Manipulating photogenerated electron flow in nickel single‐atom catalysts for photocatalytic CO2 reduction into tunable syngas

Abstract

The key to designing photocatalysts is to orient the migration of photogenerated electrons to the target active sites rather than dissipate at inert sites. Herein, we demonstrate that the doping of phosphorus (P) significantly enriches photogenerated electrons at Ni active sites and enhances the performance for CO₂ reduction into syngas. During photocatalytic CO₂ reduction, Ni single‐atom‐anchored P‐modulated carbon nitride showed an impressive syngas yield rate of 85 μmol g_cat⁻¹ h⁻¹ and continuously adjustable CO/H₂ ratios ranging from 5:1 to 1:2, which exceeded those of most of the reported carbon nitride‐based single‐atom catalysts. Mechanistic studies reveal that P doping improves the conductivity of catalysts, which promotes photogenerated electron transfer to the Ni active sites rather than dissipate randomly at low‐activity nonmetallic sites, facilitating the CO₂‐to‐syngas photoreduction process.

INTRODUCTION

Driven by solar energy, photoreduction of CO₂ into value-added chemical products provides a green and promising avenue to obtain energy in the future.^1–4 Among various products of CO₂ photoreduction, syngas (CO and H₂) is regarded as a versatile feedstock for significant chemical industry, such as methanol synthesis and Fischer–Tropsch synthesis reactions.^5–7 Therefore, the development of efficient catalysts to convert CO₂ into syngas is an alternative process with excellent potential.^8–10

Due to the advantages of the low cost, thermal stability, and the response of visible light, graphitic carbon nitride (g-C₃N₄) has attracted considerable attention for photocatalysis.^11–13 g-C₃N₄ was first applied for photocatalytic H₂ evolution by Wang et al.¹⁴ and has been widely used in various photocatalytic reactions in recent years.^15–18 In particular, single-atom catalysts (SACs) anchored on a C₃N₄ substrate have been widely used in the field of CO₂ photoreduction.^19–21 For instance, single cobalt sites were loaded on g-C₃N₄ nanosheets to promote CO₂ activation and accordingly enhance the yield rate of syngas during CO₂ photoreduction.²² Unfortunately, the activity of C₃N₄-based SACs is generally limited by insufficient photogenerated electrons accumulated at metal single atoms.^23,24 Owing to the sluggish separation of photogenerated carriers and the low conductivity of C₃N₄, considerable number of photogenerated electrons were dissipated at inert sites rather than enriched at metal active sites during the migration process. If more photogenerated electrons could be oriented to the target metal single atoms, the photogenerated electrons' density at metal sites would significantly increase, thus improving the CO₂ photoreduction performance.

Herein, we report that the increased photogenerated electron density for Ni active sites by doping phosphorus (P) in Ni single-atom-decorated C₃N₄ catalysts (Ni₁/P-CN) enhanced the catalytic performance of CO₂ photoreduction into syngas. Impressively, Ni₁/P-CN showed a remarkable syngas yield rate of 85 μmol g_cat⁻¹ h⁻¹ and continuously adjustable CO/H₂ ratios ranging from 5:1 to 1:2, which exceeded those of most of the reported C₃N₄-based SACs. Mechanistic studies imply that the incorporation of Ni single atoms provided additional target sites for carrier transfer and CO₂ activation. Meanwhile, P doping improved the conductivity of the catalyst, which accelerated the transfer of photogenerated electrons to Ni sites, suppressing the random dissipation of electrons at C sites during the migration process. Moreover, the CO/H₂ ratios of syngas could be adjusted by orienting the migration of photogenerated electrons to the target active sites (Scheme 1).

[IMAGE OMITTED. SEE PDF]

EXPERIMENTAL SECTION

Instruments

Transmission electron microscopy (TEM) and high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images were collected using Hitachi H-7700 and JEM ARM-200F field-emission TEM, respectively. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) tests were performed using an ICP-mass spectrometer (Thermo iCAP RQ). X-ray powder diffraction (XRD) patterns were collected on a multifunctional rotating-anode X-ray diffractometer (MiniFlex600). We collected photoluminescence (PL) spectra using a luminescence spectrometer (Fluorolog-3-Tau and DeltaFlex). The ultraviolet–visible (UV–vis) diffuse reflectance spectra were collected using a UV–vis-near infrared spectrophotometer (SolidSpec-3700). For the photoelectrochemical experiments, we used the standard three-electrode setup (CHI 760E). A Pt foil, an fluorine-doped tin oxide glass coated with catalysts, and an Ag/AgCl electrode were used as the counter electrode, working electrode, and reference electrode, respectively. Ultraviolet photoelectron spectroscopy, X-ray photoelectron spectroscopy (XPS), and in situ diffuse reflective infrared Fourier transform spectroscopy (DRIFTS) were conducted at the beamline BL10B of the National Synchrotron Radiation Laboratory (NSRL).

Catalytic tests

Typically, 500 μg of photocatalyst and 2 mL of acetonitrile (MeCN) solution containing 0.2 M 1,3-diethyl-2-phenyl-2,3-dihydro-1H-benzo[d]imidazole (BIH) and 0.2 M trifluoroethanol (TFE) were added to a photocatalytic reactor. MeCN, BIH, and TFE were used as the solvent, electron donor, and proton source, respectively.²⁸ The reactor was further bubbled with pure CO₂ for 0.5 h. During the reactions, a xenon lamp was used under visible light (>400 nm). We used gas chromatography (GC-4000A; EWAI) to detect CO and H₂ products. The possible liquid products were detected by liquid chromatography. ¹³CO₂ was used as the feeding gas during ¹³C isotope labeling experiments at BL04B beamline of NSRL.

In situ soft X-ray absorption spectroscopy (sXAS) measurements

We conducted in situ sXAS at beamline BL10B of NSRL. The Ni L-edge, C K-edge, and N K-edge spectra were collected in the range of 845–880, 280–295, and 395–410 eV, respectively. During the measurements, the spectra were collected under darkness or light irradiation.³⁶

RESULTS AND DISCUSSION

Pristine C₃N₄ and Ni single-atom-decorated C₃N₄ (Ni₁/CN) were prepared using the one-step thermal polymerization method (see SupportingInformation). P-doped C₃N₄ (P-CN) and Ni₁/P-CN were obtained from C₃N₄ and Ni₁/CN by an additional phosphating process (Figure S1). The Ni mass loadings of Ni₁/CN and Ni₁/P-CN determined by ICP-AES were 2.7% and 2.5%, respectively. Figure 1A shows an HAADF-STEM image of the Ni₁/P-CN sample. Apparently, the individual Ni atoms were uniformly dispersed on the C₃N₄ substrate. Moreover, XRD patterns of C₃N₄, P-CN, Ni₁/CN, and Ni₁/P-CN are well indexed to the hexagonal phase of g-C₃N₄ (Figure S2). Meanwhile, no Ni nanoparticles or clusters were observed in Ni₁/CN and Ni₁/P-CN. In addition, elemental mapping images of Ni₁/CN and Ni₁/P-CN demonstrate the homogeneous distribution of various elements (Figure S3). We further conducted XPS tests. For the C 1s spectra, the peaks located at 284.5 and 288.0 eV were assigned to C═C and N═C–N species, respectively (Figure S4).²⁵ For the N 1s spectra, the peaks at 398.6 and 400.4 eV corresponded to C═N–C and N–(C)₃ species, respectively (Figure S5).²⁵ This was further confirmed by N K-edge spectra (Figure S6). Moreover, the P 2p XPS spectra were further recorded. As shown in Figure 1B, P 2p_3/2 and 2p_1/2 signals were observed at 133.2 and 134.0 eV, respectively, resulting from the P–N bond.^25,26 Apparently, the doped P substituted the partial C atoms in the C₃N₄ framework. In addition, the similar features of P 2p spectra for P-CN and Ni₁/P-CN suggest that the doped P was not directly coordinated with Ni metal sites in Ni₁/P-CN.

[IMAGE OMITTED. SEE PDF]

We further used X-ray absorption fine structure (XAFS) spectroscopy to study Ni species. As depicted in the Ni K-edge X-ray absorption near-edge structure spectra (Figure S7), the absorption edge positions of Ni₁/CN and Ni₁/P-CN were between those of the Ni foil and NiO, indicating the oxidized states of Ni species in Ni₁/CN and Ni₁/P-CN.²⁷ The corresponding R-space curves of the as-obtained samples are shown in Figure 1C. Ni₁/CN showed an obvious peak at around 1.5 Å, which corresponds to the position of Ni–N bonds in standard NiPc and thus could be ascribed to the Ni–N coordination. For Ni₁/P-CN, similar Ni–N coordination was also obtained. Meanwhile, no obvious peaks of Ni–Ni bonds could be observed, suggesting the atomic dispersion of Ni species in Ni₁/CN and Ni₁/P-CN. Besides, compared with Ni₂P, the absence of Ni–P coordination indicates that the doped P was not directly bonded with Ni atoms. The above points were also confirmed by wavelet transform analysis. The peak locations of Ni₁/CN and Ni₁/P-CN show distinct negative shifts in the wave vector of k relative to the Ni foil, implying that the atomic number of coordination element is much lower than that of the Ni element. Thus, the coordination element could be attributed to N atoms (Figure 1D–F). We further carried out extended X-ray absorption fine structure (EXAFS) spectra fitting analyses (Figures S8 and S9). The best-fitting results for Ni₁/CN demonstrate that the peak originated from Ni–N coordination (Table S1). For Ni₁/P-CN, fourfold Ni–N bonds with the same bond distance were obtained, as illustrated in Figure 1C.

We further evaluated the photocatalytic properties of the as-obtained catalysts for producing syngas. Each reaction was performed in a CO₂-saturated acetonitrile solution.²⁸ The generated gas products were detected by gas chromatography. With only pristine C₃N₄, H₂ was the main product and the generated CO was nearly negligible. The photocatalytic activity and selectivity of P-CN were similar to those of C₃N₄. In comparison, Ni₁/CN and Ni₁/P-CN showed a marked enhancement in the activity of CO production (Figure 2A). Specifically, for Ni₁/CN, the yield rates of H₂ and CO were 34 and 17 μmol g_cat⁻¹ h⁻¹, respectively. For Ni₁/P-CN, a yield rate of 14 μmol g_cat⁻¹ h⁻¹ was found for H₂ and a yield rate of 71 μmol g_cat⁻¹ h⁻¹ was found for CO, which are comparable to those of the C₃N₄-based photocatalysts reported recently (Tables S2 and S3). No liquid products were obtained for Ni₁/P-CN during photocatalytic CO₂ reduction reactions. Obviously, the incorporation of Ni atoms promoted the catalytic conversion of CO₂ into CO, which may be attributed to the efficient adsorption and activation of CO₂ in Ni-active sites. In addition, P doping significantly changed the catalytic selectivity of Ni-containing catalysts. Thus, we further explored the effect of different P contents in Ni₁/P-CN catalysts on catalytic products. A series of Ni₁/P-CNs with various P concentrations (denoted as Ni₁/P-CN-X) was prepared for different phosphating times. As shown in Figure 2B, with the increase of the P content, the CO yield gradually increased, indicating that the doped P promoted CO generation. The maximum CO activity and selectivity were achieved with a phosphating time of 3 h. More importantly, the H₂/CO production ratio could be adjusted for Ni₁/P-CN by simply controlling the phosphating time. The wide range from 2:1 to 1:5 covers the desirable composition of syngas toward methanol synthesis and Fischer–Tropsch synthesis reactions, demonstrating excellent potential in terms of application.²⁹ Besides, the apparent quantum yield of the CO product for Ni₁/P-CN was calculated to be 0.32% at 400 nm (Table S2).

[IMAGE OMITTED. SEE PDF]

Furthermore, we conducted a serious of controlled experiments to confirm the authenticity of CO₂ photoreduction reactions. As shown in Figure 2C, the CO and H₂ products for Ni₁/P-CN increased almost linearly with time. In addition, CO and H₂ were either undetectable or detected only in small amounts when the reaction was carried out without Ni₁/P-CN, without light irradiation, without TFE, without BIH, or on replacing CO₂ with Ar (Figure S10). Meanwhile, no liquid products were detected during reactions. Moreover, an isotope labeling experiment with ¹³CO₂ gas was carried out, in which the products were detected by synchrotron-based vacuum ultraviolet photoionization mass spectrometry (SVUV-PIMS). The photon energy of 14.2 eV was used to avoid the interference of possible complex fragment ions (Figure S11). As shown in Figure 2D, the products showed an obvious signal of ¹³CO, reconfirming that the CO product originated from the photocatalytic process. Besides, the stability of Ni₁/P-CN was also investigated. After five cycles, Ni₁/P-CN retained almost 100% of the initial catalytic performance, demonstrating its outstanding photocatalytic stability (Figure 2E). The stability of Ni₁/P-CN was also confirmed by XPS spectra and XAFS tests (Figure S12 and Table S1).

To explore the mechanism of CO₂ photoreduction, the energy band structures were first investigated. The values of the band gap of C₃N₄, P-CN, Ni₁/CN, and Ni₁/P-CN were calculated to be 2.72, 2.63, 2.68, and 2.66 eV in UV–vis spectra, respectively (Figure S13). We also investigated the valence band positions of catalysts using synchrotron radiation photoemission spectroscopy. The valence band maxima for C₃N₄, P-CN, Ni₁/CN, and Ni₁/P-CN were 1.77, 1.82, 1.72, and 1.70 eV below the Fermi level, respectively (Figure S14). Apparently, the band structures of the obtained samples had sufficient thermodynamic force to trigger CO₂-to-CO photoreduction (Figure S15). Also, the band structures of the catalysts did not change significantly after the incorporation of Ni atoms and P element.

We further used room-temperature PL spectroscopy to investigate the separation efficiency of photogenerated carriers for the obtained samples. As depicted in Figure 3A, all samples showed an obvious peak at ∼465 nm, corresponding to the intrinsic radiative recombination of photogenerated carriers. Ni₁/P-CN showed the lowest intensity relative to those of C₃N₄, P-CN, and Ni₁/CN, implying that the incorporation of Ni and P together promoted the separation of photogenerated carriers in carbon nitride catalysts.³⁰ Based on the previous studies, we inferred that the optimized separation and utilization of photogenerated carriers in Ni₁/P-CN resulted from two aspects: P doping can improve the electrical conductivity of carbon nitride materials,³¹ while the incorporation of metal single atoms provided additional target sites for carrier transfer.³² Furthermore, we conducted electrochemical impedance spectra (EIS) measurements. As shown in Figure 3B, Ni₁/P-CN showed the smallest semicircle among the samples, indicating the outstanding charge-transfer kinetics of Ni₁/P-CN.³³ In addition, Ni₁/P-CN showed the highest value of transient photocurrent response, further demonstrating the efficient separation of photogenerated carriers for Ni₁/P-CN (Figure 3C).^34,35

[IMAGE OMITTED. SEE PDF]

Considering that the Ni₁/P-CN catalyst showed remarkable separation efficiency of photogenerated carriers, another question arises: for the Ni₁/P-CN catalyst, how could the selectivity of CO₂ photoreduction be controlled by changing the content of P doping to generate different proportions of syngas? To obtain an answer to this question, we further conducted in situ sXAS tests. The transfer trend of photogenerated carriers in tested samples was deduced by comparing the difference between light and darkness of sXAS spectra intensity (Figure S16).³⁶ The peaks at 285 and 288 eV of the C K-edge spectra were attributed to the π* transition of C═C and N═C–N, respectively (Figure 3D). For the Ni L-edge (Figure 3E), the peaks for 2p_3/2 to 3d (L₃) and 2p_1/2 to 3d (L₂) excitation were observed at 855 and 872 eV, respectively.

The peak intensity in the C K-edge spectra of both C₃N₄ and P-CN decreased significantly under irradiation (Figure S17), which was attributed to the transition of photogenerated electrons to the unoccupied C 2p orbital of the lowest unoccupied molecular orbital. For Ni₁/CN, the extent of decline of the C K-edge peak intensity was observed. Meanwhile, the intensity of Ni L_2,3 peaks reduced slightly, which was attributed to the partial photogenerated electron transfer toward the unoccupied Ni 3d orbital. In comparison, Ni₁/P-CN showed the lowest decreased extent of C K-edge peaks under irradiation. Significantly weakened intensity was also obtained with light irradiation for Ni₁/P-CN in the Ni L-edge. Furthermore, we quantified the changing amplitude of peaks for in situ sXAS tests (Figure 3F). Apparently, the incorporation of Ni single atoms provided additional target sites for photogenerated electrons transfer in Ni₁/CN. For Ni₁/P-CN, further P doping changed the direction of electron flow, leading to the accumulation of more photogenerated electrons at Ni sites. We also conducted in situ XPS to monitor the flow direction of photogenerated holes (Figure S18). Notably, the electron transfer trends of different samples, reflected from in situ sXAS and in situ XPS tests, conformed well to the catalytic properties of different carbon nitride catalysts. Specifically, for C₃N₄ and P-CN, photogenerated electrons were enriched at C sites. Due to the weak activation ability of CO₂ molecules in C sites, C₃N₄ and P-CN showed low activity and selectivity of CO₂ photoreduction. After introducing Ni single atoms, the partial electrons transferred to the Ni sites, which provided more efficient active sites toward the activation of CO₂ than C sites, promoting the generation of CO from CO₂. Besides, as the photogenerated electrons were mainly concentrated at the C sites, H₂ evolution was still the main reaction for Ni₁/CN. However, for Ni₁/P-CN, most photogenerated electrons accumulated at Ni sites, facilitating CO₂ photoreduction and accordingly increased the yield rate of CO generation.

We further used in situ DRIFTS to monitor the CO₂ photoreduction process for the Ni₁/P-CN sample. As shown in Figure 4A, the bands at 1410 and 1446 cm⁻¹ corresponded to CO₃²⁻ and m-CO₂⁻, respectively.^37,38 Besides, the bands at 1383 and 1617 cm⁻¹ corresponded to COOH*, which was regarded as an important intermediate species for CO₂ reduction into CO.^39–41 Combined with in situ sXAS and in situ DRIFTS results, we identified the possible reaction pathway of CO₂ photoreduction with the Ni₁/P-CN catalyst. As shown in Figure 4B, the CO₂ molecule was first adsorbed and preliminarily activated at the Ni site. Subsequently, the H atom was coupled with the activated CO₂ to generate the COOH* group. Next, COOH* was transformed into CO* plus a H₂O molecule through the proton–electron coupling step, and CO* was finally released from the catalyst surface as a free CO product.

[IMAGE OMITTED. SEE PDF]

CONCLUSION

In conclusion, the photocatalytic CO₂ reduction activity and selectivity of Ni₁/P-CN could be tailored by manipulating the enrichment trend of photogenerated electrons. Ni₁/P-CN showed an excellent syngas yield rate of 85 μmol g_cat⁻¹ h⁻¹ during the reactions. Moreover, adjustable CO/H₂ ratios ranging from 5:1 to 1:2 were obtained by adjusting the degree of phosphatization. This work not only proposes a strategy for increasing photocatalytic performance by manipulating electron flow but also advances the understanding of the relation between photogenerated electrons and the catalytic selectivity of photocatalysts.

ACKNOWLEDGMENTS

Yida Zhang and Qingyu Wang contributed equally to this work. The authors acknowledge the financial support provided by the National Natural Science Foundation of China (Grant Nos. 12222508, U1932213), the Fundamental Research Funds for the Central Universities (Grant No. WK2060000016), the National Key R&D Program of China (Grant No. 2023YFA1506304), the USTC Research Funds of the Double First-Class Initiative (Grant No. YD2310002005), and the Youth Innovation Promotion Association CAS (Grant No. 2020454).

CONFLICT OF INTEREST STATEMENT

The authors declare that there are no conflicts of interests.

References

Wang K, Du Y, Li Y, et al. Atomic‐level insight of sulfidation‐engineered aurivillius‐related Bi₂O₂SiO₃ nanosheets enabling visible light low‐concentration CO₂ conversion. Carbon Energy. 2023;5(2): [eLocator: e264].

Wang Q, Zhang Y, Lin M, et al. Photoinduced metastable asymmetric Cu single atoms for photoreduction of CO₂ to ethylene. Adv Energy Mater. 2023;13(42): [eLocator: 2302692].

Ding Y, Maitra S, Wang C, et al. Vacancy defect engineering in semiconductors for solar light‐driven environmental remediation and sustainable energy production. Interdiscip Mater. 2022;1(2):213‐255.

Zhang X, Huang W, Yu L, et al. Enabling heterogeneous catalysis to achieve carbon neutrality: directional catalytic conversion of CO₂ into carboxylic acids. Carbon Energy. 2023;6(3). [DOI: https://dx.doi.org/10.1002/cey2.362]

Zhang Y, Hou T, Xu Q, et al. Dual‐metal sites boosting polarization of nitrogen molecules for efficient nitrogen photofixation. Adv Sci. 2021;8(13): [eLocator: 2100302].

Wan S, Xu J, Cao S, Yu J. Promoting intramolecular charge transfer of graphitic carbon nitride by donor‐acceptor modulation for visible‐light photocatalytic H₂ evolution. Interdiscip Mater. 2022;1(2):294‐308.

Wang Q, Xiao Y, Yang S, et al. Monitoring electron flow in nickel single‐atom catalysts during nitrogen photofixation. Nano Lett. 2022;22(24):10216‐10223.

Bi W, Zhang L, Jiang H, Li C, Hu Y. Construction of CuInS₂/C/TiO₂ hierarchical tandem heterostructures with optimized CO₂ photoreduction under visible light. Chem Eng J. 2022;433(3): [eLocator: 133679].

Lei B, Cui W, Chen P, Chen L, Li J, Dong F. C‐doping induced oxygen‐vacancy in WO₃ nanosheets for CO₂ activation and photoreduction. ACS Catal. 2022;12(15):9670‐9678.

Ban C, Duan Y, Wang Y, et al. Isotype heterojunction‐boosted CO₂ photoreduction to CO. Nano‐Micro Lett. 2022;14(1):74.

Li XX, Zhang L, Yuan L, et al. Constructing crystalline redox catalyst to achieve efficient CO₂ photoreduction reaction in water vapor. Chem Eng J. 2022;442(1): [eLocator: 136157].

Gao Y, Sun L, Bian J, Zhang Z, Li Z, Jing L. Accelerated charge transfer of g‐C₃N₄/BiVO₄ Z‐scheme 2D heterojunctions by controllably introducing phosphate bridges and Ag nanocluster co‐catalysts for selective CO₂ photoreduction to CO. Appl Surf Sci. 2023;610: [eLocator: 155360].