Multistage Polymerization Design for g-C3N4 Nanosheets with Enhanced Photocatalytic Activity by Modifying the Polymerization Process of Melamine.

Graphene-like g-C3N4 nanosheets (NSs) have been successfully synthesized with a modified polymerization process of melamine by cocondensation with volatile salts. Volatile ammonium salts such as urea-NH4Cl/(NH4)2SO4/(NH4)3PO4 were added with melamine to modulate the thermodynamic process during polymerization and optimize the structure formation in situ. The surface area, surface structure, and surface charge state of the obtained g-C3N4 NSs could be controlled by simply adjusting the mass ratio of the melamine/volatile ammonium salt. As a consequence, the g-C3N4 NSs exhibited much higher activity than bulk g-C3N4 for the photocatalytic degradation of target pollutants (rhodamine B, methylene blue, and methyl orange), and it also exhibited greater hydrogen evolution under visible light irradiation with an optimal melamine/volatile ammonium salt ratio. The as-prepared g-C3N4 NSs with melamine-urea-NH4Cl showed the highest visible light photocatalytic H2 production activity of 1853.8 μmol·h-1·g-1, which is 9.4 times higher than that of bulk g-C3N4 from melamine. The present study reveals that the synergistic effect of the enhanced surface area, surface structure, and surface charge state is the key for the enhancement of photocatalytic degradation and hydrogen evolution, which could be controlled by the proposed strategy. The result is a good explanation for the hypothesis that adding properly selected monomers can truly regulate the polymerization process of melamine, which is beneficial for obtaining g-C3N4 NSs without molecular self-assembly. Considering the inexpensive feedstocks used, a simple synthetic controlling method provides an opportunity for the rational design and synthesis, making it decidedly appealing for large-scale production of highly photocatalytic, visible-sensitizable, metal-free g-C3N4 photocatalysts.

Introduction

Graphite carbon nitride (g-C3N4) is one of the most promising photocatalysts and has gained much more attention because of its high chemical stability, appealing electronic band gap, and optical features as a metal-free, visible-light-driven semiconductor photocatalyst for solar energy conversion and environmental remediation.[1−5] However, bulk g-C3N4 still suffers from low surface area and high recombination rate of photogenerated electrons and holes, which result in low photocatalytic activity.[6,7] Much effort has been made to enhance the photocatalytic activity of g-C3N4, including doping with heteroatoms,[8−11] compositing with other semiconductors and controlling morphology.[12−15] Nevertheless, a simple development of highly efficient g-C3N4 still remains a significant challenge.[16,17]

Porous or nanosheet (NS) structures possess higher surface area and improved mass transferability compared to other types of photocatalysts, which is favorable for enhancing the photocatalytic efficiency of g-C3N4. Presently, different approaches have been developed to prepare porous g-C3N4, and hard-template or soft-template approaches are the two classic ways to achieve this goal.[18] Although the soft- and hard-template methods have achieved good results in controlling the shape, there are also some problems. For hard-template approaches, some hazardous and poisonous substances such as hydrogen fluoride (HF) and aqueous ammonium bifluoride (NH4HF2) have been used to remove the silica template. On the other hand, using conventional surfactants has required complicated post-treatments and the residual carbon species were hard to completely remove, which may influence the photocatalytic activity.[19]

Supramolecular chemistry provides a great opportunity for the synthesis of nanostructured materials without any further templating techniques.[20] The structure of the final product can be controlled by choosing the appropriate monomers and solvents for the synthesis. Supramolecular aggregates of melamine with cyanuric acid,[21,22] oxalic acid,[23,24] barbituric acid,[25] cyanuric acid and barbituric acid,[26,27] cyanuric acid and urea,[28] cyanuric acid and ethylene glycol,[29] or hydrogen peroxide and ammonium chloride,[30] connected by hydrogen bonds, have been adopted to prepare nanostructured g-C3N4. A diverse morphology containing an ordered structure and a relatively higher surface area with enhanced photocatalytic performance by molecular self-assembly has become a hot topic. However, the additive monomers and solvents or the supramolecular structure itself may modify the polycondensation process, and the chemical structure of g-C3N4 was rarely of concern or discussed. To simplify the system and focus on the role of the monomer itself in the polycondensation process and the chemical structure, monomers without molecular self-assembly are needed. A hypothesis made herein was that g-C3N4 with various morphologies and surface charge states toward high photocatalytic performance would be created through cocondensation of melamine with volatile salts. The volatile salts could modify the polymerization process of melamine by the bubbles that are generated at different stages of polymerization through decomposition of the additive volatile salts and without molecular self-assembly.

We hypothesized that if the decomposition of volatile salts could be in combination with the polymerization process of melamine and bubbles can be generated at different stages of polymerization, then g-C3N4 with different surface areas or different surface structures should be obtained. To achieve this goal, designing mixed volatile salts with different decomposition temperatures is a critical step. Under prepolymerization of melamine at a lower temperature, the volatile salt melts to exhibit flow dynamics and begins to partially decompose, creating bubbles that could promote sufficient mixing of the entire system and expand the prepolymerized precursor. Then, with the continued polymerization of melamine at a higher temperature, the viscosity of the system is further increased, and the volatile salts encapsulated in the prepolymerization precursor with high decomposition temperature begin to decompose to generate bubbles, which can effectively avoid agglomeration of melamine in the high-temperature polymerization process. Then, g-C3N4 with a large specific surface area and different surface structures could be obtained. The mixture of melamine–urea–inorganic ammonium salt was chosen as the target system. In addition, the volatilization of urea and ammonium ions can avoid the unnecessary metal-ion doping to the structure. The melting points and decomposition temperatures of urea and inorganic ammonium salt are different, which can effectively regulate the polymerization process of melamine. In addition, the volatile and decomposable properties of urea and inorganic ammonium salt can also serve as gas bubble templates, which is favorable for producing a higher surface area. As a consequence, the in situ method of cocondensation of melamine and volatilized ammonium salt to modulate the thermodynamic process during polymerization and optimize the structure is proposed.

Herein, we report a facile one-step method for the synergic achievement of large-quantity and high-quality g-C3N4 NSs by in situ polymerization of melamine–urea–inorganic ammonium salt. Urea and volatile inorganic ammonium salts (NH4Cl, (NH4)2SO4, and (NH4)3PO4) were used as the additive monomers that can modify the polycondensation process by faster mass transfer and serve as the bubble template. The bubbles from the decomposition of the additive volatile salts can continuously release gas such as NH3(g),[31] SO2(g),[32] HCl(g),[33] and H2O(g) during the polymerization process of melamine. The surface area, surface structure, and surface charge state of g-C3N4 can be controlled by adjusting the mass ratio of the melamine/volatile molten salt and the calculated temperature. The g-C3N4 NSs synthesized by this method possessed enhanced specific surface area and achieved improved electron transportability as well as an efficient separation rate of electrons and holes. The g-C3N4 NSs exhibit much higher activity than bulk g-C3N4 in the photocatalytic degradation of the target pollutants [rhodamine B (RhB), methylene blue (MB), and methyl orange (MO)] and hydrogen evolution under visible light irradiation.

Results and Discussion

Structure and Morphology of Photocatalysts

Powder X-ray diffraction (XRD) analysis was used to investigate the crystal structure of the obtained g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600 (shown in Figure ). From Figure , it can be seen that two obvious peaks were observed in all samples. The strongest XRD peak (002) is at approximately 27.3°, which is attributable to the interplanar stacking peak of aromatic systems. The lower-angle reflection peak at 13.0° is derived from the lattice planes parallel to the c-axis.[34] It is clear that the (002) peak intensity of g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600 clearly are weaker, compared with that of g-C3N4-M-550 and g-C3N4-M-600. In addition, the (002) peak position slightly shifts toward a higher angle, which could be attributed to the creation of the porous structure or the NS structure observed in g-C3N4 by Brunauer–Emmett–Teller (BET) and transmission electron microscopy (TEM) analysis.[18] The thermogravimetric (TG)–differential scanning calorimetry (DSC) curve of melamine and melamine–urea–NH4Cl is shown in Figure . In the DSC curve of melamine, the strongest endothermic peak, at approximately 321 °C, corresponds to the sublimation and thermal condensation of melamine.[35] For melamine–urea–NH4Cl, about four strong endothermic peaks appeared from 109 to 368 °C. Furthermore, the TG analysis has been performed to illustrate the thermal behaviors of melamine and melamine–urea–NH4Cl, as displayed in Figure . For melamine, it experiences most of the weight loss from 295 to 325 °C and shows a single weight loss step. However, the curve of melamine–urea–NH4Cl shows the most weight loss from 197 to 375 °C and presents multiple weight loss steps along the entire TG process, corresponding to the continuous decomposition of the mixed precursor. A strong endothermic peak appeared at 109 °C with nearly no weight loss from the TG and DSC curve of melamine–urea–NH4Cl, indicating melting of the mixture. The TG–DSC curve of melamine–urea, melamine–urea–(NH4)2SO4, and melamine–urea–(NH4)3PO4 is also shown in Figure S1. It is clearly shown that when the temperature ranges from 200 to 360 °C, nearly no obvious endothermic peaks appeared, which indicates that the added (NH4)2SO4, (NH4)3PO4 salt cannot regulate the polymerization process of melamine during this stage. Besides, the exothermic peaks in the DSC curve of melamine–urea–(NH4)2SO4 or melamine–urea–(NH4)3PO4 at higher temperature may indicate part of the g-C3N4 local structure self-combustion decomposition reaction (shown in Figure S1).

Figure 1

XRD patterns of bulk and mesoporous g-C3N4 materials.

Figure 2

TG–DSC thermograms for heating of melamine and the mixture.

The pore structure and N2 adsorption–desorption isotherms of the as-prepared bulk g-C3N4 and g-C3N4 NSs are shown in Figure . As shown in Figure and Table , g-C3N4-M-U-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600 exhibited an increased specific surface area (72.8, 53.5, and 47.4 m2/g, respectively) compared to bulk g-C3N4-M-550 (10.8 m2/g) and g-C3N4-M-600 (22.2 m2/g). Further, the specific surface area of g-C3N4-M-U-Cl-600 increased to 103.3 m2/g, which was 9.6, 4.7, 3.5, and 1.4 times higher than the values of bulk g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-Cl-600, and g-C3N4-M-U-600, respectively. However, the specific surface areas of g-C3N4-M-U-S-600 and g-C3N4-M-U-P-600 were lower than g-C3N4-M-U-600, which indicates that the properly selected monomer is important for the improved surface area. Compared with the other three samples, g-C3N4-M-U-Cl-600 has a larger specific surface area, which is reasonable. According to TG characterizations, in the system of melamine–urea–NH4Cl mixture, the decompositions mainly occurred in the temperature range between 200 and 360 °C. Because melamine polymerizes in this temperature range, similar to a polymerization foaming process, the gases produced by the decompositions should help to form bubbles in the system and thus increase the specific surface area. However, also according to the TG characterizations, in the other three systems, the decompositions took place mainly in the temperature range between 400 and 550 °C, when the melamine polymerization was largely finished and the melamine polymer solidified. Therefore, it is imaginable that the gases formed in the high temperature range have limited effects on the specific surface area. The surface area of g-C3N4-M-U-Cl-600 was as large as that of g-C3N4 obtained by the supramolecular aggregates method (shown in Table ).[32,33,36] As a consequence, the cocondensation of melamine and volatilized ammonium salt to modulate the thermodynamic process during polymerization and optimize the structure by the in situ method was successful. In addition, the raw materials were cheaper to obtain and easier to control during the polymerization process of melamine compared with the supramolecular aggregates method. This kind of enhancement reflects that a synergistic effect could be obtained by selecting two suitable volatilized ammonium salts that can serve as bubble-forming agents simultaneously. The synergistic effect most likely resulted from the generation of gas bubbles containing nitrogen oxides,[17] NH3, and H2O[16] continuously that can then modify the polycondensation process of melamine.

Figure 3

(a) Nitrogen adsorption–desorption isotherms and (b) corresponding PSD of bulk g-C3N4 and g-C3N4 NS materials.

Table 1

Surface Area and Pore Diameter of Bulk g-C3N4 and g-C3N4 NS Materials

sample	SBET (m2/g)	pore diametermax (nm)
g-C3N4-M-550	10.8	2.6
g-C3N4-M-600	22.2	2.7
g-C3N4-M-U-600	72.8	2.8
g-C3N4-M-Cl-600	29.8	2.9
g-C3N4-M-P-600	19.7	2.9
g-C3N4-M-U-Cl-600	103.3	2.8
g-C3N4-M-U-S-600	53.5	2.7
g-C3N4-M-U-P-600	47.4	2.9

Table 2

Comparison of the g-C3N4 Samples Synthesized from Melamine via Different Additive Monomers

raw material	additive monomer	solvent	monomer interaction	SBET (m2/g)	refs
melamine	cyanuric acid	dimethyl sulfoxide	supramolecular aggregates	77.0	(21)
melamine	cyanuric acid	ethanol	supramolecular aggregates	45.0	(22)
melamine	oxalic acid	hot deionized water	supramolecular aggregates	32.0	(24)
melamine	barbituric acid	hot deionized water	supramolecular aggregates	55.1	(25)
melamine	cyanuric acid–barbituric acid	deionized water	supramolecular aggregates	70.0	(26)
melamine	cyanuric acid–barbituric acid	ethanol	supramolecular aggregates	179.0	(27)
melamine	cyanuric acid–urea	ethanol	supramolecular aggregates	97.4	(28)
melamine	cyanuric acid–ethylene glycol	ethylene glycol	supramolecular aggregates	93.9	(29)
melamine	H2O2–NH4Cl	hydrogen peroxide	supramolecular aggregates	139	(30)
melamine	(NH4)2SO4	none	none	75.0	(36)
melamine	urea	none	none	50.0	(37)
melamine	urea–NH4Cl	none	none	103.3	this work

The morphology and microstructure of the as-prepared bulk g-C3N4 and g-C3N4 NSs samples were characterized by TEM. As shown in Figure a–e, bulk g-C3N4 possesses dense and thick layers to construct a massive two-dimensional sheetlike structure (Figure a,b), while the urea and inorganic ammonium salt as gas bubble-assisted g-C3N4 NSs show a flaky structure with bubble-like or irregular morphology and a significant decrease in thickness, as shown in Figure c–f. The TEM image of g-C3N4-M-U-600 in Figure c shows the characteristics of a crinkly structure. This kind of crinkly structure could be preserved, and a typical ultrathin NS-like architecture with a crinkly structure could be observed for g-C3N4-M-U-Cl-600 (Figure d). This phenomenon was reasonable because of the continuous gases emitted during the formation of g-C3N4 NSs with urea–inorganic ammonium salt as the additive monomers. These results suggest that large-scale thin g-C3N4 NSs with crinkly structures were successfully prepared by the modified polymerization process of melamine with urea–inorganic ammonium salt as the additive monomer.

Figure 4

Typical TEM images of as-prepared samples: (a) g-C3N4-M-550, (b) g-C3N4-M-600, (c) g-C3N4-M-U-600, (d) g-C3N4-M-U-Cl-600, (e) g-C3N4-M-U-S-600, and (f) g-C3N4-M-U-P-600.

The proposed formation process of multistage polymerization design for g-C3N4 NSs is showing in Figure . In the prepolymerization of melamine–urea–NH4Cl at the lower temperature, the volatile urea melts to exhibit flow dynamics and then partially decomposes, creating bubbles that could promote sufficient mixing of the entire system and expand the prepolymerized precursor. Then, with the continued polymerization of melamine at a higher temperature, the viscosity of the system is further increased, and NH4Cl encapsulated in the prepolymerization precursor with higher decomposition temperature begins to decompose to generate bubbles, which can effectively avoid the agglomeration of melamine in the higher-temperature polymerization process. Then, the g-C3N4 NSs could be obtained.

Figure 5

Proposed formation process of multistage polymerization design for g-C3N4 NSs.

X-ray photoelectron spectroscopy (XPS) measurements were carried out to reveal the surface chemical compositions of the obtained materials and are shown in Figure a–d. The XPS survey spectrum in Figure a shows that the g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, and g-C3N4-M-U-S-600 samples contain only C, N, and O species without other impurities, while g-C3N4-M-U-P-600 contains only C, N, P, and O species, which indicates that P could be introduced. For the C 1s spectra (Figure b), most of the samples showed similar two characteristic peaks located at 284.7 and 288.0 eV. The peak at 284.7 eV can be attributed to the signal of carbon impurities.[38] The main peak with a binding energy of 288.0 eV can be identified as an sp2-hybridized carbon in an N-containing aromatic ring (N–C=N).[39] In addition, there is a higher peak located at 288.5 eV in the C 1s spectrum for g-C3N4-M-U-P-600, which is attributed to O–C=O.[40] From the N 1s spectrum (Figure c), it can be seen that the major peak is located at 398.7 or 399 eV. The major peak located at 398.7 eV is assigned to sp2-hybridized nitrogen in C-containing triazine rings (C–N=C), whereas the peak at 399.0 eV is usually attributed to the tertiary nitrogen N–(C)3 groups.[38] The P 2p signal (Figure d) can be deconvoluted into three peaks located at approximately 133.3, 134, and 134.9 eV, respectively. The peaks at 133.3 and 134 eV are typical for P–N species,[41,42] which indicates that the P atoms may replace C atoms in the C–N framework of g-C3N4. The peaks at 133.3 and 134 eV can be attributed to the P–N and P=N bonds, respectively.[38] In comparison, the peak at 134.9 eV can be correlated with the P=O bond because of the reaction between P and O2 during the copolymerization reaction in air at high temperature, which is generally denoted as P2O5 in the literature.[43] The carbon and nitrogen XPS analyses further confirm the triazine or triazine heterocyclic ring structure of the obtained samples.

Figure 6

High-resolution XPS spectra, (a) complete XPS spectra, (b) C 1s spectra, (c) N 1s spectra, and (d) P 2p spectra.

The Fourier transformed infrared (FTIR) spectra are shown in Figure . In the spectrum of the obtained g-C3N4 samples, three characteristic bands were found similar to that of the typical g-C3N4 structure from the previous reports.[44−46] The absorption peak at 808 cm–1 corresponds to the characteristic breathing mode of triazine units,[44] while the strong band at 1200–1600 cm–1 is associated with the stretching vibration of the C–N and C=N heterocycles.[45] The broad absorption peak in the region of 3000–3500 cm–1 is attributed to the stretching vibrations of terminal N–H or N–H2 originated from uncondensed amino groups.[46] The peak centered at 980 cm–1 is clearly observed in the spectra of g-C3N4-M-U-P-600. This peak is assigned to the P–N stretching mode or vibration mode, which suggests that phosphorus atoms are doped into the crystal lattice of g-C3N4 successfully.[47,48] The vibrations of the P-related group were hardly observed in previous investigations, and when they were observed, they were attributed to either the low phosphorus content or an overlapping of its vibration by the C–N bond. The observed obvious P–N-related peak indicates that a higher concentration of P-doped g-C3N4 could be obtained by our method. In addition, there is a distinct adsorption peak at 2170 cm–1 in g-C3N4-M-U-P-600 corresponding to the defects of cyano group stretch caused by the incomplete polymerization and the loss of ammonia, which indicates that the cyano groups were successfully introduced into g-C3N4-M-U-P-600.[49,50] The cyano groups were detected only in the FTIR of g-C3N4-M-U-P-600 that was prepared by melamine–urea–ammonium phosphate, which indicates that the introduction of cyano groups is related to the phosphate ions in (NH4)3PO4. This is the first report that the cyano groups can be introduced by a phosphate anion, which is different from previously reported cyano groups that were formed by cations such as Na+ or K+ from NaCl or KCl/LiCl molten salt.[49,50]

Figure 7

FTIR spectra of g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600.

Optical and Photoelectrochemical Properties of the g-C3N4 Samples

The optical property of the obtained g-C3N4 samples were analyzed by UV–vis diffuse reflectance spectra. As shown in Figure a, it can be clearly observed that all of the samples had a strong absorption from UV to visible wavelengths. Compared to the bulk g-C3N4-M-550 or g-C3N4-M-600, the absorption edges of the g-C3N4 samples (g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600) exhibited a systematically slight blue shift with the added urea and volatile inorganic ammonium salt. Accordingly, the band gap energies of g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600 were determined to be 2.55, 2.57, 2.71, 2.78, 2.68, and 2.65 eV, respectively (Figure b). This kind of blue-shift performance of g-C3N4 NSs can presumably be ascribed to its nanoporous or NS structure, which can result from the quantum confinement effect.[51,52]

Figure 8

UV–vis diffuse reflectance spectra of the as-prepared samples’ (a,b) relationship between (Ahν)1/2 vs (hν) for the band gap energy of g-C3N4 samples.

The charge carrier’s separation abilities of as-synthesized samples were monitored by photoluminescence (PL) analysis. As depicted in Figure a, it can be observed that an obvious decrease can be observed with g-C3N4 by adding urea and a volatile inorganic ammonium salt, which may be caused by the different structure. The g-C3N4-M-U-Cl-600 and g-C3N4-M-U-P-600 samples exhibited the lowest PL emission intensity. Generally, the lower PL intensity indicates a decrease in the recombination of photoinduced electrons and holes.[52] Therefore, the lower PL emission intensity of g-C3N4 NSs is probably related to the introduced surface defects, porous structure, and thin NSs,[18,53] which can improve the charge carrier transfer rate and benefit the improvement of photocatalytic activity. For g-C3N4-M-U-Cl-600, the thin NSs structure may reduce the carrier diffusion distance and thus decrease the possibility of recombination of the photoinduced electrons and holes. However, for g-C3N4-M-U-P-600, the cyano groups were introduced to the structure and the cyano groups as an electron-withdrawing group may affect the surface charge distribution of the catalyst and then inhibit photoinduced carrier recombination to some extent. From this point of view, it is reasonable for the above two samples (g-C3N4-M-U-Cl-600 and g-C3N4-M-U-P-600) to own the lowest PL emission intensity.

Figure 9

(a) PL emission spectra; the excited wavelength was 330 nm. (b) Periodic ON/OFF photocurrent response in 0.1 M Na2SO4 electrolyte under visible light irradiation (λ > 400 nm) at 0.0 V vs SCE electrode.

Furthermore, the charge carrier transfer process and visible light response were sequentially probed by transient photocurrent–time plots. As shown in Figure b, all the samples exhibit a sensitive photocurrent response under visible light illumination. In the first on–off switch, the g-C3N4 samples prepared by adding urea and a volatile inorganic ammonium salt show the higher photocurrent compared with that of the bulk g-C3N4-M-550 or g-C3N4-M-600. When the on–off switch was operated for a longer time, g-C3N4-M-U-Cl-600 shows the highest photocurrent, which is 4.4 times higher than that of g-C3N4-M-550. This observation indicates more efficient transportation of photogenerated electron–hole pairs on g-C3N4-M-U-Cl-600 than on that of bulk g-C3N4, which contributed to the enhanced photocatalytic activity.

Photocatalytic Activity of the g-C3N4 Samples

The photocatalytic activity of g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600 was evaluated by the degradation of RhB (20 ppm), MO (10 ppm), and MB (15 ppm), which are the hazardous dyes, under visible light irradiation. As the prepared catalyst has different adsorption characteristics for different dyes, the adsorption capacity for RhB, MB, and MO is sequentially decreased. In order to maintain the concentration of the system after adsorption, different initial concentrations are set. In the process of comparison, each of the catalyst has been used to test the degradation ability of the above three dyes. For a certain catalyst prepared in the present study, it has different adsorption capacities for different dyes, RhB, MB, and MO. To fix the effective concentration of different types of dyes after the adsorption by the catalyst, different initial concentrations of RhB, MB, and MO dyes are used. However, for a certain dye of the three dyes, the initial concentrations for the different catalysts are the same. Actually, the relative catalytic activity of a certain catalyst was determined by comparing the degradation activities of the different catalysts for the same dye at the same concentration. Figure a shows the photodegradation of RhB solutions of g-C3N4 obtained with different urea and volatile inorganic ammonium salts under visible light irradiation. The RhB solutions containing g-C3N4 NSs underwent significant degradation and became nearly transparent within 50 min (shown in Figure a). It should be noted that the sample g-C3N4-M-U-P-600 showed the highest photocatalytic oxidation on RhB. After only 20 min of irradiation with visible light, approximately 96% RhB was already degraded by the catalyst (g-C3N4-M-U-P-600), as shown in Figure a. However, the bulk g-C3N4 obtained without adding any urea or volatile inorganic ammonium salt showed poorer photocatalytic activity, only about 19 and 31% of RhB was degraded by g-C3N4-M-550 and g-C3N4-M-600, respectively. The original strategy of adding urea and a volatile inorganic ammonium salt into melamine to enhance the photocatalytic was successful.

Figure 10

Photodegradation of dye solutions by using g-C3N4 obtained with different mixed molten salts under visible light irradiation in neutral suspension, (a) 20 ppm RhB, (b) 10 ppm MO, (c) 15 ppm MB, and (d) removal efficiency of different dyes by adsorption.

To further confirm whether the obtained sample has a similar activity for degradation of other dyes, we also compared the photocatalytic activity of the degradation of MO and MB (shown in Figure b,c). Surprisingly, the trend is not the same; it even showed opposite changes. For MO degradation, g-C3N4-M-U-Cl-600 showed the highest photocatalytic activity. After 100 min of illumination under visible light, nearly 78% of MO was degraded by g-C3N4-M-U-Cl-600. However, the photocatalytic activity of g-C3N4-M-U-P-600 was reduced to the same level of pure g-C3N4 (g-C3N4-M-550 or g-C3N4-M-600); approximately 53 and 46% MO were degraded by g-C3N4-M-U-P-600 and g-C3N4-M-600, respectively. For MB degradation, g-C3N4-M-U-Cl-600 still showed the highest photocatalytic activity, and approximately 98% of the MB was eliminated after 50 min. However, the photocatalytic activity of g-C3N4-M-U-P-600 still decreased to the same level of g-C3N4-M-600 and approximately 67% MB was degraded after 50 min.

This anomalous phenomenon prompted us to think more deeply about the reasons that led to this phenomenon. We realized that to explain this phenomenon, we needed to consider the combination of the nature of the dye itself, the specific surface area of the catalyst, and its surface charge state. There is a mutual correlation between the three factors and even the constraints. From the dye itself, we know that MB, RhB, and MO typically represent the cationic dye, the amphoteric dye, and the anionic dye, respectively. Therefore, the surface charge state of the catalyst may affect the adsorption and degradation properties of different types of dyes.

In general, high surface areas can endow the catalysts with more available active adsorption sites, thus resulting in a better adsorption performance. The dye adsorption property of the obtained g-C3N4 NSs was investigated by removing the cationic dye MB, the amphoteric dye RhB, and the anionic dye MO molecules from water under dark conditions, and the results are shown in Figure d. In theory, the physical adsorption of a dye by the surface of the catalyst should be similar for different dyes if the surface area is similar. However, from Figure d, we can see that significant differences exist for the different dyes. For example, the BET surface area of g-C3N4-M-U-Cl-600 is 2 times higher than that of g-C3N4-M-U-P-600. However, the two samples show similar adsorption characteristics for MB (77 and 68% at 30 min, respectively), while showing different trends for RhB and MO. Although the BET surface areas of g-C3N4-M-U-S-600 and g-C3N4-M-U-P-600 are the same, they show the different adsorption characteristics for MB (57 and 68% at 30 min, respectively). From the above discussion, we may suppose that except for physical adsorption, other factors need to be considered, such as electrostatic attractive interactions.

To distinguish the type of surface charge of the obtained samples, a zeta potential test for all the samples was carried out in deionized water. From the zeta potential, we know that the surface charge is −13, −31, −29, −22, −23, and −38 mv for g-C3N4-M-550, g-C3N4-M-600, g-C3N4-M-U-600, g-C3N4-M-U-Cl-600, g-C3N4-M-U-S-600, and g-C3N4-M-U-P-600, respectively (shown in Figure S2). The zeta potential of g-C3N4-M-U-P-600 was the most negative, which indicates that the incorporation of the P atom has greatly changed the electronic properties of g-C3N4. This is reasonable; when the P atom is inserted into the framework, four of the five electrons form covalent bonds with the N neighbors to adopt into the planar structure. The remaining lone electron of the P atom will delocalize into the p-conjugated triazine ring, thus creating an electron-rich state of the P-doped g-C3N4.[34,38] From the tested negative zeta potential, we may guess that electrostatic interactions could be strengthened between the cationic dye (MB) or the anionic dye (MO), which may promote adsorption for the cationic dye (MB) and reduce adsorption for anionic dye (MO) by electrostatic attraction or repulsion interaction. This conjecture is confirmed by the adsorption performance shown in Figure d and appropriately explains the abnormalities observed in the experiment. From the above discussions, it can be concluded that the improved adsorption and degradation ability of the obtained g-C3N4 composites results from the synergy of the BET surface area and the surface charge state.

The photocatalytic activity of g-C3N4 NSs was also investigated by evaluating photocatalytic hydrogen evolution under visible light. As shown in Figure a, the hydrogen evolution rates of g-C3N4-M-550 and g-C3N4-M-600 are 197.7 and 432.0 μmol·h–1·g–1, respectively. After the addition of urea and volatile inorganic ammonium salt, g-C3N4-M-U-600 (1395.4 μmol·h–1·g–1), g-C3N4-M-U-Cl-600 (1853.8 μmol·h–1·g–1), g-C3N4-M-U-S-600 (1167.8 μmol·h–1·g–1), and g-C3N4-M-U-P-600 (1193.7 μmol·h–1·g–1) all displayed enhanced H2 evolution activity that is greater than that of g-C3N4-M-550 or g-C3N4-M-600. The g-C3N4-M-U-Cl-600 sample showed the highest H2 evolution rate of 1853.8 μmol·h–1·g–1, which was 9.4 times higher than that of g-C3N4-M-550. On the basis of the above results, the enhanced photocatalytic activity of g-C3N4 NSs can be ascribed to the synergetic effect of the enhanced surface area, surface structure, and surface charge state. From the above discussion, we know that both the synergy of BET surface area and the surface charge should be considered for dye adsorption and degradation. For hydrogen evolution, we find that the synergy of the two still existed, and the BET surface area played a dominant role.

Figure 11

(a) Photocatalytic H2 evolution as a function of reaction time and (b) stability test of H2 evolution for g-C3N4-M-U-Cl-600 under visible light for five circles.

Moreover, the hydrogen production rate over the g-C3N4-M-U-Cl-600 photocatalyst after constant irradiation for five cycles could be sustained, which shows good stability (Figure b). This high stability without deterioration in the photocatalytic activity is indispensable for practical applications of this photocatalyst.

Conclusions

In conclusion, g-C3N4 NSs that exhibit much improved photocatalytic activity has been successfully synthesized with the modified polymerization process of melamine by cocondensation of melamine with volatile salts. The as-fabricated g-C3N4 NSs possessed high photocatalytic activity with variable high specific surface areas and surface charge states, which can be controlled by adjusting the mass ratio of urea–inorganic ammonium salt and calcination temperature. The result is a good explanation for the hypothesis that adding monomers can truly regulate the polymerization process of melamine when there is no molecular self-assembly. Moreover, the revelation of this phenomenon is that molecular self-assembly is beneficial for the regulation of the ordered structure. However, we also need to pay attention to the role of the additive monomers and solvents; otherwise, the supramolecular structure itself may influence the polycondensation process. With the properly selected monomer, we can obtain excellent nanostructures of g-C3N4 by simply adjusting the polymerization process. Considering the inexpensive feedstocks used and the simple synthetic controlling method make it highly appealing for large-scale production of highly photocatalytic, visible-sensitizable, metal-free g-C3N4 photocatalysts. In addition, the obtained g-C3N4 NSs show good stability and can be repeatedly used without significant reduction in the photocatalytic activity.

Experimental Section

Sample Preparation

Materials Used

All starting materials were purchased from Sinopharm Chemical Reagent Corp, P. R. China, and used without further purification.

Synthesis of g-C3N4 Photocatalysts

Synthesis of Bulk g-C3N4

Bulk g-C3N4 was formed with 6 g of melamine. In a detailed experiment, 6 g of melamine was placed into an alumina crucible with a cover and heated at 550 °C or 600 °C for 4 h in a semiclosed system with a ramp rate of 5 °C/min. When the sample was cooled to room temperature, the sample was removed from the muffle furnace. The obtained samples were called g-C3N4-M-550 or g-C3N4-M-600.

g-C3N4 NSs

g-C3N4 NSs were synthesized using 6 g of melamine, 10 g of urea, and 4 g of ammonium sulfate, which were called g-C3N4-M-U-S-600; g-C3N4 NSs synthesized with 6 g of melamine, 15 g of urea, and 4 g of ammonium phosphate were called g-C3N4-M-U-P-600; g-C3N4 NSs synthesized with 6 g of melamine, 15 g of urea, and 4 g of ammonium chloride were called g-C3N4-M-U-Cl-600. In addition, the samples obtained with 6 g of melamine and 15 g of urea, 4 g of ammonium sulfate, or 4 g of ammonium chloride were called g-C3N4-M-U-600, g-C3N4-M-P-600, and g-C3N4-M-Cl-600, respectively. The mixture was ground with an agate mortar. Then, the mixture was put in an alumina crucible with a cover and heated at 600 °C for 4 h in air with a ramp rate of 5 °C/min. When the sample was cooled to room temperature, it was removed from the muffle furnace.

Material Characterization

Powder XRD analysis was carried out using a Rigaku D/MAX 2500X X-ray powder diffractometer with Cu Kα radiation. TEM observations were conducted on a JEOL-2010 electron microscope at an acceleration voltage of 200 kV. Zeta potentials of the samples dispersed in water were analyzed by a Malvern zeta analyzer (ZS90-2026). XPS measurements were performed on a PerkinElmer PHI 5000C ESCA with Al Kα radiation operated at 250 W. BET-specific surface areas and porosity of the g-C3N4 samples were calculated based on nitrogen adsorption isotherms measured at 77 K using a gas adsorption apparatus 3H-2000PS4. The specific surface area was obtained by the BET method. Pore size distribution (PSD) was calculated from the adsorption branch of the isotherm using the Barrett–Joyner–Halenda model. Ultraviolet–visible diffuse reflection spectra (UV–vis DRS, Cary Eclipse) were recorded on an Agilent Cary 5000 UV–vis–NIR spectrophotometer in the range of 200–800 nm at room temperature. FTIR spectra were recorded with KBr disks holding the powder sample with an FTIR spectrometer (PE Spectrum Two). The PL spectra of the mesoporous g-C3N4 samples were examined at room temperature after excitation with incident light of 330 nm by a spectrofluorophotometer (Cary Eclipse), and the slit widths at the excitation and emission of the spectroflurometer were 2.5 and 2.5 nm, respectively.

Photocatalytic Hydrogen Production

The photocatalytic activities of bulk g-C3N4 and g-C3N4 NSs were compared by water splitting for hydrogen evolution under visible light irradiation. Before the hydrogen production, 6.4 mL of H2PtCl6 aqueous (3 wt % Pt) solution, 20 mL of methanol, and 80 of mL deionized water were combined. Then, 200 mg of the g-C3N4 sample was added into the solution. Finally, the mixture was photo-reduced in situ during the photocatalytic reaction for 1 h under full arc. During the hydrogen production, 50 mg of the photocatalyst was dispersed ultrasonically in 100 mL of an aqueous solution containing 10 vol % of triethanolamine solution as a sacrificial agent. Then, the system was degassed, and the solution was agitated for 30 min. Next, the solution was irradiated for 3 h with a 300 W xenon lamp (CEAULIGHT) with a 400 nm cutoff filter. The gas was extracted every 30 min and the photocatalytic H2 evolution rate was analyzed by a GC-7900 (CEAULIGHT), using high-purity Ar as the carrier gas.

Photocatalytic Degradation Test

The photocatalytic degradation test is also a way to evaluate the photocatalytic properties of different samples. RhB (20 mg/L), MO (10 mg/L), and MB (15 mg/L) were used as the target pollutants. In each run, 100 mg of the photocatalyst was dispersed in 100 mL of the target pollutant solutions. Prior to the photocatalytic degradation test, the adsorption–desorption equilibrium was achieved by stirring the mixture in a dark environment for 30 min (the dye adsorption vs time in the dark condition of g-C3N4 shown in Figure S3). A 300 W xenon lamp with a 400 nm cutoff filter irradiated the dye solution for the purpose of degradation. After a certain period of time, approximately 5 mL of the suspension was removed for detection. The UV–visible absorption spectra of the supernatant solution were analyzed by a UV–visible spectrometer (Shimadzu UV-2550).

Photoelectrochemical Measurements

Photoelectrochemical measurements were performed using a PGSTAT204 (Metrohm) electrochemical analyzer, employing a three-electrode system with a saturated calomel electrode (SCE) as the reference electrode. A 0.1 M Na2SO4 aqueous solution was employed as the electrolyte. The light source was a 300 W xenon lamp with a 400 nm cutoff filter. Then, 30 mg of samples was mixed with 1200 μL of ethanol, 800 μL of isopropanol, and 100 μL of PEDOT:PSS (Aladdin). The solution was then subjected to ultrasonication for 30 min. After 2 min, drops of the obtained solution were deposited onto a clean fluorine-doped tin oxide glass electrode. Finally, the electrodes were heated at 150 °C for 1 h under vacuum.