Synthesis of Porous Carbon Nitride Nanobelts for Efficient Photocatalytic Reduction of CO2

Sustainable conversion of CO2 to fuels using solar energy is highly attractive for fuel production. This work focuses on the synthesis of porous graphitic carbon nitride nanobelt catalyst (PN-g-C3N4) and its capability of photocatalytic CO2 reduction. The surface area increased from 6.5 m2·g−1 (graphitic carbon nitride, g-C3N4) to 32.94 m2·g−1 (PN-g-C3N4). C≡N groups and vacant N2C were introduced on the surface. PN-g-C3N4 possessed higher absorbability of visible light and excellent photocatalytic activity, which was 5.7 and 6.3 times of g-C3N4 under visible light and simulated sunlight illumination, respectively. The enhanced photocatalytic activity may be owing to the porous nanobelt structure, enhanced absorbability of visible light, and surface vacant N-sites. It is expected that PN-g-C3N4 would be a promising candidate for CO2 photocatalytic conversion.


Introduction
Significantly increased CO 2 concentration in the atmosphere has caused problems such as air pollution and global warming in the past decades, posing a serious threat to our future generations [1,2]. In order to alleviate these issues, innovative and sustainable technologies are needed to effectively capture and convert CO 2 . Sustainable conversion of CO 2 to high value-added products not only helps to reduce the content of CO 2 in the atmosphere but also promotes the carbon cycle [3,4]. However, due to the lack of efficient, stable, and selective catalysts, the research on CO 2 photoreduction is still progressing slowly.
In bulk g-C 3 N 4 , the stacked 2D single layers are held together in place by weak van der Waals forces of attraction [21]. Aiming to provide more reaction sites, exfoliation has been of particular interest for modifying g-C 3 N 4 in recent years. g-C 3 N 4 is treated with thermal exfoliation [20], concentrated acid treatment [22], and ultrasonic exfoliation [23] to synthesize nanoribbon or nanobelt samples. In this study, we used melamine as the precursor and a mixed solution of distilled water and ethylene glycol as the solvent. Melamine molecularly dissolved in the mixed aqueous solution at elevated temperature and polymerized into porous carbon nitride nanobelt (PN-g-C 3 N 4 ), as displayed in Figure 1. The thin layer may be formed due to the polyols introduced into the interlayer [24] in the hydrothermal process and ultrasonic-induced exfoliation [25] in the washing stage of supramolecular precursor.
Melamine molecularly dissolved in the mixed aqueous solution at elevated temperatur and polymerized into porous carbon nitride nanobelt (PN-g-C3N4), as displayed in Figur 1. The thin layer may be formed due to the polyols introduced into the interlayer [24] i the hydrothermal process and ultrasonic-induced exfoliation [25] in the washing stage o supramolecular precursor. Meanwhile, during thermal calcination process the released gas and volume shrinkage of precursor would create many pores on the layers, finall producing porous few-layer C3N4. The morphologies, microstructures, and physicochem ical properties of the photocatalyst were studied. CO2 was used as the raw material t evaluate PN-g-C3N4 photocatalytic performance under visible light and simulated sun light. The obtained PN-g-C3N4 was confirmed to be an efficient photocatalyst in the con version of CO2.  Figure 2 presents the morphologies and micro-structures of PN-g-C3N4 and g-C3N samples. The as-obtained PN-g-C3N4 nanobelts are characterized with thickness of 30-8 nm with a lateral size of micrometers. It can be seen that PN-g-C3N4 has loose nanobel structures with pores in its framework (Figure 2a,b). The existence of a large number o edges and pores in the obtained porous nanobelt structures is extremely important fo improving the photochemical and catalytic performance of carbon nitride. On the con trary, g-C3N4 has bulk structure (Figure 2c,d), which is formed by lamellar structure stacking with each other.  Figure 2 presents the morphologies and micro-structures of PN-g-C 3 N 4 and g-C 3 N 4 samples. The as-obtained PN-g-C 3 N 4 nanobelts are characterized with thickness of 30-80 nm with a lateral size of micrometers. It can be seen that PN-g-C 3 N 4 has loose nanobelt structures with pores in its framework (Figure 2a,b). The existence of a large number of edges and pores in the obtained porous nanobelt structures is extremely important for improving the photochemical and catalytic performance of carbon nitride. On the contrary, g-C 3 N 4 has bulk structure (Figure 2c,d), which is formed by lamellar structures stacking with each other.

SEM Analysis
The nitrogen adsorption-desorption isotherms of the samples are presented in Figure 3a. Both g-C 3 N 4 and PN-g-C 3 N 4 exhibited a type IV isotherm with a hysteresis loop at P/P o = 0.6-1.0. The pore sizes in PN-g-C 3 N 4 are about 3-20 nm, which is attributed to the pores formed in the porous nanobelt structures (Figure 3b). The BET surface areas of g-C 3 N 4 and PN-g-C 3 N 4 were calculated to be 6.5 m 2 ·g −1 , and 32.94 m 2 ·g −1 , respectively. The specific surface area of PN-g-C 3 N 4 increases greatly due to its porous nanobelt structures, which is beneficial for the exposure of more active catalytic sites. The nitrogen adsorption-desorption isotherms of the samples are presented in Figur 3a. Both g-C3N4 and PN-g-C3N4 exhibited a type IV isotherm with a hysteresis loop at P/P = 0.6-1.0. The pore sizes in PN-g-C3N4 are about 3-20 nm, which is attributed to the pore formed in the porous nanobelt structures (Figure 3b). The BET surface areas of g-C3N4 and PN-g-C3N4 were calculated to be 6.5 m 2 ·g −1 , and 32.94 m 2 ·g −1 , respectively. The specifi surface area of PN-g-C3N4 increases greatly due to its porous nanobelt structures, whic is beneficial for the exposure of more active catalytic sites.

IR and UV-Vis DRS Analysis
The FTIR spectrum for g-C3N4 and PN-g-C3N4 ( Figure 4) showed a peak at 807 cm − typical for the out-of-plane bending mode of heptazine rings, whilst peaks locked betwee 800 and 1800 cm −1 originated from N-C=N heterorings [26]. The peak at 3000-3500 cm − corresponded to N-H stretching vibrations. For the PN-g-C3N4 samples, a new peak cen  The nitrogen adsorption-desorption isotherms of the samples are presented in Figure  3a. Both g-C3N4 and PN-g-C3N4 exhibited a type IV isotherm with a hysteresis loop at P/Po = 0.6-1.0. The pore sizes in PN-g-C3N4 are about 3-20 nm, which is attributed to the pores formed in the porous nanobelt structures (Figure 3b). The BET surface areas of g-C3N4 and PN-g-C3N4 were calculated to be 6.5 m 2 ·g −1 , and 32.94 m 2 ·g −1 , respectively. The specific surface area of PN-g-C3N4 increases greatly due to its porous nanobelt structures, which is beneficial for the exposure of more active catalytic sites.

IR and UV-Vis DRS Analysis
The FTIR spectrum for g-C3N4 and PN-g-C3N4 ( Figure 4) showed a peak at 807 cm −1 typical for the out-of-plane bending mode of heptazine rings, whilst peaks locked between 800 and 1800 cm −1 originated from N-C=N heterorings [26]. The peak at 3000-3500 cm −1 corresponded to N-H stretching vibrations. For the PN-g-C3N4 samples, a new peak centered at 2173 cm −1 is found in the spectrum, which is assigned to an asymmetric stretching

IR and UV-Vis DRS Analysis
The FTIR spectrum for g-C 3 N 4 and PN-g-C 3 N 4 ( Figure 4) showed a peak at 807 cm −1 typical for the out-of-plane bending mode of heptazine rings, whilst peaks locked between 800 and 1800 cm −1 originated from N-C=N heterorings [26]. The peak at 3000-3500 cm −1 corresponded to N-H stretching vibrations. For the PN-g-C 3 N 4 samples, a new peak centered at 2173 cm −1 is found in the spectrum, which is assigned to an asymmetric stretching vibration of C≡N triple bond. The other change was the decrease in the intensity of the N-H stretching peaks between 3000 and 3300 cm −1 . The results suggest the synthesis of PN-g-C 3 N 4 decreases the concentration of N-H groups and introduces C≡N groups. The existence of C≡N groups in PN-g-C 3 N 4 is supposed to increase the electron delocalization and adjust band structures, beneficial for visible-light absorption and photon-generated carrier separation [27].
vibration of C≡N triple bond. The other change was the decrease in the intensity o H stretching peaks between 3000 and 3300 cm −1 . The results suggest the synthesi g-C3N4 decreases the concentration of N-H groups and introduces C≡N groups. T ence of C≡N groups in PN-g-C3N4 is supposed to increase the electron delocaliza adjust band structures, beneficial for visible-light absorption and photon-genera rier separation [27]. The optical absorption properties of the photocatalyst have a great effect on tocatalytic performance. In order to investigate the optical absorption propertie samples, diffuse reflectance absorption spectra were recorded on UV-Vis system. tical absorption spectra of g-C3N4 and PN-g-C3N4 are displayed in Figure 5a. The tion edge of g-C3N4 was at around 460 nm. However, the absorption spectrum o C3N4 extends to the more visible light region from 420 nm to 800 nm. The results strate that PN-g-C3N4 has enhanced optical adsorption of the visible light, which to large number of edges and pores in the obtained porous nanobelt structures. T gaps of g-C3N4 and PN-g-C3N4 are presented in Figure 5b. The band gap of PN (2.50 eV) is lower than that of g-C3N4 (2.68 eV). These changes are related to q confinement effect, due to excitation into the lower energy defect states [28]. The re low band gap of PN-g-C3N4 allows it to absorb a good number of photons in th domain of the solar spectrum, which is the most important for an effective photo

XRD and XPS Analysis
The XRD pattern for pristine g-C3N4 ( Figure 6a) showed two characteristic 13.0° and 27.4°, which can be assigned to the (100) and (002) crystal planes of g-C resenting in-plane packing and interfacial stacking of g-C3N4 sheets, respectively [ The optical absorption properties of the photocatalyst have a great effect on the photocatalytic performance. In order to investigate the optical absorption properties of the samples, diffuse reflectance absorption spectra were recorded on UV-Vis system. The optical absorption spectra of g-C 3 N 4 and PN-g-C 3 N 4 are displayed in Figure 5a. The absorption edge of g-C 3 N 4 was at around 460 nm. However, the absorption spectrum of PN-g-C 3 N 4 extends to the more visible light region from 420 nm to 800 nm. The results demonstrate that PN-g-C 3 N 4 has enhanced optical adsorption of the visible light, which ascribe to large number of edges and pores in the obtained porous nanobelt structures. The band gaps of g-C 3 N 4 and PN-g-C 3 N 4 are presented in Figure 5b. The band gap of PN-g-C 3 N 4 (2.50 eV) is lower than that of g-C 3 N 4 (2.68 eV). These changes are related to quantum confinement effect, due to excitation into the lower energy defect states [28]. The relatively low band gap of PN-g-C 3 N 4 allows it to absorb a good number of photons in the visible domain of the solar spectrum, which is the most important for an effective photocatalyst.
vibration of C≡N triple bond. The other change was the decrease in the intensity of the N H stretching peaks between 3000 and 3300 cm −1 . The results suggest the synthesis of PN g-C3N4 decreases the concentration of N-H groups and introduces C≡N groups. The exist ence of C≡N groups in PN-g-C3N4 is supposed to increase the electron delocalization and adjust band structures, beneficial for visible-light absorption and photon-generated car rier separation [27]. The optical absorption properties of the photocatalyst have a great effect on the pho tocatalytic performance. In order to investigate the optical absorption properties of th samples, diffuse reflectance absorption spectra were recorded on UV-Vis system. The op tical absorption spectra of g-C3N4 and PN-g-C3N4 are displayed in Figure 5a. The absorp tion edge of g-C3N4 was at around 460 nm. However, the absorption spectrum of PN-g C3N4 extends to the more visible light region from 420 nm to 800 nm. The results demon strate that PN-g-C3N4 has enhanced optical adsorption of the visible light, which ascrib to large number of edges and pores in the obtained porous nanobelt structures. The band gaps of g-C3N4 and PN-g-C3N4 are presented in Figure 5b. The band gap of PN-g-C3N (2.50 eV) is lower than that of g-C3N4 (2.68 eV). These changes are related to quantum confinement effect, due to excitation into the lower energy defect states [28]. The relativel low band gap of PN-g-C3N4 allows it to absorb a good number of photons in the visibl domain of the solar spectrum, which is the most important for an effective photocatalyst

XRD and XPS Analysis
The XRD pattern for pristine g-C3N4 (Figure 6a) showed two characteristic peaks a 13.0° and 27.4°, which can be assigned to the (100) and (002) crystal planes of g-C3N4, rep resenting in-plane packing and interfacial stacking of g-C3N4 sheets, respectively [29]. Th peak at 27.4° of PN-g-C3N4 is weaker and wider, suggesting that the interlayer structur

XRD and XPS Analysis
The XRD pattern for pristine g-C 3 N 4 (Figure 6a) showed two characteristic peaks at 13.0 • and 27.4 • , which can be assigned to the (100) and (002) crystal planes of g-C 3 N 4 , representing in-plane packing and interfacial stacking of g-C 3 N 4 sheets, respectively [29]. The peak at 27.4 • of PN-g-C 3 N 4 is weaker and wider, suggesting that the interlayer structure of g-C 3 N 4 has been weakened, which agrees well with the changes in the micro-morphology.
Molecules 2022, 27, x FOR PEER REVIEW 5 of 1 of g-C3N4 has been weakened, which agrees well with the changes in the micro-morphol ogy. The survey XPS spectra of g-C3N4 and PN-g-C3N4 samples are shown in Figure 6b The XPS data showed a decrease in the N/C ratio from 1.29 (g-C3N4) to 1.20 (PN-g-C3N4 on the surface, suggesting the introduction of surface N defects. High-resolution XPS peaks of C1s spectra of the g-C3N4 sample in Figure 6c are deconvoluted into three peak for C-C (285.0 eV), C-N (286.5 eV), and N-C=N (288.5 eV) bonds. Moreover, the high-res olution XPS peaks of N1s spectrum (Figure 6d) is deconvoluted into four peaks. The firs peak at 398.9 eV represents the C=N-C bond. The other peaks at 400.3, 401.5, and 404.8 eV belongs to N-(C)3, C-N-H, and π excitation bonds, respectively (Figure 6d). Compared to g-C3N4, PN-g-C3N4 showed a slight shift in all the peaks of C1s (0.1-0.2 eV) and N1s (0.1-0.3 eV) spectra, which may be caused by the defects in the carbon nitride network. Inter estingly, the intensity of C-C peaks of PN-g-C3N4 slightly increased. The peak area ratio between C-C and N-C-N of C1s spectra were calculated to be 0.18 and 0.25 for g-C3N4 and PN-g-C3N4 samples, respectively. Similarly, the peak area ratios between C=N-C and N (C)3 peaks in the N1s spectra were determined to be 4.5 (g-C3N4) and 3.6 (PN-g-C3N4) respectively. It is strong evidence that C=N-C vacancies are formed on the surface of PN g-C3N4, which can act as entrapping points for charges, yielding longer lifetimes for the charge carrier photoexcitons [30].

Photocatalytic Performance
The photocatalytic activities of as-prepared g-C3N4 and PN-g-C3N4 are shown in Fig  ure 7. In the range of 420-800 nm, simulated sunlight has the similar profile with visible light. However, simulated sunlight has energy distribution at UV zone (360-420 nm) and NIR zone (820 nm) while these parts of visible light are cut off. Remarkably, the sample PN-g-C3N4 exhibits an excellent CO evolution rate (29.8 μmol·h −1 ·g −1 ), which is about 5.7 times that of g-C3N4 (5.2 μmol·h −1 ·g −1 ) under the visible light (Figure 7a). The CO evolution rate under simulated sunlight catalyzed by PN-g-C3N4 is 52.6 μmol·h −1 ·g −1 , which is abou The survey XPS spectra of g-C 3 N 4 and PN-g-C 3 N 4 samples are shown in Figure 6b. The XPS data showed a decrease in the N/C ratio from 1.29 (g-C 3 N 4 ) to 1.20 (PN-g-C 3 N 4 ) on the surface, suggesting the introduction of surface N defects. High-resolution XPS peaks of C1s spectra of the g-C 3 N 4 sample in Figure 6c are deconvoluted into three peaks for C-C (285.0 eV), C-N (286.5 eV), and N-C=N (288.5 eV) bonds. Moreover, the high-resolution XPS peaks of N1s spectrum (Figure 6d) is deconvoluted into four peaks. The first peak at 398.9 eV represents the C=N-C bond. The other peaks at 400.3, 401.5, and 404.8 eV belongs to N-(C) 3 , C-N-H, and π excitation bonds, respectively (Figure 6d). Compared to g-C 3 N 4 , PN-g-C 3 N 4 showed a slight shift in all the peaks of C1s (0.1-0.2 eV) and N1s (0.1-0.3 eV) spectra, which may be caused by the defects in the carbon nitride network. Interestingly, the intensity of C-C peaks of PN-g-C 3 N 4 slightly increased. The peak area ratios between C-C and N-C-N of C1s spectra were calculated to be 0.18 and 0.25 for g-C 3 N 4 and PN-g-C 3 N 4 samples, respectively. Similarly, the peak area ratios between C=N-C and N-(C) 3 peaks in the N1s spectra were determined to be 4.5 (g-C 3 N 4 ) and 3.6 (PN-g-C 3 N 4 ), respectively. It is strong evidence that C=N-C vacancies are formed on the surface of PN-g-C 3 N 4 , which can act as entrapping points for charges, yielding longer lifetimes for the charge carrier photoexcitons [30].

Photocatalytic Performance
The photocatalytic activities of as-prepared g-C 3 N 4 and PN-g-C 3 N 4 are shown in Figure 7. In the range of 420-800 nm, simulated sunlight has the similar profile with visible light. However, simulated sunlight has energy distribution at UV zone (360-420 nm) and NIR zone (820 nm) while these parts of visible light are cut off. Remarkably, the sample PN-g-C3N4 exhibits an excellent CO evolution rate (29.8 µmol·h −1 ·g −1 ), which is about 5.7 times that of g-C 3 N 4 (5.2 µmol·h −1 ·g −1 ) under the visible light (Figure 7a). The CO evolution rate under simulated sunlight catalyzed by PN-g-C 3 N 4 is 52.6 µmol·h −1 ·g −1 , which is about 6.3 times that of g-C 3 N 4 (8.3 µmol·h −1 ·g −1 ). The results confirmed porous nanobelt structures of PN-g-C 3 N 4 can extremely enhance the specific surface area and provide more space for mass transfer and reaction, which in turn improves the photocatalytic activity of the samples. The superior activities of PN-g-C 3 N 4 can also be attributed to enhanced visible-light absorption and N defects.
6.3 times that of g-C3N4 (8.3 μmol·h −1 ·g −1 ). The results confirmed porous nanobelt struc tures of PN-g-C3N4 can extremely enhance the specific surface area and provide more space for mass transfer and reaction, which in turn improves the photocatalytic activity o the samples. The superior activities of PN-g-C3N4 can also be attributed to enhanced visi ble-light absorption and N defects.
As presented in Figure 7b, the yield of CO is stable without any significant deactiva tion after five cycles (15 h illumination), which indicates high photostability of PN-g-C3N for the CO2 reduction. It is worth noting that no other gas products such as CH3OH or CH generated by PN-g-C3N4 were detected by gas chromatography. Electrochemical tests were performed in a three-electrode cell with a g-C3N4-coated working electrode to further understand the dynamics of electron transfer at the PN-g C3N4 surface. Figure 8a shows the current of the electrochemical cell with pulsed ligh excitation. Under visible light illumination, both g-C3N4 and PN-g-C3N4 generated signif icant photocurrent, implying efficient photogeneration of charge carriers in both material that is then transferred to the working electrode. Furthermore, the PN-g-C3N4 showed higher photocurrent intensity than that of g-C3N4, suggesting the higher separation rate of photogenerated charge carriers in the PN-g-C3N4. Additionally, the photocurrent can reproducibly increase and recover in every on-off cycle of irradiation, demonstrating the high stability in practical applications. The photogenerated electrons and holes are likely separated more efficiently in PN-g-C3N4 than in g-C3N4. To test this, photoluminescence (PL) measurements were performed to study the separation of photogenerated electron and holes in g-C3N4 and PN-g-C3N4. Figure 8b displays the PL spectra of the two sample under 380 nm excitation at room temperature. The strong emission peak of g-C3N4 around 465 nm was derived from the direct band transition. By contrast, the PL intensity of PN g-C3N4 was 65% lower, indicating the higher efficiency in separation of the photogener ated charge carriers. Furthermore, the morphology change from multi-layer structure (g C3N4) to thin nanobelts (PN-g-C3N4) would shorten the distance for the photogenerated electrons to reach the surface, thus facilitating the charge separation. As presented in Figure 7b, the yield of CO is stable without any significant deactivation after five cycles (15 h illumination), which indicates high photostability of PN-g-C 3 N 4 for the CO 2 reduction. It is worth noting that no other gas products such as CH 3 OH or CH 4 generated by PN-g-C 3 N 4 were detected by gas chromatography.
Electrochemical tests were performed in a three-electrode cell with a g-C 3 N 4 -coated working electrode to further understand the dynamics of electron transfer at the PNg-C 3 N 4 surface. Figure 8a shows the current of the electrochemical cell with pulsed light excitation. Under visible light illumination, both g-C 3 N 4 and PN-g-C 3 N 4 generated significant photocurrent, implying efficient photogeneration of charge carriers in both materials that is then transferred to the working electrode. Furthermore, the PN-g-C 3 N 4 showed higher photocurrent intensity than that of g-C 3 N 4 , suggesting the higher separation rate of photogenerated charge carriers in the PN-g-C 3 N 4 . Additionally, the photocurrent can reproducibly increase and recover in every on-off cycle of irradiation, demonstrating the high stability in practical applications. The photogenerated electrons and holes are likely separated more efficiently in PN-g-C 3 N 4 than in g-C 3 N 4 . To test this, photoluminescence (PL) measurements were performed to study the separation of photogenerated electrons and holes in g-C 3 N 4 and PN-g-C 3 N 4 . Figure 8b displays the PL spectra of the two samples under 380 nm excitation at room temperature. The strong emission peak of g-C 3 N 4 around 465 nm was derived from the direct band transition. By contrast, the PL intensity of PN-g-C 3 N 4 was 65% lower, indicating the higher efficiency in separation of the photogenerated charge carriers. Furthermore, the morphology change from multi-layer structure (g-C 3 N 4 ) to thin nanobelts (PN-g-C 3 N 4 ) would shorten the distance for the photogenerated electrons to reach the surface, thus facilitating the charge separation.
The VB XPS spectra (Figure 8c) shows that the band gap of g-C 3 N 4 and PN-g-C 3 N 4 between the valence band (VB) and Fermi level (E f ) are 2.38 and 2.25 eV [31], respectively. The Mott-Schottky plot (Figure 8d) of g-C 3 N 4 and PN-g-C 3 N 4 illustrates that the flat band potentials are −0.80 and −0.69 V, versus the saturated calomel electrode (SCE). The Fermi levels of g-C 3 N 4 and PN-g-C 3 N 4 are −0.58 and −0.47 V (vs. NHE) [32]. Therefore, the CB and VB potentials of g-C 3 N 4 can be calculated to −0.88 and 1.80 eV, respectively, while the CB and VB potentials of PN-g-C 3 N 4 were equal to −0.72 and 1.78 eV, respectively [33]. The potential position change between g-C 3 N 4 and PN-g-C 3 N 4 is shown in Figure 9, and the band gap structures and charge migration of g-C 3 N 4 and PN-g-C 3 N 4 are illustrated. The VB XPS spectra (Figure 8c) shows that the band gap of g-C3N4 and PN-g-C3N between the valence band (VB) and Fermi level (Ef) are 2.38 and 2.25 eV [31], respectively The Mott-Schottky plot (Figure 8d) of g-C3N4 and PN-g-C3N4 illustrates that the flat band potentials are −0.80 and −0.69 V, versus the saturated calomel electrode (SCE). The Ferm levels of g-C3N4 and PN-g-C3N4 are −0.58 and −0.47 V (vs. NHE) [32]. Therefore, the CB and VB potentials of g-C3N4 can be calculated to −0.88 and 1.80 eV, respectively, while the CB and VB potentials of PN-g-C3N4 were equal to −0.72 and 1.78 eV, respectively [33]. The potential position change between g-C3N4 and PN-g-C3N4 is shown in Figure 9, and the band gap structures and charge migration of g-C3N4 and PN-g-C3N4 are illustrated.    [32]. Therefore, the CB and VB potentials of g-C3N4 can be calculated to −0.88 and 1.80 eV, respectively, while the CB and VB potentials of PN-g-C3N4 were equal to −0.72 and 1.78 eV, respectively [33]. The potential position change between g-C3N4 and PN-g-C3N4 is shown in Figure 9, and the band gap structures and charge migration of g-C3N4 and PN-g-C3N4 are illustrated.

Possible Mechanism
The possible reaction mechanism is discussed for photocatalytic CO2 reduction with water into CO over PN-g-C 3 N 4 as depicted in Figure 10. Generally, the photocatalytic CO 2 reduction reaction involves the following three steps: (i) CO 2 adsorption and activation; (ii) photo-produced charge carriers' excitation and transfer to the catalyst surface; and (iii) photocatalytic reaction [34]. Upon illumination with light, the photocatalyst generated electrons (e -) in the CB and holes (h + ) in the VB, as shown in Equation (1). Further, the eare exploited to reduce CO 2 to its radical (CO 2 •− ), as shown in Equation (2) [7]. The water (H 2 O) oxidation arises at VB of the catalyst to produce the energetic protons (H + ) and oxygen, as shown in Equation (3). The CO 2 •− , H + , and efurther boosted the rate CO generation, as shown in Equation (4). In the present investigation, CO was developed, which involves an 2e -/2H + reduction process [35], as shown in Equation (5).
In semiconductors, the numerous eand H + transfer by proton-coupled electron transfer mechanism is feasible for multi ereduction reaction.
PN-g-C 3 N 4 + hυ → PN-g-C 3 N 4 * + h + + e − (1) photocatalytic reaction [34]. Upon illumination with light, the photocatalyst generate electrons (e -) in the CB and holes (h + ) in the VB, as shown in Equation (1). Further, the e are exploited to reduce CO2 to its radical (CO2 •− ), as shown in Equation (2) [7]. The wate (H2O) oxidation arises at VB of the catalyst to produce the energetic protons (H + ) and ox ygen, as shown in Equation (3). The CO2 •− , H + , and efurther boosted the rate CO genera tion, as shown in Equation (4). In the present investigation, CO was developed, whic involves an 2e -/2H + reduction process [35], as shown in Equation (5). In semiconductors the numerous eand H + transfer by proton-coupled electron transfer mechanism is feas ble for multi ereduction reaction.  Figure 10. A possible photocatalytic CO2 reduction mechanism for CO production.

Materials
The chemical reagents used for the synthesis of PN-g-C3N4 were commercially avai able reagents. Melamine, hydrochloric acid, and ethylene glycol were purchased from Sigma-Aldrich. All the chemicals were used as received without further purification.

Synthesis of Catalysts
The PN-g-C3N4 was synthesized using a simple one-pot hydrothermal method. Firs in a typical synthesis procedure, melamine (2 g, 99%) was dissolved in the mixture o distilled water (40 mL) and ethylene glycol (20 mL, 99%) to make a clear solution at 60 °C Then, 2.4 mL concentrated hydrochloric acid (36.5%) was added into 60 mL of this solu tion by stirring for 10 min. Then, the mixed solution was transferred into a Teflon-line autoclave and heated at 150 °C for 12 h. The mixture was filtered to remove the solven and the precipitate was washed several times with ethanol and deionized water unde ultrasonication, followed by drying overnight at 60 °C in vacuum oven. The resultin solid was heated at 600 °C for 2 h with a heating rate of 3 °C·min −1 . The g-C3N4 was syn thesized by directly heating melamine at 500 °C for 2 h with a heating rate of 3 °C·min −1 . Figure 10. A possible photocatalytic CO 2 reduction mechanism for CO production.

Materials
The chemical reagents used for the synthesis of PN-g-C 3 N 4 were commercially available reagents. Melamine, hydrochloric acid, and ethylene glycol were purchased from Sigma-Aldrich. All the chemicals were used as received without further purification.

Synthesis of Catalysts
The PN-g-C 3 N 4 was synthesized using a simple one-pot hydrothermal method. First, in a typical synthesis procedure, melamine (2 g, 99%) was dissolved in the mixture of distilled water (40 mL) and ethylene glycol (20 mL, 99%) to make a clear solution at 60 • C. Then, 2.4 mL concentrated hydrochloric acid (36.5%) was added into 60 mL of this solution by stirring for 10 min. Then, the mixed solution was transferred into a Teflon-lined autoclave and heated at 150 • C for 12 h. The mixture was filtered to remove the solvent and the precipitate was washed several times with ethanol and deionized water under ultrasonication, followed by drying overnight at 60 • C in vacuum oven. The resulting solid was heated at 600 • C for 2 h with a heating rate of 3 • C·min −1 . The g-C 3 N 4 was synthesized by directly heating melamine at 500 • C for 2 h with a heating rate of 3 • C·min −1 .

Characterization
XRD spectra were recorded on a Bruker D8 Advance diffractometer (Cu Ka radiation). The IR spectra were collected with a Thermo Nicolet iS50 FTIR spectrometer, equipped with an attenuated total reflection (ATR) setup. Diffuse reflectance absorption spectra were recorded on a Varian Cary 4E UV-Vis system equipped with a Labsphere diffuse reflectance accessory. X-ray Photoelectron Spectroscopy (XPS) experiments were performed on Thermo ESCALAB 250 using monochromatized Al Kα at hυ = 1486.6 eV. Bandgap energy (Eg) of the g-C 3 N 4 and PN-g-C 3 N 4 samples was calculated according to the formula below: (αhv) 1/n = C(hυ − Eg) where α, υ, and C are the absorption coefficient, light frequency, and a constant, respectively. The parameter n is a pure number corresponding to different electronic transitions (n = 2 or 1/2 for indirect-allowed or direct-allowed transitions, respectively.

Photoactivity Meaasurements
The photocatalytic CO 2 reduction test was performed using a batch process under visible light with a 300 W Xenon lamp. In addition, a 420 nm cutoff filter was used to prevent the UV light and Am1.5 filter was used to simulate solar spectral. In this experiment, as-prepared photocatalyst (10 mg) was ultrasonically dispersed in 10 mL of deionized water using a 50 mL round-bottom quartz photo-reactor. Then, the reactor was tightly closed with a silicone rubber septum and the solution was saturated with CO 2 gas for 30 min before the light illumination. After illumination, the gaseous product such as CO was analyzed by gas chromatography.

Conclusions
In summary, porous g-C 3 N 4 nanobelts were synthesized via a facile hydrothermal method. The obtained PN-g-C 3 N 4 had abundant pores and edges, high specific surface areas, and possessed C≡N groups and vacant N on the surface, which dramatically improved the photocatalytic performance. This catalyst displayed enhanced optical absorption in the visible range. It efficiently and selectively catalyzed CO 2 reduction to CO under both visible light and simulated sunlight illumination. Enhanced visible light absorption and the existence of vacant N-sites on the surface also contributed to the photocatalytic activity of PN-g-C 3 N 4 . The successful synthesis of PN-g-C 3 N 4 opens up a new way to improve the photochemical performance of carbon nitride-based catalyst.
Author Contributions: Z.J.: investigation; writing-review and editing. Y.S.: investigation; methodology. Y.Y.: supervision; funding acquisition; writing-review and editing. All authors have read and agreed to the published version of the manuscript.