Photoelectrochemical properties of TiO₂ nanotube arrays: effect of electrolyte pH and annealing temperature

Abstract

The effect of electrolyte pH and annealing temperature on the formation of TiO₂ nanotube arrays in connection with the photoelectrochemical response was investigated in this article. Well-aligned TiO₂ nanotube arrays were fabricated by anodisation of Ti foil in an electrolyte consisting of 1 M of glycerol (85 wt% of glycerol and 15 wt% of water) with 0.5 wt% of NH₄F at 30 V for 30 min. The pH of the electrolyte was varied from pH 1 to 7. With the increase of electrolyte pH to neutral condition, the length of the nanotube arrays was increased from ∼320 to 1100 nm. As-anodised TiO₂ nanotube arrays were amorphous in nature. However, anatase phase was observed after annealing at 400°C and polycrystalline anatase and rutile phase could be observed by heating up to 500°C in air atmosphere. Based on the results obtained, the length and crystalline phases of TiO₂ nanotube arrays affect the performance of photoelectrochemical response and photoconversion efficiency significantly.

Keywords:

• TiO₂ nanotube arrays
• anodisation
• electrolyte pH
• annealing temperature, photoelectrochemical

1. Introduction

Today, our world's hunger for energy is a growing matter of concern. In recent years, the rise of the carbon dioxide (CO₂) emission is one of the most serious global environmental problems due to the increase in energy consumption and lack of long-term sustainability of existing energy sources 1–5. In these aspects, we have to find out an alternate non-polluting and renewable energy source to meet our world's future needs. Many scientists believe that it should be superseded by an economy based on hydrogen 3,6–9. This hydrogen economy has been established as a potential future energy and would have numerous benefits over the existing system. The novel feature of hydrogen is that it can be produced from water using solar energy. Both the energy (solar light) and reactive media (water) are readily available and are renewable 7–12. However, there are several large technological, social and economic barriers impeding the emergence of hydrogen energy technology. In order to steer the economy away from its dependency on fossil fuels and towards a hydrogen-based renewable energy, improvement of photoconversion efficiency in photoelectrochemical (PEC) system will be a monumental challenge.

PEC system is one of the most attractive methods to produce hydrogen which has high potential in hydrogen economy 13–16. N-type semiconducting titanium dioxide (TiO₂) with band gap of approximately 3.2 eV has been widely investigated to meet the requirement of the PEC 14–17. TiO₂ has attracted great attention owing to their excellent characteristic such as great capacity for oxidation, wide band gap, non-toxicity, low cost and long-term stability 15,16. In spite of its promising properties, TiO₂ nanoparticles are still far away from becoming a potential semiconductor for PEC cell. Despite its low cost, making high efficient PEC cell using TiO₂ nanoparticles as photoanode is a challenge unless two important issues are addressed 18,19. It is recognised that the performance of the PEC cell is restricted by electron loss in nanoparticle network and limitation of the photon absorption from illumination due to the small surface area. Therefore, achievement of high efficiency of TiO₂ photoanode requires a suitable architecture that minimises the electron loss at nanostructure connections and maximises photons absorption 18–21. Such architecture ought to have high surface area in contact with the electrolyte species, such as nanotubes, nanowires and nanorods 18,20–24.

Among these architectures, considerable attention has been focused on the development of TiO₂ nanotube arrays by anodisation method. Studies have indicated that vertically oriented and highly ordered TiO₂ nanotubes are of great importance due to their one-dimensional nature, ease of handling and simple preparation 25,26. Moreover, the geometrical features of the anodised nanotube arrays can be tailored by controlling the process parameters. Therefore, in recent years, considerable efforts have been focused on synthesising highly ordered TiO₂ nanotube arrays to achieve specific light absorption and propagation characteristics. In this article, a detailed study has been performed to evaluate the surface morphology of TiO₂ nanotube arrays in different electrolyte pH values and discuss the influence of tube length and crystalline phase on the PEC performance. This consecutively will aid on building up a base level of knowledge regarding a suitable architecture of TiO₂ nanotube arrays by utilising an electrolyte with different pH values and crystalline phases for better photoconversion efficiency.

2. Experimental

TiO₂ nanotube arrays were grown by anodic oxidation of Ti foils (4 cm × 1 cm) with 99.6% purity of 0.127 mm thickness. Prior to anodisation, the Ti foils were degreased by sonication in ethanol for 30 min. The Ti foils were then rinsed in deionised water and dried with nitrogen stream. Anodisation was performed in a two-electrode configuration bath with the Ti foil as the anode and platinum foil as the counter electrode at 30 V for 30 min. The electrolyte was homogenised via magnetic stirring. The electrolyte was 1 M of glycerol (85 wt% of glycerol and 15 wt% of water) containing 0.5 wt% of NH₄F. The pH of the electrolyte was adjusted to 1, 3, 5 and 7, with the addition of 0.1 M sulphuric acid (H₂SO₄) and 0.1 M sodium hydroxide (NaOH). The anodised Ti foils were cleaned using acetone and dried in nitrogen stream. The second set of experiment involves anodisation of the sample in electrolyte with pH 7 for further investigation. The effect of annealing temperature on the crystallinity of TiO₂ nanotube arrays was investigated. The annealing temperature was varied from 200°C, 400°C and 500°C for 2 h in air atmosphere.

The morphology of the anodised Ti foils was viewed using a field-emission scanning electron microscope (FESEM) Zeiss SUPRA 35VP at a working distance of around 1 mm. In order to obtain the thickness of the anodic oxide formed, cross-sectional measurements were carried out on mechanically bent samples, where a partial lift-off of the anodic layer occurred. Phase determination of the anodic layer was determined by X-ray diffraction (XRD) analysis by using Philips, PW 1729, operated at 45 kV and 40 mV.

The PEC properties of the samples were characterised using a three electrodes’ PEC cell with TiO₂ nanotube arrays as the working photoelectrode, platinum rod as the counter electrode and saturated calomel electrode (SCE) was used as the reference electrode. 1 M potassium hydroxide (KOH) with 1 wt% of ethylene glycol solution was used as the electrolyte in this experiment. All the three electrodes were connected to a potentiostat (µAutolab III). A 800 W/m² xenon lamp (Zolix LSP-X150) was used to produce a largely continuous and uniform spectrum and 100% transmittance of the light was permitted by the quartz glass as the xenon lamp shined on the TiO₂ nanotube arrays (photoanode). The xenon lamp was switched on after the three electrodes were connected to the potentiostat. Then, a linear sweep potentiometry was sweeped from −0.5 to 1.0 V at a scan rate of 5 mV/s and corresponding photocurrent was measured.

3. Results and discussion

3.1. Effect of electrolyte pH on the formation of TiO₂ nanotube arrays

In this part of experiment, the effect of electrolyte pH on the morphology of TiO₂ nanotube arrays is discussed. Figure 1(a)–(d) shows the FESEM images of the cross-section morphology of the anodised Ti foils in electrolyte that consists of 1 M of glycerol (85 wt% of glycerol and 15 wt% of water) with 0.5 wt% of dissolved NH₄F at 30 V for 30 min with different electrolyte pH value. Insets are the surface morphology of the TiO₂ nanotube arrays. For sample anodised at pH 1 (Figure 1a), nanotube with diameter of ∼100 nm and length of ∼320 nm was produced. With further increase in the electrolyte pH to 3 (Figure 1b), the length of the nanotube became longer, ∼730 nm with an average diameter of 100 nm. The length of the nanotubes increased up to ∼1000 nm with the diameter of ∼105 nm when the electrolyte pH reached 5 (Figure 1c). In this condition, the shape of the nanotubes became much more uniform compared with those produced at a low pH value. Upon further increase of the electrolyte pH to 7 (Figure 1d), TiO₂ nanotube arrays exhibited an average diameter of 105 nm and the length was ∼1100 nm. From these FESEM images, it is concluded that pH value of the electrolyte significantly affects the growth rate of the nanotube arrays but the diameter of the nanotube arrays is not strong dependence on the electrolyte pH value.

Figure 1. FESEM micrographs of the cross-section view of the TiO₂ nanotube arrays formed by anodisation of titanium foil at different electrolyte pH values: (a) pH 1, (b) pH 3, (c) pH 5 and (d) pH 7. The inset is the top view showing the diameter of TiO₂ nanotube arrays at different electrolyte pH values.

The TiO₂ nanotube arrays length with electrolyte pH value is shown in Figure 2. The growth rate of TiO₂ nanotube arrays was accelerated with increasing electrolyte pH to neutral condition (pH 7). At initial stage of the anodisation, a compact oxide layer was formed due to the migration of the O²⁻ ions towards the Ti/TiO₂ interface which induces further growth of the barrier layer (Equations (1) and (2)). Pits were formed in the barrier layer and were attacked by the F⁻ ions in the electrolyte to induce chemical dissolution (Equation (3)). The pits will grow inwards to form the tube structure. The formation of TiO₂ nanotube arrays can be explained using the following equations 22,23:

Figure 2. Variation of TiO₂ nanotube arrays length as a function of electrolyte pH value.

Low pH electrolyte (acidic condition) will produce high concentration of H⁺ ions, which reduces the hydrolysis ability of the electrolyte (Equation (1)). This subsequently results in lower concentration of oxygen and eventually leading to the formation of less dense oxide layer (Equation (2)). Therefore, the growth rate of nanotube arrays in lower pH (acidic condition) was decreased due to increasing chemical dissolution of TiO₂ occurs at the tip and inside of the nanotubes (Equation (3)). The concentration of [TiF₆]²⁻ complex and H⁺ ions was higher at lower pH value (acidic condition), which eventually accelerate the chemical etching at the tip of the nanotube, leading to the formation of shorter tubes 24,27. A schematic illustration regarding shorter nanotubes obtained at lower pH value (acidic condition) due to excessive H⁺ ions that increase the dissolution rate at the tip of nanotube and cause the local acidification and etching at the wall and inside of the tubes is shown in Figure 3.

Figure 3. Schematic illustration showing the high dissolution rate at the tip of nanotube in low pH (acidic condition) due to excessive H⁺ ions: (a) before chemical etching TiO₂ of nanotube arrays and (b) after chemical etching of TiO₂ nanotube arrays.

3.2. Effect of annealing temperature

In this part of experiment, XRD was used to investigate the effect of annealing temperature on the crystal growth and phase transition of TiO₂ nanotube arrays. Figure 4 exhibits the crystal structure of the as-anodised and annealed TiO₂ nanotube arrays at different annealing temperatures in air atmosphere. The result shows that the crystal structure of TiO₂ is dependent on the annealing temperature. The XRD pattern of the as-anodised TiO₂ nanotube arrays exhibited Ti phase only, which indicates the amorphous in nature of TiO₂ (Figure 4a). The Bragg reflection of Ti substrate was detected at 2θ values of 35.09°, 38.43°, 40.17° and 53.01° in all the XRD patterns correspond to (1 0 0), (0 0 2), (1 0 1) and (1 0 2) of Ti substrate. However, the TiO₂ nanotube arrays treated at 200°C showed only the Ti peaks and no characteristic peaks of anatase and rutile phase were observed. This result indicates that the annealing temperature was not sufficient to transform the amorphous phase into crystalline phase (Figure 4b). The presence of anatase phase was detected in the sample subjected to the annealing temperature 400°C (Figure 4c). The anatase phase present was identified by ICDD file 00-021-1272. At 2θ value of 25.5°, the peak corresponds to (1 0 1) anatase peak. In addition, several indistinct anatase peaks were found at 38.67°, 48.21°, 54.10° and 55.26° which correspond to (1 1 2), (2 0 0), (1 0 5) and (2 1 1) crystal planes for the anatase, respectively. The polycrystalline anatase- and rutile-type TiO₂ were obtained by raising the annealing temperature up to 500°C (Figure 4d). Rutile phase is vaguely detected at 2θ value of 27.4° in the XRD spectra. It was reported that rutile grows at the interface between the barrier layer and Ti metal by thermal oxidisation of Ti metal because the constraints imposed by the nanotube walls make it difficult for the anatase crystals situated there to undergo phase transformation to rutile phase 28.

Figure 4. XRD patterns of (a) as-anodised TiO₂ nanotube, (b) annealed at 200°C, (c) annealed at 400°C and (d) annealed at 500°C (A, anatase; R, rutile; Ti, titanium).

3.3. PEC response

To evaluate the PEC response, selected TiO₂ nanotube arrays from the previous experiments were used as photoanode in the PEC process. The experimental conditions for the selected anodic samples with different nanotube length and crystal structure by varying the electrolyte pH and annealing temperature are summarised in Table 1. Figure 5 shows the photocurrent density in a linear sweep potential (−0.5 to 1.0 V) corresponding to light off and light on condition, with a light intensity of approximately 800 W/m².

Figure 5. The I–V characteristics of TiO₂ nanotube arrays with different tube length and crystal structure: (a) sample C, (b) sample D, (c) sample A and (d) sample B.

Table 1. Experimental conditions used to achieve different tube length and crystal structure of TiO₂ nanotube arrays.

Sample	Anodisation condition	Crystal structure	Tube length (nm)	Diameter (nm)
A	pH 1, annealed at 400°C	Anatase	320	100
B	pH 7, without annealed	Amorphous	1100	105
C	pH 7, annealed at 400°C	Anatase	1100	105
D	pH 7, annealed at 500°C	Anatase + rutile	1100	105

TiO₂ nanotube arrays were immersed in 1 M of KOH with the addition of 1 wt% of ethylene glycol under irradiation. In this case, ethylene glycol was used as ‘sacrificial agent’ to reduce electron–hole recombination losses which significantly improved the photocurrent density of the integrated TiO₂ nanotube photoanode 29. Electron–hole pairs were generated by the absorption of photons with energies greater than band gap and charges transfer occurred at the interface due to the difference in tendency of the two phases to gain and lose electrons 30. The net result is the formation of electric field of the depletion region which separates electron and hole. The excited electrons move to the counter electrode undergoes reduction process, whereas the generated holes move towards the semiconductor/electrolyte interface where water oxidation takes place 31.

All samples exhibited insignificant dark currents under light off condition. This indicates that no photo-induced electrons were transported through the sample due to inactive photocatalysis reaction 32. However, the photocurrent density increased under illumination (light on). Thus, TiO₂ nanotube arrays can be concluded as a good photo-response semiconductor for the transfer and decay of the photo-induced electrons. Based on the results obtained, amorphous TiO₂ nanotubes (Sample B) showed poor photocurrent density of 0.03 mA/cm². The reason may be attributed to the low degree of anatase crystallinity, which increases the number of charge carrier for recombination centre. Besides, the amorphous phase comprises of various defects such as impurities, dangling bonds and micro-voids, which act as recombination centre and result in a decrease of photocurrent density. Most importantly, an amorphous TiO₂ nanotube is basically unable to develop a regular depletion region. This suggests that improved crystallinity of the samples can effectively increase the photocurrent density due to the development of a depletion layer and the reduction in recombination centres 31,33.

In contrast, the photocurrent density was found to increase for annealed samples. The sample C annealed at 400°C showed the highest photocurrent density throughout the potential window, suggesting that an efficient charge separation occurred and the photocurrent reached a high value of 0.85 mA/cm² compared to sample D annealed at 500°C which is 0.65 mA/cm². The increase in photocurrent density is likely attributed to the higher photogenerated electron/hole pairs caused by the higher content of anatase phase. This is in agreement with the results of Sclafani and Herrmann 34, which proposed that anatase phase is more photoactive than the rutile phase due to its larger specific surface area and easy hydroxylation 35. The difference in specific surface area, crystal size and pore size distribution between the anatase and rutile phase may cause defect sites. The charge transfer was obviously decreased due to the recombination centre of electron–hole pairs from the defect sites. This condition indicates that less photo-induced electrons can effectively transfer to the conduction band of TiO₂ and eventually generates lower photocurrent response.

Among of the annealed samples, the shorter nanotube (sample A) fabricated in acidic electrolyte (pH 1) showed a lower photocurrent density, which implies that effective surface area plays an important role in PEC response. Sample A has shorter tube length (∼320 nm), which will eventually lead to a reduction in the effective surface area to a great extent and thus decreasing the number of active reactions sites. Based on this result, it can be concluded that less effective surface area of TiO₂ nanotube arrays exhibited poor photon absorption from illumination. This is clearly to show that less photo-induced electrons have been effectively generated at both sides of the tube walls 36,37.

3.4. Photoconversion efficiency

To estimate the quantitative correlation of light absorption to the different length and crystal structure of the TiO₂ nanotubes, the photoconversion efficiency (η) of light to hydrogen energy was measured (Figure 6). The photoconversion efficiency η was calculated based on Equation (4):

where j _p is the photocurrent density (mA/cm²); j _p the total power output; j _p|E _app| the electrical power input; I _o the power density of the incident light (mW/cm²); the standard reversible potential (1.23 V/SHE); E _app = E _mean − E _aoc; E _mean the electrode potential (versus SCE) of the working electrode where the photocurrent was measured under illumination; and E _aoc the potential (versus SCE) of the working electrode at open-circuit condition 28.

Figure 6. The corresponding photoconversion efficiencies of TiO₂ nanotube arrays with different tube length and crystal structure: (a) sample C, (b) sample D, (c) sample A and (d) sample B.

The highest photoconversion of about 1.4% was obtained for sample C annealed at 400°C. The decrease in photocurrent efficiency was as follows, 1.0%, 0.65% and 0.05% corresponding to sample D, sample A and sample B. The results show that there is a distinct dependency of the PEC properties on length of the nanotube and crystal structure of TiO₂ nanotube arrays. In this aspect, the 400°C-annealed TiO₂ nanotube arrays (sample C) exhibited the best PEC response.

4. Conclusion

From the experimental results of this research work, well-aligned and longer TiO₂ nanotube arrays were successfully fabricated by anodisation in 1 M of glycerol with 0.5 wt% of dissolved NH₄F at 30 V for 30 min under neutral condition (pH = 7). In neutral electrolyte, dissolution rate at the top of the nanotubes can be controlled by reducing the amount of H⁺ ions and thus longer TiO₂ nanotube arrays can be fabricated at this pH value. Besides, it was found that by varying the annealing temperature on TiO₂ nanotubes, it is possible to essentially modify the phase formation and transformation to tailor its PEC behaviour. Annealing at 200°C under air atmosphere did not reveal the crystalline peaks. However, transformation of the amorphous structure into crystalline anatase phase was observed after annealing at 400°C whilst polycrystalline anatase and rutile phase could be observed by heating up to 500°C. In this study, longer TiO₂ nanotubes with high crystalline anatase phase showed higher photocurrent response with photoconversion efficiency of 1.4% and it has been acknowledged as the most efficient form for the PEC performance.

Acknowledgements

The authors thank Universiti Sains Malaysia for sponsoring this study under RU grant 814075 and the Fellowship USM and Research University Postgraduate Research Grant Scheme (Grant no. 80430146).