Degradation of AZ31 and AZ91 magnesium alloys in different physiological media:Effect of surface layer stability on electrochemical behaviour

Emmanuel Mena-Morcillo,Lucien Veleva

Applied Physics Department,Center for Investigation and Advanced Study,CINVESTAV-IPN,Carr.Ant.a Progreso Km.6,97310 Merida,Yucatan,Mexico

Received 12 December 2019;received in revised form 15 February 2020;accepted 23 February 2020 Available online 11 June 2020

Abstract This work deals with the degradation of AZ31 and AZ91 magnesium alloys when they are exposed to three types of physiological media for seven days at 37 °C:Ringer's,Hanks',and simulated body flui(SBF)solutions.A combination of immersions tests and surface characterisation methods were employed to evaluate the attack on the surface,and the stability of the formed corrosion product layers for each alloy/electrolyte system.Measurements of the Mg-ion released into the electrolytes were also carried out in order to be correlated with the degradation of the alloys.Electrochemical impedance spectroscopy(EIS)and potentiodynamic polarisation(PDP)techniques were employed to compare the performance of the alloys in these different aggressive electrolytes.According to the obtained results,the Mg-alloys exposed to Hanks'media were the less affected,which fact was attributed to a higher stability of the corrosion products layer formed in this medium,in comparison of those formed in Ringer's and SBF solutions.In addition,the corrosion damage was lower for AZ91 than for AZ31 alloy in all environments due to its higher Al content.The mass loss rates calculated from both immersion tests and electrochemical methods followed the same trend for comparative purposes between alloys.? 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:Mg-alloys;AZ-series;Surface layer;Physiological media;Mass loss rate;Degradation.

1.Introduction

Magnesium alloys have shown to be relevant materials for the development of novel bio-degradable medical implants[1-3].Nevertheless,when these alloys are exposed to physiological environments,they exhibit high degradation rates[4-6],attributed to the presence of chloride ions,which transform the magnesium hydroxide corrosion layer(formed on the alloy surface)into highly soluble MgCl2[5-7].Moreover,the corrosion rate and morphology of Mg-alloy surfaces after being attacked directly depend on their microstructure[8-11].The Mg-Al-Zn alloys(AZ-series)have been investigated in both in vitro and in vivo conditions for their potential application as medical implant materials[12-17].The microstructure of these alloys is typically formed by theα-matrix,theβ-phase(Mg17Al12)[18,19],as well as by other intermetallic particles,such as Al11Mn4,Al8Mn5,ε-AlMn,andβ-Mn(Al)[20].The choice of the chemical environment for the characterisation of the anodic dissolution of biodegradable materials is always important for an understanding of their corrosion activity and is sometimes chosen for the sake of convenience.For in vitro experiments,there are several synthetic solutions that aim to simulate the physiological environment.However,the difference in each solution composition impacts on the stability of the corrosion products layer that is formed on the tested material.

The aim of this work was to assess the degradation of two Mg-Al-Zn alloys,AZ31 and AZ91,after their exposure for seven days to three types of physiological media:Ringer's,Hanks',and simulated body flui(SBF)solutions,at 37 °C.Each solution was replaced every 8 h,to maintain their initial characteristics,taking into account their continuous fl w in the human body.The six alloy/electrolyte systems selected in this work were considered of interest for biomaterial science,where the effect of the solution composition in the corrosion products layer stability,and the electrochemical behaviour of the alloys were the focus of this research.It must be noted that the methods applied in this work consider a uniform corrosion mechanism for the investigated systems.

Table 1Chemical composition(wt%)of AZ31 and AZ91 Mg-alloys.

Table 2Reagents employed for the preparation of the physiological solutions(g L?1).

2.Materials and methods

2.1.Samples and solutions preparation

Table 1 lists the chemical composition of both AZ31(Alfa Aesar,Ward Hill,MA,USA)and AZ91(Magnesium Elektron Ltd.,Swindon,England)magnesium alloys.

Both materials were cut into square samples(1 cm2),some of which were used for immersion tests and others were embedded in epoxy resin for electrochemical experiments.Prior to measurements/tests,all samples were abraded with 400,800,and 1200 grit SiC papers,polished with 0.3μm Al2O3in ethanol,sonicated,and dried in warm air.

Three standard physiological test media were employed in this study,namely Ringer's solution(pH=6.1),Hanks'solution(pH=7.3),and simulated body flui(SBF,pH=7.5)[21-24].They were prepared with analytical grade reagents and ultrapure deionised water(18.2 MΩcm).The composition of the electrolytes is shown in Table 2.

2.2.Immersion tests and characterisation of the surface

Specimens by triplicate of AZ31 and AZ91 were placed in closed plastic containers and exposed to the physiological media(20 mL)for 7 days,following ISO 16428 standard[22].During the immersion tests,the temperature of the solutions was kept at 37 °C by employing a stable thermostatic bath(Model SC-15,Ecoshel Technology Ltd.,McAllen,TX,USA)equipped with a home-made holder for the containers.Besides,the electrolytes were replaced every 8 h to maintain a constant pH value and avoid alkalisation.The wastes were stored in independent containers for their further analysis by photometry(HI83200,Hanna Instruments,Woonsocket,RI,USA),in order to measure the concentration of the released Mg-ions.At the end of the test,the samples were withdrawn,rinsed with deionised water,and dried in air at room temperature.Scanning electron microscopy coupled with energy dispersive X-ray spectrometry(SEM-EDS,Philips XL-30,Amsterdam,the Netherlands)was performed on the alloys surface,in order to investigate their microstructure and morphologic changes after the exposure to the electrolytes.X-ray photoelectron spectroscopy(XPS,K-Alpha,Thermo Scientific Waltham,MA,USA)was used for the characterisation of the products formed on the surfaces after immersion tests.Mass loss measurements were carried out according to ASTM G31-12a standard[25].

2.3.Electrochemical experiments

Electrochemical impedance spectroscopy(EIS)and potentiodynamic polarisation(PDP)were performed by triplicate with an Interface-1000E potentiostat(Gamry Instruments,Inc.,Philadelphia,PA,USA).A three-electrode cell configuratio was employed for all measurements,with the alloys samples as working electrodes,a platinum mesh(Alfa Aesar,Ward Hill,MA,USA)as auxiliary,and a saturated Ag/AgCl/KCl reference electrode(CH Instruments Inc.,Austin,TX,USA).The impedance spectra were obtained using a perturbation amplitude of±10 mV(vs stabilised OCP after 2 h),and a frequency interval from 100 kHz to 50 mHz.EIS data was analysed and fitte to electric equivalent circuits with Gamry Echem Analyst software(Gamry Instruments,Inc.),where the polarisation resistance(Rp)was calculated from the f tting parameters.The potentiodynamic scans were carried out immediately after EIS measurements,starting from?0.3 V to+0.5 V vs OCP,at a scan rate of 1 mV s?1.

3.Results and discussion

3.1.Immersion test results

3.1.1.SEM-EDS analysis

Fig.1.SEM images of Mg-alloys surface before exposure:(A)AZ31,(B)AZ91;and after their exposure to Ringer's:(C)AZ31,(D)AZ91;Hanks:(E)AZ31,(F)AZ91;and SBF:(G)AZ31,(H)AZ91 solutions at 37°C.

Fig.1 presents SEM images of AZ31 and AZ91 surfaces before and after their exposure to each kind of physiological solution.The AZ31 alloy microstructure(Fig.1A)consisted of theα-matrix,and some intermetallics[26].In the case of AZ91(Fig.1B)theβ-phase(Mg17Al12)was present[27].It has been reported that the volume fraction of Mg17Al12phase in AZ91 will determine its degradation rate by acting as an anodic barrier or as a galvanic cathode[28-31].

After 7 days of exposure(Fig.1C-H),the corrosion layers on AZ31 and AZ91 surfaces had different morphologies and composition,depending on the solution in which they were immersed.For instance,for both alloys in Ringer's solution,their corrosion products(CP)layer(Fig.1C and D)did not allow to observe theβ-phase or the Al-Mn intermetallics.For Hanks'solution(Fig.1E and F),the CP layers were less irregular,and some cracks appeared.Different in appearance is the surface of both alloy surfaces after their exposure to SBF media(Fig.1G and H):the attacks seem homogeneous,but the CP layers are very cracked.On AZ91 surface(Fig.1F and H)the grey areas corresponding to theβ-phase(Mg17Al12),and few white particles related to Al-Mn intermetallics are still observable,which fact confirm their function as local cathodes.Table 3 presents the EDS elemental quantificatio results of the layers formed on AZ31 and AZ91 Mg-alloys surfaces(cf.Fig.1),after their exposure for 7 days to different physiological media.

Fig.2.High-resolution XPS spectra of:(A)Mg2p;(B)O1s;(C)Ca2p;and(D)P2p after AZ31(solid line)and AZ91(dashed line)samples exposure to different physiological media for 7 days at 37°C.The employed solutions are indicated by the following colors:(black)Ringer's;(blue)Hanks';and(red)SBF.

The analysis indicated that the elemental composition of the layers formed on each alloy surface in Ringer's solution is virtually similar,with a double percentage of oxygen in relation to magnesium.As regards Hanks'and SBF solutions,besides the quantificatio values of O and Mg,the percentages for Ca and P elements were significant

In order to correlate the elemental quantificatio analysis(Table 3)with the composition of the phases formed on each magnesium alloy surface,XPS was performed on exposed samples.The high-resolution spectra for Mg,O,Ca,and P is shown in Fig.2.

The spectrum for Mg showed one peak centred between 49.5 eV and 50.5 eV,whereas the spectrum for O displayeda peak centred between 531.1 eV and 531.5 eV,where both peaks could be associated to a mixture of MgO and Mg(OH)2phases[32,33],but with the magnesium hydroxide as dominant surface product.On the other hand,the spectrum for Ca split into two peaks centred around 347.4 eV and 350.8 eV respectively,as a result of spin orbit splitting,whereas the spectrum for P exhibit one peak between 133.2 eV and 133.6 eV,originated from bonding between(PO4)3?and Ca in the form of Ca10(PO4)6(OH)2phase[34,35].

Table 3EDS elemental composition(wt%)of corrosion layers formed on AZ31 and AZ91 Mg-alloy surfaces after exposure for 7 days to different physiological media at 37 °C.

Fig.3.Cross-section of the Mg alloys after immersion tests at 37°C:AZ31(A)and AZ91(B)Mg-alloy exposed to Ringer's solution;AZ31(C)and AZ91(D)to Hanks'solution;AZ31(E)and AZ91(F)to SBF solution.

Therefore,according to both EDS and XPS results,the corrosion products layer formed on both alloys when they were immersed in Ringer's solution was mainly composed of MgO/Mg(OH)2.As regards to the alloys that were immersed in Hanks'and SBF solutions,the products layer also contained Ca10(PO4)6(OH)2phase.

In a complementary way,Fig.3 presents the cross-section of each alloy/electrolyte system after immersion tests at 37 °C.The corrosion attack on AZ31 and AZ91 displayed cavities when the alloys were exposed to Ringer′s(Fig.3A and B)and SBF(Fig.3E and F)solutions,whereas for AZ31 and AZ91 exposed to Hanks'solution(Fig.3C and D)the corrosion attack seems more homogeneous.The cracks on the corrosion products f lm are probably caused by dehydration of the phase Mg(OH)2·nH2O to Mg(OH)2,when it was irradiated with the electron beam in SEM[35].The presence of phosphates in Hanks′and SBF media help to induce passivation on Mg-alloys due to the precipitation of calcium phosphate,promoted by the local rinse in pH during the anodic dissolution of Mg.In contrast to magnesium hydroxide,the calcium phosphate is inherently dense and compact,and it is barely affected by the Cl?ions[36].It was reported that in the presence of HEPES(acid buffering agent)in SBF,the phosphates passivating effect diminished[37].For these reasons,the CP layers of AZ31 and AZ91 alloys formed in Hanks'solution were more protective as they exhibited more stability against the formation of cavities in comparison with Ringer′s and SBF solutions,even when the phases present in the CP layers are similar.

Fig.4.EIS Nyquist plots of AZ31 and AZ91 Mg-alloys immersed for 2 hours in the electrolytes:(A)Ringer's,(B)Hanks',and(C)SBF.The respective equivalent electric circuit employed to fi the EIS spectra for each solution is located at the right of each plot.

Table 4Mass loss rates and Mg-ion release rates in the physiological media after 7 days of immersion tests(at 37 °C).The standard deviation was included.

3.1.2.Mass loss measurements and analysis of the solutions after the exposure of AZ31 and AZ91 surfaces

Table 4 presents the values of mass loss(MR)and Mg-ion release rates after 7 days of immersion tests at 37 °C for each alloy/electrolyte system.

The mass loss rate(MR)was calculated according to the following equation:

where,Δmis the mass loss(g),Ais the exposed area(m2),andtis the exposure time(days).

It can be seen(Table 4)that the higherMRcorresponds to AZ31 exposed to SBF and Ringer's solutions,whereas the values for AZ91 for the same solutions were lower and almost similar.It must be noted that each alloy exposed to Hanks'solution displayed the lowest values ofMR.As the degradation of AZ-Mg-alloys occursviaMg-ions release[38],the measurement of their concentration was also a subject of this study.Usually,corrosion takes place at preferential sites in the partially protective CP layer,where undermining of particles may occur[39].Also,Cl?ions increase the dissolution of such CP layer by reacting with Mg2+ions to form MgCl2compound,which displays high solubility in aqueous media(54.2 g in 100 mL of water[5-7]:

The calculated Mg-ion release rates after 7 days of immersion tests for both alloys are also presented in Table 4.The obtained values correspond to Mg-ion that remained in the solutions and did not react to form the CP layers,or were dissolved through action of chlorides,as stated above(Eq.(2)).Mg-ion release rates were calculated as follows:

where,Δ[Mg2+]is the difference between initial and fina concentration of Mg-ion(mg L?1),V is the volume of the solution(mL),A is the exposed area(m2),and t is the exposure time(days).It can be noted that the Mg-ion release rate values from the highest to lowest is as follows:AZ31/SBF＞AZ31/Ringer's＞AZ91/SBF＞AZ91/Ringer's＞AZ91/Hanks'＞AZ31/Hanks'.

The results in Table 4 suggest that the alloys exposed to SBF and Ringer's solutions were the most affected owing to a higher tendency of their CP layers to dissolve(or detach)than the CP layer on Hanks'solution.

3.2.Electrochemical analysis

3.2.1.EIS Nyquist diagrams

Fig.4 illustrates the EIS Nyquist diagrams of AZ31 and AZ91,after 2 h of exposure to the physiological media,as well as their respective equivalent electrical circuits.The Nyquist diagrams display capacitive loops at high and medium frequencies:one in the case of Hanks'solution,and two for Ringer's and SBF media.Those capacitive loops are usually attributed to charge transfer,fil effects,and mass transport[36].

The equivalent electrical circuits employed in this study(Fig.4E and F),were proposed after careful inspection of previous studies with EIS on Mg alloys[40-45],whereRswas related to the resistance of the solutions,R1to the initial corrosion stage,andR2to the discharge of intermediate adsorbed species[46].The elements CPE1,and CPE2represent the capacitive response of the possible corrosion products f lm,and the capacitance formed by the double layer at the metal/electrolyte interface.Constant phase elements(CPE)were employed instead of capacitors,owing to reported dispersion effects that may be produced by microscopic roughness of the alloy surface[47].

Table 5Fitting parameters obtained from the EIS spectra of AZ31 and AZ91 Mg-alloy exposed for 2 h to Ringer's,Hanks'and SBF solutions.The associated error of the f tting was included.

Fig.5.Potentiodynamic polarisation curves of:AZ31 and AZ91 Mg-alloys immersed in Ringer's,Hanks'and SBF solutions.

The quantificatio of the parameters representing the EIS equivalent circuits is provided in Table 5.

The values reflec the specific of the metal-electrolyte interfaces,created at both alloy surfaces,as well as the difference in the tested media compositions.The magnitude of the diameter of the semicircle in the high frequency range may help to estimate qualitatively the corrosion resistance of the alloys[48-51].In all cases,the diameter of AZ91 alloy semicircle was greater than that of AZ31,which fact indicated that,the corrosion resistance of AZ91 appears to be higher than that of AZ31 surface,at least at the initial corrosion stage.In addition,semicircle diameters were higher in Hanks'solutions than in the other media,which fact was attributed to the previously suggested stability of the CP layers formed on the alloys surface when exposed to this media:SEM images of the cross-section(Fig.3C and D)showed that the attack were less aggressive and each alloy displayed the lowest values of Mg2+release concentration(Table 4).

The polarisation resistance(Rp)is define as the resistance of the specimen to oxidation during the application of an external potential.The calculation ofRpis of interest in corrosion studies,since it is inversely proportional to the corrosion current(higher polarisation resistance means lower corrosion current).TheRpvalues can be extracted from EIS data by calculating the equivalent resistance of each equivalent electrical circuit[40-42]:R1for Hanks'solution,andR1+R2for Ringer's and SBF media.It can be observed(Table 5)that theRpvalues confir the behaviour expected by the qualitative analysis of the magnitude of the diameter of the semicircles in Nyquist plots.

3.2.2.Polarisation curves

The acquired curves for AZ31 and AZ91 alloy surfaces in the physiological media are plotted in Fig.5.

It may be noted that the anodic currents of both alloys immersed in Hanks'and SBF solutions tend to reach a limiting value,related with a semi-passivation process occurring at the alloy surface,followed by a sharp increase in the current which is associated with the dissolution of the oxide layer.This passive region was not observed in Ringer's media.Such difference observed in PDP curves is caused by the absence of PO43?,SO42?,CO32?anions in Ringer's solution,which promote the formation of phosphate,sulphate,and carbonate salts on the surface layer of the alloys[52-55].

Table 6Corrosion parameters obtained from PDP and EIS measurements,and calculated values of corrosion current(jcorr)and mass loss rate(MR).The associated error of the fittin was included.

Although it has been reported that PDP method is questionable to acquire the corrosion current density(jcorr)of magnesium-based materials from Tafel extrapolation[56,57],it is possible to simply extract the Tafel slopes from the pre-Tafel region in the vicinity of the corrosion potential(Ecorr)[58].The above was done in order to only consider the activation-controlled kinetics of the system.Then,if the polarisation resistance(Rp)is also acquired(as it was done by EIS technique),thejcorrvalue can be calculated through the Stern-Geary relation[59]:

where the Stern-Geary constant,B,is calculated using both anodic(βa)and cathodic(βc)Tafel slopes,as shown in the following equation[59]:

Once thanjcorris obtained,mass loss rate(MR)can be estimated from Faraday's law[59]:

whereKis a constant(K=8.954×10?3g cm2μA?1m?2d?1),andEwis the equivalent weight for each alloy(12.2 g/eq for AZ31 and 12.6 g/eq for AZ91 approximately).

Table 6 shows the corrosion parameters obtained from PDP and EIS measurements,as well as the calculated values of corrosion current(jcorr)and mass loss rate(MR).

TheMRs calculated by electrochemical methods indicate that AZ31 is less resistant to corrosion than AZ31 in Ringer's and SBF electrolytes,whereas the difference between theMRs of both AZ31 and AZ91 in Hanks'solution was quite similar.The lower degradation rate of AZ91 with respect to that of AZ31 was attributed to its higher content in Al,which promotesβ-phase(Mg17Al12)formation.The distribution of this phase plays a significan role in such protective behaviour.The distribution of this phase plays a significan role in such protective behaviour.Previous studies[28-31]reported that theβ-phase acts as a galvanic cathode,which promotes an acceleration of the corrosion process only if the volume fraction is small.In addition to this fact,large volume fractions may act as an anodic barrier and the overall corrosion diminishes.In this way,theβ-phase reduces the reactive surface area,so less area of the alloy is available to be corroded[28].

Results from Table 6 follow the same trend as theMRs obtained by immersion tests(Table 4).In general,mass loss rates calculated from electrochemical experiments do not match with those calculated by immersion tests.The above may be caused by two reasons:i)differences in the exposed area[60]:for immersion tests usually a 3D sample is exposed,whereas for electrochemical experiments a 2D area is restricted,which avoid edge contributions;and ii)the evolving hydrogen and products formation during Mg corrosion[61]:the reactions at the Mg surface may have been decoupled from the electrochemical measurement.

Thus,MRs calculated by immersion tests are generally higher than those obtained by electrochemical methods.It must be noted that for SBF this was not the case,though this exception would need further study.However,theMRs calculated from electrochemical methods showed to be reliable to compare the performance of alloys and give an insight of which material is prone to degrade faster.

4.Conclusion

The initial stages of degradation behaviour of AZ31 and AZ91 Mg-alloys were studied after their exposure for seven days to three types of physiological media:Ringer's,Hanks',and simulated body flui(SBF)solutions.Surface characterisation results revealed that the composition of the corrosion products(CP)layers was mainly a mixture of MgO/Mg(OH)2,but with the presence of Ca10(PO4)6(OH)2phase in Hanks'and SBF solutions,promoted by their composition.Moreover,the attack on the surface was less aggressive for both alloys immersed in Hanks'solution.The mass loss and Mg-ions release rates obtained from immersion tests indicated that the CP layers are unstable and prone to detachment,but this effect was diminished when the alloys were exposed to Hanks'solution.The mass loss rates(MRs),estimated from electrochemical methods,indicated that AZ31 presented less corrosion resistance than AZ91 in all electrolytes,which agrees with the trend ofMRs from immersion tests.The above was related to the higher Al content in AZ91,which causes an increase in the resistance of the CP layer.All obtained results indicated that Hanks'electrolyte causes less harm than the other physiological solutions,attributed to the formation of a more protective surface fil in this environment.This work is expected to be useful for researchers exploring the relevance of the used methods in this study,to assess the in vitro degradation of magnesium alloys.

Declaration of Competing Interest

The authors declare that they have no known competing financia interests or personal relationships that could have appeared to influenc the work reported in this paper.

Acknowledgements

Emmanuel Mena-Morcillo gratefully thanks CONACYT for his scholarship as Ph.D.student at CINVESTAV-IPN.The authors acknowledge LANNBIO-CINVESTAV for permitting the use of their facilities,as well to Dora A.Huerta-Quintanilla,and Wilian J.Cauich-Ruiz for their technical assistance.