On the biodegradability, mechanical behavior, and...

On the biodegradability, mechanical behavior, and cytocompatibility ofamorphous Mg72Zn23Ca5 and crystalline Mg70Zn23Ca5Pd2 alloys astemporary implant materials

E. Pellicer,1 S. Gonz�alez,1 A. Blanquer,2 S. Suri~nach,1 M. D. Bar�o,1

L. Barrios,2 E. Ib�a~nez,2 C. Nogu�es,2 J. Sort3

1Departament de Fısica, Facultat de Ciencies, Universitat Aut�onoma de Barcelona, E-08193 Bellaterra, Spain2Departament de Biologia Cel�lular, Fisiologia i Immunologia, Universitat Aut�onoma de Barcelona, E-08193 Bellaterra, Spain3Instituci�o Catalana de Recerca i Estudis Avancats (ICREA) and Departament de Fısica, Universitat Aut�onoma de Barcelona,

E-08193 Bellaterra, Spain

Received 16 March 2012; revised 23 May 2012; accepted 18 June 2012

Published online 28 August 2012 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/jbm.a.34347

Abstract: The evolution of microstructure and mechanical

properties of almost fully amorphous Mg72Zn23Ca5 and crys-

talline Mg70Zn23Ca5Pd2 alloys during immersion in Hank’s bal-

anced salt solution (HBSS), as well as their cytocompatibility,

are investigated in order to assess the feasibility of both mate-

rials as biodegradable implants. Though the crystalline

Mg70Zn23Ca5Pd2 sample shows lower wettability and more

positive corrosion potential, this sample degrades much faster

upon incubation in HBSS as a consequence of the formation

of micro-galvanic couples between the nobler Pd-rich den-

drites and the surrounding phases. After 22-h immersion, the

concentration of Mg ions in the HBSS medium containing

the Mg70Zn23Ca5Pd2 sample is six times larger than for

Mg72Zn23Ca5. Due to the Zn enrichment and the incipient

porosity, the mechanical properties of the Mg72Zn23Ca5 sam-

ple improve within the first stages of biodegradation

(i.e., hardness increases while the Young’s modulus

decreases, thus rendering an enhanced wear resistance). Cyto-

compatibility studies reveal that neither Mg72Zn23Ca5 nor

Mg70Zn23Ca5Pd2 are cytotoxic, although preosteoblast cell ad-

hesion is to some extent precluded, particularly onto the sur-

face of Mg70Zn23Ca5Pd2, because of the relatively high

hydrophobicity. Because of their outstanding properties and

their time-evolution, the use of the Pd-free alloy in temporary

implants such as screws, stents, and sutures is envisioned.

VC 2012 Wiley Periodicals, Inc. J Biomed Mater Res Part A: 101A: 502–

517, 2013.

Key Words: magnesium, biodegradation, mechanical proper-

ties, metal ion release, cytotoxicity

How to cite this article: Pellicer E, Gonz�alez S, Blanquer A, Suri~nach S, Bar�o MD, Barrios L, Ib�a~nez E, Nogu�es C, Sort J. 2013. Onthe biodegradability, mechanical behavior, and cytocompatibility of amorphous Mg72Zn23Ca5 and crystalline Mg70Zn23Ca5Pd2

alloys as temporary implant materials. J Biomed Mater Res Part A 2013:101A:502–517.

INTRODUCTION

The need for temporary implants such as plates, screws,pins, stents, and sutures has recently prompted a renewedinterest and, in turn, stimulated a lively research activity inthe field of biodegradable materials. Polymers and metalshave been advocated as potential bioabsorbable implants.1,2

Compared with polymers (the poly-L-lactic acid (PLLA) is byfar the most studied), metals show superior mechanicalproperties (e.g., higher strength to bulk ratio) which makethem the material of choice in many applications, for exam-ple for replicating the properties of SS316L—the referencematerial for coronary stents3—or as internal bone fixationscrews/pins.4 Among the metallic candidates, Mg-basedalloys have captured a great deal of attention because of

their outstanding properties such as appropriate corrosionrates in physiological media and Young’s modulus close tothat of bone tissue (Ebone ¼ 3-20 GPa). Pure Mg implantshad been tested in the 1940s as bone plates and screws butwere ultimately abandoned mostly due to the generation ofhydrogen gas during in vivo implantation.5 Alloying Mg withsuitable elements has been demonstrated as an effectivemeans to minimize or even block hydrogen gas releasewhile improving the mechanical properties of the basemetal. Furthermore, the corrosion rates of pure Mg, whichare exceedingly large in physiological conditions, arereduced. In this line, Mg alloys including Mg-Al-Zn (e.g.AZ31), Mg-Al-Zn-Mn (e.g. AZ91D), Mg-Zn-Mn, and rare-earth(RE) containing alloys (e.g. WE43 and LAE442) have been

Correspondence to: E. Pellicer; e-mail: [email protected] or S. Gonz�alez; e-mail: [email protected] or C. Nogu�es; e-mail: Carme.

[email protected]

Contract grant sponsor: Generalitat de Catalunya and Spanish Ministry of Economy and Competitiveness; contract grant numbers: 2009-SGR-

1292, 2009-SGR-282, MAT2011-27380-C02-01, MINAHE3 (TEC2008-06883-C03-03)

Contract grant sponsors: Juan de la Cierva contract (Spanish Ministry of Science and Innovation), ICREA Academia award

502 VC 2012 WILEY PERIODICALS, INC.

investigated as re-absorbable bone implants in rats andguinea pigs with rather great success.6–9 Yet, some of themcontain aluminum and/or RE elements that preclude theirmedical use in clinical patients. Al is known to have adverseeffects on neurons and osteoblasts and has indeed beenlinked to dementia and Alzheimer’s disease, while RE mightlead to hepatotoxicity.8 Hence, there is currently a need forMg alloys free from Al, RE or other heavy metals that showgood biocompatibility and acceptable degradability rates inphysiological conditions. Some investigators have shownthat binary Mg-Zn and ternary Mg-Zn-Ca alloys are two ofthe most promising candidates.10,11 Compared with otheralloying elements, which could cause serious adverse effectswhen released into the blood serum, zinc and calcium arerather harmless. Zinc is an essential trace element to thebody and is vital to keep a healthy immune system, whereascalcium is present in large amounts in bone tissues.

Examples of either fully crystalline or amorphous (i.e.,metallic glass) alloy compositions have been reported forMg-Zn12,13 and Mg-Zn-Ca14,15 families. Because of the lackof dislocations, metallic glasses can exhibit mechanical prop-erties that are often better than those encountered in theircrystalline counterparts. In addition, they typically showimproved corrosion performance due to both their chemicalhomogeneity and the absence of grain boundaries.16 In aprevious work we studied the mechanical properties ofMg72-xZn23Ca5Pdx (2 � x � 6) alloys for use in biomedicalapplications.17 It was found that alloying Pd with Mg-Zn-Caglassy alloy gave rise to changes in the mechanical perform-ance of the material in the as-cast state. The addition of 2 at%Pd was particularly favorable in terms of improving hardnessand wear resistance. However, the biodegradability of thenewly developed alloys was not explored in detail, neither theevolution of the mechanical properties of these materialsupon immersion in bodily solutions. Furthermore, the cyto-toxicity remained mostly overlooked. This is a serious con-cern when it comes to Pd since, compared with other metals,the available information on the release and accumulation ofPd into the body is rather scarce. Pd biocompatibility hasbeen analyzed mostly in relation to dentistry applicationwhere it is used for dental inlays, crowns and bridges, usuallymixed with other materials (gold, silver or copper and zinc).Several reports have described oral symptoms (inflammationof gingival and periodontal tissues) in individuals exposed todental material containing Pd. Similarly, skin allergy has beenreported in persons after ear-piercing with jewels containingPd.18 Other detrimental effects in Pd-sensitized patientsinclude: a specific Interferon (IFN)-c production in peripheralblood mononuclear cells (PBMC) challenged with a Pd com-pound, an increase of Pd concentration in biological fluids inpatients with dental prosthesis, and skin sensitization toPd.19 Scarce articles have tested the toxicity of Pd in vitro.One of them demonstrated that pure Pd metal did notdecrease cell viability after 24 h in contact with human fibro-blasts-keratinocytes cocultures,20 while another reportedthat Pd2þ produced a 50% of cell mortality in mouse fibro-blast cultures at a concentration of 0.3 mM.21 Hence, theeventual release of Pd into the body from biodegradable Pd-

containing alloys poses some risk. The goal of the presentwork is to assess the feasibility of Mg72Zn23Ca5 andMg70Zn23Ca5Pd2 alloys as temporary implants by focusing ontwo aspects: (i) biodegradability in a simulated body fluid,including electrochemical characterization, solution pH evolu-tion with time, concomitant monitoring of release of metalions to the adjoining medium and evolution of the mechanicalproperties, and (ii) mouse preosteoblast cells adhesion ontoalloy surfaces and cytocompatibility. The present study pro-vides baseline information to fathom out the mechanisms re-sponsible for in vitro biocorrosion of Mg72Zn23Ca5 andMg70Zn23Ca5Pd2 alloys and gives some insights regardingtheir cytotoxicity in a physiological environment.

MATERIALS AND METHODS

MaterialsIngot alloys were obtained by melting a mixture of pure ele-ments (>99.9 at. %) of Mg, Zn, Ca and Pd in an inductionfurnace. Rod samples with diameter of 2 mm and up to 20mm length were obtained by remelting the master alloy in aquartz tube and subsequent injection into copper mould inan inert gas atmosphere. For most experiments (corrosion,mechanical testing, cytotoxicity) the disks were about 1 mmthick. They were ground with SiC paper up to 4000 grit andthen thoroughly rinsed in ethanol and distilled water priorto the experiment.

Structural characterizationThe structure of the samples was studied by X-ray diffraction(XRD) (Philips X’Pert) with monochromated Cu Ka radiation(30–90� 2y range, counting time: 10 s, step size: 0.030�). Themicrostructure was observed with a scanning electron micro-scope (SEM) (Zeiss Merlin) equipped with energy dispersive X-ray (EDX) analysis. The surfaces of the samples after beingimmersed in Hank’s balanced salt solution (HBSS, purchasedfrom Aldrich) were observed both before and after removal ofthe corrosion products. For removal of the corrosion products,the specimens were soaked in a CrO3 solution (180 g/L) for afew minutes, and afterwards thoroughly rinsed with ultrapurewater followed by ethanol and finally dried with nitrogen gas.

Contact angle measurementsTo assess the wettability of the alloy surfaces, the contactangle of aqueous drops (<5 lL) deposited onto the surfaceof the alloys was measured using a contact Angle MeasuringSystem DSA 100 from KRUSS at room temperature (sessiledrop technique). Both HBSS and ultrapure (MiliPore MilliQ)water were tested.

Electrochemical measurementsElectrochemical experiments were performed in a thermost-atized, one-compartment three-electrode cell filled with 50mL HBSS at 37�C. A double junction Ag|AgCl reference elec-trode was used with 3M KCl inner solution and 1M NaClouter solution. A Pt sheet acted as a counter electrode. Ini-tially, the specimens (3.14 mm2 exposed area) wereimmersed in the HBSS to determine the open circuit poten-tial (OCP). The potential became usually stable after 3 h of

ORIGINAL ARTICLE

JOURNAL OF BIOMEDICAL MATERIALS RESEARCH A | FEB 2013 VOL 101A, ISSUE 2 503

immersion. Immediately afterwards, the potential was sweptat a rate of 0.1 mV s�1 from 300 mV below the OCPtowards 300 mV above the OCP. The corrosion current den-sity (jcorr) values were determined by extrapolation of theanodic and cathodic Tafel slopes to the corrosion potential(Ecorr). Four replicas per sample were tested.

Metal cations release and pH evolutionIn order to evaluate the time-dependent release of metalcations from the alloy samples and pH evolution, disk sam-ples were introduced in sterilized plastic containers filledwith 45 mL of HBSS (free from CaCl2 and MgCl2) from Invi-trogen. The containers were then sealed and placed in athermostatized bath at 37�C. The pH was monitored with apH-meter from Metrohm. At certain time intervals aliquotsof 5 mL HBSS were pipetted off and placed in tube tests forinductively coupled plasma-optical emission spectroscopy(ICP-OES) measurements. Both tests (metal cations releaseand pH evolution) were extended over 22 h.

Mechanical characterizationThe dependence of mechanical properties of the two alloysamples with the immersion time in HBSS was evaluated bynanoindentation at room temperature in the load controlmode. Prior to the initial test (i.e., before immersion of thesamples in HBSS), the specimens were polished until thetwo flat surfaces of the disks exhibited a mirror-like appear-ance. Then, the disks were immersed in HBSS for differenttimes. After removing the samples from the HBSS bath thedisks were not polished again (i.e., polishing was only car-ried out once, before the immersion experiments), but thecorrosion products were chemically etched away using theCrO3 solution.15,22 Nanoindentation tests of the as-cast andimmersed specimens were performed on one of the twofaces of the disks using a diamond Berkovich-type tip. Theindentation function consisted of a loading segment of 150s, to a maximum load of 500 mN, followed by a load holdingsegment of 50 s and an unloading segment of 150 s. Thethermal drift was kept below 0.05 nm s�1.

The results shown in this work correspond to the aver-age of a total of 30 indentations per sample. The hardness(H) and reduced elastic modulus (Er) values were derivedfrom these load-displacement curves using the method ofOliver and Pharr,23 while the wear resistance was esti-mated as the ratio between H and Er, H/Er. Proper calibra-tions for the instrument compliance, tip blunting, thermaldrift and initial penetration depth were applied. It shouldbe noted that Er takes into account the elastic contribu-tions from both the sample and the indenter tip. However,due to the high Young’s modulus of the indenter tip, Er isusually very close to the Young’s modulus of the investi-gated sample.24

Cytotoxicity studiesMouse preosteoblasts (MC3T3-E1, ATCC) were maintainedin a-minimum essential medium (Invitrogen) supplementedwith 10% foetal bovine serum in standard conditions (37�C,humidified atmosphere and 5% CO2). Sample disks were

first glued individually onto glass coverslips using silicone,and the coverslips introduced into a 4-multiwell cultureplate. Samples were sterilized by ultraviolet light for at least2 h. Preosteoblasts were then seeded into each well (75,000cells/well) and cultured for 4 and 8 h. Control cells wereseeded directly on the glass coverslip in the absence of thealloy and cultured also for 4 and 8 h. Alloy cytotoxicity wasevaluated by detecting the activity of intracellular esterases,using the live/dead Viability/Cytotoxicity Kit for mammaliancells (Invitrogen) according to the manufacturer’s protocol.This kit allows distinguishing live cells (showing green colorbecause of their esterase activity) from dead ones (present-ing red color because of permeability of their damagedplasma membrane to propidium iodide). Samples were ana-lyzed under an Olympus IX71 inverted microscope equippedwith epifluorescence. Several images from different regionsof each sample were captured and live/dead cells werecounted (minimum 400 cells per experiment). Differences inthe number of live/dead cells between the experimentalgroups were analyzed using the Fisher test. Statistically sig-nificance was considered when p < 0.05.

Cell adhesion analysisAfter the cell viability assay, control cells and cells grownonto disk alloys were processed for SEM analysis. Cellswere rinsed twice in 0.1M sodium cacodylate buffer (SCB),fixed in 2.5% glutaraldehyde in 0.1M SCB for 45 min atroom temperature and rinsed twice in SCB. Cell dehydrationwas performed in a series of ethanol washes (50, 70, 90,and two 100%) for 8 min each. Dehydrated samples werethen dried using a substitute for critical point drying: 50%of hexamethyl disilazane (HMDS) in absolute ethanol fol-lowed by 100% of HMDS. Finally samples were mounted onstubs and examined using the SEM (Zeiss Merlin) equippedwith EDX analysis.

RESULTS

Microstructural characterizationFigure 1(a,c) shows the electron back-scattered (EBS) SEMimages of Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloy samples,respectively. While the former is mostly featureless and justa few scratches from the polishing procedure are seen, aclear microstructure consisting of graded-contrast featuresis observed in the latter. Such a drastic change in the micro-structure is ascribed to the addition of Pd to the alloy,which is known to decrease the glass-forming ability. Thecorresponding XRD patterns [see Fig. 1(b,d)] are in agree-ment with the EBS images. Namely, a broad halo centered ataround 2y ¼ 39� is observed for the Mg72Zn23Ca5 sample.This confirms the mainly amorphous nature of the Pd-freealloy. Conversely, several reflections appear in the XRD pat-tern of the Mg70Zn23Ca5Pd2 sample. These peaks match theangular positions expected for hexagonal CaZn5, hexagonalMgZn, cubic Mg6Pd, and trigonal Mg6Zn3Ca2 phases.25 Thelocal compositional variations within each region of theMg70Zn23Ca5Pd2 sample were evidenced by EBS observa-tions. Figure 2 shows a representative EDX mapping corre-sponding to a central area of the Mg70Zn23Ca5Pd2 disk.

504 PELLICER ET AL. AMORPHOUS Mg72Zn23Ca5 AND CRYSTALLINE Mg70Zn23Ca5Pd2 ALLOYS

Apparently, Pd is concentrated within the bright dendrites,which are thus attributed to Mg6Pd. Instead, Ca and Zn ele-ments mainly contribute to the grey and black zones sur-rounding the bright dendrites, together with Mg element.Hence, the hexagonal CaZn5 and MgZn phases and the trigo-nal Mg6Zn3Ca2 phase are presumably located in the regionssurrounding the dendrites.

Electrochemical and contact-angle measurementsFigure 3(a) shows representative potentiodynamic polariza-tion curves for Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloys.Though the jcorr values are comparable (1.7 mA cm�2 and2.1 mA cm�2, respectively), Ecorr for the Mg70Zn23Ca5Pd2alloy is shifted 110 mV towards more anodic values, thussuggesting a delayed onset of material degradation in physi-ological conditions. Indeed, it has been demonstrated thatthe formation of superhydrophobic surfaces onto magne-sium alloys improves their corrosion resistance.26 Severaltreatments have been pursued to this end, such as anodicoxidation27 or chemical vapor deposition.28 As a result ofsuch treatments, an Ecorr shift towards more positive valuesis typically observed.29 Figure 3(b,c) shows photographs ofdroplets of HBSS deposited onto the surfaces ofMg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloys. The measuredcontact angles are 86.7� and 102.7� , respectively. A similartrend was found when depositing ultrapure water droplets,that is, a higher contact angle is measured forMg70Zn23Ca5Pd2 (100.4�) compared with the Pd-free sample(82.7�). Although Mg-based and Mg-containing crystallinematerials and, in general, multiphase crystalline alloys typi-cally exhibit poorer corrosion resistance compared withtheir homologous amorphous counterparts, there are someexamples in the literature where the opposite trend isobserved.30,31 In this study, not only the microstructurechanges from amorphous to crystalline but also the chemi-cal composition varies. Hence, both the evolution from a sin-gle amorphous structure to a multiphase crystalline alloyand the change in the nominal chemical composition needto be taken into account.

pH dependence and metal cations releaseon immersion timeIn order to get a better understanding on the degradationperformance of the alloy samples, as-polished disk speci-mens were immersed into HBSS at 37�C under nonpolarizedconditions. Figure 4 shows the time evolution of HBSS pHfor the two samples. An increase in the bulk pH of the solu-tion was observed in both cases, though higher values weremeasured for the Pd-containing alloy from the very begin-ning. The alkalinization effect stems from the release ofOH� ions due to the oxidation of Mg metal to Mg2þ accord-ing to the following reaction:

Mgþ 2H2O ! Mg2þ þ 2OH� þ H2 (1)

After 21.5 h, the pH readings were 7.93 and 8.58 for theMg72Zn23Ca5 and Mg70Zn23Ca5Pd2 incubated samples,respectively. Though it has been reported that local pH atsample surface may differ in several pH units comparedwith bulk pH32, these results point to a higher corrosionrate of the Pd-containing sample, which seems inconsistentwith the previous electrochemical polarization measure-ments. Furthermore, after 4 h of immersion, theMg70Zn23Ca5Pd2 disk disintegrated into a few pieces as aconsequence of the formation of deep cracks that pro-gressed across the disks. As will be described in detail later

FIGURE 1. (a) and (c) SEM images taken at the disk center of

Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloys, respectively. (b) and (d) are

the corresponding XRD patterns.

ORIGINAL ARTICLE


such apparent contradiction between the results fromimmersion experiments (non-polarized conditions) and theelectrochemical data (polarized conditions) can be under-stood on the basis of micro-galvanic couples originatingbetween the nobler Mg6Pd phase and the surrounding Mg-Zn matrix that leads to selective dissolution of phase boun-daries in the Mg70Zn23Ca5Pd2 alloy. Such a selective etchingspreads all over the sample ultimately leading to its rapiddisintegration. Figure 5(a,b) shows the SEM morphologies ofthe corroded surfaces after incubation of the alloys in HBSSfor 2 h. While the Mg70Zn23Ca5Pd2 sample displays arougher surface and, apparently, larger fraction of corrosionproducts, the Mg72Zn23Ca5 alloy shows more uniform corro-sion morphology, similar to that observed in other Mg-Zn-Caalloy compositions.15 Differences in the surface morphologyfor the two investigated samples are also observed afterselective etching of the corrosion products. As shown in Fig-ure 5(c), immersion of the Mg72Zn23Ca5 alloy in HBSS for 2h already causes certain surface corrugation as a result of

the dealloying process. The effects are much more pro-nounced in the Mg70Zn23Ca5Pd2 sample [Fig. 5(d)], wheredeep pores with sizes in the micrometer range are seen.This large porosity is closely related to the occurrence ofmicro-galvanic couples, which causes severe dissolution ofMg metal.

The effects of dealloying have been carefully investigatedby assessing the levels of metal cations released from thealloy disk samples. Figure 6 shows the concentration of Mg,Zn, Ca (and Pd) ions in the HBSS extraction medium as afunction of the immersion time. First of all, the concentrationof Mg cations released from the Pd-containing sample notori-ously increases from 4 h onwards and at 21.5 h of immersionis six times higher than for the Mg72Zn23Ca5 sample, whichconfirms its faster degradation. The release of Zn ions fromMg70Zn23Ca5Pd2 also surpasses the levels of the Mg72Zn23Ca5sample. Remarkably, no Pd was detected in the aliquotspipetted off from the Mg70Zn23Ca5Pd2-containing HBSS me-dium. This means that Pd is not released into the solution

FIGURE 2. EDX mapping of a central region of Mg70Zn23Ca5Pd2 disk sample. [Color figure can be viewed in the online issue, which is available

at wileyonlinelibrary.com.]

FIGURE 3. (a) Potentiodynamic polarization curves of Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloys in HBSS at 37�C. Photographs of as-deposited

HBSS droplets onto the surface of (b) Mg72Zn23Ca5 and (c) Mg70Zn23Ca5Pd2 alloys. [Color figure can be viewed in the online issue, which is

available at wileyonlinelibrary.com.]


but rather remains in the metallic form. These results rein-force the hypothesis that micro-galvanic couples exist in theMg70Zn23Ca5Pd2 sample. The compositional analyses, per-formed by EDX on the alloy samples, as a function of HBSSimmersion time (see Table I) are in agreement with the fasterdissolution rate of Mg compared with the other constituentelements. The disintegration of the Mg70Zn23Ca5Pd2 alloy in

a short period of incubation (around 4 h) seriously precludesits use in hard-tissue repair. This means that mechanical in-tegrity of the implant is spoiled before the tissue has suffi-cient time to heal the bone.33

Mechanical propertiesShown in Figure 7(a,b) are the dependences of the hard-ness, H, and the reduced Young’s modulus, Er, on the immer-sion time in HBSS for the Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2alloys. Results from macroscopic compression experiments(not shown) revealed that the two investigated alloys arerather brittle but confirmed that the Young’s modulus ofMg72Zn23Ca5 is lower than for Mg70Zn23Ca5Pd2. Interest-ingly, while Er tends to simply decrease with the immersiontime in both cases, the dependence of H is more complex,particularly in the case of Mg72Zn23Ca5. For this alloy, Hsteeply increases after short-term immersion, decreases forlonger times and increases again, but only slightly, forimmersion times longer than 16 h.

From a tribological point of view, it is known that thewear resistance of a given material not only depends on itshardness (harder materials typically exhibit a higher wear re-sistance) but also on the reciprocal of the Young’s modulus.Actually, the elastic strain to failure, which is related to theH/Er ratio measured by nanoindentation, has been shown tobe a suitable parameter to estimate the wear resistance, evenmore than the hardness itself.34,35 Figure 7(c,d) reveals thatlarger H/Er are observed for the amorphous Mg72Zn23Ca5alloy than for the crystalline Mg70Zn23Ca5Pd2 sample.

FIGURE 4. Bulk pH evolution of HBSS as a function of immersion

time for Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 disk samples at 37�C.

[Color figure can be viewed in the online issue, which is available at

wileyonlinelibrary.com.]

FIGURE 5. SEM images of surface morphology of (a,c) Mg72Zn23Ca5 and (b,d) Mg70Zn23Ca5Pd2 disk samples after immersion in HBSS at 37�C

for 2 h. Note that panels (a) and (b) correspond to the samples with the corrosion products, whereas panels (c) and (d) are obtained after remov-

ing the corrosion products using the CrO3 solution.

ORIGINAL ARTICLE


Remarkably, a peak in H/Er is observed after 2h of immer-sion (due to the increase of hardness) and this ratio againincreases for immersion times longer than 16h, indicatingthat the wear resistance shows a nonmonotonic behaviorduring corrosion of the Mg72Zn23Ca5 alloy. Conversely, due toits high corrosion rate and concomitant large porosity, thecrystalline Mg70Zn23Ca5Pd2 alloy practically only exhibits adecrease in H/Er with immersion time.

Cytotoxicity analysisThe potential toxic effect of Mg72Zn23Ca5 orMg70Zn23Ca5Pd2 alloys on preosteoblasts cultures was ana-lyzed through the live/dead Viability/Cytotoxicity Kit. For

each sample, fluorescence images of cells growing on top ofthe disk and of cells growing on top of the coverslip weretaken (Fig. 8). No significant differences were found in thenumber of live cells between cultures in the presence ofMg70Zn23Ca5Pd2 alloy and control cultures after 4 or 8 h ofincubation, neither when comparing control cultures withcultures in presence of Mg72Zn23Ca5 after 8 h of incubation(Fig. 8). In all cases more than 93% of cells were aliveregardless of the presence or absence of the alloy. However,for both alloys and time-points tested, fewer cells werefound growing on top of the disks than on the coverslips.Whereas after 8 h of culture many cells were still attachedto the Mg72Zn23Ca5 disk, and 59.9% were still alive, veryfew cells were attached to the Mg70Zn23Ca5Pd2 disk andmost of them were dead. In fact, very few live cells wereseen on the top of Mg70Zn23Ca5Pd2 after 4 h of culture.

Cell adhesion studiesAdhesion studies of cells growing on top of the coverslipsand of cells growing on top of the disks were also carriedout. Figure 9 shows the morphology and adhesion degree ofcells growing on top of the coverslip at 4 and 8 h. After 4 hin culture, some control cells were completely attached tothe coverslip and thus presented a flat morphology, whereasothers appeared to be in the process of attachment. Cellconfluence was about 60–80% depending on the area. Com-pletely and partially attached cells were also observed onthe coverslip of culture containing the Mg70Zn23Ca5Pd2 disk;preosteoblasts appeared healthy with the only differencethat cells were covered by debris of soft appearance. After 8h of culture, almost all control cells were completelyattached to the coverslip, and in a large number of cellsnucleoli could be observed in their nuclei (Fig. 9). The mor-phology of cells attached to the coverslip of cultures in pres-ence of Mg72Zn23Ca5 or Mg70Zn23Ca5Pd2 at 8 h was similarto that of cells from the control culture, also presentingnuclei with several nucleoli. Soft debris was also observedin cultures in presence of Mg72Zn23Ca5 and in lower quan-tity in the presence of the Mg70Zn23Ca5Pd2 alloy. The celldensity was similar in all cultures analyzed after 8 h andwas equivalent to that seen after 4 h of culture.

The morphology and adhesion degree of cells growingon top of the alloy disks at 4 and 8 h can be seen in Figure10. Cells adhered to the Mg70Zn23Ca5Pd2 disk after 4 h of

FIGURE 6. Mg, Ca, Zn (and Pd) elements concentration in 5-mL ali-

quots withdrawn from the HBSS extraction medium at different peri-

ods of immersion for (a) Mg72Zn23Ca5 and (b) Mg70Zn23Ca5Pd2 disk

samples at 37�C. The inset in (a) shows a zoomed detail of the con-

centration profiles. Volume corrections have been applied to account

for the successive withdrawn of aliquots from the test solution. [Color

figure can be viewed in the online issue, which is available at

wileyonlinelibrary.com.]

TABLE I. Summary of Compositional Results, Obtained

Using EDX, on the Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2

Samples in Their As-Cast States and after Being Immersed in

HBSS for 2 and 50 h (the Latter Only for Mg72Zn23Ca5)

Sample Mg Zn Ca Pd

Mg72Zn23Ca5 as-cast 72.0 23.0 5.0 –Mg70Zn23Ca5Pd2 as-cast 70.0 23.0 5.0 2.0Mg72Zn23Ca5 2 h HBSS 63.5 32.0 4.5 –Mg70Zn23Ca5Pd2 2 h HBSS 62.0 31.5 4.5 2.0Mg72Zn23Ca5 50 h HBSS 50.5 46.0 3.5 –

For such analyses the corrosion products were previously removed

using a CrO3 solution. Data reported has been normalized to 100% af-

ter neglecting the O content (around 2%).


culture [Fig. 10(a,b)] were difficult to distinguish because ofthe large amount of debris. The disk had visible cracks andcells were completely covered by debris, which looked morecompact than that covering the cells grown onto the cover-slips. Protrusions that seemed to correspond to a cell wereanalyzed using EDX analysis. Such protrusions displayedlarger amounts of carbon than the surroundings, hence con-firming the presence of cells [Fig. 10(e,f)]. After 8 h of cul-ture, many healthy-looking cells were found attached to thedisk of Mg72Zn23Ca5 alloy, but most of them were not firmlyattached. Cells were also covered with soft debris [Fig.10(c,d)]. After 8 h of culture, the Mg70Zn23Ca5Pd2 alloy diskwas completely fractured, and only small fragments couldbe seen scattered on top of the coverslip with no cellsadhered to them.

DISCUSSION

Wettability and corrosionA correlation is usually observed between the wettability ofa given material and its anticorrosion performance. Materi-als with more hydrophobic character are typically more re-sistant against corrosion in aqueous environments. Thisaccounts for crystalline as well as for amorphous solids.36

The wettability can be determined by means of the sessiledrop technique from the measurement of the contact anglebetween a liquid drop and the polished surface of the stud-

ied material. In a first approximation, if the wetting is non-reactive, the Young’s equation correlates the contact anglewith the existing interfacial tensions:

cos h ¼ rSV � rSL

rLV(2)

where rSV, rSL, and rLV are the interfacial tensions betweensolid-gas, solid–liquid and liquid–gas phases, respectively.Our results indicate that the crystalline Mg70Zn23Ca5Pd2alloy exhibits markedly larger contact angles than the amor-phous Mg72Zn23Ca5, in spite of their close composition.Although the wettability of metallic glasses can varydepending on the existing amount of free volume frozen inthe glassy structure during the casting,37 it has been shownthat, quite often, metallic glasses are more hydrophilic(lower contact angles) than crystalline materials with thesame composition.38,39 Since rLV depends mainly on the liq-uid used for the sessile drop experiments, its value can beregarded as constant. Hence, the change in contact anglebetween amorphous and crystalline structures of the samecomposition has to be attributed to variations in rSV and/orrSL. It has been argued that since the structure of metallicglasses is metastable, their atomic activity is rather highand this would tend to favor the spreading of the liquiddrop on the solid surface, hence decreasing rSL with respectto the value in the crystalline alloy.38,39 In turn, due to the

FIGURE 7. Dependence of the hardness, H, and reduced Young’s modulus, Er, on the immersion time in HBBS for: (a) Mg72Zn23Ca5 and (b)

Mg70Zn23Ca5Pd2 samples. Dependence of the H/Er ratio on the immersion time for: (c) Mg72Zn23Ca5 and (d) Mg70Zn23Ca5Pd2 samples. [Color fig-

ure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

ORIGINAL ARTICLE


high atomic activity and the presence of not-fully bondedstates at the surface of metallic glasses, the value of rSV

could be higher than for the crystalline alloy.38 Conse-quently, the value of cos y would turn out to be larger forthe amorphous structure [see Eq. (2)], resulting in a lower

contact angle. It should be pointed out that in some highlycorrosive Ca-based metallic glasses, suitable surface treat-ments, such as coating with appropriate materials (e.g., fluo-roalkylsilane or pure Fe films) has been reported to resultin a significant enhancement of hydrophobicity and

FIGURE 8. Cytotoxicity analyses of Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloys on preosteoblasts cultured for 4 and 8 h. Live cells, with plasma

membrane integrity and esterase activity, emit green fluorescence whereas dead cells, with damaged membranes, emit red fluorescence. No-

menclature: ‘‘control’’ (cells on the coverslip in the absence of the alloys), ‘‘Mg72Zn23Ca5 coverslip’’ (cells on the coverslip in the presence of

Mg72Zn23Ca5 disk), ‘‘Mg72Zn23Ca5 disk surface’’ (cells on the Mg72Zn23Ca5 disk surface), ‘‘Mg70Zn23Ca5Pd2 coverslip’’ (cells on the coverslip in the

presence of Mg70Zn23Ca5Pd2 disk) and ‘‘Mg70Zn23Ca5Pd2 disk surface’’ (cells on the Mg70Zn23Ca5Pd2 disk surface). [Color figure can be viewed in

the online issue, which is available at wileyonlinelibrary.com.]

FIGURE 9. SEM images showing the morphology and adherence of cells growing on top of the coverslips (indirect analyses) after 4 h and 8 h

in culture, in the absence (‘‘control’’) and in the presence of Mg72Zn23Ca5 and Mg70Zn23Ca5Pd2 alloys.


improvement of the corrosion resistance.40,41 In this sense,the nobler character of Pd, as compared with Mg or Ca,could assist to provide an effective anti-corrosion protectionof the Mg72Zn23Ca5.

Aside from the wettability, one would expect instead thecrystalline Mg70Zn23Ca5Pd2 alloy to show more cathodicEcorr and larger jcorr values than the amorphousMg72Zn23Ca5 if their microstructure is considered. Actually,when dissimilar metallic phases are in intimate electricalcontact in a corrosive environment, the less noble metal ispreferentially corroded at the expense of the more noblemetal. This phenomenon is termed galvanic corrosion and

could in principle develop in the Mg70Zn23Ca5Pd2 alloy.However, the occurrence of galvanic corrosion can be re-tarded if the material gets covered by a protective passivelayer, which acts as efficient barrier to corrosion. In such acase, the resulting Ecorr is typically near to that of the morenoble metal (or phase). Hence, the shift in Ecorr with theaddition of Pd to the alloy is, to some extent, to be expectedand is in agreement with the sessile drop technique meas-urements. However, if just the Ecorr values are taken intoaccount, one can underestimate the hazards of galvanicattack.42 Indeed, the jcorr values retrieved by Tafel extrapola-tion of the potentiodynamic polarization curves can be

FIGURE 10. SEM images showing the morphology and adherence of cells growing on top of alloy disks (direct analyses). (a) Cells attached to

the Mg70Zn23Ca5Pd2 disk after 4 h in culture, detail in (b). (c) Cells attached to the Mg72Zn23Ca5 disk after 8 h in culture, detail in (d). The EDX

analyses of the zones indicated with ‘‘1’’ and ‘‘2’’ in (b) are displayed in panels (e) and (f), respectively. [Color figure can be viewed in the online

issue, which is available at wileyonlinelibrary.com.]

ORIGINAL ARTICLE


regarded as ‘‘instant’’ values; that is, they do not reflect thechanging nature of corrosion rate with time.43

Results from immersion experiments in the HBSS indicatea higher corrosion rate of the Pd-containing sample thatseems inconsistent with the previous electrochemical polar-ization measurements and contact angle tests. To understandthis apparent contradiction, one should bear in mind that ina potentiodynamic polarization test, both cathodic and an-odic reactions are artificially accelerated, which induces a sig-nificant disturbance of material surface. This has been dem-onstrated to be particularly tricky in Mg-based alloys due tothe so-called ‘‘abnormal polarization behavior’’.44 This impliesthat the polarization curve might not necessarily follow theTafel equation; henceforth, using the Tafel extrapolationmethod or the Stern-Geary equation to calculate the jcorr val-ues can provide biased data. There is another feature in thepolarization curves shown in Figure 3(a) that further provesthe vulnerability of Mg70Zn23Ca5Pd2 sample. The anodicbranch corresponding to this sample is especially noisy, or atleast is noisier than the curve corresponding to Mg72Zn23Ca5.This can be attributed to the slow scan rate used (0.1 mV/s),which makes the potential to be forced away from the steadyvalue for extended periods, thereby allowing significant cor-rosion to occur.43 A proposed scheme of the natural biode-gradation process (i.e., under non-polarized conditions) forthe Mg70Zn23Ca5Pd2 alloy is depicted in Figure 11. When thesample is immersed in the HBSS, preferential anodic dissolu-tion of the Mg-Zn matrix occurs at the expense of the Pd-con-taining dendrites, so that the release of Mg2þ ions (and ofZn2þ to a less extent) into the medium is accompanied byintense hydrogen gas evolution [Fig. 11(a,b)]. This brings

about the formation and progress of cracks across the disksample. The divalent ions precipitate onto the sample surfaceforming a Mg(OH)2 (and Zn(OH)2) passive layer, and theirformation is promoted as the pH of the HBSS increases [Fig.11(c)]. However, the chloride anions of the HBSS make thepassive layer extremely vulnerable and cause its rapid solubi-lization [Fig. 11(d)].15 The protective layer is thus progres-sively thinned and the release of the ions coming from theless noble regions of the sample (i.e., Mg2þ, Zn2þ and Ca2þ)is increased [Fig. 11(e)]. At some point, the disk completelybreaks due to mechanical disintegration rendering nondis-solved Pd-containing fragments as final products [Fig. 11(f)].

Mechanical propertiesAlthough the evolution of the mechanical properties withtime during in vitro or in vivo experiments using biodegrad-able implants is of upmost importance for the applicability ofsuch materials, this issue has been only investigated in somedetail in bioabsorbable polymers45 but has been largely over-looked in biodegradable metallic alloys. The results pre-sented in Figure 7 show that the hardness in Mg72Zn23Ca5increases by roughly 50% during the first 2 h of immersion,whereas that of Mg70Zn23Ca5Pd2 mainly decreases with time,although a modest increase is also observed after immersingthe alloy for a few minutes in the HBSS. The observedincrease of H after short-term immersion could be mainlyattributed to the fast dissolution of Mg into the HBSS. Indeed,the dissolution rate of Mg is faster than that of Zn and Ca(see Fig. 6), therefore an enrichment in Zn occurs at the sur-face of both materials as they get progressively corroded [seeFig. 12(a) and Table I]. Since the hardness of Zn is almost

FIGURE 11. Illustration of the corrosion process for Mg70Zn23Ca5Pd2 sample upon immersion in HBSS: (a) to (b) anodic dissolution of the less

noble regions takes place accompanied by hydrogen gas release; (b) to (c) the Mg2þ ions precipitate onto the sample surface as Mg(OH)2 pas-

sive layer; (c) to (d) the chloride anions destroy the hydroxide layer; (d) to (e) anodic dissolution unavoidably proceeds so that more and more

Mg2þ (and Zn2þ and Ca2þ) ions are released into the HBSS; (e) to (f) sample disintegration occurs leaving the noble Pd-containing sample pieces

undissolved. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]


twice that of Mg46, this could therefore induce a mechanicalhardening effect. However, for longer immersion times thedissolution progress is such that both alloys not only undergosurface chemical change but also a physical modificationwith the formation of corrugations and pores (Fig. 5). Whenthese physical flaws become large enough, their softeningeffect can compensate the hardening caused by Mg dissolu-tion and the hardness decreases.

The lowering of reduced Young’s modulus is probablyrelated also to the increase of porosity as the samples getprogressively dealloyed, as evidenced in Figure 12(b,c) forMg72Zn23Ca5 and in Figure 5(d) for Mg70Zn23Ca5Pd2.Indeed, it is well known that the porosity level has a stronginfluence on the measured values of Young’s modulus. In afirst approximation, if the porous material is treated as anopen-cell metallic foam, the Young’s modulus, Eporous, can bedescribed as follows47–49:

Eporous ¼ C1Ebulkqporousqbulk

� �2

(3)

where Ebulk is the Young’s modulus of the bulk, nonporousmetal, C1 is a geometry constant close to 1 and qporous/qbulkis the relative density of the foam.

Remarkably, while both H and Er decrease rather fast inthe Mg70Zn23Ca5Pd2 alloy (after 2 h both parameters deteri-orate considerably), in the Mg72Zn23Ca5 alloy relatively largevalues of H and Er are retained even after 30 h of immer-sion. This can be understood bearing in mind the faster deg-radation rate (and larger degree of porosity) inMg70Zn23Ca5Pd2 as compared with the Mg72Zn23Ca5 alloy[cf. Figure 5(d) and 12(b)].

At this point it should be noted that enrichment in Znmay also bring about an increase of Er, since the Young’smodulus of Zn is higher than that of Mg. This is notobserved experimentally, probably because the effects of po-rosity overcome any eventual increase of Er. It is worthmentioning that while the H/E ratio tends to slightlydecrease with the porosity level in ceramic materials, theopposite trend is often encountered in metals.50,51 In ourcase, H/Er increases from around 0.053 (in as-castMg72Zn23Ca5) to �0.1 (in Mg72Zn23Ca5 immersed in HBSSfor 2 h), thus indicating that the effect of porosity on theYoung’s modulus after short-term immersion is more pro-nounced than on hardness, which is in agreement withwhat is observed in many other metallic alloys.51,52

Despite the large porosity level, the Young’s modulus ofthe Mg72Zn23Ca5 alloy does not decrease drastically, even af-ter long-term immersion. Actually, the value at 50 h immer-sion time (Er ¼ 37 GPa) is very close to that of humanbone. In turn, H tends to slightly increase again after long-term immersion in HBSS. Such trends can be ascribed tothe densification effect that occurs in this alloy during nano-indentation, particularly when the porosity level is suffi-ciently high [see Fig. 12(c)]. Densification is known to causean increase of the compressive yield stress47 and could alsoslow down the decrease in Er. The dissolution rate of Mg,which is particularly high for long immersion times [seeFigs. 6(a) and 12(a)], could also contribute to the observedincrease in H.

Figure 13 compares the mechanical properties (com-pressive yield stress, ry,C, Young’s modulus, E, and resil-ience, which is proportional to r2

y,C/E) of different familiesof materials suitable for bioabsorbable implants, includingbiodegradable polymers, metallic alloys and ceramics.Mechanical hardness is directly proportional to ry,C, that is,

FIGURE 12. (a) EDX profiles corresponding to the Mg72Zn23Ca5 sam-

ple in the as-cast state and after immersion in HBBS during 2 and 50

h. Secondary-electron SEM image corresponding to indentation

impressions performed on the Mg72Zn23Ca5 sample after immersing it

in HBBS during (b) 2 h and (c) 50 h. Note that the corrosion products

were removed using a CrO3 solution prior to the nanoindentation

tests. [Color figure can be viewed in the online issue, which is avail-

able at wileyonlinelibrary.com.]

ORIGINAL ARTICLE


H � C ry,C (where C is the so-called constraint factor and isusually around 3 for crystalline metallic alloys and a bitlarger for metallic glasses53). Figure 13(a) reveals that Mg-Zn-Ca bulk metallic glasses exhibit rather large values ofyield stress, thus suggesting that these alloys are amongstthe hardest biodegradable materials so far reported in theliterature. In turn, Mg-Zn-Ca glassy alloys exhibit Young’smodulus values closer to that of cortical bone (Ebone ¼ 3–20 GPa) than most synthetic hydroxyapatites or crystallineMg-based alloys (which are also mechanically softer). Takinginto account that Mg70Zn23Ca5Pd2 is fully crystalline, itshardness is also much larger than that of most Mg-Zn-Cacrystalline alloys. Mg-based bulk metallic glasses are alsoharder than most Fe-based biodegradable alloys, which havethe additional drawback that they are often ferromagneticat room temperature (except FeMn, which is antiferromag-netic), hence precluding the use of nuclear resonance imag-ing techniques for diagnostics purposes. After long-termimmersion in HBSS, the Young’s modulus of Mg72Zn23Ca5approaches that of Ca-based or Sr-based biodegradable me-tallic glasses, as well as that of polymeric materials rein-forced with glassy fibers. At the same time, the Mg72Zn23Ca5alloy remains mechanically harder than all these materials.

Resilience is another important mechanical parameter inmaterials intended to be used as biological implants. Thisproperty refers to the ability of a given material to absorbenergy when it is deformed elastically and then release itupon unloading (i.e., the maximum energy per unit volumethat can be elastically stored in the material, without caus-ing any plastic deformation) and it is determined as theratio r2

y,C/2E. Biomaterials such as articular cartilage orintervertebral disks require high mechanical resilience.Figure 13(c) reveals that Mg-Zn-Ca bulk metallic glassesshow rather high values of resilience, similar to Ca-basedmetallic glasses or some biodegradable polymers, particu-larly those reinforced with glassy fibers. The high resilienceof Mg-Zn-Ca glasses, as compared with their homologouscrystalline alloys, stems from their rather high yield strain.

Cytotoxicity and cell adhesionCytotoxicity can be measured indirectly, that is, culturingcells in the presence of a solution that has been previouslyin contact with a specific material, or directly, growing thecells on top of the specific material.56 In this work, cellswere seeded in wells of 2 cm2 where the alloy (3.14 mm2)was glued on top of a coverslip with nearly the same areaof the well. Cells were distributed randomly onto the alloydisk and the coverslip, allowing to carry out indirect anddirect analyses at the same time. Cells growing on the cov-erslip were exposed to metal cations released from the cor-roded alloy but were not directly in contact with it (indirectstudies), whereas cells growing on the alloy surface wereexposed to the corroded material and in close contact withthe material (direct studies).

Since the corrosion rate of the alloys tested was verydifferent, cytotoxicity analyses were carried out at differenttime-points depending on the rate. For the Pd-free alloy,with a slow corrosion rate, cell adhesion and viability was

FIGURE 13. Comparison of the mechanical properties (compressive

yield stress, ry,C, versus Young’s modulus, E) in (a) logarithmic scale

and (b) linear scale for different families of biodegradable implant

materials, including metallic alloys (Mg-based, Ca-based, Sr-based, or

Fe-based), ceramics (e.g., synthetic hydroxyapatites) and polymers.

Panel (c) compares the dependence of ry,C2/E (which is proportional

to the resilience) as a function of E. Note that BMG stands for ‘‘bulk

metallic glass,’’ whereas WE43 designates the commercial Mg-Y-RE

(RE ¼ rare earth) alloy. The acronyms PLLA, PMMA, PGA, and PCL

designate, respectively, Poly-L-(lactic acid), Polymethylmethacrylate,

Poly(glycolic acid) and Polycaprolactone polymeric materials. Data are

taken from Refs. 4,10,15,33,45,54,55. [Color figure can be viewed in

the online issue, which is available at wileyonlinelibrary.com.]


checked at 8 h of culture, whereas for the Pd-containingsample, with a faster corrosion rate, analyses were carriedout at 8 h but, in addition, at 4 h to be sure to find cells ontop of the alloy before its disintegration.

Regarding indirect studies (cells growing on coverslips),no differences in viability were found when comparing thenumber of live/dead cells at different time-points and in thepresence or absence of the two alloys. Culture density wasalso similar in all cultures analyzed and cell population didnot increase due to the short time in culture. At 4 h of cul-ture nearly all cells were already adhered (osteoblasts needto adhere in order to survive57) both in control culturesand in the presence of the Pd-containing alloy, and at 8 hcells displayed prominent nucleus containing several nucle-oli. The presence of nucleoli is a sign of cellular activitybecause the nucleolus is the site of rRNA transcription andof preribosomal assembly. Thus, cells cultured in the pres-ence of both alloys are able to adhere, resume their nuclearactivity and restore contacts with neighboring cells (see thecellular membrane extensions in Fig. 9). These results are inagreement with those reported by Schmalz et al.20, who didnot observe any negative effect in the viability of fibroblast-keratinocyte cells cultured onto pure Pd during 24 h.

When direct cultures (cells grown onto the alloy) wereanalyzed, significant differences were found in viability andcell adhesion between both alloys and between these andcontrol cultures. The total number of cells observed (8 hculture) on top of the Mg72Zn23Ca5 alloy was lower thanover the coverslip in the presence of the same alloy. More-over, the number of dead cells was higher on the alloy thanon the coverslip. This decrease in the number of cells wasobserved both in fluorescence and SEM images. The Pd-freealloy also presented fewer cells than the corresponding cov-erslip, and the cells were not completely adhered to the sur-face, as both round cells and fusiform cells could beobserved. It has been reported that one of the most impor-tant items to be considered when culturing cells is thehydrophobicity of the surface on which the cells aregrown.58,59 Osteoblast adhesion increases with an increaseof surface wettability. For the Mg72Zn23Ca5 alloy the contactangle is around 87� and it has been shown that it is insuffi-cient to allow a good cell spreading.60 Thus, some cellsremained attached, but not spread, after 8 h of culture andsome probably died because they could not be attached.

These differences are much more dramatic when cellsgrown over Mg72Zn23Ca5Pd2 alloy are analyzed. The numberof cells observed on top of Pd-containing alloy was alreadyscarce at 4 h of culture and most of them were dead. At 8 hthe alloy had disintegrated into pieces and it was very diffi-cult to find a piece with cells faced up. The small fragmentsthat were analyzed presented a high number of dead cells.No fragments with attached cells were found when analyz-ing the alloy by SEM, but that could be due to the protocolused to fix and dehydrate cells, which involves several solu-tion changes. If the contact angle of the Mg72Zn23Ca5 alloywas too hydrophobic for the growth of cells, the contactangle of the Pd-containing alloy was even more hydropho-bic, around 103�, and according to Wei and collaborators60

the adherence is very poor in these surfaces. In fact, the Pd-containing alloy prevents osteoblast adhesion, as few cellsfound on the alloy surface (4 h culture) maintained a roundmorphology and no signs of spreading were observed.Another issue that hinders the adhesion of cells to the sub-strate is the roughness of the sample.58,61 As mentionedbefore, the Pd-containing alloy is more prone to corrosionthan the Pd-free alloy, i.e., after incubating both alloys inHBSS for 2 h the former displays a rough surface withroughness of the order of micrometers (see Fig. 5) whereasthe latter shows a more uniform corrosion. In addition, thePd-containing alloy disintegrates into pieces after 4 h of cul-ture. It has been described that osteoblast adhesion andproliferation is progressively inhibited when roughness(micropore size from 200 nm to 8 lm) is increased.61

According to the same authors, surface discontinuities pro-duced in this case by the track-etched pores could preventcell adhesion. In round-shaped cells stretching is inhibitedbecause no focal adhesions can be formed.

Cell adhesion to the extracellular matrix or to a surfaceis a complex process mediated by transmembrane proteinscalled integrins, which cluster to form focal adhesions. Sev-eral authors have described the induction of anoikis (a formof apoptosis) when cells are prevented to adhere. Forinstance, when human intestinal ephithelial cells are kept insuspension, 90% of the cells activate anoikis after 3 h.62 Onthe other hand, it has been shown that hydrophobic andrough materials, such as the Mg72Zn23Ca5Pd2 alloy, preventintegrin clustering and the formation of focal adhesion com-plexes due to the spatial conformation of the adsorbed mol-ecules in the material.59 Thus, we believe that the increasein dead cells found after 8 h in culture in the presence ofthe Mg70Zn23Ca5Pd2 alloy is not due to alloy toxicity but tothe hindrance of osteoblasts adhesion onto the alloy surface,which may cause osteoblasts to undergo apoptosis after 8 hin culture.

CONCLUSIONS

In this work, a rational framework to understand the mech-anisms responsible for the biodegradation of amorphousMg72Zn23Ca5 and crystalline Mg70Zn23Ca5Pd2 alloys in aphysiological environment is provided. Moreover, apparentinconsistencies between different characterization tests thatmight lead to dissimilar conclusions are outlined and thor-oughly discussed. It is shown that the microstructural char-acteristics of the materials (i.e., amorphous vs. crystalline)bring about differences at various levels (wettability, degra-dation rate, mechanical properties, preosteoblast cell adhe-sion, etc.). Due to the occurrence of micro-galvanic couplesbetween the crystalline phases, the Mg70Zn23Ca5Pd2 alloydegrades too fast in physiological conditions. Conversely,Mg72Zn23Ca5 shows a slower degradation rate and animprovement of the mechanical properties within the firsthours of immersion in the bodily fluid, which makes thisparticular alloy certainly appealing for use as temporaryimplants such as screws, stents and sutures. In any case,none of these two alloys are cytotoxic when preosteoblastcells are cultured in their presence, though cell adhesion

ORIGINAL ARTICLE


onto alloy surfaces is significantly better in the amorphousPd-free alloy due to its lower hydrophobicity. Appropriatesurface plasma treatments could be performed to increasethe hydrophilic character of the Pd-containing alloy. Thepresent article demonstrates that the crystallographic na-ture of two alloys with a very similar composition clearlydetermines their biodegradation and cell compatibility, aresult that can be used as a valuable roadmap in futureworks.

ACKNOWLEDGMENTS

The authors acknowledge O. Castell from the Servei de Micro-sc�opia of the Universitat Aut�onoma de Barcelona for technicalsupport during SEM observations.

REFERENCES1. Lendlein A, Langer R. Biodegradable, elastic shape-memory poly-

mers for potential BIOMEDICAL applications. Science 2002;296:

1673–1676.

2. Yun Y, Dong Z, Lee N, Liu Y, Xue D, Guo X, et al. Revolutionizing

biodegradable metals. Mater Today 2009;12:22–32.

3. Hermawan H, Dub�e D, Mantovani D. Developments in metallic

biodegradable stents. Acta Biomater 2010;6:1693–1697.

4. Hermawan H, Ramdan D, Djuansjah JRP. Biomedical engineer-

ing—From theory to applications. In: Fazel-Rezai R, editor. Metals

for Biomedical Applications, INTECH, InTech: Rijeka, Croatia,

Chapter 17. (ISBN 978-953-307-637-9), August 2011, p 411–430.

5. McBride ED. Absorbable metal in bone surgery. J Am Med Assoc

1938;111:2464–2467.

6. Zhang E, Xu L, Yu G, Pan F, Yang K. In vivo evaluation of biode-

gradable magnesium alloy bone implant in the first 6 months im-

plantation. J Biomed Mater Res A 2009;90:882–893.

7. Witte F, Fischer J, Nellesen J, Crostack HA, Kaese V, Pisch A,

et al. In vitro and in vivo corrosion measurements of magnesium

alloys. Biomaterials 2006;27:1013–1018.

8. Zhang S, Zhang X, Zhao C, Li J, Song Y, Xie C, et al. Research on

an Mg–Zn alloy as a degradable biomaterial. Acta Biomater 2010;

6:626–640.

9. Kraus T, Fischerauer SF, H€anzi AC, Uggowitzer PJ, L€offler JF,

Weinberg A-M. Magnesium alloys for temporary implants in

osteosynthesis: In vivo studies of their degradation and interac-

tion with bone. Acta Biomater 2012;8:1230–1238.

10. Zberg B, Uggowitzer PJ, L€offler JF. MgZnCa glasses without clini-

cally observable hydrogen evolution for biodegradable implants.

Nature Mater 2009;8:887–891.

11. Wang YB, Xie XH, Li HF, Wang XL, Zhao MZ, Wang EW, et al.

Biodegradable CaMgZn bulk metallic glass for potential skeletal

application. Acta Biomater 2011;7:3196–3208.

12. Zhang S, Li J, Song Y, Zhao C, Zhang X, Xie C, et al. In vitro deg-

radation, hemolysis and MC3T3-E1 cell adhesion of biodegrad-

able Mg–Zn alloy. Mater Sci Eng C 2009;29:1907–1912.

13. Matsuda T, Mizutani U, Sato H. Thermoelectric power of Mg-

based simple metallic glasses. J Phys F: Met Phys 1986;16:1005.

14. Wang HX, Guan SK, Wang X, Ren CX, Wang LG. In vitro degrada-

tion and mechanical integrity of Mg–Zn–Ca alloy coated with Ca-

deficient hydroxyapatite by the pulse electrodeposition process.

Acta Biomater 2010;6:1743–1748.

15. Gu X, Zheng Y, Zhong S, Xi T, Wang J, Wang W. Corrosion of,

and cellular responses to Mg–Zn–Ca bulk metallic glasses. Bioma-

terials 2010;31:1093–1103.

16. Miracle DB. A structural model for metallic glasses. Nat Mater

2004;3:697–702.

17. Gonz�alez S, Pellicer E, Fornell J, Blanquer A, Barrios L, Ib�a~nez E,

Solsona P, Suri~nach S, Bar�o M D, Nogu�es C, Sort J. Improved

mechanical performance and delayed corrosion phenomena in

biodegradable Mg-Zn-Ca alloys through Pd-alloying. J Mech

Behav Biomed Mater 2012;6:53–62.

18. Faurschou A, Menn�e T, Johansen JD, Thyssen JP. Metal allergen

of the 21st century—A review on exposure, epidemiology and

clinical manifestations of palladium allergy. Contact Dermatitis

2011;64:185–195.

19. Cristaudo A, Bordignon V, Petrucci F, Caimi S, De Rocco M, Pic-

ardo M, et al. Release of palladium from biomechanical prosthe-

ses in body fluids can induce or support PD-specific IFNgamma T

cell responses and the clinical setting of a palladium hypersensi-

tivity. Int J Immunopathol Pharmacol 2009;22:605–614.

20. Schmalz G, Schuster U, Schweikl H. Influence of metals on IL-6

release in vitro. Biomaterials 1998;19:1689–1694.

21. Syverud M, Dahl JE, Heru H, Morisbak E. Corrosion and biocom-

patibility testing of palladium alloy castings. Dent Mater 2001;17:

7–13.

22. Nguyen TL, Blanquet A, Staiger MP, Dias GJ, Woodfield TBF.

2012. On the role of surface roughness in the corrosion of pure

magnesium in vitro. J Biomed Mater Res Part B. 2012;100B:

1310–1318.

23. Oliver WC, Pharr GM. An improved technique for determining

hardness and elastic modulus using load and displacement sens-

ing indentation experiments. J Mater Res 1992;7:1564–1583.

24. Fischer-Cripps AC. In: Nanoindentation. Ling FF, editor. New York:

Springer; 2004.

25. Jardim PM, Sol�orzano G, Sande V, John B. Precipitate crystal

structure determination in melt spun Mg-1.5wt%Ca-6wt%Zn alloy.

Microsc Microanal 2002;8:487–496.

26. Ishizaki T, Sakamoto M. Facile formation of biomimetic color-

tuned superhydrophobic magnesium alloy with corrosion resist-

ance. Langmuir 2011;27:2375–2381.

27. Lee W, Park BG, Kim DH, Ahn DJ, Park Y, Lee SH, et al. Nano-

structure-dependent water-droplet adhesiveness change in super-

hydrophobic anodic aluminum oxide surfaces: From highly

adhesive to self-cleanable. Langmuir 2010;26:1412–1415.

28. Song XY, Zhai J, Wang YL, Jiang L. Fabrication of superhydro-

phobic surfaces by self-assembly and their water-adhesion prop-

erties. J Phys Chem B 2005;109:4048–4052.

29. Yin B, Fan L, Hu J, Tang A-Q, Wei W-H, He J. Preparation and

properties of super-hydrophobic coating on magnesium alloy.

Appl Surf Sci 2010;257:1666–1671.

30. Wang Y, Tan MJ, Pang J, Wang Z, Jarfors, AWE. In vitro corro-

sion behaviors of Mg67Zn28Ca5 alloy: From amorphous to crystal-

line. Mater Chem Phys 2012;134:1079–1087.

31. Wang DP, Wang SL, Wang, JQ. Relationship between amorphous

structure and corrosion behaviour in a Zr–Ni metallic glass. Corro-

sion Science 2012;59:88–95.

32. Yamamoto A, Hiromoto S. Effect of inorganic salts, amino acids

and proteins on the degradation of pure magnesium in vitro.

Mater Sci Eng C 2009;29:1559–1568.

33. Staiger MP, Pietak AM, Huadmai J, Dias G. Magnesium and its

alloys as orthopedic biomaterials: A review. Biomaterials 2006;27:

1728–1734.

34. Leyland A, Matthews A. On the significance of the H/E ratio in

wear control: A nanocomposite coating approach to optimised tri-

bological behaviour. Wear 2000;246:1–11.

35. Rebholz C, Leyland A, Schneider JM, Voevodin AA, Matthews A.

Structure, hardness and mechanical properties of magnetron-

sputtered titanium–aluminium boride films. Surf Coat Technol

1999;120–121:412–417.

36. Wang YB, Li HF, Zheng YF, Wei SC, Li M. Correlation between

corrosion performance and surface wettability in ZrTiCuNiBe bulk

metallic glasses. Appl Phys Lett 2010;96:251909.

37. Ma GF, Zhang HF, Li H, Hu ZQ. Influence of structural relaxation

on wetting behavior of molten In–Sn alloy on Cu40Zr44Al8Ag8 bulk

metallic glass. J Alloys Compd 2012;513:273–276.

38. Ma GF, Zhang HF, Li H, Hu ZQ. Contact angles and diffusion of

molten Bi–Sn alloy on amorphous and crystalline Fe78B13Si9.

Mater Lett 2009;63:1605–1607.

39. Shen P, Sun JX, Yang J, Qi Y, Jiang QC. Wettability of amor-

phous and nanocrystalline Fe78B13Si9 substrates by molten Sn

and Bi. Nanos Res Lett 2011;6:318.

40. Zhao K, Liu KS, Li JF, Wang WH, Jiang L. Superamphiphobic

CaLi-based bulk metallic glasses. Scripta Mater 2009;60:225–227.

41. Li HF, Wang YB, Cheng Y, Zheng YF. Surface modification of

Ca60Mg15Zn25 bulk metallic glass for slowing down its biodegra-

dation rate in water solution. Mater Lett 2010;64:1462–1464.


42. Virtanen S, Milosev I, Gomez-Barrena E, Trebse R, Salo J, Kontti-

nen YT. Special modes of corrosion under physiological and simu-

lated physiological conditions. Acta Biomater 2008;4:468–476.

43. Kirkland NT, Birbilis N, Staiger MP. Assessing the corrosion of

biodegradable magnesium implants: A critical review of current

methodologies and their limitations. Acta Biomater 2012;8:

925–936.

44. Song G. Recent progress in corrosion and protection of magne-

sium alloys. Adv Eng Mater 2005;7:563–586.

45. Claes LE, Mechanical characterization of biodegradable implants.

Clin Mater 1992;10:41–46.

46. Available at: http://www.webelements.com/

47. Queheillalt DT, Katsumura Y, Wadley HNG. Synthesis of stochas-

tic open cell Ni-based foams. Scripta Mater 2004;50:313–317.

48. Biener J, Hodge AM, Hamza AV, Hsiung LM, Satcher JH. Nanopo-

rous Au: A high yield strength material. J Appl Phys 2005;97:

024301.

49. Bellet D, Lamagnere P, Vincent A, Br�echet Y. Nanoindentation

investigation of the Young’s modulus of porous silicon. J Appl

Phys 1996;80:3772.

50. Xie G, Fukuhara M, Louzguine-Luzgin DV, Inoue A. Ultrasonic

characteristics of porous Zr55Cu30Al10Ni5 bulk metallic glass fabri-

cated by spark plasma sintering. Intermetallics 2010;18:2014–2018.

51. Tancret F, Osterstock F. Modelling the toughness of porous sin-

tered glass beads with various fracture mechanisms. Phil Mag

2003;83:125–136.

52. Pellicer E, Pan�e S, Panagiotopoulou V, Fusco S, Sivaraman KM,

Suri~nach S, et al. Localized electrochemical deposition of porous

Cu-Ni microcolumns: Insights into the growth mechanisms and

the mechanical performance. Int J Electrochem Sci 2012;7:

4014–4029.

53. Fornell J, Concustell A, Suri~nach S, Li WH, Cuadrado N, Gebert A,

Bar�o MD, Sort J, Yielding and intrinsic plasticity of Ti-Zr-Ni-Cu-Be

bulk metallic glass. Int J Plast 2009;25:1540–1559.

54. Sabir MI, Xu X, Li L. A review on biodegradable polymeric materi-

als for bone tissue engineering applications. J Mater Sci 2009;44:

5713–5724.

55. Schinhammer M, H€anzi AC, L€offler JF, Uggowitzer PJ. Design

strategy for biodegradable Fe-based alloys for medical applica-

tions. Acta Biomater 2010;6:1705–1713.

56. Wang YB, Zheng YF, Wei SC, Li M. In vitro study on Zr-based

bulk metallic glasses as potential biomaterials. J Biomed Mater

Res B: Appl Biomater 2011;96:34–46.

57. Triplett JW, Pavalko FM. Disruption of a-actinin-integrin interac-

tions at focal adhesions renders osteoblasts susceptible to apo-

ptosis. Am J Physiol Cell Physiol 2006;291:C909–C921.

58. Chang H-I, Wang Y. Cell responses to surface and architecture of

tissue engineering scaffolds. In: Eberli D, editor. Regenerative

Medicine and Tissue Engineering—Cells and Biomaterials, InTech:

Rijeka, Croatia, chapter 27, INTECH (ISBN 978-953-307-663-8), Au-

gust 2011, p 569–588.

59. Bacakova L, Filova E, Parizek M, Ruml T, Svorcik V. Modulation of

cell adhesion, proliferation and differentiation on materials

designed for body implants. Biotechnol Adv 2011;29:739–767.

60. Wei J, Igarashi T, Okumori N, Igarashi T, Maetani T, Liu B, et al.

Influence of surface wettability on competitive protein adsorption

and initial attachment of osteoblasts. Biomed Mater 2009;4:045002.

61. Lee SJ, Choi JS, Park KS, Khang G, Lee YM, Lee HB. Response of

MG63 osteoblast-like cells onto polycarbonate membrane surfa-

ces with different micropore sizes. Biomaterials 2004;25:

4699–4707.

62. Grossmann J, Walther K, Artinger M, Kiessling S, Sch€olmerich J.

Apoptotic signaling during initiation of detachment-induced apo-

ptosis (‘‘Anoikis’’) of primary human intestinal epithelial cells. Cell

Growth Differ 2001;12:147–155.

ORIGINAL ARTICLE


On the biodegradability, mechanical behavior, and...

Documents

Transcript of On the biodegradability, mechanical behavior, and...