A Green Synthesis of Ru Modified g-C3N4 Nanosheets for Enhanced Photocatalytic Ammonia Synthesis

Centre for Technology in Water and Wastewater (CTWW), School of Civil and Environmental Engineering, University of Technology Sydney (UTS), Sydney, NSW 2007, Australia State Key Laboratory of Rare Earth Resource Utilization, Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, Changchun 130022, China Laboratory of Advanced Catalysis for Sustainability, School of Chemistry, The University of Sydney, Sydney 2006, Australia Centre for Advanced Materials & Industrial Chemistry (CAMIC), School of Science, RMIT University, Melbourne, VIC 3000, Australia Australian Centre for Microscopy and Microanalysis, The University of Sydney, Sydney 2006, Australia Beijing Engineering Research Center of Sustainable Urban Sewage System Construction and Risk Control, Beijing University of Civil Engineering and Architecture, Beijing 102612, China


Introduction
Nitrate (NO 3 -) is a necessary nutrient for plants, whereas high-level of NO 3 enters the food chain via water sources can be toxic and hazardous to the ecosystem [1][2][3]. The main culprits of NO 3 pollution include agricultural runoff, fertilizer abuse, septic systems, industrial plants, and irrigation systems [4]. Currently, NO 3 contamination has been a serious environmental concern as it can affect the quality of groundwater and surface water, causing health problems to humans [5,6]. In an oxygen-deficient environment, such as in the digestive tract, NO 3 can be reduced to more toxic nitrite [7,8]. Nitrite can oxidize low ferritin into methemo-globin in the human body, causing the loss of oxygen transportation. It can also react with secondary amine compounds to produce carcinogenic nitrosamines [9]. Therefore, it is urgent to develop a novel strategy to remove NO 3 pollution with high efficiency and selectivity. Ammonia (NH 3 ) is an important industrial ingredient widely used in fertilizers, pharmaceutical industries, and other areas [10][11][12]. More importantly, it is also an essential energy source. The decomposition of ammonia is a low-cost and facile process; therefore, using ammonia as a hydrogen carrier is able to solve the hydrogen storage problem to a certain extent [13,14]. With the increase of the global population, the demand for NH 3 keeps increasing. Currently, NH 3 is mainly produced via the Haber process, in which gaseous nitrogen (N 2 ) and water gas are converted to NH 3 under high temperature and high pressure with the assist of catalysts [15]. Every year, the synthesis of NH 3 consumes about 2% of global energy, leading to serious carbon dioxide emission [16][17][18]. Thus, it is urgent to develop a green synthesis of NH 3 under ambient conditions.
In recent years, photocatalytic synthesis of NH 3 has become a hot research frontier, where researchers are mainly focusing on the reduction of gaseous nitrogen [19][20][21][22]. However, the research is still limited in the laboratory, because the strong bond energy of N ≡ N bonds and the poor solubility of N 2 make the catalytic NH3 yield efficiency low [16]. Compared with N 2 , the required energy for the cleavage of N=O is only 21.68% as that of N ≡ N bonds. In addition, the solubility of NO 3 is 40000 times higher than that of N 2 [23]. Therefore, if the waste NO 3 can be converted to NH 3 , the conversion of environmental hazards to valuable energy resources will be more sustainable for the planet.
Graphitic carbon nitride (g-C 3 N 4 ) is a metal-free photocatalyst, which is a rising star in photocatalysis because it has the advantages of facile synthesis, wide sources, and good photocatalytic activity for redox reactions [24][25][26]. In recent years, several kinds of g-C 3 N 4 -based photocatalysts have been prepared and applied in NO 3 wastewater treatment, such as TiO 2 /g-C 3 N 4 [27], Mn 2 O 3 /g-C 3 N 4 [28], and g-C 3 N 4 @AgPd heterojunction [29]. However, the main aim is to convert NO 3 to N 2 to reduce environmental hazards. Compared with the generation of N 2 , the reduction of NO 3 to NH 3 is more challenging in terms of kinetics and thermodynamics, as it is an eight-electron process with multiple steps [30]. On the other hand, many ruthenium-containing compounds exhibit good catalytic properties because of their unique electronic structure [31][32][33][34]. Meanwhile, ruthenium (Ru) can absorb the light in the visible spectrum, and it is being actively researched for solar energy technologies [35,36]. Nevertheless, toxic reducing agents like ethylene glycol are usually required to reduce Ru 3+ to Ru 0 , making the synthesis complex, dangerous, and unecofriendly [37]. Until now, Ru modified g-C 3 N 4 photocatalysts for highly efficient NO 3 reduction to NH 3 have never been reported, and the reaction mechanism is not clear.
In this work, we report a green synthesis of Ru modified g-C 3 N 4 nanosheets with significantly enhanced photocatalytic activity on the reduction of NO 3 to NH 3 . Herein, waste green tea bags were used to reduce the Ru 3+ , and the photocatalytic activity of the optimized sample was 2.93-fold as that of bulk g-C 3 N 4 under simulated sunlight irradiation. The material preparation process was ecofriendly without using strong reducing reagent. Based on experimental and theoretical studies, the introduction of Ru to g-C 3 N 4 can not only boost the light absorption, the adsorption of NO 3 -, but also accelerate the separation of electron-hole pairs. The thermodynamic energy barrier for the rate determining step in NO 3 reduction to the NH 3 process is calculated to be less than 0.75 eV, which is much lower than the competing H 2 generation (0.98 eV) and N 2 formation (1.36 eV), leading to the preference of generating NH 3 . This work provides a novel approach to synthesize metallic particle-based photocatalysts for highly efficient photocatalytic NO 3 reduction to synthesize NH 3 .
2.2. Synthesis of g-C 3 N 4 Nanosheets. g-C 3 N 4 was prepared via the thermal polymerization of dicyandiamide. Firstly, 5 g of dicyandiamide was put in a 50 mL corundum crucible without the lid. Then, the crucible was transferred into the oven and heated from 25 to 540°C with the heating rate of 4°C/min. After heating at 540°C for 4 h in air, the bulk g-C 3 N 4 was obtained. The g-C 3 N 4 nanosheets were synthesized via the ultrasonic peeling of bulk g-C 3 N 4 in water.
In details, 500 mg of bulk g-C 3 N 4 was added in 800 mL water and put under ultrasonic bath for 24 h. Finally, the ultrathin g-C 3 N 4 nanosheets were successfully obtained and named as 2D-CN.
2.3. Synthesis of Ru/g-C 3 N 4 . Herein, we used waste green tea as the reduction regent to prepare Ru modified g-C 3 N 4 (Scheme 1). To get simulated waste tea bags, 5 tea bags were put in boiled water for 20 min. Then, the 5 used tea bags were put in 500 mL and boiled for 30 min. After that, the tea water was gone through a filter for further use. A certain amount of RuCl 3 and 0.5 g of g-C 3 N 4 nanosheets was put into 150 mL prepared tea water and under magnetic stir for 24 h. Finally, the powder was washed by water and ethanol for 5 times and then dried at 60°C for 4 h. The amounts of RuCl 3 used were 0.005, 0.01 and 0.038 g, and the obtained samples were named as CN-Ru-0.5, CN-Ru-1, and CN-Ru-3, respectively.

2.4.
Characterizations. The X-ray diffraction (XRD) patterns of the samples were tested with a Bruker D8 Discover XRD with intense Cu Kα radiation (40 kV and 40 mA, λ = 1:5406 Å) at room temperature. The morphology observation on the materials was carried out using a Zeiss Supra 55VP scanning electron microscope (SEM), with an operating voltage of 5-15 kV. The images of SEM-energydispersive X-ray spectroscopy (EDS) for elemental mapping were obtained with the Oxford Ultim Max. High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and EDS mapping were performed on a double-corrected FEI Themis-Z 60-300 transmission electron microscopy (TEM) equipped with ChemiSTEM EDS detector system for ultra-high count rates. The X-ray photoelectron spectra (XPS) of the samples were tested with a Thermo Fisher Scientific K-Alpha+X-ray photoelectron spectroscopy. The UV-vis diffuse reflectance spectra (DRS) of prepared samples were obtained from a Perkin Elmber Lambda 950 UV/VIS/NIR spectrometer, using high-purity 2 Energy Material Advances barium sulfate (BaSO 4 ) as the blank reference. Photoluminescence (PL) spectra were obtained from a Shimadzu RF-6000 fluorescence spectrometer excited at 325 nm. The photocurrent, electrochemical impedance spectroscopy (EIS) and Mott-Schottky curves of the samples were obtained from an electrochemical working station (CHI-760E) in 0.1 M Na 2 SO 4 solution (details shown in supplementary materials).
2.5. Photocatalytic Activity Testing. The photocatalytic nitrate reduction performance was carried out with a selfmade quartz reactor under simulated sunlight irradiation. During the reaction process, the light intensity was 600 mW/cm 2 (HSX-F300), and it was measured by a radiometer. The distance between the light and the reactor was 10 cm. Firstly, 20 mg of the samples was weighed and put into the reactor. Then, 95 mL of NaNO 3 (10 mg/L) solution and 5 mL of formic acid (200 μL/L) were added. Before the reaction, argon was purged into the solution to remove air. After 2 h strong stir to make the powder catalysts uniformly dispersed in the solution, the light was turned on and the light irradiation lasted 4 h. In each hour, 3 mL of liquid was sampled from the reactor. The concentration of ammonium was measured by UV-vis spectrometer using Nessler's reagent method and was further verified by 1H-Nuclear magnetic resonance spectroscopy (NMR) analysis. The concentration of NO 3 and NO 2 was measured by a high-pressure integrated capillary anion chromatography (IC). The exact details of Nessler's reagent method, NMR, and IC are shown in the supplementary materials ( Figure S1-S4).
2.6. Density Functional Calculation. The calculations of geometry structures and energies were carried out by density functional theory (DFT) with Vienna ab initio simulation package (VASP) [38][39][40][41]. The exact details are presented in supplementary materials.

Results and Discussion
3.1. Structure and Morphology. In the XRD patterns of g-C 3 N 4 ( Figure S5 in supporting information, SI), two distinct peaks at 13.1 and 27.1°can be assigned to the (100) and (002) crystal plane diffraction (JCPDS 87-1526) [42]. After peeled off by ultrasonic powder, the peak intensity decreased, indicating the reduced crystallinity.
Since the loaded amount of Ru is not high, it cannot be detected by XRD. Then, HRTEM and SEM were used to explore the morphology of the prepared samples. As shown in Figure 1(a), the g-C 3 N 4 nanosheets showed an ultrathin film layer structure. The CN-Ru-1 sample had the same morphology as 2D-CN with Ru particles loaded on the g-C 3 N 4 nanosheets (dark spots in Figure 1(b)). The size of the Ru particles was less than 5 nm. To further investigate the lattice, the Ru particles were zoomed in and displayed in Figure 1(c). The lattice distance of 0.234 nm

Energy Material Advances
can be indexed to the (100) of metallic Ru. Additionally, the high-angle annular dark-field (HAADF) image and energy dispersive spectroscopy (EDS) for Ru element mapping were obtained to further confirm the component of the CN-Ru sample. As exhibited in Figures 1(d) and 1(e) and Figure S6, Ru particles are uniformly distributed on g-C 3 N 4 nanosheets. SEM image was also employed to study the morphology of CN-Ru-1 (Figure 1(f)). In the SEM image, we can only see the nanosheet structure of g-C 3 N 4 , since the size of Ru particles was too small to be observed by SEM.
Then, XPS was used to study the components of the prepared samples. Figures 2(a) and 2(b) are the narrow C 1 s spectra of 2D-CN and CN-Ru-1. The three peaks of 2D-CN at 293.6, 287.9, and 284.6 eV can be assigned to the conjugated π electrons and sp 2 -hybridized C and C-C bonds [10]. Compared with 2D-CN, two new peaks appeared in the narrow XPS spectra of C of CN-Ru-1. The peak at 286.1 eV was attributed to C-O which came from the green tea (Figure 2(b)) [43]. The other new peak at 280.2 eV is associated with Ru 0 [44]. Both 2D-CN and CN-Ru-1 have 3 peaks in the narrow N 1 s spectra. In Figure 2(c), the three peaks at 404.4, 400.0, and 398.4 eV of 2D-CN are attributed to the charging effect, tertiary nitrogen, and sp 2 -hybridized nitrogen. Notably, the three peaks of N of CN-Ru-1 shifted to 404.9, 400.1, and 398.6 eV, which was caused by the interaction between 2D-CN and Ru [45].
Based on the XRD, SEM, TEM, and XPS results, it can be concluded that metallic Ru particle modified g-C 3 N 4 nanosheets were successfully fabricated.

Enhanced Photocatalytic NO 3
-Reduction to NH 3 . Herein, we used photocatalytic NO 3 reduction to NH 3 to evaluate the catalytic activity of samples. After 4-hour simulate sunlight irradiation, all the samples exhibited apparent photocatalytic activity on the reduction of NO 3 to NH 3 (Figure 3(a)). The NH 3 yield rate of 2D-CN was 1.126 mg/h/g cat , and it was 1.26-fold as that of bulk g-C 3 N 4 . When metallic Ru was used to modify the 2D-CN, the photocatalytic activity increased remarkably. Among all the samples, CN-Ru-1 was the most active one (2.627 mg/h/g cat ), and its activity is 2.93 times higher than bulk g-C 3 N 4 . Cycle stability is crucial for the study and application of photocatalysts. After the fourth cycle, the NH 3 yield of CN-Ru-1 was still 2.32 mg/h/g cat , which was 88.16% as the fresh catalyst, indicating good cycle stability for reusage. The use of formic acid did not influence of the detection by the accuracy of Nessler's reagent method ( Figure S7).

Energy Material Advances
To confirm the source of NH 3 generated, we carried on the 15 N isotope experiment with Na 15 NO 3 . A series of 15 NH 4 Cl solutions with different concentration were firstly used to generate a standard line, and then the concentration of photocatalytic reactions can be analyzed. As shown in Figure 3(c), the standard 1H-NMR spectrum of 15 NH 4 + had two peaks with a coupling constant of 73.14 Hz. In the solution after photocatalytic reaction of CN-Ru-1 with 15 NO 3 -, two characteristic 15 NH 3 peaks were also found, and the calculated concentration of NH 3 was 2.05 mg/L, which well matched the results of UV-vis spectrophotometry. The NMR results can confirm that the formation of NH 3 is attributed to the photocatalytic reduction of NO 3 -. Besides NH 3 , there might be some other products of the NO 3 reduction. Herein, we used ion chromatography to measure the concentration of NO 3 and NO 2 -. After 4 h light irradiation, bulk CN reduced 48.85% of the NO 3 -(10 mg/L), 50.77% of which was converted to NH 3 . Under the same condition, CN-Ru-1 eliminated 92.85% of NO 3 -, and the NH 3 selectivity was 77.9%. The UV-vis spectrophotometry, 1H-NMR, and ion chromatography results confirmed that both the NO 3 conversion rate and NH 3 selectivity were significantly increased by CN-Ru-1, compared with bulk CN.
In Table S1, we summarized several typical g-C 3 N 4 , TiO 2, and other materials for photocatalytic NO 3 reduction in literature for comparison. Many materials can achieve highly efficient NO 3 removal, whereas the ammonia selectivity is quite low. Pd/TiO 2 can get a high NH 3 selectivity of 76.9%, but its catalytic activity is really poor [46]. Nevertheless, the CN-Ru-1 reported in this work can simultaneously realize both high NO 3 removal and high NH 3 selectivity.

Mechanism of the Enhanced Performance.
Light absorption is an essential step for photocatalytic reaction. 5 Energy Material Advances significantly enhanced. The increased light absorption enabled to make the catalysts absorb more photons, boosting the solar energy utilization ratio. The density of states of g-C 3 N 4 and CN-Ru is calculated by DFT (Figure 4(b)). Compared with g-C 3 N 4 , the introduction of metallic Ru can bring some metallicity to the composite catalysts, which is beneficial to the transfer and separation of photocatalytic generated electron-hole pairs [47].
In order to explore more direct evidence about the transfer and separation of photogenerated electron-hole pairs, the photocurrent response, EIS, and PL spectra of bulk CN, 2D-CN, and CN-Ru samples were tested. Figure 4(c) is the photocurrent response image. All the tested samples had apparent and rapid photocurrent response under light irradiation. Among all the samples, the photocurrent intensity of CN-Ru-1 was the strongest, which followed the same trend as the photocatalytic activity, indicating that CN-Ru-1 had the fastest separation and transfer of electrons and holes. The photocurrent slightly decreased in 200 s, which was caused by the decreased interaction between the powder catalysts and the ITO glass [48]. It does not mean the photocatalysts have poor stability. In the EIS Nyquist plot, the bulk CN had the largest arc radius because of its poor conductivity. The 2D-CN got a smaller arc radius as it had better conductivity than bulk CN. When the 2D-CN got modified by metallic Ru, the arc radius of the EIS Nyquist plot significantly decreased, because Ru had strong conductivity. The more metallic Ru loaded, the better conductivity the sample got. The PL spectra of bulk CN, 2D-CN, and CN-Ru samples were displayed in Figures 4(e) and 4(f). Bulk CN showed a broad and strong emission spectrum with profiles slightly deviating from a perfect Guanine curve centered at about    Energy Material Advances 460 nm. Compared with bulk CN, the PL intensity of the other samples decreased significantly, which mean the recombination of photoinduced charge carriers was obstructed, and it is helpful for the enhanced photocatalytic activity [49]. Experimental studies on NO 3 reduction have shown that Ru modified g-C 3 N 4 plays an essential role in improving NH 3 production compared with pure g-C 3 N 4 . To further understand the reaction mechanism and the origin of high activity and selectivity of CN-Ru, the density functional theory calculation is performed. NO 3 adsorption, the first step of NO 3 reduction, was first calculated, and the optimized adsorption structures, as well as the adsorption energies, are listed in Figures 5(a) and 5(b). The adsorption energy of NO 3 on g-C 3 N 4 was -1.85 eV, and it was -3.75 eV on CN-Ru. It is clearly seen that Ru modification increased the NO 3 stability by 2.03 times compared with   Energy Material Advances pristine g-C 3 N 4 , indicating the following NO 3 reduction step proceeds more easily, consistent with the experimental observations. Furthermore, metallic Ru particles involve a high spin density on Ru cluster and positive Bader charge, as well as a zero gap, which can enhance the electron transfer ability (Figures 5(c) and 5(d)). Since Ru cluster got a positive Bader charge, the photocatalytic generated electrons for g-C 3 N 4 can be accumulated on Ru, making it as the active sites. At the same time, the CN-Ru system got some magnetic properties, and the spin density is concentrated on Ru. All these electronic structure results are responsible for the improved photocatalytic activity of CN-Ru.
To further understand the role of metallic Ru particles in the NO 3 reduction process, we calculated the free energy diagrams of NO 3 reduction on g-C 3 N 4 and CN-Ru. Compared with the generation of NH 3 , there are two main com-peting reactional products for the photocatalytic reduction of NO 3 -including H 2 and N 2 . For g-C 3 N 4 , it is very easy to take part in hydrogen generation reaction, while the formation of N 2 is very difficult because a higher thermodynamic energy barrier of 2.33 eV should be overcome (Figure 6(a)). As for CN-Ru, the thermodynamic energy barriers of the formation of H 2 and N 2 are 0.98 and 1.36 eV (Figure 6(b)). All free energy curves are downhill from NO 3 to HNO 2 , and although the reaction pathway is different, the thermodynamic energy barrier for the rate determining step in this process is all calculated to be less than 0.75 eV (Figure 6(c)), which is much lower than those of 2H ⟶ H 2 (0.98 eV) and 2 N ⟶ N 2 (1.36 eV), suggesting that induced Ru cluster played a key role in the enhancement of the photocatalytic activity and selectivity for NO 3 reduction to form NH 3 , in good agreement with the experimental observations.   Figure 6: Calculated free energy differences between the formation of H 2 and N 2 on g-C 3 N 4 (a) and CN-Ru (b). Calculated free energy changes of NO 3 reduction to NH 3 on g-C 3 N 4 and CN-Ru (c).

Energy Material Advances
Based on the experimental and calculated results, the mechanism for the enhanced photocatalytic NO 3 reduction to NH 3 activity can be summarized as follows (Scheme 2). Firstly, the introduction of metallic Ru caused a redshift of the absorption edge as well as broad absorption in visible light range, which boosted the absorption of photons and the utilization of solar light. Secondly, CN-Ru samples have much better conductivity and higher separation efficiency of photogenerated electron-hole pairs than g-C 3 N 4 . Thirdly, the high spin density of Ru cluster and its positive Bader charge can accumulate electrons, making the adsorption and cleavage of NO 3 easier. More importantly, the thermodynamic energy barrier for the rate determining step in this process is all calculated to be less than 0.75 eV (Figure 6(c)), which is much lower than the competing H 2 generation (0.98 eV) and N 2 formation (1.36 eV), leading to the preference of generating NH 3 and higher activity.
Using green tea as reducing reagent to prepare nanomaterials is promising because it is cost-effective. More importantly, the operation and storage are much safer than using strong reducing reagent like sodium borohydride. The tea polyphenols of green tea are regarded as active ingredients to achieve reduction reactions [50,51]. Meanwhile, the average reduction potential of green tea is about 0.219 V, indicating that the approach reported in this work can be expanded to the synthesis of other metal-based materials [52]. In Figure S8, we summarized the elements that can be reduced to zero valent theoretically, including Tc, Ru, and Cu, which would potentially use the proposed approach [53].

Conclusion
In summary, we used a facile and green approach to synthesize novel metallic Ru modified g-C 3 N 4 nanosheets as photocatalysts for enhanced NO 3 reduction to NH 3 . The optimal sample (CN-Ru-1) had the highest NH 3 yield rate of 2.627 mg/h/g cat , and it was 2.93 times as that of bulk CN. After 4-hour light irradiation, CN-Ru-1 eliminated 92.85% of NO 3 -, and the NH 3 selectivity was 77.9%. After four cycles, the sample still had good photocatalytic performance (88.16% as the fresh catalyst). NMR and 15 N isotope labeling provided solid evidence that the N of NH 3 was from the reduction of NO 3 -. The g-C 3 N 4 nanosheets modified by metallic Ru particles have stronger light absorption, better conductivity, and more rapid separation of electron-hole pairs. With the enhanced adsorption energy of NO 3 and the low thermodynamic energy barriers, the photocatalytic activity and selectivity were increased significantly. The results and findings of this work may provide a new platform for the facile and green synthesis of metal particle modified photocatalysts for reducing NO 3 to NH 3 under ambient conditions.