The properties and photocatalytic performance comparison of Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ composite catalysts

Abstract

Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ composite catalysts were prepared by a simple method using melamine and ferric nitrate as precursors. X-ray diffraction, UV-Vis spectroscopy, Fourier transform infrared spectra, scanning electron microscopy, photoluminescence, X-ray photoelectron spectroscopy and photocurrent measurements were used to characterize the prepared catalysts. The results indicated that Fe³⁺ was inserted at an interstitial position of g-C₃N₄ by coordinating to N atoms, which exist in a form completely different from those in the Fe₂O₃/g-C₃N₄ composite. The different form of Fe species caused the difference in optical properties, energy band structure and electrons–holes separation rate. The activities of Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ composite catalysts were tested by the photocatalytic degradation of Rhodamine B (RhB) under visible light. The rate constant of Fe³⁺-doped g-C₃N₄ was 2.5 and 1.8 times higher than that of the g-C₃N₄ and Fe₂O₃/g-C₃N₄ composite. The possible mechanism is proposed.

---

Introduction

Global energy and environmental crisis is a big challenge that every country has to face. With continuous efforts among researchers, new energy sources such as solar, wind and hydrogen energy are gradually replacing traditional fossil energy sources on the stage of history. In the past decades, semiconductor photocatalysis, as a “green” technology, has been widely used for the degradation of organic compounds, treatment of toxic gases and reduction of heavy metal ions.¹ However, the most commonly used metal oxide photocatalysts based on TiO₂ suffer from no or limited visible light absorption due to their large band gap energies (3.0 and 3.2 eV for rutile and anatase). Therefore, most work is focused on the search of visible-light sensitive photocatalysts. One method is TiO₂ modification such as by metal ion doping,² non-metal doping,³ dye sensitization⁴ and semiconductor coupling.⁵ These approaches have resulted in improvements in visible light absorption and catalytic performance but only to a limited extent. Another method is required to develop new and efficient visible-light sensitive semiconductors.

Recently, a polymeric semiconductor, graphite-like carbon nitride (g-C₃N₄) has attracted extensive interest. As a ‘‘sustainable’’ photocatalyst, g-C₃N₄ has numerous advantages such as good thermal and chemical stability, being metal-free and having a tunable electronic structure. Moreover, the moderate band gap energy (2.7 eV) allows it to directly utilize visible light. These unique properties make g-C₃N₄ a valuable material for various potential applications such as energy conversion,⁶ organic synthesis,⁷ pollutant treatment,⁸ hydrogen production⁹ and carbon dioxide storage.¹⁰ Nevertheless, the low efficiency caused by a high recombination rate of the photogenerated charges limits the practical application of g-C₃N₄.

Several routes have been developed to solve this problem such as metal and nonmetal doping,^9,11–16 preparation of porous g-C₃N₄,¹⁷ protonating it by strong acids¹⁸ and designing composites with other semiconductors.¹⁹ Among these strategies, doping is one of the most convenient and effective methods. Zhang et al.⁹ synthesized I-doped g-C₃N₄ for hydrogen evolution. They proposed that the modification enhanced optical absorption, enlarged surface area and accelerated charge carrier transfer rate, as well as increased its hydrogen production ability. Yan et al.¹¹ prepared B-doped g-C₃N₄ and suggested that the enhanced RhB degradation performance resulted from the improvement of its dye-adsorption and light-absorption ability. Zhang et al.¹² synthesized phosphorus-doped g-C₃N₄ using an ionic liquid [Bmim]PF₆ as the phosphorus source. The obtained material exhibited significantly improved electrical conductivity and photocurrent. Noble metals, such as Pt, Pd and Ag, are effective dopants that trap the photogenerated electrons, thereby improving the separation rate of photogenerated electrons and holes. Bu et al.¹³ prepared Ag-modified carbon nitride and found that the separation efficiency was significantly improved, thereby enhancing the photoelectric conversion performance. Chang et al.¹⁴ synthesized Pd/mpg-C₃N₄ and used it for photodegradation of bisphenol A. They found that most of the Pd presented in the Pd⁰ state acted as an electron trap, leading to the improved separation efficiency and photocatalytic performance. However, the high price of noble metals inhibits their practical application. Wang et al.^15,16 found that the optical and electronic properties of g-C₃N₄ could also be changed by transition-metal doping such as with Fe. They suggested that Fe-doping strongly modified the electronic properties of g-C₃N₄ and enlarged the optical absorption range, thus leading to better photocatalytic performance.

Fe₂O₃, as a semiconductor with narrow band-gap energy (2.1–2.2 eV), is widely used as a visible light photocatalyst.^20,21 Fe₂O₃ has also been used to prepare composite photocatalysts with other semiconductors, such as TiO₂, Bi₂O₃ and SnO₂, to promote charge transfer across the heterojunction, resulting in improved photocatalytic performance.^22–24 However, to the best of our knowledge, few studies concerning the Fe₂O₃/g-C₃N₄ composite catalyst have been reported. The properties and photocatalytic performance of Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ have also not been compared. In this work, Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ composite catalysts were prepared and the possible Fe doping site is discussed. The photocatalytic activity was evaluated for the degradation of RhB under visible light. The properties and photocatalytic performance of Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ were compared.

Experimental

Preparation and characterization

In a typical experiment, 3 g melamine was dispersed in 15 ml deionized water under stirring. Then, 0.006–0.12 g Fe(NO₃)₃·9H₂O was added. The obtained suspension was heated to 100 °C to remove water. The solid product was dried at 80 °C in an oven, followed by milling and annealing at 520 °C for 2 h (at a rate of 5 °C min⁻¹). The obtained product was denoted as Fe(x%)–CN, where x is the mass percent of Fe(NO₃)₃·9H₂O to melamine. For comparison, neat g-C₃N₄ was prepared following the abovementioned procedure in the absence of Fe(NO₃)₃·9H₂O.

α-Fe₂O₃ was prepared by a solvothermal method. Typically, 0.6 g Fe(NO₃)₃·9H₂O and 0.5 g urea were dissolved in a mixture of glycol and distilled water. The solution was transferred into a 30 ml Teflon-lined autoclave, and then heated at 150 °C for 12 h. The solid product α-Fe₂O₃ was collected and washed by centrifugation. The Fe₂O₃/g-C₃N₄ composite catalyst was prepared as follows: the desired amount of α-Fe₂O₃ and 4.082 g melamine were dispersed in 20 ml deionized water under stirring. Then, the suspension was heated to 100 °C to remove water. The solid product was dried at 80 °C in an oven, followed by milling and annealing at 520 °C for 2 h (at a rate of 5 °C min⁻¹). The obtained product, with the same Fe/g-C₃N₄ mass ratio as Fe(x%)–CN, was denoted as FeO(x%)–CN.

XRD patterns of the prepared TiO₂ samples were recorded on a Rigaku D/max-2400 instrument using Cu-Kα radiation (λ = 1.54 Å). UV-Vis spectroscopy measurement was carried out on a JASCO V-550 model UV-Vis spectrophotometer using BaSO₄ as the reflectance sample. FT-IR spectra were obtained on a Nicolet 20DXB FT-IR spectrometer. The morphology was observed by a scanning electron microscope (SEM, JSM 5600LV, JEOL Ltd). XPS measurements were conducted on a Thermo Escalab 250 XPS system with Al Kα radiation as the exciting source. Photoluminescence (PL) spectra were measured at room temperature with a fluorospectrophotometer (FP-6300) using an Xe lamp as excitation source. Working electrodes were prepared as follows: 0.3 g of sample was ground with 0.7 g of ethanol to prepare a slurry, and then coated on an indium–tin oxide glass by the doctor-blade method. Photocurrents were measured by electrochemical analyzer (CHI 618C Instruments) in a standard three-electrode system in which the prepared sample film was used as the working electrode, a Pt flake as the counter electrode, and Ag/AgCl as the reference electrode. A 500 W Xe-lamp was used to irradiate the working electrode from the back. A 1.0 M Na₂SO₄ solution was used as the electrolyte.

Photocatalytic reaction

RhB was selected as the model compound to evaluate the photocatalytic performance of the prepared g-C₃N₄-based catalysts in an aqueous solution under visible light irradiation. 0.05 g catalyst was dispersed in 200 ml aqueous solution of RhB (10 ppm) in an ultrasound generator for 10 min. The suspension was then transferred into a self-designed glass reactor and stirred for 30 min in darkness to achieve adsorption equilibrium. In the photoreaction under visible light irradiation, the suspension was exposed to a 250 W high-pressure sodium lamp with main emission in the range of 400–800 nm, and air was bubbled through the solution at 130 ml min⁻¹. The UV light portion of the sodium lamp was filtered by 0.5 M NaNO₂ solution. All the runs were conducted at ambient pressure and 30 °C. At given time intervals, 4 ml of the suspension was removed and immediately centrifuged to separate the liquid samples from the solid catalyst. The concentration of RhB before and after the reaction was measured by UV-Vis spectrophotometer at a wavelength of 550 nm.

Results and discussion

Fig. 1 shows the XRD patterns of the synthesized Fe₂O₃, g-C₃N₄, Fe(x%)–CN and FeO(x%)–CN. The patterns of g-C₃N₄ and Fe(x%)–CN exhibit two peaks at 13.1° and 27.3°, corresponding to (100) and (002) crystal planes of g-C₃N₄ (Fig. 1a), respectively.²⁵ No peak for Fe₂O₃ is observed in the Fe(x%)–CN pattern, whereas an obvious shift toward a higher 2θ value (about 0.2°) is observed for all the Fe(x%)–CN catalysts. Dong et al. prepared carbon self-doping g-C₃N₄ and observed a similar phenomenon.²⁶ This crystal lattice distortion may be due to the Fe³⁺ doping in g-C₃N₄. In Fig. 1b, no peak for Fe₂O₃ is observed in the pattern of FeO(x%)–CN, which is probably because the Fe₂O₃ amount is low. In addition, compared with g-C₃N₄, no peak shift in the pattern of FeO(x%)–CN is observed. This indicates that the existence form of Fe species differs in Fe(x%)–CN and FeO(x%)–CN.

Fig. 1 XRD patterns of synthesized Fe₂O₃, g-C₃N₄, Fe(x%)–CN and FeO(x%)–CN.

The light absorption property of the as-prepared g-C₃N₄, FeO(x%)–CN and Fe(x%)–CN catalysts was studied by UV-Vis spectra; the results are shown in Fig. 2. g-C₃N₄ shows typical semiconductor absorption, originating from charge transfer response of g-C₃N₄ from VB populated by N 2p orbital to CB formed by C 2p orbital.²⁷ For FeO(x%)–CN, only a slight shift of absorption band to visible light region is observed, which is due to the low amount of Fe₂O₃. However, the obvious red shifts of absorption band are observed for all the Fe(x%)–CN catalysts, indicating that the band gap energies decreased. This obvious difference is due to the different forms of Fe species that exist in Fe(x%)–CN and FeO(x%)–CN. Loading of Fe₂O₃ did not influence the optical property of synthesized g-C₃N₄ because of the low Fe₂O₃ amount. However, Fe³⁺ doped into g-C₃N₄ lattice could affect its electronic structure, thus changing the optical property of g-C₃N₄. The color of the Fe(x%)–CN changed from primrose yellow to orange as the iron content increased from 0.5 to 4%. The band gap energy calculated based on the Oregan and Gratzel method indicated that the value for g-C₃N₄ was 2.70 eV, which is consistent with previous results.^28,29 For FeO(x%)–CN, the band gap energies decreased slightly to 2.69 eV. However, for Fe(0.5%)–CN, Fe(1.5%)–CN and Fe(4%)–CN, these values were 2.53, 2.51 and 2.48 eV, respectively, indicating that Fe³⁺ content strongly influenced the optical property of the synthesized Fe(x%)–CN catalysts.

Fig. 2 UV-Vis diffuse reflectance spectra of g-C₃N₄, Fe(x%)–CN and FeO(x%)–CN catalysts.

FT-IR spectra can provide plentiful structural information regarding synthesized g-C₃N₄-based catalysts (Fig. 3). In g-C₃N₄, a series of peaks ranging between 1200 and 1600 cm⁻¹ are assigned to the typical stretching modes of CN heterocycles, whereas the sharp peak located at 810 cm⁻¹ is assigned to the bending vibration of heptazine rings, indicating that the synthesized g-C₃N₄ is composed of heptazine units. The broad absorption band around 3200 cm⁻¹ originated from the stretching vibration of the N–H bond, which was associated with uncondensed aminogroups.³⁰ The characteristic peaks of g-C₃N₄ do not move in Fe(0.5%)–CN and FeO(0.5%)–CN curves, indicating that Fe modification did not change the structural features of g-C₃N₄. For comparison, Fe₂O₃ was measured (Fig. 3). The peaks at 556 and 471 cm⁻¹ are the characteristic stretching vibrations of the Fe–O bond in hematite particles.³¹ These peaks are not detected in FeO(0.5%)–CN due to the low amount of Fe₂O₃ loading.

Fig. 3 FT-IR spectra of synthesized g-C₃N₄, Fe₂O₃, FeO(0.5%)–CN and Fe(0.5%)–CN.

The morphology of the prepared g-C₃N₄-based catalysts was investigated by SEM. Fig. 4 shows the SEM images of the g-C₃N₄, FeO(0.5%)–CN and Fe(0.5%)–CN. All these catalysts exhibit a layered structure similar to that of their analogue graphite. No obvious difference was observed between g-C₃N₄ and FeO(0.5%)–CN, indicating that loading Fe₂O₃ did not influence the morphology of g-C₃N₄. For Fe(0.5%)–CN, the particle size decreased remarkably compared with that of g-C₃N₄. This indicated that Fe³⁺ doping could inhibit the crystal growth of graphitic carbon nitride.

Fig. 4 SEM images of g-C₃N₄ (A), FeO(0.5%)–CN (B) and Fe(0.5%)–CN (C).

In order to confirm the structure of the prepared catalysts and to further identify the chemical state of iron, g-C₃N₄, FeO(0.5%)–CN and Fe(0.5%)–CN were characterized by XP spectra. In Fig. 5a and b, the spectra of g-C₃N₄ in both the N 1s and C 1s region can be fitted with three contributions. In the N 1s region (Fig. 5a), three contributions located at 398.2, 399.0 and 400.4 eV were assigned to the sp² hybridized aromatic nitrogen atoms bonded to carbon atoms (C–N=C), tertiary nitrogen N–(C)₃ groups linking structural motif or aminogroups carrying hydrogen ((C)₂–N–H) in connection with structural defects and incomplete condensation, and nitrogen atoms bonded to three carbon atoms in the aromatic cycles.³² In Fig. 5b, three components located at 284.6, 285.8 and 287.9 eV for g-C₃N₄ were attributed to the C–C bond that originated from sp² C atoms bonded to N in an aromatic ring as (N–C=N), C=N or C=N, which could be ascribed to defect-containing sp² hybridized carbon atoms present in graphitic domains and pure graphitic sites in a CN matrix.^33,34 Fig. 5 indicates that the spectrum of Fe(0.5%)–CN in the N 1s and C 1s regions can also be fitted with three contributions. However, obvious shifts to higher binding energies were observed for Fe(0.5%)–CN in the C 1s and N 1s regions compared with that of g-C₃N₄. This is probably due to the change in chemical environment after Fe³⁺ doping. The electrons of the lone electron pair of nitrogen in g-C₃N₄ may be partially transferred to iron atoms, leading to the decreased electron density of nitrogen. Accordingly, the electrons on C–N bond in the g-C₃N₄ framework tend to be closer to the nitrogen atoms, thus the electron density of carbon atoms also decreased. These decreased electron densities cause the higher binding energies of Fe(0.5%)–CN in the C 1s and N 1s regions.

Fig. 5 XP spectra of synthesized g-C₃N₄, FeO(0.5%)–CN and Fe(0.5%)–CN in the region of C 1s (a), N 1s (b), Fe 2p (c) and VB XPS (d).

In Fig. 5c, the binding energies of Fe(0.5%)–CN and FeO(0.5%)–CN in the Fe 2p region are located at 711.9 and 711.2 eV, respectively. It has been reported that the binding energies in the Fe 2p_3/2 region are located at 710.8 and 711.1 eV in the formation of Fe₃O₄ and Fe₂O₃, respectively.^32,35 This confirms that Fe exists as Fe₂O₃ in FeO(0.5%)–CN. For Fe(0.5%)–CN, this binding energy should result from Fe³⁺ being inserted at the interstitial position and stabilized in the electron-rich g-C₃N₄ structure through the Fe–N bonds (Fig. 5c inset), as reported in previous studies.^16,17 The valence band XP spectra of g-C₃N₄, FeO(0.5%)–CN and Fe(0.5%)–CN are shown in Fig. 5d. Compared with the spectrum of g-C₃N₄, an obvious shift is observed in Fe(0.5%)–CN, which should be attributable to the Fe³⁺ doped into the g-C₃N₄ lattice. The binding energy difference between them is ∼0.2 eV, which is close to the calculation result from the UV-Vis spectra. No obvious binding energy difference between g-C₃N₄ and FeO(0.5%)–CN is revealed.

Fig. 6 shows the room temperature PL spectra of g-C₃N₄, FeO(0.5%)–CN and Fe(x%)–CN under the excitation wavelength of 380 nm. For g-C₃N₄, the broad PL band is around 465 nm, which can be attributed to the band–band PL phenomenon with the energy of light approximately equal to the band gap of g-C₃N₄, as calculated by UV-Vis spectra.³⁴ Such band–band PL signal is attributed to excitonic PL, which mainly results from the n–π* electronic transitions involving lone pairs of nitrogen atoms in g-C₃N₄.³⁶ FeO(x%)–CN shows a curve similar to that of g-C₃N₄, whereas the peak intensities decreased. In theory, the CB and VB of g-C₃N₄ are −1.12 V and +1.57 V, respectively.²⁷ For Fe₂O₃, the CB and VB are +0.28 V and +2.48 V, respectively.³⁷ Therefore, the interfacial charge transfer from g-C₃N₄ to Fe₂O₃ could occur, decreasing the recombination of photogenerated electron–hole pairs. The PL intensity of FeO(x%)–CN decreased at first and then increased with increasing Fe₂O₃ content. This indicated that over-loading Fe₂O₃ could increase the recombination rate of photogenerated electron–holes. For Fe(x%)–CN, the peak intensities decreased sharply. This indicates that doping is more favorable for separation of photogenerated electron–hole pairs compared with Fe₂O₃ loading. It is known that the reduction potential of Fe²⁺/Fe³⁺ is below the conduction band of g-C₃N₄.¹⁶ After Fe³⁺ doping, the photogenerated electrons could be trapped by the Fe³⁺ doping site, leading to the reduced recombination of photogenerated electron–hole pairs. With greater Fe³⁺ content, more photogenerated electron trapping sites exist, leading to the further decrease of PL intensity.

Fig. 6 Photoluminescence emission spectra of g-C₃N₄, FeO(x%)–CN and Fe(x%)–CN.

The outstanding carrier separation ability of g-C₃N₄-based catalysts was confirmed by the photocurrent responses measurement (I–t curve, Fig. 7). Undoubtedly, g-C₃N₄ exhibits a low photocurrent because of the high photogenerated electrons–holes recombination rate. The photocurrent value of FeO(0.5%)–CN is ∼1.8 times as high as that of g-C₃N₄, which can be attributed to the intimate interaction existing on the Fe₂O₃/g-C₃N₄ interface where photogenerated electrons and holes are efficiently separated, leading to the decreased photoinduced carrier recombination, corresponding to its enhanced photocurrent. For Fe(0.5%)–CN, the photocurrent value is ∼3 times as high as that of g-C₃N₄, which is much higher than that of FeO(0.5%)–CN. This indicates that Fe³⁺ doping could inhibit the recombination more effectively, which is consistent with PL results. In addition, the photocurrent responses did not decay with increased illumination time, indicating that the prepared catalysts could provide stable amount of electrons and holes during irradiation.

Fig. 7 The photocurrent responses of g-C₃N₄, FeO(0.5%)–CN and Fe(0.5%)–CN.

The photocatalytic performance of g-C₃N₄-based catalysts in the degradation of RhB under visible light irradiation is shown in Fig. 8. Control experiment results indicated that the RhB degradation performance can be ignored in the absence of either irradiation or photocatalysis, indicating that RhB was degraded via a photocatalytic process. 70% RhB was degraded over g-C₃N₄ in 120 min. Fe(x%)–CN showed clearly higher activities than that of g-C₃N₄ (Fig. 8a). This increased photocatalytic performance should be due to the synergistic effect of decreased band gap energy and reduced recombination rate of photogenerated electrons–holes pairs caused by Fe³⁺ doping. Fe(0.5%)–CN exhibited the highest activity of more than 99% RhB degradation in 120 min. When the Fe content exceeded 0.5%, the activity of the catalyst began to decrease; the possible reason is discussed below. In Fig. 8b, the activity of g-C₃N₄ increased after Fe₂O₃ loading. The activity increased in the order of FeO(0.1%)–CN < FeO(0.5%)–CN < FeO(0.2%)–CN, which is opposite to the PL intensity order. Considering that no obvious difference in structural and optical properties among FeO(x%)–CN catalysts was observed, such activity order should be attributable to the different photogenerated electrons–holes separation rate (Fig. 6). To investigate the catalytic stability of Fe³⁺-doped g-C₃N₄, the photocatalytic performance of Fe(0.5%)–CN was investigated in three cycles (Fig. 9). No obvious decrease in activity was observed after three cycles, indicating that prepared Fe³⁺-doped g-C₃N₄ is stable under visible-light irradiation.

Fig. 8 Photocatalytic performances of g-C₃N₄ based catalysts in the degradation of RhB under visible light irradiation.

Fig. 9 The catalytic stability test of Fe(0.5%)–CN.

The reaction rate constant k was obtained by assuming that the reaction followed first order kinetics.^38,39 In a batch reactor, the performance equation is −ln(C/C₀) = kt, where C₀ and C represent the concentrations of RhB dye before and after photocatalytic degradation, respectively. If a linear relationship is established when −ln(C/C₀) is plotted against t (reaction time), the rate constant k can be obtained from the slope of the line. The calculated results indicated that the rate constant k was 0.0095, 0.0183, 0.0236, 0.0125, 0.0116, 0.0135, 0.0122, 0.0145 and 0.0132 min⁻¹ for g-C₃N₄, Fe(0.2%)–CN, Fe(0.5%)–CN, Fe(1%)–CN, Fe(1.5%)–CN, Fe(4%)–CN, FeO(0.1%)–CN, FeO(0.2%)–CN and FeO(0.5%)–CN, respectively. Fe(0.5%)–CN exhibited the highest rate constant, which is 2.5 and 1.8 times as high as that of g-C₃N₄ and FeO(0.5%)–CN.

To clarify the reaction mechanism in detail, the active species generated during the reaction process were identified by a hole and free radical trapping experiment. EDTA–2Na, tert-butyl alcohol (t-BuOH) and 1,4-benzoquinone (BQ) were used as the hole (h⁺) scavenger, hydroxyl radical (˙OH) scavenger and superoxide radical (˙O₂⁻) scavenger, respectively.⁴⁰ Fig. 10 shows the influence of various scavengers on the photocatalytic performance of g-C₃N₄, Fe(0.5%)–CN and FeO–CN. For g-C₃N₄, the photodegradation rate of RhB decreased slightly after the addition of t-BuOH, indicating that hydroxyl radicals are not the main active species in the current photocatalytic systems. When BQ was added, the degradation of RhB decreased sharply, indicating ˙O₂⁻ is mainly responsible for the RhB degradation. In theory, the CB and VB of g-C₃N₄ were −1.12 V and +1.57 V, respectively. The redox potentials of ˙OH/OH⁻ and O₂/˙O₂⁻ have been determined as +1.99 V and −0.33 V, respectively.⁴¹ Therefore, the reduction potential of CB electrons in g-C₃N₄ was more negative than the redox potential of O₂/˙O₂⁻, which can reduce O₂ to form ˙O₂⁻, whereas the VB holes in g-C₃N₄ were not sufficiently positive to generate ˙OH. Therefore, the main active species in the current system should be ˙O₂⁻ and not ˙OH, which is consistent with our experimental results. In the presence of EDTA–2Na, the RhB degradation rate was obviously increased, which is completely different from previous results.⁴⁰ Zhang et al.⁴⁰ prepared a C₃N₄/Bi₅Nb₃O₁₅ heterojunction catalyst for 4-CP photodegradation. The photodegradation rate decreased significantly after the addition of EDTA–2Na, indicating that the photogenerated holes were the main active species in their system. In our investigation, although the redox potential of RhB is reported to be 1.43 V,⁴² which is higher than the VB of g-C₃N₄, the direct photogenerated hole oxidation did not occur. On the contrary, the addition of EDTA–2Na to trap the h⁺ could promote the separation rate of e⁻–h⁺ pairs, leading to the formation of more ˙O₂⁻. Thus, the photocatalytic performance increased. For Fe(0.5%)–CN, a similar trend has been obtained, indicating that the main oxidative species is the same as that of g-C₃N₄. Fe³⁺ doping did not change the reaction mechanism. However, when the Fe³⁺ content exceeded the optimal value of 0.5%, the photocatalytic performance decreased significantly, as shown in Fig. 8a, which is possibly due to competitive capture existing between the adsorbed O₂ molecule and Fe³⁺ doping site. In this case, excess Fe³⁺ doping sites could trap more photogenerated electrons, leading to fewer photogenerated electrons reacting with adsorbed O₂ molecules. Accordingly, the superoxide radical content reduced, thus leading to the decreased photocatalytic performance. For FeO(0.5%)–CN, the general trend is not changed but some difference is observed. When t-BuOH was added to trap ˙OH, the activity decreased by 9.4%, which was much higher than that of g-C₃N₄ (2.2%). The VB of Fe₂O₃ is +2.48 V, which is sufficiently positive to generate ˙OH. Therefore, we deduced that not only ˙O₂⁻, but ˙OH is also one of the active species in the FeO(0.5%)–CN system. In addition, when EDTA–2Na was added to trap the holes, the activity increased (∼96%) but not as high as that of g-C₃N₄ (100% in 90 min). Trapping h⁺ could promote the separation rate of e⁻–h⁺ pairs, leading to the formation of more ˙O₂⁻. However, the formation of ˙OH was suppressed, and therefore, the activity was lower than that of g-C₃N₄.

Fig. 10 Influence of various scavengers on the visible light photocatalytic activity of g-C₃N₄, Fe(0.5%)–CN and FeO(0.5%)–CN.

Conclusions

Fe³⁺-doped g-C₃N₄ and Fe₂O₃/g-C₃N₄ composite catalysts were prepared by a simple method using melamine and ferric nitrate as precursors. Fe³⁺ was inserted at the interstitial position of g-C₃N₄ by coordinating to N atoms, which exist in a form that is completely different from a Fe₂O₃/g-C₃N₄ composite. Such a different form of Fe species caused the difference in optical properties, energy band structure and electrons–holes separation rate. The rate constant of Fe(0.5%)–CN for RhB photodegradation was 2.5 and 1.8 times higher than that of the g-C₃N₄ and the Fe₂O₃/g-C₃N₄ composite, respectively. When the Fe³⁺ content exceeded 0.5%, the activity of the catalyst began to decrease. This is due to the competitive capture existing between the adsorbed O₂ molecules and the Fe³⁺ doping site, which suppresses the formation of ˙O₂⁻. The superoxide radical ˙O₂⁻ was the active species in g-C₃N₄ and Fe³⁺ doped g-C₃N₄ systems. For Fe₂O₃/g-C₃N₄ system, both ˙O₂⁻ and ˙OH were the main active species.

Acknowledgements

The authors are grateful for the financial support from Program for New Century Excellent Talents in University (NCET) (no. NCET-11-1011), the National Natural Science Foundation of China (no. 21103077), Natural Science Foundation of Liaoning Province (no. 201202123) and the Natural Science Foundation of Liaoning Shihua University (no. 2011XJJ-021).

Notes and references

1. M. R. Hoffmann, S. T. Martin, W. Choi and D. W. Bahnemann, Chem. Rev., 1995, 95, 69–96.
2. S. Klosek and D. Raftery, J. Phys. Chem. B, 2001, 105, 2815–2819.
3. R. Asahi, T. Morikawa, T. Ohwaki, A. Aoki and Y. Taga, Science, 2001, 293, 269–271.
4. D. Chatterjee and A. Mahata, Appl. Catal., B, 2001, 33, 119–125.
5. S. Z. Hu, F. Y. Li, Z. P. Fan and J. Z. Gui, J. Power Sources, 2014, 250, 30–39.
6. A. Vinu, Adv. Funct. Mater., 2008, 18, 816–827.
7. J. Xu, H. T. Wu, X. Wang, B. Xue, Y. X. Li and Y. Cao, Phys. Chem. Chem. Phys., 2013, 15, 4510–4517.
8. S. N. Talapaneni, S. Anandan, G. P. Mane, C. Anand, D. S. Dhawale, S. Varghese, A. Mano, T. Mori and A. Vinu, J. Mater. Chem., 2012, 22, 9831–9840.
9. G. G. Zhang, M. W. Zhang, X. X. Ye, X. Q. Qiu, S. Lin and X. C. Wang, Adv. Mater., 2014, 26, 805–809.
10. Q. A. Li, J. P. Yang, D. Feng, Z. X. Wu, Q. L. Wu, S. S. Park, C. S. Ha and D. Y. Zhao, Nano Res., 2010, 3, 632–642.
11. S. C. Yan, Z. S. Li and Z. G. Zou, Langmuir, 2010, 26, 3894–3901.
12. Y. J. Zhang, T. Mori, J. H. Ye and M. Antonietti, J. Am. Chem. Soc., 2010, 132, 6294–6295.
13. Y. Y. Bu, Z. Y. Chen and W. B. Li, Appl. Catal., B, 2014, 144, 622–630.
14. C. Chang, Y. Fu, M. Hu, C. Y. Wang, G. Q. Shan and L. Y. Zhu, Appl. Catal., B, 2013, 142-143, 553–560.
15. X. F. Chen, J. S. Zhang, X. Z. Fu, M. Antonietti and X. C. Wang, J. Am. Chem. Soc., 2009, 131, 11658–11659.
16. X. C. Wang, X. F. Chen, A. Thomas, X. Z. Fu and M. Antonietti, Adv. Mater., 2009, 21, 1609–1612.
17. F. Goettmann, A. Fischer, M. Antonietti and A. Thomas, Angew. Chem., Int. Ed., 2006, 45, 4467–4471.
18. Y. J. Zhang, A. Thomas, M. Antonietti and X. C. Wang, J. Am. Chem. Soc., 2009, 131, 50–51.
19. C. S. Pan, J. Xu, Y. J. Wang, D. Li and Y. F. Zhu, Adv. Funct. Mater., 2012, 22, 1518–1524.
20. Q. Zhang, X. W. Lu, L. Y. Chen, Y. X. Shi, T. Xu and M. L. Liu, Mater. Lett., 2013, 106, 447–451.
21. Y. Liu, C. Y. Yu, W. Dai, X. H. Gao, H. S. Qian, Y. Hu and X. Hu, J. Alloys Compd., 2013, 551, 440–443.
22. H. Tang, D. Zhang, G. G. Tang, X. R. Ji, W. J. Li, C. S. Li and X. F. Yang, Ceram. Int., 2013, 39, 8633–8640.
23. J. Z. Li, J. B. Zhong, X. Y. He, S. T. Huang, J. Zeng, J. J. He and W. L. Shi, Appl. Surf. Sci., 2013, 284, 527–532.
24. Y. Mi, Y. H. Cao, X. L. Liu, J. B. Yi, H. R. Tan, P. Ma, H. Hao, X. Zhang and H. M. Fan, Mater. Chem. Phys., 2013, 143, 311–321.
25. Y. Wang, X. C. Wang and M. Antonietti, Angew. Chem., Int. Ed., 2012, 51, 68–89.
26. G. H. Dong, K. Zhao and L. Z. Zhang, Chem. Commun., 2012, 48, 6178–6180.
27. X. C. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, K. Domen and M. Antonietti, Nat. Mater., 2009, 8, 76–80.
28. B. Oregan and M. Gratzel, Nature, 1991, 353, 737–740.
29. G. Q. Li, N. Yang, W. L. Wang and W. F. Zhang, J. Phys. Chem. C, 2009, 113, 14829–14833.
30. S. C. Yan, Z. S. Li and Z. G. Zou, Langmuir, 2009, 25, 10397–10401.
31. L. Xu, J. X. Xia, H. Xu, S. Yin, K. Wang, L. Y. Huang, L. G. Wang and H. M. Li, J. Power Sources, 2014, 245, 866–874.
32. Y. W. Zhang, J. H. Liu, G. Wu and W. Chen, Nanoscale, 2012, 4, 5300–5303.
33. L. Ge and C. Han, Appl. Catal., B, 2012, 117-118, 268–274.
34. W. Lei, D. Portehault, R. Dimova and M. Antoniettit, J. Am. Chem. Soc., 2011, 133, 7121–7127.
35. X. S. Zhou, B. Jin, R. Q. Chen, F. Peng and Y. P. Fang, Mater. Res. Bull., 2013, 48, 1447–1452.
36. V. N. Khabashesku, J. L. Zimmerman and J. L. Margrave, Chem. Mater., 2000, 12, 3264–3270.
37. Y. Xu and M. A. A. Schoonen, Am. Mineral., 2000, 85, 543–556.
38. C. Zhang and Y. F. Zhu, Chem. Mater., 2005, 17, 3537–3545.
39. H. B. Fu, C. S. Pan, W. Q. Yao and Y. F. Zhu, J. Phys. Chem. B, 2005, 109, 22432–22439.
40. S. Q. Zhang, Y. X. Yang, Y. N. Guo, W. Guo, M. Wang, Y. H. Guo and M. X. Huo, J. Hazard. Mater., 2013, 261, 235–245.
41. G. Liu, P. Niu, L. C. Yin and H. M. Cheng, J. Am. Chem. Soc., 2012, 134, 9070–9073.
42. T. Shen, Z. G. Zhao, Q. Yu and H. J. Xu, J. Photochem. Photobiol., A, 1989, 47, 203–212.

---