SURFACE AND INTERFACE ANALYSIS Surf. Interface Anal. 26, 1027È1034 (1998) XPS Studies of Low-temperature Plasma-produced Graded Oxide–Silicate–Silica Layers on Titanium Michael Steveson, Pawittar S. Arora and Roger St. C. Smart* Ian Wark Research Institute, University of South Australia, The Levels Campus, Mawson Lakes, South Australia 5095 A method has been developed to produce a silica coating bonded to an oxidized titanium substrate, graded in composition and structure through the surface layer, providing the base for deposition of bioactive glasses and hydroxyapatite. A low-temperature air/water/tetraethoxysilane plasma coating method is reported. Characterization of the graded surface layer using XPS and SEM is compared with previously reported studies of similar layers on nickel substrates. Depth proÐling and the modiÐed Auger parameter were used to elucidate the structure of the interphase region between the titanium and silica. A functionally-graded layer, from the titanium metal through the oxide, orthosilicates, pyrosilicates, chain silicates and layer silicates to bulk silica, was demonstrated. The thickness of the layer was in the range 30–50 nm, depending on the duration of oxidation and tetraethoxysilane deposition. ( 1998 John Wiley & Sons, Ltd. KEYWORDS : XPS ; x-ray photoelectron spectroscopy ; Ti ; SiO ; Auger parameter ; depth proÐling 2 INTRODUCTION Owing to their strength, metals (e.g. CoCr alloy, stainless steel, titanium) are the materials of choice for stressbearing medical and dental implants. When an implant is Ðtted, the way the gap between the implant and the bone is Ðlled by the body is determined by the type of material used. The body surrounds materials such as stainless steel, referred to as a biotolerant material, with a layer of Ðbrous tissue. This prevents direct boneÈ implant contact and reduces the rigidity of the joint, thus stainless-steel implants have been replaced in many instances by titanium implants. The spaces between a bioinert material, such as titanium, and the bone are Ðlled by growing bone, producing a much more rigid joint.1 In the 1980s, bioactive materials, e.g. bioglass and hydroxyapatite, were developed for implant use.1,2 Bone integrates with the surface of these compounds resulting in the strongest interface possible.1 These ceramic or glass bioactive materials have lower mechanical strength than the metals previously used for implant manufacture and hence are generally applied as surface coatings to the metals rather than as bulk materials. With the development of high strength at the bone/implant interface with modern bioactive materials, there is increased need for correspondingly highstrength adhesion between the coating and the metal implant. Currently, titanium implants are coated with hydroxyapatite using plasma spraying. In this process, a high-speed, high-temperature (15 000 ¡C) stream of partially melted hydroxyapatite particles is Ðred at the titanium surface, resulting in a relatively thick ([10 lm) coating that is usually poorly bound to the titanium * Correspondence to : R. St. C. Smart, Ian Wark Research Institute, University of South Australia, The Levels Campus, Mawson Lakes, South Australia 5095 CCC 0142È2421/98/131027È08 $17.50 ( 1998 John Wiley & Sons, Ltd. surface. It is brittle and often contains a variety of unwanted compounds (e.g. CaO) produced by thermal decomposition of the hydroxyapatite. One study found that the load required for failure of the titanium/apatite interface was 12 MPa, whereas the bone/apatite interface failed at 17 MPa.1 Hence, the advantages of the new high-strength bioactive materials was not being realized due to poor coating quality. Other methods to coat titanium with bioactive materials have been suggested, e.g. sol-gel-derived coatings,3 but bonding at the metal/ceramic interface remains weak and hydrolysis can detach the layer. It is evident that this is a research area requiring more development. Arora and Smart4 reported a low-temperature plasma reaction using a radio frequency source (100 W) at \40 ¡C to produce compositionally and structurally graded oxide/silicate/silica layers on nickel substrates. The surface of the nickel was cleaned in an air or argon plasma and, then air saturated with water vapour was introduced into the plasma to oxidize the nickel surface. When tetraethoxysilane (TEOS, (C H O) Si) vapour 2 5 4 as deterwas introduced, SiO radicals were produced, mined by mass spectroscopy of the plasma gas, which reacted with the oxidized surface to produce silicon oxides. As the reaction proceeded, orthosilicate (SiO ) 4 structures were formed, then pyrosilicates, followed by layer silicates, chain silicates and bulk silica. Surface analysis by XPS, with silicate structures elucidated from modiÐed Auger parameter data, was used to characterize the surface layers as the reaction proceeded. The integrity of the layer was tested by dissolution testing in pH 2 nitric acid. The silica coating reduced the nickel dissolution rate by more than two orders of magnitude over [30 days, demonstrating the continuity and coherence of the layer. In this paper we will report the use of a similar method to coat titanium with silica, producing a coherent, continuous coating and strong metalÈceramic bonding. The transition from titanium, through the Received 1 May 1998 Accepted 14 August 1998 1028 M. STEVESON, P. S. ARORA AND R. St. C. SMART oxide and silicates to silica is characterized by XPS and SEM, as previously.4 The silica-coated titanium can be used as a substrate for hydroxyapatite, bioactive glass or ceramic coating. METHODS Titanium sheet (99.99%) and tetraethoxysilane (or tetraethylorthosilicate) (98%) were supplied by Aldrich Chemical Co. All water introduced into the plasma was puriÐed in a Milli-Q system. Before use, the titanium sheet was cut into D1.5 cm square pieces that were cleaned following the method described by Wilson,5 i.e. washing in 5 wt.% KOH solution followed by a water rinse, then a warm 1.6 wt.% HF/33 wt.% HNO solution for 2È3 min followed by another water rinse. 3 The plasma reaction method has been described previously.4 The plasma was generated in a Harrick Plasma Cleaner/Sterilizer (PDC-32G) run at 100 W, frequency 13.5 MHz and D2 Torr pressure, as measured by an Edwards High Vacuum Model 8/1 pressure gauge. The sample was placed on a Petri dish to support it in the centre of the plasma. In the Ðrst step of the reaction, air is introduced to the chamber to clean impurities (particularly hydrocarbons) o† the metal surface. In the second step, water vapour was introduced into the plasma by bubbling the incoming air through water. In the third step, silicon species were added to the plasma by adding TEOS to the water with air bubbling through it. A typical treatment was 10 min of air plasma, 20 min of water plasma and 10 min of TEOS/water plasma. The treated titanium surfaces were analysed by x-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM). A Physical Electronics PHI 5600 ESCA system with a magnesium (hl \ 1253.6 eV) x-ray source operating at 300 W (15 kV, 20 mA) was used for XPS surveys (pass energy 99.3 eV) and elemental regions (pass energy 58.7 eV). The bremsstrahlung radiation of the source was used for Auger electron spectroscopy of the D1600 eV Si KLL transition. The vacuum in the analysis chamber was maintained at \10~8 Torr during analysis. All spectra were corrected for charging e†ects using the hydrocarbon contamination peak (uncharged) at 284.8 eV. Depth proÐles were obtained by etching the surface with an argon ion gun, with the etching rate calibrated using a 100 nm tantalum oxide layer on a tantalum substrate. At 3 kV, the etch rate was determined to be 1.2 nm min~1. The inelastic mean free path (IMFP) can be calculated from the Seah and Dench relationship for inorganic compounds.6 For the O 1s XPS signal, using the bulk density of rutile TiO , the IMFP is calculated to be 0.43 nm. As 95% of the2 analysed photoelectrons are emitted from a depth less than or equal to 3 ] IMFP, this depth is taken to represent the analysis depth. For the angle-resolves XPS measurements, the analysis depth used was 3IMFPsinh. The values of these analysis depths are summarized in Table 1. For more general XPS measurements, an analyser angle of 45¡ was used. Information on the chemical nature of the siliconÈ oxygen bonds in the surface coating was determined using the modiÐed Auger parameter. The Auger paramSurf. Interface Anal. 26, 1027È1034 (1998) Table 1. Average analysis depths for O1s corresponding to XPS analysis angles of between 10 and 90Ä Analysis Angle IMFP (nm) 3 Ã IMFP (nm) 10 30 45 60 75 90 0.074 0.21 0.30 0.37 0.41 0.43 0.22 0.63 0.90 1.1 1.2 1.3 eter, introduced by Wagner,7,8 is now commonly expressed in the modiÐed form [email protected] \ KE ] BE where KE is the kinetic energy of the most intense Auger electron and BE is the binding energy of the main photoelectron from the same element in the same XPS spectrum. The modiÐed Auger parameter has the advantage that surface charging is removed by the equal and opposite shifts in each of the two contributing energies. More importantly, it can be shown that the modiÐed Auger parameter is a measure of the energy changes due to extra-atomic relaxation and is therefore often more sensitive to the stoichiometry and strength of bonding to the next nearest neighbours of the emitting atom than the XPS BE shifts.9,10 Hence, the modiÐed Auger parameter for silicon in the silicateÈ silica coatings provides information about the changes in siliconÈoxygen bonding as the layer is formed by plasma reaction. Scanning electron microscopy was performed using a high-resolution, Ðeld emission CAMSCAN CS44FE microscope operating at 20 kV. The samples were not coated before examination in the microscope. RESULTS Initial oxidized titanium surface Although the characteristics of titanium surfaces exposed to air are already well documented, our analyses of these surfaces are brieÑy summarized here for comparison with the changes produced by air plasma cleaning described in the next section. Electron microscopy of the initial, naturally oxidized titanium surface shows that this oxide layer has many impurities and irregularities, as indicated by the pits, scratches and di†erentiated regions in Fig. 1. Chemically-cleaned titanium can form nonstoichiometric oxides and sub-oxides, TiO , where x is x Table 2. Titanium 2p 3/2 XPS binding energy for the naturally oxidised titanium surface compared to literature values Compound Ti 2p 3/2 Peak (eV) Reference TiO 2 Ti O 2 3 TiO Ti Oxidised Surface 458.5 456.8 454.7 453.89 485.5 Í18Ë Í18Ë Í18Ë Í17Ë ( 1998 John Wiley & Sons, Ltd. LOW-TEMPERATURE PLASMA-PRODUCED GRADED SILICA COATING ON Ti Figure 1. The naturally oxidized titanium surface showing an irregular surface and pitting. \2, on reoxidation.11 X-ray photoelectron spectroscopy analysis of the natural oxide layer revealed only the dioxide (Table 2). There was no shoulder on the Ti 2p 458.5 eV peak to indicate suboxides. An overlayer of [email protected] organic carbon contamination was also indicated by the XPS spectrum. Surface atomic concentrations calculated from XPS peak areas give 2È10% titanium, 15È30% oxygen and 60È80% carbon, varying from sample to sample depending on their history of exposure to the environment. The hydrocarbon layer contains some oxygen, as evidenced by ether (e.g. 287 eV) and carbonyl or carbonate (e.g. near 288.5 eV) oxidized hydrocarbon peaks on the high-binding-energy side of the main C 1s peak and by the presence of three peaks in the surface oxygen spectrum, the Ðrst relating to the oxide in TiO (BE \ 529.7 eV), the second to OH 2 eV) and the third oxygen as CxO on species (D531.5 carbonate (BE \ 533 eV) (Fig. 2). These species have been found previously on untreated titanium implant Figure 2. Comparison of the O 1s XPS peak of the naturally oxidized titanium at different analysis angles. ( 1998 John Wiley & Sons, Ltd. 1029 materials.12 The relative intensity of the oxygen peaks changes as the analysing depth is increased by angle resolution, with the OH and CxO oxygens being more prominent in the more surface-speciÐc spectra (i.e. 10¡). With the instrument set to record from the deepest level, i.e. 90¡, the organic oxygen peak is seen only as a shoulder on the side of the much larger oxide peak (Fig. 2). The depth of analysis at each angle is recorded in Table 1. Even at 90¡, the depth of analysis is less than the oxide layer thickness ; hence, an ion-etched depth proÐle was used to determine the oxide thickness. Figure 3 shows an ion-etched depth proÐle of the natural oxide layer. As expected, the carbon signal from surface contamination drops to a constant level after the Ðrst 30 s. The etch rate of 1.2 nm min~1 gave a contamination layer thickness of \0.6 nm, in reasonable agreement with the angle-resolved XPS results. The surface oxygen concentration reaches a plateau at D20 at.% at a depth of D15 nm. This depth corresponds to the stabilization of the Ti signal, predominantly as the metal, at D65 at.%. It should be noted that the carbon and oxygen atomic concentrations stabilize at around 10% and 20%, respectively, rather than going to zero. There are a number of reasons for this observation. The surface of the titanium used is rough (Fig. 1) and some faces of the surface undulations are shadowed from the angled ion beam. The pressure in the instrument during measurement (i.e. 10~8È10~9 Torr) was too high to maintain a clean surface, i.e. reaction with residual H O 2 and CO is likely. The titanium may have oxidized down grain boundaries or the oxide may be of uneven thickness on di†erent facets. The 3 kV ion beam may also have knocked some of the surface carbon and oxygen further into the titanium. The oxide depth measured in this proÐle appears to be in the range 10È15 nm. Figure 3. X-ray photoelectron spectroscopy ion-etched depth profile of the naturally oxidized titanium surface. Sputter rate was 1.2 nm minÉ1 ; scale is atomic concentration (%). Surf. Interface Anal. 26, 1027È1034 (1998) 1030 M. STEVESON, P. S. ARORA AND R. St. C. SMART Plasma-modiÐed titanium surfaces Air plasma cleaning. Figure 4 shows a titanium surface after 10 min in an air plasma. The pits seen in the surface before plasma treatment (Fig. 1) have been removed. The air plasma has also reduced the surface organic carbon layer. The carbon concentration, measured at 45¡, is 60È80 at.% for the natural oxide. After plasma cleaning, the amount of carbon contamination varies between samples, due to their di†erent exposure times to the atmosphere before XPS analysis, but is generally 30È40 at.%. Also indicative of the removal of organic carbon contamination is that the 0 ls peak at 45¡ becomes a single peak at 529.7 eV with no OH or CxO shoulder on the high-energy side found before plasma cleaning (Fig. 2). This is evidence that the plasma actually removes almost all of the surface contamination but the titanium samples are recontaminated when they are reoxidized on exposure to air during transferral between the plasma chamber and the XPS. A series of samples were prepared in which contamination was minimized by backÐlling the plasma chamber with argon and transferring the sample from plasma to XPS as fast as possible (\30 s) in an argon-Ðlled vessel. These samples had only 7È8 at.% surface carbon and, signiÐcantly, only 16È17 at.% oxygen. The titanium surface has been cleaned extensively and is likely to be highly reactive on re-exposure to air. Hence, the hydrocarbon contamination can be reduced signiÐcantly, supporting the proposition that contamination in the plasma chamber is very low. The surface of titanium on exposure to air is pock-marked with irregularities. The air plasma removed these markings to leave a smooth, even surface. As the response of tissue to a metal implant surface is inÑuenced by the chemical and structural properties of the surface,12 e.g. hydrocarbon contamination and microscopic structure (e.g. pits), these variations may lead to an equally variable boneÈimplant bond strength. Thus, if the sample surface is not removed before plasma coatings are applied, the air cleaning should lead to better implant performance. Subsequent plasma treatments (as described below) are then applied to this cleaned, reactive surface. Figure 4. Naturally oxidized titanium surface after 10 min of air plasma cleaning, showing no pitting. Surf. Interface Anal. 26, 1027È1034 (1998) Air/water vapour plasma oxidation. The aim of air/water vapour plasma oxidation was to increase controllably the depth of the oxide layer. Figure 5 shows a depth proÐle of the titanium surface layers after 10 min in an air plasma and then 20 min in air/water vapour plasma. The depth of the oxide layer was D30 nm, after which both the oxygen and titanium concentrations reach stable values. This indicates a 67È100% increase in the depth of the oxide due to this plasma reaction. The SEM images of the air/water vapour plasmaoxidized titanium surface are closely similar to Fig. 4. vapour plasma reactions. The introduction of tetraethoxysilane (TEOS) vapour into the plasma, along with air and water vapour, results in a silica reaction with the metal surface.4 The modiÐed Auger parameter in XPS provides a means of elucidating the structure of the silica surface coating produced on titanium. The data can be displayed on a silicon chemical state plot as in Fig. 6. This plot shows the Si 2p binding energy on the x-axis, the Si KLL Auger kinetic energy on the y-axis and, hence, the modiÐed Auger parameter on diagonals. Figure 6 is a silicon chemical state plot with a series of siliconÈoxygencontaining standards analysed by Wagner et al.,13 with the energies adjusted to a C ls standard of 284.8 eV rather than the original 284.6 eV. It shows that the different SiwO bond types from orthosilicates (SiO ) 4 through Si O species to SiO can be determined from x y 2 this XPS data. Various TEOS plasma reaction times (up to 12 min) were used to prepare a series of samples with surface silicon concentrations of between 2.2 and 28 at.%. Allowing for some hydrocarbon contamination, 25È30% silicon is the practical maximum for SiO . In the samples that contain this amount of silicon,2 the Air/H O/tetraethoxysilane 2 Figure 5. X-ray photoelectron spectroscopy ion-etched depth profile of water vapour plasma-treated titanium. The spike in the spectrum at 4.5 nm was caused by an electrical fault whilst recording the Ti spectrum. ( 1998 John Wiley & Sons, Ltd. LOW-TEMPERATURE PLASMA-PRODUCED GRADED SILICA COATING ON Ti 1031 Figure 6. Silicon chemical state plot showing the distribution of a number of silicon–oxygen species. The silicon–oxygen stoichiometry is shown in parentheses. (Adapted from Wagner et al .13). plasma reaction was stopped when the silicon content had just reached this level rather than being allowed to continue to deposit as bulk silica. Figure 7 shows the XPS silicon atomic concentrations after increasing reaction time superimposed on the chemical state plot. The chemical state plot gives two di†erent kinds of information. As indicated by the plot of the siliconÈoxygen standards,13 SiwO bonds in di†erent structures are distributed along the Auger parameter diagonals, i.e. compounds with the same modiÐed Auger parameter Figure 7. Silicon chemical state plot showing spread of silicon–oxygen structures with varying surface silicon concentrations (shown next to data points). The TEOS plasma reaction times ranged from 0 to 12 min. ( 1998 John Wiley & Sons, Ltd. Surf. Interface Anal. 26, 1027È1034 (1998) 1032 M. STEVESON, P. S. ARORA AND R. St. C. SMART can have di†erent structure. It is evident from Fig. 7 that there is a change in the structure of the siliconÈ oxide bonds with increased silicon deposition. As the silicon concentration rises, there is a shift from orthosilicate-like structure to silica-like structure via the intermediate pyro-, chain and layer silicates. The second type of information on a chemical state plot is the stoichiometry of the silicon oxide. A series of papers by Alfonsetti and colleagues14h16 have derived a parameter, b, that can be related siliconÈoxygen stoichiometry. The di†erence between the modiÐed Auger parameter for elemental silicon and that of the silicon oxide SiO , x the b parameter, is related to the stoichiometric ratio, x, in SiO via the following equations from refs 14È16, x b \ 1716.0 [ [email protected] SiOx b \ 2.25x indicating that the higher the value of [email protected] is, the fewer oxygens per silicon. The [email protected] parameters for quartz (i.e. 1711.5 eV) and pure Si (i.e. 1716.0 eV) were used to calibrate the calculated b values as in our previous paper.4 The points in Fig. 7 lie in a line of lower slope than y \ x (constant Auger parameter), indicating a change in SiwO stoichiometry from SiO at 2.8% Si to SiO at 26% Si. These results may 1.4 seem contradictory 1.8 structural information in Fig. 7, which indicates to the that at 2.8% silicon the oxide is orthosilicate-like, i.e. x \ 4, but this is not the case because both the low silicon/oxygen ratio and the orthosilicate structure can be explained by the mechanism of the plasma reaction. Arora and Smart4 used mass spectroscopy of the plasma to deduce the mechanism of the reaction on nickel oxide. When TEOS was added to the plasma, the m/e signals at 22 and 44 (singly and doubly charged SiO) increased markedly. The peaks corresponding to O, OH, H O, SiH and SiOH increased less dramatically. The 2silicon atom from SiO apparently occupies the tetrahedral interstice in the surface oxide structure, which in their case is nickel oxide but in this case is titanium dioxide. The resulting structure has silicon atoms connected to four oxygen atoms by orthosilicatelike bonds, hence the position of the 2.8% Si point on Fig. 7, but the oxygen atoms are bonded to titanium atoms as well, reducing the siliconÏs “shareÏ of the four oxygens and explaining the apparent silicon/oxygen ratio of 1.4. It was concluded that SiO was the species responsible for the initial formation of SiO4~ in the 4 nickel oxide.4 Ion etching depth proÐles were performed on some of the samples to determine the depth of the siliconcontaining layer. Figures 8È11 show that samples with surface silicon concentrations of 6.1, 13, 18 and 26 at.% had silicon concentrations down to depths of D0.8, 2.5, 6 and 18 nm, respectively. The results of these depth proÐles, combined with the chemical state plot (Fig. 6), indicate that as the TEOS plasma reaction proceeds, the silicon-containing layer becomes thicker and more silica-like, i.e. a compositionally graded structure exists in the layer, as suggested by Arora and Smart.4 Further proof of this structure is found in the modiÐed Auger parameter depth proÐle, using ion beam etching, through a completely coated (Si \ 28 at.%) surface. Figure 12 shows that there is a systematic shift of both SiwO structure and stoichiometry SiO as the x Surf. Interface Anal. 26, 1027È1034 (1998) Figure 8. X-ray photoelectron spectroscopy ion-etched depth profile of TEOS plasma-treated titanium with 6.1 at.% surface silicon. ion etch continues through the layer. The nature of the SiwO bonds on the outer surface are SiO -like, moving 2 through the layer silicates, the chain silicates, the pyrosilicates and Ðnally the orthosilicates close to the titanium oxide (as the Si signal disappears). It should be noted that as the ion etching cleaned the hydrocarbon o† the surface within the Ðrst minute, no charge correction of the points against the C ls peak at Figure 9. X-ray photoelectron spectroscopy ion-etched depth profile of TEOS plasma-treated titanium with 13 at.% surface silicon. ( 1998 John Wiley & Sons, Ltd. LOW-TEMPERATURE PLASMA-PRODUCED GRADED SILICA COATING ON Ti Figure 10. X-ray photoelectron spectroscopy ion-etched depth profile of TEOS plasma-treated titanium with 18 at.% surface silicon. 284.8 eV was performed. This does not reduce the relevance of these results because the modiÐed Auger parameter is self-correcting for charge. If the surface becomes positively charged due to the loss of electrons, the binding energy would increase and the kinetic energy will decrease equally. The 30È50 mm silicate/silica layers are completely deformable (to repeated 180¡ bending) without evidence 1033 Figure 11. X-ray photoelectron spectroscopy ion-etched depth profile of TEOS plasma-treated titanium with 26 at.% surface silicon. of disbonding or cracking at the interface. Mechanical tests on the hardness, tensile strength and frictional forces are in progress. CONCLUSION The low-temperature air/water/tetraethoxysilane plasma coating method has been shown to produce Figure 12. Silicon chemical state plot showing the change in silicon–oxygen stoichiometry and structure within a 27.7 at.% surface silicon coating. Produced from an XPS ion-etched depth profile. Sputter rate was 1.2 nm minÉ1 ; sputter times (in min) are shown next to the data points. ( 1998 John Wiley & Sons, Ltd. Surf. Interface Anal. 26, 1027È1034 (1998) 1034 M. STEVESON, P. S. ARORA AND R. St. C. SMART compositionally and structurally graded layers from the titanium metal through the oxide, orthosilicate, pyrosilicates, chain silicates and layer silicates to bulk silica, as was found previously for nickel substrates. The initial surface of the titanium metal in air has a thin layer of hydrocarbon contamination (D0.6 nm) overlying a layer of TiO 10È15 nm thick. The low-temperature air 2 plasma removed the bulk of the hydrocarbon contamination and substantially smoothed the surface by annealing pits and scratches. An air/water vapour plasma treatment increased the depth of the oxide layer to 25È30 nm. The addition of tetraethoxysilane vapour to the air/water plasma resulted in a surface reaction producing the silicate structures with increasing SiwO bonding as the reaction proceeded. The compositional and structural gradation of the layer has been demonstrated in XPS depth proÐling and by information from the modiÐed Auger parameter measurements from the layer during deposition and in ion beam etching after deposition to an overlayer of bulk SiO . 2 These bonded layers provide the basis for deposition of overlayers of oxides, glasses and hydroxyapatite using solÈgel formulations.1h3 These depositions have been accomplished successfully and the results will be reported in a separate publication. Acknowledgement The authors would like to thank Dr Bill Skinner for helpful advice and Mr Len Green for his assistance on the scanning electron microscope. Funding for this research was provided by the Australian Research Council through a large grant and a Senior Research Fellowship (R.St.C.S.). REFERENCES 1. K. Soballe, Acta Orthop . Scand . Suppl . 64 (255), 1 (1993). 2. L. L. Hench and H. A. Paschall, J . Biomed . Mater . Res . Symp . 4, 25 (1973). 3. C. Chai and B. Ben-Nissan, Int . Ceram . Monogr . 1 (1 , Proceed ings of the International Ceramics Conference , 1994 ) p. 66 (1994). 4. P. Arora and R.St.C. Smart Surf . Interface Anal . 24, 539 (1996). 5. D. H. Wilson, in Wood, W. G. (ed) Metals Handbook , 9th Edn, Vol. 5, edited by W. G. Wood. American Society for Metals, Ohio, USA (1982). 6. M. P. Seah and W. A. Dench, Surf . Interface Anal . 1(2), 2 (1979). 7. C. D. Wagner, Anal . Chem . 44(6), 967 (1972). 8. C. D. Wagner, Anal . Chem . 47(7), 1201 (1975). 9. C. D. Wagner, Faraday Discuss . Chem . Soc . 60, 291 (1975). 10. S. D. Waddington, Practical Surface Analysis , 2nd Edn, Vol. 1, edited by D. Briggs and M. P. Seah. Wiley, Chichester (1990). Surf. Interface Anal. 26, 1027È1034 (1998) 11. A. F. Carley, P. R. Chalker, J. C. Riviere and M. W. Roberts, J . Chem . Soc . Faraday Trans . 1 83, 351 (1987). 12. J. Lausmaa and B. Kasemo, Appl . Surf . Sci . 44, 133 (1990). 13. C. D. Wagner, D. E. Passoja, H. F. Hillery, T. G. Kiniski, H. A. Six, W. T. Jansen and J. A. Taylor, J . Vac . Sci . Technol . 21(4), 933 (1982). 14. R. Alfonsetti, L. Lozzi, M. Passacantando, P. Picozzi and S. Santucci, Thin Solid Films 213, 158 (1992). 15. R. Alfonsetti, L. Lozzi, M. Passacantando, P. Picozzi and S. Santucci, Appl . Surf . Sci . 70/71, 222 (1993). 16. R. Alfonsetti, G. De Simone, L. Lozzi, M. Passacantando, P. Picozzi and S. Santucci, Surf . Interface Anal . 22, 89 (1994). 17. M. V. Kuznetsov, Ju.F. Zhuravlev, V. A. Zhilyaev and V. A. Gubanov, J . Electron Spectrosc . Relat . Phenom . 58, 1 (1992). 18. D. Briggs and M. P. Seah (eds), Practical Surface Analysis , 2nd Edn, Vol. 1. Wiley, Chichester (1990). ( 1998 John Wiley & Sons, Ltd.