Thermal Exfoliation and Phosphorus Doping in Graphitic Carbon Nitride for Efficient Photocatalytic Hydrogen Production

Abstract

Photocatalytic H₂ evolution has been regarded as a promising technology to alleviate the energy crisis. Designing graphitic carbon nitride materials with a large surface area, short diffusion paths for electrons, and more exposed reactive sites are beneficial for hydrogen evolution. In this study, a facile method was proposed to dope P into a graphitic carbon nitride framework by calcining melamine with 2-aminoethylphosphonic acid. Meanwhile, PCN nanosheets (PCNSs) were obtained through a thermal exfoliation strategy. Under visible light, the PCNS sample displayed a hydrogen evolution rate of 700 μmol·g⁻¹·h⁻¹, which was 43.8-fold higher than that of pure g-C₃N₄. In addition, the PCNS photocatalyst also displayed good photostability for four consecutive cycles, with a total reaction time of 12 h. Its outstanding photocatalytic performance was attributed to the higher surface area exposing more reactive sites and the enlarged band edge for photoreduction potentials. This work provides a facile strategy to regulate catalytic structures, which may attract great research interest in the field of catalysis.

1. Introduction

Solar energy is regarded as the most promising candidate for renewable green energy to avoid environmental pollution and address the global severe energy crisis [1,2]. As an attractive strategy to utilize solar energy, photocatalytic water splitting has been considered a sustainable method for generating clean hydrogen [3,4]. To date, graphitic carbon nitride has been regarded as a promising semiconductor photocatalytic material in the field of photocatalysis. Importantly, graphitic carbon nitride possesses an excellent electronic structure, outstanding physiochemical properties, and a suitable band gap (2.7 eV) [5,6]. Therefore, graphitic carbon nitride has gradually become a competitive nano-semiconductor material for hydrogen energy production.

Nevertheless, pristine g-C₃N₄ suffers from several drawbacks, including poor electric conductivity, a low specific surface area, and the rapid recombination of photogenerated electron–hole pairs [7]. These drawbacks result in the inferior photocatalytic activity of g-C₃N₄. To address these drawbacks, several kinds of strategies have been developed to engineer the chemical composition and structure of g-C₃N₄, for example, cocatalyst loading [8,9], elemental doping [10,11], controlling its morphology [12,13], the construction of heterojunctions with other semiconductors [14,15], molecule incorporation [16], defect engineering [17,18], and nanostructure design [19]. Among these, element doping offers an effective strategy to regulate its electronic structure and extend its light absorption, further enhancing the photocatalytic activity of g-C₃N₄. Currently, non-metal elements (Br, B, S, O, and P) with different electronegativities can be doped into the g-C₃N₄ framework to tune the band gap structure, further accelerating the photogenerated carriers’ separation [20]. Phosphorus is the most earth-abundant non-metal element rich in electrons that can serve as an electron donor. Chen et al. reported doping P into a g-C₃N₄ framework to form P–N bands, which can accelerate the charge separation and transfer efficiency and further improve the hydrogen evolution [21]. P doping into g-C₃N₄ (PCN) has achieved remarkable enhancements of its photocatalytic activities, but it still fails to satisfy the practical requirement of synthesis tuning the band gap structure of the photocatalyst. Therefore, it is urgent to seek a new method to prepare a large-surface area photocatalyst to create more reactive sites and further enhance the photocatalytic activity of g-C₃N₄.

A two-dimensional (2D) g-C₃N₄ nanosheet exhibits a large specific surface area, more reactive sites, a short charge transfer distance, and quantum effects. Various strategies have been developed to prepare large-specific surface area g-C₃N₄ nanosheets, such as water steam exfoliation, sonification exfoliation, and ball milling. However, these strategies use solvents for the exfoliation process, which is time-consuming and inefficient. Qiu et al. reported a thermal exfoliation method to prepare single- or few-layered nanosheets with strengthened surfaces and semiconductor properties. Therefore, combining element doping and a thermal exfoliation method to prepare g-C₃N₄ with a large specific surface area improves its photocatalytic performance.

In this work, we successfully developed a green and facile method to prepare PCNSs with a large specific surface area. As expected, thermal exfoliation could effectively exfoliate the bulk PCN into few-layered nanosheets. The PCNS sample exhibited the highest photocatalytic hydrogen evolution rate of 700 μmol·g⁻¹·h⁻¹, which was 43.8-fold higher than that of g-C₃N₄. Its apparent quantum efficiency reached 0.49% under visible light of 420 nm. Meanwhile, the photocatalyst was subjected to four successive cycles and exhibited excellent photostability. This outstanding photocatalytic performance was attributed to its large surface area, which could afford more reactive sites for hydrogen evolution. Meanwhile, the introduction of P–N bonds could accelerate the charge carrier separation and transfer efficiency, leading to more efficient photocatalytic hydrogen production. This work paves a new way to construct high-surface area PCNSs by integrating element doping and thermal exfoliation.

2. Results

The synthesized materials were investigated by X-ray diffraction (XRD) to measure their crystal phases and compositions. As shown in Figure 1a, the diffraction peaks located at 13° and 27° were ascribed to the (100) and (002) crystal planes [22], respectively. The former (100) plane was ascribed to the in-plane structure stacking pattern [23], while the latter (002) plane corresponded to the interlayer stacking [24,25]. Through the thermal exfoliation reaction process, the CNS photocatalyst maintained the original structure of the bulk g-C₃N₄ (Figure 1b). Compared with g-C₃N₄, the intensities of the two peaks of PCN were reduced. The diffraction peak in the (002) plane was slightly shifted to a small angle (Figure 1c). The reduction in the diffraction angle revealed the increased (002) interplane distance, which was ascribed to the radius of P being much bigger than that of C or N. As presented in Figure 1d, the PCNS sample maintained the original structure of the bulk g-C₃N₄. Meanwhile, the characteristic peak intensity became weaker and broader.

The morphologies and microscopic structures of all the samples were examined by SEM and TEM. As shown in Figure 2a, the pristine g-C₃N₄ consisted of two-dimensional nanosheets with curling edges, which was attributed to the multilayer structure of the graphitic properties of carbon nitride. As presented in Figure S1a, the CNS maintained the two-dimensional nanosheet structure after the thermal exfoliation reaction process. After doping with the P element, the SEM image of PCN exhibited nearly no changes, revealing that the structure remained intact after the elemental doping (Figure S1b). Moreover, the PCNS sample still retained the two-dimensional structure after the thermal exfoliation reaction process (Figure 2b). In addition, the PCNS’s specific surface area reached 76.99 m²·g⁻¹, which was 8, 1.35, and 9.4 times higher than that of PCN, CNS, and g-C₃N₄ (Figure S2). The above results further confirm that thermal exfoliation can enlarge the surface area, afford more reactive sites, and reduce the carriers’ diffusion distance, further improving the photocatalytic performance of g-C₃N₄.

The chemical states and compositions of g-C₃N₄ and the PCNS were studied by XPS. The XPS survey spectra in Figure 3a reveal that the two samples contained C and N elements. Moreover, the element of P was observed in the PCNS. As presented in Figure 3b, it can be observed that g-C₃N₄ exhibited three typical C 1s peaks located at 284.78, 286.38, and 288.08 eV, attributed to the physically absorbed carbon species or sp² C–C bonds, C–NH₂ species, and sp²–bond carbon (N–C=N) in the g-C₃N₄ aromatic ring [26]. The high-resolution spectrum of N 1s in g-C₃N₄ was divided into three peaks located at 398.48, 400.38, and 404.28, eV. The peak centered at 398.48 eV was attributed to the sp²-hybridized nitrogen (C–N=C group) [27,28]. The peak located at 400.38 eV was assigned to the tertiary N (C₃–N or C₂–N–H) [29,30]. The peak at 404.28 eV was related to the amino functional group (C–N–H) [8,31,32,33]. Compared with pristine g-C₃N₄, the C 1s and N 1s spectra of the PCNS were downshifted to low binding energies, which was ascribed to the P-doped change in the surface charge distribution. In addition, the characteristic peak of P 2p at 133.38 eV corresponded to the P–N bond in the PCNS [34,35,36,37].

To gain insight into their molecular structures, all the photocatalysts were investigated by FTIR. As presented in Figure 4, the peak located at 810 cm⁻¹ was assigned to the out-of-plane bending mode of the heptazine units [38]. The peak in the range of 880–1640 cm⁻¹ could correspond to the N–C=N heteroaromatic rings [39,40,41]. A characteristic peak for the aromatic C–N and C=N stretching vibrational model was shown at 1200–1700 cm⁻¹ [42]. The peak centered at 3160–3440 cm⁻¹ was attributed to the N–H stretching vibrations [43,44,45]. After the P-doping and thermal exfoliation reaction processes, the CNS, PCN, and PCNS photocatalysts maintained the original molecular structure of g-C₃N₄.

The optical properties of all samples were measured by UV–vis absorption spectroscopy. As displayed in Figure 5a, the absorption edge of g-C₃N₄ was mainly located at 460 nm. In comparison, the absorption of CNS exhibited an obvious blueshift, which was ascribed to the quantum size effect. After P doping, the PCN sample absorption band edge had a minor redshift, expanding to the visible-light region. After the thermal exfoliation reaction process, the PCNS featured a blueshift, which corresponded to the quantum size effect. In addition, according to the Tauc method, the band gaps of g-C₃N₄ and the PCNS were 2.72 and 2.75 eV, respectively. As presented in Figure 5c, the VB potentials of g-C₃N₄ and the PCNS were 2.05 and 1.95 eV, respectively. Thus, the VB values of g-C₃N₄ and the PCNS were calculated to be 2.15 and 2.05 eV, respectively, according to the following equation:

E_VB-NHE = Ψ + E_VB-XPS − 4.44

(1)

Ψ: the electron work function of the XPS analyzer; E_VB-XPS: the VB value tested by the VB-XPS plots; and E_VB-NHE: the standard hydrogen electrode potential. Then, the valence band position was measured by the following equation:

E_CB = E_VB − Eg

(2)

The results exhibit that the E_VB values of g-C₃N₄ and the PCNS were −0.57 and −0.7 eV, respectively. The energy band positions of g-C₃N₄ and the PCNS are schematically displayed in Figure 5d. The shifted CB and VB position of the PCNS led to a larger thermodynamic driving force for photocatalytic redox reactions.

Photoelectrochemical analysis was conducted to gain further insight into the charge separation and transfer efficiency. Figure 6a presents the periodic on/off photocurrent responses of the CN and PCNS electrodes at 0.2 V (vs. Ag/AgCl) with a 1.0 M Na₂SO₄ electrolyte under visible-light illumination (λ > 420 nm). The photocurrent density of the PCNS was higher than that of CN, indicating the fast separation and transfer efficiency of the photogenerated electron–hole pairs after the P-doping and thermal exfoliation strategies.

Meanwhile, the electrochemical impedance spectra of the CN and PCNS electrodes using 10 mM K₃[Fe(CN)₆]/K₄[Fe(CN)₆] as the electrolyte in the dark are shown in Figure 6b. The PCNS shows a smaller arc radius than pristine CN, revealing its lower charge transfer resistance and fast charge separation (Figure 6b). Subsequently, the CN sample shows a strong PL emission peak intensity. The PL intensity of the PCNS significantly decreased, confirming that the recombination of photogenerated electron–hole pairs was effectively suppressed (Figure S3). Based on the above results, the P-doping and thermal exfoliation reaction processes can promote the separation and transfer efficiency of photogenerated hole–electron pairs.

The photocatalytic activity of all samples was tested under visible-light irradiation using 100 mL of an aqueous solution containing 10% TEOA as a sacrificial agent. No hydrogen gas was detected without irradiation or a photocatalyst. As depicted in Figure 7a, the hydrogen evolution rate of pristine g-C₃N₄ was negligible (16 μmol·g⁻¹·h⁻¹). Then, after the thermal exfoliation reaction process, the CNS exhibited a high value of 55 μmol·g⁻¹·h⁻¹. After P doping, PCN showed that its photocatalytic performance was 81.4 μmol·g⁻¹·h⁻¹, which was 5.1 and 1.5 times higher than that of g-C₃N₄ and CNS. Meanwhile, the PCNS exhibited a high photocatalytic activity of 700 μmol·g⁻¹·h⁻¹, which was 43.8 times higher than that of CN. To further prove that the P-doping and thermal exfoliation reaction processes can enhance the photocatalytic performance of g-C₃N₄, comparisons between the photocatalytic abilities of the photocatalysts in the related reports and the as-prepared photocatalysts are listed in Table S1. Our work exhibited an improved H₂ evolution rate compared with other reported photocatalysts. The remarkable photocatalytic performance of the PCNS was ascribed to its higher surface area affording more reactive sites. In addition, the AQE reached 0.49% under 420 nm light irradiation for the PCNS. Photostability is an important parameter for a photocatalyst’s application. As depicted in Figure 7b, the H₂ evolution rate almost kept the same value during the four-cycle reaction, revealing excellent photocatalytic stability. Moreover, the XRD and FTIR patterns of the PCNS before and after the reaction did not change, confirming the favorable stability of the PCNS (Figure 7c,d).

3. Materials and Methods

3.1. Materials

Melamine, H₂PtCl₆·6H₂O (37.5 wt% Pt), and 2-aminoethylphosphonic acid (AEP) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China).

3.2. Preparation of g-C₃N₄

An amount of 10 g of melamine was placed into an alumina crucible with a cover. Then, the powder was calcined at 550 °C for 4 h with a ramp rate of 5 °C/min in a muffle furnace under an Ar atmosphere (50 mL/min). After cooling to room temperature, the obtained sample was denoted as g-C₃N₄ (CN).

3.3. Preparation of PCN

In a typical process, 10 g of melamine was added to 30 mL of deionized water. Next, 0.28 g of 2-aminoethylphosphonic acid was added to the mixed solution. Then, the mixed solution was kept at 100 °C under magnetic stirring at 100 rpm overnight. Finally, the obtained powder was calcined at 550 °C for 4 h under an Ar atmosphere (50 mL/min). The obtained brown powder was denoted as PCN.

3.4. Preparation of PCNS

An amount of 600 mg of PCN was placed into a porcelain boat and then sent into a tube furnace and calcined at 550 °C for 4 h under an Ar atmosphere (50 mL/min). After cooling to room temperature, the obtained powder was labeled as PCNS. The CNS sample was obtained through the same method.

3.5. Characterizations

The crystal structures and phase compositions of the photocatalysts were recorded by a Bruker XRD advance X-ray diffractometer (Salbruken, Germany) system using a Cu Kα X-ray source (λ = 0.15406 nm). The morphologies and microstructures were characterized by SEM (FE-SEM, JSM-6701F, JEOL, Tokyo, Japan ) and TEM (Tecnai Model G2 F20 S-TWIN, Peabody, MA, USA). The chemical states of the elements were analyzed by XPS (XPS, VG ESCALAB 250, Thermo Fisher Scientific, East. Grinstead, UK) with 150 W Al Ka X-ray radiation. C 1s (284.8 eV) was calibrated as the binding energy. The UV–vis diffusion spectra of all the photocatalysts were obtained with a Cary500 spectrophotometer with BaSO₄ powder as the reflectance standard in the range of 800–4000 cm⁻¹. The photoluminescence spectra (PL) were measured with a fluorophotometer (Edinburgh FL/FS900, Livingston, Scotland, UK) with an excitation wavelength of 400 nm. The functional groups of all samples were characterized by the FTIR spectra using a Nicolet 670 (Salbruken, Germany). The nitrogen adsorption–desorption isotherms were obtained using a nitrogen adsorber (ASAP 2020, Norcross, GA, USA). The photoelectrochemical values were measured in a standard three-electrode system using a CHI660E electrochemical workstation (CHI-660, Shanghai, China). A Pt wire and Ag/AgCl were the counter electrode and reference electrode. The working electrode was prepared by dropping 10 uL of a 5 mg/mL photocatalyst suspension onto the conductive surface of ITO glass and then dried in air. The transient photocurrent response was tested in a 1.0 M Na₂SO₄ aqueous solution irradiated by a 300 W xenon lamp. The electrochemical impedance spectra (EIS) were obtained over a frequency range from 0.01 to 10⁵ at an applied potential of 0.2 V using a 10 mM K₃[Fe(CN)₆]/K₄[Fe(CN)₆] aqueous solution.

3.6. Evaluation of Photocatalytic Activity

Photocatalytic hydrogen production was performed in a vacuum-closed gas circulation system (Lab-solar 6A, perfectlight, Beijing, China). In a typical photocatalytic process, 10 mg of each photocatalyst was dispersed into 100 mL of an aqueous solution containing 10% TEOA as a sacrificial agent. Pt of 3 wt% (theoretical amount) was loaded onto the photocatalysts as a cocatalyst by in situ photo-deposition. Before illumination, the reaction system was evacuated for 30 min to remove air, and the hydrogen yield was measured by online gas chromatography (PANNA, A91, Changzhou, China) using Ar as a carrier gas.

The apparent quantum efficiency (AQE) was measured under the same conditions. The light irradiation area was 19 cm². The amount of hydrogen evolution was obtained using a 300 W Xe lamp (CEL-HXF300-T3, Beijing, China) as a light source and a 420 nm band-pass filter to allow the corresponding wavelength photons to pass through. The AQE was calculated by the following equation:

$$\mathrm{AQE} (\%)={\displaystyle \frac{\mathrm{Number} \mathrm{of} \mathrm{reacted} \mathrm{electrons}}{\mathrm{Number} \mathrm{of} \mathrm{incident} \mathrm{photons}}} \times 100\%={\displaystyle \frac{{\mathrm{Number} \mathrm{of} \mathrm{evolved} \mathrm{H}}_2 \mathrm{molecules} \times 2}{\mathrm{Number} \mathrm{of} \mathrm{incident} \mathrm{photons}}}$$

(3)

4. Conclusions

In conclusion, we designed a PCNS photocatalyst with a large specific surface area through P-doping and thermal exfoliation reaction processes. The hydrogen evolution rate of 700 μmol·g⁻¹·h⁻¹ of the PCNS was 43.8 times higher than that of g-C₃N₄. Its apparent quantum efficiency reached up to 0.49%. The PCNS displayed excellent cycling and hydrogen evolution stability. Its outstanding photocatalytic performance was ascribed to its thin 2D structure and large surface area affording more reactive sites, further promoting the photogenerated hole–electron pairs’ separation and transfer. This work provides a facile and effective method to prepare PCNSs with excellent photocatalytic activity for hydrogen evolution, and this method confirms the potential of using non-metal doping and thermal exfoliation for the future optimization of high-performance solar-driven water-splitting catalysts.