Thermal Ramping Rate during Annealing of TiO2 Nanotubes Greatly Affects Performance of Photoanodes

Herein, highly ordered TiO2 nanotube (NT) arrays on a Ti substrate is synthesized in a fluoride‐containing electrolyte, using the electrochemical anodization method, which yields amorphous oxide tubes. The effects of different thermal annealing profiles for the crystallization of the amorphous TiO2 NTs are studied. It is found that the temperature ramping rate has a significant impact on the magnitude of the resulting photocurrents (incident photon‐to‐current conversion efficiency [IPCE]) from the tubes. No appreciable changes are observed in the crystal structure and morphology of the TiO2 NTs for different annealing profiles (to a constant temperature of 450 °C). The electrochemical properties of the annealed TiO2 NTs are investigated using intensity‐modulated photocurrent spectroscopy (IMPS), open‐circuit potential decay, and Mott–Schottky analysis. The results clearly show that the annealing ramping rate of 1 °C s−1 leads to the highest IPCE performance. This beneficial effect can be ascribed to a most effective charge separation and electron transport (indicating the least amount of trapping states in the tubes). Therefore, the results suggest that controlling the annealing ramping rate is not only a key factor affecting the defect structure but also a powerful tool to tailor the physical properties, and photocurrent activity of TiO2 NTs.

DOI: 10.1002/pssa.202100040 Herein, highly ordered TiO 2 nanotube (NT) arrays on a Ti substrate is synthesized in a fluoride-containing electrolyte, using the electrochemical anodization method, which yields amorphous oxide tubes. The effects of different thermal annealing profiles for the crystallization of the amorphous TiO 2 NTs are studied. It is found that the temperature ramping rate has a significant impact on the magnitude of the resulting photocurrents (incident photon-to-current conversion efficiency [IPCE]) from the tubes. No appreciable changes are observed in the crystal structure and morphology of the TiO 2 NTs for different annealing profiles (to a constant temperature of 450 C). The electrochemical properties of the annealed TiO 2 NTs are investigated using intensity-modulated photocurrent spectroscopy (IMPS), open-circuit potential decay, and Mott-Schottky analysis. The results clearly show that the annealing ramping rate of 1 C s À1 leads to the highest IPCE performance. This beneficial effect can be ascribed to a most effective charge separation and electron transport (indicating the least amount of trapping states in the tubes). Therefore, the results suggest that controlling the annealing ramping rate is not only a key factor affecting the defect structure but also a powerful tool to tailor the physical properties, and photocurrent activity of TiO 2 NTs.
Among the various methods, electrochemical anodization is the most straightforward approach to produce highly ordered arrays of closely packed, vertically oriented, size-controlled, and back contacted NTs on metal substrate (Ti). [26] The synthesis of TiO 2 NTs is based on a simple electrochemical anodization of metallic titanium in fluoride-containing electrolytes in the presence of self-organizing conditions. [27,28] The size of the tubes can be altered in dimension (i.e., length and diameter) by changing the anodization conditions. However, such electrochemical anodization provides only an amorphous form of TiO 2 NTs. For many applications, the tubes need to be converted to an anatase/rutile or anatase-rutile mixture with suitable heat/annealing protocols. [29] In general, annealing of TiO 2 NTs in the presence of air at T > 300 C results in anatase phase, whereas heating at T > 500-600 C alters them to rutile phase or anatase/rutile mixture. [30] It has been reported that the introduction of point/structural defects in TiO 2 NTs will enhance the activity for various applications. [31][32][33] Such point defects can also be introduced in TiO 2 NTs under different annealing conditions [34,35] that also leads to a modification of the composition and electron transport properties, and accordingly to changes in physical properties of TiO 2 NTs. Numerous studies show for photoelectrochemical applications, that a most effective conversion of amorphous tubes to anatase by an annealing treatment at 450-500 C for 1-3 h. However, annealing temperature and time are generally well investigated, [36] some other crucial factor such as the thermal ramping/heating rate, is much less explored. Generally, annealing treatments are performed in a furnace, but this can also be carried out in rapid thermal annealers (RTAs) if annealing with specific ramping/heating rates is desired.
While some early reports address some effects of the temperature ramping rate, [37] we provide herein a systematic investigation of the photoresponse from a most typical anodic NTs morphology of anatase using different annealing ramping/heating rate. The causes for the observed significant effect of the ramping rate were examined.

Reagents and Chemicals
Titanium foil (125 μm) with 99.6% purity was purchased from Advant Research Materials England, ethylene glycol was obtained from Sigma-Aldrich, ammonium fluoride (NH 4 F) was purchased from MERCK, Germany. We used deionized water (DI) in all experiments.

Growth of TiO 2 NTs
For the growth of TiO 2 NTs, we used titanium foil of 125 μm thickness. For all experiments, the obtained titanium foil was cut into 1.5 Â 1.5 cm 2 pieces and cleaned with acetone, ethanol, and DI water using sonication for 30 min followed by drying in a nitrogen (N 2 ) stream. Fresh electrolyte was prepared using ethylene glycol (EG) with the addition of 0.15 M NH 4 F containing 3.0 wt% DI water. The highly ordered TiO 2 NTs were synthesized using a two-electrode potentiostatic electrochemical cell configuration with a platinum electrode as a cathode and titanium foil (1.5 Â 1.5 cm 2 ) as an anode, which was mounted at the bottom of the cell. Anodization was carried out by applying a constant voltage (60 V) from the open-circuit potential (OCP) for 30 min using VOLTRAFT VLP2403 Pro DC power supply. After completion of the anodization experiment, the samples were washed with ethanol and DI water followed by drying in an N 2 stream. All experiments were carried out at room temperature. The obtained TiO 2 NTs were amorphous and annealed at 450 C for 1 h in air using RTA (Jipelec Jetfirst 100) for obtaining anatase TiO 2 NTs, as shown in Figure 1b. To see the effect of annealing temperature and time, we annealed all samples at different ramping/heating rates. The samples were annealed at 50, 10, 1, and 0.1 C ramping rates with different heating times such as 9, 43, 430, and 4300 s, respectively, and named as T-50, T-10, T-1, and T-0.1, respectively. www.advancedsciencenews.com www.pss-a.com
where D is the crystallite size, K is the Scherrer constant, λ is the wavelength of the used X-ray radiation, β is the full width at halfmaximum (FWHM) in X-ray diffraction reflections of the anatase peak (101) of TiO 2 , and θ is the Bragg's angle. The morphology of the prepared NTs was characterized using field emission scanning electron microscopy (Hitachi, FE-SEM S4800). The crosssectional images were obtained by mechanically scratching the samples to examine the length of tubes. The chemical analysis of the synthesized TiO 2 NTs were carried out by energydispersive X-ray spectroscopy (EDX) using EDAX Genesis, fitted to the Hitachi FE-SEM S4800. All electrochemical measurements IPCE, IMPS, OCP, and Mott-Schottky [M-S] analysis) were carried out in a threeelectrode cell configuration, with Ag/AgCl (3 M KCl) as the reference electrode, platinum sheet as the counter electrode, and the TiO 2 NTs as the working electrode. All experiments were performed in 0.1 M Na 2 SO 4 aqueous solution as an electrolyte. Moreover, all assessments either in dark or light (IPCE, IMPS, OCP, and M-S) were carried out after the stabilization of current at a given potential.
The IPCE analysis was investigated using a LOT-Oriel 6365 with 150W Xe arc lamp equipped with a 1/8m Oriel cornerstone 7400 monochromator as a light source with a constant potential of 500 mV in the range of 650-300 nm in 5 nm steps. The IPCE value for each wavelength was obtained using Equation (2).
where J ph is the photocurrent density (mA cm À2 ), P is the power density of incident light (W cm À2 ), hν is the photon energy of the incident light (1240 eV nm À1 ), and λ is the wavelength of the light (nm). Intensity-modulated photocurrent spectroscopy (IMPS) analysis was carried out under constant potential 0.5 V using a Zahner IM6 (Zahner Elektrik Kronach, Germany) in the presence of UV-modulated light-emitting diode (λ ¼ 358 nm). The frequency analyzer (FRA) was used for frequency modulation. The photovoltage and photocurrent of the cell were examined using an electrochemical interface and send feedback to FRA. The frequency range of oscillations was 0.1-1 kHz with varying power densities of the light emitting diode (LED) (171.3, 85.5, and 9.7 μW cm À2 ). The incident light intensity on the cell was examined using a calibrated Si photodiode. The corresponding charge transport time (t rs ) constants were investigated using Equation (3), which is inversely proportional to specific frequency ( f IMPS ).
The OCP decay analysis was also investigated by illuminating TiO 2 NTs for 10 s and then the light was switched off. The OCP decay of photoinduced electrons as a function of time was recorded during relaxation from the irradiation steady-state to the dark equilibrium over a period of 20 min.
The M-S analysis was done using a Zahner IM6 system (Zahner Elektrik, Kronach, Germany) under dark conditions at a fixed frequency of 1 Hz. The applied voltage was in the range from 0.5 to À0.4 V with an amplitude of 10 mV. The donor density (N d ) was obtained using Equation (4).
where e 0 is the charge of the electron (1.60 Â 10 À19 C), ε is the dielectric constant of TiO 2 (ε ¼ 55 for anatase thin film), ε 0 is the permittivity of vacuum (8.85 Â 10 À12 Fm À1 ), and C is the capacitance obtained from the electrochemical impedance at each potential (V). However, d(1/C 2 )/dV values were calculated from the slope of the linear parts of the M-S plots. [38] 3

. Results and Discussion
A series of TiO 2 NT samples were produced by the anodizing Ti as described in Experimental Section. The samples were then annealed using different thermal ramping rates.

Morphology and Crystal Structure
The surface morphology of the as-formed and annealed TiO 2 NTs was investigated using FE-SEM. The cross-sectional SEM of the as annealed sample shows that the NTs used here are 6.7 μm in (see Figure 2a). The tube layers are made up of open tubes with smooth inner and outer wall morphologies and have well-defined tube tops. Moreover, tubes show a typical honeycomb closedpacked arrangement of discrete NTs. To obtain crystallized tubes, the anodized TiO 2 NTs were annealed at 450 C for 1 h in air at different ramping/heating rates. [39] Figure 2b-e shows the morphology of tubes annealed with ramping rates of 50, 10, 1, and 0.1 C s À1 (T-50, T-10, T-1, and T-0.1). Figure 2a shows that the tubes maintain their geometry to a large extent. Namely, the tube length, tube diameter, and the outer wall remain virtually unchanged. However, the tube tops exhibit an initiation layer [40] and some titanate decoration [41,42] which can be ascribed to the modifications when converting the amorphous tubes into the crystalline form. [43] From the top-view SEM images of crystalline NTs, the diameter and thickness of the NT walls were found to be in the range of 41-45 nm.
To confirm the conversion of the amorphous phase to anatase, XRD patterns of the annealed samples were examined. The XRD patterns of T-50, T-10, T-1, and T-0.1 are shown in Figure 2f [43,44] However, peaks at 40.1 , 53.0 , and 76.1 are indexed to (101), (102), and (201) of titanium substrate. [43,44] This illustrates that the annealing treatment changes the original amorphous phase of TiO 2 NTs to a wellcrystallized anatase phase. The crystallite size of all samples was calculated using the Scherrer Equation (1) by taking the www.advancedsciencenews.com www.pss-a.com FWHM of (101) anatase reflex at 2θ ¼ 25.3 , which is given in Table 1. The results show a similar degree of crystallization to anatase with a crystallite size of %20-23 nm.
To evaluate the chemical composition of the TiO 2 NTs in dependence of the annealing ramping rate, EDX measurements were carried out. Figure 2g-j shows the EDX spectra of T-50, T-10, and T-1, indicating the presence of carbon, oxygen, fluorine, and titanium. The presence of carbon in TiO 2 NTs originates from the electrolyte and environment. [45] Table 1 shows the percentage of different elements present in the TiO 2 NT matrix.

Photocurrent (IPCE) Analysis
To examine the photoresponse of annealed TiO 2 NTs at different ramping rates at 450 C for 1 h in air, the IPCE measurements were taken under monochromatic light illumination, as shown in Figure 3a. The IPCE values calculated from photocurrent analysis as a function of incident wavelength are highest (37%) for T-1, followed by T-0.1 (31%), T-10 (19%), and T-50 (10%). The results indicate that increasing the heating time leads to enhancement in the IPCE values up to T-1, and then reduction in IPCE values is observed. This improvement in the IPCE values with heating time may be due to the introduction of point/structural defects. [46] To see the effect of heating and cooling times, we annealed samples at different heating, cooling times and observed that only heating rate matters. The IPCE data were further plotted for the evaluation of the bandgap (E g ) of TiO 2 NTs annealed at 450 C for 1 h in air for different ramping/heating rates. Figure 3b shows the graph of (IPCE Â hν) 1/2 versus photon energy (hν), which was used to calculate the bandgap, as shown in Table 2. [47,48]

IMPS Measurement
To examine the kinetics of electron-hole recombination and charge transport (mobility and trapping) of TiO 2 NTs annealed at 450 C for 1 h in air at different ramping/heating rates, IMPS measurements were carried out. IMPS is a perturbation  technique, which uses sinusoidal oscillation of intensity of illuminated light to provide information on the transport times of photogenerated charge carriers, viz. holes to the semiconductor/electrolyte junction and electrons through the tubes to the back contact. [49] Figure 2c-e shows the normalized IMPS spectra of the as-prepared TiO 2 NTs annealed at different ramping rates at 0.5 V with different power densities of LED (171.3, 85.5, and 9.7 μW cm À2 , respectively) by plotting the photocurrent efficiency response in complex Nyquist diagram. All the spectra show one semicircle, whereas a second semicircle of the opposite sign can be discerned in some cases at high light intensities. Figure 3f shows the photogenerated charge transport time constants (t rs ) of the TiO 2 NTs annealed at different ramping/heating rates obtained from IMPS analysis and calculated using Equation (3). The results show that T-1 has the highest photocurrent efficiency, and also exhibits the shortest transport time. This indicates a smaller extent of charge trapping/detrapping in T-1. [50] Therefore, T-1 exhibits less recombination of electrons and holes during illumination, indicating less interfering electronic defects in the structure. Thus the enhanced charge transport or lower recombination may be ascribed to less point/structural defects. [51,52]

OCP Decay Analysis
To calculate the lifetime, transient photoconductivity, charge transfer efficiency, and recombination kinetics of photogenerated electrons, the OCP decay analysis was conducted. OCP decay analysis was done according to the technique reported by Zaban et al. [53,54] Figure 4a shows that after turning off the illumination, the voltage shows an exponential decay as a function of time due to recombination of charge carriers. As shown in Figure 4a that T-1 exhibits the fastest electron transfer time among all samples, leading to a longer lifetime of charge carriers. Therefore, the number of free-electron which can migrate to the surface is highest for the T-1 sample. This is well in line with the findings from IMPS measurements and their interpretation in terms of point/structural defects in TiO 2 NTs at an optimized annealing ramping rate. [52,55]

M-S Measurements
To further examine the impact of the point/structural defects generated by the effect of ramping/heating time on the TiO 2 NTs, the flat band potential (V fb ) and charge carrier density (N d ) of the prepared annealed TiO 2 NTs were calculated using M-S analysis. Figure 4b shows the M-S plots of annealed TiO 2 NTs at different ramping rates for 1/C 2 as a function of the applied potential under dark conditions. Figure 4b shows that  all samples display a positive slope, indicating n-type semiconducting behavior with electrons as the majority carrier. [56,57] The values of V fb were calculated from the X-intercept of the linear region of the M-S plot, by extrapolating the 1/C 2 ¼ 0 graph for TiO 2 NTs annealed at different ramping/heating rates. The calculated V fb values are shown in Table 2. The T-1 shows a lower V fb as compared with other NTs, suggesting more efficient charge separation and transport as discussed in previous work. [58] Moreover, the N d of different TiO 2 NTs was also calculated using Equation (4) and shown in Table 2. [59] The calculated N d values for all TiO 2 samples are found to be in the range of reported donor density values in literature for TiO 2 anatase (approximately %10 17 -10 19 cm À3 ). [58,60]

Conclusion
In this work, a widely used electrochemical anodization method was applied to synthesize highly ordered TiO 2 NT arrays of %6.7 μm length on a Ti foil in a fluoride-containing electrolyte. These NT arrays are amorphous-they were then crystallized to anatase at 450 C using different ramping/heating rates. Ramping/heating rates have a drastic impact on the IPCE of the different anatase photoelectrodes. The effect of annealing ramping/heating rates was investigated using XRD, EDX, IMPS, OCP decay, and M-S techniques. The experimental results reveal that among all samples, an optimum ramping rate (1 C s À1 ) that exhibits the highest photocurrent activity. This beneficial effect can be ascribed to the most effective electron transport-shortest transport time-and the least amount of trapping states present in these tubes. Therefore, an optimal annealing temperature and heating/cooling time is necessary for TiO 2 to achieve a minimum of point defects and as a result, the anatase nanotubular structure of TiO 2 can yield a higher photocurrent activity. We also observe that the reported effects depend only on the heating rates. Remarkably, these effects have hardly been previously reported in the literature.