Boosting the photocatalytic H₂ evolution activity of Fe₂O₃ polymorphs (α-, γ- and β-Fe₂O₃) by fullerene [C₆₀]-modification and dye-sensitization under visible light irradiation

Abstract

C₆₀/Fe₂O₃ nanocomposites are successfully prepared, well characterized, and employed in visible-light-driven photocatalytic H₂ production. The Fe₂O₃ polymorphs show obvious broad-spectrum absorption, even close to the near infrared region (780–900 nm). The H₂ production rates of β-Fe₂O₃ and γ-Fe₂O₃ are almost 2.1 times and 3.1 times higher than α-Fe₂O₃ (which itself is close to that of g-C₃N₄). This demonstrates that carefully controlling the polymorphs can tune the photocatalysts' H₂ production properties. After modifying the Fe₂O₃ polymorphs with C₆₀, the sample with 0.5 wt% C₆₀/β-Fe₂O₃ has the optimum photocatalytic activity. This result indicates that the strength of the interaction and interfacial contact between C₆₀ and Fe₂O₃ polymorphs plays an important role in the enhancement of photocatalytic activity, which can improve the transmission efficiency of photogenerated electrons via a conjugated three-dimensional π system. Fluorescein is introduced as a photosensitizer and the optimum mass ratio of fluorescein + 0.5C₆₀/β-Fe₂O₃ is 1 : 1, which significantly boosts the photocatalytic H₂ evolution rate of 0.5C₆₀/β-Fe₂O₃ from 321.8 to 1665.0 μmol g⁻¹ h⁻¹. Meanwhile, the composites exhibit high stability and reusability.

---

Introduction

Solar hydrogen generation is a very attractive area of research. Being free from emission of pollutants and greenhouse gases, it allows the production of clean and renewable energy to compensate for the depletion of fossil fuels and represents a possible solution to environmental pollution and the emerging energy crisis.^1–4 In recent decades, several basic methods have been developed to improve the photocatalytic H₂ evolution properties of materials, such as doping with other materials,⁵ surface modification,⁶ formation of heterostructures,⁷ formation of composites,^8–10 and so on. However, the photocatalytic activity can also be regulated by carefully controlling the phase composition. Due to its well-suited band gap, natural abundance, nontoxicity, environmental compatibility, low cost, and chemical stability, α-Fe₂O₃ has attracted considerable interest for potential applications.^11–13 Iron oxide has four crystal phases: α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, and ε-Fe₂O₃.¹⁴ According to the literature, α-Fe₂O₃ has a corundum structure, while γ-Fe₂O₃ has a defect spinel structure.^15,16 β-Fe₂O₃ has a bixbyite structure.¹⁷ Compared with α-Fe₂O₃, information on the photocatalytic H₂ production rate and synthesis of γ-Fe₂O₃ and β-Fe₂O₃ are still scarce. Therefore, motivated by this situation that few studies have focused on the Fe₂O₃ polymorphs, (α-, γ-, β-Fe₂O₃) are obtained by different synthesis methods. Furthermore, the pure α-Fe₂O₃ material still suffers from unsatisfactory photocatalytic efficiency owing to the high recombination rate of photogenerated electrons and holes and its low absorption coefficient.¹⁸

A suitable cocatalyst to modify the metallic oxide can increase the transmission efficiency of photogenerated electrons and provide additional active sites for photocatalytic H₂ evolution.^19–22 Fullerene [C₆₀] was discovered in 1985 by Kroto et al.²³ The C₆₀ fusion structure (diameter 0.7 nm) shows stability, anti-compression performance, high electron affinity, and lower material cost, potentially making it advantageous for application in photocatalytic H₂ evolution.²⁴ However, relatively few studies have investigated the use of fullerene as a cocatalyst when compared to the number of studies using highly priced precious metal cocatalysts. It may be difficult to form stronger interactions and interface contacts by using C₆₀-modified semiconductors.

Dyes usually do not show photocatalytic H₂ production activity due to the fast recombination of the photogenerated electron–hole pairs under visible light illumination despite their strong ability to absorb visible light.^25–27 However, the photocatalytic H₂ evolution activity of a photocatalyst is greatly increased under visible light illumination if dye molecules are adsorbed on the surface of the photocatalyst.^28–30 Recently, Schmuttenmaer et al. proposed that the overall efficiencies of water-splitting dye-sensitized photoelectron chemical cells (WS-DSPECs) remain low in large part because of poor electron injection into the conduction band (CB) of the oxide support.³¹ The electrons transfer from the LUMO of the dye to the CB of the oxide support. The dye's HOMO must be more positive than the sacrificial reagent's redox potential and its LUMO must be more negative than the semiconductor's CB.³² Therefore, present research knowledge of the photocatalytic H₂ evolution activity of dye-sensitized semiconductors is still insufficient. Furthermore, many metal complex dyes are not suitable for the potential practical applications of capturing and converting solar energy because of their high cost. Fortunately, metal-free organic dyes, such as fluorescein, exhibit low cost, low toxicity, absorption in the visible light region, and high electron injection rates, making them suitable to enhance photocatalytic H₂ evolution.^33–35

Herein, we report the facile synthesis and detailed characterization of C₆₀/Fe₂O₃ composites. Iron(III) oxide polymorphs exhibit significantly different photocatalytic H₂ evolution properties. The fluorescein-sensitized 0.5C₆₀/β-Fe₂O₃ composites deliver an extremely high solar-driven hydrogen production rate (1665.0 μmol g⁻¹ h⁻¹) under visible light illumination (λ > 420 nm). Fluorescein is used as a photosensitizer. Fullerene [C₆₀] is exploited to improve the transmission efficiency of photogenerated electrons across the interface of the semiconductor.

Experimental section

Reagents

Ferric nitrate (Fe(NO₃)₃·9H₂O, 99.9%), citric acid (C₆H₈O₇, 99.5%), urea (CH₄N₂O, 99.5%), ferric chloride (FeCl₃·6H₂O, 99.0%), ethanediol (C₂H₆O₂, 98.0%), methylbenzene (C₇H₈, 99.8%), triethanolamine (C₆H₁₅NO₃, 99.0%), ammonia (NH₃, 25.0%), and fullerene (C₆₀, 99.9%) were purchased from Aladdin Co. Ltd. The materials were all utilized without further purification and were all analytically pure. Deionized water was utilized throughout the experiments. Pure g-C₃N₄ was synthesized directly by calcination of melamine at 550 °C for 4 h at a heating rate of 10 °C min⁻¹.

Synthesis of the α-Fe₂O₃ nanoporous material

The α-Fe₂O₃ nanoporous material was obtained by the sol–gel method. 16.16 g Fe(NO₃)₃·9H₂O was dissolved in 6.8 mL deionized water to prepare metal ion solution A, and 2.56 g citric acid was dissolved in 4.0 mL ethanediol to prepare the clear and transparent solution B. Next, solution A was dropwise added to solution B to prepare solution C (with a metal ion concentration of 2.5–3.5 mol L⁻¹) under vigorous magnetic stirring. NH₃·H₂O was used to regulate the pH value of solution C within the range 3–4 and a gel was formed at 70 °C by continuously stirring for 4 h. After gel formation, the gel was dried under vacuum at 70 °C for 10 h to form a xerogel. Finally, the xerogels were heat-treated at 900 °C for 1 h under air atmosphere.

Synthesis of the γ-Fe₂O₃ nanoparticles

20.00 g Fe(NO₃)₃·9H₂O was completely dissolved in 40 mL ethanol under magnetic stirring. 19.60 g urea was added to the Fe(NO₃)₃·9H₂O ethanol solution. The ethanol solution was continually stirred for 2.5 h. After urea was added, the solution became light green and a light green powder was observed to grow on the bottom of the vessel. The light green powder was washed with anhydrous ethanol. After filtration, the powder (Fe(CON₂H₄)₆]·(NO₃)₃) was dried at room temperature in a Petri dish with a filter paper cover. Finally, the powder was ground and heat-treated at 200 °C for 1 h under air atmosphere.

Synthesis of low-dimensional tablet of β-Fe₂O₃

A low-dimensional tablet made from β-Fe₂O₃ was synthesized by a hydrothermal reaction. At room temperature, 1.35 g FeCl₃·6H₂O and 0.3 g urea were slowly added to 10 mL deionized water. Then, 2.6 mL NH₃·H₂O was added to the clear and transparent solution under vigorous magnetic stirring. Finally, the mixture solution was put in a hydrothermal cell and heated for 6 h at 150 °C in the oven. Then, the obtained solid was washed with ethanol followed by deionized water and dried at 70 °C.

Synthesis of the C₆₀/Fe₂O₃ polymorphs

The C₆₀-modified Fe₂O₃ nanocomposites were obtained as follows: a known amount of C₆₀ was dissolved completely in 30 mL toluene by sonication for 1 h. Next, to the solution was added the as-synthesized powder (ca. 0.3 g), which was vigorously magnetically stirred for 6 h at ambient temperature. Finally, an opaque powder was obtained by volatilization of the toluene. The opaque powder was washed with ethanol followed by deionized water and dried for 12 h at 80 °C in a vacuum drying oven. In addition, samples synthesized with different mass ratios (wt%) of C₆₀/Fe₂O₃ (1C₆₀/α-Fe₂O₃, 1C₆₀/γ-Fe₂O₃, 1C₆₀/β-Fe₂O₃, 0.25C₆₀/β-Fe₂O₃, 0.5C₆₀/β-Fe₂O₃, 2C₆₀/β-Fe₂O₃) were prepared by following the same method as above.

Characterization

To characterize the structural variation of the materials, X-ray diffraction (XRD) patterns (Bruker D8 Advance) were obtained using graphite monochromatized Cu-Kα (λ = 1.5406 Å) radiation in the 2θ range from 5° to 80°. Fourier transform infrared (FTIR) spectra of the materials were measured on a Bruker Tensor 27 spectrometer with the KBr tabletting method. Raman spectra were measured by a micro-Raman spectrometer (Renishaw InVia) in the backscattering geometry using an Ar ion laser (532 nm) as an excitation source at room temperature. The Brunauer–Emmett–Teller (BET) specific surface area was determined by a surface area and porosity analyzer (ASAP 2020, American) based on nitrogen adsorption and desorption isotherms measured at 77 K. The degasification temperature and the drying time of the samples were 200 °C and 2 h, respectively. Microstructural characterization and the elemental mapping of desired regions were performed by field-emission scanning electron microscopy (FESEM, Hitachi, S-4800) and energy dispersive X-ray analysis (EDX), respectively. Transmission electron microscopy (TEM) images were collected on an FEI Tecnai G2 F20 TEM. The UV-visible diffuse reflectance spectra (DRS, Hitachi U-3010 spectrophotometer) were obtained by the Kubelka–Munk approach with BaSO₄ as a reference at room temperature. The room-temperature magnetization curves were measured by means of a vibrating sample magnetometer (VSM, Lakeshore Model 7304). X-ray photoelectron spectroscopy (XPS, Axis Ultra DLD, Kratos) measurements were carried out with a monochromatic X-ray source (Al Kα, 15 kV, 200 W).

Photocatalytic H₂ production

Photocatalytic experiments for H₂ evolution were carried out in a Pyrex reaction cell connected to a closed gas evacuation and circulation system. The composites (0.005 g) were sonicated for 5 min in a 78 mL triethanolamine aqueous solution (10 vol% TEOA, sacrificial reagent). Then the aqueous suspension was degassed for 1 h and irradiated by a 300 W Xe lamp with an ultraviolet cut-off filter (λ > 420 nm) (PLS-SXE300, Trusttech). A flow of cooling water was used to maintain the reaction temperature at room temperature. The content of H₂ generated was determined by online gas chromatography (GC7900, Tian Mei, Shanghai) equipped with a 5 Å molecular sieve column and a thermal conductivity detector (TCD) using nitrogen as the carrier gas, as shown in Fig. S1.†

Photoelectrochemical measurements

The electrochemical measurements (CHI660C, Electrochemical Instruments) utilized a standard three-electrode system with Na₂SO₄ (1 mol L⁻¹) aqueous solution as the electrolyte. Ag/AgCl (saturated KCl) and platinum flake were used as the reference electrodes and counter electrodes, respectively. The sample electrodes separately served as the working electrode, which was obtained by using the doctor blade coating method to deposit the suspensions onto indium tin oxide (ITO). The visible light source was a 300 W Xe lamp with an ultraviolet cut-off filter (λ > 420 nm) (PLS-SXE300, Trusttech). The working electrodes were prepared as follows: the nanocomposite (0.05 g) was ground with 0.5 mL deionized water and 0.02 g polyethylene glycol to make a slurry. Next, the slurry was coated onto ITO glass electrodes and these electrolytes were dried at 120 °C for 2 h.

Results and discussion

Crystal structure and composition

By comparison with reference inorganic XRD patterns, it was confirmed that the desired Fe₂O₃ polymorphs (α-Fe₂O₃ (Fig. S2(a), γ-Fe₂O₃ (b), β-Fe₂O₃ (c)†)) had been obtained. It was found that β-Fe₂O₃ and α-Fe₂O₃ had similar diffraction peaks. However, those of γ-Fe₂O₃ were significantly different, suggesting a very different crystal structure for this polymorph. According to the results of the Scherrer equation shown in Table S1,† the particle size of α-Fe₂O₃ was 47.10 nm, of γ-Fe₂O₃ was 27.87 nm, and of β-Fe₂O₃ was 42.48 nm. As shown in Fig. 1, the XRD patterns of the 0.5C₆₀/β-Fe₂O₃ (wt%) and 1C₆₀/β-Fe₂O₃ (wt%) samples were unchanged from that of β-Fe₂O₃, which shows that the adsorption of C₆₀ did not affect the lattice structure of β-Fe₂O₃. Due to the minimal C₆₀ content, no XRD diffraction peak of C₆₀ was observed. The grinding method was used to mix fluorescein with 0.5C₆₀/β-Fe₂O₃. In comparison with 0.5C₆₀/β-Fe₂O₃, the XRD diffraction spectrum of the fluorescein + 0.5C₆₀/β-Fe₂O₃ sample shows additional diffraction peaks corresponding to the characteristic peaks of fluorescein.

Fig. 1 X-ray diffraction patterns of C₆₀, fluorescein, 0.5C₆₀/β-Fe₂O₃, 1C₆₀/β-Fe₂O₃, and Fl + 0.5C₆₀/β-Fe₂O₃ (fluorescein + 0.5C₆₀/β-Fe₂O₃ at a mass ratio of 1 : 1).

Fig. S3† depicts the FTIR spectra of the Fe₂O₃ polymorphs. It shows several bands at 462, 560, 1383, 1680, 2341, and 3430 cm⁻¹. The Fe–O–Fe stretching vibration is responsible for the peaks at 462 and 560 cm⁻¹,³⁶ but the intensities of this stretching vibration are not the same for all samples, which implies that the different synthetic methods produced different Fe₂O₃ polymorphs. Fig. S4 and S5† present the FTIR spectra of the C₆₀-modified β-Fe₂O₃ material and fluorescein-sensitized 0.5C₆₀/β-Fe₂O₃ composites. As can be seen, no prominent infrared absorption peaks were exhibited by pure C₆₀, but the peak intensity of β-Fe₂O₃ increased after modification with C₆₀. The fluorescein component led to four additional absorption peaks at 1114 cm⁻¹, 1260 cm⁻¹, 1460 cm⁻¹, and 1593 cm⁻¹, which is consistent with the infrared spectrum of pure fluorescein. However, there was no enhancement of the peak intensity of 0.5C₆₀/β-Fe₂O₃. This fact indicates that C₆₀ and fluorescein have different effects on β-Fe₂O₃-based systems.

Fig. 2(a) shows that the α-Fe₂O₃ and β-Fe₂O₃ samples have similar Raman spectra. For the Fe–O stretching vibration, the primary characteristic peaks appear at 221, 240, 288, 290, 406, 410, and 608 cm⁻¹. However, the peaks at 288 and 406 cm⁻¹ for α-Fe₂O₃ are at slightly different wavenumbers than the corresponding peaks at 290 and 410 cm⁻¹ for β-Fe₂O₃. Moreover, the characteristic peaks of γ-Fe₂O₃ are at 660 and 724 cm⁻¹. The vibration frequencies of the Hg(7), Ag(2), and Hg(8) modes for pristine C₆₀ are 1412, 1460, and 1560 cm⁻¹, respectively, in agreement with the previous report.³⁷ Aside from a major band (Ag(2)) at 1460 cm⁻¹, it was noticeable that the other bands for C₆₀ disappeared (Fig. 2(b)). It is known that C₆₀ can adsorb on the surface of β-Fe₂O₃ in composite materials.³⁸

Fig. 2 Raman spectra of Fe₂O₃ polymorphs (a); Raman spectra of C₆₀, β-Fe₂O₃, and 0.5C₆₀/β-Fe₂O₃ (b).

As expected, the O1s spectrum can be decomposed into two peaks as shown in Fig. 3(b). The peaks for lattice oxygen and chemisorbed oxygen (hydroxyl groups) are at 529.5 and 530.8 eV, respectively.³⁹ As shown in Fig. 3(c), the main peak at 284.6 eV is ascribed to sp²-hybridized carbon and adventitious carbon from the C₆₀. Defect-containing sp²-hybridized carbon and carboxyl carbon (O=C=O) lead to the peaks located at 286.0 and 288.4 eV, respectively.⁴⁰ The spectrum of Fe2p (Fig. 3(d)) shows two peaks at 710.6 and 724.1 eV (with a splitting energy of 13.5 eV), which is very close to the reported signal.⁴¹ This result shows that Fe is in the trivalent state in the 0.5C₆₀/β-Fe₂O₃ composites.

Fig. 3 XPS analysis of the 0.5C₆₀/β-Fe₂O₃ sample. Full-scale XPS spectrum (a), high resolution XPS spectra of O1s (b), C1s (c), and Fe2p (d), respectively.

Morphology

The scanning electron micrograph of the α-Fe₂O₃ material shows a porous structure, which facilitates the photocatalytic H₂ evolution (Fig. 4(a) and (b)). As shown in Fig. 4(c) and (d), the morphology of γ-Fe₂O₃ consists of a small aggregation of grains or irregular particles. Fig. 4(e) and (f) show the high-magnified images of the β-Fe₂O₃ composite. The growth of the material in a uniform low-dimensional form was unexpected. A good dispersion of particles can supply more reactive sites for the photocatalytic H₂ evolution than aggregated particles can.⁴² The uniform shape and low-dimensional form of the β-Fe₂O₃ composite did not change after C₆₀ modification (Fig. 4(g)). In addition, the EDS (Fig. 4(h)) and element mapping from SEM (Fig. 4(i)) indicate the presence of Fe, O, and C.

Fig. 4 SEM images of the as-prepared composites: α-Fe₂O₃ ((a), (b)), γ-Fe₂O₃ ((c), (d)), β-Fe₂O₃ ((e), (f)), and 0.5C₆₀/β-Fe₂O₃ (g); (h) EDS spectrum of 0.5C₆₀/β-Fe₂O₃ sample; (i) elemental mapping pattern of 0.5C₆₀/β-Fe₂O₃ sample.

TEM and HRTEM analyses were performed to further investigate the shape and the lattice properties of the β-Fe₂O₃ and 0.5C₆₀/β-Fe₂O₃ samples. Consistent with the FESEM observations, Fig. 5(a) and (c) show that β-Fe₂O₃ did indeed grow in a uniform low-dimensional form. The TEM (HRTEM) images of β-Fe₂O₃ (Fig. 5(e)) show that the samples are highly crystalline. The lattice spacing was measured as d = 0.249 nm, which is consistent with the (110) plane of β-Fe₂O₃ (JCPDS no. 89-2810). As shown in Fig. 5(b) and (d), it was found that the crystal structure of β-Fe₂O₃ was not affected by C₆₀ adsorption, which is consistent with the other characterization results. However, the outer boundary of the β-Fe₂O₃ material became distinctly different upon C₆₀ adsorption (Fig. 5(f)). It is clear that during the adsorption process, the two phases formed an intimate heterogeneous interface, which has a beneficial effect on the transmission efficiency of photogenerated electrons.⁴³

Fig. 5 TEM images of the as-prepared samples: (a) β-Fe₂O₃, (c) magnification of β-Fe₂O₃, and (e) high-resolution TEM image of β-Fe₂O₃; (b) 0.5C₆₀/β-Fe₂O₃, (d) magnification of 0.5C₆₀/β-Fe₂O₃, and (f) high-resolution TEM image of 0.5C₆₀/β-Fe₂O₃.

BET specific surface area and pore textural analysis

The specific surface areas of α-Fe₂O₃, γ-Fe₂O₃, and β-Fe₂O₃ respectively were 1.5418 m² g⁻¹, 4.7966 m² g⁻¹, and 13.6316 m² g⁻¹ (Table S1†). After modification, the surface areas of α-Fe₂O₃ and γ-Fe₂O₃ increased, which not only facilitates the transfer of photogenerated electrons, but also provides additional active surface sites.⁴⁴ However, the instrument was unable to detect the corresponding pore volume and pore size because the surface areas of α-Fe₂O₃ and γ-Fe₂O₃ are too small. The pore volume and pore size of β-Fe₂O₃, respectively, were 0.03114 cm³ g⁻¹ and 9.1375 nm. Compared with β-Fe₂O₃, the C₆₀-modified β-Fe₂O₃ samples have smaller BET specific surface areas. This is because the introduction of C₆₀ molecules occupied some of the space and filled some of the pores of β-Fe₂O₃.

Optical absorption and magnetic properties

The Fe₂O₃ polymorph materials obtained by different methods exhibit strong absorption of visible light and even the closest part of the near-infrared region (780–900 nm) (Fig. 6(a)). This excellent light absorption performance is better than that of any existing single photocatalyst at present, and is conducive to the conversion and application of solar energy. As shown in Fig. S6,† the band gaps of α-Fe₂O₃, γ-Fe₂O₃, and β-Fe₂O₃ were respectively 1.60 eV, 1.61 eV, and 1.91 eV. Primarily due to C₆₀ being a visible light photosensitizer, the visible light absorption intensity of β-Fe₂O₃ material is known to be increased by C₆₀.⁴⁵ Furthermore, β-Fe₂O₃ itself already has better light absorption properties than pure fluorescein. Therefore, the light absorption strength of the 0.5C₆₀/β-Fe₂O₃ sample was not further improved by the inclusion of fluorescein molecules. The effects of fluorescein and C₆₀ on the β-Fe₂O₃ composite are thus notably different.

Fig. 6 UV-vis diffuse reflectance spectra of the as-prepared samples (a); in-plane measurements of the magnetic properties of the as-prepared samples (b).

The magnetization curve of the as-prepared sample at room temperature is shown in Fig. 6(b). The magnetization values obtained at 2 × 10⁴ Oe are approximately 0.11, 131.53, 1.34, and 0.48 emu g⁻¹ for α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, and 0.5C₆₀/β-Fe₂O₃, respectively. As can be seen, the nano-sized γ-Fe₂O₃ exhibits strong magnetic properties at room temperature. The magnetic strength order of Fe₂O₃ polymorphs is γ-Fe₂O₃ > β-Fe₂O₃ > α-Fe₂O₃. Furthermore, the magnetic properties of the as-prepared samples allow them to be easily separated from the aqueous solutions compared with other, nonmagnetic photocatalysts, which facilitates their practical application as photocatalysts.

Photocatalytic H₂ production and stability

The photocatalytic H₂ evolution ability of the C₆₀/Fe₂O₃ and fluorescein-sensitized 0.5C₆₀/β-Fe₂O₃ composites were compared in 10 vol% triethanolamine (TEOA) aqueous solution under visible light illumination (λ > 420 nm), as shown in Fig. 7. As shown in Fig. 7(a), the photocatalytic H₂ evolution rate of α-Fe₂O₃ under these conditions was 80.6 μmol h⁻¹ g⁻¹, that of γ-Fe₂O₃ was 252.8 μmol h⁻¹ g⁻¹, and that of β-Fe₂O₃ was 169.4 μmol h⁻¹ g⁻¹. The H₂ production rates of β-Fe₂O₃ and γ-Fe₂O₃ are almost 2.1 and 3.1 times higher than α-Fe₂O₃ (which itself is close to that of g-C₃N₄), respectively. The different light absorbing characteristics, specific surface areas, and particle sizes are responsible for the different photocatalytic H₂ evolution activities of the Fe₂O₃ polymorphs. The photocatalytic H₂ evolution activity of all the Fe₂O₃ polymorphs was improved by C₆₀ modification. This result indicated that the intense interaction between C₆₀ and Fe₂O₃ polymorphs played a key role in the improvement of photocatalytic activity, by improving the transmission efficiency of photogenerated electrons. Therefore, the 0.5C₆₀/β-Fe₂O₃ sample exhibits a more efficient H₂ release rate than pure β-Fe₂O₃. Next, the β-Fe₂O₃ materials modified with different contents of C₆₀ as cocatalyst were evaluated for photocatalytic H₂ evolution under visible light irradiation (Fig. S7†). With the addition of C₆₀ in amounts below 0.5 wt%, the rates of H₂ evolution increase with the C₆₀ content. In contrast, the rate of H₂ production decreases with increasing C₆₀ content when the C₆₀ is present at more than 0.5 wt%. This may be because excess C₆₀ increased photon absorption and scattering.⁴⁶ Thus, the 0.5C₆₀/β-Fe₂O₃ sample is the best performing photocatalyst. However, because C₆₀ does not have the ability to produce hydrogen under visible light irradiation by itself, its presence in the β-Fe₂O₃ material may limit the photocatalytic H₂ evolution activity of the 0.5C₆₀/β-Fe₂O₃ composite. This drawback was effectively overcome by the fluorescein-sensitized 0.5C₆₀/β-Fe₂O₃ sample. As shown in Fig. 7(b) and S8,† the Fl + 0.5C₆₀/β-Fe₂O₃ (1 : 1) sample is the best-performing of all the tested catalysts with an H₂ production rate of 1665.0 μmol g⁻¹ h⁻¹. This implies that fluorescein molecules are adsorbed on the surface of the low-dimensional β-Fe₂O₃ material, thus widening the electron transport channels.^47,48 As shown in Fig. 8(a), to evaluate the cycle stability of this sample, repeat cycles of photocatalytic H₂ production (each cycle lasting 3 h in an experimental vacuum) were performed. The rate of photocatalytic H₂ evolution was not observed to reduce after six consecutive cycles, indicating that the Fl + 0.5C₆₀/β-Fe₂O₃ composite has good cycle stability. Its structural stability can be clearly seen from the XRD diffraction peaks and infrared spectrum peaks (Fig. S9†).⁴⁹

Fig. 7 Photocatalytic activity of the investigated samples in an aqueous solution containing 10 vol% TEOA as the sacrificial agent under visible light illumination: (a) photocatalytic H₂ evolution rates of the as-synthesized samples (irradiation time 3 h); (b) H₂ evolution rates of β-Fe₂O₃; 1C₆₀/β-Fe₂O₃; 0.5C₆₀/β-Fe₂O₃; fluorescein + 0.5C₆₀/β-Fe₂O₃ at a mass ratio of 1 : 1; and fluorescein + β-Fe₂O₃ at a mass ratio of 1 : 1 in a 70 mL aqueous solution with 7.8 mL sacrificial agent (irradiation time = 3 h).

Fig. 8 Cycling test of photocatalytic H₂ generation for 0.5C₆₀/β-Fe₂O₃ (wt%) with fluorescein (0.005 g) and 10 vol% TEOA for photocatalytic hydrogen production (each cycle lasting 3 h in an experimental vacuum) (a); steady-state photoluminescence (PL) spectra of α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, 0.5C₆₀/β-Fe₂O₃, and Fl + 0.5C₆₀/β-Fe₂O₃ composites (b); transient photocurrent response of α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, 0.5C₆₀/β-Fe₂O₃, and Fl + 0.5C₆₀/β-Fe₂O₃ composites in 1 mol L⁻¹ Na₂SO₄ aqueous solution under visible light irradiation ((c and d)).

Clarification of the mechanism

The emission spectra of the α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, 0.5C₆₀/β-Fe₂O₃, and Fl + 0.5C₆₀/β-Fe₂O₃ composites have similar broad emission bands, but the samples differ in the intensity of emissions (Fig. 8(b)). When the intensity of emissions is higher, it can be concluded that the recombination of the photogenerated electron–hole pairs is increased. The results show that C₆₀ improved the transmission efficiency of photogenerated electrons, thus reducing the recombination of the photogenerated electron–hole pairs. Furthermore, the emission intensity of the Fl + 0.5C₆₀/β-Fe₂O₃ composite is lower than 0.5C₆₀/β-Fe₂O₃, which is consistent with the greater H₂ production rate of the former. To provide further detail, the transient photocurrent responses were recorded for α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, and 0.5C₆₀/β-Fe₂O₃ sample photoelectrodes for several on–off cycles under visible light irradiation, as shown in Fig. 8(c). The higher the photocurrent value, the higher the separation efficiency of photogenerated electrons–holes. The photocurrents quickly increased to a certain value after the lamp was turned on, and quickly returned to zero when the lamp was turned off. This method was repeated five times. As can be seen, the photocurrent values of α-Fe₂O₃, γ-Fe₂O₃, β-Fe₂O₃, and 0.5C₆₀/β-Fe₂O₃ are consistent with the rate of photocatalytic H₂ evolution. However, upon adding fluorescein, the current shape of the 0.5C₆₀/β-Fe₂O₃ sample changes (Fig. 8(d)). The photocurrent value does not return to zero when the lamp is switched off. Under visible light irradiation, fluorescein molecules are excited to generate electrons in the LUMO, and these electrons migrate from fluorescein to the CB of β-Fe₂O₃. Finally, via a long, three-dimensionally conjugated π bond, the fluorescein electrons are transported to the ITO support by C₆₀. Some of these electrons may still be traveling along the transmission path when the visible light is switched off. This would explain the observed photocurrent behavior.

On the basis of the above experiments, proposed transfer mechanisms of the photo-generated electrons in the fluorescein-sensitized 0.5C₆₀/β-Fe₂O₃ composites are shown in Fig. 9. By linearly extrapolating the leading edge of the valence band (VB) of the XPS of β-Fe₂O₃ to the base line (Fig. S10†), its VB level was determined to be +1.39 eV. The band gap of β-Fe₂O₃ was determined to be 1.91 eV, with the corresponding CB position at −0.52 eV. According to previous reports, the LUMO of fluorescein is more negative than the CB of β-Fe₂O₃, while the HOMO of fluorescein is more positive than the redox potential of TEOA.⁵⁰ Under visible light irradiation, fluorescein and β-Fe₂O₃ are excited to produce electrons (e⁻) in the LUMO of fluorescein and the CB of the β-Fe₂O₃ composite, leaving corresponding holes (h⁺) in the HOMO and VB. The fluorescein electrons are then injected into the CB of β-Fe₂O₃. After this, the electrons tend to move rapidly towards the surface of the C₆₀ via the intimate heterogeneous interface, which has a positive effect on the transmission efficiency of the photogenerated electrons. Finally, the electrons react with H⁺ to generate H₂ on the C₆₀ surface. Due to the unique electronic properties of C₆₀, containing a highly three-dimensionally delocalized π electron system, it is advantageous for accelerating electron transfer, thereby increasing the rate of H₂ evolution. In the meantime, the fluorescein and β-Fe₂O₃, which accumulate holes, can react with TEOA. Fig. S11 and S12† summarize the specific H₂ evolution process.

Fig. 9 Schematic illustration of the possible mechanism of charge separation and transfer over Fl + 0.5C₆₀/β-Fe₂O₃ composite. Also shown are B3LYP/6-31G calculated molecular orbital amplitude plots of the HOMOs and LUMOs of fluorescein.

Conclusions

In summary, iron(III) oxide polymorphs are respectively obtained by different synthetic methods, which are easier, cheaper, and consume less energy than alternative methods. Even close to the near infrared region (780–900 nm), the Fe₂O₃ polymorphs clearly show broad visible light absorption. The magnetic properties of materials based on these Fe₂O₃ polymorphs are favorable for potential practical application in photocatalytic H₂ evolution. The particle size is probably the most important factor among the variety of factors found to affect the H₂ production activity of Fe₂O₃ polymorphs. The 0.5C₆₀/β-Fe₂O₃ (wt%) sample has the optimum photocatalytic activity among the C₆₀-modified Fe₂O₃ polymorphs. With conjugated three-dimensional π systems, formation of interfacial contacts, and strong C₆₀–Fe₂O₃ interactions, the C₆₀-modified Fe₂O₃ polymorphs show improved transmission efficiency of photo-generated electrons. If fluorescein is introduced as a photosensitizer, the optimum mass ratio of fluorescein + 0.5C₆₀/β-Fe₂O₃ is 1 : 1, which significantly boosts the photocatalytic H₂ evolution of 0.5C₆₀/β-Fe₂O₃ (321.8 μmol h⁻¹ g⁻¹) to 1665.0 μmol g⁻¹ h⁻¹. Moreover, the composites remain highly stable and can withstand repeated use. We hope that this research will attract close attention to the use of Fe₂O₃ polymorphs for the potential practical applications of capturing and converting solar energy.

Acknowledgements

We are extremely grateful to the research fund of the Key Laboratory of Fuel Cell Technology of Guangdong Province, and the National Natural Science Foundation of China (No. 21571064, 21371060) for financial support. We thank Benjamin J. Deibert for proof-reading this article.

Notes and references

1. A. Fujishima and K. Honda, Nature, 1972, 238, 37.
2. X. Chen, S. Shen, L. Guo and S. S. Mao, Chem. Rev., 2010, 110, 6503.
3. M. S. Dresselhaus and I. L. Thomas, Nature, 2001, 414, 332.
4. T. Song, L. Zhang, P. Y. Zhang, J. Zeng, T. T. Wang, A. Ali and H. P. Zeng, J. Mater. Chem. A, 2017, 5, 6013.
5. J. R. Ran, T. Y. Ma, G. P. Gao, X. W. Du and S. Z. Qiao, Energy Environ. Sci., 2015, 8, 3708.
6. Y. Choi, H. Kim, G. H. Moon, S. W. Jo and W. Y. Choi, ACS Catal., 2016, 6, 821.
7. Y. P. Xie, Z. B. Yu, G. Liu, X. L. Ma and H. M. Cheng, Energy Environ. Sci., 2014, 7, 1895.
8. K. Maeda, M. Eguchi and T. Oshima, Angew. Chem., Int. Ed., 2014, 53, 13164.
9. X. Wang, Q. Xu, M. R. Li, S. Shen and C. Li, Angew. Chem., Int. Ed., 2012, 51, 13089.
10. Y. Liao, W. Que, Q. Jia, Y. He, J. Zhang and P. Zhong, J. Mater. Chem., 2012, 22, 7937.
11. J. X. Zhu, Z. Y. Yin, D. Yang, T. Sun, H. Yu, H. E. Hoster, H. H. Hng, H. Zhang and Q. Y. Yan, Energy Environ. Sci., 2013, 6, 987.
12. G. Y. Rao, Q. Y. Zhang, H. L. Zhao, J. T. Chen and Y. Li, Chem. Eng. J., 2016, 302, 633.
13. S. Sakurai, A. Namai, K. Hashimoto and S. Ohkoshi, J. Am. Chem. Soc., 2009, 131, 18299.
14. G. Carraro, D. Barreca and P. Fornasiero, Adv. Funct. Mater., 2014, 24, 372.
15. K. F. Zhao, H. L. Tang, B. T. Qiao, L. Li and J. H. Wang, ACS Catal., 2015, 5, 3528.
16. L. Q. Guo, F. Chen, X. Q. Fan, W. D. Cai and J. L. Zhang, Appl. Catal., B, 2010, 96, 162.
17. A. Rathi, M. B. Gawande, J. R. Pechousek and R. Zborila, Green Chem., 2016, 18, 2363.
18. M. Mishra and D. M. Chun, Appl. Catal., A, 2015, 498, 126.
19. I. Paramasivam, H. Jha, N. Liu and P. Schmuki, Small, 2012, 8, 3073.
20. R. B. Jiang, B. X. Li, C. H. Fang and J. F. Wang, Adv. Mater., 2014, 26, 5274.
21. Z. k. Zheng, T. Tachikawa and T. Majima, J. Am. Chem. Soc., 2015, 137, 948.
22. H. B. Fu, T. G. Xu, S. B. Zhu and Y. F. Zhu, Environ. Sci. Technol., 2008, 42, 8064.
23. H. W. Kroto, J. R. Heath, S. C. O. Brien, R. F. Curl and R. E. Smalley, Nature, 1985, 318, 162.
24. S. E. Zhu, F. Li and G. W. Wang, Chem. Soc. Rev., 2013, 42, 7535.
25. Y. Bai, I. MoraSeró, F. D. Angelis, J. Bisquert and P. Wang, Chem. Rev., 2014, 114, 10095.
26. R. Abe, K. Shinmei, N. Koumura, K. Hara and B. Ohtani, J. Am. Chem. Soc., 2013, 135, 16872.
27. J. R. Swierk and T. E. Mallouk, Chem. Soc. Rev., 2013, 42, 2357.
28. A. Kruth, S. Hansen, T. Beweries, V. Brüser and K. D. Weltmann, ChemSusChem, 2013, 6, 152.
29. T. Toyao, M. Saito, S. Dohshi, K. Mochizuki, M. Iwata, H. Higashimura, Y. Horiuchi and M. Matsuoka, Chem. Commun., 2014, 50, 6779.
30. X. H. Zhang, U. Veikko, J. Mao, P. Cai and T. Y. Peng, Chem.–Eur. J., 2012, 18, 12103.
31. J. R. Swierk, N. S. McCool, C. T. Nemes, T. E. Mallouk and C. A. Schmuttenmaer, J. Phys. Chem. C, 2016, 120, 5940.
32. X. H. Zhang, T. Y. Peng and S. S. Song, J. Mater. Chem. A, 2016, 4, 2365.
33. D. P. Wu, K. Cao, F. J. Wang, H. J. Wang, Z. Y. Gao, F. Xu, Y. M. Guo and K. Jiang, Chem. Eng. J., 2015, 280, 441.
34. C. Kong, S. X. Min and G. X. Lu, ACS Catal., 2014, 4, 2763.
35. Y. P. Yuan, L. S. Yin, S. W. Cao, G. S. Xu, C. H. Li and C. Xue, Appl. Catal., B, 2015, 168, 572.
36. H. Y. Zhu, R. Jiang, L. Xiao and W. Li, J. Hazard. Mater., 2010, 179, 251.
37. X. J. Bai, L. Wang, Y. J. Wang, W. Q. Yao and Y. F. Zhu, Appl. Catal., B, 2014, 152, 262.
38. J. P. Huo and H. P. Zeng, J. Mater. Chem. A, 2015, 3, 6258.
39. A. E. Kandjani, Y. M. Sabri, S. R. Periasamy, A. Nafady and S. K. Bhargava, Langmuir, 2015, 31, 10922.
40. B. K. Vijayan, N. M. Dimitrijevic, D. F. Shapiro, J. S. Wu and K. A. Gray, ACS Catal., 2012, 2, 223.
41. P. M. Rao and X. L. Zheng, Nano Lett., 2011, 11, 2390.
42. X. Zhang, M. I. Zhou and L. Lei, Carbon, 2005, 43, 1700.
43. S. H. Yang, X. F. Song, P. Zhang, J. Sun and L. Gao, Small, 2014, 10, 2270.
44. R. M. Navarro, M. C. A. Galvan, J. A. V. DelaMano, S. M. A. Zahrani and J. L. G. Fierro, Energy Environ. Sci., 2010, 3, 1865.
45. M. X. Sun, Y. Wang, Y. L. Fang, S. F. Sun and Z. S. Yu, J. Alloys Compd., 2016, 684, 335.
46. S. B. Zhu, T. G. Xu, H. B. Fu, J. C. Zhao and Y. F. Zhu, Environ. Sci. Technol., 2007, 41, 6234.
47. J. C. D. Faria, A. J. Campbell and M. A. McLachlan, Adv. Funct. Mater., 2015, 25, 4657.
48. M. W. Mara, D. N. Bowman, E. Jakubikova and L. X. Chen, J. Am. Chem. Soc., 2015, 137, 9670.
49. P. Y. Zhang, T. Song, T. T. Wang and H. Zeng, Appl. Catal., B, 2017, 206, 328.
50. X. H. Zhang, T. Y. Peng, L. J. Yu, R. J. Li, Q. Q. Li and Z. Li, ACS Catal., 2015, 5, 504.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ra03451b

---