Potent Charge‐Trapping for Boosted Electrocatalytic Oxygen Reduction

Metal‐free carbon‐based materials are considered to be one of the most promising alternatives to precious metal Pt‐based electrocatalysts. However, the electrocatalytic activity of heteroatom‐modulated carbon rarely reaches the level of metal‐based electrocatalysts. Here, electron‐rich carbon and abundant pyridinic‐N adjacent to C vacancies decorated with carbon nanosheets (E‐NC‐V) are synthesized and used as the host for boosting efficient oxygen reduction reaction. Rich pyridinic‐N structures adjacent to C vacancies work in synergy with electron‐rich carbon, which promotes the sharp decrease of |ΔGO*|, resulting in the balanced adsorption and dissociation of oxygen intermediates, and thus activating OO. This can be attributed to the abundant vacancies and d–p orbital hybridization between Zn and N/C. The E‐NC‐V catalyst drives the oxygen reduction reaction (ORR) via a 4e− transfer‐dominated pathway with a half‐wave potential of 0.87 V versus RHE in the alkaline solution, even superior to Pt/C. The assembled Al–air battery exhibits a high peak power density of 113 mW cm−2. This promising strategy sheds light on the design and fabrication of robust, rich‐density, and high‐performance active sites for the ORR. The work is expected to inspire future work on the role of electronic structure modulation and defect engineering for enhanced reaction kinetics.

tively replace the precious metal Pt catalysts. Therefore, it is essential to develop intramolecular structural modulation strategies. [6] The strong electronic effect among carbon networks plays a crucial role in preventing active site toxicity to achieve multiple electron transfers. [7] Especially for electrocatalytic reduction reactions, electron-rich active sites facilitate the reduction process of the reactants. [7b,8] Additionally, the electron state of catalytic centers is highly related to the adsorption strength of oxygen intermediates on the surface. [9] According to the Sabatier principle, suitable adsorption strength is required to allow easy adsorption and desorption of reaction intermediates. [10] The majority of current approaches to tuning adsorption energy in ORR rely on electronic and geometric effects (e.g., strain tuning, [11] ensemble effect, [12] and surface distortion). [13] Nevertheless, the electronic effect is always generated by heterogeneous elements, which are accompanied by visible geometric changes due to lattice mismatch. [14] This could mitigate and confuse the significant role of the electronic effect. Moreover, N atom doping changes the electronic structure of the carbon atoms, in which pyridinic-N profits the ORR process. [15] Meanwhile, planar defect sites also cause in-plane transfer localization of charge, interfering with the local density π-electrons and modulating electrocatalytic activity. Therefore, the construction of defect density with precisely controlled density and Metal-free carbon-based materials are considered to be one of the most promising alternatives to precious metal Pt-based electrocatalysts. However, the electrocatalytic activity of heteroatom-modulated carbon rarely reaches the level of metal-based electrocatalysts. Here, electron-rich carbon and abundant pyridinic-N adjacent to C vacancies decorated with carbon nanosheets (E-NC-V) are synthesized and used as the host for boosting efficient oxygen reduction reaction. Rich pyridinic-N structures adjacent to C vacancies work in synergy with electron-rich carbon, which promotes the sharp decrease of |ΔG O* |, resulting in the balanced adsorption and dissociation of oxygen intermediates, and thus activating OO. This can be attributed to the abundant vacancies and d-p orbital hybridization between Zn and N/C. The E-NC-V catalyst drives the oxygen reduction reaction (ORR) via a 4e − transfer-dominated pathway with a half-wave potential of 0.87 V versus RHE in the alkaline solution, even superior to Pt/C. The assembled Al-air battery exhibits a high peak power density of 113 mW cm −2 . This promising strategy sheds light on the design and fabrication of robust, rich-density, and high-performance active sites for the ORR. The work is expected to inspire future work on the role of electronic structure modulation and defect engineering for enhanced reaction kinetics.

Introduction
The oxygen reduction reaction (ORR) at the cathode of fuel cells/metal-air batteries is vital to deal with the cli-electron-rich environments are used to investigate the extent of electronic influence under real applications.
Herein, a potent charge-trapping strategy is proposed to regulate the adsorption strength of the oxygen intermediate. The Zn-induced d-p orbital hybridization effect was used to construct electron-rich environments with optimal electronic properties and chemical activity. Meanwhile, the evaporation of Zn produced an abundance of defects. This as-obtained electron-rich carbon and abundant pyridinic-N adjacent to N vacancies decorated with carbon nanosheets (E-NC-V) catalyst exhibit superior ORR activity (e.g., positive half-wave potential, small Tafel slope, and perfect four-electron pathway) to the market-available noble metal Pt/C catalyst. The rechargeable Al-air battery (AAB) with the E-NC-V cathode shows ranking performance with a high discharge voltage and gravimetric energy density. This is because the E-NC-V catalyst owns abundant pyridinic-N, C vacancies, and electron-rich carbon atom centers and features optimized electronic structure, leading to increased active sites and enhanced intrinsic activity. Theoretical calculations demonstrate that the enhancement of ORR catalytic activity of the E-NC-V catalyst stems from the highdensity pyridinic-N adjacent to electron-rich C site decreases the adsorption energy barrier of both *O and *OH. This work provides a simple design and fabrication of high-density defects CNMs and the essentiality of charge-trapping configurations for enhanced kinetics.

Results and Discussion
CNMs possess suitable hydrophobicity in line with ORR reactions, but suffer from low intrinsic activity due to limited active sites and difficult O 2 adsorption. In the first step, N atoms are introduced to dope to change the electronic structure of C atoms and construct abundant NC active sites to overcome the problem of limited active sites. To further enhance the intrinsic activity, the d orbitals of Zn atoms are utilized to hybridize with the P orbital of N/C, thus adjusting the electron assignment of the C/N, which facilitates the coupling with the P orbital of O and further activates OO. Finally, removing the Zn would generate defects and further enhance the performance of the catalyst (Figure 1a). To achieve the above multi-step enhancement strategy, the Zn-assisted pyrolysis with rich nitrogen coordination was adopted in this work, as shown in Figure 1b. Zn-containing organonitrogen precursors are obtained by a surfactant (P123) polymerization process. g-C 3 N 4 dispersion favors the avoidance of aggregation of Zn nanoparticles at high temperatures during the following reactions. The thickness of the nanosheets is increased in precursors. During the pyrolysis, the g-C 3 N 4 precursor decomposed into gaseous nitrogen-containing carbonaceous species (e.g., C 2 N 2 + , C 3 N 2 + , and C 3 N 3 + ), and Zn became melting particles. [16] In their Transmission electron microscope (TEM) images ( Figure S1, Supporting Information), C, N, and Zn elements are uniformly distributed in the nanosheets. All these facts illustrate the successful introduction of Zn into the carbon network. The gas release continuously generates defects. P123 is pyrolyzed to form a highly conductive carbon network. Meanwhile, the gas produces a large number of nanobubbles in the carbon matrix prompting the stretch of surrounding atoms and an increase in surface curvature. Density-functional theory (DFT) calculations are used to explore the possibility of controlling defect synthesis by selecting different defect sites with specific pyridinic-N doping configurations. [17] The formation energy (E f ) of each model was calculated and the feasibility of the theory was verified. Specifically, removing one C or graphite-N atom near pyridinic-N creates a vacancy. However, individual vacancies are unstable and tend to migrate and merge into more stable vacancies. According to this rule, the conversion E f from graphite-N to E5 is calculated to be 6.91 eV. This high value indicates that it may require more energy input, such as a higher heating temperature, to make the transition. The removal of C is much easier to take place due to their much lower formation energy, below 4 eV. The calculations in Figure S2, Supporting Information, theoretically explain the rational design of defects in the carbon structure, to achieve comprehensive control of specific N configurations and corresponding confirmed defect types.
Experimentally, different defect densities and electron-rich materials were chosen to verify the above statement, corresponding to nanocarbon flakes with E-NC-V and those with low-density C vacancies (NC-V and L-NC-V), respectively. The electron paramagnetic resonance (EPR) patterns of E-NC-V exhibit a stronger peak signal at a magnetic field value of 2.003 G compared with NC-V and L-NC-V. This indicates high-density defects of C in E-NC-V, which is in line with our designed defect strategy to introduce the electron-rich interface (Figure 2a). The deeper insights into as-prepared catalysts are obtained in combination with multiple structural characterizations. Figure   Figure 2. Structural characterization. a) EPR spectra, b) Raman spectra, c) Defect density, and d) PALS spectra, of E-NC-V and L-NC-V; e) HAADF image of E-NC-V with an acceleration voltage of 80 kV (Hexagons, pentagons, and heptagons were labeled in orange, green, and red, respectively); f) AFM image of E-NC-V; g,h) TEM; i) High-resolution TEM of E-NC-V.
S3, Supporting Information, shows that only the characteristic (002) and (100) diffraction peaks of graphitic carbon at around 25.5° and 44.0° are present in the X-ray diffraction patterns of E-NC-V, NC-V, and L-NC-V. This finding is supported by the metal Zn content (0.12 wt%) from the inductively coupled plasma optical emission spectrometer analysis. It is noted that the (002) peaks of E-NC-V and NC-V became broader, demonstrating a large number of graphene sheets in these two samples. [18] Meanwhile, the broad (002) diffraction peak indicates the defective nature of E-NC-V, which can be further demonstrated by Raman spectroscopy (Figure 2b). The band intensity ratios of the D band to the G band (I D /I G ) for E-NC-V, NC-V, and L-NC-V are 0.99, 0.96, and 0.95, respectively. D-band (≈1350 cm −1 ) and G-band (≈1580 cm −1 ) correspond to the defects and the vibration of sp 2 -bonded carbon atoms, respectively. [3d] This indicates that Zn created more effective defects as a measurable metal reduction medium during the carbonization process. We calculated the defect density of E-NC-V, NC-V, and L-NC-V based on the Raman spectrum ( Figure 2c). Remarkably, the E-NC-V exhibits the highest defect density (nD). Moreover, the defect density of E-NC-V can be further normalized according to the Brunauer-Emmett-Teller (BET) surface of the mesopores, which has a density of 4.6 × 10 12 cm −2 (Figures S4 and S5, Supporting Information). The defect density of NC-V and L-NC-V are 2.0×10 12 and 1.4×10 12 cm −2 , respectively, the increase in defect density facilitates the exposure of more active sites. Furthermore, the type and concentration of vacancies in the prepared samples were further determined by positron annihilation lifetime spectroscopy. Figure 2d shows the positron lifetime spectra of E-NC-V and L-NC-V, which have similar spectral shapes. Table S2, Supporting Information, shows the three lifetime components of E-NC-V. Thereinto, the longest lifetime τ3 (=1.5 ns) is attributed to the annihilation of positrons for large-size defects in the catalyst. Larger defects possess lower average electron density than smaller ones, which would lead to decreased annihilation rate and consequently increases the positron lifetime. The shortest lifetime of τ1 is generally ascribed to the free annihilation of positrons in a crystal. [19] However, the existence of small defects, for example, some mono-vacancies in disordered crystals can decrease the surrounding electron density, which result in the increased lifetime of τ1. [20] According to the theoretically calculated lifetime of positrons, the shorter lifetime component τ1 of E-NC-V is attributed to the C vacancies.
Aberration-corrected high-resolution transmission electron microscopy was used to visualize the defect regions of the catalyst. At low magnification (Figure 2e and Figure S6, Supporting Information), the representative images of the E-NC-V revealed the formation of holes in the graphene sheet presumably resulting from the removal of C or N. Meanwhile, various structural defects were observed proximal to the lattice vacancies. Combined with the atomic force microscopy (AFM) results in Figure 2f, the thickness of the E-NC-V nanosheets is about 1.6 nm, which is equivalent to ≈5 layers of graphene sheets. According to N 2 sorption isotherms, E-NC-V, NC-V, and L-NC-V, all exhibit combinational type I/IV curves, their respective BET specific surface areas are 393.1, 495.4, and 80.2 m 2 g −1 . E-NC-V has a similar total pore volume as NC-V (2.51 cm 3 g −1 ), but a substantially larger meso/macropores ratio than L-NC-V (0.41 cm 3 g −1 ). The structural properties of high-density defects can also change some physical properties of amorphous carbon materials. For instance, in comparison to the electrical conductivity (0.15 S cm −1 ) of L-NC-V, the electrical conductivity of E-NC-V was much higher (1.06 S cm −1 ). Scanning electron microscope images display that L-NC-V appears as a dense and regular lamellar structure ( Figure S7a, Supporting Information). Transmission electron microscope (TEM) images further verify nanosheet layer structure and HRTEM shows disordered lattice fringe ( Figure S7b,c, Supporting Information). In addition, the carbon source was replaced by P123, the morphology changed to an uneven surface and multilayer folded structure with more distorted and disordered lattice stripes. However, with the intervention of metallic Zn media, the morphology changes again to a nanosheet with an uneven but ordered surface, and the HRTEM results still show distorted and disordered lattice stripes (Figure 2g,h). P123 not only serves as a carbon source but also effectively regulates the pore size distribution. Interestingly, the specific surface area of the samples showed a positive correlation with their micromorphology. The uneven and folded structure of the surface can effectively increase the specific surface area, which is consistent with the BET results. As marked by yellow circles in Figure 2i, a large number of defects appeared on the surface of the carbon mesh. The vicinity of these defects can provide additional edges on the carbon mesh surface and overall increase the number of exposed edges. Besides, uniform C and N distribution on the basal plane is disclosed by elemental mapping (Figure S8, Supporting Information).
The surface chemistry and bonding configuration of the E-NC-V, NC-V, and L-NC-V catalysts were investigated by X-ray photoelectron spectroscopy (XPS). The XPS spectra also confirmed that the prepared catalysts contained only carbon, nitrogen, and oxygen due to the removal of metallic zinc by several times of hydrochloric acid wash ( Figure S9, Supporting Information). The high-resolution N 1s spectrum in Figure 3a exhibits four fitted peaks at about 398.3, 399.5, 400.8, and 402.3 eV, which are connected to, respectively, pyridinic-N, pyrrolic-N, graphitic-N, and oxidized-N. [21] The overall N content decreases from 11.69 to 8.27 at%. Specifically, E-NC-V possesses the highest proportion of pyridinic-N (32%) among these samples (Figure 3c). It has been speculated that zinc power prefers to react with defects of nitrogen-doped carbon during thermal treatment, which promotes the formation of pyridinic-N. Compared with E-NC-V, the content of N dopant in NC-V has decreased (8.27 at%), but the overall defective N content remains at a high percentage, with higher pyrrolic-N content. In addition, the content of pyridinic-N in L-NC-V (14%) is significantly reduced and the graphitic-N occupies the most. Notably, its nanosheet carbon network is due to the in situ replacement of C atoms by N atoms in the process, and the degree of defects is lower than that of E-NC-V and NC-V. The spectra of the C1s core level in Figure 3b can be well fitted to sp 2 CC (284.7 eV), sp 3 CC (285.9 eV), CN/CO (288.8 eV), and CN/CO (289.9 eV). [3d] Thereinto, sp 3 level reflects the defect degree of graphene. Notably, as for E-NC-V, the ratio of sp 3 CC is the highest, implying that the vacancies appeared at the sp 2 -hybridized C site (Figure 3d), which demonstrated that the electron density in heptazine rings was decreased due to the redistribution of electrons caused by vacancies, [22] consistent with Raman and EPR characterization. These results confirm the preparation of carbon nanosheets with high-density pyridinic-N due to the use of Zn powders. In addition, N-doped carbon materials always contain oxygen-containing groups, derived from thermally stable groups and adsorbed oxygen on the carbon surfaces.
To further explain the superiority of pyridinic-N, C vacancies and electron-rich nanosheets, a slab model of the E-NC-V is constructed based on the above results. By using the contours of the electron localization function, the charge density distributions are shown in Figure S10 asymmetric energy levels close to the Fermi level, indicating that there are plenty of unpaired electrons in this catalyst, which is consistent with the EPR results. In addition, the occurrence of abundant energy levels above the Fermi level of C indicates that there are many empty orbitals in the E-NC-V catalyst, which could accept the lone-pair electrons of O 2 . Conversely, each orbital of the C site is in a fully occupied state, indicating that C is an electron-rich state. [23] This unique defective dominant and strong electronic interaction structure is expected to produce a highly active ORR catalyst.
The as-prepared catalysts are evaluated for ORR catalytic performance using a rotating disk electrode system. Cyclic voltammetry (CV) tests were first performed on E-NC-V, NC-V, L-NC-V, and 20% Pt/C in 0.1 m KOH solution. According to Figure S11, Supporting Information, E-NC-V exhibits the most positive onset potential (E one , 1.0 V) and reduction peak potential (0.87 V), and the largest peak current density (2.5 mA cm -2 ), suggesting its superior ORR activity. Meanwhile, linear sweep voltammogram (LSV) curves further confirmed an excellent ORR activity of E-NC-V catalyst (Figure 4a). Additionally, E-NC-V reaches a more positive half-wave potential (E 1/2 ) of 0.87 V compared to 20% Pt/C (0.85 V), NC-V (0.81 V), and L-NC-V (0.63 V). The J k of E-NC-V (21.5 mA cm −2 ) at 0.85 V is higher than that of 20% Pt/C (10.2 mA cm −2 ), NC-V (4.2 mA cm −2 ), and L-NC-V (0.9 mA cm −2 ) (Figure 4b). Combined with XPS and EPR results, the N content (11.69 vs 8.27 at%) in the catalyst has negligible effects on the ORR activity, while the pyridinic-N vacancies showed a positive correlation with the ORR activity. It has been reported that the content of pyridinic-N in the carbon network can be modulated according to the pyrolysis temperature. [24] The annealed temperature of E-NC-V was increased to 1000 °C to obtain the E-NC-V (1000) catalyst. Its overall N content was 7.97 at%, still maintaining a highly graphitized structure with a pyridinic-N ratio of 33 at% ( Figure S12, Supporting Information).
The smallest Tafel slope of E-NC-V (70.6 mV dec −1 ) in Figure 4c demonstrates a good kinetic process. Similar ORR kinetic processes on E-NC-V, NC-V, L-NC-V, and 20% Pt/C are also revealed by the values of the electron transfer number (n) (Figure 4d and Figure S13, Supporting Information). The n of, according to the Koutecky-Levich (K-L) equation, the calculated value of electron transfer for the E-NC-V catalyst was close to 4.0, indicating a perfect four-electron ORR pathway with high catalytic efficiency (Figure 4d). [25] The electrochemically active surface areas of as-prepared catalysts are compared by testing the behavior of the double-layer capacitance (C dl ) in the non-Faraday region (Figures S14 and S15, Supporting Information). The C dl value of E-NC-V (8.2 mF cm −2 ) is higher than that of 20% Pt/C (7.3 mF cm −2 ). Differently, L-NC-V with the high N content shows an ultra-low C dl value of 0.7 mF cm −2 , which confirmed that the E-NC-V with electron-rich carbon atoms and defective pyridinic-N can accelerate the ORR process. Furthermore, when non-metals (ammonia) act as reduct- ants to replace Zn, the pyridinic-N content and ORR catalytic activity did not achieve the desired results ( Figure S16, Supporting Information). To systematically investigate the relationship between defect density, pyridinic-N content, and catalytic activity, we modulated the variables of temperature, ratio, and reductant ( Figure S17, Supporting Information). Interestingly, the defect density and pyridinic-N content contribute to the ORR performance individually (Figure 4e). To obtain insights into the active sites of E-NC-V, SCN − anions, which are known as a poisoning agent for blocking metal catalytic active sites, were introduced into the electrolyte. [26] The ORR current for E-NC-V shows inappreciable change upon adding SCN − anions ( Figure S18, Supporting Information), showing the absence of metal related active sites in E-NC-V. Therefore, the influence of 0.12 wt% Zn during the preparation process as well as the contribution to electrochemical performance of the catalyst can be ignored.
Electrochemical impedance spectroscopy results showed the lowest interfacial charge transfer resistance for E-NC-V ( Figure S19, Supporting Information), suggesting an increased pyridinic-N content in E-NC-V. This phenomenon enhances the electronic interaction between O 2 molecules and active molecules, thus promoting its ORR activity. When 3 m methanol is injected into the electrolyte, the Pt/C catalyst current immediately increased due to the oxidation reaction of Pt in methanol, however, only a slight disturbance of the current on the E-NC-V catalyst is observed (Figure 4g). The stability of the E-NC-V and Pt/C catalysts at 0.7 V is further tested using chronoamperometric (Figure 4h). After a long period of test over 24 h, current retention rate of E-NC-V (90%) is significantly better than Pt/C (80%). These performance indexes also rank E-NC-V among the top reported ORR catalysts in the alkaline electrolyte ( Figure 4f and Table S5, Supporting Information). These results demonstrated that the E-NC-V has not only superior activity but also excellent stability, which can be attributed to high-density pyridinic-N vacancies and electron-rich carbon active sites.
To further explore the origin of ORR activities, E-NC-V, NC-V, and L-NC-V models (Figure 5a) were established and DFT calculations were performed based on the 4e − ORR pathway. During the synthesis, the d orbitals of Zn will hybridize with adjacent p orbitals of C and N (Figure 5b). Five d orbitals are arranged into four d states: d z2 , d xy , d xz/yz (twofold degenerate), and d x2−y2 . Based on the orbital symmetry, the d Z2 orbital hybridizes with the p Z orbital of C, forming a strong energy splitting σ and σ* bond. Moreover, the d xz/yz orbital hybridizes with the p x /p y orbital of N to form two relatively weak π and π* bonds. The remaining d orbitals do not participate in the hybridization. [27] Compared to E-NC-V, the p states of N in L-NC-V is less localized, especially for p x and p y orbitals. The electronic states near the Fermi level of active C atoms in a variety of doped metal-free carbon-based materials are predominately contributed by the p z orbital for graphene. Meanwhile, the strongest p z orbitals are favorable for the adsorption of oxygen-containing intermediates. The distribution on the π orbital of the N atom just matches the electronic orbital distribution of the pyridinic N atom, thus demonstrating the suitable active site of defective pyridine N. The DOS results show that the π* orbital of O 2 molecule in E-NC-V model is the closest to the Fermi level, ascribing to the p orbital of O 2 interacting with the p orbital of C atom and forming hybrid states. [5,28] As demonstrated in Figure 5c,d, more electrons in E-NC-V are transferred from the C site to the O 2 molecule than that of L-NC-V and NC-V, corresponding to a strong bonding strength and thus full activation of O 2 . These results suggest that E-NC-V has significant ORR catalytic properties.
The free energy diagram of ORR elementary steps is further investigated at U = 1.23 V (Figure 5e). The limiting potentials of ORR on E-NC-V, NC-V, and L-NC-V are 1.13, 1.24, and 1.4 V, respectively, which indicate that the defect density and pyridinic-N content boost the ORR performance. The specific structural diagram of the four-electron reaction process is shown in Figure S20, Supporting Information. During the ORR process, functional groups that receive electrons are more likely to be active sites. Combined with electron localization function contour map ( Figure S21, Supporting Information), the CN and CC bonds are typical covalent bonds in all systems. The doped N and the vacancy offered the lone pair electrons which improve the chemical activity of the system. [29] The structure C gets more electrons than N, and it indicates better catalytic performances of C sites adjacent to pyridinic-N. Figure 5f-h shows that the C atom adjacent to N atom has electron enrichment. Therefore, the C atom becomes a more competitive active site, while the N atom may become a potential active site. Meanwhile, it is observed that the C atom at the position of multiple defective pyridinic NC structures receives more electrons, so it is easier to recombine with the oxygen-containing intermediates. As for E-NC-V and NC-V, the charge of adjacent C atoms increases by 1.24 and 1.10 e, respectively, while the L-NC-V only increases by 0.58 e. The proper defect density can break the linear proportionality of the carbon material, and thus the C atom adjacent to pyridinic N will be activated as the main site to adsorb *OH. Although the carbon defect structures and pyridinic-N sites have been demonstrated as active sites, respectively, the variation of ORR performance with C vacancies content is still indeterminate. The electron-rich C favors the activation of OO, and the high-density C vacancies provide more highly active sites to accelerate the ORR.
Based on ideal ORR performance of E-NC-V, the performance of the air-cathodes prepared from as-synthesized catalysts was further evaluated in a flowing AAB presented in Figure 6a. One AAB system consists of an air cathode, an aluminum anode, and the electrolyte, where the ORR is the main factor process that determines the overall battery performance. The E-NC-V-based AAB exhibits a consistent and high open-circuit voltage (OCV, 1.68 V). As expected, the power density of the E-NC-V cathode (113 mW cm −2 ) was substantially higher compared to 20% Pt/C (106 mW cm −2 ), NC-V (93 mW cm −2 ), and L-NC-V (85 mW cm −2 ) (Figure 6b). The discharge performance of as-prepared catalysts was evaluated at 100 mA cm −2 (Figure 6c). The E-NC-V cathode displays a higher discharging voltage than that of Pt/C and NC. Moreover, the energy density of E-NC-V-based AAB is 2893 Wh kg −1 normalized by the mass of consumed Al electrode, which exceeds that of 20% Pt/C-based AAB (2602 Wh kg −1 ) (Figure 6d). L-NC-V-based AAB underperformed in terms of energy density and discharge voltage, showing trends consistent with ORR performance. E-NC-V-based AAB was prepared to work stably during the discharge process by only replacing the electrolyte and aluminum anode. Interestingly, no significant voltage attenuation was noticed for E-NC-V-based AAB through of 50 h at a current density of 25 mA cm −2 (Figure 6e). L-NC-V catalyst has poor oxygen reduction performance due to the lack of effective active sites. Meanwhile, L-NC-V is assembled as air cathode to discharge an aluminum-air battery, the wettability with the aqueous electrolyte is poor and the stabilization process is longer. At the same time, the discharge voltage is consistently better than that of 20% Pt/C-based AAB during long cycles with less degradation. To fully evaluate the electrochemical performance, E-NC-V-based AAB has a high discharge voltage and excellent energy density compared to current commercial manganese dioxide and many recently reported air-cathodes ( Figure 6f and Table S6, Supporting Information). These results demonstrated the superior performances of as-fabricated ORR electrocatalysts, which can attribute to abundant active sites and structurally stable carbon networks.

Conclusion
In summary, the Zn auxiliary method as a potent charge-trapping strategy is utilized to produce stable carbon nanosheets with rich pyridinic-N adjacent to carbon vacancies and electronrich carbon atoms. The optimized E-NC-V with extreme defect density (4.6×10 12 cm −2 ) and promotional pyridinic-N content (32%) shows superior activity compared to most of reported metal-free electrocatalysts for ORR in an alkaline electrolyte (E 1/2 = 0.87 V). Also, as-assembled AAB shows excellent discharge ability and durability. More importantly, electrochemical measurements and DFT calculations confirm that the C atoms near the aggregated defective pyridinic-N efficiently activate the OO bond, facilitating the reduction of the adsorption potential of the decisive step (*OH) for ORR. The methodology developed in our work provides a "two birds with one stone" strategy to effectively create abundant defective and strongly charged atoms as active sites, which is an advancement in the development of CNMs for ORR. Prospectively, strong charge effect sites can also develop practical multifunctional electrocatalysts by coupling other sites, while facilitating the development of next-generation economical electrocatalysts.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.