Background

Recently, H-doped TiO2 (H-TiO2) has triggered broad research interest due to its enhanced photocatalytic properties. Chen et al. reported that H-TiO2 nanoparticles were obtained by high-pressure hydrogen gas annealing. The nanoparticles contained an unique crystalline core and a disordered shell homo-interface which led to a narrowed band gap and enhanced photocatalytic behavior [1]. Wang et al. reported a new approach assisted by hydrogen plasma to synthesize H-doped black titania with core-shell nanostructures, superior to the high pressure hydrogenation process [2]. So far, the photocatalytic reports of H-TiO2 in literature are mainly limited to nanoparticle systems [3]. There are only a few investigations about H-TiO2 films on conductive substrates which can be used as photoanodes for three-electrode photoelectrochemical (PEC) studies [3]. Notably, highly oriented H-TiO2 nanorods (NRs) and nanotubes (NTs) have been demonstrated to be highly efficient photoanodes for solar light driven water splitting [4, 5]. Such an unidirectional nanostructure decouples the processes of light absorption and charge collection, which can benefit the charge carrier separation and transport [6,7,8]. However, the progress of hydrogen processing methods and their influence on the structural, optical, and photoelectrochemical behaviors of H-TiO2 is rarely reported due to lack of a practical hydrogenation method with excellent controllability on the processing parameters. Wang et al. reported a pioneer work of H-TiO2 nanorods grown on the fluorine-doped tin oxide (F:SnO2; FTO) substrate by high temperature hydrogen gas annealing in a tube furnace [4]. They studied the relation between the annealing temperature and photoelectrochemical properties. Due to the degradation issue of FTO substrate, the photocurrent density decreases at hydrogenation temperatures beyond 350 °C, an intrinsic relation between the annealing temperature and photoelectrochemical properties of H-TiO2 could not be indicated. The degradation issue of H-TiO2 nanorods/FTO material system will restrict its applications such as photocatalysis, hydrogen generation from water splitting and solar energy conversion.

Since the hydrogen treatment can strongly influence the structural and photocatalytic properties of H-TiO2 [9], a precise control of processing parameters (temperature, time, flux etc.) will play an important role to reproduce the core-shell structure and enhanced photocatalytic properties of H-TiO2 in order to identify the process–structure–PEC property relationship. It is known that rapid thermal annealing (RTA) is a standard semiconductor processing technique where the processing parameters can be precisely controlled by a PC [10, 11]. It has become essential to the fabrication of advanced semiconductors such as oxidation, annealing and deposition. It can provide fast heating and cooling to process temperatures of 300–1200 °C with ramp rates typically 10–250 °C/s, combined with excellent gas ambient control, allowing the creation of sophisticated multistage processes within one processing recipe. To our best knowledge, no work of H-TiO2 nanorods hydrogenated by RTA is reported till now. In comparison to the conventional hydrogen gas annealing, RTA allows the local thermal annealing on the samples. The RTA chamber is cooled down with cycled water, only the sample and sample holder (usually Si wafer) are locally heated by a set of infrared lamps (Fig. 1). Furthermore, several sharp quartz tips are used to support the sample and Si sample holder in order to prevent the thermal loss. It is evident that RTA consumes less energy than that of conventional hydrogen gas annealing.

Fig. 1
figure 1

Schematic of hydrogenation of TiO2 nanorods by rapid thermal annealing (RTA) with controlled temperature recipe

Full size image

In this work, we report for the first time the use of the RTA method to successfully prepare H-TiO2 nanorods grown on FTO substrate. The relationship of the processing parameters and morphology with the optical and photoelectrochemical properties is further illustrated with systematic characterization.

Methods

Growth of Ordered TiO2 Nanorods on FTO (F:SnO2) Substrate

TiO2 nanorods were directly grown on the FTO substrate via a previously reported hydrothermal method [12,13,14]. Typically, 0.35 ml of tetra-tert-butoxy titanate (Ti(OtBu)4) was dissolved in 30 ml of 6 M HCl, and the solution was transferred into a steel-lined Teflon autoclave, where a cleaned FTO substrate was placed. The autoclave was constant at 160 °C for 18 h and the coated FTO substrate was washed several times with deionized water and ethanol. To remove the chemical residues and improve the crystallinity of TiO2 nanorods and their electric contacts with FTO glass, the samples were annealed at 550 °C for 3 h in air.

Hydrogenation of TiO2 Nanorods by Rapid Thermal Annealing

In comparison to conventional hydrogen treatment in a tube furnace, we applied rapid thermal annealing (RTA) to make use of the controlled temperature recipes. Figure 1 shows the standard configuration of a typical rapid thermal annealing system. The samples were placed on a silicon wafer and heated by infrared lamps in hydrogen atmosphere. The temperature was measured by a pyrometer and controlled by a PID loop. All processing parameters (temperature, time, and hydrogen flux) were controlled precisely by a programmable recipe; therefore, a reproducible hydrogenation on TiO2 nanorods could be realized. For our experiments we chose a hydrogen flux of 50 sccm, 1 bar pressure, a temperature ramping rate of 5 °C/s, and the annealing temperatures of 350, 400, and 450 °C for 1 h each.

Characterizations of the Samples

The H-TiO2 nanorods were investigated by using a scanning electron microscope, a X-Ray diffractometer, an optical absorption spectrometer and a high resolution transmission electron microscope (HR-TEM) Jeol J2010F FEG operating at 200 kV and coupled to a Gatan GIF filter for EELS analysis.

Photoelectrochemical (PEC) Studies

A Newport solar simulator 150 W Xe lamp with AM 1.5G filter acted as a light source for the PEC test. The lamp power was adjusted using a reference silicon solar cell to obtain 100 mW/cm2 (1 sun). All J-V curves were recorded with a Princeton Applied Research 2273 potentiostat using 1 M KOH as electrolyte in three-electrode configuration (Ag/AgCl in 3 M KCl as a reference electrode and platinum wire as a counter electrode). The IPCE were performed based on Oriel QE-PV-SI (Newport Corporation) and electrochemical workstation (CHI660E, China). The M-S measurements were performed on the electrochemical workstation (Gamry, USA).

Results and Discussion

To study the possible change of morphology and crystal phase of H-TiO2 nanorods, scanning electron microscopy (SEM) images and X-ray diffraction (XRD) spectra were collected. Figure 2b shows the in-plane and cross-sectional SEM images of H-TiO2 nanorods treated at 400 °C. No changes of morphology and alignment are observed before and after RTA treatment. The nanorods show a good vertical alignment and similar length indicating high quality of H-TiO2 nanorod arrays. Figure 2c shows the XRD patterns of H-TiO2 nanorods treated at 350, 400, and 450 °C. The (101) and (002) diffraction peaks confirmed the good alignment of TiO2 nanorods on the FTO substrate [12,13,14]. After the hydrogenation, the intensities of (101) and (002) diffraction decrease indicating the generation of disordered shell around the TiO2 nanorods. In comparison to the (101) peak, the intensity of (002) decreases more when the annealing temperature increases. It indicates that the tips of TiO2 nanorods are treated stronger than the sidewalls near to the FTO substrate.

Fig. 2
figure 2

SEM images of (a) as-prepared TiO2 and (b) H-TiO2 nanorods treated at 400 °C. c XRD patterns of H-TiO2 nanorods treated at different temperatures

Full size image

The microstructures of single H-TiO2 nanorod and the outmost disordered shell were further investigated by high-resolution transmission electron microscopy (TEM). Figure 3 shows the TEM images of as-prepared and H-treated TiO2 nanorods. The as-prepared TiO2 nanorods are confirmed to be single crystalline (Figs. 3a–c) whereas the H-treated TiO2 nanorods show a single crystalline core and disordered shell heterostructure [15]. The amorphous shell of the sample treated at 400 °C (Figs. 3d–f) is ca. 4 nm whereas the shell of the sample-treated at 450 °C (Figs. 3g–i) becomes thicker and defective.

Fig. 3
figure 3

TEM, HR-TEM and FFT images of (a–c) as-prepared TiO2 nanorods and H-TiO2 treated at 400 °C (d–f) and 450 °C (g–i). The scale bar of Figs. 3 a, d, g is 50 nm whereas the scale bar of Figs. 3 b, e, h is 5 nm

Full size image

The optical absorption spectra of as-prepared TiO2 and H-TiO2 were recorded in order to study the modified optical absorption behavior of H-TiO2. Additional file 1: Figure S1a shows the small red-shift of band edge and gradual increase of visible absorption with increasing the annealing temperature. The Tauc plot (Additional file 1: Figure S1b) shows the band gap narrowing from 3.0 to 2.9 eV. The disordered shell contains a lot of oxygen vacancies and hydrogen atoms fill up partially in positions of oxygen vacancies which is assumed to result in the slight deformation of the rutile lattice cells and the band structure [16]. It is noted that the as-prepared TiO2 nanorods show the increased visible absorption in the range of 400–850 nm (Additional file 1: Figure S1a). This is due to the multi-step scattering of photons between the 3D nanorods. The visible absorption in the range of 400–850 nm of H-TiO2 nanorods increases further with increasing hydrogenation temperature. It could be attributed to the rough and porous shells and the altered absorption properties of the disordered shell of H-TiO2 nanorods to trap more visible photons.

To evaluate the photoelectrochemical behavior of H-TiO2 nanorods, J-V and IPCE measurements were performed in a three-electrode electrochemical system. Figure 4 shows that H-TiO2 nanorods exhibit a significant change in PEC performance with a maximum photocurrent density after hydrogen treatment at 400 °C. The photocurrent density of H-TiO2 treated at 400 °C saturates at a lower potential of −0.4 V vs Ag/AgCl whereas the density of as-prepared TiO2 saturates at −0.2 V vs Ag/AgCl. The photocurrent density of H-TiO2 treated at 400 °C reaches ca. 3.7 mA/cm2 at 0.23 V vs Ag/AgCl, which is more than four times larger than that of the as-prepared TiO2. The corresponding photoconversion efficiency is shown in Additional file 1: Figure S2.

Fig. 4
figure 4

J-V curves of pristine TiO2 and H-TiO2 nanorods in 1 M KOH solution in the dark and under solar illumination (light)

Full size image

To clarify the influence of annealing temperature on the photoactivity, IPCE measurements were performed on the as-prepared and H-treated samples at 0.23 V vs Ag/AgCl (Fig. 5). IPCE can be expressed by the equation (eq. 1):

$$ IPCE=\frac{1240{J}_p}{\uplambda {J}_{\uplambda}}\mathrm{x}100\% $$
(1)
Fig. 5
figure 5

IPCE spectra of pristine TiO2 and H-TiO2 nanorods measured at the bias potential 0.23 V vs. Ag/AgCl

Full size image

where J p is the measured photocurrent density at a specific wavelength, λ is the wavelength of incident light, and J λ is the measured irradiance at a specific wavelength. Figure 5 shows that the as-prepared TiO2 exhibits a maximum of about 45% IPCE at a wavelength of around 390 nm, consistent with literature data of rutile TiO2 nanorods. It is interesting to note that the IPCE of H-TiO2 shows an overall enhancement in the 300–600 nm range. The IPCE curves of H-TiO2 have three interesting features: (i) large increase of IPCE in the UV range from 300 to 400 nm; (ii) The band edge is red-shifted from 410 to 425 nm which is also observed in optical absorption spectra (Additional file 1: Figure S1); (iii) The IPCE of H-TiO2 samples show enhancement of IPCE in the visible (400–600 nm) range, in comparison to that of as-prepared samples. The 400 °C treated sample shows the best IPCE values in the range of 300–600 nm.

The influence of annealing temperature on H-TiO2 was further studied by Mott-Schottky (M-S) plots (Fig. 6). It is expected that the slope of H-TiO2 decreases with increasing annealing temperature indicating the increased donor density and conductivity according to the Mott − Schottky equation (eq. 2)

Fig. 6
figure 6

Mott-Schottky plots of pristine TiO2 and H-TiO2 nanorods

Full size image
$$ {N}_d=\left(\frac{2}{{\mathrm{e}}_0{\upvarepsilon}_0{\upvarepsilon}_{\mathrm{r}}}\right){\left[\frac{d\left(1/{C}^2\right)}{ d V}\right]}^{-1} $$
(2)

where Nd is the donor density, e0 is the electron charge, εr is the dielectric constant of TiO2 nanorods, ε0 is the permittivity of vacuum, and V is the applied bias voltage. The calculated donor densities of H-TiO2 are 5.2 ± 0.1 × 1017 cm−3 (350 °C), 2.1 ± 0.1 × 1018 cm−3 (400 °C) and 3.0 ± 0.2 × 1017 cm−3 (450 °C) which are higher than that of the as-prepared TiO2 (1.5 ± 0.1 × 1017 cm−3) [17]. The 450 °C treated sample shows abnormal M-S plot which may be attributed to the more defective structures which is observed in the TEM image (Fig. 3g). Additionally, the flat band potentials (V fb) of as-prepared and H-TiO2 were determined by the extrapolation of straight line to the abscissa in the Mott-Schottky plots (Fig. 6). The Vfb of H-TiO2 treated at 350 °C is similar to that of as-prepared TiO2, however, the Vfb was observed to decrease from −0.95 V (pristine TiO2) to −0.97 V for the H-TiO2 treated at 400 °C. The negative shift of Vfb after RTA hydrogenation at 400 °C could be attributed to the substantially increased donor density (Additional file 1: Table S1), which could consequently shift the Fermi level (EF) of H-TiO2 towards to the conduction band (Ec). The Vfb was observed to increase from −0.95 V (pristine TiO2) to −0.71 V for the H-TiO2 treated at 450 °C. It could be due to the inhomogeneous defective structure (Fig. 3g) and decreased donor density (Additional file 1: Table S1).

To study the possible degradation of the FTO electrode in the RTA process, the sheet resistance of FTO treated at 350, 400, and 450 °C was measured by the four-probe method. In comparison to untreated FTO (~18.13 Ohm/sheet), the sheet resistances of the FTO after RTA treatment are unchanged (Fig. 7). Wang et al. reported the investigation of H-TiO2 nanorods by hydrogen gas annealing in a furnace tube and observed the degradation of the FTO layer when the temperature was higher than 350 °C (Fig. 7) [4]. In that study, it was difficult to study the effect of temperature-dependent hydrogen gas treatment due to the degradation issue of FTO layer. As no FTO degradation is observed in RTA process, the relation between RTA processing temperature, H-TiO2 structure and its properties can be discussed here in order to understand the mechanism of enhanced PEC performance. Our results could be related to the following factors:

Fig. 7
figure 7

Sheet resistances of FTO and RTA treated FTO substrates compared with literature

Full size image
  1. (i)

    3D morphology of H-TiO2 nanorods

In comparison to 2D films, 3D nanorods with subwavelength-scale induce the light-trapping effect [18,19,20,21]. When the light is incident on the surface of 3D nanorods, the photons will be reflected back and forth between the nanoscale gaps resulting in an increase of light absorption (Scheme 1). Higher optical absorption and lower reflection are the important factors in achieving higher photocurrent density. Cho et al. compared the current density with the best data of 2D and 3D TiO2 from literature [22]. It is evident that 3D TiO2 can achieve the photocurrent density 1 mA/cm2, three times higher than that of 2D TiO2 (ca. 0.3 mA/cm2) because 3D TiO2 can simultaneously offer larger surface area with electrolyte, higher optical absorption and shorter diffusion path length of holes from bulk via the surface into the electrolyte than those of 2D TiO2.

Scheme 1
scheme 1

Light absorption and carrier charge transport in (a) 2D TiO2 film; (b) untreated 3D TiO2 nanorods and (c) 3D H-TiO2 nanorods, the outmost disordered shell is drawn with black line and the enhanced light scattering is marked with thick arrows. Relation between the diameter d of nanorods and the depletion region width W: (d) d > > W; (e) d > W and (f) d ~ W

Full size image

Furthermore, the optical spectra of H-TiO2 (Additional file 1: Figure S1) show a band gap narrowing and an enhanced visible absorption after the RTA hydrogenation. Cho et al. prepared branched TiO2 nanorods and co-doped C, W-TiO2 to achieve an enhancement of the visible optical absorption and PEC performance [22, 23]. It is evident that the 3D surface of TiO2 nanorods will be used not only for morphology change, but also for the change of optical absorption via chemical doping or hydrogenation in our work.

  1. (ii)

    The role of disordered shell (TiO2-xHy)

The HR-TEM results show that the thickness of disordered shell increases with increased annealing temperature which is rarely reported in literature (Fig. 3). The EELS line-scan (Additional file 1: Figure S3) shows the O/Ti ratios across a single nanorod of as-prepared TiO2 and H-TiO2 treated at 400 °C, respectively. The O/Ti ratio of as-prepared TiO2 remains constant whereas the O/Ti ratio of H-TiO2 treated at 400 °C decreases gradually in the ca. 4 nm range of disordered shell. Recently Lü et al. prepared TiO2 homo-junction films consisting of an oxygen-deficient amorphous layer on top of a highly crystalline layer in order to simulate the crystalline core-disordered shell configuration of black TiO2 nanoparticles [24]. They observed that metallic conduction was achieved at the crystalline − amorphous homo-interface via electronic interface reconstruction, which may be the main reason for the enhanced electron transport of black H-TiO2. Since the composition of the disordered shell is important to understand the crystalline core-disordered shell homo-interface, detailed structural analysis will be performed in future studies.

We compared the IPCE data with literature in order to understand the improved PEC performance. Wang et al. reported nearly 100% IPCE above band-gap and less than 0.5% IPCE in the visible light range for the best samples treated at 350 °C [4]. Yang et al. reported ca. 80% IPCE above band-gap and less than 1.5% IPCE in the visible range [25]. Our IPCE spectrum of H-TiO2 treated at 400 °C shows nearly 100% IPCE in the UV range and ca. 2.5% IPCE in the visible range, additionally the band edge is red-shifted indicating a bandgap narrowing from 3.0 to 2.9 eV. This results in higher photocurrent density during operation, in comparison to literature.

The influence of the disordered shell on the PEC performance could be: (i) efficient electron – hole separation: Fabrega et al. studied the relationship between the donor density (Nd) and the corresponding depletion region width (W) [17]. If the donor density is too high, the depletion region width is more localized nearby the nanorods surface which results in lower electron–hole separation efficiency (Scheme 1d). Therefore the control and optimization of donor density by hydrogenation will play a key role to achieve high PEC performance. Additional file 1: Table S1 shows the Nd values as a function of RTA annealing temperatures. It is evident that the donor density initially increases with increasing annealing temperature up to 400 °C. The depletion region width of 400 °C treated sample is ca. 78 nm which is close to the diameter of TiO2 nanorods. It means that the H-TiO2 nanorods are fully depleted and the electron–hole separation takes place in the whole nanorods (Scheme 1f). For higher temperatures the increasing disorder leads to a decrease in the PEC properties indicating an optimum treatment temperature at 400 °C. (ii) efficient hole injection: Pesci et al. performed transient absorption (TA) to study the lifetime of photo-generated holes [26]. Effective suppression of microsecond to seconds charge carrier recombination could be the key factor to improve the photoelectrochemical activity of H-TiO2. In comparison to the diameter of TiO2 nanorods, the less than 5 nm thick disordered shell is more like a surface co-catalyst [27, 28] which can (i) remove the surface electron trapping states; and (ii) promote hole injection into the electrolyte.

  1. (iii)

    Hydrogenation issues of H-TiO2 nanorods/FTO system

In recent years, one-dimensional (1D) TiO2 nanorod arrays on FTO substrate received broad interest due to their utilization in important applications such as photocatalysis, hydrogen generation from water splitting and solar energy conversion [3]. It is clear that TiO2 nanorods/FTO system is not suitable for high temperature hydrogenation. The decreased conductivity of the FTO substrate is due to the formation of Sn metal upon reductive high temperature hydrogenation, the photocurrent of H-TiO2 nanorod photoanode is limited by the decreased FTO conductivity [4]. Our investigation shows that RTA approach is better than the conventional hydrogen gas annealing and could be a suitable solution for the hydrogenation of TiO2 nanorods / FTO system. Here, the H-TiO2 nanorods/FTO system without FTO degradation by RTA treatment could provide better performance in the above-mentioned applications [3].

The physical insight for the decreased PEC performance of H-TiO2 upon high temperature hydrogenation is not entirely understood. Leshuk et al. proposed that high temperature hydrogenation could be counterproductive to improving the photocatalytic activity of TiO2 because of its propensity to form bulk vacancy defects [29]. In comparison to TEM image of 400 °C treated sample, the H-TiO2 treated at 450 °C shows inhomogeneous contrast and more defective structures (Fig. 3g) indicating strong structure destruction of TiO 2 during the hydrogenation process at higher temperature. It is clear that a controlled and suitable hydrogenation process is required for the development of H-TiO2 systems because their PEC properties are strongly determined by the art of treatment.

Conclusions

In summary, we presented a controlled and local RTA hydrogenation method to treat TiO2 rutile nanorods and studied the effects of annealing temperature on the structural, optical and PEC properties of H-TiO2 nanorods. Our investigation shows that the improvement of PEC performance could be attributed to (i) band gap narrowing from 3.0 to 2.9 eV; (ii) improved optical absorption in the visible range induced by the three-dimensional (3D) morphology and rough surface of the disordered shell; (iii) increased proper donor density; (iv) enhanced electron–hole separation and injection efficiency due to the formation of disordered shell after hydrogenation. As there is an optimum of the PEC properties with respect to the hydrogenation process, the RTA method with its precise control over the processing parameters could be used as one of the standard hydrogenation processes for large-scale industrial applications. The features of RTA and conventional hydrogen gas annealing are summarized in Table 1. It indicates that RTA is more efficient than the conventional hydrogen gas annealing.

Table 1 RTA vs. conventional hydrogen gas annealing
Full size table