Synthesis of Bi2O3/g-C3N4 for enhanced photocatalytic CO2 reduction with a Z-scheme mechanism

Bi2O3/g-C3N4 nanoscale composites with a Z-scheme mechanism were successfully synthesized by high temperature calcination combined with a hydrothermal method. These synthesized composites exhibited excellent photocatalytic performance, especially the 40 wt% Bi2O3/g-C3N4 composite, which produced about 1.8 times the CO yield of pure g-C3N4. The obtained products were characterized by X-ray diffraction (XRD) patterns, X-ray photoelectron spectroscopy (XPS), scanning electron microscope (SEM), transmission electron microscopy (TEM), Brunauer–Emmett–Teller (BET), UV-vis diffuse reflectance spectroscopy (UV-vis DRS) and so on. Characterization results revealed that Bi ions had well covered the surface of g-C3N4, thus restraining the recombination of electron–hole pairs and resulting in a stronger visible-light response and higher CO yield. In addition, the electron transfer process through the Z-scheme mechanism also promoted the photocatalytic activity.


Introduction
In recent years, the continuous development of industry and the serious destruction of forest vegetation have led to difficulties in controlling CO 2 emission, thus seriously threatening the survival and development of mankind. [1][2][3][4][5][6] Recently, the eld of CO 2 photocatalytic reduction which translates CO 2 into some useful substances has attracted many research teams. [7][8][9][10][11][12] As we all know, many narrow band gap semiconductor based photocatalysts have been used in CO 2 photocatalytic reduction. [13][14][15] Among them, g-C 3 N 4 is the most widely used one due to its suitable band gap, simple and convenient preparation and good visible light response. 16,17 However, there are still many obvious shortcomings associated with this catalyst, such as its small range of light response, low separation rate of electron-hole pairs and unsatisfactory performance in the photocatalytic reaction. Lots of methods have been developed to improve the photocatalytic performance of g-C 3 N 4 , such as element modi-cation, 18-20 texture fabrication 21,22 and heterojunction formation. [23][24][25][26][27] Fabrication of a heterojunction between two semiconductors is a promising way among these methods, in which the formation of an internal electric eld and the promoted separation of charge carriers effectively restrains the recombination rate of electron-hole pairs. 28 As an important semiconductor material, Bi 2 O 3 has been applied in many elds like electronic ceramics, optoelectronic devices, high-temperature superconductors, catalysts, and sensors due to its special physical properties and crystal structure. [29][30][31][32][33] What's more, Bi 2 O 3 also has been used as a normal photocatalyst in water splitting and photodegradation pollutant. Unfortunately, its photoreduction performance is poor owing to the low migration of photo-charges, which leads to a fast combination of photogenerated electron-hole pairs. [34][35][36][37] For the improvement of photocatalytic ability over Bi 2 O 3 , the construction of heterojunction is considered as an effective measure to reduce the probability of electron-hole pairs recombination, like BiOCl/Bi 2 O 3 , 38 BiVO 4 /Bi 2 O 3 (ref. 39) and Bi 2 O 3 /TiO 2 . 40 It is known that an internal electric eld is produced aer the formation of a p-n heterojunction. The photocarriers are greatly accelerated by the electric eld, which effectively suppresses the reux of photogenerated carriers and thus improves the photocatalytic performance. 41 Due to the presence of heterojunctions, the excitation wavelength of light is also extended at the same time. 42 From the above analysis, it can be presumed that there exists a possibility to construct a heterostructure between Bi 2 O 3 and g-C 3 N 4 through their mutual activation. However, the use of Bi 2 O 3 /g-C 3 N 4 heterojunctions for CO 2 photoreduction has not been reported. Therefore, Bi 2 O 3 /g-C 3 N 4 compounds with different Bi 2 O 3 contents were prepared and used in CO 2 photocatalytic reduction in this study. It was found that the yields of Bi 2 O 3 /g-C 3 N 4 compounds was higher than single Bi 2 O 3 or g-C 3 N 4 . Moreover, the visible light response is also enhanced. Based on the results of all the characterization techniques, the possible promotion mechanism is also raised.

Chemicals
Urea (H 2 NCONH 3 ) and bismuth nitrate pentahydrate (Bi(NO 3 ) 3 $5H 2 O) were purchased from Sinopharm Chemical Reagent Corp, P. R. China. All of these purchased reagents are of analytical grade and used directly.

Synthesis
Pure g-C 3 N 4 was obtained by heating urea in air at 550 C for 2 h. The Bi 2 O 3 /g-C 3 N 4 composites with different Bi 2 O 3 mass ratio (0, 20, 40, 60 and 80 wt%) were gained through the following procedure: g-C 3 N 4 (0.5 g) was dissolved in 60 mL of ethylene glycol and sonicated for 30 min to yield the g-C 3 N 4 sheet. Subsequently, a certain amount (0.104, 0.208, 0.312, 0.416 g) of Bi (NO 3 ) 3 $5H 2 O and urea (1 g) were added in the solution and stirred for 1 h. The solution was then transfer to Teon-lined autoclave (100 mL) and hydrothermally treated at 180 C for 12 h. The obtained product was rinsed several times with deionized water and dried overnight at 80 C. The hydrothermal product was then calcined in air at 380 C for 2 h to get the nal product. Pure Bi 2 O 3 was obtained by the same procedure without the adjunction of g-C 3 N 4 .

Characterization
The crystal phase of catalyst was determined by X-ray diffraction (XRD, Bruker D8, Cu Ka radiation), and the morphology of the synthesized samples were investigated by scanning electron microscope (SEM, Phillips XL-30 FEG/NEW) and transmission electron microscope (TEM, Phillips Model CM200). The chemical elements of the compositions were analyzed by X-ray photoelectron spectroscopy (XPS, ESCALAB 250xi, USA) and Al Ka radiation sources. The Brunauer-Emmett-Teller (BET, Quantachrome Autosorb-iQ-AG instrument) pore structure and surface area were measured by N 2 adsorption-desorption at À196 C. UV-vis diffuse reectance spectrum was analyzed in the range of 250-800 nm on a spectrophotometer (SHIMADZU UV-3600, Japan) using BaSO 4 as the reectance standard material. Photoluminescence (PL) was measured by a uorescence spectrophotometer (Hitachi F-4600, 325 nm excitation wavelength).

Photoelectrochemical
Photoelectrochemical included electrochemical impedance spectroscopy (EIS) and transient photocurrent responses analysis were carried out on an electrochemical instrument (CHI 660E). A standard three-electrode system was immersed in a Na 2 SO 4 electrolyte solution (0.5 M). The Pt and Ag/AgCl electrodes were used as counter and reference electrodes, respectively. FTO conductive glass covered by the synthesized sample was used as working electrode, and it was gained by the following method: mixture solution was consisted of Naon (20 mL, 5%) and ethanol (1 mL). Aer that, 10 mg of the synthesized sample was dropped into the mixed solution and ultrasonically dispersed (2 h), then drop the slurry (0.1 mL) on 1 Â 1 cm FTO glass. Therefore, the sample was well attached to the surface of the glass piece aer evaporation of the ethanol. In addition, photocurrent and EIS measurements were performed under the illumination of the simulated solar light.

Photoactivity
The CO 2 photoreduction experiment was performed with a reactor (500 mL) in a gas-enclosed circulation system. During the reaction, a Xenon lamp (300 W) was used as the light source. The experimental procedure was designed as follows: 50 mg sample was dispersed in 100 mL deionized water, then magnetic stirring was performed at an appropriate rotation speed. The reactor was vacuum treated and 100 kPa of high purity CO 2 was passed into the reactor under the throttling of airow. Then this pressure was kept for 30 min to obtain the balance of adsorption-desorption. Throughout the experiment, the temperature was maintained at 25 C. The gas (0.15 mL) in reactor was obtained and analyzed by a gas chromatography (GC-2010 Plus, SHIMADZU, Japan) in the course of the reaction.

SEM, XRD and TEM
The SEM patterns of g-C 3 N 4 , Bi 2 O 3 and Bi 2 O 3 /g-C 3 N 4 composites are exhibited in Fig. 1. Clearly, the g-C 3 N 4 exhibits a curled layered structure (Fig. 1a), and the Bi 2 O 3 shows a distinct rodlike structure (Fig. 1b). Fig. 1c shows the SEM image of 40 wt% Bi 2 O 3 /g-C 3 N 4 sample and its corresponding elemental mapping images. It could be observed from Fig. 1c that 40 wt% Bi 2 O 3 /g-C 3 N 4 composite perfectly retains the form of Bi 2 O 3 and g-C 3 N 4 , showing a curled sheet structure. In addition, the mapping images of 40 wt% Bi 2 O 3 /g-C 3 N 4 composite in Fig. 1c also reveals the coexistence of C, N, O and Bi elements.
The XRD patterns of all composites are displayed in Fig. 2. In the pattern of g-C 3 N 4 , the diffraction peak observed at 27.3 could be attributed to the (002) plane of g-C 3 N 4 (JCPDS 87-1526). The peak at 12.7 is the (100) plane of g-C 3 N 4 , corresponding to the in-plane structure packing motif of tri-s-triazine units. 43 The synthesized Bi 2 O 3 represents six major peaks at 2q ¼ 27.  421) and (423) planes respectively, which is in consistent with the b-Bi 2 O 3 (JCPDS 65-1209). As the content of Bi 2 O 3 in Bi 2 O 3 /g-C 3 N 4 samples increase, the intensity of g-C 3 N 4 peak becomes weaker. As a contrast, the peak intensity of Bi 2 O 3 becomes stronger, especially for the diffraction peak at 27.8 , revealing the increased crystallinity of Bi 2 O 3 .
The TEM images of samples are demonstrated in Fig. 3. Fig. 3a and b reveal the microscale morphology of g-C 3 N 4 and Bi 2 O 3 , which are characterized by a curled edge and a rod-like structure respectively, as also revealed by the SEM patterns. The TEM pattern of 40 wt% Bi 2 O 3 /g-C 3 N 4 composite shows that Bi 2 O 3 and g-C 3 N 4 are combined together uniformly (Fig. 3c). In addition, the 0.318 nm interplanar spacing corresponding to the (201) crystal plane of the cubic Bi 2 O 3 is visible in the high resolution TEM image (Fig. 3d). 44 The results show that g-C 3 N 4 and the rod-like Bi 2 O 3 have deeply combined together in 40 wt% Bi 2 O 3 /g-C 3 N 4 composite.

XPS, BET and UV
The compositions of chemical elements on g-C 3 N 4 , Bi 2 O 3 and 40 wt% Bi 2 O 3 /g-C 3 N 4 were analyzed by XPS. As the XPS spectra show (Fig. 4a)    and Bi 4f are also exhibited in Fig. 4. In the C 1s spectrum of g-C 3 N 4 , two sub-bands centered at 287.4 and 284.0 eV could be observed (Fig. 4b), corresponding to the N-C]N group and the C-C bond, respectively. 45  Nitrogen adsorption-desorption measurements were performed at À196 C to analyze the textural features of the Bi 2 O 3 , g-C 3 N 4 and 40 wt% Bi 2 O 3 /g-C 3 N 4 composite. All samples show a type IV isotherm with a hysteresis loop (Fig. 5), revealing the presence of mesoporous structure into the composite. The specic surface area and average pore size of photocatalysts are listed in Table 1. Obviously, 40 wt% Bi 2 O 3 /g-C 3 N 4 composite possesses the largest specic surface area (136.1 m 2 g À1 ), while that for g-C 3 N 4 and Bi 2 O 3 are 66.7 m 2 g À1 and 53.2 m 2 g À1 respectively. Correspondingly, more active sites are available on 40 wt% Bi 2 O 3 /g-C 3 N 4 composite. The BJH pore size distribution of g-C 3 N 4 , Bi 2 O 3 and 40 wt% Bi 2 O 3 /g-C 3 N 4 is shown in the inset of Fig. 5. Obviously, all the three samples exhibit a fairly narrow and limited pore size distribution between 3.6-4.1 nm.
The optical properties of Bi 2 O 3 , g-C 3 N 4 , and Bi 2 O 3 /g-C 3 N 4 composites were tested by UV-vis measurement, and the relevant data were converted from the Kubelka-Munk equation. The 450 nm absorption edge of g-C 3 N 4 could be observed in Fig. 6a, while the Bi 2 O 3 shows a similar absorption properties. It can be found that Bi 2 O 3 /g-C 3 N 4 samples could obtain more photons during the reaction according to the vertical coordinates, which is good to the photoreduction of CO 2 . The band gap energy of Bi 2 O 3 and g-C 3 N 4 could be gured out by the formula: ahv ¼ A (hn À E g ) n/2 , where a, h, n, A and E g are the

Photocatalytic activity, PL and photoelectrochemical
The photocatalytic activities of all samples are tested and CO was found as the main product. In addition, the control experiments were also conducted. No hydrocarbon products are tested without simulated solar light or sample, revealing that the conditions mentioned above are essential for the CO 2 photocatalytic reduction. Fig. 7a demonstrated the amount of CO production over the photocatalysts during the light irradiation. It could be clearly seen that the pure Bi 2 O 3 sample has not produced any CO, which means that pure Bi 2 O 3 cannot reduce CO 2 to CO under photocatalytic conditions, due to the fact that the CB edge of Bi 2 O 3 is lower than the reduction potential of CO 2 /CO (À0.52 V vs. NHE). 54 The yield of g-C 3 N 4 is also not high, indicating that the photocatalytic performance of pure catalyst is not very well, which can be ascribed to the fast recombination rate of electron-hole pairs in reaction. A great enhancement of CO yield is detected in terms of the Bi 2 O 3 /g-C 3 N 4 composites. Moreover, the photocatalytic activity enhances as the increasing Bi 2 O 3 content in the composites, which is owing to the increased quantity of the heterojunctions. During the reaction, the highest CO yield (22.5 mmol g À1 ) is achieved on 40 wt% Bi 2 O 3 /g-C 3 N 4 , which is about 1.8 times the CO yield of g-C 3 N 4 . However, the photocatalytic performance would decline along with the excessive addition of Bi 2 O 3 , which might be aroused by the reduce of the specic surface area, and the excessive Bi 2 O 3 might act as the recombination core of electron-hole pairs   during the reaction. 55,56 The photocatalytic stability of all composites and g-C 3 N 4 are shown in the Fig. S1, † the yield of CO keeps stable during three experimental runs, revealing that the performance of all composites and g-C 3 N 4 are stable during the reaction. To further probe the truth, the separation efficiency was characterized by PL. As shown in Fig. 7b, the emission peak of Bi 2 O 3 is almost invisible in the 450-500 nm range, which may be due to a trace of surface defects or surface oxygen vacancies of Bi 2 O 3 . Compared with Bi 2 O 3 , a strong peak concentrated at 450 nm could be noticed in the g-C 3 N 4 spectrum, which reects the fast recombination rate of electrons-hole pairs. 57 Different from g-C 3 N 4 , the Bi 2 O 3 /g-C 3 N 4 composites exhibit a moderate peak intensity in the pattern, suggesting that heterojunctions are built among Bi 2 O 3 and g-C 3 N 4 , so the holes and electrons are effectively separated. Therefore, the recombination rate is greatly reduced, which is good to enhance the activity of CO 2 photoreduction. In addition, the 40 wt% Bi 2 O 3 /g-C 3 N 4 has the lowest peak intensity among all the synthesized Bi 2 O 3 /g-C 3 N 4 samples.
In addition to PL measurements, photo-electrochemical measurements are used to further analyze the properties of photoelectrons in Bi 2 O 3 , g-C 3 N 4 and 40 wt% Bi 2 O 3 /g-C 3 N 4 . Fig. 8a shows the photocurrent response of the samples under visible light absorption. Transient photocurrent measurements could show more evidence for the rapid electron transfer efficiency in the photocatalyst. In the light-on and light-off tests, all the photocatalysts show rapid and intense photocurrent response, indicating that the photocatalytic activity is relatively stable. Both Bi 2 O 3 and g-C 3 N 4 exhibit a rapidly increasing current when light is on and an abruptly decreasing signal when the light is off. This transient response behavior might be resulted from the capture and release of electrons caused by surface defects. 58 In addition, the photocurrent intensity of 40 wt% Bi 2 O 3 /g-C 3 N 4 heterostructured composite is 2.5 and 1.5 times the intensity of Bi 2 O 3 and g-C 3 N 4 , respectively, indicating that 40 wt% Bi 2 O 3 /g-C 3 N 4 has higher charge separation efficiency and better photocatalytic properties than pure catalysts. To conrm this inference, the EIS experiment was also performed. The semicircular Nynquist plots of the samples are presented in Fig. 8b. Obviously, a smaller semicircle diameter could be detected on the plot of 40 wt% Bi 2 O 3 /g-C 3 N 4 than that of pure Bi 2 O 3 and pure g-C 3 N 4 , indicating that the electron transfer rate is faster in 40 wt% Bi 2 O 3 /g-C 3 N 4 , 59 which is corresponding to the previous photocatalytic activity results.

Mechanism
According to the characterization and CO 2 photocatalytic reduction test results of the prepared samples, the reaction mechanism of CO 2 reduction over Bi 2 O 3 /g-C 3 N 4 composites were proposed. As shown in Fig. 9a, the valence band (VB) and conduction band (CB) of a semiconductor could be determined using the following equation: 60 E 0 CB ¼ c À E C À 1/2E g . Where c is the absolute electronegativity of the semiconductor (c for Bi 2 O 3 is 5.986 eV), E C is the energy of free electrons in the hydrogen size (4.5 eV) and E g is the band gap of the semiconductor. 61,62 Combined with the results in Fig. 6b, the conduction band (CB) and valence band (VB) values for the pure Bi 2 O 3 are 0.07 eV and 2.90 eV, respectively. For g-C 3 N 4 , the VB is À1.57 eV, and the CB is 1.15 eV. 63,64 The charge transfer in the interface can follow a double-transfer mode or a Z-scheme transfer mechanism. Double-transfer means that the CO 2 reduction reaction would happen on the CB of Bi 2 O 3 , and the H 2 O oxidation reaction would occur on the VB of g-C 3 N 4 . In fact, the CB of Bi 2 O 3 is lower than the reduction potential level of CO 2 /CO (À0.52 V vs. NHE), thus the reduction of CO 2 would not proceed and CO would not be generated, which means that double-transfer mechanism is not suitable here to explain. Z-scheme transfer mechanism is different from the double-transfer mechanism, as shown in Fig. 9b, the photo-induced electrons on the surface of Bi 2 O 3 can easily be transferred to g-C 3 N 4 through the interface, the holes on VB of Bi 2 O 3 are captured by H 2 O molecules to produce O 2 and protons. At the same time, the CO 2 molecules react with the electrons on the CB of g-C 3 N 4 to generate CO and H 2 O with the participation of protons. Moreover, the formation of other products including CH 4 , H 2 , and O 2 were also observed in the experimental process, and the charge balance table is shown in Table 2. The enhanced activity of 40 wt% Bi 2 O 3 /g-C 3 N 4 in CO 2 reduction is mainly due to the Z-scheme heterostructure, which effectively delays the rapid recombination of the electron-hole pairs in Bi 2 O 3 and g-C 3 N 4 , therefore improving the photocatalytic activity.

Conclusions
In general, Bi 2 O 3 /g-C 3 N 4 nanoscale composites with heterojunction was successfully synthesized by the combination of high temperature calcination and hydrothermal method. The 40 wt% Bi 2 O 3 /g-C 3 N 4 composite has the highest CO yield in the CO 2 photocatalytic reduction, which is 1.8 times the yield of that g-C 3 N 4 . This obvious improvement in photoactivity is mainly due to the effective separation of electron-hole pairs and successive charge transfer through the interface. Moreover, the enhancement of specic surface area and visible light response also upgrade the photocatalytic performance of Bi 2 O 3 /g-C 3 N 4 composite. This study gives some valuable opinion for the research of g-C 3 N 4 -related photocatalysts.

Conflicts of interest
There are no conicts to declare.