1 Introduction

The nearly equiatomic Ni–Ti alloy (Nitinol) has been attracting considerable interest for biomedical applications due to its shape memory and superelastics properties and biocompatibility. The shape memory effect involves the recovery of a certain deformation induced at low temperature and as a consequence, a controlled change of shape when the alloy is heated. This phenomenon of the Ni–Ti alloy, first described by Buehler and Wang [1], is the result of a phase transition form the low-temperature phase martensite to the high-temperature phase austenite. In the low-temperature phase the alloy can easily be deformed plastically. When the alloy is exposed to temperatures above the transition temperature, the alloy changes to the high-temperature phase and will revert to its original shape. In this way, deformations even up to 8% strain can be completely recovered. The special properties of Ni–Ti alloy can be put to excellent use for various biomedical applications. For example, at present, the properties of Ni–Ti alloy are already being used clinically in wires for orthodontic tooth alignment, osteosynthesis staples, vena cava filters and other vascular applications [2–5]. In spite of the intriguing properties of the Ni–Ti alloy and its possible interesting biomedical applications, the implantation of nickel containing materials in human body requires caution. The biocompatibility of a metallic implant is strongly related to its resistance to corrosion processes. Many studies have reported that metallic implant alloys ions can be released in the body owing to these processes. Because released metallic ions such as nickel may promote toxic, allergic, and potentially carcinogenic effects [6–8], specific concerns have been raised relative to the corrosion behavior of nickel containing alloys such Ni–Ti and stainless steels. Bass et al. [9] reported conversion cases of nickel-nonsensitive subjects into nickel-sensitive subjects following the use of Ni–Ti arch wires. Similar nickel sensitivity reactions have been reported with stainless steel plates [10]. Moreover, it has been suggested that allergic reactions to nickel may trigger in-stent restenosis after Ni–Ti stents implantation [11]. In addition, subtoxic amounts of metallic ions have been reported to affect the release of cytokines, which may be a concern for specific applications such as cardiovascular stents [12]. Therefore, in order to pursue better safety, the development of Ni-free shape memory alloys is necessary.

According to studies on both the cytotoxicity of pure metals and polarization resistance-tissue reaction of surgical implant materials, shape memory alloys consisting of Ti, Zr, Nb, Ta and Sn elements would be recommended for biological applications [13]. Hanada et al. developed Ti–Nb–Sn alloys consisting of non-cytotoxic elements aiming for alternative biomaterial of Ni containing Ti alloys [14, 15]. The newly developed alloys exhibit superelasticity or low Young’s modulus of approximately 40 GPa, which mechanism is attributed to phase stability of metastable bcc β phase to hcp α, martensite α and ω. The low Young’s modulus in these alloys is for mechanical compatibility with human cortical bone (10–30 GPa), which enables the alloys to be used for orthopedic implants [16–18]. Despite the high amount of reported research on the microstructure and mechanical properties of Ti–Sn–Nb alloys [19–26], their corrosion behaviour in simulated physiologic media has scarcely been explored [27].

In this work, the in vitro corrosion resistance of two Ti–Nb–Sn shape memory alloys, Ti–16Nb–5Sn and Ti–18Nb–4Sn (mass%) was assessed and compared with that of the nearly equiatomic Ni–Ti alloy. The in vitro corrosion behavior of the alloys was investigated in an artificial physiological solution using three electrochemical techniques: open circuit potential measurements, potentiodynamic polarization and electrochemical impedance spectroscopy (EIS).

2 Experimental details

The metals used were Ti grade 2 (L. Klein SA, Chemin du Long-Champ 110, 973 CH-2501 Biel–Switzerland), Nb (99.6 mass%, H.C.Starck, Inc., 45 Industrial place, Newton, MA 02461–1951/USA), and Sn 99.999 mass% (Newmet Koch, Waltham Abbey, Essex, UK). The Ti–16Nb–5Sn and Ti–18Nb–4Sn alloys, with a mass of about 2 g each, were prepared by arc melting from starting metals in pieces, in inert Ar atmosphere. The samples were melted and reversed at least five times, so obtaining compact and homogenous ingots with neither weight loss nor oxidation.

Electrochemical experiments were performed in a standard three-electrode cell with 0.25 cm2 of exposed area in the working electrode, having a platinum mesh as a counter electrode and a saturated calomel electrode (SCE) as reference. Electrochemical tests were conducted in naturally aerated Ringer’s solution (8.61 g L−1 NaCl + 0.30 g L−1 KCl + 0.49 g L−1 CaCl2) at (37 ± 1)°C to simulate physiological conditions representative of what a biomedical component would experience [27–31]. Before each measurement, the samples’ surface was first ground with 320, 400 and 600 μm SiC abrasive papers in anhydrous ethyl alcohol and then automatically polished up to 1 μm using diamond pastes and non aqueous lubricants until a mirror-bright surface was achieved. The Ti–Nb–Sn alloys were studied on as-cast conditions.

The corrosion potential, E corr, with time was measured since the first minutes of electrodes exposure to Ringer’s solution. The samples remained immersed for 360 h in naturally aerated Ringer’s solution without stirring, and the corrosion potential was measured as a function of time, until its variation with time became negligible.

Potentiodynamic polarization studies were performed after 24 h exposure to Ringer’s solution for each alloy. Potentiodynamic polarization curves were recorded using a Solartron 1286 electrochemical interface, by scanning from a cathodic potential of −1,000 mV up to 1 V vs. SCE, at a scan rate of 0.5 mV s−1.

Electrochemical impedance spectroscopy (EIS) measurements were carried out at open-circuit potential using a GAMRY EIS300 electrochemical frequency response analyzer (FRA) system. The impedance spectra were acquired in the frequency range from 10 kHz to 10 mHz with a perturbation signal of 10 mV. EIS spectra were acquired at different exposure times to the aggressive environment.

For comparison, all tests were also performed on the nearly equiatomic Ni–Ti shape memory alloy (Nitinol) supplied by Johnson Matthey, London, UK. Measurements were performed thrice as to ensure reproducibility of results.

3 Results and discussion

The corrosion potentials, E corr, variation with exposure time of Ni–Ti, Ti–16Nb–5Sn and Ti–18Nb–4Sn alloys to Ringer’s solution at 37°C for a period of 360 h are shown in Fig. 1. The time profiles of E corr obtained for the samples are quite similar. As can be seen, the corrosion potential changes quickly towards more positive potentials during the first 10 h. After that, E corr changes more slowly until it reaches a quasi-stationary value at 80 h, not changing significantly after that. This fact indicates that both the Ti–Nb–Sn alloys as well as the nearly equiatomic Ni–Ti alloy undergo spontaneous passivation due to spontaneously formed oxide film passivating the metallic surface, in Ringer’s solution.

Fig. 1
figure 1

Corrosion potential vs. time profile for Ni–Ti and Ti–Nb–Sn alloys after different exposure times to naturally aerated Ringer’s physiological solution at 37°C

Full size image

Similar behaviour was found by Cai et al. [32], where an initial increase in the E corrs during the early hours followed by stabilization (−205 to −85 mV/SCE) observed on all the specimens (CP Ti, Ti–6Al–4V, Ti–6Al–7Nb and Ti–13Nb–13Zr) suggests that a protective film formed rapidly on the metal surfaces in the artificial saliva and remained stable during the entire exposure period (6 × 104 s) [32].

By comparing the results reported in Fig. 1, it can be observed that the Ti–18Nb–4Sn alloy presents the most positive corrosion potential values, the Ti–16Nb–5Sn alloy exhibits lower values, while the nearly equiatomic Ni–Ti alloy shows the most negative E corr values. These behaviours indicate that the Ti–18Nb–4Sn alloy displays the greatest tendency to spontaneous oxide film formation in Ringer’s solution. Zheng et al. [27] performed E corr measurements on Ti–16Nb–5Sn alloy in Hank’s solution, pH 7.4, at 37°C for a period up to 10 h. The corrosion potential was found to increase towards the noble direction with increasing time, then stabilizing after 6 h exposure to a final value of −200 mV/SCE [27] which is significantly lower than that obtained in the present work for the same alloy after 10 h exposure (E corr = 50 mV/SCE).

Figure 2 compares typical potentiodynamic polarization curves of Ni–Ti, Ti–16Nb–5Sn and Ti–18Nb–4Sn alloys recorded after 24 h exposure to naturally aerated Ringer’s solution at 37°C. The average corrosion potential estimated from these curves are −400, −295 and −185 mV(SCE), for the Ni–Ti, Ti–16Nb–5Sn and Ti–18Nb–4Sn alloys, respectively. The corrosion potentials determined from the polarization curves are significantly lower than those obtained from the open circuit potentials measurements. This is expected, as the polarization test were started at a cathodic potential relatively to the corrosion potential, so that the passive oxide film at the surface was at least partially removed due to the highly reducing initial potentials.

Fig. 2
figure 2

Potentiodynamic polarization curves of Ni–Ti and Ti–Nb–Sn alloys recorded after 24 h exposure to naturally aerated Ringer’s physiological solution at 37°C

Full size image

The nearly equiatomic Ni–Ti alloy exhibits a passive potential range, from −300 to 450 mV(SCE), characterized by low current density values (not exceeding 6 μA cm−2), followed by a sharp increase in current density due to the breakdown of passivity. The potential at which this sharp increase in current by two or three orders of magnitude occurs is defined as breakdown potential, which corresponds to the crack of the passive oxide film formed at the surface of the specimen.

The polarization curve of the Ti–16Nb–5Sn alloy, in Fig. 2, reveals that current density increases with potential from the corrosion potential until approximately −200 mV(SCE). In the potential range, −200 ÷ 1,000 mV(SCE), the current density remains substantially unchanged as the potential increases, indicating typical passive behaviour. The current density of the Ti–16Nb–5Sn alloy in this passive region (i pp) is approximately 4 μA cm−2.

A similar polarization behaviour is observed for the Ti–18Nb–4Sn alloy. The current density increases with potential from the corrosion potential until approximately −125 mV(SCE). In the potential range, −125 ÷ 1,000 mV(SCE), the current density remains substantially unchanged as the potential increases, indicating typical passive behaviour. However, the Ti–18Nb–4Sn alloy is characterized by a lower i pp value, around 2 μA cm−2.

These results indicate that the passive film present on the Ti–Sn–Nb alloys’ surface has better corrosion protection characteristics than the one formed on Nitinol.

Potentiodynamic polarization measurements were also performed by Zheng et al. [27] on Ti–16Nb–5Sn alloy in Hank’s solution, pH 7.4, at 37°C. The shape of polarization curve is very similar to that obtained in the present work for the same alloy. A wide passive region, extending from 0 to about 1,000 mV/SCE, was observed. The primary passive current density value was about 10 μA cm−2 [27] which is slightly higher than that obtained in the present work (i pp ≈ 4 μA cm−2).

The enhanced passive behaviour of Ti–Nb–Sn alloys with respect to the nearly equiatomic Ni–Ti alloy can mainly be ascribed to the presence of Nb. The effects of Nb as an alloying element in stabilizing the surface films on Ti based alloys have been reported [33]. Moreover, the Nb cations improve the passivation properties of the surface layer by decreasing the concentration of anion vacancies present on titanium oxide film. These anion vacancies are generated by the presence of lower titanium oxidation states [34, 35].

Figure 3 reports the impedance spectra, presented as Bode diagrams (phase angle vs log f and Z modulus vs log f), of the nearly equiatomic Ni–Ti alloy and the Ti–Nb–Sn alloys at the corrosion potential after different exposure times to naturally aerated Ringer’s solution at 37°C. The quality of the impedance data was checked using the Kramers–Kronig transforms that indicate excellent agreements, except for the small discrepancies at low frequencies. These discrepancies are due to the “tails” problem that arises from the fact that the impedance data are measured over a finite bandwidth whereas the transforms are defined over an infinite frequency bandwidth [36].

Fig. 3
figure 3

Bode diagrams for Ni–Ti and Ti–Nb–Sn alloys after different exposure times to Ringer’s physiological solution at 37°C

Full size image

It can be seen in Fig. 3 that the modulus of impedance (∣Z∣) increases with exposure time indicating an improvement in corrosion resistance of the spontaneous oxide film in the chloride-containing environment, for both Ni–Ti and Ti–Nb–Sn alloys investigated. However, the impedance values of Ti–16Nb–5Sn and Ti–18Nb–4Sn alloys are higher than those of the Ni–Ti alloy. Moreover, the phase-angle (θ) diagrams of Ti–Nb–Sn alloys indicate a highly capacitive behaviour, typical of passive materials, with phase angles approaching −90° in the frequency range from 100 to 0.1 Hz. This result suggests a highly stable passive film on both Ti–16Nb–5Sn and Ti–18Nb–4Sn alloys in Ringer’s solution. A lower capacitive behaviour is observed for the nearly equiatomic Ni–Ti alloy, supporting the results obtained from the other electrochemical techniques.

The impedance spectra obtained have been interpreted in terms of an equivalent circuit, reported in Fig. 4, with the circuit elements representing electrochemical properties of the alloy and its surface oxide film. A very good agreement between experimental and theoretical data was obtained. A standard deviation χ-quadrate was in the order of 10−5 [37, 38], and the relative error was less than 5%.

Fig. 4
figure 4

Proposed equivalent circuit for modeling experimental impedance data performed after different exposure times to Ringer’s physiological solution at 37°C. R e = electrolyte resistance; R p = porous layer resistance; Q p = porous layer capacitance; R b = barrier layer resistance; Q b = barrier layer capacitance

Full size image

The equivalent circuit shown in Fig. 4 is constituted by a parallel combination of resistance and capacitance elements (RC) that are in series with R e, which represents electrolyte (Ringer’s solution) resistance. In Bode diagrams Re is expressed in a high frequency limit (f > 1 Hz). The model assumes that the oxide layer on the alloys experimentally examined consists of a dense inner layer and a porous outer layer [37–40]. The high frequency parameters R p and Q p represent the properties of the reactions at the outer porous passive film/solution interface. The symbol Q signifies the possibility of a non-ideal capacitance (CPE, constant phase element). The impedance the CPE is given by [36]:

$$ Q = Z_{\text{CPE}} = \left[ {C\left( {{\text{j}}\omega } \right)^{n} } \right]^{ - 1} $$

where for n = 1, the Q element reduces to a capacitor with a capacitance C and, for n = 0, to a simple resistor. The parameter R b coupled with Q b describes the processes at the inner barrier layer at the electrolyte/dense passive film interface.

The variation of the porous outer layer resistance, R p, and the inner barrier layer resistance, R b, with time in Ringer’s solution is illustrated in Fig. 5. In all cases the polarization resistances R p and R b increase with exposure time to the aggressive environment, demonstrating that the spontaneously formed oxide film on the alloys becomes more resistive with increasing exposure time to Ringer’s solution. On the other hand, the polarization resistances of the inner barrier layer (R b) for all three alloys are significantly higher than R p (outer porous layer). This indicates that the protection is predominantly provided by the inner barrier layer, as also observed by Assis et al. for Ti-alloys [38]. Comparisons between porous outer layer and inner barrier layer resistances (R p and R b, respectively) for Ni–Ti and Ti–Nb–Sn alloys as a function of exposure time to Ringer’s solution, permit to observe that the nearly equiatomic Ni–Ti alloy possesses lower Rp and Rb values than those of the Ti–Nb–Sn alloys, as shown in Fig. 5. This indicates that the Ni–Ti alloy exhibits poorer corrosion behaviour [37, 39, 41, 42]. Finally, the highest polarization resistances, R p and R b, values for Ti–18Nb–4Sn show that this alloy has better corrosion resistance and higher stability than the other metals studied at 360 h exposure to Ringer’s solution, which agrees with and confirms the results obtained from the open circuit potential and potentiodynamic polarization studies (Figs. 1, 2).

Fig. 5
figure 5

(a) Barrier layer resistance, R b, and (b) porous layer resistance, R p, for Ni–Ti and Ti–Nb–Sn alloys after different exposure times to Ringer’s physiological solution at 37°C

Full size image

4 Conclusions

The in vitro corrosion behavior of two Ti–Nb–Sn shape memory alloys, Ti–16Nb–5Sn and Ti–18Nb–4Sn (mass%), has been assessed and compared with that of Nitinol, a nearly equiatomic Ni–Ti alloy employed in the medical and dental fields owing to its superelastic properties. The in vitro corrosion resistance was investigated in Ringer’s physiological solution at 37°C by corrosion potential and electrochemical impedance spectroscopy (EIS) measurements as a function of exposure time, and potentiodynamic polarization curves. The results can be summarized as follows:

  1. 1.

    Corrosion potential, E corr, values indicated that both Ni–Ti and Ti–Nb–Sn alloys undergo spontaneous passivation due to spontaneously formed oxide film passivating the metallic surface, in the aggressive environment. Ti–18Nb–4Sn alloy presented the highest E corr values, while the nearly equiatomic Ni–Ti alloy exhibited the lowest E corr values, indicating that the tendency for the formation of a spontaneous oxide is greater for the Ti–18Nb–4Sn alloy.

  2. 2.

    Significantly low anodic current density values were obtained from the polarization curves, indicating a typical passive behaviour for all investigated alloys, but Nitinol exhibited breakdown of passivity at potentials above approximately 450 mV(SCE), suggesting lower corrosion protection characteristics of its oxide film compared to the Ti–Nb–Sn alloys.

  3. 3.

    High impedance values were obtained for all samples, and its increase with exposure time to the aggressive solution indicated an improvement in corrosion resistance of the spontaneous oxide film. The obtained EIS spectra were analyzed using an equivalent electrical circuit representing a duplex structure oxide film, composed by an outer and porous layer (low resistance), and an inner barrier layer (high resistance) mainly responsible for the alloys corrosion resistance. The resistance of passive film present on the metals’ surface increases with exposure time displaying the highest values to Ti–18Nb–4Sn alloy.

All these electrochemical results are connected and suggest that Ti–Nb–Sn alloys are promising materials for biomedical applications.