Advances of the functionalized carbon nitrides for electrocatalysis

Abstract

Over the past few decades, the design and development of carbon materials have occurred at a rapid pace. In particular, these porous graphene‐like carbon nitride materials have received considerable attention due to their superior structures and performances in the energy transformation field. In this review, nitrogenated holey two‐dimensional graphene and polymeric carbon nitride will be discussed in depth. The structural properties, synthetic methods, and applications including electrocatalytic reactions, such as hydrogen evolution reaction, oxygen reduction reaction, oxygen evolution reaction, and nitrogen reduction reaction, will be presented in detail. Finally, we will present the outlooks on the current obstacles to the development of carbon nitride materials. This comprehensive understanding will help guide and motivate researchers to develop and modify carbon nitride materials with better properties in the future.

INTRODUCTION

Two-dimensional (2D) materials have witnessed a rapid development in the fields of electronics, optoelectronics, catalysis, batteries, supercapacitors, biomedicine, sensing, and ferromagnetism in the recent decades due to the unique 2D extended topological structure.^1–7 Various novel 2D materials have been developed, such as hexagonal boron nitride, transition metal (TM) dichalcogenides, layered metal oxides, and carbon nitride (CN) materials. As one of the most popular 2D materials, graphene shows remarkable electron mobility and mechanical property.^8–12 However, graphene's semimetallic properties and the resulting loss of the electronic bandgap have severely hindered its wider applications.^13,14 Introduction of hetero elements in the P-block is a promising way to decorate the electronic structure and adjust the bandgap. Among the hetero elements, nitrogen is the most optimized because of the appropriate atomic size, five-electron valence structure (sp² hybridization), similar electronegativity, electron affinity beneficial to modifying the bandgap, and the adjustable conductivity.^15,16 Various types of doped nitrogen such as graphitic-N, pyridinic-N, pyrrolic-N, and oxidized-N could be assembled into the carbon sp² matrix, modifying the physical and chemical properties of the carbon materials.¹⁷

The development history of CN materials can be traced back to the 19th century naming melon that is a 1D chains of amine-linked heptazine units by Berzelius and Liebig.¹⁸ To date, many different types of CN materials have been synthesized experimentally or predicted theoretically,^19–23 indicating the advantage that applying nitrogen to replace part of the carbon in the carbon materials to form the CN topological matrix further improves their mechanical strength, electrical stability, and thermodynamic stability.²⁴ Among them, research on melon-based polymeric carbon nitride (PCN) materials has never stopped due to their relatively simple synthesis and good physical and chemical properties. However, many published papers have inappropriately named and elucidated melon-based polymerized carbon–nitrogen materials. The pseudographite phase produced in carbon–nitrogen materials preparation from carbon and nitrogen molecular compounds such as urea, dicyandiamide, and melamine is a major reason for the controversy. In addition, in the process of chemical or physical exfoliation of PCN, not only simple π–π accumulation but also hydrogen bonds will be destroyed. As PCN is a hydrogen-bonding material, it is difficult to successfully exfoliate it to obtain a layered material.^25,26 Moreover, its chemical composition is closer to C₆H₃N₉; thus, it should not be denoted as g-C₃N₄.²⁷ Therefore, on the basis of the report of Kessler's team, we name this type of melon-based CN materials PCN.²⁶ The optical bandgap of melon-based materials is 2.7 eV, which is also an important indicator to distinguish it from other CN materials.²⁸ After a long period of development, currently, a variety of preparation methods for PCN have been reported, among which the universal and classic strategies of the template-assisted method and supramolecular preorganization method (the self-templating method) will be discussed in this review. In realistic applications, PCN can be used as a qualified substitute for heterogeneous materials in catalysis, biomedicine, and gas detection. The excessively large intrinsic bandgap of the pristine PCN severely hinders its wide application in the electrochemical field. Therefore, PCN is usually modified by the methods of anchoring metals, doping nonmetals and compounding with other materials to fulfill the demands of applications.

The pursuit of CN materials with better intrinsic properties never stops. In recent years, a novel CN material with a clear structure has been successfully prepared by Baek who used the simple bottom-top wet chemical method in 2015 and named C₂N,²³ with rich nitrogen atoms at the edge of the uniformly dispersed hexagonal vacancy, which is highly sought after due to its excellent optical, thermal, mechanical, electronic, and magnetic properties. Its PBE-PAW (Perdew-Burke-Ernzerhof–projector augmented wave method) nonmagnetic semiconductor bandgap is apparently smaller than that of PCN (2.7 eV) and other CNs.^29–32 Being superior to PCN, the C₂N layer achieves higher electrical conductivity, resulting in its enormous potential in electrochemistry. Besides, the cohesive energy (E_coh) of C₂N (6.56 eV) per atom is smaller than that of graphene (7.95 eV per atom) and about twice that of silicon (3.71 eV per atom), indicating the excellent bonding strength of the C₂N monolayer network. Besides, based on the direct bandgap of 1.96 eV and the reduction potential through the electrochemical test, the LUMO (lowest unoccupied molecular orbital) is found to be −3.63 eV and the final calculated HOMO (highest occupied molecular orbital) is −5.59 eV. Furthermore, the bandgap of C₂N can be tuned in the range from 2.4 to 2.9 eV by adjusting the surface strain.^33–35 The valence charges of C and N in C₂N are different from that in PCN, which further implies that the two support materials will form different bonds with the anchored atoms and thus lead to different properties.^36–39

In this review, we systematically focus on the preparation and structural characterization of C2N and PCN, and specifically describe the application in electrocatalysis. At the end of this paper, it clearly presents the main challenges in the research of carbon nitride materials and the prospects in the future.

PREPARATION AND CHARACTERIZATIONS OF CN MATERIALS Preparation of PCN

Several synthesis methods of PCN have been previously reported, including chemical vapor deposition, plasma sputtering reaction deposition, and so forth. In this part, we mainly introduce the supramolecular preorganization approach and the hard/soft template-assisted methods due to their simple preparation process and the widespread applicatios.^24,40–51

The soft-template synthesis

As a relatively green method, the soft template process can not only achieve highly porous nanostructured PCN but also simplify the synthesis route.⁵² In soft-templating routes, the soft structure-directing agent and the guest species are naturally driven to reduce the interface to complete the cooperative assembly and form an ordered mesoporous arrangement.⁵³ The composition and properties of the template are critical to the formation of the microstructure. For instance, mesoporous PCN was successfully constructed by using P123 as a soft template.⁵⁴ The high-resolution transmission electron microscopy (HRTEM) image shows the mesoporous PCN sample prepared in the presence of P123 has a layered, plate-like surface morphology and a worm-like porous structure. When there is no P123 almost no such structure is observed in prepared sample. The results indicate that P123 is essential for the formation of a worm-like mesoporous structure (Figure 1A). The UV–vis diffuse reflection spectra of samples prepared with different amounts of P123 clearly shows a significant red shift in light absorption (Figure 1B). Figure 1C shows that the nitrogen adsorption/desorption isotherms of PCN samples prepared under different experimental conditions presenting type IV behavior, indicating the presence of mesopores. The Brunauer–Emmett–Teller (BET) surface area of the sample synthesized without P123 (27 m²/g) is much smaller than that with P123 (90 m²/g). The Barret–Joyner–Halender method was also used to calculate the pore size distribution of the samples prepared with P123, and it was highly uniform (Figure 1D).

Figure 1. (A) Transmission electron microscopy image of polymeric carbon nitride (PCN) prepared by using the Pluronic P123 surfactant. (B) UV–vis diffuse reflection spectra of mesoporous PCN prepared with different amounts of P123. (C) Nitrogen adsorption/desorption isotherm and (D) corresponding pore size distribution of mesoporous PCN prepared without or with the P123 surfactant (Reproduced with permission: Copyright 2012, Royal Society of Chemistry54). (E) Schematic illustration of the synthesis of ordered porous PCN. (F) Field emission scanning electron microscopy images of PCN. Reproduced with permission: Copyright 2011, Wiley46). (G) Synthetic path of PCN. (H) SEM images of production at different condensation times (Reproduced with permission: Copyright 2013, American Chemical Society55)

The hard template synthesis

The hard template (inorganic nanoparticles [NPs])-assisted method is also used to prepare nanostructured PCN materials by casting or replicating ordered mesoporous inorganic substrates. This method usually uses different morphologies of nanostructured silica substrate or silicon as a hard template. Many other synthetic methods are essentially derived from this method. The regular arrangement is determined by the ordered-structured template. The method avoids the randomness of the prepared material structure caused by the collaborative assembly process of the structure-directing agent and the guest, and therefore, it has been very successfully applied. Fukasawa et al.⁴⁶ used self-assembled silica nanospheres (SNS) with a close-packed configuration as a starting template for the synthesis of ordered porous PCN. The ordered porous PCN was obtained by the infiltration and polymerization of cyanamide within the interparticle voids of SNSs (uniform-sized [<50 nm] SNSs), followed by removing silica (Figure 1E). The scanning electron microscope (SEM) images show the inverse opal structure of the PCN products with the spherical pores (50 and 80 nm) connecting to the adjacent moiety determined by the size of the SNSs (Figure 1F).

Supramolecular preorganization

Supramolecular preorganization is a self-templating method in which the precursors spontaneously form stable aggregates by noncovalent bonds under equilibrium conditions.⁵⁵ Different from the template method, in this approach, due to the reversibility, specificity, and directionality of noncovalent interactions, hydrogen bonds play a vital role in the process of generating ordered structures.⁵⁶ Shalom et al.,⁵⁵ through supramolecular preorganization, found that PCN prepared with different solvents has different morphologies. Figure 1G shows that supramolecular aggregates composed of hydrogen bonds are a cyanuric acid–melamine (CM) complex. The PCN was obtained after calcining the precursor at 550°C for 8 or 12 h under a nitrogen atmosphere at a heating ramp of 2.3 K/min. A slight distinction in the morphologies depending on the reaction time can be observed in SEM images (Figure 1H). The pancake-like morphology of the CM complex before heating resulted in hollow replicas consisting of CNs). This represents the representative growth of material on the surface of the precursor structure. The hole size is strictly related to the primary structure size, and the order of PCN is driven by the solvent of the precipitated CM. The calcination time seems to govern the ripening and coarsening of structures. Moreover, SEM images of PCN formed from CM in either water or chloroform also clearly show different morphologies. This also shows that the type of solvent has an important influence on the morphology of the material.

Preparation of C₂N

Compared with PCN, C₂N has a relatively short research history and fewer synthetic methods. Herein, we introduce a relatively mature bottom-up wet chemical method for preparing C₂N.

A bottom-up wet-chemistry reaction

The preparation route of C₂N is shown in Figure 2A, which shows that by reacting hexaaminobenzene (HAB) hydrochloride and hexaketocyclohexane (HKH) octahydrate in N-methyl-2-pyrrolidone, the nitrogenated holey 2D (C₂N-h2D) crystal can be obtained.²³ Then, the processed sample solution is poured into the silica substrate for high-temperature calcination. It can be seen that the calcining sample is very bright. Finally, the sample film is transferred to a flexible polyethylene terephthalate substrate. Single transition metal (STM) clearly reveals the overall structure of the hexagonal image of C₂N, which is in good agreement with the theoretically derived image (Figure 2B,C). Figures 2D and 2E, respectively, show the band structure along the symmetry line of the Brillouin zone and the electronic density of states. Combined with the density functional theory (DFT), the effective bandgap of the material is calculated to be 1.70 eV, which is smaller than the optically measured value. This stems from an underestimation of the Kohn–Sham value of DFT. The benzene ring in C₂N–h2D is bridged by a pyrazine ring, which is composed of a six-membered D2h ring and two nitrogen atoms facing each other. Thus, the p-electron structure of the benzene ring is isolated, so a flat band different from the graphene and a cone-shaped band appear near the edges of the valence band and the conduction band (Figure 2F). The conduction band is composed of the dispersive band and local p orbitals of nitrogen atoms. The maximum value of the valence band is mainly derived from the nonbonded s state at the nitrogen atom. This feature can be used to design flat band edges to generate a useful value. Therefore, it can also provide complementary properties for the more widely studied graphene, which has a vanishing bandgap (conductor). The field-effect transistor devices prepared by C₂N-h2D crystals prove that the semimetal behavior of the C₂N-h2D crystal is attributable to the unintentional doping effect caused by the adsorption behavior of the porous structure, which also shows that it has adjustable electronic properties.

Figure 2. (A) Schematic illustration of the synthesis of the nitrogenated holey two-dimensional (2D) (C2N–h2D) crystal. The digital photographs represent an as-prepared C2N–h2D crystal; a solution-cast C2N–h2D crystal on a SiO2 surface after heat treatment and a C2N–h2D crystal film transferred onto a polyethylene terephthalate substrate. (B) An atomic-resolution single transition metal topography image of the C2N–h2D crystal on Cu (111). The top-left inset shows the structure of the C2N–h2D. The bottom-right inset shows the 2D fast Fourier transform. (C) Simulated image. (D) Band structure from the zone center to the M point of the 2D triangular lattice. (E) Density of electronic states. (F) Conduction band and the doubly degenerate valence-band maximum state (Reproduced with permission: Copyright 2015, The Nature Publishing Group23)

Structural characterization

Many advanced characterization techniques are now used to characterize the morphology, element composition, and the electronic state of CNs, which facilitate a more intuitive understanding of the structure and performance of materials. The combination of AFM and TEM can be used to initially observe and analyze the crystallinity, agglomeration morphology, pore size, and thickness of CNs.^57–59 HRTEM, with the advantage of higher resolution, can enable further intuitive exploration of the atomic coordination structure of CNs, such as amorphous graphite texture, porous structure, and defective carbon characteristics with a certain degree of crystallinity.⁶⁰ For example, STM clearly reveals the overall hexagonal structure of C₂N, in good agreement with the theoretically derived image.^23,60 Fourier transform infrared spectroscopy is normally applied to verify the degree of polymerization of CNs.^17,61,62 X-ray photoelectron spectroscopy (XPS) explores the surface composition of the material and indicates the hybrid state of carbon and nitrogen.^63,64 Powder X-ray diffraction (XRD) determines the crystal structure and long-range order. For example, the peak position can be used to judge whether the carbon and nitrogen materials are successfully synthesized, as well as calculate layer spacing.^28,65 The structures such as heptazine in CN materials can be further confirmed by solid-state NMR (nuclear magnetic resonance spectroscopy).^35,66,67 Electron paramagnetic resonance spectroscopy confirms the formation of nitrogen or oxygen vacancies.^66,68

APPLICATION OF CN MATERIALS

CNs are widely used in electrocatalysis fields such as oxygen reduction reaction (ORR), oxygen evolution reaction (OER), hydrogen evolution reaction (HER), nitrogen reduction reaction (NRR), and carbon dioxide reduction reaction (CO₂RR) due to their excellent adsorption capacity, high chemical stability, excellent electrical conductivity, and controllable structure. Therefore, the recent related researches are briefly summarized in Table 1 for reference and comparison.

Table 1 Summary of conventional electrocatalysis

Reaction	Catalysis	Electrolyte	Half-wave potential (E1/2)	References
ORR	NP Co3O4/Fe@C2N	0.1 M KOH	0.87 V (V vs. RHE)	[69]
	S-C2NA	0.1 M KOH	0.88 V (V vs. RHE)	[70]
	Au1Nx/PCN	0.1 M KOH	0.76 V (V vs. RHE)	[71]
	Co−PCN/CNT	0.1 M KOH	0.85 V (V vs. RHE)	[72]
	NiCo2S4@PCN-CNT	0.1 M KOH	0.76 V (V vs. RHE)	[73]
	PCN@PPy-C-900	0.1 M KOH	−0.21 V (V vs. Ag/AgCl)	[74]
	Fe@C2N	0.1 M KOH	0.876 V (V vs. RHE)	[75]
	Catalysis	Electrolyte	${\mathit{E}}_{\mathbf{10}\mathbf{}\mathbf{mA}\mathbf{/}{\mathbf{cm}}^{\mathbf{2}}}$
HER	PCN@N-graphene	0.5 M H2SO4	80 mV (V vs. RHE)	[76]
	PCN@NG	0.5 M H2SO4	240 mV (V vs. RHE)	[77]
	PCN nanoribbon-G	0.5 M H2SO4	207 mV (V vs. RHE)	[78]
	PCN@S–Se-pGr	0.5 M H2SO4	300 mV (V vs. RHE)	[79]
	Ru@C2N	1 M KOH	17 mV (V vs. RHE)	[80]
OER	PCN NS–CNT	0.1 M KOH	1580 mV (V vs. RHE)	[81]
	PCN–CFP	0.1 M KOH	1630 mV (V vs. RHE)	[82]
	Co−O−PCN@CS	1 M KOH	230 mV (V vs. RHE)	[83]
	Co−PCN/CNT	1 M KOH	161 mV (V vs. RHE)	[72]
	NP Co3O4/Fe@C2N	1 M KOH	430 mV (V vs. RHE)	[69]
	Catalysis	Electrolyte	Faradaic efficiency (%)
CO2RR	PCN/MWCNTs	0.1 M KHCO3	60 (−0.75 V vs. RHE)	[84]
	Au-CDots-PCN	0.5 M KHCO3	79.8 (−0.5 V vs. RHE)	[85]
	Au/PCN	0.5 M KHCO3	90 (−0.45 V vs. RHE)	[86]
NRR	Au1/PCN	0.05 M H2SO4	11.1 (−0.1 V vs. RHE)	[87]
	B-C2N	/	/(−0.15 V vs. RHE)	[88]
	AuSAs-NDPCs	0.1 M HCl	12.3 (−0.2 V vs. RHE)	[89]

Abbreviations: CO₂RR, carbon dioxide reduction reaction; HER, hydrogen evolution reaction; NRR, nitrogen reduction reaction; OER, oxygen evolution reaction; ORR, oxygen reduction reaction; PCN, polymeric carbon nitride.

Application of PCN Electrocatalytic OER

OER is a significant half-reaction for water splitting.⁹⁰ Most of the efficiency loss in the electrolytic cell in the hydrolysis reaction occurs from the anode side not only because of the high thermodynamic potential of OER but also because of the slow reaction kinetics involved in the multiproton-coupled electron transfer step. To improve the ability of electrolyzing water, it is necessary to reduce the overpotential of the reaction.^91,92 PCN takes advantage of its abundant and uniform nitrogen ligands to provide a large number of lone pairs of electrons to capture metal ions.^72,93–95 A series of molecular-scale PCN coordinated TM catalysts for OER have been developed.⁷² In particular, Song et al.⁸³ reported that Co-PCN@CS (Co atoms are anchored on PCN@carbon spheres to produce Co–O bonds and have extremely high dispersion) can significantly reduce the overpotential in alkaline solutions and show extraordinary OER performance. This is attributed to the precise coordination structure of O–Co–N and N in the PCN@CS support.⁸³ As shown in Figure 3A, PCN@CS plays a vital role in the preparation of the catalyst because N with a lone–pair valence electron is capable of capturing Co²⁺ and ultimately promotes the formation of Co–N and Co–O bonds in the calcination atmosphere of Ar and air atmosphere, respectively. The thermally favorable configuration of Co–X coordination predicted by DFT calculation can help promote an in-depth understanding of the source of the high performance. Determining the most stable binding site of the metal atom on the support is more conducive to precisely studying the interaction between the metal and the support. Therefore, based on nonequivalent nitrogen elements and the void rings, four stable and distinct adsorption sites for Co on the PCN@CS were initially constructed through geometric optimization (Figure 3B). In-depth analysis shows that the most stable structures of Co-PCN@CS and Co-O-PCN@CS are both Co-embedded in the empty ring to form the CoN₃C₂ ring with two adjacent pyridinic–N atoms and the CoON₂C ring with two nearby pyrimidine–N atoms, respectively (Figure 3C). These theoretical calculations and structural characterizations are consistent with the conclusion that the active site originates from the interaction between the metal atom and the support. This is due to the excellent affinity between nitrogen and metal ions, which leads to the formation of M–N bonds and further facilitates interface electron transfer (ET). In addition, the M–O bond can facilitate the transfer of electrons between the metal cation and the oxygen adsorbate to accelerate the OER rate.^72,96 The kinetic test also confirms that the charge transfer resistance (Rct) and diffusion resistance (Zw) of the catalyst are smaller, which is more conducive to the desorption of oxygen. The linear sweep voltammetry (LSV) curve shows that the overpotential of Co-O-PCN@CS at the current density of 10 mA/cm² is lower than other catalysts (Figure 3D). In Figure 3E, the energy profile pattern is used to estimate the catalytic process under different potentials. The high electronegativity of O in O–Co–N bond effectively promotes the conversion of O* into OOH*, and O–Co–N bond is more useful for ET at a low potential. In Figure 3E, Co–O–PCN@CS still has a high potential barrier, which proves that N–Co–N bond has a greater potential to transfer electrons than O–Co–N bond. The excessive adsorption strength of OOH* hinders the subsequent desorption process and decreases the performance of OER. This result is consistent with the demonstrated OER property. Recently, another novel nanostructured electrocatalyst (Co–B/PCN) was reported for OER, further suggesting that PCN sheets can serve as excellent support for the dispersion of active sites to be used in energy conversion applications.⁹⁷

Figure 3. (A) Schematic illustration of the formation mechanism of Co−PCN@CS and Co−O–PCN@CS. (B) Illustration of the starting position for Co adsorption on the surface of the PCN@CS support. (C) Corresponding calculated binding energy of Co−PCN@CS and Co−O−PCN@CS. (D) Oxygen evolution reaction (OER) polarization curves. (E) Free-energy diagram of the OER process on Co−PCN@CS and (F) Co−O−PCN@CS surfaces, respectively. The red balls and white balls represent O atoms and H atoms, respectively (Reproduced with permission: Copyright 2019, American Chemical Society83). PCN, polymeric carbon nitride

Electrocatalytic ORR

The ORR, as a half-reaction in metal-air batteries, occupies an important position. However, power delivery is often limited by the relatively sluggish ORR, so it is necessary to develop high-performance ORR catalysts.^98–101 PCN with high nitrogen content and rich active reaction sites has always been considered as a feasible metal-free catalyst for ORR. However, the ORR performance of the pristine PCN is not ideal, which is ascribed to the limited ET ability to promote the two-electron path and lead to a large accumulation of the reaction intermediate OOH^—. To solve this problem, PCN@CMK-3 (a highly ordered mesoporous carbon) was obtained by one pot precursor impregnation. It has also been proven to effectively increase the active site and thus promote ET.¹⁰² As shown in Figure 4A, ORR processes include three paths with different ET numbers (zero, two, and four electrons). It can be seen that the PCN compounded with the conductive support is more conducive to the participation of four electrons, which reduces the energy barrier and promotes the spontaneous reaction to produce OOH^— and unobstructed conversion into OH^—. Moreover, the hexagonal pattern of the P6mm symmetric mesoporous structure can be clearly seen in the image of the HRTEM, which indicates that the PCN in CMK-3 is uniformly distributed and PCN is perfectly compounded (Figure 4B). In addition, other structural characterizations also show that PCN is filled into CMK-3. LSV further reveals that PCN@CMK-3 has better onset potential and higher ORR current density than pristine PCN and the physically mixed PCN + CMK-3 (Figure 4C). It also proves that CMK-3 is not only the substrate of the restricted PCN catalyst but it can also significantly boost the electron accumulation on the surface of PCN and then markedly improve the catalytic activity. Other corresponding electrochemical characterizations show that the nano-confinement of PCN@CMK-3 improves ORR reaction kinetics, decreases the diffusion limit, and has high four-electron selectivity.

Figure 4. (A) Free-energy plots of oxygen reduction reaction and optimized configurations of adsorbed species on the polymeric carbon nitride (PCN) surface with zero-, and four-electron routes: I, II, and III. Gray, blue, red, and white small spheres represent C, N, O, and H, respectively, and corresponding schemes of ORR's pathway (red areas represent the active sites). (B) Typical high-resolution transmission electron microscopy images of ordered mesoporous PCN@CMK-3 nanorods. The inset represents a schematic illustration (yellow: PCN; black: carbon). (C) Linear sweep scanning voltammetry of various electrocatalysts (Reproduced with permission: Copyright 2011, ACS102)

Electrocatalytic HER

The research and development of a universal and efficient hydrogen evolution catalyst with the advantages of low overpotential, excellent conductivity, and good stability represent an urgent task.^103–107 In-depth exploration of the catalytic mechanism and structure–activity relationship is necessary for obtaining high-performance catalysts. However, many recent experimental and theoretical studies on electrocatalytic HER have only focused on the elucidation of the surface properties of metal catalysts due to the influence of metal-H bond, whether nonmetallic materials can show similar catalytic performance or even outperform metal-based catalysts has seldom been evaluated. The controversy of whether PCN is a 2D CN material has now been resolved by proving that it is a one-dimensional polymeric chain. Since large number of published papers still utilizted the 2D model of PCN for simulation, here these simulation works are objectively introduced although the 2D structure is incorrect. Zheng et al.⁷⁷ prepared PCN@NG by directly growing PCN on GO and further reducing it in an ammonia gas stream for hydrogen production. The structure of PCN@NG nanosheets was analyzed in detail through a series of electron microscopy characterizations (Figure 5A). The fine structures of carbon K-edges confirm the existence of defects (283.4 eV), graphitic (285.5 eV), and sp³ (290.2 eV) carbon species. Electron energy loss spectroscopy (EELS) mapping precisely presents the distribution areas of different carbon species. In-depth exploration shows that the defective carbon species is due to the strong interaction between PCN and the N-graphene substrate that leads to the breakage of the N–3C bridge bond at the edge of PCN. These characterizations indicate that the formation of chemical bonds arises from the coupling of PCN and N-graphene. NEXAFS (Near Edge X-ray Absorption Fine Structure) and XPS were implemented to further explore the interfacial interaction model between PCN and N-graphene, proving that the interlayer bond between PCN and N-graphene can promote the redistribution of electrons between PCN and N-graphene to provide a resistance-free path for fast ET, thereby accelerating the electrocatalytic HER kinetics. Besides, the DFT calculation also further confirmed the interlayer electronic coupling effect between PCN and N-graphene. The comparison of polarization curves proved that the performance of PCN@NG is better than the PCN/NG mixture, N-graphene, and PCN (Figure 5B). A stability test was subsequently carried out and it was found that the PCN@NG catalyst changed almost negligibly after 1000 cycles, which is necessary for the continuous production of hydrogen (Figure 5C). The DFT calculation also illustrates that the Gibbs free-energy uphill of PCN@NG is 0.19 eV, which is lower than that of the two individual moiety samples. This is consistent with the experimental results and confirms the synergy between the two assembled moieties (Figure 5D). Adsorption configuration analysis shows the mediated adsorption–desorption behavior caused by the chemical coupling of PCN and N-graphene. This is due to the fact that one H* in the PCN molecule is bonded to two pyridine nitrogen atoms in the tri-s-triazine periodic unit. Meanwhile, electrons are transferred from N-graphene to the active PCN layer, rapidly reducing the adsorbed H* into molecular hydrogen. This also illustrates from the atomic level that the high electrocatalytic activity of PCN@NG is due to the coupling synergistic effect of chemistry and electron, which enhances proton adsorption/reduction kinetics. Besides, according to the characteristics of electrocatalytic i₀ and thermodynamic ∆GH*, the HER performance of PCN@NG is comparable with that of the nanostructured MoS₂ electrocatalyst and even more than many nonprecious metal catalysts via a volcano-shaped plot (Figure 5E). At a low overpotential, one PCN@NG unit only adsorbs one H* due to its smallest |∆G_H*|, resulting in a low coverage rate of 1/3 (Figure 5F). The structure-oriented H* adsorption on PCN@NG is completely different from Pt. In addition, the activity of PCN@NG catalysts is relatively lower than that of platinum-based catalysts due to fewer active sites being exposed. Exposure to more active sites would greatly improve their electrocatalytic performance. The above findings suggests that PCN@NG with excellent electrocatalytic HER performance is comparable to or even better than that of traditional metallic catalysts. Although their finding inspires the development of nonmetallic catalysts, its activity is lower than that of Pt/C, also indicating the bottleneck of nonmetallic catalysts. However, the mechanism derived from this study has brought attention to the need for the development of new nonmetallic catalysts.

Figure 5. (A) Electron microscopy characterization of a PCN@NG nanosheet. (B) Hydrogen evolution reaction (HER) polarization curves. (C) Polarization curves recorded for the PCN@NG hybrid. (D) The calculated free-energy diagram of HER at the equilibrium potential for three metal-free catalysts and the Pt reference. (E) Volcano plots of i0 as a function of the ∆GH*. (F) Free-energy diagram of HER on the surface of PCN@NG (Reproduced with permission: Copyright 2014, The Nature Publishing Group77). PCN, polymeric carbon nitride

Electrocatalytic CO₂RR

With the rapid development of industry, the emission of carbon dioxide (CO₂) has increased sharply, which markedly violates the concept of sustainable development.^108,109 Electrochemical reduction of CO₂ has attracted widespread attention as an effective way to achieve carbon neutrality.^110–114 To obtain higher energy efficiency and scalability, the process should not only be efficient but also highly selective at a low overpotential. Adjusting the electronic structure of catalysts is a crucial way to accelerate the CO₂RR process.^85,115–118 For example, Zhang et al.⁸⁶ reported that Au/PCN was applied for electrocatalytic reduction of CO₂. XRD result shows that gold has a cubic crystal structure and a characteristic peak of PCN at 27.6° (Figure 6A). TEM image clearly shows the uniform distribution of gold NPs in PCN, and no large agglomerations appear (Figure 6B). The CO₂RR electrocatalytic activity test of Au/PCN and Au/C was carried out in a gas-tight two-compartment cell. Au/PCN has better CO₂RR performance, which is clearly recognized by its high CO partial current density (j_CO) (Figure 6C). In addition, although the j_CO growth of Au/PCN in the high potential area slows down, it may inhibit the mass transfer process due to the low CO₂ concentration in the solution. Structural characterization shows that the enhanced CO₂RR activity originates from the Au/PCN interaction promoting the formation of a negatively charged gold surface, which plays a critical role in stabilizing the key intermediate *COOH. An optimized structure is constructed to study the charge transfer between Au₈ clusters, PCN, and graphitic carbon. Although Au₈ forms an N–Au bond on PCN, it only presents a weak van der Waals interaction with the carbon substrate. In addition, Bader charge analysis shows that there is only a phenomenon of charge transfer at the edge of the N-heterocyclic ring, which is consistent with the experimental results (Figure 6D). It can be seen from Figure 6E that the energy required to convert CO₂RR into CO by Au^δ−/PCN is lower than that of Au⁰/C, which is also the origin for its higher catalytic activity. Although the PCN model constructed in this study is still controversial, this study further broadens the application range of PCN and provides a potential way for the conversion of CO₂ by enhancing the electronegativity of the catalyst surface to improve the binding ability to key intermediates.

Figure 6. (A) X-ray diffraction patterns. (B) Transmission electron microscopy images of Au/polymeric carbon nitride (PCN). (C) CO partial current densities for Au/PCN, Au/C, and PCN at different potentials. (D) Top and side (inset) views of the optimized Au8 cluster adsorbed on PCN and graphene–carbon support. (E) Calculated Gibbs free-energy profile of CO2RR (carbon dioxide reduction reaction) into CO catalyzed by Auδ−/PCN and Au0/C, respectively. Yellow, while, red, gray, and blue balls represent Au, H, O, C, and N atoms, respectively (Reproduced with permission: Copyright 2018, ACS86)

Electrocatalytic NRR

NH₃ is the most important chemical raw material due to the existence of extremely stable nitrogen–nitrogen triple bonds, which is difficult to prepare by destroying the N₂ bond by conventional means under normal conditions.^119–123 At present, the synthesis of NH₃ mainly relies on the traditional Haber–Bosch method, but this can pollute the environment and requires harsh conditions of high temperatures and high pressures.^124–128 The electrochemical nitrogen reduction reaction (eNRR) has attracted considerable attention due to the mild reaction conditions. Recently, to facilitate an in-depth understanding of the reaction mechanism and rational design of efficient catalysts, Liu et al.¹²⁷ innovatively proposed DFT calculations to show the full picture of TM-SACs (single atom catalyst) @PCN used in eNRR instead of focusing on one point unilaterally. From the limiting potential summarized in Figure 7A, Co@PCN, Ru@PCN, and W@PCN are potentially good candidates for eNRR, which is also convenient for further exploration. Figure 7B,C not only estimates the free-energy change of each basic step but also further explains the influence of different metal centers on the limiting potential by establishing the relationship between the limit potential and ΔEN*. With the change of the metal center, the adsorption strength and ΔEN* of the reaction intermediate also vary. The ligand is another factor that may affect the activity of the catalyst. Therefore, to fully understand the impact of the interaction between the metal center and the support (ligand) on the eNRR activity, a color contour map of the limiting potential was constructed (Figure 7D). From this, we can see that SACs@PCN shows better performance than other NC supports and individual metals. This is probably due to the fact that the support (ligand) adjusts the ratio of intermediate adsorption, which indirectly affects the eNRR activity, as the key adsorption intermediates *NNH and *NH₂ have a crucial influence on the potential-determining step. Besides, calculations based on thermodynamics also further explored the influence of the selectivity and stability of SACs@NC on the catalytic performance and show that N-doped carbon with four coordinated nitrogen atoms (N4) is the most stable (Figure 7E). Once again, the CPN model constructed in this study is controversial, but whether the simulation results have guiding significance for the experiment still needs to be fully verified.

Figure 7. (A) Summary of limiting potentials for the electrochemical nitrogen reduction reaction. (B) For the early transition metal and (C) for the late transition metal. (D) Compreh ensive comparisons of the limiting potential of SACs consisting of different metal centers and supports and the pure metal (111) surface is included as a reference. (E) Decomposition energies of the N-doped substrate (Reproduced with permission: Copyright 2019, American Chemical Society127)

Application of C₂N Electrocatalytic ORR

C₂N with inherent hexagonal vacancies can produce abundant trapping sites for metal species, which are potential substitutes for the noble metal-based commercial catalysts.¹²⁹ Li et al.¹³⁰ reported that the double-TM atoms could be stably anchored on C₂N. Later, Mahmood et al.⁷⁵ developed a novel indirect-contact electrocatalyst of Fe NPs stably encased in C₂N for ORR by the in situ sandwiching of Fe³⁺ precursors in C₂N layers. As schematically shown in Figure 8A, in the presence of ferric chloride, HAB HKH octahydrate underwent a condensation reaction. On embedding Fe³⁺ into the C₂N layer, the Fe³⁺@C₂N was subsequently reduced into iron oxide (Fe_xO_y) NPs. The 800°C heat treatment process promoted the formation of a core–shell structure of the Fe@C₂N catalyst. It can be clearly seen in the atomic-resolution transmission electron microscopy image that the distance between the Fe NPs' core lattice is 0.204 nm and the distance between the encapsulation layers is 0.34 nm, which also proves that the Fe NPs are perfectly encapsulated in the graphite nitride shell (Figure 8B). In addition, XRD pattern analysis shows that C₂N has a certain degree of graphitization and the iron element mainly exists in the form of metallic iron with only a very small proportion of iron carbide (Fe₃C) species (Figure 8C). Later, the electronic structure of Fe in the catalyst was further analyzed by the extended X-ray absorption fine structure. As can be seen, the edge structures of the sample before and after annealing are different, and Fe@C₂N moves to the low-energy region after annealing, which is similar to pure metal Fe (Figure 8D). The Fourier transform radial distribution of the Fe@C₂N catalyst shows that it moves to a lower energy region after annealing, which is also similar to that of pure metal Fe and matches the previous results (Figure 8E). In addition, the Fe@C₂N catalyst shows the same length of Fe–Fe bonding at about 2.55 ± 0.01 Å, which again confirms that the metallic Fe core is encapsulated in the graphite nitride shell. On the basis of the structural information shown, it was also shown that the Fe@C₂N catalyst has excellent stability. Attributable to the encapsulation of the metallic Fe core by the graphite nitride thin shell, it can prevent impurities from poisoning and deactivating the catalyst. The Fe@C₂N catalyst shows an onset potential close to that of Pt/C (1.021 V) via the efficient four-electron path, and the half-wave potential is better than that in the alkaline solution, as shown in Figure 8F. On the basis of these results, it can be determined that electron tunneling occurs from the inner metallic Fe core to the surface of the graphite nitride shell of Fe@C₂N NPs, and it also can be known that the polar C₂N can effectively adsorb and activate O₂. In addition, a zinc (Zn)–air battery was prepared using a Fe@C₂N catalyst. The polarization and power density curves are shown in Figure 8G. The Fe@C₂N (1.47 V) catalyst and Pt/C (1.45 V) have similar open-circuit voltages. The maximum power density provided by the catalyst is 123 mW/cm at 143 mA/cm, which is better than that of Pt/C (112 mW/cm at 128 mA/cm). Therefore, the Fe@C₂N catalyst is a kind of nonprecious metal-based catalyst that can be used in rechargeable Zn–air batteries and has great potential for use in the future.

Figure 8. (A) Schematic representation of the structural evolution of the Fe@C2N catalyst. (B) Atomic-resolution transmission electron microscopy (AR-TEM) image of Fe@C2N catalyst. The inset shows corresponding fast Fourier transform (FFT) of an Fe nanoparticle core. (C) Powder X-ray diffraction pattern of the Fe@C2N catalyst compared with those of Fe3C and pure Fe. (D) X-ray absorption near-edge structure spectra of FexOy@C2N (before annealing), the Fe@C2N catalyst (after annealing), and reference materials (Fe and FexOy). (E) Corresponding Fourier transform the radial distribution of the samples. (F) Polarization curves for Fe@C2N and Pt/C electrocatalysts. (G) Polarization and power density curves of Zn–air cells using the Fe@C2N catalyst and Pt/C (Reproduced with permission: Copyright 2018, Elsevier75)

Electrocatalytic HER

Hydrogen is attractive because of its advantages of high energy conversion efficiency and pollution-free nature. Therefore, the electrolysis of water to produce hydrogen has also become a research hotspot nowadays.^131–134 Pristine C₂N has a certain HER activity, which, however, is much lower than that of platinum-based catalysts.^135,136 Therefore, it needs to be doped and modified to meet the demand. Experimental and theoretical researches also prove that anchoring metal can effectively enhance the HER performance of C₂N. M@C₂N such as Co@C₂N and Ru@C₂N present remarkable activity in HER.^80,136–138 XRD is used as a conventional method to explore the crystal structure of materials. It can be clearly seen that the hexagonal Ru (PCPDF No. 89-4903) and the peak at 25.09° belong to the C₂N plane (Figure 9A). The TEM image (Figure 9B) shows that the Ru NPs with narrow size distribution are uniformly distributed in the C₂N substrate inset in Figure 9B. Also, scanning transmission electron microscopy-coupled energy-dispersive X-ray spectrum (STEM-EDS) element mapping again confirmed the uniform distribution of Ru NPs (Figure 9C). EELS spectroscopy (Figure 9D) further revealed the existence of C, N, and Ru. BET showed that the specific surface area of Ru@C₂N (400.1 m² g⁻¹) is larger than that of pure C₂N (280.5 m² g⁻¹) (Figure 9E). This also confirmed that the nucleation and growth of Ru NPs occur on the C₂N substrate. Electrochemical tests showed that the HER performance of Ru@C₂N in either 1 M KOH or 0.5 M H₂SO₄ solution is comparable to or even better than that of commercial Pt/C (Figure 9F,G). After 10,000 cycles of the stability test, the change in the catalyst is smaller than that of Pt/C, which also indicates that the catalyst is suitable for long-term operations (Figure 9H). Subsequent DFT calculations also shows that the high electrocatalytic activity of Ru@C₂N is due to the proper binding energy of Ru–H, which promotes the adsorption of hydrogen intermediates and the desorption of products. The above findings indicate that Ru@C₂N with excellent electrocatalytic activity can potentially be used as a highly valuable HER catalyst in the future.

Figure 9. (A) X-ray diffraction pattern of Ru@C2N. (B) Transmission electron microscopy (TEM) image of Ru@C2N. Inset: corresponding particle size distribution of the Ru nanoparticles. (C) Scanning TEM image of Ru@C2N. Scale bar = 20 nm. (D) Electron energy-loss spectroscopy spectrum of Ru@C2N. (E) N2 adsorption/desorption isotherms of Ru@C2N. The top inset shows the corresponding pore size distribution. (F) Polarization curves of Co@C2N, Ni@C2N, Pd@C2N, Pt@C2N, and Pt/C in H2SO4. (G) Polarization curves of Co@C2N, Ni@C2N, Pd@C2N, Pt@C2N, and Pt/C electrocatalysts in 1 M KOH. (H) Durability test of Pt/C and Ru@C2N. The polarization curves were recorded before and after 10,000 potential cycles (Reproduced with permission: Copyright 2017, The Nature Publishing Group80)

Electrocatalytic OER

OER occurs in critical half-reaction metal–air batteries.^139–141 Precious metals are used in OER due to their good catalytic activity but are limited due to their scarcity. The emergence of carbon-based materials can solve the problem, because the strong carbon/nitrigon-metal bond endows the delicate dispersion of precious metals on the carbon support and significantly decline the usage. The normal carbon-based materials tend to cause agglomeration of metal atoms to reduce utilization efficiency and catalytic activity. Therefore, there is an urgent need to develop more suitable support materials to overcome this problem.^142,143 The new type of carbon-based material C₂N with a rich nitrogen content and a stable structure has the function of stabilizing metal atoms, which can help solve this difficulty. Zhang et al.¹⁴⁴ designed a catalyst of TM atom-anchored C₂N–h2D monolayers and explored the electronic structure by computing. The possible anchor sites of the single TM atom as well as the two TM atoms on the C₂N monolayer are clearly presented in Figure 10A. On the basis of these models, further exploration and analysis are convenient. Some researches have shown that excellent electrical conductivity is completely conducive to charge transfer to promote the reaction. Calculation and analysis of the energy band structure are regarded as common methods to determine the conductivity of the material. Pristine C₂N has a large bandgap that is not conducive to charge transfer (Figure 10B). To solve this problem, anchoring of metal is a common method to reduce the bandgap and increase electrical conductivity to improve catalytic activity. The value of the free energy of adsorption can also be used as another indicator of OER reaction activity. Calculations confirm that the binding of O is stronger than the OH binding for Mn₁@C₂N (Figure 10C). It is more conducive to the adsorption of oxygen free radicals on the surface and promotes OER. Later, Kim et al.⁶⁹ assembled Co₃O₄ NPs and Fe@C₂N together to achieve a synergetic effect between the two moieties. Fe NPs are encapsulated on the nitrogenated 2D network to form a Fe@C₂N catalyst and then grow Co₃O₄ NPs on Fe@C₂N by the hydrothermal process to prepare Co₃O₄/Fe@C₂N. The interlayer spacing observed by HR-TEM images proves the existence of Co₃O₄, metallic Fe, and the graphite layer of the encapsulating shells. Also, combined with STEM, it is shown that the Fe NPs and Co₃O₄ NPs encapsulated in the N-doped graphite shell are evenly distributed and successfully anchored in the C₂N framework (Figure 10D,E). Then, it is found that the performance of Co₃O₄/Fe@C₂N NPs is better than that of IrO₂ and Fe@C₂N. In addition, the electrochemical testing of Co₃O₄ anchored on various substrates showed improved catalytic performance which also indicated that Co₃O₄ is the main active center of OER (Figure 10F). In Figure 10G, the XPS spectrum of Co₃O₄/Fe@C₂N shows the elemental composition of C, N, O, Fe, and Co. The corresponding high-resolution XPS showed that the relative proportions of Co³⁺ and Co²⁺ in Co₃O₄/Fe@C₂N NPs are slightly higher than that of the physical mixture of Co₃O₄ and Fe@C₂N. From the XPS spectrum changes of other elements, it was confirmed that the high activity mainly arises from the junction sites at which Co₃O₄ is anchored onto the N-doped carbon substrate. The lower O_lattice/O_ad ratio in Co₃O₄/Fe@C₂N NP obtained through an in-depth analysis indicates that the catalyst has a strong adsorption capacity for oxygen species, thus improving the electron donor capacity and promoting the synergy between Co₃O₄ NP and C₂N. These are all conducive toward improving the catalytic activity. Besides, the cycle performance comparison of Co₃O₄/Fe@C₂N NP and Pt/C + IrO₂ composites was performed at a constant current density of 0.5 mA/c² (Figure 12H). This demonstrated that its stability is better than that of Pt/C + IrO₂ after 25 h. This also indirectly proved that the catalyst is suitable for long-term stable operations.

Figure 10. (A) Several possible anchoring sites for a single transition metal (TM) atom. (B) Band structures and projected density of states (PDOS) of the C2N monolayer. (C) Computed negative overpotential (−η) against ΔG (O*) − ΔG (OH*) on TMx@C2N catalysts (Reproduced with permission: Copyright 2018, The Royal Society of Chemistry144). (D) Schematic illustration of the fabrication of NP Co3O4/Fe@C2N nanocomposites. (E) High-resolution TEM image of Co3O4/Fe@C2N and magnified images of Co3O4 and Fe@C2N. (F) Linear sweep scanning voltammetry plots for oxygen evolution reaction. (G) X-ray photoelectron spectroscopy survey spectrum of NP Co3O4/Fe@C2N. (H) Cycling performance of hybrid Li–air cells using NP Co3O4/Fe@C2N and Pt/C + IrO2 at a current density of 0.5 mA/cm2 (Reproduced with permission: Copyright 2019, American Chemical Society69)

Electrocatalytic NRR

As an irreplaceable step in the industrial synthesis of NH₃, the reduction of N₂ has attracted widespread attention.^146,147 How to activate the N≡N triple bond and reduce the reaction energy represent top priorities. The emergence of new catalysts may potentially help solve the problem. Zhang et al.¹⁴⁸ explored the catalytic performance of TM_x@C₂N for N₂ fixation by first-principles computations. Chen et al.¹⁴⁹ also designed TM–C₂N and TM₂–C₂N for NRR through theoretical calculations.¹⁴⁹ The geometrically optimized structures of TM–C₂N and TM₂–C₂N are shown in Figure 11A, and it can also be clearly seen that the redistribution of the charges between the TM–N bonds results in the shortening of the chemical bonds and an increase in the chemical bond strength, which also indirectly proves that the stability of the catalyst increases due to the strong interaction between the metal atoms and C₂N. Due to the existence of the strong N≡N triple bond, it was once considered that N₂* + H⁺ + e⁻ → NNH* was the rate-determining step (RDS) of NRR. To clearly identify the activation mechanisms of N₂ on TM_n–C₂N, Chen et al. did further research and found that the E_ad-NNH* value of the TM₂–C₂N system is higher, which is conducive to the first proton transfer and promotes the RDS. This also indicates that the bridge site of TM₂–C₂N makes the adsorbed N₂ more active and reduces the system energy of NNH* (Figure 11B,C). The ΔG value of this reaction path decreases, which means that the first proton transfer no longer hinders the progress of NRR. The excellent electrocatalytic performance of TM₂–C₂N was further confirmed by studying the enzymatic, distal, and alternating mechanisms of the entire reaction process. The free-energy changes of the catalytic pathways on Mn–C₂N and Mn₂–C₂N and the corresponding structural configurations of each intermediate state are compared. Once again, it clearly shows that at a given potential, Mn₂–C₂N has a high current density and superior catalytic activity (Figure 11D). Later, Cao et al.¹⁵⁰ designed other types of nonmetallic catalysts: B@C₂N and B₂@C₂N for NRR. As can be seen, the optimized B@C₂N and B₂@C₂N geometric structure models show that the doped atoms do not destroy the structure of C₂N (Figure 11E). The calculation results intuitively demonstrate that the ΔGmax values of B@C₂N and B₂@C₂N are only 0.45 and 0.35 eV, which are much lower than those of other single-metal atom catalysts (Figure 11F). Due to the consumption of protons and electrons, HER as a major side reaction will reduce the Faradic efficiency of NRR. The calculation result reveals that B@C₂N and B₂@C₂N have poor HER performance and high selectivity for NRR, so they are very promising NRR catalysts (Figure 11G,H). Recently, Yin et al.¹⁵¹ further studied the nature of the optimal active B site for NRR, demonstrating that C₂N provides an ideal platform to identify whether sp²- or sp³-hybridized B is optimal for NRR. As a novel material, C₂N has not been experimentally verified in NRR, but in the current theoretical calculation work, an in-depth analysis of potential catalysts was conducted, and thus more promising catalyst materials were selected.

Figure 11. (A) Structures of the TM–C2N and TM2–C2N. Red and yellow shadows represent the electron accumulation and loss, respectively. White, gray, blue, and purple balls indicate H, C, N, and Mn atoms, respectively. (B) Adsorption energy (Ead) of NNH* and (C) reaction free energy (∆G) of N2* + H+ + e− → NNH*. (D) Free-energy diagram of nitrogen reduction reaction (NRR) on Mn–C2N and Mn2–C2N (Reproduced with permission: Copyright 2019, Wiley149). (E) Structure of B@C2N and B2@C2N. (F) Volcano diagrams of nine catalysts for NRR. (G) Free-energy diagram for the hydrogen evolution reaction (HER) on B@C2N and B2@C2N. (H) The energy request for HER and NRR. ΔGmaxHER versus ΔGmaxNRR. (Reproduced with permission: Copyright 2019, IOP Publishing150)

Electrocatalytic CO₂RR

Reducing carbon dioxide emission is the topic of future global development. As a potentially rich carbon raw material, CO₂ has also attracted attention because it can be used to prepare various fuels and chemical substances.^{113,118,152–154} The first step of CO₂RR is to activate the CO₂ molecule, but CO₂ with a very high negative redox potential is very stable. It is necessary to select a suitable catalyst so that it can be reduced at a lower potential.¹⁵⁵ Now that it is clear that single-atom catalysts play an increasingly important role in the catalysis field, their role in CO₂RR is also becoming increasingly significant.^156–159 Continuous supply of reactants and rapid product release are conducive to maintaining the high efficiency of the reaction, that is, the mass transfer rate is very important to the catalytic reaction. Therefore, how to adjust the adsorption strength between intermediates and catalysts to obtain ideal catalysts has always been a hot topic for CO₂RR.^{104,145,160–162} To this end, Huang et al.¹⁶³ reported that NiCo@C₂N promotes catalysis for CO₂RR by DFT calculations combined with a computational hydrogen electrode model. It was found that the asymmetric N₂Ni–CoN₂ bonding configuration is more stable than the symmetric N₂Ni–CoN₂ bonding configuration. The clear display of the main reaction path diagram from CO₂RR to CH₄ clearly indicates that moderate to strong CO₂ adsorption promotes contact with active sites, which is a prerequisite for breaking of O═C═O double bonds (Figure 12A). In addition, the amount of charge transferred to the antibonding orbital of CO₂ molecules adsorbed on diatomic sites is greater than that on single atoms, showing that the dimer site is more conducive to the progress of the reaction. Besides, further predictions indicate that compared with single Ni atomic sites, the dimer sites achieve better selectivity for CO₂RR. From the atomic scale, the different bonding strength of Ni–C and Co–C, Ni–O, and Co–O lies in their uniqueelectronic resonance of 3d-states between different metals and homometallic dimers near the Fermi level (Figure 12B,C). In addition, the bond length of the heterometallic dimer is even shorter than that of the solid phase, which is also closely related to the adjustment of the electronic structure of the dimer pair. This can also indicate that the synergistic effects of the heterometallic Ni–Co dimer sites are more favorable for the enhanced CO₂RR activity than the corresponding homometallic dimer. HER has always been considered a reaction that competes with CO₂RR in kinetics and inhibits the selectivity of CO₂RR. It can be seen from Figure 12D that Ni–Co has better selectivity for CO₂RR, which better facilitates the progress of the reaction. The above computational studies show that the dimer sites anchored in the C₂N thin layer are more stable, providing a promising strategy for optimizing the anchoring of metal dimers on CN materials for efficient CO₂RR.

Figure 12. (A) Density functional theory (DFT-) optimized structures along the primary reaction pathway of CO2RR (carbon dioxide reduction reaction) toward CH4 formation. Red rectangle: potential-determining step. (B) Difference in limiting potentials for CO2RR and HER. Partial density of states (PDOS) of 3d-orbitals of Ni2, Co2, and NiCo@C2N. (C) Monomer. (D) Dimer (Reproduced with permission: Copyright 2019, American Chemical Society163)

SUMMARY AND OUTLOOK

In this review, we have presented an in-depth and comprehensive discussion on the structure, properties, synthesis methods, and practical applications of C₂N and PCN. Although wider commercialization still faces many challenges, an in-depth understanding of the basic structure–property relationship is indispensable for the design of high-performance electrocatalysts in the future. Therefore, more efforts should be devoted to the study of C₂N and PCN to address the following issues:

First, as a new CN material, C₂N suffers from the disadvantages of a rigorous synthetic process, low yield, and requirement of the use of expensive raw materials. Therefore, there is an urgent need to develop a large-scale, versatile, simple, and affordable preparation method to produce C₂N.

Second, PCN is usually prepared by the high-temperature calcination of urea, melamine, and other conventional reagents. However, the fabrication process of the reconstruction of the small molecules has not been revealed. The atomic self-assembly of C and N topologically forming the structure of PCN, nevertheless other possible topologies are still unclear. Thus, more efforts should be devoted to the in situ atomic structure study in the fabrication process of PCN. The current cognitive error in the structure of PCN (which is 1D structure instead of 2D structure) has been presented in this review, and thus the corrected 1D structure is also expected to be widely recognizyed in the future.

Finally, the electrochemical reaction mechanism of heteroatom-doped PCN and C₂N has rarely been elucidated. In the published papers, experimental techniques, which aim at investigating the reaction mechanism, are always absent; instead, DFT calculations are reported. Nevertheless, direct experimental evidence to support the proposed mechanism is indispensable. In this respect, two major bottlenecks impede the underlying experimental studies. On the one hand, the atomic configuration structure of the catalyst, which is the fundamental aspect of structure–property relations, is still unclear. On the other hand, the state-of-the-art in situ techniques to uncover the structure–property relations are newly developed and not easy to perform, such as in situ electrochemical process studies on image spherical aberration-corrected TEM.

This mini-review presents a systematic approach for the synthesis, characterization, and application of topological CN. Looking ahead, topological CN materials with excellent conductivity and stability can be used as excellent substrates for metal moieties, with bright application prospects in advanced energy conversion, such as fuel cells and catalytic water splitting. With the ever-growing demand for efficient carbon-based catalysts, large-scale realistic applications of topological CNs will become available in the near future.

ACKNOWLEDGMENTS

The authors acknowledge the financial support from the National Natural Science Foundation of China (22001228), the “Double-First Class” University Construction Project (C176220100022 and C176220100042), the Major Science and Technology Project of Precious Metal Materials Genetic Engineering in Yunnan Province (2019ZE001-1 and 202002AB080001), and the International Joint Research Center for Advanced Energy Materials of Yunnan Province (202003AE140001). Guangzhi Hu is grateful to the Double Tops Joint Fund of the Yunnan Science and Technology Bureau and Yunnan University (2019FY003025).

CONFLICT OF INTEREST

The authors declare no conflict of interest.