1 Introduction

Titanium-based alloys are one of the most important alloys in orthopaedics, and there have been numerous attempts to further improve their properties, especially those related to elastic modulus, corrosion resistance and biocompatibility. Vanadium and aluminium have been replaced by more bio-acceptable metals like zirconium, niobium and tantalum, with the aim of improving the biocompatibility of these alloys. The latter elements affect the microstructure of titanium alloys by acting as β-stabilizers and lowering the elastic modulus. Further, these valve metals are highly corrosion resistant, forming protective oxide layers under various conditions of pH, solution composition and temperature. Studies devoted to alloying titanium with these metals to develop various binary, ternary and quaternary alloys have intensified in the last decade. Different types of oxides can be produced by anodic oxidation of binary alloys containing Zr or Nb. Anodic oxidation of Ti–40Nb alloy in 5 M NaOH resulted in the formation of oxide layers composed of TiO2 and Nb2O5 with 3D nano-porous structure [1]. Compared to pure TiO2, the elastic modulus of this mixed oxide layer is decreased, to 53.6 GPa, due to the incorporation of niobium oxide. The structure of oxide films formed in Ti–Zr alloys depends strongly on alloy composition [2]. Amorphous anodic films grew at voltages above 40 V on Ti–23Zr and Ti–42Zr, while further increase in Zr content resulted in the formation of an anodic film containing a crystalline monoclinic ZrO2 phase. In chloride-containing simulated physiological solutions the binary alloys are less resistant than the ternary alloys, thus the Ti–13Nb–13Zr alloy remained corrosion resistant, but the binary Ti–50Zr alloy was highly susceptible to pitting corrosion under these conditions [3]. Of the ternary alloys, Ti–13Nb–13Zr alloy exhibited lower passivation current density than Ti–6Al–7Nb and Ti–15Zr–4Nb alloys [4]. Further, Khan et al. [5] have suggested that proteins interact with the repassivation process to a greater extent for Ti–6Al–4V and Ti–6Al–7Nb than for Ti–13Nb–13Zr.

The corrosion resistance of quaternary Ti-based alloys is reported to be better than that of commercial ternary alloys [6] and of binary Ti–15Mo alloy [7]. The native oxide layer formed spontaneously at the surface of ternary and quaternary alloys contained TiO2 as a major oxide component, together with oxides of other metal components of the alloy that are often enriched at the surface relative to their bulk content [8–11]. An approximately fivefold increase in Zr concentration was observed for the Ti–13Nb–13Zr and Ti–15Nb–4Zr alloys at the outermost surface [9]. The native layer was a mixture of different oxides in different oxidation states [10].

Although immersion tests provide a relevant basis for predicting long-term behaviour of metallic materials in vivo, little has been reported for ternary and quaternary alloys. Okazaki and Gotoh [12] reported the concentrations of metal ions released from Ti–6Al–4V, Ti–6Al–7Nb and Ti–15Zr–5Nb–4Ta alloys during a 1 week’s immersion in various solutions. The quantities of Ti released here from the Ti–15Zr–5Nb–4Ta alloy were ~30 % of those released from the Ti–6Al–4V alloy. The total quantity of Zr, Nb and Ta was much smaller than that of elements released from the Ti–6Al–4V and Ti–6Al–7Nb alloys. No similar study has been reported for other ternary or quaternary alloys intended for orthopaedic applications.

Biocompatibility is another important property to be considered when searching for new alloys. In terms of toxicity of, refractory metals like Ti, Nb,Ta and Zr are found to have good biocompatibility and osteoconductivity [13–15] with no known adverse effect on human [16, 17]. The Ti–Nb–Zr–Ta alloys were reported to exhibit comparable cell proliferation but greater cell differentiation than Ti–6Al–4V ELI alloy [6]. Proliferation of the osteoblast-like cell MG63 was reported to be lower on Ti–30Ta than on Ti–30Nb and Ti–6Al–4V; however, it is not clear whether the difference was statistically significant [18]. Initial attachment of mouse osteoblast cells was significantly higher on the Ti–50Zr alloy, which had the lowest roughness, than on Ti and Ti–50Nb, making it superior to the other two materials [19]. It appears that minor differences in surface roughness and/or surface energy regulate the degree of cell adherence. On the other hand, no significant differences in terms of morphology, adherence, viability, or cell growth were observed between Ti and Ti–5Ta alloys [20]. The differences in biocompatibility between the various titanium alloys are probably subtle and should be investigated further in order to identify the effects originating from alloying elements.

The electrochemical behaviour and microstructure and mechanical properties of Ti–20Nb–10Zr–5Ta (TNZT) alloy have been studied [21, 22]. The improved passivity characteristics of the TNZT alloy compared to Ti and commercial Ti alloys was ascribed to the incorporation of alloying oxides of Nb, Zr and Ta [21]. Furthermore, with low Young’s modulus of 59 GPa and ultimate tensile strength of 883 MPa this material exhibits better functional properties than Ti [22]. In the present work we have studied the behaviour of this alloy during 25 days immersion in simulated physiological solutions. An integrated approach was employed to study the film formation, dissolution and biocompatibility of the alloy surface exposed to Hanks’ simulated physiological solution (HS) and minimum essential medium (MEM). While HS is broadly accepted inorganic solution which simulates physiological conditions, MEM contains inorganic and organic compounds (vitamins, sugars and amino acids). Formation was studied by analyzing the composition and morphology of the layer formed at the alloy surface during immersion, using X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM) combined with energy dispersion X-ray spectrometry (EDS). Dissolution was followed by measuring the quantities of dissolved metal ions, using inductively coupled plasma mass spectrometry (ICP-MS). Finally, biocompatibility was assessed using two human cell lines, analyzed in terms of cell survival, induction of apoptosis, growth delay and alkaline phosphatase activity.

2 Experimental part

2.1 Materials

Ti–20Nb–10Zr–5Ta (composition in wt%) alloy (denoted as TNZT) was obtained as ingots by semi-levitation melting in a high frequency induction furnace with a cold crucible, type FiveCell. The alloy was synthesized in two steps, melting and re-melting at 2,000 °C, with cooling stages inside the furnace [22]. Comparative measurements were performed on titanium (purity 99.6 %, annealed) supplied by Goodfellow (Cambridge Ltd., UK) in the shape of a 2 mm-thick foil. Samples were cut from the ingot or foil in the shape of discs of 15 mm diameter. Prior to immersion, tests samples were mechanically ground under water with 320- up to 4.000-grit SiC papers, cleaned with ethanol in an ultrasonic bath for two minutes, double-rinsed with Milli-Q water, and finally dried in a stream of nitrogen.

2.2 Immersion test

Samples of titanium and TNZT alloy were immersed in 100 mL glass containers containing simulated Hanks’ physiological solution (HS) or MEM. HS comprised 8 g/L NaCl, 0.4 g/L KCl, 0.25 g/L NaH2PO4·2H2O, 0.35 g/L NaHCO3, 0.06 g/L Na2HPO4·2H2O, 0.19 g/L CaCl2·2H2O, 0.4 g/L MgCl2·6H2O and 0.06 g/L MgSO4·7H2O. The pH was adjusted to 7.4 with HCl or NaOH (1 M) solution. MEM containing inorganic salts (6.8 g/L NaCl, 0.4 g/L KCl, 0.158 g/L NaH2PO4·2H2O, 0.264 g/L CaCl2·2H2O, 0.2 g/L MgSO4·7H2O and 2.2 g/L NaHCO3), various vitamins, amino acids and other compounds (dextrose, sodium bicarbonate and phenol red sodium salt) was supplied by Advanced MEM (Gibco®, Carlsbad, CA, USA).

Containers were placed in a thermostat bath maintained at 37 ± 0.1 °C. The immersion test lasted for 25 days. Each experiment was performed in duplicate. Glass containers with only MEM served as blank samples where needed. The experimental conditions were as for the test samples.

2.3 X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) was performed with a TFA Physical Electronics Inc. spectrometer using non- and mono-chromatized Al Kα radiation (1,486.6 eV) and a hemispherical analyzer. The mono-chromatized radiation used for high-resolution spectra yields a resolution of 0.6 eV, as measured on an Ag 3d5/2 peak. These spectra were used to differentiate between various species, whereas the spectra obtained using non-monochromatized radiation were used for quantifying the chemical composition. The take-off angle used, defined as the angle of emission relative to the surface, was 45°. The energy resolution was 0.5 eV. Survey scan spectra were recorded at a pass energy of 187.85 eV, and individual high-resolution spectra at a pass energy of 23.5 eV with an energy step of 0.1 eV. After taking the surface spectra, depth profiling of the oxidized layers was performed. An Ar+ ion beam, with an energy level of 3 keV and a raster of 4 mm × 4 mm, was used for sputtering. Sputter rate, determined relative to SiO2 standard, was 1.7 nm/min.

Details about the fitting of Ti 2p, Nb 3d, Zr 3d, Ta 4f and O 1 s have been published elsewhere [8, 21, 23]. The 2p3/2 peak for Ti metal state is centred at 454.2 eV, and that for Ti in the form of TiO2 at 458.7–59.1 eV. Additionally, two sub-oxides, TiO and Ti2O3, were used to deconvolute the experimental spectra. The centre of the Nb 3d5/2 peak of metallic niobium appears at 203.5 eV. Two niobium oxides were considered: NbO x (comprising NbO and NbO2) at 204.5–204.8 eV, and Nb2O5 at 207.0–207.5 eV. The centre of the XPS 3d5/2 peak of zirconium metal is located at 178.9–179.0 eV and that of ZrO2 at 182.1–183.1 eV. The XPS Ta 4f7/2 peak of tantalum metal is centred at 21.6–21.9 eV, of TaO x (comprising TaO and TaO2) at 24.6–24.8 eV, and of Ta2O5 at 26.8–27.0 eV.

The position of the centre of the O 1 s peak shifts from oxide species (O2−) at 530.1 eV, over hydroxide (OH−) at 531.4 eV, to water and/phosphate (H2O/PO4 3−) at 532.5 eV, to organic (Oorg) oxygen species at 533.6 eV. Spectra recorded in HS were deconvoluted using three component peaks, O1 (O2−), O2 (OH−) and O3 (H2O/PO4 3−) and those recorded in MEM using four component peaks, O1, O2, O3 and O4 (Oorg). Carbon C 1 s spectra were deconvoluted using three peaks: aliphatic carbon C1 (C–C/C–H) at 284.8–285.0 eV, carbon C2 bonded to oxygen (C–O) at 286.0–286.2 eV, and carbon C3 in carboxylic group and/or carbon bonded to nitrogen in organic compounds (COO−/C–N) at 288.4–288.7 eV. The nitrogen N 1 s peak was deconvoluted using three peaks related to nitrogen in organic compounds, N1 at 397.8–398.3 eV, N2 at 400.0–400.8 eV and N3 at 402.0–403.5 eV. The phosphorus peak P 2p was deconvoluted using three component peaks, P1 at 132.8–133.1 eV, P2 at 133.6–134.0 eV and P3 134.6–136.0 eV. It is suggested that these three peaks are related to different bondings between phosphorus and oxygen (P–O−, P–OH, P=O) [24, 25].

2.4 Scanning electron microscopy and energy dispersive X-ray spectroscopy

Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) were carried out with a FEI Quanta 3D FEG Dual Beam system equipped with an EDX spectrometer. The surface structure was analyzed in secondary (SE) and backscattering (BSE) electron modes.

2.5 Inductively coupled plasma-mass spectroscopy

In order to follow the biodegradation, the concentrations of metal ions (Ti, Nb, Zr and Ta) dissolved from Ti metal and TNZT alloy in MEM were measured using inductively coupled plasma-mass spectroscopy (ICP-MS). During the immersion period, 1 mL of solution aliquot was taken from each solution after 3, 7, 14, 21 and 25 days. Each aliquot was added to a polypropylene tube (Sarstedt, Nümbrecht, Germany) containing 1 mL 2 % HNO3 solution and stored for ICP-MS analysis.

Concentrations of dissolved ions were measured using an Agilent 7500ce ICP-MS instrument (Agilent Technologies, Palo Alto, USA) equipped with a MicroMist pneumatic nebulizer and a Peltier-cooled spray chamber working at 1,500 W RF power. An Ar carrier gas was used at 0.85 L min−1 and A make-up gas at 0.2 L min−1. An octopole reaction system (ORS) in kinetic energy discrimination mode, with 5 mL min−1 He collision gas, was used to reduce the effect of polyatomic interferences from the sample matrix. To further reduce the matrix influence, a method of standard additions was used for calibration (11 points). Results were expressed as mean concentrations (cavg) of duplicate measurements ± standard deviation (σ), in μg L−1. The concentrations of metal ions measured in blank samples were subtracted from the concentrations measured in test samples. Limit of detection (LOD) was calculated as the threefold standard deviation of replicate blank measurements. For titanium the value of LOD was 0.017 μg/L for Ti, 0.04 μg/L for Zr, 0.01 μg/L for Nb and 0.01 μg/L for Ta.

2.6 Assessment of biocompatibility

Human osteosarcoma (HOS), a human osteoblast-like cell line (ATCC no. CRL-1543), and human immortalized pulmonary fibroblasts (Wi-38 VA-13, ATCC no. CCL-75.1) were used for testing in vitro biocompatibility. They were cultured routinely in Advanced MEM, supplemented with 5 vol% foetal bovine serum (FBS) and 15 μg/mL gentamicin (all Gibco®, Carlsbad, CA, USA), at 37 °C in a humidified atmosphere with 5 vol% CO2.

For biocompatibility assessment of the novel TNZT alloy, steady alloy extraction in FBS-free Advanced MEM was performed at 37 °C, as previously described [26] but with an extended extraction period of 25 days. Simultaneously, pure copper and pure titanium extractions were performed to provide positive and negative controls. A sample of MEM without any metal was incubated under the same conditions and used for non-exposure controls. Following extraction, metal samples were aseptically removed. Extracts were supplemented with 5 % FBS and applied to cells.

All cell-based experiments were repeated at least three times. Data are presented as mean ± standard error. Student’s t test and One-Way ANOVA were used for statistical analysis and statistical significance was set at p < 0.05.

2.6.1 Cell survival (clonogenic assay)

Both cell lines were seeded on 12-well tissue culture plates (TPP, Trasadingen, Switzerland) at empirically pre-determined densities and grown in either unexposed control medium or metal extract media for 48, 96 or 144 h. After treatment, cell survival was evaluated using the clonogenic assay as previously described [27]. Treated HOS and Wi-38 cells were harvested, counted manually and seeded at low densities (500 and 1,000 per plate, respectively) on 6-cm Petri dishes (TPP) in 4 mL fresh complete growth medium, then incubated for 7–12 days, until the formation of macroscopic colonies. Subsequently, cells were stained with crystal violet (Sigma, München, Germany) and colonies were counted manually. Only colonies with more than 50 cells were counted. The survival fraction (SF) was determined by calculating plating efficiencies (PE = ratio between colonies counted and cells initially seeded) for all experiments and controls and subsequently normalising the PEs of experiments to those of unexposed controls.

2.6.2 Induction of apoptosis

Apoptosis induction was assessed by measurement of caspase 3 and 7 (Ca3/7) activities using the luminescence-based CaspaseGlo 3/7 assay (Promega, Madison, WI, USA), following the manufacturer’s protocol. 1,000 HOS or Wi-38 cells, suspended in 50 μL of the appropriate metal extract or control medium, were seeded in the wells of a white, flat-bottom, luminescence-compatible 96-well cell culture plate (Greiner Bio-One, Frickenhousen, Germany) and grown for 48, 96 and 144 h. The same volume of reagent was then added to each well and the plates incubated in the dark at room temperature for 2 h, followed by measurement of luminescence (1 s integration time) with a GENios plate reader (Tecan, Männedorf, Switzerland).

The measured relative luminescence units (RLU) indicating Ca3/7 activities were normalized to the respective surviving fraction (as determined by the clonogenic assay) of cells exposed to the same metal extract. Ca3/7 fold induction was subsequently calculated as the ratio of metal-extract-exposed cells to unexposed controls.

2.6.3 Growth delay

Growth curves of the TNZT alloy, copper and titanium extract-exposed and unexposed control cells were assessed using the fluorescence-based AlamarBlue® Assay (Invitrogen) according to the manufacturer’s protocol. Briefly, 2,500 HOS or Wi-38 cells, suspended in 90 μL of the appropriate metal extract or control medium, were seeded in the wells of a black, flat-bottom, fluorescence-compatible 96-well cell culture plate (Greiner) and grown for 6 days. Cell-free medium was used for blanks. Every 24 h, 10 μL of AlamarBlue reagent was added to each well and the plates incubated for 2 h at 37 °C. Fluorescence was measured with a GENios plate reader (Tecan) and growth curves plotted as net fluorescence versus time in days. Doubling times for exposed cells in exponential growth phases were calculated using the web-based tool Doubling time [28]. The start and end points of the exponential growth phase (Day 1 and 4, respectively) were used as the basis for the calculations.

2.6.4 Alkaline phosphatase activity

To measure intracellular alkaline phosphatase (ALP) activity, HOS cells were seeded in a 12-well clear cell culture plate (Greiner) in 3 mL of each metal extract or control medium and incubated for 48, 96 and 144 h. After incubation, cells were harvested and counted, and aliquots of 1 × 104 cells prepared for ALP activity measurement.

Differences in ALP activity between exposed cells and unexposed controls were assessed using the fluorimetric Alkaline Phosphatase Assay Kit (Abcam, Cambridge, UK) according to the manufacturer’s protocol. Since HOS cells have been shown empirically to exhibit high ALP activity, the total number of cells used per assay was lower than in the manufacturer’s indication. 1 × 104 cells were lysed in 110 μL of Assay Buffer and centrifuged. Supernatants were transferred into the wells of a black, flat bottomed 96-well cell culture plate (Greiner) and 20 μL of the ALP substrate methylumbelliferyl phosphate disodium salt (MUP) was added to each well. Plates were incubated in the dark for 30 min at room temperature. After incubation, reactions were stopped by adding 20 μL of Stop Solution to each well and fluorescence at 360 nm was measured using a GENios microplate reader (Tecan). ALP activity was obtained using a calibration curve (ALP enzyme supplied in the kit).

Background controls were obtained in the same way as for the experimental cell lysate, but using unexposed cells and stopping the reaction immediately after addition of MUP. Net ALP activity was calculated by subtracting the average fluorescence of background controls from the average fluorescence of each experiment. Relative ALP activity was calculated by normalizing the net activities of exposed cells to unexposed controls.

3 Results and discussion

3.1 Formation of layers at the surface of Ti and TNZT alloy during immersion

3.1.1 Composition of the layers

The chemical composition of the surface layers formed on Ti and TNZT alloy during 25 days immersion in HS and MEM was determined based on XPS survey spectra (Table 1). The carbon content of the surface is due to air contamination during the transfer of the sample from the cell to the XPS chamber. The main oxide component of the surface layer formed during immersion is titanium oxide. Other alloy components are also present, mainly Nb and low concentrations of Zr and Ta. The total content of these elements in the surface layer however does not exceed 2.2 at.%. The surface layer also contains calcium phosphate (Table 1), indicated by ratios of Ca/P ≤1.0. In MEM, the content of Ca and P is lower than in HS. The layer also contains nitrogen, originating from the organic compounds in MEM (amino acids and vitamins). In both media, other elements originating from the solution itself were also detected (Mg, Cl, Na).

Table 1 Surface composition of the layers formed on Ti and Ti–20Nb–10Zr–5Ta alloy during 25 days immersion in Hanks’ simulated physiological solution (HS) and MEM
Full size table

The content of particular species in the layer was determined by deconvolution of high resolution XPS spectra (see Fig. 1 for the spectra recorded at the surface of TNZT alloy immersed in MEM for 25 days). The intensities of peaks for metal, sub-oxide and oxide obtained from Ti and TNZT alloy immersed in HS and MEM are presented in Table 2. The location of the Ti 2p3/2 peak indicates the formation of titanium(IV) oxide (Fig. 1a). No formation of sub-oxide peaks (TiO and/or Ti2O3) was detected (Table 2). The surface layers formed on the TNZT alloy also contain small amounts of Nb, Zr and Ta (Table 1). Deconvolution of the Nb 3d peak showed that the layer contained sub-oxide NbO x (NbO and/or NbO2) and Nb2O5 (Fig. 1b). The contribution of the sub-oxide NbO x peak to the total intensity of that for Nb 3d5/2 is slightly over 10 % (Table 2). Titanium and niobium are not present in the form of metals although Zr, and especially Ta, were detected in the metal state also (Fig. 1c, d). The intensity of the Zr metal peak at 180.0 eV contributes less than 30 % of the total intensity of the Zr 3d peak obtained in HS and MEM, the remainder being the intensity due to ZrO2 (Fig. 1c; Table 2). In the case of tantalum, the contribution of the metal peak is higher at 51–55 %, the rest being divided between the sub-oxide TaO x (~one-third of the intensity) and the oxide Ta2O5 (<one-fifth of the intensity). Thus, in the surface layer, Ti is completely oxidized to TiO2, Nb is present as a mixture of NbO x and Nb2O5 (the latter being predominant), Zr is present as metal and ZrO2, and Ta is present mainly as metal, sub-oxide TaO x and pentoxide Ta2O5, the latter being the least intense component (Table 2). Incomplete oxidation of Nb, and, especially Ta, in the native oxide layers has also been observed at the surface of Ti–29Nb–13Ta–4.6Zr alloy [10]. Further, similar behaviour was observed for the TNZT samples oxidized electrochemically at different potentials for 1.5 h [21].

Fig. 1
figure 1

Deconvoluted a Ti 2p b Nb 3d c Zr 3d and d Ta 4f XPS spectra recorded at the surface of Ti–20Nb–10Zr–5Ta alloy formed during immersion for 25 days in MEM at 37 °C. Symbols measured curves, solid line fitted curve, dotted/dashed lines component peaks

Full size image
Table 2 The intensity of component peaks (metal, suboxide and oxide) expressed as percentages of the total intensity of the XPS peaks (Ti 2p3/2, Nb 3d5/2, Zr 3d5/2 and Ta 4f7/2) recorded at the surface of Ti and of Ti–20Nb–10Zr–5Ta alloy after 25 days immersion in Hanks’ simulated physiological solution (HS) and MEM
Full size table

Examples of deconvoluted spectra for the non-metal elements oxygen, nitrogen, carbon and phosphorus are presented in Fig. 2 for the spectra recorded at the surface of the TNZT alloy immersed in MEM. The intensities of the component peaks were obtained by deconvolution of all samples, Ti and TNZT alloy, each immersed in HS and MEM (Table 3). Depending on the solution, the O1 s spectra were deconvoluted using three (HS) or four (MEM) peaks. The most intense component peaks in MEM are the oxide and hydroxide peaks O1 and O2, followed by the phosphate peak O3 and the organic peak O4 (Table 3). A nitrogen peak is present only at the surface of samples immersed in MEM (Tables 1 and 3), originating from compounds present in MEM (all amino acids and vitamins B1, B2, B3, B6, B9, B12 and H). The N2 peak at 400.0 eV predominates, while peaks N1 and N3 are minor (Fig. 2b; Table 3). Thus, during immersion, nitrogen forms chemical bonds with other components, i.e. oxygen (peak O4) and carbon (peak C3) ascribed to carboxylic carbon and/or carbon bonded to nitrogen (Fig. 2c; Table 3). The intensity of the aliphatic carbon peak C1 predominates in all samples, followed in intensity by peak C2 (C–O) and peak C3 (COO−/C–N). Since the intensity of the latter was higher for samples immersed in MEM, it can be assumed that this is due to increased contribution of C–N bonding in MEM. These results indicate that component peaks O4, N2 and C3 are related to the formation of species containing organic components from MEM solution.

Fig. 2
figure 2

Deconvoluted a O 1 s, b N 1 s, c C 1 s and d P 2p XPS spectra recorded at the surface of Ti–20Nb–10Zr–5Ta alloy formed during immersion for 25 days in MEM at 37 °C. Symbols measured curves, solid line fitted curve, dashed lines component peaks (symbols used are defined in Table 3)

Full size image
Table 3 The intensity of component peaks expressed as percentages of the total intensity of the XPS peaks (O 1 s, N 1 s, C 1 s and P 2p) recorded at the surface of Ti and Ti–20Nb–10Zr–5Ta (TNZT) alloy after 25 days immersion in Hanks’ simulated physiological solution (HS) and MEM
Full size table

The surface layers formed in HS and MEM contained calcium phosphate (Table 1). The position of the Ca 2p3/2 peak is between 347.6 and 348.0 eV, consistent with the formation of Ca8H2(PO4)6·5H2O (347.2 eV), CaHPO4 (347.6 eV), and Ca10(PO4)6(OH)2 (347.8 eV) [29]. The peak corresponding to phosphorus, P 2p, appears at 134.0–134.2 eV and can be related to the following compounds: Ca3(PO4)2 at 132.9 eV and CaHPO4 at 133.8 eV [29]. The phosphorus peak P 2p was deconvoluted using three component peaks, P1, P2 and P3 (Fig. 2d). These peaks can be related to different bondings between phosphorus and oxygen (P–O−, P–OH, P=O) although determination of the exact composition of individual phosphorus peaks is beyond the scope of the present paper. Phosphate compounds are based upon a tetrahedral arrangement around phosphorus. The binding energy of the P 2p peak depends on the type of phosphorus–oxygen compound. Depending on the type of calcium phosphate, phosphorus is bonded to more OH− (Ca(H2PO4)2), or more O2− (Ca3(PO4)2) ions, respectively [25]. The similar but lower binding energy peak at 131.3 ± 0.2 eV can be ascribed to the P–P or P–H bonding, while the higher energy peak at 133.4 ± 0.2 eV is associated with P–O bonding [30, 31]. In the layers formed on Ti and TNZT alloy, the peak P2 at 133.6–134.0 eV is the most intense, followed by peaks P1 and P3 (Table 3). P1 and P2 are therefore suggested to be related to P–O− and P–OH bonding, while peak P3 may be associated to P=O bonding.

Compositions of the layers formed on Ti and TNZT alloy during immersion are revealed by XPS analysis combined with ion sputtering. The depth profiles are presented in Figs. 3 and 4. For the sake of clarity, only metal elements, oxygen and carbon are presented in the depth profiles. Minor elements, like Ca, P, and N, are not included. As the sputtering process proceeds and the surface oxide is removed, the content of oxygen decreases progressively, while that of titanium increases (Fig. 3). Carbon is removed as the sputtering progresses towards the bottom of the layer. On the alloy, the contents of Nb, Zr and Ta likewise increase (Fig. 4).

Fig. 3
figure 3

XPS depth profiles of the layers formed on Ti during immersion for 25 days in a Hanks’ simulated physiological solution (HS) and b MEM

Full size image
Fig. 4
figure 4

XPS depth profiles of the layers formed on Ti–20Nb–10Zr–5Ta alloy during immersion for 25 days in a Hanks’ simulated physiological solution (HS) and b MEM

Full size image

High resolution XPS spectra were recorded at different depths of the layer during the sputtering process. At the surface of TNZT alloy immersed in MEM, metal components Ti, Nb and Zr are present mainly in the oxide form, whilst Ta is present in both metal and oxide forms (Fig. 5). As the sputtering process proceeds, the peak centres shift towards the metal range due to progressive removal of the oxide layer. This process can also be followed in the progressive reduction of intensity of the oxygen peak (Fig. 6a). In contrast to oxygen, which is present throughout the layer depth, nitrogen and carbon are present mainly at the surface, as their intensities decrease much faster with sputtering time than those of oxygen (Fig. 6b, c). The intensity of phosphorus—and of calcium (results not shown)—also decrease rapidly with sputtering time, indicating that the calcium phosphate is present mainly at the layer surface. Similar behaviour is observed for all other samples, i.e. Ti in HS and MEM, and TNZT in HS. The interior of the layer formed during immersion therefore consists mainly of oxide while, at the surface, the oxide is mixed with calcium phosphate and carbon and other minor species from solution.

Fig. 5
figure 5

XPS a Ti 2p, b Nb 3d, c Zr 3d and d Ta 4f spectra recorded at different depths of the layer formed on Ti–20Nb–10Zr–5Ta alloy during immersion for 25 days in MEM. Numbers at the spectra indicate the sputtering time. Dashed lines denote the position of peak centres at the surface (oxide) and after the layer is sputtered away (metal)

Full size image
Fig. 6
figure 6

XPS a O 1 s, b N 1 s, c C 1 s and d P 2p spectra recorded at different depths of the layer formed on Ti–20Nb–10Zr–5Ta alloy upon immersion for 25 days in MEM. Numbers at the spectra indicate the sputtering time. Dashed lines denote the position of peak centre at the surface and after the layer is sputtered away

Full size image

3.1.2 Morphology and composition of the surface layers

The SE image mode of the SEM micrographs reveals the surface topography of the alloy, with some superficial pores and line scratches; the surface was homogeneous in composition, in agreement with the nominal alloy composition, with wt% contents of 65 ± 5, 20 ± 2, 10 ± 2, and 5 ± 1 for Ti, Nb, Zr and Ta [21]. Compared to as-prepared surface showing lines due to abrading procedure and superficial pores [21], the SE image of the SEM micrographs reveals clear changes in the surface topography of the alloy immersed in MEM solution for 25 days (Fig. 7a, b). The surface pores and line scratches have disappeared under a new surface layer, largely homogeneous. The BSE-images of the same surface regions (Fig. 7c, d) shows that the new surface layer masked the compositional contrast in the alloy in the macro-scale images (Fig. 7b), indicating that the surface layer composition is the same in all zones while, on the micro-scale, two distinctive features are visible. One is the presence of grey areas, a few to ten microns wide and distributed evenly across the surface. These features correspond to the topographical features (Fig. 7b, d). This can be interpreted as, at these points, the surface layer being thicker. One of these thicker points is shown in SE and BSE-images in Fig. 8a, c. The EDX spectra show only the alloy elements and O (Fig. 8e), indicating that they correspond to a passivating oxide layer. The second new feature of the alloy immersed in MEM was the presence of some scattered particles, tens of microns in size, lying on top of the surface layer (Fig. 8b, d). The black particles in the BSE-image (Fig. 8d) differed clearly in composition, with lower atomic weights than in the alloy. The EDX spectra detected Ca, P, O and C as the main elements, which can be attributed to a carbonated calcium phosphate.

Fig. 7
figure 7

SEM/EDX analysis of the surface layer formed during immersion of the Ti–20Nb–10Zr–5Ta alloy in MEM for 25 days: SEM images recorded in (a and b) the SE mode under magnification a ×100 and b ×600, and (c and d) the BSE mode under magnification c ×100 and d ×600

Full size image
Fig. 8
figure 8

SEM/EDX analysis of the thicker areas (grey areas in Fig. 9d) at the surface layer formed during immersion of the Ti–20Nb–10Zr–5Ta alloy in MEM for 25 days: SEM images recorded in the (a, b) SE mode under magnification a ×5,000 and b ×500, and (c, d) BSE mode under magnification c ×5,000 and d ×500. EDX spectra were taken at the (e) oxide region (image c) and (f) phosphate region (black regions in image d)

Full size image

The SEM/EDS analysis of the alloy immersed in HS showed similar features as in MEM, both in the topographical features and in the compositional homogeneity. The BSE images showed the presence of grey spots points with thicker oxide layer, and Ca and P rich black particles. Besides, the signal from the alloy in the subsurface has greatly attenuated, thus indicating that a thicker surface layer has formed in HS than in MEM.

3.2 Dissolution of Ti and TNZT alloy during immersion

Samples of Ti and TNZT alloy were, separately, immersed in MEM for 25 days. During that time the concentrations of ions in solution were measured by ICP-MS (Fig. 9a). The quantities of titanium released from the Ti and the TNZT alloy was low—in μg/L (ppb) range (Fig. 9a). It increased with time of immersion and after 25 days reached similar values for both materials—0.12 μg/L for Ti and 0.17 μg/L for the alloy (Fig. 9a). These low concentrations demonstrate that titanium exhibits a low tendency to dissolve, from both the metal and the alloy. Of the other elements present in the alloy, zirconium dissolves in the largest quantity, reaching 1.13 μg/L, i.e. approximately six times greater than the level of Ti (Fig. 9b). These values are still very low indicating high stability of the alloy. The dissolution of niobium and tantalum is even lower than that of Ti, reaching values of 0.01 μg/L for Nb and 0.02 μg/L for Ta at the end of the test (Fig. 9b). Thus, during immersion in MEM, the metals dissolve from the alloy in the following order: Zr > Ti > Ta–Nb. Increasing dissolution of zirconium compared to other metal components reflects the order of corrosion resistance of these metals, as observed also in the electrochemical measurements [21]. Based on the obtained results it can be stated that under simulating physiological conditions zirconium metal is the least, and tantalum the most corrosion resistant. In β titanium alloys the addition of zirconium contributes to solid solution hardening [32]. The formation of a thin ZrO2 layer on Zr metal occurs in a broad range of pH [33]. The growth rate of oxide increased with increasing pH which was ascribed to the decrease of the homogeneity of oxide growth [33]. Localized corrosion phenomenon was observed for mechanically polished Zr samples in 1 M HCl and 1 M HClO4 [33] and in borate buffer containing chloride ions at pH 8 [34]. The depassivation, i.e. onset of localized corrosion, occurs close to the thermodynamic potential of oxygen evolution. In acidic solutions containing chloride ions the breakdown of the passive ZrO2 film may be induced either by flaking of zirconium oxide from the surface or weakening of the oxide film by H3O+ ions, resulting in porous ZrO2 [33, 34]. In alkaline solutions, loss of structural water at potentials close to the potential of oxygen evolution leads to the transformation of amorphous structure, which incorporates OH− groups [33], to crystalline structure [35]. Two different values of dielectric constant were obtained for the zirconium oxides formed at potential below or above oxygen evolution potential, presumably due to incorporation of anions from the solution to the oxide layer thus leading to its higher conductivity [36]. Crystallization of the amorphous passive layer causes the formation of small cracks which exposes the metal surface to the electrolyte. This process leads to increased dissolution of Zr during prolonged immersion in chloride solution, as observed by ICP-MS (Fig. 9b). Further, it may be related to the presence of peak of Zr metal in XPS spectra (Fig. 1c).

Fig. 9
figure 9

Concentration measured by ICP-MS of a titanium dissolved from Ti and Ti–20Nb–10Zr–5Ta alloy and b titanium, niobium, zirconium and tantalum dissolved from the TNZT alloy during 25 days immersion in MEM. Symbols and bars represent mean values and standard deviation, respectively

Full size image

Under physiological conditions niobium is expected to form stable passive layer. In the present work the layer formed after 25 immersion in MEM was composed primarily of Nb2O5 and some NbO x (Fig. 1b). Similarly, the film formed in 0.15 M NaCl after 6 days was composed of NbO2 and different forms of Nb2O5 [37]. The incorporation of niobium cations in the oxide lattice of TiO2 resulted in a decrease in concentration of defects and improved the corrosion resistance of titanium alloy Ti–6Al–7Nb [8, 38]. Similar mechanism may be operative also for the TZNT alloy. Tantalum is present in the smallest amount in the surface layer (Table 1; Fig. 1d). As tantalum also has the highest standard potential among the metal components of the TZNT alloy [39], the dissolution of tantalum into solution is low (Fig. 9b).

3.2.1 Indication of biocompatibility of titanium metal and TNZT alloy

Current in vitro cytotoxicity assays for implant alloys are described in the ISO 10993-5:2009 standard for Biological evaluation of medical devices [40], but since they involve short-term extractions (72 h), they might not necessarily be adequate for materials that are intended to be present in the body for years [41]. Thus, a modified extraction test was performed with longer (25 days) exposure of metal samples to MEM in an attempt to better simulate the long-term effects of implant alloys in a recipient’s body.

3.2.2 Cell survival

Human osteosarcoma (HOS), a human osteoblast-like cell line, and human immortalized pulmonary fibroblasts (Wi-38) were incubated in extracts of metals, together with controls, for 48, 96 and 144 h, followed by assessment of survival with the clonogenic assay. No significant differences in survival fraction (SF) of the two lines were observed between cells exposed to extracts of titanium and of the TNZT alloy (Fig. 10). Regardless of the measurement time-point, SF remained between 0.85 and 1.18 and p values ranged from 0.35 to 0.76 for HOS and from 0.26 to 0.81 for Wi-38 cells. Moreover, we observed that duration of incubation with either TNZT alloy or titanium extracts was not associated with significant differences in SF (p > 0.17 for HOS and p > 0.223 for Wi-38).

Fig. 10
figure 10

Survival fraction (SF) of a HOS and b Wi-38 cells exposed to titanium, Ti–20Nb–10Zr–5Ta (TNZT) alloy and copper extracts. SF values are plating efficiencies of exposed cells normalized to the plating efficiency of unexposed controls. Dotted reference line unexposed controls SF = 1. *p < 0.05

Full size image

Thus, TNZT alloy and titanium extracts do not significantly affect survival of these cell lines over a period of 144 h. In contrast, a significant reduction in SF was observed for cells exposed to copper extracts, in which SF ranged between 0.003 and 0.01, with p values ranging from <0.001 to 0.006, confirming the growth inhibiting properties of copper.

3.2.3 Induction of apoptosis

Cells were incubated in extracts of titanium and of TNZT alloy and in control medium for 48, 96 and 144 h, followed by luminescent measurement of caspase Ca3/7 activity.

Exposure to titanium extracts did not significantly affect Ca3/7 activity in HOS cells, whereas a slight, but statistically significant increase in Ca3/7 activity was observed in HOS cells exposed for 144 h to extracts of the TNZT alloy (p = 0.017; Fig 11a).

Fig. 11
figure 11

Ca3/7 fold induction after exposure to titanium, Ti–20Nb–10Zr–5Ta (TNZT) and copper extract. Fold induction was calculated by normalizing data to Ca3/7 activity in unexposed controls. Dotted reference line unexposed controls Ca3/7 fold induction = 1. *p < 0.05

Full size image

Results for Wi-38 cells were somehow different, since prolonged exposure to the extracts of TNZT alloy appeared to decrease the levels of apoptosis in comparison to unexposed controls (Fig. 11). A statistically significant increase in Ca3/7 activity was observed in cells that were exposed to the TNZT alloy extract for 48 h (p = 0.03), but the same parameter decreased slightly (albeit non-significantly; p = 0.831) after 96 h. However, after 144 h, the decrease in apoptosis between TNZT-exposed cells and unexposed controls was significant once more (p < 0.001).

Exposure of Wi-38 cells to copper extracts caused a significant increase in Ca3/7 activity, regardless of incubation time (p < 0.05), whereas an interesting pattern was observed in HOS cells exposed to copper—compared to Ca3/7 activity of unexposed cells, a marked increase was observed after 48 and 96 h of incubation (p < 0.05), but a lower Ca3/7 activity was detected in cells exposed for 144 h (p < 0.001).

Regardless of the fact that overall cell survival (as determined by the clonogenic assay) was not significantly affected by the metal extracts (Fig. 10), there were some slight differences in the rates of apoptotic cell death, as determined by Ca3/7 activities, between metal-exposed and unexposed cells, but the trend differed between the cell lines (Fig. 11). While for HOS a small but statistically significant increase in Ca3/7 activity was observed for the TNZT extract, a small but statistically significant decrease was noticed for Wi-38 cells. Since HOS cells are human osteosarcoma (thus, malignant) osteoblast-like cells, and Wi-38 are immortalized pulmonary (non-malignant) fibroblasts, their mechanisms of apoptosis induction might be different, thus also affecting the response of these cells to metal extracts. Since Wi-38 are not malignantly transformed, their response to any sort of alteration in their growth conditions may be more relevant in terms of induction of apoptosis.

3.2.4 Growth delay

Rates of population growth were observed for cells exposed to control medium, copper, titanium and TNZT alloy extracts (Fig. 12). With the exception of the copper extract, no significant differences in growth were observed between cells grown in TNZT alloy, titanium extract or control medium (p = 0.114 for HOS and p = 0.251 for Wi-38 cells).

Fig. 12
figure 12

Growth curves for a HOS and b Wi-88 cell lines grown in either control medium or metal extracts. Alamar Blue fluorescence was measured at the start of the experiment (Day 0) and every 24 h for 5 subsequent days. *p < 0.05

Full size image

Doubling times were calculated using the web-based Doubling Time Online calculator [28]. Since cell death was apparent in the populations exposed to copper extracts, the “doubling time” values were negative, thus resulting in the calculation of the population half-life. No statistically significant differences in doubling times were observed between cells grown in the extracts of the TNZT alloy and titanium or control medium (p > 0.05 for both cell lines).

3.2.5 Alkaline phosphatase activity

Since intracellular ALP activity is a well-established marker for the initial stages of osteoblast differentiation and is relevant in terms of bone cell function and extracellular matrix turnover, its activity was tested in HOS cells only. Cells were incubated in metal extracts and control medium for 48, 96 and 144 h, followed by fluorimetric measurement and calculation of ALP activity.

After 48 h, there were no significant differences in ALP activity between cells exposed to metal extracts and unexposed controls (Fig. 13). After 96 h, the ALP activity in cells exposed to TNZT alloy decreased significantly compared to unexposed controls (p < 0.001), but the difference between the response to Ti extract and to unexposed control was not significant. At this point of investigation we cannot find a reasonable explanation for the statistically significant reduction after 96 h of incubation. After prolonged incubation (144 h) with metal extracts, the ALP activity appeared to increase for both Ti and TNZT but the difference was not statistically significant compared to unexposed controls (p > 0.05). In vitro ALP activity varied with time, since it is affected by changes in medium composition and pH, and by the growth phase of the cell population [42]. Therefore, a comparison between ALP activities at different incubation time-points does not make sense in this context. Nevertheless, we believe that the observed ALP activity increase after prolonged incubation is an encouraging finding that might indicate that TNZT alloy affects HOS cells similarly to Ti.

Fig. 13
figure 13

ALP activity in unexposed control, titanium, the Ti–20Nb–10Zr–5Ta (TNZT) alloy and copper extract exposed HOS cells

Full size image

Contrary to the two Ti-based materials, copper extract-exposed cells had significantly lower ALP activity (p = 0.003 for 48 and 0.020 for 144 h, respectively).

4 Conclusions

The behaviour of Ti–20Nb–10Zr–5Ta alloy and Ti metal during 25 days immersion in Hanks’ simulated physiological solution and MEM were investigated in terms of formation of the surface layers, dissolution and biocompatibility.

The layer formed on titanium metal comprised TiO2 as oxide component. The layer formed on the alloy was a mixed one consisting of TiO2 as the major oxide component and sub-oxides and oxides of the alloy components: NbO x , Nb2O5, ZrO2, TaO x and Ta2O5. Titanium was completely, and niobium and zirconium almost completely, oxidized to the highest-valence oxide. The oxidation of tantalum was retarded compared to oxidation of other metal components.

The layers formed during immersion in MEM and HS were homogeneous in composition but its thickness varied along the surface. Particles of carbonated calcium phosphate were identified. In MEM, species containing nitrogen, carbon and oxygen organic species originating from solution components were also observed at the surface.

Dissolution of titanium in both Ti metal and TNZT alloy in MEM was low, i.e. up to 20 ng/L. The quantity of dissolved metal ions from the TNZT alloy decreased in the following order: Zr > Ti > Ta–Nb.

In terms of biocompatibility, the TNZT alloy was shown to behave similarly to titanium, as no detrimental effects on cell survival, apoptosis induction, growth delay, or alkaline phosphatase activity were observed. Considering better functional properties including higher corrosion resistance, hardness and ultimate tensile strength and lower elastic modulus [21, 22], this study confirms that also biocompatible properties of quaternary alloy are suitable for the use in biomedical applications.