Photoreduction of CO2 to CH4 over Efficient Z-Scheme γ-Fe2O3/g-C3N4 Composites

A series of composite γ-Fe2O3/g-C3N4 (denoted as xFeCN with x equal 5, 10, 15, and 20 of γ-Fe2O3 percentage in weight) was prepared by calcination and precipitation-impregnation methods. X-ray diffraction (XRD), Fourier transform infrared (FTIR), and X-ray photoelectron spectrometry (XPS) characterizations indicated the successful synthesis of Z-scheme FeCN composites. A red shift of the light absorption region was revealed by UV-vis diffuse reflectance spectroscopy (UV-DRS). In addition, photoluminescence spectroscopy (PL) spectra showed an interface interaction of two phases Fe2O3 and g-C3N4 in the synthesized composites that improved the charge transfer capacity. The photocatalytic activity of these materials was studied in the photoreduction of CO2 with H2O as the reductant in the gaseous phase. The composites exhibited excellent photoactivity compared to undoped g-C3N4. The CH4 production rate over 10FeCN and 15FeCN composites (2.8 × 10−2 and 2.9 × 10−2 μmol h−1 g−1, respectively) was ca. 7 times higher than that over pristine g-C3N4 (0.4 × 10−2 μmol h−1 g−1). This outstanding photocatalytic property of these composites was explained by the light absorption expansion and the prevention of photogenerated electron-hole pairs recombination due to its Z-scheme structure.


Introduction
Carbon dioxide from fossil energy consumption is the most important source of greenhouse gas emissions to the atmosphere, causing global warming [1,2]. Carbon dioxide capture and storage (CCS) or utilization (CCU) has largely been studied during the last decades [3,4]. Among them, photocatalytic CO 2 conversion into valuable compounds, such as CH 4 and CH 3 OH, is an attractive route [5][6][7]. us, many semiconductors have been investigated as photocatalysts with a particular focus on the development of photocatalytic heterojunction systems by combining various semiconductor materials to form different photocatalyst types such as type I, type II, or especially Z-scheme systems [8][9][10][11][12]. In the type I and type II photocatalysts, there occurs the photogenerated electrons transfer between conduction bands and holes, one between valance bands of each component in a composite. While in the Z-scheme type, this transfer takes place between the conduction and valance band of two adjacent components. By this characteristic, the Z-scheme photocatalyst has the advantage of mobilizing the potential position of the conduction band or valance band of each semiconductor when they are simultaneously combined for targeted reaction. ereby, Z-scheme photocatalysts are expected to have stronger redox properties, better charge transfer, and higher improved light absorption yield than those of other simple photocatalyst types.
Recently, graphitic carbon nitride, g-C 3 N 4 , a semiconductor, has attracted researchers as a potential photocatalyst thanks to its easy synthesis, low cost, and moderate bandgap energy of 2.7 eV, which allows the absorption of visible light [13,14]. In particular, the conduction band (CB) position of g-C 3 N 4 is quite negative (−1.14 eV), which is favorable for the photoreduction of CO 2 into almost valuable hydrocarbons such as CH 4 , HCOOH, CH 3 OH, and C 2 H 5 OH [15]. However, one of the major disadvantages of g-C 3 N 4 is the fast electron-hole pair recombination, the relative high bandgap energy to be able to absorb most of the visible light, occupying 44% solar spectra. To improve these drawbacks, one of the strategies is to combine g-C 3 N 4 with other semiconductors [16][17][18][19][20].
Among the semiconductors used as photocatalysts, iron oxide, Fe 2 O 3 , is an interesting candidate [21]. One of the outstanding features of this semiconductor is its low synthesis cost and relatively low band gap energy, E g � 2.2 eV, which allows broad absorption in the visible region of sunlight.
erefore, Fe 2 O 3 oxide has been extensively studied as a photocatalyst through the photodegradation of polluted organic compounds in water [22,23]. However, for the photoreduction of CO 2 , the conduction band potential (CB) is positive (+1.58 eV). So, Fe 2 O 3 oxide is not able to reduce this molecule. Combining Fe 2 O 3 oxide with another semiconductor having enough negative conduction band potential is required to create an efficient photocatalyst for CO 2 photoreduction [24][25][26][27][28][29]. e improving photocatalytic activity by adding c-Fe 2 O 3 was also reported in some other works. In the research of Ding et al., for the photoreduction of CO 2 in liquid phase to CH 3 OH [30], a-Fe 2 O 3 /g-C 3 N 4 with the weight ratio a-Fe 2 O 3 : g-C 3 N 4 of 60 : 40 had the best photocatalytic activity with 2.9-fold than pristine g-C 3 N 4 . Duan and Mei [31] also reported the highest amount of CH 3 OH obtained on 15%Fe 2 O 3 /g-C 3 N 4 (>3.5 fold than pristine g-C 3 N 4 ). For CO 2 photoreduction in the gas phase, Wong et al. [32] observed the CO formation on dendritestructured a-Fe 2 O 3 /g-C 3 N 4 (27.2 μmol h −1 g −1 ), which was about 2.2 times higher than the one on pure g-C 3 N 4 (10.3 μmol h −1 g −1 ). On the same type of catalyst with urchin-like a-Fe 2 O 3 morphology, Yong Zhou et al. also recognized a CO production rate of 17.8 μmol g −1 h −1 , 3 times higher than that of pristine g-C 3 N 4 (6.1 μmol g −1 h −1 ) [33]. Besides, Fe 2 O 3 /g-C 3 N 4 composites also showed higher photoactivity than pure a-Fe 2 O 3 and g-C 3 N 4 in the photodegradation of organic pollutants (Direct Red 81, Rhodamin B, and tetracycline hydrochloride) [34][35][36]. ese studies showed that the morphology of Fe 2 O 3 oxide seems to play an important role in the activity of catalyst. e different morphologies may change interface interaction between Fe 2 O 3 and g-C 3 N 4 phase, leading to the change in charge carrier, a key factor of photocatalysis.
Based on the analysis above, in this study, we have synthesized Z-scheme c-Fe 2 O 3 /g-C 3 N 4 photocatalysts for CH 4 production from CO 2 photoreduction in the gas phase.
is new Z-scheme materiel could take advantage of the low bandgap energy of c-Fe 2 O 3 oxide and the rather negative conduction band potential of g-C 3 N 4 as well.

Synthesis of g-C 3 N 4 and c-Fe 2 O 3 /g-C 3 N 4 Composite.
Carbon graphitic nitride, g-C 3 N 4 , was synthesized by calcination of melamine at 550 o C for 3 hours under the nitrogen atmosphere.
A series of x%c-Fe 2 O 3 /g-C 3 N 4 composites was prepared by the precipitation-impregnation method. First, 1 g of g-C 3 N 4 , which were synthesized from melamine calcination, was added in a 100 ml solution of 0.1 M NaOH. e mixture was covered by paraffin paper, stirred, and kept at 60 o C for 2 hours. en, a calculated amount of FeSO 4 .7H 2 O was slowly added in the solution above, and the pH was adjusted to 10 with 0.1 M NaOH solution, leading to the formation of a composite material of iron oxide precipitate and g-C 3 N 4 . e latter was filtered and washed by deionized water and dried at 70 o C. Four Z-scheme c-Fe 2 O 3 /g-C 3 N 4 photocatalysts, denoted as xFeCN, with x equal to 5, 10, 15, or 20 wt.% of c-Fe 2 O 3 were obtained.

Photocatalytic Procedure.
In a typical test, 100 mg of catalyst was added in 15 mL deionized water in a beaker with a 5 cm diameter. e solvent in the mixture was completely evaporated in an oven at 70 o C to well disperse the photocatalyst on the bottom of the beaker. After that, the catalyst-containing beaker was placed in a closed stainless reactor equipped with a quark window ( Figure 1). e high purity (99,999%) CO 2 flow of 500 ml/minute bubbled in a closed stainless contains deionized water kept at 25 o C before entering the stainless reactor for 30 minutes. en, a 150W Xenon lamp (Newport model 67005) was switched on for 18 h.  shows the FTIR spectra of all fresh photocatalysts. All these materials have similar characteristic signals of g-C 3 N 4 structure. e peak at 804 and 887 cm −1 is attributed to the heptazine stretching mode [37], while those at 1203-1630 cm −1 correspond to stretching vibrations of C�N and C-N bonds of the heterocycle [38,39]. For the broad band from 3000 to 3500 cm −1 , the signals were ascribed to the vibration of NH 2 and NH functional groups of g-C 3 N 4 and the OH group of absorbed water [40]. In addition, the weak signals observed at586 cm −1 originated from the Fe-O bond vibrations [41]. ese results confirmed the formation of the two expected crystalline phases of g-C 3 N 4 and c-Fe 2 O 3 in the photocatalysts synthesized. In order to confirm the formation of composite materials between these two components, XPS analysis was performed for one photocatalyst containing 10 wt% of c-Fe 2 O 3 ( Figure 3). e general XPS spectrum (Figure 3(a)) shows the presence of Fe, O, C, and N on the surface of this photocatalyst. On the high-resolution Fe elemental spectrum (Figure 3(b)), there are two major peaks at the binding energy 711.2 and 724.7 eV corresponding to Fe 3+ 2p 3/ 2 and Fe 3+ 2p 1/2 , respectively. ese two peaks are characteristics of the Fe 3+ oxidation state as observed on the XPS spectrum of Fe 2 O 3 oxide [42,43]. Besides, two satellite peaks can be detected at positions 718.2 and 731.0 eV. With the C1s spectrum ( Figure 3(e)), there are three peaks at 284.9, 286.5, and 288.5 eV. e first peak at 284.9 eV is assigned to C-C/C�C bonds, the second to the carbon of the C-NH 2 group, and the third to bonded C in the heptazine structure (N-C�N) [44,45]. In the N1s spectrum (Figure 3(d)), the peak deconvolution shows the presence of 3 peaks at 399.0, 400.3, and 401.2 eV, which correspond to Sp 2 bonded N in triazine ring (C-N�C), the tertiary nitrogen N-(C) 3 group in the heptazine structure, and N in the C-N-H group, respectively [39,40]. For the oxygen element (Figure 3(c)), the peak deconvolution of O1s indicates the existence of three components: lattice oxygen of Fe-O bond (529.7 eV), oxygen of surface hydroxyl-OH (531.0 eV), and the one of adsorbed water H 2 O (533.3 eV) [43]. Hence, the results of XPS, XRD, and IR proved that c-Fe 2 O 3 /g-C 3 N 4 composites were successfully synthesized.

Results and Discussion
In order to predict the light absorption ability, all composites and g-C 3 N 4 were measured by the UV-DRS method. Figures 4(a) and 4(b) 5FeCN, 10FeCN, 15FeCN, 20FeCN, and c-Fe 2 O 3 , respectively. So, the increase of c-Fe 2 O 3 content led to decrease of the bandgap energy. In other words, it means that the light absorption region of catalysts shifted more in the visible light wavelength by increasing the c-Fe 2 O 3 content. Hence, the photocatalytic activity of composites is expected to be improved.
To estimate the recombination of electron-hole pair phenomena, the PL spectroscopy was carried out for g-C 3 N 4 , c-Fe 2 O 3 , and 10FeCN composites ( Figure 5(a)). As observed, the intensity of PL spectrum of g-C 3 N 4 is nearly twice in comparison with the one of 10FeCN. No signal was observed on PL spectra of Fe 2 O 3 at activated wavelength.
is result reflected that the presence of c-Fe 2 O 3 and its interface interaction with g-C 3 N 4 seem to inhibit the photogenerated electron-hole pairs recombination, which could improve photoactivity. Figures 5(b), 5(c), and 5(d) show the SEM image and different element distribution of the 10FeCN sample. It is obvious that Fe and O elements or Fe 2 O 3 phase were quite homogenously dispersed on the surface of g-C 3 N 4 . e elemental composition is given in Table 1. It is noted that the experimental composition is quite. e morphology of composite 10FeCN was characterized by TEM ( Figure 6). It is obvious that the particle Fe 2 O 3 is in cubic form with average size of 10 nm. Apart from c-Fe 2 O 3 particles dispersed on the g-C 3 N 4 surface, it seems that some these particles were covered by g-C 3 N 4 layer to form core-shell structure Fe 2 O 3 @g-C 3 N 4 . e photocatalytic activity was evaluated through photoreduction of CO 2 . Figure 7(a) shows the obtained results. Methane was the main production of the reaction, while CO was not noticeable. No product was detected on c-Fe 2 O 3 .
is is explained by the more positive value of its conduction band potential in comparison of that reduction potential of CO 2 /CH 4 (−0.24 V) [2]. It was noted that the presence of c-Fe 2 O 3 remarkably impacted the CH 4 formation. A volcano-like evolution of methane formation as a function of c-Fe 2 O 3 is observed. Concretely, under the same operational conditions, the amount of CH 4 formed is 0.4 × 10 −2 , 2.1 × 10 −2 , 2.8 × 10 −2 , 2.9 × 10 −2 , and 1.0×10 −2 μmol g −1 h −1 for g-C 3 N 4 , 5FeCN, 10FeCN, 15FeCN, and 20FeCN, respectively. Hence, the maximum CH 4 formation quickly increased when rising the c-Fe 2 O 3 content and reached the maximum with 10FeCN and 15FeCN composites. us, compared to pristine g-C 3 N 4 , the produced amount of CH 4 over 10FeCN and 15FeCN composites was 7 fold. However, at higher c-Fe 2 O 3 content, e.g., 20wt.% Fe 2 O 3 , CH 4 production dropped rapidly to 1.0×10 −2 μmol g −1 h −1 . In comparison with the research on urchin-like Fe 2 O 3 /g-C 3 N 4 and dendrite-structured Fe 2 O 3 /g-C 3 N 4 catalysts for CO 2 photoreduction in the gas phase, CH 4 gas was the preferred product on this catalyst, instead of CO [32,33]. Hence, the obtained results show interesting photocatalytic activity of the synthesized composite materials and also prove the good combination of two phases, c-Fe 2 O 3 and g-C 3 N 4 , to make new active photocatalysts. e small formed CH 4 amount was probably due to the low power of xenon lamp (only 150W). To justify these results, a blank test (without catalyst) and a test on g-C 3 N 4 under N 2 gas were carried out. No CH 4 amount was detected for these ones. In addition, the test was performed also on the mixture of 10% (wt) of c-Fe 2 O 3 with g-C 3 N 4 (Figure 7(a)), and the CH 4 product is under detectable limit that could be evidence for the absence of phase interaction between cFe 2 O 3 and g-C 3 N 4 as well as the presence of c-Fe 2 O 3 hindering the illumination on g-C 3 N 4 .
Generally, the outstanding photoactivity of c-Fe 2 O 3 /g-C 3 N 4 composite materials compared to g-C 3 N 4 is assigned to better light-harvesting ability and charge transfer than those of single phase of c-Fe 2 O 3 or g-C 3 N 4 . It should be emphasized that the charge transfer process in the Z-scheme structure not only reduces photoelectron-hole pair recombination but also makes photocatalyst, the redox potential, stronger: CB potential more negative and VB potential more positive. In addition, it seems that the quite small size   Figure 7: CH 4 production on g-C 3 N 4 and different xFeCN composites (a); proposed mechanism of CO 2 photoreduction on Z-scheme structure of c-Fe 2 O 3 /g-C 3 N 4 (b).  Journal of Analytical Methods in Chemistry cover g-C 3 N 4 surface and inhibits the irradiation on this phase. Hence, the c-Fe 2 O 3 quantity of 10% (wt) was the most suitable to obtain the highest photoactivity. From UV-DRS spectrum and Mulliken electronegativity theory, the valance band (VB) and conduction band (CB) of c-Fe 2 O 3 and g-C 3 N 4 were calculated [30]. According to this theory, E CB � X−E c −0.5E g , where E CB is the conduction band edge energy, X is the electronegativity of the semiconductor (equal 5.825 eV for Fe 2 O 3 and 4.72 eV for g-C 3 N 4 ), E c , equal 4.5 eV, is the energy of free electrons with hydrogen scale, and E g is the bandgap energy of semiconductor. Basing on this equation and concrete value of each parameter, it is found that VB and CB of c-Fe 2 O 3 and g-C 3 N 4 are +2.23, +0.43 eV and +1.53, −1.10 eV, respectively. Figure 7(b) shows a possible photoreduction mechanism of CO 2 photoreduction into CH 4 . According to this Z-scheme composite, all the two phases, c-Fe 2 O 3 and g-C 3 N 4, harvested the irradiated light and generated electron-hole pairs. e photogenerated electron located on CB of c-Fe 2 O 3 migrated and recombined with photogenerated holes located on the valance band (VB) of g-C 3 N 4 . is process allows broadening the light absorption region and also prevents recombination of photogenerated electron-hole pairs that both improved the photocatalytic activity. e details of this process are presented through the following equations:

Conclusions
In this work, different Z-scheme photocatalysts, c-Fe 2 O 3 /g-C 3 N 4, were synthesized by simple methods of calcination and impregnation-precipitation. XRD, IR, and XPS characterizations of these materials confirmed the formation of composite structure of these photocatalysts, wherein c-Fe 2 O 3 grew on g-C 3 N 4 surface. SEM analysis highlighted a good distribution of Fe on the surface of g-C 3 N 4 support. e UV-DRS and PL spectra of the photocatalyst containing 10 wt.% c-Fe 2 O 3 evidenced interface interaction of the two phases of c-Fe 2 O 3 and g-C 3 N 4 . e catalytic performance of these materials was evaluated through the CO 2 photoreduction in the gaseous phase. Methane was found as the main product, while no trace of CO was observed. According to CH 4 production, the photocatalytic activity followed the following order: 15FCN (2.9×10 −2 μmol g −1 h −1 ) 10FeCN (2.8×10 −2 μmol g −1 h −1 ) > 5FeCN (2.1×10 −2 μmol g −1 h −1 ) > 20FeCN (1.0 × 10 −2 μmol g −1 h −1 ) > g-C 3 N 4 (0.4×10 −2 μmol g −1 h −1 ). Hence, 10FeCN and 15FeCN composites had the highest photoactivity, which is approximately 7 times higher than that on bulk g-C 3 N 4 . is could be explained by a synergy combination and interaction of the two phases c-Fe 2 O 3 and g-C 3 N 4 , leading to the Z-scheme structure composites, which improved light absorption with red-shift light and also diminishes photogenerated electron-hole pairs recombination. ese outstanding results are promising for the design of a low-cost and highly efficient photocatalyst for CO 2 reduction on the basis of c-Fe 2 O 3 and g-C 3 N 4 .

Data Availability
e data used to support the findings of this study are included within the article.