The field of heterogeneous photocatalysis has developed rapidly in the last four decades as governments and scientists research green and sustainable technologies [1, 2]. Semiconductor-based photocatalysis requires only light as a driving force. A suitable semiconductor as a photocatalyst has been used in catalytic reactions for a variety of applications, such as hydrogen production from water splitting [3], CO₂ reduction into hydrocarbonfuels [4], decomposition and mineralization of organic pollutants [5], selective organic synthesis [6], and disinfection of bacteria [7]. Semiconductor-based photocatalysis has emerged with valuable metal-based semiconductors; however, they are active only in the ultraviolet region and have moderate performances.

Graphitic carbon nitride (g-C₃N₄) is the most stable allotrope among various carbon nitrides (CNs) under ambient conditions. Unlike TiO₂, which is only active in the UV region, g-C₃N₄ possesses a bandgap of ca. 2.7 eV [8, 9], which enables it to be a visible-light active photocatalyst for a range of reactions. More importantly, g-C₃N₄ is only composed of two earth- abundant elements: carbon and nitrogen, suggesting that it can be easily prepared at low cost [10, 11]. Moreover, its polymeric nature allows control over the surface chemistry via molecular-level modification and surface engineering. The unique aforementioned characteristics of g-C₃N₄ make this material a very promising photocatalyst for various applications [12]. Great and fruitful efforts have been made on g-C₃N₄-based photocatalysis [13]. However, pristine g-C₃N₄ still suffers from unsatisfactory photocatalytic efficiency because of its restricted visible-light harvesting capacity, ready recombination of charge carriers, and low surface area [13-21].

Many researchers have put great effort of preparing g-C₃N₄ nanostructures (synthetic routes [22-24], thermal exfoliation [25-27], and templates [28]). For example, Shalom et al. [23] reported a new and simple synthetic pathway to form ordered, hollow CN structures, using a cyanuric acid-melamine complex in ethanol as a starting product. Yang et al. [24] demonstrated a biotic precursor approach of g-CNX polymers synthesized from urea and nucleobases. Yang et al. [27] reported a facile and green approach to prepare few-layered polymeric CN semiconductors by a one-step carbon/nitrogen steam reforming reaction. g-C₃N₄ can be imprinted with a twisted hexagonal rod-like morphology by a nanocasting technique using chiral silicon dioxides as templates [28]. Ou et al. [29] fabricated crystalline CN nanosheets by exfoliation of bulk tri-s-triazine- based crystalline CN powder in isopropanol via sonication for 15 h. Guo et al. [30] presented a facile synthesis method (the mixture of glucose, boric acid and urea) to produce a porous structure of two-dimensional boron CN nanosheets. In the results of the above-mentioned studies, g-C₃N₄ exhibited dramatically enhanced visible-light photocatalytic activity toward hydrogen evolution and pollutant degradation. A potential scale method for preparing g-C₃N₄ nanosheets remains a challenge.

In this work, we developed a simple method to prepare g-C₃N₄ nanosheets by thermal polymerization of cyanuric acid and melamine in air. These g-C₃N₄ nanosheets may show superior photocatalytic activities compared with the bulk g-C₃N₄. The photocatalytic activity of g-C₃N₄ is evaluated toward phenol degradation under visible-light irradiation and compared with that of g-C₃N₄.

Melamine, cyanuric acid, phenol, and methanol were purchased from Sinopharm (Shanghai, China). All chemicals were of analytical grade and used without further purification. All aqueous solutions were freshly prepared with deionized water.

All g-C₃N₄ samples were synthesized using melamine and cyanuric acid as the starting material through a stage programming heating approach. The starting material was heated to 550 ℃ at a heating rate of 2 ℃/min and held at this temperature for 4 h. The sample was denoted as g-C₃N₄(x), where x refers to the molar ratio of melamine and cyanuric acid.

For comparison, bulk g-C₃N₄ was prepared through a widely-used one-step polycondensation process. Briefly, 2 g of MA was directly heated to 550 ℃ in air and kept for 2 h.

Scanning electron microscopy was performed with a FEI Nove NanoSEM 230 field emission system on loose and lightly pressed samples. X-ray diffraction patterns were measured on a Rigaku 2500 diffractometer with Cu-K_α radiation (λ = 0.15406 nm) at a scan rate of 8°/min. X-ray photoelectron spectroscopy (XPS) spectra were obtained on a ThermoFisher- VG Scientific instrument with an Al-K_α (1486.6 eV) monochromatic X-ray radiation (operated at 200 W) from a twin anode in the constant analyzer energy mode with an energy of 30 eV. The UV-vis absorption spectra were measured on a Shimadzu UV2550 spectrophotometer using BaSO₄ as the reflectance standard.

Photodegradation of a phenol solution (50 mg/L) was performed to evaluate the photocatalytic performance of the synthesized catalysts in a top-window Pyrex cell with the temperature maintained at 20 ℃ by a circulating water system. The catalyst (50 mg) was added into the phenol solution (100 mL). Prior to irradiation, the suspension was magnetically stirred in the dark for 30 min to ensure phenol adsorption/desorption equilibrium. The suspension was irradiated by a 300 W Xe lamp with a cut off filter ( < 400 nm) and an irradiation intensity of 100 mW/cm². At given time intervals, aliquots of the irradiated suspension were collected, centrifuged, and analyzed on a Shimadzu LC-20AT high-performance liquid chromatography system with an SPD-20A column. The detection wavelength was 280 nm. The mobile phase was a mixture of methanol and water with a volume ratio of 70:30 and a flow rate of 1 mL/min.

The one-step pyrolysis of the precursor melamine showed decreased yields of the products as the molar ratio of melamine to cyanuric acid was varied from 1:0 to 1:9. Cyanuric acid is totally decomposed in these conditions. The appearance of the sample looked like messaline with characteristics associated with the delamination and crystal-structural alternation within the CN polymers [26, 31].

Fig. 1 shows the scanning electron microscopy images of all samples. Many flakes with laminar morphologies were observed. However, the bulk g-C₃N₄ showed no exfoliation with irregular particles. The product after thermal exfoliation presented in Fig. 1 displayed a layer structure with some fabric-like surface. Fig. 2 shows a representative atomic force microscopy image of the g-C₃N₄(1:9) nanosheets. The lateral size of these sheets ranged from tens of nanometers to several micrometers. The thickness analysis of the nanosheets revealed a thickness of about 3.0 nm.

As cyanuric acid decomposed, the appearance of CN became mainly atomic layers with a thin, glossy, and transparent texture (Fig. 1). Moreover, from the detailed view of the sample, a distinct curved nanodomain was displayed at the edge of the large plane (Fig. 1). Such a structural distortion of 2D crystals has been demonstrated to stabilize the 2D structure, as also observed in single-layered graphene, which may be stabilized by the formation of finite-sized ripples [32, 33].

Fig. 3 shows X-ray diffraction patterns of all the samples with a characteristic peak at 27.48° (d = 0.326 nm), corresponding to the (002) interlayer reflection of g-C₃N₄. The peak at 27.48° was weaker than those of the sample obtained by bulk polymerization at 550 ℃, which indicated that the interlayer structure was destroyed after thermolysis. Moreover, the peak at 27.48° (attributed to the in-plane repeated tri-s-triazine units) become narrow with increased reaction temperature. The evolution of the curved shape and curls during the assembly process in a fluidic medium may disturb the long-range packing of the tri-s-triazine rings (C₃N₃) in a layer along the (002) direction, as shown with the formation of the nanosheets topology at g-C₃N₄(1:9). The weak peak at 13.1° corresponds to the (100) plane, which is the in-plane structural packing of tri-s-triazine units caused by the decreased planar size of the layers during the thermal oxidation etching of bulk g-C₃N₄[26].

The chemical structures of the samples were further analyzed by Fourier transform infrared spectroscopy, and the results are shown in Fig. 4. The broad peaks between 3000 and 3500 cm⁻¹ correspond to the N–H bond. The peaks in the region from 900 to 1800 cm⁻¹ are attributed to either trigonal C–N(–C)–C (full condensation) or bridging C–NH–C units, and these bands became sharp because of the ordered packing of hydrogen bonds in the long strands of polymeric melamine units after thermal oxidation etching in the layers of nanosheets. The peaks at 1251, 1325, 1419, 1571, and 1639 cm⁻¹ are the stretching modes of CN heterocycles. In addition, the characteristic breathing mode of triazine units at 810 cm⁻¹ was observed [34]. There were no differences of peaks between all samples, which indicated that the chemical structure of g-C₃N₄ (different molar ratios of melamine to cyanuric acid) was the same as that of bulk g-C₃N₄.

The optical absorption properties of the g-C₃N₄ were examined with UV-vis diffuse reflectance spectroscopy (Fig. 5). The texture evolution at the nanoscale can induce planar g-C₃N₄ warping spontaneously, and accordingly, an important modification of the optical absorption property [35, 36]. As shown in Fig. 5, the samples exhibited one absorption edge around 450 nm in the blue region, which was identified as the intrinsic electronic transition from the HOCO-2 to LUCO in g-C₃N₄ polymers (π → π* transitions). Because of the extended 2D electron delocalization, the band edge slightly red-shifted at a high temperature [26, 31, 37].

Using Brunauer-Emmett-Teller theory, the surface area of g-C₃N₄(1:9) was estimated as 103.24 m²/g, which is much larger than those of g-C₃N₄ prepared from urea (≈58.0 m²/g) [38] and melamine (8.0 m²/g) [39]. The g-C₃N₄(1:9) exhibited an enhanced performance in pollutant removal, such as phenol. The g-C₃N₄(1:9) showed a higher adsorption capacity of phenol than that of the g-C₃N₄(all) in the first 30 min. However, the g-C₃N₄ cannot remove the pollutants after the adsorption equilibrium was reached at 3 h. The g-C₃N₄ with a low molar ratio of melamine to cyanuric acid showed a low adsorption capacity and photocatalytic degradation ability, indicating a low pollutant removal ability. Because of the simultaneously high adsorption capacity and photocatalytic degradation ability, the high molar ratio enhanced the performance of pollutant removal via a cooperative effect of adsorption and photocatalytic degradation, which was 10 times that of pure g-C₃N₄ (Fig. 6(c)). Photocatalyst stability is important in application, so the reusability of the g-C₃N₄(1:9) for phenol degradation was evaluated (Fig. 6(d)). Over five consecutive cycles, no obvious deactivation of the photocatalyst was observed. To further test the mineralization degree of phenol during the photodegradation process, the evolution of total organic compounds during light irradiation was investigated (Fig. 6(e)). About 91% of phenol was mineralized to CO₂ within 210 min with the g-C₃N₄(1:9), which is three times that of the g-C₃N₄(1:0) during the same period. This result indicates that the g-C₃N₄(1:9) has a much higher mineralization efficiency for phenol photodegradation than g-C₃N₄(1:0).

As shown in Fig. 7(a), the addition of benzoquinone (BQ) and isopropanol (IPA) caused a notable change in photodegradation efficiency, suggesting the importance of •O₂⁻ radicals and •OH radicals. The addition of ethylenediaminetetraacetic acid (EDTA) showed almost no effect on photodegradation efficiency, suggesting a low importance of h⁺ radicals. The Mott-Schottky plot for the g-C₃N₄(1:9) in Fig. 7(b) is linear to a reverse potential of −1 V. The x-intercept of the Mott-Schottky plot indicates a flat band voltage of −0.96 V. The temporal evolution of phenol and the hydroxylated phenolic intermediates for the g-C₃N₄(1:9) in Fig. 7(c) suggest that the major photodegradation products are catechol and hydroquinone at the beginning. As shown in Fig. 7(d), the photogenerated holes in the valance band of g-C₃N₄ (E_VB = +1.30 V, vs. NHE) are incapable of oxidizing hydroxyl groups into •OH radicals (E(•OH/OH⁻) = +1.99 V, vs. NHE) because of the more negative valance band potential, suggesting that the observed •OH radicals were generated from the •O²⁻ radicals by a photochemical reaction. When phenol molecules are adsorbed on the surface of excited g-C₃N₄ nanosheets, there is activation of the phenol molecules by reaction with an •OH radical. The hydroxyl radical shows electrophilic character and prefers to attack electron rich ortho or para carbon atoms of phenol, forming dihydroxycyclohexadienyl radicals that undergo further reaction with dissolved oxygen to yield dihydroxy benzenes with simultaneous generation of an •OH radical. Dihydroxycyclohexadienyl radicals are also converted to phenoxy radicals. These phenoxy radicals can react with •OH to form benzoquinone, hydroquinone, which are colored intermediates and also dihydroxy benzenes. The direct combination of two phenoxy radicals can form intermediates with two aromatic rings attached to each other by a single bond.

The synergy of adsorption and photocatalytic degradation by high temperature is further explored by photocurrent analysis. As shown in Fig. 8, a low photoelectric response of agar was observed. The photocurrent of low molar ratio cyanuric acid samples is similar to that of pure g-C₃N₄, whereas the high molar ratio cyanuric acid materials show an enhanced photoelectric response. Another interesting observation was the slow photocurrent response of the g-C₃N₄ sample during the on–off irradiation cycles. Upon light irradiation, the conduction band of the g-C₃N₄ sample serves as electron reservoirs to store the photogenerated electrons, so only a portion of the photoelectrons are transported to the back-contact electrode until the equilibrium state is reached. Thus, a gradually rising photocurrent response occurs.

Structurally distorted CN polymers (g-C₃N₄(1:9)) were fabricated during the delamination process of the layers. The minimum band gap is sensitive to the configuration of the g-C₃N₄. As a wide-range visible-light photocatalyst, buckled CN converts light ( > 495 nm) to run chemical reactions such as phenol removal. The basic units and connecting mode of CN show no major alternation, but the charge separation is promoted because of the 2D buckled structure and high polycondensation. This paper provides a new and simple method to synthesize high-performance g-C₃N₄ nanosheets, which explores a new photocatalytic oxidation catalyst for environmental remediation and energy conversion.