Revisiting SrTiO₃ as a photoanode for water splitting: development of thin films with enhanced charge separation under standard solar irradiation

Abstract

Strontium titanate (SrTiO₃) is an n-type semiconductor with high chemical and photochemical stability. This wide band gap oxide has a band gap energy of about 3.2 eV as well as a favorable energy for photocatalysis. In this study, we demonstrate an alternative and superior method to produce Nb-doped and undoped SrTiO₃ photoanode thin films based on a colloidal deposition process which possess good activity under standard solar illumination conditions. Methanol was used as “hole scavenger,” and the results showed that the semiconductor–liquid junction (SCLJ) charge accumulation is not an important mechanism to control the photocurrent density and overpotential. In addition, experimental results suggest that the dominance of photocurrent density is controlled by the potential at the surface space charge layer for the Nb-doped SrTiO₃ and by recombination at the depletion layer for the undoped oxide.

---

Introduction

Since the seminal study of Fujishima and Honda,¹ much research has been focused on the conversion of sunlight into hydrogen as a clean and renewable energy source. In this classical work, these authors showed that it is possible to induce water-splitting by light using a TiO₂ semiconductor as a photoanode and proved that it is possible to use sunlight to produce light-to-chemical energy conversion with a photoelectrochemical cell (PEC). In the last decade, several semiconductor oxides such as Fe₂O₃,^2,3 WO₃⁴ and BiVO₄⁵ have been considered as promising materials for photoanodes in PEC devices. These oxide semiconductors had a high photocurrent, and generally their performance is directly associated with morphological control at nanoscale² which eliminated problems such as a small optical absorption coefficient and rapid electron (e⁻)–hole (h⁺) recombination due to short minority carrier (h⁺) diffusion lengths. However, these semiconductors exhibit a major drawback of a large overpotential for water oxidation which is a critical parameter for the design of PEC reactors because it can define the number of tandem devices and, consequently, how complicated the project design and fabrication of the PEC reactor will be.³

Defining the pseudo-overpotential for water oxidation (η_OX) as the difference between the onset voltage for photocurrent (V_on) under standard illumination condition (see comments in the ESI†) and the flat band potential (V_fb), i.e. η_OX = V_on − V_fb, a larger η_OX is reported for Fe₂O₃ (0.5–0.6 V) and BiVO₄ (0.4–0.5 V).³ This large η_OX is especially critical for Fe₂O₃ with a V_fb ranging from 0.4–0.5 V_RHE;³ TiO₂ and WO₃ have a η_OX of about 0.2 V.³ For the Fe₂O₃ and BiVO₄, the electrochemistry pseudo-overpotential origin can be related to slow surface reaction kinetics for water oxidation or electronic surface traps which cause a type of Fermi level pinning.³

Strontium titanate (SrTiO₃) is an n-type semiconductor with high chemical and photochemical stability. This wide band gap oxide has a band gap energy of about 3.2 eV which is a favourable energy for photocatalysis with conduction band edge of about 200 mV higher than TiO₂. In the 1970s, this oxide semiconductor was considered a promising photoanode for water splitting.^6–8 Most of the photoelectrochemical data reported in that period were collected by using SrTiO₃ single crystals as photoanodes and reached a maximum efficiency of 20% when irradiated with 330 nm light (V_appl = 0.25–0.4 V).^6–8V_FB measured by the Mott–Schottky method had a value of −0.15 V_RHE at pH = 13.6,⁹ which is one of the smallest anodic value reported in the literature when taking into account several oxide semiconductors.² More recently, Zou and co-authors¹⁰ reported a high photocurrent density using Nb-doped SrTiO₃ single crystal photoanodes for the water splitting. These authors reported an incident photon-to-current conversion efficiency (IPCE) of 15.67% for a SrTiO₃ single crystal doped with 0.07 mol% of Nb when irradiated with 298.2 nm light at an applied potential of 1.5 V_SCE at pH 5.9 (Na₂SO₄ electrolyte-0.1 M).

Most recent research related to SrTiO₃ for water splitting are focused on the development of photocatalytic materials for an overall water splitting process.^11,12 Very few literature reports describe the use of SrTiO₃ as a photoanode in the thin film shape.¹³ For instance, Shen and co-authors reported the effect of carbon quantum dots in the photoelectrochemical response of SrTiO₃ by observing that this composite material is capable of converting near infrared photon energy to a photocurrent.¹³ Moreover, recent reports have described the use of Rh-doped SrTiO₃ as a p-type semiconductor with good performance as a photocathode for water splitting.^14,15 With the exception of good results reported for a Nb-doped SrTiO₃ single crystal, a complete study of the development of this material as a photoanode thin film is lacking.

In this study, we demonstrate an alternative and promising way to produce Nb-doped and undoped SrTiO₃ photoanode thin films with good activity under standard solar illumination conditions. In this approach, we processed undoped and Nb-doped SrTiO₃ thin films using a colloidal dispersion of amorphous titanate nanoparticles as the precursor. Using a colloidal dispersion for the preparation of a photoanode thin film is an attractive and facile method and has been used by our research group to process hematite and WO₃ photoanodes with very good activity.^16,17 Moreover, a qualitative model is employed to explain the effect of donor doping impurities on photoanode performance for water splitting which brings new insights into the role of electronics and morphological parameters in photocurrent density.

Experimental section

Materials

Strontium isopropoxide (99.99%), titanium(IV) butoxide (97%), oleic acid (99%), niobium(V) chloride (99%), and oleyl alcohol were purchased from Aldrich chemical Co.; acetone and toluene were purchased from Tedia Company, Inc. SnO₂:F deposited on aluminoborosilicate glass or pure silica glass (Solaronix) was used as a transparent conductor oxide (TCO) substrate.

Nanoparticles synthesis

6 mmol of strontium isopropoxide, titanium(IV) butoxide and oleic acid were dissolved in 20 mL of oleyl alcohol. After complete dissolution, the yellow solution was heated at 260 °C during 48 hours with vigorous stirring. The colloidal which formed was cooled to room temperature and then added to 30 ml of acetone to provoke nanocrystal flocculation. The flocculate was separated via centrifugation and washed five times with mixture of high purity acetone and toluene. For the Nb-doped SrTiO₃ synthesis, we considered the SrTi_1−XNb_XO₃ stoichiometry, and we studied the following Nb concentrations: X = 0.05; 0.10 and 0.15. Nb was added as niobium(V) chloride to the strontium isopropoxide, titanium (IV) butoxide, oleic acid and oleyl alcohol solution. Then, after complete dissolution, the reaction was heated at 260 °C for during 48 hours. The nanoparticle purification was performed by the same procedure described for undoped strontium titanate.

Preparation of photoanode

Prior to the deposition, the TCO substrate was washed with acetone and isopropyl alcohol for 30 min at 70 °C and then dried in a nitrogen flow. For the colloidal solution preparation, synthesized nanoparticles were washed and re-dispersed in toluene which resulted in a very stable colloidal suspension (see Fig. S1 in the ESI†). Dip-coating deposition (withdrawal speed of 1.0 cm min⁻¹) was used to perform the solution deposition on the TCO substrate. After deposition, the substrate was dried on a hot plate and then sintered at several temperatures ranging from 600–1000 °C during one hour. The sintering was performed with an oxygen atmosphere flow.

Characterization

The film morphology and thickness (by cross-sectional analysis of the cleaved sample) were characterized by field emission scanning electron microscopy (FE-SEM-FEI Inspec F-50). The crystalline phases were identified by X-ray diffraction (XRD) (Rigaku D-Max 200 using CuKα radiation). For transmission electron microscopy (TEM) and X-ray energy dispersive analysis coupled with the TEM (TEM/EDX) analysis, a TECNAI F20 FEI microscopy was used which operated at 200 kV.

Ultraviolet visible (UV-Vis) absorption spectra of the films were obtained using a Cary 5E UV-Vis spectrophotometer. Photoelectrochemical measurements were carried out in a standard three-electrode cell using the Nb-doped and undoped SrTiO₃ films as the working electrode (0.19 cm²), Ag/AgCl in a KCl saturated solution as the reference electrode and platinum wire as a counter electrode. A 1.0 M NaOH (Merck) pro-analysis in high pure water (pH = 13.6, 25 °C) solution was used as the electrolyte. A scanning potentiostat (Potentiostat/Galvanostat μAutolab III) was used to measure the dark and illuminated currents at a scan rate of 10 mV s⁻¹. Sunlight (1000 W m⁻²) was simulated with a 450 W xenon lamp (Osram, ozone free) and AM1.5 filter. The light intensity was set at 100 mW cm⁻². The IPCE was measured as a function of the excitation wavelength using a 300 Xe lamp (Newport 74125). The IPCE was calculated by considering the following equation:

IPCE = [J_ph/P] × [1240/λ] × 100 (1)

where J_ph is the photocurrent density (μA cm⁻²), P is the incident light power or irradiance (μW cm⁻²) and λ is the wavelength (nm). Absorbed photon-to-current efficiency (APCE) measurements was calculated by dividing IPCE by the UV-Vis absorption spectra.

The Mott–Schottky analysis was obtained by an electrochemical impedance spectroscopy (EIS) measurement using an autolab PGSTAT302N with a three-electrode configuration in 1 M of NaOH solution. Frequencies ranging from 100 KHz to 1 Hz, 10 mV of amplitude potential and a bias voltage of 0.8 to 1.8 V vs. RHE were applied. The EIS was performed in the absence of light, and the Nyquist plot was used to simulate the equivalent circuit to obtain the semiconductor space charge layer capacitance.

Results and discussion

Undoped and Nb-doped SrTiO₃ nanoparticle syntheses were based on a non-hydrolytic procedure developed by Niederberger et al.¹⁸ In a series of experiments, we optimized the synthesis conditions to obtain amorphous nanoparticles with good control over the chemical stoichiometry (SrTi_1−XNb_XO₃, X = 0; 0.05; 0.10 and 0.15) and very good colloidal stability in an organic solvent. XRD analysis of the as-synthesized nanoparticles (see Fig. S2a in the ESI†) revealed the formation of amorphous compounds up to the doping level for X = 0.10. For X = 0.15, we observed the crystallization of several niobate phases in addition to the amorphous phase. Furthermore, the amount of niobium has a direct impact on the colloidal stability. Increasing the doping level produced a decrease in the colloidal stability (see Fig. S1†) which should be related to a decrease in the amount of organic ligand chemically bonded in the nanoparticles surface as represented in the thermogravimetric analysis of the as-prepared doped and undoped material (see Fig. S3 in the ESI†).

After synthesis, the colloidal dispersion was used to process the thin film deposition by dip coating. The stable colloidal suspension was deposited by dip coating which formed a homogeneous and continuous thin film on the TCO substrate. After the deposition step, the amorphous titanate layer is transformed into an SrTiO₃ film through a suitable combination of multi-step deposition and sintering under an oxygen atmospheric flow. During the deposition process, we optimized the Nb concentration and the sintering temperature using the photocurrent density at 1.23 V_RHE as figure of merit. As shown in Fig. 1a, the highest photocurrent density was achieved at X = 0.10. This first optimization facilitated the selection of the SrTi_0.90Nb_0.10O_3−δ stochiometry for further photoelectrochemistry characterization. This optimization was performed at a sintering temperature of 600 °C. After this first set of experiments, we also optimized the sintering temperature, since this parameter has a strong influence on the PEC performance for photoanodes prepared by the colloidal deposition approach.^16,17,19 As illustrated in Fig. 1b, the optimum sintering temperature for the Nb doped SrTiO₃ was 600 °C and the best sintering temperature for the undoped SrTiO₃ was 800 °C. As illustrated in Fig. S2b† (in the ESI), XRD analysis of the sintered films prepared in the optimized doping concentration and temperature revealed a SrTiO₃ single phase. From this point, all structural and photoelectrochemistry characterizations performed in this work will be relative to thin films processed following the optimized conditions described previously; i.e., X = 0.10 and a sintering temperature of 600 °C for the doped film (Nb–STO) and sintering temperature of 800 °C for the undoped film (STO).

Fig. 1 (a) Photocurrent density as function of the Nb concentration (X) for the SrTi_1−XNb_XO₃ thin films sintered at 600 °C; (b) photocurrent density as function of the sintering temperature for the undoped and Nb doped (X = 0.10) SrTi_1−XNb_XO₃.

Lattice parameter measurements for doped and undoped materials, measured by XRD analysis as well as the elemental chemical analysis performed by TEM/EDX are reported in Table 1. These measurements were performed in powder calcined at the same time and temperature used for the thin film preparation. The Nb content and the [Sr]/[Ti + Nb] ratio of the doped and undoped samples are summarized in Table 1; these values are close to the nominal composition. The lattice parameter value obtained for STO is in very good agreement with typical values for undoped SrTiO₃ (3.905 Å) reported in the literature.^20,21 For the doped material, the lattice parameter showed a positive deviation from the undoped material value. This variation is consistent with the crystal ionic radii of Nb⁺⁵ (78 pm) in relation to the Ti⁺⁴ radii (74.5 pm)²² which suggest the formation of solid solution in the Nb–STO system. The substitution of Ti⁺⁴ by Nb⁺⁵ in the B-site of the STO will generate a charge compensation that can be either ionic or electronic.²³ In both charge compensation mechanisms, the incorporation of Nb⁺⁵ as a solute in SrTiO₃ will act as a doping donor which results in an n-type wide band gap semiconductor.

Table 1 Lattice parameter measurements as well as the elemental chemical analysis performed by TEM/EDX for doped and undoped materials

Sample	Nominal [Nb]/[Ti] ratio	Measured [Nb]/[Ti] ratio	Nominal [Sr]/[Nb + Ti] ratio	Measured [Sr]/[Nb + Ti] ratio	Lattice parameter (nm)
STO	0.0	0.0	1.00	1.07 ± 0.09	0.3905
Nb–STO	0.10	0.103 ± 0.016	1.00	1.03 ± 0.11	0.3918

---

The deposition method used in this work produced transparent STO and Nb–STO photoanodes. Fig. 2 and 3 illustrate the cross section and the top-view analysis of the films morphology characterized by FE-SEM. The SrTiO₃ film (see Fig. 2a–c) has a thickness of ∼170 nm with a uniform and continuous morphology. The top view analysis exhibits the nanostructured nature of the film, which is formed by equiaxial grains with a mean grain size of 20 nm (see high magnification FE-SEM image in the Fig. S4 in the ESI†). Fig. 3 displays the Nb–STO film FEM-SEM analysis; this film has a thickness of ∼160 nm and equaxial grains with a mean grain size of 23 nm (see high magnification FE-SEM image in the Fig. S5 in the ESI†). In both samples, a good interface between TCO and STO films is apparent. Note that the doped film has a grain size similar to the undoped film even when sintered at lower temperature which is indirect evidence that the Nb is not segregating during the sintering process. In general, the cation segregation causes a reduction in the grain boundary mobility during the sintering process in ceramic oxides and results in a final microstructure formed by a smaller grain size.²⁴

Fig. 2 FE-SEM analysis of the STO sample: (a) backscattering image of the cross section; (b) top-view secondary electron image; (c) cross section secondary electron image.

Fig. 3 FE-SEM analysis of the Nb–STO sample: (a) backscattering image of the cross section; (b) top-view secondary electron image; (c) cross section secondary electron image.

Fig. 4 displays STO and Nb–STO current potential curves under front-side illumination. As a general trend, the SrTiO₃ photoanode samples under illumination manifest a sharp increase in the photo current around 0.2 V_RHE, which reached a plateau before the onset of the dark current. The STO photoanode displayed a smaller photocurrent density at 1.23 V_RHE and reached a value of 0.04 mA cm⁻². The incorporation of Nb in the SrTiO₃ structure has a significant impact on the photoelectrochemical property of this compound. The Nb–STO photoanode evidenced a photocurrent density of 0.13 mA cm⁻² (at 1.23 V_RHE). The Nb–STO photoanode thin film processed by colloidal deposition had an impressive photocurrent with a performance equivalent to a Nb-doped SrTiO₃ single crystal.¹⁰ IPCE measurements (see Fig. 5) confirm the high efficiency of the Nb–STO photoanode under front illumination; this thin film has IPCE values of 25% around 290 nm. The IPCE measurement confirms that SrTiO₃ films (STO and Nb–STO) possess high activity for water oxidation in a similar spectral window; i.e., between 250 and 375 nm. APCE measurements (see Fig. S6 in the ESI†) also confirm the high efficiency of the Nb–STO photoanode. For instance, this thin film achieved APCE values of 55% around 315 nm. To verify the value of the photocurrent at 1.23 V_RHE for the Nb–STO photoanode, the IPCE as a function of photon wavelength was examined. Integrating the overlap of IPCE data with the standard solar spectrum (AM 1.5/100 mW cm⁻²), gives a calculated value of the photocurrent of 0.122 mA cm⁻² (see Fig. S7 in the ESI†) which is similar to the value measured by the current–potential curve (0.13 mA cm⁻² at 1.23 V_RHE) and confirms that the light source employed simulated the AM 1.5 solar emission.

Fig. 4 Current potential curves under front-side illumination and in dark condition for the STO and Nb–STO samples.

Fig. 5 IPCE measurement for the STO and Nb–STO samples, performed at 1.23 V_RHE.

To obtain a better understanding of the charge transport in STO and Nb–STO photoanodes, a Mott–Schottky analysis was performed to estimate the donor density (N_D) and the flat band potential (V_fb). Based on the depletion layer model, the capacitance of the semiconductor space charge layer (C_SC) is dependent upon the applied potential (V_appl) and can be described by the Mott–Schottky equation:

(1/C)⁻² = (2/ε_rε_oeN_d)(V_appl − V_fb − k_bT/e) (2)

where e is the charge of the electron, ε_r is the semiconductor dielectric constant (200 for the SrTiO₃), ε_o is the vacuum permittivity, T is the absolute temperature, and k_b is the Boltzmann constant. Fig. 6 displays the Mott–Schottky plot for doped and undoped SrTiO₃ thin films recorded in the absence of light. A good linear fit (with a R² > 0.99 for both samples) was obtained in the bias range from 0.8 to 1.6 V vs. RHE with a positive slope, which is typical of an n-type semiconductor. N_D and V_fb values measured for both films are listed in Table 2. Clearly, the effect of the Nb in the N_D value increases the ionized donor density by almost one order of magnitude in relation to the undoped film (STO sample). This result agrees with the XRD data analysis and confirms the incorporation of the Nb in the SrTiO₃ structure acting as a donor doping impurity. V_fb values listed in the Table 2 ranging from −0.07 to −0.19 V_RHE; these values are similar to values reported in the literature.⁹ Considering the onset voltage for a photocurrent (V_on) under standard illumination conditions and the flat band potential (V_fb), we can estimate the η_OX for both film in the range of 0.25 to 0.40 V.

Fig. 6 Mott–Schottky plot for doped and undoped SrTiO₃ thin films recorded in the absence of light.

Table 2 N _D, V_fb, W_O and W_O/G values measured for both films

Sample	N D (cm^−3)	V fb (VRHE)	W O (nm)	W O/G
STO	1.8 × 10 (ref. 19)	−0.07	35.1	1.75
Nb–STO	1.5 × 10 (ref. 20)	−0.19	13.9	0.60

---

To understand the critical parameters that control the η_OX and photocurrent density in SrTiO₃ thin films processed by colloidal dispersion, we performed several photoelectrochemical experiments using electrolytes with the addition of a “hole scavenger”. The use of H₂O₂ as a hole scavenger is well established for the study of overpotential in photoanodes for water splitting;^25,26 however, in the present study, the H₂O₂ produced an intensive corrosion of STO and Nb–STO films during the transient photoelectrochemical experiments. Thus, we decided to use an alternative hole scavenger molecule. Organic compounds are widely used as electron donors for photo-catalytic hydrogen production as they can be oxidized by valence bond holes.²⁷ Among the organic compounds, methanol (CH₃OH) is an interesting molecule because it is transparent to visible and UV light, it does not corrode the SrTiO₃, and it oxidation from the thermodynamics and kinetics point of view is easer than water oxidation by SrTiO₃.²⁸ Therefore, we used methanol as the hole scavenger probe molecule in this study. In this experiment, a 1 M NaOH + 0.5 M CH₃OH solution was used as the electrolyte. Fig. 7 illustrates chopped current potential curves of Nb–STO in different electrolytes. The introduction of methanol in the NaOH electrolyte solution promotes a small modification in the current potential curve with a small cathodic shift and the elimination of spikes at low potential (for a potential smaller then 0.5 V_RHE). The elimination of spikes can be better observed in photocurrent transient experiments illustrated in Fig. S8† (see ESI). This experiment confirms that, even without the methanol, spikes are quite small; after the addition of methanol, spikes are eliminated. STO samples produced similar results. Spikes at a low potential are associated with charge accumulation (mainly holes) at the semiconductor–liquid junction (SCLJ).²⁶ Based on the research of Dotan et al.,²⁵ the catalytic efficiency for water oxidation (η_Cat) can be defined as η_Cat = J_H2O/J_H2O+MeOH where J_H2O is the photocurrent measured in 1 M NaOH water solution, and J_H2O+MeOH is the photocurrent measured in 1 M NaOH + 0.5 M CH₃OH water solution. We observed that the catalytic efficiency is near zero up to the photocurrent onset potential and rises abruptly up to values of η_Cat higher than 90%. The η_Cat reported here for the Nb–STO is very high compared to the values reported for hematite, which is a semiconductor photoanode where the charge (hole) accumulation at SCLJ is very relevant for the water splitting performance.²⁵ These experiments suggest that the charge accumulation in the SCLJ is not a relevant issue for water splitting performance in the SrTiO₃ photoanode and should not be considered as an important source for the photoelectrochemical pseudo-overpotential. This is an interesting results, indicating the SrTiO₃ as a promising catalytic or trap passivation layer material to be used in others n-type semiconductors oxide's in order to improve it performance for water oxidation.

Fig. 7 Chopped current potential curves of Nb–STO in different electrolytes (with and without methanol).

The formulation developed by Dotan et al.²⁵ quantified the carrier separation efficiency. In their work, the total photocurrent density (J_ph) is expressed as:

J_ph = J_abs × η_sep × η_Cat, (3)

where J_abs is the photoabsorption rate expressed as current density, and η_sep is the carrier-separation efficiency. Using the light absorption measurement (see Fig. S9 in the ESI†) and the AM1.5G solar spectrum, J_abs was estimated as 0.22 mA cm⁻². Defining the η_sep as η_sep = J_H2O+MeOH/J_abs in this study, we calculated the carrier-separation efficiency for STO and Nb–STO samples (see Fig. 8). As expected, the carrier-separation efficiency for Nb–STO is higher than the carrier-separation efficiency of the STO sample. The Nb–STO sample reached a carrier-separation efficiency of 60% which indicates that the incorporation of Nb in the SrTiO₃ lattice decreased the recombination rate.

Fig. 8 Carrier-separation efficiency as a function of the applied potential for STO and Nb–STO samples.

For a wide band gap semiconductor, where the hole diffusion length (L_P) is much shorter than the depletion layer width (W_SC) (L_P ≪ W_SC), the photocurrent is primarily due to the carriers generated in the depletion layer which indicates that enhancing W_SC should improve the charge separation efficiency.⁸ The incorporation of a donor doping impurity (such as Nb in SrTiO₃), increase the N_D and the donor density has a direct impact on the W_SC formed at the semiconductor electrode–electrolyte interface. According to eqn (4), increasing N_D produces a decreasing in the W_SC.

W_SC = W_O(V_SC)^1/2 (4)

where W_O = (2ε_rε_o/eN_D)^1/2, and V_SC is the potential at the surface space charge layer. Table 2 contains W_O values calculated for Nb–STO and STO samples with an expressive difference; the STO sample has a W_O value higher then twice the value reported for Nb–STO.

To better understand the Nb doping effect in the photocurrent, we modified the formulation develop by Dotan et al.,²⁵ combining with the formulation proposed by Warren et al.²⁹ Writing the η_sep as η_sep = f(ν_active/ν_total), where ν_active is the active volume (ν_active = W_SCA, where A is the photoanode area), ν_total is the material volume and ƒ is the solid volume fraction and considering the planar configuration for the film, as described in the Scheme 1, we can write:

J_ph = J_absη_Cat(W_O/nG)f(V_appl − V_fb − V_GB)^1/2 (5)

where n is the number of grains in a series, G in the titanate grain size, V_appl is the applied potential, and V_GB is the grain boundary potential (V_SC = V_appl − V_fb − V_GB). As proposed by Warren,²⁹ we included the influence of the back-to-back Schottky barrier at the grain boundary in the J_ph. These barriers modify the overpotential and the photocurrent density because the V_GB decreases V_SC, which decreases W_SC and consequently results in a smaller J_ph.

Scheme 1 Schematic representation of the STO (undoped SrTiO₃) and Nb–STO (Nb–SrTiO₃) samples, taking into account the electronics parameters and morphological features.

The eqn (5) shows a dependence of the photocurrent density with W_O and the smallest feature size of the semiconductor microstructure (in this study, the titanate grain size – G). We are postulating that the ratio W_O/G is an important parameter to consider in PEC performance for a polycrystalline and nanometric photoanode.

If W_O/G < 1, the total photocurrent density (J_ph) will be controlled by V_SC, with a direct impact of the V_GB. The Nb incorporation in the SrTiO₃ lattice leads to an increase of N_D, resulting in a smaller W_O. Since the G < W_O, it is possible to increase the V_SC up to W_SC reach a magnitude close to the grain size (G). As consequence, a better charge separation can be obtained. Therefore, in the Nb–STO sample (where W_O/G < 1), the V_SC will control the charge-separation process.

When W_O/G > 1, the grain size becomes fully depleted, and eqn (5) is not valid. Here we can postulate two hypotheses to explain the poor charge separation efficiency when W_O/G > 1. The first hypothesis is the absence of band bending due to the nanosized semiconductor aspect;^30,31 the second hypothesis is based on potential barriers (such as the back-to-back Schottky barrier) in the semiconductor grain boundary (solid–solid interface). These barriers can provoke an increase in the overpotential as described before. Besides the overlap of the depletion layer W_SC (formed in the solid–liquid interface) with the potential barrier at grain boundary (formed at solid/solid interface) must increase the recombination rate. In this hypothesis, the charge separation efficiency is dominated by recombination in the depletion layer (see Scheme 1). As displayed in Table 2, the STO sample confirms that W_O/G > 1. Since this sample was described well by the Mott–Schottky model (see Fig. 6), the absence of band bending is not a plausible explanation for poor charge separation efficiency. On the other hand, the formation of potential barriers in the polycrystalline SrTiO₃ grain boundary is well documented.^32,33 Thus, we can assume that in the undoped sample, the charge separation efficiency is dominated by recombination at the depletion layer.

The influence of the electronic (W_O, N_D, etc.) and morphologic parameters in the SrTiO₃ photoanode performance is summarized in Scheme 1. Based on this model, the W_O/G ratio is an important parameter to control the photocurrent. In a photoanode where the charge (holes) accumulation in the SCLJ is not relevant, we can postulate that for a W_O/G < 1, the V_SC will control the photocurrent density. On other hand, for W_O/G > 1, the recombination at the depletion layer will control the photocurrent density. Actually, note that we are introducing a new interpretation for the use of the donor doping impurities; i.e., N_D can be used to fit the electronic parameter (W_O) with the smallest feature size of the semiconductor microstructure. The results discussed and reported in this study for the n-type SrTiO₃ semiconductor can be extended to other metal oxide photoanodes and will contribute to the development of better materials for PEC devices.

Conclusion

The colloidal synthetic route as well as the colloidal deposition process reported in this study is appropriate to process undoped and Nb-doped SrTiO₃ photoanode thin films for water splitting with very good carrier-separation efficiency under standard solar irradiation. Methanol was used as the hole scavenger, and the transient photoelectrochemical experiments suggested that the charge (holes) accumulation in the SCLJ is not a important mechanism to control the photocurrent density and overpotential. In addition, the W_O/G ratio was used as a parameter to predict the dominance of photocurrent density controlled by V_SC or by recombination at the depletion layer.

Acknowledgements

The financial support of FAPESP (projects CEPID - 2013/07296-2), FINEP, CNPq (INCT program) and CAPES (all Brazilian agencies) is gratefully acknowledged. We would also like to thank researcher Ricardo H. Gonçalves for his helpful discussion and suggestions.

Notes and references

1. A. Fujishima and K. Honda, Nature, 1972, 238, 37.
2. K. Sivula, F. LeFormal and M. Graetzel, ChemSusChem, 2011, 4, 432–449.
3. K. Sivula, J. Phys. Chem. Lett., 2013, 10, 1624–1633.
4. R. Solarska, A. Krolikowska and J. Augustynski, Angew. Chem., 2010, 122, 8152–8155 ( Angew. Chem. Int. Ed. , 2010 , 49 , 7980–7983 ).
5. K. Sayama, A. Nomura, Z. Zou, R. Abe, Y. Abe and H. Arakawa, Chem. Commun., 2003, 2908–2909.
6. M. S. Wrighton, A. B. Ellis, P. T. Wolczanski, D. L. Morse, H. B. Abrahamson and D. S. Ginley, J. Am. Chem. Soc., 1976, 98(10), 2774–2779.
7. J. G. Mavroides, J. A. Kafalas and D. F. Kolesar, Appl. Phys. Lett., 1976, 28, 241–243.
8. A. J. Bard, J. Photochem., 1979, 10, 59–75.
9. R. K. Quinn, R. D. Nasby and R. J. Baughman, Mater. Res. Bull., 1976, 11, 1011–1018.
10. J. Yin, J. Ye and Z. Zou, Appl. Phys. Lett., 2004, 84, 689–691.
11. T. K. Townsend, N. D. Browning and F. E. Osterloh, ACS Nano, 2012, 8, 7420–7426.
12. T. K. Townsend, N. D. Browning and F. E. Osterloh, Energy Environ. Sci., 2012, 5, 9543–9550.
13. F. Wang, Y. Liu, Z. Ma, H. Li, Z. Kang and M. Shen, New J. Chem., 2013, 37, 290–294.
14. S. Kawasaki, N. Nakatsuji, J. Yoshinobu, F. Komori, R. Takahashi, M. Lippmaa, K. Mase and A. Kudo, Appl. Phys. Lett., 2012, 101, 033910.
15. K. Iwashina and A. Kudo, J. Am. Chem. Soc., 2011, 113, 13272–13275.
16. R. H. Gonçalves, B. H. R. Lima and E. R. Leite, J. Am. Chem. Soc., 2011, 133, 6012–6019.
17. R. H. Gonçalves, L. D. T. Leite and E. R. Leite, ChemSusChem, 2012, 5, 2341–2347.
18. M. Niederberger, G. Garnweitner, N. Pinna and M. Antonietti, J. Am. Chem. Soc., 2004, 126, 9120–9126.
19. K. Sivula, R. Zboril, F. Le Formal, R. Robert, A. Weidenkaff, J. Tucek, J. Frydrych and M. Grätzel, J. Am. Chem. Soc., 2010, 132, 7436–7444.
20. JCPDS Powder Diffraction File (Card 35–0734, JCPDS International Centre for Diffraction Data, Swarthmore, 1986).
21. P. Blennow, A. Hagen, K. K. Hansen, L. R. Wallenberg and M. Mogensen, Solid State Ionics, 2008, 179, 2047–2058.
22. R. D. Shannon, Acta Crystallogr., Sect. A: Found. Crystallogr., 1976, 32, 751–767.
23. Y.-M. Chiang, D. Birnie III and W. D. Kingery, Physical Ceramics – Principles for Ceramics Science and Engineering, Jonh Wiley, New York, 1997.
24. E. R. Leite, A. P. Maciel, I. T. Weber, P. N. Lisboa-Filho, E. Longo, C. O. Paiva-Santos, A. V. C. Andrade, C. A. Pakoscimas, Y. Maniette and W. H. Schreiner, Adv. Mater., 2002, 14, 905–908.
25. H. Dotan, K. Sivula, M. Grätzel, A. Rothschild and S. C. Warren, Energy Environ. Sci., 2011, 4, 958–964.
26. F. Le Formal, K. Sivula and M. Grätzel, J. Phys. Chem. C, 2012, 116, 26707–26720.
27. M. Ni, M. K. H. Leung, D. Y. C. Leung and K. Sumathy, Renewable Sustainable Energy Rev., 2007, 11, 401–425.
28. Y. Kuo and K. J. Klabunde, Nanotechnology, 2012, 23, 294001.
29. S. C. Warren, K. Voïtchovsky, H. Dotan, C. M. Leroy, M. Cornuz, F. Stellacci, C. Hérbert, A. Rothschild and M. Grätzel, Nat. Mater., 2013, 12, 842–849.
30. A. Goossens, J. Electrochem. Soc., 1996, 143, L131–L133.
31. R. van de Krol, Principles of Photoelectrochemical Cells, in Photoelectrochemical Hydrogen Production, ed. R. van de Krol and M. Grätzel, Springer, New York, 2012.
32. S. V. Kalinin and D. A. Bonnell, Phys. Rev. B: Condens. Matter Mater. Phys., 2004, 70, 235304.
33. P. Prabhamirashi, V. P. Dravid, A. R. Luoini, M. F. Chisholm and S. J. Pennycook, Appl. Phys. Lett., 2005, 87, 121917.

---

Footnote

† Electronic supplementary information (ESI) available: Additional colloidal stability, XRD, TGA, UV-Vis spectroscopy and photoelectrochemical experiments. See DOI: 10.1039/c3ra45066j

---