Controlling morphological parameters of a nanotubular TiO2 coating layer prepared by anodic oxidation

A promising modification route to improve osseointegration of dental and medical titanium devices is a nanostructured titanium oxide coating layer in the form of self-ordered vertically aligned nanotubes (or nanotubular TiO2). In this work, we report a detailed investigation of nanotubular TiO2 coating layer on metallic Ti substrate prepared by anodic oxidation. The main goal was to determine an optimized and reproducible route to produce a nanotubular TiO2 layer with homogenous morphology, narrow distribution and accurate control of the nanotube diameter. The influence of electrolyte temperature, anodizing time and applied voltage were studied, comparing three different electrolytes: 1.5 wt% HF, 0.5 wt% HF, and 0.5 wt% HF + 1 mol l−1 H3PO4. Samples were analyzed by SEM, EDS, FIB, and XPS techniques. The most favorable result was achieved by using 0.5 wt% HF + 1 mol l−1 H3PO4 electrolyte, for anodizing time of about 90 min, temperature of 20 °C, and anodizing potential from 1 to 25 V. Using these parameters, a uniform self-organized nanotubular TiO2 layer was prepared with a fine control of the nanotube diameter value over a wide range (10 to 100 nm).


Introduction
Titanium and its biomedical alloys have been extensively used in implantology due to their favorable characteristics such as fatigue and corrosion resistance, biocompatibility, low density and relatively low modulus of elasticity. The reported long-term clinical success for titanium and its alloys has made titanium the standard material for the fabrication of dental implants [1,2]. Metallic titanium and its alloys oxidize immediately upon exposure to air. However, at ambient temperatures, the formation of a passive oxide coating protects the bulk metal from further oxidation. This layer is biologically inert, which impairs osseointegration because it induces the formation of a fibrous tissue between bone and implant, which may cause implant loss [3,4]. To avoid implant encapsulation and increase the effectiveness of the osseointegration process, modifications are usually introduced on the implant surface. Common surface modifications can be made through coatings with materials with osteoinductive characteristics such as hydroxyapatite or other calcium phosphate compounds, or with morphological modifications such as blasting (which will produce a micrometer roughness) or anodic oxidation [1,5]. Anodic oxidation or anodizing is a promising modification used to produce a nanostructured titanium oxide (TiO 2 ) coating layer in the form of self-ordered vertically aligned nanotubes (or nanotubular TiO 2 ) on titanium implant devices [6].
Zwilling et al, in a pioneering study, has produced nanotubular TiO 2 arrays on metallic titanium via anodic oxidation in a chromic acid electrolyte containing hydrofluoric acid [7]. Thereafter, anodizing has become a well-known method for the production of nanotubular TiO 2 [6,8]. This is an attractive procedure due to its ease of implementation and low cost. Besides that, it enables to control the main properties of nanostructured TiO 2 overlayer (i.e. morphology, thickness and internal pore diameter) by tuning the anodizing variables such as applied potential and electrolyte composition.
The mechanism of nanotubular titanium oxide formation by the anodization of Ti metal has previously been detailed [9,10] and it will be briefly discussed here. The anodizing procedure is typically carried out in a two-electrode configuration with clean metallic titanium at the anode. Once a sufficient large potential is applied (about 1 V), two processes start to occur at the anode: (i) the oxidation of the metallic Ti to Ti 4+ and (ii) the diffusion of O 2− , formed by deprotonation of water present in the electrolyte, toward the surface of the metal, as describe by equation (1).
The result is the formation of a thin layer of titanium oxide on the exposed metal surface. Oxide dissolution occurs at the oxide/electrolyte interface because Ti-O bonds are weakened due to polarization. In the presence of fluoride ions, localized dissolution of titanium oxide takes place, forming water-soluble [TiF 6 ] 2− complexes that initiates the nanotube formation, as describe by equation (2).
The O 2− ions remain diffusing through the TiO 2 layer and reacts with the metal to produce fresh oxide at the bottom of the nanotubes. The rate of oxide growth at the metal-oxide interface and the rate of oxide dissolution at the oxide-electrolyte interface equalizes, thus the thickness of the oxide layer at the base of the nanotube essentially remains unchanged as it moves further into the metal making the pore deeper [11,12].
It is well acknowledged that titanium covered with a nanotubular TiO 2 overlayer presents a large potential for clinical applications mainly due to its exceptional biocompatibility. There are many works reporting an increase in cell adhesion and proliferation in the presence of nanostructured surface compared to smooth titanium surface. However, there is no consensus about which nanotube diameter is more effective for cell response [6,[13][14][15][16].
Since Bauer et al [17] showed the dependence of tube diameter with applied voltage during anodizing process, authors have been studying the in vitro effects of nanotubes of different diameters and the results are controversial. One group has related the best biological performance with nanotubular titania layer with an internal nanotube diameter smaller than 30 nm [18][19][20][21]. Conversely, other authors have reported an increasing on the biological performance for nanotubular titania surfaces with nanotube diameter around 100 nm [22,23].
The cellular behavior on a nanotubular TiO 2 surface is a matter of great interest and have been intensively investigated, especially in comparison with a smooth or a blasted surface, aiming to use in implantology [21,[23][24][25]. The vast majority of the reports had shown favorable results for a nanotubular TiO 2 surface. However, there is still no consensus on what would be the most appropriate TiO 2 nanopore diameter to use in order to improve the osseointegration character of Ti implant devices, requiring further studies on this subject. Regardless of the nanotube diameter best suited for the cell response, the production of a nanotubular titania coating on titanium with a narrow distribution of nanotube diameter and well controlled thickness is a key issue for a detailed study of its biological performance.
Tailoring and controlling the anodic oxidation preparation route for large-scale production is not an easy task due to the large number of thermodynamic parameters associated to the process. While the process duration, potential, temperature and fluoride concentration are the main parameters that control the nanostructured titania layer thickness, diameter, and growth rate, different electrolyte features including pH, viscosity, conductivity, and organic additives can affect the process, leading to the formation of a nanotubular coating layer with undesired morphology or even to the non-formation of a nanostructured oxide layer [5,8,12]. In general, the most commonly used electrolytes for titanium anodization can be divided into two groups: organic, such as glycerol [26] and ethylene glycol [24]; and inorganic, such as HF [27], HF+H 3 PO 4 [17], NH 4 F+(NH 4 ) 2 SO 4 [28], H 2 SO 4 +HF [29].
Here we present a detailed investigation of the preparation of nanotubular TiO 2 surface coating obtained by anodic oxidation of metallic Ti substrates. The influence of electrolyte temperature, anodizing time and applied voltage were studied, comparing three different electrolytes: 1.5 wt% HF, 0.5 wt% HF, and 0.5 wt% HF+1 mol l −1 H 3 PO 4 . These electrolytes were chosen due to previous results that have demonstrated reproducibility [17,18,27,30]. In addition, we also evaluated the effect of the presence of H 3 PO 4 on morphology of TiO 2 nanotubes as compared to pure HF electrolytes. The aim of this work was to establish the most favorable condition for obtaining a uniform layer of titanium oxide nanotube on metallic Ti substrate by anodic oxidation, with homogenous morphology, narrow distribution and accurate control of nanotube diameter.

Sample preparation
Ti substrates were obtained by machine cutting of a circular rod made of commercially pure titanium grade IV (ASTM F67) in form of discs with 12.7 mm of diameter and 2.0 mm thick. After machine cutting, titanium substrates were cleaned as described here. Firstly, Ti discs were subjected to a degreasing bath, washed using warm water (≈80°C) and biodegradable detergent in ultrasonic bath cleaner. Then, discs were chemical cleaned using sodium hydroxide (NaOH), rinsed with deionized water (dH 2 O), immersed in aqueous solution of hydrofluoric and nitric acids for 5 min and rinsed again with dH 2 O in ultrasonic bath cleaner. Finally, discs were dried in an oven at 70°C by 20 min Immediately before anodizing process, titanium substrates were cleaned using acetone (CH 3 COCH 3 −99.5%, Anidrol) and deionized water, and, with the purpose of removing a natural layer of TiO 2 , discs were immersed in a solution containing 2 v% HF+3 v% HNO 3 65%, Vetec, for 5 min and rinsed with dH 2 O. Millipore Milli-Q system.
The anodizing setup consisted of a homemade electrochemical cell, composed of two electrodes, an adjustable DC power supply (Agilent 6634B) and a monitoring system developed on LabVIEW platform that allows storing the main electrochemical parameters (applied potential, current, and electrolyte temperature) and setting the anodizing duration.
The titanium sample is set as the anode, with 0.994 cm 2 of area exposed to anodizing electrolytes and the stainless-steel mesh set as the cathode. The distance between anode and cathode is about 2.5 cm. All the anodizing procedures were performed in the potentiostatic mode.
A summary of sample preparation parameters to evaluate the effect of anodizing time and temperature on the nanotubular TiO 2 coating layer obtained by anodic oxidation is presented in table 1. For HF 1.5 wt% electrolyte, a set of samples were prepared at 18°C and anodizing time of 8 min in order to compare with the results presented by Gong et al [27].
The influence of anodizing voltage on the nanotube diameter were explored by fixing time and temperature (60 min and 12°C for HF 0.5% electrolyte and 90 min and 20°C for HF+H 3 PO 4 electrolyte, respectively) and varying anodizing voltage in the range from 1 to 25 V.
After anodizing, each sample was washed in dH 2 O, dried rapidly with a heat blower and then placed in an oven at 100°C for 30 min in order to achieve complete drying. Once ready, the samples were stored in desiccator. SEM/EDS analyzes were performed in a SEM-FEG microscope (SIGMA-VP, brand Carl Zeiss, CDTN). From the SEM micrographs, the average diameter of the TiO 2 nanotubes was determined using the Quantikov [31] software. About 10 samples of each anodizing voltage were used to estimate the average nanotube diameters.

Sample characterization
A FIB/SEM microscope (Quanta FEG 3D FEI, CM/UFMG) was used to obtain cross sections of titanium samples after nanotubular TiO 2 surface preparation. A beam of Ga + ions, with energy of 30 keV and current of 0.5 nA, swept a delimited region in sample in order to excavate a wedge-shaped surface pit (nominal dimensions of 7×5×2 μm).
In order to evaluate the surface chemical composition after nanotubular TiO 2 coating layer preparation, samples were investigated by XPS (Phoibos 150, Specs). XPS is a surface sensitive technique that provides information about chemical state of constituting species. Here, the differences in the chemical state of O, F and Ti species were evaluated for the three different preparation routes. The results were analyzed in detail and correlated with the observed surface morphology of the TiO 2 nanotubular coatings. Titanium samples were pretreated by Ar + sputtering for 10 min (0.8 keV ion energy), in order to remove surface contamination, especially carbon. The experimental XPS spectra were analyzed using CasaXPS software. High-resolution spectra were deconvoluted with Gaussian-Lorentzian mixed function, with peak intensity ratio of 30/70.

Results and discussion
3.1. 1.5 wt% HF electrolyte Gong et al [27] have reported the formation of a nanotubular TiO 2 layer by anodic oxidation of Ti substrates using HF 1.5 wt%, with a potential of 20 V and 18°C. However, in our experiments, after anodization under similar conditions (HF concentration, electric potential, duration time and temperature), a nanostructured oxide layer was not obtained (figure 1(a)), while anodizing performed at 12°C, for 10 to 60 min produced a nanotubular layer with low degree of organization ( figure 1(b)). In contrast, regardless of anodization time (8 to 60 min) and temperature (18 and 24°C), the anodic oxidation of Ti using 1.5 wt% HF and 20 V has resulted in a surface covered with a compact oxide film, with uneven height and no indication of nanotubular oxide layer.
A reason for no formation of a nanotubular TiO 2 surface layer by anodizing in 1.5 wt% HF electrolyte, 20 V, at 18 and 24°C, may be related to the concentration of fluoride ions in the electrolyte. A high concentration of F − (greater than 1 wt%) retard the titanium oxide formation, as the available Ti 4+ immediately reacts with the fluoride ions, forming [TiF 6 ] 2− . The final result would be similar to an electrolytic polishing process [12,32]. 3.2. 0.5 wt% HF electrolyte SEM images of Ti samples anodized by 30 min at different temperatures show different results: a compact oxide layer was produced at 24°C, while a self-ordered nanotubular layer was produced at lower temperatures (18°C and 12°C, as shown on figures 2(a) and (b), respectively). No significant morphological differences were observed between these temperatures. This result is not in agreement with authors that bring the information that an ideal temperature range to TiO 2 nanotube formation is around room temperature (20°C-24°C) [10,12]. According to Ocampo and Echeverría [12], higher temperatures (∼60°C) reduce electrolyte viscosity promoting disorderly oxide dissolution; conversely, very low solution temperature (∼5°C) increases viscosity harming oxide dissolution, resulting in a TiO 2 layer with low degree of organization.
By comparing anodized samples, prepared at the same temperature (12°C), it was observed that longer anodizing time improves the uniformity of nanotubes. For anodizing time of 10 min, a nanotubular layer with low uniformity was formed. The organization of the nanotubular layer turns even better when anodizing time increases to 30 min and, then, to 60 min (figures 2(b) and (c), respectively). After 60 min, the organization degree of the nanotubular coating layer was already quite satisfactory. Following the formation of TiO 2 nanotubes, the surface morphology increases with anodizing time, which reaches a stabilization after a time limit. Beyond this limit, in which self-organization degree of nanotubes is already high, the anodizing time no longer influences [10].
Regarding the nanotubular titanium oxide thickness, no significant changes were observed for samples anodized for 30 min and 60 min The average thickness of the nanotubular TiO 2 layer was determined as 262±30 nm and 275±10 nm, for anodizing time of 30 and 60 min, respectively (figure 3). This observation is consistent with previous results that, for a determined anodizing condition, the growth of the nanotubular titanium oxide layer reaches a saturation regime and, therefore, the layer thickness does not dependent significantly on anodizing time, but self-ordering can be enhanced, until a certain limit, obtaining a nanotubular layer more homogeneous [10,33].
EDS results indicate the presence of titanium, oxygen and fluorine in both samples, what suggests the fixation of elements that constitute the anodizing solution in the nanotubular layer formed on the titanium surface.
For these samples, prepared with same anodizing time and temperature, but at different anodizing potential, SEM analyzes revealed that the morphology of the nanotubular TiO 2 layer was not uniform for the whole range of values (figure 4). For anodizing conducted at 10 V, a compact TiO 2 layer was observed ( figure 4(a)), for 15 V and 20 V (figures 4(b) and (c), respectively), a nanotubular layer of TiO 2 was formed, but only for 20 V, a uniform layer with a high degree of self-organization was observed. For 25 V ( figure 4(d)), a layer of disorganized oxide was formed. This behavior was not expected since some authors have reported to obtain TiO 2 nanotubes in this same voltage range using fluorine-containing electrolytes [17,27,34]. Figure 5(a) shows XPS survey spectrum of a sample anodized in 0.5 wt% HF electrolyte, for 60 min, at 12°C and anodizing voltage of 20 V. Despite the sputtering process, XPS spectrum ( figure 5(a)) presents a C 1 s peak, although in much smaller intensity, compared to the spectrum of as-prepared sample. After sputtering procedure, the observed ratio of O 1 s and Ti 2p peak areas, corrected by their relative sensitivity factors [35], is about 2.32. Ar and F peaks were also observed in the XPS survey spectrum. The observed Ar is probably due to ion implantation on the sample's surface during the sputtering process, while F species come from anodizing process with a fluorine-based electrolyte. Figures 5(b)-(d) present high-resolution XPS spectra of nanotubular TiO 2 surface. The O 1 s peak, presented in figure 5(b), was fitted with three components, centered at the energies of 531.0 eV, 532.1 eV and 533.2 eV. The peak component at 531.0 eV is assigned to oxygen species bonded to The average thickness of the obtained nanotubular oxide layer was in the range of 500 to 580 nm (figures 7(a) and (b), respectively), a higher value if compared to the nanotubular layer obtained by anodizing in 0.5 wt% HF electrolyte. The reason may be due to the presence of H 3 PO 4 in the electrolyte, which acts as buffering specie [17]. It has been previously established that the final nanotubular layer thickness is essentially the result of an equilibrium between electrochemical formation of TiO 2 , at the pore bottom, and the chemical dissolution of   this TiO 2 in a solution containing Fions [33]. Thus, a higher dissolution rate of oxide will result in a thin oxide layer. According to Macák, Tsuchiya and Patrick [40], the dissolution rate of TiO 2 depends on the pH values of the solution. The more acidic the solution, the greater oxide dissolution. Using a buffer solution, the acidification caused by oxidation of Ti inside the nanotube can be controlled and a thicker layer of TiO 2 can be obtained.
EDS analysis of Ti samples anodized using HF+H 3 PO 4 electrolyte indicates the presence of titanium and oxygen at the samples, even for Ti samples where a compact layer (without nanotube) was noticeable (figure 8). In addition, a small amount of fluorine and phosphor, elemental species from the electrolyte, was observed for the Ti samples that presented a nanotubular coating layer. In the case of phosphorous, a component of the major calcium phosphate compound found in bone and dental tissues, hydroxyapatite [Ca 10 (PO 4 ) 6 (OH) 2 or HA], it is suggested that the presence of this element may help to accelerate the osseointegration process, according to previous literature results [41][42][43]. It is also observed that Ti peak intensity decreases with increasing anodization time. This is mainly due to the increasing of titanium oxide layer with anodizing time, which progressively buries the metallic titanium substrate. Figure 9 pictures SEM micrographs of anodized Ti samples at 25 V for 90 min, with different temperature (10°C to 24°C), confirming the production of a nanotubular titanium oxide layer. However, Ti samples anodized at 20°C and 24°C presented a surface morphology with better organization and homogeneity, with the best result obtained for 20°C. For samples anodized at 10°C and 15°C, the substrate surface was covered with a non-uniform nanotubular TiO 2 layer. SEM micrographs of Ti substrates after anodizing at 20°C for 90 min (figure 10) confirmed the production of a self-ordered nanotubular TiO 2 layer for the entire anodizing potential range and revealed the effect of anodizing potential on the nanotube diameter. The average nanotube diameter and standard deviation values are presented in table 2. Figure 11 shows a plot of the average nanotube diameter as a function of the anodizing potential with a linear fitting curve confirming an approximately linear dependence, in agreement with previous results [17].
The applied voltage directly affects the diameter of TiO 2 nanotube during the anodizing process. A possible explanation to this dependence is, since the formation of nanotubes is caused by the action of Fions in the oxide layer, and the migration of these ions is under the influence of the applied voltage, the higher the applied voltage, the larger the electric field and hence, the greater the amount of ions acting on a same area in the oxide layer. As a consequence, the larger the applied voltage the larger the nanotube diameter [10,27].    shown in figure 12(b). They are related to oxygen bonded to titanium ions (Ti 4+ ) in TiO 2 species, titanium oxide in Magnéli phases and OH groups on the surface, respectively. The energy shift of about 0.4 eV observed in the peak component associated to Magnéli phases, as compared to the spectrum of anodized sample in 0.5 wt% HF, is an indication of the change on the composition ratio of the different oxide species, i.e., TiO, TiO 2 and Ti 2 O 3 . This change is confirmed by the Ti 2p high-resolution spectrum (figure 12(c)), which was deconvoluted into two components: a doublet centered at 458.8, eV and 464.3 eV, assigned to Ti 4+ in TiO 2 species, and a second doublet centered at 457.1 eV and 462.6 eV, related to Ti 3+ in Ti 2 O 3 species. Therefore, unlike the sample anodized with 0.5 wt% HF, TiO was not identified for the sample prepared in 0.5 wt% HF+1 mol l −1 H 3 PO 4 electrolyte. F 1 s high-resolution spectrum, presented in figure 12(d), was fitted by a component centered at 685.4 eV, identified as metal fluorine [39].
Summarizing, our results showed that by using either 0.5 wt% HF or 0.5 wt% HF+1 mol l −1 H 3 PO 4 electrolytes a uniform nanotubular TiO 2 layer can be produced by anodic oxidation of metallic titanium substrates. However, when using 0.5 wt% HF, a uniform self-organized nanotubular layer was achieved after 60 min, conversely, it take longer to get the same result (about 90 min) with the HF+H 3 PO 4 electrolyte. On the other hand, the anodizing process in 0.5 wt% HF electrolyte produced a thinner layer as compared to a sample anodized in HF+H 3 PO 4 , that presented a final thickness 50% higher. It was shown that the anodizing temperature is another key issue to control the production of a nanotubular TiO 2 layer with uniform morphology. Our experimental results revealed that by using the HF+H 3 PO 4 electrolyte, the temperature window to produce a uniform nanotubular TiO 2 layer is from 15°C to 24°C, a broader range as compared to the results obtained using 0.5 wt% HF electrolyte, which requires to control the anodizing temperature in the range of 12°C to 18°C. Finally, an accurate control of the average TiO 2 nanotube diameter was achieved only for the anodic oxidation of titanium by using the HF+H 3 PO 4 electrolyte.

Conclusions
An optimized and reproducible route for obtaining a uniform layer of nanotubular titanium oxide on metallic Ti substrate by anodic oxidation using an aqueous electrolyte was established. It was observed a strong dependence on surface morphology of the preparation conditions applied in the anodizing process: electrolyte composition, anodizing potential and temperature. The results allowed to identify the individual contribution of each parameter on the nanotubular TiO 2 layer morphology. The HF concentration is a key point: high concentration of Fions increases the oxide dissolution rate, producing an excessively thin oxide layer or completely dissolving it, as with 1.5 wt% HF electrolyte. Evaluating the effect of anodization time, it was observed an equilibrium between oxide formation and dissolution is reached, and after this point, it does not significantly influence the morphology of the TiO 2 nanotubular layer. It has also been observed that a different temperature range is  required for each electrolyte solution, in order to form a nanotubular coating, rather than a compact titanium oxide layer. For 0.5 wt% HF electrolyte, it was observed that the ideal temperature for nanotubular layer formation is around 12°C, while for 0.5 wt% HF+1 mol l −1 H 3 PO 4 electrolyte, it forms in the range of 20°C-24°C. As previously observed, the average diameter of TiO 2 nanotubes varies linearly with the applied voltage, which allows a thorough control of the oxide layer morphology. This dependence was verified for 0.5 wt% HF+1 mol l −1 H 3 PO 4 electrolyte but not for 0.5 wt% HF electrolyte. It was still observed that presence of H 3 PO 4 increased the oxide layer thickness, behaving as a buffer in solution, decreasing the dissolution of the titanium oxide.  In conclusion, the best result was obtained for anodic oxidation of Ti substrates using 0.5 wt% HF+1 mol l −1 H 3 PO 4 electrolyte, anodizing time of about 90 min and temperature of 20°C. Using these parameters, a uniform self-organized nanotubular TiO 2 layer was produced with narrow distribution and accurate control of the nanotube diameters over a broad range (10 to 100 nm) and average layer thickness of 550 nm. The elemental analysis performed by EDS confirmed the production of a titanium oxide layer on the Ti substrate. In addition, the presence of phosphorus suggests the fixation of electrolyte species on the nanotubular oxide layer, which could be explored to tailor, in a controlled fashion, the chemical composition of the nanotubular oxide layer on Ti implant devices. XPS analysis showed that nanotubular titanium oxide coating layer is composed mostly by TiO 2 . However, significant amounts of Magnéli phases TiO, and TiO 3 may also be present.