Transition metal single atom-optimized g-C₃N₄ for the highly selective electrosynthesis of H₂O₂ under neutral electrolytes

Abstract

Neutral electrosynthesis of H₂O₂via the 2e⁻ ORR is attractive for numerous applications, but the low activity and high cost of electrocatalysts have become important constraints. Therefore, the development of cheap and efficient electrocatalysts for the 2e⁻ ORR is necessary. Herein, we report the embedding of transition metal single atoms (TM SAs) in g-C₃N₄ nanosheets (CNNS). The introduction of TM SAs increases the N–C=N content and reduces the C–C/C=C content in CNNS, which contributes to the increased selectivity of TM SA/CNNS for the 2e⁻ ORR. TM SA is the main reason for the enhanced activity of the 2e⁻ ORR. Based on the results obtained by replacing a series of TM SA, the Ni_0.10 SA/CNNS with optimal N–C=N content exhibited the best selectivity (∼98%) and highest yield of H₂O₂ (∼503 mmol g_cat⁻¹ h⁻¹), which is ∼14.6 times higher than that of CNNS (∼34.4 mmol g_cat⁻¹ h⁻¹). Other TM SA/CNNS also exhibited high activity and selectivity. This study demonstrates the ability of TM SA to modulate the selectivity and activity of CNNS, making it a promising candidate for the 2e⁻ ORR and providing more reference ideas for the preparation of H₂O₂.

---

New concepts

Transition metal single atom-optimized g-C₃N₄ have been successfully synthesized via a two-step thermal oxidation etching method for the efficient neutral electrosynthesis of H₂O₂. Compared with CNNS, the H₂O₂ selectivity of Ni_0.10 SA/CNNS increased from 61% to 98% in 0.1 M PBS. The H₂O₂ production rate of Ni_0.10 SA/CNNS is 503 mmol g_cat⁻¹ h⁻¹, which is 14.6 times that of CNNS (34.4 mmol g_cat⁻¹ h⁻¹). The excellent electrochemical activity is due to the introduction of TM SA, optimising the structure of CNNS and promoting the increase in N–C=N sites and the decrease in C–C/C=C sites in the CNNS, and improving the selectivity of the electrocatalysts for the 2e⁻ ORR. The high electrochemical activities exhibited by TM SA/CNNS are due to the isolated TM SA on CNNS, which enhanced the 2e⁻ ORR activity of electrocatalysts.

1. Introduction

As a high-value and environmentally friendly oxidant, hydrogen peroxide (H₂O₂) has a wide range of applications in various industrial environments such as chemical synthesis, environmental remediation, and paper/pulp bleaching.^1–4 Currently, the synthesis of H₂O₂ relies on anthraquinone processes.⁵ However, anthraquinone processes are limited by complex infrastructure, large energy input, and the generation of large quantities of waste chemicals.⁶ In addition, the high concentration of H₂O₂ (up to 70 wt%) obtained by the anthraquinone method has huge safety hazards in transportation and does not match the end-user demand that requires low concentrations of H₂O₂.⁷ Based on this, green and sustainable electrochemical synthesis (two-electron oxygen reduction reaction, 2e⁻ ORR) that is powered by renewable electricity is highly sought-after for the direct on-site production of H₂O₂.⁸ However, the electrochemical synthesis of H₂O₂ is constrained by the four-electron oxygen reduction reaction (4e⁻ ORR),⁹ so electrocatalysts with high activity and selectivity are essential. At present, noble metal catalysts (e.g., Pt, Pd, Ag, Au, Pd–Hg, and Au–Pd) exhibit excellent selectivity and activity for the synthesis of H₂O₂^10–16 but these catalysts are expensive, scarce, and have inferior stability, which restrict their large-scale production.^13,17 Therefore, it is of great interest to develop highly selective, efficient, and durable electrocatalysts that can replace noble metals for electrochemical H₂O₂ synthesis.

Single-atom electrocatalysts (SACs) benefit from the desirable properties of isolated metal sites to enable the end-on type of adsorption of O₂, thereby reducing the probability of the O–O bond cleavage and improving their selectivity for electrochemical H₂O₂ synthesis.^18,19 At the same time, the realization of atomical dispersion, the utilization of the unique geometry, and the regulation of the electronic structure of SACs are all conducive to the further improvement of the electrocatalytic activity and selectivity.^7,15 Therefore, the study of carbon-supported non-noble metal single-atom electrocatalysts with tailored coordination environments for electrocatalytic H₂O₂ synthesis is of great interest. To date, non-noble metal SACs (e.g., Fe, Mn, Cu, Zn, Mo, Co, and Ni) based on carbon-based supports have emerged as promising schemes for the electrocatalytic synthesis of H₂O₂ in alkaline solutions.^20–25 However, H₂O₂ has the problem of easy decomposition in alkaline solution, which is detrimental to the preservation of H₂O₂.¹⁰ Therefore, the electrosynthesis of H₂O₂ under neutral conditions is very attractive for many applications where neutral pH operation is required, such as H₂O₂ fuel cells, disinfection, environmental purification, and other on-site H₂O₂ applications.²⁶ However, the low activity for the electrochemical synthesis of H₂O₂ at neutral pH has become an important factor restricting its development.^27,28 This is due to the relatively low concentration of hydroxide ions in the neutral electrolyte hindering the efficient incorporation of OOH*, a key intermediate of the 2e⁻ ORR, resulting in the inadequate performance of non-noble metal single-atom electrocatalysts with high overpotentials and low selectivity.^29,30 Although the selectivity of carbon-supported non-noble metal SACs for the 2e⁻ ORR can be enhanced by the selection of appropriate precursors such as metal–organic frameworks, the preparation of such electrocatalysts requires expensive precursors (such as terephthalic acid) and harsh synthesis conditions (such as strong acids).^30,31 Furthermore, the resulting carbon-supported non-noble metal SACs still suffer from inefficiencies.^31,32 Therefore, the selection, optimization, and modification of carbon-based supports are important for enhancing the 2e⁻ ORR selectivity of non-noble metal SACs under neutral conditions because carbon-based materials are not only excellent supports but also excellent 2e⁻ ORR catalysts.

It is well known that nitrogen doping is a way to increase the 2e⁻ ORR capability of carbon-based catalysts.^31,33 There is a linear relationship between the activity of the 2e⁻ ORR and the nitrogen content of nitrogen-doped carbon materials.³⁴ Various carbon–nitrogen coordination structures exist in nitrogen-doped carbon materials, such as pyridine-N, pyrrole-N, and N-oxide.^11,35,36 The carbon–nitrogen coordination can effectively improve the adsorption energy of OOH* and generate more active sites, which can improve the selectivity for both H₂O₂ and the 2e⁻ ORR activity.³⁶ Some studies have shown that pyridine N is suitable for the 2e⁻ ORR, and it has been reported that carbon atoms near pyridine N are the active centers of the 2e⁻ ORR.¹ Recently, the formation of nitrogen-doped graphene oxide on a carbon paper was reported by Kim et al. The authors identified the active center of the catalyst through various characterization techniques and observed that the sp² carbon sites near the region of sp³ carbon after nitrogen doping were the catalytic active centers of the 2e⁻ ORR.^1,3 In addition, it was found that partially reduced graphene oxide achieved 99% H₂O₂ selectivity in acidic and basic solutions;^2,3 however, its activity for H₂O₂ electrosynthesis at neutral pH is still limited. Based on this, graphitic carbon nitride (g-C₃N₄) has infinite potential in electrocatalytic H₂O₂ synthesis via the 2e⁻ ORR because g-C₃N₄ nanosheets (CNNS) have abundant carbon–nitrogen coordination, including sp² carbon bond (N–C=N), sp³ carbon bond (C–NH_X) and the graphite-like carbon (C–C/C=C) and so on, which can create more active conditions for the 2e⁻ ORR and achieve the stable anchoring of isolated non-noble metal single atoms through the formation of metal–nitrogen bonds.^37,38 CNNS also has the advantages of being low-cost, non-toxic and harmless, with a large surface area and good stability of the carbon–nitrogen structure.^38,39

Herein, relying on a simple and universal two-step thermal oxidation etching method, transition metal single atoms (TM SA: Ni, Fe, Co, Mn, Cu, Zn) have been successfully anchored on CNNS (TM SA/CNNS) for efficient H₂O₂ synthesis via the 2e⁻ ORR in neutral electrolytes. Compared with CNNS, all the TM SA/CNNS exhibited high selectivity for the 2e⁻ ORR in the following order: Ni_0.10 SA/CNNS (98%) > Mn_0.10 SA/CNNS (95%) > Zn_0.10 SA/CNNS (93%) > Cu_0.10 SA/CNNS (90%) > Fe_0.10 SA/CNNS (86%) > Co_0.10 SA/CNNS (76%) > CNNS (61%). The high selectivity of TM SA/CNNS may be ascribed to the introduction of TM SA, which optimized the N–C=N (increased) and C–C/C=C (decreased) contents in CNNS, which can improve the selectivity of TM SA/CNNS for the 2e⁻ ORR. This demonstrates that the N–C=N site favors the 2e⁻ ORR, while the C–C/C=C site hinders the electrochemical synthesis of H₂O₂. The linear relationship between the content of the N–C=N bond in TM SA/CNNS and its 2e⁻ORR selectivity further supports these conclusions. In addition, TM SA further promotes the enhancement of 2e⁻ ORR activity. Secondly, the introduction of TM SA lowers the reaction potential of the 2e⁻ ORR and improves the electrocatalytic activity of TM SA/CNNS, which is one of the important reasons why H₂O₂ yields could reach industrial levels. All of these are verified by characterization methods including high-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM), X-ray photoelectron spectroscopy (XPS), density functional theory (DFT), and so on.

2. Results and discussion

2.1 Synthesis and characterization of electrocatalysts

The synthetic scheme for TM SA/CNNS (TM: Ni, Mn, Fe, Co, Cu, Zn) is as follows: three precursors (urea, dicyandiamide (DCDA), NiCl₂·6H₂O and other transition metal chlorides) were used. In the first step, urea, DCDA and transition metal chlorides were thoroughly and uniformly mixed by mechanical grinding. The obtained mixture was treated at 550 °C for 4 h in the atmosphere to obtain the initial TM SA/g-C₃N₄ nanocomposites. Finally, the initial TM SA/g-C₃N₄ nanocomposites were further heated to 500 °C for 2 h with 10% Ar/H₂ flow to achieve TM SA/g-C₃N₄ nanosheets (TM SA/CNNS). Based on the excellent activity and selectivity of Ni SA/CNNS electrocatalysts, a series of samples with different Ni contents (0.05, 0.10, and 0.20 mmol) were also synthesized to explore the optimal loading of Ni single atoms. They were denoted as CNNS, Ni_0.05 SA/CNNS, Ni_0.10 SA/CNNS, Ni_0.20 SA/CNNS, respectively. Among them, Ni_0.10 SA/CNNS exhibited the best 2e⁻ ORR activity and selectivity. Similar to the method of Ni_0.10 SA/CNNS, 0.10 mmol of transition metal chlorides (TM: Mn, Zn, Cu, Fe, Co) were used as the metal source to prepare Mn_0.10 SA/CNNS, Fe_0.10 SA/CNNS, Co_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Zn_0.10 SA/CNNS; all of these were named TM SA/CNNS. It can be seen that the synthesis method is simple, safe, universal and suitable for the industrial production of TM SA on CNNS.

The XRD results of CNNS and TM SA/CNNS are displayed in Fig. 1(a) and Fig. S1 (ESI†). The characteristic peaks exhibited at 13.2° and 27.6°, respectively, correspond to the in-plane structural packing motif of the continuous heptazine network and the interlayer stacking of aromatic segments in g-C₃N₄, which belongs to the (100) and (002) diffraction plane of g-C₃N₄ (JCPDS no. 87-1526).^37,40 There were no obvious diffraction peaks of Ni, Mn, Zn, Cu, Fe, and Co among the XRD, which might be due to the lower transition metal content or the formation of transition metal single atoms or clusters on CNNS. Based on XRD patterns, the Debye–Scherrer equation was used to calculate the crystallite size of CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS. As shown in Table S1 (ESI†), the crystallite sizes of CNNS, Ni_0.10 SA/CNNS, Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS and Co_0.10 SA/CNNS were 3.45 nm, 4.10 nm, 5.88 nm, 5.27 nm, 5.91 nm, 5.44 nm, 4.75 nm, respectively. The results show that TM SA enhanced the crystallite size of CNNS, but the change in the crystallite size of TM SA/CNNS is inconsistent with its corresponding electrochemical properties. It can be seen that with the increase in crystallite size of CNNS via TM SA, it is possible to achieve improved electrocatalytic performance but the crystallite size is not the main cause of the electrocatalytic performance.

Fig. 1 (a) XRD patterns of CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS. TEM images (insets show the HRTEM images) of (b) CNNS, (c) Ni_0.10 SA/CNNS, (d) Mn_0.10 SA/CNNS, (e) Zn_0.10 SA/CNNS, (f) Cu_0.10 SA/CNNS, (g) Fe_0.10 SA/CNNS and (h) Co_0.10 SA/CNNS. HAADF-STEM images and EDS mapping of (i) Ni_0.10 SA/CNNS, (j) Mn_0.10 SA/CNNS, (k) Zn_0.10 SA/CNNS, (l) Cu_0.10 SA/CNNS, (m) Fe_0.10 SA/CNNS and (n) Co_0.10 SA/CNNS.

As shown in Fig. S2a (ESI†), the Raman spectroscopy of CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS was conducted using a 785 nm laser. CNNS, Ni_0.10 SA/CNNS, Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS and Co_0.10 SA/CNNS exhibited similar Raman vibration modes at 900–1700 cm⁻¹. The Raman spectra at 900–1400 cm⁻¹ were related to the typical heptazine units. The peaks located at 987 cm⁻¹, 1168 cm⁻¹, 1294 cm⁻¹, 1429 cm⁻¹, 1520 cm⁻¹ and 1646 cm⁻¹ originated from the vibration modes of CN heterocycles in CNNS. In detail, the diffraction peak at 1294 cm⁻¹ is caused by the =C (sp²) bending vibration. The two peaks at 1429 cm⁻¹ and 1520 cm⁻¹ are assigned to the D and G bands of a typical graphitic structure, which are related to the sp² carbon materials of CNNS. This means that the TM SA does not make other significant changes to the CNNS.

Further research on materials by Fourier Transform infrared (FT-IR) spectroscopy revealed the functional groups present in CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS (Fig. S2b, ESI†). Several major bands centered at about 700–1000 cm⁻¹, 1100–1800 cm⁻¹ and 2800–3700 cm⁻¹ were observed. In comparison with CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS showed similar spectra. The absorption peak at 812 cm⁻¹ is caused by the out-of-plane bending vibration of the triazine/heptazine ring in CNNS. The band at 888 cm⁻¹ is due to the deformation vibrations of the N–H bonds. The absorption peaks at 1237 cm⁻¹, 1323 cm⁻¹ 1402 cm⁻¹, 1461 cm⁻¹, 1569 cm⁻¹ and 1643 cm⁻¹ arise from the stretching vibrations of the C–N bonds. Several broad bands near 3182 cm⁻¹ are caused by stretching vibrations of the N–H or O–H bonds of the residual amino groups or adsorbed H₂O molecules. These show that all materials maintain the classical structure of g-C₃N₄.

The morphologies and structures of CNNS and TM SA/CNNS were characterized via SEM, TEM, and HRTEM. As shown in Fig. 1(b)–(h) and Fig. S3, S4 (ESI†), CNNS and TM SA/CNNS have similar structures. The SEM images (Fig. S3 and S4, ESI†) show that CNNS and TM SA/CNNS are composed of ultrathin flexible nanosheets with wrinkles, and a large number of pore structures are observed in the TEM images (Fig. 1(b)–(h)). All these indicate the formation of ultrathin flexible nanosheets with porous structures in CNNS and TM SA/CNNS. The insets of Fig. 1(b)–(h) show the HRTEM of these electrocatalysts, no obvious transition metal nanoparticles were observed on the surface of CNNS. However, the elemental mapping of TM SA/CNNS (Fig. 1(i)–(n)) shows that C, N, and transition metals (Ni, Mn, Zn, Cu, Fe, and Co) are present and uniformly distributed. These prove that Ni, Mn, Zn, Cu, Fe, and Co may also exist in TM SA/CNNS in the form of single atoms or clusters. From HAADF-STEM images, the isolated bright dots are observed in TM SA/CNNS (Fig. 2(a)–(f)), which are identified as transition metal single atoms considering the atomic number of transition metals is larger than that of C and N.^7,19,41 In addition, XAFS (X-ray absorption fine structure) was used to further demonstrate the presence of Ni single atoms in Ni_0.10 SA/CNNS, which has the best 2e⁻ ORR performance in all TM SA catalysts. The detailed electronic structures and coordination information of Ni single atoms in Ni_0.10 SA/CNNS were also obtained. The Ni k-edge XANES (X-ray absorption near-edge structure) spectra in Fig. 2(g) show that the white-line intensity of Ni_0.10 SA/CNNS is between Ni foil and NiO, indicating that Ni single atoms are positively charged. The EXAFS (extended X-ray absorption fine structure) data in Fig. 2(h) show that the peak at 2.18 Å in Ni foil is a metallic Ni–Ni bond. This is not found in Ni_0.10 SA/CNNS, indicating that Ni does not exist as clusters or nanoparticles. Meanwhile, the main peak of Ni_0.10 SA/CNNS at 1.44 Å is assigned to the Ni–N bond. As shown in Fig. S5 and Table S2 (ESI†), the fitting results of Ni_0.10 SA/CNNS show Ni–N₃ coordination. Combined with HAADF-STEM, all these results indicate that Ni in Ni_0.10 SA/CNNS does exist in the form of single atoms.

Fig. 2 HAADF-STEM images of (a) Ni_0.10 SA/CNNS, (b) Mn_0.10 SA/CNNS, (c) Zn_0.10 SA/CNNS, (d) Cu_0.10 SA/CNNS, (e) Fe_0.10 SA/CNNS and (f) Co_0.10 SA/CNNS. (g) The normalized XANES spectra at the Ni K-edge of Ni foil, Ni_0.10 SA/CNNS and NiO. (h) Ni R-space of the EXAFS for Ni foil, Ni_0.10 SA/CNNS and NiO. (i) Ni k-space of the EXAFS for Ni foil, Ni_0.10 SA/CNNS and NiO.

To further clarify the interactions between the TM SA and CNNS, the chemical states of CNNS and TM SA/CNNS were investigated by XPS. Fig. S6 and S7 (RDI), show that CNNS and TM SA/CNNS are composed of C and N. At the same time, Ni, Mn, Zn, Cu, Fe, and Co were found in TM SA/CNNS, respectively. The spectra of Ni 2p (Fig. S7a, ESI†) can be decomposed into four peaks at 853.40 and 870.90 eV, 860.10, and 878.80 eV, which are ascribed to Ni 2p_3/2 and Ni 2p_1/2, and the satellite peaks of Ni 2p_3/2 and Ni 2p_1/2, which are similar to the binding energy of NiO.^39,40 Similarly, the chemical states of atomically dispersed Mn, Zn, Cu, Fe and Co species in TM SA/CNNS are close to (II and III), (II), (II), (II and III), and (II) (Fig. S7b–f, ESI†), respectively.^39,42–50 All of these prove the existence of the isolated transition metal single atoms. The high-resolution C 1s spectra (Fig. 3 and Fig. S8, ESI†) show three component peaks at 284.7 eV, 284.40 eV, and 288.15 eV for CNNS and TM SA/CNNS, which correspond to C–C/C=C, C–NH_X and N–C=N, respectively.^37,49 The typical components around 284.7 eV, 284.40 eV, and 288.15 eV in CNNS and TM SA/CNNS are almost the same. These results indicate that the chemical state of the carbon remains almost unchanged, meaning that carbon atoms hardly interact with the metal ions. Fig. S9 and S10 (ESI†) display the N 1s spectra of CNNS and TM SA/CNNS; four peaks were observed at 398.60, 399.80, 400.94, and 404.30 eV. The peaks at 398.60 eV, 399.80, 400.94, and 404.30 eV correspond to the N sp² bond, the tertiary nitrogen group (N–C₃), the quaternary N three carbon atom amino functional group (N–H_X) in the aromatic ring and π excitation, respectively.^38,40 It is notable that nitrogen peaks appeared at around 398.20, 397.82, 398.10, 297.90, 397.90, and 398.00 eV in the N 1s spectra of TM SA/CNNS, which could be attributed to N–TM.^40,44 Besides, as shown in Table S3 (ESI†), the contents of N–C=N in CNNS and TM SA/CNNS follow the order of Ni_0.10 SA/CNNS (92%) > Mn_0.10 SA/CNNS (90%) > Zn_0.10 SA/CNNS (86%) > Cu_0.10 SA/CNNS (85%) > Fe_0.10 SA/CNNS (80%) > CNNS (73%) > Co_0.10 SA/CNNS (67%). The C–C/C=C content has the exact opposite pattern. These indicate that the introduction of TM SA increases the content of N–C=N sites and reduces the C–C/C=C content. This may be beneficial to improve the selectivity of TM SA/CNNS, which will be further verified by electrochemical properties, as well as DFT calculations.

Fig. 3 High-resolution XPS of the C 1s spectra of (a) Ni_0.10 SA/CNNS, (b) Mn_0.10 SA/CNNS, (c) Zn_0.10 SA/CNNS, (d) Cu_0.10 SA/CNNS, (e) Fe_0.10 SA/CNNS and (f) Co_0.10 SA/CNNS.

2.2 The electrochemical properties of catalysts

To verify the above conclusions, the electrochemical properties of CNNS and TM SA/CNNS were analyzed. The 2e⁻ ORR performances of CNNS and TM SA/CNNS were determined in 0.1  M PBS. As shown in Fig. S11 (ESI†), the linear scan voltammogram (LSV) profiles were obtained by subtracting the capacitive current in the N₂-saturated electrolyte from the O₂-saturated ORR current. The most promising Ni SA/CNNS electrocatalysts were optimized by adjusting the Ni content (Fig. S12, ESI†). Among them, Ni_0.10 SA/CNNS exhibited the highest 2e⁻ ORR activity and selectivity (the selectivity was ∼98% at 0.3 V vs. RHE; the faradaic efficiency (FE%) was ∼97%). Based on this, other TM SA/CNNS (Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS, and Co_0.10 SA/CNNS) were also synthesized. Fig. 4(a) shows the LSV curves of CNNS and TM SA/CNNS, which contain the disk and ring currents. Compared with CNNS (∼0.54 V vs. RHE) and Ni_0.10 SA/CNNS (∼0.55 V vs. RHE), other TM SA/CNNS have similar initial potentials. These are below the equilibrium potential of the 2e⁻ ORR (0.70 V vs. RHE), indicating that CNNS and TM SA/CNNS are close to the thermodynamic equilibrium potential of the 2e⁻ ORR, which was also confirmed by the ORR selectivity. The selectivity, n, and FE% of CNNS and TM SA/CNNS are presented in Fig. 4(b) and Fig. S13 (ESI†). The selectivities of CNNS, Ni_0.10 SA/CNNS, Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS, and Co_0.10 SA/CNNS are ∼61%, ∼98%, ∼95%, ∼93%, ∼90%, ∼86%, and ∼76% at 0.3 V vs. RHE (Fig. 4(b)), where n is ∼2.73, ∼2.02, ∼2.11, ∼2.14, ∼2.19, ∼2.27, and ∼2.48, (Fig. S13a, ESI†) respectively. FE% is ∼44%, ∼97%, ∼89%, ∼86%, ∼82%, ∼76%, and ∼61%, respectively (Fig. S13b, ESI†). These reflect that the electrocatalytic activities of CNNS and TM SA/CNNS follow the order Ni_0.10 SA/CNNS > Mn_0.10 SA/CNNS > Zn_0.10 SA/CNNS > Cu_0.10 SA/CNNS > Fe_0.10 SA/CNNS > Co_0.10 SA/CNNS > CNNS, which is consistent with the changes in the N–C=N and C–C/C=C contents in the high-resolution C 1s spectra, except for Co_0.10 SA/CNNS. Tafel plots of CNNS and TM SA/CNNS (Fig. 4(c)) were calculated from the RRDE data. TM SA/CNNS exhibited significantly faster kinetics for the 2e⁻ ORR than CNNS under neutral conditions in the order of Ni_0.10 SA/CNNS > Mn_0.10 SA/CNNS > Zn_0.10 SA/CNNS > Cu_0.10 SA/CNNS > Fe_0.10 SA/CNNS > CNNS > Co_0.10 SA/CNNS. The electrochemically active surface areas (ECSA) of CNNS and TM SA/CNNS were measured via the electrochemical double-layer capacitance (Fig. 4(d) and Fig. S14, ESI†). The ECSA of Ni_0.10 SA/CNNS, Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS and Co_0.10 SA/CNNS are 21.9 mF cm⁻², 21.1 mF cm⁻², 19.0 mF cm⁻², 18.0 mF cm⁻², 16.4 mF cm⁻² and 15.4 mF cm⁻², respectively. These results also clearly display the higher electrochemical activity of TM SA/CNNS for the 2e⁻ ORR as compared with CNNS (7.80 mF cm⁻²); Ni_0.10 SA/CNNS showed the highest electrochemical activity. This means that the isolated TM SA could improve the kinetics of the 2e⁻ ORR and the electrochemically active area of TM SA/CNNS. The activities of the electrocatalytic H₂O₂ reduction reaction (H₂O₂RR) of CNNS and TM SA/CNNS were measured in N₂-saturated 0.1 M PBS containing 10 mM H₂O₂. For Fig. 4(e), when the potential of Ni_0.10 SA/CNNS is 0.3 V vs. RHE, the H₂O₂RR current density of Ni_0.10 SA/CNNS is about −0.03 mA cm⁻². Meanwhile, the H₂O₂RR current densities of Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS, and Co_0.10 SA/CNNS were about −0.042 mA cm⁻², −0.050 mA cm⁻², −0.087 mA cm⁻², −0.125 mA cm⁻², and −0.143 mA cm⁻², respectively. This means that TM SA/CNNS have poorer activity towards the H₂O₂RR as compared to CNNS (−0.18 mA cm⁻² at 0.3 V vs. RHE), which indicates that TM SA/CNNS are more conducive to the retention of H₂O₂ during the 2e⁻ ORR. Therefore, the introduction of the isolated TM SA can improve the yield of H₂O₂ and is also beneficial to the synthesis of high-concentration H₂O₂ solutions. From the Tafel slopes, ECSA and H₂O₂RR, it was found that the introduction of the isolated TM SA effectively enhanced the activity and selectivity for the 2e⁻ ORR. Combined with the relationship between the change in carbon species and H₂O₂ selectivity in electrocatalysts (Fig. 4(f)), it was observed that the isolated TM SA could optimize the structure of CNNS, adjust the content of N–C=N and C–C/C=C in CNNS and improve the selectivity for the 2e⁻ ORR. Although the N–C=N content of Co_0.10 SA/CNNS is significantly lower than that of CNNS, it still possesses better activity and selectivity for the 2e⁻ ORR. It might be due to the introduction of Co SA optimising the charge distribution, the lowering of the reaction potential of the 2e⁻ ORR and the improvement of the electrochemically active area, which can compensate for the loss of N–C=N content. This indicates that N–C=N plays an important role in improving the 2e⁻ ORR selectivity, but it is not the only factor affecting the activity and selectivity of electrocatalysts.

Fig. 4 Electrocatalytic performances of CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS in 0.1 M PBS: (a) ORR disk current density together with the ring currents at a fixed potential of 1.20 V vs. RHE. (b) H₂O₂ selectivity (H₂O₂%). (c) Tafel plots. (d) Current density differences at 0.75 V vs. the scan rates. (e) LSV curves of H₂O₂ reduction reaction recorded in N₂-saturated 0.1 M PBS containing 10 mM H₂O₂. (f) The atomic percentages of carbon species in high-resolution C 1s spectra of CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS and their corresponding H₂O₂ selectivities at 0.3 V vs. RHE. (g) Stability measurement of Ni_0.10 SA/CNNS at a fixed disk potential of 0.3 V vs. RHE. (h) The H₂O₂ production rate of CNNS, Ni_0.10 SA/CNNS and other TM SA/CNNS. (i) Stability measurement and H₂O₂ concentration of Ni_0.10 SA/CNNS at 0.3 V vs. RHE.

The stability of Ni_0.10 SA/CNNS was also further assessed. The chronoamperometric currents of the ring and disk electrodes were recorded in O₂-saturated 0.1 M PBS, when they were biased at 1.2 and 0.3 V vs. RHE, respectively. Fig. 4(g) shows that both i_disk and i_ring remained stable through the 16 h test with negligible decay. The selectivity for H₂O₂ was maintained in the range of 96.6–98.8% during the entire test (Fig. S15, ESI†), indicating that the Ni_0.10 SA/CNNS has excellent electrochemical stability. The physical stability of Ni_0.10 SA/CNNS was also tested by comparing the 2e⁻ ORR activity and selectivity of the Ni_0.10 SA/CNNS ink before and after 68 days at room temperature. Fig. S16 (ESI†) shows that the H₂O₂ selectivity of Ni_0.10 SA/CNNS after standing for 68 days was 94%, which is 4% lower than the original Ni_0.10 SA/CNNS (the selectivity for H₂O₂ was 98%), indicating that the Ni_0.10 SA/CNNS has high physical stability. Lastly, to demonstrate the feasibility of CNNS and TM SA/CNNS for the 2e⁻ ORR in the fuel cell, we deposited CNNS and TM SA/CNNS electrocatalysts (1 mg cm⁻²) on carbon paper as the working electrode and constructed an H-cell for the direct synthesis of H₂O₂. The titration method was used to obtain the amount of H₂O₂ produced in an H-cell (Fig. S17–S24, ESI†). As shown in Fig. 4(h), the H₂O₂ production rate of Ni_0.10 SA/CNNS is also highly promising with up to ∼503.0 mmol g_cat⁻¹ h⁻¹ at 0.3 V vs. RHE, which is 14.6 times that of CNNS (∼34.4 mmol g_cat⁻¹ h⁻¹). Therefore, the H₂O₂ production rates of Mn_0.10 SA/CNNS, Zn_0.10 SA/CNNS, Cu_0.10 SA/CNNS, Fe_0.10 SA/CNNS, and Co_0.10 SA/CNNS (Fig. 4(h)) were also measured, which were ∼425.5 mmol g_cat⁻¹ h⁻¹, ∼385.9 mmol g_cat⁻¹ h⁻¹, ∼332.0 mmol g_cat⁻¹ h⁻¹, ∼205.6 mmol g_cat⁻¹ h⁻¹, and ∼188.4 mmol g_cat⁻¹ h⁻¹ at 0.3 V vs. RHE, respectively. Compared with CNNS, TM SA/CNNS exhibited a significantly improved yield of H₂O₂, which is consistent with the test results of the three-electrode system. Fig. 4(i) shows the stability measurements and H₂O₂ concentration of Ni_0.10 SA/CNNS at 0.3 V vs. RHE for 24 h. In the i–t test, the change in the current density was negligible, indicating that the Ni_0.10 SA/CNNS has very good stability. The accumulated H₂O₂ concentration increased with the operation time and approached 4.5 g L⁻¹ for 24 h (the titration method was also used, as shown in Fig. S25 (ESI†)). The rate of H₂O₂ was also maintained in the range of 0.43–0.51 mol g⁻¹ h⁻¹ (Fig. S26, ESI†) in the i–t test, indicating that the production of H₂O₂ increased linearly with time, which further demonstrated that Ni_0.10 SA/CNNS has high activity, selectivity, and stability in the electrochemical synthesis of H₂O₂. Consequently, the TM SA/CNNS are considered the candidates with the greatest potential for electrocatalytic H₂O₂ preparation due to their high electrochemical activity and selectivity, low H₂O₂RR activity, and excellent stability, which surpass most of the reported work (Table S4, ESI†).

2.3 The DFT calculations of catalysts

The above results indicate that N–C=N is the main active site for the 2e⁻ ORR, while C–C/C=C sites can hinder the electrochemical synthesis of H₂O₂. The mechanisms of C–C/C=C and N–C=N sites in CNNS were investigated by DFT calculations (Fig. 5(a)–(d)). Fig. 5(e) and (f) show the Gibbs free energies (ΔG) of C–C/C=C and N–C=N at different potentials (U). The ΔG of OOH* on C–C/C=C is about −3.70 eV. The high free energy is more favorable for the adsorption of O₂ and the breakage of the O–O bond, which inhibits the formation of the OOH* intermediates in the 2e⁻ ORR process. The ΔG of OOH* on N–C=N is ∼0.46 eV, and the lower free energy is favorable for the formation of OOH* in the 2e⁻ ORR process. These verify that N–C=N is an important active site for the electrochemical preparation of H₂O₂, while C–C/C=C hinders the formation of H₂O₂. Therefore, it is clear that N–C=N in CNNS is more inclined to the 2e⁻ ORR but it needs to overcome a potential barrier of 0.46 eV during the formation of OOH*, which affects the reaction rate of the 2e⁻ ORR. To further understand the role of TM SA, a model of Ni SA/CNNS was constructed. As shown in Fig. 6(a) and (b), Ni SA in CNNS can improve the adsorption energy of N–C=N sites for OOH* (where the yellow part shows the increase of charge density and the blue part shows the decrease of charge density). It is known that the rate-determining step (RDS) of the ORR depends on the thermodynamic step with the smallest change in free energy. From the Gibbs free energy of Ni SA/CNNS (Fig. 6(c)), the ΔG of OOH* on Ni SA/CNNS is −0.65 eV in the 2e⁻ ORR and all reaction steps are thermodynamically decreasing, so the process of OOH* is the RDS of the 2e⁻ ORR. ΔG of OOH* is closer to the equilibrium potential of the 2e⁻ ORR (U_equilibrium = 1.4/2 = 0.7), indicating that Ni SA/CNNS has excellent capability for the 2e⁻ ORR. During the 4e⁻ ORR, the overpotential of Ni SA/CNNS is ∼0.58 eV (Fig. 6(d)), and the high overpotential leads to its slow reaction kinetics. The introduction of Ni single atoms improves the activity and selectivity for the 2e⁻ ORR and hinders the 4e⁻ ORR. Fig. 6(e) and (f) show the Gibbs free energy of Ni SA/CNNS and CNNS in the 2e⁻ ORR pathway, respectively. On comparing the Gibbs free energy of Ni SA/CNNS and CNNS, it was found that Ni SA/CNNS has a negligible barrier potential in the RDS of the 2e⁻ ORR. This indicates that Ni SA can improve the adsorption energy of N–C=N sites for OOH*, and reduce the overpotential of OOH* intermediates, thus enhancing the reaction rate of the 2e⁻ ORR.

Fig. 5 (a)–(d) The optimized configuration and (e), (f) calculated free energy (U = 0 and 0.7 eV) of C–C/C=C and N–C=N species in CNNS for the synthesis of H₂O₂.

Fig. 6 (a) and (b) The optimized configuration of Ni_0.10 SA/CNNS. (c) and (d) The calculated free energy of Ni_0.10 SA/CNNS for the synthesis of H₂O and H₂O₂. (e) and (f) The calculated free energy (U = 0 and 0.7 eV) of N–C=N species in CNNS and Ni_0.10 SA/CNNS for the synthesis of H₂O₂.

3. Conclusions

TM SA/CNNS have been successfully synthesized via a simple and universal two-step thermal oxidation etching method for electrochemical H₂O₂ synthesis in neutral electrolytes. Among them, Ni_0.10 SA/CNNS showed the highest selectivity, which is up to ∼98%. The H₂O₂ yield of Ni_0.10 SA/CNNS is ∼503 mmol g_cat⁻¹ h⁻¹, which is ∼14.6 times higher than that of CNNS (∼34.4 mmol g_cat⁻¹ h⁻¹). Notably, compared with CNNS, TM SA/CNNS showed efficient electrochemical H₂O₂ production. Combined with the characterization of XPS and HAADF-STEM, electrochemical tests and theoretical calculations, it was found that the N–C=N and C–C/C=C sites in CNNS belong to the 2e⁻ ORR and 4e⁻ ORR pathways, respectively. Therefore, the highly efficient electrochemical activities exhibited by TM SA/CNNS are due to the isolated TM SA on CNNS, which effectively enhances the activity of the 2e⁻ ORR of electrocatalysts. On the other hand, the introduction of TM SA optimizes the structure of CNNS and promotes the increase of N–C=N sites and the decrease of C–C/C=C sites in CNNS, thereby further improving the 2e⁻ ORR selectivity. The results indicate that TM SA/CNNS are promising candidates for the 2e⁻ ORR, and provide more reference ideas for the preparation of H₂O₂ by 2e⁻ ORR.

Author contributions

Hongcen Yang and Fei Ma: conceptualization, methodology, data curation, writing – original draft, formal analysis, investigation, and visualization. Niandi Lu and Kun Tao are in the analysis of data fitting. Hong Zhang performed HAADF-STEM analysis. Shuhao Tian and Guo Liu: investigation, visualization, and so ware analysis. Ying Wang, Zhixia Wang and Di Wang: validation, data curation, and investigation. All authors discussed the results and participated in writing the manuscript. Shanglong Peng directed the research.

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

All the DFT calculation works were supported by the Supercomputing Center of Lanzhou University. All SEM, TEM and HAADF-STEM were supported by the Electron Microscopy Centre of Lanzhou University. This work was financially supported by the Gansu Provincial Natural Science Foundation of China (No. 21JR7RA493, 17JR5RA198, 2020HZ-2, 21JR7RA470), the Cooperation project of Gansu Academy of Sciences (2020HZ-2), the Fundamental Research Funds for the Central Universities (No. lzujbky-2018-119, lzujbky-2018-ct08, lzujbky-2019-it23), and Key Areas Scientific and Technological Research Projects in Xinjiang Production and Construction Corps (No. 2018AB004). All the DFT calculation works were supported by the Supercomputing Center of Lanzhou University. All SEM, TEM and HAADF-STEM were supported by the Electron Microscopy Center of Lanzhou University.

References

1. Y. Bu, Y. Wang, G. F. Han, Y. Zhao, X. Ge, F. Li, Z. Zhang, Q. Zhong and J.-B. Baek, Adv. Mater., 2021, 33, 2103266.
2. K. Dong, Y. Lei, H. Zhao, J. Liang, P. Ding, Q. Liu, Z. Xu, S. Lu, Q. Li and X. Sun, J. Mater. Chem. A, 2020, 8, 23123–23141.
3. N. Wang, S. Ma, P. Zuo, J. Duan and B. Hou, Adv. Sci., 2021, 8, 2100076.
4. F. Zhang, Y. Zhu, C. Tang, Y. Chen, B. Qian, Z. Hu, Y. C. Chang, C. W. Pao, Q. Lin, S. A. Kazemi, Y. Wang, L. Zhang, X. Zhang and H. Wang, Adv. Funct. Mater., 2022, 32, 2110224.
5. W. Zhou, L. Xie, J. Gao, R. Nazari, H. Zhao, X. Meng, F. Sun, G. Zhao and J. Ma, Chem. Eng. J., 2021, 410, 128368.
6. X. Yang, Y. Zeng, W. Alnoush, Y. Hou, D. Higgins and G. Wu, Adv. Mater., 2022, 34, 2107954.
7. E. Zhang, L. Tao, J. An, J. Zhang, L. Meng, X. Zheng, Y. Wang, N. Li, S. Du, J. Zhang, D. Wang and Y. Li, Angew. Chem., Int. Ed., 2022, 61, e202117347.
8. C. Zhang, R. Lu, C. Liu, J. Lu, Y. Zou, L. Yuan, J. Wang, G. Wang, Y. Zhao and C. Yu, Adv. Sci., 2022, 9, 2104768.
9. C. Yan, Q. Ma, F. Wang, L. Zhou, X. Lv, J. Du, B. Zheng and Y. Guo, Chem. Eng. J., 2022, 433, 133651.
10. H. Li, P. Wen, D. S. Itanze, Z. D. Hood, S. Adhikari, C. Lu, X. Ma, C. Dun, L. Jiang, D. L. Carroll, Y. Qiu and S. M. Geyer, Nat. Commun., 2020, 11, 3928.
11. S. Yang, A. Verdaguer-Casadevall, L. Arnarson, L. Silvioli, V. Čolić, R. Frydendal, J. Rossmeisl, I. Chorkendorff and I. E. L. Stephens, ACS Catal., 2018, 8, 4064–4081.
12. A. Verdaguer-Casadevall, D. Deiana, M. Karamad, S. Siahrostami, P. Malacrida, T. W. Hansen, J. Rossmeisl, I. Chorkendorff and I. E. L. Stephens, Nano Lett., 2014, 14, 1603–1608.
13. J. S. Jirkovský, I. Panas, E. Ahlberg, M. Halasa, S. Romani and D. J. Schiffrin, J. Am. Chem. Soc., 2011, 133, 19432–19441.
14. C. Yang, S. Bai, Z. Yu, Y. Feng, B. Huang, Q. Lu, T. Wu, M. Sun, T. Zhu, C. Cheng, L. Zhang, Q. Shao and X. Huang, Nano Energy, 2021, 89, 106480.
15. N. Wang, X. Zhao, R. Zhang, S. Yu, Z. H. Levell, C. Wang, S. Ma, P. Zou, L. Han, J. Qin, L. Ma, Y. Liu and H. L. Xin, ACS Catal., 2022, 12, 4156–4164.
16. R. Shen, W. Chen, Q. Peng, S. Lu, L. Zheng, X. Cao, Y. Wang, W. Zhu, J. Zhang, Z. Zhuang, C. Chen, D. Wang and Y. Li, Chem, 2019, 5, 2099–2110.
17. E. Jung, H. Shin, B. H. Lee, V. Efremov, S. Lee, H. S. Lee, J. Kim, W. Hooch Antink, S. Park, K. S. Lee, S. P. Cho, J. S. Yoo, Y. E. Sung and T. Hyeon, Nat. Mater., 2020, 19, 436–442.
18. X. Song, N. Li, H. Zhang, L. Wang, Y. Yan, H. Wang, L. Wang and Z. Bian, ACS Appl. Mater. Interfaces, 2020, 12, 17519–17527.
19. X. Li, S. Tang, S. Dou, H. J. Fan, T. S. Choksi and X. Wang, Adv. Mater., 2022, 34, 2104891.
20. C. Liu, H. Li, F. Liu, J. Chen, Z. Yu, Z. Yuan, C. Wang, H. Zheng, G. Henkelman, L. Wei and Y. Chen, J. Am. Chem. Soc., 2020, 142, 21861–21871.
21. C. Tang, Y. Jiao, B. Shi, J. N. Liu, Z. Xie, X. Chen, Q. Zhang and S. Z. Qiao, Angew. Chem., Int. Ed., 2020, 59, 9171–9176.
22. Z. Zhou, Y. Kong, H. Tan, Q. Huang, C. Wang, Z. Pei, H. Wang, Y. Liu, Y. Wang, S. Li, X. Liao, W. Yan and S. Zhao, Adv. Mater., 2022, 34, 2106541.
23. H. Wu, T. He, M. Dan, L. Du, N. Li and Z. Q. Liu, Chem. Eng. J., 2022, 435, 134863.
24. C. Liu, H. Li, J. Chen, Z. Yu, Q. Ru, S. Li, G. Henkelman, L. Wei and Y. Chen, Small, 2021, 17, 2007249.
25. J. Zhang, W. Liu, F. He, M. Song, X. Huang, T. Shen, J. Li, C. Zhang, J. Zhang and D. Wang, Chem. Eng. J., 2022, 438, 135619.
26. J. Lee, S. W. Choi, S. Back, H. Jang and Y. J. Sa, Appl. Catal., B, 2022, 309, 121265.
27. H. Shen, N. Qiu, L. Yang, X. Guo, K. Zhang, T. Thomas, S. Du, Q. Zheng, J. P. Attfield, Y. Zhu and M. Yang, Small, 2022, 18, 2200730.
28. M. Gao, Z. Y. Wang, Y. R. Yuan, W. W. Li, H. Q. Liu and T. Y. Huang, Chem. Eng. J., 2022, 434, 134788.
29. Q. Zhao, Y. Wang, W. H. Lai, F. Xiao, Y. Lyu, C. Liao and M. Shao, Energy Environ. Sci., 2021, 14, 5444–5456.
30. Y. Pang, K. Wang, H. Xie, Y. Sun, M. M. Titirici and G. L. Chai, ACS Catal., 2020, 10, 7434–7442.
31. Y. Sun, I. Sinev, W. Ju, A. Bergmann, S. Dresp, S. Kühl, C. Spöri, H. Schmies, H. Wang, D. Bernsmeier, B. Paul, R. Schmack, R. Kraehnert, B. Roldan Cuenya and P. Strasser, ACS Catal., 2018, 8, 2844–2856.
32. Q. Yang, W. Xu, S. Gong, G. Zheng, Z. Tian, Y. Wen, L. Peng, L. Zhang, Z. Lu and L. Chen, Nat. Commun., 2020, 11, 5478.
33. J. Zhang, G. Zhang, S. Jin, Y. Zhou, Q. Ji, H. Lan, H. Liu and J. Qu, Carbon, 2020, 163, 154–161.
34. Z. Lu, G. Chen, S. Siahrostami, Z. Chen, K. Liu, J. Xie, L. Liao, T. Wu, D. Lin, Y. Liu, T. F. Jaramillo, J. K. Nørskov and Y. Cui, Nat. Catal., 2018, 1, 156–162.
35. K. Jiang, S. Back, A. J. Akey, C. Xia, Y. Hu, W. Liang, D. Schaak, E. Stavitski, J. K. Nørskov, S. Siahrostami and H. Wang, Nat. Commun., 2019, 10, 3997.
36. Y. Wang, Y. Zhou, Y. Feng and X. Y. Yu, Adv. Funct. Mater., 2022, 32, 2110734.
37. H. Yang, J. Yin, R. Cao, P. Sun, S. Zhang and X. Xu, Sci. Bull., 2019, 64, 1510–1517.
38. H. Yang, R. Cao, P. Sun, J. Yin, S. Zhang and X. Xu, Appl. Catal., B, 2019, 256, 117862.
39. R. Du, K. Xiao, B. Li, X. Han, C. Zhang, X. Wang, Y. Zuo, P. Guardia, J. Li, J. Chen, J. Arbiol and A. Cabot, Chem. Eng. J., 2022, 441, 135999.
40. Z. Teng, W. Cai, W. Sim, Q. Zhang, C. Wang, C. Su and T. Ohno, Appl. Catal., B, 2021, 282, 119589.
41. W. Liu, J. Feng, R. Yin, Y. Ni, D. Zheng, W. Que, X. Niu, X. Dai, W. Shi, F. Wu, J. Yang and X. Cao, Chem. Eng. J., 2022, 430, 132990.
42. L. Wang, Y. Zhu, D. Yang, L. Zhao, H. Ding and Z. Wang, Appl. Surf. Sci., 2019, 488, 728–738.
43. J. Ma, N. Jia, C. Shen, W. Liu and Y. Wen, J. Hazard. Mater., 2019, 378, 120782.
44. S. Hu, X. Qu, P. Li, F. Wang, Q. Li, L. Song, Y. Zhao and X. Kang, Chem. Eng. J., 2018, 334, 410–418.
45. A. Malik, P. K. Prajapati, B. M. Abraham, S. Bhatt, P. Basyach and S. L. Jain, Catal. Sci. Technol., 2022, 12, 2688–2702.
46. C. Lai, H. Yan, D. Wang, S. Liu, X. Zhou, X. Li, M. Zhang, L. Li, Y. Fu, F. Xu, X. Yang and X. Huo, Chemosphere, 2022, 293, 133472.
47. Z. Huang, M. Shen, J. Liu, J. Ye and T. Asefa, J. Mater. Chem. A, 2021, 9, 14841–14850.
48. W. Niu, K. Marcus, L. Zhou, Z. Li, L. Shi, K. Liang and Y. Yang, ACS Catal., 2018, 8, 1926–1931.
49. X. Yan, Z. Jia, H. Che, S. Chen, P. Hu, J. Wang and L. Wang, Appl. Catal., B, 2018, 234, 19–25.
50. C. Wang, S. Zhao, X. Song, N. Wang, H. Peng, J. Su, S. Zeng, X. Xu and J. Yang, Adv. Energy Mater., 2022, 12, 2200157.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d2nh00564f

‡ Hongcen Yang and Fei Ma contributed equally to this work.

---