Proton exchange membrane fuel cells (PMFCs) and metal-air batteries with desired energy density have exhibited considerable advantages as promising energy equipment substitute [1-3]. Oxygen reduction reaction (ORR) is the pivotal process on the cathodes of PMFCs and metal-air batteries, which yields large overpotential due to the kinetic retardation [4-7]. Unfortunately, Pt nanoparticles supported on carbon black (Pt/C), as the most effective commercial catalyst for ORR, suffer from unavoidable imperfection including unsatisfied high-cost and low-stability, impeding the practical transformational applications in electrical equipment [8-12]. To overcome the aforementioned shortcomings and serious environment pollution and energy crisis, highly active non-precious metal catalysts have been received wide attention for ORR, especially carbon-based nanomaterials [13-15].

The ORR performance can be promoted by optimizing the components, regulating the structures and morphologies, adjusting the physicochemical properties such as defects and crystal planes. For example, non-metal elements (B, N, P and S) have been added into various carbon-based nanomaterials (carbon nanotubes, carbon nanocages, carbon nanospheres, and graphene) to enhance the ORR performances [16-18]. Since Dai's group created vertically aligned N-doped CNTs as efficient ORR catalyst in 2009, the N-doping carbon materials have been rapidly developed for more than ten years [19-21]. The incorporated N atoms can redistribute the charge of carbon systems due to the larger electronegativity of N, which further activate the O₂ adsorption on the C atoms adjacent to N [22]. Transition metals (Fe, Co, Ni and Mo) incorporation into N doped carbon can fabricate the highly active M–N sites and further boost the ORR [23-28]. For example, FeCo–N_x carbon material with inner connected bimetal-N_x is highly active for ORR and Zn–air battery [29]. And well-accepted Ni(Fe)OOH and Ni/Fe–N–C active species for ORR were integrated together as high-efficiency oxygen catalysts [30]. Among them, Co and N co-doped carbon materials have been demonstrated as the most potential ORR catalysts with Co–N–C as the ORR active centers [31-36]. The central metals tend to disrupt the O═O bond and subsequently decrease the activation barriers for the oxygen reduction [37]. For example, Zhu et al. fabricated Co, N co-doped porous carbon nanosphere with polydopamine as the carbon and nitrogen source. The prepared catalyst is exactly effective for ORR with the half-wave potential of 0.93 V in alkaline media [38]. Hu et al. synthesized N-doped porous carbon nanofibers encapsulated with Co nanoparticles via electrospinning and high-temperature shock techniques [39]. The well-defined catalyst performs desired activity for ORR (half-wave potential of 0.93 V) and Zn–air battery (power density of 292 mW/cm²). In addition, the well-defined porous structures can greatly increase the density of available active sites to enhance ORR. Silicon is the most practical and common pore-forming template, which can realize the controllable design of porous structures [40-44]. The large surface area and pore volume stemmed from the elimination of SiO₂ template can facilitate the charge and mass transportation between the catalyst and the electrolyte [45]. However, preparing Co, N-codoped porous carbon with SiO₂ template as efficient ORR catalyst needs to be further explored to lay the foundation for commercial applications in Zn–air batteries.

Herein, N-doped mesoporous carbon loaded with cobalt nanoparticles (CoMCN) was synthesized as robust electrocatalysts for ORR in alkaline/acidic solution and Zn–air batteries. Inspiringly, plenty of mesopores on the carbon substrate can provide more accessible active sites and profitable charge/mass transport for ORR. The high content of pyridinic and graphitic N is beneficial for promoting O₂ adsorption and reduction. The smaller I_D/I_G value illuminates the higher degree of graphitization of CoMCN, providing better electronic conductivity. With the aforementioned superiorities, CoMCN performs satisfied ORR activity and durability under both alkaline and acidic conditions. In addition, the Zn–air battery composed of CoMCN performs the larger power density and open-circuit voltage than battery composed of Pt/C, indicating the potential application in energy conversion systems.

As depicted in Fig. 1a, precursor was firstly prepared via the assembly of pyrrole in the presence of SiO₂ nanospheres. The above precursor and Co(NO₃)₂·6H₂O were then dispersed into ethanol. The precursor contained Co (Co-precursor) was obtained by evaporating the ethanol under stirring at room temperature. Ultimately, CoMCN (Fig. 1b) was prepared after carbonization and elimination of SiO₂ template. TEM image (Fig. 1c) displays that CoMCN is uniformly dispersed with the diameter of about 200 nm. The pore distribution curve (Fig. S1 in Supporting information) exhibits the pore size of CoMCN is 12 nm, which is consistent with the diameter of SiO₂ template (Fig. S2 in Supporting information). The high-resolution TEM (HRTEM) image shows the crystal distance is 0.177 nm, which is belonging to the (200) crystal plane of metallic Co (Fig. 1d). The carbon layers outside the Co nanoparticles can be observed to be 1 to 3 layers. The around carbon layers can be activated by Co nanoparticles to enhance the ORR activity. Moreover, the surrounding graphitic layers would prevent the cobalt nanoparticles from corroding and dissolving into the harsh electrolyte. The XRD pattern (Fig. 1e) illustrates that the typical planes at 44.2°, 51.5° and 75.9° are belonged to metallic Co phase (PDF#15-0806). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) elemental mapping images (Fig. 1f) display C and N are uniformly distributed across the whole zone of CoMCN, and Co is concentrated on certain parts of Co nanoparticles. Therefore, the aforementioned characterization demonstrates the successful fabrication of CoMCN.

Subsequently, the chemical composition of CoMCN was employed by the XPS spectrum. The high resolution C 1s spectrum (Fig. 2a) can be divided into C—C and C—N peaks at 284.5 and 285.2 eV, respectively. The high resolution N1s spectrum can be concluded into pyridinic N (398.3 eV), Co-N_x (399.2 eV), pyrrolic N (400.4 eV), and graphitic N (400.9 eV), in which pyridinic N and graphitic N account for a large proportion of 76% (Fig. 2b). It is worth noting that pyridinic N and graphitic N are reported as profitable N types to boost ORR and effective attaching sites for cobalt [46, 47]. As displayed in Fig. 2c, the high-resolution Co 2p_3/2 spectrum can be fitted into three typical peaks. The peak located at 779.4 eV represents for Co⁰, which is consistent with previous work [48]. The peaks at 780.7 and 782.1 eV are indexed to CoC_xN_y and Co–N_x species, respectively, which can be attributed to oxidized Co by the air [49, 50]. The data demonstrate that Co is hybridized into the graphitized carbon and coordinated with N atoms. It was reported that cobalt coordinating with nitrogen can increase the density of state at Fermi level and reduce the work function, which subsequently facilitate the ORR process [51]. The Co K-edge X-ray absorption near edge structure (XANES) curves demonstrate that the chemical state of Co is between Co⁰ and Co³⁺ (Fig. 2d), suggesting the partial oxidation of cobalt. The Fourier transformed extended X-ray absorption fine structure (FT-EXAFS) curves in Fig. 2e suggest the co-existence of Co–Co (2.2 Å) and Co–N (1.5 Å) scattering paths in CoMCN [52]. Moreover, the N-K edge spectrum of CoMCN displays characteristic peak at 400.1 eV, which belonging to π* band (Fig. S3 in Supporting information). The split peak at 402.4 eV indicates the boosting of substitutional N in an asymmetric configuration of C–N [35].

Meanwhile, CoMCN and MCN were also detected by the Raman spectra (Fig. 2f). The typical peaks at 1358 and 1589 cm⁻¹ are corresponding to defective carbon and graphitic carbon, respectively. The value of I_D/I_G for CoMCN (0.85) is smaller than that of MCN (0.89), manifesting the superior degree of graphitization for CoMCN. The introduction of cobalt can promote the formation of graphitic carbon. The higher degree of graphitization offers better electronic conductivity, which may boost the charge transfer for ORR. The N₂ adsorption-desorption isotherms (Fig. S4 in Supporting information) were conducted to obtain the Brunauer-Emmett-Teller (BET) surface area (227.4 m²/g) and pore volume (0.35 cm³/g) of CoMCN. The hysteresis of the desorption curve implies the abundant mesopores in CoMCN, which is consistent with the result of pore distribution curve. The mesoporous structure of CoMCN with large BET surface area and pore volume can enhance the ability of mass transport and then promote the ORR performances.

The rotating ring-disk electrode (RRDE) curves of the catalysts annealed at different temperatures (Fig. S5 in Supporting information) suggest the sample annealed at 800 ℃ (CoMCN) performs the best ORR activity. The RRDE curves of CoMCN, MCN, and Pt/C in Fig. 3a prove that CoMCN performs more positive half-wave potential (E_1/2, 0.865 V), compared with that of MCN (0.735 V) and Pt/C (0.841 V). The onset potential (E_onset) of CoMCN is 0.964 V, which is more positive than that of MCN (0.875 V). CoMCN is one of the best Co, N co-doped carbon materials for ORR in alkaline solution (Table S1 in Supporting information). Tafel plot is the key parameter to evaluate the dynamic process for ORR, which equals to b in the equation η = a + blgj. As shown in Fig. 3b, the Tafel plots of CoMCN, MCN, and Pt/C are −109.8, −116.9 and −101.6 mV/dec, respectively. The smaller Tafel plot of CoMCN than that of MCN manifests the more favourable dynamic process for ORR on the CoMCN catalyst. The better catalytic activity and dynamic process of CoMCN than those of MCN can be ascribed to the introduction of abundant Co nanocrystals. The electron transfer number (n) and H₂O₂ yield of CoMCN (Fig. 3c) are evaluated to be 3.81–3.97 and below 4.7%, which are close to those of the commercial Pt/C. The above electrochemical data indicates that CoMCN displays efficient ORR activity and 4e⁻ dominated ORR process. The methanol tolerance of CoMCN (Fig. 3d) was examined by the addition of 1.0 mol/L CH₃OH solution into the electrolyte of 0.10 mol/L KOH. The insignificant transform of the CV curves before and after the addition of CH₃OH verifies that CoMCN is resisted with the disturbance of methanol. However, the typical oxidation peaks of methanol occurred to Pt/C (Fig. 3e) under the same operations. The results illustrate the better selectivity of CoMCN than that of the standard Pt/C.

Moreover, the stability of CoMCN was measured through the accelerated 5000 CV scans. There is slight shift of E_1/2 in the RRDE curves (Fig. 3f) for CoMCN. The slight activity decay of CoMCN may be the result of leaching out of superficial Co into the electrolyte during the reaction. However, there is 20 mV change of half-wave potential for the commercial Pt/C after 3000 CV cycles (Fig. S6 in Supporting information). The results indicate the better durability of CoMCN. In addition, TEM image and XPS data of CoMCN after stability test were acquired. As displayed in Fig. S7a (Supporting information), the morphology of CoMCN after stability test is basically consistent with that before test. The high resolution XPS spectra of C (Fig. S7b in Supporting information) and N (Fig. S7c in Supporting information) are also agreed well with those before stability test. Compared with that before test, the typical Co⁰ peak at 779.4 eV (Fig. S7d in Supporting information) accounts for a much smaller proportion, which can be attribute to further oxidation of Co. The results can demonstrate the desired stability of CoMCN for ORR.

RRDE curves of CoMCN (Fig. 4a) were acquired to evaluate the ORR activity in acidic solution. The E_1/2 of CoMCN is tested to be 0.730 V, which is approach to that of the commercial Pt/C (0.805 V). Tafel slope of CoMCN (−83.5 mV/dec) is comparable to that of the Pt/C (−76.7 mV/dec) catalyst, suggesting the favorable dynamic process for ORR in acidic solution (Fig. 4b). The H₂O₂ yield and n of CoMCN (Fig. 4c) are evaluated to be below 2.2% and 3.91-3.95, which are close to those of the standard Pt/C, evidencing the 4e⁻ dominated ORR pathway in acidic solution. Further, the tolerances of CoMCN (Fig. 4d) and Pt/C (Fig. 4e) toward methanol were characterized by pouring of 1.0 mol/L CH₃OH into the acidic electrolyte. The almost unchanged CV curves indicate the great anti-methanol ability of CoMCN in comparison with the commercial Pt/C. CoMCN also performs great durability for ORR in acidic solution. As shown in Fig. 4f, the successive 5000 CV cycles were conducted on the CoMCN modified electrode with almost no shift of E_1/2, illustrating the excellent durability of CoMCN catalyst. As a contrast, a significant decrease in activity of the commercial Pt/C can be observed after 3000 CV cycles (Fig. S8 in Supporting information).

The electrochemical double layer capacitances (C_dl) were obtained by conducting CVs of the CoMCN (Fig. S9a in Supporting information) and MCN (Fig. S9b in Supporting information) electrodes with a series of scan rates. The C_dl of the CoMCN and MCN electrodes are 40 and 29 mF/cm² via calculating the slopes of the lines in Fig. S10a (Supporting information). The results verify that CoMCN has larger effective electrochemical area with more available active sites than MCN, which can be helpful to increase the ORR activity to a certain degree. Besides, the Nyquist plots of CoMCN and MCN gained by EIS measurements in 0.10 mol/L KOH were shown in Fig. S10b (Supporting information). By measuring the radius of arc, the charge transfer resistance (R_ct) of CoMCN is 79 Ω, which is smaller than that of MCN (121 Ω). The data proves that CoMCN possesses better electronic conductivity which indicates a faster electron transfer for ORR. The introduction of metallic Co into the carbon framework gives rise to better electron conductivity and larger electrochemical surface area which can facilitate the ORR process.

To further prove the experimental conclusion, the physical mechanism for ORR will be investigated by the theoretical calculation with atomic level. To fully meet the experimental configuration, cobalt encapsulated by one graphene layer (Co@1layer) and three graphene layers (Co@3layers) will be considered with pure graphene model as contrast (Fig. S11 in Supporting information). The calculation hydrogen electrode (CHE) mode, which based on free energy changes, is employed to estimate the activity for ORR. The free energy curves for ORR on graphene, Co@3layers and Co@1layer are shown in Fig. 5a, while the geometric structures for intermediates are shown in Fig. S12 (Supporting information). For adsorption structures, it is found that the intermediate OOH will be favorable to adsorb at the C atom rather than the N atom. This result indicates that the main active site is C atom. For the free energy curves, the Co@1layer will show higher activity than the other two configurations, while the potential determining step is O₂ → OOH with the ΔG of −0.62 eV. For graphene and Co@3layers, the ORR activity will also be limited by the uphill process of O₂ → OOH with the ΔG of 0.10 and 0.03 eV, respectively. The results confirm that the ORR activity would be improved by cobalt encapsulation, which is well in agreement with the experimental observation. Furthermore, the electronic structures were calculated to explore the physical mechanism of high activity for Co@1layer. For the partial density of state (PDOS), it is noted that the p orbital contributes the main state around the Fermi level, indicating that the p orbital will affect the adsorption of intermediate. Moreover, it is seen that the PDOS for the p orbital of Co@1layer is obviously more delocalization than the other two configurations (Figs. 5b-d). This suggests that the conductivity of Co@1layer will be improved. The differential charge displays that there is strong electron transfer between Co and the nearest graphene layer (Figs. 5e and f). In other word, electron will transfer from the low electronegativity of Co atom (1.9) to the high electronegativity of C (2.5) or N (3.0) atom, and improve the conductivity of C atom in graphene. Based on the above results, it can be concluded that the encapsulation of Co can facilitate the electron transfer from Co to the carbon layer and increase the electronic DOS at Fermi level. Herein, the graphene layer is activated and the ORR activity is improved, which strongly support the experimental conclusion. However, such effect will be obviously reduced as the graphene layers increased (≥ 2 layers).

The desired oxygen reduction activity of CoMCN urges us to further determine its performance in Zn–air batteries. The battery was constructed by using a CoMCN air cathode and a Zn anode (Fig. 6a). As shown in Fig. 6b, the battery composed of CoMCN air cathode performs a high open-circuit voltage of 1.464 V, which is exactly higher than the Pt/C-based one (1.353 V, Fig. S13 in Supporting information). More importantly, the CoMCN air cathode also delivers relatively high specific capacity of 628.3 mAh/g and gravimetric energy density of 721.7 Wh/kg_Zn, which outperforms the commercial Pt/C cathode (specific capacity of 585.6 mAh/g, energy density of 651.9 Wh/kg_Zn, Fig. 6c). The CoMCN enables battery to yield a maximum power density of 102.8 mW/cm² at 110.0 mA/cm², which is bigger than that of standard Pt/C (70.4 mW/cm² at 120.0 mA/cm², Fig. 6d). In addition, the voltage plateaus for the CoMCN cathode under a current density range of 1–30 mA/cm² are specifically bigger than those of the Pt/C-based system (Fig. 6e). As an example of practical application, a red-light-emitting diodes (LED) lamp can be steadily illuminated without brightness decay over 2 h (Fig. 6f).

In summary, the well-designed mesoporous structure of CoMCN together with the large BET surface area and pore volume is in favor of exposing more active sites for ORR. The better electron conductivity and larger electrochemical surface area can be attributed to the incorporation of Co nanoparticles into the carbon nanosphere substrate. CoMCN not only performs better ORR activity, stability, and methanol tolerance than the commercial Pt/C in basic solution, but also shows desired ORR activity in acidic solution. In addition, the desired performances of the CoMCN-based Zn–air battery make it as one promising candidate in the application of commercial energy conversion devices.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The authors are grateful for the financial support from the National Natural Science Foundation of China (Nos. 21974097 and 21675147), the Education Department of Guangdong Province (Nos. 2020KSYS004 and 2020ZDZX2015), Science and Technology Bureau of Jiangmen (No. 2019030102360012639), and the Science Foundation for High-Level Talents of Wuyi University (No. 2018RC50).

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.107815.