Development of open-porosity magnesium foam produced by investment casting

Honort Kp?on,Crsten Blwert,Jek Ch?emnowski,Krzysztof Nploh

a Faculty of Mechanical Engineering,Wroc?aw University of Science and Technology,Ul.Smoluchowskiego 25,Wroc?aw 50-370,Poland

b Institute of Surface Science,Helmholtz-Zentrum Geesthacht,Max-Planck-Straβe 1,Geesthacht,21502,Germany

c Faculty of Chemistry,Wroc?aw University of Science and Technology,Wybrze˙ze Stanis?awa Wyspia′nskiego 27,Wroc?aw,50-370,Poland

Abstract High-porosity,open-cell AZ91 magnesium alloy foams of two pore sizes were fabricated by means of investment casting technology,using PUR foam patterns.Foam casting variables such as pressure,mould temperature and metal pouring temperature were thoroughly investigated to define the most optimal casting conditions.The mechanical properties of the fabricated foams were measured in compression tests.A potential application for the foams considered is temporary bioresorbable bone implants,therefore the mechanical properties of the foams were compared with those of cancellous bone tissue.Foams with smaller pore size and lower porosity(20 PPI and 80%–87%)exhibited mechanical properties in the lower regions of the cancellous bone property range(Young’s modulus 36.5–77.5 MPa),while foams with higher pore size and porosity(10 PPI and～90%)were found to have insufficient compression strength(Young’s modulus 11.65–23.8),but thickening their walls and lowering their porosity below 90% yielded foams with Young’s modulus between 36.5 and 77.5 MPa.Foam fractures were also investigated to determine their collapse mechanism.A series of corrosion tests in stimulated body fluid was carried out to determine their applicability as a biomaterial.The Plasma Electrolytic Oxidation(PEO)process was used in a feasibility study to examine the microstructure and chemical composition of foams with protective coating.

Keywords:Magnesium foams;AZ91 investment casting;Bone implants;Plasma electrolytic oxidation.

1.Introduction

Metal foams exhibit properties that are unique amongst metal materials.Due to their high surface-to-volume ratio and lower weight,they have found applications mostly in the automotive and aircraft industries,but also as lightweight construction materials,silencers,heat exchangers,filters and catalysts in the chemical industry,or as shock absorbing material in military vehicles[1].On top of that,their porous structure offers the possibility of developing biomaterials applicable as biomedical implants.

Magnesium,as a material with good biocompatibility,seems like the perfect candidate for this application.It also has similar density and mechanical properties to bone tissue:magnesium density ranges from 1.74 to 2.0 g/cm3,while bone tissue density ranges from 1.8 to 2.1 g/cm3;elastic modulus for Mg ranges from 41 to 45 GPa,while for bone tissue,it ranges from 3 to 20 GPa[2].This means that their mechanical properties are not the same,but much closer than those of steel,titanium or Co-Cr alloys.These properties are an important factor because they decrease the stress in the boneimplant interface,resulting in a more stable environment for bone tissue growth[3].

Another advantage of magnesium over other materials is biodegradability,which is very important for application as short-term implants.The use of biodegradable material spares the patient the additional steps of removing the implant at the end of treatment.One of the biggest drawbacks of magnesium–low corrosion resistance–poses no threat to the human body since its corrosion products are soluble,non-toxic and can be easily excreted in the urine.Magnesium implants can be gradually dissolved with no harm to the patient’s health,as opposed to steel,titanium or Co-Cr implants that can release toxic products during the corrosion process.Its ions are naturally present in the human body and essential to its metabolism.It is estimated that about 1 mol of Mg is stored in the body of a healthy 70 kg person,half of which is stored in bone tissue[4].One problem with this process is that magnesium corrosion locally increases hydrogen concentration and creates hydrogen bubbles near the implant,which can cause segregation of tissue layers and tissue necrosis,leading to delayed healing[5].

In order to extend the life of Mg foam implants before they dissolve,special coatings can be developed.Plasma Electrolytic Oxidation(PEO)has proved to be a very efficient method of creating protective coatings on metals such as aluminium[6]or magnesium[7].Their porous structure was found to have both advantages and disadvantages in biomedical applications.It gives less protection against corrosion,but on the other hand,pores have a significant role in hydroxyapatite growth and in better adhesion to bone structure through mechanical interlocking[8].That is why further research in this area is a promising field of study.

The best known and widely used method for the fabrication of magnesium foams is the space-holder method,which uses a negative pattern(e.g.,salt particles)pre-form that is infiltrated with liquid metal and cleaned.This method,employing salt particle patterns,was used for example in research on Mg-Zn-Ca scaffolds with cubic NaCl particles as a space holder,obtained by Dong et al[9]..Other salt patterns,shaped on 3D-printed polymer scaffolds and after model firing,were infiltrated with pure Mg by Kirkland et al[10]..Jiang and He on the other hand,used a 3D titanium wire structure that was infiltrated with Mg and then removed in a HF solution[11].A significant problem with the space holder method is the residues that cannot be removed from the deep regions of magnesium structures and in the case of salt patterns,they increase corrosion process.Another popular technique for magnesium foam processing is powder metallurgy.Seyedraoufi and Mirdamadi[12]fabricated Mg-Zn alloy foams using powder metallurgy,with carbamide as a space holder,while Yang et al.fabricated closed-cell foams using powder metallurgy with CaCO3as the blowing agent from Mg-Al alloy in[13,14],and Mg-Al-Zn alloy in[15].

Investment casting can be another method for fabricating magnesium foams.3D structures made with wax or polymer are prepared(for example through fused deposition modelling),put in investment moulds and burnt away.Maier et al.[16]made such models from wax using FDM technology and fabricated magnesium 3D scaffold structures.A different kind of pattern,with much bigger porosity and less preparation effort,can be obtained by using polyurethane foam.Yamada et al.[17]investigated such foams for AZ91 alloy,with a pore size of around 4.5 mm and a density of 0.05 g/cm3.However,they revealed lower mechanical properties(plateau stress of 0.11 MPa),which means that the foams need to be reinforced for this method to be applicable in biomedical implants.The method presented in this research uses a polyurethane foam pattern coated with a wax slurry in order to increase its density and mechanical properties.This technique produces foams with controlled morphology(as opposed to the salt particle space holder method),by using easily available uniform materials and less specialised equipment than the methods mentioned above.

2.Materials and methods

2.1.Foam casting process

The magnesium-based foams were cast with AZ91 alloy(9 wt% of Al,0.13 wt% of Mn,and 0.7 wt% of Zn)widely used in die casting[18].Foam fabrication was based on the investment casting technology combined with evaporation of polyurethane(PUR)foam-pattern.The entire process is illustrated in Fig.1.

A specific mesh size defined by the PPI parameter(pores per inch)was maintained in two sizes:10 PPI with approximately 6 mm pore size and 20 PPI with approximately 3 mm pore size.The porosity parameter was changed by coating the cell walls with a wax-based slurry.Two 10 PPI foams were coated with a thicker wax layer to study the difference in wall thickness which can be obtained with the same PPI number.The foam structure is presented in Fig.2.The moulds for the research were made in?100 mm casting flasks,using Ransom&Randolph Argentum gypsum-based investment.The moulds were subjected to heat treatment during which the foams were burnt away to create a cavity:an exact copy of the pattern.The heat treatment was carried out at the following temperatures:

·dehydrating and dewaxing at temperatures rising from 150 to 370 °C,

·burning and firing at 730 °C for 2 h,

·cooling down to a pouring temperature of 550 °C.

The AZ91 magnesium alloy was cast in 660–680 °C,in SF6protective gas atmosphere.The mould and metal temperatures during pouring were selected based on calculations followed by experimental trials.However,the process conditions and outcomes may vary depending on multiple factors,such as the shape and size of the gating system,the vacuum system,and the casting flask size,etc.

Fig.1.Diagram illustrating the process of AZ91 foam investment casting.

Fig.2.Macro-scale photograph of AZ91 magnesium alloy foam with 20 PPI.

The temperature of the mould and of the liquid metal,as well as the vacuum value are the most important pouring parameters to ensure effective penetration of intricate mould cavity and simultaneously minimise the adverse reaction between the metal and the ceramic material.The process variables can be classified into three groups related to(1)the characteristics of the casting mould,(2)the pouring conditions and(3)the properties of the alloy.As the molten metal front penetrates through very narrow channels,it may cool quickly and crystallise prematurely.The key issue is the thermal diffusivity of the mass.With low thermal conductivity and average specific heat that characterize investment casting moulding materials,it can be assumed that with fast pouring the flowing metal heats only a thin layer of the mould cavity.Therefore,a sufficiently high-pressure gradient(high vacuum in the autoclave)must be provided so that the flowing metal loses only a small amount of heat and reaches the longest possible distance.The pressure,calculated from Eq.(1),should be exerted on the metal front in order to overcome the resistance of the mould-Pmould,resulting from the existing gas inside the cavity and the formation of a new metal–mould interface.Moreover,the meniscus-shaped metal front of radiusRwill be inhibited by the surface tension acting on it.

Fig.3.AZ91 10 PPI foams:(a)non-coated,porosity 92%;(b)wax slurry-coated,porosity 85%.

The generated vacuum can reduce the amount of gas in the cavity as long as the mould is permeable.To improve this parameter,fine graphite flakes were added to the commercial ceramic mass and later gasified during the heat treatment of the mould,making the mass more porous.After spiral casting tests with varying path width,it was possible to define appropriate temperature parameters.With a narrow path,the length of the flow was similar,and the influence of metal temperature was small.For the real mould,it was necessary to select the lowest possible temperature values so that,on the one hand,the metal completely filled the cavity,to avoid misruns,and on the other hand,to limit the reaction of the liquid metal with the mass.Fig.3 shows the structure of the AZ91 alloy foam with an open porosity of 92% or 85% both based on the original 10 PPI PUR foam size.

During pouring,a small amount of air and gas penetrates through the walls of the mould placed in the autoclave.Accelerated oxidation causes an increase in the temperature,which may locally exceed 1200°C.Then one of the components of the moulding mass–calcium sulphate-can decompose according to formula(1)[19]:

and oxygen is released,which intensifies the reaction and finally the mould is violently ruptured by large amounts of sulphur dioxide.Hence,it was necessary to develop a protective cover by feeding a gas mixture(SF6/CO2)and by controlling the temperature parameters very precisely both during the pouring and cooling of the casting.The interaction of the liquid and cooling magnesium with the mould mass should be limited to inhibit the reaction of magnesium with silicon oxide and the formation of Mg2Si as in formula(2)[20]:

Such Mg2Si crystals were observed right next to the casting surface(see Fig.4).These inclusions triggered intergranular corrosion and accelerated the biodegradation process.Additionally,MgAl2O4and Mg2SiO4can form on the casting surface,as observed by Vyas and Sutaria[21]when AZ91 interacts with alumina or silica binder present in some shell moulds.In order to produce a good quality magnesium foam,the process conditions should allow for the formation of a continuous MgS layer by reaction with protective gas or compounds,and possibly a short contact of the liquid alloy with the ceramic mould.

Measurements of temperature and pressure near the mould cavity show that the temperature decreases very slowly during solidification(dT/dt≈4.5 °C/min for the first minutes)and the applied low pressure forces metal to fill the entire volume in a fraction of a second.Therefore,the crucial factor affecting the chemical interaction between metal and ceramic is their contact time while the metal is still in the liquid state since this is when harmful reactions and deterioration of the casting surface can occur.The total energyQreleased from the solidifying metal is the sum of the latent heat of solidification and the superheat,as Eq.(2)shows:

The thermo-physical properties of AZ91 alloy and mould materials with assigned symbols are shown in Table 1.The flux of this heat transferred through the interface metal/mould at timetcan be calculated from the Eq.(3)according to Rundman[22]:

Table 1The thermo-physical properties of AZ91 alloys and the investment casting mould.

Table 2Parameters for PEO foam treatment.

using the integral of determined function in Eq.(4):

and comparing the total heat dissipated through the casting area interface A(calculated according to the method proposed by Chen et al.[23]),the solidification timetscan be calculated for different temperatures of the poured metal and the preheated mould according to Eq.(5)[22]:

Fig.4.Microstructure and EDS spectrum of the foam with Mg2Si formed during interaction with moulding material at high process temperature(620°C mould/740°C metal).

Since two opposing requirements must be met in the casting process(solidification time long enough to allow the metal to fill the entire mould and short enough to prevent a metalmould material reaction),it is crucial to find the range of mould and metal temperatures sufficient to reach solidification time between 1 and 15 s(ideally 5 s).The calculated solidification times are presented in Fig.5,with mould temperatures ranging from 250 to 550 °C and metal pouring temperatures ranging from 600 °C to 720 °C.As can be seen,solidification time rises exponentially with increasing mould temperature,which suggests the most optimal temperature range of 500 °C–525 °C for the 10 PPI foam(square points on the plot)and 525–550°C for the 20 PPI foam(circle points on the plot).The effect of the pouring temperature on solidification time is relevant when the mould is preheated above 500 °C and is most visible in mould temperatures over 550°C.Below 500 °C,metal solidifies almost immediately irrespectively of temperature.At a mould temperature of 450 °C for 10 PPI foams,the solidification time allows(in ideal conditions)filling the mould;however,in 20 PPI foams,tsis shorter than 1 s.For mould temperatures over 525°C,the pouring temperature in 10 PPI foams has a significant impact on the solidification time and should be kept below 660 °C.The 20 PPI foams,due to their larger interface area,solidify faster and in mould temperatures up to 550 °C have shorttsregardless of the pouring temperature.Consequently,temperatures up to 720 °C can be used since it is the upper limit for AZ91 casting,above which oxidation and combustion danger increases[25].The separately calculated length of metal flow in the cylindrical mould channel using Fleming’s formula[26]indicates that a mould temperature in the range of 440–500 °C is required to fill a 6-cm foam-shaped cavity.Finally,the calculated process parameters allow the production of AZ91 foam blocks with maximum dimensions of 80×80×40 mm,with a homogeneousα-Mg-Mg17Al12structure as seen in Fig.6.

2.2.Testing procedures

Compression tests on uncoated foams were carried out on a universal testing machine(Tinius Olsen H25KT).The specimens had a cross-section of 35×23 mm(±2 mm)and a height of 40 mm(±2 mm).They were extracted from larger blocks and then weighed and measured.Their porosity,which was measured using Archimedes’principle,ranged from 83% to 92%.The compression tests were performed with a displacement range set to 10 mm/min,using a 10 kN load cell.

The microstructures of coatings,casts and fractures were examined with a scanning electron microscope(SEM,Hitachi TM-3000)and analysed with an EDS detector to determine their shapes,sizes,thickness,composition and compression characteristics.

Fig.5.Calculated solidification time of AZ91 foam casts poured into preheated ceramic moulds at the temperature of 600–720 °C.

Fig.6.SEM micrographs of the α-Mg-Mg17Al12 structure of AZ91 foam.

Fig.7.Stress-strain curves of AZ91 foams under compression tests:(a)10 PPI foams with thin wax coat(A1-A5,90–91% porosity)and thick wax coat(A6-A7,82–85% porosity),(b)20 PPI foams with thin wax coats.(For interpretation of the references to color in this figure,the reader is referred to the web version of this article.)

The foam surface was treated by Plasma Electrolytic Oxidation(PEO)processing,using different treatment parameters(Table 2).A pulsed DC power source was used to treat the foams for 20 min under constant current.Different electrolytes with and without particles,and different current densities were tested to identify problems with foam processing and to improve the performance conditions.A set of foam specimen was cleaned prior to PEO treatment for 30 s in an aqueous solution of 150 g/L acetic acid.The foam was the anode in the process and a stainless steel tube cathode was integrated in the cooling system in an effort to maintain the temperature of the electrolyte at 20±2 °C.

ICP-OES studies–analysis of the elemental composition of SBF elemental analysis in body solutions was performed using the technique of ICP-OES(Inductively Coupled Plasma Optical Emission Spectrometry).Under experimental conditions,the samples(in five replications)were immersed in 100 ml of SBF.

The content of the elements entering the matrix of the magnesium alloy and the elements forming the corrosive solution during the exposure of the samples to SBF was analysed.Knowing the concentration of the alloying elements after exposure to SBF makes it possible to determine the biocompatibility of the magnesium alloy as an implant,while the rate of loss of calcium and phosphorus from the SBF solution leads to the assumption that a thin layer of apatite ceramics was formed on the surface of the magnesium alloy,which slowed down the corrosion process of the metallic material.

The measurements were conducted in an SBF solution,which imitates the corrosiveness of the biological aqueous environment of the human body.The SBF was prepared according to the methodology established by Kokubo and Takadama[27](8.035 g NaCl,0.355 g NaHCO3,0.225 g KCl,0.231 g K2HPO4·3H2O,0.311 g MgCl2·6H2O,39 mL 1.0 mol/L HCl,0.292 g CaCl2,0.072 g Na2SO4,6.118 g(HOCH2)3CNH2and 0–5 ml of 1.0 mol/L HCl).

3.Results and discussion

3.1.Mechanical properties

The analysis of the magnesium foam compression tests shows a correlation between the porosity of the foam and its mechanical properties.The foams from B series(with 20 PPI)had lower porosities and higher compression strength than the thin wax-coated foams from A series(10 PPI)that had higher porosities and bigger pores and struts.The thick-coated 10 PPI foams,described in Section 2,had similar porosities and mechanical properties as the 20 PPI foams.Fig.7 shows the stress-strain curves from both series,with a linear part,plateau and densification area.This kind of compression test curve is typical for metal foams–initially the stress increases linearly until the struts start to break.As the foam slowly collapses,the stress remains stable,creating plateau of the stress-strain curve.Eventually the crushed foam density increases and the material starts to behave like a monolyth.Due to magnesium’s microstructural and crystallographic properties it has brittle collapse mechanism,therefore the stress-strain curve lines are rigid.The properties of all the tested samples are summarized in Table 3.

Table 3Properties of selected AZ91 foams test samples presented on Fig.7.

The compression test results showed that all fabricated foams achieved compressive strengths no higher than 2.2 MPa and Young’s moduli below 80 MPa.These results vary greatly from those obtained by other methods.Posada et al.fabricated foams with Young’s moduli in the range of 0.8–1.8 GPa,while Rúa obtained foams with Young’s moduli of about 1.45 GPa[28,29].However,the porosity and morphology of their foams significantly differed from those produced in this research.They used the salt space holder method,withwhich Posada obtained foams with porosities ranging from 60 to 79%,while Rúa fabricated foams with porosities ranging from 36 to 53%.The lower mechanical properties of the foams in this study may be due to the very high porosity and very low foam strut thickness.The compression process was recorded on video and analysed frame-by frame to determine how the foam deformation processed.Fig.8 shows photographs of the foam compression process in different displacement phases.The red lines highlight the areas where deformations occurred.A frame-by-frame analysis shows that the first breakage occurs on the surface touching the compression plates and then new deformation lines appear on the thin struts adjacent to the areas strengthened by pattern modification.The collapse mechanism of these foams is slightly different from that of foams with lower porosities and with more closed cells,such as those developed by Bayani and Mirbahgerri[30].Due to thinner and more fragile struts,first fractures occur sooner and at the interfaces of both compression plates.Then,instead of a propagating crack near the compression plates,deformations appear in random weak spots within the foam material.A similar mechanism was also reported for Duocel foams by Zhou and Soboyejo[31].

Fig.8.Magnesium foam at different cracking stages during a compression test and the stages’locations on a stress-strain curve.(For interpretation of the references to color in this figure,the reader is referred to the web version of this article.)

As Fig.9 shows,both compressive strength and Young’s modulus drop significantly when porosity levels go over 90%because the metal weight is no longer capable of supporting the structure.Two samples of 10 PPI foam with lower porosities(82.6% and 85.7%)were fabricated in order to compare foams with similar densities and different pore sizes.Their Young’s modulus and compressive strengths were found to better than in 20 PPI foams of similar density.However,their compressive behaviour proved to be more unpredictable than that of 10 PPI foams with lower densities.Moreover,their stress-strain curves are more fluctuating and the structure crack points are clearly visible.Precise methods of controlling the density and uniformity of foam in the wax slurry coating process still need to be developed.At the current stage of technological development,as the wax content increases there is a smaller chance of obtaining foams with regular density and open pores.Thus,it is more beneficial to increase foam density by selecting ductile foams of smaller pore sizes than by thickening the wax layer in the pattern.Overall,the mechanical properties of the obtained foams are within the limits of properties of cancellous bone structures:Young’s modulus within 0.01–2 GPa and compressive strength within 0.2–80 MPa[32].

Fig.9.Compression test results:Young’s modulus(a)and compressive strength(b)vs.porosity.

The energy absorption properties of the foams were also calculated in order to determine which type of foam performs better in this regard.Specific energy absorption(SEA)was calculated using the following Eq.(6)[33]:

Wheresfis crush length andmfis mass of the crushed material.The results of the calculation are shown in Fig.10.The SEA-porosity plot shows that foams with 20 PPI perform better at absorbing energy than 10 PPI foams.One possible explaination for this tendency is that the compression force in 20 PPI foams is spread over a higher amount of pores and strouts than in 10 PPI foams.The graph also indicates that,at least in the porosity range between 80 and 95%,increasing foam densitiy results in better energy absorption.The thickcoated 10 PPI foams of lower porosities(82.6 and 85.7%)performed worse than the 20 PPI foams with similar densities.This further supports the conclusion that minimizing pore size is a better way of increasing the density and mechanical properties of foam than increasing the mass of foam with a larger pore size.

3.2.Fractography analysis

In order to study the fracture behaviour of AZ91 foams,post-compression samples were analysed with an SEM microscope.Foams of such complex morphology undergo all kinds of deformations during compression tests.They not only become compressed,but also stretched,bent and sheared when the neighbouring struts get deformed.That is why analysing their fractographs is much more complicated than in the case of solid materials.The images presented in Fig.11 show that foam struts exhibit not only behaviours leading to simple compression fracture and fracture in the brittle mechanism caused by theα-Mg hexagonal structure[34]and/or relatively large grain size,but also leading to failure under bending and flexing in multiple directions(red arrows)in the ductile state,which is usually induced by the homogenous microstructure and the fine grain size.Another explanation for the two different types of collapse can be structural defects such as air bubbles or pre-existing cracks coming from shrinkage stresses in some of the struts which induce their brittle cracking.

Analysis of the fractures on the struts shows that the foam has two kinds of fracture mechanisms:brittle and ductile.Fig.12 shows a strut with visible necking,but brittle cracks can be seen on the broken surface at a smaller scale.Similar fracture modes have been observed by other authors.Onck et al.studied Duocel aluminium foams[35];however,they obtained results with more ductile and less brittle collapse mechanisms,which can be explained by the better plasticity of aluminium compared to magnesium.The cracked surface of the magnesium alloy foam strut also has pre-existing voids and oxide impurities that initiated the failure.Such defects can be explained by the fact that during the casting process,molten AZ91 fills multiple channels at once,resulting in two liquid metal fronts meeting in one channel,with oxides and gas bubbles remaining at the connection point.Additionally,the casting parameters were adjusted to minimize the solidification time for finer grain size and reduce the formation of Mg2Si on the surface through the reaction between the metal and the mould material.Fast solidification creates a solid coating on metal and stops the vacuum from removing gas bubbles.Unfortunately,such problems are difficult to overcome in the casting process and can only be mitigated by improving casting conditions such as the vacuum environment and mould temperatures.

Fig.10.Specific energy absorption(SEA)vs.foam porosity.

Fig.11.SEM fractographs of compressed AZ91 foams with 20 PPI,showing both brittle strut breaking(green arrows)and ductile strut bending and/or flexing(red arrows).(For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

Figs.12 and 13a show multiple small cracks at the surfaces of strut fractures,while Fig.13b shows a ruptured foam wall.Such cracks which appear on most of the deformed surfaces of the foam,as well as the ruptures in the struts and walls suggest that the collapse mechanism of the foam is brittle rather than ductile,which is a fairly common problem in magnesium and magnesium-based alloys that underlies the hexagonal crystallographic structure and the dislocation mechanisms of magnesium,which has been properly described by Wu and Curtin[34].Wall ruptures similar to those in Fig.13b were also observed in closed-cell aluminium foams by Onck et al[36].shortly before the final failure.Their foam,however,exhibited more brittle fracture behaviour,with metal debris visible on the SEM images,while AZ91 foam appears to fail in a“cleaner”manner,revealing the two remaining matching wall pieces.

Fig.12.Broken strut tip of a 20 PPI AZ91 foam with visible necking and surface cracks.

Fig.13.Brittle cracks at the tip of a broken strut(a)and a ruptured foam wall(b).

Fig.14.Initial crack in the centre of a strut base(a)and a propagating crack at the connection of struts.

Most of the observed cracks and fractures appear in areas with gas bubbles inclusions as defects,such as those shown in Figs.12b and 14.These cracks seem to be appearing far away from the weakest foam structure areas in the centre of the struts.Bigger gas bubbles initiate cracks obviously not only in struts,but also in stronger areas of the foam:at strut bases and strut connections.Therefore,compressive failure of the foam,while overwhelmingly occurring in the centre of individual struts,can be also initiated at strut connections,foam walls and strut bases,depending on the defects in the material structure.

Fig.15.A bent strut of AZ91 10 PPI foam with cracks on the outer surface and wrinkling on the inner surface.

It can be observed from Fig.15 that some of the struts show a more ductile deformation mechanism,with some of them bending almost 180° without fracture.However,many small cracks can be seen on both the outer and the inner bending surfaces.On the outer bending surface,major cracks propagate from defects,thus creating points for potential future strut fractures.Cracks of such shape on the outer bending surface were also observed by Zhou et al.[31]when they investigated aluminium foams.However,they observed more ductile failure modes as those cracks had less rigid edges than those on AZ91 foams because the hexagonal microstructure of magnesium causes more brittle collapse mechanisms.The inner bending surface of the strut,on the other hand,shows folds due to material build-up,initiating cracks and material layer detachment.

3.3.Corrosive behaviour analysis

Tests for the degradation of magnesium alloys are performed in various aquatic environments,for example,in solutions reflecting the chemical composition of body fluids.Most often,the corrosive medium is Hank’s or Ringer’s solution or simulated body fluid(SBF)[37–39].For magnesium alloys in the form of porous scaffolds,corrosion tests encounter a number of difficulties,such as those related to determining the size of the surface being tested[38].Therefore,the rate of degradation of the magnesium alloy(in the form of a scaffold)was determined by measuring changes in the concentration of the individual alloying elements entering the SBF.

The aim of the research was to measure both the migration of the alloying elements to the SBF(Fig.16)and the components of SBF deposited on the surface of the metallic material(Fig.17)as a function of the exposure time.

In aquatic environments,magnesium and its alloys react with water,resulting in the formation of Mg(OH)2and the release of hydrogen(H2).The reaction can be described by formula(3)[37]:

The reaction in formula(3)causes changes in the concentration of individual elements in SBF(Figs.16 and 17).The changes in the concentration of Mg and Al(the main alloy components)in the SBF are shown in Fig.16a;the remaining alloy additions,the proportions of which in the alloy do not exceed 0.9% by weight are shown in Fig.16b.In SBF,the magnesium alloy(in the form of a scaffold)decomposes,which results in an increase in the concentration primarily of the main components of the Mg-Al alloy(Fig.16a).The release rate of magnesium ions was much faster than that of Al ions.After 216 h of exposure to SBF,the concentration of aluminium ions in the solution was almost 9 times lower than that of magnesium ions.

The daily requirement of a healthy adult woman for magnesium is about 300–320 mg,and for men this dosage goes up to 400–420 mg.Thus,the average amount of magnesium released from the alloy into solution by the end of the experiment did not exceed these values.

As a result of the dissolution of the melt and the evolution of H2,the pH value of the solution increased:after 216 h,the pH value was about 9.5–9.6,which is consistent with studies carried out by other researchers[40].At such a pH value,local alkalisation of the solution occurs,which promotes passivation of the surface of the magnesium alloy and,consequently,slows down the oxidation process of the metallic material.The observed slowdown in the migration of magnesium and aluminium to the SBF may have been the result of local passivation of the alloy surface(Fig.16a).

The migration of Be,Ni,Cu,and Fe from the matrix to the SBF was negligible:after 216 h of exposure to SBF,their content in the solution was about 0.02–0.04 mg/dm3(Fig.17).Such a small concentration of these elements in the SBF does not exceed the acceptable daily requirement in an adult human;for example,for iron it is 1–3 mg,for copper 700 mg,and for nickel 0.3–0.5 mg.Both Zn and Si are necessary microelements for the proper functioning of the body.The daily requirement for zinc in an adult is a minimum of 5 mg,but the body should take it in higher doses,from 15 to 20 mg.In the case of silicon,it is assumed that the daily dosage for an adult should be 20–30 mg.The average daily consumption of silicon in the European and North American population is 20–50 mg,though it is higher in China and India(140–200 mg/day),where grains,fruits and vegetables make up the majority of the diet.Thus,the content of the elements Zn and Si released from the matrix of the magnesium alloy to the SBF(about 20 mg/dm3and about 5 mg/dm3,respectively,after 216 h of the alloy’s exposure to the solution)did not exceed the accepted doses for an adult human(Fig.17).The case was similar with Mn.

Fig.16.Changes in the concentration of elements contained in the magnesium alloy after 216 h of exposure to SBF(a)for the main alloying elements,Mg and Al,and(b)for the elements whose proportion within the matrix does not exceed 1% by weight.

Fig.17.Changes in the concentration of SBF elements during the exposure of the magnesium alloy to the solution.

The changes in the concentration of ions comprising SBF were small and occurred mainly during the first 48 h of exposure(Fig.17).The lowered concentration of calcium and phosphorus indicates the adsorption of ions of these elements onto the surface of the scaffold.Data from the literature show that the presence of apatite ceramics is a desirable phase for corrosion protection of Mg alloys[41–43].In the following hours of exposure to the Mg alloy,the concentrations of Ca,P,K,and Na did not change significantly,which indicates the stability of SBF despite the increase in the pH of the solution.

Fig.18.Phosphate-based PEO coating(a)and aluminate-based PEO coating with agglomerated CuO nanoparticles(b)deposited on AZ91 foams.

Fig.19.Fracture of a single foam strut(bar)after PEO treatment in phosphate-based electrolyte(a)and a cross-section of the produced oxide layer with a typical outer porous and dense inner protective layer(b)20 min,2 A,foam as cast,electrolyte up to 44 °C.

Fig.20.EDS analysis of PEO-coated AZ91 foam surface composition.

3.4.Outlook PEO processing of foams

Adding ceramic surface layers can affect the mechanic properties of the coatings on the one hand and,on the other hand,it can improve the degradation behaviour as well as provide opportunities for surface functionalisation e.g.adding antibacterial properties.However,the current work is just a first feasibility study to explore the conditions for applying such coatings by plasma electrolytic oxidation.

The main intention was to screen a variety of different PEO parameters in order to identify a suitable range of parameters to uniformly cover the surface of the foams.The main challenge was to find a current density(actually a current value),since the real surface area is unknown as is the surface condition inside the foams.It turned out that the foam in its raw state needs a high current density to build up a dielectric layer in order to reach breakdown potential and start the discharges.In spite of the cooling of the electrolyte,the heat input was much higher and thus temperatures well beyond 40 °C were reached at the end of the 20 min treatments.At a final temperature of 60 °C,no stable PEO processing was possible anymore,with coating dissolution faster than its growth.Even at a final temperature of 40 °C,a typical PEO coating was produced(Fig.19b).Without cleaning the foam,it was not possible to further lower the current density.Therefore,a cleaning procedure was developed to remove possible surface contaminations from the surface based on an acidic cleaning step(a 30 s dip in aqueous 150 g/L acetic acid)and a rinse in deionised water.After cleaning,stable discharge conditions were obtained at much lower current densities and at low enough input energy levels that the cooling system was able to control the temperature at 20 °C,achieving conditions under which reproducible surface treatments were possible.Finally,the test was extended to different base electrolytes:besides phosphate also silicate and aluminate-based electrolytes were successfully used to apply uniform coatings on the foams.Fig.18a shows a typical surface of a phosphate coating and Fig.18b illustrates a typical aluminate coating with co-deposition of CuO nanoparticles(agglomerated).Latter was considered as a first attempt to see if the foams can be functionalised by particle additions.

Fig.19 shows a foam element fracture with a protective oxide coating formed in the phosphate-based electrolyte,which includes an outer porous layer and a dense layer adjacent to the metal core.Typically,the coating contains elements from the electrolyte and the substrate,which are converted by the discharges into the ceramic-like coating Fig.20.shows the EDS analysis of the PEO coating composition,showing the expected coating forming elements(Mg,P and O).

4.Conclusions

·Open-cell AZ91 Mg alloy foams were manufactured with an elaborated method based on investment casting,using wax-thickened PUR foam with two pore sizes(20 PPI and 10 PPI).20 PPI foams achieved porosity between 80.91 and 87%,while 10 PPI foams achieved porosity between

90.55 and 91.82% in normal coated foams and between 82.55 and 85.74% in thick-coated foams.

·The thin-coated 10 PPI foams did not achieve mechanical properties sufficient for desired applications(Young’s modulus 11.65–23.8 MPa),while thick-coated foams performed well with a Young’s modulus of 36.5–77.5 MPa and a compressive strength of 1.15–2.20,although they turned out to be quite unpredictable and their fracture mechanism was more brittle.

·20 PPI foams performed satisfactorily in mechanical tests,with Young’s modulus ranging from 21 to 60 MPa and compressive strength ranging from 0.53 to 1.12 MPa,which fits the range of mechanical properties of cancellous bone structure,although further research on lower pore sizes and porosities should be conducted for a better understanding of the connection between foam structure and properties.

·Fractured foams SEM analysis showed that the foam collapsing mechanism consists mostly of brittle fracture and cracks.The fractures occur mostly in the middle of the struts,in areas with microstructural defects.Ductile deformations such as bending and necking were also observed.

·After the magnesium alloy was exposed to SBF,the concentration of elements entering the metal matrix did not exceed the acceptable daily requirement in an adult human.The slowdown in the release of the alloy elements may have been the result of both local passivation of the scaffold surface and the deposition of a thin layer of apatite ceramics on the implant.

·PEO coatings were applied successfully to add various ceramic type of surface layers to the foams.The dense inner layer in such coatings may provide good corrosion resistance,while the porous outer layer may be beneficial for bone implants application,both by providing a larger scaffold surface for hydroxyapatite growth and by improving adhesion.Particles can be used to add additional functionalities to the coatings e.g.antibacterial or photocatalytic properties.However,how the additional PEO coatings affect biocompatibility,mechanical integrity and degradation behaviour is a question of further studies.

Conflic of interest

The authors declare that they have no conflicts of interests.

001.
[6]C.Yang,J.Zhu,S.Cui,P.Chen,Z.Wu,Z.Ma,Z.Wu,Surface Coat.Technol.409(2021)126885,doi:10.1016/j.surfcoat.2021.126885.

[7]A.G.Rakoch,E.P.Monakhova,Z.V.Khabibullina,M.Serdechnova,C.Blawert,M.L.Zheludkevich,A.A.Gladkova,J.Mag.Alloy.8(3)(2020)587–600,doi:10.1016/j.jma.2020.06.002.

[8]G.Barati Darband,M.Aliofkhazraei,P.Hamghalam,N.Valizade,J.Mag.Alloy.5(1)(2017)74–132,doi:10.1016/j.jma.2017.02.004.

[9]Q.Dong,Y.Li,H.Jiang,X.Zhou,H.Liu,M.Lu,J.Bai,J.Mag.Alloy.9(4)(2021)1329–1338,doi:10.1016/J.JMA.2020.05.022.

[10]N.T.Kirkland,I.Kolbeinsson,T.Woodfield,G.J.Dias,M.P.Staiger,Mater.Sci.Eng.B Solid State Mater.Adv.Technol.176(2011)1666–1672 Elsevier,doi:10.1016/j.mseb.2011.04.006.

[11]G.Jiang,G.He,Mater.Sci.Eng.C 43(2014)317–320,doi:10.1016/j.msec.2014.07.033.

[12]Z.S.Seyedraoufi,S.Mirdamadi,J.Mech.Behav.Biomed.Mater.21(2013)1–8,doi:10.1016/j.jmbbm.2013.01.023.

[13]D.Yang,Z.Hu,W.Chen,J.Lu,J.Chen,H.Wang,A.Ma,J.Manuf.Process.22(2016)290–296,doi:10.1016/J.JMAPRO.2016.04.003.

[14]D.Yang,W.Chen,J.Lu,Z.Hu,Y.Feng,J.Chen,H.Wang,Metal Powder Rep.72(2)(2017)124–127,doi:10.1016/J.MPRP.2016.02.053.

[15]D.Yang,S.Guo,J.Chen,C.Qiu,S.O.Agbedor,A.Ma,L.Wang,J.Alloy.Compd.857(2021)158112,doi:10.1016/J.JALLCOM.2020.158112.

[16]H.J.Maier,S.Julmi,S.Behrens,C.Klose,A.K.Gartzke,P.Wriggers,A.Meyer-Lindenberg,JOM 72(5)(2020)1859–1869,doi:10.1007/s11837-020-04078-8.

[17]Y.Yamada,K.Shimojima,Y.Sakaguchi,M.Mabuchi,M.Nakamura,T.Asahina,K.Higashi,Adv.Eng.Mater.2(4)(2000)184–187,doi:10.1002/(SICI)1527-2648(200004)2:4〈184::AID-ADEM184〉3.0.CO;2-W.

[18]J.R.Davis,Metals Handbook,Desk Edition,2nd Ed.,ASM International,1998 Retrieved from https://app.knovel.com/web/toc.v/cid:kpMHDEE004/viewerType:toc/?toc-within=AZ91.

[19]Z.Q.Yan,Z.A.Wang,X.F.Wang,H.Liu,J.R.Qiu,Trans.Nonferrous Met.Soc.China 25(2015)3490–3497,doi:10.1016/S1003-6326(15)63986-3.

[20]X.Hu,D.Mayson,M.R.Barnett,J.Alloy.Compd.589(2014)485–490,doi:10.1016/J.JALLCOM.2013.11.092.

[21]A.V.Vyas,M.P.Sutaria,Int.J.Met.20(4)(2020)139–144,doi:10.1007/s40962-020-00530-2.

[22]K.B.Rundman,Metal Casting(Reference Book for MY4130),Dept.of Materials Science and Engineering,Michigan Tech.University,2000,pp.49–52.

[23]J.Chen,X.Zhang,C.Li,X.Zhang,Y.Ren,J.He,J.Chen,Int.J.Photoenergy(2020)2020,doi:10.1155/2020/2478579.

[24]Matweb.Plaster;Molded,Dry.Retrieved April 30,2021,from http://www.matweb.com/search/DataSheet.aspx?MatGUID=0e684bc25a1642d0a90aabe41f59b917

[25]Q.Le,Z.Zhang,J.Cui,S.Chang,in:Materials Science Forum,610–613,Trans Tech Publications Ltd,2009,pp.754–757,doi:10.4028/www.scientific.net/MSF.610-613.754.

[26]K.R.Ravi,R.M.Pillai,K.R.Amaranathan,B.C.Pai,M.Chakraborty,J.Alloy.Compd.(2008)Elsevier,doi:10.1016/j.jallcom.2007.02.038.

[27]T.Kokubo,H.Takadama,Biomaterials 27(15)(2006)2907–2915,doi:10.1016/J.BIOMATERIALS.2006.01.017.

[28]V.M.Posada,J.Ramírez,J.P.Allain,A.R.Shetty,P.Fernández-Morales,J.Mater.Eng.Perform.29(1)(2020)681–690,doi:10.1007/s11665-020-04566-7.

[29]J.M.Rúa,A.A.Zuleta,J.Ramírez,P Fernández-Morales,Surface Coat.Technol.360(2019)213–221,doi:10.1016/j.surfcoat.2018.12.106.

[30]H.Bayani,S.M.H.Mirbagheri,Mater.Charact.113(2016)168–179,doi:10.1016/j.matchar.2016.01.017.

[31]J.Zhou,W.O.Soboyejo,Mater.Sci.Eng.A 369(1–2)(2004)23–35,doi:10.1016/j.msea.2003.08.009.

[32]M.Yazdimamaghani,M.Razavi,D.Vashaee,K.Moharamzadeh,A.R.Boccaccini,L.Tayebi,Mat.Sci.Eng.C(2017)Elsevier Ltd,doi:10.1016/j.msec.2016.11.027.

[33]M.Stamenovi′c,S.Puti′c,M.Zrili′c,L.J.Milovi′c,J.Pavlovi′c-Krsti′c,Metalurgija 50(3)(2011)197–200.

[34]Z.Wu,W.A.Curtin,Nature 526(7571)(2015)62–67,doi:10.1038/nature15364.

[35]P.R.Onck,R.van Merkerk,J.T.M.De Hosson,I.Schmidt,Adv.Eng.Mater.6(6)(2004)429–431,doi:10.1002/adem.200405156.

[36]P.R.Onck,R.van Merkerk,A.Raaijmakers,J.T.M.De Hosson,J.Mater.Sci.40(22)(2005)5821–5828,doi:10.1007/s10853-005-4996-7.

[37]N.I.Zainal Abidin,B.Rolfe,H.Owen,J.Malisano,D.Martin,J.Hofstetter,A.Atrens,Corros.Sci.75(2013)354–366,doi:10.1016/J.CORSCI.2013.06.019.

[38]M.H.Kang,T.S.Jang,S.W.Kim,H.S.Park,J.Song,H.E.Kim,H.D.Jung,Mater.Sci.Eng.C 62(2016)634–642,doi:10.1016/J.MSEC.2016.01.085.

[39]N.I.Zainal Abidin,A.D.Atrens,D.Martin,A.Atrens,Corros.Sci.53(11)(2011)3542–3556,doi:10.1016/J.CORSCI.2011.06.030.

[40]A.Atrens,G.L.Song,M.Liu,Z.Shi,F.Cao,M.S.Dargusch,Adv.Eng.Mater.17(4)(2015)400–453,doi:10.1002/ADEM.201400434.

[41]X.B.Chen,N.Birbilis,T.B.Abbott,Corros.Sci.55(2012)226–232,doi:10.1016/J.CORSCI.2011.10.022.

[42]X.B.Chen,N.Birbilis,T.B.Abbott,Corros.Sci.53(6)(2011)2263–2268,doi:10.1016/J.CORSCI.2011.03.008.

[43]S.Hiromoto,A.Yamamoto,Electrochim.Acta 54(27)(2009)7085–7093,doi:10.1016/J.ELECTACTA.2009.07.033.