Chunyn Zhng ,Shiyu Zhng ,Dongwei Sun ,Jiji Lin ,Fncheng Meng ,Huinn Liu,*

a School of Materials Science &Engineering,Chongqing University of Technology,No.69 Hongguang Avenue,Chongqing 400054,China

b Department of Bioengineering,and the Materials Science and Engineering Progra m,University of California,Riverside,900 University Avenue,MSE 205,Riverside,CA 92521,USA

Abstract A micro-nano structure CaF2 chemical conversion layer was prepared on fluoride-treate AZ31 alloy,then the composite fluorid conversion fil (CaF2/MgF2) was modifie by stearic acid (SA) and fabricated a superhydrophobic surface.The fluoride-treate magnesium,fluorid conversion fil and superhydrophobic coating were characterized by SEM,EDS,XRD and FTIR.The properties of coatings’ adhesion and corrosion resistance were evaluated via tape test and electrochemical measurement.The cytocompatibility of the MgF2,CaF2 and superhydrophobic CaF2/SA surface was investigated with bone marrow-derived mesenchymal stem cells (BMSCs) by direct culture for 24h.The results showed that the superhydrophobic fluorid conversion coating composed of inner MgF2 layer and the outer CaF2/SA composite layer had an average water contact angle of 152°.SA infiltrate into the micro-nano structure CaF2 layer and formed a strong adhesion with CaF2 layer.Furthermore,the super-hydrophobic coating showed higher barrier properties and corrosion resistance compared with the fluorid conversion fil and fluoride-treate AZ31 alloy.The BMSC adhesion test results demonstrated MgF2 CaF2 and CaF2/SA coatings were all nontoxic to BMSC.At the condition of in direct contact with cells,MgF2 showed higher cell density and enhanced the BMSCs proliferation,while CaF2 and CaF2/SA coating showed no statistically difference in cell density compared with glass reference but the CaF2 and CaF2/SA coating were not conducive to BMSCs adhesion.

Keywords: Biodegradable;Magnesium alloy;Superhydrophobic;Fluoride conversion film Cytocompatibility.

1.Introduction

Magnesium and its alloys have been widely studied as bioabsorbable implant materials due to their excellent performances,such as similar density and elastic modulus to human bones,high strength-weight ratio and good biocompatibility [1,2].However,the rapid corrosion rate and the localized/pitting corrosion behavior degraded their mechanical stability as the implant,which seriously limits the further clinical application of magnesium alloy for orthopedic fixatio accessories [3,4].

In recent years,many surface coatings have been applied to improve the corrosion resistance of Mg and its alloys [5],such as chemical conversion coatings [6],polymer coatings[7],micro arc-oxidation (MAO) [8],calcium phosphate coatings [9] and so on.But up to now,it is still a challenge for present coatings to meet the requirement of gradient degradation for magnesium alloy bone implants.In other words,the degradation rate of magnesium-based implants should be very low during the firs 12-18 weeks to keep the mechanical integrity during the bone tissue healing,and then gradually dissolves and becomes absorbed by the body [1].

Superhydrophobic surfaces,which exhibit special functions as anti-corrosion [10] and anti-bacteria [11] have attracted great attention for their potential medical applications and have been actively studied on their interaction with tissues,cells,biological flui and biological molecular[12].Stearic acid (SA,CH3(CH2)16COOH) is nontoxic and biocompatible [13] and has been applied to magnesium or coated magnesium alloy to gain low surface energy and improve the corrosion resistance [14-18].Wang reported the superhydrophobic Mg(OH)2fil modifie by stearic acid could reduce the toxicity of Mg alloy and resist the adhesion of the bacteria [15].Zhang et al.deposited superhydrophobic calcium stearate coating on plasma electrolytic oxidation(PEO) pretreated magnesium substrate,and the result showed that the superhydrophobic coating reduced the proliferation,but enhances the differentiation of osteoblast and formed hydroxyapatite on its surface [16].Moreover,fluorid treatment has been used on magnesium alloy because the protective fil(MgF2) formed on it is nontoxic to BMSCs,but the MgF2coating can still not meet the application of Mg-based implant material [19-22].

The present study is an attempt to modify fluorid conversion fil with SA to form a superhydrophobic surface on the AZ31 substrate to further improve the corrosion resistance of Mg alloys.However,the results from our previous exploratory experiments indicated that it is difficul to coat the MgF2fil by SA due to the comparatively smooth surface of the fil and different chemical polarity of two materials.Generally,to fabricate a superhydrophobic surface needs two signifi cant requirements,including to make a rough surface with micro-nano structure and to modify the surface by applying a hydrophobic component layer to develop low surface energy[23].To overcome the poor adhesion of SA,a two-step process was applied in this study.The fluoride-treate AZ31 sample was firs immersed in Ca(OH)2solution to form a CaF2layer,which provided a micro-nano structure and a surface containing calcium ions [24] to make it easier for SA to react with Ca ion and form a chemical bond.Then the composite fluorid conversion fil (CaF2/MgF2) was modifie with SA and fabricated a superhydrophobic surface.Moreover,the properties of surface wettability,corrosion resistance and BMSCs adhesion of the coatings obtained under each preparation step were studied and discussed in this paper.

2.Experimental procedures

2.1.Material preparation

AZ31 magnesium alloy disks of 10mm diameter and 1mm thickness were polished with SiC abrasive paper up to grade #1200,ultrasonically cleaned for 10min in acetone and ethanol separately.Then the blank AZ31 disks were immersed in hydrofluori acids (40wt%) for 5 days to prepare the fluorid passivation film

2.2.Preparation of the superhydrophobic fluorid conversion coating

A two-step process was used to prepare the superhydrophobic coating on the fluoride-treate AZ31 alloy.Firstly,the fluoride-treate samples were soaked in a saturated Ca(OH)2solution for 24h to get a CaF2chemical conversion layer with a micro-nano structure.Subsequently,the AZ31 samples with fluorid conversion layer were placed separately into Teflo containers with 0.05mol/L Stearic acid (SA,(CH3(CH2)16COOH)) ethanol solution for 4h at 70°C to form a low energy surface.After modificatio with SA,the samples were taken out and rinsed with ethanol and dried at room temperature.

2.3.Characterization

The morphologies of samples after each step of surface modificatio were characterized by scanning electron microscopy (SEM,ΣIGMA HDTM).X-ray diffraction meter(XRD,Empyrean DX-2500),fourier transform infrared spectra (FT-IR,NicoletIs10) and energy dispersive X-ray spectroscopy (EDS,Oxford Isis) were used to analyze the crystal structure,organic group and chemical composition of the samples modifie by each step,respectively.Surface wettability of the samples were examined by measuring the contact angle in Hank’s solution (a simulated body fluid using a drop shape analyser (DropMeter A-100p) at ambient temperature.The composition of Hank’s solution:8g/l NaCl,0.4g/l KCl,0.14g/l CaCl2,0.35g/l NaHCO3,1.0g/l C6H6O6,0.1g/l MgCl2·6H2O,0.06g/l MgSO4·7H2O,0.06g/l KH2PO4and 0.06g/l Na2HPO4·12H2O,with a pH of 7.4.

2.4.Electrochemical measurement

In order to evaluate the corrosion resistance properties of coatings,electrochemical measurements (Gamy3000) were conducted in Hank’s solution using traditional three-electrode cell,coated specimens as a working electrod,platinum as the counter electrode and saturated calomel electrode (SCE)as a reference electrode.The specimen was firs immersed in Hank’s solution for 20min to acquire a stable open circuit potential (OCP).Then electrochemical impedance spectroscopy (EIS) was measured in the frequency range from 100kHz to 0.01Hz under 10mV amplitude of the perturbation signal.The potentiodynamic polarization tests was followed at the scan rate of 0.5mVs-1after EIS measurement.The EIS results were fitte and analyzed by Gamry Echem Analyst software.The corrosion currents were calculated by Tafel extrapolating.

2.5.Coating adhesion test

The coating’s adhesion to the substrate was qualitatively evaluated by tape test according to GB/T 9286-1998.In the tape test,the coating was cross engraved into a lattice pattern of 1mm×1mm.Then a pressure-sensitive adhesive tape was attached to the coatings.Finally,both ends of the tape were removed at the same time in the direction perpendicular to the sample surfaces.The adhesion strength was assessed by comparison with descriptions and illustrations in the Standard.

2.6.BMSC adhesion test

2.6.1.Direct culture of BMSCs

BMSCs were harvested from the marrow cavity of the femur and tibia of rat and cultured in Dulbecco’s Modifie Eagle’s medium (DMEM) according to the procedures published in reference [25].Before cell seeding,AZ31 alloy substrate,coated samples as well as the glass reference (Fisher Scientific 1mm thick,10mm in diameter) were placed in standard 12-well cell culture plate and disinfected under ultraviolet (UV) radiation for 1h on each side.Then BMSCs(P2) were seeded onto the disinfected samples at a density of 1×104cells·cm-2and incubated in 3mL DMEM under cell culture conditions ((37°C,5%/95% CO2/air,humidified sterile environment) for 24h.BMSCs cultured in DMEM and DMEM without any samples served as the cell-only reference group and the media-only reference.After a 24-hours cell culture,the media was extracted and measured pH value immediately by a meter (Symphony TM,Model SB70P,VWR).The samples were transferred to another disinfected 12-well plate to fi and dye the cells on the sample (direct contact)and the plate (indirect contact) separately.

2.6.2.Adhesion and morphology of BMSC

After 24-h BMSC culture,BMSCs attached on the samples and plates were fi ed with 4%formaldehyde for 20 min.Then Alexa Flour 488 for F-actin and DAPI were used to stain the BMSCs fi ed on the surface of sample.BMSCs attached to each sample (direct contact with the sample) and each plate(indirect contact surrounded each sample) were imaged via a fluorescenc microscope (Eclipse Ti,Nikon).At least six images of random field were taken to ensure the reliability of the cell statistical results.Image J analysis software was applied to quantify the Cell Density (the number of adherent cell per square centimetre),Spreading Area per Cell (the area of cell outline for each BMSC) and Aspect Ratio (the ratio of the length of the major axis and the length of the minor axis for the elliptical fittin of a cell outline).

The individual cell morphology and the element distributions on the sample surface were characterized by SEM and EDS.After 24-h BMSC culture,the BMSCs on the sample were fi ed by 10% glutaraldehyde in phosphate-buffered saline solution for 2h.Then BMSCs were dehydrated by dipping in a series of ethanol solutions,from 30%,50%,60%,70%,80%,90%,95% to 100% for each dipping with 15min.The dehydration process in 100% ethanol was repeated twice.Finally,the samples were sputter-coated with Pt/Pd for 90s for the SEM analysing.

2.6.3.Statistical analysis

Fig.1.XRD patterns of AZ31 substrate (a) and coated AZ31 samples (b.MgF2,c.MgF2/CaF2,d.CaF2/SA).

All numerical data used in this study were obtained from experiments run in triplicate and examined using a one-way analysis of variance followed by the Tukey post-hoc test for detecting statistical differences.When comparing two different groups,p<0.05 was considered to demonstrate statistical differences,p <0.01 was considered that the two group has obviously statistical difference andp <0.001 was considered that the difference was very significantl,symbolized as*,**,***in diagram,respectively.

3.Results

3.1.Microstructure and composition of the coatings

The XRD patterns of the AZ31 alloy,fluoride-treated flu oride and calcificatio treated AZ31 alloy and stearic acidmodifie samples afterward are shown in Fig.1.The characteristic peaks of MgO and MgF2were observed in the fluoride-treate sample.The weak intensities and broad line widths of diffraction peaks appeared at 2θ=27.2°,40.4° and 53.5° (PDF Number:35-0816) indicated that MgF2had an amorphous structure.After calcificatio treatment,the MgF2peaks disappeared and CaF2peaks at 2θ=27.2°,40.4° and 53.5° (PDF Number:70-0212) were observed.The weak intensities of the diffraction peaks demonstrated that the crystallinity of CaF2was not high.After being modifie with SA,some new peaks appeared at 2θ=8.52°,25.9° and 32.6°.According to the references [16,26,27],the peaks were from the reflectio of CaSA.

Fig.2 shows the FT-IR spectra of the SA modifie fluo ride conversion coated sample as well as the SA.The absorption peaks between 1190 cm-1and 1350 cm-1correspond to the carboxylic acid groups of long-chain fats.The bands at 1430 cm-1are due to the presence of C-O.The symmetric stretching vibration peaks of carboxylate and C=O of stearic acid were at 1470 cm-1and 1700 cm-1,respectively.The bands at 2850 cm-1are attributed to C-H symmetric stretching vibrations (-CH2-),while the bands at 2920 cm-1and 2960 cm-1are due to the C-H asymmetric stretching vibrations (-CH3-) [28,29].After modification the band of -COO from SA modifie coating appears at 1470 cm-1and 1540 cm-1[13].And two adsorption peaks at about 3690 cm-1and 3440 cm-1are from O-H and H2O stretching vibration.The changes of the infrared absorption peak of stearic acid reflec the activation of decomposition and grafting behavior.

Fig.2.FT-IR spectra of stearic acid and the stearic acid modifie CaF2 coating.

Fig.3 shows the surface SEM morphology,cross-section SEM morphology and the corresponding EDS analysis of coating.After fluorid treatment,a dense MgF2layer of about 2.5μm thick was formed on the surface of AZ31 alloy as shown in Fig.3a.Fig.3b displays that fla e crystals composed of Ca,F and O was formed on MgF2 fil after further calcificatio treatment.Most of the crystals grew almost perpendicular to the surface.The fla e crystals are a few microns long and a few dozens of nanometers thick,which formed a micro-nano scale structure on surface.The sphere-shape deposits on the surface contained more Ca and O than the fla e structure.Combine with the results of XRD,the following chemical reactions occurred on surface of MgF2:

The CaF2layer was composed of fla y crystals with about 1 um long and dozens of nanometers thick that interlaced nearly perpendicular to the surface (Fig.3-b1).However,only the outer surface of the MgF2fil reacted with Ca(OH)2.Therefore,the fluorid conversion coating (MgF2/CaF2) was composed of two layers,the upper CaF2layer with a thickness of about 1.5μm and the lower MgF2layer about 2.0μm in thick,as shown in Fig.3-b2 and Fig.3-b3.After modifi cation by stearic acid,in addition to the elements of Ca and F,41.86 at% C,and 5.98 at% O were detected on the sample surface (Table 1).The fla e-shape structure was covered by SA but the sphere-shape structure could still be observed(Fig.3-c1).The cross-section image in Fig.3-c2 and composition distribution diagram in Fig.3-c3 show that element C was distributed within a range of 1.5mm on the surface(in the range of CaF2layer) and no element C was detectedin the range of MgF2film indicating the infiltratio of SA to CaF2layer during stearic acid treatment but no infiltratio into the MgF2layer.Combine with the results of XRD and FT-IR,the stearic acid reacted Ca2+ions and formed CaSA[30].Therefore,the SA modifie fluorid conversion composite coating was composed of two layers,the inner MgF2layer and the outer CaF2/SA composite layer.

Table 1 EDS results of the marked areas in Fig.3,at%.

3.2.properties of the coatings

3.2.1.Surface wettability

Fig.4 shows the images of water droplets and contact angle values of the samples.The surface of the AZ31 and MgF2coated AZ31 alloy are hydrophilic and the values of contact angle are about 39.25° and 49.75°,respectively.After the calcificatio treatment,the surface of CaF2coating with a micro-nano structure became much more hydrophilic and the value of a contact angle decreased to 9.85°.However,after further modifie with SA,a water droplet on the coating surface forms almost a round sphere with a high contact angle of 152° (Fig.4(d)).Moreover,the water droplet barely sticks to the surface and rolls off easily,which demonstrated that the fluorid conversion composite coating (MgF2/CaF2) on AZ31 alloy acquired a superhydrophobic property by using SA.

The Superhydrophobicity of CaF2could be explained by the Cassie-Baxter equation [23]:

Whereθandθ0are the contact angle for the rough surface and the smooth surface,respectively,andf sis the area fraction of solid at the droplet interface.The increases of microscopic roughness of surface make the hydrophilic surface more hydrophilic and the hydrophobic surface more hydrophobic.Therefore,the superhydrophobic property is due to the combination of a micro-nano structure of CaF2coating and the hydrophobic tail on the surface of the coating achieved by SA modification

3.2.2.Coatings adhesion

Fig.5a1-c1 shows after the tape test the whole MgF2coating almost remained on the AZ31 subtrates and a small amount of coating exfoliated at the edge of the cut of CaF2and CaF2/ SA coated samples.The EDS results display that there were more F and less Ca at the position of coating’s peeling (Spectrum 5 and Spectrum 6),indicating the exfoliation position of CaF2/MgF2composite coating was on the interface of CaF2and MgF2instead of the interface of MgF2and substrate.For the CaF2/ SA coated AZ31 sample,the position of the coating’s peeling (Spectrum 9 and Spectrum 10)were mainly composed of element Mg,F,and C,demonstrating SA and CaF2exfoliated together as a whole,which indicated the adhesion between SA and CaF2was fairly strong.

Fig.3.Surface SEM morphology (a1-c1),cross-section SEM morphology (a2-c2) and the corresponding EDS analysis of coating cross-section (a3-c3) of the coated AZ31 samples (a.MgF2,b.MgF2/CaF2,c.CaF2/SA).

3.2.3.Corrosion properties

Fig.6 reveals the potentiodynamic polarization curves of samples in Hank’s solution.The corrosion current density icorris obtained by Tafel extrapolation of the cathodic branch of the polarization curve,and icorris related to the average corrosion rate [31].

The electrochemical parameters are presented in Table 2.Corrosion current density represents that once the coating is damaged by Ebk,the electrochemical corrosion occurs.The value of Icorrindicates the corrosion rate of the Mg substrate.Ecorrand Ebkof the MgF2coated AZ31 alloy was increased significantl by the superhydrophobic compositecoating from -1.48V to -1.40V and from -0.95V to-0.21V,respectively.The corrosion rates (Pi) of the MgF2coated (15.19mm/y) and MgF2/CaF2coated (13.61mm/y)AZ31 are close to each other and nearly 7 times lower than those of AZ31 alloy (97.11mm/y).The sample with superhydrophobic CaF2/SA coating shows the lowest icorrand Pi(0.94mm/y) nearly 100 times lower than that of AZ31 alloy.The results above demonstrate that fluorin treatment reduced the corrosion tendency and corrosion rate of AZ31 magnesium alloy.And the corrosion property of fluorid conversion fil was further improved by SA modificationFig.7 displays the electrochemical impedance spectrum(EIS) of the AZ31 and coated AZ31 alloy samples in Hank’s solution.According to the number and position of the Phase Angle peak,as well as the diameter of the capacitive or inductive arc,the corresponding circuit elements could be estimated to match the coating structure,and the appropriate equivalent circuit diagram could be selected for the simulation of the electrode reaction process,as shown in Fig.7d.The data fit ting results are listed in Table 3.Parameters related to the coating corrosion resistance mainly include coating resistance(Rf) and constant phase element of coating (Q1),reaction resistance (Rct) and constant phase element of electric doublelayer (Q2),and inductance (L).

Table 2 Parameters of the untreated and surface modifie AZ31 alloy samples by fittin polarization curves.

Table 3 EIS date according to the fittin of equivalent circuits.

Fig.4.Contact angle of AZ31 substrate (a) and coated samples (b.MgF2,c.MgF2/CaF2,d.CaF2/SA) in Hank’s solution.

Fig.5.The macroscopic images (a1-c1),SEM images (a2-c2) and composition distribution (a3-c3) of the coated samples after tape test.(a.MgF2,MgF2/CaF2,c.CaF2/ SA).

Fig.6.The Potentiodynamic polarization curves of AZ31 and coated AZ31 samples in Hank’s solution.

Fig.7.EIS plots of AZ31 and coated AZ31 alloy samples in Hank’s solution (a.Bode -Phase,b.Bode -Mod,c.EIS -Nyquist and d.Equivalent circuit for the simulated electrode process (d1.R(QR)(QRL) (AZ31 alloy),d2.R(QR)(QR) (MgF2 coated,CaF2/ SA coated),d3.R(QR)(QR(QR))W (MgF2/CaF2 coated)).

Fig.8.Fluorescence images of BMSC on the sample surface (direct contact with the sample) (a1-f1),on the culture plate surrounding each corresponding sample (indirect contact with the sample) (a2-e2) after in vitro culture for 24h.a.AZ31 control,b.c.d.MgF2,CaF2 and CaF2/SA coated AZ31,e.Glass reference,f.BMSCs reference.

Due to the existence of passivation fil (MgO),just as the MgF2fil on AZ31 alloy,the EIS plots of the AZ31 also exhibits two Phase Angular peaks (Fig.7a) and two capacitance loops (Fig.7c) at high and medium frequencies separately,corresponding to the information of fil (Rfand Q1) and interface between fil and substrate (Rctand Q2),respectively.The existence of inductive arc at low frequencies in Nyquist plots of AZ31 alloy (Fig.7c) indicates that the AZ31 sample entered the induction period of pitting,but no pitting corrosion were formed.The Bode-Phase diagram of the MgF2/CaF2coated sample has three Phase Angle peaks and displays the phenomenon of“Weber impedance”in Nyquist plots which might be caused by the difference concentration in solution due to the porous structure of CaF2layer.The“weber impedance”is represented by the equivalent circuit element“W”,shown as Fig.7-d3.Rpand Q3represent the porous resistance and constant phase element of the pore layer of the coating,respectively.The superhydrophobic CaF2/SA composite coated samples presents one phase peak at the medium frequencies and the low frequencies respectively (Fig.7a),corresponding to the information of composite coating (Rfand Q1) and interface between coating and substrates (Rctand Q2),respectively.The equivalent circuit was shown in Fig.7-d2.Table 3 shows Rfand Rctof the CaF2/MgF2coated sample significantl elevated after SA modification demonstrating the effectiveness of SA in improving the corrosion resistance of magnesium alloy.

The Nyquist plots show that the radius of the capacitive reactance arcs was:AZ31 magnesium alloy

3.3.Cytocompatibility in direct culture

3.3.1.Fluorescence images in direct and indirect contact

The fluorescenc microscopy images (Fig.8) illustrate that more BMSCs attached on MgF2coating,and the BMSCs attached displayed the typical morphology for proliferating(spindle-like morphology) [32].BMSCs attached on the glass substrate and on plate surrounded the glass presented polygonal cell shape which is typical for non-proliferating cell morphology [32].Further,most of the BMSCs restrained and the cell spheroids could be observed on the surface of the CaF2and CaF2/SA coating,which leading to disabled cell mobility.This indicates that the BMSCs adhered to CaF2and CaF2/SA coating were inhibited.Under the indirect contact conditions,the representative fluorescenc images show that the BMSCs around CaF2coating had a less density than other groups.Cells surrounding the CaF2/SA superhydrophobic composite coating appeared no difference from other groups.

3.3.2.SEM morphology and EDS analysis of individual BSMC in direct contact

Fig.9 shows the morphologies of Cellular pseudopod and individual BMSC on the samples,the overlaid images of SEM and EDS for each group,respective elemental maps and the composition distributions charts on the surface.Fig.9a1-d1 displays that the pseudopods of BMSC on the surface of AZ31 alloy were like stubby fingers and the pseudopods of BMSC on MgF2coating cross over each other,making a branching network,which indicated that BMSCs could adhere to the AZ31 alloy and MgF2coating surface well and could move fl xibly to complete their functions.However,the pseudopods of BMSC on CaF2and CaF2/SA coatings were sparse and thin,demonstrating unfavorable for cell adhesion.As demonstrated in Fig.9a2-d2,a BMSC attached on AZ31 alloy spread at multiple directions and demonstrated a spreading morphology;the cell on MgF2coating exhibited a more spreading morphology.By contrast,the BMSC on the surface of CaF2and CaF2/SA coating had a more round and less spreading morphology.The overlap of the locations of BMSCs in the SEM images with the regions of intensifie C element distribution in the EDS maps indicated that the C element came from the cells.Except the element C from cell and Mg from the substrate,a large amount of O and P (6.1 at%)and a small amount of Ca (1.2 at%) were detected on the surface of AZ31 alloy.It was reported that at the initial stage of corrosion,it is easy for Mg2+to bind withand then inducing and controlling the formation of calcium phosphate [33].Therefore,the spherical-shape sediment on the surface of magnesium alloy might be the deposition of calcium phosphate.Some cracks were also observed on the surface of the coated samples.The cracks on the coating sprawled gradually during the SEM experiment,so the cracks might be caused by dehydration in the vacuum process and electron beam bombardment.In addition to carbon from the cells and the elements from the coatings,only a small amount of element P was detected on the surfaces of the three coated samples compared to that on the AZ31 alloy.Moreover,4.8 at% Mg were detected on the surface of CaF2/SA coating.Fig.9-d3 shows that Mg was mainly distributed at the cracks of the coating.According to the results of the adhesion test,the deep crack on the surface of the coating resulted in the exposing of MgF2coating instead of Mg substrate.

Fig.9.Surface Cell SEM morphologies and elemental distribution of (a) AZ31 control,(b) MgF2 coated,(c) CaF2 coated and (d) CaF2/SA coated AZ31 alloys with the direct-contact BMSCs.SEM images of Cellular pseudopod morphologies and Cells morphologies are shown in a1-d1) and row 2 (a2-d2);and the overlaid SEM and EDS maps (a3-d3).The distributions of carbon,magnesium,fluorine calcium and phosphate are also shown in separate columns.Composition distribution charts (at%) are shown in row 7.(a7-d7).

Fig.10.The quantitative analysis after BMSC culture for 24h with the coated AZ31 alloys (MgF2,CaF2 and CaF2/SA),controls (AZ31),and references(DMEM,BMSC and glass).(a) adhesion density of BMSCs,(b) spreading area per BMSC,(c) aspect ratio of BMSCs on the sample surface (direct contact)and on the culture plate surrounding each corresponding sample (indirect contact).Values are mean ± standard deviation (n=3).*p < 0.05,**p < 0.01,***p< 0.001.

3.3.3.Cell Adhesion Density and Spreading Area per BMSC

Fig.10a displays the cell adhesion density on samples(direct) and plates (indirect).Among all groups,there was no significan difference in the cell density under direct and indirect for the same group of samples.MgF2coating in the direct culture demonstrated higher cell density than glass reference.CaF2and CaF2/SA coating displayed lower cell density than the MgF2coated sample.Statistically,in direct contact,MgF2coating showed higher cell density,while the CaF2and CaF2/SA coating groups showed no difference to glass reference.Moreover,AZ31 control and MgF2coated AZ31 of indirect culture showed higher cell density than CaF2coating and glass reference.Fig.10b displays the spreading area per BMSC in the direct and indirect contact conditions.Each group in direct contact showed a small spreading area per BMSC comparison with indirect contact condition.Under direct contact condition,the spreading area per BMSC of CaF2and CaF2/SA coating groups were statistically smaller than other groups.Under the indirect contact conditions,the groups of AZ31 alloy and MgF2coating showed statistically larger area per BMSC than BMSC groups.However,the average spreading area per BMSC around the CaF2and CaF2/SA coating has no statistically significan difference by comparison with glass and BMSC reference groups.Fig.10c displays the Aspect Ratio of BMSCs.CaF2/SA coating group presents a statistically lower Aspect Ratio than glass reference in the direct contact condition.No statistically significan difference in the Aspect Ratio was found among the AZ31 control and other three coated AZ31 groups.

3.3.4.The change of pH,Mg2+and Ca2+concentration in the post culture media

Fig.11a displays the pH values of the media after BMSCs were cultured for 24 h.After 24h cultures,all the coated and non-coated AZ31 alloys demonstrated more alkaline pH compared with the BMSC cell control,DMEM medium reference and glass reference.The pH values of coated magnesium alloy were higher than that of reference groups (DMEM,glass and BMSC group).Specificall,the pH of the culture media with non-coated AZ31 (8.13±0.01) was statistically higher(p<0.05) than that of DMEM (7.81±0.01).The pH value of three coating groups has no statistically significan difference compared with non-coated AZ31.Fig.11b shows Mg2+ion concentrations in the media after BMSCs culture.The original Mg2+ionic concentration in DMEM is 0.81mM.There was no significan change of Mg2+ionic concentrations in the collected media of DMEM,BMSC,and glass reference.As expected,a statistically significan higher Mg2+ion concentration was presented in the culture media of AZ31 controls than those of glass,BMSC and DMEM references.Specifi cally,all coated AZ31 alloys groups showed lower Mg2+ion concentration in their culture media.Although Mg2+ion concentrations in culture media of the MgF2group were higher than that of CaF2and CaF2/SA coating groups,no statistically significan difference was found among the coating groups.Fig.11c shows Ca2+ion concentrations in the collected media after BMSCs were cultured.The original Ca2+ionic concentration in DMEM is 1.8mM.Ca2+ionic concentrations in culture media of DMEM,BMSCs and glass reference also changed little.The AZ31 controls displayed lower Ca2+ion concentrations in their respective culture media than other references and coated groups,but no statistically significan difference was found among the coated groups and controlled AZ31.

4.Discussion

4.1.Effects of coatings on degradation of AZ31 alloy

The measurement of the polarization curve is the accelerated corrosion process under the applied voltage.The measurement of EIS is the assessment of the shielding performance of the coatings.Usually,the value of Rctand Zmodrepresents the insulation state of the coating.The results of electrochemical test showed that the corrosion resistance property of magnesium alloy was improved gradually after every step of coating preparation.

The significantl higher pH value and Mg2+ion concentrations in the culture media of AZ31 alloy than those of the references groups (Fig.11) was obviously resulted from the degradation of Mg alloys which reacts with water to form OH-ions.Furthermore,AZ31 alloy group also showed a statistically lower average Ca2+ion concentration in the culture media than those of the reference groups.It was reported that Mg2+ion could induce Ca-containing mineral deposition in alkaline environment [25].Therefore,the significantl lower Ca2+ion concentrations might be caused by Mg-induced Ca deposition.As compared with the non-coated AZ31 group,the three coating groups showed lower pH and statistically lower Mg2+ion concentration in their post-culture media,which indicated that the coatings enhanced the corrosion resistance of AZ31 alloy.Among the MgF2,MgF2/CaF2and CaF2/SA coating samples,MgF2coating samples showed the highest Mg2+ion concentration,but no significan difference of the pH value,likely because of the release of the Mg2+ions of MgF2coating into the media.The Mg2+and Ca2+ion concentration in the culture media of MgF2/CaF2and CaF2/SA coating groups changed little compared with the BMSC reference group,which indicated that the release of the Ca2+ions likely from CaF2and CaF2/SA coating and the Mg2+ions from Mg substrate into the media could be ignored.Therefore,it could be concluded that the solubility of the CaF2and CaF2/SA coating is very low in the culture media.

From the point of solubility product (Ksp),Kspof CaSA,CaF2,and MgF2are 2×10-20,4.0×10-11and 6.4×10-9(25°C),respectively [34,35].Ksp(MgF2)>Ksp(CaF2)>Ksp(CaSA),Theoretically,during the immersion,the dissolution rate of the coating increases gradually from outside to inside (CaSA,CaF2and MgF2).The superhydrophobic composite coated AZ31 may show a tendency for gradient degradation.However,due to solvent swelling and defects of the coating,exfoliation of coating and local corrosion may occur during the immersion.Therefore,thein vitroimmersion experiment is needed to further study the degradation behavior of the coating,and relevant experiments and data compilation are still in progress.

Fig.11.Post-culture media analyses after BMSC culture for 24h with the coated AZ31 alloys (MgF2,CaF2 and CaF2/SA),controls (AZ31),and references(DMEM,BMSC and glass).(a) pH values,(b) Mg2+ ion concentration,(c) Ca2+ ion concentration,Values are mean ± standard deviation (n=3).*p < 0.05,**p < 0.01,***p < 0.001.

4.2.Possible factors affecting BMSC adhesion and spreading

4.2.1.pH value

The media pH increase might affect BMSC adhesion and morphology.In this study,the average pH values of the culture media for the three coating groups were similar (from 8.02 to 8.06).However BMSC on the surface of MgF2coating had much better cell adhesion,highest cell density and largest area per cell than those on the MgF2/CaF2and CaF2/SA coatings (Fig.10a and b).Cipriano et al.reported that an increase of initial media pH to 9 did not affect BMSC adhesion and viability [36].Thus the influenc of pH on the adhesion and spreading of BMSC for both direct contact and indirect contact in this study should be minimal.

4.2.2.Media composition

Media composition,such as the ions released from the degradation of the samples,could also affect the cell morphology and adhesion on the surfaces of samples (under direct contact condition) but might play a more important role around the samples (under indirect contact condition) [37].In this study,the Cell density and Area per BMSC around AZ31 alloy under indirect contact condition was higher than that of glass and BMSC reference groups.Moreover,AZ31 alloy group showed a statistically higher Mg2+ion concentration and lower average Ca2+ion concentration in postculture media than those of other reference groups.It was reported that osteogenic differentiation could be stimulated,when magnesium ion concentration reached 50ppm [37] and the extracellular matrix (ECM) mineralization of BMSCs was enhanced when the concentration of Mg2+was 5mM [38].Furthermore,when the Ca2+concentration exceeds 1mM,the cell adhesion could be inhibited [25].Therefore,the moderate Mg2+ion concentration (3.67mM,88ppm) and the decrease of Ca2+ion concentration in the culture media caused by Mginduced Ca deposition could benefi BMSCs adhesion on the surface of AZ31 alloys.

The MgF2coating samples also showed higher Cell density and Area per BMSC under indirect condition than that of glass and BMSC reference groups.Moreover,the cell adhesion density on MgF2coating under direct contact was over 10,000 cell/cm2and significantl higher than that of glass and BMSC,which indicated that the BMSCs on MgF2coating proliferated during the 24-hour direct culture.Fluoride had biphasic effects.At a low-dose,fluorid could stimulate the proliferation and differentiation of osteoblasts [39,40].Yu et al.reported the MgF2coated AZ31 scaffolds improved the proliferation and attachment of rat bone marrow stromal cells(rBMSCs) [41].Mg2+ion concentration in the culture media of MgF2coating group was higher than that of reference caused by the release of the Mg2+ions from MgF2coating into the media.So there should be low levelF-ions in the media which might benefi the higher cell adhesion density and the spreading cell morphology.

In contrast,Cell density and Area per BMSC of the CaF2coating under indirect culture condition were close to that of glass and BMSC reference groups (Fig.10).Compared with the reference group,Ca2+ion concentration in the postculture media of CaF2coating group barely changed indicated that the CaF2coating hardly dissolved during the immersion in culture media for 24h.SoF-ions in the solution could be ignored,which might explain the cell density and Area per BMSC of the CaF2coating group was lower than that of MgF2coating group but similar to that of reference groups under indirect culture condition.

On the whole,under indirect contact condition,there was no significan difference of cell aspect ratio among the references groups (BMSC and glass),the control group (AZ31)and all the coating groups (Fig.10c).The cell density of AZ31 and three coating groups were higher than those of glass reference and no statistically difference to the BMSC reference.Furthermore,the area per BMSC of AZ31 alloy and MgF2coating groups were higher than those of BMSC reference group but no statistically difference among BMSC,CaF2and CaF2/SA coating groups.The results under indirect contact condition above indicated that the MgF2,CaF2and CaF2/SA coating were nontoxic to BMSC.

4.2.3.Surface topography

Compared with the effects of composition change in media caused by Mg degradation and salt precipitation,the effects of surface chemistry and topography might be more remarkable for direct adhesion of BMSCs on the sample surface[42].The cell density on the surface of AZ31 alloy under direct contact condition was higher than that of glass and BMSC control groups,which could be contributed to the oxide-containing degradation layer on AZ31 plates [43].

In comparison with non-coated AZ31 alloy,the MgF2,CaF2,and CaF2/SA coatings changed surface chemistry,microstructure and wettability,which directly affected BMSCs attached to the surface.The significantl high cell adhesion density and area per BMSC of MgF2coating might also benefi from then fluorid ions in the media.Although the cell adhesion densities on CaF2and CaF2/SA coating were significantl lower than that on the MgF2coating,there was no statistically difference among the glass reference group,AZ31 control group,CaF2coating and CaF2/SA coating.However,the area per BMSC showed significantl statistically difference between the glass reference group and CaF2coating and CaF2/SA coating groups.The BMSC on CaF2presented elongated and poorly spreading cell morphology (Fig.9).The broaden diffraction peaks of MgF2and the low Peak Strength Ratio of the highest Diffraction Peaks of MgF2and Mg shown in Fig.1 demonstrated that crystallinity of MgF2was low and presented an amorphous state.While the results of XRD show that CaF2has a good crystallinity.It was reported that more crystalline morphology of coatings could lead to reduced cell activities [44].Furthermore,the crystals on CaF2coating exhibited a micro-nano scaled fla e structure.Some studies indicated that the micro-or nano-structure could accelerate osteogenic differentiation of BMSCs [45-48].However,a research results by Meirelles showed that the nanostructure alone might not be the optimal structure to induce bone integration [49].The micro-and nano-structure have a different effect on stimulating bone regeneration [50].Washburn reported that cells are significantl sensitive to changes of topography and roughness in nanometer-scale [48].The fla e-like CaF2crystals grew almost vertical to the surface and interlaced.Maybe the sharp topography on CaF2coating inhibited the adherent and proliferation of cells.Moreover,the BMSC on the CaF2/SA coating presented a round cell morphology and ever smaller average area per cell than that on CaF2coating.The reason may be that the CaF2/SA coating is hydrophobic and has no recognition sites for BMSC attachment,which prevented the cell-sample interactions and cellular responses [51].

It was reported that surface topography,surface charges,crystalline morphology,and functional groups could impact cell adhesion and cell spreading on the surface of materials[52].How these characteristics worked together and influ enced the BMSCs adhesion and spreading on the surface of CaF2and CaF2/SA coatings would require further research.

5.Conclusion

The superhydrophobic fluorid conversion coating composed of inner MgF2layer and outer CaF2/SA composite layer was fabricated by chemical conversion and stearic acid modification The SA diffused into the micro-nano structure of the CaF2coating and formed CaSA.There was a strong adhesion of SA and CaF2.The superhydrophobic CaF2/SA coating could lower the corrosion tendency,postponed the corrosion process and lower the corrosion rate of MgF2coated AZ31 alloy.

Under indirect contact condition,the cell density on MgF2,CaF2and CaF2/SA coating were higher than those of glass reference and no statistically difference to the BMSC reference.The area per BMSC on MgF2coating was statistically higher than those of BMSC group and no statistically difference of CaF2coating and CaF2/SA coating groups compared with BMSC group,demonstrating that the MgF2,CaF2and CaF2/SA coating were nontoxic to BMSC.

Under direct contact condition,MgF2coating showed significantl higher cell density than glass and BMSC,implying that the MgF2coating benefite to the BMSCs proliferation during the 24-hour direct culture.The cell density of CaF2coating and CaF2/SA coating groups showed no statistical difference to glass reference in direct contact.However,the cells on CaF2and CaF2/SA coatings showed shrunken morphology and presented statistically smaller spreading area per BMSC than glass and BMSC reference groups,indicating that the CaF2and CaF2/SA coatings are not conducive to cell adhesion.

Declaration of Competing Interest

None.

Acknowledgements

This work was supported by the National Natural Science Foundation of China [Grant No.51201192] and Natural Science Foundation of Chongqing [Grant No.cstc2018jcyjA2285].