Materials Science and Engineering C 49 (2015) 567–578

Contents lists available at ScienceDirect

Materials Science and Engineering C journal homepage: www.elsevier.com/locate/msec

Titania nanotubes from weak organic acid electrolyte: Fabrication, characterization and oxide film properties Balakrishnan Munirathinam ⁎, Lakshman Neelakantan Department of Metallurgical and Materials Engineering, Indian Institute of Technology Madras, Chennai 600036, India

a r t i c l e

i n f o

Article history: Received 30 September 2014 Received in revised form 3 December 2014 Accepted 8 January 2015 Available online 10 January 2015 Keywords: Titanium Anodic oxidation TiO2 nanotubes Wettability Impedance Mott Schottky analysis

a b s t r a c t In this study, TiO2 nanotubes were fabricated using anodic oxidation in fluoride containing weak organic acid for different durations (0.5 h, 1 h, 2 h and 3 h). Scanning electron microscope (SEM) micrographs reveal that the morphology of titanium oxide varies with anodization time. Raman spectroscopy and X-ray diffraction (XRD) results indicate that the as-formed oxide nanotubes were amorphous in nature, yet transform into crystalline phases (anatase and rutile) upon annealing at 600 °C. Wettability measurements show that both as-formed and annealed nanotubes exhibited hydrophilic behavior. The electrochemical behavior was ascertained by DC polarization and AC electrochemical impedance spectroscopy (EIS) measurements in 0.9% NaCl solution. The results suggest that the annealed nanotubes showed higher impedance (105–106 Ω cm2) and lower passive current density (10−7 A cm−2) than the as-formed nanotubes. In addition, we investigated the influence of post heat treatment on the semiconducting properties of the oxides by capacitance measurements. In vitro bioactivity test in simulated body fluid (SBF) showed that precipitation of Ca/P is easier in crystallized nanotubes than the amorphous structure. Our study uses a simple strategy to prepare nano-structured titania films and hints the feasibility of tailoring the oxide properties by thermal treatment, producing surfaces with better bioactivity. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The synthesis of nanostructured titanium oxide through anodic oxidation for biomedical applications has been investigated by many authors owing to their interesting features like large surface area, biocompatibility/bioactive properties, non-toxicity, corrosion resistance and adequate mechanical properties [1,2]. This method can control the titanium oxide formation to nanoscale dimensions. Over the years, four generations of fluoride containing electrolytes — aqueous solution of acidic pH [3], aqueous solution of neutral pH [4], organic [5,6] and chloride in both aqueous [7] and organic [8] based electrolytes have been utilized to develop nanotubes of varied dimensions. TiO2 nanotubes grown on titanium render increased surface area, which accelerates bone growth in orthopedic and dental implants [9]. Such a surface layer enhances cell adhesion, cell proliferation and protein adsorption [10,11]. Hydroxyapatite formation, which is considered to be an essential step for the bone-binding ability of biomaterials, is enhanced in porous surfaces as compared to flat compact surface [9,12]. Controlling the size and arrangement of pores helps bone growth and this promotes the use of titania nanotubes for biomedical applications. A mixed crystalline nanotube (anatase + rutile) structure showed better biocompatibility than the amorphous nanotubes [12]. Various researchers focused on the effect of anodization time and voltage on ⁎ Corresponding author. E-mail address: [email protected] (B. Munirathinam).

http://dx.doi.org/10.1016/j.msec.2015.01.045 0928-4931/© 2015 Elsevier B.V. All rights reserved.

the surface morphology and studied the film properties [5,13]. The passive film behavior of titanium and its alloys plays a crucial role for its suitability as an implant material. The kinetics of corrosion process can be reduced by the passive film which blocks the diffusion of aggressive ions present in the body fluids to the substrate metal [14]. It has been shown that the blood compatibility of heart valve materials is influenced by the semi-conducting nature of non-stoichiometric titanium oxide films [15]. Hence, the rate of active dissolution of ions strongly depends on the chemical composition and defective nature of the oxide layer. Though few studies have reported on the passive film and semiconducting properties of the titanium oxide in a biomedical perspective [15–17], there is no report which provides an overall picture on the synthesis and on the variation of semiconducting properties of titania nanotubes subjected to thermal treatments (annealing). In this respect, we experimentally investigated and compared the as-formed and annealed nanotubes and an attempt is being made to characterize the variation in the morphology and passive film properties of the oxide nanotubes in saline solution. 2. Experimental procedure 2.1. Sample preparation and anodization High purity titanium sheets (Grade 2, 99.99%) of dimension 20 × 10 × 1 mm3 were employed in this work. The sheets were

568

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

metallographically prepared, ultrasonically cleaned with acetone and then rinsed in distilled water. The electrolyte used in this work was 0.1 M citric acid with 0.5 wt.% NaF at an initial pH 4. Before anodizing, the polished titanium surfaces were etched in a solution containing 1:4:5 ratio of HF:HNO3:H2O. All solutions were prepared from reagent grade chemicals and distilled water. A surface area of 2 cm2 was exposed to the electrolyte. Oxide films were grown by anodizing in a two electrode electrochemical cell with titanium as working electrode and graphite rod as counter electrode. The experiments were performed at room temperature and a constant potential of 20 V was applied for varying time durations of 0.5, 1, 2 and 3 h, respectively. After anodization, the samples were rinsed with distilled water and dried. Another set of samples were heat treated at 600 °C by holding them for 3 h and subsequently furnace cooled. 2.2. Surface characterization The surface morphology and cross sectional features of anodized surfaces were examined using field emission scanning electron microscopy (FE-SEM FEI Quanta FEG 400). Energy-dispersive X-ray spectroscopy (EDS) fitted to SEM was used to analyze the composition of the oxide films. Surface roughness of the anodic oxide layers were measured using 3D noncontact surface optical profiler (Bruker ContourGT Inmotion). Image analysis and processing was performed using computer-aided software analysis (vision analysis). X-ray diffraction (XRD) measurements (X'Pert PRO, PANalytical) were carried out using a incident Cu Kα radiation (λ = 1.54 Å), tube voltage 30 kV at a scan speed of 5°/min. Raman spectrum was recorded using a micro Raman Spectroscopy (Horiba Jobin Yvon, model HR800UV) instrument with He–Ne laser (632.81 nm) and analyzed with grating of 600 lines/in. The acquisition time of 5 and 30 s were used in the analysis with 10 mW incident power. Contact angle measurements were carried out on the as-formed and annealed surfaces using a contact angle detect system (Easy Drop, KRUSS, Germany) with distilled water. Water droplet (0.5 μl) was dispensed onto the surface with a syringe and the digital images of the droplet silhouette were captured with a CCD camera (KB-1380) within 10 s after wetting. All data reported here are expressed as mean ± standard error, and analyzed statistically by appropriate t-test. A value of p b 0.05 was considered significant.

of 50 mVs−1. The electrochemical measurements were performed at least three times to confirm reproducibility. 2.4. Evaluation of apatite inducing ability As-formed and annealed samples were immersed in simulated body fluid (SBF) with pH 7.4, whose ion concentrations (Na+ 142.0, K+ 5.0, 2− Mg2 + 1.5, Ca2 + 2.5, Cl− 147.8, HCO− 1.0, SO24 − 0.5 mM) 3 4.2, HPO4 are nearly equal to those of human blood plasma at 36.5 °C [18]. The SBF was prepared by dissolving reagent grade chemicals of NaCl, NaHCO3, KCl, K2HPO4·3H2O, MgCl2·6H2O, CaCl2 and Na2SO4 in distilled water and buffered at pH 7.4 with tris(hydroxymethyl)aminomethane and 1 M HCl at 37 °C. 30 mL of SBF was added to the beaker to maintain a ratio of the specimen surface area per volume (SA/V) of 0.1 cm−1 [18]. After soaking in this medium for 5 days, the samples were removed, washed with distilled water and then dried. 3. Results and discussion 3.1. Morphological characteristics of TiO2 nanotubes Fig. 1a–d shows the surface morphology of the samples anodized for different time durations (0.5, 1, 2 and 3 h) in 0.1 M citric acid containing 0.5 wt.% of NaF electrolyte. In order to convert amorphous oxides into crystalline, all samples were annealed at 600 °C for 3 h. The SEM micrographs of the annealed samples are shown in Fig. 2b. The average inner and outer pore diameters (Di and Do), interpore distance (w), wall thickness (t) and tube length (L) were obtained using image J software. In pore diameter calculations, pore shapes were assumed to be perfect circles. The total surface area of the nanotube (A), pore density (n), porosity of the lattice (P), specific surface area (As) can be estimated from the following expression [19,20],   2 2 A ¼ 2π Do  Di þ 2πLðDo þ Di Þ

n¼

1014 ð3Þ0:5  D2o

ð1Þ

ð2Þ

 2 D P ¼ 0:90 i Do

ð3Þ

2.3. Electrochemical characterization

As ¼ n  A:

ð4Þ

Potentiodynamic polarization and impedance spectroscopy were performed using a Gamry Potentiostat (Reference 600–14083) controlled by a personal computer and softwares (Gamry Framework and Gamry Echem Analyst). A conventional three electrode cell was used for all the electrochemical measurements. A saturated calomel electrode (SCE) was used as the reference electrode, graphite rod as counter electrode and titanium as the working electrode. All experiments were carried out at room temperature in an electrolyte of 0.9% NaCl at pH 7.2. The working electrode was pressed against an O-ring sealing from outside of the cell exposing an area of 0.5 cm2. All potentials reported in this paper are with reference to SCE. Before the electrochemical tests, all samples were immersed in electrolyte for a time period of 3 h in order to attain a stable open circuit potential (OCP). Potentiodynamic polarization measurements were performed from −0.5 VSCE to +1 VSCE at a scan rate of 1 mVs−1. Electrochemical impedance measurements were performed under open circuit conditions in the frequency range of 0.1 MHz–10 mHz with an excitation voltage of 10 mV (peak-to-peak). The measured spectra were fitted with an equivalent circuit using Gamry Echem Analyst software. The space charge capacitance measurement was carried out using the Mott Schottky analysis at a frequency of 1 kHz by sweeping the potential in the negative direction from 0.6 VSCE to −0.5 VSCE at a scan rate

The morphological parameters of anodized titanium grown at different time intervals are presented in Table 1. Tube length does not vary with anodization time even when the specimens were anodized for 2 h, whereas the other dimensions vary. The levels of porosity measured were in the range of 35–50%. After annealing the as-formed nanotubes, the wall thickness and porosity decreases and this will be explained in later sections. Standard deviations of the values are reported and the difference observed is insignificant. Further, BET measurements would be helpful in comparing the total surface area of the fabricated nanotubes with the existing calculations. The cross-section and bottom views of TiO2 nanotubes are shown in Fig. 2a. Variation in wall thickness is attributed to the current oscillations which arise due to the change in pH at the pore tip and thereby transitory increase in the dissolution rate [21]. The obtained nanotubes are straight and hollow with wrinkled walls. Increased water content in the electrolyte is responsible for the formation of ripples in the tube walls. From the bottom view, it is clear that the semispherical tubes generate dimple structures on the titanium surface underneath and closed with a barrier oxide layer. Hence the bonding strength between these interfaces can be improved due to the ‘cup- and cone-like’ structure formation. Also, it was observed that nanotube arrays maintain highly ordered structures when the film is removed from the substrate.

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

569

Fig. 1. FE-SEM micrographs of the surface morphologies of TiO2 nanotubes anodized for different durations a) 0.5 h, b) 1 h and c) 2 h d) 3 h.

Good adhesion strength between titanium/oxide interfaces is essential for implant applications. 3.2. Crystallinity of TiO2 nanotubes Annealing was carried out at 600 °C for 3 h to crystallize the amorphous nanotube arrays formed as a result of anodization. FE-SEM micrographs presented in Fig. 2b reveal that heat treatment has a significant influence on the nanotube arrays. After annealing, the surface structure of TiO2 nanotubes tend to shrink and the tube diameter decreases whereas, the tube wall thickness increases. Since sintering is an activation phenomenon, the surface with higher surface area and surface energy tends to be thermodynamically metastable and will change its morphology when it is thermally activated. At very high temperature, collapse and cracking of nanotubes was observed and this tends to destroy the nanotubular structure [22]. Phase transformation from an amorphous to crystalline structure in the anodic nanotube film was further investigated by micro Raman spectroscopy and X-ray diffraction technique. Fig. 3A shows the Raman spectrum of as-formed and annealed samples. The absence of Raman peaks for the as-formed sample indicates the amorphous nature of the oxide. However, the anatase and rutile signals are evident for all anodized surfaces subjected to annealing and the enhancement of crystallinity is observed with

anodization time. For the annealed samples, all the spectra consist of a strong peak around 146.6 (± 2.13) cm− 1 are assigned to the Eg phononic mode and four bands around 198.1 (± 1.2), 396.1 (± 3.1), 511.6 (± 1.1) and 633 (±2.4) cm−1, are attributed to the Eg, B1g, A1g or B1g and Eg modes of anatase phase, respectively. The Raman bands at 436 (± 3.1) and 610 (± 2.4) cm− 1 correspond to the Eg and A1g modes of rutile phase. Peak positions observed are in good agreement with those previously reported by other authors [23,24]. In order to see the non-apparent features of the spectrum, the acquisition time was increased to 30 s and the Raman spectrum was recorded in preference to less intensity peaks as shown in Fig. 3B. The co-existence of anatase and rutile phase is clearly established at higher acquisition time. The increase in anodization time enhances the crystallinity and the oxide thickness as inferred from the FWHM and intensity of Raman peaks. In addition, the dominant mode (146.6 cm−1) also shifts towards a lower wavenumber with anodization time which can be attributed to modulation of unit cell through increase in crystallinity. Our observation is in accordance with earlier reports of the effect of strain and degree of crystallinity on Raman peak [25]. The XRD pattern of the as-formed and heat treated specimens are shown in Fig. 4. The as-formed sample showed peaks of titanium indicating that the oxide obtained is amorphous. Both rutile and anatase phases were observed when annealed at 600 °C. The weight fractions

570

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

Fig. 2. FE-SEM micrographs of the cross sectional view of the TiO2 nanotubes obtained for 0.5 h and 2 h b) surface morphology of 1 h and 2 h anodized samples followed by annealing at 600 °C for 3 h.

of anatase (XA) and rutile (XR) phases obtained for different anodization times were calculated from the intensity of XRD peaks using Spurr and Myers' equation [16,26], XR ¼

1

  I 1 þ 1:26 R IA

X A ¼ 1‐X R

ð5Þ

ð6Þ

where XA and XR are percentage contents of rutile and anatase, IR and IA are intensities integral intensities of (110) anatase and (101) rutile, respectively. The amount of anatase and rutile content are as follows: XA-0.72, XR-0.28 (0.5 h); XA-0.58, XR-0.42 (1 h); XA-0.47, XR-0.53 (2 h); and XA-0.49, XR-0.51 (3 h). The amount of rutile and anatase phase variation can be correlated with increasing and decreasing rutile and anatase signals. With increasing anodization time, the amount of rutile phase increases but there is no significant difference in samples anodized for 2 h and 3 h, respectively (p N 0.05). Thickening of barrier layer (metal–nanotube interface) is possible at titanium-nanotube due to thermal oxidation at 600 °C. We expect that barrier layer thickness would have increased with increasing anodic oxidation time and further heat treatment favors rutile phase nucleation at metal-nanotube interface and may not be formed in nanotube structure. A similar observation on the rutile formation was recorded by Regonini [27]. Rutile nucleation will not happen at the inner wall of the TiO2 nanotubes having a wall thickness less than 30 nm [28]. Anatase phase tends to exist at this

annealing temperature and this phase transformation is initiated at the defect sites. Activation energy for the formation of anatase can be correlated with wall thickness and surface area. The nanotubes with thicker walls and having a higher surface area possess a higher density of distribution of defect sites in the bulk that aids the heterogeneous nucleation of anatase crystallites, and make the anatase particles grow quickly. Morphological changes promoted by annealing are due to the quick growth of anatase crystallites and transformation of anatase to rutile happened earlier at 600 °C due to the coalescence of anatase crystallites [27]. It is reported that, the TiO2 layers with mixed anatase and rutile structure are more favorable for apatite formation than an amorphous structure, when used as bio-implant [12].

3.3. Roughness measurement and wettability studies of nanotubes 2D and 3D noncontact surface profile micrographs are presented in Fig. 5a–b which clearly represents the peaks and valleys of the tubular features. As-formed anodic oxide nanotubes had a surface roughness Ra in the range of 0.5 μm–1.5 μm. After annealing, we observed that the values of Ra were slightly lower than the as-formed structure but significant difference was not observed (p N 0.05). Annealing alters amorphous oxide into crystalline (anatase and rutile) structure resulting in a decrease of porosity levels; hence the Ra value slightly decreases. From a bioimplant perspective, it has been reported that the implant material with surface roughness of 10 nm to 10 μm can significantly improve the cell–material interactions [29].

18.74 ± 2.87 24.31 ± 1.97 28.92 ± 2.76 29.45 ± 3.24

Annealed As-formed

23.58 ± 3.30 29.44 ± 3.21 31.75 ± 4.21 31.77 ± 1.54 4.71 ± 0.49 4.47 ± 0.54 3.52 ± 0.76 3.79 ± 0.38

Annealed As-formed

5.27 ± 1.45 4.50 ± 1.23 4.05 ± 1.62 3.26 ± 0.92 0.34 ± 0.015 0.35 ± 0.023 0.38 ± 0.01 0.32 ± 0.01 307.8 ± 7.27 462.7 ± 5.81 641.6 ± 7.30 636.8 ± 4.3 301.3 ± 12.83 452.2 ± 5.86 636 ± 6.32 623 ± 7.11

Annealed As-formed Annealed As-formed

29.82 ± 2.32 25.45 ± 3.45 28.46 ± 2.96 26.86 ± 4.63 11.45 ± 1.73 12.34 ± 2.56 11.53 ± 1.24 10.26 ± 1.09

Annealed Annealed

109.89 ± 14.32 119.78 ± 12.34 124.77 ± 18.05 122.37 ± 19.21 104.63 ± 15.34 113.09 ± 20.41 119.38 ± 11.21 133.03 ± 19.53

As-formed

68.94 ± 12.30 76.54 ± 11.53 79.61 ± 15.05 73.91 ± 11.05

Annealed

Fig. 6a–b shows the wettability of the titanium dioxide nanotubes produced by varying the anodization time. Anodized surface morphology for 0.5 h showed a contact angle of 80° and with increasing anodization time the contact angle decreases significantly (p b 0.05). The water contact angle for anodized samples of the nanotube layer showed a hydrophilic behavior that is wetting of water on the entire surface and into the pores. The emergence of nanotubular morphology caused an increase in surface roughness with anodization time, thereby resulting in a greater net free energy decrease to induce rapid wetting. An

85.65 ± 13.17 91.04 ± 18.07 92.80 ± 12.39 105.26 ± 15.51

As-formed

Fig. 3. A) Micro-Raman spectra of as-formed and annealed titania nanotubes recorded at an acquisition time of 5 s. The anodization time varied from 0.5 h to 3 h. All samples were annealed at 600 °C for 3 h. B) Micro-Raman spectrum of annealed titania nanotube recorded at an acquisition time of 30 s.

0.5 h 1h 2h 3h

As-formed

P L (nm) W (nm) Do (nm) Di (nm) Time

Table 1 Morphological parameters for the as-formed and annealed TiO2 nanotubes.

571

0.40 ± 0.03 0.58 ± 0.01 0.54 ± 0.02 0.56 ± 0.02

n 109 (pores/cm2)

As (cm2/per 1 cm2)

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

Fig. 4. XRD patterns of as-formed and crystallized titania nanotubes.

572

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

Fig. 5. 2D and 3D non-contact surface profile micrographs of a) as-formed and b) annealed TiO2 nanotubes for 1 h anodized specimen.

Fig. 6. Water droplet on the as-formed TiO2 nanotube obtained by anodizing for a) 0.5 h, b) 3 h and c), d) subsequently annealed TiO2 nanotubes.

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

increased surface roughness and decreased contact angle are preferable for biocompatibility. Apart from surface roughness, the surface chemistry is also considered to be a significant factor for the contact angle on a biomaterial surface. The observed hydrophilic behavior of TiO2 nanotubes is attributed to the density of hydroxyl groups on the surface and high polarity of O\Ti\O bonds. As-formed nanotubes are partially hydroxylated and show lower contact angles [30]. This contact angle is related to the surface energy, which can be calculated [31], γ sv ‐ γ sl ¼ γ vl cosθ

ð7Þ

Es ¼ Evl cosθ

ð8Þ

573

phase generally possesses more surface hydroxyls in the oxide compared to the rutile phases [33]. More the surface hydroxyl, the better will be the wettability behavior. From Table 1, it can be seen that the effective surface area and porosity values of annealed nanotubes are lower than the as-formed nanotubes. As-formed nanotubes showed lower contact angle due to its higher surface energy values than the annealed nanotubes. Annealing makes the crystallites grow, which causes shrinkage of nanotubes thereby leading to increase in contact angle. Even though porosity and surface energy decrease after annealing, hydrophilicity tends to persist due to the presence of mixed crystalline phases which possess optimum surface hydroxyl. Cells tend to attach better on hydrophilic surfaces [11], hence the cell adhesion would be favored onto the formed nanotubular structure.

where, γsv-γsl = energy, Es, required to form a unit area of solid–liquid interface; Evl is the surface energy between water and air under ambient condition (i.e., 72.8 mJ/m2 at 20 °C); and θ is the static contact angle. The contact angle and the calculated surface energy for as-formed and annealed nanotubes are shown in Fig. 7A–B, which shows a significant difference (p b 0.05). It is also clear from Fig. 7A–B that annealing of TiO2 nanotubes also induce hydrophilicity. Annealing of as-formed nanotubes introduces clean surface, phase transition and change of porosity level [32]. These three aspects alter the surface chemistry of nanotubular structures. It is reported from XPS measurements that the amount of contaminants (F and C) decreases and oxygen content increases after annealing compared to as grown TiO2 nanotubes [32]. The increase of oxygen content after annealing is due to the reduction of oxygen vacancies and conversion of Ti3 + to Ti4 +, which in turn makes the surface with high superficial activity. Similar to as-formed nanotubes, annealed nanotubes also showed a lower contact angle exhibiting hydrophilic nature. It is mainly due to the presence of mixed crystalline phases (anatase and rutile), which significantly affects hydrophilicity. Both phases possess surface hydroxyl but the anatase

Impedance measurements were carried out under open circuit conditions in 0.9% NaCl to characterize the oxide film properties of the as-formed and annealed nanotubes. It can be observed from Fig. 8A–B the Bode plots, reveal the presence of two time constants at lower and intermediate frequency ranges. This can be attributed to the presence of inner barrier and outer porous TiO2 nanotube layer. TiO2 nanotubes showed two time constants at different frequency ranges and this can be due to the change in surface morphology and thickness changes with increasing anodization time. The decrease in phase angles at the high-frequency region may be related to the porous nature of the outer layer. At intermediate frequencies, |Z| vs log f plot (figure not shown) is a straight line with slope varying from − 0.7 to − 0.8 exhibiting capacitive behavior and at higher frequency plot remains horizontal displaying resistive behavior due to the uncompensated solution resistance. From the Bode impedance plot, higher impedance

Fig. 7. A) Contact angle measurements of as-formed and annealed TiO2 nanotubes. B) Surface energy values of as-formed and annealed TiO2 nanotubes (statistically significant difference was observed between as-formed and annealed samples (p b 0.05)).

Fig. 8. Bode phase representations of EIS spectra obtained for TiO2 nanotubes: A) as-formed and B) annealed.

3.4. Electrochemical impedance spectroscopy measurements

574

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

data using the equivalent circuit, a chi-quadrate (χ2) value of approximately 10−4–10−3 was observed, which suggests reasonably good fit. In this model, Ru is the uncompensated electrolytic resistance, Rb and Ro are the resistance of the inner barrier and outer porous layer; W is the Warburg impedance. The capacitance of the inner and outer layer is related to constant phase elements Qb and Qo with coefficients nb and no. The impedance of CPE is defined as,  n −1 Z CPE ¼ Q o ð jwÞ

ð9Þ

where, Z is the impedance of CPE; j is the imaginary number (j2 = -1); w is the angular frequency (rad s−1); and Q and n are the CPE parameters. The resistance values of the as-formed nano-tubular oxide differ from the annealed nanotubes. Interestingly, Rb shows higher values compared to Ro in as-formed nanotubes whereas porous layer exhibits higher resistance in annealed nanotubes. Since annealing promotes shrinkage of nanotubes, tube wall thickness tends to increase with reduction in surface area of the nanotubes and hence, resistance increases. The nb values of barrier layers for all the electrodes are in the range of 0.80–0.99, which indicates that the barrier layers of all the samples show near capacitive behavior regardless of the heat treatment, while no value of tubular layers are in the range of 0.60–0.97 which can be associated with a distribution of the relaxation times as a result of heterogeneities present. Values of n around 0.30–0.50 were reported by many authors in the case of porous electrodes [35]. Diffusion impedance can also contribute to the lower values of no. Annealed nanotubes comprising mixed anatase and rutile phases showed more stable behavior than the as-formed nanotubes. 3.5. Mott Schottky analysis

Fig. 9. Nyquist representations of EIS spectra obtained for TiO2 nanotubes: A) as-formed and B) annealed. Inset in Fig. 9A: Equivalent circuit employed to fit the experimental curves.

is observed at low frequency for the thermally treated nanotubes than the as-anodized electrodes. Nyquist plots shown in Fig. 9A–B for as-formed and annealed TiO2 nanotubes display a semicircle, which indicates near capacitive response and the presence of two time constants, can also be identified. Surface anodized for longer duration showed a larger capactive loop which can be related to increase in thickness of the oxide layer. At high frequency, semicircles show that the transfer resistance dominates the impedance and at lower frequencies, Warburg impedance is observed which is due to diffusion of titanium ions to the porous layer. It has been reported that upon immersion in bio-fluids, Warburg impedance tends to decrease at earlier stages and finally it reaches a constant value [34]. This behavior is due to the increase in the buildup of titanium ions in the electrolyte such that the diffusion of titanium ions back into the porous layer takes place. Equivalent circuit, schematized in Fig. 9A as inset was used to fit the experimental data and the deduced values for all the samples are shown in Table 2. Upon fitting the experimental

Mott Schottky analysis for as-formed and annealed nanotubes shown in Fig. 10A–B, were performed to identify the semiconducting properties of the oxide film, such as the electronic type, the effective donor density and the flat band potential. Since the dissolution and the diffusion rate of the passive film depends on the semiconducting properties of the grown oxide, it is important to analyze the oxide nature for clear understanding of their corrosion properties. Fig. 10a–b displays positive slopes, indicating the n-type semiconductor behavior for all anodized specimens. The changes in C−2 with the applied potentials indicate the electronic properties of the passive film close to the electrolyte/oxide interface. This is because when the applied potential is first increased, the region close to the electrolyte is depleted of electrons and thus a region of uniform donor density is located close to the metal/film interface [36]. The capacitance value decreases with voltage implying that the contribution of capacity due to space charge is dominant and hence Mott Schottky analysis is valid. The Mott Schottky can be expressed using the well known equation [17,18,36], 1 2 ¼ C 2sc ε0 εr eN D A2

  kT V−V fb − e

ð10Þ

where ε0 is the vacuum permittivity (8.85 × 10−14 F/cm), εr is the dielectric constant (εr = 50–80), A is the surface area of the electrode

Table 2 Impedance parameters obtained for the nanotubular layers grown at different durations.

As-formed

Annealed

Time

Ru (Ω cm2)

Ro (Ω cm2)

Qo (Ω−1 sn cm2)

no

Rb (Ω cm2)

Qb (Ω−1 sn cm2)

nb

W (Ω−1 s0.5 cm2)

GOF

0.5h 1h 2h 3h 0.5 h 1h 2h 3h

18.29 5.72 16.39 9.87 26.82 15.97 7.11 26.12

231.12 43.7 19.04 149.3 1.78E+05 7.39E+04 5.30E+05 1.98E+06

2.12E−04 5.91E−05 9.62E−06 3.89E−04 6.87E−05 4.49E−05 3.94E−05 3.51E−05

0.66 0.74 0.89 0.99 0.75 0.88 0.86 0.73

1.06E+ 04 2.16E+ 04 1.54E+ 05 1.40E+ 05 375.5 270.7 205 409

1.46E−04 2.51E−04 9.62E−06 2.10E−05 1.23E−05 4.47E−06 6.30E−06 3.75E−06

0.94 0.90 0.89 0.98 0.8 0.88 0.99 0.87

4.00E−05 2.28E−05 1.70E−05 8.77E−05 4.52E−05 6.47E−05 1.55E−05 2.97E−05

1.69E−03 2.95E−03 8.70E−04 1.82E−03 7.49E−04 2.50E−02 2.45E−02 1.19E−03

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

575

Fig. 10. Mott Schottky plots of A) as-formed and B) annealed samples.

−19

2

Fig. 11. Potentiodynamic polarization plots of TiO2 nanotubes: A) as-formed and B) annealed. Inset: SEM micrographs after polarization in saline medium.

interface (cm ), e is the charge of the electron (1.6 × 10 C), ND is the donor concentration over a volume (cm−3), kT/e is about 25.0 mV at ambient temperature and Vfb is the flatband potential (VSCE). The Mott Schottky plot may show a wide variation in data with frequency for porous and defective oxide structures and makes it difficult to choose a frequency value for analysis. It has been recommended by authors [36,37] that, a frequency greater than 1 kHz with relatively higher scan rate can be used to perform Mott Schottky analysis, where the ionic conductivity is negligible and the response is only due to electronic conductivity. The above equation shows a linear dependence between C−2 and V. However, non-linearity was observed experimentally, which according to Tomkiewicz [38] is due to oxidation of the electrode itself by holes in the depletion region, non-homogeneous doping and deep doping levels. Despite the existence of non-linear behavior, it has been shown that ND and Vfb values can be determined from the slope and the intercept of the linear portion. From Table 3, it is clear that ND values decrease with increasing anodization time and the values obtained are in the order of 1018–1019 cm−3. The obtained ND values are consistent with other reports of TiO2 nanotube arrays [17,19,20]. In general, non-ideal (anodically grown) oxide films are always non-stoichiometric with an excess of metal cations or a deficiency of oxygen anions. According to the point defect model (PDM), n-type semiconducting behavior indicates that the defects in all the samples are the oxygen vacancies and/or few titanium ion interstitials and the evidence for this defective nature

has been reported earlier [37]. Mixed phases of anatase and rutile showed less ND values compared to the as-formed TiO2 nanotubes. After thermal treatment, oxygen to titanium ratio increases resulting in the reduction of oxygen vacancies, thereby the ND values decrease and hence the rate of dissolution decreases. According to AcevedoPena [17], the values of donor concentrations tend to decrease with the rutile formation in the oxide film and a similar trend was observed in our work, which can be confirmed with amount of rutile fraction calculated from the XRD analysis. Anions (Cl−) present in the electrolyte tend to interact with electrode when the presence of oxygen vacancies in the oxide film is high and may destruct the oxide film. Calculated Vfb showed more negative values due to the presence of a surface state capacitance shift the intercept on the vertical axis and on the horizontal axis and yielding too negative Vfb values [36]. From the table, it is clear that annealed nanotubes showed less negative Vfb values compared to as-formed nanotubes, which is due to the elimination of F− ions or organic species after thermal oxidation. In general, Vfb represents the position of the Fermi level with respect to the potential of the reference electrode. So, the decrease of Vfb can be related to the increase of donor concentrations which affect the band position [19]. It is evident that a more negative Vfb and lower ND make the electrode surface less reactive to the anions in the electrolyte [18].

Table 3 Semiconducting properties obtained from Mott Schottky measurements.

Table 4 Electrochemical parameters measured from Tafel plots.

Time

0.5 h 1h 2h 3h

As-formed

Annealed

Efb (VSCE)

ND (×10

−0.68 −0.77 −0.57 −0.75

1.80 2.0 0.60 1.77

19

cm

−3

)

Time

Efb (VSCE)

ND (×10

−0.47 −0.35 −0.375 −0.15

1.55 0.45 0.39 0.29

18

cm

−3

) 0.5 h 1h 2h 3h

Ecorr (mVSCE)

ipass (A cm−2)

Epass (mVSCE)

As-formed

Annealed

As-formed

Annealed

As-formed

Annealed

−391.3 −376.5 −314.8 −293.2

−165.9 −276.6 −311.1 −267.8

−165.9 −209.4 −187.7 −169.4

−35.11 −117.4 −146.7 −141.6

5.42E−7 3.87E−7 3.63E−7 3.78E−7

4.70E−7 2.65E−7 0.79E−7 0.64E−7

576

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

3.6. Potentiodynamic polarization studies Fig. 11A–B displays potentiodynamic polarization plots recorded for the nanotubes in 0.9% NaCl solution at a scan rate of 1 mVs−1. Before

polarizing the samples, the system was allowed to stabilize for 3 h in order to reach the steady state open circuit potential. The plots recorded show typical behavior as often seen for a highly passive system. The passive current density, ipass, and corrosion potential, Ecorr were deduced

Fig. 12. A) FE-SEM and EDS analyses of a) as-formed and b) annealed samples after immersion in SBF at 37 °C for 5 days. B) EDS elemental mapping of Ca/P-annealed TiO2 nanotubes: a) SEM image, b) overall mapping, c) Ca mapping, d) P mapping, e) O mapping, and f) Ti mapping.

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

from the polarization curves and is tabulated in Table 4. Anodization time and annealing has significant influence on passive current density, ipass. All the samples are characterized by the negligible extent of the active region and wide passive region. The complete passivation occurs at a potential, Epass and the estimated values are close to −0.5 VSCE which is consistent with other reports [4,20]. No breakdown potential is observed, which signifies that the nanotubes are protective. Nanotubular structures may act as more effective channels for the electrolyte to reach the interface and their porous nature acts as perfectly passive pits possibly due to the higher tube and the barrier oxide thickness. The increased effective surface area of the oxide layer enables more surface reaction to take place and hence it favors more charge transfer reactions. Annealed TiO2 nanotubes containing less surface area (Table 1) than as-formed nanotubes showed better corrosion resistance exhibiting lower ipass. This can be confirmed by surface area measurements since extended surface area will tend to shift the passive current density to a higher value [20]. Also Epass showed more negative values compared to the values obtained for as-formed nanotubes. The stable nature of the oxide layer was confirmed by SEM images taken after polarization indicating no significant damage of the nanotubular oxide layer formed yet the surface is covered with corrosion products. It can be found that ipass values decreased with increase in anodization time. The passive current density observed was in the order of 0.5– 5 × 10−7 A cm−2; this low passive current density was observed even at higher anodic over potentials. 3.7. Bioactivity tests Fig. 12A shows the SEM/EDS results of as-formed and heat treated samples on soaking in SBF for 5 days. The plate-like particles were formed on as-formed samples whereas on annealed samples rod-like morphology was observed. The EDS spectra reveal the presence of Ca, P, O and Ti and the elemental mapping on annealed nanotubes (Fig. 12B), demonstrates a relatively homogeneous distribution of elements throughout the surface. Hence the deposit on the surface should correspond to calcium phosphate. The good coverage of Ca/P on annealed surface is due to the presence of crystalline phases which favors the formation of apatite crystals. The anatase phase promotes nucleation and growth of apatite than the rutile phase due to its better lattice fit with hydroxyapatite crystals [39]. On the other hand, the presence of rutile phase improves the dissolution resistance because of its close packed structure, which hinders ionic diffusion compared to anatase [40]. From this study it is evident that, the corrosion properties of TiO2 nanotubes tend to improve significantly due to annealing. Studies to find out the effect of annealing time on the passive film and semiconducting properties of TiO2 nanotubes would be the scope of future work, as thermal treatment for different time periods would tend to change the morphology and the fraction of rutile/anatase phases formed. We believe that corrosion resistant TiO2 nanotubes can be further employed for improving nanomechanical properties, tribocorrosion properties and cell culture studies. Nevertheless, further work is essential to throw light on this subject. 4. Conclusions This investigation outlines the synthesis of TiO2 nanotubes in citric acid containing solution. It compares the oxide properties of the asformed and annealed titanium oxide nanotubes. Surface morphologies tend to vary with anodization time and with thermal annealing. The following conclusions can be drawn from this work. 1. XRD studies on annealed oxides revealed variation in the fraction of rutile and anatase phase percentage related to the variation with anodization time. Raman bands become sharper and more intense when the anodization time increases. The correlation between the

2.

3.

4.

5.

6.

577

specific geometry of tubular structures and their surface properties was established. Both as-formed nanotubes and nanotubes with mixed anatase and rutile phases showed hydrophilic behavior with contact angles in the range of 45°–75°. TiO2 nanostructures had a surface roughness, Ra, lower than 1.5 μm and porosity in the range of 40–50%. The properties of the barrier oxide layer and outer porous layer were evaluated using EIS technique. Annealed TiO2 nanotubes had a higher impedance magnitude compared to as-formed nanotubes. The results of the capacitance response indicate that the TiO2 nanotube layer behaves as n-type semiconductor. Annealed nanotubes exhibited lower donor density (1018 cm−3) as compared to the asformed nanotubes (1019 cm−3). Polarization measurement confirmed the wide passive behavior of all TiO2 surfaces with lower ipass values in the range of 0.5– 5 × 10−7 A cm−2. Annealed nanotubes showed enhanced precipitation of Ca/P during bioactivity tests. This study of surface properties of amorphous and crystalline nanotubes will open up the new possibilities to tune the oxide properties for biomedical applications.

Acknowledgments The authors thank Prof. T.S. Sampath Kumar for kindly permitting us to utilize the wettability testing facility. We thank Dr. Ratna Sunil and Mr. G. Logesh for their help during contact angle measurements and SEM analysis. Useful discussions with Miss. P. Haveela and Miss. R. Lavanya are greatly acknowledged. We thank Dr. J. Magesh from the Department of Physics, IIT Madras for their assistance in carrying out micro-Raman spectroscopy. References [1] L. Reclaru, R. Lerf, P.Y. Eschler, A. Blatter, J.M. Meyer, Biomaterials 24 (2003) 3027–3038. [2] F. Schmidt-Stein, S. Thiemann, S. Berger, R. Hahn, P. Schmuki, Acta Mater. 58 (2010) 6317–6323. [3] R. Beranek, H. Hildebrand, P. Schmuki, Electrochem. Solid State Lett. 6 (2003) B12–B14. [4] M. Balakrishnan, R. Narayanan, Thin Solid Films 540 (2013) 23–30. [5] R. Narayanan, T.Y. Kwon, K.H. Kim, Mater. Chem. Phys. 117 (2009) 460–464. [6] W. Yang, X. Xi, Q. Ran, P. Liu, Y. Hu, K. Cai, Mater. Sci. Eng. C 34 (2014) 410–416. [7] R. Hahn, J.M. Macak, P. Schmuki, Electrochem. Commun. 9 (2007) 947–952. [8] Z.B. Xie, D.J. Blackwood, Electrochim. Acta 56 (2010) 905–912. [9] S. Oh, S. Jin, Mater. Sci. Eng. C 26 (2006) 1301–1306. [10] S. Oh, K.S. Brammer, K.S. Moon, J.M. Bae, S. Jin, Mater. Sci. Eng. C 31 (2011) 873–879. [11] I. Roman, R.D. Trusca, M.L. Soare, C. Fratila, E. Krasicka-Cydzik, M.S. Stan, A. Dinischiotu, Mater. Sci. Eng. C 37 (2014) 374–382. [12] H. Tsuchiya, J.M. Macak, L. Muller, J. Kunze, F. Muller, P. Greil, S. Virtanen, P. Schmuki, J. Biomed. Mater. Res. Part A 77 (2006) 534–541. [13] N. Caliskan, C. Bayram, E. Erdal, Z. Karahaliloglu, E.B. Denkbas, Mater. Sci. Eng. C 35 (2014) 100–105. [14] P. Marcus, Electrochim. Acta 43 (1998) 109–118. [15] I.U. Petersson, J.E. Loberg, A.S. Fredriksson, E.K. Ahlberg, Biomaterials 30 (2009) 4471–4479. [16] K. Venkateswarlu, N. Rameshbabu, D. Sreekanth, M. Sandhyarani, A.C. Bose, V. Muthupandi, S. Subramania, Electrochim. Acta 105 (2013) 468–480. [17] P. Acevedo-Peña, J.E. Carrera-Crespo, F. González, I. González, Electrochim. Acta (2014). http://dx.doi.org/10.1016/j.electacta.2014.06.056. [18] T. Kokubo, H. Kushitani, S. Sakka, T. Kitsugi, T. Yamamuro, J. Biomed. Mater. Res. 24 (1990) 721–734. [19] L. Aïnouche, L. Hamadou, A. Kadri, N. Benbrahim, D. Bradai, Electrochim. Acta 133 (2014) 597–609. [20] S.A. Ali Yahia, L. Hamadou, A. Kadri, N. Benbrahim, E.M.M. Sutter, J. Electrochem. Soc. 159 (2012) K83–K92. [21] G.D. Sulka, J. Kapusta-Kołodziej, A. Brzózka, M. Jaskuła, Electrochim. Acta 55 (2010) 4359–4367. [22] B. Yang, C.K. Ng, M.K. Fung, C.C. Ling, A.B. Djurisic, S. Fung, Mater. Chem. Phys. 130 (2011) 1227–1231. [23] X.Y. Zhang, L.D. Zhang, W. Chen, G.W. Meng, M.J. Zheng, L.X. Zhao, Chem. Mater. 13 (2001) 2511–2515. [24] D. Regonini, A. Jaroenworaluck, R. Stevensa, C.R. Bowen, Surf. Interface Anal. 42 (2010) 139–144. [25] H. Cheng, J. Ma, Z. Zhao, L. Qi, Chem. Mater. 7 (1995) 663–671. [26] R.A. Spurr, H. Myers, Quant. Anal. Chem. 29 (5) (1957) 760–762.

578

B. Munirathinam, L. Neelakantan / Materials Science and Engineering C 49 (2015) 567–578

[27] D. Regonini, Anodised TiO2 Nanotubes: Synthesis, Growth Mechanism and Thermal Stability(PhD Thesis) University of Bath, Materials Research Centre, UK, 2008. [28] J. Xiong, Y. Wang, Y. Li, P.D. Hodgson, Thin Solid Films 526 (2012) 116–119. [29] C. Vasilescu, P. Drob, E. Vasilescu, I. Demetrescu, D. Ionita, M. Prodana, S.I. Drob, Corros. Sci. 53 (2011) 992–999. [30] E. Balaur, J.M. Macak, H. Tsuchiya, P. Schmuki, J. Mater. Chem. 15 (2005) 4488–4491. [31] B.R. Sunil, A.A. Kumar, T.S.S. Kumar, U. Chakkingal, Mater. Sci. Eng. C 33 (2013) 1607–1615. [32] Q. Ye, P.Y. Liu, Z.F. Tang, L. Zhai, Vacuum 81 (2007) 627–631. [33] Q.J. Liu, X.H. Wu, B.L. Wang, Q. Liu, Mater. Res. Bull. 37 (2002) 2255–2262.

[34] R. Narayanan, Coating of Ti–6Al–4V for Orthopaedic Applications by Anodic Oxidation and Cathodic Deposition(PhD Thesis) IIT Madras, India, 2005. [35] D. Qu, Electrochim. Acta 48 (2003) 1675–1684. [36] E. Sikora, J. Sikora, D.D. Macdonald, Electrochim. Acta 41 (1996) 783–789. [37] Z. Jiang, X. Daia, H. Middleton, Mater. Chem. Phys. 126 (2011) 859–865. [38] M. Tomkiewicz, Electrochim. Acta 35 (1990) 1631–1635. [39] J.S.L. Leach, B.R. Pearson, Corros. Sci. 28 (1988) 43–56. [40] X. Zhu, K.H. Kim, Y. Jeong, Biomaterials 22 (2001) 2199–2206.