Zhiwen
Kang
a,
Xiaochen
Wang
a,
Dan
Wang
a,
Bing
Bai
b,
Yafei
Zhao
a,
Xu
Xiang
c,
Bing
Zhang
*a and
Huishan
Shang
*a
aSchool of Chemical Engineering, Zhengzhou University, Zhengzhou, 450001, P. R. China. E-mail: zhangb@zzu.edu.cn; shanghs@zzu.edu.cn
bKey Lab For Special Functional Materials, Ministry of Education, School of Materials, Henan University, Kaifeng, 100029, China
cState Key Laboratory of Chemical Resource Engineering, Beijing University of Chemical Technology, Beijing 100029, China
First published on 28th April 2023
Slow oxygen reduction reaction (ORR) kinetics is the main factor restricting the development of fuel cells and metal–air batteries. Carbon-based single-atom catalysts (SACs) have the advantages of high electrical conductivity, maximal atom utilization, and high mass activity, thus showing great potential in exploring low-cost and high-efficiency ORR catalysts. For carbon-based SACs, the defects in the carbon support, the coordination of non-metallic heteroatoms, and the coordination number have a great influence on the adsorption of the reaction intermediates, thus significantly affecting the catalytic performance. Consequently, it is of vital importance to summarize the impacts of atomic coordination on the ORR. In this review, we focus on the regulation of the central atoms and coordination atoms of carbon-based SACs for the ORR. The survey involves various SACs, from noble metals (Pt) to transition metals (Fe, Co, Ni, Cu, etc.) and major group metals (Mg, Bi, etc.). At the same time, the influence of defects in the carbon support, the coordination of non-metallic heteroatoms (such as B, N, P, S, O, Cl, etc.), and the coordination number of the well-defined SACs on the ORR were put forward. Then, the impact of the neighboring metal monomers for SACs on the ORR performance is discussed. Finally, the current challenges and prospects for the future development of carbon-based SACs in coordination chemistry are presented.
In order to improve the catalytic activity of ORR catalysts, it is necessary to study the reaction mechanism of the ORR process. In 1976, Wroblowa proposed the most effective explanation of the complex ORR step mechanism.27 The ORR process can proceed through two reaction paths, one through a four-electron (4e−) reaction path to produce OH− (in alkaline medium) and H2O (in acidic medium), and the other through a two-electron (2e−) reaction path to produce HO2− (in alkaline medium) and H2O2 (in acidic medium).28 No hydrogen peroxide is produced in the 4e− reaction, avoiding adverse effects on the catalyst, and the current efficiency is higher, which is ideal for the cathodic reduction reaction of fuel cells. Therefore, it is preferable for the oxygen reduction reaction to proceed by a 4e− reaction mechanism.
In the past few decades, SACs have been deeply explored by researchers due to their advantages of cost-effectiveness, maximal atom utilization, and high intrinsic activity.29–31 So far, this has become one of the most popular research areas in the field of electrocatalysis.32–35 SACs consist of metal single atoms bonded to atoms on carriers with an unsaturated coordination configuration.36–39 The binding energy of the chemical intermediates can be greatly affected by the coordination structure.40 So, small variations in ligand structure have an impact on the performance of the catalysts, even if they are similar-structure catalysts. Coordination engineering is achieved mainly through the modulation of the active central atoms, the distribution of defects in the carriers, and the regulation of the coordination atoms. Accurate and advanced characterization tools such as synchrotron radiation X-ray absorption fine structure (XAFS) and density functional theory (DFT) calculations also play important roles in the modulation of the coordination patterns.41–44 Up to now, many excellent electrocatalysts for the ORR have been designed and synthesized. Many excellent reviews have provided comprehensive summaries of various approaches to improve ORR activity.45–48 However, most of these are expounded from the macroscopic aspects such as the synthesis method, selection of the active center metal, catalyst morphology, etc. In-depth coordination engineering research studies and summaries are rare.
In this review, we comprehensively summarize and discuss the effects of coordination engineering for SACs on ORR performance. Firstly, we discuss the typical synthetic techniques and structural features of carbon-based SACs. Then, the types of central atom in SACs are summarized, which involve the most active noble metal Pt and a series of transition metals with the most development potential, as well as the seldom investigated group metals (Fig. 1). Especially, the effect of defects in the carrier, the coordination of non-metallic heteroatoms (such as B, N, P, S, O, Cl, etc.), and the coordination environment of the SACs on the reaction path, reduction products, ORR catalytic activity, and durability are discussed. In addition to this, we also summarize the influence of adjacent metal monomers for SACs on ORR performance (Fig. 2). Finally, a brief outlook on the current challenges and prospects of SACs for ORR reactions is also presented.
![]() | ||
Fig. 3 (a and b) Schematic illustration of the formation and ORR polarization curves of Fe1/N,S-PC. Reproduced with permission ref. 55. Copyright 2023, Tsinghua University Press. (c) Schematic illustration for the synthesis of Co@NCB. Reproduced with permission ref. 56. Copyright 2020, The Royal Society of Chemistry. (d) Illustration of the formation of a Sb SAC. Reproduced with permission ref. 57. Copyright 2021, Wiley-VCH. (e) Schematic illustration for the synthesis of a Sn SAC. Reproduced with permission ref. 58. Copyright 2023, Wiley-VCH. (f) SEM image of Fe1-PAN-NFs. Reproduced with permission ref. 59. Copyright 2020, Elsevier. |
![]() | ||
Fig. 4 (a) Schematic illustration of the synthetic procedure of Co-SA/N-C900. Reproduced with permission ref. 74. Copyright 2023, Elsevier. (b) Schematic illustrations of the synthesis of CoTPyP@Im-RGO. Reproduced with permission ref. 75. Copyright 2022, Elsevier. (c and d) Synthetic procedure and LSV curves of the Co–N,B-CSs. Reproduced with permission ref. 79. Copyright 2018, American Chemical Society. (e) Schematic illustration of the synthesis process for the Fe-NSDC. Reproduced with permission ref. 80. Copyright 2019, WILEY-VCH. (f–h) The synthetic process, ORR polarization curves and energy diagram of Fe-ISA/SNC. Reproduced with permission ref. 84. Copyright 2018, WILEY-VCH. |
Interestingly, recent studies have shown that heteroatoms that are not directly coordinated to the metal active sites can also enhance the catalyst activity.81–83 Although the heteroatoms are not directly attached to the metal active center, the heteroatoms doped in the carbon carrier will modulate the electronic structure of the active site by remote off-domain interaction, thus enhancing the ORR activity.46 For example, Li et al.84 designed a novel pyrrole-thiophene copolymer pyrolysis strategy to obtain Fe-ISA/SNC with S,N co-doping (Fig. 4f). By increasing the amount of S doping, the ORR performance of the Fe-ISA/SNC showed a volcano-shaped curve change. The E1/2 of optimal Fe-ISA/SNC was 55 mV more positive than Pt/C (Fig. 4g). XAFS analysis and DFT calculations (Fig. 4h) showed that the introduction of S enriched the charge around the Fe–N active site, which facilitated the rate-limiting reductive release of OH* and ultimately promoted the ORR process. Similar results were obtained in the study of Chen et al.,85 which demonstrated the positive effect of heteroatom doping on enhancing ORR activity.
Li et al.96 synthesized a Pt1/NPC catalyst with isolated Pt atoms on N-doped porous carbon (Fig. 5a). The loading of platinum was 3.8 wt%, as measured by inductively coupled plasma photoemission spectroscopy (ICP-OES). In Fig. 5b, it also exhibited high activity on an ORR, with an E1/2 = 0.887 V. Studies showed that this was attributed to the abundance of PtN4 sites and the introduction of N atoms to enhance the electron transfer. Similarly, Liu et al.97 reported a N-doped carbon Pt SAC (Pt1–N/BP) for highly efficient 4e− ORR. In contrast to Li et al.,96 the Pt loading was only 0.4 wt%. The Pt1/BP without N showed Pt agglomeration, suggesting that the introduction of N could help disperse the Pt atoms. Under acidic conditions, Pt1–N/BP had a high ORR performance (Fig. 5c), but the Pt loading is lower, demonstrating its economical nature. The assembled H2/O2 fuel cell of Fig. 5d showed a power density of 680 mW cm−2. More in-depth theoretical calculations using DFT showed that the adsorption of O2 on g-P–N1–Pt1 was stronger than that of CO, verifying its CO tolerance. The free energy was also calculated in Fig. 5e and f, the ORR superpotential at the g-P–N1–Pt1 site was 1.74 V, and the desorption of OH was the rate-limiting step. At the g-P–N1–Pt1 position, the entire ORR process was much faster than at the g-P–N1 position due to the small energy barrier of the preceding exothermic step and the rate-limiting step (*OH dissociation). In another study, Song et al.98 synthesized Pt SACs on the MOF-derived N-doped carbon. From the X-ray absorption near-edge structure (XANES) spectra in Fig. 5g, the platinum atoms were linked to N-electronegative atoms from the support, resulting in easier electron transfer from the Pt to the support, thereby increasing the 5d vacancy at the platinum sites. This was also the reason why Pt SAs-ZIF-NCs had the highest ORR performance. Furthermore, the adsorption of platinum atoms on pyrrolic N-containing ZIF-NCs was determined to be the most advantageous when compared with other forms of N (graphitic N, and oxidized N).
![]() | ||
Fig. 5 (a and b) Schematic illustration, ORR polarization curves of Pt1/NPC. Reproduced with permission ref. 96. Copyright 2018, American Chemical Society. (c–f) RRDE polarization curves, the voltages and power densities of H2/O2 fuel cells, the free energy diagram, side view, and bond lengths of Pt1–N/BP. Reproduced with permission ref. 97. Copyright 2017, Nature. (g) The XANES spectra of Pt SAs–ZIF-NC. Reproduced with permission ref. 98. Copyright 2020, Wiley-VCH. (h) The preparation of the PtNPC. Reproduced with permission ref. 51. Copyright 2021, Wiley-VCH. (i) Schematic illustration of Ir/N–C. Reproduced with permission ref. 103. Copyright 2019, Wiley-VCH. (j) Illustration of Ru–N/G. Reproduced with permission ref. 105. Copyright 2017, American Chemical Society. |
Inspired by N coordination, many scientists have placed heteroatoms into the Pt center, such as P and S, to alter the local coordination environment. Zhu et al.99 synthesized PtNPC catalysts by introducing P (Fig. 5h). The optimization of the 5d orbitals, a left shift of the projected density of states (PDOS) of the 5d orbitals, and a deeper d-band center of PtN3–PO, made it exhibit an excellent ORR performance. The NC, NPC and PtNC had poor ORR performance, but the PtNPC-0.5 had high ORR activity (E1/2 = 0.85) with the introduction of P. Theoretical calculations showed the P atoms promoted the kinetics of the 4e− route and prohibited the formation of peroxide species, further demonstrating that the P-coordinated Pt–N3 sites were the active sites. It also showed the feasibility of the introduction of heteroatoms for enhancing ORR activity.
Chen et al.100 loaded Pt onto S-doped graphitic carbon nitride (SGCN), and S-doping enabled the neighboring C and N atoms of Pt to be deficient in electrons, thereby enhancing the metal–support interaction. 20Pt/SGCN-550 had an ORR performance with E1/2 = 0.91 V. DFT calculations showed that 20Pt/SGCN-550 had a lower d-band center than 20Pt/GCN, indicating that the binding of O-containing intermediates is weak, so the adsorption binding energy decreases. And the ΔG of 20Pt/GCN-550 to generate OH− is also lower. Taken together, the introduction of S enhanced the interfacial interaction between the Pt and the carrier to enhance the ORR activity.
Pt can not only generate O2 through the four-electron pathway of oxygen reduction but also generate H2O2 through the two-electron pathway. Theoretical studies suggested that two ORR pathways share a common intermediate of *OOH. So, in theory, the two-electron and four-electron processes can be selected by regulating the adsorption capacities of the catalyst and oxygen intermediates. Zhao et al.101 successfully controlled the selectivity of the products by adjusting the coordinating heteroatoms around the Pt. The maximum hydrogen peroxide selectivity of Pt–S–C reaches 88%, which exceeded that of Pt–N–C (72.5%). The analysis of DFT calculations matched well with the experiment about H2O2 selectivity, indicating that adjusting the local coordination environment of the Pt can affect the intermediate the adsorption strength of the body determines the oxygen reduction pathway.
Therefore, for Pt-based SACs, there are significant differences in the performance depending on the Pt atomic coordination. Furthermore, coordination engineering of Pt-based SACs can efficiently adjust the 2e− or 4e− process of ORR to meet our urgent needs.
Catalyst | Active site | E onset (V vs. RHE) | E 1/2 (V vs. RHE) | Stability | Ref. |
---|---|---|---|---|---|
Fe SAC/N–C | FeN5 | — | 0.890 | Only minute differences after 5k CV cycles | 115 |
Fe-SAs–N/C-20 | Edge-hosted FeN4 | — | 0.915 | A negligible negative shift after 10k successive cycles | 116 |
FeN4–O–NCR | FeN4–O | 1.050 | 0.942 | Only a 5 mV degradation in E1/2 after 5000 CV cycles | 77 |
Fe–N/P–C | Fe–N3P | 0.941 | 0.867 | A negligible negative shift for 36![]() |
78 |
CoN4/NG | CoN4 | 0.980 | 0.870 | A high relative current of 92% after scanning 36![]() |
117 |
Co SAs/N–C(900) | CoN2C2 | — | 0.881 | No obvious decay in E1/2 after 5000 continuous potential cycles | 118 |
Ni SAs–NCs | NiNx | — | 0.850 | After continuous working for 20 h, the current retention rate is still more than 90% | 119 |
Cu-SA/N–C | CuN4 | 0.990 | 0.895 | No significant reduction in E1/2 was observed after 5000 continuous potential cycles | 120 |
ZIFs are assembled from transition metal ions and imidazole rings via tetrahedral coordination, and they can uniformly pre-arrange with metal–N4 linkages within an ordered 3D support.122,123 Many ZIFs not only exhibit excellent thermal and chemical stability, but also contain an abundant source of nitrogen in the imidazole ligand, and after pyrolysis, the ZIF is converted into regular nitrogen-doped carbon carriers.124 Inspired by ancient overhang-eave architectures, Hou et al.125 designed and synthesized overhang-eave structures containing isolated Fe atoms using silica as a medium (Fe/OES, Fig. 6a), which can provide more extended edges as three-phase exchange points (Fig. 6b), not only accelerating the mass transport of the ORR but also maximizing the exposure of the atomically dispersed/catalytically sites. As shown in Fig. 6c, Fe/OES exhibited excellent ORR activity with E1/2 = 0.85 V. As shown in Fig. 6d, for Fe–N4–C, step (ii) and step (v) required an external force to overcome the activation energy barrier, and step (v) is considered the rate-determining step (RDS) of the ORR reaction. Step (i) required a substantially higher endothermic energy for VFe–N4–C (VFe symbolizes Fe vacancy) as the contrastive sites, which was the RDS of ORR. The above shows that Fe–N4–C sites do participate in the ORR reaction as active sites. Similar conclusions were also reached by the work of Lu et al.,126 when they prepared Fe and N co-doped porous carbon nanotubules (MF–Fe-T, T stands for pyrolysis temperature, Fig. 6e). Electrochemical tests showed that the Fe,N-doped nanotubes synthesized by the same method were more electrocatalytically active for ORR than the N-doped nanotubes alone, implying that Fe–N4 promoted the association of oxygen species. The catalyst prepared at 800 °C is the most effective in the series, with Eonset = 0.98 V (Fig. 6f). The MF–Fe-T catalyst also exhibited good durability: in Fig. 6g there was only a 7 mV decrease in E1/2 after 5000 cycles. As can be seen in Fig. 6h, steps (3) or (4) for FeN4-doped graphene can be recognized as the RDS, denoting an advantageous binding with oxygen species. Stone–Wales Fe–N4 activates the neighboring C atom so that it also contributes to the ORR reaction. From the experiment and theory, the FeN4 sites played the important role in ORR activity.
![]() | ||
Fig. 6 (a–d) Synthesis process, STEM images, LSV curves, and ORR free-energy paths of Fe/OES. Reproduced with permission ref. 119. Copyright 2020, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim. (e–h) TEM images, RRDE voltammograms, the durability tests, and free energy diagram of MF-Fe-800. Reproduced with permission ref. 120. Copyright 2017, American Chemical Society. (i) Preparation process of a high-purity pyrrole-type FeN4 structure. (j and k) Deconvoluted features of peak a and peak b of the N K-edge spectra. (l and m) ORR polarization curves and free energy diagram of HP-FeN4. Reproduced with permission ref. 123. Copyright 2020, Royal Society of Chemistry. |
Among the M–N–C SACs, the usual nitrogen species are pyridinic, pyrrolic, and graphitic N. Among them, pyridinic N is the most common, but recent experiments and theoretical calculations have shown that pyrrolic N is more ORR-catalytic active.127–129 For example in Fig. 6i, high-purity pyrrole-type FeN4 sites (HP-FeN4) were successfully prepared by Zhang et al.129 Soft X-ray absorption spectroscopy (sXAS) proofed Fe–N sites of HP-FeN4 were confirmed to be high purity as pyrrole-type coordination (Fig. 6j and k). In Fig. 6l, HP-FeN4 exhibited a higher Eonset than FeN4 and a more positive E1/2 than FeN4. According to DFT calculations, the charge distribution of pyrrole-type FeN4 and pyridine-type FeN4 was noticeably different, and the iron's valence state was more positive. In the free energy diagram in Fig. 6m, pyrrole-type FeN4 was preferable for ORR because of the lower Gibbs free energy differences between O2 and OOH*. Furthermore, in a 4e− route, pyrrole-type FeN4 had a lower thermodynamic overpotential (0.35 eV) than pyridine-type (0.67 eV). In particular, on the pyrrole-type FeN4 structure, the reduction of OOH* to H2O2 is 0.2 eV higher compared with the pyridine-type FeN4 structure (1.77 eV vs. 1.57 eV), indicating that the 2e− reduction pathway is largely inhibited. Overall, pyrrole-type FeN4 would be the ideal sites for ORR because of the lower limiting potential and is more conducive to 4e− reactions.
However, the intermediates have an excessively high adsorption energy on the FeN4 sites and this affects their desorption to participate in the subsequent reactions, so this does not show the best activity.130,131 In addition to the Fe–N4 structure, the researchers found that FeN2 also has excellent ORR performance and conducted in-depth research on this because of the receding of the adsorption energy of the intermediates. Shen et al.132 introduced a number of dispersed FeN2 sites on the N-doped carbon. Efficient ORR catalysts (FeN2/NOMC) with iron atoms dispersed on the framework surface can be obtained by the subsequent removal of SBA-15 and agglomerated Fe-based particles with HF etching, leaving only the FeN2 sites (Fig. 7a). As shown in Fig. 7b, the FeN2/NOMC exhibit superior activity with an E1/2 of 0.863 V. The extended X-ray absorption fine structure (EXAFS) data analysis of Fig. 7c showed the coordination number of N is 2.0, and FeN2/NOMC-3 exhibits an almost total absence of Fe–Fe and Fe–C bonds, indicating that the majority of the Fe atoms are scattered as mononuclear particles in the FeN2 fraction. The DFT in Fig. 7d indicates that the overpotential of FeN2 is reduced by 0.12 V compared with FeN4, and the reduced interaction with *O2 and *OH intermediates is the cause of the increased ORR activity. And the existence of FeN2 sites was more beneficial for enhanced electron transport. The same idea was also obtained in the experiments of Song et al.133 They successfully synthesized iron-based ORR catalysts containing multiple active sites. The DFT calculations and tests showed that the following active sites on Fe–Nx/C catalysts for ORR have a structure–activity relationship: Fe–N2C > Fe–N4C > Fe4–NC > N–C > Fe4–C > C. These calculations provide theoretical support for subsequent studies of Fe coordination engineering.
![]() | ||
Fig. 7 (a–d) The synthetic procedure, ORR polarization, EXAFS spectra, and free energy diagram for FeN2/NOMC. Reproduced with permission ref. 126. Copyright 2017, Elsevier. (e–h) LSV curves, XANES spectra, three models with various coordination environments, and free energy diagrams of Fe SAC/N–C. Reproduced with permission ref. 128. Copyright 2019, Wiley-VCH. (i) Free energy diagrams of the N–Fe–N4, Fe–N4, and Fe–N2 in acid media. Reproduced with permission ref. 129. Copyright 2017, American Chemical Society. |
Recently, Lin et al.115 first introduced metal ions into UiO(bpdc) with pyrolysis and acid leaching. Fe SAC/N–C catalysts with FeN5 sites were obtained. The E1/2 of Fe SAC/N–C was 0.89 V (Fig. 7e). DFT calculations were performed to completely comprehend the superb ORR activity that the FeN5 site exhibited. Three models in Fig. 7g, Fe-4pN, Fe-4pN-OH, and Fe-4pN-py, were investigated in depth. In Fig. 7h, for the ideal Fe-4pN model, all the reaction steps are thermodynamically favorable, but in practice, the ideal Fe-p4N is converted to Fe-p4N-OH due to abundant OH−. For Fe-4pN-OH, the formation of *OOH is the RDS. The energy barrier of Fe-p4N-py would be lowered to 0.11 eV if the extra OH group were replaced with pyridine. Although the whole reaction process was still limited by the formation of *OOH, the catalytic activity of Fe atoms coordinated to five pyridine nitrogen atoms for ORR is very considerable. And in a study, the five-coordinated Fe–Nx configuration prepared by Lai et al.134 exhibited better ORR activity in acidic media compared with the lower-coordinated Fe–Nx configuration. Three theoretical models, N–Fe–N4, Fe–N4, and Fe–N2, were constructed to explain the reasons in Fig. 7i. For N–Fe–N4, and Fe–N4, the free energies of all the successive intermediate steps at 0 V are reduced compared with Fe–N2, but the N–Fe–N4 structure showed the lowest energy barrier, indicating that the O2 molecules were more easily reduced to H2O completely on the N–Fe–N4 structure. In addition, when the OH* adsorption energy was compared, the N–Fe–N4 structure's OH adsorption energy was 2.88 eV, which was much lower than the adsorption energies of Fe–N4 and Fe–N2. The lower hydroxyl adsorption energy protects the active center from deactivation. The above indicates that even if the coordination mode is the same, different synthesis methods, carbon carriers, and defect richnesses can affect the actual performance of the catalyst. Therefore, research on catalysts should not only focus on the theory.
Defective Fe active sites have been extensively studied. For example, Jiang et al.116 prepared catalysts with Fe–N4 sites anchored on 3D layered porous carbon (Fe-SAs–N/C-20, Fig. 8a and b). Fe-SAs–N/C-20 exhibited excellent ORR performance with a half-wave potential of 0.915 V, in Fig. 8c. 13C solid state nuclear magnetic resonance (ssNMR) showed that the integrated intensity ratios of sp3- to sp2-hybridized carbons (Isp3/Isp2) grew with defect increases, indicating that defects make it easier for C–N bonds to break. Six possible conformations of Fe–N4–C were calculated by DFT. As shown in Fig. 8e, the least overall reaction free energy change is shown in the defective Fe–N4-6r-c2 near the gap's edge, suggesting that selective CN bond cleavage (SBC) can adjust the coordination environment of pyridine N. In another study, Xiao et al.135 incorporated FeN4 edge sites into graphene (Fe/N–G-SAC) by a new self-sacrificing template method (Fig. 8f). This method used an extremely excess amount of Fe precursors, and the Fe clusters not only promoted the formation of graphitized structures but also promoted the generation of FeN4 partially near the Fe clusters. After etching the Fe clusters, edge-enriched FeN4 sites were obtained. Due to these structural properties, the Fe/N–G-SAC had excellent activity and stability. The assembled zinc vacancy cells also exhibit excellent performance. It was calculated by DFT that the closer the distance between the Fe cluster and the Fe–N4 part, the lower the relative formation energy (Ef). Therefore, the introduction of excessive Fe clusters during the synthesis process made it easier to form the Fe cluster/Fe–N4-1 structure. This eventually leads to the formation of edge-dominated FeN4 sites in the sample after the Fe clusters are removed (Fig. 8g). The conformational differences between the edge site and the in-place site were compared. Due to the decrease in the number of N coordinations, a significant charge redistribution occurs at the Fe–N4 edge sites, allowing a more favorable electron transfer from the Fe atom to the neighboring N atom, and the higher Bader charge estimated also demonstrates this. The partial density of states (DOS) of the Fe calculation showed that with the negative shift of the d-band center on the edge FeN4 site, the binding strength of the adsorbates on the edge site is also relatively diminished. The RDSs on the in-plane Fe–N4 sites were known to be the OH desorption; however, after a thorough investigation, it was shown that the adsorption free energy of OH* (GOH*) on the edge site was larger, indicating a faster OH desorption process that was beneficial for the ORR (Fig. 8h). Similar conduct had also been observed by Chen et al.136 The sur-FeN4-HPC catalyst they synthesized using the edge effect also has excellent ORR activity.
![]() | ||
Fig. 8 (a–e) Schematic illustration, HRTEM, LSV curves, five possible atomic configurations, and free energy diagram of Fe-SAs–N/C. Reproduced with permission ref. 130. Copyright 2018, American Chemical Society. (f–h) Scheme of synthesis, structure, and free energy diagrams of Fe/N–G-SAC. Reproduced with permission ref. 131. Copyright 2020, Wiley-VCH. |
The introduction of other heteroatoms will also significantly enhance the ORR activity of the catalyst. For example, Mun et al.137 doped S functionalities into the carbon plane to modulate t-doped S functionalities into the carbon plane to modulate the electron-absorbing/electron-giving properties of the Fe–N4 sites (Fig. 9a). X-ray photoelectron spectroscopy (XPS), shown in Fig. 9b, indicated that the S existed in both thiophene-like S (C–S–C) and oxidized S (C–SOx) forms. The C–S–C enriched the carbon plane with electrons, thus enhancing the adsorption of the ORR intermediate on Fe–N4, while the introduction of C–SOx reduces the Fe d-band center. Interestingly, similar trends were seen in the E1/2 of the FeNC–S–MSUFC catalysts with regard to the ratios of oxidized S and thiophene-like S (Fig. 9c), demonstrating that the kind of S affected the ORR activity of the Fe–N4.
![]() | ||
Fig. 9 (a and b) Schematic representation of the synthesis and S 2p XPS spectra of FeNC–S–MSUFC. (c) Relationship between the ratio of oxidized S and thiophene-like S and ORR activity. Reproduced with permission ref. 133. Copyright 2019, American Chemical Society. (d and e) Scheme of the synthesis, E1/2 and jk of the FeN4–O–NCR. Reproduced with permission ref. 77. Copyright 2022, Wiley-VCH. (f–i) Synthesis, LSV curves, the durability tests, and illustration of proposed various structures of Fe(Zn)–N–C. (j) Gibbs free energy diagrams at 1.23 V on FeN4, 2L-Mid, and 2L-Up sites. Reproduced with permission ref. 134. Copyright 2020, Wiley-VCH. |
In another study combining experimental and theoretical, Peng et al.77 synthesized FeN4–O–NCR in which the Fe–N4 sites were modulated by Fe–O bonds (Fig. 9d). As shown in Fig. 9e, FeN4–O–NCR had good ORR activity due to its special coordination structure. DFT calculations similarly demonstrated that FeN4–O–NCR was beneficial for the ORR process. When simple axial O ligands are present, this led to a weaker binding of FeN4/C and FeN4/NC to *OH, thereby improving the ORR performance. A later study from Gong et al.138 in Fig. 9f also introduced O into FeN4, forming Fe(Zn)–N–C catalysts with Fe@O@Fe bridge bonds. Compared with the single-atom FeN4 site, the O bridge-bonded Fe sites have 10 times higher activity in terms of turnover frequency (TOF, 3.2 per es per sites vs. 0.32 per es per sites) and have promising half-wave potential (0.83 V), in Fig. 9g. Meanwhile, after 10000 continuous cycles, the E1/2 reduced only by 14 mV (Fig. 9h). Gibbs free energies were calculated in Fig. 9i for the three models (2L-Mid, 2L-Up, and FeN4). As shown in Fig. 9j, 2L-Up had the lowest overpotential and the final *OH dissociation step was the only heat-absorption process, suggesting a facilitative effect of the indicated bridge-bonded O on the improvement of catalytic activity. The electron flow from the Fe d orbital to O was found using Bader charge analysis, resulting in a weaker adsorption of the remaining d electrons on O2 and ultimately excellent ORR performance. A later study from Yuan et al.78 showed that the introduction of P heteroatoms was beneficial for the adsorption/desorption of oxygen intermediates, boosting the catalytic ORR performance. Three N atoms and one P-anchored Fe (Fe–N3P) were found to make up the compound Fe–N/P–C. It can be inferred from a comparison of the free energy distributions of Fe–N4 and Fe–N3P that the O2 molecule caught by the Fe atom went along a downward trajectory. However, the fact that Fe–N3P's free energies were lower than those of Fe–N4 suggests that the O2 molecules can easily adsorb and form a strong bond on Fe–N3P active sites.
To date, the study of Co–Nx coordination configurations can be widely divided into two categories: experimental and theoretical. Although Co–N–C has a variety of coordination modes, CoN4 is the most commonly used model.143,144 Zhang et al.145 investigated in depth the CoN4–graphene catalysts with pyridine-N and pyrrole-N coordination by DFT calculations (Fig. 10a and b). Despite both CoN4–G having ORR activity, the specific catalytic mechanisms are different. The CoN4–G(A) catalysts exhibited higher activity, whereas the CoN4–G(B) catalyst, owing to the high response barrier of H2O2 formation (1.07 eV), exhibited higher 4-electron selectivity (CoN4–G(A) and CoN4–G(B) are the names given to the Co atoms coordinated by four pyridinic-N atoms and four pyrrolic-N atoms, respectively). In an earlier study, Yang et al.117 prepared the single-atom Co–N–C catalysts with atomically dispersed CoN4 (denoted as CoN4/NG, Fig. 10c). In Fig. 10d and e, XANES and EXAFS analyses showed that one Co atom in CoN4/NG was coordinated to four surrounding nitrogen atoms, forming a CoN4 group, which was the main reason for its excellent ORR performance. As shown in Fig. 10f, Wang et al.146 further designed a locally distorted CoN4 conformation, where the charge can be transferred more rapidly from the Co atom to the N due to the disruption of the symmetric electron distribution. Compared with plane CoN4, it showed a lower energy barrier for the RDS and thus had a better ORR performance (Fig. 10g). In addition, the free energy diagrams of locally distorted Co–N3C1 and Co–N2C2 structures were also calculated in Fig. 10h. The RDS of Co–N3C1 and Co–N2C2 is the reduction of OH*, where the ΔGmax values are 0.43 eV for Co–N3C1 and 0.54 eV for Co–N2C2, suggesting that Co–N3C1 has higher ORR activity. For a more comprehensive summary comparison, four types of Co–N contribution categorised as CoNx (x = 1–4) were estimated by Sun et al.,147 in Fig. 10i. The CoN4 profiled most nearly resembles the optimum free-energy pathways. The overpotential was further calculated and found to increase in the order of CoN4 < CoN3 < CoN2 < CoN1, indicating that CoN4 has better catalytic performance (Fig. 10i).147
![]() | ||
Fig. 10 (a and b) Top and side views of the optimized structures of CoN4–G(A) and CoN4–G(B). Reproduced with permission ref. 141. Copyright 2020, Wiley-VCH. (c–e) Schematic illustration, XANES Co K-edge spectra, and Fourier-transforms of Co K-edge spectra for CoN4/NG. Reproduced with permission ref. 142. Copyright 2018, Elsevier. (f) Distorted Co–N4 structure. (g) Free energy diagram on different structures of CoN4. (h) Free energy diagram for the ORR on CoN2C2 and CoN3C1. Reproduced with permission ref. 143. Copyright 2022, Elsevier. (i) Free-energy diagrams for the ORR and OER pathways on CoNx (x = 1–4). Reproduced with permission ref. 144. Copyright 2019, Elsevier. (j and k) Schematic illustration of CoN4 and CoN2. (l) Free-energy diagram for the reduction of O2 to H2O2 on the CoN4 defect. Reproduced with permission ref. 145. Copyright 2019, Royal Soc Chemistry. |
In a theoretical study, a computational study of the ORR mechanism of Co–N4 electrocatalysts with pyridine-N coordination by Kattel et al.148 had verified again the significant effect caused by different N coordinations on the Co–N4 activity. Compared with graphitic CoN2 defects, graphitic CoN4 defects were energetically favorable and had better stability. Therefore, CoN4 should be the more promising catalytic center (Fig. 10j and k). Due to the defective configuration of Co–N4 with pyrrole-N coordination and the O2 bound in an end-pair configuration, it is more inclined to produce H2O2 than the Co–N4 with pyrrole-N coordination. As shown in Fig. 10l, the adsorption of OOH− on CoN4 was weak, and a second catalytically active center was required to further reduce H2O2, which is also consistent with the experimental results.118,149,150 The 4e− reaction without OOH− intermediate generation occurred in the CoN2 part, which is similar to the catalytic process at the Fe–N2 site.151 The later work of Sun et al.152 also drew similar conclusions.
Another study from Yin et al.118 found that Co–N2 species interacted more strongly with peroxides than Co–N4 and promoted the four-electron reduction process of ORR (Fig. 11a). Experiments also demonstrate that Co–N2 sites can exhibit superior ORR performance. By regulating the pyrolysis temperature, Co SAs/N–C(800) with CoN4 sites and Co SAs/N–C(900) with CoN2C2 sites were obtained. In Fig. 11b, the ORR activity of Co SAs/N–C(900) was better than Co SAs/N–C(800).
![]() | ||
Fig. 11 (a and b) The formation and RRDE polarization curves of Co SAs/N–C. Reproduced with permission ref. 148. Copyright 2016, Wiley-VCH. (c–e) Top and side views of three structures of CoN2–G. Reproduced with permission ref. 151. Copyright 2017, The Electrochemical Society. (f) Scheme of synthesis of the CoSA/NPC. (g and h) The Gibbs free energy diagram, charge density differences and Bader charge analysis of Co/N4 and Co/N3P. Reproduced with permission ref. 154. Copyright 2021, Royal Society of Chemistry. (i and j) SEM image and schematic illustration, and RRDE polarization curves of Co–NBG. Reproduced with permission ref. 155. Copyright 2019, Elsevier. (k) Schematic illustration of single-atom Co-SAs/NSC. Reproduced with permission ref. 76. Copyright 2019, American Chemical Society. (l–n) Scheme of synthesis, free energy diagram, and ORR polarization curves of CoPS@SPNC. Reproduced with permission ref. 160. Copyright 2022, Elsevier. |
Zhang et al.153 performed DFT calculations on the detailed kinetic and thermodynamic behavior of the ORR on three different CoN2–G models (Fig. 11c–e). Similar to the conclusions obtained by Kattel et al.148 all three CoN2–G models favored the ORR response, but with subtle differences. The catalyst with the highest levels of catalytic activity and four-electron selectivity is CoN2–G(A). In contrast, the two-electron pathway is favored due to the weak adsorption energy of the H2O2 generated on CoN2–G(B) (−0.39 eV) which is easily desorbed.
Similar to Fe–N–C catalysts, doping with nonmetallic heteroatoms (B, P, S, etc.) also enhances the Co–N–C ORR performance by tuning the electronic properties of neighboring carbon atoms.154,155 For example, Liu et al.156 anchored Co single atoms to ultrathin N- and P-doped porous carbon nanosheets to obtain CoSA/NPC catalysts (Fig. 11f). DFT were carried out to theoretically comprehend the ORR reaction mechanism in order to understand the impacts brought about by P doping. As shown in Fig. 11g, the RDS of Co/N3P is smaller than that of Co/N4 (0.445 eV vs. 0.466 eV) and the charge distribution at the Co active site is unbalanced due to P doping (Fig. 11h), which is beneficial for the adsorption of oxygen species. As a result, the ORR catalytic activity of Co/N3P is better compared with CoN4 (E1/2 = 0.87 V).
In another study by Xu et al.,157 B was introduced into Co–N–C to obtain Co–NBG (Fig. 11i), and this sample exhibited excellent ORR performance. In Fig. 11j, the E1/2 of Co–NBG was 0.792 V. The effect of B doping on the ORR catalytic activity of Co–gN4 based on DFT was investigated by Fu et al.158 Because the electronegativity of B is lower than C, when graphite was doped with B, an area close to Co–gN4 developed that was positively charged, requiring the Co sites to lose additional electrons in order to achieve equilibrium. Additionally, the adsorption of Co sites to intermediates was less as the B concentration increased, which increased the reactivity. In addition to B and P, Zhang et al.76 synthesized Co single atoms on the porous N,S-co-doped carbon (Co-SAs/NSC, Fig. 11k). The coordination states of the S atoms at the single atom core were shown to be crucial in lowering the reaction barrier, which increased the ORR kinetics, according to DFT calculations. Therefore, the ORR activity of the catalyst can be optimized by adjusting the electronic structure of the catalyst by selecting different electronegativity heteroatoms.
It was also found that co-doped graphene with two or more heteroatoms would be more electrocatalytically active due to synergistic effects compared with single heteroatom-doped graphene.159–161 As shown in Fig. 11l, Xu et al.162 obtained CoPS@SPNC through carbonization of ZIF-67 followed by a simultaneous sulfidation and phosphatization. The introduction of heteroatoms decreased the overpotential of the ORR to enhance the catalytic activity, as calculated by DFT (Fig. 11m), which is consistent with the experimental results (Fig. 11n).
In addition to the 4e− ORR discussed above, Co SAC was found to have the best performance in the production of H2O2 among the transition metal SACs. As shown in Fig. 12a, Jung et al.163 found that ΔGOOH* could be adjusted by attaching a functional group to the Co–N4 molecule ΔGOOH*. GOOH* rises from 3.9 eV to 4.1 eV when electron-rich species, such as O*, are adsorbed close to the Co–N4 molecule (Co–N4(O)), approaching the ideal value for the formation of H2O2. It showed that the 2e− ORR produced H2O2 with a high degree of selectivity when O* was present near the Co atom. Therefore, they synthesized Co1–NG(O) for the electrochemical production of H2O2 in Fig. 12b. Using the Koutecky–Levich equation, the kinetic current density for H2O2 production from Co1–NG(O) is 2.8 ± 0.2 mA cm−2 at 0.65 V (Fig. 12c). Similarly, Li et al.164 conducted a mechanistic study of the synthesized Co–POC–O for H2O2 production. As shown in Fig. 12d, they found that the high H2O2 production rate was due to the synergistic effect of the Co–Nx–C and OfG sites responsible for ORR reactivity and two-electron selectivity, respectively. More recently, Shen et al.165 compared in detail the catalytic activity and selectivity for the H2O2 production of five-coordinated O–Co–N2C2 and Co–N4 by DFT calculations, in Fig. 12e and f. The ΔGOOH* of Co in O–Co–N2C2 (Fig. 12g) was closer to 4.2 eV than that of CoN4. Beyond this, at the equilibrium potential for H2O2 production, the RDS energy on O–Co–N2C2 was much lower than that on Co–N4, indicating that the penta-coordinated O–Co–N2C2 was inherently more active and selective than traditional tetra-coordinated Co–N4 for the generation of H2O2 (Fig. 12h).
![]() | ||
Fig. 12 (a) Calculated catalytic activity volcanoes. (b) Scheme of synthesis of Co1–NG(O) SAC. (c) Summary of H2O2 production activity. Reproduced with permission ref. 161. Copyright 2020, Nature. (d) Schematic illustration of atomic Co–Nx–C sites and oxygen functional groups for H2O2. Reproduced with permission ref. 162. Copyright 2019, Wiley-VCH. (e–h) Deformation densities, calculated catalytic activity volcanoes, and calculated reaction energetics of the 2e− ORR of O–Co–N2C2 and Co–N4. Reproduced with permission ref. 163. Copyright 2022, Wiley-VCH. |
In general, the optimization of Co-based catalysts was similar to that of Fe-based catalysts. The coordination structure of the Co–N–C part can be modified and controlled by adjusting the coordination number of the N, and the non-metallic heteroatom doping can also enhance the ORR activity by adjusting the electronic structure to the central Co atom. Different from the Fe-based catalysts, Co-based catalysts can be used for the production of H2O2 due to the inherent properties of the Co metal.
Jiang et al.119 synthesized nitrogen-doped carbon sheets (Ni SAs–NCs) with dispersed Ni atoms by pyrolysis (Fig. 13a). Ni SAs–NCs have good ORR activity with an E1/2 of 0.85 V (Fig. 13b). In addition, the stability of Ni SAs–NC is excellent; after continuous working for 20 h, the current retention rate was still more than 90%. Its high activity was thought to be due to the transition of O2 to O* intermediates facilitated by the Ni–Nx center.168 Qiu et al.169 in-depth investigated the four-electron ORR catalytic performance of N,Ni-co-doped graphene in alkaline conditions by DFT calculations. O2 reacts at the active sites and is successively converted to oxygenated intermediates OOH*, O*, and OH*, eventually producing water. For this catalyst, the rate-limiting step is the generation of OOH* intermediates, the adsorption energy of OOH* on N,Ni-co-doped graphene was higher than that on N single-doped graphene and Ni-doped graphene, and the O–O bond was longer in N,Ni-co-doped graphene. The above constitute the reasons for the high ORR activity of Ni–N–C. In addition, compared with pure graphene, doping N on graphene promotes the anchoring of Ni, which enhanced the ORR activity by increasing the density of Ni active sites.
![]() | ||
Fig. 13 (a and b) Scheme of synthesis and LSV curve of Ni SAs–NC. Reproduced with permission ref. 166. Copyright 2022, Elsevier. (c and d) Free energy diagrams for the ORR and OER on four-coordinated and on three-coordinated Ni–N–C-gra. Reproduced with permission ref. 169. Copyright 2020, Royal Society of Chemistry. |
In a theoretical study, Liang et al.170 established 15 different Ni–N–C coordinated doped graphene systems using nitrogen-doped graphene as a carrier, and comprehensively and theoretically studied the important effect of the coordination environment of Ni-based catalysts on ORR activity. As shown in Fig. 13c, the adsorption capacity of tetracoordinate Ni–N–C-gra for oxygenated intermediates was weak, while the adsorption capacity and ηORR increased with the decrease of N doping amount. In contrast, three-coordination Ni–N–C-gra highly adsorbed intermediates showed that OOH* rapidly dissociates and that there was an excess amount of OH* being adsorbed (Fig. 13d). In addition, it was also found that on the Ni–N–C sites modified with OH functional groups, the adsorption strength of the intermediate product was reduced, thereby enhancing the ORR activity. The ηORR of Ni(OH)N3 was even lower than for Pt (0.42 V vs. 0.45 V). In summary, these theoretical studies have provided a solid theoretical basis for future research on Ni-based ORR catalysts.
![]() | ||
Fig. 14 (a–d) Scheme of synthesis, copper 2p XPS spectra, EXAFS fitting curve, and RRDE polarization curves of Cu-SAs/N–C. Reproduced with permission ref. 171. Copyright 2018, Springer Nature. (e) Scheme of synthesis of CuN3 and CuN4. (f and g) View of the model of the CuN4 model and CuN3 model. (h and i) Free energy diagrams for the ORR process on CuN4 and CuN3 at different overpotentials. Reproduced with permission ref. 173. Copyright 2020, Wiley-VCH. |
The chemical valence state of the central Cu atom also plays a significant role in enhancing the ORR activity of Cu-based ORR catalysts. Sun et al.174 prepared a series of Cu-based ORR catalysts (Cu-SA/NC(meso)) with different ratios of Cu1+ and Cu2+ using ammonia as the reducing agent (Fig. 15a). Testing found that ORR activity increased with the addition of Cu1+ sites. The Cu1+-SA/NC(meso)-7 achieved excellent ORR activity with E1/2 = 0.898 V (Fig. 15b) and Jk = 5.36 mA cm−2 (Fig. 15c). As shown in Fig. 15d, DFT calculations indicated that the energy required for the rate-limiting step of Cu1+-SA is 0.46 eV, which is lower than that of Cu2+-SA (0.71 eV). Moreover, the Cu single catalysts rich in Cu1+ sites had good binding energies with OOH adsorption in the ORR process, resulting in high electrocatalytic activity in alkaline media. From the DFT calculations, it can be concluded that the coordination state of the S atom plays an important role for the monoatomic centre in lowering the reaction potential barrier and facilitating the onset of ORR kinetics. The study of the CuN2C2 site by Han et al.175 also demonstrated the presence of high ORR activity. They simulated three different CuN2C2 models. Although the geometries are different, the potential determining step is the same, which is the protonation of *O2 to *OOH. The calculations revealed that more negative charges were transferred to *O2 on Cu/CNT-8 due to the different substrate structures of the three models. In addition, the maximum structural deformation occurred when O2 was adsorbed on Cu/CNT-8, which was also considered to be the reason for its high activity. Therefore, it was very comprehensive to correlate the metal valence state of the central site with the coordination mode to study the reason for its high activity towards ORR.
![]() | ||
Fig. 15 (a–d) Schematic diagram, LSV curves, the comparison of Jk and E1/2, and the free energy diagram of Cu-SA/NC(meso). (e–h) Fourier-transformed EXAFS spectra, ORR polarization curves, Bader charge and corresponding chemical valence, volcano plot between ORR activity and ΔE0 for Cu–N©C-60. Reproduced with permission ref. 174. Copyright 2016, Royal Society of Chemistry. (i–k) Schematic diagram, free energy diagrams, and LSV curves of CuN2+2/C. Reproduced with permission ref. 181. Copyright 2022, Elsevier. (l–o) SEM image, polarization curves, Cu K-edge XANES spectra, and ORR overpotential volcano plot of S–Cu-ISA/SNC. Reproduced with permission ref. 182. Copyright 2020, Nature. |
Wu et al.176 synthesized N-doped graphene (Cu–N©C) with Cu(I)–N active sites by pyrolysis of ligand-saturated copper phthalocyanines. XPS showed that most of the Cu is present in the form of Cu+. As shown in Fig. 15e, the results of EXAFS showed that the number for the Cu–N©C-60 coordination was about 2 and Cu was present as CuN2 sites. Interestingly, treatment of Cu–N©C-60 with HNO3 solution resulted in a significant reduction in ORR activity after conversion of some of the Cu(I) sites to Cu(II) sites, indicating that Cu(I)–N(CuN2) was the active site for catalytic ORR (Fig. 15f). As shown in Fig. 15g, Bader charge analyses showed that the Cu valence states in the CuN2 and CuN4 fractions are closer to Cu3N (Cu(I)) bulk and CuPc (Cu(II)). The top of the volcano plot was Cu–N2, indicating that the Cu–N2 sites were the active sites (Fig. 15h).
The high ORR activity of CuN2 was also demonstrated by Li et al.177 It was well known that the adsorption energy of O2 and OOH* on the active sites and the elongation of the O–O bonds had a decisive influence on oxygen reduction.32,178,179 They investigated in depth the chemisorption capacity between O2 and Cu–N2. It showed that Cu–N2 had a suitable O2 adsorption energy and that long O–O distances were readily available on Cu–N2, making it easier for O2 to be activated. Xie et al.180 also compared the adsorption preferences of O2 on CuI@N2, CuII@N4, and C atoms: CuI@N2 > C atoms > CuII@N4. As a result, the ORR kinetics of Cu–N–C catalysts with CuI@N2 and CuII@N4 fractions were different. While CuII cannot directly engage in the ORR process, CuI may, and the Cu–N–C bond contains a C atom that is implicated in oxygen reduction. The ORR activity of Cu–N–C was therefore mainly derived from CuI@N2, which was also in general agreement with the findings of Wu et al.176
In addition to this, scientists have also investigated other CuNx conformations. Hu et al.181 synthesized a N-doped carbon support (CuN2+2/C) with CuN2+2 sites (Fig. 15i). DFT theoretical calculations showed that, in Fig. 15j, the changed electronic structure of Cu and the resultant orbital overlap caused by proper symmetry with the degenerate π* orbital of the adsorbed oxygen molecule allow the CuN2+2 sites to enable O2 activation. The PDOS of the Cu1 3d orbital showed that the d-band center εd of CuN2+2 was higher than that of CuN4, indicating that the Cu–O bonding on CuN2+2 is tighter. Also, the protonation barrier for PDS on CuN2+2 was lower. Therefore, CuN2+2/C had excellent ORR activity. As shown in Fig. 15k, CuN2+2/C had a more favorable half-wave potential.
Coordination with other non-metallic heteroatoms also resulted in stronger ORR activity of the Cu active center. In Fig. 15l, Shang et al.182 synthesized S–Cu-ISA/SNC with an unsymmetrically coordinated Cu–S1N3 moiety. S–Cu-ISA/SNC reflected a half-wave potential of 0.918 V (Fig. 15m). The Cu K-edge in situ XANES spectra for S–Cu-ISA/SNC were further investigated, and the edge position was gradually relocated to the lower energy, along with a reduction in the intensity of the white line, illuminating the significance of Cu valence alterations in ORR reaction (Fig. 15n). As shown in the volcano-type plot (Fig. 15o), the Cu atom in Cu–S1N3 had the best ORR activity, which was consistent with the experimental results.
![]() | ||
Fig. 16 (a and b) Aberration-corrected HAADF-STEM image and ORR polarization curve of the Zn–N–C-1. (c) Free-energy diagrams for Zn(OH)2, Fe(OH)2, and Fe(OH)3. Reproduced with permission ref. 183. Copyright 2019, Wiley-VCH. (d and e) Schematic models and comparison of the ORR performances of WN3, WN4, and WN5. (f) Illustration of the limiting potential (UL) “volcano” as a function of the change in ΔGOH. Reproduced with permission ref. 184. Copyright 2019, Elsevier. |
Mn SACs are also popular for ORR.185 Lin et al.186 prepared a Mn SAC (Mn-SA) with Mn-pyridinic-N4 sites with an E1/2 of 0.870 V. Comparing the three models of Mn–Nx (x = 1, 2, 3), it has been discovered that the RDS of Mn-pyridinic-N4 only had an energy uphill of 0.3 V at η = 0.4 V, indicating that the reason for the high ORR activity is the appropriate adsorption of intermediates. Shang et al.187 designed a carbon-based Mn–N2C2 electrocatalyst (Fig. 17a). In 0.1 M KOH, the E1/2 of this catalyst is 0.915 V (Fig. 17b). Four types of Mn active center with different N coordination numbers were constructed. DFT calculations showed that MnN2C2 was stable and the whole ORR process was exothermic, indicating a better ORR performance for Mn–N2C2 (Fig. 17c). In another study, by Shang et al.188 (Fig. 17d), the S-modified MnSAs/S-NC catalysts had high half-wave potentials (E1/2 = 0.916 V). The high ORR activity of MnSAs/S-NC was discovered to be the product of an electronic and atomic synergistic interaction between the Mn and the co-modified S,N carbon carrier. Mn-based catalysts containing Mn–N3O1 sites were subsequently produced by Yang et al.189 With the addition of O atoms, the d-bind center may be moved to a better location, resulting in purposeful ORR dynamics. Mn metal is employed for ORR reactions in an acidic environment, just like other transition metals like Fe, Co, etc. In Fig. 17e, the Mn–N–C catalyst prepared in the study of Li et al.190 showed excellent ORR activity in acidic conditions with E1/2 = 0.80 V (Fig. 17f). The DFT calculations result were the same as those observed by XAS, and MnN4 was the best active site (Fig. 17g). To prevent the oxidative corrosion of the catalyst caused by the Fenton reaction, Luo et al.191 obtained a Cr–N–C catalyst (Fig. 17h and i). In acidic conditions, the E1/2 reached 0.773 V in Fig. 17j. Additionally, Cr/N/C-950 showed greater ORR stability than Fe–N–C after 20000 cycles, with just a 15 mV decrease in the E1/2 being noticed (Fig. 17k). This also provided a new idea for solving the stability problem of traditional transition metal catalysts.
![]() | ||
Fig. 17 (a–c) Schematic illustration for the preparation, ORR polarization curve, and theoretical ORR and OER activity of MnSAC. Reproduced with permission ref. 187. Copyright 2020, American Chemical Society. (d) Schematic illustration for the preparation of MnSAs/S–NC. Reproduced with permission ref. 188. Copyright 2020, The Royal Society of Chemistry. (e–g) Schematic illustration for the preparation, ORR polarization curves, and calculated free-energy evolution diagram of Mn-NC-second. Reproduced with permission ref. 190. Copyright 2018, Springer Nature. (h–k) Schematic illustration, SEM, ORR polarization curves, and durability tests of the Cr SACs. Reproduced with permission ref. 191. Copyright 2019, Wiley-VCH. |
In contrast to precious metals and transition metals, the main group metals were often considered to be devoid of ORR catalytic activity.192 However, in nature, enzymes that play a key role in many important metabolic pathways and nucleic acid biochemistry had been found to contain group metal magnesium (Mg) cofactors,193,194 indicating that Mg metals have excellent potential to be used as efficient ORR catalysts. More recently, Bisen et al.195 produced an Mg–N–C catalyst by a very simple method using only Mg precursor and dicyandiamide (DCDA) (Fig. 18a). The Mg–N–C catalyst had ORR performance (Fig. 18b). Excitingly, Mg–N–C also exhibited excellent cycling durability: in Fig. 18c there was a loss of 16 mV in E1/2 after 10k cycles. Combined analysis by microscopy and spectroscopy confirmed that Mg–N–C was partially responsible for its high activity. Moreover, the ORR activity of Mg–N–C was proportional to the pyridine N and pyrrole N content at a certain ratio, and the defects on graphitic carbon also caused a faster electron transfer, which reduced the Rct and enhanced the ORR activity. DFT calculations by Liu et al.196 showed that the adsorption strength of oxygen intermediates on Mg is lower compared with Ca and Al, which were also main group metals. Additionally, the coordination of the main group metal Mg with N atoms can affect its catalytic activity, and Mg coordinated with two nitrogen atoms had close to optimal adsorption strength with intermediate oxygen species, thus obtaining the optimal ORR activity (Fig. 18d and e). To verify their calculations, Mg–N–C catalysts with MgN2C were synthesized (Fig. 18f). In Fig. 18g, the E1/2 of Mg–N–C was 0.91 V and the Eonset was 1.03 V. The above findings remind us that some methods that can enhance the ORR activity of transition metal–N–C can also be tried on the main group metals, which may lead to surprising discoveries.
![]() | ||
Fig. 18 (a–c) HAADF-STEM images, ORR polarization curves, and accelerated stability test of Mg–N–C/800. Reproduced with permission ref. 195. Copyright 2020, Elsevier. (d) The zoomed-in view of the onset potential versus ΔGOH*. (e) Adsorption free energies of OH as a function of the εp position of metal atoms for Mg cofactors, and also as a function of the highest O-occupied state of hydroxyl after interaction. (f and g) Schematic illustration and LSV curve of Mg–N–C. Reproduced with permission ref. 196. Copyright 2020, Nature. (h and i) Schematic illustration, free-energy diagram of ANG. Reproduced with permission ref. 198. Copyright 2018, American Chemical Society. (j) Schematic illustration of the Ca-N,O/C. Reproduced with permission ref. 199. Copyright 2021, Wiley-VCH. |
Al is the most prevalent metal on Earth, and can be found everywhere in our daily life. And Al is in the same family as B with low electronegativity, so the vacant 3pz orbital of Al can extract π electrons and form a positively charged Al active center like B.197 When O2 is adsorbed on Al, the O–O bond will be longer, indicating that Al-doped graphene may have good ORR activity. Qin et al.198 synthesized Al- and N-codoped graphene (ANG) by pyrolyzing Al precursor on N-doped graphene (NG) shown in Fig. 18h. It was found that ANG had an E1/2 of 0.85 V. And ANG had excellent stability: after 20000 s, as the current retention was 90%. Calculations, in Fig. 18i, found that among the three structures, ortho-ANG exhibited the lowest overpotential of ortho-, meta-, and para-ANG, demonstrating that the Al and N structure with direct bonding had the best performance and the lowest energy barrier. Liu et al.199 used O to replace the coordinating N atom, which induces higher positive charge density in the Ca central atom (Fig. 18j). The DFT calculations showed that the introduction of oxygen changed the electronic structure of the Ca active center, which can enhance ORR activity by enhancing the adsorption of O intermediates. The highest E1/2 = 0.90 V was in an alkaline environment.
These works demonstrate to us that the main-group-metal elements can be highly active and extremely durable ORR catalysts.
![]() | ||
Fig. 19 (a–c) TEM images, the durability test, mass activity and specific activity of PtCo@NC-10. Reproduced with permission ref. 207. Copyright 2017, Tsinghua University Press. (d and e) Schematic illustration, LSV and the number of transferred electrons of LP@PF-2. Reproduced with permission ref. 208. Copyright 2018, AAAS. (f) Schematic illustration of A-CoPt-NC. (g and h) Model of the configuration and top view of the charge densities of a(Co–Pt)@N8V4. Reproduced with permission ref. 209. Copyright 2018, American Chemical Society. (i) C-ZIF-CuPt catalyst preparation process. Reproduced with permission ref. 211. Copyright 2021, Elsevier. (j) Schematic illustration of PtFeNC. (k) Specific capacities of primary ZAB based on PtFeNC. Reproduced with permission ref. 212. Copyright 2018, American Chemical Society. |
Zhang et al.209 gained insight into the catalytic mechanism of their PtCo catalyst (A-CoPT-NC, Fig. 19f) by DFT calculations. After screening five different coordination structures, this revealed that the configuration with the lowest overpotential, a(Co–Pt)@N8V4, had the highest energy efficiency in Fig. 19g. The up-shifted d orbital in relation to the Fermi level will provide a strong binding between the catalyst and the adsorbate, and vice versa, according to the d band center theory of Nørskov et al.210 It was clear that a(Co–Pt)@N8V4 and oxygen had a stronger link because the energy of the Co 3d orbital in this compound was closer to the Fermi level than the energy of the Pt 5d orbital in this compound. As shown in Fig. 19h, due to the asymmetric distribution of platinum and cobalt, which polarized the surface charge near the active sites, the O2 reduction process will be accelerated by the electrons stored on the Co site, improving the ORR performance.
In addition to Co, scientists have doped other transition metal atoms (Cu, Fe, etc.) with Pt to obtain high-performance ORR catalysts. In Fig. 19i, Geng et al.211 calcined the Cu-doped ZIF-8 material at high temperatures and later loaded it with Pt atoms in an oriented manner, resulting in Cu–Pt C-ZIF-CuPt. The E1/2 of C-ZIF-CuPt was 0.874 V (vs. RHE). Zhong et al.212 encapsulated Pt into Fe-doped zeolite imidazolium salt skeleton cavities and subsequently obtained atomically dispersed PtFe SACs (PtFeNC) by subsequent pyrolysis (Fig. 19j). Importantly, the Zn–air battery (ZAB) showed a specific capacity of up to 807 mA h g−1 at a discharge current density of 10 mA cm−2 (Fig. 19k).
![]() | ||
Fig. 20 (a and b) Schematic diagram and free energy evolution diagram of Fe&Mn/N–C. Reproduced with permission ref. 223. Copyright 2022, Wiley-VCH. (c) Optimized atomic structures of Fe,Mn/N–C. Reproduced with permission ref. 214. Copyright 2021, Springer. (d–f) Schematic illustration, RDE polarization curves, energies of intermediates and transition states in the mechanism of ORR of (Fe,Co)/N–C. Reproduced with permission ref. 224. Copyright 2017, American Chemical Society. (g–i) Graphic illustration, HAADF-STEM, and LSV polarization curve of FeNi SAs/NAC. Reproduced with permission ref. 225. Copyright 2022, Elsevier. |
Yang et al.214 successfully prepared Fe,Mn/N–C electrocatalysts with bimetallic atom dispersion by implanting Mn–N in Fe/N–C through pre-polymerization and pyrolysis processes. Similar to the Fe&Mn/N–C prepared by Cai et al.,223 it showed an excellent ORR performance (E1/2 = 0.928 V in 0.1 M KOH and E1/2 = 0.804 V in 0.1 M HClO4). The results of DFT calculations showed that when O2 is adsorbed by Fe/Mn atom partners, it led to suitable bond lengths and suitable binding energies, thus reducing the dissociation energy barrier (Fig. 20c). Additionally, it can quickly break the bonds between M–OH and efficiently capture oxygen-containing intermediates, ensuring the regeneration of O* and OH* intermediates and boosting ORR kinetics by effectively reducing peroxide formation.
In another study, in Fig. 20d, Wang et al.224 constructed (Fe,Co)/N–C catalysts with Fe–Co double sites embedded on N-doped porous carbon. These exhibited excellent ORR performance (E1/2 = 0.863 V, Fig. 20e). More importantly, the activity of (Fe,Co)/N–C was significantly higher than that of the single active metal Fe–Nx–C and Co–Nx–C, indicating that the Fe,Co bimetallic site plays a key role in the ORR reaction. As shown in Fig. 20f, DFT calculations showed that the O–O bond was more prone to break due to the strong binding of O2 on the Fe–Co dual site. Besides, the potential barriers for the dissociation of O2 and OOH* into O* and OH* were lower than those of Fe–Nx–C and Co–Nx–C, which is consistent with the experimental results.
FeNi SAs/NAC catalysts were obtained from the pyrolysis of ZIF-8@FeNi(mIm)x precursor by Bai et al.225 (Fig. 20g). HAADF-STEM results showed that FeNi diatoms and Fe or Ni single atoms were distributed evenly on N-doped carbon (Fig. 20h). Consequently, the E1/2 of FeNi SAs/NAC was 0.91 V in 0.1 M KOH (Fig. 20i). The Fe/Ni metal 3d orbital interaction was considered to be the cause of its exceptional ORR activity. Furthermore, the Bader charge of Fe with Ni is higher than that of Fe without Ni, implying an electron transfer occurring between Fe and Ni. Furthermore, the FeNi SAs/NAC catalysts showed faster ORR dynamics in the RDS of four-electron transfer.
Bi element can effectively optimize the activity of peroxides and noble metals, but it is quite uncommon for it to increase the oxygen reduction activity of SACs catalysts. Jin et al.226 successfully synthesized a Fe/N–C catalyst with Bi–O forms (Fe/Bi-RNC) by using rod-like C3N4 templates (Fig. 21a). Due to its porous rod-like structure, the RNC not only exposed numerous active sites, but also facilitated charge transport. Also, the additional nitrogen doping introduced by the C3N4 template facilitates improved the redox kinetics of the carbon material. The optimal Fe/Bi-RNC catalyst had an E1/2 of 0.899 V (Fig. 21b). According to the calculations of DFT in Fig. 21c, the presence of Bi–O adjusts the electronic structure of Fe–N4–BiO, resulting in a significant reduction of the band gap and their synergistic impact speeds up the electron transfer in the step that determines the ORR rate, leading to a better oxygen reduction performance.
![]() | ||
Fig. 21 (a–c) Preparation process, LSV curve, and ORR free energy diagram of Fe/Bi-RNC. Reproduced with permission ref. 226. Copyright 2022, Royal Society of Chemistry. (d) Illustration of the formation of Zn/CoN-C. (e) Optimized geometry of O2 adsorption configuration on the ZnN4, CoN4, and ZnCoN6(OH) systems. Reproduced with permission ref. 221. Copyright 2019, Wiley-VCH. (f–h) Proposed synthetic protocol, LSV curve, and free energy diagram of Co&Ni@N/C. Reproduced with permission ref. 220. Copyright 2019, Science Press and Dalian Institute of Chemical Physics. (i) Schematic illustration of Fe1Se1-NC. Reproduced with permission ref. 227. Copyright 2022, Elsevier. |
In addition to the diatomic catalysts involving Fe, other combinations were found to have excellent ORR activity as well. In Fig. 21d, the Zn/Co–N–C catalyst prepared by Lu et al.221 exhibited high ORR performance with an E1/2 of 0.861 V. It was found that the electronic structure was adjusted due to the co-coordination of bimetallic Zn and Co with N and the distance between the O–O bonds increased, leading to easier cracking and thus enhanced ORR activity (Fig. 21e). In another research study, Mao et al.220 prepared Ni-doped Co–N/C catalysts using Ni-doped ZIF-67 as a precursor (Fig. 21f). In Fig. 21g, the Ni&Co@N/C exhibited an E1/2 of 0.895 V. The PDS of the 4e− ORR process was lowered by the synergy of Ni–Co bimetals according to calculations using DFT, and the adsorption energy on the catalyst surface was reduced due to the introduction of nickel metal (Fig. 21h).
1. Lack of advanced characterization techniques. Various analysis techniques have been used to characterize the fine structure of single-atom active centers, but they can only get an average coordination structure. The local structure of the single metal sites still vague. If metal centers with different coordination modes are present in one catalyst at the same time, it is difficult to determine the true active center. DFT calculations are often used to explore the reaction mechanisms of SACs in depth. However, the construction of accurate theoretical models also requires a defined local structure. Therefore, the development of more advanced and accurate characterization techniques is necessary to deeply explore the structure and mechanism of SACs.
2. The controllable synthesis of SACs is difficult. Although many effective methods such as coprecipitation, wet impregnation, etc., have been developed to prepare SACs, with the existing synthesis methods it is difficult to precisely synthesize the desired coordination configurations. Additionally, the increase of metal amount during the synthesis process can easily lead to aggregation, which hinders the preparation of high-density active sites. To resolve these limitations, new synthesis strategies should be developed to better control the structure of the SACs and increase the density of active sites.
3. There is still a gap to actual applications. Although the ORR activity and stability of SACs are now substantially improved, there are still a number of problems in practical applications, such as sluggish mass transport, large resistance, poor durability, low catalytic activity in acidic conditions, etc., which makes them even more limited in application. Therefore, more advanced SACs should be developed to meet the industrial demand.
4. More efficient support should be developed. For an electrocatalyst, on the one hand, the support can play the role of dispersing the metal active centers. On the other hand, the inherent properties of the support, such as the structure, specific surface area, electrical conductivity and stability have a significant impact on the catalytic performance. Therefore, selecting the appropriate support is critical for increasing the catalytic performance of SACs. Currently, carbon-based materials are the most commonly used support for SACs. More excellent materials with the characters of better electrical conductivity, a larger specific surface area and greater stability than the carbon-based materials should be explored to support SACs.
This journal is © The Royal Society of Chemistry 2023 |