Efficiently enhancing the photocatalytic activity of faceted TiO₂ nanocrystals by selectively loading α-Fe₂O₃ and Pt co-catalysts

Abstract

The loading of oxidation and/or reduction co-catalysts onto the surface of semiconductor nanomaterials is one of the most efficient methods of improving the performance of semiconductor-based photocatalysts. However, in most of cases, the enhancing effect can be weakened by a random co-catalyst loading method because the different roles of photocatalyst facets in the photocatalytic process are not simultaneously considered. In this paper, a ternary composite photocatalyst, Fe₂O₃–TiO₂–Pt, with α-Fe₂O₃ and Pt nanoparticles selectively deposited onto the {001} and {101} facets of TiO₂, respectively, was successfully constructed using the facet-induced photogenerated electrons and holes of well-faceted anatase TiO₂ nanocrystals as natural redox agents. The overall photocatalytic activity of this well-designed composite photocatalyst in H₂ production has been enhanced greatly by as much as 2.2 times and 30 times compared to the photocatalysts loaded randomly and without a co-catalyst, respectively. The enhanced photocatalytic activity of Fe₂O₃–TiO₂–Pt was attributed to the remarkably enhanced separation of photogenerated charge carriers with the excitation of UV light.

---

1. Introduction

Semiconductor-based photocatalysis has attracted increasing attention recently not only for the generation of solar fuels (e.g., by splitting water for H₂ production and reducing CO₂ for CH₄ production) but also for the treatment of gas and water contaminants.^1–6 The photocatalytic activity of a semiconductor catalyst primarily depends on its ability to create electron–hole pairs. The most effective strategy for promoting the photocatalytic activity of semiconductors is, therefore, to improve the separation efficiency of photogenerated electron–hole pairs.^7,8 Noble metal (such as Pt) and/or metal oxide nanoparticles (such as NiO) have been loaded onto semiconductor nanomaterials as co-catalysts to successfully improve the performance of semiconductor photocatalysts.^9,10 For these binary and ternary composite photocatalysts, the separation between electrons and holes can be considerably enhanced under the driving force of a built-in electric field at the heterojunction interface.^11–14 Unfortunately, the performance of the composite photocatalysts have been far lower than anticipated, despite much higher than that of unloaded photocatalysts. This lower-than-expected performance might be the result of random co-catalyst loading. The exploration of rationally engineered co-catalysts at desired locations on photocatalyst surfaces is needed to optimize the photocatalytic activities of these composite photocatalysts.

Recently, photogenerated electrons and holes were found to be spatially separated onto different facets of well-faceted semiconductor nanocrystals (NCs), leading to different redox reactivities towards photo-activated chemical reactions.^15–18 These findings suggest a route towards loading co-catalysts at specific locations on photocatalyst surfaces. In other words, noble metal or metal oxide co-catalysts can be selectively loaded onto the specific reductive or oxidative facets of semiconductor nanocrystals (NCs) using the facet-induced photogenerated electrons and holes as natural redox agents. Recently, this synthetic strategy has been demonstrated with many semiconductors such as TiO₂, BiVO₄, BiOCl, and AgI.^19–22 Despite great success in rationally engineering locations of co-catalyst, we are still facing a problem how to design band position matched heterostructure semiconductors, in which the synergy between the heterojunction-induced and facet-induced effects for the separation of photogenerated electrons and holes can be achieved.

Hematite (α-Fe₂O₃) is a narrow bandgap semiconductor (E_g: 2.0–2.2 eV) and thought to be a promising photocatalytic and photovoltaic material, because of its natural abundance and excellent stability in aqueous media. Limited by its high recombination and the very short lifetime of its photogenerated charge carriers, α-Fe₂O₃ is often employed as a oxidation co-catalyst for the wide bandgap semiconductor photocatalyst (such as TiO₂) in photocatalytic reaction, instead of main photocatalyst.^23–27 In previous studies, the deposition of α-Fe₂O₃ onto the surface of semiconductor photocatalysts was typically achieved using either impregnation or hydrothermal methods, which usually result in randomly loading of α-Fe₂O₃ nanoparticles.^28,29 In this paper, a ternary composite photocatalyst was designedly constructed in which α-Fe₂O₃ and Pt co-catalysts were loaded on the photo-oxidative {001} facets and photo-reductive {101} facets, respectively, of tetragonal bipyramidal anatase TiO₂ NCs. Our experimental results demonstrated that this well-designed composite photocatalyst was significantly more active in H₂ evolution than the photocatalyst loaded randomly or with only one co-catalyst, due to its more efficient electron–hole separation, as we expected.

2. Experimental section

2.1 Chemicals

Potassium hydroxide (KOH, 90%), hexamethylenetetramine (HMTA, 99%), methanol (CH₃OH, 99.5%), ethanol (C₂H₅OH, 99.9%), potassium iodate (KIO₃, 99%), iron nitrate nonahydrate (Fe(NO₃)₃·9H₂O, 99%) and ferrous sulfate (FeSO₄·7H₂O, 99%) were purchased from Sinopharm Chemical Reagent Co., Ltd. Potassium hexachloroplatinate (K₂PtCl₆·6H₂O, 99.95%) was purchased from Alfar Aesar. The commercial photocatalyst Degussa P25 was purchased from Shanghai Haiyi Co. All reagents were used as received without further purification.

2.2 Synthesis of samples

2.2.1 Synthesis of tetragonal bipyramidal anatase TiO₂ NCs with exposed {101} and {001} facets. Tetragonal bipyramidal anatase TiO₂ NCs with exposed {001} and {101} facets were prepared via a hydrothermal route, using potassium titanate nanowires (KTNWs) as precursor and HMTA as shape regulator. KTNWs were prepared at 200 °C under hydrothermal conditions for 24 h using Degussa P25 and high concentration KOH solution as raw materials.³⁰ Then, KTNWs (0.01 g) and HMTA (0.70 g, 5 mmol) were in order dispersed in distilled water (6 mL) under intense ultrasonic treatment. The resulting solution was transferred into a 25 mL Teflon-lined stainless-steel autoclave and was kept at 200 °C for 12 h. After reaction, the white precipitate in the autoclave was collected by centrifugation, washed with distilled water and ethanol 3 times, and finally dried in an oven.

2.2.2 Deposition of α-Fe₂O₃ nanoparticles on the photo-oxidative {001} facets of TiO₂ NCs. The deposition of α-Fe₂O₃ on the photo-oxidative {001} facets of TiO₂ NCs was achieved via a photo-activated oxidation process using FeSO₄ as the precursor and KIO₃ as the electron scavenger. Typically, 50 mg TiO₂ NCs were suspended in a 100 mL KIO₃ solution (0.01 mol L⁻¹), and then 1.0 mL FeSO₄ solution (30 mmol L⁻¹) was added. The resulting suspension was irradiated by a 300 W Xe lamp under continuous stirring. After 1 h, the product was collected by centrifugation, washed with de-ionized water for more than 3 times, and finally calcined at 450 °C for 3 h. According to the composition of deposits on the {001} facets, the as-prepared samples before and after calcination were denoted as FeO_x(OH)_3−2x–TiO₂ (x = 0–1) and Fe₂O₃–TiO₂, respectively.

2.2.3 Deposition of Pt nanoparticles on the photo-reductive {101} facets of Fe₂O₃–TiO₂ NCs. Pt nanoparticles were selectively deposited on the photo-reductive {101} facets of the as-prepared Fe₂O₃–TiO₂ NCs via a photoreduction process using K₂PtCl₆ as the precursor.¹⁹ Typically, 50 mg Fe₂O₃–TiO₂ NCs were dispersed in 100 mL of deionized water, and then 1 mL K₂PtCl₆ aqueous solution (2.5 mmol L⁻¹) was impregnated into the above solution. The as-resulting suspension was stirred magnetically, and irradiated under a 300 W Xe lamp for 1 h at room temperature. The final product was collected by high-speed centrifugation, and washed with water and ethanol three times. The prepared sample was denoted as Fe₂O₃–TiO₂–Pt.

2.2.4 Deposition of α-Fe₂O₃ nanoparticles randomly on the surface of TiO₂ NCs. α-Fe₂O₃ nanoparticles were randomly deposited on the surface of TiO₂ NCs by a wet impregnation, drying, ethanol washing, and calcination process.³¹ Typically, 1 mL of 0.03 mol L⁻¹ Fe(NO₃)₃·9H₂O in ethanol was added to 50 mg of TiO₂ NCs, and the resulted suspension was stirred for 30 min at room temperature and sonicated for 30 min. After ethanol was evaporated at 50 °C, the sample was calcined at 300 °C for 10 min. And then, the sample was washed by ethanol, and finally calcined at 300 °C for 6 h. The prepared sample was denoted as n-Fe₂O₃–TiO₂.

2.2.5 Deposition of Pt nanoparticles on the {101} facets of n-Fe₂O₃–TiO₂ NCs. The procedure of depositing Pt nanoparticles on the {101} facets of n-Fe₂O₃–TiO₂ NCs was mostly as same as the procedure of depositing Pt nanoparticles on the {101} facets of Fe₂O₃–TiO₂ NCs except changing Fe₂O₃–TiO₂ to n-Fe₂O₃–TiO₂ NCs. The prepared sample was denoted as n-Fe₂O₃–TiO₂–Pt.

2.3 Characterization of samples

The morphology and structure of all samples were observed by transmission electron microscopy (TEM, JEOL JEM-2100) at a 200 kV accelerating voltage and scanning electron microscopy (SEM, Hitachi S-4800) at a 10 kV accelerating voltage. The UV-vis diffuse reflectance spectra (DRS) of samples were measured by UV-vis spectrophotometry (Varian Cary-5000). The specific surface areas of samples were measured by the Brunauer–Emmett–Teller (BET) method using nitrogen adsorption and desorption isotherms on a Micromeritics Instrument Corporation sorption analyzer (TriStar II 3020). The high resolution X-ray photoelectron spectroscopy (XPS) study was performed using a Thermo Scientific K-alpha photoelectron spectrometer (PHI Quantum-2000) using monochromatic Al Kα radiation; all peak positions were calibrated with the C 1s peak (284.5 eV) as an internal standard. The effective loading amounts of Pt and Fe₂O₃ on TiO₂ NCs were determined by inductively coupled plasma-atomic emission spectrometry (ICP-AES, Baird PS-4).

2.4 Photocatalytic activity measurement in H₂ evolution

The photocatalytic H₂ production experiments were conducted in a Labsolar-III (AG) photocatalytic H₂ production system (Perfectlight Co. Ltd). Typically, 25 mg TiO₂ sample was well dispersed in a mixed solution of deionized water (90 mL) and methanol (10 mL) under magnetically stirring. The reaction temperature was maintained at 10 °C by a cooling water circulation system. A 300 W Xe lamp (PLS-SXE-300UV, Beijing Trusttech Co. Ltd) with the wavelength of 300–2500 nm was employed as light source, and the power density was measured to be 1200 and 560 mW cm⁻² at the light outlet and actual location of reactor, respectively. Before irradiation, the system was pumped for 1 h to remove the air. The amount of H₂ evolved was determined by using a gas chromatography (GC 2060) with a TCD detector, and high purity Ar gas (99.999%) was used as carrier gas.

3. Results and discussion

For anatase TiO₂, the {001} and {101} facets become hole- and electron-rich, respectively, upon irradiation, thereby providing oxidation and reduction sites for chemical reaction, as depicted in Scheme 1a. Because of this unique photochemical behavior of anatase TiO₂, a step-by-step photo-activated chemical deposition process was developed to construct a special ternary composite photocatalyst with co-catalysts at desired locations of TiO₂ NCs. As illustrated in Scheme 1b, α-Fe₂O₃ nanoparticles can be selectively deposited on the hole-rich {001} facets of TiO₂ through the photooxidation of an Fe(II) salt (FeSO₄) (step 1) and a subsequent calcination process at 450 °C (step 2). And then Pt nanoparticles were deposited on the electron-rich {101} facets of TiO₂ via the photoreduction of K₂PtCl₆ (step 3). In this ternary composite photocatalyst, Pt nanoparticles deposited on the photo-reductive {101} facets of TiO₂ NCs not only act as sinks for photogenerated electrons, but also provide effective proton reduction sites for H₂ production. In the meanwhile, α-Fe₂O₃ nanoparticles deposited on the photo-oxidative {001} facets of TiO₂ NCs could induce a transfer of photogenerated holes from TiO₂ to Fe₂O₃, because the valence band of α-Fe₂O₃ is higher than that of TiO₂. Since the heterojunction-induced and facet-induced effects for the separation of photogenerated electrons and holes work cooperatively, this well-designed ternary composite photocatalyst is anticipated to exhibit a photocatalytic activity superior to that of a photocatalyst loaded randomly or with only one co-catalyst.

Scheme 1 (a) Schematic illustration of facet-induced spatial separation of photogenerated charges over tetragonal bipyramidal anatase TiO₂ NCs with exposed {101} and {001} facets. (b) Schematic illustration of the procedures of selectively depositing α-Fe₂O₃ and Pt onto the {001} and {101} facets of TiO₂ NCs, respectively.

In this study, anatase TiO₂ NCs with a truncated tetragonal bipyramidal morphology were used as a model photocatalyst, in which the top/bottom surfaces were {001} facets and the side surfaces were {101} facets. For the truncated tetragonal bipyramidal TiO₂ NCs used, the average lengths of the maximum and minimum sides of the square {001} faces in every particle were 250 nm and 200 nm, respectively, corresponding to 27.9% {001} facets exposed on the NCs' surfaces (Fig. S1†). Fig. 1 shows typical morphologies of TiO₂ NCs after the selective deposition of α-Fe₂O₃ and Pt co-catalysts, i.e. Fe₂O₃–TiO₂ and Fe₂O₃–TiO₂–Pt. In Fe₂O₃–TiO₂, the {001} facets of TiO₂ NCs were decorated with a large number of spherical nanoparticles with an average diameter of 4–8 nm, while the {101} facets remained smooth (Fig. 1a and b). These spherical nanoparticles were actually α-Fe₂O₃ and they resulted from the dehydration of amorphous ferric oxyhydroxide (FeO_x(OH)_3−2x, x = 0–1) nanoparticles that were formed on the {001} facets of TiO₂ NCs via the hydrolysis of Fe³⁺ after the photooxidation of Fe²⁺ (Fig. S2†).³² After the photoreduction process, the originally smooth {101} facets of TiO₂ NCs became very coarse, indicating the presence of nanoparticles (Fig. 1c and d). Unfortunately, these nanoparticles on the {101} facets were outside the resolution limit of the SEM due to their small size.

Fig. 1 Low magnification and high magnification SEM images of (a and b) Fe₂O₃–TiO₂, and (c and d) Fe₂O₃–TiO₂–Pt. Insets in (b) and (d) are corresponding models to show the sample morphologies.

In order to determine the composition of the deposits, XRD analysis was performed. However, the XRD patterns of these samples were the same as that of pure anatase TiO₂ (PDF no. 84-1286), and the characteristic peaks of Fe₂O₃ and Pt were not detected, as shown in Fig. 2a. This may be attributed to the low loading content and small size of the deposited nanoparticles.²⁵ ICP-AES results confirmed that the loading amounts of Fe and Pt were only ca. 3.17 wt% and 0.64 wt%, respectively, lower than their raw material ratios (ca. 5 wt% and 1 wt% for Fe and Pt, respectively).

Fig. 2 (a) XRD patterns of TiO₂, Fe₂O₃–TiO₂, and Fe₂O₃–TiO₂–Pt; TEM images of (b) Fe₂O₃–TiO₂ and (c) Fe₂O₃–TiO₂–Pt. Insets are high resolution TEM images recorded from the zones marked with “1” and “2”.

More information about the composition of the deposits on each facet was provided by TEM. As shown in Fig. 2b, in Fe₂O₃–TiO₂, 8–15 nm nanoparticles with a higher contrast appeared on the {001} facets of the TiO₂ NCs. The lattice fringes of 0.27 nm spacing were observed on the nanoparticles, which corresponded to the {104} plane of α-Fe₂O₃ (Fig. 2b). These α-Fe₂O₃ nanoparticles presented good crystallinity and close contact with the TiO₂ surface, which would facilitate charge transfer between α-Fe₂O₃ and TiO₂. In Fe₂O₃–TiO₂–Pt, the lattice fringes taken from the nanoparticles deposited on the {001} facets likewise corresponded to α-Fe₂O₃. Besides that, some very small nanoparticles of 2–3 nm in size were observed on the {101} facets of the TiO₂ NCs. The lattice spacing of these small nanoparticles was measured to be 0.224 nm, which corresponded to the {111} planes of metallic Pt (Fig. 2b). Therefore, the deposits on the {101} facets of the TiO₂ NCs were Pt nanoparticles.

To confirm the exact chemical states of Fe and Pt species deposited on the surface of the TiO₂ NCs, high resolution XPS were further measured, and the results taken from Fe₂O₃–TiO₂–Pt are shown in Fig. 3. The Ti 2p XPS spectrum displayed doublet peaks at 458.7 eV and 464.3 eV, which corresponded to Ti 2p_3/2 and Ti 2p_1/2, respectively (Fig. 3a). The splitting width between the two peaks was 5.7 eV, indicating a normal state of Ti⁴⁺. In the Fe 2p XPS spectrum (Fig. 3b), two peaks were observed near 710.4 eV (Fe 2p_3/2) and 724.2 eV (Fe 2p_1/2) with a weak shake-up satellite peak, which represented the presence of Fe³⁺–O.^33,34 The O 1s XPS spectrum displayed a slightly asymmetric single peak centered at 530.0 eV, which was deconvoluted by Gaussian fitting into three components (Fig. 3c). The dominant component (O_L) centered at 529.8 eV was attributed to the lattice oxygen in the oxides. The weak peak located at 530.7 eV was attributed to the component (O_V) associated with O²⁻ ions in oxygen-deficient regions (oxygen vacancies), and the weaker component (O_C) at approximately 532.5 eV was attributed to chemisorbed oxygen species (e.g., OH⁻).³⁵ It should be noted that Fe₂O₃–TiO₂–Pt contained a much lower proportion of both O_V and O_C compared to the FeO_x(OH)_3−2x–TiO₂ sample without calcination processing (Fig. S3†). This indicated that the crystallinity of ferric oxide nanoparticles deposited on the {001} facets of the TiO₂ NCs was improved by calcination. The Pt 4f XPS spectrum showed two peaks centered at 74.9 eV and 71.8 eV, which corresponded to the 4f_5/2 and 4f_7/2 signals of metallic Pt, respectively (Fig. 3d).

Fig. 3 High resolution XPS survey of (a) Ti 2p, (b) Fe 2p, (c) O 1s, and (d) Pt 4f for Fe₂O₃–TiO₂–Pt.

According to the above characterizations, there was no doubt that the nanoparticles selectively deposited on the {001} and {101} facets of TiO₂ in Fe₂O₃–TiO₂–Pt were α-Fe₂O₃ and Pt, respectively, as was intended. According to our design, a synergy was expected between the facet-induced and heterojunction-induced effects for the separation of photogenerated carriers in Fe₂O₃–TiO₂–Pt, which would efficiently enhance the photocatalytic activity of photocatalysts.

To validate the effectiveness of the fabrication strategy used, the photocatalytic performances of bare TiO₂, TiO₂ loaded only with Fe₂O₃ on {001} facets (Fe₂O₃–TiO₂), and TiO₂ loaded with dual co-catalysts (Fe₂O₃–TiO₂–Pt) for H₂ production were systematically evaluated. For comparison, TiO₂ loaded only with Pt on {101} facets (TiO₂–Pt), which has been previously proven to exhibit an excellent photocatalytic activity in H₂ evolution, were specifically prepared and measured as reference (Fig. S4†).¹⁹ Fig. 4a shows the H₂ production rate (μmol h⁻¹) in the presence of four samples under irradiation by a 300 W Xe lamp for 4 h, which followed the order of Fe₂O₃–TiO₂–Pt > TiO₂–Pt ≫ Fe₂O₃–TiO₂ > TiO₂. In the absence of co-catalysts, TiO₂ NCs exhibited very poor photocatalytic activity, and the hydrogen production rate was only 3.6 μmol h⁻¹. When loaded with α-Fe₂O₃, the photocatalyst displayed an enhanced photocatalytic activity, and the H₂ production rate increased to 7.1 μmol h⁻¹, near twice that of bare TiO₂. The hydrogen production rate of TiO₂–Pt was 20 times that of the bare TiO₂ NCs and 8 times that of Fe₂O₃–TiO₂. Excitingly, the hydrogen production rate of the ternary photocatalyst Fe₂O₃–TiO₂–Pt increased up to 90.0 μmol h⁻¹, which was 150% that of TiO₂–Pt. This result was entirely consistent with our expectation.

Fig. 4 (a) Comparison of hydrogen production rates of bare TiO₂, TiO₂ loaded only with one co-catalyst (Fe₂O₃–TiO₂ and TiO₂–Pt) and Fe₂O₃–TiO₂–Pt. (b) Comparison of hydrogen production rates of bare TiO₂, Fe₂O₃–TiO₂, Fe₂O₃–TiO₂–Pt and their counterparts with randomly loading (n-Fe₂O₃–TiO₂ and n-Fe₂O₃–TiO₂–Pt). A 300 W Xe lamp was used as the light source.

In fact, the importance of rationally engineering the locations of co-catalysts to improve photocatalytic activity can be further highlighted by comparing the photocatalytic performances between the composite photocatalysts with co-catalysts selectively deposited and their counterparts with randomly deposited. We have systematically studied the effect of Pt nanoparticles selective and random deposition on the surface of TiO₂ NCs in our previous study.¹⁹ The results indicated that the selective Pt deposition greatly enhanced the photocatalytic performance of TiO₂ NCs whereas this enhancement was reduced by randomly depositing Pt nanoparticles on the surface of TiO₂ NCs. In order to explore whether the deposition location of Fe₂O₃ nanoparticles have similar effect, two photocatalysts with α-Fe₂O₃ randomly deposited, n-Fe₂O₃–TiO₂ (Fig. S5†) and n-Fe₂O₃–TiO₂–Pt (Fig. S6†) was specifically synthesized, in which α-Fe₂O₃ were deposited on the TiO₂ NCs via a traditional impregnation–calcination method.³¹ It should be noted that the photocatalytic activity of TiO₂ NCs relates to the size of Fe₂O₃ nanoparticles. Therefore, different precursors and calcination temperatures were optimized in our study to ensure that the size distribution range of Fe₂O₃ selectively and randomly deposited on the surface of TiO₂ NCs were similar. As shown in Fig. S8,† the size distribution range of Fe₂O₃ in Fe₂O₃–TiO₂ and n-Fe₂O₃–TiO₂ sample are 6.4 ± 1.0 nm and 6.0 ± 1.0 nm, respectively. ICP-AES result confirmed the loading amounts of Fe (ca. 3.62 wt%) and Pt (ca. 0.69 wt%) in n-Fe₂O₃–TiO₂–Pt were similar as Fe₂O₃–TiO₂–Pt. As shown in Fig. 4b, the H₂ production rate of n-Fe₂O₃–TiO₂ was 5.1 μmol h⁻¹, which was slightly higher than that of bare TiO₂ (3.6 μmol h⁻¹) but lower than that of Fe₂O₃–TiO₂ (7.2 μmol h⁻¹). Even after the deposition of Pt nanoparticles, the H₂ production rate of n-Fe₂O₃–TiO₂–Pt was only 44.0 μmol h⁻¹, which was approximately half that of Fe₂O₃–TiO₂–Pt. Obviously, the photocatalysts with co-catalyst selectively deposited are much superior to their counterparts with co-catalysts deposited randomly in the photocatalytic activity. This indicated that the enhanced photocatalytic activity of Fe₂O₃–TiO₂–Pt indeed originated from the rational locations of dual co-catalysts on the TiO₂ NCs. In other words, the composite photocatalysts could not hit peak photocatalytic efficiency, if co-catalysts are randomly located on the surface of photocatalysts.

It should be pointed out that the photocatalytic activity of photocatalysts is actually influenced by many factors, such as the specific surface area of photocatalysts and the ability to absorb light. However, there was very few difference in their specific surface areas between the measured samples (10.35, 12.03, 11.80, 12.64 m² g⁻¹ for TiO₂, Fe₂O₃–TiO₂, TiO₂–Pt, and Fe₂O₃–TiO₂–Pt, respectively), and the specific surface area of Fe₂O₃–TiO₂–Pt was just slightly higher than Fe₂O₃–TiO₂ and TiO₂–Pt. Therefore, the significant enhancement of Fe₂O₃–TiO₂–Pt in photocatalytic activity compared to Fe₂O₃–TiO₂ and TiO₂–Pt has no relationship with the differences in their specific surface areas. On the other hand, a significant change in the color of the TiO₂ NCs occurred after the loading of the co-catalysts (insets to Fig. 5a). The UV-vis DRS of the samples revealed that the bare TiO₂ NCs absorbed only UV light below 400 nm, while Fe₂O₃–TiO₂ and Fe₂O₃–TiO₂–Pt not only absorbed light more strongly in the UV region but also had a considerable absorption in the visible region. The absorption edges of two samples were significantly red-shifted to approximately 620 nm, which is consistent with the bandgap of Fe₂O₃ (ca. 2.00 eV). Therefore, the extension of the light absorption range in the visible light region for Fe₂O₃–TiO₂ and Fe₂O₃–TiO₂–Pt can be attributed to α-Fe₂O₃ deposited on the TiO₂ NCs. Previous studies have demonstrated that TiO₂–Fe₂O₃ nanocomposites exhibited a pretty good photocatalytic activity under visible light, which was usually attributed to the visible light driven electron transfer.^23,26,27,29 That is, the electrons in the valence band of α-Fe₂O₃ could be excited to higher energy levels of the conduction band with the excitation of high-energy visible light (400–550 nm), and these high energy electrons, in which energy level was even higher than the conduction band position of TiO₂, could thermodynamically transfer to the conduction band of TiO₂, thereby enhancing photocatalytic activity of photocatalysts.²³ To determine the contribution from the visible-light driven photocatalytic activity of α-Fe₂O₃, the H₂ production rates of TiO₂, Fe₂O₃–TiO₂, TiO₂–Pt, and Fe₂O₃–TiO₂–Pt were measured under irradiation by visible light (λ ≥ 400 nm) using a UV400 cut-off filter. As shown in Fig. 5b, these photocatalysts performed badly under visible-light irradiation, and the photocatalytic H₂ evolution rates were orders of magnitude lower than those illuminated by an unfiltered Xe lamp. Such low visible light photocatalytic activities revealed that the contribution from the visible light driven electron transfer from α-Fe₂O₃ to TiO₂ were minimal. However, the photocatalytic activities of the different photocatalysts followed the same order under visible and UV-visible irradiation. Therefore, the rational locations of co-catalysts was exactly the underlying cause for the enhanced photocatalytic activities of Fe₂O₃–TiO₂–Pt.

Fig. 5 (a) UV-vis DRS of TiO₂, Fe₂O₃–TiO₂ and Fe₂O₃–TiO₂–Pt. Insets are photographs of the samples to show their color. (b) Comparison of hydrogen production rates of TiO₂, TiO₂–Pt, Fe₂O₃–TiO₂, and Fe₂O₃–TiO₂–Pt under irradiation by visible light (λ ≥ 400 nm).

It should be noted Fe₂O₃–TiO₂–Pt possessed not only remarkably enhanced photocatalytic ability but also an excellent stability for H₂ production. As shown in Fig. 6, the photocatalytic activity of Fe₂O₃–TiO₂–Pt decreased by less than 10% after four H₂ production cycles. Previous studies revealed that the photogenerated electrons in the conduction band of α-Fe₂O₃ could reduce Fe³⁺ to Fe²⁺, resulting in the dissolution of α-Fe₂O₃ in a liquid photocatalytic environment, which greatly restricts stability and recyclability of Fe₂O₃-based photocatalyst.^36–38 However, XPS analysis of the photocatalyst after the photocatalytic H₂ production reaction confirmed that Fe did not undergo a change of valence state (Fig. S7†). Therefore, the slight decrease in photocatalytic activity might have been caused by the incomplete recovery of the photocatalyst in these experiments.

Fig. 6 Multiple cycles of photocatalytic H₂ production in the presence of Fe₂O₃–TiO₂–Pt under irradiation by a 300 W Xe lamp. The amount of photocatalyst was 25 mg and the irradiation time was 4 h.

Given the above results, Fe₂O₃–TiO₂–Pt exhibited the best photocatalytic activity among all the photocatalysts and good stability for the H₂ evolution reaction as was expected. The rational engineering of the locations at which co-catalysts are deposited on a photocatalyst, i.e. the selective deposition of reduction and oxidation co-catalysts onto the relevant reduction and oxidation facets of TiO₂ NCs, was indeed able to effectively achieve an optimal photocatalytic activity. To interpret the remarkably enhanced photocatalytic performance of Fe₂O₃–TiO₂–Pt, the transfer of photogenerated electrons and holes in Fe₂O₃–TiO₂–Pt is illustrated in Scheme 2, where both the facet- and heterojunction-induced effects are fully considered. Note that the electrons and holes in the conduction bands and valence bands of TiO₂ and α-Fe₂O₃ are dominantly generated by the excitation with UV light of Xe lamp, since the contribution from the visible light has proven to be negligible. Previous studies have revealed that the positions of both the conduction and valence band of the {101} facets of anatase TiO₂ are slightly lower than those of the {001} facets.^17,39,40 Photogenerated electrons prefer to transfer toward the {101} facets, while photogenerated holes transfer toward the {001} facets (the facet-induced effect). As a result, the {101} and {001} facets provide reduction and oxidation sites, respectively, in photochemical reactions. When Pt nanoparticles are selectively deposited on the {101} electron-rich facets, the photogenerated electrons transfer to the Pt nanoparticles from the {101} facets of TiO₂, causing the Pt nanoparticles to become the catalytic centers for H₂ production (the semiconductor/metal heterojunction-induced effect). When α-Fe₂O₃ nanoparticles are deposited on the {001} facets of TiO₂ NCs, the holes concentrated on the {001} facets are likewise easily transferred to α-Fe₂O₃ because of higher valence band position of α-Fe₂O₃ (the semiconductor/semiconductor heterojunction-induced effect). In contrast, there is no pathway for the electrons, which are concentrated on the {101} facets of TiO₂ NCs, to transfer to α-Fe₂O₃ located on the {001} facets of TiO₂ NCs due to spatial separation, although α-Fe₂O₃ is lower than TiO₂ in conduction band positions. That is, if only co-catalysts locate on the right facets, the facet-induced effect acts in cooperation with the heterojunction-induced effect, thereby leading to an optimized separation efficiency of photogenerated electrons and holes. For the randomly loaded photocatalysts, in which some co-catalyst nanoparticles locate on the facets with an opposite effect, the recombination of photogenerated electrons and holes would become serious, because the heterojunction-induced effect is in conflict with the facet-induced effect.

Scheme 2 Schematic diagram of the proposed mechanism for the enhanced photocatalytic activity of Fe₂O₃–TiO₂–Pt under UV light.

4. Conclusions

A specific ternary heterojunction photocatalyst was designedly synthesized, in which two co-catalysts, α-Fe₂O₃ and Pt, were selectively loaded onto the photo-oxidative {001} and photo-reductive {101} facets, respectively, of TiO₂ NCs by utilizing the facet-dependent photochemical reactivity of TiO₂ NCs. Compared to the photocatalysts loaded randomly and without co-catalyst, this well-designed ternary photocatalyst, Fe₂O₃–TiO₂–Pt, exhibited superior photocatalytic activity (as high as 2.2 times and 30 times, respectively) and excellent recycling stability for H₂ production. The fundamental cause was that the separation of photogenerated charge carriers with the excitation of UV light were remarkably improved due to the synergy between the facet-induced and heterojunction induced effects. This photocatalyst engineering strategy, which was based on the selective deposition of co-catalysts, allows the intrinsic effects of specific facets and heterojunctions to be probed for their role in the enhanced photocatalytic performance of the composite photocatalysts. In addition, since the proposed enhancement mechanism is universal, the synthesized photocatalysts are expected to perform best in all photocatalytic processes.

Acknowledgements

This work was supported by the National Basic Research Program of China (2015CB932301), the National Natural Science Foundation of China (21473146 and 21333008) and the Fundamental Research Funds for the Central Universities (20720160026).

Notes and references

1. T. Hisatomi, J. Kubota and K. Domen, Chem. Soc. Rev., 2014, 43, 7520–7535.
2. H. Tong, S. Ouyang, Y. Bi, N. Umezawa, M. Oshikiri and J. Ye, Adv. Mater., 2012, 24, 229–251.
3. B. Oregan and M. Gratzel, Nature, 1991, 353, 737–740.
4. M. S. Dresselhaus, G. Chen, M. Y. Tang, R. Yang, H. Lee, D. Wang, Z. Ren, J.-P. Fleurial and P. Gogna, Adv. Mater., 2007, 19, 1043–1053.
5. C. K. Chan, H. Peng, G. Liu, K. McIlwrath, X. F. Zhang, R. A. Huggins and Y. Cui, Nat. Nanotechnol., 2008, 3, 31–35.
6. X. Chen, L. Liu, P. Y. Yu and S. S. Mao, Science, 2011, 331, 746–750.
7. C. Chen, W. Ma and J. Zhao, Chem. Soc. Rev., 2010, 39, 4206–4219.
8. R. Li, H. Han, F. Zhang, D. Wang and C. Li, Energy Environ. Sci., 2014, 7, 1369–1376.
9. R. Marschall, Adv. Funct. Mater., 2014, 24, 2421–2440.
10. W. Fan, Q. Zhang and Y. Wang, Phys. Chem. Chem. Phys., 2013, 15, 2632–2649.
11. K. Woan, G. Pyrgiotakis and W. Sigmund, Adv. Mater., 2009, 21, 2233–2239.
12. M. Murdoch, G. I. N. Waterhouse, M. A. Nadeem, J. B. Metson, M. A. Keane, R. F. Howe, J. Llorca and H. Idriss, Nat. Chem., 2011, 3, 489–492.
13. Y. Ohko, T. Tatsuma, T. Fujii, K. Naoi, C. Niwa, Y. Kubota and A. Fujishima, Nat. Mater., 2003, 2, 29–31.
14. S. Sakthivel, M. V. Shankar, M. Palanichamy, B. Arabindoo, D. W. Bahnemann and V. Murugesan, Water Res., 2004, 38, 3001–3008.
15. T. Ohno, K. Sarukawa and M. Matsumura, New J. Chem., 2002, 26, 1167–1170.
16. T. Tachikawa, S. Yamashita and T. Majima, J. Am. Chem. Soc., 2011, 133, 7197–7204.
17. J. Yu, J. Low, W. Xiao, P. Zhou and M. Jaroniec, J. Am. Chem. Soc., 2014, 136, 8839–8842.
18. M. D'Arienzo, M. V. Dozzi, M. Redaelli, B. D. Credico, F. Morazzoni, R. Scotti and S. Polizzi, J. Phys. Chem. C, 2015, 119, 12385–12393.
19. C. Liu, X. Han, S. Xie, Q. Kuang, X. Wang, M. Jin, Z. Xie and L. Zheng, Chem.–Asian J., 2013, 8, 282–289.
20. R. Li, F. Zhang, D. Wang, J. Yang, M. Li, J. Zhu, X. Zhou, H. Han and C. Li, Nat. Commun., 2013, 4, 1432.
21. L. Zhang, W. Wang, S. Sun, D. Jiang and E. Gao, Appl. Catal., B, 2015, 162, 470–474.
22. Q. Kuang, X. Zheng and S. Yang, Chem.–Eur. J., 2014, 20, 2637–2645.
23. P. Luan, M. Xie, X. Fu, Y. Qu, X. Sun and L. Jing, Phys. Chem. Chem. Phys., 2015, 17, 5043–5050.
24. K. E. deKrafft, C. Wang and W. Lin, Adv. Mater., 2012, 24, 2014–2018.
25. S. J. A. Moniz, S. A. Shevlin, X. An, Z.-X. Guo and J. Tang, Chem.–Eur. J., 2014, 20, 15571–15579.
26. T. Madrakian, A. Afkhami, R. Haryani and M. Ahmadi, Rsc Adv., 2014, 4, 44841–44847.
27. P. Luan, M. Xie, D. Liu, X. Fu and L. Jing, Sci. Rep., 2014, 4, 6180.
28. G. H. Du, Q. Chen, P. D. Han, Y. Yu and L. M. Peng, Phys. Rev. B: Condens. Matter Mater. Phys., 2003, 67, 035323.
29. L. Peng, T. Xie, Y. Lu, H. Fan and D. Wang, Phys. Chem. Chem. Phys., 2010, 12, 8033–8041.
30. G. H. Du, Q. Chen, P. D. Han, Y. Yu and L. M. Peng, Phys. Rev. B: Condens. Matter Mater. Phys., 2003, 67, 035323.
31. W. Zhou, H. Fu, K. Pan, C. Tian, Y. Qu, P. Lu and C.-C. Sun, J. Phys. Chem. C, 2008, 112, 19584–19589.
32. I. Jang, K.-E. You, Y. C. Kim and S.-G. Oh, Appl. Surf. Sci., 2014, 316, 187–193.
33. T. Yamashita and P. Hayes, Appl. Surf. Sci., 2008, 254, 2441–2449.
34. Y. Luo, J. Luo, J. Jiang, W. Zhou, H. Yang, X. Qi, H. Zhang, H. J. Fan, D. Y. W. Yu, C. M. Li and T. Yu, Energy Environ. Sci., 2012, 5, 6559–6566.
35. P. T. Hsieh, Y. C. Chen, K. S. Kao and C. M. Wang, Appl. Phys. A, 2008, 90, 317–321.
36. B. Palanisamy, C. M. Babu, B. Sundaravel, S. Anandan and V. Murugesan, J. Hazard. Mater., 2013, 252, 233–242.
37. Y. Wang, W. Du and Y. Xu, Langmuir, 2009, 25, 2895–2899.
38. Z. Li, J. Sheng, Y. Wang and Y. Xu, J. Hazard. Mater., 2013, 254, 18–25.
39. J. Pan, G. Liu, G. Q. Lu and H.-M. Cheng, Angew. Chem., Int. Ed., 2011, 50, 2133–2137.
40. Y.-F. Li, Z.-P. Liu, L. Liu and W. Gao, J. Am. Chem. Soc., 2010, 132, 13008–13015.

---

Footnote

† Electronic supplementary information (ESI) available: XRD pattern and SEM images of TiO₂ NCs; SEM, TEM images and XPS spectra of FeO_x(OH)_3−2x–TiO₂; SEM and TEM images of TiO₂–Pt; SEM and EDX analysis of n-Fe₂O₃–TiO₂ and n-Fe₂O₃–TiO₂–Pt NCs; XPS spectra of Fe₂O₃–TiO₂–Pt. See DOI: 10.1039/c6ra04552a

---