Temperature-controlled morphology evolution of graphitic carbon nitride nanostructures and their photocatalytic activities under visible light

Abstract

A series of g-C₃N₄ photocatalysts were synthesized by changing the temperature through air-assisted thermal polymerization and etching. The detailed characterization and analysis clearly revealed that the texture, surface structure, optical properties, and band structure of g-C₃N₄ depend on the preparation conditions such as atmosphere and temperature. The morphology of samples obtained at different temperatures undergoes sequential evolution from bulk to nanosheet with multilayer, nanosheet with few layer, moderately rolled nanosheet, tailored nanotube, nanoflakelet and even nanoparticles. Accordingly, the BET surface area increased dramatically from 4.5 m² g⁻¹ of bulk g-C₃N₄ to 210.1 m² g⁻¹ of modified sample obtained at 540 °C. The different textural and structural properties resulted in the different optical and electronic features. The exfoliated and tailored samples obtained by this method presented better photocatalytic performance as compared with bulk g-C₃N₄. Factors affecting the photocatalytic activity of g-C₃N₄ were discussed in details on the basis of the textural and structural characterization as well as the photocatalytic activity results. We found that the surface area, surface defects, and light absorption have great impacts on the photocatalytic activities of g-C₃N₄.

---

1. Introduction

Metal-free semiconductors (such as graphene and its derivatives as well as carbon nitride and C₃N₃S₃ organic conjugated polymers) with a great deal of advantages (e.g. non-toxic and abundant) have been developed as photocatalysts.^1–5 Among these semiconductors, graphitic carbon nitride (g-C₃N₄) has been considered as a suitable and promising photocatalyst for the splitting of water into hydrogen or oxygen using solar light. However, condensed bulk g-C₃N₄ cannot give rise to a high photocatalytic performance due to its unsatisfied texture and surface structure. Considering the graphitic structure of g-C₃N₄ and inspired by the discovery and preparation of graphene,^6–8 delaminating, exfoliating, and tailoring the structure of g-C₃N₄ at the nanoscale is a very important research direction in photocatalytic community to change the microstructure and surface property and thus to improve photocatalytic activity of g-C₃N₄, and is also a key to design highly efficient metal free photocatalysts.

As expected, recently, in the research of modification on the microstructure of g-C₃N₄, many breakthrough discoveries have been reported, ranging from bulk to nanosheets, nanotubes, and nanoparticles, from three-dimensional to two-dimensional or zero-dimensional growth. The methods for exfoliating and tailoring laminar structure of g-C₃N₄ include liquid exfoliation,^9–12 hydrothermal approach,¹³ protonation,¹⁴ and so on. For example, Zhang et al.⁹ utilized ultrasonic exfoliation route in water to prepare the ultrathin g-C₃N₄ nanosheets. The bulk g-C₃N₄ can be also exfoliated in 1,3-butanediol after ultrasonic processing.¹¹ Similarly, the crystalline C₃N₄ nanosheets was prepared by one-step liquid exfoliation from layered bulk poly(triazine imide) in water.¹⁰ Wang et al.¹⁴ reported that protonation can breaks up both sheets and stacks, and thus provides a facile method to modify the morphology and microstructure of g-C₃N₄. Besides, heat treatment of bulk g-C₃N₄ in air was also found be an impactful method to achieve the modification of g-C₃N₄.^15,16 All above mentioned methods for exfoliating and tailoring bulk g-C₃N₄ seem a bit complicated, and are consisting of two or more steps including the preparation of bulk g-C₃N₄ and subsequent post-treatment. It should be finding a simple alternative method to attempt one-step preparation to obtain nanosheets and/or tailored g-C₃N₄ by directly precursor heating.

A large number of studies showed that the exfoliation and tailoring of g-C₃N₄ can not only increase the surface area but also influence the surface structure, optical properties, electronic structure and energy band of g-C₃N₄.^9,11,13–15 Moreover, during the exfoliating process by thermal etching, both the atmosphere and calcination temperature greatly influence the morphology, structure, and optical properties of as-obtained samples.^15–17 All of these can affect the photocatalytic activity of g-C₃N₄. Under certain conditions, there are some contradictions between effects of some factors on the photocatalytic activity. For example, after exfoliation, the specific surface area increased but light absorption decreased.^9,11,13,15 Moreover, bandgap narrowing due to the modification of the defects such as nitrogen vacancies leads to the enhanced light absorption but lowered redox ability as a result of the down-shifted conduction band and up-shifted valence band.^18,19 Hence, an in-depth understanding on the main factors influencing the photocatalytic activity of g-C₃N₄ is very important for the synthesis of efficient g-C₃N₄ based photocatalysts.

Herein, in the present work, we use the method of air-assisted thermal polymerization and etching to obtain a series of g-C₃N₄ photocatalysts by changing the calcination temperature, and have systemically investigated the effects of calcination temperature on exfoliation process of g-C₃N₄, morphology evolution, optical properties, band-gap engineering, and photocatalytic activity for water splitting. Factors affecting the photocatalytic activity of g-C₃N₄ are discussed in details on the basis of the textural and structural characterization as well as the photocatalytic activity test. This work not only provides a simple strategy for exfoliating and tailoring bulk g-C₃N₄, but also offers in-depth understanding on structure–activity relationship of graphitic carbon nitride.

2. Experimental

2.1. Preparation of samples

Preparation of bulk g-C₃N₄-W reference. All chemicals used in the experiments were reagent grade and no further purification is needed before use. The bulk g-C₃N₄ reference was synthesized via thermal polycondensation. Typically, 6.0 g melamine was put in a cylindrical alumina crucible (∼5 cm in diameter) with cover. The crucible was heated up to 520 °C for 2 h and then extended to 540 °C for 2 h with a heating rate of 5 °C min⁻¹ in a muffle furnace. The product was finally grounded within a mortar and pestle to get g-C₃N₄ photocatalysts (denoted as g-C₃N₄-W).

Preparation of modified g-C₃N₄. The modified g-C₃N₄ with desired nanoscale and different morphology can be directly synthesized by air-assisted thermal polymerization and etching (denoted as g-C₃N₄-T, where T is the synthesis temperature at 470, 500, 520, and 540 °C, respectively). Typically, a certain amount of melamine (6.0 g) was loaded at the centre of an opened combustion boat (alumina crucible with 50 mL volume, 137 × 27 × 21 mm). Afterwards the combustion boat was heated to a desired temperature in a muffle furnace for 2 h and then extended to a higher temperature for 2 h. After cooling to room temperature, the sample with specific colour and physical condition was deposited on the both ends of crucible boat, where the temperature is lower. The temperature curve versus time and a group of different temperature settings are compiled and shown in Fig. S1 (in ESI†).

2.2. Characterizations

The X-ray diffraction (XRD) patterns of all samples were collected on XRD-6000 X-ray diffractometer (Cu Kα source) at a scan rate of 2° min⁻¹. Fourier Transform Infrared spectra (FTIR) were derived from Perkin Elmer Fourier Transform Infrared Spectrometer GX. The X-ray photoelectron spectra (XPS) were determined on a VG ESCALAB 250 XPS system with a monochromatized Al Kα X-ray source (15 kV, 200 W, pass energy = 20 eV). The scanning electron microscopy (SEM) images were obtained by JEOL JSM-7600F scanning electron microscope at an acceleration voltage of 5 kV. The transmission electron microscopy (TEM) images were obtained by a JEOL JEM-2010 EX instrument at an accelerating voltage of 200 kV. The N₂ absorption–desorption isotherms were determined at 77 K with a Micromeritics ASAP 2010 instrument. Electron paramagnetic resonance (EPR) signals of samples were examined with a Bruker ESP 300E spectrometer. With Lambda 750 UV/vis/NIR spectrophotometer (Perkin-Elmer, USA) using BaSO₄ as reference, UV-vis absorption was measured. Mott–Schottky curves were measured using a three-electrode system in 0.5 M Na₂SO₄ solution. Photoluminescence (PL) spectra were accomplished in solid with Shimazu RF5301 Spectrofluorophotometer with an excitation wavelength of 380 nm.

2.3. Photocatalytic properties

Photocatalytic H₂ evolution. The photocatalytic hydrogen production was carried out in a quartz reactor. Typically, 10 mg of the Pt (3.0 wt%) loaded sample was dispersed into 10 mL triethanolamine (TEOA, 15 vol%) aqueous solution. Before the irradiation with visible light, the suspension was degassed with N₂ for 15 minutes to remove the O₂ in the system. A 300 W Xenon Lamp (MAX-302, Asahi Spectra, USA) coupled with a UV cut-off filter (λ > 420 nm) was adopted to provide the visible light. The amount of produced H₂ was analyzed at a regular interval (1 h) by gas chromatograph (Agilent 7890A) with TCD detector. All reactions were accomplished at room temperature.

Photocatalytic O₂ evolution. Firstly, The RuO_x cluster loaded samples were prepared by hydrolysis method. Typically, 20 mg as-prepared sample was added to mixture solution containing 25 mL water and 25 mL ethanol with constant stirring for 0.5 h. Then, 20 μL of RuCl₃ solution (20 mg Ru mL⁻¹) was added under constant stirring for 0.5 h. Subsequently, NaOH solution (0.04 M) was slowly added to adjust the pH to 7 and the mixed solution was then stirred overnight. The products were washed with water and ethanol, dried at 60 °C, and finally calcinated at 150 °C for 2 h.

Then, the photocatalytic oxygen evolution was carried out in a quartz reactor at room temperature. Typically, 10 mg of the as-prepared RuO_x loaded sample was dispersed into 10 mL AgNO₃ (0.1 M) solution. The suspension was degassed with Ar for 15 min to remove the O₂ in the system prior to irradiation under a 300 W xenon lamp (MAX-302, Asahi Spectra, USA) with a UV cut-off filter (λ > 420 nm). The produced O₂ was analyzed at a regular interval (1 h) by gas chromatograph (Agilent 7890A) with TCD detector.

3. Results and discussion

3.1. Evolution of structure and morphology

The crystal structure, surface composition and chemical state of as-prepared samples were firstly characterized by XRD, FTIR, and XPS, respectively. Fig. 1a shows XRD patterns of g-C₃N₄-T samples (g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540) obtained at different temperature under air atmosphere and bulk g-C₃N₄-W reference prepared by general thermal condensation of melamine. The XRD pattern of the g-C₃N₄-470 presents two diffraction peaks at 13.1° and 27.3°, similar to that of bulk g-C₃N₄-W reference, which are attributed to the in-planar packing and the stacking of the conjugated aromatic system, respectively.^2,20–22 No diffraction peaks belonging to melamine (XRD pattern of melamine is shown in Fig. S2 in ESI†) can be observed. This indicates that the polymerization reaction of melamine can occur at 470 °C to form graphitic carbon nitride, which can be further confirmed by FTIR spectra, as shown in Fig. 1b and S3.† The intensity of diffraction peak at 13.1° in XRD pattern of g-C₃N₄-470 is much higher than that of bulk g-C₃N₄-W reference, indicating the perfect g-C₃N₄ structure where the tri-s-triazine unit is connected through bridging nitrogen atoms, as shown in Fig. S4.† However, it is found that the XRD peak at 27.3° of g-C₃N₄-470, attributed to packing between the layers, is lower than that of bulk g-C₃N₄-W (27.4°), indicating the larger interlayer distances due to the low degree of condensation and packing between the layers at low heating temperature. With increasing the preparation temperature, the diffraction intensity at the lower angle (13.1°) decreases and the diffraction peak at higher angle (27.4°) shifts to the higher angle. This indicates that the increased heating temperature leads to seriously damage of the periodicity in the parallel direction and a denser packing between the conjugated aromatic systems. Fig. 1b shows the FTIR spectra of g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, g-C₃N₄-540, and g-C₃N₄-W reference. Similar with the g-C₃N₄-W reference, the FTIR spectra of g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540 samples display several characteristic bands belonging to the fingerprint region (1200–1650 cm⁻¹) of s-triazine ring (C–NH–C and C=N stretching vibrations), two bands caused by the N–H stretches at 3328 and 3168 cm⁻¹, and a band corresponding to the out of plane bending of the s-triazine ring at 806 cm⁻¹.^17,21 The FTIR spectra of g-C₃N₄-T samples are not changed along with the different preparation temperature. According to the XRD and FTIR results, we can conclude that the g-C₃N₄-T samples also have poly tri-s-triazine structure but with destroyed planar structure and denser interlayer distances.

Fig. 1 XRD patterns (a) and FTIR spectra (b) of g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540.

The XPS analyses results for the g-C₃N₄-T samples and the bulk g-C₃N₄-W reference are shown in Fig. 2. The C 1s signals (Fig. 2a) of all samples can be fitted into two peaks centered at around 284.6 and 288.0 eV, which can be attributed to the aromatic carbon atom and the sp² C=N bond in the s-triazine ring, respectively.^15,23–25 As shown in Fig. 2b, the N 1s signals in the binding energy region of 396–402 eV for these samples can be fitted into three peaks located at 398.5 eV, 399.9 eV, and 400.9 eV for sp²-hybridized nitrogen in triazine rings (C–N=C), tertiary nitrogen N–(C)₃ groups, amino functions caring hydrogen (C–N–H), respectively.^26,27 A weak peak of N 1s at 404.1 eV for all samples is attributed to charging effects.²⁶ Comparing to bulk g-C₃N₄-W sample, no obvious binding energy shift of C 1s and N 1s signals for g-C₃N₄-T can be observed, suggesting that the surface chemical states of both C and N in the g-C₃N₄-T samples are the same as in the bulk g-C₃N₄ reference. The binding energies of O 1s peaks at 531.8 and 533.4 eV of bulk g-C₃N₄-W sample (Fig. 2c) are ascribed to adsorbed oxygen species (water and oxygen).^19,28 For g-C₃N₄-T synthesized in an opened crucible, the peak at low binding energy gradually shifted to the higher binding energy along with higher preparation temperature (531.8, 531.9, 532.0, and 532.1 eV for g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540, respectively), which might be due to the oxidation of some carbon atoms into C=O species at high temperature under air atmosphere.²⁹ The percentages of elements (C and N) on the surface of various carbon nitride samples are determined by XPS. For g-C₃N₄-470, the surface atomic ratio of C/N is about 0.69, which is close to that of bulk g-C₃N₄-W (0.70). The surface C/N ratio of g-C₃N₄-T increases with increasing of temperature (0.73 for g-C₃N₄-500, 0.81 for g-C₃N₄-520, and 0.82 for g-C₃N₄-540), which suggests that the surface of g-C₃N₄-T obtained at high temperature is poor in nitrogen. This is might be due to the thermal oxidation at high temperature, which leads to exfoliation and tailoring of bulk structure of g-C₃N₄. These findings are in good consistence with the XRD, SEM, TEM, and N₂ adsorption results.

Fig. 2 High resolution XPS spectra of (a) C 1s, (b) N 1s, and (c) O 1s for bulk g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540.

The morphology and microstructure of the obtained carbon nitride are changed dramatically with increasing of the preparation temperature. As shown in Fig. 3, the SEM and TEM images of bulk g-C₃N₄ reference (Fig. 3a and f) and g-C₃N₄-470 obtained at lower heating temperature (Fig. 3b and g) represent typical monolith like morphology. When the temperature increased to 500 °C, the g-C₃N₄-500 became sheet-like (Fig. 3c and h), suggesting the exfoliation of bulk carbon nitride materials to form the two-dimensional nanosheets by direct heating melamine in air atmosphere. With reaction temperature further increased, nanotube type g-C₃N₄ is synthesized at 520 °C, which achieves a one-dimensional nanotube growth of g-C₃N₄ (shown in Fig. 3d and i). This indicates that the carbon nitride nanosheets can be rolled up to minimize the surface free energy.^16,30 Additionally, some destroyed tubes and fragments can be observed, indicating the further tailoring by the thermal etching, which can be confirmed by the fact that the tailored nanotubes, nanoflakelets, nanoparticles, even nanodots were formed when the calcination temperature rised up to 540 °C (Fig. 3e and j).

Fig. 3 FESEM and TEM images of bulk g-C₃N₄-W reference (a and f), g-C₃N₄-470 (b and g), g-C₃N₄-500 (c and h), g-C₃N₄-520 (d and i), and g-C₃N₄-540 (e and j), respectively.

N₂ adsorption at 77 K was used to further characterize the microstructural change of obtained carbon nitride with increasing the temperature. The N₂ adsorption–desorption isotherm and the corresponding BJH pore-size distribution curve of the g-C₃N₄-T samples and bulk g-C₃N₄-W reference are shown in Fig. 4. The obtained specific BET surface area, pore volume, and pore size of these samples are summarized in Table 1. The specific BET surface area of g-C₃N₄-470 sample is relatively low (6.0 cm² g⁻¹), which is highly comparable to that of bulk g-C₃N₄-W reference (4.5 cm² g⁻¹). Accordingly, there is no obvious hysteresis loop observed in the N₂ adsorption–desorption isotherm of g-C₃N₄-470 and bulk g-C₃N₄-W reference (Fig. 4), suggesting that no mesopores in the g-C₃N₄-470 and bulk g-C₃N₄-W reference, which can be confirmed by corresponding BJH pore-size distribution result (inset of Fig. 4). The N₂ adsorption–desorption isotherm of g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540 can be identified as type IV according to the IUPAC classification.^31,32 The obtained specific BET surface area are 41.5, 173.6, and 210.1 cm² g⁻¹, respectively, for g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540 samples, indicating the increased specific BET surface area. For g-C₃N₄-500 sample obtained at 500 °C, the most probable distribution of pore is centered at around 3.5 nm (inset in Fig. 4), which can be attributed to the mesopores between exfoliated carbon nitride nanosheets. When the treatment temperature is further increased, the exfoliated carbon nitride nanosheets rolled up and then the formed nanotubes can be further tailored, therefore the pore distribution centered at 20 nm are observed in the BJH pore-size distribution curves of g-C₃N₄-520 and g-C₃N₄-540 samples, supporting the SEM and TEM observation. These results further confirm that increasing polymerization temperature leads to exfoliation and tailoring of laminar structure of g-C₃N₄. Most importantly, g-C₃N₄ material with mesopores and large surface area can be obtained by using this simple method even though the collection yields decrease at higher preparation temperature (as shown in Table 1). Therefore, according to the results obtained from the SEM, TEM, and N₂ adsorption, we can conclude that the increased temperature in our study gives rise to significant changes in the textural properties of carbon nitride.

Fig. 4 The N₂ adsorption–desorption isotherm and the corresponding BJH pore-size distribution curve (inset) of bulk g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540, respectively.

Table 1 Summary of the physicochemical characteristics of as-prepared samples

Samples	Yields	Absorption edge (nm)	Eg (eV)	ECB (V)	EVB (V)	Emission peak (nm)	SBET (cm^2 g^−1)	Pore volume (mL g^−1)	Pore size (nm)
g-C3N4-W	33.4%	464	2.79	−1.29	1.50	458	4.5	0.035	40.2
g-C3N4-470	40.7%	451	2.87	−1.26	1.61	448	6.0	0.02	35.2
g-C3N4-500	16.2%	454	2.85	−0.88	1.97	449	41.5	0.14	9.2
g-C3N4-520	1.70%	448	2.89	−1.03	1.86	439	173.6	0.77	15.6
g-C3N4-540	0.35%	447	2.93	−0.99	1.94	436	210.1	0.94	16.5

---

3.2. Effects of synthesis temperature on optical property and electronic structure

The optical property and electronic structure of samples were studied by the combined analysis of the electron paramagnetic resonance (EPR), UV-visible absorption spectra, Mott–Schottky, and photoluminescence (PL) spectra. It has been reported that the semiconductor structure of g-C₃N₄ originates from the formation of an extended π-conjugated system composed of s-triazine units.¹⁷ Therefore, the EPR was used to evaluate the semiconductor band structure of g-C₃N₄-T obtained at different preparation temperature. The melamine and melem do not show any observable EPR signal. As shown in Fig. 5a, the EPR spectrum of g-C₃N₄-470 shows a signal at g value of 2.003 for the delocalized electrons in the conduction band,^17,33 indicating the polymerization of melamine at 470 °C, which is in agreement with the XRD result (Fig. 1a). The g-C₃N₄-500 sample obtained at 500 °C presents stronger signal of at g value of 2.003 as compared to g-C₃N₄-470, suggesting that higher preparation temperature leads to higher polymerization extent.^20,34,35 Further increasing the preparation temperature led to decreased EPR signal intensity because the connected π-conjugated system can be destroyed by thermal exfoliation and tailoring at high temperature under air atmosphere (as shown in Fig. 3), and consequently, the concentration of delocalized electrons in g-C₃N₄ decreases. In comparison, the g-C₃N₄-W reference has very high polymerization extent without exfoliation, thus it exhibits stronger EPR signals than all g-C₃N₄-T samples.

Fig. 5 (a) EPR spectra, (b) UV-vis absorption spectra, (c) Tauc plot, (d) Mott–Schottky, (e) energy band diagram, and (f) PL spectra of bulk g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540.

Fig. 5b shows the UV-visible absorption spectra of bulk g-C₃N₄ reference and g-C₃N₄-T samples. The g-C₃N₄-470 sample prepared at lower temperature exhibits an absorption edge at 451 nm, which is shorter than that of bulk g-C₃N₄-W reference (464 nm). This might be due to the low degree of condensation and packing between the layers at low heating temperature.^14,20 Increased preparation temperature leads to the enhanced condensation and packing between the layers, and therefore the obtained g-C₃N₄-500 presents a longer adsorption edge (454 nm), but it is still shorter than that of bulk g-C₃N₄-W reference, which might be due to quantum size effect.^9,17,20,26 With increasing preparation temperature from 500 to 540 °C, the light absorption thresholds of obtained samples gradually shift from 454 to 447 nm (Fig. 5b and Table 1), which should also be attributed to quantum size effect.^9,17,20,26 Fig. 5c shows the Tauc plot as a function of light energy of all samples. According to Fig. 5b and c, the estimated bandgaps are 2.79, 2.87, 2.85, 2.89, and 2.93 eV, corresponding to bulk g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540, respectively, as shown in Table 1. Based on the Mott–Schottky plots (Fig. 5d) the conduction band potentials (E_CB) of bulk g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540 are estimated as −1.29 V, −1.26 V, −0.88 V, −1.03 V, and −0.99 V, respectively (Table 1). Accordingly, the band structures of these samples can be depicted as Fig. 5e.

The photoluminescence (PL) spectra of the samples were tested at room temperature with an excitation wavelength of 380 nm. As shown in Fig. 5f, the g-C₃N₄-470 shows a very strong emission peak centered at 448 nm, which is shorter than that of bulk g-C₃N₄-W reference (458 nm), suggesting the lower condensation and packing. This is consistent with the blue-shift of absorption band edge, as shown in Fig. 5b. In addition, the intensity of peak at 448 nm for g-C₃N₄-470 is larger than that for g-C₃N₄-W reference, suggesting the strong radiative transition of g-C₃N₄-470. When the preparation temperature increased to 500 °C, the obtained sample showed a red-shifted PL peak at 449 nm, but it is still shorter than that of bulk g-C₃N₄-W reference. The former is due to the enhanced condensation and packing between the layers caused by the increased temperature, and the latter is due to quantum size effect,^9,20 which can be verified by the fact that further increasing preparation temperature leads to the gradual blue-shift of PL peak from 449 to 436 nm (Fig. 5f and Table 1). Increasing preparation temperature from 470 °C to 520 °C leads to the decreased PL intensity, indicating the enhanced non-radiative recombination at trapping states or defects on carbon nitride structure due to the thermal exfoliation and tailoring (shown in SEM and TEM images). However, further increase to 540 °C results in an increase of PL intensity, which might be attributed to the PL of very small g-C₃N₄ nanostructures (Fig. 3e and j) with quantum size effect.^36–38 These results indicate that the fluorescence spectrum of carbon nitride strongly depends on their structure and texture. It is worth noting that the variation of PL intensity of samples is in good agreement with the photocatalytic activity results, suggesting that the structural defects and textural property of samples have great impacts on their photocatalytic performance as discussed in details below.

3.3. Effects of synthesis temperature on photocatalytic activity

The H₂ generation reaction was firstly carried out in an aqueous solution of TEOA (15 vol%) as the sacrificial agent under visible light (>420 nm) to evaluate the photocatalytic activities of all samples to study the effect of sample preparation temperature on photocatalytic activities. Prior to the photocatalytic test, all samples were loaded with Pt (3.0 wt%) nanoparticles by using a wetness impregnation method as reported in our previous works.^21,39 The TEM image of Pt-loaded carbon nitride is shown in Fig. S5.† Among all modified g-C₃N₄-T samples, g-C₃N₄-470 exhibits the lowest photocatalytic H₂ evolution activity with an H₂ evolution rate of ∼0.29 μmol h⁻¹, which is even lower than that of bulk g-C₃N₄ reference (0.80 μmol h⁻¹) due to the bulk structure with low condensation (Fig. 6a and b). A volcano-type histogram of H₂ evolution rate depending on the preparation temperature is observed in Fig. 6b, and the maximum H₂ evolution rate of 6.89 μmol h⁻¹ is achieved on the g-C₃N₄-520 sample prepared at 520 °C, which is nearly 24 times higher than that of g-C₃N₄-470. It is worth noting that the photocatalytic H₂ evolution rates of the all exfoliated and tailored samples (2.97 μmol h⁻¹, 6.89 μmol h⁻¹, and 3.79 μmol h⁻¹ for g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540, respectively) are higher than that of bulk g-C₃N₄ reference, suggesting that the thermal exfoliation and tailoring for carbon nitride are effective for improving the photocatalytic H₂ evolution activity.

Fig. 6 (a) Time resolved hydrogen evolution amount and (b) hydrogen generation rate, (c) time resolved oxygen evolution amount and (b) oxygen generation rate of bulk g-C₃N₄-W reference, g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540.

Then, we also evaluated the photocatalytic O₂ evolution of RuO_x clusters (2.0 wt%) loaded carbon nitride samples (Fig. S6, see ESI†) from aqueous solution with AgNO₃ as the sacrificial agent under visible light irradiation. The results are shown in Fig. 6c and d. All samples exhibit visible light activity for O₂ evolution, although the order of magnitude for O₂ evolution rates is lower than that for H₂ evolution (comparison between Fig. 6b and d). Similar to H₂ evolution, the O₂ evolution rates increase with increasing preparation temperature from 470 °C to 520 °C and reach a maximum at 520 °C, followed by a decrease when the temperature increases to 540 °C. The g-C₃N₄-470 sample with bulk structure also shows the lowest O₂ evolution rate (0.28 μmol h⁻¹). The exfoliated and tailored samples present the enhanced photocatalytic O₂ evolution activity (1.20 μmol h⁻¹, 1.24 μmol h⁻¹, and 0.72 μmol h⁻¹ for g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540, respectively) as compared to g-C₃N₄-W reference (0.32 μmol h⁻¹).

3.4. Factors affecting the photocatalytic activity

XRD, FTIR, XPS, SEM, TEM, and N₂ adsorption results showed that the polymerization conditions such as the atmosphere, vapour pressure, and polymerization temperature have great influences on textural and structural properties of graphitic carbon nitride. When the melamine precursor was sealed in a cylindrical alumina crucible, because of the lower concentration of oxygen, flowability of produced gases (such as NH₃, CO, and CO₂ etc.) during the polymerization process and vapor pressure of melamine, the g-C₃N₄ reference obtained at 520 °C with stone-like morphology has very low specific surface area, as shown in Fig. 3 and 4 and Table 1. If the precursor is placed on an opened combustion boat, the polymerization of melamine takes place in air atmosphere and thus the sample obtained at same temperature (520 °C) exhibits very different microstructure with tube like morphology, which might be due to the evaporation of precursor, flowing of produced gases, and thermal etching of oxygen, and thus has very large surface area (173.6 m² g⁻¹) as compared to the bulk g-C₃N₄ reference (4.5 m² g⁻¹). The photocatalytic activity (both H₂ evolution and O₂ evolution) of g-C₃N₄-520 sample is much better than that of the bulk g-C₃N₄-W reference obtained at same temperature. We also found that the temperature has great impact on the polymerization and thermal etching process. The morphology of samples obtained at different temperatures undergoes sequential evolution from bulk to nanosheet with multilayer, nanosheet with few layer, moderately rolled nanosheet, tailored nanotube, and even nanodots. Accordingly, the specific surface area of samples increases gradually with increasing of temperature. The S_BET of g-C₃N₄-470, g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540 samples are 6.0, 41.5, 173.6 and 210.1 cm² g⁻¹, respectively, as shown in Table 1. For g-C₃N₄-470, g-C₃N₄-500, and g-C₃N₄-520 samples, their photocatalytic activities increase with larger specific surface area (Fig. 6). Based upon the above analyses, we can conclude that the surface area is a very important factor influencing the photocatalytic activity of graphitic carbon nitride. The exfoliated and tailored structures can shorten the migration distance of photogenerated charge carriers, which is beneficial to reduce the chance of charge recombination. In addition, the larger surface area can also provide more surface active sites for reactions.

During the process of heat treatment, the layer structure can be gradually exfoliated and the conjugated poly-triazine ring structure is damaged to a certain extent (as shown in Fig. 3). Accordingly, the g-C₃N₄ surface structure will be changed, such as increased surface defects, which can be confirmed by PL results (Fig. 5f). The photocatalytic activity results showed that surface defects are helpful to improve the activity because the surface defects can capture the photogenerated carriers, inhibiting the radiative recombination (as shown in PL spectra). But redundant defects can act as recombination centers of photogenerated charge carriers, promoting charge recombination and thus reducing the photocatalytic activity. This can be proved by the fact that the g-C₃N₄-540 sample exhibits reduced photocatalytic activities.

The different textural and structural properties result in the different optical and electronic characterization (EPR, UV-vis absorption, Mott–Schottky, and PL), as shown in Fig. 5. Light absorption of semiconductor photocatalyst is also one of the important factors affecting the photocatalytic activity. Generally, the longer optical absorption edge allows more light be absorbed by the materials and thus leads to the enhanced photocatalytic activity. Because of the low polymerization degree, g-C₃N₄-470 sample shows the narrow light absorption range as compared to the bulk g-C₃N₄-W reference, which is also one of the reasons for its low activity. As shown in SEM and TEM images (Fig. 3), the size and thickness of g-C₃N₄ gradually decrease as a result of the thermal cutting and stripping effect. Therefore, the optical absorption edge of g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540 samples shift to short wavelength as compared to g-C₃N₄-W reference and further gradually blue shift to with increasing of temperature (Fig. 5b and c) due to the quantum effect. The activity results indicate that their activities displayed a volcano-type of H₂ and O₂ evolution rates instead of linear decrease. But the photocatalytic activities of each sample (g-C₃N₄-500, g-C₃N₄-520, and g-C₃N₄-540) are still much higher than that of bulk C₃N₄-W reference. Hence one can see that only when the temperature is increased up to 540 °C, effects of light absorption on the photocatalytic activity predominated. So, the reasons for decreased photocatalytic activity of g-C₃N₄-540 sample with the largest specific surface area are attributable to the too many surface defects as the recombination centers and the blue shift of absorption edge.

4. Conclusions

In summary, the modified g-C₃N₄ photocatalysts with exfoliated and tailored structure are successfully obtained by using air-assisted thermal polymerization and etching. The effects of calcination temperature on exfoliation process of g-C₃N₄, morphology evolutions, optical properties, band-gap engineering, and photocatalytic activity for water splitting are systemically studied and clarified. The morphology of samples obtained at different temperatures undergoes sequential evolution from bulk to nanosheet with multilayer, nanosheet with few layer, moderately rolled nanosheet, tailored nanotube, and nanoflakelet even nanoparticles. The samples prepared at different temperatures also have different surface structure, optical properties, and energy band structures. The exfoliated and tailored samples obtained by this method present better photocatalytic activities as compared to that of bulk g-C₃N₄. On the basis of the textural and structural characterization as well as the photocatalytic activity results, we can conclude that the surface area, surface defects, light absorption have great effects on the photocatalytic activity of g-C₃N₄.

Acknowledgements

This work was financially supported by NTU seed funding for Solar Fuels Laboratory, MOE AcRF-Tier1 (RG 44/11), MOE AcRF-Tier2 (MOE2012-T2-2-041, ARC 5/13), and CRP (NRF-CRP5-2009-04) from NRF Singapore.

Notes and references

1. Q. Xiang, J. Yu and M. Jaroniec, Chem. Soc. Rev., 2012, 41, 782.
2. X. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M. Carlsson, K. Domen and M. Antonietti, Nat. Mater., 2009, 8, 76.
3. Y. P. Yuan, L. S. Yin, S. W. Cao, L. N. Gu, G. S. Xu, P. W. Du, H. Chai, Y. S. Liao and C. Xue, Green Chem., 2014, 16, 4663.
4. J. Xu, L. Luo, G. Xiao, Z. Zhang, H. Lin, X. Wang and J. Long, ACS Catal., 2014, 4, 3302.
5. J. Liu, Y. Liu, N. Liu, Y. Han, X. Zhang, H. Huang, Y. Lifshitz, S.-T. Lee, J. Zhong and Z. Kang, Science, 2015, 347, 970.
6. D. A. Dikin, S. Stankovich, E. J. Zimney, R. D. Piner, G. H. B. Dommett, G. Evmenenko, S. T. Nguyen and R. S. Ruoff, Nature, 2007, 448, 457.
7. J. Shen, Y. Hu, M. Shi, X. Lu, C. Qin, C. Li and M. Ye, Chem. Mater., 2009, 21, 3514.
8. W. Zhao, M. Fang, F. Wu, H. Wu, L. Wang and G. Chen, J. Mater. Chem., 2010, 20, 5817.
9. X. Zhang, X. Xie, H. Wang, J. Zhang, B. Pan and Y. Xie, J. Am. Chem. Soc., 2013, 135, 18.
10. K. Schwinghammer, M. B. Mesch, V. Duppel, C. Ziegler, J. Senker and B. V. Lotsch, J. Am. Chem. Soc., 2014, 136, 1730.
11. X. She, H. Xu, Y. Xu, J. Yan, J. Xia, L. Xu, Y. Song, Y. Jiang, Q. Zhang and H. Li, J. Mater. Chem. A, 2014, 2, 2563.
12. S. Yang, Y. Gong, J. Zhang, L. Zhan, L. Ma, Z. Fang, R. Vajtai, X. Wang and P. M. Ajayan, Adv. Mater., 2013, 25, 2452.
13. T. Sano, S. Tsutsui, K. Koike, T. Hirakawa, Y. Teramoto, N. Negishi and K. Takeuchi, J. Mater. Chem. A, 2013, 1, 6489.
14. Y. Zhang, A. Thomas, M. Antonietti and X. Wang, J. Am. Chem. Soc., 2009, 131, 50.
15. P. Niu, L. Zhang, G. Liu and H.-M. Cheng, Adv. Funct. Mater., 2012, 22, 4763.
16. S. Wang, C. Li, T. Wang, P. Zhang, A. Li and J. Gong, J. Mater. Chem. A, 2014, 2, 2885.
17. J. Sun, J. Zhang, M. Zhang, M. Antonietti, X. Fu and X. Wang, Nat. Commun., 2012, 1139.
18. P. Niu, L.-C. Yin, Y.-Q. Yang, G. Liu and H.-M. Cheng, Adv. Mater., 2014, 26, 8046.
19. J. Li, B. Shen, Z. Hong, B. Lin, B. Gao and Y. Chen, Chem. Commun., 2012, 48, 12017.
20. Y. Wang, X. Wang and M. Antonietti, Angew. Chem., Int. Ed., 2012, 51, 68.
21. Q. Gu, Y. Liao, L. Yin, J. Long, X. Wang and C. Xue, Appl. Catal., B, 2015, 165, 503.
22. K. Maeda, R. Kuriki, M. Zhang, X. Wang and O. Ishitani, J. Mater. Chem. A, 2014, 2, 15146.
23. Y. P. Yuan, W. T. Xu, L. S. Yin, S. W. Cao, Y. S. Liao, Y. Q. Tng and C. Xue, Int. J. Hydrogen Energy, 2013, 38, 13159.
24. J. Ding, Q. Liu, Z. Zhang, X. Liu, J. Zhao, S. Cheng, B. Zong and W.-L. Dai, Appl. Catal., B, 2015, 165, 511.
25. L. S. Yin, Y. P. Yuan, S. W. Cao, Z. Y. Zhang and C. Xue, RSC Adv., 2014, 4, 6127.
26. J. Zhang, M. Zhang, C. Yang and X. Wang, Adv. Mater., 2014, 26, 4121.
27. S. W. Cao, X. F. Liu, Y. P. Yuan, Z. Y. Zhang, J. Fang, S. C. J. Loo, J. Barber, T. C. Sum and C. Xue, Phys. Chem. Chem. Phys., 2013, 15, 18363.
28. D. Wei, Y. Liu, Y. Wang, H. Zhang, L. Huang and G. Yu, Nano Lett., 2009, 9, 1752.
29. F. Xiao, J. Miao and B. Liu, J. Am. Chem. Soc., 2014, 136, 1559.
30. P. Cendula, S. Kiravittaya, I. Mönch, J. Schumann and O. G. Schmidt, Nano Lett., 2011, 11, 236.
31. C. Yuan, X. Zhang, L. Su, B. Gao and L. Shen, J. Mater. Chem., 2009, 19, 5772.
32. A. Vinu, Adv. Funct. Mater., 2008, 18, 816.
33. Y. Zhang, Z. Schnepp, J. Cao, S. Ouyang, Y. Li, J. Ye and S. Liu, Sci. Rep., 2013, 3, 2163,  DOI:10.1038/srep02163.
34. V. W.-H. Lau, M. B. Mesch, V. Duppel, V. Blum, J. Senker and B. V. Lotsch, J. Am. Chem. Soc., 2015, 137, 1064.
35. M. J. Bojdys, J.-O. Müller, M. Antonietti and A. Thomas, Chem.–Eur. J., 2008, 14, 8177.
36. J. Zhou, Y. Yang and C.-Y. Zhang, Chem. Commun., 2013, 49, 8605.
37. S. Barman and M. Sadhukhan, J. Mater. Chem., 2012, 22, 21832.
38. S. Liu, J. Tian, L. Wang, Y. Luo and X. Sun, RSC Adv., 2012, 2, 411.
39. Q. Gu, J. Long, H. Zhuang, C. Zhang, Y. Zhou and X. Wang, Phys. Chem. Chem. Phys., 2014, 16, 12521.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra07284k

---