Laser-assisted modification of titanium dioxide nanotubes in a tilted mode as surface modification and patterning strategy

Electrochemical anodization is regarded as a facile and easily scalable fabrication method of titania nanotubes (TiO2NTs). However, due to the extended duration of calcination and further modifications, much faster alternatives are highly required. As a response to growing interest in laser modification of nanotube arrays, a comprehensive investigation of pulsed-laser irradiation and its effect onto TiO2NT properties has been carried out. The impact of irradiation onto the surface being placed at different angles in respect to the laser beam was studied and evaluated. The usage of the motorized table enables formation of laser-treated traces over the selected area. SEM and TEM analysis provides insight into morphological changes and shows partial melting of nanotubes surface, which is accompanied by the decrease of internal TiO2 tube diameter just below the melted region. Although structural and optical analysis consisting of Raman, photoluminescence and UV–Vis data indicate that presented method does not result in complete material crystallization, it promotes creation of advantageous localized states within TiO2 bandgap that may play a crucial role in charge separation. Moreover, impressive improvements to the mechanical properties resulting from the laser-modification are presented.


Introduction
The titanium dioxide nanotubes (TiO 2 NTs) without a doubt can be included in the group of the most investigated functional nanomaterials since the number of paper dedicated to them increases at an impressive rate [1]. They are typically obtained via electrochemical oxidation of titanium foil and their properties are highly dependent on morphology, such as their final length [2], spacing [3] or pore size [4]. However, they can be relatively easily tailored by adjusting anodization parameters and electrolyte composition. Owing it to easily controlled fabrication and quite low equipment requirements, TiO 2 NTs have found a lot of success in research involving photocatalysis [5], solar cells [6], filtering systems [7], and gas sensors [8]. However, as it is known, asanodized TiO 2 NTs are in the amorphous phase, limiting their immediate usability. In order to obtain crystalline material, they are usually annealed in a furnace in about 450°C for at least one hour. Nevertheless, considering heating and cooling time, the whole process can take up a whole working day. Thus, the main drawback of this method is the heat treatment duration to achieve crystalline phase, while the size of the material placed in the oven is limited by its dimensions.
As an alternative to traditional calcination process, low-temperature approaches towards phase transition have been proposed in the literature. Until now, successful crystallization of TiO 2 NTs was reported using plasma [9] and vapour treatment [10], alternating voltage [11] and transmission electron microscope (TEM) [12]. According to Yu et al. [10], different methods used for phase transformation exhibit a great influence onto the crystallinity, morphology, architecture as well as photoactivity and the biological response of TiO 2 NTs layer. Therefore, it should be kept in mind that the change of phase is not the only result of the particular treatment of as-anodized titania nanotubes. Recently published results reporting on plasma-assisted approach [9], show that such conditions induce crystallization within only a few seconds of exposure to oxygen plasma, without changing the morphology of NTs. At the certain plasma conditions, namely, forward power of 800 W applied only for 1 s is sufficient for reaching anatase phase, but the temperature of the whole sample rapidly increases up to 1030°C, while the maintenance of temperature below 1000°C at lower plasma power does not ensure crystallization. Consequently, for a successful phase transition, the whole substrate is treated at a much more elevated temperature than in the furnace. In the case of water vapour treatment [13], as the exposure exceeds 30 min, NTs finally convert into the porous rod-like structure along with a significant change in the recorded FTiR, Raman and UV-vis spectra. Other method concerns the immersion of Ti foil covered by TiO 2 NTs in water at 92°C for over 6 h and results indicate that water plays a crucial role in the rearrangement processing of the TiO 6 2− octahedral. As Liao et al. [14] reported, during hot water treatment the surface roughness increases and the initially ordered morphology is not observed anymore. Unfortunately, upon sample immersion, the tubular layer peels off the substrate and is finally used in the form of powder for photocatalytic degradation tests. It should be also underlined, that those methods are difficult to scale-up since dimensions of water batch chamber and furnace are limited. Moreover, regardless of the dimensions of the samples and their number designated to simultaneous treatment, the cost of larger size equipment itself increases significantly. On the contrary to above-described methods, laser irradiation has been proposed as an easy, fast and scalable way of inducing phase transition in TiO 2 NTs as well as surface modification. In the past, laser treatment was used towards crystallization of thin layers [15,16] as well as welding the interface between TiO 2 and fluoride tin oxide electrode towards DSSC efficiency improvement [17]. Following promising results on flat titania surface, a laser beam was applied for tubular layer treatment. Nevertheless, there have been only few reports describing changes of morphology, structure or properties of the nanotubes. Xu et al. [18] proved that melting of upper parts of the anatase NTs layer upon laser irradiation occurs and impacts significantly their photoelectrochemical response.
On the contrary, Hsu et al. [19] focused on the annealing of amorphous nanotubes. The process was performed when the samples were rotated in order to homogenize modified surface. During the rotation, simultaneous tilt mode was applied to improve crystallinity across the NTs layer since the penetration depth of the laser beam is limited. Nevertheless, further materials characterization concerned only analysis of topography and crystalline phase of some selected samples. Moreover, application of the sample rotation enabling uniform surface modification still cannot be applied for larger substrates or those exhibiting irregular shape.
In order to overcome listed problems, arising when conventional calcination is performed or laser treatment within a small area is carried out, herein we propose laser irradiation of as-anodized titania, where the sample is moving relative to the beam. Such an approach enables laser beam scanning along the precisely determined path and in consequence modification over an area of any shape and size. A homogeneous modification was achieved by moving TiO 2 NT substrate located at the motorized stage both in the perpendicular plane and under tilted mode in respect to the direction of the laser beam. Such an approach affects the size of the laser spot since its area grows with increasing angle between the beam and the substrate. We present a comprehensive analysis of physical and morphological changes induced in amorphous TiO 2 NTs by nanosecond UV laser irradiation. While morphological transformations were captured by SEM both at the surface and at the cross-section of the material, certain topographical features were better visualized through TEM imaging. The change in crystallinity was observed using Raman spectroscopy and optical properties were discussed based on UV-Vis and photoluminescence measurements. Moreover, the mechanical properties of reference and modified material were determined via nanoindentation, as it has been shown that pillars covered by TiO 2 can endure stress well and show improved mechanical properties under certain conditions [20].

Sample preparation
Ordered titanium dioxide nanotubes were synthesized by the means of electrochemical oxidation carried out in the cylinder shape electrochemical cell with an outer jacket. First, titanium foil (99.7%, Strem) was cut into 35 × 25 mm pieces and cleaned ultrasonically for 10 min in acetone (p. a. Protolab), ethanol (96%, Chempur) and deionized water (0.08 µS, Hydrolab), respectively. After drying in air, the foil was used as a working electrode in the anodization process, while platinum net (20 × 25 mm) acted as a counter electrode. The electrolyte solution contained 9.2 wt% NH 4 F (p. a. Chempur), 15 vol% deionized water and 85% of ethylene glycol (p.a. Chempur). The procedure was performed in 23°C (controlled by thermostat Julabo F-12), under 40 V for 45 min. Ramp-up and ramp-down voltages were set to 0.1 V/s and controlled by in-house built hardware equipped with a constant voltage generator (MPS-600-5L-2) and precisely adjusted using dedicated software. After the process, the foil was rinsed with and submerged in ethanol for 1 h, after which it was left to dry in air. Substrates obtained that way underwent thermal treatment in the pulsed and continuous mode.
Laser modification of the TiO 2 NTs was performed using Nd:YAG pulsed laser (6 ns, Quantel), equipped with 4th harmonics generator crystals (λ = 266 nm) and set to 2 Hz pulse repetition rate. The laser beam was homogenized and focused on the surface of the sample, creating a 2.4 × 2.4 mm spot at 0°angle of incidence (AOI). The fluence of laser pulse was changed between 25 and 75 mJ/cm 2 and measured by Coherent FieldMaxII Top radiometer. The sample was placed onto the computer-controlled 4-axis stage (SmarAct) moving precisely during the whole modification process. The tilted mode applied to the motorized table enables change of the angle between the plane of the sample surface and the incident laser beam. Following that, the angle was increased from 0°to 80°, changing the area exposed to the laser radiation. It should be noted, that with increased angle of irradiation, the spot size grows from 2.4 up to 14.1 mm, following inversely cosine of the angle of incidence (a = a 0 /cosθ). In accordance with this change, the laser fluence was established. At each angle, the sample moved along the axis parallel to the shorter spot edge at a constant speed of 3.5 mm/min two times -back and forth. The whole process was performed in ambient conditions. Continuous thermal treatment was carried out in order to obtain the reference sample. It was prepared with the use of electric furnace (Nabertherm) in 450°C for 2 h with the heating rate set to 2°C/min. Cooling down was performed freely to the room temperature. The whole process takes over 14 h, including heating up and cooling down.

Characterisation
Scanning Electron Microscope pictures were taken with FE-SEM, FEI Quanta FEG 250 equipped with a secondary electron detector. The SEM images were taken at the surface, at the cross-section, and in different places of the sample surface to confirm its uniformity.
TEM images were taken using JEOL ARM 200F transmission electron microscope working at 80 kV. TiO 2 NTs were prepared by Focus Ion Beam (FIB) using a JEOL JIB-4000 instrument, with Ga ion source at different voltages (3-30 kV). Cross-sections were placed on commercially available Cu grids for analysis.
Raman characterization was performed using Renishaw Raman spectrometer equipped with 514 nm argon ion laser with emission power of 50 mW. Measurements were taken using ×50 objective, 10 s exposure time and utilizing 10% of laser power in the range of 100-2000 cm −1 . The Raman spectra were collected in three different places of the sample, in the centre and towards vertical edges when titled mode was applied.
UV-Vis measurements were taken in reflectance mode with PerkinElmer dual-beam spectrophotometer in the range of 300-1100 nm. Scanning speed was set to 120 nm/min. Photoluminescence (PL) spectra of the samples were measured at room temperature using a SHAMROCK -SR-303I-A Spectrograph (Andor Technology) equipped with a 450 mW LED as the excitation light source (λ = 365 nm) and ICCD camera as a detector. Optical filters, FGUV-11 (Thorlabs) and GG 400 (Schott AG) were used to eliminate undesired light modes during measurements of the PL excitation spectra. Overall, 200 scans in the range of 370-800 nm were collected and their sum was used for further evaluation.
Nanoindentation tests were performed with the NanoTest™ Vantage (Micro Materials) using a Berkovich three-sided pyramidal diamond. Fifteen independent measurements of indentation were carried out for each sample and two references specimens. One curve for a medium hardness was selected for analysis. The maximum applied force was equal to 0.45 mN, the loading time was set up at 5 s, the dwell period at maximum load was 2 s and unloading time was set up at 5 s. The distances between the subsequent indents were 20 μm. Based on the load-penetration curves and used the Oliver and Pharr method, surface hardness (H) and reduced Young's modulus (E) were calculated using the integrated software.

Results and discussion
3.1. Morphology   Fig. 1A and 1C present SEM images of as-anodized TiO 2 nanotubes, while Fig. 1B and 1D show furnace annealed sample used here as a reference one. According to the top and cross-section images, the geometrical parameters could be determined. The nanotube length was ca. 2.1 µm with the internal diameter of 100 nm, whereas wall thickness and spacing between crowns -of ca. 15 nm for both as-anodized and annealed samples.
The effects of differing angle and various laser fluences onto nanotubes are visualized in Figs. 2, 3, and 4 which confirm transformation of the surface morphology. Although in all cases some degree of surface melting is visible, the effect is stronger with increased pulse fluence. In the SEM image obtained for TiO 2 NTs modified at the fluence of 25 mJ/ cm 2 ( Fig. 2A-C), some nanotubular morphology is still visible underneath thin melted layer, whereas for higher pulse fluences melted region tightly covers ordered architecture (Fig. 2D, E). Interestingly, different kind of structure emerges from the melted layeruniform morphology enriched with micropores (Fig. 2G, H). Samples modified at higher angles of incidence (Fig. 2F, I), however, require significantly higher fluences in order to transform their morphology into structures similar to those obtained at lower angles.
After irradiation, the tubular structure shortened by 0.5, 0.8 and 1 µm, for laser fluences of 25, 50 and 75 mJ/cm 2 , respectively (see Fig. 3). It is noteworthy that, in the intermediate region between nanotubes that are left intact and melted top layer, the nanotubes tend to form necks as presented in Fig. 4 A. Although the degree of contraction varies heavily, it narrows the width of each nanotube by about 50% and is likely caused by the upward material flow, towards the melted region. This effect is especially visible for the higher fluences with more developed melted region.
Introducing varying angle to the conversion process has a significant effect onto the final morphology, as illustrated in Figs. 2 and 3. As indicated in the experimental section, change in the incident angle influences the sample area exposed to the laser radiation. Although the difference between the length of melted nanotubes during modification is relatively small, transforming nanotubes at an angle decreased the total length of melted nanotubes by about 20%. This affects topological changes, which indicate that the surface of the samples modified at an angle are much less transformed compared to the case when sample plane was perpendicular to the laser beam. Interestingly, the surface of tilted (80°) altered with pulses of 75 mJ/cm 2 represents similar degree of changes than non-tilted samples treated with 25 mJ/cm 2 . In addition, modification in tilted mode results in a directional orientation of micropores due to non-uniform irradiation of any given nanotube, as presented in Fig. 4B, C. This effect was not observed by Hsu et al. [19], likely due to rotation incorporated into reported modification method. In the case of irradiation of tilted TiO 2 NTs, substantially more light is being absorbed by the lower part of nanotubes inner wall than its outer wall on the opposite side, which is covered by lower wall of the neighbouring nanotube. This causes each nanotube to melt nonuniformly, resulting in the directional orientation of micropores visible in Fig. 4B.

Structure
Raman spectra were gathered from as-anodized, thermally annealed  and laser-modified nanotube arrays and shown in Fig. 5. As-anodized titania nanotubes exhibit broad spectra without any clear peaks, which are characteristic for amorphous TiO 2 nanotubes [13,21,22]. Those wide humps appearing in the range of 120-310 cm −1 correspond to the TieO bending, while the band at 400-745 cm −1 comes from TieO stretching vibrations [23]. In Raman spectra of calcined material, clear maxima at 144 (E g(1) ), 197 (E g (2) ), 395 (B 1g ), 519 (B 1g ), and 635 (E g (3) ) cm −1 are visible, which are characteristic for anatase phase [24]. In addition, faint peak at about 1600 cm −1 was recorded which can be associated with the presence of amorphous carbon [25] originating from the thermal decomposition of the organic compound present in the electrolyte used for anodization. Presented results go in line with  those obtained by other investigators [13,22]. Although in laser-modified material all anatase peaks can be detected, their intensity decreases with increased fluence and the angle of modification. In addition, the E g(1) mode shifts towards 159 cm −1 , which indicates non-stoichiometric composition through i.e. presence of oxygen vacancies and is sometimes referred to as self-doping [26]. The signals attributed to the anatase phase become nearly undetectable for samples treated with 75 mJ/cm 2 for all modification angles. While lowering of the intensity could be explained by changes in surface geometry, the disappearance of anatase peaks might be associated with laser-induced disordering of the structure during rapid temperature changes [27,28]. The HR-TEM and SAED analysis confirms laser-induced partial crystallisation of TiO 2 NTs to anatase, as shown in Fig. 6.

Optical properties
In order to derive bandgap values of the modified material, the UV-Vis absorption spectra were measured and Tauc plots for respective samples were prepared (Fig. 7). Indirect allowed transitions were assumed, which according to Tauc's relations are also valid for noncrystalline materials [30,31]. The typical bandgap value for TiO 2 found in the literature [14,32] is about 3.2 eV, however, the calculated value in both of our reference samples equals 3.15 eV. Because E bg is the same for as-anodized and calcined material, the 0.05 eV lower value compared to other reports is likely due to the geometrical aspect of the nanotubes, which is preserved during calcination. E bg values determined for each of laser-modified samples (Table 1) are significantly lower in comparison to thermally annealed titania NTs. In addition, samples modified using 50 mJ/cm 2 laser pulses tend to have significantly lower bandgap values, than the other investigated cases (Fig. 7B, C). As an exception, within samples modified at an 80°angle, the one modified with 75 mJ/cm 2 pulses had the lowest bandgap (Fig. 7D). Interestingly, those samples share similar morphology (see Fig. 2D, E, I), which indicates favourable modification parameters for the creation of localized states within the bandgap of the material [33]. This could potentially improve their photoactivity by i.e. allowing the material to absorb longer wavelengths.
In order to describe the optical and electronic properties of titania NTs, the photoluminescence measurements were performed. An example of the recorded photoluminescence (PL) spectrum of the sample treated with a fluence of 50 mJ/cm 2 at the normal angle of incidence is given in Fig. 8 B, whereas PL for the calcined titania NTS is presented in Fig. 8 A. Comparing both spectra, one may observe clear difference especially in the contribution of the blue emission. It shows the main maximum at around 530 nm with a strong shoulder on the low wavelength side and a long tail on the other. Taking into account resolved PL spectra reported by other authors [34][35][36][37][38], the distinctive emission bands should be found at around λ blue ≈ 400-480 nm, λ green ≈ 500-540 nm, and λ red ≈ 590-660 nm. The most accurate deconvolution was performed using parameters listed in Table. The blue photoemission could be attributed to recombination of electrons trapped in oxygen vacancies below the conduction band (CB) with mobile holes present in valence band [37]. Green photoluminescence might rely on the recombination of trapped holes on Ti 3+ ions or conduction band electrons [36,38]. According to Pallotti et al. [38], excitation of red photoemission can be observed using belowbandgap activation, therefore, additional radiative channel involving trapped electrons should be present in the material which would be responsible for red photoemission between 600 and 650 nm.
Photoluminescence measurements spectra have shown that laser annealing of the TiO 2 nanotubes changes their emissivity and influences the shape of their spectrum. The strongest change relates to the emission of red light. As seen in Fig. 9 A, the lowest intensity of red emission is observed for the laser beam guided parallel to nanotubes grow direction and an increment of the angle is followed by the stronger red emission. On the other hand, the green emission is enhanced by the laser annealing and the laser beam angle is not relevant, what is shown in Fig. 9B. The blue photoemission in laser-modified samples is weaker than in reference samples and changes inversely with the angle of modification for samples modified with 50 and 75 mJ/cm 2 pulses (Fig. 9C).
While calcination leads to increased blue emissivity and decrement to green and red emissions, the laser treatment ensures lower emissions in red and blue range while simultaneously increasing green photoluminescence.
The increment of the green PL caused by the laser treatment indicates a growing number of holes trapped within the material, whereas lower emissions in red and blue suggest decreasing the number of Ti 3+ ions and oxygen vacancies through which electrons dissipate energy in as-anodized TiO 2 NTs. Schematic representation of the summary of PL mechanisms within the material is shown in Fig. 10.

Mechanical properties
In Fig. 11 it is possible to observe slight steps for both calcined and as-anodized nanotubes. They indicate that under the applied load, nanotubes first bend elastically and, as the depth increases, they break. Consequently, the material becomes denser, producing higher resistance that grows with indentation depth.
Amorphous TiO 2 NTs show higher hardness and reduced Young's modulus in comparison to calcined TiO 2 NTs. The latter relation is in agreement with other reports [39], as amorphous materials tend to have higher hardness and elasticity modulus than their crystalline counterparts.
Some parameters of laser treatment show vastly improved mechanical properties in comparison with the reference samples. Table 3 presents results of nanoindentation tests. As can be seen from the results in Table 3, almost all laser machining methods have increased hardness in comparison to calcined TiO 2 NTs (Table 3). In some cases, laser processing also increased the hardness and elasticity modulus in comparison with amorphous nanotubes obtained after anodization. A particularly visible increase in hardness and elastic modulus occurred after remelting under the highest fluence of 75 mJ/cm 2 and for 80°angle. This mode of laser processing yielded more than three-fold increase in hardness and over 50% increase in the elastic modulus in comparison with amorphous TiO 2 nanotubes. The increase of mechanical properties after laser treatment can be associated with a higher dislocation density and compressive residual stress caused by laser remelting. In addition, the hysteresis loop for each sample has been recorded (Fig. 11), indicating their ability to dissipate energy elastically [39,40].

Conclusions
In summary, we have demonstrated the effects of laser-treatment onto TiO 2 nanotube arrays, considering different modification angles. Angle-and fluence-dependent changes onto morphology were described and discussed along with, structural, and mechanical properties induced through laser irradiation. We have found that upon laser radiation top part of titania nanotubes has become porous at the expense of the thickness of an ordered tubular layer. Additionally, samples subjected to higher-energy pulses, which exhibit appreciable melted region, have a very clear contraction in their diameters in the region adjacent to the melted area. Particularly important is that such modification can be done in a precisely defined area and angle through the use of a motorized table. Although laser-treatment alone is not sufficient for phase transition of the whole titania layer material, it promotes the creation of advantageous localized states within band-gap of the material. Among others, titania NTs treated with 75 mJ/cm 2 laser pulses show superior mechanical properties compared to both amorphous and calcined samples.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to Fig. 8. Deconvoluted photoluminescence spectra of calcined TiO 2 NTs (A) and NTs modified with 50 mJ/cm 2 laser pulses at 30°angle of incidence (B).

Table 2
The fitting parameters used to perform deconvolution of photoluminescence spectra given in Fig. 8, recorded for sample modified with 50 mJ/cm 2 at 0°.  influence the work reported in this paper.  Table 3 Hardness and reduced Young's modulus for reference specimens and samples treated by the laser beam. H stands for hardness, E is a reduced Young's modulus and ± indicates standard deviation.