Introduction

It is well known that titanium (Ti) and Ti alloys have been used extensively in dental and orthopedic implants. This is ascribed to a thin but dense natural TiO2 passive layer formed spontaneously on their surface that leads to excellent corrosion resistance and biocompatibility [1,2,3]. Despite the remarkable biocompatibility of the native oxide layer of Ti, long-term implant failure or infection may occur due to the poor osseointegration, corrosion, wear, tribocorrosion, the release of metallic wear debris, etc. [4,5,6,7,8]. Therefore, various surface modifications of Ti implants have been developed to enhance osseointegration and other properties [5,6,7,8,9,10]. It has been well documented that the formation of nanostructures, such as nanotubes and nanowires, on the surface of Ti and its alloys can improve their biocompatibility and osseointegration [11,12,13,14].

On the other hand, since body fluids are chloride-containing solutions and always kept at 37 °C, the physiological corrosion of Ti-based implants may occur, resulting in the possible release of toxic metallic ions. In particular, the low pH value and existence of lipopolysaccharide in saliva can increase corrosion rate of Ti dental implants [15]. Hence, it is highly desirable to achieve better corrosion resistance for Ti-based implants.

Recently, TiO2 nanotube array films obtained by anodization have gained considerable attention owing to greater control over film morphology, ease of preparation and good adhesion to the underlying substrate [8,9,10,11, 16]. The anodized nanotubular surfaces have been demonstrated to exhibit enhanced osteoblast or bone-forming cell adhesion and function [17, 18], improved growth of hydroxyapatite [19], and better cellular behavior and tissue integration [20] compared to their unanodized counterparts, suggesting great promise as biomaterials. Since TiO2 nanotube films on Ti substrate possess larger surface areas, they may change the corrosion resistance of Ti. It is therefore not surprising that much attention has been directed to exploring the corrosion resistance properties of nanotubes in recent years [21,22,23,24,25,26]. For instance, Saji et al. [21] investigated the electrochemical passivation behavior of nanoporous and nanotubular oxide layers on Ti–35Nb–5Ta–7Zr alloy in Ringer’s solution at 37 °C and found that the corrosion resistance of the alloys was in increasing order of nanotubular < bare < nanoporous ones. The nanotubular alloy showed a lower corrosion resistance property compared to the bare alloy. Nevertheless, Yu et al. [22] found that Ti with TiO2 nanotube films exhibited higher corrosion resistance and lower corrosion current density in Hank’s solution than the smooth Ti. They also examined the electrochemical behaviors of Ti with two different diameter nanotube layers in two test solutions: phosphate-buffered saline (PBS) and Dulbecco’s minimum essential medium with serum proteins. Their results showed that Ti with the nanotube layers of two different diameters had increased corrosion resistances in comparison with the smooth Ti and the corrosion resistance of samples in the test solution containing serum proteins was better than that in PBS [23]. Mazare et al. [26] reported on the influence of annealing temperatures on the corrosion behavior and antibacterial activity of ordered TiO2 nanotube layers. Their work demonstrated that the TiO2 nanotubes subjected to an optimum annealing process exhibit excellent corrosion resistance and antibacterial activity as well as improved hemocompatibility.

However, in the above-mentioned studies, TiO2 nanotube films were all prepared by the commonly employed single-step anodization so as to yield similar film morphology. Actually, the morphology of nanotubes, especially the interfacial structure between TiO2 nanotubes and Ti substrate, can potentially influence the corrosion resistance properties of the Ti substrate with nanotubes [11]. Hence in the present work, to get more insight into the effect of film morphology on the corrosion behavior, five different types of TiO2 nanotube films were produced on the polished Ti foils using multistep anodization procedures in a fluoride-containing electrolyte, or further in a fluoride-free electrolyte (for details see below). The film morphologies of the anodized Ti samples before and after thermal annealing were compared. The corrosion behaviors of Ti covered by these different nanotube films were examined in detail in an artificial saliva. Previous studies have indicated that the adsorption of proteins might cause a change in corrosion behavior of Ti in a simulated body fluid containing proteins [27, 28]. Hence, the artificial saliva containing bovine serum albumin was used in this study to better reflect actual technical applications.

Experimental

Materials and reagents

Ti foils (99.9% purity, 0.1 mm thickness) were purchased from Shanghai Shangmu Technology Co., Ltd. (China), and bovine serum albumin was obtained from Shanghai yuanye Bio-Technology Co., Ltd. (China). All other chemical agents were purchased from Sinopharm Chemical Reagent Co., Ltd. (China), which were analytical purity and used without further purification. Solutions used in this work were prepared freshly using deionized water.

TiO2 nanotube film preparation

TiO2 nanotube array films were fabricated by anodization of Ti foils (1 cm × 2 cm) in a two-electrode configuration with a carbon rod as cathode. Anodization was carried out in an ethylene glycol (EG) electrolyte containing 0.5 wt% NH4F and 2 vol% deionized water at 20 °C. In this process, the electrolyte was stirred continuously using a magnetic stirrer. Before anodization, Ti foils were chemically etched and polished for 20 s in a mixed solution of HF (40 wt%), HNO3 (65–68 wt%) and deionized water (1:1:2 in volume) and then cleaned with deionized water thoroughly using an ultrasonic bath. Subsequently, five different types of TiO2 nanotube films on the polished Ti foils were fabricated. Sample a: TiO2 nanotube films were obtained by anodization of the polished Ti foils at 30 V for 10 min in the EG solution of NH4F. That is, Sample a was obtained by single-step anodization. Sample b: TiO2 nanotube films were fabricated by two-step anodization. The first-step anodization was the same as sample a. The as-obtained nanotube films were removed from the Ti foil with adhesive tape. The second-step anodization was performed under the same condition as the first-step one. Sample c: TiO2 nanotube films were fabricated also by two-step anodization. The first-step anodization was also the same as sample a. Subsequently, the Ti foils with the formed nanotube films were additionally anodized at 60 V for 5 min in a fluoride-free EG electrolyte containing 5 wt% H3PO4 to generate a compact oxide layer between TiO2 nanotubes and Ti substrate [29]. Sample d: TiO2 nanotube films were obtained by three-step anodization. The first-step anodization was the same as sample a but for 5 min anodization. Then, a compact oxide layer was prepared by the same way as sample c. Finally, the third-step anodization was performed for 5 min under the same condition as the first-step one. Sample e: TiO2 nanotube films were fabricated by two-step anodization. The first-step anodization was the same as sample a but for 9 min anodization. Subsequently, the anodization voltage was decreased from 30 to 20 V, the Ti foils with the formed nanotube films were anodized again at 20 V for 5 min in the same electrolyte. A DC power supply (Agilent Technologies, N5772A) was used for the sample preparation with a limiting current of 1.0 A. Anodization conditions and procedures of these TiO2 nanotube samples are summarized in Table 1 for clarity. Before electrochemical tests, all samples were annealed at 550 °C for 3 h with a heating rate of 5 °C min−1 in air.

Table 1 Fabrication conditions and procedures of five different TiO2 nanotube films
Full size table

Characterizations and electrochemical measurements

The morphology of the as-prepared samples was characterized by a field-emission scanning electron microscope (FESEM, FEI Quanta 250FEG). The elemental composition of the annealed samples was determined by energy-dispersive X-ray spectroscopy (EDS) using an Oxford INCA MAX-50 detector incorporated into the FESEM system. All electrochemical measurements were performed with Autolab PGSTAT302N/FRA2 in a three-electrode configuration, where a saturated calomel electrode (SCE) was used as reference electrode and a Pt sheet as counterelectrode. The potentiodynamic polarization behaviors were recorded after immersed for 90 min in a test electrolyte. The scan range of the potentiodynamic polarization was − 250 mV to + 250 mV (vs. open-circuit potential) at a scanning rate of 0.167 mV s−1. Corrosion parameters such as corrosion potential (Ecorr) and corrosion current density (icorr) of each sample were obtained from polarization curves with the support of NOVA (V1.1) software. Electrochemical impedance spectroscopy (EIS) measurements were carried out over a frequency range of 100 kHz to 0.1 Hz at the open-circuit potential in the same test electrolyte. The amplitude of AC signal was 10 mV and 10 points per decade were used. The ZSimpWin (V3.10) software was used for fitting the impedance spectra to equivalent circuit models. An artificial saliva based on Fusayama’s solution was used as the test electrolyte with a composition as follows [5, 30, 31]: 0.4 g L−1 NaCl, 0.4 g L−1 KCl, 0.795 g L−1 CaCl2·2H2O, 0.78 g L−1 NaH2PO4·2H2O, 1 g L−1 uree, 5 g L−1 bovine serum albumin. All tests were maintained at 37 ± 1 °C and performed in triplicate for each sample.

Results and discussion

Figures 1 and 2 show the representative SEM images for various types of as-anodized TiO2 nanotubes and the corresponding thermally annealed ones. For sample a and sample b prepared by single-step anodization and two-step anodization, respectively, they exhibit nearly identical morphology of nanotubes with an average outer diameter of ~ 65 nm and a length of 1 μm because the two-step anodization has the same anodization parameters as the single-step anodization (Fig. 1a, c). The insets of Fig. 1 display the surface morphology of nanotube films. By comparing the insets of Fig. 1a, c, it is seen that sample b exhibits more well-defined and uniformly distributed nanopores on its surface than sample a, but both samples have a similar average pore diameter of ~ 38 nm along the surface area. The difference in the surface topography of two samples can be ascribed to the effect of the two-step anodization. However, after the thermal annealing, sample a (single-step anodization) undergoes a pronounced morphological change (Fig. 1b), whereas sample b (two-step anodization) shows little change in the nanotube morphology (Fig. 1d). As shown in Fig. 1b, the annealed sample a presents cracked nanotube walls, and the bottom part of nanotubes is converted to a layer of nanoparticulate aggregations with a thickness of ~ 140 nm. The morphological difference between sample a and sample b after the thermal annealing may be attributed to differences in the interfacial adhesion between TiO2 nanotubes and Ti substrate [29]. Besides, after the thermal annealing, both samples present similar nanopores with square-shaped structures on the surface due to crystallization of as-anodized TiO2, as shown in the insets of Fig. 1b, d.

Figure 1
figure 1

SEM images of various TiO2 nanotube films before (left) and after (right) thermal annealing: a, b sample a (single-step anodization); c, d sample b (two-step anodization). Insets: top view of the corresponding nanotube films

Full size image
Figure 2
figure 2

SEM images of various TiO2 nanotube films before (left) and after (right) thermal annealing: a, b sample c (the second-step anodization was carried out in the fluoride-free electrolyte); c, d sample d (three-step anodization); e, f sample e (the second-step anodization was carried out in the fluoride-containing electrolyte at 20 V). Insets: top view of the corresponding nanotube films

Full size image

Unlike sample b, sample c was prepared by performing the second-step anodization in 5 wt% H3PO4 EG solution instead of the NH4F EG solution. As a result, a ~ 94-nm-thick compact oxide layer between TiO2 nanotubes and Ti substrate appears due to the second-step anodization in the fluoride-free electrolyte [11], as shown in Fig. 2a. Obviously, the hemispherical bottoms of nanotubes are buried in the compact oxide layer with no appreciable change in the morphology of the upper part of the nanotubes. As can be seen from the inset of Fig. 2a, sample c has a similar surface topography to sample a. Further, the nanotube layer shows no apparent change in the thickness and morphology before and after the thermal annealing (Fig. 2a, b). It is seen, however, that the compact oxide layer is transformed into closely packed grain structures upon the thermal annealing (Fig. 2b).

Sample d was originally designed to form a triple-layer structure, namely nanotube layer–compact oxide layer–nanotube layer structure. In other words, another nanotube layer was added under the compact oxide layer of sample c. To ensure that the thickness of the triple-layer structure was also close to that of sample c (~ 1 μm), the anodization durations for the fabrication of the upper and lower nanotube layers of sample d were halved compared to sample c (Table 1). However, as shown in Fig. 2c, the expected triple-layer structure cannot be observed and sample d still exhibits the same morphological structure as that of sample c except for a reduced film thickness. The thickness of the upper layer nanotubes for sample d is only ~ 550 nm due to the halved anodization time, whereas the compact oxide layer is ~ 80 nm thick. While the third-step anodization for sample d was performed under the same voltage (30 V) as the first-step one, no appreciable current was found during the anodization duration of 5 min because of the barrier effect of the compact oxide layer. Therefore, the triple-layer structure was not realized for sample d due to the growth failure of lower layer nanotubes. It is worth mentioning that sample d shows almost no change in morphology after the thermal annealing (Fig. 2d). In addition, from the inset of Fig. 2c, it is seen that sample d shows slightly smaller diameter nanopores on its surface compared with other samples, which should be associated with the third-step anodization. Although the third-step anodization of sample d has failed, the diameter of nanopores on its surface would become smaller due to the immersion of sample d in the fluoride-containing electrolyte during anodization.

Unlike sample d, sample e was obtained by the two-step anodization instead of three-step anodization process, i.e., the second-step anodization was carried out directly in the same fluoride-containing electrolyte after the fabrication of the first-layer nanotubes. Although the voltage of the second-step anodization was lower than that of the first one, because of the absence of the compact oxide layer, distinct currents corresponding to nanotube growth were observed during the second-step anodization. Accordingly, a ~ 136-nm-thick nanotube layer with smaller tube diameters appears under the first-layer nanotubes after 5 min anodization at 20 V, as shown in Fig. 2e. After the thermal annealing, the nanotube layer with smaller tube diameters is converted to a layer of nanoparticulate aggregations and the first-layer nanotubes exhibit cracked nanotube walls (Fig. 2f). Comparing Fig. 2f with Fig. 1b, it appears that the annealed sample e has a similar morphology to the annealed sample a. Besides, by comparison of the surface morphology of five different types of samples, it can be seen that nanopores on the surface for all samples display the transformation of circular structures to square-shaped ones after thermal annealing except for sample c and sample d (Insets of Figs. 1, 2). This implies that thermal annealing-induced crystallization can lead to morphological changes of the anodized amorphous TiO2 nanotubes except for sample c and sample d. Interestingly, there are no cracks in the nanotube walls of sample c and sample d after thermal annealing, suggesting that sample c and sample d undergo a small volume change associated with the crystallization process. It is worthy to note that, unlike the other samples, both sample c and sample d have performed the second-step anodization in the fluoride-free electrolyte, which is presumably related to little morphological change of nanotubes before and after the thermal annealing. We suppose that the anodization process in the fluoride-free electrolyte may eliminate or decrease the internal stress generated during the growth of TiO2 nanotubes, resulting in a small volume change and little morphological change of nanotubes accompanied by the crystallization process.

The chemical compositions of the nanotube samples were confirmed by semiquantitative EDS analysis. Figure 3 shows the EDS spectra acquired from the nanotube surfaces of the samples after thermal annealing. It is evident that C and Pt elements are also detected in addition to Ti and O elements for all samples. The residual carbon originates from organic electrolyte decomposition products [11] and the existence of Pt element is because samples were sputter-coated with platinum prior to SEM observation. Unlike the as-anodized TiO2 nanotubes formed in NH4F electrolyte [8,9,10], no F element can be detected in these annealed samples owing to the effect of thermal annealing. As mentioned above, both sample c and sample d have performed the second-step anodization in the H3PO4 electrolyte, but P element can only be detected in sample c (Fig. 3c). The fact that no P element was detected in sample d may be a result of the dissolution of P species in fluoride-containing electrolyte during the third-step anodization.

Figure 3
figure 3

EDS spectra of the annealed samples: a sample a, b sample b, c sample c, d sample d and e sample e

Full size image

Figure 4 displays the change in open-circuit potential with immersion time for Ti sheets with five different TiO2 nanotube films and the chemically polished Ti sheet in the artificial saliva at 37 ± 1 °C. The presented curves are representative curves for the open-circuit potential test. After 2 h of testing, the open-circuit potentials for all samples reach their steady-state values. Obviously, the chemically polished Ti sheet (i.e., bare Ti without TiO2 nanotube films) has the lowest open-circuit potential. After the anodization treatments, five Ti samples coated with different TiO2 nanotube films all exhibit the markedly enhanced open-circuit potentials, suggesting that they possess better corrosion resistance properties than the bare Ti. Further, for Ti sheets with five different TiO2 nanotube films, the final open-circuit potentials for sample b (two-step anodization) and sample d (three-step anodization) are significantly higher than those of other samples, whereas sample e (the second-step anodization was carried out in the fluoride-containing electrolyte at 20 V) has the lowest open-circuit potential.

Figure 4
figure 4

Open-circuit potential variation with time curves of samples in the artificial saliva

Full size image

Figure 5 shows the typical potentiodynamic polarization curves acquired for bare Ti and Ti samples coated with five different TiO2 nanotube films. The corresponding corrosion parameters are summarized in Table 2. It is seen that the bare Ti sheet shows the lowest Ecorr value and the highest icorr value, demonstrating that five Ti samples treated by anodization have enhanced corrosion resistance properties. Among five different TiO2 nanotube films, sample d (three-step anodization) displays the lowest corrosion rate, whereas sample e (the second-step anodization was carried out in the fluoride-containing electrolyte at 20 V) has the highest corrosion rate. The corrosion rate in the artificial saliva is increased in the order of sample d < sample c < sample b < sample a < sample e (Table 2). As described above, sample d obtained by three-step anodization still possesses a double-layer structure (i.e., nanotube layer and compact oxide layer) rather than the triple-layer structure (Fig. 2c, d). However, unlike sample c (the second-step anodization was carried out in the fluoride-free electrolyte), the thermal annealing appears to have no effect on its morphology. The compact oxide layer between TiO2 nanotubes and Ti substrate still exists after thermal annealing. Therefore, it is concluded that the superior corrosion resistance properties of sample d can be attributed to the existence of the compact and defect-free oxide layer. As for sample a (single-step anodization) and sample e, their relatively decreased corrosion resistance properties may be associated with the cracked nanotube walls and the less-dense nanoparticulate layer (Figs. 1b, 2f), because this type of film morphology can facilitate the permeation of artificial saliva.

Figure 5
figure 5

Typical diagrams of the potentiodynamic polarization for various samples

Full size image
Table 2 Corrosion parameters from polarization plots
Full size table

Their corrosion resistance properties were further investigated by EIS technique. Figure 6 shows the impedance spectra for various samples in the form of Bode plot. The fitting curves are represented by a solid line and the experimental data are denoted by an individual symbol. A best fitting of the experimental impedance data can be obtained by using equivalent circuits as shown in the insets of Fig. 6a, e. For bare Ti, the inset of Fig. 6a shows its equivalent circuit. As for five different TiO2 nanotube films, they all can be seen as a double-layer structure of nanotube layer and compact layer, which has been demonstrated by the above SEM observations (Figs. 1, 2 right). Hence, they can be modeled by the same equivalent circuit shown in the inset of Fig. 6e. In the models, Re is the resistance of the test solution. Rt and Rb denote the resistances of the upper tube layer and lower barrier layer of TiO2 nanotube films, respectively. Because of the presence of heterogeneities for the nanotube films, the measured capacitive behavior is not generally ideal. A constant-phase element (CPE) is then introduced in this work in addition to an ideal capacitor. It is known that the impedance of CPE is given by ZCPE = Y−1(jω)−n, where Y is a frequency-independent fit parameter, n the CPE exponent, ω (= 2πf) the angular frequency. The exponent n has the meaning of a phase shift, which lies between 0 and 1. At n = 1, the CPE describes an ideal capacitor with an impedance of (jωY)−1.

Figure 6
figure 6

Bode plots of samples in the artificial saliva at 37 ± 1 °C: a bare Ti, b sample a (single-step anodization), c sample b (two-step anodization), d sample c (the second-step anodization was carried out in the fluoride-free electrolyte), e sample d (three-step anodization), f sample e (the second-step anodization was carried out in the fluoride-containing electrolyte at 20 V). Insets show the equivalent circuits used to fit experimental data

Full size image

Table 3 lists the fitted parameter values for bare Ti and Ti samples coated with five different TiO2 nanotube films. As expected, the Re (solution resistance) value is similar for all these samples because of the utilization of the same test solution (artificial saliva). It has been generally accepted that a higher barrier layer resistance (Rb) suggests a better corrosion resistance [22, 23, 32]. As can be seen from Table 3, bare Ti has the lowest Rb value, indicating very poor corrosion resistance properties. For five TiO2 nanotube films, the Rb value is decreased in the order of sample d < sample c < sample b < sample a < sample e, which is also fully consistent with the trend observed in previous corrosion rate (Table 2). Similarly, sample d (three-step anodization) has the highest Rb value among five TiO2 nanotube films, also demonstrating the optimized corrosion resistance.

Table 3 Best fit impedance parameter values for the samples in artificial saliva
Full size table

Conclusions

We found that after the anodization treatments, five Ti samples coated with various TiO2 nanotube films all exhibit the markedly enhanced corrosion resistance properties compared to bare Ti. Sample d (three-step anodization) and sample c (two-step anodization) possess superior corrosion resistance among all samples, both of which have performed the second-step anodization in the fluoride-free electrolyte. It is believed that the excellent corrosion resistance ability can be ascribed to the formation of the compact oxide layer between nanotubes and Ti substrate during the second-step anodization in the fluoride-free electrolyte. Besides, sample a (single-step anodization) and sample b (two-step anodization) prepared with the identical anodization parameters have distinctly different corrosion resistance, although they exhibit nearly identical morphology of nanotubes for the as-anodized samples. The better corrosion resistance of sample b as compared with sample a can be attributed to the morphological difference between them after thermal annealing. For sample e obtained by two-step anodization in the same fluoride-containing electrolyte at different voltages, the underlying nanotube layer with smaller tube diameters was converted to a layer of nanoparticulate aggregations after thermal annealing, which is likely responsible for the relatively poor corrosion resistance.