2019 Volume 60 Issue 3 Pages 422-428
Ti–xNb–ySn alloys with different Nb and Sn contents, of x = 24, 35, 42 mass% and y = 4, 7.9 mass%, were synthesised directly from TiO2, Nb2O5 and SnO2 mixtures via the FFC-Cambridge process. Compacted powder discs were employed as the cathode versus graphite as the anode in molten CaCl2 as the electrolyte. XRD analysis of the as-synthesised alloys showed that the two alloys with Nb content of 24 mass% were dual-phase α/β-Ti whereas the other four alloys with Nb contents of 35 and 42 mass% were single-phase β-Ti. SEM analysis showed that the alloys were highly porous, and that particle size decreased with increase in Nb and Sn contents. Alloy samples of each composition were subjected to short-term and long-term immersion tests in Hanks’ simulated body fluid solution. XPS studies then identified a passive oxide film on the surface of the alloy, and a hydroxyapatite layer on top of the oxide. Potentiodynamic polarisation studies revealed excellent corrosion resistance with very small corrosion current densities despite high open porosities. Furthermore, alloy samples were subjected to heat treatment in vacuum. Mechanical testing of these identified a substantial increase of elastic modulus and Vickers hardness. Overall, the experimental programme has brought out that the properties of the Ti–Nb–Sn alloys prepared are influenced markedly by their Nb and Sn contents, and that single-phase β-Ti–35Nb–4Sn holds promise as a candidate material for body implant applications.
Ti-based alloys are becoming increasingly attractive in the medical field as implant materials because of their superior biocompatible, corrosion-resistant and mechanical properties when compared with traditional implant materials such as stainless steels and Co–Cr alloys.1–5) Ti–6Al–4V has been widely employed in the past, but recent studies have now shown that Al and V are deleterious to human body cells and may lead to various diseases.5) Therefore, Ti alloys are being developed at present that contain exclusively elements that are considered non-cytotoxic, such as Nb, Ta, Zr and Mo.1–6)
The key mechanical parameter of implant materials is the elastic, or Young’s, modulus. Metallic materials are conventionally made by arc melting and casting, which renders them dense and stiff. The elastic modulus of dense α-phase Ti alloys is typically in the range of 50–120 GPa,1–3,5,6) whereas that of bone is much lower in the range of 0.1–20 GPa.7) This mismatch is detrimental, because it leads to loosening of the implant from the bone over the course of time due to non-uniform transfer of load at the implant-bone junction.8) The β-phase of Ti has a lower elastic modulus than the α-phase.1,3,5,6) Consequently, a currently pursued approach of lowering the elastic modulus of Ti alloys consists in the addition of major amounts of non-cytotoxic β-stabilising elements so as to attain dual-phase α/β-Ti or single-phase β-Ti compositions. Nb seems to have received most attention as the alloying element thus far. An additional problem is that there is a strong tendency in some Ti alloys towards the precipitation of the ω-phase which then increases the elastic modulus. In 2004, Ozaki et al.9) demonstrated that this can be avoided through the addition of small quantities of Sn. In 2005, Matsumoto et al.10) synthesised Ti–35Nb–4Sn and Ti–35Nb–7.9Sn alloys by arc melting and casting, followed by diverse postprocessing procedures including homogenisation, solution treatment, quenching and cold rolling, and they achieved an elastic modulus of 43 GPa. Since then, many more studies have described the preparation of Ti–Nb–Sn alloys with a composition of or near Ti–35Nb–4Sn by arc melting and postprocessing in various ways.11–18) All studies have reported low elastic moduli, with the lowest one reached to date being 36 GPa,15) which is much below that of most other Ti alloys, albeit still above that of bone.
Surface morphology is another key parameter of implant materials. This is because a suitable degree of porosity facilitates cell growth from the outside to the inside of the implant and thereby enables efficient osseointegration.3–5) Porous metal bodies can be made by mechanical milling and sintering,19,20) but this process is difficult to control, does not easily lead to homogeneous mixing at the atomic level, and is sensitive to oxidation.
Molten salt methods like the FFC-Cambridge process21,22) and the OS process23,24) have been gaining importance in recent years for the synthesis of porous Ti alloys. In both processes, the oxide mixture to be reduced to the target alloy is made the cathode versus a graphite anode in a CaCl2 molten salt electrolyte. In the former process, the potential is controlled such that no Ca deposition occurs and direct oxygen ionisation becomes the dominating reduction process, while in the latter process, the potential is controlled such that Ca deposition occurs and CaO formation becomes the prevailing process. Both processes have already been used for the preparation of porous metals for the biomedical sector. Peng et al.25) and Yu et al.26) synthesised, respectively, Ti–Zr and Ti–13Nb–13Zr alloys with low elastic modulus by means of the FFC-Cambridge process, while Osaki et al.27) synthesised Ti–29Nb–13Ta–4.6Zr alloy by the OS process.
The purpose of the present study has been the preparation of porous Ti–xNb–ySn alloy bodies with varying contents of Nb and Sn (x = 24, 35, 42 and y = 4, 7.9) directly from sintered TiO2–Nb2O5–SnO2 mixed oxide precursors via the FFC-Cambridge process, and their evaluation in terms of crystallographic composition, morphology, corrosion behaviour and mechanical performance.
Powders of TiO2, Nb2O5 and SnO2 were weighed in the required amounts to yield the final compositions of Ti–24Nb–4Sn, Ti–35Nb–4Sn, Ti–42Nb–4Sn, Ti–24Nb–7.9Sn, Ti–35Nb–7.9Sn, and Ti–42Nb–7.9Sn (numbers in mass%). The powders were mixed, added into isopropanol containing 1 mass% of polyethylene glycol binder and 0.5 mass% of polyvinyl alcohol plasticiser, ground with pestle and mortar, and dried in an oven. Grinding and drying were repeated several times to achieve uniform mixtures. The dried oxide powders were uniaxially pressed into discs of mass 3 g and diameter 25 mm, and these were sintered at 1223 K in air for 3 h. Anhydrous CaCl2 salt was prepared from CaCl2·2H2O (98+%, Sigma Aldrich) by first drying in air for 48 h and then in vacuum for 24 h, both at 443 K.
For each experiment, 500 g of molten CaCl2 was kept in an alumina crucible that was located inside an argon-flushed Inconel retort at 1173 K. Pre-electrolysis of the CaCl2 melt, to remove redox-active impurities, was performed by applying a DC voltage of 2.8 V for typically 3–12 h between a Ni coil cathode and a graphite rod anode. Electro-deoxidation was then carried out by applying a DC voltage of 3.1 V for 22 h between a sintered TiO2–Nb2O5–SnO2 disc, tied with a Ni wire, as the cathode and the same graphite rod as the anode. After the experiment, the processed cathode sample was taken out of the melt and cooled to room temperature, and then thoroughly washed with distilled water and dried in vacuum.
The crystallographic phases present in the precursor oxide samples and in the processed alloy samples were identified by means of X-ray diffraction analysis (XRD) (Philips PW 1830). The microstructures and local elemental compositions of the samples were examined by low-resolution scanning electron microscopy (SEM) in conjunction with energy-dispersive X-ray spectroscopy (EDX) (JEOL JSM-5800LV). In selected cases, high-resolution SEM (Nova-Nano SEM 450) and EDX (Bruker X Flash 6I100) were employed. Open porosities were measured by impregnation in water using Archimedes’ method. Alloy samples of each composition were polished and immersed in Hanks’ simulated body fluid solution (HS)28) for 1 h, 7 days or 15 days. Their surfaces were after that analysed by SEM, EDX and X-ray photoelectron spectroscopy (XPS) (Thermo VG K-alpha). The solution was replaced every 48 h throughout the long-term tests to preclude any possible effects due to its slow degradation.
The corrosion properties of the alloys prepared were evaluated in HS at a pH of 7.4 and a temperature of 310 K by means of potentiodynamic polarisation studies. In the electrochemical cell, the alloy was employed as the working electrode and polarised by means of a potentiostat (Autolab PGSTAT-030 with GPES software) versus a Ag/AgCl reference electrode within the potential range from −0.8 V to 3.0 V, using Pt foil as the counter electrode. From this, the corrosion current density, icorr, the corrosion potential, Ecorr, and the polarisation resistance, Rp, were determined.
The mechanical properties of the alloys were assessed by determining the elastic modulus and the Vickers hardness. For elastic modulus measurements, the alloy samples were cut into rectangular bars, and tests were then carried out with a 4-point bending tester (Tinius Olsen H5KS) according to ASTM D790-10.29) For hardness measurements, the alloy samples were sectioned, hot-mounted in resin and polished up to 1200 grit, and indentations were then created along the cross-sections with a Vickers hardness tester (Laizhou Huayin 200HV-5) using a load of 200 g for a dwell of 15 s and averaging at least 10 individual tests. The indentations were imaged with a metallurgical optical microscope (Olympus BHM).
The alloys were subjected to a specific heat treatment. For this, the alloy samples were sealed in quartz tubes in vacuum, heated at 1473 K for 3 h, and cooled to room temperature at a rate of 5 K/min or less. Thereafter porosity and mechanical properties were re-measured.
Figure 1(a) presents the XRD patterns of the mixed oxide discs sintered at 1223 K in air. All samples contained TiO2 in the rutile form as the major phase and TiNb2O7 as a second phase. The intensity of the TiNb2O7 peaks increased with increasing Nb2O5 content. The TiNb2O7 originated from the chemical reaction of TiO2 and Nb2O5 at elevated temperature. No peaks arising from SnO2 could be seen. This is because SnO2 with its rutile-like tetragonal structure has extensive solubility in solid rutile TiO2.30)
(a) XRD patterns of the mixed oxide discs sintered at 1223 K in air for 3 h and used as the precursors to (i) Ti–24Nb–7.9Sn, (ii) Ti–35Nb–7.9Sn, (iii) Ti–42Nb–7.9Sn, (iv) Ti–24Nb–4Sn, (v) Ti–35Nb–4Sn, and (vi) Ti–42Nb–4Sn. (b) SEM images of the cross-sections of the same mixed oxides (MOs). (c) Current vs time curves obtained during electro-deoxidation of the mixed oxide discs in CaCl2 melt at 3.1 V and 1173 K for 22 h. Insets are photographs of a typical oxide disc before electro-deoxidation and of various alloy discs after electro-deoxidation. (d) XRD patterns of the as-prepared Ti–xNb–ySn alloys (x = 24, 35, 42 and y = 4, 7.9) in the same order as in (a). (e) SEM images of the cross-sections of the same alloy samples in the same order as in (b). (f) EDX area maps for Ti, Nb and Sn from polished surface of Ti–42Nb–4Sn alloy.
Figure 1(b) displays SEM images of the cross-sections of the sintered oxide discs. The microstructures were characterised by individual oxide particles with size in the range of 0.15–3.5 µm. High-resolution images further revealed that, at lower Nb2O5 contents, the particle size was at 0.4–3.5 µm irrespective of the SnO2 content. At higher Nb2O5 contents, the particle size was smaller, at 0.15–2.0 µm, for higher SnO2 contents and larger, at 0.2–2.5 µm, for lower SnO2 contents. This may be attributed to the different sintering behaviours of the TiO2–SnO2 solid solution and the TiNb2O7. In overall, particle size of the mixed oxides decreased with increase in Nb2O5 and SnO2 contents. The microstructures contained significant degrees of open porosity. This was measured to be in the range of 28–35% for the oxide precursors to Ti–xNb–4Sn (x = 24, 35, 42) and 23–31% for the oxide precursors to Ti–xNb–7.9Sn (x = 24, 35, 42), and was found to increase with increasing Nb2O5 content and decreasing SnO2 content.
The current versus time curves recorded during the electro-deoxidation of the mixed oxide discs are given in Fig. 1(c). The curves show the typical behaviour, in that there was an initial peak which was followed by a gradual decline towards a small constant value. The initial current peak occurs because Ca is inserted at a fast rate into the oxide cathode, through the reaction with Ca2+ ions from the electrolyte and electrons from the negative terminal, so that Ca metallates and metal suboxides are formed simultaneously. This has been investigated in detail in studies on the reduction of the pure oxides, and it is known that the main transient phases in the TiO2 and Nb2O5 systems include, respectively, CaTiO3, Ti3O5, Ti2O3 and TiO,31) as well as CaNb2O6, NbO2 and NbO.32) Once the reactions involving the uptake of Ca are exhausted, the governing reaction is the removal of oxygen from these compounds, and this is associated with the gradual decline of the current. All this will be similar in the reduction of mixed oxides. Upon completion of electro-deoxidation, the originally white-coloured mixed oxide discs had been converted into metallic discs with black coatings. After polishing, a shining metallic lustre was visible, as is seen in the inset of Fig. 1(c).
Figure 1(d) presents the XRD patterns of the as-reduced metallic discs. For the Ti–24Nb–4Sn and Ti–24Nb–7.9Sn alloys, i.e., those with Nb content of 24 mass%, both the α-phase and the β-phase of Ti were observed. For the other alloys, i.e., those with Nb contents of 35 and 42 mass%, only the β-phase of Ti was found. Given that Nb is β-stabilising and that Sn is neutral,33) the Ti–Nb phase diagram34) predicts that the minimum amount of Nb required for the exclusive formation of the β-phase is 29 mass%. The experimental results are therefore in accordance with expectations and suggest that the Nb plays the crucial role in the phase selection in the given type of alloy. In line with the present study, Matsumoto et al.10) detected the β-phase alone in Ti–35Nb–4Sn and Ti–35Nb–7.9Sn alloys made by arc-melting and homogenised at 1423 or 1223 K, whereas Praveen et al.18) found an α/β-phase mix in Ti–35Nb–4Sn alloy made by arc-melting and homogenised at 1273 K.
Figure 1(e) displays SEM images of the cross-sections of the as-reduced metallic discs. The microstructures were characterised by interconnected particles with size in the range of 0.4–10 µm. These were considerably bigger than those in the original oxides due to in-situ sintering during and after electro-deoxidation. High-resolution images further revealed that the average particle size decreased with increasing Nb content and with increasing Sn content. Porosity was measured to be in the range of 41–50% and increased with increasing Nb content.
EDX analysis of the as-reduced alloys showed only Ti, Nb and Sn and no oxygen. Quantitative results for the alloys are compiled in Table 1. These indicate that the compositions of the alloys were close to their target values. EDX elemental maps for Ti, Nb and Sn taken from a polished surface of an as-reduced Ti–42Nb–4Sn alloy sample are given in Fig. 1(f). The images confirm the uniform distribution of the three elements without any enrichment, depletion or segregation. It is worthwhile noting that the target values for Sn were reached very closely. Sn is much nobler than Nb and Ti, with its oxide decomposition potential being 0.58 V positive to that of Nb at 1173 K.35) It is therefore likely that Sn was released first from the oxide and, since it is liquid at the operating temperature, remained inside the solid matrix without flowing, or being leached, into the molten salt electrolyte until it could finally alloy with the subsequently released Nb and then the Ti.
Figures 2(a), (b), (c) display SEM images of the surface of a Ti–42Nb–4Sn alloy sample before its immersion in HS and after its exposure for 7 and 15 days, respectively. The corresponding EDX analyses show Ti, Nb and Sn in all three cases, as well as additional Ca, P and O in the latter two cases. Figure 2(d) presents an XPS survey spectrum from the surface of the Ti–42Nb–4Sn alloy after 15 days of immersion in HS. The spectrum displays the peaks for Ti 2p, Nb 3d, Sn 3d, Ca 2p, P 2p and O 1s. Figures 2(e)–(j) present additional high-resolution XPS spectra. These show that the characteristic peaks observed at 458.4 and 464.2 eV for Ti, 206.8 and 209.6 eV for Nb, and 486.6 and 495.3 eV for Sn, were all in close agreement with those reported in the literature for oxidic environments.36,37) They also show that the peaks at 347.2 and 351.0 eV for Ca, 133.2 and 134.5 eV for P, and 529.8, 531.7 and 533.2 eV for O, were all in agreement with those reported for hydroxyapatite, Ca10(PO4)6(OH)2, on the surface of Ti alloy.38) It can hence be concluded that, during immersion of the alloy in HS, a surface film of the oxides TiO2, Nb2O5 and SnO2 developed on the metal bulk, and that a layer of Ca10(PO4)6(OH)2 grew on top of the oxide film. Because of the similarity in composition, such results are also expected for the other alloy compositions. Overall these findings are very important, because the oxide film passivates the metal surface and the hydroxyapatite film permits cell growth, both of which are vital for the use of alloys as biomedical implant materials.
SEM images of the polished surface of Ti–42Nb–4Sn alloy (a) before immersion, (b) after immersion for 7 days, and (c) after immersion for 15 days in Hanks’ simulated body fluid solution at 310 K. Insets are EDX analyses of the surfaces. (d) XPS survey spectrum of the surface of Ti–42Nb–4Sn alloy after immersion in Hanks’ solution for 15 days. (e) to (j) XPS high-resolution spectra of Ti 2p, Nb 3d, Sn 3d, Ca 2p, P 2p and O 1s. Potentiodynamic polarisation curves of (k) Ti–xNb–7.9Sn and (l) Ti–xNb–4Sn alloys (x = 24, 35, 42) after 1 h of immersion in Hanks’ solution and (m) Ti–42Nb–7.9Sn alloy after immersion in Hanks’ solution for 1 h, 7 days and 15 days.
Figures 2(k), (l) show the current-voltage curves obtained during the potentiodynamic polarisation of the Ti–xNb–4Sn and Ti–xNb–7.9Sn alloys (x = 24, 35, 42) in HS at 310 K. The Ecorr and icorr values were of the same order of magnitude for all six alloys. When polarising the alloys away from their corrosion potentials in the more positive direction, in each case the current density increased slightly, before it stabilised up to the potential of 3.0 V versus Ag/AgCl. This indicated that passivation of the alloys took place and that no breakdown of the protective oxide films occurred. Figure 2(m) presents the current-voltage curves for the Ti–42Nb–7.9Sn alloy after different durations of immersion. Closer inspection revealed a small decrease in icorr with time, pointing to the slow growth of the surface film. All values obtained from the various potentiodynamic polarisation curves for Ecorr, icorr and Rp are compiled in Table 2. The icorr values were in the range of 2–9 µA/cm2 and indicated that all six alloys exhibited excellent corrosion resistance in HS even after long immersion times. In view of the inevitable scatter of the results, it is not possible to give preference to one specific candidate material based on the corrosion studies alone. There are as yet no studies in the literature that have investigated the corrosion behaviour of Ti–Nb–Sn alloys of the same compositions as those prepared in the present work. However, alloys of similar composition, made by arc melting, have likewise demonstrated good passivation behaviour in artificial physiological solutions.14,17,37,39)
Figure 3(a) presents the force versus displacement curves obtained in the 4-point bending tests of the alloy bars in the as-prepared state. Each alloy gave a linear response until reaching the breakdown point. Similar responses were also recorded with the alloys in the heat-treated state (not shown). All elastic moduli measured are depicted graphically in Figs. 3(b), (c) and summarised in Table 3. The data show that the moduli of the as-prepared samples were all of a similar order of magnitude, ranging within 5–10 GPa, while those of the heat-treated alloys were approximately twice as large. The key result of the bending tests therefore is that the elastic moduli of all samples were within or near the bone-matching range. Since the elastic modulus is a complex function of several properties, including chemical composition, phase make-up, porosity and microstructure, of which the latter can be further modified by heat treatment, it is not possible to derive any specific correlations between the various parameters involved.
(a) Force vs displacement curves of 4-point bending tests of Ti–xNb–ySn alloys (x = 24, 35, 42 and y = 4, 7.9) in the as-prepared state. Elastic modulus of the same alloys (b) in the as-prepared state and (c) in the heat-treated state (1473 K in vacuum for 3 h). Vickers hardness of the alloys (d) in the as-prepared state and (e) in the heat-treated state. Optical micrographs of Vickers indentations on the alloys (f–k) in the as-prepared state and (l–n) in the heat-treated state. Inset in (c) is a photograph of a vacuum-sealed alloy sample, inset in (d) is a photograph of a resin-mounted polished alloy sample.
As already mentioned, it has not been possible yet to reach values lower than 36 GPa for the elastic modulus of arc-melted Ti–Nb–Sn alloys. The reason for the much smaller values found in the present study is the difference in porosity. Arc-melted samples are virtually dense while the electrochemically prepared ones are highly porous. The impact of porosity also explains the increase of the elastic modulus of the present samples after heat treatment because this leads to partial densification and thus lower porosities, as is evident from Table 3. It is noted that XRD analysis of the alloys showed that phase composition stayed the same after the heat treatment, i.e., α/β-Ti phases for the low Nb content and β-Ti phase for the higher Nb contents.
Figure 3(d) presents the results of the hardness tests for the alloys in the as-prepared state. Hardness values were rather low, ranging within 28–125 HV. This is again a consequence of the high porosity of the samples. In accordance with expectations, the hardness of the dual-phase α/β-Ti alloys was higher than that of the single-phase β-Ti alloys. Figure 3(e) shows the results of the hardness tests for the alloys in the heat-treated state. These are significantly higher, ranging within 185–335 HV. This may again be ascribed to the lower porosity of the samples after heat treatment. All measured hardness values are summarised in Table 3. Figures 3(f)–(n) display light optical images of Vickers indentations created in the hardness tests. The images show that the indentations were well defined without crack formation at the edges and corners. They furthermore illustrate, in line with the above findings, that the size of the indentations increased with increasing Nb content and decreasing Sn content, and that the indentations on the heat-treated samples were smaller than those on the as-prepared samples of the same compositions. Some hardness values have been reported in the literature for practically dense Ti–35Nb–4Sn alloy samples. The hardness of arc-melted and homogenised α/β-phase Ti–35Nb–4Sn alloy was measured as 404 ± 11 HV for the α-phase and as 715 ± 15 HV for the β-phase,18) and the hardness of arc-melted Ti–30Nb–ySn alloy (y = 2, 5, 8) was found to range within 455–559 HV.40) These high values are again due to low porosity.
Porous Ti–xNb–ySn alloys (x = 24, 35, 42 and y = 4, 7.9) have been successfully synthesised from TiO2–Nb2O5–SnO2 mixed oxide precursor discs by electro-deoxidation in molten CaCl2. The alloys with the low Nb content were dual-phase α/β-Ti, whereas those with the higher Nb contents were single-phase β-Ti. Microstructural studies showed that all alloys had nodular microstructures with considerable porosity. The chemical compositions of the alloys were confirmed to be close to the target values. Immersion of the alloy samples in Hanks’ simulated body fluid solution gave rise to the formation of an oxide film on the alloy surface, and of a hydroxyapatite layer on top of the oxide film, which is imperative for the use of these alloys in biomedical applications.
Corrosion tests in Hanks’ solution proved the excellent corrosion resistance of the alloys with very low corrosion current densities of 2–9 µA/cm2 despite high open porosities. Mechanical testing showed that the elastic moduli of all samples were similar to those of bone. The elastic modulus and the Vickers hardness could both be increased through suitable heat treatment in vacuum. Overall, the corrosion properties and the elastic moduli of all six alloys prepared were appropriate, and with hardness and strength scaling nearly linearly, it would be expedient to select the alloy with the highest hardness, i.e., phase-pure β-Ti–35Nb–4Sn, for further optimisation and trials in the biomedical sector.
This study was funded through the Research Chair Grant Program of The Research Council of the Sultanate of Oman.
