- Email: [email protected]
Gradient Ti-doping in hematite photoanodes for enhanced photoelectrochemical performance
Journal of Power Sources xxx (xxxx) xxx
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Gradient Ti-doping in hematite photoanodes for enhanced photoelectrochemical performance Fan Feng a, Can Li a, Jie Jian b, Fan Li b, Youxun Xu b, Hongqiang Wang b, **, Lichao Jia a, * a
Key Laboratory of Applied Surface and Colloid Chemistry, National Ministry of Education, Shaanxi Key Laboratory for Advanced Energy Devices, Shaanxi Engineering Lab for Advanced Energy Technology, School of Materials Science and Engineering, Shaanxi Normal University, 620 West Chang’an Street, Xi’an, Shaanxi, 710119, China b State Key Laboratory of Solidification Processing, Center for Nano Energy Materials, School of Materials Science and Engineering, Northwestern Polytechnical University and Shaanxi Joint Labortary of Graphene, Xi’an, 710072, PR China
H I G H L I G H T S
G R A P H I C A L A B S T R A C T
� Gradient Ti doping is affected by tem perature rather than the location of Ti. � Fe2O3–Ti/Ti–Fe2O3 exhibit almost similar photocurrent of 1.50 mA cm 2 at 1.23 V. � The ηsurface of Fe2O3–Ti/Ti–Fe2O3 can reach to 96% at 1.30–1.50 V vs. RHE.
A R T I C L E I N F O
A B S T R A C T
Keywords: Charge separation Gradient doping Photoanode Fe2O3 Water splitting
Hematite based photoanode is promising for solar water splitting while suffers from poor charge transport and separation efficiency that limit its practical application. Herein, we demonstrate two types gradient Ti-doped Fe2O3 photoanodes (Fe2O3–Ti (TiO2 deposited on the top of Fe2O3) and Ti–Fe2O3 (Fe2O3 on the top of TiO2)) to enhance charge transport and separation efficiency. Interestingly, Fe2O3–Ti and Ti–Fe2O3 photoanodes exhibit almost identical PEC performance with photocurrent of 1.50 mA cm 2 at 1.23 V vs. the reversible hydrogen electrode (RHE). However, the onset potential of Ti–Fe2O3 photoanode displays a cathodic shift of 80 mV comparing with that of Fe2O3–Ti photoanode. Moreover, the charge separation efficiencies on the surface (ηsurface ) for Fe2O3–Ti and Ti–Fe2O3 can reach up to 96% at the higher potential range from 1.30 to 1.50 V vs. RHE, which are among the top values in the record of hematite-based photoanodes without co-cocatalyst. Further investigation demonstrates the forming of gradient Ti doping is not dependent on the location of the TiO2 layer, but mainly affected by the high annealing temperature. The enlarged contact area between hematite photoanode and the electrolyte, the improved charge separation efficiency, and increased charge carrier density are responsible for the enhanced PEC water splitting.
* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (H. Wang), [email protected] (L. Jia). https://doi.org/10.1016/j.jpowsour.2019.227473 Received 16 September 2019; Received in revised form 2 November 2019; Accepted 14 November 2019 0378-7753/© 2019 Elsevier B.V. All rights reserved.
Please cite this article as: Fan Feng, Journal of Power Sources, https://doi.org/10.1016/j.jpowsour.2019.227473
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
1. Introduction
photoanodes shows a significant increase to more than 96% at higher potential range from 1.30 to 1.50 V vs. RHE, which are among the top values in the record of hematite-based photoanodes without cococatalyst.
Photoelectrochemical (PEC) water splitting is considered as a promising approach to utilize solar energy for hydrogen generation [1–3]. In recent decades, various n-type semiconductors have been extensively explored as photoanode materials for PEC water oxidation [4–9]. Among a large selection of photocatalysts for solar water split ting, hematite (α-Fe2O3) has drawn great research interests owing to its suitable bandgap, ample availability, stability in a broad pH range, nontoxicity and low cost [10,11]. With a bandgap width of 1.9–2.2 eV (depending on the method of preparation), hematite would absorb suf ficient visible light and hold the potential to reach a solar to hydrogen efficiency up to 16.8% and a maximum photocurrent of 12.6 mA cm 2 under AM 1.5G illumination (100 mW cm 2 light intensity) based on the theoretical calculation [12,13]. Although these remarkable merits, the reported efficiencies of hematite are far from the satisfactory due to the sluggish charge carrier separation efficiency, which is mainly resulted from the physical and chemical nature of α-Fe2O3 including the extremely short hole-diffusion length (2–4 nm), low minority carriers mobility (0.2 cm2 V 1 s 1) and poor photogenerated charge carriers lifetime (<10 ps) [14–16]. Additionally, the slow kinetics of the oxygen evolution reaction is another obstacle to restrict the PEC performance [17]. To improve the PEC performance of Fe2O3 photoanodes, element doping, such as Si4þ [18], Zr4þ [19], Sn4þ [20], Ti4þ [21] etc., has been extensively employed to address these issues. Comparing with the normal homogenous doping, the gradient-doping could not only induce a more upward band bending within a very wide region, but also could substantially increase the electron density of hematite to improve the conductivity and charge separation efficiency of bulk Fe2O3 [22], thus received great attentions. For instance, Gong and co-workers described that gradient doping of phosphorus can increase the width of band bending in Fe2O3 photoanode inducing promoted charge separation efficiency in the bulk [22]. Schmuki and co-workers demonstrated that gradient Sn4þ doping in Fe2O3 could result in a decrease of the onset potential and an increase in the photocurrent density due to the decreased charge recombination and the high conductivity [23]. Particularly, the radius of Ti4þ (0.61 Å) is smaller than that of high spin Fe3þ (0.65 Å), which suggests that Ti incorporation would cause lattice contraction and a reduction in the unit cell volume, leading to the enhanced hopping probability of charge carriers [17]. Hence, Ti stands out as a promising candidate for hematite doping. Until now, most of research works have been mainly focused on the TiO2 induced Fe2TiO5 layer or the surface TiO2 passivation layer on the Fe2O3 [24–26]. Even the photocurrent density can be improved but the surface separation efficiency is still quite low. Recently, Dass and co-workers displayed gradient-Ti-doped hematite photoanodes used a multi-layer spray py rolysis method for the first time and found that a gradient concentration of Ti dopant in a Fe2O3 resulted in an enhanced photoelectrochemical response attributing to the values of high flat band potential, and high conductivity as well as low carriers recombination rate [27]. While the gradient Ti-doping in hematite has displayed preliminary success, seeking facile method, clarifying the formation mechanism while improving the surface separation efficiency is still a highly desirable challenge. Herein, we demonstrate a simple and facile method for synthesizing gradient-Ti-doped hematite photoanode by magnetron sputtering deposition of TiO2 and subsequent hydrothermal treatment. Further investigation reveals that an identical gradient Ti doping decreasing from top to bottom can be achieved whether the TiO2 located at the surface of Fe2O3 (Fe2O3–Ti) or in between Fe2O3 and FTO (Ti–Fe2O3). Both Fe2O3–Ti and Ti–Fe2O3 photoanodes exhibit pronounced PEC performance with almost identical photocurrent of 1.50 mA cm 2 at 1.23 V vs. RHE, while the onset potential of Ti–Fe2O3 photoanode demonstrates a negative shift of 80 mV comparing with that of Fe2O3–Ti photoanode. Additionally, the ηsurface of the Fe2O3–Ti and Ti–Fe2O3
2. Experimental section 2.1. Materials Ferric chloride (FeCl3, 97%) was obtained from Sigma-Aldrich Co. LLC. Sodium nitrate (NaNO3, 99%), acetone (C3H6O, 99.5%) and hy drochloric acid (HCl, 36–38 wt%) were purchased from Sinopharm Chemical Reagent Co., Ltd. 1 M KOH standard aqueous solution (pH ¼ 13.6) was provided by Alfa Aesar Co. LLC. Ethanol (C2H5OH, 99.7%) was supplied by Tianjin Guangfu Fine Chemical Research Institute. Deionized water with a resistance of 18.2 MΩ cm was produced by a Water Purification System (Merck Millipore, Direct-Q5 UV) and used for all reactions. Fluorine-doped tin oxide (FTO, 14 Ω, 2.2 mm) glass sub strates were purchased from Nippon Sheet Glass Co., Ltd. Before sample preparation, all FTO substrates were ultrasonically pre-cleaned in ethanol, acetone, ethanol and deionized water for 30 min sequentially, then treated by a plasma cleaner (Diener, ZEPTO) at 50 W in air for 10 min to increase the surface hydrophilicity. 2.2. Preparation of the Fe2O3 photoanodes The Fe2O3 photoanodes were synthesized through a versatile hy drothermal method as reported previously [26]. Briefly, 0.1 M FeCl3 and 1 M NaNO3 were added into a Teflon heating-clave, then proper quantity of HCl was used to adjust pH to 1.5. Afterwards, the heating-clave was transferred into an oven and maintained at 95 � C for 12 h, a yellow FeOOH film was deposited onto the FTO substrates. Next, the samples were rinsed in deionized water for several times and dried by N2 flow. After annealing at 550 � C for 2 h and at 800 � C for 10 min, the Fe2O3 nanorod-array photoanodes can be obtained. 2.3. Preparation of the Fe2O3–Ti and Ti–Fe2O3 film TiO2 deposition was performed at room temperature by a radio fre quency (RF) magnetron sputtering system using a pure TiO2 as sputter target. During the deposition, the sputtering power was fixed at 80 W and the basic pressure was maintained at 5 � 10 3 Pa with the steady Ar flow of 18 sccm. The deposition time was varied from 10 to 60 min. For the sample of Fe2O3–Ti, the TiO2 was deposited on the top of as-prepared Fe2O3 nanorod array and then annealed at 800 � C for 10 min. For the sample of Ti–Fe2O3, the TiO2 was deposited on FTO substrate firstly and then Fe2O3 nanorod array was synthesized on the top of TiO2 layer ac cording to the above hydrothermal method, after the finally annealing process at 550 � C for 2 h and 800 � C for 10 min, the prepared sample can be obtained. 2.4. Films characterization X-ray diffraction analysis (XRD) was used to investigate the crystal structures and composition of the as-synthesized samples over a 2θ range of 20� ~ 70� with a Cu Kα radiation (λ ¼ 1.54056 Å) at 45 kV and 200 mA equipped. The field emission scanning electron microscope (FESEM, SU8220, Japan, 3 kV) and transmission electron microscopy (TEM, JEOL JEM-2800, 200 kV) were performed to clarify the detailed morphology and microstructure of the prepared samples. The surface chemical state was evaluated by the X-ray photoelectron spectroscopy (XPS, Axis Ultra, Kratos Analytical Ltd). The binding energy was cali brated using the C 1s photoelectron peak at 284.6 eV. UV–vis absorption spectra were measured by a UV–vis spectrophotometer (UV-3600, Kra tos Analytical Ltd.). Raman spectra were recorded on a Renishaw via micro-Raman spectroscope with 532 nm laser. 2
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
2.5. PEC measurements
further annealing at 800 � C is beneficial to improve the crystallinity [30]. After the introduction of element Ti into the Fe2O3, the hexagonal structure hematite with obvious (110) peak still can be confirmed for Fe2O3–Ti and Ti–Fe2O3 samples as exhibited in Fig. 1a. When enlarging the diffraction peak of (110), a small high angle shift can be observed for the Fe2O3–Ti comparing with pure Fe2O3, as demonstrated in Fig. 1b, manifesting the unchanged orient growth direction and incorporation of Ti in the crystal lattice of Fe2O3 [31]. Interestingly, the situation is totally different for the Ti–Fe2O3, no any angle shift can be observed. Instead, the peak intensity of (300) decreases significantly and the new (104), (012) and (024) peaks appear, which would be resulted from the formation of multiple Fe2O3 crystal planes due to the incorporation of TiO2 [20]. Moreover, no other distinct peaks correlated with TiO2 or Ti-based compounds can be found from the Fe2O3–Ti and Ti–Fe2O3 samples, which could be ascribed to the relatively low concentration of Ti [29]. In order to confirm the incorporation of Ti in the hematite, Raman spectrum was explored as presented in Fig. 1c. Evidently, six intrinsic Raman peaks (224, 245, 292, 410, 499 and 613 cm 1) belonging to Fe2O3 can be observed from Fe2O3–Ti and Ti–Fe2O3 samples, which is in agreement well with reported results in the literature [29]. After careful examination, the pronounced Raman peak locating around 660 cm 1 in Fe2O3–Ti and Ti–Fe2O3 can be seen, which should derive from the disordered Fe2O3 phase induced by the Ti [32,33]. From the UV–vis absorbance spectra of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 films demonstrated in Fig. S2a, similar light absorption behavior with the absorption edge around 600 nm can be clearly seen, which corresponds to the bandgap of 2.1 eV, and is consistent well with the previously reports [12,34]. Further, it should be pointed that the introduction of Ti has no obvious effects on the absorption of Fe2O3. The XPS was also employed to investigate the surface chemical state of pure Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes. Distinct Fe, Ti, O and Sn signals can be observed from XPS full scan spectrum for all the samples (Fig. S2b). Here, the Sn signal could be ascribed to the selfdoping from fluorine-doped tin oxide (FTO) due to the high
The PEC properties were performed using a typical three-electrode configuration with prepared film as photoanodes, an Ag/AgCl (satu rated KCl) as reference electrode, and a Pt wire as counter electrode (CH Instrument, CHI 760E). 1 M KOH standard aqueous solution of PH 13.6 by N2 purging for 30 min was employed as electrolyte. 500 W xenon lamp (Beijing Perfectlight Technology Co. Ltd., CEL-S500) equipped with an AM 1.5G filter with a power density of 100 mW cm 2 was used as the simulated light source. Photocurrent vs. voltage (J-V) curves of samples were measured with a scan rate of 10 mV s 1 from cathodic to anodic potentials with the photoanode exposing area of 0.2 cm2. All potentials were converted to the reversible hydrogen electrode (RHE) potential [28]: ERHE ¼ EAg=AgCl þ 0:1976 þ 0:0591 � pH The incident photon-to-current conversion efficiency (IPCE) was investigated under a 300 W Xe light (CEL-HXF300) in combination with a grating monochromator (CEL-IS151). The electrochemical impedance spectra (EIS) measurements were carried out at frequency range of 1 MHz–10 Hz under 1.23 V vs. RHE. Mott-Schottky (M S) plots were measured at a frequency of 1000 Hz under dark condition. The elec trochemical active surface area (EASA) was calculated based on the electrochemical double-layer capacitance as presented in previously reports [29]. 3. Results and discussion The crystal structure and composition of as-prepared samples were determined by the X-ray diffraction (XRD) shown in Figs. S1 and 1a. Clearly, after the high temperature annealing at 550 � C and 800 � C, the FeOOH peaks disappear and Fe2O3 signals emerge. The pronounced (110) diffraction peak suggests that the Fe2O3 is highly oriented along the (110) direction perpendicular to the FTO substrate [26]. Addition ally, the much stronger diffraction peaks of (110) and (300) indicate that
Fig. 1. (a) XRD pattern, (b) Enlarged (110) peak of XRD pattern and (c) Raman spectra of the pristine TiO2, pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples, respectively. XPS (d) Fe 2p, (e) O 1s, and (f) Ti 2p spectrum collected from the pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples, respectively. 3
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
temperature treatment at 800 � C [21,35]. Fe 2p3/2 and Fe 2p1/2 peaks located at 710.5 eV and 723.9 eV (Fig. 1d) are associated with Fe3þ in Fe2O3, consisting well with previously reported value [36,37]. Compared with pure Fe2O3, the Fe 2p peak of the Fe2O3–Ti and Ti–Fe2O3 exhibits higher energy shift of around 0.4 eV, which could be correlated with the incorporation of Ti4þ ions in the Fe2O3 crystal lattice [38]. The O 1s XPS spectrum (Fig. 1e) can be splitted into two peaks at 529.5 eV and 530.8 eV for Fe2O3 sample, corresponding to lattice oxygen and surface hydroxyl oxygen respectively [39]. The different degree of high energy shifts for Fe2O3–Ti and Ti–Fe2O3 samples may suggest varying doping quantity of Ti4þ into the Fe2O3 lattice. In Fig. 1f, a similar small negative banding energy shift for Ti 2p3/2 and Ti 2p1/2 peaks can be easily distinguished from Fe2O3–Ti and Ti–Fe2O3, comparing with the pure TiO2. This result manifests that chemical environment of the Ti for both two films is almost the same and Ti4þ ions were inserted into the Fe2O3 matrix [26]. To examine the morphologies and microstructures of the as-prepared pristine Fe2O3, Fe2O3–Ti and Ti– Fe2O3 films, field emission scanning electron microscopy (FE-SEM) measurements were carried out as demonstrated in Fig. 2. As shown in Fig. 2a-c, all of the films are composed of uniform and homogeneous wormlike particles. After the incorporation of Ti in the Fe2O3, the average particle size exhibits an evidently decrease from 72 nm to 35 and 60 nm, and the Fe2O3–Ti contains the smallest particle size with around 35 nm (see Fig. S3). Meanwhile, the Ti–Fe2O3 film demonstrates the better dispersity and lower particle density when comparing with Fe2O3–Ti film. Further, from the cross-section displayed in Fig. 2d-f, one can confirm that Ti incorporation has significantly influence on the morphology and the average diameter of the Fe2O3–Ti nanorod (56 nm) is smaller than that of Ti–Fe2O3 nanorod (76 nm). But the length of nanorod arrays does not show big difference and all films possess similar thickness of about 340 nm. This can be confirmed by the corresponding TEM images shown in Fig. 3a and 3b. Moreover, the crystal lattice with interplanar spacing of 0.25 nm observed from the HRTEM in Fig. 3c and 3d could be well assigned to (110) plane of Fe2O3 (JCPDS card Number 89–0597) for both samples. To figure out the reason for the reduced particle size comparing with pure Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples without annealing at 800 � C were also checked (see Figs. S4a–c). Visibly, the particle size has almost no change, while the dispersity of Ti–Fe2O3 is much higher than that of pure Fe2O3. Interestingly, for Fe2O3–Ti, that is totally similar with the
sample of pure Fe2O3 treated at 550 � C followed by Ar plasma treatment for 30 min (see Fig. S4d). As shown in Figs. S4e and S4f, after further annealing at 800 � C, the particle size of the plasma treated Fe2O3 sample is slightly increased, but much smaller than that of pure one. Combining the results of Fe2O3–Ti and Ti–Fe2O3 samples under high temperature treatment at 800 � C (see Fig. 2b and 2c), the smaller particle size of Fe2O3–Ti sample could be ascribed to the synergistic effect of plasma treatment during the deposition and the Ti-based growth restriction on the particle surface. The grain size reduction of Ti–Fe2O3 sample could only result from the Ti-based reason. To further evaluate the Ti element distribution, X-ray photoelectron spectroscopy (XPS) depth analysis in Fig. 3e and 3f were carried out via employing argon-ion etching. Surprisingly, the Ti concentration de creases from top of the samples to the bottom with the increase of argonion etching time, presenting a gradient profile in both Fe2O3–Ti and Ti–Fe2O3. Additionally, the reducing of Ti concentration in the Fe2O3–Ti is much faster than that of Ti–Fe2O3 sample. In order to clarify the originate of the Ti gradient distribution for Ti–Fe2O3 sample, EDS elemental mapping of Ti–FeOOH, Ti–Fe2O3 without annealing at 800 � C, Ti–Fe2O3 and Fe2O3–Ti samples were performed as shown in Fig. S5. Clearly, less quantity of Ti in Ti–FeOOH and Ti–Fe2O3 sample without annealing at 800 � C can be distinguished on the surface (Figs. S5d and S5h), which may be resulted from the partially dissolution of the sput tered TiO2 in the acid hydrothermal solution with 1.5 pH, leading to uniform doping of Ti4þ in Fe2O3. Fig. S5i-S5p display uniform and dense distribution of Fe, O and Ti elements in Ti–Fe2O3 and Fe2O3–Ti samples along the nanoparticles on the surface. Further, it should be pointed that the calculated Ti concentration is around 0.05 atom % in Ti–Fe2O3 sample without annealing at 800 � C, which is only one eighth of pre pared Ti–Fe2O3 samples (0.40 atom %). This can also be verified by the XPS results of Fe2O3–Ti and Ti–Fe2O3 samples after annealing at 550 � C (Fig. S6). Combining with the above SEM, TEM and XPS results, the gradient Ti doping obtained here could be ascribed to the high tem perature annealing at 800 � C, which could cause the melt and recrys tallization of the oxide [29]. As we know, the mass density of TiO2 (3.90 g cm 3) is smaller than that of Fe2O3 (5.26 g cm 3) [40], that means the TiO2 could very easily locate near surface during the melting process and form a gradient doping from top to bottom during rapid recrystallization. In order to evaluate the influence of Ti incorporation in the Fe2O3 photoanodes, PEC measurements were employed in 1 M KOH electrolyte
Fig. 2. (a, b, c) SEM top view, and (d, e, f) the corresponding cross section images of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 thin films. 4
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
Fig. 3. (a, b) TEM images, (c, d) HRTEM images, and (e, f) XPS depth profile signal of Ti 2p in Fe2O3–Ti and Ti–Fe2O3 samples, respectively.
Fig. 4. (a) Chopped photocurrent density-potential (J–V) characteristics measured in 1 M KOH solution under AM 1.5 G illumination (100 mW cm 2), (b) Transient photocurrent measurements at 1.23 V vs. RHE, (c) IPCE spectra collected at the incident wavelength range from 340 to 660 nm at 1.23 V vs. RHE, and (d) ABPE spectra of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes, respectively. 5
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
using a three-electrode electrochemical cell under simulated AM 1.5G irradiation. It is noticeable that the pristine Fe2O3 photoanode treated at 800 � C was used as control sample. The J-V curves of the Ti–Fe2O3 and Fe2O3–Ti with different TiO2 sputtering time are displayed in Fig. S7. Obviously, all the samples exhibit the neglected dark current densities from 0.60 to 1.60 V vs. RHE. The PEC performance is significantly enhanced comparing with the pure hematite while changes with the TiO2 deposition time. The Fe2O3 with 30 min and 40 min TiO2 treatment demonstrate the best photocurrent density value for the Fe2O3–Ti and Ti–Fe2O3 photoanodes at 1.23 V vs. RHE, respectively. From the SEM images shown in Fig. S8 and Fig. S9, the smaller particle size can be confirmed for both two samples, which could result in more surfaceactive site and interfacial contact area between photoanode materials and electrolyte, inducing the improved PEC performance [29]. Further, when comparing the Fe2O3–Ti with Ti–Fe2O3 samples (as illustrated in Fig. 4a), one can found that both two photoanodes display almost the similar water oxidation behavior and the photocurrent densities could reach to 1.50 mA cm 2 at 1.23 V vs. RHE, which is about two times higher than that of pristine Fe2O3 (0.80 mA cm 2 at 1.23 V vs. RHE). Interestingly, the onset potential of Ti–Fe2O3 photoanode displays a cathodic shift of 80 mV comparing with that of Fe2O3–Ti photoanode. Meanwhile, the photocurrent is significantly larger at low potential range from 0.80 to 1.20 V vs. RHE, which may be resulted by the remained interfacial TiO2, reducing the lattice mismatch between FTO and hematite [32,41]. Combining with the above XPS results, it is obvious that the sputtered TiO2 in Ti–Fe2O3 sample not only can act as a source for Ti4þ gradient doping but also can work as interfacial layer to reduce the interfacial charge recombination, and improve the hematite conductivity (see Fig. 5c) [32]. Moreover, in order to assess the exposing active sites, the electrochemical active surface area (EASA) has been investigated and shown in Fig. S10. Evidently, the linear slopes (see Fig. S10d) are 3.5, 5.5 and 5.4 μF cm 2 for pristine Fe2O3, Fe2O3–Ti and
Ti–Fe2O3 photoanodes, respectively, manifesting both Fe2O3–Ti and Ti–Fe2O3 films can provide almost identical active sites during the photoelectrochemical process. This can be ascribed to the balanced ef fect between dispersity and particle size of Ti–Fe2O3 and Fe2O3–Ti sample as showed in Fig. 2b and 2c. The transient photocurrent measurements of Fe2O3–Ti and Ti–Fe2O3 photoanodes were investigated to estimate the charge recombination behaviors at the semiconductor-electrolyte junction at 1.23 V vs. RHE under chopped light illumination (as illustrated in Fig. 4b). Evidently, the sharp photocurrent spike of pristine hematite is greatly restrained after the incorporation of Ti in Fe2O3, indicating the weaken charge carries recombination during their transport to the surface of the semiconductor. To further investigate the photo-response properties, the incident photon-to-electron conversion efficiency (IPCE) was measured as a function of incident light wavelength in 1 M KOH at 1.23 V vs. RHE, as illustrated in Fig. 4c. Apparently, the IPCE of both Fe2O3–Ti and Ti–Fe2O3 photoanodes demonstrates the significantly enhancement in almost the whole examined wavelength from 340 nm to 660 nm comparing with that of pristine hematite. The best value from Ti–Fe2O3 photoanodes could reach to 39% at 340 nm, which is around 1.25 times higher than that of Fe2O3–Ti (31%). The higher IPCE of Ti–Fe2O3 pho toanode when the wavelength is below 450 nm suggests that the absorbed photons can be utilized more efficiently, which may be owing to the increased hematite conductivity and charge carrier density. Moreover, the applied bias photon-to-current efficiency (ABPE) was also carried out to quantitatively assess the PEC water splitting efficiency of these samples. As presented in Fig. 4d, an optimum photoconversion efficiency of 0.13% can be obtained for the Ti–Fe2O3 photoanode at 1.10 V vs. RHE, which is higher than that of Fe2O3–Ti photoanode (0.10% at 1.10 V vs. RHE) and around two times over comparing with that of pristine Fe2O3 (0.06% at 1.10 V vs. RHE). Clearly, the improved PEC performance is closely related with the
Fig. 5. (a) Surface charge separation efficiency (ηsurface ), (b) Bulk charge transfer efficiency (ηbulk ), (c) Electrochemical impedance spectra (EIS) at 1.23 V vs. RHE, and (d) Mott-Schottky plots measured at 1 kHz frequency of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples, respectively. 6
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
effect of enhanced charge separation caused by the incorporation of Ti in hematite. To verify this, J-V measurements were performed using 0.1 M H2O2 as a hole scavenger in 1 M KOH electrolyte (Fig. S11). The photocurrent density of both Fe2O3–Ti and Ti–Fe2O3 photoanodes can reach up to 1.60 mA cm 2 at 1.23 V vs. RHE, which is significantly higher than that of pristine Fe2O3 with 1.30 mA cm 2 at 1.23 V vs. RHE. Based on the above results, charge separation efficiencies on the surface (ηsurface ) and in the bulk (ηbulk ) can be calculated and plotted as shown in Fig. 5a and 5b. Clearly, The ηsurface for Fe2O3–Ti and Ti–Fe2O3 can reach up to 96% at the higher potential range from 1.30 to 1.50 V vs. RHE, which is much larger than that of pristine Fe2O3 (60% at 1.23 V vs. RHE), and are among the top values in the record of hematite-based photoanodes without co-cocatalyst (See Table S1). This result illus trates that the Ti gradient doping is favorable to alleviate the severe surface recombination and sluggish surface reaction for Fe2O3. How ever, when the potential is less than 1.04 V vs. RHE, the ηsurface of Fe2O3–Ti is extremely low, which may be due to the high interfacial charge recombination comparing with Ti–Fe2O3 photoanode. This phenomenon is well consistent with that of PEC results shown in Fig. 4a. The plot of ηbulk exhibits the totally different trend as illustrated in Fig. 5b. The ηbulk of pristine Fe2O3 photoanode is as low as 7.9% at 1.23 V vs. RHE, which is comparable to the previously reported value [42]. After the incorporation of Ti, the ηbulk of Fe2O3–Ti and Ti–Fe2O3 pho toanodes achieves the same level of 14.7% at 1.23 V vs. RHE, which is two times higher comparing with that of pristine Fe2O3, confirming that the Ti gradient doping can effectively facilitate the bulk charge transport and separation. To gain insight into the charge transfer efficiency of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes, the electrochemical impedance spectroscopy (EIS) analysis was performed. In the Nyquist plots (Fig. 5c), the radius of each arc hints the behavior of charge transfer at the photoanode-electrolyte interface and a smaller radius implies a lower charge transfer resistance [26]. Obviously, the Fe2O3–Ti and Ti–Fe2O3 photoanodes possess a much smaller arc radius than that of pristine Fe2O3 photoanode, manifesting that Ti gradient doping can remarkably improve the hematite conductivity and enhance the PEC water splitting performance. Interestingly, the arc radius of Ti–Fe2O3 photoanode is much smaller than that of Fe2O3–Ti photoanode, which is most likely due to the reduced interfacial charge recombination caused by the remained interfacial TiO2 between Fe2O3 and FTO substrate. Additionally, the Mott-Schottky (M S) analysis was implemented to clarify the charge carrier densities of these photoanodes. As plotted in Fig. 5d, the positive slopes can confirm the n-type semiconductor nature of all photoelectrodes [21]. The carrier densities calculated from the slopes of the plots are 1.0 � 1018 cm 3, 6.8 � 1018 cm 3 and 9.0 � 1018 cm 3 for pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes, respec tively. Evidently, the distinctly smaller slope of Fe2O3–Ti and Ti–Fe2O3 implies the enhanced charge carrier density after the gradient Ti-doping. The stability of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes has been also checked as illustrated in Fig. 6. The chronoamperometry curves of all Ti-based hematite photoanodes are performed for 4 h continuously at 1.23 V vs. RHE under AM 1.5G illumination. Evidently, Fe2O3–Ti and Ti–Fe2O3 photoanodes show a remarkable stable photo current density with retaining nearly 92% and 95% of their original activity in the end, respectively. The worse stability of Fe2O3–Ti pho toanode comparing with that of Ti–Fe2O3 could result from the partly fall off of Ti source in the Fe2O3–Ti photoanode surface during the PEC measurement.
Fig. 6. Stability test with of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photo anodes collected under AM 1.5 G irradiation at 1.23 V vs. RHE.
location of the TiO2 layer, but mainly affected by the high annealing temperature. The corresponding PEC performances of two different location types of TiO2 (Fe2O3–Ti and Ti–Fe2O3) have been systematically investigated. Fe2O3–Ti and Ti–Fe2O3 photoanodes exhibit almost iden tical PEC performance with 1.50 mA cm 2 at 1.23 V vs. RHE. However, the onset potential of Ti–Fe2O3 photoanode displays a cathodic shift of 80 mV comparing with that of Fe2O3–Ti photoanode, which can be ascribed to the remained interfacial TiO2, reducing the lattice mismatch between FTO and hematite. Importantly, the ηsurface for Fe2O3–Ti and Ti–Fe2O3 can reach up to 96% at the higher potential range from 1.30 to 1.50 V vs. RHE, which are among the top values in the record of hematite-based photoanodes without co-cocatalyst. Further analysis demonstrates that the enlarged contact area between hematite photo anode and the electrolyte, the improved charge separation efficiency, and increased charge carrier density are the main reasons for the enhanced PEC water splitting. We believe that this technique of element gradient doping paves a facile way to address the charge transport and separation, further improving the PEC performance of hematite. Declaration of competing interest None. Acknowledgements L. Jia acknowledges the financial support from the Fundamental Research Funds for the Central Universities (No. GK201702007), the National Natural Science Foundation of China (No. 51872179). H. Wang acknowledges the financial support from the Fundamental Research Funds for the Central Universities (G2017KY0002), Natural Science Foundation of Shaanxi Province (2017JM5028), the National Natural Science Foundation of China (Nos. 11811530635, 51872240, and 51672225), and the 1000 Youth Talent Program of China. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.jpowsour.2019.227473. References
4. Conclusions
[1] J. Jian, G.S. Jiang, R. van de Krol, B.Q. Wei, H.Q. Wang, Nano Energy 51 (2018) 457–480. [2] J. Liu, H.Q. Wang, Z.P. Chen, H. Moehwald, S. Fiechter, R. van de Krol, L.P. Wen, L. Jiang, M. Antonietti, Adv. Mater. 27 (2015) 712–718. [3] H.Q. Wang, L.C. Jia, P. Bogdanoff, S. Fiechter, H. M€ ohwald, D. Shchukin, Energy Environ. Sci. 6 (2013) 799–804. [4] L.C. Jia, P. Bogdanoff, M. Schmid, U. Bloeck, S. Fiechter, H.Q. Wang, ACS Appl. Energy Mater. 1 (2018) 6748–6757.
In conclusion, gradient Ti doping in hematite has been achieved by combining the magnetron sputtering deposition of TiO2 with hydro thermal treatment of Fe2O3. Further investigation reveals that gradient Ti doping decreasing from top to bottom is not dependent on the 7
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx [24] M.G. Ahmed, I.E. Kretschmer, T.A. Kandiel, A.Y. Ahmed, F.A. Rashwan, D. W. Bahnemann, ACS Appl. Mater. Interfaces 7 (2015) 24053–24062. [25] X.B. Chai, H.F. Zhang, Q. Pan, J.L. Bian, Z.F. Chen, C.W. Cheng, Appl. Surf. Sci. 470 (2019) 668–676. [26] C.C. Li, T. Wang, Z.B. Luo, S.S. Liu, J.L. Gong, Small 12 (2016) 3415–3422. [27] A. Srivastav, A. Verma, A. Banerjee, S.A. Khan, M. Gupta, V.R. Satsangi, R. Shrivastav, S. Dass, Phys. Chem. Chem. Phys. 18 (2016) 32735–32743. [28] T.W. Kim, K.S. Choi, Science 343 (2014) 990–994. [29] F. Feng, C. Li, J. Jian, X.K. Qiao, H.Q. Wang, L.C. Jia, Chem. Eng. J. 368 (2019) 959–967. [30] D. Chen, Z.F. Liu, ChemSusChem 11 (2018) 3438–3448. [31] R. Zhang, L. Yang, X.N. Huang, T. Chen, F.L. Qu, Z.A. Liu, G. Du, A.M. Asiri, X. P. Sun, J. Mater. Chem. 5 (2017) 12086–12090. [32] Z.B. Luo, T. Wang, J.J. Zhang, C.C. Li, H.M. Li, J.L. Gong, Angew. Chem. Int. Ed. 56 (2017) 12878–12882. [33] D.M. Satoca, M. B€ artsch, C. F� abrega, A. Genç, S. Reinhard, T. Andreu, J. Arbiol, M. Niederberger, J.R. Morante, Energy Environ. Sci. 8 (2015) 3242–3254. [34] Y.P. Zhang, J.D. He, Q.L. Yang, H.F. Zhu, Q.Q. Wang, Q.Z. Xue, L.Q. Yu, J. Power Sources 440 (2019) 227120. [35] A. Annamalai, A. Subramanian, U. Kang, H. Park, S.H. Choi, J.S. Jang, J. Phys. Chem. C 119 (2015) 3810–3817. [36] B.J. Tan, K.J. Klabunde, P.M.A. Sherwood, Chem. Mater. 2 (1990) 186–191. [37] F. Li, J. Li, F.W. Li, L.L. Gao, X.F. Long, Y.P. Hu, C.L. Wang, S.Q. Wei, J. Jin, J. T. Ma, J. Mater. Chem. 6 (2018) 13412–13418. [38] D.W. Ding, B.T. Dong, J. Liang, H. Zhou, Y.C. Pang, S.J. Ding, ACS Appl. Mater. Interfaces 8 (2016) 24573–24578. [39] Q. Rui, L. Wang, Y.J. Zhang, C.C. Feng, B.B. Zhang, S.R. Fu, H.L. Guo, H.Y. Hu, Y. P. Bi, J. Mater. Chem. 6 (2018) 7021–7026. [40] R. van de Krol, M. Gr€ atzel, Photoelectrochemical Hydrogen Production, Springer, New York, 2012. [41] D.G. Wang, G.L. Chang, Y.Y. Zhang, J. Chao, J.Z. Yang, S. Su, L.H. Wang, C.H. Fan, L.H. Wang, Nanoscale 8 (2016) 12697–12701. [42] P.S. Bassi, R.P. Antony, P.P. Boix, Y. Fang, J. Barbe, L.H. Wong, Nano Energy 22 (2016) 310–318.
[5] J. Jian, Y.X. Xu, X.K. Yang, W. Liu, M.S. Fu, H.W. Yu, F. Xu, F. Feng, L.C. Jia, D. Friedrich, R. van de Krol, H.Q. Wang, Nat. Commun. 10 (2019) 2609. [6] T. Higashi, H. Nishiyama, Y. Suzuki, Y. Sasaki, T. Hisatomi, M. Katayama, T. Minegishi, K. Seki, T. Yamada, K. Domen, Angew. Chem. Int. Ed. 58 (2019) 2300–2304. [7] Z.Z. Ma, K. Song, L. Wang, F.M. Gao, B. Tang, H.L. Hou, W.Y. Yang, ACS Appl. Mater. Interfaces 11 (2019) 889–897. [8] H.Q. Wang, N. Koshizaki, L. Li, L.C. Jia, K. Kawaguchi, X.Y. Li, Adv. Mater. 23 (2011) 1865–1870. [9] H.Q. Wang, M. Miyauchi, Y. Ishikawa, A. Pyatenko, N. Koshizaki, Y. Li, L. Li, X. Li, Y. Bando, D. Golberg, J. Am. Chem. Soc. 133 (2011) 19102–19109. [10] L.C. Jia, K. Harbauer, P. Bogdanoff, I.H. Geppert, A. Ramírez, R. van de Krol, S. Fiechter, J. Mater. Chem. 2 (2014) 20196–20202. [11] T.K. Sahu, M.K. Mohanta, M. Qureshi, J. Power Sources 445 (2020) 227341. [12] L.C. Jia, P. Bogdanoff, A. Ramírez, U. Bloeck, D. Stellmach, S. Fiechter, Adv. Mater. Interfaces 3 (2016) 1500434. [13] H.J. Ahn, K.Y. Yoon, M.J. Kwak, J.H. Jang, Angew. Chem. Int. Ed. 55 (2016) 9922–9926. [14] L.C. Jia, K. Harbauer, P. Bogdanoff, K. Ellmer, S. Fiechter, J. Mater. Sci. Technol. 31 (2015) 655–659. [15] S.S. Yi, B.R. Wulan, J.M. Yan, Q. Jiang, Adv. Funct. Mater. 29 (2019) 1801902. [16] M.S. Park, D. Walsh, J.F. Zhang, J.H. Kim, S. Eslava, J. Power Sources 404 (2018) 149–158. [17] C.C. Li, Z.B. Luo, T. Wang, J.L. Gong, Adv. Mater. 30 (2018) 1707502. [18] A. Kay, I. Cesar, M. Gr€ atzel, J. Am. Chem. Soc. 128 (2006) 15714–15721. [19] C.C. Li, A. Li, Z.B. Luo, J.J. Zhang, X.X. Chang, Z.Q. Huang, T. Wang, J.L. Gong, Angew. Chem. Int. Ed. 56 (2017) 4150–4155. [20] H.Q. Ma, M.A. Mahadik, J.W. Park, M. Kumar, H.S. Chung, W.S. Chae, G.W. Kong, H.H. Lee, S.H. Choi, J.S. Jang, Nanoscale 10 (2018) 22560–22571. [21] Y.F. Yuan, J.W. Gu, K.H. Ye, Z.S. Chai, X. Yu, X.B. Chen, C.X. Zhao, Y.M. Zhang, W. J. Mai, ACS Appl. Mater. Interfaces 8 (2016) 16071–16077. [22] Z.B. Luo, C.C. Li, S.S. Liu, T. Wang, J.L. Gong, Chem. Sci. 8 (2017) 91–100. [23] H.J. Ahn, A. Goswami, F. Riboni, S. Kment, A. Naldoni, S. Mohajernia, R. Zboril, P. Schmuki, ChemSusChem 11 (2018) 1873–1879.
8
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Gradient Ti-doping in hematite photoanodes for enhanced photoelectrochemical performance Fan Feng a, Can Li a, Jie Jian b, Fan Li b, Youxun Xu b, Hongqiang Wang b, **, Lichao Jia a, * a
Key Laboratory of Applied Surface and Colloid Chemistry, National Ministry of Education, Shaanxi Key Laboratory for Advanced Energy Devices, Shaanxi Engineering Lab for Advanced Energy Technology, School of Materials Science and Engineering, Shaanxi Normal University, 620 West Chang’an Street, Xi’an, Shaanxi, 710119, China b State Key Laboratory of Solidification Processing, Center for Nano Energy Materials, School of Materials Science and Engineering, Northwestern Polytechnical University and Shaanxi Joint Labortary of Graphene, Xi’an, 710072, PR China
H I G H L I G H T S
G R A P H I C A L A B S T R A C T
� Gradient Ti doping is affected by tem perature rather than the location of Ti. � Fe2O3–Ti/Ti–Fe2O3 exhibit almost similar photocurrent of 1.50 mA cm 2 at 1.23 V. � The ηsurface of Fe2O3–Ti/Ti–Fe2O3 can reach to 96% at 1.30–1.50 V vs. RHE.
A R T I C L E I N F O
A B S T R A C T
Keywords: Charge separation Gradient doping Photoanode Fe2O3 Water splitting
Hematite based photoanode is promising for solar water splitting while suffers from poor charge transport and separation efficiency that limit its practical application. Herein, we demonstrate two types gradient Ti-doped Fe2O3 photoanodes (Fe2O3–Ti (TiO2 deposited on the top of Fe2O3) and Ti–Fe2O3 (Fe2O3 on the top of TiO2)) to enhance charge transport and separation efficiency. Interestingly, Fe2O3–Ti and Ti–Fe2O3 photoanodes exhibit almost identical PEC performance with photocurrent of 1.50 mA cm 2 at 1.23 V vs. the reversible hydrogen electrode (RHE). However, the onset potential of Ti–Fe2O3 photoanode displays a cathodic shift of 80 mV comparing with that of Fe2O3–Ti photoanode. Moreover, the charge separation efficiencies on the surface (ηsurface ) for Fe2O3–Ti and Ti–Fe2O3 can reach up to 96% at the higher potential range from 1.30 to 1.50 V vs. RHE, which are among the top values in the record of hematite-based photoanodes without co-cocatalyst. Further investigation demonstrates the forming of gradient Ti doping is not dependent on the location of the TiO2 layer, but mainly affected by the high annealing temperature. The enlarged contact area between hematite photoanode and the electrolyte, the improved charge separation efficiency, and increased charge carrier density are responsible for the enhanced PEC water splitting.
* Corresponding author. ** Corresponding author. E-mail addresses: [email protected] (H. Wang), [email protected] (L. Jia). https://doi.org/10.1016/j.jpowsour.2019.227473 Received 16 September 2019; Received in revised form 2 November 2019; Accepted 14 November 2019 0378-7753/© 2019 Elsevier B.V. All rights reserved.
Please cite this article as: Fan Feng, Journal of Power Sources, https://doi.org/10.1016/j.jpowsour.2019.227473
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
1. Introduction
photoanodes shows a significant increase to more than 96% at higher potential range from 1.30 to 1.50 V vs. RHE, which are among the top values in the record of hematite-based photoanodes without cococatalyst.
Photoelectrochemical (PEC) water splitting is considered as a promising approach to utilize solar energy for hydrogen generation [1–3]. In recent decades, various n-type semiconductors have been extensively explored as photoanode materials for PEC water oxidation [4–9]. Among a large selection of photocatalysts for solar water split ting, hematite (α-Fe2O3) has drawn great research interests owing to its suitable bandgap, ample availability, stability in a broad pH range, nontoxicity and low cost [10,11]. With a bandgap width of 1.9–2.2 eV (depending on the method of preparation), hematite would absorb suf ficient visible light and hold the potential to reach a solar to hydrogen efficiency up to 16.8% and a maximum photocurrent of 12.6 mA cm 2 under AM 1.5G illumination (100 mW cm 2 light intensity) based on the theoretical calculation [12,13]. Although these remarkable merits, the reported efficiencies of hematite are far from the satisfactory due to the sluggish charge carrier separation efficiency, which is mainly resulted from the physical and chemical nature of α-Fe2O3 including the extremely short hole-diffusion length (2–4 nm), low minority carriers mobility (0.2 cm2 V 1 s 1) and poor photogenerated charge carriers lifetime (<10 ps) [14–16]. Additionally, the slow kinetics of the oxygen evolution reaction is another obstacle to restrict the PEC performance [17]. To improve the PEC performance of Fe2O3 photoanodes, element doping, such as Si4þ [18], Zr4þ [19], Sn4þ [20], Ti4þ [21] etc., has been extensively employed to address these issues. Comparing with the normal homogenous doping, the gradient-doping could not only induce a more upward band bending within a very wide region, but also could substantially increase the electron density of hematite to improve the conductivity and charge separation efficiency of bulk Fe2O3 [22], thus received great attentions. For instance, Gong and co-workers described that gradient doping of phosphorus can increase the width of band bending in Fe2O3 photoanode inducing promoted charge separation efficiency in the bulk [22]. Schmuki and co-workers demonstrated that gradient Sn4þ doping in Fe2O3 could result in a decrease of the onset potential and an increase in the photocurrent density due to the decreased charge recombination and the high conductivity [23]. Particularly, the radius of Ti4þ (0.61 Å) is smaller than that of high spin Fe3þ (0.65 Å), which suggests that Ti incorporation would cause lattice contraction and a reduction in the unit cell volume, leading to the enhanced hopping probability of charge carriers [17]. Hence, Ti stands out as a promising candidate for hematite doping. Until now, most of research works have been mainly focused on the TiO2 induced Fe2TiO5 layer or the surface TiO2 passivation layer on the Fe2O3 [24–26]. Even the photocurrent density can be improved but the surface separation efficiency is still quite low. Recently, Dass and co-workers displayed gradient-Ti-doped hematite photoanodes used a multi-layer spray py rolysis method for the first time and found that a gradient concentration of Ti dopant in a Fe2O3 resulted in an enhanced photoelectrochemical response attributing to the values of high flat band potential, and high conductivity as well as low carriers recombination rate [27]. While the gradient Ti-doping in hematite has displayed preliminary success, seeking facile method, clarifying the formation mechanism while improving the surface separation efficiency is still a highly desirable challenge. Herein, we demonstrate a simple and facile method for synthesizing gradient-Ti-doped hematite photoanode by magnetron sputtering deposition of TiO2 and subsequent hydrothermal treatment. Further investigation reveals that an identical gradient Ti doping decreasing from top to bottom can be achieved whether the TiO2 located at the surface of Fe2O3 (Fe2O3–Ti) or in between Fe2O3 and FTO (Ti–Fe2O3). Both Fe2O3–Ti and Ti–Fe2O3 photoanodes exhibit pronounced PEC performance with almost identical photocurrent of 1.50 mA cm 2 at 1.23 V vs. RHE, while the onset potential of Ti–Fe2O3 photoanode demonstrates a negative shift of 80 mV comparing with that of Fe2O3–Ti photoanode. Additionally, the ηsurface of the Fe2O3–Ti and Ti–Fe2O3
2. Experimental section 2.1. Materials Ferric chloride (FeCl3, 97%) was obtained from Sigma-Aldrich Co. LLC. Sodium nitrate (NaNO3, 99%), acetone (C3H6O, 99.5%) and hy drochloric acid (HCl, 36–38 wt%) were purchased from Sinopharm Chemical Reagent Co., Ltd. 1 M KOH standard aqueous solution (pH ¼ 13.6) was provided by Alfa Aesar Co. LLC. Ethanol (C2H5OH, 99.7%) was supplied by Tianjin Guangfu Fine Chemical Research Institute. Deionized water with a resistance of 18.2 MΩ cm was produced by a Water Purification System (Merck Millipore, Direct-Q5 UV) and used for all reactions. Fluorine-doped tin oxide (FTO, 14 Ω, 2.2 mm) glass sub strates were purchased from Nippon Sheet Glass Co., Ltd. Before sample preparation, all FTO substrates were ultrasonically pre-cleaned in ethanol, acetone, ethanol and deionized water for 30 min sequentially, then treated by a plasma cleaner (Diener, ZEPTO) at 50 W in air for 10 min to increase the surface hydrophilicity. 2.2. Preparation of the Fe2O3 photoanodes The Fe2O3 photoanodes were synthesized through a versatile hy drothermal method as reported previously [26]. Briefly, 0.1 M FeCl3 and 1 M NaNO3 were added into a Teflon heating-clave, then proper quantity of HCl was used to adjust pH to 1.5. Afterwards, the heating-clave was transferred into an oven and maintained at 95 � C for 12 h, a yellow FeOOH film was deposited onto the FTO substrates. Next, the samples were rinsed in deionized water for several times and dried by N2 flow. After annealing at 550 � C for 2 h and at 800 � C for 10 min, the Fe2O3 nanorod-array photoanodes can be obtained. 2.3. Preparation of the Fe2O3–Ti and Ti–Fe2O3 film TiO2 deposition was performed at room temperature by a radio fre quency (RF) magnetron sputtering system using a pure TiO2 as sputter target. During the deposition, the sputtering power was fixed at 80 W and the basic pressure was maintained at 5 � 10 3 Pa with the steady Ar flow of 18 sccm. The deposition time was varied from 10 to 60 min. For the sample of Fe2O3–Ti, the TiO2 was deposited on the top of as-prepared Fe2O3 nanorod array and then annealed at 800 � C for 10 min. For the sample of Ti–Fe2O3, the TiO2 was deposited on FTO substrate firstly and then Fe2O3 nanorod array was synthesized on the top of TiO2 layer ac cording to the above hydrothermal method, after the finally annealing process at 550 � C for 2 h and 800 � C for 10 min, the prepared sample can be obtained. 2.4. Films characterization X-ray diffraction analysis (XRD) was used to investigate the crystal structures and composition of the as-synthesized samples over a 2θ range of 20� ~ 70� with a Cu Kα radiation (λ ¼ 1.54056 Å) at 45 kV and 200 mA equipped. The field emission scanning electron microscope (FESEM, SU8220, Japan, 3 kV) and transmission electron microscopy (TEM, JEOL JEM-2800, 200 kV) were performed to clarify the detailed morphology and microstructure of the prepared samples. The surface chemical state was evaluated by the X-ray photoelectron spectroscopy (XPS, Axis Ultra, Kratos Analytical Ltd). The binding energy was cali brated using the C 1s photoelectron peak at 284.6 eV. UV–vis absorption spectra were measured by a UV–vis spectrophotometer (UV-3600, Kra tos Analytical Ltd.). Raman spectra were recorded on a Renishaw via micro-Raman spectroscope with 532 nm laser. 2
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
2.5. PEC measurements
further annealing at 800 � C is beneficial to improve the crystallinity [30]. After the introduction of element Ti into the Fe2O3, the hexagonal structure hematite with obvious (110) peak still can be confirmed for Fe2O3–Ti and Ti–Fe2O3 samples as exhibited in Fig. 1a. When enlarging the diffraction peak of (110), a small high angle shift can be observed for the Fe2O3–Ti comparing with pure Fe2O3, as demonstrated in Fig. 1b, manifesting the unchanged orient growth direction and incorporation of Ti in the crystal lattice of Fe2O3 [31]. Interestingly, the situation is totally different for the Ti–Fe2O3, no any angle shift can be observed. Instead, the peak intensity of (300) decreases significantly and the new (104), (012) and (024) peaks appear, which would be resulted from the formation of multiple Fe2O3 crystal planes due to the incorporation of TiO2 [20]. Moreover, no other distinct peaks correlated with TiO2 or Ti-based compounds can be found from the Fe2O3–Ti and Ti–Fe2O3 samples, which could be ascribed to the relatively low concentration of Ti [29]. In order to confirm the incorporation of Ti in the hematite, Raman spectrum was explored as presented in Fig. 1c. Evidently, six intrinsic Raman peaks (224, 245, 292, 410, 499 and 613 cm 1) belonging to Fe2O3 can be observed from Fe2O3–Ti and Ti–Fe2O3 samples, which is in agreement well with reported results in the literature [29]. After careful examination, the pronounced Raman peak locating around 660 cm 1 in Fe2O3–Ti and Ti–Fe2O3 can be seen, which should derive from the disordered Fe2O3 phase induced by the Ti [32,33]. From the UV–vis absorbance spectra of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 films demonstrated in Fig. S2a, similar light absorption behavior with the absorption edge around 600 nm can be clearly seen, which corresponds to the bandgap of 2.1 eV, and is consistent well with the previously reports [12,34]. Further, it should be pointed that the introduction of Ti has no obvious effects on the absorption of Fe2O3. The XPS was also employed to investigate the surface chemical state of pure Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes. Distinct Fe, Ti, O and Sn signals can be observed from XPS full scan spectrum for all the samples (Fig. S2b). Here, the Sn signal could be ascribed to the selfdoping from fluorine-doped tin oxide (FTO) due to the high
The PEC properties were performed using a typical three-electrode configuration with prepared film as photoanodes, an Ag/AgCl (satu rated KCl) as reference electrode, and a Pt wire as counter electrode (CH Instrument, CHI 760E). 1 M KOH standard aqueous solution of PH 13.6 by N2 purging for 30 min was employed as electrolyte. 500 W xenon lamp (Beijing Perfectlight Technology Co. Ltd., CEL-S500) equipped with an AM 1.5G filter with a power density of 100 mW cm 2 was used as the simulated light source. Photocurrent vs. voltage (J-V) curves of samples were measured with a scan rate of 10 mV s 1 from cathodic to anodic potentials with the photoanode exposing area of 0.2 cm2. All potentials were converted to the reversible hydrogen electrode (RHE) potential [28]: ERHE ¼ EAg=AgCl þ 0:1976 þ 0:0591 � pH The incident photon-to-current conversion efficiency (IPCE) was investigated under a 300 W Xe light (CEL-HXF300) in combination with a grating monochromator (CEL-IS151). The electrochemical impedance spectra (EIS) measurements were carried out at frequency range of 1 MHz–10 Hz under 1.23 V vs. RHE. Mott-Schottky (M S) plots were measured at a frequency of 1000 Hz under dark condition. The elec trochemical active surface area (EASA) was calculated based on the electrochemical double-layer capacitance as presented in previously reports [29]. 3. Results and discussion The crystal structure and composition of as-prepared samples were determined by the X-ray diffraction (XRD) shown in Figs. S1 and 1a. Clearly, after the high temperature annealing at 550 � C and 800 � C, the FeOOH peaks disappear and Fe2O3 signals emerge. The pronounced (110) diffraction peak suggests that the Fe2O3 is highly oriented along the (110) direction perpendicular to the FTO substrate [26]. Addition ally, the much stronger diffraction peaks of (110) and (300) indicate that
Fig. 1. (a) XRD pattern, (b) Enlarged (110) peak of XRD pattern and (c) Raman spectra of the pristine TiO2, pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples, respectively. XPS (d) Fe 2p, (e) O 1s, and (f) Ti 2p spectrum collected from the pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples, respectively. 3
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
temperature treatment at 800 � C [21,35]. Fe 2p3/2 and Fe 2p1/2 peaks located at 710.5 eV and 723.9 eV (Fig. 1d) are associated with Fe3þ in Fe2O3, consisting well with previously reported value [36,37]. Compared with pure Fe2O3, the Fe 2p peak of the Fe2O3–Ti and Ti–Fe2O3 exhibits higher energy shift of around 0.4 eV, which could be correlated with the incorporation of Ti4þ ions in the Fe2O3 crystal lattice [38]. The O 1s XPS spectrum (Fig. 1e) can be splitted into two peaks at 529.5 eV and 530.8 eV for Fe2O3 sample, corresponding to lattice oxygen and surface hydroxyl oxygen respectively [39]. The different degree of high energy shifts for Fe2O3–Ti and Ti–Fe2O3 samples may suggest varying doping quantity of Ti4þ into the Fe2O3 lattice. In Fig. 1f, a similar small negative banding energy shift for Ti 2p3/2 and Ti 2p1/2 peaks can be easily distinguished from Fe2O3–Ti and Ti–Fe2O3, comparing with the pure TiO2. This result manifests that chemical environment of the Ti for both two films is almost the same and Ti4þ ions were inserted into the Fe2O3 matrix [26]. To examine the morphologies and microstructures of the as-prepared pristine Fe2O3, Fe2O3–Ti and Ti– Fe2O3 films, field emission scanning electron microscopy (FE-SEM) measurements were carried out as demonstrated in Fig. 2. As shown in Fig. 2a-c, all of the films are composed of uniform and homogeneous wormlike particles. After the incorporation of Ti in the Fe2O3, the average particle size exhibits an evidently decrease from 72 nm to 35 and 60 nm, and the Fe2O3–Ti contains the smallest particle size with around 35 nm (see Fig. S3). Meanwhile, the Ti–Fe2O3 film demonstrates the better dispersity and lower particle density when comparing with Fe2O3–Ti film. Further, from the cross-section displayed in Fig. 2d-f, one can confirm that Ti incorporation has significantly influence on the morphology and the average diameter of the Fe2O3–Ti nanorod (56 nm) is smaller than that of Ti–Fe2O3 nanorod (76 nm). But the length of nanorod arrays does not show big difference and all films possess similar thickness of about 340 nm. This can be confirmed by the corresponding TEM images shown in Fig. 3a and 3b. Moreover, the crystal lattice with interplanar spacing of 0.25 nm observed from the HRTEM in Fig. 3c and 3d could be well assigned to (110) plane of Fe2O3 (JCPDS card Number 89–0597) for both samples. To figure out the reason for the reduced particle size comparing with pure Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples without annealing at 800 � C were also checked (see Figs. S4a–c). Visibly, the particle size has almost no change, while the dispersity of Ti–Fe2O3 is much higher than that of pure Fe2O3. Interestingly, for Fe2O3–Ti, that is totally similar with the
sample of pure Fe2O3 treated at 550 � C followed by Ar plasma treatment for 30 min (see Fig. S4d). As shown in Figs. S4e and S4f, after further annealing at 800 � C, the particle size of the plasma treated Fe2O3 sample is slightly increased, but much smaller than that of pure one. Combining the results of Fe2O3–Ti and Ti–Fe2O3 samples under high temperature treatment at 800 � C (see Fig. 2b and 2c), the smaller particle size of Fe2O3–Ti sample could be ascribed to the synergistic effect of plasma treatment during the deposition and the Ti-based growth restriction on the particle surface. The grain size reduction of Ti–Fe2O3 sample could only result from the Ti-based reason. To further evaluate the Ti element distribution, X-ray photoelectron spectroscopy (XPS) depth analysis in Fig. 3e and 3f were carried out via employing argon-ion etching. Surprisingly, the Ti concentration de creases from top of the samples to the bottom with the increase of argonion etching time, presenting a gradient profile in both Fe2O3–Ti and Ti–Fe2O3. Additionally, the reducing of Ti concentration in the Fe2O3–Ti is much faster than that of Ti–Fe2O3 sample. In order to clarify the originate of the Ti gradient distribution for Ti–Fe2O3 sample, EDS elemental mapping of Ti–FeOOH, Ti–Fe2O3 without annealing at 800 � C, Ti–Fe2O3 and Fe2O3–Ti samples were performed as shown in Fig. S5. Clearly, less quantity of Ti in Ti–FeOOH and Ti–Fe2O3 sample without annealing at 800 � C can be distinguished on the surface (Figs. S5d and S5h), which may be resulted from the partially dissolution of the sput tered TiO2 in the acid hydrothermal solution with 1.5 pH, leading to uniform doping of Ti4þ in Fe2O3. Fig. S5i-S5p display uniform and dense distribution of Fe, O and Ti elements in Ti–Fe2O3 and Fe2O3–Ti samples along the nanoparticles on the surface. Further, it should be pointed that the calculated Ti concentration is around 0.05 atom % in Ti–Fe2O3 sample without annealing at 800 � C, which is only one eighth of pre pared Ti–Fe2O3 samples (0.40 atom %). This can also be verified by the XPS results of Fe2O3–Ti and Ti–Fe2O3 samples after annealing at 550 � C (Fig. S6). Combining with the above SEM, TEM and XPS results, the gradient Ti doping obtained here could be ascribed to the high tem perature annealing at 800 � C, which could cause the melt and recrys tallization of the oxide [29]. As we know, the mass density of TiO2 (3.90 g cm 3) is smaller than that of Fe2O3 (5.26 g cm 3) [40], that means the TiO2 could very easily locate near surface during the melting process and form a gradient doping from top to bottom during rapid recrystallization. In order to evaluate the influence of Ti incorporation in the Fe2O3 photoanodes, PEC measurements were employed in 1 M KOH electrolyte
Fig. 2. (a, b, c) SEM top view, and (d, e, f) the corresponding cross section images of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 thin films. 4
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
Fig. 3. (a, b) TEM images, (c, d) HRTEM images, and (e, f) XPS depth profile signal of Ti 2p in Fe2O3–Ti and Ti–Fe2O3 samples, respectively.
Fig. 4. (a) Chopped photocurrent density-potential (J–V) characteristics measured in 1 M KOH solution under AM 1.5 G illumination (100 mW cm 2), (b) Transient photocurrent measurements at 1.23 V vs. RHE, (c) IPCE spectra collected at the incident wavelength range from 340 to 660 nm at 1.23 V vs. RHE, and (d) ABPE spectra of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes, respectively. 5
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
using a three-electrode electrochemical cell under simulated AM 1.5G irradiation. It is noticeable that the pristine Fe2O3 photoanode treated at 800 � C was used as control sample. The J-V curves of the Ti–Fe2O3 and Fe2O3–Ti with different TiO2 sputtering time are displayed in Fig. S7. Obviously, all the samples exhibit the neglected dark current densities from 0.60 to 1.60 V vs. RHE. The PEC performance is significantly enhanced comparing with the pure hematite while changes with the TiO2 deposition time. The Fe2O3 with 30 min and 40 min TiO2 treatment demonstrate the best photocurrent density value for the Fe2O3–Ti and Ti–Fe2O3 photoanodes at 1.23 V vs. RHE, respectively. From the SEM images shown in Fig. S8 and Fig. S9, the smaller particle size can be confirmed for both two samples, which could result in more surfaceactive site and interfacial contact area between photoanode materials and electrolyte, inducing the improved PEC performance [29]. Further, when comparing the Fe2O3–Ti with Ti–Fe2O3 samples (as illustrated in Fig. 4a), one can found that both two photoanodes display almost the similar water oxidation behavior and the photocurrent densities could reach to 1.50 mA cm 2 at 1.23 V vs. RHE, which is about two times higher than that of pristine Fe2O3 (0.80 mA cm 2 at 1.23 V vs. RHE). Interestingly, the onset potential of Ti–Fe2O3 photoanode displays a cathodic shift of 80 mV comparing with that of Fe2O3–Ti photoanode. Meanwhile, the photocurrent is significantly larger at low potential range from 0.80 to 1.20 V vs. RHE, which may be resulted by the remained interfacial TiO2, reducing the lattice mismatch between FTO and hematite [32,41]. Combining with the above XPS results, it is obvious that the sputtered TiO2 in Ti–Fe2O3 sample not only can act as a source for Ti4þ gradient doping but also can work as interfacial layer to reduce the interfacial charge recombination, and improve the hematite conductivity (see Fig. 5c) [32]. Moreover, in order to assess the exposing active sites, the electrochemical active surface area (EASA) has been investigated and shown in Fig. S10. Evidently, the linear slopes (see Fig. S10d) are 3.5, 5.5 and 5.4 μF cm 2 for pristine Fe2O3, Fe2O3–Ti and
Ti–Fe2O3 photoanodes, respectively, manifesting both Fe2O3–Ti and Ti–Fe2O3 films can provide almost identical active sites during the photoelectrochemical process. This can be ascribed to the balanced ef fect between dispersity and particle size of Ti–Fe2O3 and Fe2O3–Ti sample as showed in Fig. 2b and 2c. The transient photocurrent measurements of Fe2O3–Ti and Ti–Fe2O3 photoanodes were investigated to estimate the charge recombination behaviors at the semiconductor-electrolyte junction at 1.23 V vs. RHE under chopped light illumination (as illustrated in Fig. 4b). Evidently, the sharp photocurrent spike of pristine hematite is greatly restrained after the incorporation of Ti in Fe2O3, indicating the weaken charge carries recombination during their transport to the surface of the semiconductor. To further investigate the photo-response properties, the incident photon-to-electron conversion efficiency (IPCE) was measured as a function of incident light wavelength in 1 M KOH at 1.23 V vs. RHE, as illustrated in Fig. 4c. Apparently, the IPCE of both Fe2O3–Ti and Ti–Fe2O3 photoanodes demonstrates the significantly enhancement in almost the whole examined wavelength from 340 nm to 660 nm comparing with that of pristine hematite. The best value from Ti–Fe2O3 photoanodes could reach to 39% at 340 nm, which is around 1.25 times higher than that of Fe2O3–Ti (31%). The higher IPCE of Ti–Fe2O3 pho toanode when the wavelength is below 450 nm suggests that the absorbed photons can be utilized more efficiently, which may be owing to the increased hematite conductivity and charge carrier density. Moreover, the applied bias photon-to-current efficiency (ABPE) was also carried out to quantitatively assess the PEC water splitting efficiency of these samples. As presented in Fig. 4d, an optimum photoconversion efficiency of 0.13% can be obtained for the Ti–Fe2O3 photoanode at 1.10 V vs. RHE, which is higher than that of Fe2O3–Ti photoanode (0.10% at 1.10 V vs. RHE) and around two times over comparing with that of pristine Fe2O3 (0.06% at 1.10 V vs. RHE). Clearly, the improved PEC performance is closely related with the
Fig. 5. (a) Surface charge separation efficiency (ηsurface ), (b) Bulk charge transfer efficiency (ηbulk ), (c) Electrochemical impedance spectra (EIS) at 1.23 V vs. RHE, and (d) Mott-Schottky plots measured at 1 kHz frequency of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 samples, respectively. 6
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx
effect of enhanced charge separation caused by the incorporation of Ti in hematite. To verify this, J-V measurements were performed using 0.1 M H2O2 as a hole scavenger in 1 M KOH electrolyte (Fig. S11). The photocurrent density of both Fe2O3–Ti and Ti–Fe2O3 photoanodes can reach up to 1.60 mA cm 2 at 1.23 V vs. RHE, which is significantly higher than that of pristine Fe2O3 with 1.30 mA cm 2 at 1.23 V vs. RHE. Based on the above results, charge separation efficiencies on the surface (ηsurface ) and in the bulk (ηbulk ) can be calculated and plotted as shown in Fig. 5a and 5b. Clearly, The ηsurface for Fe2O3–Ti and Ti–Fe2O3 can reach up to 96% at the higher potential range from 1.30 to 1.50 V vs. RHE, which is much larger than that of pristine Fe2O3 (60% at 1.23 V vs. RHE), and are among the top values in the record of hematite-based photoanodes without co-cocatalyst (See Table S1). This result illus trates that the Ti gradient doping is favorable to alleviate the severe surface recombination and sluggish surface reaction for Fe2O3. How ever, when the potential is less than 1.04 V vs. RHE, the ηsurface of Fe2O3–Ti is extremely low, which may be due to the high interfacial charge recombination comparing with Ti–Fe2O3 photoanode. This phenomenon is well consistent with that of PEC results shown in Fig. 4a. The plot of ηbulk exhibits the totally different trend as illustrated in Fig. 5b. The ηbulk of pristine Fe2O3 photoanode is as low as 7.9% at 1.23 V vs. RHE, which is comparable to the previously reported value [42]. After the incorporation of Ti, the ηbulk of Fe2O3–Ti and Ti–Fe2O3 pho toanodes achieves the same level of 14.7% at 1.23 V vs. RHE, which is two times higher comparing with that of pristine Fe2O3, confirming that the Ti gradient doping can effectively facilitate the bulk charge transport and separation. To gain insight into the charge transfer efficiency of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes, the electrochemical impedance spectroscopy (EIS) analysis was performed. In the Nyquist plots (Fig. 5c), the radius of each arc hints the behavior of charge transfer at the photoanode-electrolyte interface and a smaller radius implies a lower charge transfer resistance [26]. Obviously, the Fe2O3–Ti and Ti–Fe2O3 photoanodes possess a much smaller arc radius than that of pristine Fe2O3 photoanode, manifesting that Ti gradient doping can remarkably improve the hematite conductivity and enhance the PEC water splitting performance. Interestingly, the arc radius of Ti–Fe2O3 photoanode is much smaller than that of Fe2O3–Ti photoanode, which is most likely due to the reduced interfacial charge recombination caused by the remained interfacial TiO2 between Fe2O3 and FTO substrate. Additionally, the Mott-Schottky (M S) analysis was implemented to clarify the charge carrier densities of these photoanodes. As plotted in Fig. 5d, the positive slopes can confirm the n-type semiconductor nature of all photoelectrodes [21]. The carrier densities calculated from the slopes of the plots are 1.0 � 1018 cm 3, 6.8 � 1018 cm 3 and 9.0 � 1018 cm 3 for pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes, respec tively. Evidently, the distinctly smaller slope of Fe2O3–Ti and Ti–Fe2O3 implies the enhanced charge carrier density after the gradient Ti-doping. The stability of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photoanodes has been also checked as illustrated in Fig. 6. The chronoamperometry curves of all Ti-based hematite photoanodes are performed for 4 h continuously at 1.23 V vs. RHE under AM 1.5G illumination. Evidently, Fe2O3–Ti and Ti–Fe2O3 photoanodes show a remarkable stable photo current density with retaining nearly 92% and 95% of their original activity in the end, respectively. The worse stability of Fe2O3–Ti pho toanode comparing with that of Ti–Fe2O3 could result from the partly fall off of Ti source in the Fe2O3–Ti photoanode surface during the PEC measurement.
Fig. 6. Stability test with of pristine Fe2O3, Fe2O3–Ti and Ti–Fe2O3 photo anodes collected under AM 1.5 G irradiation at 1.23 V vs. RHE.
location of the TiO2 layer, but mainly affected by the high annealing temperature. The corresponding PEC performances of two different location types of TiO2 (Fe2O3–Ti and Ti–Fe2O3) have been systematically investigated. Fe2O3–Ti and Ti–Fe2O3 photoanodes exhibit almost iden tical PEC performance with 1.50 mA cm 2 at 1.23 V vs. RHE. However, the onset potential of Ti–Fe2O3 photoanode displays a cathodic shift of 80 mV comparing with that of Fe2O3–Ti photoanode, which can be ascribed to the remained interfacial TiO2, reducing the lattice mismatch between FTO and hematite. Importantly, the ηsurface for Fe2O3–Ti and Ti–Fe2O3 can reach up to 96% at the higher potential range from 1.30 to 1.50 V vs. RHE, which are among the top values in the record of hematite-based photoanodes without co-cocatalyst. Further analysis demonstrates that the enlarged contact area between hematite photo anode and the electrolyte, the improved charge separation efficiency, and increased charge carrier density are the main reasons for the enhanced PEC water splitting. We believe that this technique of element gradient doping paves a facile way to address the charge transport and separation, further improving the PEC performance of hematite. Declaration of competing interest None. Acknowledgements L. Jia acknowledges the financial support from the Fundamental Research Funds for the Central Universities (No. GK201702007), the National Natural Science Foundation of China (No. 51872179). H. Wang acknowledges the financial support from the Fundamental Research Funds for the Central Universities (G2017KY0002), Natural Science Foundation of Shaanxi Province (2017JM5028), the National Natural Science Foundation of China (Nos. 11811530635, 51872240, and 51672225), and the 1000 Youth Talent Program of China. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.jpowsour.2019.227473. References
4. Conclusions
[1] J. Jian, G.S. Jiang, R. van de Krol, B.Q. Wei, H.Q. Wang, Nano Energy 51 (2018) 457–480. [2] J. Liu, H.Q. Wang, Z.P. Chen, H. Moehwald, S. Fiechter, R. van de Krol, L.P. Wen, L. Jiang, M. Antonietti, Adv. Mater. 27 (2015) 712–718. [3] H.Q. Wang, L.C. Jia, P. Bogdanoff, S. Fiechter, H. M€ ohwald, D. Shchukin, Energy Environ. Sci. 6 (2013) 799–804. [4] L.C. Jia, P. Bogdanoff, M. Schmid, U. Bloeck, S. Fiechter, H.Q. Wang, ACS Appl. Energy Mater. 1 (2018) 6748–6757.
In conclusion, gradient Ti doping in hematite has been achieved by combining the magnetron sputtering deposition of TiO2 with hydro thermal treatment of Fe2O3. Further investigation reveals that gradient Ti doping decreasing from top to bottom is not dependent on the 7
F. Feng et al.
Journal of Power Sources xxx (xxxx) xxx [24] M.G. Ahmed, I.E. Kretschmer, T.A. Kandiel, A.Y. Ahmed, F.A. Rashwan, D. W. Bahnemann, ACS Appl. Mater. Interfaces 7 (2015) 24053–24062. [25] X.B. Chai, H.F. Zhang, Q. Pan, J.L. Bian, Z.F. Chen, C.W. Cheng, Appl. Surf. Sci. 470 (2019) 668–676. [26] C.C. Li, T. Wang, Z.B. Luo, S.S. Liu, J.L. Gong, Small 12 (2016) 3415–3422. [27] A. Srivastav, A. Verma, A. Banerjee, S.A. Khan, M. Gupta, V.R. Satsangi, R. Shrivastav, S. Dass, Phys. Chem. Chem. Phys. 18 (2016) 32735–32743. [28] T.W. Kim, K.S. Choi, Science 343 (2014) 990–994. [29] F. Feng, C. Li, J. Jian, X.K. Qiao, H.Q. Wang, L.C. Jia, Chem. Eng. J. 368 (2019) 959–967. [30] D. Chen, Z.F. Liu, ChemSusChem 11 (2018) 3438–3448. [31] R. Zhang, L. Yang, X.N. Huang, T. Chen, F.L. Qu, Z.A. Liu, G. Du, A.M. Asiri, X. P. Sun, J. Mater. Chem. 5 (2017) 12086–12090. [32] Z.B. Luo, T. Wang, J.J. Zhang, C.C. Li, H.M. Li, J.L. Gong, Angew. Chem. Int. Ed. 56 (2017) 12878–12882. [33] D.M. Satoca, M. B€ artsch, C. F� abrega, A. Genç, S. Reinhard, T. Andreu, J. Arbiol, M. Niederberger, J.R. Morante, Energy Environ. Sci. 8 (2015) 3242–3254. [34] Y.P. Zhang, J.D. He, Q.L. Yang, H.F. Zhu, Q.Q. Wang, Q.Z. Xue, L.Q. Yu, J. Power Sources 440 (2019) 227120. [35] A. Annamalai, A. Subramanian, U. Kang, H. Park, S.H. Choi, J.S. Jang, J. Phys. Chem. C 119 (2015) 3810–3817. [36] B.J. Tan, K.J. Klabunde, P.M.A. Sherwood, Chem. Mater. 2 (1990) 186–191. [37] F. Li, J. Li, F.W. Li, L.L. Gao, X.F. Long, Y.P. Hu, C.L. Wang, S.Q. Wei, J. Jin, J. T. Ma, J. Mater. Chem. 6 (2018) 13412–13418. [38] D.W. Ding, B.T. Dong, J. Liang, H. Zhou, Y.C. Pang, S.J. Ding, ACS Appl. Mater. Interfaces 8 (2016) 24573–24578. [39] Q. Rui, L. Wang, Y.J. Zhang, C.C. Feng, B.B. Zhang, S.R. Fu, H.L. Guo, H.Y. Hu, Y. P. Bi, J. Mater. Chem. 6 (2018) 7021–7026. [40] R. van de Krol, M. Gr€ atzel, Photoelectrochemical Hydrogen Production, Springer, New York, 2012. [41] D.G. Wang, G.L. Chang, Y.Y. Zhang, J. Chao, J.Z. Yang, S. Su, L.H. Wang, C.H. Fan, L.H. Wang, Nanoscale 8 (2016) 12697–12701. [42] P.S. Bassi, R.P. Antony, P.P. Boix, Y. Fang, J. Barbe, L.H. Wong, Nano Energy 22 (2016) 310–318.
[5] J. Jian, Y.X. Xu, X.K. Yang, W. Liu, M.S. Fu, H.W. Yu, F. Xu, F. Feng, L.C. Jia, D. Friedrich, R. van de Krol, H.Q. Wang, Nat. Commun. 10 (2019) 2609. [6] T. Higashi, H. Nishiyama, Y. Suzuki, Y. Sasaki, T. Hisatomi, M. Katayama, T. Minegishi, K. Seki, T. Yamada, K. Domen, Angew. Chem. Int. Ed. 58 (2019) 2300–2304. [7] Z.Z. Ma, K. Song, L. Wang, F.M. Gao, B. Tang, H.L. Hou, W.Y. Yang, ACS Appl. Mater. Interfaces 11 (2019) 889–897. [8] H.Q. Wang, N. Koshizaki, L. Li, L.C. Jia, K. Kawaguchi, X.Y. Li, Adv. Mater. 23 (2011) 1865–1870. [9] H.Q. Wang, M. Miyauchi, Y. Ishikawa, A. Pyatenko, N. Koshizaki, Y. Li, L. Li, X. Li, Y. Bando, D. Golberg, J. Am. Chem. Soc. 133 (2011) 19102–19109. [10] L.C. Jia, K. Harbauer, P. Bogdanoff, I.H. Geppert, A. Ramírez, R. van de Krol, S. Fiechter, J. Mater. Chem. 2 (2014) 20196–20202. [11] T.K. Sahu, M.K. Mohanta, M. Qureshi, J. Power Sources 445 (2020) 227341. [12] L.C. Jia, P. Bogdanoff, A. Ramírez, U. Bloeck, D. Stellmach, S. Fiechter, Adv. Mater. Interfaces 3 (2016) 1500434. [13] H.J. Ahn, K.Y. Yoon, M.J. Kwak, J.H. Jang, Angew. Chem. Int. Ed. 55 (2016) 9922–9926. [14] L.C. Jia, K. Harbauer, P. Bogdanoff, K. Ellmer, S. Fiechter, J. Mater. Sci. Technol. 31 (2015) 655–659. [15] S.S. Yi, B.R. Wulan, J.M. Yan, Q. Jiang, Adv. Funct. Mater. 29 (2019) 1801902. [16] M.S. Park, D. Walsh, J.F. Zhang, J.H. Kim, S. Eslava, J. Power Sources 404 (2018) 149–158. [17] C.C. Li, Z.B. Luo, T. Wang, J.L. Gong, Adv. Mater. 30 (2018) 1707502. [18] A. Kay, I. Cesar, M. Gr€ atzel, J. Am. Chem. Soc. 128 (2006) 15714–15721. [19] C.C. Li, A. Li, Z.B. Luo, J.J. Zhang, X.X. Chang, Z.Q. Huang, T. Wang, J.L. Gong, Angew. Chem. Int. Ed. 56 (2017) 4150–4155. [20] H.Q. Ma, M.A. Mahadik, J.W. Park, M. Kumar, H.S. Chung, W.S. Chae, G.W. Kong, H.H. Lee, S.H. Choi, J.S. Jang, Nanoscale 10 (2018) 22560–22571. [21] Y.F. Yuan, J.W. Gu, K.H. Ye, Z.S. Chai, X. Yu, X.B. Chen, C.X. Zhao, Y.M. Zhang, W. J. Mai, ACS Appl. Mater. Interfaces 8 (2016) 16071–16077. [22] Z.B. Luo, C.C. Li, S.S. Liu, T. Wang, J.L. Gong, Chem. Sci. 8 (2017) 91–100. [23] H.J. Ahn, A. Goswami, F. Riboni, S. Kment, A. Naldoni, S. Mohajernia, R. Zboril, P. Schmuki, ChemSusChem 11 (2018) 1873–1879.
8