Porous Oxygen-Doped g-C3N4 with the Different Precursors for Excellent Photocatalytic Activities under Visible Light

Antibiotic contamination has received widespread attention globally. In this work, the oxygen-doped porous graphite carbonitride (g-C3N4) was prepared with urea and ammonium oxalate (CNUC) or urea and glycine (CNUG) as precursors by thermal polymerization. Using bisphenol A (BPA) as a probe and CNUC or CNUG as photocatalysts, the removal performance test was carried out. Meanwhile, all prepared photocatalysts were characterized by XRD, FT-IR, SEM, TEM, XPS, UV-Vis DRS, PL and EIS. Under visible light irradiation, both CNUC and CNUG exhibited about seven and five times greater photocatalytic activity than that of pure g-C3N4, respectively. The radical capture experiments verified that superoxide radicals (•O2−) and holes (h+) were the main active species in the photocatalytic degradation of BPA by CNUC, and the possible photocatalytic mechanism of CNUC was proposed. In addition, all these results indicate that CNUC catalyst can effectually inhibit the photocorrosion and keep superior stability. The proposed technique provides a prospective approach to develop nonmetal-modified photocatalysts for future applications.


Introduction
Bisphenol A (BPA), also named diphenol-based propane, is a low-toxic compound and an important raw material for the production of polycarbonate, epoxy resin, phenolic resin, some polysulfones, and some special materials [1]. Because of its light weight, transparency, and toughness, BPA is widely used in manufacturing notebooks, mobile phones, baby bottles, and other food and beverage containers [2]. However, experiments have proved that BPA has an endocrine disruptor effect, which can be transferred to the environment through plastics, and it is increasingly detected in water environments, posing a serious threat to the ecosystem [3]. At present, a variety of treatment technologies have been studied, such as chemical reactions [4], biodegradation [5], and physical absorption [6]. The removal of BPA could reduce harm to the human body and could reduce water pollution. Therefore, removing BPA is the focus of current research. Among various methods, photocatalytic technology is the most likely and effective method to remove phenolic pollutants in water [7]. Photocatalytic technology converts low-density solar energy into high-density chemical energy and can directly use solar energy to degrade and mineralize various organic pollutants in water and air. It has the advantages of mild reaction conditions, no pollution, and low cost [8], making it an ideal piece of technology for environmental pollution control and clean energy production. However, the current problem with photocatalysts is the rapid recombination of photogenerated electron-hole pairs and the narrow photoresponse range of the photocatalyst. [9]. Therefore, the development of high-efficiency photocatalysts is still one of the focuses and core issues of photocatalysis scientific research at present, and it will likely still be in the future.
Graphite carbonitride (g-C 3 N 4 ) has gradually become a research hotspot since its discovery in 2009 as an n-type nonmetallic polymer semiconductor for the photocatalytic

Materials and Reagents
All chemical reagents and materials were purchased and used without further purification. Urea ((NH 2 ) 2 CO) was bought from Profile (Wuxi, China). Ammonium oxalate ((NH 4 ) 2 C 2 O 4 ) and glycine (C 2 H 5 NO 2 ) were purchased from Aladdin (Shanghai, China). Ultrapure water (>18.2 MΩ) was used in this study.

Preparation
The detailed preparation processes were provided in the Supplementary Materials section.

Characterization
The crystal structure of the prepared samples was characterized by X-ray diffraction (XRD), and the diffraction system was German BRUKER-AXS D8(Beijing Beishide Instrument Co., Ltd., Beijing, China). The chemical composition and element bonding of the prepared samples were analyzed by X-ray photoelectron spectroscopy (XPS) (Beijing Oubotong Optical Technology Co., Ltd., Beijing, China), and the XPS spectrum was recorded on the ESCALAB250Xi system equipped with a monochromatic Al Kα X-ray source. The chemical composition of the sample and the nature of the chemical bond are characterized by Fourier Transform Infrared (FT-IR) Spectrometer, Nicolet380 (Beijing Beishide Instrument Co., Ltd., Beijing, China.). The surface and micro morphology of the sample were characterized by using a scanning electron microscope (SEM) (S-3400) and a transmission electron microscope (TEM) (Tecnai) of Hitachi, Japan. The optical properties of the material were characterized by a diffuse reflectance spectrophotometer (UV-vis DRS, UV-3900) manufactured by Beijing General Analysis Co., Ltd., Beijing, China. The electrochemical properties of the material were recorded using the CHI 660B electrochemical system (Beijing Pofila Co., Ltd., Beijing, China) to record the impedance spectroscopy (EIS) Nyquist plot.

Photocatalytic Degradation of BPA
BPA was used as the target pollutant to evaluate the photocatalytic activity of the catalyst. We used a wavelength above 420 nm and a 500 W xenon lamp as a light source. First, we dispersed 50 mg of photocatalyst in 50 mL of BPA aqueous solution (20 mg/L), and magnetically stirred for 10 min in the dark to ensure that the adsorption-desorption equilibrium was reached before irradiation. During the reaction, stirring was continued to ensure that the mixture was in suspension. Every 10 min, we used a pipette to take out 5 mL of the solution and filter it with a 0.22 µm microporous membrane to remove the catalyst. Finally, high-performance liquid chromatography (HPLC) was used to measure the bisphenol A solution. We eluted bisphenol A with mobile phase methanol and water (55:45 by volume). The degradation rate of BPA was calculated using the following Equation (1): where C t is the concentration of BPA at time t and C 0 is the initial concentration of BPA before the reaction.

Photocatalytic Performances Test
Under the irradiation of visible light, using the prepared catalyst as a probe and BPA as the target pollutant, the photocatalytic properties of pure g-C 3 N 4 , CNUC and CNUG were explored. As shown in Figure 1a, the degradation rate of BPA with pure g-C 3 N 4 as a photocatalyst was only 14% after a reaction of 60 min, indicating that only 14% BPA was removed. Compared with pure g-C 3 N 4 , the four CNUC 1-4 samples presented higher photocatalytic efficiency for BPA degradation. The degradation rate of pure BPA is 11%. The degradation rate of BPA with CNUC 1 , CNUC 2 , CNUC 3 , and CNUC 4 as photocatalysts was 48%, 52%, 81%, and 53%, respectively, corresponding to 3, 4, 7, and 4 times higher rates than that of BPA with pure g-C 3 N 4 , as a photocatalyst. The photodegradation performance was observed to increase with the increase in (NH 4 ) 2 C 2 O 4 loading from 1 to 3 mmol and deteriorated with a further increase in (NH 4 ) 2 C 2 O 4 loading to 4 mmol, probably due to the blocking effect of light absorption when a higher amount of (NH 4 ) 2 C 2 O 4 was incorporated. Similar phenomena were observed for CNUG samples, as shown in Figure 1b. The degradation rate of BPA with g-C 3 N 4 , CNUG 1 , CNUG 2 , CNUG 3 , and CNUG 4 as photocatalysts was 35%, 62%, 72%, and 42%, respectively, corresponding to a 2.5, 4, 5, and 3 times higher rate than that of BPA with pure g-C 3 N 4 as a photocatalyst. Among all of the CNUC and CNUG samples, CNUC 3 presented the best performance. The photocatalytic degradation kinetics of BPA could be well expressed by pseudo-first-order reaction using the following equation: where C t is the concentration of BPA at a specific time t, C 0 is the original concentration, and k is the rate constant. As shown in Figure 1c, these BPA photodegradation reactions followed first-order kinetics closely. The value of k was calculated to be 0.0125, 0.0165, 0.0278, and 0.0183 min −1 for BPA with CNUC 1 , CNUC 2 , CNUC 3 , and CNUC 4 as photocatalysts, respectively, corresponding to 5, 7, 12, and 8 times faster than that of BPA with pure g-C 3 N 4 as a photocatalyst. Figure 1d presents the pseudo-first-order fitting for BPA with CNUG as a photocatalyst. The obtained rate constant for CNUG 1 , CNUG 2 CNUG 3 , and CNUG 4 was 0.0090, 0.0091, 0.0207, and 0.0072 min −1 , respectively. The highest rate constant observed for BPA with CNUG 3 as a photocatalyst was 9 times higher than that for BPA with pure g-C 3 N 4 as a photocatalyst. removed. Compared with pure g-C3N4, the four CNUC1-4 samples presented higher photocatalytic efficiency for BPA degradation. The degradation rate of pure BPA is 11%. The degradation rate of BPA with CNUC1, CNUC2, CNUC3, and CNUC4 as photocatalysts was 48%, 52%, 81%, and 53%, respectively, corresponding to 3, 4, 7, and 4 times higher rates than that of BPA with pure g-C3N4, as a photocatalyst. The photodegradation performance was observed to increase with the increase in (NH4)2C2O4 loading from 1 to 3 mmol and deteriorated with a further increase in (NH4)2C2O4 loading to 4 mmol, probably due to the blocking effect of light absorption when a higher amount of (NH4)2C2O4 was incorporated. Similar phenomena were observed for CNUG samples, as shown in Figure 1b. The degradation rate of BPA with g-C3N4, CNUG1, CNUG2, CNUG3, and CNUG4 as photocatalysts was 35%, 62%, 72%, and 42%, respectively, corresponding to a 2.5, 4, 5, and 3 times higher rate than that of BPA with pure g-C3N4 as a photocatalyst. Among all of the CNUC and CNUG samples, CNUC3 presented the best performance. The photocatalytic degradation kinetics of BPA could be well expressed by pseudo-first-order reaction using the following equation: where Ct is the concentration of BPA at a specific time t, C0 is the original concentration, and k is the rate constant. As shown in Figure 1c, these BPA photodegradation reactions followed first-order kinetics closely. The value of k was calculated to be 0.0125, 0.0165, 0.0278, and 0.0183 min −1 for BPA with CNUC1, CNUC2, CNUC3, and CNUC4 as photocatalysts, respectively, corresponding to 5, 7, 12, and 8 times faster than that of BPA with pure g-C3N4 as a photocatalyst. Figure 1d presents the pseudo-first-order fitting for BPA with CNUG as a photocatalyst. The obtained rate constant for CNUG1, CNUG2 CNUG3, and CNUG4 was 0.0090, 0.0091, 0.0207, and 0.0072 min −1 , respectively. The highest rate constant observed for BPA with CNUG3 as a photocatalyst was 9 times higher than that for BPA with pure g-C3N4 as a photocatalyst.

Catalysis Structure and Morphology Analysis
The phase composition and the crystal structure of g-C 3 N 4 , CNUC 3 , and CNUG 3 were characterized by XRD, as shown in Figure 2a. All samples revealed a typical g-C 3 N 4 layered structure, and the result show the presence of two diffraction peaks at 13.0 • and 27.5 • , which were attributed to the (100) and (002) crystal planes, respectively [24]. The diffraction peak at 2θ = 27.5 • was attributed to the interlayer stacking of aromatic compounds, and the diffraction peak at 2θ = 13.0 • was derived from the in-plane stacking of the tris-s-triazine ring. [25]. Compared with that of g-C 3 N 4 , the (002) diffraction peak of CNUC 3 and CNUG 3 was shifted from 27.61 • to 27.31 • , indicating that the interplanar stacking distance increased after O-doping.

Catalysis Structure and Morphology Analysis
The phase composition and the crystal structure of g-C3N4, CNUC3, and CNUG3 were characterized by XRD, as shown in Figure 2a. All samples revealed a typical g-C3N4 layered structure, and the result show the presence of two diffraction peaks at 13.0° and 27.5°, which were attributed to the (100) and (002) crystal planes, respectively [24]. The diffraction peak at 2θ = 27.5° was attributed to the interlayer stacking of aromatic compounds, and the diffraction peak at 2θ = 13.0° was derived from the in-plane stacking of the triss-triazine ring. [25]. Compared with that of g-C3N4, the (002) diffraction peak of CNUC3 and CNUG3 was shifted from 27.61° to 27.31°, indicating that the interplanar stacking distance increased after O-doping. Figure 2b shows the FT-IR spectra of g-C3N4, CNUC3, and CNUG3. All of the samples revealed similar FT-IR patterns. g-C3N4 had an obvious characteristic peak at about 809 cm −1 , which was assigned to the tri-s-triazine ring. The multiple diffraction peaks within 1200 to 1700 cm −1 represented the stretching vibration of the C-N heterocyclic ring [26], and the broad peak between 3000 to 3500 cm −1 was caused by the stretching vibration of N-H and the water molecules adsorbed on the catalyst surface. [27]. Figure 2b shows that the enhanced peak intensity of CNUC3 at between 3000 and 3500 cm −1 indicated the presence of more −OH bonds on the surface of CNUC3. The morphologies of the g-C3N4, CNUC3, and CNUG3 were characterized by SEM and TEM. As shown in Figure 3a,d,e, pure g-C3N4 showed large bulk particles, a smooth surface and a thick 2D sheet-like structure, while CNUC3 had a porous structure with a skeleton shape, owing to the release of gases (NH3, H2O, CO, and CO2) due to the decomposition of (NH4)2C2O4 during the pyrolysis process [28]. CNUG3 showed obvious flakelike structures with some disordered pores, which might have resulted from the evolution of carbonaceous gas from glycine oxidation at high temperatures [29]. Compared with CNUC3, CNUG3 had fewer holes. In addition, the porous structure provided the CNUC3 and CNUG3 catalysts with a higher specific surface area and a larger pore volume, thereby exposing more active sites and resulting in enhanced photocatalytic activity. The TEM image showed that the as-prepared CNUC3 samples consisted of the ultrathin 2D layers with abundant mesoporous layers compared to g-C3N4 and CNUG3 with a compact microstructure. From the morphological point of view, consistent with the TEM image, the pore structure of CNUG3 was not as good as that of CNUC3. The knitting of ultrathin and surface porous CNUC3 nanosheets could shorten the distance between the body and the surface diffuse migration of photo-generated charges, thus inhibiting the photogenerated charges in the internal recombination. Figures S1 and S2 show the energy-dispersive Xray analysis maps of the CNUC3 and CNUG3 sample. The C, N, and O elements were uniformly dispersed on the CNUC3 and CNUG3 photocatalysts.  Figure 2b shows the FT-IR spectra of g-C 3 N 4 , CNUC 3 , and CNUG 3 . All of the samples revealed similar FT-IR patterns. g-C 3 N 4 had an obvious characteristic peak at about 809 cm −1 , which was assigned to the tri-s-triazine ring. The multiple diffraction peaks within 1200 to 1700 cm −1 represented the stretching vibration of the C-N heterocyclic ring [26], and the broad peak between 3000 to 3500 cm −1 was caused by the stretching vibration of N-H and the water molecules adsorbed on the catalyst surface. [27]. Figure 2b shows that the enhanced peak intensity of CNUC 3 at between 3000 and 3500 cm −1 indicated the presence of more −OH bonds on the surface of CNUC 3 .
The morphologies of the g-C 3 N 4 , CNUC 3 , and CNUG 3 were characterized by SEM and TEM. As shown in Figure 3a,d,e, pure g-C 3 N 4 showed large bulk particles, a smooth surface and a thick 2D sheet-like structure, while CNUC 3 had a porous structure with a skeleton shape, owing to the release of gases (NH 3 , H 2 O, CO, and CO 2 ) due to the decomposition of (NH 4 ) 2 C 2 O 4 during the pyrolysis process [28]. CNUG 3 showed obvious flake-like structures with some disordered pores, which might have resulted from the evolution of carbonaceous gas from glycine oxidation at high temperatures [29]. Compared with CNUC 3 , CNUG 3 had fewer holes. In addition, the porous structure provided the CNUC 3 and CNUG 3 catalysts with a higher specific surface area and a larger pore volume, thereby exposing more active sites and resulting in enhanced photocatalytic activity. The TEM image showed that the as-prepared CNUC 3 samples consisted of the ultrathin 2D layers with abundant mesoporous layers compared to g-C 3 N 4 and CNUG 3 with a compact microstructure. From the morphological point of view, consistent with the TEM image, the pore structure of CNUG 3 was not as good as that of CNUC 3 . The knitting of ultrathin and surface porous CNUC 3 nanosheets could shorten the distance between the body and the surface diffuse migration of photo-generated charges, thus inhibiting the photogenerated charges in the internal recombination. Figures S1 and S2 show the energy-dispersive X-ray analysis maps of the CNUC 3 and CNUG 3 sample. The C, N, and O elements were uniformly dispersed on the CNUC 3 and CNUG 3 photocatalysts. pore volumes, which was consistent with the result of TEM. Furthermore, the surface area (SBET) of CNUC3 was 48.28 m 2 g −1 , which was higher than that of CNUG3 (45.27 m 2 g −1 ) and much higher than that of g-C3N4 (14.93 m 2 g −1 ). The increase in the surface area of the CNUC3 sample was conducive to photocatalytic performance. According to the existing literature, a larger area could increase the number of adsorption and photocatalytic reaction sites, which was conducive to photocatalytic reactions because it enhanced mass transfer and light collection efficiency. The elemental compositions and the chemical binding states of the photocatalyst composites were analyzed by XPS. The XPS spectra of g-C3N4, CNUC3, and CNUG3 are presented in Figure 4a. In Figure 4b, it could be seen that the C 1s XPS spectrum of g-C3N4 was mainly divided into three characteristic peaks at 284.60, 288.1 and 293.5 eV. The binding energy at 284.6 eV was indefinite carbon (C-C/C=C) [31]. The characteristic peak at 288.1 eV was attributed to the sp 2 bonding carbon in the s-triazine ring (N-C=N), and this binding energy was the main carbon characteristic peak of g-C3N4 [32]. The peak at 293.5 eV could be assigned to π electronic excitation [33]. The new peaks at 285.9 eV confirmed the generation of C-O bonds in CNUC3 and CNUG3 samples [34]. The peak area of the C-O bond was noted to be the largest for CNUC3, indicating that the amount of oxygen doped in CNUC3 was the highest. The N 1s spectrum of the sample was divided into four The N 2 adsorption-desorption isotherms of the as-prepared samples are shown in Figure S3. The isotherm of the sample was type IV and the hysteresis loop was type H3 [30]. As displayed in Table S1, compared with g-C 3 N 4 and CNUG 3 , CNUC 3 showed higher pore volumes, which was consistent with the result of TEM. Furthermore, the surface area (S BET ) of CNUC 3 was 48.28 m 2 g −1 , which was higher than that of CNUG 3 (45.27 m 2 g −1 ) and much higher than that of g-C 3 N 4 (14.93 m 2 g −1 ). The increase in the surface area of the CNUC 3 sample was conducive to photocatalytic performance. According to the existing literature, a larger area could increase the number of adsorption and photocatalytic reaction sites, which was conducive to photocatalytic reactions because it enhanced mass transfer and light collection efficiency.
The elemental compositions and the chemical binding states of the photocatalyst composites were analyzed by XPS. The XPS spectra of g-C 3 N 4 , CNUC 3 , and CNUG 3 are presented in Figure 4a. In Figure 4b, it could be seen that the C 1s XPS spectrum of g-C 3 N 4 was mainly divided into three characteristic peaks at 284.60, 288.1 and 293.5 eV. The binding energy at 284.6 eV was indefinite carbon (C-C/C=C) [31]. The characteristic peak at 288.1 eV was attributed to the sp 2 bonding carbon in the s-triazine ring (N-C=N), and this binding energy was the main carbon characteristic peak of g-C 3 N 4 [32]. The peak at 293.5 eV could be assigned to π electronic excitation [33]. The new peaks at 285.9 eV confirmed the generation of C-O bonds in CNUC 3 and CNUG 3 samples [34]. The peak area of the C-O bond was noted to be the largest for CNUC 3 , indicating that the amount of oxygen doped in CNUC 3 was the highest. The N 1s spectrum of the sample was divided into four characteristic peaks located at 398.5, 399.6, 400.9 and 404.2 eV. The characteristic peak of 398.5 eV was the pyridine nitrogen of the triazine ring in C=N-C. The peak with binding energy at 399.6 eV corresponded to the bridging N atom in tertiary nitrogen (N-(C) 3 ). The binding energy was at 400.9 and 404.2 eV. It is attributed to the presence of nitrogen atoms in the amino group (C-N-H) and π excitation, respectively [35]. The peak area of N 1s peak at 398.5 eV attributed to the C=N-C group for CNUG 3 was reduced compared with that of pure g-C 3 N 4 and CNUG 3 . This might be due to the sp 2 hybridized triazine structure introduced into the N-containing aromatic ring in CNUC 3 , which caused the N atom to be replaced by the multi-electron O atom and gradually destroyed. In the spectrum of O 1s in Figure 4d, the new peaks at 531.5 eV for CNUC 3 and CNUG 3 might be attributed to the formation of O-C species in the crystal lattice by O doping [36]. The peak area at 531.5 eV for CNUC 3 was higher than that of CNUG 3 , indicating that the O atom in ammonium oxalate easily replaced the N atom in the g-C 3 N 4 structure to form C-O. characteristic peaks located at 398.5, 399.6, 400.9 and 404.2 eV. The characteristic peak of 398.5 eV was the pyridine nitrogen of the triazine ring in C=N-C. The peak with binding energy at 399.6 eV corresponded to the bridging N atom in tertiary nitrogen (N-(C)3). The binding energy was at 400.9 and 404.2 eV. It is attributed to the presence of nitrogen atoms in the amino group (C-N-H) and π excitation, respectively [35]. The peak area of N 1s peak at 398.5 eV attributed to the C=N-C group for CNUG3 was reduced compared with that of pure g-C3N4 and CNUG3. This might be due to the sp 2 hybridized triazine structure introduced into the N-containing aromatic ring in CNUC3, which caused the N atom to be replaced by the multi-electron O atom and gradually destroyed. In the spectrum of O 1s in Figure 4d, the new peaks at 531.5 eV for CNUC3 and CNUG3 might be attributed to the formation of O-C species in the crystal lattice by O doping [36]. The peak area at 531.5 eV for CNUC3 was higher than that of CNUG3, indicating that the O atom in ammonium oxalate easily replaced the N atom in the g-C3N4 structure to form C-O.

Optical and Electrochemical Properties
To confirm the light absorption and electronic structure of the catalysts, UV-vis spectroscopy was performed on g-C3N4, CNUC3, and CNUG3. From the UV-vis spectra of g-C3N4, CNUC3, and CNUG3 shown in Figure 5a, we could see that all samples showed strong absorption in the visible light region. The absorption edges of g-C3N4 were located at around 489 nm. Both CNUC3 and CNUG3 samples exhibited a redshifted in the absorption edges at 696 and 650 nm, respectively. These results indicate that CNUC3 and CNUG3 could absorb a wider spectrum and more light photon energy, leading to an increase in the production of electrons and holes. Compared with CNUG3, CNUC3 showed a wider spectrum range, indicating a possibly improved visible light photocatalytic activity. The corresponding optical bandgap energy (Eg) was calculated using the following formula:

Optical and Electrochemical Properties
To confirm the light absorption and electronic structure of the catalysts, UV-vis spectroscopy was performed on g-C 3 N 4 , CNUC 3 , and CNUG 3 . From the UV-vis spectra of g-C 3 N 4 , CNUC 3 , and CNUG 3 shown in Figure 5a, we could see that all samples showed strong absorption in the visible light region. The absorption edges of g-C 3 N 4 were located at around 489 nm. Both CNUC 3 and CNUG 3 samples exhibited a redshifted in the absorption edges at 696 and 650 nm, respectively. These results indicate that CNUC 3 and CNUG 3 could absorb a wider spectrum and more light photon energy, leading to an increase in the production of electrons and holes. Compared with CNUG 3 , CNUC 3 showed a wider spectrum range, indicating a possibly improved visible light photocatalytic activity. The corresponding optical bandgap energy (Eg) was calculated using the following formula: where α, hν, A, and Eg represent the absorption coefficient, light energy, constant, and bandgap, respectively. The bandgap of g-C 3 N 4 , CNUC 3 , and CNUG 3 was determined to be 2.77, 2.65, and 2.45 eV, respectively, as shown in Figure S4. Chen [18], 1.5% SO-g-C 3 N 4 , had a corresponding band gap of 2.63 eV. Compared with Chen, the CNUC 3 and CNUG 3 prepared in this paper significantly reduced the forbidden band width of the catalyst, and the wavelength shifts to the visible light direction, so that the catalyst could make better use of visible light. The narrow band gap could enable the catalyst to respond to visible light, shift the wavelength to the visible light direction, and the catalyst could capture more visible light photons, which helped to improve the photocatalytic performance of the catalyst. The separation efficiency and the transferability of photoexcited electrons and holes were evaluated using the photoluminescence (PL) emission response and EIS. Figure 5b presents the EIS spectra of g-C 3 N 4 , CNUC 3 , and CNUG 3 . The smaller arc radius in the EIS Nyquist diagram, illustrated a stronger photogenerated charge separation and transferability. It could be seen from the EIS Nyquist diagram that CNUC 3 had the smallest arc radius, followed by CNUG 3 , and g-C 3 N 4 had the largest arc radius. This result shows that CNUC 3 had a good photo-generated carrier separation ability and further confirms that CNUC 3 had good photocatalytic activity, which was consistent with the results discussed previously. Figure 5c presents the PL spectra of g-C 3 N 4 , CNUC 3 , and CNUG 3 . Generally, the stronger PL intensity indicated the faster combination of photogenerated charge. As shown in the steady-state PL spectra, g-C 3 N 4 depicted an obvious PL peak with a strong emission intensity. The incorporation of either (NH 4 ) 2 C 2 O 4 or C 2 H 5 NO 2 led to a significant reduction in the intensity of the PL peaks for CNUC 3 and CNUG 3 , indicating the efficient separation of light-induced electron-hole pairs. Since the PL intensity was mainly derived from the recombination of photogenerated carriers in the photocatalyst, the reduction in PL intensity was crucial evidence for increasing the charge separation rate. CNUC 3 showed the lowest fluorescence intensity compared with the other two samples, which was consistent with the EIS spectra. Meanwhile, the emission peaks were red-shifted from 465 to 539 nm due to the O doping, which could promote the π-electron delocalization of the g-C 3 N 4 network.
In order to explore the intermediates and possible degradation pathways generated by BPA in the photocatalytic degradation process, LC−MS was used to analyze the mineralization evolution process of the catalyst degradation of BPA. Similar work had been carried out in the author's previous research [22]. In Figure 6, two possible BPA degradation pathways were proposed [37,38]. According to reports in the literature, the current active oxidative free radicals degrade BPA molecules through two possible mechanisms: C−C bond cleavage and hydroxylation. In the first approach, the benzene ring in BPA undergoes hydroxylation reaction to form dihydroxylated BPA (wt: 259), and the OH group on the dihydroxylated BPA molecule could be further oxidized to form tetracarbonylation BPA (wt: 255), and tetracarbonylation BPA was further oxidized to produce [2E,4Z)-3-(2-(3,4dioxane)hexane-1,5-ethylene-1-yl)] hexane-2,4-dienedioic acid (wt: 287). In the second way, the C−C bond connected to the benzene ring in BPA reacts with the active material to form 4-isopropylphenol (wt: 135), and 4-isopropylphenol was further oxidized to form 1-(2,4-dihydroxy-yphenyl) ethanone (wt: 151). Finally, all of the above aromatic compounds undergo ring-opening reactions to produce small molecular compounds, such as 1-dihydroxybutyric acid (wt: 119), valeric acid (wt: 102), and acetic acid (wt: 60), etc. These small molecular compounds with further mineralization generate CO 2  In order to explore the intermediates and possible degradation pathways generated by BPA in the photocatalytic degradation process, LC−MS was used to analyze the mineralization evolution process of the catalyst degradation of BPA. Similar work had been carried out in the author's previous research [22]. In Figure 6, two possible BPA degradation pathways were proposed [37,38]. According to reports in the literature, the current active oxidative free radicals degrade BPA molecules through two possible mechanisms: C−C bond cleavage and hydroxylation. In the first approach, the benzene ring in BPA undergoes hydroxylation reaction to form dihydroxylated BPA (wt: 259), and the OH group on the dihydroxylated BPA molecule could be further oxidized to form tetracarbonylation BPA (wt: 255), and tetracarbonylation BPA was further oxidized to produce [2E,4Z)-3-(2-(3,4-dioxane)hexane-1,5-ethylene-1-yl)] hexane-2,4-dienedioic acid (wt: 287). In the second way, the C−C bond connected to the benzene ring in BPA reacts with the active material to form 4-isopropylphenol (wt: 135), and 4-isopropylphenol was further oxidized to form 1-(2,4-dihydroxy-yphenyl) ethanone (wt: 151). Finally, all of the above aromatic compounds undergo ring-opening reactions to produce small molecular compounds, such as 1-dihydroxybutyric acid (wt: 119), valeric acid (wt: 102), and acetic acid (wt: 60), etc. These small molecular compounds with further mineralization generate CO2 and H2O.

Mechanisms of Enhanced Photocatalytic Performance
The photocatalytic mechanism of the degradation of organic contaminants in CNUC3 samples under visible light irradiation was explored by free radical trapping experiments. tert-Butyl alcohol (t-BA), p-benzoquinone (BQ), and KI were taken as the quenchers for

Mechanisms of Enhanced Photocatalytic Performance
The photocatalytic mechanism of the degradation of organic contaminants in CNUC 3 samples under visible light irradiation was explored by free radical trapping experiments. tert-Butyl alcohol (t-BA), p-benzoquinone (BQ), and KI were taken as the quenchers for •OH, •O 2 − , and h + , respectively. It could be seen from Figure 7a that t-BA could slightly inhibit the degradation of BPA, and BQ and KI had a significant inhibitory effect on BPA, reducing BPA by 66% and 47.8%, respectively. The above results show that •O 2 − and h + were the main active substances in the degradation of BPA and played a major role in photocatalytic degradation. The bandgaps of g-C 3 N 4 , CNUC 3 , and CNUG 3 were discussed earlier by UV-vis diffuse reflectance spectra. The bandgap of solid solutions of g-C 3 N 4 , CNUC 3 , and CNUG 3 was 2.77, 2.45, and 2.65 eV, respectively. The XPS valence band spectra of g-C 3 N 4 , CNUC 3 and CNUG 3 are shown in Figure 7b. The valence band (VB) spectra for g-C 3 N 4 , CNUC 3 , and CNUG 3 were 0.98, 1.27, and 1.13 eV, respectively, and the corresponding conduction band (CB) calculated from the equation E CB = E VB − E g , was −1.79, −1.18, and −1.52 eV, respectively. The band structures of g-C 3 N 4 , CNUC 3 , and CNUG 3 are presented in Figure 7c. According to previous literature reported [39], the N1 and N4 sites in g-C 3 N 4 were replaced by oxygen atom in CNUC 3 catalyst prepared by ammonium oxalate as precursor, while the N1 site in g-C 3 N 4 was replaced by glycine as precursor [29]. Therefore, CNUC 3 has a narrower band gap, which improves the visible light capture ability and photocatalytic activity.  Figure 7c. According to previous literature reported [39], the N1 and N4 sites in g-C3N4 were replaced by oxygen atom in CNUC3 catalyst prepared by ammonium oxalate as precursor, while the N1 site in g-C3N4 was replaced by glycine as precursor [29]. Therefore, CNUC3 has a narrower band gap, which improves the visible light capture ability and photocatalytic activity. Based on the above experiment and analysis, a possible mechanism for the photodegradation of BPA by CNUC3 under visible light irradiation was proposed, as shown in Figure 8. First, under the irradiation of visible light, photogenerated e − transferred from the VB to the CB of CNUC3 and formed a conjugated delocalized π system, because the potential on CNUC3 CB (−1.18 to −1.79 eV) was more negative than the O2/•O2 − (−0.33 eV vs. NHE). The eon CB could directly react with O2 on the surface of the catalyst to produce •O2 − . Due to the strong oxidation ability of •O2 − radicals, it could directly mineralize BPA into CO2 and H2O. Besides, the VB potential of CNUC3 was lower than the standard oxidation-reduction potential of OH − /•OH (+2.38 eV vs. NHE) [27], so that h + left on the Based on the above experiment and analysis, a possible mechanism for the photodegradation of BPA by CNUC 3 under visible light irradiation was proposed, as shown in Figure 8. Due to the strong oxidation ability of •O 2 − radicals, it could directly mineralize BPA into CO 2 and H 2 O. Besides, the VB potential of CNUC 3 was lower than the standard oxidation-reduction potential of OH − /•OH (+2.38 eV vs. NHE) [27], so that h + left on the VB of CNUC 3 could not oxidize OH − to produce •OH kinetically. However, the h + on CNUC 3 VB could directly react with BPA to produce CO 2 and H 2 O. On the one hand, O atom doping was introduced and porous structure built an inter-bandgap, leading to a narrow bandgap, which could increase the visible light absorption and light quantum efficiencies and further enhanced the photocatalytic activity under visible light. On the other hand, O-doped g-C 3 N 4 could effectively improve the separation of photo-generated carriers and accelerated the improvement of charge transfer efficiency, thereby enhancing its photocatalytic activity. In addition to investigating the photocatalytic activity of the catal of the photocatalyst was also the key to its industrialization. Five phot periments were carried out on CNUC3 to evaluate its light stability and 9a was a photocatalytic cycle experiment diagram. It could be seen fr after five cycles of the catalyst, the degradation efficiency of BPA wa reaction process, a small amount of CNUC3 powder would be removed uous stirring process, and a small amount of catalyst would also be re washing and drying process of the catalyst. It could be seen from the f photocatalytic reaction was cycled three times, the catalyst activity tend phenomenon showed that CNUC3 had good stability and reusability. In to explore whether the crystal form and chemical species of the catalyst cycles, the CNUC3 was characterized by XRD and FT-IR. Figure 9b wa after five cycles. From the figure, it could be seen that after five cycles crystal faces hardly changed. Figure 9c was the FT-IR diagram after five show that the characteristic peaks of the FT-IR spectrum before and hardly changed. From the above results, CNUC3 had good chemical s fectively inhibit photo-corrosion performance. In addition to investigating the photocatalytic activity of the catalyst, the reusability of the photocatalyst was also the key to its industrialization. Five photocatalytic cycle experiments were carried out on CNUC 3 to evaluate its light stability and reusability. Figure 9a was a photocatalytic cycle experiment diagram. It could be seen from the figure that after five cycles of the catalyst, the degradation efficiency of BPA was 75%. During the reaction process, a small amount of CNUC 3 powder would be removed due to the continuous stirring process, and a small amount of catalyst would also be removed during the washing and drying process of the catalyst. It could be seen from the figure that after the photocatalytic reaction was cycled three times, the catalyst activity tends to be stable. This phenomenon showed that CNUC 3 had good stability and reusability. In addition, in order to explore whether the crystal form and chemical species of the catalyst changed after five cycles, the CNUC 3 was characterized by XRD and FT-IR. Figure 9b was the XRD pattern after five cycles. From the figure, it could be seen that after five cycles of the catalyst, all crystal faces hardly changed. Figure 9c was the FT-IR diagram after five cycles. The results show that the characteristic peaks of the FT-IR spectrum before and after the reaction hardly changed. From the above results, CNUC 3 had good chemical stability and can effectively inhibit photo-corrosion performance.
after five cycles. From the figure, it could be seen that after five cycles of the catalyst, all crystal faces hardly changed. Figure 9c was the FT-IR diagram after five cycles. The results show that the characteristic peaks of the FT-IR spectrum before and after the reaction hardly changed. From the above results, CNUC3 had good chemical stability and can effectively inhibit photo-corrosion performance.

Conclusions
In summary, we successfully fabricated O-doped porous g-C 3 N 4 nanosheet photocatalysts via thermal copolymerization with urea and ammonium oxalate or glycine as raw materials. A large number of pores were formed by gas generated from the copolymerization decomposition of ammonium oxalate, resulting in an enlarged specific surface area. At the same time, O doping adjusted the g-C 3 N 4 bandgap engineering to narrow the bandgap and expand the visible light response.
(1) Under visible light irradiation, the optimum photocatalytic efficiency of CNUC 3 and CNUG 3 nanosheets was enhanced by nearly seven and five times, respectively, more than that of g-C 3 N 4 for BPA degradation in 60 min. CNUC 3 was found to present high stability and excellent recycling in the photocatalytic degradation processes.  Figure S1: Mapping images of CNUC3; Figure S2: Mapping images of CNUG3; Figure S3: N2 adsorption-desorption isotherms of g-C3N4 CNUC3 and CNUG3; Table S1: Textural properties of the prepared samples; Figure S4: Bandgap value, estimated using a related curve of (αhν) 1/2 versus photon energy.