Improving Photocatalytic Hydrogen Production over Pd Nanoparticles Decorated with g-C₃N₅ Photocatalyst

Abstract

A uniform distribution of Pd nanoparticles on g-C₃N₅ layers is achieved through wet-chemical reduction (Pd/g-C₃N₅-EG). Pd/g-C₃N₅-EG demonstrates superior hydrogen production yield compared to heterojunctions synthesized via photoreduction (Pd/g-C₃N₅-PR), attributed to improved metal–support interaction. Pd/g-C₃N₅-EG demonstrates an average hydrogen evolution rate of 891.304 µmol g⁻¹ h⁻¹, representing a 7.8-fold increase over Pd/g-C₃N₅-PR and a 34.5-fold increase over bare g-C₃N₅. Pd/g-C₃N₅-EG also maintains its hydrogen production capability after five recycling cycles, with only a slight decrease in efficiency.

1. Introduction

Nitrogen-rich carbon nitride g-C₃N₅ recently received much attention as a newly emerged photocatalyst. In comparison to g-C₃N₄, g-C₃N₅ exhibits a narrower band gap energy and a higher potential of having a conductive band (CB) due to the nitrogen-rich matrix. In particular, the presence of a triazole unit with N–N coupling within the g-C₃N₅ framework enhances its ability to absorb light, extending its absorption range into the visible spectrum—substantially broader than that of g-C₃N₄ [1,2]. As a result, g-C₃N₅ has been increasingly considered as a viable alternative to g-C₃N₄ in various applications, including photocatalytic degradation of organic compounds [3] and H₂ production [4]. However, similar to single-component photocatalysts, sluggish charge transfer may hinder its broader application. The synthesis of composites containing g-C₃N₅ and metallic/bimetallic nanoparticles could provide a versatile approach to enhancing the H₂ evolution of g-C₃N₅ [5]. Under irradiation, electrons in the valence band (VB) of g-C₃N₅ undergo excitation, moving to the CB of g-C₃N₅ and leaving behind a hole in the VB of g-C₃N₅. The efficient electron/hole separation is preferably achieved when metallic/bimetallic nanoparticles are anchored on a g-C₃N₅ server as electron acceptors. Electron acceptors play a crucial role in capturing electrons from the CB of g-C₃N₅, thereby initiating the hydrogen evolution reaction on the conduction band of g-C₃N₅ layers. During this process, protons (H⁺) react with captured electrons to produce H₂, attributed to the more positive CB value of g-C₃N₅ in comparison to the standard reduction potential (H⁺/H₂).

The distribution and loading of metallic/bimetallic nanoparticles on the g-C₃N₅ layers has been achieved through various methods, such as wet-chemical reduction, photochemical reduction, and high-temperature reduction under an inert atmosphere. Among these, the wet-chemical reduction method using organic capping ligands enables the controllable size and uniform distribution of metallic and bimetallic nanoparticles on g-C₃N₅ layers, which is expected to significantly enhance photocatalytic activity compared to other synthesis methods [6,7].

In our approach, we employ wet-chemical reduction using ethylene glycol (EG) acting as both a capping and reduction agent for loading Pd° nanoparticles (NPs) on the g-C₃N₅ matrix, ensuring a better metal–support interaction. For comparison, we also synthesized Pd/g-C₃N₅ via the photoreduction method. In addition, the Pd/g-C₃N₅ composite, prepared according to these methods and employing PdCl₂ as the Pd source, may be more cost-effective than using Pt. This approach avoids the introduction of additional impurities, allows for easy control of the reaction conditions, maintains a low production cost, and facilitates straightforward large-scale production.

2. Materials and Methods

2.1. Materials

All the chemicals were used without further purification. 3-amino-1,2,4-triazole (C₂H₄N₄, 99%), palladium (II) chloride (PdCl₂, 99%), and Nafion (5 wt.% in lower aliphatic alcohols and water, contains 45% water) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ethylene glycol (EG, C₂H₆O₂, 99%), methanol (CH₄O, 99%), ethanol (C₂H₆O, 99%), potassium ferricyanide (K₃Fe(CN)₆, 99%), sodium sulfate (Na₂SO₄, 99%), potassium cyanide (KCN, 98%), and triethanolamine (TEOA, C₆H₁₅NO₃, 99%) were purchased from Junsei (Tolyo, Japan).

2.2. Synthesis of g-C₃N₅ Photocatalyst by a Thermal Condensation Method

The g-C₃N₅ photocatalyst was fabricated using a thermal condensation method with 3-amino-1,2,4-triazole as the starting material [8]. Specifically, 3 g of 3-amino-1,2,4-triazole powder was thoroughly ground into an aluminum crucible with a lid. The crucible was then heated to 550 °C in a muffle furnace for 4 h under ambient conditions, with a heating rate of 5 °C min⁻¹. The as-obtained sample was ground into a fine powder, and after dispersing deionized water, it was stirred continuously for 2 h. The final g-C₃N₅ sample was collected by centrifuge and dried under vacuum at 60 °C overnight.

2.3. Synthesis of Pd/g-C₃N₅ Composite

The Pd/g-C₃N₅ heterojunction was synthesized by the wet-chemical method, referred to as Pd/g-C₃N₅-EG. In a typical procedure, 300 mg of g-C₃N₅ was dissolved in 100 mL of EG, sonicated for 30 min, and then placed in the ice-water bath with continuous stirring. After stirring for 30 min, 0.6 mL of PdCl₂ solution (2.5 mg Pd/mL) was added dropwise into the g-C₃N₅-EG suspension, followed by additional stirring for 2 h. The Pd content was set at 0.5 wt.%. Then, the solution was transferred into a flask and heated in an oil bath at 140 °C for 2 h. After cooling to room temperature, the Pd/g-C₃N₅-EG solid was collected by centrifugation, washed with deionized water and ethanol at least three times, and finally dried under vacuum overnight at 60 °C.

For comparison, the Pd/g-C₃N₅ composite was also synthesized via a photoreduction method, denoted as Pd/g-C₃N₅-PR. Firstly, 300 mg of g-C₃N₅ was suspended, dispersed in 100 mL of 20% (v/v) methanol solution, and sonicated for 30 min to form a suspension. Then, the mixture was transferred into the double-layer beaker, and 0.6 mL of PdCl₂ solution (2.5 mg mL⁻¹) was added dropwise to the suspension. The Pd content was set at 0.5 wt.%. The suspension was stirred for the next 2 h before it was illuminated under a 300 W Xenon lamp with a 420 nm filter cutoff for 2 h. The Pd/g-C₃N₅-PR product was collected by centrifuge, rinsed with ethanol and DI water several times, and dried at 60 °C overnight.

2.4. Characterization

The crystal structure of photocatalysts was analyzed using X-ray diffraction (XRD) spectra from an X’Pert3-Powder X-Ray Diffractometer (PANalytical, Malvern, UK) with a 2θ range of 10–80°. The morphologies and structures of as-prepared catalysts were investigated by field emission transmission electron microscopy (FE-TEM) and energy-dispersive spectroscopy (EDS, Tokyo, Japan) analysis using JEM-F200 (JEOL, Tokyo, Japan). UV-Vis diffuse reflection spectra (UV-Vis DRS) of g-C₃N₅ and Pd/g-C₃N₅ photocatalysts were recorded using Shimadzu UV-Vis spectroscopy (UV-2600, Tokyo, Japan). The photoluminescence (PL) spectra of the as-prepared samples were obtained at room temperature using a photofluorescence luminescence spectrophotometer system (Fluorolog-QM, HORIBA, Kyoto, Japan), with the excitation wavelength set at 360 nm.

2.5. Electrochemical Analysis

The electrochemical measurements were performed using a ZIVELAB single-channel electrochemical workstation (Seoul, Republic of Korea). The Ag/AgCl electrode, Pt wire, and ITO substance with the sample were used as the reference, counter, and working electrodes, respectively, for the three-electrode system. The working electrodes were prepared as follows: the as-prepared photocatalyst (10 mg) was mixed with a solution containing 20 μL Nafion and 1 mL isopropanol solution (isopropanol/H₂O = 1/3 v/v). Then, the mixture was sonicated for 60 min before being coated on an ITO glass (5 mm × 15 mm). Electrochemical impedance spectroscopy (EIS) Nyquist was performed with the frequency set in the range of 0.1 Hz to 1 MHz and an amplitude of 5 mV. The electrolyte solution was a mixture of K₃Fe(CN)₆ and KCN at concentrations of 2 mM and 1 M, respectively. The Mott–Schottky test was performed at a frequency of 1 kHz, and the amplitude was in the range of −1 V and 1 V, using 0.5 M of Na₂SO₄ (pH = 6.97) as the electrolyte. Electrochemical measurements were performed in the dark. The data were analyzed using ZMAN 2.5 software.

2.6. Catalytic Activity Tests

The photocatalytic activity of the as-synthesized samples was evaluated by measuring their efficiency of H₂ production. In brief, 50 mg of the photocatalyst was evenly dispersed in 10 mL of TEOA and 90 mL of DI water through sonication. This mixture was then transferred to a 250 mL flask, which was sealed with a rubber stopper. The gas was removed by vacuuming for 15 min, followed by N₂ gas purging for an additional 15 min. After that, the flask was illuminated with a 300 W Xenon lamp with the 420 nm cutoff filter for 3 h to facilitate H₂ production. The generated H₂ gas was determined using an HP gas chromatograph (GC5890 series II, Santa Clara, CA, USA) equipped with a thermal conductivity detector (TCD), using 5 Å molecular sieve columns and Ar as the carrier gas. The calibration of the TCD signal for H₂ using various dilutions of H₂ in N₂ is shown in Figure 1. To confirm the stability of the photocatalyst, a recycling test was conducted. The catalyst was washed with ethanol and DI water after each cycle, and their photocatalytic performance was re-evaluated.

3. Results

The Pd/g-C₃N₅ composites were prepared following the procedure outlined in Figure 2. Initially, g-C₃N₅ was synthesized by heating 3-amino-1,2,4-triazole at 550 °C under an ambient atmosphere. Subsequently, the wet-chemical reduction and photoreduction methods were applied to distribute metallic Pd nanoparticles onto the g-C₃N₅ layer. In both routers, a PdCl₂ solution served as the precursor solution to form the metallic Pd nanoparticles. In the wet-chemical reduction, the Pd²⁺ ion in the solution was reduced to Pd by EG. In the photoreduction method, the Pd²⁺ ion in the solution was reduced to Pd by the electrons from the conduction band of g-C₃N₅ during the photocatalytic reaction. In addition, the presence of EG as an organic capping ligand facilitates precise control over the shape and size of Pd nanoparticles, which is expected to enhance their photocatalytic activity.

The XRD patterns of all catalysts are consistent with those reported in previous studies on g-C₃N₅ (Figure 3A) [1,9]. The introduction of metallic Pd does not appear to significantly alter the crystalline structure of g-C₃N₅. Additionally, the distinct peaks of the Pd° NPs were not observed, due to the tiny atomic diameter of metallic Pd° nanoparticles and their relatively low concentration in the composite. Pd° and Pd²⁺ species were detected in both Pd/g-C₃N₅-EG and Pd/g-C₃N₅-PR samples (Figure 3B). Notably, the Pd° content was highest in Pd/g-C₃N₅-EG, indicating that the wet-chemical reduction method employing EG as a reducing agent was particularly effective in promoting Pd²⁺ reduction. In addition, compared to g-C₃N₅-PR (Figure 3(C1,C2)), Pd/g-C₃N₅-EG (Figure 3(D1,D2)) shows a finely dispersed metallic Pd homogeneously distributed on g-C₃N₅ layers, underscoring the superiority of the wet-chemical reduction method for synthesizing metallic Pd over the photoreduction method. This outcome is anticipated to result in enhanced photoelectrochemical properties.

Figure 4 illustrates the N₂ adsorption–desorption isotherms and pore width distribution for catalysts. The g-C₃N₅ possesses a BET surface area of 12 m² g⁻¹ with a pore size of 4 to 6 nm. The introduction of Pd° NPs in g-C₃N₅ did not significantly change the textural properties of the g-C₃N₅.

Figure 5A depicts the ultraviolet–visible diffuse reflectance spectra of pure and Pd°-decorated g-C₃N₅ samples. The g-C₃N₅ exhibits robust absorption across the solar spectrum, with a band edge wavelength of 650 nm, which is attributed to its narrow band gap [1,5,9]. The absorption spectra of g Pd/g-C₃N₅ composites appear to undergo no significant change upon the introduction of metallic Pd. The band gap (E_g) values for pure and Pd°-decorated g-C₃N₅ samples were determined from the plots of (αhν)^1/2 against photon energy (Figure 5B). The estimated E_g values were 1.95 eV for g-C₃N₅, 1.87 eV for Pd/g-C₃N₅-EG, and 1.91 eV for Pd/g-C₃N₅-PR.

Figure 5C illustrates the photoluminescence of g-C₃N₅ and Pd/g-C₃N₅ composites. The g-C₃N₅ samples exhibit an emission peak in the range of 400–600 nm, which is attributed to the band-to-band photoluminescence [1]. The Pd/g-C₃N₅ composites show a lower photoluminescence intensity compared to g-C₃N₅, indicating that the recombination rate has been effectively suppressed due to the introduction of Pd° NPs.

In the Nyquist plot, a higher curvature corresponds to lower charge transfer resistance. The curvature of the Nyquist curve follows the order of Pd/g-C₃N₅-EG > Pd/g-C₃N₅-PR > g-C₃N₅, indicating that Pd/g-C₃N₅-EG has the smallest impedance (Figure 5D). This finding verifies the improved charge transfer mechanism at the interface between Pd° NPs and g-C₃N₅, facilitating the movement of charge carriers. The excellent electron transport efficiency from g-C₃N₅ to Pd° NPs enhances the transmission and separation of charge carriers, thereby improving photocatalytic hydrogen production performance. Moreover, the Pd° NP-decorated g-C₃N₅ layer synthesized via wet-chemical reduction demonstrated a more effective metal–support interaction compared to the photoreduction method, leading to an expected higher hydrogen production yield. As discussed in the introduction section, the wet-chemical reduction method, employing organic capping ligands, facilitates the precise control of nanoparticle size and ensures a uniform distribution of Pd° NPs on g-C₃N₅ layers. This approach is anticipated to significantly affect the interface properties between Pd° NPs and g-C₃N₅ compared to the photoreduction method.

The conduction band level (E_CB) was determined using Mott–Schottky plots (Figure 5E). The E_CB values for all samples were determined to be −0.32 eV (vs. NHE), as previously demonstrated [10]. The valence band position (E_VB) was then calculated using the following equation: Eg = E_VB − E_CB. Based on this, the E_VB values are 1.55 eV for Pd/g-C₃N₅-EG, 1.59 eV for Pd/g-C₃N₅-PR, and 1.63 eV for g-C₃N₅. The energy band diagrams for pure and Pd°-decorated g-C₃N₅ samples are illustrated in Figure 5F.

Figure 6A shows the photocatalytic H₂ evolution rates over g-C₃N₅ and Pd/g-C₃N₅ composites. The average hydrogen evolution rate for Pd/g-C₃N₅-EG is approximately 891.304 µmol g⁻¹ h⁻¹ under visible light irradiation, representing a 34.5-fold increase compared to the pristine g-C₃N₅ at 23.753 µmol g⁻¹ h⁻¹ and a 7.8-fold increase compared to Pd/g-C₃N₅-PR at 113.418 µmol g⁻¹ h⁻¹. Figure 6B shows photocatalytic H₂ evolution rates over Pd/g-C₃N₅-EG composites after five reaction cycles. The hydrogen evolution performances of Pd/g-C₃N₅-EG remain stable across the testing cycles, highlighting the excellent stability of the catalysts.

The schematic band structure diagram of g-C₃N₅ and Pd/g-C₃N₅ composites (Figure 5F) reveals minimal differences in the CB and VB values between g-C₃N₅ and Pd/g-C₃N₅, suggesting the establishment of the heterojunction between Pd° NPs and the g-C₃N₅ layer. In this configuration, Pd° NPs play a role as electron traps, accumulating electrons from g-C₃N₅ due to the smaller work function of g-C₃N₅ (5.2 eV vs. vacuum [11]) compared to Pd° (5.6 eV vs. vacuum [12]). The establishment of a heterojunction effectively prevents the reverse migration of electrons trapped by Pd° NPs, thus promoting enhanced electron-transfer pathways. In addition, the Pd nanoparticle-decorated g-C₃N₅ layer, obtained through wet-chemical reduction, exhibited an excellent hydrogen production yield compared to the Pd° nanoparticle-decorated g-C₃N₅ layer obtained by photoreduction. The smaller size and uniform distribution of Pd° NPs in the wet-chemical reduction process can enhance the effective metal–support interaction with g-C₃N₅ layers. As a result, the favorable electronic transmission properties of uniformly distributed Pd° NPs in the Pd/g-C₃N₅-EG enable them to act as electron acceptors, effectively capturing photogenerated electrons and accelerating the separation of electron–hole pairs, thereby enhancing hydrogen production yield.

Because the energy level of the photogenerated electrons on the CB of g-C₃N₅ (−0.32 eV) is higher than the H⁺/H₂ redox level (0.0 eV), these electrons produced by g-C₃N₅ under visible light irradiation can effectively reduce H⁺ to generate H₂. However, the presence of O₂ and photogenerated holes can hinder the H₂ evolution process. To mitigate their effects, O₂ was purged from the reactor using N₂ gas and the holes were trapped by triethanolamine (TEOA). In this inert atmosphere, the H₂ evolution process can be described by Equations (1)–(3).

$$\mathrm{g}-{\mathrm{C}}_3{\mathrm{N}}_5\stackrel{\mathrm{h}\mathsf{\theta }}{\to }{\mathrm{e}}^{-}+{\mathrm{h}}^{+}$$

(1)

$${\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}{\mathrm{H}}_2$$

(2)

$$\mathrm{T}\mathrm{E}\mathrm{O}\mathrm{A}+{\mathrm{h}}^{+}\to {\mathrm{T}\mathrm{E}\mathrm{O}\mathrm{A}}^{+}$$

(3)

However, the rapid recombination of photogenerated carriers may limit the photocatalytic activity of g-C₃N₅. To overcome this, Pd° NPs, acting as electron accepters, can capture the photogenerated electrons from the CB of g-C₃N₅. Then, these electrons migrate to the Pd° NP surface, where they participate in the reduction of H⁺ to generate H₂.

4. Conclusions

We have successfully synthesized Pd° NPs decorated on the Pd/g-C₃N₅ layers using two methods: (1) wet-chemical reduction with EG serving as both a capping and reduction agent and (2) a photoreduction method employing a 10% v/v methanol solution under irradiation of a 300 Xenon light lamp. The results indicated that Pd/g-C₃N₅ synthesized via wet-chemical reduction exhibited enhanced optical and electrochemical properties compared to Pd/g-C₃N₅ obtained by the photoreduction method. This improvement is due to the wet-chemical reduction method, which facilitates the uniform deposition of Pd° NPs and promotes a more effective metal–support interaction, highlighting the superiority of the wet-chemical reduction method over the photoreduction method for synthesizing Pd° NPs. As a result, the Pd NP-decorated g-C₃N₅ layer, obtained through wet-chemical reduction, exhibited an excellent H₂ production yield compared to heterojunctions obtained by photoreduction. In addition, the performance of the Pd/g-C₃N₅ material for hydrogen production was evaluated over five recycling cycles. Although a slight decrease in hydrogen after five recycling cycles may be attributed to the loss of catalysts, the photocatalyst maintained its utility for sustained hydrogen production. Our study presents an approach that enables easy control of reaction conditions, maintains a low production cost, and facilitates the straightforward large-scale synthesis of metal–photocatalyst heterojunctions to improve H₂ production.