Purified Oxides with Novel Morphologies formed from Ti-alloys

The current invention relates to a purified oxide which is possessed of a novel in surface morphology, and a method of producing the same from a Ti-alloy mixed with sulphuric and phosphoric acid electrolytes.

Background to the Invention

Spark anodization of c.p.-titanium (commercially pure) for biomedical surface preparation, also know as micro(-)arc oxidation (MAO), micro-plasma oxidation (MPO), anodic spark deposition (ASD) and plasma electrolyte oxidation (PEO), has been the subject of numerous recent scientific studies. Considerably less data is available on spark anodisation of the Ti6A14V alloy, and about the increasingly used Ti6A17Nb alloy. Spark anodisation of Ti or Ti- alloys under appropriate conditions can lead to rough and porous anodic films which are up to several tens of micrometers thick, whilst at the same time being free of sharp edges. The high porosity opens up new possibilities with regard to incorporation of bioactive or bioresorbable species and the preparation of "bioactive surfaces", which is particularly interesting for implant applications where accelerated bone in-growth is important. In 1986, Branemark and co-workers published the patent US 4,330,891, describing a micro-pitted (Ti)-surface with pore size between

10-1000 nm, ideally equal or smaller than 300 nm, in a TiO ₂ rutile layer. Without further specification, it is evident that such surfaces were produced using a spark anodisation process.

The electrochemical oxide growth on cp-titanium in several acid and alkaline electrolytes, addressing anodic forming voltage, forming rate and growth constant, was studied by SuI et al. [SuI YT, Johansson CB, Jeong Y, Albrektsson T. The electrochemical oxide growth behaviour on titanium in acid and alkaline electrolytes. Med Eng & Phys 2001 ;23: 329-346]. Working in acetic acid, the same authors found that spark anodized turned surfaces, with an oxide thickness more than 600 nm, demonstrated a stronger histomorphometrical bone response and a more pronounced osteoinductivity compared to non-oxidized control implants after 6 weeks of insertion in rabbit bone. By comparing removal torque values of implants with different surface oxide chemistry resulting from anodisation in sulphuric acid, phosphoric acid and calcium containing electrolytes, it was concluded that surface chemistry and topography play an important role in the bone response to oxidized implants. The possibility of biochemical bonding between P and Ca containing Ti-oxides and bone was pointed out. WO 02/096475 was based on this data, which further claimed Ti-alloys, describing a spark anodized layer containing Ca, P or S and having a double layer structure of an upper porous layer and lower compact barrier layer. By comparing spark anodized (H ₂ SO ₄ ) Ti surfaces, subjected to a post thermal treatment (600°C), to "bioactive" Ti surfaces produced by alkali immersion and heat treatment, Liang and co-workers [Liang B, Fujibayashi S, Neo M, Tamura J, Kim HM, Uchida M, Kokubo T, Nakamura T. Histological and mechanical investigation of the bone-bonding ability of anodically oxidized titanium in rabbits. Biomaterials 2003;24 (27): 4956-4966] observed faster bone-in-growth and higher failure loads on anodized surfaces in the early stage of implantation. The failure in improving bone-bonding ability compared to the chemically treated surfaces at later implantation stages was attributed to the low oxide porosity and hence, the superficial apatite-like deposition into the pores. The importance of the presence of crystalline TiO ₂ phase on the bioactivity of

sulphuric acid anodized titanium has also been underlined. An increase of anatase and even more significantly, rutile phase, obtained by either applying a higher anodisation potential or a post heat-treatment, resulted in accelerated apatite formation upon immersion in simulated body fluid (SBF). This has been attributed to the matching structure of rutile (101) and apatite (0004) planes, acting as nuclei for apatite crystal growth. JP 2003 190272, describes the formation of a coating with high bioaffmity on Ti and Ti-alloys by combined anodisation and post heat treatment at around 600°C. A very similar approach of anodisation (phosphate solution) and heat treatment for titanium or titanium-based micro-alloy containing at least 98 w-% Ti is discussed in US 5,354,390.

Considerable effort has been made to incorporate Ca and P species during spark anodisation of c.p.-Ti surfaces for a further enhancement of the bioactivity of titanium implants, without having the mechanical drawbacks of hydroxyapatite (HA) coatings. US 4,846,837 describes a process of spark anodizing a previously roughened (100-400 μm) implant metal in a

solution containing CaP, leading to 5 to 30 μm thick oxides having a CaP gradient throughout the

oxide. Electrolytes containing β-glycerophosphate (β-GP) and calcium acetate (CA) have been used and the anodisation process combined with a post-hydrothermal treatment at 300°C, leading to HA precipitation on the anodic oxide film, as described in US 478,237. JPl 1 043 799 and JP 2000 178791 describe a method for anodizing Ti in an electrolyte containing a glycerophosphate and a metal acetate followed by a hydrothermal treatment in water or steam, during which phosphorus or sulphur and the added metal dissolve and leave an adjustable pore size and highly wettable films. Working in a very similar electrolyte, WO 2004/000378 describes a biomimetic surface and a spark process for Ti, Ta and their alloys with incorporation of Ca and P. The 3-step process consists of spark anodisation in Ca-GP, a second spark deposition in Ca(OH) ₂ followed by immersion in NaOH. By pulsed anodisation in β-GP and CA containing solutions, oxides

containing CaTiO ₃ , β-Ca ₂ P ₂ O ₇ and α-Ca ₃ (Pθ ₄ ) ₂ have been produced above 450 V incorporated in a mixed anatase/rutile matrix with Ca/P ratios between 1.3-1.8. Without the need for hydrothermal treatment, these oxides show the capability to induce bone-like apatite in SBF. Working under very similar conditions, an increase of ALP activity and a decrease of cell proliferation rate has been found with increasing anodisation potential. Via a complex of Ca with EDTA and working at pH 14, a Ca/P ratio up to 1.3 has been seen. The fully or partly soluble CaP phases were embedded in a matrix of amorphous and anatase TiO ₂ , with further evidence that this process can be applied to Ti6A14V and Ti6A17Nb alloys. This anodisation procedure is described in WO 02/078759 mentioning pore densities between 10 ⁴ -10 ⁸ /mm ² and Ca/P ratios from 0.5 to 2.0.

Porous anodized films on Ti6A14V α/β alloy have been initially found advantageous for bonding purposes, and more recently for wear- and corrosion-resistant applications, biocidal or catalytic properties, sensors and also as biomedical coatings.

Spark anodisation of Ti6A14V in aluminate and silicate solutions containing Na ₂ SiO ₃ - KOH-(NaPOs) ₆ , Na ₂ SiO ₃ -KOH-(NaPO ₃ ) _O -NaAlO ₂ , (NaP0 ₃ ) ₆ -NaF-NaA10 ₂ and NaAlO ₂ has been shown to provide thick and wear-resistant porous coatings. Coating hardnesses of 8.54 GPa up to 13 GPa were reported due to the presence of hard TiAl ₂ O ₅ phase within TiO ₂ rutile matrix. Pulsed anodisation with 8% duty cycle was found to be advantageous compared with constant voltage, in terms of creating smaller pores in anodized films and less apparent cracking. A good combination of wear- and corrosion resistance of duplex Cr(N)ATiO ₂ coatings formed on Ti6A14V by microarc oxidation in sodium phosphate combined with magnetron sputtered Cr(N) as top layer has also been observed. By anodising Ti-alloys in vanadate-tungstate and vanadate- tungstate-phosphates containing solutions, oxides enriched with P, V and W having catalytic properties have been prepared. Of interest for sensor applications, nano-crystalline rutile films on

Ti6A14V substrate via spark anodisation in aqueous Na ₂ CO ₃ solution have been prepared and their mechanical properties investigated.

Using a 2-step process, spark anodisation of Tϊ6A14V in a solution of phosphate salt has been combined with anodic electrophoresis, creating a HA-TiO ₂ coating with a dense TiO ₂ film as an inner layer and HA as top layer. A combination of bioactivity, chemical stability and mechanical integrity was reported. Incorporation of Ca and P during spark anodisation of Tϊ6A14V as well as the effect of electrolyte composition and anodisation potential on structure, Ca/P ratio and film morphology have been studied. Using a combination of anodisation and hydrothermal treatment as known for c.p.-Ti, oxides containing fine HA or apatite crystals on Ti6A14V alloy were prepared. Working under alkaline conditions, the TiMAX ^® process for implant surfaces, property of DePuy ACE, USA, is based on the anodisation of Ti6A14V material followed by a mechanical post-treatment, resulting in a high fatigue strength.

The Ti6A17Nb alpha-beta alloy has been increasingly used for the production of cementless femoral stems and elastically deformable cup shells, often in combination with an oxygen diffusion hardening process reaching a surface hardness up 1200 HV. Investigations of the composition of passivated Ti6A17Nb surfaces, have shown an incorporation of Al and Nb in the oxide layer, with an enrichment of Al in the α and Nb in the β phase, reflecting the distribution in the underlying substrate. By comparing the electrochemical behaviour of air formed passive layers and anodic layers on cp-Ti, Ti6A14V and Ti6A17Nb at different pH, a significantly lower donor density and a higher Faraday efficiency of oxide formation is found for Ti6A17Nb compared with the other materials. As such, in the case of an inflammatory response associated with a pH decrease in the surrounding tissue, a significant increase of the electronic conductivity is expected for Ti6A14V, but not for Ti6A17Nb, underlying the excellent biocompatibility of this alloy. Investigations of the corrosion resistance of both α/β alloys in 2M

H ₂ SO ₄ and 2M NaOH, the Ti6A17Nb alloy has shown a higher resistance in sulphuric acid but a lower resistance in sodium hydroxide. Using a multi-step chemical and thermal treatment, Ti6A17Nb oxide surfaces with good wettability, low ion release and moderate bone-like apatite induction ability have been prepared.

The chemical composition of anodic oxide films with respect to substrate alloying elements is of great importance for biomedical applications, since many alloying elements such as vanadium (V), aluminium (Al), nickel etc. are today recognized as undesired, harmful or even toxic (like V) elements in the body environment. There is an obvious need for a spark anodization process capable of decreasing the amount of alloying elements in the metal oxide layer. Only thin anodic oxides on Ti, Ti6A14V and Ti6A17Nb below the dielectric breakdown have been investigated. Based on XPS and SIMS depth profiles, a concentration level of typically 12-17 w-% Al, being significantly higher than expected from the bulk composition, was observed in native oxides on Ti6A14V and Ti6A17Nb. A SIMS depth profile through a Ti6A14V sample anodized in 1 M H2S04 at 45 V, well below the breakdown limit, showed a higher Al concentration in the oxide than in the metal, where V was depleted in the outermost surface region but enriched in the interfacial region [Lausmaa J. Mechanical, Thermal, Chemical and Electrochemical Surface Treatment of Titanium. In: Brunette DM, Tengvall P, Textor M, Thomsen P, editors. Titanium in Medicine. Berlin & Heidelberg: Springer; 2001. p. 248-2551. An accumulation of alloying elements in the metal oxide layer compared to the metal substrate for native or spark-less anodized oxides on Ti-alloys is to be avoided with respect to the biocompatibility of such oxide layers. Today, no information is available on the distribution of alloying elements within spark anodized thick oxide layers on e.g. Ti-alloys, which is problematic when considering the aforementioned efforts to incorporate bioactive species within a "biocompatible" matrix.

In light of the above, there is still a great deal of scope for producing bio-compatible materials, which will give an enhanced morphology for incorporation of bone-growth stimulating agents. Further, such surfaces should preferentially be formed with a reduction in the amount of chemicals toxic to the human body, whilst additionally being formed in a repeatable and straightforward manner.

Summary of the Invention

In order to address the problems as disclosed above, a material and method of production thereof is presented. The material is a metal alloy upon which an oxide layer is formed. The metal oxide layer and especially its outermost part contains significantly less alloying elements than are present in the metal alloy substrate. Preferentially, this oxide layer is continuous, and is partially intersected by multiple, narrow and elongate recesses in the surface. In another preferred embodiment, this metal oxide film, has a crystalline structure which is comprises a brookite.

Description of the Figures

Fig. 1. SEM micrographs of etched Ti (a), Ti6A17Nb (b) and Ti6A14V (c) surfaces showing the microstructures of the used substrates. The bright phase in (b) and (c) corresponds to the β phase.

Fig. 2. Current density and anodisation potential as a function of time during anodisation of Ti and Ti-alloys in (a) 0.5 M H ₂ SO ₄ and (b) 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ .

Fig. 3. SEM micrographs of selected anodisation conditions, (a) Ti / 0.5 M H ₂ SO ₄ / 165 V, (b) TiAlNb / 0.5M H ₂ SO ₄ / 170 V, (c) TiAlV / 0.5 M H ₂ SO ₄ / 135 V, (d) Ti / 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 185 V, (e) TiAlNb / 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 200 V, (f) TiAlV / 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 190 V. The worm-like topography in (e) and (f) is very different from the nano/micro- porous oxides in (b) and (c).

Fig. 4. FIB cross-sections of spark anodised Ti6A17Nb substrates, (a) Micro/nano-porous morphology in 0.5 M H ₂ SO ₄ at 170 V, (b) Worm-like oxide in 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ at l70 V.

Fig. 5. Raman laser spectra of the spark anodized surfaces shown in Fig. 3. (a) Ti / 0.5 M H ₂ SO ₄ /

165 V, (b) Ti / 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 185 V, (c) TiAlNb / 0.5M H ₂ SO ₄ / 170 V,

(d) TiAlNb / 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 200 V ₅ (e) TiAlV / 0.5 M H ₂ SO ₄ / 135 V, (f)

TiAlV / 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 190 V. Fig. 6. Quantitative GD-OES depth profile of a spark anodized Ti surface, (a) 0.5 M H ₂ SO ₄ / 165

V, (b) 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 185 V. The inserts show a zoom of the first 200 nm.

At the intercept of Ti and O signals (Ti/O= 1 : 1 ), an oxide thickness of (a) 2.04 μm and (b)

1.67 μm is obtained. Fig. 7. Quantitative GD-OES depth profile of a spark anodized Ti6A17Nb surface, (a) 0.5 M

H ₂ SO ₄ / 170 V, (b) 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 200 V. The inserts show a zoom of the first 200 nm. At the intercept of Ti and O signals (Ti/O=l:l), an oxide thickness of (a)

2.24 μm and (b) 3.14 μm is obtained. Fig. 8. Quantitative GD-OES depth profile of a spark anodized Ti6A14V surface, (a) 0.5 M

H ₂ SO ₄ / 135 V, (b) 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ / 190 V. The inserts show a zoom of the first 200 nm. At the intercept of Ti and O signals (Ti/O=l:l), an oxide thickness of (a)

2.15 μm and (b) 3.14 μm is obtained. Fig. 9. Recombined micrograph from 4-quadrant BSE detector imaging showing a sandblasted, chemically etched and finally spark anodized Ti6A17Nb surface (0.5 M H ₂ SO ₄ + 0.5 M

H ₃ PO ₄ / 200 V).

Description of the Preferred Embodiment

1. Sample preparation and discussion of techniques

The following sections discuss the choice of materials and preparation of such. The methods disclosed in the preparation and certain equipment used, is given purely as an example of the technique. It would be clear to the skilled person, that other equipment options and different methods for performing the experiments can be utilized to generate the same results, and that

these are considered as being covered by the following disclosure. Further discussion is presented of the specifics of certain tests performed on the samples for characterisation of its mechanical and physical properties.

1.1. Substrate preparation and morphological characterization

Metal discs (15 mm diameter, 1 mm thickness) made of c.p. Ti (commercially pure, ISO 5832-2, CH), Ti6A14V (ISO 5823-3, UK) as well as Ti6A17Nb (ISO 5832-11, CH) were used as substrates. Ti6A17Nb samples were machined and electro-eroded from hip implants, although this is presented purely as an example, samples of such a compound can be obtained by any normal manner. All samples were mechanically polished to 2000 grit followed by a chemical attack in a pickling solution containing (NH ₄ )HF ₂ , HNO ₃ and H ₂ O, removing the top 30 μm and any possible change induced by mechanical polishing. For combining macro- and micro- roughness, some samples were sandblasted and chemically etched (pickled) before anodizing.

In order to contrast the α / α-β microstructure of the substrates, a TiAlNb and TiAlV substrate, respectively, was prepared as described above, additionally etched in a solution containing HF and HCl (Petzov Ti-M4) for approximately 60s. A c.p. Ti sample was contrasted by short electropolishing in 3M H ₂ SO ₄ in methanol at 0°C.

1.2. Electrochemical spark anodisation

The following electrolytes were tested: 0. IM H ₂ SO ₄ (0.1 M HS), 0.5M H ₂ SO ₄ (0.5 M HS), 0.1 M H ₃ PO ₄ (0.1 M HP), 0.5M H ₃ PO ₄ (0.5 M HP), 0.1 M H ₂ SO ₄ + 0.5M H ₃ PO ₄ (0.1-0.5 M HSHP) and 0.5M H ₂ SO ₄ + 0.5 M H ₃ PO ₄ (0.5 M HSHP). Spark anodisations were performed at a starting temperature of 25 ⁰ C in a 1 L glass cell with water jacket, the temperature being controlled by a heater/cryostat circuit. A temperature increase up to 35°C was observed for the

conditions with the largest anodisation currents. The working electrode was fixed to a Teflon® holder, facing up and leaving a small recess (d=0.5 mm) on the exposed surface (A=I.539 cm ). A Labview controlled power source (6 kW, Xantrex XDC 300-20) working in a 2-electrode setup with a cylindrical titanium mesh as counter electrode was used. Anodisation was performed potentio-dynamically with 5 V/s from OV up to the anodisation potential δE _max , followed by 60s

hold time at δE _max . The resulting current was precisely read back over selectable shunt resistors.

Before anodisation, samples were thoroughly rinsed with ultra-high purity water (18 Mωcm) and

transferred wet into the anodisation cell. δE _max was systematically varied in order to find optimal conditions in terms of oxide topography and morphology, uniformity and mechanical properties (low abrasion, good adhesion). Anodised samples were copiously rinsed with high purity water and dried under nitrogen flow. Whilst the above is presented as one system for performing spark anodisation, it would be clear to the skilled person that other systems and experimental set-ups could be utilised for generating the same results.

1.3. Tensile adhesion test

Tensile adhesion tests were carried out following DIN EN 582 using HTK Ultra Bond 100 as glue (measured tensile strength 110 MPa), hardening during a heat treatment with a maximum at 210 °C. The substrates with the anodized surface were glued between 2 sandblasted cylindrical steel supports that were fixed to the Epprecht-Multitest tensile testing machine. The backsides of the samples were roughened by grit-blasting to avoid failure at the interface backside/glue during the test. For a valid test, the failure upon load must occur within the oxide (cohesive) or at the interface substrate/oxide (adhesive). It was verified that the fracture morphology was not influenced by possible infiltration of the glue into open pores in the oxides.

1.4. Miller particle shedding test

In this test, carried out following ASTM G75, the anodized samples with 15 mm diameter were glued to a steel support connected to the moving arms of the Miller test equipment. Within one cycle of the test, the sample, rubbing against Neopren with Shore-Hardness 3, is moving linearly approximately 25 cm forth and 25 cm back. The mass loss, being a measure for the abrasion resistance of the coatings, is measured as a function of applied load and cycle frequency for a given number of cycles. This test was performed during 1000 cycles at a specific load of 0.113 N/mm ² at 0.78 Hz.

2. Results of performing the above experiments

In the following sections, the results of performing the experiments according to section 2.2 detailed above, on the various samples disclosed in section 2.1 are presented.

2.1 Microstructure

The microstructures of the substrates are presented in Fig. 1. The hexagonal α grains in

the c.p. titanium have diameters between 10-20 μm and are not preferentially oriented. A clear

orientation of the darker hexagonal α and the brighter cubic β phase, due to the forging of the

TiAlNb hip implant material, is presented in Fig. Ib. The plate-like α/β microstructure of TiAlV has smaller grains than in the TiAlNb alloy with no clear orientation (Fig. Ic). By EDX analysis an average of 6.2 w-% Al + 4.1 w-% Nb in the α and 3.0 w-% Al + 22.3 w-% Nb in the β phase

of the TiAlNb alloy was found. In the TiAlV alloy, the α phase contained 5.1 w-% Al + 3.0 w-%

V, whereas the β phase contained 4.4 w-% Al + 10.1 w-% V + 1.2 w-% Fe. This demonstrates

the role of V and Nb as β stabilizing and Al as α stabilizing elements. The iron contamination

accumulated in the α phase of the Ti6A14V alloy can be explained by the tolerated maximum 0.3 w-% Fe in this alloy, according to ISO 5832-3 (1996).

2.2 Spark anodisation

The current density and the anodisation potential δE as a function of time for optimized anodisation conditions are shown in Fig. 2. In 0.5M H ₂ SO ₄ (Fig. 2a), the behaviour of TiAlV differed significantly from the other materials. Its anodisation current increased very rapidly at potentials above 100 V, reaching a maximum at E _max (135 V) and then only slowly decreasing to a remaining high value during the hold-time. Also Ti and TiAlNb showed the current maximum when reaching E _max . However, the current then decreased rapidly towards almost zero for Ti and towards a low plateau value for TiAlNb, which was most probably due to the oxidation of H ₂ O on top of the oxide. The anodisation of Ti and TiAlNb was found to be self-limiting in this electrolyte, whereas TiAlV continued to oxidize upon polarisation at E _max . At almost identical chosen anodisation potentials of 165V and 170 V for Ti and TiAlNb, respectively, Ti showed significantly lower currents. At low potentials however, Ti is the first substrate to exhibit sparking, followed by TiAlV and then TiAlNb, as can also be seen from the current increase in the initial stage of polarization. This behaviour at low potentials is identical upon anodisation in the electrolyte containing 0.5 M H ₂ SO ₄ + 0.5 M H ₃ PO ₃ (Fig. 2b). Sparking first appeared on Ti, then on TiAlV and TiAlNb. The maximum anodisation current for Ti, found before reaching the anodisation potential E _max , was even more significantly lower with respect to the alloys than in Fig. 2a, even so all three anodisation potentials were chosen quite close together. The maximum currents measured on the alloys were similar, TiAlV reaching a maximum when reaching E _max , whereas TiAlNb reproducibly exhibited a 2-peak behaviour with the maximum current shifted to the hold-time. All three substrates demonstrate a complete passivation within the hold-time.

In both electrolytes, the spark anodisation current was clearly larger for Ti-alloys than for c.p. Ti, which points towards a lower (electro-)chemical stability of the α/β microstructure compared to

the Ti-α phase. It is interesting that the addition of phosphoric acid to 0.5 M H ₂ SO ₄ resulted in lower anodisation currents in Fig. 2b compared to Fig. 2a, even so the anodisation potentials E _max were higher for the presented conditions in the mixture. Phosphoric acid seems to act as an "inhibitor" for spark anodisation of Ti-alloys.

2.3 Oxide morphology

The following sections provide details of the effects of different electrolytes on the structures of the samples.

2.3.1 H2SO ₄ containing electrolytes

In 0.1 M HS, a potential of at least 220 V is necessary to obtain oxides on Ti with uniformly distributed nano-pores, having pore diameters of around 100-200 nm. The pore size and simultaneously the roughness increases when going to higher potentials, leading to an increasing amount of protruding features and particles in the surface. In 0.5 M HS, the evolution of oxide morphology with growing potential was shifted to lower potentials compared to 0.1 M HS, due to the higher aggressiveness of this electrolyte. E.g., uniformly porous oxides on Ti appeared around 125-150 V. Generally, pores were found to be larger with higher electrolyte concentration. An example at 165 V is given in Fig. 3a. TiAlNb spark anodised in 0.1 M HS does not result in a homogeneous distribution of pores, even at potentials up to 250 V. Suitable morphologies are obtained in 0.5 M HS at potentials between 170-200 V, similar to the oxides on c.p. Ti (Fig. 3b). Further increasing the potential results in very rough and thick, but fragile and powdery looking surfaces. Spark anodisation of the TiAlV alloy in both sulphuric acid containing electrolytes behaves clearly differently from the other materials, as could be expected from the

current-potential behaviour in Fig. 2a. Around 135 V in 0.5 HS, an oxide with small nano-pores was obtained (Fig. 3c). This oxide has a powdery, uneven appearance with fragile outstanding features and its colour is bright brownish and very different from the dark grey of the other materials. At 135 V, the surface is inhomogeneous, and at higher potentials the powdery aspect and brownish colour is more pronounced.

2.3.2 H ₃ PO ₄ containing electrolytes

With applied voltages up to 300 V, hardly any sparking is observed in 0.1 M HP with all three materials. In 0.5 M HP, sparking at potentials above 250 V leads to larger pores than seen in 0.5 M HS. Pore density and pore distribution are inhomogeneous on the entire surface, leading to the conclusion that phosphoric acid alone does not provide satisfactory results.

2.3.3 H ₂ SO ₄ - HsP O ₄ containing electrolytes

The addition of phosphoric acid to sulphuric acid completely changes the oxide morphologies on all three materials compared with spark anodisation in H ₂ SO ₄ alone. For a given potential, the anodisation current is lower in the mixture than in the electrolyte containing only the sulphuric acid part, H ₃ PO ₄ acting as an inhibitor. Spark anodisation in 0.1-0.5 M HSHP results in similar oxides as obtained in 0.5 M HSHP, but again the lower acid concentration requires higher anodisation potentials and in general the final oxides morphology is less homogeneous. Spark anodisation of Ti in 0.5 M HSHP at 185 V leads to worm-like voids rather than pores (Fig. 3d). Above 200 V, the morphology changes to large pores with thick edges that seem to stand out of the surface, very similar to the TiUnite ^® surface [Hall J, Lausmaa J, WO0072776, WO0072777, WO0072775] on titanium. At 250 V, the pores have a diameter of approximately 5 μm. Almost identical oxide morphologies are found upon anodisation of both

Ti-alloys in 0.5 M HSHP. For TiAlNb, uniform worm-like structures ere found at potentials

around 200-225 V, becoming again more pore-like at 250 V. For TiAlV, the formation of worms begins at around 175 V, to become also more pore-like at 250 V. Ideal conditions are found at with E _max of 200 V for TiAlNb (Fig. 3e) and 190 V for TiAlV (Fig. 3f).

Based on morphological considerations, the following conditions leading to homogeneous and suitable morphologies are chosen: Ti / 0.5 M HS / 165 V, Ti / 0.5 M HSHP / 185 V, TiAlNb / 0.5 M HS / 170 V, TiAlNb / 0.5 M HSHP / 200 V, TiAlV / 0.5 M HSHP / 190 V.

2.3.4 FIB cross-sections

The cross-section of a focused ion beam (FIB) milled oxide on TiAlNb with a porous morphology, grown in 0.5 M HS at 170 V and corresponding to Fig. 3b, is presented in Fig. 4a. Fig. 4b shows the cross-section of an oxide with a worm-like morphology on the same substrate, but grown in 0.5 M HSHP at 200 V (Fig. 3e). An average thickness of 2.21 ± 0.15 μm for the

porous and 3.39 ± 0.72 μm for the worm-like oxide was measured. The coating in Fig. 4a is very similar to oxides currently found upon spark anodisation of c.p. Ti and contains closed pores within the oxide, whereas the surface pores do not reach very deep. This may be the reason for the superficial bone in-growth observed in in-vivo tests with anodized Ti implants having a similar porous morphology, not showing an advantage over chemically treated Ti surfaces at a later stage of implantation. The cross-section of the oxide in Fig. 4b reveals approximately 2 μm deep channels between the oxide "worms", the oxide itself being almost free of closed porosity. In contrast to Fig ¹ . 4a, the substrate was preferentially dissolved below the channels, leading to a significantly rougher interface metal/oxide compared to the porous oxide. The preferential dissolution of specific spots of the α/β microstructure in the presence of the acid mixture seems to be the reason for the observed oxide morphology. The oxide under the channels has a remaining thickness of 500-1000 nm. In spite of the dissolution of substrate material upon

anodisation, the anodisation current was smaller during formation of the worm-like oxide (Fig. 2a) than for the porous oxide on TiAlNb (Fig. 2b). Both cross-sections reveal some nano-porosity at the interface metal/oxide.

2.4 Raman structure

The Raman spectra of c.p. Ti, TiAlNb and TiAlV spark anodized under optimized conditions in 0.5 M HS and 0.5 M HSHP are shown in Fig. 5. Crystalline titania can exist in the three polymorphs anatase, rutile and brookite. The main vibrational modes of anatase monocrystals are located at 144, 399, 519 and 639 cm ^'1 , with 144 cm ^"1 being the strongest band. The Raman lines of monocrystalline rutile are found at 232, 447 and 612 cm ^"1 . An intense and very intense band at 128 and 153 cm ^"1 , a strong and medium band at 247 and 636 cm ^"1 as well as a shoulder peak at 396 cm ^"1 are reported for brookite.

The spectra in Fig. 5a for Ti anodized at 165 V in 0.5 M HS can be ascribed to a mixture of anatase and rutile, with anatase being the dominant phase. After spark anodisation of Ti at 185 V in the acid mixture, anatase structure (Fig. 5b) without any signs of rutile was observed. A strong rutile signature was found for TiAlNb oxidized at 170 V in 0.5 M HS (Fig. 5c). The spectra contains also anatase and brookite signals, with a strong brookite line at 157 cm ^"1 , being too far from 144 cm ^"1 to be ascribed to anatase. A frequency blue shift and linewidth increase compared with single crystals is generally explained by non-stoichiometry by oxygen deficiencies, disorder induced by minor phases or pressure and phonon confinement effects. Anatase is clearly present at 518 cm ^"1 , and the shoulders at approximately 396 and 638 cm ^"1 are due to anatase as well as brookite, having Raman active bands very close. Anodized in the acid mixture at 200 V (Fig. 5d), the brookite/anatase mixture becomes dominant, with a small amount of rutile still present. Very similar spectra as for TiAlNb are obtained for spark anodized TiAlV samples. Here, the rutile phase after anodisation in 0.5 M HS at 135 V is even more dominant,

mixed with signals from the anatase and brookite phase (Fig. 5e). In the mixture anodized at 190 V, the obtained spectrum is virtually identical to Fig. 5d, showing a mixed brookite/anatase structure containing a small amount of rutile.

It is well known that for a given electrolyte, an increase in the anodisation potential results in higher anodisation currents and more vigorous sparking and hence, to an increased local temperature, enabling the formation of anatase, brookite and finally rutile phase, the latter being the high temperature phase of TiO ₂ . The fact that the amount of rutile is clearly lower for anodized c.p. Ti than for the alloys in Fig. 5, can be explained by the considerably lower anodisation currents for Ti compared with the alloys (see Fig. 2) under the chosen conditions. The same argumentation can explain the difference in the amount of rutile between the alloys anodised in sulphuric acid or in the sulphuric/phosphoric acid mixture. Particularly interesting is the observation of the brookite phase within the coatings on Ti-alloys.

2.5 Chemical analysis with GD-OES

In Fig. 6, quantitative GD-OES depth profiles of c.p. Ti spark anodised under optimized conditions in 0.5 M HS (6a) and 0.5 M HSHP (6b) are shown, corresponding to the oxides in Fig. 3a and 3d and being a reference for the Ti-alloys. A stoichiometric oxide with Ti/O = 1 :2 with a rather sharp interface metal /oxide and a thickness of 2.04 μm, defined at the point where Ti/O = 1:1, is found in Fig. 6a. A carbon contamination of about 1 at-% throughout the oxide with an increase at the outer surface due to atmospheric contamination is found. The sulphur contamination of maximum 1.7 at-% is located mainly at the interface metal/oxide, possibly due to the mentioned interfacial nano-pores that might contain some electrolyte. However, no sulphur is detected in the first 200 nm of the oxide (insert in Fig. 6a). The sharp increase of oxygen and decrease of Ti towards the surface is systematically been observed and is most probably due to adsorbed humidity and OH-groups, which is also supported by an increase of the hydrogen signal

(not shown). The 1.67 μm thick oxide grown in the acid mixture in Fig. 6b shows a spreading of surface and interface signals, due to the larger roughness of the obtained oxide morphology (Fig. 3d). Sulphur and carbon contamination are very similar to Fig. 6a, and an additional incorporation of 6-8 at-% P throughout the oxide is present, increasing to 12 at-% in the outermost 100 nm, possibly due to chemisorbed phosphates as supported by the simultaneous increase of the oxygen content. The higher oxide roughness compared with Fig. 6a results in a peak enlargement for species present at the surface, as seen for P and C in Fig. 6b.

The oxide grown on Ti6A17Nb in 0.5 M HS in Fig. 7a, corresponding to Fig. 3b and 4a, shows the same sulphur and carbon contamination as seen for c.p. Ti (Fig. 6a). Its interface is sharp due to the small roughness of the nano-porous oxide morphology. The amount of the alloying elements Al and Nb is clearly diminished within the oxide with respect to the substrate composition, yet both elements are still detectable in the outermost 200 nm (see insert Fig. 7a). The insert shows that Al remains with around 2 at-% and then decreases rapidly at the surface. From GD-OES depth profiling, a thickness of 2.24 μm is obtained, which is in excellent

agreement with an average thickness of 2.21 ± 0.15 μm measured at 7 spots in Fig. 4a. GD-OES depth profiling on the TiAlNb substrate spark anodized in 0.5 M HSHP in Fig. 7b, leading to the worm-like morphology in Fig. 3e and Fig. 4b, shows smeared signals at the surface and the interface, due to their larger roughness. From the GD-OES profile, a thickness of 3.14 μm is obtained (Ti/O = 1:1), which is again in good agreement with the average thickness of 3.39 ± 0.72 μm measured at 7 spots in Fig. 4b, in spite of the large thickness variations. The decrease of the alloying elements in the oxide is even more significant than in Fig. 6a. Within the first 200 nm of the oxide (see insert in Fig. 7b), the Nb content is below 0.5 at-%, and also Al decreases continuously to reach <0.5 at-% at the surface. Moreover, a significant and reproducible decrease in S contamination at the interface (<0.5 at-%) is found in the presence of phosphoric acid,

possibly due to a lower interfacial porosity. A phosphorous content of about 8 at-% throughout the oxide is present, as seen already in Fig. 6b. This is not considered as contamination, but is regarded advantageous for the bioactivity of titanium oxides, especially when simultaneously present with Ca.

A sharp interface due to the good thickness uniformity thickness of the porous oxide (2.15 μm) and a decrease of the alloying elements Al and V, is measured for Ti6A14V anodized in 0.5 M HSHP (Fig. 8a). As observed before for Ti6A17Nb in Fig. 7a (insert), the Al content remains at 2-3 at-% before almost disappearing at the surface. Vanadium decreases to 0.5-1 at-% in the outer 200 nm of the oxide, but remains clearly detectable. The GD-OES depth profile of Ti6A14V spark anodized in the acid mixture (Fig. 8b) is very similar to Fig. 7b, with vanadium replacing niobium. Again the interfacial region appears larger than after anodisation in sulphuric acid, due to the enlarged interfacial roughness due to the local dissolution of the substrate upon spark anodisation. A thickness of 3.14 μm was found, identical to the oxide grown on TiAlNb in Fig. 7b. The aluminium concentration decreases throughout the oxide and is lower in the outer 200 nm compared with Fig. 8a. The vanadium content rapidly decreases within the oxide and is hardly detectable in the outermost oxide (see insert of Fig. 8b). This is of great interest for biomedical applications, where vanadium is considered toxic. The S contamination at the interface was significantly lower (< 1 at-%) compared with Fig. 8a, and P is present as seen for the other oxides grown in 0.5M HSHP. A small P peak before the interface metal/oxide, superposed to the phosphorous contained in the oxide and coinciding with the S contamination peak, supports the assumption of trapped electrolyte being present in a nano-porous zone close to the interface.

2.6. Mechanical properties

Tensile adhesion tests according to DIN EN 582 are summarized in Table 1 for certain anodisation conditions. The adhesion strength of 40-50 MPa for the porous oxide morphology on c.p. Ti and TiAlNb from anodisation in 0.5 M HS is slightly better than the almost identical values around 34 MPa measured for the worm-like morphology formed in 0.5M HSHP mainly on Ti-alloys. A mixed adhesive/cohesive failure is observed for all conditions. Good adhesion of the coatings is expected, since a layer formed by plasma chemical oxidation of the substrate grows from inside and outside and does not contain an artificial interface as deposited coatings do. An adhesive/cohesive failure and an adhesion strength of 40 MPa is found for spark-anodized coatings on Ti6A14V in (NaPO ₃ ) ₆ containing alkaline electrolytes, using a shear strength test. A value of 37 ± 3 MPa and a cohesive failure is known for Ti6A14V spark anodized in Na ₂ CO ₃ solution. For comparison, the minimal adhesion strength required for plasma-sprayed biomedical HA coatings according to ISO 13779-2 is 15 MPa, which is clearly lower than the present values. With respect to the large thickness variations in the worm-like oxide morphology on Ti-alloys (Fig. 4b), it is not surprising that the failure load for the mixed fracture is slightly lower compared to the porous morphology in Fig. 4a, where the perpendicular force is distributed more uniformly within the oxide. Good cohesion leading to low abrasion during implantation and during the life time of the implant, avoiding micron-sized particles that are difficult to phagocyte, is of fundamental importance. The Miller particle shedding test according to ASTM G75 allows for a quantitative investigation of the abrasion. Table 1 shows the mass loss of the present oxides in comparison to a pickled c.p. Ti reference surface and a plasma-sprayed Ti surface of 16 mm diameter (1 mm more than the other surfaces). All tested oxides perform at least four times better than the metallic c.p. Ti surface, leading to a larger mass loss during abrasion. The worm-like oxide on TiAlV shows particularly low abrasion, probably as a result of its hardness. The comparison to the approximately 250 μm thick commercial VPS-Ti surface (consisting of ~50

μm dense and -200 μm porous layer, 20-40 % open porosity, Rt = 150 μm) is even more dramatic, showing far better behaviour of the anodic coatings, probably also due to the smaller thickness.

2.7. Combination of grit-blasting and spark anodisation

A combination of micrometer and/or nanometre scale topographies on implantable orthopaedic devices is known to affect the cell behaviour in terms of cell adhesion, proliferation, differentiation, morphology, production of local factors etc. Spark anodisation was performed on grit-blasted substrates, followed by an additional pickling of the surface. In order to verify that the induced micro-roughness from grit-blasting is not significantly altered by the studied spark anodisation conditions due to a possible planarisation of the surface upon spark formation, the evolution of the relevant surface roughness parameters for c.p. Ti, TiAlNb and TiAlV after grit- blasting, chemical etching (pickling) and after spark anodisation was measured (Table 2). The pickling attack was found to reduce significantly the incorporated Al ₂ O ₃ contamination on the surface from grit-blasting. In order to have the same initial surface chemistry before anodisation, the pickling attack was applied to all substrates in this study. The commonly cited roughness value Ra is slightly reduced during spark anodisation of grit-blasted substrates for all anodisation conditions and substrates, and also Rz follows this trend, however, enough roughness is left. The induced micro/nano roughness from to the porous or worm-like oxide morphologies can obviously not be accounted for with the applied profϊlometric method, having a good z-resolution of about 50 nm but an insufficient lateral resolution of 1 μm. Fig. 9 represents a 4-quadrant BSE detector micrograph of a Ti6A17Nb surface after grit-blasting, pickling and anodisation in 0.5 M HSHP at 200 V, representing at least two levels of roughness.

3. Discussion of the above results

As seen in Fig. 6-8, all investigated spark anodized surfaces contain a lower concentration of alloying elements Al, V and Nb than present in the substrate. On c.p. Ti, qualitative GD-OES depth profiles of Ca/P containing spark anodised show a more or less constant distribution of these elements and a phosphorous peak at the surface, possibly from adsorbed phosphates similar to Fig. 6b-8b. Based on XPS spectra, it is suggested that sulphated and phosphated titanium oxides (e.g. TiSO ₄ , Ti ₂ (SO ₄ ) ₃ , TiPO ₄ ) are present. Based on the GD-OES analysis of the oxides, it is likely that sulphur is not reacting with titania and is solely present in the form of remaining electrolyte contamination in interfacial oxide pores, whereas P is mostly incorporated in the form of a phosphated titanium oxide. Based on XPS and SIMS depth profiles, a concentration level of typically 12-17 w-% Al, being significantly higher than expected from the bulk composition, is observed in native oxides on Tϊ6A14V and Ti6A17Nb. Many findings for native or spark-less anodized oxides on Ti-alloys differ from the present findings for anodisation above the breakdown limit, which underlines the obvious differences in the formation mechanism in the presence of sparks and the locally very high temperature.

The depth profiles in Fig. 7 and 8 show that the amount of the biomedically problematic alloying elements and sulphur contaminations in spark anodized Ti-alloys can further be decreased when working in a mixture of sulphuric and phosphoric acid. This is clearly correlated to the different electrochemical behaviour in Fig. 2 when using the acid mixture, resulting in the specific morphological characteristics in Fig. 3 and 4 and probably a smaller interfacial porosity and decreased incorporation of electrolyte residuals. An oxide morphology on Ti-alloys similar to Fig. 3e & f and Fig. 4b is quite different from the nano- or micro-porous oxides (Fig. 3a-c) often reported, and seems particularly interesting with respect to the large topographical features in the range of 1-20 μm as well as the lack of protruding features and sharp edges. The worm-like topography is less pronounced on c.p. Ti, where it is mixed with porous structures. Hence, the

particular and very similar oxide morphology forming on both Ti-alloys seems related to the presence of the alloying elements, but less to the α/β microstructure. A preferential dissolution of

the β phase of both Ti-alloys, being rich in V or Nb, would explain the measured decrease of Nb

and V, but not the decrease of Al being an α stabilizing element. Moreover, the preferential dissolution of one phase would lead to different oxides on Ti6A14V and Ti6A17Nb, due to the different substrate microstructure (Fig. Ib and Ic), which is in contradiction to the similarities in Fig. 3e and 3f. It is apparent that the higher anodisation currents for Ti-alloys in Fig. 2 are due to the purifying preferential dissolution of alloying elements, having an inferior electrochemical stability compared to TiO ₂ . It seems most probable that the obtained round oxide features are the result of an electrolyte-specific vortex generated by micro-convection phenomena at the surface due to gas and heat formation upon spark discharge.

4. Conclusions

Spark anodisation of the α/β Ti alloys Tϊ6A14V and Ti6A17Nb for biomedical surface preparation has been performed in various H ₂ SO ₄ , H ₃ PO ₄ and mixed acid containing electrolytes and is compared to spark anodized c.p. Ti (α). Higher anodisation currents for the alloys compared to Ti indicate a lower electrochemical stability. A novel oxide morphology on the Ti- alloys featuring a "worm-like" topography, open cavities in the μm-range and a roughened interface metal/oxide was found when working in mixed H ₂ SO ₄ and H ₃ PO ₄ containing electrolytes, the formation of this special morphology being less pronounced on α-Ti. The oxides

with a typical feature size of 1-20 μm are very different from the well-known and reported nano- or micro-porous morphology commonly found after spark anodisation. By GD-OES depth profiling, the sparking process is found to produce oxides with significantly lower amounts of alloying elements than present in the substrates, which is highly interesting for biomedical

applications. The worm-like oxides formed in the acid mixture show a particularly low amount of Al, V and Nb. The sulphur contamination from the electrolyte is strongly reduced in the presence OfH ₃ PO ₄ , while constant incorporation of around 8 at-% P throughout the oxide is measured. The particular morphological and chemical phenomena seem to be linked and could be the result of a specific solution vortex upon sparking in the acid mixture. From Raman, a mixture of anatase, rutile and brookite was identified in the coatings on Ti-alloys, where the latter has never been reported on spark anodized Ti materials. Bond strengths of 34 MPa for the worm-like and 40-50 MPa for the porous oxide morphology and mixed adhesive/cohesive failure are found. The oxides reveal an excellent abrasion resistance in a particle shedding test. It has further been shown that these anodic spark layers can readily be applied to grit-blasted and additionally etched surfaces, creating a surface topography with at least two levels of roughness. The described alloying element deficient oxides having a particular morphology suitable for incorporation of bone- growth stimulating agents are highly interesting for orthopaedic surfaces.