*Corresponding author. Fax: 852-2358-1543.

Biomaterials 21 (2000) 1339}1343

Chemical gradient in plasma-sprayed HA coatings

Jing Wen!,*, Yang Leng", Jiyong Chen!, Chenge Zhang"!Engineering Center of Biomaterials Research, Sichuan University, Chengdu, People+s Republic of China

"Department of Mechanical Engineering, Hong Kong University of Science and Technology, Hong Kong, People+s Republic of China

Received 10 December 1998; accepted 30 November 1999

Abstract

The microstructure and inhomogeneous features of plasma-sprayed hydroxyapatite coatings on titanium substrates have beenexamined using the time-of-#ight secondary ion mass spectroscopy (ToF-SIMS), micro-Raman spectroscopy and nano-indentationtechniques. The crystalline and amorphous areas in coatings can be identi"ed by the elastic modulus di!erence. The concentrationgradient of O and OH ions was detected in the through-thickness direction of coatings. Lack of O and OH ions near the titaniuminterface implies the existence of phases other than HA, and might result in excessive adsorption of the coatings near the interface inHA-coated Ti implants. ( 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Hydroxyapatite; Coatings; Implants; Microstructure and chemical gradient

1. Introduction

Plasma-sprayed hydroxyapatite (HA) coatings on tita-nium (Ti) alloy provide excellent biocompatibility andbioactivity for skeletal and dental Ti implants. The idealHA coatings on a Ti implant should be mechanicallydurable in the physiological environment and inducenatural bone growth around the implants. These e!ectscan be achieved by (1) an in growth of bone into HAcoatings, and (2) a reciprocal dissolution/re-precipitationreaction between the bone and HA coatings through ionexchange. In clinical applications, it was found that HA-coated Ti implants su!er mechanical failure, often nearthe HA coatings/Ti alloy interfaces, by follow-up opera-tions after a certain period of implantation [1,2]. Relia-bility of HA coatings depends on the microstructure andchemistry of the coatings, which are a!ected by coatingprocessing. In the plasma spray processing, rapid heatingby plasma and rapid cooling on metal substrate cancause phase transformations of HA crystalline powder.Thus, the di!erent thermal history of individual HAparticles which impact the Ti substrates at di!erent timesgenerates inhomogeneous microstructure in HA coatings[3}6]. Considering the di!erence in cooling rates of mol-ten HA powder located from near Ti interface of Ti to the

coating surface, certain chemical or physical changes incoatings through the thickness direction should be ex-pected. The HA grain size di!erence with distance fromthe interface was observed by transmission electronmicroscopy [7]. This study focused on developing anunderstanding of inhomogeneous features in plasma-sprayed HA coatings on Ti implants using variouscharacterization techniques. The aim is to improve thecoating process based on such an understanding, and toachieve better performance of HA-coated implants.

2. Materials and experimental

A Ti}6Al}4V substrate was sand-blasted to a rough-ness of R

!3.5}4.0 lm before the HA coating. Then, the

substrate was cleaned ultrasonically in petroleum etherand in alcohol. Pure crystalline HA powder havinga spherical shape with an average diameter of 52 lm wasused for plasma spray coating. The plasma spray processwas performed using a Metco MN system with arc gas ofnitrogen. The arc current and voltage were 360}400 Aand 60}70 V, respectively. The thickness of HA coatingwas controlled 40 and 400 lm. The as-received coatingsamples were mounted in epoxy resin for preparing thecross-sectional specimens. The cross-section of HA coat-ings on Ti alloy substrate was polished with aluminumoxide powder with a size of 0.5 lm for observation inscanning electron microscope (JEOL JSM 6300F).

0142-9612/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved.PII: S 0 1 4 2 - 9 6 1 2 ( 9 9 ) 0 0 2 7 3 - 2

Fig. 1. SEM micrograph of HA coating cross-section shows the rib-bon-like region surrounded by the relatively smooth area.

Fig. 2. High-magni"cation SEM micrograph reveals that the ribbon-like region is granular.

Table 1Di!erence of Young's modulus in HA coatings!

Microscopicareas

1 2 3 4 Average(GPa)

A 83.9 84.0 83.5 82.1 83.4B 115.5 139.5 126.9 132.5 128.6

!The values were obtained by averaging four measurements as shownin the table.

Time-of-#ight secondary ion mass spectrometer (Phys-ical Electronic PHI 7200) was used to examine thechemical variations in HA coatings. The spectrometerequipped with two ion guns (Cs` for high mass-resolu-tion spectroscopy and 69Ga` for spatially resolved imag-ing), and a re#ection ToF SIM analyzer. The SIMSmapping was acquired in both positive and negative ionmodes using a 69Ga` beam of 25 keV. To obtain highspatial resolution mapping and control the surface charg-ing, an ion pulse length of 20 ns was used. Each pulsedelivered about 125 ions and the beam size was about0.5 lm. The time to digital converter was 10 ns.

Micro-Raman (Renishaw 3000) with a spatial resolu-tion of 1 lm was also used for chemical analysis. Thespectrum resolution was about 0.2 cm~1. The laserpower at specimens was 40 mw, and exciting wavelengthwas 512.5 nm of argon ion laser. The spectral slit widthwas 2.7 cm~1.

Nanoindenter (Nano Instrument, NanoIndenter II)was used for obtaining the Young's modulus of micro-areas in the HA coatings. The load-controlled mode wasadopted for indentation tests with the maximum force of3 mN.

3. Results and discussion

3.1. Crystalline versus amorphous

The microstructure of HA coatings on metal substrateswas complicated as a result of the plasma spray process-ing. The examinations of the coating cross-section revealthat there are two distinguishable microstructural areasas shown in Fig. 1: a ribbon-like region (pancake shape inthree dimensions) surrounded by a relatively smootharea. The higher magni"cation micrograph (Fig. 2) indi-cates that the ribbon-like region (area B in Fig. 2) has

a granular surface appearance. No signi"cant di!erenceof chemical composition between the two regions wasdetected by SIM spectroscopy. The local elastic modulusdetermined by the nano-indentation tests is shown inTable 1. The Young's modulus of area B is considerablyhigher than that of area A (&83 GPa versus&128 GPa).A similar di!erence in the Young's modulus has beenfound in HA on stainless steel [8]. Considering the rapidcooling of melt HA on metal substrates, we believe thatarea A is likely the amorphous phase which should ex-hibit a lower elastic modulus than its crystalline counter-part due to extra free volume associated with amorphousstructure. To identify the crystalline phase in microscopicarea remains a challenge, since the most convincingmethod, X-ray di!raction, cannot provide su$cient spa-tial resolution. Note that the micro-Raman spectroscopywith high spatial resolution might provide indirect in-formation of HA crystallinity [9]. Fig. 3 shows themicro-Raman spectra of areas A and B. The spectra ofboth areas shows the phosphate vibration mode l

2(&430 cm~1), and mode l

4(&596 cm~1). The phosphate

mode l1

near 964 cm~1 is notably shifted to lower wavenumber side in the spectrum of area A. Such a shift canalso be detected at the band of mode l

2. More-over, all

the vibration bands of phosphate are broadened in the

1340 J. Wen et al. / Biomaterials 21 (2000) 1339}1343

Fig. 4. ToF SIMS maps of Ti/HA coatings of 40 lm. The intensity of chemical element signals is proportional to brightness. The left-hand side of mapsis the Ti substrate.

Fig. 3. Micro-Raman spectra of the ribbon-like region (B) and of thesmooth area (A).

spectrum of area A. It has been reported that theamorphous phase generally widens the HA vibrationbands and enhances the low-frequency side of bands[10]. A direct comparison of HA vibrational spectra ofcrystalline and amorphous phases [10], which were iden-ti"ed by the X-ray di!raction, supported such character-istics of amorphous in vibrational spectrum. Combiningthe evidence of elastic modulus and Raman spectra, we

believe that area A is amorphous and area B is crystallinecalcium phosphate as appear in the cross-section of HAcoatings. Note that the bands at 950 and at 1653 cm~1

did not appear in that of pure HA [11], and this mightindicate the existence of calcium phosphate phases otherthan HA.

3.2. Composition gradient

ToF SIMS provides information of chemical composi-tion changes through thickness direction of HA coatings.Figs. 4 and 5 show the SIMS scan of coating sampleswith thicknesses of 40 and 400 lm, respectively. Theimage of Ti ions shows the clear interface of Ti/HAcoatings. The intensity of Ca ions is uniform in thecoating side, forming an image of coatings. The SIMSintensities of O and OH ions are not uniform in coatingsof 400 and 40 lm. The intensity of both ions decreaseswith distance from the surfaces of coatings, and reach theminimum at the interface between Ti and HA. ToFSIMS, however, failed to provide information of phos-phorus ions distribution due to the phosphorus' lowelectronegativity. Thus we could not obtain Ca/P ratiochanges through the thickness.

The gradient of OH ions was examined by the Ramanspectroscopy. Fig. 6 shows the Raman spectra obtainedat three locations in the coatings: near the interface (a), inthe middle (b) and near the surface (c). The band of OH

J. Wen et al. / Biomaterials 21 (2000) 1339}1343 1341

Fig. 5. ToF SIMS maps of Ti/HA coatings of 400 lm. The intensity of chemical element signals is proportional to brightness. The right-hand side ofmaps is the Ti substrate.

Fig. 6. Micro-Raman spectra of HA coating obtained in di!erentlocations: (a) near Ti/HA interface; (b) near the middle position betweenthe interface; and surface; and (c) near the surface of coating.

stretch vibration (&3579 cm~1) cannot be detected nearthe interface, but is obvious near the surface. The inten-sity of OH band is intermediate in the middle of coatings(spectrum b). The consistent results from both ToF SIMSand Raman spectroscopy reveal the concentration gradi-ent of O and OH in the plasma-sprayed HA coatings.Note that the vibrational spectroscopy (both IR and

Raman) has been used to examine the calcium phosphatecoatings [12], however, the examinations, commonlyfrom the top of coating surfaces, cannot reveal the chem-ical changes through the thickness direction. What ismore harmful is that the examinations on the coatingsurfaces, including X-ray di!raction, may providea wrong impression of phase compositions and crystal-linity in plasma-sprayed HA coatings.

It is known that HA might transform into other cal-cium phosphate phase during plasma spray. The lack ofO and OH ions might indicate possible phase transformfrom HA to tricalcium phosphate (TCP) and oxyhyd-roxyapatite (OHA). Our examination results suggest thatsuch phase transformations are enhanced in HA coatingsclose to the Ti interface during plasma spray. A possibleexplanation of the concentration gradient is that thephase transformations are sensitive to cooling rates, con-sidering decrease of cooling rates with the formation ofcoatings after the molten HA particles impact the sub-strate. The high dynamic process, due to high-speedimpact and rapid solidi"cation of HA particles on the Tisubstrate, might also generate extra driving force ofsolid-phase transformations after solidi"cation. Figs. 4and 5 seem to rule out the possibility of chemical changesdue to absorption and di!usion of O and OH ions fromcoating surfaces because the chemical gradient apparent-ly is independent of coating thickness. Otherwise, we

1342 J. Wen et al. / Biomaterials 21 (2000) 1339}1343

would see much less chemical gradient in 40 lm thicksample than in 400 lm sample if the gradient would haveresulted from surface ion absorption and ion di!usion insolid. The higher content of TCP or OHA phase near theinterfaces, however, needs to be further con"rmed byother techniques.

The chemical gradient and possible phase gradient inplasma- sprayed HA coatings should a!ect the bioactiv-ity and mechanical stability of coated metal implants.The interface adhesion of Ti and HA coatings relies onmechanical bonding. Weak mechanical bonding oftencauses the early failure of coated implants due to themetal}substrate and interface separation of metal sub-strate and coatings. The lack of OH near the interfacemight also contribute to early failure due to higher ab-sorption rates of non-HA phases such as TCP in livingbody. Further in vivo tests are needed to evaluate thee!ects of chemical gradient in HA coatings.

4. Conclusions

Plasma-sprayed HA coatings on Ti implants exhibittwo metallographically distinguishable areas. Combina-tion of the nano-indentation and micro-Ramantechniques reveals that ribbon-like crystalline areas aresurrounded by the amorphous phase.

The OH and O concentrations decrease with the dis-tance from the surface of HA coatings, reaching theminimum at the interface between HA and Ti alloys.E!ects of this chemical gradient in HA coatings on bioac-tivity and mechanical stability should be addressed infuture study.

Acknowledgements

This work was supported by a competitive earmarkedgrant for research, Project No. HKUST799/96E fundedthe Research Grants Council of Hong Kong. The authorsthank Dr. Lutao Weng in the Materials Characterizationand Preparation Facility for assistance in ToF SIMSexperiments.

References

[1] Yang CY, Wang BC, Chang E, Wu BC. J Mater Sci: Mater Med1995;6:258}65.

[2] Sergo V, Sbaizero O, Clarke DR. Biomaterials 1997;18:477}82.[3] Ducheyne P, Van Raemdonck W, Heughebaert JC, Heughebaret

M. Biomaterials 1986;7:97}103.[4] LeGeros RZ, Zheng R, Kuikowska R, Fan D, LeGeros JP. In:

Horowitz E, Parr JE, editors. ASTM STP 1196. Philadelphia:American Society for Testing Materials, 1994. p. 43}53.

[5] Tong WD, Chen JY, Zhang XD. Biomaterials 1995;16:829}32.[6] Chen JY, Wolke JGC, de Groot K. Biomaterials 1994;15:396}9.[7] Ji HX, Ponton CB, Marquis PM. J Mater Sci: Mater Med

1992;3:283}7.[8] Zhang CG, Chan WL, Leng Y, Tong P, Zhang XD. In: Zhang

XD, Ikada Y, editors. Biomedical research in the far eastern III.Kyoto, Japan: Kobunshi Kankokai, 1997. p. 170.

[9] Arias JL, Mayor MB, Garcia-Sanz FJ, Pou J, Leng Y, Leon B,Merez-Amor M. Thin calcium phosphate coatings produced bypulsed laser deposition with tailored physicochemical properties.Presented in 14th European Conference on Biomaterials, Hague,The Netherlands, September 1998.

[10] De Mul FF, Hottenjuis MH, Bouter P, Greve J, Arends J, BoschJJT. J Dent Res 1986;3:437}40.

[11] Blakslee KC, Condrate RA. J Amer Ceram Soc 1971;54:559}63.[12] Dasarathy H, Riley C, Coble HD. J Biomed Mater Res

1993;27:477}82.

J. Wen et al. / Biomaterials 21 (2000) 1339}1343 1343