Development and screening of (Ca-P-Si-F)-PEO coatings for biodegradability control of Mg-Zn-Ca alloys

Lara Moreno,Marta Mohedano,Raul Arrabal,Endzhe Matykina

Departamento de Ingeniería Química y de Materiales,Facultad de Ciencias Químicas,Universidad Complutense de Madrid,28040 Madrid,Spain

Abstract PEO coatings were developed on MgZnxCay alloys for biomedical applications using different treatment times and a novel transparent electrolyte containing Ca,P,Si and F.Surface characteristics and chemical composition of studied materials were evaluated using scanning electron microscopy,X-ray diffraction and optical metrology.Corrosion investigations were performed for uncoated and coated materials using potentiodynamic polarization,electrochemical impedance spectroscopy and hydrogen evolution measurements in three media (0.9wt.%NaCl, α-MEM and inorganic part of α-MEM).Findings revealed that Zn-rich intermetallics favour the formation of large voids in the coatings.Ca,P,F and Si are successfully incorporated into the coatings.Coatings consist of crystalline phases (MgO,MgF2 and Ca,Si-rich compounds) and amorphous material.Aminoacids and organic components of α-MEM solution accelerate corrosion of the bare substrates but have no significant influence on the corrosion resistance of coated specimens.Therefore,inorganic α-MEM can be considered as an appropriate medium for fast screening of PEO coatings.The relatively small amount of F content in the barrier layer facilitates early failure of the coatings,which show undercoating corrosion.Zn-rich intermetallic particles in the undercoating crevice induce a more aggressive corrosive environment and enhanced micro-galvanic couple effect with the Mg matrix.This leads to faster degradation of the coated alloys compared to the bare substrates.

Keywords: Magnesium;Biomaterials;Plasma electrolytic oxidation;Corrosion.

1.Introduction

During the last decade magnesium alloys have been actively investigated as materials for temporary biomedical implants;magnesium is abundant in the human body,it can be absorbed and metabolized without causing any adverse effect and is known to induce bone cell growth[1,2,3].The main limitation is the high corrosion rate of Mg causing excessive mass loss of the implant,accumulation of hydrogen gas bubbles(originating from the cathodic reaction)and the collateral localized pH increase in the physiological environment[4].These phenomena can lead to premature loss of mechanical integrity of the implant and an inflammatory response of the surrounding tissues[5].

In order to decrease the corrosion rate of Mg,two main strategies are employed: alloy design and surface treatments.There are a few considerations for element selection when developing Mg alloys as the degradation products must be non-toxic and absorbable by the human body[6].In the last years,Mg alloys incorporating aluminium and rare earth (RE)elements (Y,Nd,Gd,La,Nd,Zr) have been commercialized as implant biomaterials due to their low corrosion rate,excellent mechanical properties and favourable growth of cells on their surface[7,8].However,some studies have shown that both Al and RE elements may lead to harmful effects in the human body during implantation period[9,10,11].Therefore,it is necessary to research and commercialize Mg alloys based on non-toxic alloying elements.Zinc and calcium have been pinpointed as promising biocompatible alloying elements as they allow for improvement of mechanical properties and corrosion resistance[12].Anin vivostudy and a clinical trial in human patients demonstrated that a Mg5Zn1Ca alloy implant initiated bone calcification and within a year was completely replaced by the new bone,although hydrogen accumulation was notable during the first 4 weeks post-surgery,inhibiting cancellous bone formation around the implant during first 3 months[13].This demonstrates that the degradation rate of Mg-Zn-Ca alloys is still relatively high,possibly due to submicron and nano-size secondary phases (i.e.Mg2Ca and Ca2Mg6Zn3) forming microgalvanic couples with the matrix[11,14,15,16,17].

In order to improve the corrosion resistance of Mg alloys,a range of surface treatments such as anodizing techniques,conversion coatings,ion implantation,plasma spraying with generation of hydroxyapatite,bioceramics and bioactive glasses are available [2,12,18].Among them,plasma electrolytic oxidation (PEO) stands out due to its versatility and environmental friendliness.This technique consists on anodizing the material under high voltages and controlling the composition,microstructure,porosity and roughness by adjusting the electrical parameters and electrolyte composition[19].PEO coatings not only improve the corrosion resistance and tribocorrosion performance[20] of Mg alloys,but also facilitate osteoblast cell proliferation and adhesion,development of extracellular matrix and formation of hydroxyapatitein vitro[21],[22].In addition,PEO-modified implants improve the osseointegration and bone-to-implant contactin vivo[23].The main drawback of PEO coatings in terms of corrosion performance is insufficient long-term protection,which is mainly due to the solubility of magnesium oxide/hydroxides in neutral media.Additionally,defects such as pores and cracks,which are typically linked to the presence of intermetallic compounds in the alloy contribute to coating failure as they facilitate the ingress of corrosion species.

The electrolyte design is the key to incorporation of bioactive elements (Ca,P,Si and F) into the coatings in order to promote the biological response of bone cells[24,25].P-,Fand Ca-based additives in the electrolyte,in combination with appropriate electrical process parameters,lead to generation of compounds such as hydroxyapatite or fluorapatite which can improve cell adhesion and proliferation processes[26,27].Incorporation of Si into the coating enhances synthesis of collagen by osteoblast and matrix mineralization[28,24].

PEO coatings on the Mg-Zn-Ca system have been studied in a number of works.Wang et al.demonstrated that increasing the Si/P ratio in the electrolyte improved the coating growth rate on MgZnCa alloy by up to 50% and,consequently,the corrosion resistance in SBF solution by up to 25%[29].Shahri et al.reported that a～7 μm-thick PEO coating from a simple Na3PO4-and KF-based electrolyte on Mg5Zn0.4Ca alloy yielded a 2 orders of magnitude improvement of alloy’s corrosion resistance in PBS solution[14].Pan et al.developed～45 μm-thick coatings using Ca-P-F and Si-F based electrolytes and showed that silicate in the coating induced the nucleation of amorphous HA in SBF due to sorption of Ca and P ions and enhanced the corrosion resistance in SBF solutions[30,31].

The presence of F-containing species in the electrolyte enhances the corrosion resistance of coated Mg alloys,which is attributed to generation of MgF2in the ceramic coating[32,33].Moreover,in the presence of Ca-based salts and phosphates,fluoride induces the formation of CaF2and Ca5(PO4)3F (fluorapatite) which lead to considerable thickening of the coatings,further improving the corrosion resistance[34].Massive incorporation of F into the coating(24-39 at.%) and subsequent release of F-ions during immersion in biological media (up to～120 μg cm-2in the first 24 h) may bear some antibacterial effect.However,it also provokes cytotoxicity with respect to endothelial,preosteoblast and premyoblast cells (in descending order of influence)[24].Further,combination of phosphate and fluoride salts with Ca salts (e.g.calcium acetate is a typical electrolyte constituent) in the PEO electrolyte inevitably leads to electrolyte suspensions due to precipitation of calcium fluoride and calcium phosphates.Such suspensions require greater maintenance due to agglomeration of particles with time.

In the last years,several research works have been focused on electrolyte design strategy in order to explore true solutions[24,34],since suspension electrolytes are notoriously difficult in maintenance due to coagulation and coalescence of the precipitated particles (calcium phosphates).Such suspensions are not a significant handicap for laboratory studies on a small scale (and indeed are used in numerous studies,e.g.any electrolyte based on calcium acetate in combination with a phosphate salt),but are highly unadvisable for an industrial PEO technology.The poor solubility of the source Ca-containing compounds is a limitation for electrolyte transparency.Calcium acetate or calcium glycerophosphate are commonly used as a soluble Ca source.However,in combination with others elements present in the electrolyte (fluorides and phosphates) a precipitation of compounds with low solubility such as CaF2,Ca3PO4,CaHPO4,Ca(H2PO4)2typically happens.In order to avoid this precipitation,the electrolyte must include a strong complexing agent,such as ethylenediaminetetraacetic acid (EDTA)[35,36,37],that would form a strong and stable(i.e.soluble) complex calcium salt.

The selection of a corrosive media is an important part of the corrosion testing in order to understand the corrosion mechanism of Mg alloys.The common saline solutions used forin vitrocorrosion screening are phosphate buffered saline(PBS),simulated body fluid (SBF),Hank′s solution (HBSS),Ringer’s,etc.[4].In order to simulatein vivoconditions,media such as Dulbecco’s Modified Eagle’s Medium (DMEM),Modified Eagle’s Medium (MEM) andα-MEM are used[4].However,a recent review [38] have shown that organic additives in cell culture media have very little effect on corrosion rate of Mg alloys[11,12,13].This opens the possibility of using modified versions of the latter media which are stripped of organic content and,therefore,more stable and less prone to contamination.

The present study focuses on development of PEO coatings for MgZnCa and Mg3Zn0.4Ca alloys from a transparent electrolyte (i.e.a true solution) comprising Ca,P,Si and F species,known for producing highly corrosion resistant flash-PEO coatings [39] here adapted for biomedical applications using a much reduced content of fluoride.The corrosion behaviour of the coated materials is evaluated by electrochemical and hydrogen evolution measurements inα-MEM and the inorganic part ofα-MEM solutions in order to elucidate the corrosion inhibiting or activating role of organic and inorganic additives in a cell culture solution.The coatings are further screened by novelquasi-in vivo[40] hydrogen evolution measurements.It is revealed that the presence of a crevice generated by the onset of a uniform undercoating corrosion between a well-adhered coating and the substrate induces unexpectedly high degradation rates of PEO-coated systems,where the alloy microstructure warrants a strong micro-galvanic couple effect and the coating component lacks a strong barrier constituent.

2.Materials and methods

2.1.Materials

Cast ingots of MgZnCa and Mg3Zn0.4Ca (Table 1) alloys were supplied by Helmholtz-Zentrum Hereon,Institute of Surface Science (Geesthacht,Germany).The ingots were cut into 25 × 30 × 2.5 mm specimens and a 1.25 mm thread hole was drilled on one side for electrical contact.All samples were prepared by SiC papers to P1200 grit size,rinsed in isopropyl alcohol and dried in warm air.These alloys were selected as they were considered good representatives of the Mg-Zn-Ca system.

Table 1 Composition of alloys.

2.2.PEO treatment

An alternating current (AC) voltage-controlled power supply was used to carry out PEO treatment in a 2 L double jacket thermostatic glass cell.A MgZnxCayspecimen was used as a working electrode and a stainless steel cylinder mesh (AISI 316L,?15 cm) as a counter electrode.Electrical contact with the specimen was made through the threaded hole using an insulated copper rod.The transparency of the electrolyte was ensured using a calcium glycerophopshate(CaGly) complex salt and di-sodium ethylenediamine tetraacetate as a complexing agent(Table 2)[39].During PEO treatment,a square AC waveform with a 50% duty cycle was applied with an initial ramp of 60 s.Three treatment times were used in an attempt to study the effect of coating thickness(Table 3).After PEO,the samples were rinsed in deionized water,isopropyl alcohol and dried in warm air.

Table 2 Composition of the electrolyte.

Table 3 PEO process parameters.

2.3.Surface characterization

JEOL JSM-6400 and JEOL JSM-820 scanning electron microscope (SEM) equipped with an energy dispersive X-ray(EDS) detector were used to study the composition and morphology of the coatings.Cross-sections of selected specimens were prepared for examination by grinding to P1200 finish followed by polishing with 3 and 1 μm diamond pastes.Xray diffraction (XRD) analysis was carried out using a Philips X’Pert diffractometer (CuKα=0.154056 nm,0.01°/second scanning speed,2? range: 10 to 90°) in combination with X’PertHighScore software and the ICDD PDF4+database.Roughness parameters Sa(arithmetical mean height of the area) and Sz(sum of the largest peak height and the largest pit depth values within the area) were evaluated using a focus-variation optical 3D profilometer (InfiniteFocusSL,ALICONA).Coating thickness was determined as an average of ten measurements by Fischer ISOSCOPE-FMP10 portable eddy current meter and later confirmed with cross-sectional SEM micrographs.The pore population density and pore size of the coatings were estimated using ImageJ software.Image analysis of the coating surface was carried out using three SEM micrographs taken at arbitrarily locations.

2.4.Corrosive media

Corrosion testing of biomaterials is often carried out in 0.9% NaCl aqueous solution.This is a simple environment with many shortcomings.Therefore,in this study,0.9% NaCl was only used for bare alloys in hydrogen measurements in order to be able to compare them with other studies.Minimum essential medium Eagle-alpha modification (Merck),further referred to as completeα-MEM,is used in bioactivity evaluation experimentsin vitro24,41,42].Inorganic part ofα-MEM is a more recent approach and is used in order to establish an easy maintenance and reliable testing medium[43].In this study,complete and inorganicα-MEM were compared in electrochemical impedance and hydrogen evolution measurements.Inorganic only version ofα-MEM solution (6.8 g/L NaCl,2.2 g/L NaHCO3,0.4 g/L KCl,0.122 g/L Na2HPO4,0.098 g/L MgSO4,0.2 g/L CaCl2,pH adjusted to 7.4) was made in-house.

2.5.Electrochemical measurements

A GillAC potentiostat (ACM Instruments) was used to perform the electrochemical measurements at～37 °C.A conventional three electrode cell was used with a platinum or graphite counter electrode,a Ag/AgCl-3M KCl reference electrode and the specimen as the working electrode (exposed area of～1 cm2).Electrochemical impedance spectroscopy(EIS) measurements were conducted up to 5 days of immersion applying a sinusoidal perturbation of 10 mV amplitude(vs.open circuit potential,OCP) and a frequency sweep from 100 kHz to 10 mHz.ZView software was used to analyse the impedance spectra.Goodness of fit was ensured by keeping the square of the standard deviation between the original and the calculated spectrums＜0.01.The errors for the individual parameters of the equivalent circuits were ±5%.Potentiodynamic polarization curves were obtained after 1 h of immersion at a scan rate of 0.3 mV/s with potential sweep from-200 mV to +1000 mV,relative to the OCP,and a current density limit of 5 mA/cm2.The corrosion current density was obtained from analysis of cathodic Tafel slopes.The values presented are the average of two specimens to ensure repeatability.

2.6.Hydrogen evolution

Hydrogen evolution measurements were carried out during immersion of bare alloys and PEO-coated specimens in inorganicα-MEM solution thermostated at～37 °C for up to 5 days.Each material was tested by triplicate to obtain the average value.The surface area of the immersed specimen was～18 cm2.The pH of the solution in the immersion tank(～21 L) was continuously adjusted to 7.4 by a flow of CO2regulated by a switch coupled with a pH-sensor (more details on the set-up can be found elsewhere[34].The use of CO2has several advantages[38].The main one is that pH excursions are minimized and as a result the simulation is closer to the conditions foundin vivo.For comparison,experiments on bare substrates were also performed in 0.9wt.% NaCl,α-MEM and inorganicα-MEM,where the pH was adjusted by replacing the solution (250 mL) every 24 h.

It should be noted that the presented graphs show collected hydrogen.In order to calculate the dissolved Mg,the collected gas values were corrected for the dissolved (i.e.unobserved)H2,as per corresponding volume of the solution used in each experiment,using the solubility of H2in water at 37 °C (1.4.mL L-1).

The cross-sectional and plan view examinations of specimens after hydrogen evolution were carried out by SEM.

2.7.Fluoride release

F-release from the coatings was measured using fluoride ion selective electrode (ISE) during 12 days of immersion in 0.9wt.% NaCl solution.The electrode calibration procedure and further details are described elsewhere[44].Two specimens for each coated substrate (3.7 cm2of exposed area)were used to ensure repeatability.

3.Results and discussion

3.1.Characterization of MgZnxCay alloys

Fig.1 shows the SEM images of the MgZnCa and Mg3Zn0.4Ca alloys in backscattered electron (BSE) mode;the insets show higher magnification images with locations of EDS analyses (Table 4).Both alloys are constituted by dendriticα-Mg grains and Ca-Mg-Zn secondary phases.The latter are visible as a network at grain boundaries and at interdendritic spaces,often as islands or rosettes,particularly in Mg3Zn0.4Ca.According to previous works [45,46],the intermetallics that form in these specific alloys are Mg2Ca and Ca2Mg6Zn3,with Mg2Ca mainly located at grain boundaries.Both alloys contain Al as the main impurity which is mainly found in the secondary phases (Table 4).A considerable amount of Fe was also found within some particles in MgZnCa alloy.The main difference between the alloys is that MgZnCa alloy presents smaller dendritic arms (～75 vs.～170 μm) and a higher volume fraction of secondary phases(～8 vs～3.5%) compared to Mg3Zn0.4Ca.Additionally,Zn content in the second phase particles is about 10 times greater in Mg3Zn0.4Ca compared to MgZnCa.This suggests that intermetallic particles in Mg3Zn0.4Ca,although less abundant,are likely to form stronger microgalvanic couples with theα-Mg matrix than in MgZnCa alloy [47].

Table 4 Local EDS analyses of the alloy compositions,as per Figure 1.

3.2.Electrical response of PEO treatment

Fig.2(a)shows the evolution of root mean square values of voltage and current density (Urms,irms,respectively) acquired during the PEO treatment of MgZnxCayalloys.A linear voltage increase is observed during the 60 s ramp with a slight change of slope at～40 s (160 V) that occurs when the limiting current density of 100 mA cm-2is reached.A second pronounced inflexion occurs at～70 s,235 V,corresponding to visually confirmed (due to the electrolyte transparency)initiation of plasma microdischarges due to the breakdown of oxide layer and gas.The microdischarges were uniformly distributed across the surface of the material.At～90 s the Urmsvoltage attained a final constant value (～240 V),at which point the current density dropped to～50 mA cm-2.The drop indicates an increased coating impedance and a change in the coating growth rate due to higher resistance of the oxide material to mass and charge transfer.Microdischarges continued under further gradual decrease of the current density to 25-30 mA cm-2until the end of the treatment.The calculated apparent specific energy consumption for the longest treatments were 1.7 kW h m-2μm-1and 1.9 kW h m-2μm-1for Mg3Zn0.4Ca and MgZnCa alloys,respectively.

Fig.1.Backscattered electron micrographs of the microstructure of a) MgZnCa and b) Mg3Zn0.4Ca alloys.

Fig.2.Voltage-time and current-time dependencies during 900 s of PEO treatment of MgZnxCay alloys (a) and magnification of the initial 100 s period of treatment (b).

3.3.Coating characterization

The evolution of the thickness and roughness values of studied PEO coatings is observed in Fig.3.The thickness of PEO coatings increases linearly with the processing time in both alloys (Fig.3(a)),being～15-20% greater in case of Mg3Zn0.4Ca alloy at all times.The surface roughness parameters (Saand Sz) remain relatively constant with time for MgZnCa alloy and decrease for Mg3Zn0.4Ca.Notably,both parameters are much higher for Mg3Zn0.4Ca alloy,Szbeing an order of magnitude greater.The roughness decrease with time in case of the coatings formed on Mg3Zn0.4Ca alloy correlates with the decrease in their surface pore population density (Table 5).Conversely,surface pore population density increases with the processing time in the case of MgZnCa,however,without notable effect on surface roughness.The average pore size ranges within 0.7-1.8 μm and 0.3-1.6 μm for MgZnCa and Mg3Zn0.4Ca alloy,respectively.It is worth noting that the average pore size was larger after 600 s of PEO for both alloys (Figs.4(b,e)).This is due to transition from relatively weak discharges to more intense ones.By the end of the treatments (900 s)the more intense discharges provide sufficient energy to keep the molten material flowing for longer time and filling up the large pores.

Fig.3.Thickness (a) and roughness (b) of PEO coatings as a function of treatment time.

Table 5 Coating surface porosity characteristics in plan view.

Table 6 EDS analysis results (at.%) of MgZnCa/PEO system as per Figures 4 and 5.

Table 7 EDS analysis results (at.%) of Mg3Zn0.4Ca/PEO system as per Figures 4 and 6.

The surface and cross-sectional morphologies of PEO coatings with their respective surface area and local area EDS analysis are presented in Fig.4,5 and 6 and Tables 6 and 7.PEO surface morphologies in both alloys disclose similar pumice-like structures with pores localised in the centre of rather flat craters.This porous morphology is due to gas evolution through molten coating material followed by fast solidification and gas ejection out of the discharge channel.The surface morphologies of PEO at 300 s in both alloys reveal heterogeneous size pores and bulging areas between the pores as observed in Figs.4(a,d).However,the morphology of the PEO at 600 s (Figs.4(b,e) comprised complex shape surface cavities with a few small pores located inside them and microcracks,the latter formed due to thermal stresses from a quick cooling process.The complex shape cavities disappear in both alloys by 900 s (Figs.4(c,f) and the surface morphology on the whole appears more homogeneous with smaller pore size (Table 5),which can be related to re-melting and re-generation of the surface coating material under lower intensity discharges at lower current density.

Fig.4.Secondary electron micrographs of surface morphology of the PEO coatings produced on MgZnCa (a,b,c) and Mg3Zn0.4Ca (d,e,f) alloy in 300 s(a,d),600 s (b,e) and 900 s (c,f).

Fig.5.Backscattered electron micrographs of cross-sections of the PEO coatings produced on MgZnCa alloy in: 300 s (a,b),600 s (c,d) and 900 s (e,f).

Fig.6.Backscattered electron micrographs of cross-sections of the PEO coatings produced on Mg3Zn0.4Ca alloy in 300 s (a,b),600 s (c,d) and 900 s(e,f).

The cross-sectional morphology of the coatings formed on MgZnCa presents large internal pores,some of them near the barrier layer.These pores form a so-called pore band[48,49].The pores appear to be associated with the location of the intermetallics in the grain boundaries (Fig.5(a,b)).Intermetallic particles are partially oxidized,note the Zn-rich regions in the barrier layer (Fig.5,indicated by arrows).Therefore,the higher electronic conductivity of Zn-rich regions provide paths to easy and sustainable oxygen evolution at these locations during coating growth which results in formation of voids.The oxidized intermetallics are more evident in the coating on Mg3Zn0.4Ca alloy (Fig.6(b)),possibly because the alloy contains relatively large isolated intermetallic particles inside the grains (Figs.1,5 and 6).The particle size is comparable with the thickness of the coating and their Zn content is elevated (Table 4),hence their oxidation and disintegration is harder,especially at a reduced current density(see the current drop,Fig.2).The above observations are in agreement with those in the work of Martin et al.that pointed out the relation of segregated precipitates in Mg alloy with large discharge channels in the coatings[50].The H2evolution overpotential may play a significant role too,as it is known to be particularly high on Zn[51].During the cathodic pulse,less hydrogen would be generated on Zn-rich intermetallics (Mg3Zn0.4Ca alloy) than on Zn-poor intermetallics(MgZnCa alloy).This may be the reason for much less pronounced pore band in the former for a short treatment time of 300 s (Fig.6(a,b)).However,for longer treatment times,the probability that the discharges hit repeatedly the same spot on the surface increases [52,53,54],hence for both treated substrates the inner porosity increases.

According to EDS analysis (Tables 6 and 7),the electrolyte-derived species were present throughout the entire coating thickness profiles in both alloys.Often studies about incorporation of bioactive elements from the electrolytes neglect to disclose in detail the cross-sectional microstructure of the PEO coatings and the distribution of the elements along their thickness,focusing instead on the analysis of the surface[29,31,55].However,the composition of the inner part of the coating,which is not in direct contact with tissue,is equally important since the elements diffuse to the coating/electrolyte interface during the corrosion process and affect the tissue cell metabolisms.In this regard,Ca,P,Si,Na and K prevailed in the outer parts of the present coatings,whereas F dominated in the inner region,due to its well-known faster migration rate under the electric field produced during PEO.The high concentration of F at the metal/oxide interface improves the passivation of Mg alloys due to the formation of MgF2[56].Zn incorporation from the substrate remained relatively independent of the treatment time,although,understandably,Zn content in the coatings decayed outwards and was up to 10 times greater for Mg3Zn0.4Ca alloy (1.4-1.6 at.%) at the sites of the coatings close to the intermetallic particles.

Table 8 Corrosion characteristics from polarization curves.

Ca/P ratios above 2 were observed in the inner region of the coatings despite low at.% content of both elements,whereas Ca/P ratios were below 2 in the outer layer.These differences can be associated with the sources of Ca and P.Ca is readily available from the substrate and migratesoutwards,whereas P migrates inwards from the electrolyte and is barely present in the barrier layer.In both alloys,the maximum surface incorporation of Ca,P and Si was observed for the intermediate treatment time.The drop in the contents of these elements at 900 s may be related to the fact that the existing coating material is being continuously lost at the sites of the microdischarges and replaced with newly formed material[57].As a result of current decay (Fig.2(a))and,consequently,lower intensity and number of the microdischarges,the surface becomes depleted in electrolyte species.

Fig.7 displays the X-ray diffraction patterns of the studied PEO coatings where the high intensity peaks correspond to Mg proceeding from the substrate material.All coatings are showed MgO and amorphous material,as evidenced by the broad hump located between 25 and 40°.In all cases,the formation of MgF2occurred,which is expected due to the presence of F in the electrolyte.Si enrichment on the surface of coatings formed in silicate-based electrolytes is often associated with chemical precipitation of amorphous SiO2[58].In the present case the incorporated silicon formed several crystalline phases,such as magnesium and calcium silicates(MgSiO3,Mg2SiO4,Ca2SiO4,and Ca2Mg(Si2O3),which are known for their bioactivity as they form part of bioactive cements.

Silicon is known to facilitate more homogeneous PEO coatings as well as stabilize hydroxyapatite[59].However,no P-containing crystalline phases were formed,suggesting that all P is contained in amorphous material.Neither hydroxyapatite nor tricalcium phosphate was formed,since the Ca/P ratio was far below 1.67 in the electrolyte (～0.42 Ca/P) as well as in the outer part of the coating where the electrolyte species were mainly located (≤0.6 Ca/P).

3.4.Electrochemical testing

3.4.1.DC polarization curves

Fig.8 illustrates the potentiodynamic polarization curves obtained in inorganicα-MEM after 1 h of immersion at 37°C for bare substrates and PEO coatings.The corrosion performance parameters calculated from the curves are included in Table 8.

Fig.7.X-ray patterns of PEO coatings formed on MgZnCa (a) and Mg3Zn0.4Ca (b) alloys as a function of treatment time.

Fig.8.Polarization curves in inorganic α-MEM for PEO coated MgZnCa (a) and Mg3Zn0.4Ca (b) alloys.

In case of MgZnCa alloy,the PEO treatments produced a noble shift of corrosion potential (by～120-200 mV) and a reduction in corrosion current densities (by～2-9 times).Above the corrosion potential,the anodic branches of PEOcoated alloys show a steep slope due to pseudo-passivity.This behaviour,observed for～150-200 mV,is followed by a rapid current increase,indicating initiation of localized corrosion.The pitting potential,Epit,was similar for all three coatings and about 150 mV nobler than that of the bare substrate(Table 8).The coating formed in 600 s revealed the lowest icorrout of the three coatings.

The bare Mg3Zn0.4Ca alloy shows very similar behaviour with that of MgZnCa alloy (Table 8,Fig.8(b)).The effect of PEO treatment is also similar to that on MgZnCa alloy,i.e.shifting Epitto higher values and reducing icorr.The main difference is in the PEO treatments obtained at longer times(600 s and 900 s).These reveal a stronger tendency to pitting,given that Epitis very close to or practically coinciding with Ecorr(Table 8).This can be associated with heterogeneous morphology of the coatings,as discussed for Figure.6,in combination with the higher area fraction of Zn-rich intermetallics in Mg3Zn0.4Ca substrate.As a result,only the most homogeneous coating of all,formed at 300 s on Mg3Zn0.4Ca alloy (Fig.6(b)),discloses a distinct pseudo-passive region(Fig.8(b)).

In summary,all coatings on both alloys revealed similar Epitvalues (-1.33±0.05 V),regardless of the coating thickness or the alloy substrate.Based on the polarization experiments,the PEO coating formed on Mg3Zn0.4Ca after 300 s presented the lowest corrosion current density,which is explained by its more uniform morphology as evidenced by the corresponding cross-section in Fig.6.

Fig.9.Bode plots of EIS spectra for MgZnxCay alloys in a) inorganic α-MEM and b) complete α-MEM solutions after 1 h of immersion (Z’,Z’’,|Z| units:Ω cm2).

3.4.2.Electrochemical impedance spectroscopy

In order to elucidate better the differences in short term corrosion performance of bare and coated substrates,EIS measurements were carried out in inorganicα-MEM andα-MEM solutions at 37 °C.Completeα-MEM contains amino acids,vitamins and glucose (among other organic components) that would also be present in the body fluidsin vivoand would affect the passivation and degradation of Mg alloys.Comparing the Bode plots in both media (Fig.9) it is evident that completeα-MEM reduces the corrosion resistance of the alloys by about three and six times for MgZnCa and Mg3Zn0.4Ca,respectively,and practically eradicates the difference between the two (Fig.9(b)) that is observed in inorganicα-MEM (Fig.9(a)).

The equivalent circuit used to interpret the EIS results(Supplementary Figure S1) include Relec(solution resistance),R1/CPE1(ascribed to the capacitive and resistance behaviour of the deposited products on the surface of the specimens),R2/CPE2(ascribed to the capacitive and resistance behaviour of the corrosion products layer),and Rdl/CPEctrelated to the electrochemical activity at the substrate/electrolyte interface.Constant phase elements are used instead of the ideal capacitors to account for the surface roughness and heterogeneity.This equivalent circuit is in agreement with previous studies focused on the evolution of ZX11 Mg alloy in a medium similar toα-MEM[60].The fitting results are presented in Table 9,where CPE values correspond to the admittance of the constant phase element at the radial frequencyω=1 rad,in accordance with the relationshipZ=1/(CPE(jω)n),where and j is the imaginary number and -1 ≤n≤1.

It is evident from the fitting results that CPE1most likely represents a diffusion process in the outer part of the corrosion products film,which is mainly composed of deposited products from the media,given that the exponential factor n1is close to 0.5.Respectively,low values of R1indicate the weak resistance of the deposited products[60].Notably,in both alloys,the organic components ofα-MEM affected the resistance of the inner products layer,R2,and the charge transfer resistance,Rct;both decreased by three to six times compared to inorganicα-MEM.This can be related to the reduced thickness of the inner products layer.Accordingly,greater CPE2values were observed in completeα-MEM and the heterogeneity of the layer increased,as indicated by lower n2[61].This means that amino acids inα-MEM facilitate the dissolution of magnesium and impede the passivation of the surface,hence the thinner deposition products layer[62].The destabilizing effect of amino acids on the Mg/electrolyte interfaces is also reflected in low values of n3.As can be seen in Table 9,the total polarisation resistance (Rp) follows the same trend as the individual resistances.

EIS of the PEO coated specimens in inorganicα-MEM and completeα-MEM (Fig.10),revealed the following phenomena.The equivalent circuits and the fitting results are presented in supplementary information (Figure S1 and S2,Tables S1 and S2).First of all,in either of the media,the total impedance of any given PEO coating on either of the alloys did not exceed 104Ωcm2,which,as far as the protective properties of the coating go,is not a great extent,compared with |Z|0.01Hzof bare substrates (Fig.9).The coatings formed in 300 s exhibited a superior performance in both cases.Hav-ing said that,PEO coatings contributed more to the corrosion resistance of the alloys in completeα-MEM,yielding a maximum of one order of magnitude increase in case of the 300 s coatings,than in inorganicα-MEM.This means that the PEO layer rendered the alloy surface inaccessible to the depassivating effect of the amino acids [63,62,64].The latter are known to adsorb on the coating surface [65,66,67,68].Additionally,their large size prevents their migration through the ceramic pore network,[68,69].The important implication of this finding is that inorganicα-MEM can be considered as an inexpensive,easy maintenance and realistic medium for fast screening of PEO coatings.

Table 9 Fitted values of EIS equivalent circuits for bare substrates.

Fig.10.Bode plots of EIS spectra for PEO coated MgZnCa (a,c) and Mg3Zn0.4Ca (b,d) alloys in inorganic part of α-MEM (a,b) and complete α-MEM(c,d) after 1 h of immersion (Z’,Z’’,|Z| units: Ω cm2).

Fig.11.H2 evolution for bare and PEO coated substrates over 5 days of immersion: (a) comparison of bare substrates in 0.9% NaCl,inorganic α-MEM and complete α-MEM;(b) amplification of (a),(c) comparison of the substrates with and without PEO coatings formed in 300 s and 600 s in inorganic α-MEM under flowing CO2;(d) amplification of (c) for the initial hours of immersion.

The electrochemical testing (potentiodynamic polarization and EIS) offered different ranking controversial conclusions in regards to the coatings formed for 300 s and 600 s for the of MgZnCa alloy (Figs.8(a) and 10(a)).Therefore,both coatings were selected for further screening by hydrogen evolution measurements in inorganicα-MEM to determine the degradation rate and examine the development of corrosion products during the immersion.

3.5.Hydrogen evolution

Fig.11(a) shows the hydrogen volume evolved during 5 days of immersion for bare substrates in 0.9wt.%NaCl,completeα-MEM and inorganicα-MEM solutions at 37 °C,where the media were changed every 24 h in order to correct the pH.A mole of evolved hydrogen gas corresponds to one mole of corroded Mg,in accordance with the total reaction:

Fig.12.Cross-sectional backscattered electron micrographs of bare alloys after immersion in (a,b) 0.9% NaCl,(c,d) inorganic α-MEM and (e,f) α-MEM solutions during 5 days at 37 °C.Insets: macro-photographs of the specimens following the immersion.

Hence,comparing the results in Fig.11(a),it is evident that 0.9% NaCl is a much more aggressive medium than completeα-MEM.Specimens in inorganicα-MEM produced the lowest hydrogen volumes out of the three media.The corrosion rate is slightly higher in completeα-MEM because of the presence of amino acids and vitamins in the solution that at low concentrations can mildly accelerate the corrosion of Mg [64,70].Localised corrosion with deep pitting occurred in MgZnCa alloy in 0.9% NaCl (Fig.12(a)),while a relatively uniform corrosion layer developed in Mg3Zn0.4Ca alloy (Fig.12(b)).These differences are likely to be associated with the higher area fraction of secondary phases in the MgZnCa alloy.The corrosion progress in bothα-MEM solutions is slowed down by the secondary phases at grain boundaries (Fig.12(c)),and by the formation of a partially protective P-rich layer in the completeα-MEM solution (Fig.12(e,f)).Cracks observed in the corrosion products are due to volume shrinkage as corrosion products layer dehydrates.It is worth mentioning that hydrogen measurements and corrosion morphologies inα-MEM solutions did not show considerable difference between MgZnCa and Mg3Zn0.4Ca alloys.This indicates a less important role of alloy microstructure in these media.

Corrosion products layer formed in completeα-MEM comprised an outer brighter region corresponding to the Caand P-rich layer (Fig.12(f),at.%: 64.7O,8.6Mg,14.3P,12.4Ca,0.1Zn),and a darker one corresponding to mostly Mg(OH)2(Fig.12(f),at.%: 67.9O,29.8Mg,1.0P,1.2Ca,0.2Zn).For comparison,the composition of the corrosion products layer formed in inorganicα-MEM (Fig.12(c)) analysed by EDS is (at.%): 60.8Mg,27.1Mg,5.6P,3.8Ca,2.7Zn.There is an agreement that a protective corrosion layer formed on Mg in physiological solutions is constituted by magnesium hydroxide and hydroxyapatite-like compounds [70,71,38] that are precipitated when the near-electrode pH rises sufficiently as a result of the dissolution of Mg(OH)2in the presence of Cl-ions[72].It is worth noting that pH conditions favouring the excessive precipitation of calcium phosphates did not occur in the absence of organic additives (i.e.in the inorganicα-MEM).Therefore,it may be inferred that amino acids increase the solubility of Mg(OH)2,which is additionally evident from the very even,as if chemically polished,appearance of the corrosion products/resin interface in Figs.12(e,f).

Fig.11(c,d) shows the results of hydrogen evolution on bare and coated substrates in inorganicα-MEM with a constant pH 7.4 under flowing CO2.For bare substrates,the corrosion rate was slightly lower than in the periodically replaced solution (where pH fluctuated over 24 h from 7.4 to 8.5-9.0),however the difference was not significant (at 120 h: 0.19 vs.0.29 mL cm-2for MgZnCa and 0.55 vs.0.59 mL cm-2for Mg3Zn0.4Ca).Remarkably,the PEO coatings maintained their protective effect only during the first 2-5 h (Fig.11(c)),after which period the degradation rate of coated alloys accelerated.This is a highly unusual behaviour for PEO-coated Mg alloys,contrary to that observed in the authors’ previous works on other Mg alloys,such as Mg0.8Ca[34] and AZ31B[39].The coatings formed in 600 s demonstrated lower protective properties than those formed in 300 s,which is generally in agreement with the trend derived from EIS spectra in short-term immersion.It is worth mentioning that previous studies on Mg-Zn-Ca alloys [29],[30,31],typically show an improved corrosion performance of PEO coated specimens.However,these studies differ from the present one in several aspects: a constant pH of the corrosive medium is not maintained,PEO coatings show high levels of fluoride or corrosion testing (often only electrochemical) is limited to short immersion times.

Fig.13(a-h) discloses thick undercoating corrosion products layers developed on the alloys with PEO (300 s) coating during 18 h and 120 h of immersion.The first time point shows that once the PEO barrier layer is breached,undercoating corrosion progresses very quickly,e.g.～30 μm for MgZnCa alloys (Fig.13(c)).

Several important factors pertinent to the corrosion mechanism of PEO-coated MgZnxCayalloys must be highlighted here: (i) the coatings still appear well adhered even after 120 h of immersion;(ii) fewer penetrating cracks are observed in the undercoating corrosion product layers than in bare substrates (Fig.12);(iii) Zn-and Ca-rich secondary phases remain embedded in the corrosion products layer under the coating,whereas entrapped intermetallics were practically not present in the corrosion products layers developed on bare substrates (Figs.12(c,d)).Accordingly,it may be inferred that the corrosion process of the coated samples occurs as follows: a) corrosive species penetrate the pores and cracks of coating;b) noble secondary phases form galvanic couples with the matrix and accelerate the oxidation of the latter;c) a crevice forms between the well-adhered coating and the substrate,intensifying microgalvanic corrosion and,therefore,undercoating corrosion;d) the coating impedes the detachment of the corrosion products layer,resulting in a lesser number of cracks and entrapped or non-dissolved intermetallics.

Fig.13.Cross-sectional backscattered electron micrographs after immersion in inorganic α-MEM solution under flowing CO2.PEO (300 s): 18 h (a-d)and 120 h (e-h).Insets: macro-photographs of the specimens.

The non-oxidized cathodic intermetallics may possess greater effective area,hence stronger microgalvanic corrosion[73].Importantly,the intermetallic cathodes are rich in Zn (Table 4).Zn undergoes hydrolysis and produces acidification[74];as a result,the environment in the crevice (i.e.under the coating) is likely to be more aggressive,in terms of the pH and the gradient chloride concentration.For that reason,an elevated H2evolution and thicker layer of corrosion products were observed on PEO coated alloys (Figs.11(c)and 13(f,h).The important role of crevice in biodegradation of Mg implantsin vivowas highlited in[75] .Also,recently,Silva et al.showed that O2reduction can participate significantly in corrosion of Mg and its alloys[76].This may be particularly relevant to the cathodic reaction on Zn-rich intermetallics,since Zn corrodes in neutral media under oxygen depolarization.Hence,it is reasonable to expect the detrimental role of crevice in accelerated corrosion of coated Mg alloy systems like MgZnxCay.

Fig.14.Nyquist and Bode diagram of MgZnCa and MgZnCa/PEO-300s system after 1 h,24 h and 5 days of immersion at 37 °C.

The high permeability of the porous part of the coating and low resistance of its barrier layer may be associated with low F-content in the electrolyte and,consequently,insufficient amount of MgF2formed at the coating/substrate interface.For instance,Mohedano et al.showed that～40 μmthick PEO coatings on Mg0.8Ca alloy with～20at.% of F in the coating (generated in an electrolyte with 8 g/L NaF) delayed the initiation of undercoating corrosion for more than two weeks[34].However,these coatings were later demonstrated to release high amount of F-in the medium,causing a decrease in cell proliferation[24].In the present study,a reduced amount of F-in the electrolyte resulted in a decrease in coating thickness and F-content.Nevertheless,the maximum amount of liberated F-(measured in 0.9wt.% NaCl)was 150 μg cm-2and 70 μg cm-2(see supplementary material,Figure S3) for coated MgZnCa and Mg3Zn0.4Ca,respectively.These values are comparable or even somewhat higher than those previously reported for thick PEO coatings with high F content[24] .The enhanced lixiviation of F-may be attributed to high permeability and hydration of the coating.In vitro(e.g.for cell culture volume of 6 mL)this would generate concentrations exceeding the cytotoxicity levels (＞1.0 mM) for a range of cells[24].These findings imply that fluoride-free PEO electrolytes should be developed for alloy systems like MgZnxCay,whereas the suitability of lowfluoride particle-free electrolytes for other Mg alloy systems(with microstructures inducing lesser micro-galvanic coupling effect) should be the topic of future studies.

In regards to the correlation between the electrochemical and hydrogen evolution experiments,the corrosion current densities (Table 8) obtained from the polarization curves in inorganicα-MEM significantly underestimated the corrosion rate.For MnZnCa alloy,1.48 mg cm-2would be expected to dissolve in 5 d,whereas～2.56 mg cm-2was actually dissolved,as per corresponding evolved H2volume(Fig.11(a)) corrected for the amount of dissolved H2.This amount of dissolved Mg would generate about～20 μmthick layer of Mg(OH)2,which is in a good agreement with the average thickness of corrosion products layer observed in Fig.11(c).The difference between the icorr-and H2-derived values is undoubtedly related to the drastic change in corrosion rate in the initial period of immersion,as can be seen in Fig.11(c),which must be due to the enhancement of microgalvanic couples in the presence of crevice,i.e.under the coating once the barrier layer of the latter fails.This is a common drawback of polarization tests as they provide single-time data rather than an average rate[77].The abnormal polarization behaviour of Mg alloys,known as negative difference effect,also complicates interpretation [78].On the other hand,hydrogen tests does not detect non-Faradaic dissolution[77] and cannot reflect the contribution of oxygen reduction[79].

The EIS method can still be used for reliable qualitative corrosion screening of PEO coatings,even when the coated system is compromised by microgalvanic couples.This is demonstrated by the measurements performed on bare MgZnCa alloy and MgZnCa/PEO_300s system after 1 h,24 h and 5 days of immersion in inorganic part ofα-MEM at 37°C (Fig.14).The results show that initially superior corrosion resistance of the coated alloy (1 h) becomes lower than that of the bare alloy following 24 h of immersion.It further decreases after 5 days of immersion.This is consistent with the results of H2evolution experiments.

In summary,H2evolution test provides a better indication of the average corrosion rate,whereas EIS testing is more useful for direct and quick comparison of coatings.For EIS screening of different coating recipes,24 h immersion can be considered sufficient even when a strong microgalvanic coupling is involved in corrosion mechanism of the alloy/coating system.

4.Conclusion

1.Transparent Ca,P,Si and low F-containing electrolyte based on soluble calcium salt has been developed for PEO of MgZnxCayalloys.

2.The PEO process on MgZnCa and Mg3Zn0.4Ca cast alloys in the developed electrolyte under AC regime shows a linear coating growth kinetics yielding～14-15 μm-thick coatings in 900 s of treatment.The coatings on both alloys are comprised of magnesium oxide,magnesium fluoride and calcium and magnesium silicates;phosphorous is incorporated into amorphous phase.The coating morphology is strongly influenced by Zn-rich intermetallic particles that are embedded in the coatings in a partially oxidized state and cause formation of large voids nearby the Zn-rich sites.

3.Easy permeability of the coatings and insufficient stability of the barrier layer (deficiency of MgF2) formed in low-fluoride electrolyte allows for early initiation of undercoating corrosion in MgZnxCayalloys.A more aggressive corrosive environment and enhanced micro-galvanic couple effect between the Zn-rich intermetallic particles and the Mg matrix take place in the undercoating crevice,which leads to faster degradation of the coated alloys.

4.PEO coatings for Mg alloys with significant micro-galvanic coupling should comprise a strong barrier constituent in order to delay the formation of crevice for as long as possible.

5.The presence of amino acids and other organic components of completeα-MEM has no influence on Mg alloy/PEO system corrosion resistance,due to their adsorption on the coating surface and inability to migrate towards the coating/substrate interface.Therefore,the inorganic part ofα-MEM can be regarded as an inexpensive,reliable and easy maintenance corrosion screening medium for Mg alloy/PEO systems.A 24 h immersion is recommended for EIS screening of PEO coatings on alloys with a strong micro-galvanic coupling.

Declaration of competing interest

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

Acknowledgements

The funding of the ADITIMAT-CM project (S2018/NMT-4411,Regional Government of Madrid and EU Structural and Social Funds) and RTI 2018-096391-B-C33(MCIU/AEI/FEDER,UE) are gratefully acknowledged.M.Mohedano is grateful for the support of RYC-2017-21843.

Supplementary materials

Supplementary material associated with this article can be found,in the online version,at doi:10.1016/j.jma.2021.12.011.