1. Introduction
Due to their inherited unique relationship of sound biocompatibility, excellent chemical stability and high specific strength, titanium and titanium alloys exhibit a great variety of applications covering from aerospace and biomedical to nuclear waste. However, a major barrier for applications of titanium is its poor tribological characteristic, particularly when high wear resistance is recommended. Furthermore, the formed oxide films on titanium are sensitive to damage not only by halide anions but also in other media, such as sulfate solutions [
1,
2,
3]. To overcome these challenges, various surface modification methods have been applied to titanium and titanium alloys. It has been reported that different surface modifications of titanium often show better corrosion resistance. For instance, Liu et al. reported that the composite film TiC-Ti
5Si
3 fabricated on substrate Ti6Al4V shows corrosion current density one order lower in magnitude than the untreated Ti alloy in simulated artificial sea water [
4]. It is important to understand that for the treated material structural defects presented frequently (e.g., micropores and microcracks) in the coating, these structural defects provide the path by which the active species diffuse and damage the substrate. Hence, new methods for surface modification and new coatings are required to further improve both the mechanical and chemical stability of the oxide films [
5,
6].
Recently, nanostructured titanium oxide with dual phases (anatase and rutile) has attracted widespread attention to fabricate a structural coating used under harsh conditions due to its low density, excellent oxidation resistance and adequate strength to weight ratio. In the attempts to achieve nanostructured titanium oxide with dual phases, numerous surface modification methods, such as thermal spraying, electrochemical anodization and thermal oxidation, have been reported, while each of the methods exhibits its own limitations and values [
7,
8,
9]. Thermal spraying is a powerful technique in yielding thick titanium oxide coatings, but it is difficult to apply to substrates with intricate shapes. Anodization is an electrochemical technique that has been effectively demonstrated for the surface treatment of titanium and its alloys. Anodization is defined by the rate of formation and dissolution of oxide film specified by the selection of electrolyte and its temperature. By changing the parameters of the anodization process, titanium oxide films with different phases can be formed on Ti substrates [
10,
11]. In the case of sulfuric acid, the oxide formation rate is much more dominant than its dissolution rate. Through anodization, the formed crystalline oxide is denser and thicker than the native oxide, and the properties of the oxide, such as thickness, morphology, crystallinity, color and porosity, may be controlled by the operation specifications. Sul et al. reported that a reduction in the current density appeared by altering the temperature and concentration of electrolyte [
12]. Some of the advantages observed by applying anodizing processes include reduced cost (no necessity of gas shielding or costly vacuum conditions); simple and easy operation variables to fabricate the required coatings; ecological favorability, since there is no harmful emission involved in the operation; good adhesion; and easy incorporation with different surface treatments, such as thermal oxidation [
13,
14].
Increased corrosion resistance is obtained for anodized titanium by thermal oxidation. Thermal oxidation has been found to be an attractive method which allows the expansion of the anodized film to establish a dense crystalline film dependent on the applied temperature. Anodic films subjected to annealing at 550–600 °C form dual phase structures, which show efficient adhesion and higher corrosion resistance than amorphous or any single phase TiO
2 [
13,
15]. However, high temperatures beyond a certain limit and prolonged heating will minimize the functional properties. Prolonged annealing at 800 °C leads to massive building of the oxide film, which ultimately spalls off from the surface due to the difference in the thermal expansion coefficient and lattice mismatch between the oxide and Ti substrate. Hence, it is essential to modify the treatment time and temperature so as to produce thick, homogenous and adherent film, preferably with a dual phase structure [
15].
It has been found that the anodization and thermal oxidation of titanium and Ti alloys improve the resistance against degradation. For example, Cimenoglu and Guleryuz [
3] reported that commercially pure Ti subjected to thermal oxidation at 600 °C for 60 h showed good inhibition efficiency in hydrochloric acid solution. The result showed that the corrosion resistance of the oxide layer was improved by minimizing the anodic dissolution as compared with untreated Ti. John et al. [
9] discovered that Ti6Al4V subjected to anodization followed by thermal oxidation showed superior corrosion resistance to the samples treated separately with anodization or thermal oxidation. This may be because anodization was effective in producing anatase/amorphous TiO
2, while thermal oxidation efficiently converted the oxides into dense dual anatase/rutile structures, exhibiting improved degradation resistance. The above methods were employed mostly to modify Ti and its alloys for the improvement of biomedical applications, and no organized study was found on Ti and its alloys at high temperatures in strong acid environments. Our recent study [
16] found that anodized TiO
2 films followed by heat treatment at 600 °C built protective layers that showed an anatase/rutile dual-phase structure and excellent corrosion resistance to strong acidic solutions at high temperatures. Based on previous work, in the present study, the hybrid treatment conditions, particularly the heat treatment time, are further investigated to obtain the best corrosion resistance. The corrosion resistance of the obtained oxide films on Ti are studied in 4 M H
2SO
4 and 4 M HCl solutions at 100 °C, using conventional electrochemical techniques containing potentiodynamic polarization, electrochemical impedance spectroscopy (EIS) and Mott–Schottky analysis.
3. Results and Discussion
Figure 1 illustrates the surface morphology of groups X, Y
1, Y
2 and Y
3, respectively (
Table 1). Group X showed porous and irregular TiO
2 particles with uneven distribution at the whole surface. Groups Y
1, Y
2 and Y
3 showed very similar morphology. Moreover, from group Y
1 to Y
3, the amount of porous structure reduced without any change in the morphology, which implies that the grains of the film had adequate thermal stability for a prolonged time period at high temperatures. The magnified view of
Figure 1a–d shows a very clear perspective of the film structure; the layers seem to be irregular in shape and size, indicating that a rough surface was formed with a high specific surface area [
17,
18].
Figure 2 represents the EDS results for the sample surface of groups X, Y
1, Y
2 and Y
3, respectively. As seen from
Figure 2a–d, the prepared films are mainly made up of titanium and oxygen. The EDS spectrum of group X shows that Ti peaks are much higher than those of groups Y
1, Y
2 and Y
3. Groups Y
1, Y
2 and Y
3 represent enrichment in oxygen with respect to the thermal oxidation time; the maximum enrichment was seen in group Y
3. As reported elsewhere [
19], the increment in the TiO
2 particles may be due to the existence of Ti vacancies in the film. According to Vennekamp et al. [
20], the growth mode for the formation of TiO
2 is controlled by the movement of ionic species. If morphologically uneven growth occurs, then Ti ions are more mobile; once oxygen ions are mobile, stable growth is observed.
Typical X-ray diffraction spectra obtained from all groups and the comparison with the standard powder diffraction patterns for TiO
2 (anatase or rutile) are shown in
Figure 3. For group X, the sharp diffraction peaks exhibited nonstoichiometric and amorphous TiO
2 along with few peaks of α-Ti, which implies that an uneven thin film was produced. The XRD spectrum of group Y
1 was very similar to those of groups Y
2 and Y
3, except for the broadening of peaks. Moreover, some new diffraction peaks were also seen in groups Y
2 and Y
3, suggesting that the presence of different phases increased with a decrease in crystalline size [
9]. It was reported that the annealing of TiO
2 at 580–600 °C altered the structure into two polymorphic forms of TiO
2 (anatase and rutile). In groups Y
1, Y
2 and Y
3, rutile peaks were noticed at 2θ of 36.0 (101), 41.2 (111), 54.3 (211), 62.8 (002) and 70.0 (112), while the new peaks found at 2θ of 27.4 (110), 75.9 (301) and 82.1 (303) are presented as rutile and anatase phases, respectively. In group Y
1, the peak that appeared at 2θ of 38.0 (004) represents the anatase phase. Careful examination showed that the XRD spectra of Y
1, Y
2 and Y
3 moved towards lower values of the Bragg angle as compared with the standard powder diffraction pattern, suggesting the interstitial areas being saturated by oxygen. Furthermore, in group Y
3, the peak intensities of the anatase phase were obviously higher than those in groups Y
1 and Y
2, indicating that with longer treating time at the testing temperature, more anatase phase in the oxide film was formed [
21].
To gain a deeper understanding of the effect of thermal oxidation time at constant temperature, XPS spectra were measured on the surfaces of groups X, Y
1, Y
2 and Y
3. The binding energy values represent the valence state and the density of charge near the atoms. In
Figure 4a, the Ti 2p peaks for sample X show the typical binding energy values of Ti
+4, but the broadened shape on the low energy side reveals the presence of some low value ions of Ti, such as Ti
3+ and Ti
2+, similar to previous reports on anodized titanium [
14]. After thermal oxidation, the Ti 2p peaks remained in similar shapes, while the major change was observed in O 1s spectra, as presented in
Figure 4b. The oxygen spectrum of group X displayed three individual peaks: the main peak located at 529.8 eV is associated with O
2− in the oxide, and the two individual peaks clearly resolved in the range of 531–532.7 eV can be attributed to oxygen in the OH group and in H
2O in the surface layer. This is also consistent with previous authors [
2,
4]. After thermal oxidation, the peak intensity of O
2− in OH and H
2O obviously increased (
Figure 4b). A possible explanation for this is that at high temperatures, oxygen vacancies are generated, and more vacancies are produced if the heating time period is prolonged. In Nishikawa’s study [
22], in thermally oxidized Ti film, the atomic ratio of O/Ti decreased with heating, which indicated that the oxygen vacancy in TiO
2 increased with heating. The generation of Ti
3+ in TiO
2 at high temperatures also supports this conclusion. These vacancies can be filled in terms of hydroxyl groups by the dissociation of water that exists in the passive layer, which has been confirmed by the reported literature. Zhang et al. reported that anatase TiO
2 was exposed to UV light irradiation and a H
2O environment simultaneously for 60 h, and even after this, the bulk anatase TiO
2 was preserved due to the formation of an amorphous hydroxylation layer that was very stable and massively hydroxylated over time [
23]. He also showed that water plays a significant role in the transformation of titania. The OH peaks that appeared in groups Y
1, Y
2 and Y
3 also confirm this result. According to the TiO
2-H
2O Pourbaix diagram,
is the main corrosion product in TiO
2 films that increases its stability potential in a strong acidic region [
24]. The important evidence via XPS analysis shows that thermal oxidation promotes the existence of OH groups on the surface; thus, the TiO
2 structure is protected by the formation of titania hydroxide. This effectively explained why group Y
3 had a superior resistance to corrosion.
3.1. Potentiodynamic Scan
Figure 5 illustrates the polarization curves of groups X, Y
1, Y
2 and Y
3. The Tafel method was used to determine the E
corr and I
corr of various samples in 4 M H
2SO
4 and 4 M HCl solutions and shown in
Table 2. The measured corrosion potential at 100 ± 5 °C for group X in 4 M H
2SO
4 and 4 M HCl solutions was about -0.62 and -0.70 V
SCE, and active-passive behavior was seen. For groups Y
1, Y
2 and Y
3, the corrosion potential increased in both solutions, and spontaneous passivation was achieved. The passive current densities were five orders of magnitude lower than those of group X, showing that the anodizing plus thermal oxidation treatment greatly improved the corrosion resistance of groups Y
1, Y
2 and Y
3. Group Y
3 showed the lowest passive current density in both solutions, but similar passive current density was also found for group Y
2 in 4 M HCl. This may be due to the fact that TiCl
3 was formed in HCl solution, which hindered the dissolution reaction [
24,
25]. The improvement in corrosion resistance by anodizing-thermal oxidation may be due to the formation of a stable hydroxide film, which acts as a barrier layer.
3.2. Electrochemical Impedance Spectroscopy
Electrochemical impedance spectroscopy (EIS) is a common method to explore the electrochemical behavior and the reaction kinetics of complicated electrode systems, through which the properties of the electrode materials may be reflected. By careful examination of the EIS data provided by equivalent circuit (EEC) analysis, the entire figure of corrosion procedure can be drawn [
4].
Figure 6 displays the EIS plots of groups X, Y
1, Y
2 and Y
3. All the Nyquist plots showed semicircles with different diameters, indicating that all curves were obtained under similar corrosion processes, but possessed different corrosion rates. After thermal oxidation, the diameter of the semicircle increased significantly compared with that of the anodized sample. In 4 M HCl and 4 M H
2SO
4 solutions, samples Y
1 and Y
3 showed the largest semicircle diameters, respectively, indicating the highest corrosion resistance. Bode plots of groups Y
2 and Y
3 are very similar as compared with group Y
1. According to the bode phase plots of groups Y
2 and Y
3 at high and low frequencies, they reflected the capacitive features in the entire frequency range, while group Y
1 showed capacitive and resistive responses in low and high frequency regions. The corrosion rate of the specimen according to the bode plots at low frequency accelerated in the order Y
3 < Y
2 < Y
1 < X.
For the electrochemical corrosion evolution of groups X, Y
1, Y
2 and Y
3 by EIS, two time constants were seen in the bode phase plots of all groups; thus, the equivalent electrical circuit (EEC) model (R
s(Q
1(R
1(Q
2R
2)))) shown in
Figure 6e was used to calculate the parameter values. The choice of the equivalent electrical circuit is dependent on the behavior of the plots, as seen from
Figure 6c,d; two time constant were seen, and the fitted data were in good agreement with experimental data throughout the entire frequency range [
26]. As reported previously, [
4,
9] the most appropriate circuit for the analysis of TiO
2 is (R
s(Q
1(R
1(Q
2R
2)))). In the electric circuit (R
s(Q
1(R
1(Q
2R
2)))), R
s is the solution resistance, Q
1 is the constant phase element (CPE) for the double layer and R
1 belongs to resistance at the passive layer via charge transfer at the solution interface, and R
2 and Q
2 belong to the resistance and capacitance of the passive layer, respectively. It should be noted that many different justifications have been proposed regarding the circuit in the previous literature. For example, Heakal et al. [
27] proposed that the time constant at high frequency related to the capacitance of the double layer and charge transfer resistance may be attributed to the electrochemical process at the solution interface of the passive layer, while the time constant at low frequency is associated with the barrier properties of the passive layer. Some reported data have proposed that the circuit element Q
2R
2 indicates the impedance of the passive layer, while Q
1R
1 represents the electrochemical response of the passive layer. In the present study, values calculated by equivalent capacitance at high frequencies refer more to the Helmholtz double layer in lieu of the passive layer, because they are generally reported as double layer capacitance values [
28].
In
Table 3, for all specimens, the passive layer resistance (R
2) in both solutions was higher than the resistance at the solution interface (R
1) of the passive layer, indicating that the ionic transport that should rely on the passive layer was predominant, instead of resistance at the solution interface (R
1) of the passive layer. Based on the analysis of passive layer resistance, group Y
3 showed significantly higher corrosion resistance than X, Y
1 and Y
2 in both solutions. Potucek et al. [
29] suggested that the resistance values obtained from EIS data are strongly dependent on the selected solution, while the capacitance (C) is not changed by solution conditions and can yield reliable evidence for the related corrosion properties. The value of capacitance (C
2) for the inspected groups can be acquired from (Q
2) work conducted by Hsu and Mansfeld’s using Equation (1) [
30]. An approximate estimation for the passive layer thickness (
) was calculated by using Equation (2)
In Equations (1) and (2),
R and
C show passive layer resistance and capacitance, respectively. ε
0 (8.85 × 10
−14 F/cm) represents vacuum permittivity, and ε displays the passive layer’s dielectric constant (ε = 48 for anatase and 110 for rutile). In
Figure 3, group X does not show either rutile or anatase peaks, so their values of thickness for the passive layer were not computed here. The ε is taken to be 79 for groups Y
1, Y
2 and Y
3 (the average value of rutile and anatase), due to the anatase or rutile phase [
6]. The calculated thickness of the passive layer increased in the order X < Y
1 < Y
2 < Y
3.
3.3. Mott–Schottky Analysis
As is well known, the growth and perturbation potential of passive layers mostly rely on the electronic and ionic transport mechanism, and may be measured by the electronic properties of the passive layer. So, it is relevant to examine the electronic behavior of the passive layer to determine the appropriate corrosion process. The Mott–Schottky test is dependent on the experiences of the nonideal semiconductor path followed by the passive layers, and has been demonstrated to be a valid method in studying electronic properties, as defect density can be measured in the passive layer. By Mott–Schottky curves, the connection in (1/C
2 vs. E) the passive layer may be developed to measure the donor density (N
D) and flat band potential (E
FB) [
4]
Here, e shows the charge of the electron, ε
0 represents the vacuum permittivity and ε displays the passive layer dielectric constant. In
Figure 3, group X does not show either rutile or anatase peaks, so both donor density and flat band potential values were not computed here. The value of the dielectric constant for groups Y
1, Y
2 and Y
3 is considered to be 79. K stands for Boltzmann constant; T and E
FB are the temperature of the test solution and flat band potential, respectively. The term kT/e can be ignored, since it is only 32 mV at the required test solution condition. By re-arranging Equation (3), the donor density can be measured and may be obtained by the slope of the best linear fit line as a function of potential of the experimental C
−2:
The extrapolation of the linear portion to C
−2 = 0 can be used normally for the measurement of the flat band potential. The measured donor densities of groups Y
1, Y
2 and Y
3 in both test solutions are presented in
Table 4. By using this method, the measured donor density indicates the density at the interface of film, whereas the highest concentration of oxygen vacancies and interstitial metal atoms is predicted [
4]. From the point defect model (PDM), the whole corrosion process through passive the film is governed by the movement of cation vacancies/or oxygen vacancies, which deduces the resistance to corrosion. So, the main cause is the passive current density that can be determined by the conduction of ions. As reported elsewhere, the greater the concentration density of carriers, the faster the passive film conduction. Consequently, the greater density of a donor will precede higher current density values, which is comparable with the polarization measurement in
Figure 5. Normally, the literature N
D values rely on the film thickness and electrolyte, and are found to fall in the range 10
18 to 10
23 cm
−3 [
31]. The previous results are also acceptable, in agreement with the present results that group Y
3 has the minimum donor density value.
Figure 7 shows that the donor density for group Y
3 decreased significantly, which may be due to the hydroxide covered TiO
2 structure. The measured donor density values can be manipulated in both solutions from the highest to the lowest as X > Y
1 >Y
2 > Y
3. The remarkable reduction in the donor density and the prominent improvement in the flat band potential of group Y
3 imply that the anodization and thermal oxidation improved the protective ability of the films significantly.
In summary, the present work confirms that by anodizing plus thermal oxidation at 600 °C, a protective oxide film was formed on titanium. The film showed good stability in strong acidic solution at a high temperature (100 °C). With the oxidation time prolonged from 3 to 9 h at 600 °C, the corrosion resistance of the oxide showed an increase to a certain degree. The significantly improved corrosion resistance may be attributed to the formation of a rutile/anatase duplex structure, the increased hydroxides on the surface and the decreased donor density in the oxides.