3.1. Photocatalytic activity
The photocatalytic activity follows the order of C0.02CN (93.2%) > C0.01CN (86.0%) > C0.05CN (74.6%) > C0.08CN (72.5%) after 150 min of visible light irradiation (Fig. 1a), which indicated the significance of the addition of ammonium citrate. Moreover, the photocatalytic effect of C-doped g-C3N4 gradually ameliorated with the addition of ammonium citrate, and the best photocatalytic effect of the prepared C0.02CN was achieved. By contrast, the rate constants of C0.02CN for BPA degradation were about 6.7 times than that of g-C3N4 (Fig. 1b, Table. S1). Additionally, C0.02CN was used to the following experiments on account of notable photocatalytic activity. These results showed that the photocatalytic performance of CXCN was better than the g-C3N4, and thereinto, the photocatalytic performance of C0.02CN was optimal.
Solution pH is one of the key parameters affecting the degradation rate and degradation efficiency of pollutants in the photocatalytic process. Visible photocatalytic degradation experiments were carried out in different pH of BPA solution using, on account of its notable photocatalytic activity (Fig. 2a). At low pH (3-9), the degradation efficiency of BPA in solution decreased (96.0% to 88.0%) with the increasing of the initial pH, which can be related to the surface charge property as reflected by zeta potential and the isoelectric point (pHpzc) of C0.02CN was 5.6 (Fig. 2b). At pH < 5.6, the surface of C0.02CN is positively charged, while at pH > 5.6, the surface of C0.02CN is negatively charged. BPA is a neutral molecule that gradually dissociates into negatively charged ions (BPA-/BPA2-) at pH > 7.5(Dong et al. 2010). When the BPA solution is alkaline, electrostatic repulsion occurs between the surface of the C-doped g-C3N4 material and the solute with the same charge, so that its degradation efficiency becomes low at a solution pH of 9. When the initial pH of BPA solution is higher (pH 11), the photocatalytic degradation rate is the fastest and the degradation efficiency is also as high as 95.5%, which may be because C0.02CN has the best photocatalytic performance under strong alkalinity and can produce more free radicals. Based on the results, the solution pH has a significant effect on the C-doped g-C3N4 sample, and the degradation of BPA in solution is the best under strongly alkaline (pH 11) conditions. Nonetheless, considering the practical usability, the other experiments were explored at the original pH value.
The effect of the catalyst (C0.02CN) dosage on the removal rate and photodegradation efficiency was investigated. As can be seen in Fig. 3a, the photocatalytic degradation efficiency was optimal at 93.2% when the catalyst was added at 0.05 g. The photocatalytic degradation efficiency was reduced to 68.2% when the catalyst was added at 0.02g, which was ascribed to the low dosage of C0.02CN brought about insufficient photocatalytic active sites. However, in comparison with the dosage of 0.05g, when the dosage was raised to 0.08g, the photocatalytic degradation was reduced to 80.2%. This is because the excessive catalyst affected the light transmittance and reduced the photon absorption of the catalyst, leading to poor photocatalytic efficiency. Therefore, the amount of catalyst added affects the degradation efficiency of BPA, and the best photocatalytic degradation of BPA is achieved at 0.5 g/L of catalyst. Moreover, with the aim of comparing the complete mineralization efficiency of g-C3N4 and C0.02CN as catalysts for BPA in solution, TOC concentrations in the solution before and after the reaction were measured using an N/C type TOC analyzer, and the experimental results were shown in Fig. 3b. After 150 min irradiation, much higher mineralization rate of C0.02CN (27.6%) was observed in contrast with g-C3N4 (7.87%), denoting that the higher removal effect and mineralization ability were performed for C-doped g-C3N4. This phenomenon might be due to the fact that C-doped g-C3N4 could generate more reactive radicals after being excited by light, which made the degradation of organic matter in solution more complete.
After the exploration of photocatalytic activities, similar photocatalyts have been contrasted. C-doped g-C3N4 prepared in this study still possessed an outstanding performance and high photocatalytic degradation efficiency of BPA under visible light without additional reaction promotors and any metal doping (Table.S2), which showed that CXCN has a vast application prospect in environmental photocatalysis, in view of its simple in fabrication process, low-cost at preparing, environment friendly and high-activity. For purpose of evaluating the photocatalytic stability of C-doped g-C3N4, photodegradation cycle experiments of BPA were conducted under the same conditions. After five photodegradation cycles, the removal efficiency of BPA was promising and the removal rate was 73.4%, indicating a positive reusability of C-doped g-C3N4 (C0.02CN) (Fig. S2).
3.2. Characterization
For the identification of crystal structure of the CXCN samples, XRD characterization was carried out. XRD spectra of CXCN photocatalysts (Fig. 4a) possess a similar crystal structure to g-C3N4. All materials show typical XRD diffraction peaks around 12.9° (100) and 27.8° (002), which are related to the in-plane repeating N bridges linked to the triazine unit, as well as the conjugated in-plane stacking of aromatic compounds, respectively, indicating that the materials possess a graphite-like phase structure(Zhu et al. 2019). Simultaneously, the peaks in (100) and (200) planes showed that the diffraction intensity became stronger as the addition of the ammonium citrate. However, when it came to C0.05CN and C0.08CN, the diffraction intensity presented the opposite case, which indicated that the graphite-like structure was severely affected by the addition of an excess ammonium citrate. Meanwhile, the photocatalytic performance of C-doped g-C3N4 decreased due to the destruction of the triazine unit structure (Fig. 1a). Moreover, compared to the pristine g-C3N4, the shifts of the diffraction peak in (100) and (200) planes indicated the decrease of the in-plane hole-to-holes distance and the increase of the in-space layer-to-layer spacing, respectively. Furthermore, the peak intensity of (002) plane of the C0.02CN material is greater than that of g-C3N4, and the (002) peak changes from 27.8° to 27.6°, which means that the C-doped g-C3N4 material (C0.02CN) has a higher crystallinity and the layer-to-layer distance increases by 0.002 nm.
The FT-IR spectra was recorded to confirm the information about the presence of the pristine g-C3N4 and CXCN in Fig. 4b. The series of CXCN materials prepared show the typical molecular skeleton vibrational pattern of g-C3N4, proving that all materials have a graphite-like phase structure. All the prepared materials have a broad range between 3600 and 3000 cm− 1 corresponding to the stretching of N-H bonds and O-H bonds, which is due to free amino groups and hydroxyl substances adsorbed in the materials(Ojha et al. 2018). In addition, the region at 1700 − 1200 cm− 1 is the stretching vibrational modes of C-N and C = N(Chen et al. 2019) as well as the strong peak at 810 cm− 1 represents the characteristic respiration mode of the triazine ring unit(Ojha et al. 2018). There is no new peak to be detected in the spectra of CXCN materials, indicating that the addition of ammonium citrate has less effect on the structure of g-C3N4 and that the structure of all C-doped g-C3N4 materials is a graphite-like phase of g-C3N4.
In order to explore the effect of the addition of ammonium citrate on the formation process of g-C3N4, the chemical constituent and chemical state of the elements on the surface of g-C3N4 and C0.02CN were analyzed using XPS. Both the pristine g-C3N4 and C0.02CN are primarily consisted of C and N elements (Table. S3). However, the C/N atomic ratio was raised from 1.14 (g-C3N4) to 1.6 (C0.02CN), indicating that the addition of ammonium citrate introduced additional carbon atoms in g-C3N4. As depicted in Fig. 5a, the characteristic peaks at 284.8 eV and 288.2 eV correspond to the C-C and sp2-hybridized carbon (N-C = N) within the triazine ring, respectively(Hafeez et al. 2020). In comparison with the g-C3N4, the positions of the peaks (284.8 eV and 288.2 eV) of C0.02CN were not shifted, indicating that carbon doping had little effect on the triazine ring structure, which is in conformity with the FTIR analysis. Meanwhile, the C-NHX characteristic peak at the edge of the heptazine ring(Shu et al. 2020) exhibited a small shift from 286.1 eV (g-C3N4) to 286.3 eV (C0.02CN), and its relative content was reduced from 6.1% (g-C3N4) to 2.0% (C0.02CN), attributed to the higher crystallinity of C0.02CN, which is in accordance with the XRD analysis. Additionally, Fig. 5b displays the high-resolution XPS spectra of O 1s, where the peaks of g-C3N4 at 532.3 eV and 533.6 eV are consistent with C-OH and C-O caused by adsorbed H2O and CO2, respectively (Geng et al. 2020). Interestingly, in contrast to g-C3N4, a new peak appeared near 531.8 eV in the O 1s spectrum of C0.02CN, which is thought to be formed by the C-O-C bond. This phenomenon is probably because the C-OH on g-C3N4 reacted with another C-OH on exotic carbon source, then bridged each other to form C-O-C bonds through losing one H2O molecule, which caused the carbon bonded to the g-C3N4 structure. Figure 5c shows the spectra of N 1s can be divided into three peaks at 398.8 eV, 400.3 eV, and 404.4 eV, which are ascribed to N in the triazine ring via sp2 hybridization (C-N = C), N-(C)3 and charging effects, respectively(Lin et al. 2021, 2022, Ma et al. 2016, Zhou et al. 2016). In comparison with the pristine g-C3N4, the peak of N 1s in C0.02CN migrated to lower binding energies (398.6 eV, 400.1 eV), which is related to the doping of exotic carbon elements in the g-C3N4 material. These results indicated that the addition of ammonium citrate introduced additional carbon atoms, which is most probably caused by the replacement of the N atoms by C.
The surface morphological of photocatalysts is important for catalytic reaction. Hence, surface morphological characteristics of g-C3N4 and C0.02CN were analyzed using scanning electron microscopy (SEM), and the results are shown in Fig. S3. As can be seen, both g-C3N4 and C0.02CN have a layered structure with stacked nanosheets, indicating a small effect of the addition of ammonium citrate on the material structure, which is conformant with the results obtained from the FTIR spectroscopy analysis. The BET specific surface areas were 75.021 m2/g for g-C3N4 and 82.814 m2/g for C0.02CN (Fig. S4). Correspondingly, C0.02CN possessed the large pore volumes (0.886 cm3/g) than the pristine g-C3N4 (0.620 cm3/g). These phenomena showed that C0.02CN owned more active sites, in comparison with the pristine g-C3N4.
3.3. Optical properties
UV-vis diffuse reflectance spectroscopy (UV-vis DRS) was used to examine the optical absorption properties of CXCN samples and pristine g-C3N4. As shown in Fig. 6a, it can be seen that there is a red-shift in the optical absorption edge compared with that of g-C3N4. Interestingly, the photoresponse of the C-doped g-C3N4 material was extended to cover the whole visible region ascribed to the carbon doping, and the optical absorption of the obtained C0.02CN samples were enhanced due to the addition of ammonium citrate. It is widely acknowledged that with the extension of adsorption wavelength from UV-light to visible region, the intrinsic band gap of the materials tends to decrease. To thoroughly confirm this inference, according to the UV-vis DRS data, the Tauc-plot equation was carried out to calculate the band-gap, which can be seen in Fig. 6b. Unsurprisingly, the value of the forbidden band width of the pristine g-C3N4 decreases from 2.67 eV to 2.29 eV (C0.02CN) as the addition of the ammonium citrate. Furthermore, the energy band structures of the samples were analyzed using VB-XPS. From Fig. 6c, the Fermi energy levels (EF) of g-C3N4 and C0.02CN are located at the valence band edge maximum of both 2.04 eV. Thereafter, the valence band potential (VB) and the conduction band potential (CB) of both photocatalysts can be obtained on the basis of the following equations(Qin et al. 2016):
EVB = ∆E - Evac + WS (1)
ECB = EVB - Eg (2)
ECB, EVB, and Eg are the conduction band potential, valence band potential, and forbidden band width on the hydrogen scale, respectively; Evac is the energy of the free electron on the hydrogen scale, which is 4.5 eV with respect to the general hydrogen electrode; where △E is the difference between the EF and valence band edge values, both of which are 2.04 eV; and WS is the work function, which is 4.0 eV for g-C3N4. Based on Eq. (1), it can be calculated that the ECB of g-C3N4 and C0.02CN are both 1.54 eV. According to Eq. (2), the ECB of g-C3N4 and C-doped g-C3N4 are − 1.13 eV and − 0.75 eV, respectively, which could be seen from Fig. 6d.
As for the obtained C-doped g-C3N4, the expected results of its bandgap and light-harvesting capacity have been come off. In order to investigate the impacts of these variation on the photogenerated electron-hole pairs, PL spectra, photocurrent responses and EIS tests were conducted. In Fig. 7a, a strong PL emission peak was found near 457 nm for the pristine g-C3N4 under excitation wavelength λ = 372 nm, derived from the photo-generated electron-hole pairs produced by photoexcitation of g-C3N4 were more readily compounded. Unexpectedly, the intensity of the PL emission peak was remarkably reduced for the C0.02CN, which suggested that the addition of carbon elements in C0.02CN inhibited the recombination of the photogenerated electron-hole pairs of the prepared carbon-doped carbon nitride materials.
The transient photocurrent response of the pristine g-C3N4 and C0.02CN could be found in Fig. 7b, in which photocurrent is generated with the visible light irradiated, while the current value decreases rapidly with the light turned off. Furthermore, the photocurrent signal value of C0.02CN was larger than that of g-C3N4, which indicated that the photogenerated electrons and holes of C-doped g-C3N4 could be separated more quickly. In addition, Fig. 7c illustrated that the measured arc radius of C0.02CN was much smaller than that of g-C3N4, proving that the photogenerated electrons of C-doped g-C3N4 were more likely to migrate to the interface for reaction. These results indicated that the introduction of foreign carbon atoms into the g-C3N4 leaded to a more substantial improvement in the separation and transfer efficiency of C0.02CN photogenerated electron-hole pairs. Additionally, the reusability of the photocatalyst is crucial for the practical application.
3.4. Photocatalytic mechanism
To ascertain the mechanism of the photocatalytic reaction, the identification of reactive species in the photocatalytic process is of significance. In this study, EDTA-2Na, BQ, and TBA scavengers were selected to remove h+, -O2−, and hydroxyl radicals (-OH), respectively(Wang et al. 2018). From Fig. 8a, it can be observed that the degradation efficiency of BPA decreased substantially after the addition of EDTA-2Na and BQ, while the addition of TBA was inapparent. These results showed that -O2− and h+ were the main active species accountable for the degradation of BPA by CXCN under visible light (the magnitude of their effect is ·O2− greater than h+), while·OH play an insignificant part in the photodegradation process. In addition, as can be seen in Fig. 8b, the characteristic signal intensities of DMPO-O2·− were stronger than g-C3N4, which indicated that C doping is a critical factor in generating ·O2− radicals in the C-doped g-C3N4 photocatalytic system.
According to the above results, a possible photocatalytic reaction mechanism was proposed as shown in Fig. 9. Owing to CXCN structure, the visible light adsorption range was broadened and the recombination of electron-hole pairs was decreased. Under visible light irradiation, CXCN can produce more photogenerated electron-hole pairs. These exciting electrons transfer to the catalyst surface and react with dissolved oxygen to generate ·O2−, and BPA was oxidized by h+ and ·O2−.