Highly Dispersive Metal Atoms Anchored on Carbon Matrix Obtained by Direct Rapid Pyrolysis of Metal Complexes

Open AccessCCS ChemistryRESEARCH ARTICLE5 Sep 2022Highly Dispersive Metal Atoms Anchored on Carbon Matrix Obtained by Direct Rapid Pyrolysis of Metal Complexes Bing Huang, Minghao Wang, Chuxin Wu and Lunhui Guan Bing Huang CAS Key Laboratory of Design and Assembly of Functional Nanostructures, Fujian Key Laboratory of Nanomaterials, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350108 University of Chinese Academy of Sciences, Beijing 100049 , Minghao Wang CAS Key Laboratory of Design and Assembly of Functional Nanostructures, Fujian Key Laboratory of Nanomaterials, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350108 Fujian Normal University, Fuzhou 350108 , Chuxin Wu CAS Key Laboratory of Design and Assembly of Functional Nanostructures, Fujian Key Laboratory of Nanomaterials, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350108 and Lunhui Guan *Corresponding author: E-mail Address: [email protected] CAS Key Laboratory of Design and Assembly of Functional Nanostructures, Fujian Key Laboratory of Nanomaterials, Fujian Institute of Research on the Structure of Matter, Chinese Academy of Sciences, Fuzhou 350108 https://doi.org/10.31635/ccschem.021.202101353 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesDownload CitationsTrack Citations ShareFacebookTwitterLinked InEmail The oxygen reduction reaction (ORR) and oxygen evolution reaction (OER) are critical in energy conversion and storage devices. Single-atom catalysts (SAC) with atomic utilization efficiency are among the most promising candidates for oxygen reactions. Focusing on the controllable and efficient synthesis of SACs, we develop a direct rapid pyrolysis method that shortens the synthesis process from several hours to 1 min. This method can be applied to the synthesis of SACs with transition metals (Fe, Ni, Co, and Mn) with concentrations higher than 2.5 wt %. The optimized Fe, Ni-dual metal catalyst delivers excellent ORR and OER activity and stability. Significantly, the gap between the half-wave potential (E1/2) of ORR and the potential at 10 mA cm−2 of OER is only 0.69 V in 0.1 M KOH, which is one of the best results ever reported. The enhanced performance in dual-metal SAC is ascribed to the synergic effects between Fe and Ni atoms. Both breaker and solid-state zinc–air batteries based on the optimized catalysts show high powder densities and long-term cycling performance. This study opens up a new method to synthesize SACs and provide guidance for materials design. Download figure Download PowerPoint Introduction The oxygen-electrode catalyst in a rechargeable zinc–air battery (ZAB) should be active toward both the oxygen reduction reaction (ORR) and oxygen evolution reaction (OER).1,2 Unfortunately, most catalysts catalyze only the ORR or the OER and not both. Furthermore, the high overpotential in oxygen reactions (including the ORR and the OER) hinders applications in devices. Noble metals have proved to deliver excellent ORR and OER activity. For example, Pt can catalyze the ORR, while RuO2/IrO2 can catalyze the OER.1,3 However, the high cost of noble metals has limited their large-scale application. So it is vital to develop cheap metals-based materials that can provide high-performance catalysis at low costs. Recently, transition metal-based catalysts have attracted a great deal of attention due to their superior performance in catalysis.4,5 Especially, those containing highly dispersed active sites (including the single-atom sites and some sub-nano-sized metal clusters) with high atomic utilization efficiency are the most promising candidates for noble metals in electrochemical catalysis. In terms of an electrochemical catalyst for the ORR and the OER, Me-NX-C (Me = metal) single-atom sites have been widely reported.6–8 Thus, the development of Me-SAC (SAC = single-atom catalysts) is of great significance to address the crucial energy issues. Typically, thermal processes are imperative to synthesize Me-NX-C catalysts, and most researchers choose to anneal samples between 700–1100 °C in tube furnaces.9,10 This treatment deals with precursors at low heating rates, generally below 30 °C min−1, which brings about aggregations of metal atoms because the high temperatures and prolonged time facilitate the long-range diffusion of metal atoms.11 Consequently, secondary acid leaching processes are usually involved to remove the less-active metal aggregations. Then secondary pyrolysis processes are applied to heal the Me-NX-C sites damaged by the acid leaching.12 However, the above synthesis routines can only obtain a very low metal loading, mostly less than 1 wt %.12–15 Consequently, it is imperative to develop a facile and universal method to synthesize SACs with higher metal loading. According to a previous report,16 Joule heating can rapidly elevate and quench the temperature at a high rate by on–off switches, leading to the graphitization of carbonaceous materials within a short time. Hu et al.11 first designed the synthesis of SAC utilizing Joule heating. In their experiments, current pulses were used to directly heat the samples, and they proved that the transient heat helped to disperse the SAC. Similarly, Hong et al.17 conducted the rapid Joule thermal shocks on ZIF-8 (Co) precursors, and the as-synthesized Co SACs exhibited high ORR performance. However, the temperatures used in the above syntheses are too high (around 2000 °C) and uncontrollable, and thus the Me-NX-C structures may disintegrate. Besides, the metal atoms tend to aggregate under a high electric field, so it is challenging to yield a high metal-loading SAC with proper oxygen catalytic activity. Here, we develop a new rapid pyrolysis method with controllable temperatures by utilizing low-voltage welding currents. This method enables us to synthesize catalysts containing high metal-loading SAC within 1 min and produce SACs on a large scale. Specifically, this method can be applied to synthesize Fe, Ni, Co, and Mn-based SACs. A dual-metal Fe/Ni SAC prepared from the above method presents one of the best bifunctional catalysts with a gap of only 0.69 V between half-wave potential (E1/2) of ORR and potential at 10 mA cm−2 (E10) of OER in 0.1 M KOH. When assembled in breaker and solid-state ZABs, the FeNi/NC cathode showed excellent activity and durability. This work provides a new method for synthesizing SACs and guidance for their design and development. Experiments and Methods Synthesis of Me-Phenanthroline The metal salts used here were FeCl2·4H2O (Macklin, Shanghai, China), NiCl2·6H2O (Aladdin, Shanghai, China), Co(NO3)2·6H2O (Aladdin, Shanghai, China), and MnCl2·4H2O (Aladdin, Shanghai, China). 1,10-Phenanthroline (Macklin, Shanghai, China) was used as the ligand to coordinate with the metal ions. To prepare Me-Phenanthroline, 300 mg of Ketjenblack was mixed with 200 mg of 1,10-Phenanthroline. Then metal salts were added to increase the nominal metal weight from 0.8 to 4 wt %. The mixtures were carefully ground with ethanol in a mortar several times and then dried to collect the powders. Unless otherwise mentioned, the default nominal metal loading was 3.2 wt %. Rapid pyrolysis of Me-Phenanthroline The hollow graphite tubes (φinner = 4 mm, φouter = 6 mm, and length = 10 mm) were purchased from Sinosteel (Beijing, China), and iron clip electrodes were used to connect the graphite tube tightly. The low-voltage direct current (DC)-current generator (YD-400SS) was purchased from Panasonic (Osaka, Japan). After filling the graphite tubes with the precursors and fully evacuating the vacuum chamber, 50–100 A DC-currents passed through the graphite tubes for 10–100 s. After that, the currents were shut off, and the graphite tubes were cooled down to room temperature before refilling the air. The as-synthesized catalysts were denoted as Me/NC (Me = Fe, Ni, Co, and Mn). Unless otherwise mentioned, the default parameters for Joule heating were 60 A and 60 s. Since the whole process was reduced to 1 min, it is possible to be reproduced on a large scale. Materials characterization Transmission electron microscope (TEM) images were obtained with a JEOL 2100F (Tokyo, Japan). All the X-ray absorption spectra (XAS) were obtained at Shanghai Synchrotron Radiation Facility (Shanghai, China) and processed by Athena software. Raman spectra were recorded in the Horiba Jobin Yvon LabRAM ARAMIS system (Paris, France). The 532 nm laser was used as the light source, and the maximum values of G peaks normalized all the spectra. Powder X-ray diffraction (PXRD) spectroscopy was performed on a Miniflex 600 (Rigaku, Tokyo, Japan) equipped with Cu Kα radiation (λ = 0.154 nm) at the scan rate of 3° min−1. X-ray photoelectron spectroscopy (XPS) spectra were obtained on an ESCALAB 250Xi (ThermoFisher, Waltham, MA). An adventitious C1s peak at 284.8 eV was used to calibrate the spectra. Electrochemical measurement A CHI 760D Electrochemical Workstation was used in all the electrochemical tests. Typically, 5 mg of samples was dispersed in 0.95 mL of isopropanol, and then 0.05 mL of Nafion solution (5 wt %) was added. The ink was sonicated for 1 h before use. 20 μL of ink was dropped cast on the rotating disk electrode (RDE, 0.196 cm2) to form a homogenous catalyst layer. For the electrochemical ORR tests, O2 was pumped into 0.1 M KOH solution for 30 min before recording the data. Then cyclic voltammetry (CV) cycles were conducted at the scan rate of 100 mV s−1 in the potential range of 0.15–1.15 V until a stable CV diagram could be observed. After CV cycles, linear sweep voltammetry (LSV) curves were recorded at the scan rate of 5 mV s−1, and the rotation speed of 1600 rpm. LSV curves were also recorded in the N2 condition and used for background subtraction. For the electrochemical OER tests, N2 was pumped into 0.1 M KOH solutions. Then CV cycles were conducted at the scan rate of 100 mV s−1 in the potential range of 1.15–1.75 V until a stable CV diagram could be observed. After CV cycles, the OER LSV curves were recorded at the scan rate of 5 mV s−1 and the rotation speed of 1600 rpm in the potential range of 1.15–1.75 V. The solution resistances were obtained and corrected for all the LSV tests automatically by CHI software. Pt/C (JM, 20 wt %) was tested as the ORR reference material at a load mass of 0.28 mg cm−2. RuO2 (Adamas, Basilea, Switzerland) was tested as the OER reference material at a load mass of 0.1 mg cm−2. Because the Pt/C is sensitive to the Cl− ion, a salt bridge (filled with saturated KNO3 solution) was employed to alleviate impacts from the Ag/AgCl electrode. Note that all the potentials were displayed on the reversible hydrogen electrode (RHE) scale. The Faraday efficiencies (FEs) of OER were recorded on an rotating ring disk electrode (RRDE) configuration. The Pt ring was biased at a constant value of 0.2 V to fully reduce the dissolved oxygen from catalysts. The collection efficiency Nc for our RRDE was 0.34. The FE efficiency was calculated by: FE % = 100 * I R / ( I D * N c ) where the IR and ID represent the ring current and disk current, respectively. The square wave voltammetry (SWV) measurements were conducted in 0.1 M HClO4. The amplitude and frequency used here were 0.025 V and 10 Hz, respectively. It should be noted that SWV measurements were conducted in an acid solution because the OER currents severely disturb the SWV signals in alkaline solution. Zn–air batteries assembly For the breaker Zn–air batteries assembly, the solution of 6 M KOH containing 0.2 M zinc acetate was used as the electrolyte, and the Zn foil was directly employed as the counter electrode. A carbon paper (1 cm−2) was used as the support for loading the catalyst, and the load mass of the catalysts was controlled to be 1 mg cm−2 by dropping cast 200 μL ink on it as mentioned above. As the contrast sample, Pt/C and RuO2 were mixed with the ratio of 1:1, and the assembly procedures were the same. For the solid Zn–air batteries assembly, the gel in previous reports was used as the electrolyte,18 and 0.3 mm Zn foil was employed as the counter electrode. A carbon cloth (2 cm−2) was used as the support for loading the catalyst, and the load mass of the catalysts was controlled to be 1 mg cm−2 by dropping cast 400 μL ink on it as mentioned above. All the Zn–air battery tests were carried out on a CHI 660E Electrochemical Workstation. Results and Discussion The rapid pyrolysis was conducted in a home-built low-voltage DC-current heating device ( Supporting Information Figure S1). As illustrated in Figure 1a, a hollow graphite tube was used as both the container and heater for the materials inside. After the air evacuation from the vacuum chamber, a high DC-current passed through the graphite tube, and the precursor materials filling the graphite tube will be carbonized in 1 min by Joule heating. Supporting Information Figure S2 displays the real optical picture of the working reactor passing through a 60 A current. An infrared (IR) thermometer (Wuxi Shiao Technology, Jiangsu, China) was coupled to monitor the real-time temperatures, and the recorded temperatures are shown in Supporting Information Figure S3. As can be seen, unlike conventional tube furnaces with slow heating rates, the high DC current enabled rapid heating of the graphite tube to targeted temperatures within only a few seconds. Accordingly, the heating and cooling rates were calculated to be around 5000 K min−1. In addition, the temperatures (ranged from 900 to 1600 K) can be carefully controlled by heating at different DC-currents. A mediocre current of 60 A can heat the graphite tube to 1200 K, situated at the optimal temperature for synthesized Me-N-C catalysts, so this current value was chosen for the following experiments. Here, Me-Phenanthroline complexes were used as the precursors,19 and Ketjenblack was added to prevent these complexes’ aggregation during heating.20 After rapid carbonization at a 60 A DC-current, the obtained carbon catalysts were denoted as Ni/NC (from Ni-Phenanthroline+Ketjenblack), Fe/NC (from Fe-Phenanthroline+Ketjenblack), and FeNi/NC (from Fe/Ni-Phenanthroline+Ketjenblack). Details about the precursor preparations and rapid carbonizations can be found in the Experimental Methods section. Notably, the whole synthesis was reduced from several hours to 1 min. The Joule heat is sensitive to ohmic resistance, but repeatedly heating the graphite tubes may change their nature. So we also studied the Raman spectra of the graphitization degrees (related to the ohmic resistance) of the graphite tube used before and after the Joule heating. As shown in Supporting Information Figure S4, the graphite tube after five-time 60 A Joule heating showed no notable changes in its Raman spectra. However, after only one time of 80 or 100 A Joule heating, the graphite tube showed an increased defect concentration. Accordingly, we replaced the graphite tube after Joule heating three times (60 A) to prevent changes due to ohmic resistance. The morphologies of these samples were obtained by TEM. As shown in Figures 1b–1d, no metal-related clusters or crystals were observed in the samples, indicating the existence of highly dispersed metal atoms. However, the sample derived from the same precursor heated in the conventional tube furnace (900 °C for 2 h, 10 K min−1) contains many large Fe3C crystals ( Supporting Information Figure S5), implying our designed heating methods can efficiently disperse the metal atoms. In addition, the element distributions for FeNi/NC were investigated by high-angle annular dark-field scanning TEM (HAADF-STEM). As shown in Figure 1e, HAADF-STEM images demonstrate the homogeneous distributions of Fe and Ni atoms and that no metal cluster exists in the samples. Moreover, the bright spots observed in aberration-corrected HAADF-STEM image (Figure 1f) prove the existence of atomically dispersed metal sites. Figure 1 | (a) The illustration of rapid carbonization. (b–d) TEM images of Fe/NC, Ni/NC, and FeNi/NC. (e) HAADF-STEM images of FeNi/NC. (f) The aberration-corrected HAADF-STEM image of FeNi/NC. Download figure Download PowerPoint X-ray absorption near-edge structure (XANES) was used to characterize the metal atoms’ configurations. In Figure 2a, the Fe K-edge spectra of Fe/NC and FeNi/NC were plotted with Fe Foil. Before the edges, pre-edge bumps (marked in light red shadow) appear in Fe/NC and FeNi/NC, which can be attributed to the typical square-planar structure, reminiscent of the Fe-N4 structure.21 For the Ni K-edge spectra (Figure 2c), similar pre-edge bumps were also detected. Moreover, in the Fourier-transformed (FT) k1-weighted extended X-ray absorption fine structure (EXAFS) spectra (Figures 2b and 2d), only the peaks attributed to the M–N/O were observed (the XAS can not distinguish the M–N and M–O),22 verifying the single-atom natures. In their first derivative spectra ( Supporting Information Figure S6), the maximum for FeNi/NC exhibited positive shifts compared to Fe/NC or Ni/NC, implying the metal sites in FeNi/NC were more positively charged. Similar, more positively charged bi-metal SAC was reported before,23,24 and the decreased electron densities in metal sites were attributed to the spontaneous axial adsorption of oxygen-related species. Since the absorption toward oxygen species on Fe sites is too strong, the decreased electron density may alleviate the strong absorption and speed up the catalytic processes.24,25 Figure 2 | (a) Fe K-edge XANES spectra for Fe/NC, FeNi/NC, and reference Fe foil. (b) FT-EXAFS curves for Fe/NC, FeNi/NC, and reference Fe foil. (c) Ni K-edge XANES spectra for Ni/NC, FeNi/NC, and reference Ni foil. (d) FT-EXAFS curves for Ni/NC, FeNi/NC, and reference Ni foil. Download figure Download PowerPoint To synthesize the SAC from this method, one should be careful about the complex ligands, current values, total metal loading, and annealing times. Several control experiments were conducted to explore single-atom site formation during rapid pyrolysis. When using metal salts (FeCl2+NiCl2) instead of complexes as the precursors, the metal atoms tended to aggregate and form metal-related clusters or crystals ( Supporting Information Figure S7), consistent with the previous report.26 So, the nitrogen-contained ligands were crucial for forming single-atom sites during rapid pyrolysis due to these ligands separating the metal ions. When raising the DC-current to 70 A or above, some disturbing peaks related to Fe3C in XRD patterns began to appear because the over-high temperature can induce long-range atom diffusions ( Supporting Information Figure S8). When we increased nominal metal contents to 4 wt % (Fe/Ni = 3:1), new peaks attributed to Fe3C also began to show up ( Supporting Information Figure S9). Further, when lengthening the annealing time to 80 s or above, Fe3C started to appear in the XRD patterns ( Supporting Information Figure S10). Encouraged by the above successful synthesis of Fe/Ni SACs, we further expanded this method to other attractive transition metals (Co and Mn). As presented in the TEM images ( Supporting Information Figure S11) and XAS spectra ( Supporting Information Figure S12), we found no visible metal-related clusters or crystals, indicating that these designed rapid pyrolysis methods are universal for synthesizing atomically dispersed metal catalysts. XRD was employed to study the structure of the samples. As presented in Figure 3a, all the samples, including Ni/NC, Fe/NC, and FeNi/NC, presented only two peaks at 26° and 43°, the typical carbon peaks, and showed no peaks of metal species, consistent with the TEM results. Raman spectroscopy was employed to investigate the defect concentrations of the samples. In a typical Raman spectrum of carbon materials, the ratios of ID/IG and full width at half-maximum (FWHM) of D and G peaks are effective indicators for screening the graphitization degrees. Carbon materials with higher graphitization degrees tend to have lower ID/IG ratios and FWHM of D/G peaks.27,28 So, we fitted the Raman spectra with four peaks, and details are provided in Supporting Information Figure S13. Here, it should be noted that the FeNi/NC in the main text is at a Fe/Ni ratio of 3:1 (atomic ratio). As summarized in Figure 3b, when increasing the Fe/Ni ratios (atomic ratios in the precursor), both the ID/IG ratio and FWHM of D/G peaks tend to decrease. We explain the Raman results by suggesting the catalytic ability differences of different metals for carbon graphitization. Figure 3 | (a) XRD patterns of Ni/NC, Fe/NC, FeNi/NC, and carbon black. (b) The Raman ID/IG, FWHM of D peaks, and FWHM of G peaks for catalysts derived from various Fe/Ni ratios. (c and d) XPS- Fe 2p and Ni 2p spectra. Download figure Download PowerPoint XPS was also employed to explore the surface chemical environments. The Fe and Ni 2p spectra for three samples are shown in Figures 3c and 3d. As can be seen, the FeNi/NC show clear Fe and Ni signals, indicating the successful doping of the metal sites. The Fe contents determined from XPS of Fe/NC and FeNi/NC are 0.50 and 0.36 atom %. The Ni contents determined from XPS of Ni/NC and FeNi/NC are 0.46 and 0.15 atom %. The inductively coupled plasma mass spectrometry (ICP-MS) was further utilized to determine the actual metal contents (Table 1). In detail, the Fe contents in Fe/NC and FeNi/NC are 2.80 and 1.92 wt %, respectively. The Ni contents in Ni/NC and FeNi/NC are 2.51 and 0.55 wt %. It is worth mentioning that the metal contents here are higher than many reported SAC obtained by pyrolysis organic precursors. Here, we should also note that the XPS technique has limited detection depth on the surface, so the detected metal contents are slightly different from those determined from ICP. Table 1 | The Detailed Fe and Ni Contents Determined from ICP and XPS Samples Fe ( XPS/ICP)a Ni ( XPS/ICP)a Total ( XPS/ICP)a Ni/NC – 0.46/2.51 0.46/2.51 Fe/NC 0.50/2.80 – 0.50/2.80 FeNi/NC 0.36/1.92 0.15/0.55 0.51/2.47 aThe data obtained from XPS (atom %) is marked in bold, and the data obtained from ICP (wt %) is marked in italic. For all three samples, the high-resolution spectra of N1S ( Supporting Information Figure S14) can be deconvoluted into four species: the pyrrolic N, pyridinic N, graphitic N, and oxide pyridinic N species, respectively. As shown in Supporting Information Figure S10 and Table S1, the pyridinic N and graphitic N contents are analogous. Besides, the dominant pyridinic N in three samples can provide abundant coordination sites for metal atoms. In addition, the oxygen contents are also similar in the three samples (4.1 atom % for Ni/NC, 4.4 atom % for Fe/NC, and 3.8 atom % for FeNi/NC). The Brunauer–Emmett–Teller (BET) surface areas and pore size distributions were analyzed from the N2 adsorption–desorption isotherms ( Supporting Information Figure S15a). FeNi/NC exhibited a BET surface area of 425 m2 g−1, similar to Fe/NC (457 m2 g−1) and Ni/NC (405 m2 g−1). Pore size distributions were calculated from nonlocal density functional theory (NLDFT) methods. As shown in Supporting Information Figure S15b, all three samples had analogous pore size distributions centered at 5 nm. The enriched mesopores in our samples decreased the diffusion resistances and thus boosted the ORR and OER performances.29,30 As a result, our catalysts provide meaningful comparisons since the nitrogen content, the oxygen content, and the pore sizes are almost the same. The electrochemical ORR activities were evaluated in 0.1 M KOH by LSV. For the FeNi/NC sample with the best performance, it should be noted that the Fe/Ni ratio, total metal loading, heating current, and annealing times were optimized to be 3/1, 60 s, 60 A, and 3.2 wt %, respectively ( Supporting Information Figures S16–S19). As mentioned above, the higher metal loading, higher heating current, or longer annealing time would lead to the Fe3C nanocluster formations, thus decreasing the electrochemical catalytic performance. As shown in Figure 4a, Ni/NC exhibited the lowest ORR activity, while Fe/NC showed much improved ORR activity. However, FeNi/NC showed a significant enhancement of ORR activity with a half-wave potential (E1/2 = 0.89 V), similar to that of the commercial Pt/C (load mass: 0.28 mg cm−2, Johnson Matthey, 20 wt %). The calculated Tafel slopes showed the same tendencies ( Supporting Information Figure S20). Specifically, the Tafel slopes for Ni/NC, Fe/NC, and FeNi/NC were 61, 63, and 56 mV dec−1, indicating the fastest kinetics of FeNi/NC. Figure 4 | (a) ORR LSV curves. Note that the load masses for our materials are 0.5 mg cm−2, while the load mass of Pt/C was 0.28 mg cm−2. (b) OER LSV curves. Note that the load masses here are all 0.5 mg cm−2. (c) ORR LSV curves before and after the ADT test for ORR at 0.6–1.0 V. (d) OER LSV curves before and after the ADT test for ORR at 1.2–1.6 V. (e) The metal redox peaks were obtained from SWV. (f) Comparisons of the OER and ORR bifunctional activities.23,31–42 Download figure Download PowerPoint The inverse reaction of ORR is OER, which is initiated in more corrosive environments. According to previous reports, it is more challenging to acquire a satisfying OER than ORR for mono-metal SACs. This study accessed the OER performances at an RDE configuration with the catalyst load mass of 0.5 mg cm−2 in 0.1 M KOH. As shown in Figure 4b, Fe/NC and Ni/NC showed only negligible OER activity. In detail, to derive the current to 10 mA cm−2, Fe/NC requires an overpotential of 470 mV, and Ni/NC requires an overpotential of 450 mV. The overpotential required for Fe/NC and Ni/NC is much higher than that of the commercial RuO2 (Adamas, 99.9 wt %) catalyst. However, for the dual-metal FeNi/NC, the overpotential required to derive the current to 10 mA cm−2 is only 350 mV (E10 = 1.58 V), which is comparable to the commercial RuO2. It can be concluded that the incorporation of nickel atoms drastically enhances the OER activity. Intriguingly, whatever the Fe/Ni ratios were, the dual-metal FeNi/NC all enhanced OER performance compared to mono-metal NC ( Supporting Information Figure S16). Moreover, the FEs were tested under a RRDE configuration. As shown in Supporting Information Figure S21, the FeNi/NC showed a similar OER FE with that of RuO2. In practical devices, durability is an indispensable criterion to assess catalytic performance. The stability of FeNi/NC was evaluated by accelerated duration tests (ADT). The ORR ADT was cycled between 0.6–1.0 V, while the OER ADT was cycled between 1.2–1.6 V. After 3000 cycles, FeNi/NC showed 10 mV decline of the E1/2 and 5 mV decline of the E10 (Figures 4c and 4d). Compared to the FeNi/NC obtained from tube furnace heating (referred to as “Tube furnace FeNi/NC”), the rapid carbonized FeNi/NC exhibited better electrochemical ORR and OER performance ( Supporting Information Figure S22). The Tube furnace FeNi/NC had many less active metal-related crystals, so the formation of more active single-atom sites was restricted at the same metal weight. Carbon materials would be corroded during the OER ADT test, so it was also important to look at the carbon corrosion after the ADT tests. As shown in Supporting Information Figure S23, the electrochemical surface decreased a lot due to carbon corrosion. In terms of ORR activity after OER ADT tests, we found around 60 mV decrease of the E1/2. This decrease is typical for carbon materials since a similar decrease has been found in the Tube furnace FeNi/NC ( Supporting Information Figure S24). Interestingly, even with severe carbon corrosion, the E10 for OER did not decrease too much. As shown in Figure 4d, the LSV after ADT tests showed an even higher current density below 1.5 V, which is usually observed in bulk metal species. We propose that the metal atoms may have been transformed into bulk metal species, as they would be in MeOOH phases.43 It has been reported that iron-based catalysts are better than nickel-based catalysts in terms of ORR, and in our study, the Fe/NC admittedly far outperformed the Ni/NC and even some dual-metal FeNi/NC with low Fe/Ni ratios ( Supporting Information Figure S16). In previous DFT calculations, neighboring metal sites in the carbon substrates