causes a gradual increase in the pH of the solution because,although magnesium hydroxide is quite insoluble,a certain amount dissociates into ions (Mg2+,OH-) in aqueous solution.In order to keep the pH at a value of～7.4 for the duration of the tests,carbon dioxide (CO2) was bubbled through the solution.When this gas dissolves in water,a small amount readily and reversibly converts into carbonic acid (H2CO3) through the following reaction:

Fig.1.Variation of hardness with ageing time at 180 °C for the ZX11 and the ZX30 alloys.

Carbonic acid releases protons (H+) into the solution,then leading to the neutralisation of its alkalinity.

After the immersion tests,the corrosion progress in the samples was evaluated by SEM.

3.Results

3.1.Age-hardening response

Fig.2.SEM micrographs at two different magnification of the ZX11 alloy samples:a) as-cast,b) solution-treated,c) peak-aged and d) over-aged.

3.2.SEM examinations

Figs.2 and 3 show,respectively,SEM micrographs and EDS results of the ZX11 alloy in the as-cast condition as well as in three different thermal conditions,namely,solutiontreated,solution-treated and subsequently aged for 3 h(named hereafter peak-aged condition) and solution-treated and subsequently aged for 24 h (named hereafter over-aged condition).All the samples show a homogeneous grain size distribution,the mean grain size being 67±3μm [25],and a remarkable concentration of particles at the grain boundaries(GBs).According to the qualitative EDS analyses performed on the GB-precipitates,the as-cast alloy has two different kinds of particles (Figs.2a and 3a).Both contain Zn and Ca,but one of them (with bright contrast) is richer in Zn,while the other one (with dark contrast) is richer in Ca.These observations confir that,as expected for Mg-Zn-Ca alloys with low (＜1.2,approximately) Zn/Ca atomic ratios after solidifi cation [19,21],the microstructure of the as-cast alloy is composed ofα-Mg matrix grains and intergranular Ca2Mg6Zn3and Mg2Ca phases.The binary phase appears to be embraced by the ternary one because of the precipitation sequence during the solidification process[25].The solution-treated and aged samples,though,show a more disconnected configura tion of GB-particles (Figs.2b-d) and only have Ca-rich precipitates (Figs.3b-d).This suggests that,in agreement with the thermodynamic predictions within the Mg-Zn-Ca system[14,15],only the Ca2Mg6Zn3phase is dissolved during the solution treatment.This phase is expected to re-precipitate within the grains as very fine particles during ageing.It is worth noting that the Mg2Ca phase in the four ZX11 alloy samples is enriched with Zn (Fig.3).

Fig.3.High-magnification SEM micrographs and results of EDS mapping and compositional analysis of the ZX11 alloy samples:a)as-cast,b)solution-treated,c) peak-aged and d) over-aged.

Figs.4 and 5 show,respectively,SEM micrographs and EDS results of the ZX30 alloy in the as-cast condition as well as in three different thermal conditions,namely,solution-treated,during solidification solution-treated and subsequently aged for 4 h (named hereafter peak-aged condition) and solution treated and subsequently aged for 24 h(named hereafter over-aged condition).The as-cast alloy shows a microstructure consisting ofα-Mg grains,with a mean size similar to those of the as-cast ZX11 alloy(Fig.2a) and large particles preferentially located at the GBs(Fig.4a).These particles,rich in Zn and Ca,especially in the former (Fig.5a),were identified as (α-Mg+Ca2Mg6Zn3)eutectic phases [18,21,26].This is consistent with the suppression of precipitation of the Mg2Ca phase reported for the Mg-Zn-Ca alloys with high (＞1.2,approximately) Zn/Ca atomic ratios [19,21,27].In line with earlier studies [15,28],the present ZX30 alloy still exhibits a large amount of particles at the GBs after solution and ageing treatments (Figs.4b-d),verifying that precipitate dissolution is modest in this alloy.However,the GB-precipitates show a more connected configuration in the solution-treated and aged samples,which also show spherical colonies of the eutectic phase within the grains as well as coarser and more equiaxed grains.These observations suggest that a redistribution of the undissolved eutectic phases,as well as discontinuous recrystallisation or grain growth,take place during the solution treatment.The latter could be benefite by the lack of the thermally stable Mg2Ca particles [1],which might have a strong effect on restraining GB migration.

3.3.TEM examinations

The second-phase particles within the grains of the ZX11 and ZX30 alloys in the peak-aged condition were characterised by TEM(Fig.6).The microstructure of the ZX11 alloy sample comprises a heterogeneous distribution of irregularlyshaped particles with a mean size of about 1μm throughout theα-Mg matrix phase (Fig.6a).EDS analyses carried out on these particles indicated a composition of～60 at.%Mg,～36 at.% Ca and～4 at.% Zn,which is compatible with that of the Mg2Ca phase enriched with Zn [1].A fine scale characterisation revealed the presence of particles with a few nanometres in size (Fig.6b),which were identified as Ca2Mg6Zn3[14,15].The ZX30 alloy sample contains particles with a mean size of～200-500nm together with a very small number of particles with a few nanometres in size(Fig.6c).According to the qualitative EDS analyses,the larger particles are rich in Zn,but contain no virtually Ca.Earlier studies on the age-hardening response of Mg-Zn-Ca alloys with high Zn/Ca atomic ratios [16,29] revealed the coexistence of coarse Zn-rich precipitates (β1-MgZn2or Mg4Zn7) with fine Ca2Mg6Zn3particles after ageing,which is in good agreement with the present observations.The reduced number of fine Ca2Mg6Zn3precipitates in the peak-aged ZX30 alloy accounts for the limited agehardening response of this alloy compared with the ZX11 alloy.

Fig.4.SEM micrographs at two different magnification of the ZX30 alloy samples:a) as-cast,b) solution-treated,c) peak-aged and d) over-aged.

3.4.Three-point bending tests

Fig.7 shows the stress-displacement curves obtained from the three-point bending tests on the ZX11 alloy samples,respectively.The maximum bending stress of the ZX11 alloy decreases from～95MPa (Fig.7a) in the as-cast condition to～75MPa after solution treatment (Fig.7b).However,ageing causes the maximum bending stress of the alloy to increase to～135MPa in the peak-aged condition (Fig.7c) and to～110MPa in the over-aged condition (Fig.7d).The displacement to failure is lower for the as-cast condition than for the three thermally-treated samples.Accordingly,in-situ SEM examinations revealed that in the as-cast sample (Fig.8a),showing a more connected configuration of GB-precipitates(Fig.2),cracks easily propagate along the GBs,while in the rest of the ZX11 alloy samples,individual cracks nucleate at the large GB-particles,but they do not preferentially propagate along the GBs (Fig.8b).

Fig.5.High-magnification SEM micrographs and results of EDS mapping and compositional analysis of the ZX30 alloy samples:a)as-cast,b)solution-treated,c) peak-aged and d) over-aged.

The stress-displacement curves obtained from the threepoint bending tests on the ZX30 alloy samples are shown in Fig.9.The maximum bending stress decreases from～90MPa(Fig.9a) to～30MPa after solution treatment (Fig.9b),but increases back to～90MPa after ageing (Figs.9c,d).The displacement to failure is also lower for the solution-treated than for the as-cast and aged samples.In-situ SEM inspections revealed that,in the as-cast sample (Fig.10a),individual cracks nucleate at the large GB-particles,but they do not preferentially propagate along the GBs.However,in the solution-treated and aged samples,showing a more connected configuration of GB-particles (Fig.4),cracks easily propagate along the GBs (Fig.10b).This is consistent with the low displacement to failure of the solution-treated sample,but not with the higher displacement to failure of the two aged samples,which appears to be related to the presence of fine Ca2Mg6Zn3particles within the grains.Note that,unlike hardness,the ZX30 alloy exhibits lower bending stress than the ZX11 alloy in the solution-treated and peak-aged conditions.This suggests that the Mg2Ca particles,present in the latter alloy but not in the former,may pose more of an obstacle for crack nucleation and/or propagation than the Ca2Mg6Zn3particles,present in both alloys.

Fig.6.TEM images of the peak-aged alloys:(a),(b) ZX11 and (c) ZX30.

Fig.7.Stress-displacement curves from three-point bending tests of the ZX11 alloy samples:a) as-cast,b) solution-treated,c) peak-aged and d) over-aged.Discontinuities correspond to breaks done during the tests to obtain in-situ SEM images.

3.5.Electrochemical tests

The evolution of the OCP with the immersion time in Hank’s solution for the samples of the two Mg-Zn-Ca alloys under study is illustrated in Fig.11.All the OCP curves display a sharp rise during the firs 20 h,then a softer rise until～100 h of immersion and after that,a steady state,indicating the stable growth of a protective surface film Before the steady state,the OCP values of the different samples slightly differ from one to another for both alloys.However,in the steady state,the OCP reaches values of about -1.56V and-1.51V for the ZX11 (Fig.11a) and the ZX30 alloy samples (Fig.11b),respectively.Since Zn is nobler,while Ca is less noble than Mg,the higher OCP values for the ZX30 alloy samples are explainable by its higher (Zn content)/(Ca content) ratio.

Fig.8.In-situ SEM images showing crack propagation during three-point bending tests of the a) as-cast and b) peak-aged ZX11 alloy samples.

Fig.9.Stress-displacement curves from three-point bending tests of the ZX30 alloy samples:a) as-cast,b) solution-treated,c) peak-aged and d) over-aged.Discontinuities correspond to breaks done during the tests to obtain in-situ SEM images.

Fig.10.In-situ SEM images showing crack propagation during three-point bending tests of the a) as-cast and b) peak-aged ZX30 alloy samples.

Fig.11.Variation of the OCP with the immersion time in Hank’s solution:a) ZX11 and b) ZX30 alloy samples.

Fig.12.Anodic-cathodic curves after 1h of immersion in Hank’s solution:a) ZX11 and b) ZX30 alloy samples.

Fig.12 presents the anodic-cathodic curves of the ZX11 and the ZX30 alloys in the various thermal conditions after 1h of immersion in Hank’s solution.The corrosion potential(Ecorr) and current density (icorr) values extracted from the curves are compiled in Table 3.It can be seen that three thermally-treated ZX11 alloy samples show more positive Ecorrvalues than the as-cast one (Fig.12a).However,while the over-aged sample exhibits lower icorrthan the as-castsample,the opposite happens with the solution-treated and peak-aged samples.Regarding the ZX30 alloy (Fig.12b),both solution treatment and ageing tend to result in lower electrochemical corrosion resistance,the solution-treated and aged samples generally showing more negative Ecorrvalues and higher icorrvalues than the as-cast sample.Note that,in agreement with the relative OCP values (Fig.11),the Ecorrvalues are lower for the ZX11 than for the ZX30 alloy samples.

Fig.13 gives the evolution of the polarisation resistance(Rp) with immersion time in Hank’s solution for the ZX11 and ZX30 alloy samples.It is patent that the ZX11 alloy samples (Fig.13a) reach higher Rpvalues than the ZX30 alloy ones (Fig.13b),indicating a superior corrosion resistance of the ZX11 alloy.Fig.13 also shows that,for all the samples,Rpincreases during the firs few hours (～5h) of immersion.However,after these few hours in contact with the electrolyte,the corrosion resistance of the samples declines and,consequently,Rpfalls down until,after 20-50 h,the samples become protected by a stable surface film from corrosion products and then Rpreaches a steady value.

Now it so happened that on one occasion the princess s golden ball did not fall into the little hand which she was holding up for it, but on to the ground beyond, and rolled straight into the water. The King s daughter followed it with her eyes, but it vanished, and the well was deep, so deep that the bottom could not be seen. On this she began to cry, and cried louder and louder, and could not be comforted. And as she thus lamented1 some one said to her, What ails2 thee, King s daughter? Thou weepest so that even a stone would show pity. She looked round to the side from whence the voice came, and saw a frog5 stretching forth3 its thick, ugly head from the water. Ah! old water-splasher, is it thou? said she; I am weeping for my golden ball, which has fallen into the well.

Fig.13.Polarisation resistance (Rp) as a function of immersion time in Hank’s solution:a) ZX11 and b) ZX30 alloy samples.

3.6.Immersion tests

The results of the hydrogen evolution tests for the ZX11 and the ZX30 alloy samples in Hank’s solution are shown in Fig.14.All the samples display a similar increasing trend.For the ZX11 alloy (Fig.14a),the as-cast sample exhibits higher hydrogen evolution rate than the solution-treated and the over-aged samples,but lower than the peak-aged one.In agreement with the icorrvalues from the anodic-cathodic tests of the ZX11 alloy samples (Table 3),the highest hydrogen evolution rate corresponds to the peak-aged condition,while the lowest one corresponds to the over-aged one.For the ZX30 alloy (Fig.14b),the as-cast sample exhibits lower hydrogen evolution rate than the solution-treated and the two aged samples,which is consistent with the results of the anodic-cathodic tests for this alloy (Table 3).Note that,for both alloys,the highest corrosion rate corresponds to the peak-aged condition,the over-aged condition exhibiting much better corrosion behaviour.In particular,it displays the lowest corrosion rate for the ZX11 alloy and a very similar corrosion rate to that of the as-cast sample for the ZX30 alloy.Finally,if the ZX11 alloy is to be compared with the ZX30 one,it can be seen that,overall,the ZX11 alloy samples (Fig.14a) show lower hydrogen evolution rate and hence lower corrosion rate than the ZX30 alloy samples (Fig.14b).

Fig.15 presents SEM images showing the advance of corrosion through the thickness of the ZX11 alloy samples after the immersion tests in Hank’s solution for 336h.The corroded microstructure of the as-cast alloy (Fig.15a)shows extensive and preferential,although relatively uniform,attack of theα-Mg matrix,the GB-precipitates remaining relative unaffected and apparently acting as physical barriers against corrosion.The solution-treated and peak-aged samples(Figs.15b,c) show evidence of pitting corrosion after immersion(see red arrows),this being especially severe in the latter,which exhibits the highest corrosion rate (Fig.14a).Finally,in consistence with its lowest corrosion rate (Fig.14a),quite uniform and fla corrosion is observed in the over-aged sample,where preferential attack of theα-Mg matrix phase is no longer noticed (Fig.15d).The lack of localised attack at the GBs in the ZX11 alloy samples,which would be expected from the anodic role of the Mg2Ca phase in relation to theα-Mg matrix [19-21],could be ascribed to the observed enrichment of this secondary phase with Zn (Fig.3).The dissolution of Zn atoms in the Mg2Ca phase lattice has been proved to increase the corrosion potential of the Mg2Ca and thus its resistance against the corrosion attack [22,23].Additionally,in the case of the as-cast sample,the lack of preferential attack at the GBs could be also ascribed to the surrounding of the Mg2Ca phase by the Ca2Mg6Zn3one,which prevents the formation of a galvanic couple between the Mg2Ca andα-Mg phases [25].

Fig.14.Volume of hydrogen evolved during the 336-hour-long immersion tests in Hank’s solution:a) ZX11 and b) ZX30 alloy samples.

Fig.15.SEM images from the cross-sections of the ZX11 alloy samples after immersion in Hank’s solution for 336 h:a) as-cast,b) solution-treated,c) peak-aged and d) over-aged.

Fig.16.SEM pictures from the cross-sections of the ZX30 alloy samples after immersion in Hank’s solution for 336 h:a) as-cast,b) solution-treated,c) peak-aged and d) over-aged.

The advance of corrosion through the thickness of the ZX30 alloy samples after the immersion tests in Hank’s solution for 336h is displayed in Fig.16.In accordance with the cathodic action of the eutectic (α-Mg+Ca2Mg6Zn3) in relation to the matrix [19,20],the four samples show selective corrosion of theα-Mg phase.However,despite the GB-precipitates seem to act as physical barriers against corrosion progress in the solution-treated and aged samples (see yellow arrows in Fig.16c),the extent of corrosion in these three samples is much larger than in the as-cast one (Fig.16a),which shows the lowest corrosion rate (Fig.14b).Furthermore,in the solution-treated and peak-aged samples,the GB-precipitates appear attacked at the triple points (see green arrows),suggesting that,in these two samples,even theα-Mg in the eutectic is dissolved during immersion.This extra path for corrosion is consistent with the observed higher corrosion rates of the solution-treated and peak-aged conditions compared with the over-aged sample,which exhibits a corrosion rate very similar to that of the as-cast sample (Fig.14b).

4.Discussion

4.1.Mechanical behaviour

The as-cast microstructures of the present ZX11 and ZX30 alloys are characterised byα-Mg matrix grains and large inter-granular particles,these comprising both the Mg2Ca and Ca2Mg6Zn3phases for the ZX11 alloy (Figs.2a and 3a),but only the Ca2Mg6Zn3phase for the ZX30 alloy (Figs.4a and 5a).During solution treatment only the Ca2Mg6Zn3phase dissolves,the dissolution being complete in the ZX11 alloy (Figs.2b-d and 3b-d),but very scarce in the ZX30 alloy (Figs.4b-d and 5b-d).In fact,in the ZX30 alloy,the Ca2Mg6Zn3phase is redistributed to form a more connected configuration of GB-precipitates rather than dissolved.As a result,the age-hardening response of the ZX30 alloy is poorer than expected,so that peak hardness is very similar for both alloys (Fig.1),despite the higher content in alloying elements of the ZX30 alloy.This indicates that an equilibrated combination of Zn and Ca is more effective than the individual contents in Zn and Ca in strengthening Mg alloys.

In agreement with the age-hardening behaviour of both alloys(Fig.1),the ZX11 alloy shows higher maximum bending stress in the aged conditions than in the as-cast one (Fig.7),while the ZX30 alloy displays essentially the same maximum bending stress in both the as-cast and aged conditions (Fig.9).Moreover,in-situ SEM examinations during three-point bending tests(Figs.8 and 10),revealed that cracks nucleate at the fragile large GB-particles.In this way,when GB-particles exhibit a relatively connected configuration as it is observed in the as-cast ZX11 alloy (Fig.2a) and in the three thermallytreated ZX30 alloy samples(Figs.4b-d),GB-particles provide a preferential pathway for crack propagation and then cracks can easily propagate across the sample along the GBs (Figs.8a and 10b).This is consistent with the low ductility of the as-cast ZX11 alloy (Fig.7a) and the solution-treated ZX30 sample (Fig.9b),the unexpectedly high ductility of the aged ZX30 samples (Figs.9c,d) then suggesting that the fine Ca2Mg6Zn3particles formed during ageing within the grains may provide the present alloys with extra toughness.

4.2.Biocorrosion behaviour

The present results show that,favoured by the electrochemical inactivity of the Mg2Ca phase,caused by its enrichment with Zn,the corrosion behaviour of both the ZX11 and ZX30 alloys depends on the thermal condition or,in other words,on the distribution of the Ca2Mg6Zn3phase,with a cathodic role in relation to theα-Mg phase.In particular,for the ZX11 alloy,the as-cast sample exhibits higher hydrogen evolution rate than the solution-treated and the over-aged samples,but lower than the peak-aged one (Fig.14a).In contrast,for the ZX30 alloy,the as-cast sample exhibits lower hydrogen evolution rate than the solution-treated and the two aged samples(Fig.14b).

For the ZX11 alloy,the lower corrosion rate of the solution-treated sample compared with the as-cast one could be attributed to the dissolution of the large Ca2Mg6Zn3particles at the GBs (Figs.2a,b and 3a,b).That is to say,the absence of large Ca2Mg6Zn3particles in the solution-treated ZX11 sample hinders the formation of a galvanic couple between the Ca2Mg6Zn3phase and theα-Mg matrix and thus leads to a reduced corrosion rate related to the as-cast condition.On the contrary,in the case of the ZX30 alloy,the redistribution of the un-dissolved Ca2Mg6Zn3particles at the GBs during solution treatment to form a more connected configuration (Figs.4a,b and 5a,b) could promote galvanic corrosion in the solution-treated sample.It is worth mentioning that the effect of a continuous network of Ca2Mg6Zn3precipitates at the GBs seems to be twofold.On one hand,as verified by the corroded microstructures of the solutiontreated and aged ZX30 alloy samples (Figs.16b-d),a continuous network of GB-precipitates can act as effective barriers against corrosion,tending to block the advance of the corrosion front.On the other hand,a continuous network of Ca2Mg6Zn3precipitates can form a very effective galvanic couple with theα-Mg phase,which appears to be the dominant effect.The connected configuration of the GBparticles observed in the solution-treated sample remains after ageing,so that the corrosion rate of the aged samples is also higher than that of the as-cast sample.Some works[19,20,22,27] have also claimed a beneficial effect of a fine grained microstructure on the corrosion resistance of Mg-Zn-Ca alloys.So,in the case of the ZX30 alloy,the grain coarsening experienced during the solution treatment (Figs.4a,b) could contribute to the higher corrosion rate of the solution-treated and aged conditions compared to the as-cast one.

During ageing,a uniform distribution of the Ca2Mg6Zn3phase into fine precipitates takes place.Several studies[17,18,26] have reported the beneficial effect of heat treatment,involving solution treatment,quenching and ageing on the corrosion behaviour of Mg-Zn-Ca alloys due to a more uniform distribution of the cathodic sites.However,in the present work,the peak-aged condition exhibits the highest corrosion rate for both alloys (Fig.14).The corroded microstructure of the peak-aged ZX11 alloy sample(Fig.15c)suggests that the superior corrosion rate of the peak-aged samples could be ascribed to the occurrence of pitting corrosion[30,31],attributable to an excessive number of cathodic sites.Moreover,the fine Ca2Mg6Zn3precipitates formed within the grains during ageing may represent obstacles for a continuous passive film to form,ultimately leading to a remarkable increase in pitting susceptibility and a noticeable loss of corrosion resistance.During over-ageing,pitting corrosion could become more uniform or the pitting rate could decrease,presumably,due to a substantial decrease in the number density of Ca2Mg6Zn3precipitates within the grains resulting from the coarsening process [31].As a result,the over-aged condition shows the lowest corrosion rate for the ZX11 alloy and a corrosion rate very similar to that of the as-cast sample for the ZX30 alloy (Fig.14b).

This work also shows that the corrosion resistance of the ZX11 alloy is generally higher than that of the ZX30 alloy (Figs.13 and 14).Previous works [19,21] have shown an improvement in the corrosion behaviour of the Mg-Zn-Ca alloy when the Zn/Ca atomic ratio was lower than～1.2.Both works highlighted the important role of the Mg2Ca phase in the improvement of the corrosion behaviour of these alloys.Baksheshi-Rad et al.[21]suggested that the anodic role of the Mg2Ca phase with respect to the cathodic action of theα-Mg and Ca2Mg6Zn3phases causes the preferential dissolution of the Mg2Ca precipitates along the GBs,which may postpone the commencement of the corrosion attack in primaryα-Mg and,in turn,lead to a reduced corrosion rate of the alloy.Zhang and Yang[19],nonetheless,suggested that the fast corrosion of the Mg2Ca phase leads to a quick increase of Ca2+in the corrosive solution and then to an accelerated formation of a Ca-P protective film So,in both cases,the increased resistance against corrosion of the Mg-Zn-Ca with low Zn/Ca atomic ratios is based on the preferential dissolution of the Mg2Ca phase along the GBs.In the present work,the Mg2Ca phase in the ZX11 alloy samples remained uncorroded during immersion in Hank’s solution,which was attributed to an increased corrosion potential of the Mg2Ca phase caused by its enrichment with Zn (Fig.3).So,the increased corrosion resistance of the ZX11 alloy in relation to that of the ZX30 cannot be related to the selective attack of the Mg2Ca precipitates,but to a lower content of the Ca2Mg6Zn3phase and thus to reduced galvanic corrosion of the former.

5.Conclusions

The as-cast Mg-1wt.% Zn-1wt.% Ca (ZX11) and Mg-3wt.% Zn-0.4wt.% Ca (ZX30) alloys were subjected to heat treatment,comprising solution treatment at 450 °C for 24h,quenching and ageing at 180 °C.The microstructure,mechanical behavior (by hardness and three-point bending tests)and biocorrosion behavior (by electrochemical and immersion tests in Hank’s solution) were evaluated in both alloys in the as-cast,solution-treated,peak-aged and over-aged conditions.The following conclusions may be drawn from the present study:

1.After solidification both alloys exhibit a microstructure composed ofα-Mg grains separated by Mg2Ca and Ca2Mg6Zn3precipitates in the ZX11 alloy and by only Ca2Mg6Zn3precipitates in the ZX30 alloy.

2.Solution treatment results in the complete dissolution of the Ca2Mg6Zn3phase in the ZX11 alloy,but in a low dissolution degree of this phase in the ZX30 alloy.As a consequence,re-precipitation of fine Ca2Mg6Zn3precipitates within the grains during ageing and thus age-hardening response is more reduced in the ZX30 alloy.

3.Cracks were observed to nucleate at the fragile and large particles at the grain boundaries (GBs) during the threepoint bending tests.In this way,they can easily propagate across the sample along the GBs in those samples exhibiting a relatively connected configuration of GB-particles,which leads to a reduced ductility.In contrast,the fine Ca2Mg6Zn3particles,present within the grains in the aged samples,may provide the alloys with extra toughness.

4.The Mg2Ca phase in the ZX11 alloy samples remained uncorroded during immersion in Hank’s solution.This was attributed to an increased corrosion potential of the Mg2Ca phase caused by its enrichment with Zn,which limits the formation of a galvanic couple between the Mg2Ca and theα-Mg phases.

5.In the ZX11 alloy,the dissolution of the large Ca2Mg6Zn3particles at the GBs leads to lower galvanic corrosion and thus corrosion rate in the solution-treated sample than in the as-cast one.However,in the ZX30 alloy,solution treatment results in the redistribution of the undissolved Ca2Mg6Zn3precipitates to form a more connected network of GB-precipitates and,thus,in an increased galvanic corrosion and corrosion rate of the three thermally-treated samples in relation to the as-cast one.

6.The peak-aged condition exhibits the highest corrosion rate for both alloys,which may be ascribed to an excessive number density of fine Ca2Mg6Zn3precipitates.However,the over-aged condition exhibits the lowest corrosion rate for the ZX11 alloy and a corrosion rate very similar to that of the as-cast sample for the ZX30 alloy,attributable to a substantial decrease in the number density of Ca2Mg6Zn3precipitates within the grains.

7.The ZX11 alloy shows generally better biocorrosion behaviour than the ZX30 due to its lower content in the Ca2Mg6Zn3phase and thus reduced galvanic corrosion.

Acknowledgements

Financial support from the project ADITIMAT-CM S2018/NMT-4411 funded by the Madrid Regional Government is gratefully acknowledged.The research leading to these results has also received funding from the Spanish State Research Agency under the project RTI2018-096391-B-C31,which is also acknowledged.One of the authors (N.Pulido-González) thanks the Spanish Ministry of Education,Culture and Sports for an FPU fellowship.Thanks also to M.Tinoco-Rivas,from the CNME,for assistance with TEM.