Longevous Cycling of Rechargeable Zn‐Air Battery Enabled by “Raisin‐Bread” Cobalt Oxynitride/Porous Carbon Hybrid Electrocatalysts

Developing commercially viable electrocatalyst lies at the research hotspot of rechargeable Zn‐air batteries, but it is still challenging to meet the requirements of energy efficiency and durability in realistic applications. Strategic material design is critical to addressing its drawbacks in terms of sluggish kinetics of oxygen reactions and limited battery lifespan. Herein, a “raisin‐bread” architecture is designed for a hybrid catalyst constituting cobalt nitride as the core nanoparticle with thin oxidized coverings, which is further deposited within porous carbon aerogel. Based on synchrotron‐based characterizations, this hybrid provides oxygen vacancies and Co‐Nx‐C sites as the active sites, resulting from a strong coupling between CoOxNy nanoparticles and 3D conductive carbon scaffolds. Compared to the oxide reference, it performs enhanced stability in harsh electrocatalytic environments, highlighting the benefits of the oxynitride. Furthermore, the 3D conductive scaffolds improve charge/mass transportation and boost durability of these active sites. Density functional theory calculations reveal that the introduced N species into hybrid can synergistically tune the d‐band center of cobalt and improve its bifunctional activity. As a result, the obtained air cathode exhibits bifunctional overpotential of 0.65 V and a battery lifetime exceeding 1350 h, which sets a new record for rechargeable Zn‐air battery reported so far.


Introduction
The development of efficient bifunctional electrocatalysts is the top priority for rechargeable Zn-air battery technology.However, the slow kinetics and large overpotentials of oxygen reduction and evolution reactions (ORR and OER) significantly limit battery performance. [1]he precious metals (Pt, Ir, or Ru) based catalysts demonstrate unifunctional activity and insufficient stability.Furthermore, the scarcity of these precious metals significantly raises the total cost of the Zn-air system, hindering their widespread applications as the advanced energy conversion and storage technology.Extensive efforts have been devoted to developing cost-effective and environmentally benign bifunctional catalysts through employing non-noble transition metals (TM) such as Co, Ni, Fe, and Mn.The current trend of catalyst design is tuning the electronic structure of active sites and enabling their favorable adsorption/desorption behavior of electrocatalytic intermediates (such as O*, OH*, and OOH*) leading to reduced ORR/OER overpotentials.As such, various approaches to manipulating the electronic surface of catalytic sites have been developed, including alloying, [2] defect engineering, [3] and heteroatom doping. [4]Regarding electrocatalyst-design strategy, activity-wise, designing selective catalytic sites possessing excellent activity toward ORR and OER is significantly crucial for the successful development of efficient electrocatalysts.2a,5] Despite the recent progress in the improvement of ORR/OER activities, a well-balanced performance of both competitive activity and durability is still challenging, primarily due to the unfeasibility of a bifunctional active site and limited electrochemical stability.2b,3a,b,4a,5,6] Besides the voltage gap between battery charge and discharge, the harsh battery environment also leads to significant performance degradation.In this view, it is worth noting that Zn-air battery durability is considerably correlated with the electrocatalytic activity, in which the smaller overpotential would provide the electrocatalysts with more favorable environments, enabling a longer lifetime of the Zn-air battery.Meanwhile, the boosted mass transfer of reactants and reaction intermediates improves cycling stability by preventing congestion and retaining the initial active surface.This makes the design and fabrication of porous structured catalysts highly essential.Accordingly, the principal research spotlight lies on the strategic design of electrocatalysts to make a massive breakthrough in improving the Zn-air battery cycle life exceeding 1000 hours.From standpoint of durability, the catalytic stability of an active site is not guaranteed during unfavorable reactions, particularly for OER process, resulting in severe degradation and poor durability.To mitigate the stability issues, it is required to establish a hybrid catalyst involving the selective active sites that can lower overpotentials toward OER and ORR, to achieve sufficient bifunctionality. [7]More importantly, robust interaction between the selective active sites should be achieved for facilitated charge transfers from one to another and vice versa during the oxygen electrocatalysis, [2b,5] which is significantly beneficial to retaining the initial ORR/OER activities for the longevous Zn-air battery life.
Herein, a hybridization strategy is proposed for synergistic integration of engineering catalytic sites and constructing hierarchical porosity, which accomplishes an efficient bifunctional electrocatalyst for Zn-air battery with long-term cycling life.It is designed consisting of cobalt nitride nanoparticles with thin cobalt oxynitride covering (Co 4 N@CoON) deposited on porous carbon aerogel (PCGN).The as-obtained "raisin-bread" like Co 4 N@CoON/PCGN demonstrates remarkable bifunctional activities in both ORR (discharge) and OER (charge) with longterm stability in half-cell and electrically rechargeable Zn-air battery system.The synergistic hybridization of OER active "raisin" Co 4 N@CoON nanoparticle and ORR active "bread" porous carbon aerogel enables reduced overpotentials and longevous stability.The novel design not only provides robust interaction of Co 4 N@CoON particles with neighboring PCGN, but also renders robust defect-rich surfaces involving oxygen vacancies and fruitful N-containing species.The emergence of Co-N x -C active sites and the strong adhesion from enhanced charge transfer between the CoO and Co 4 N and optimize adsorption energies of oxygen species by efficiently modulating the d-band center confirmed by the first principles density functional theory (DFT) calculations, resulting in significantly improved battery performance.When integrated in the Zn-air battery, the rational hybrid electrocatalyst demonstrates remarkable bifunctional performances with 0.65 V of voltage gap and extraordinarily longevous cycle life of exceeding 1350 hours.

Design and Physicochemical Characterization of Bifunctional Hybrid
Schematic illustrations of the facile synthesis and the structure of Co 4 N@CoON/PCGN are presented in Figure 1a and Figure S1 (Supporting Information).Graphene oxide (GO) and pretreated carbon nanotube (CNT) have been adopted for constructing a three-dimensionally porous carbon aerogel via selfassembly of GO. [8] Oxygen functional groups (OFGs) such as OH − and COOH − on GO and CNT serve as anchor sites for cobalt crystallizations.Particularly, "in situ nitridation" is enabled by thermal decomposition of nitrogen-rich polymer, polyaniline (PANI), which generates highly reactive NH 3 gas during hightemperature pyrolysis. [9]The nitridation process leads to the generation of cobalt nitride (Co 4 N) nanoparticles as well as nitrogendoped porous carbon networks.Significantly, Co-N x -C active sites are formed via a strong interaction between the porous carbon aerogel and the anchored cobalt crystals.Furthermore, it has been observed that a thin oxidized layer (i.e., CoON) is formed encapsulating Co 4 N in an oxidative environment stemming from the plenty of the OFGs.Particularly, pretreated CNT not only acts as a charging bridge for faster charge transfers but also as a spacer preventing GO from self-stacking which leads to significant improvements in surface area and porosity in the resultant aerogel.Scanning and transmission electron microscopy reveal the "raisin-bread-like" structure of Co 4 N@CoON/PCGN as demonstrated in Figure 1b and Figure S2 (Supporting Information).The cobalt particles are well-distributed on the wide graphene sheet with an average particle size of ≈50 nm.Interestingly, there are in-plane pores generated near the cobalt particles (Figure 1c), which are formed probably due to Co-catalyzed gasification of carbon into carbon monoxide (CO) or dioxide (CO 2 ), where the initially electrostatically bonded cobalt ions would be thermally reduced and crystallized into cobalt nanoparticles while the adjacent carbon atoms are oxidized and released as the gases to generate holes. [10]It is also probable that NH 3 decomposition has led to the Co-catalyzed gasification of carbon to the formation of methane (CH 4 ). [11]The in-plane pores would provide additional defects and edge structures in the carbon networks enabling extra nitrogen doping, which will be beneficial to electrocatalytic activity.Moreover, the abundance of internal channels provided by the in-plane pores likely decreases the diffusion length for reactants, intermediates, and products, significantly improving mass transport.The in-plane pores are also confirmed (indicated by red arrows) in a dark-field scanning TEM image (Figure 1d) and corresponding EDS elemental maps (Figure 1e-g and Figure S3, Supporting Information).The clearly overlapped signals of Co and O exhibit the oxidized surface of cobalt nanoparticles.
Electron energy loss spectroscopy (EELS) elemental maps and corresponding spectrum (Figure 1h and Figures S4 and S5, Supporting Information) have been obtained, which reveals that atomic distributions of Co and N are exactly overlapped, indicating cobalt nitride.Particularly, O is densely located at the edge of the cobalt nanoparticle revealing a thin oxidized layer covering the center particle.The high-resolution TEM (HR-TEM) image (Figure 1i,j) reveals the crystal structure of the Co particles.The lattice spacing has been measured and indexed from multiple spots (Figure 2i and Figures S6-S8, Supporting Information).In Figure 1j, location (i) shows 3.4 Å that corresponds to the (002) plane of graphitic carbon derived from the PCGN.Interestingly, different lattice distances are observed in which locations (ii), (iii), and (iv) present 2.49, 2.07, and 1.79 Å, respectively, which likely stems from CoO x N y and CoN x .It is worth noting that these observed interplanar distances are slightly larger than the reference spacings of CoO (2.46 Å and 2.13 for (111) and ( 200) lattices, respectively, JCPDS No. 09-0402) and metallic Co (2.05 Å and 1.77 Å for ( 111) and ( 200) planes, respectively, JCPDS No. 15-0806).This elucidates that the Co lattices are expanded by the interposition of N and O atoms during particle crystallization, eventually leading to the generation of Co 4 N and oxidized shells consisting of CoO x N y .A fast Fourier transform (FFT) image obtained at this particle (Figure 1i, inset) discloses the coexistence of crystalline lattices of both Co 4 N and CoO x N y .Meanwhile, reference samples have been prepared in various conditions including CNT-free (Co 4 N@CoON/PGN), Co-free (PCGN), Co and CNT-free (PGN), PANI-free (Co@CoO x /PCG), and acidleached (Co-N x /PCGN) to investigate roles of each component in Co 4 N@CoON/PCGN (Figures S9-S13, Supporting Information).As expected, the in-plane pores are observed only in Coinvolved samples, which further verifies that Co-catalyzed gasification forms the unique porous structure.However, the size of cobalt particles is irregularly distributed and considerably larger in Co@CoO x /PCG (Figure S12, Supporting Information), which is obviously due to the absence of nitrogen (PANI).From the results, it could be concluded that N atoms significantly affect the crystal structure of cobalt nanoparticles where the crystallized CoN x restrains further growth of particles by competing with oxygen atoms.The acid-leaching process removes the cobalt particles, only leaving the in-plane pores (Figure S13, Supporting Information).
X-ray diffraction (XRD) patterns have been collected to explore specific crystal structures of the developed catalysts (Figure S14, Supporting Information).While clear graphitic carbon peaks for all samples are revealed at 26°and 42°corresponding to (002) and (100) planes, respectively, Co-included samples exhibit metallic Co (111, 200, and 220 facets, JCPDS  No. 15-0806) and a tiny portion of CoO (111 and 200 facets, JCPDS No. 09-0402).However, no cobalt peak is observed in the acid-leached sample (Co-N x /PCGN).Interestingly, in comparison with Co@CoO x /PCG exhibiting apparent metallic cobalt, Co 4 N@CoON/PCGN demonstrate slightly negative shifted diffraction angles (Figure S14e,f, Supporting Information).Such a shift is an indication of the expansion of the crystal lattice of metallic Co, identifying the formation of Co 4 N. [11] Due to the specific 4:1 ratio of Co and N atoms, the face-centeredcubic (FCC) structure of metallic Co retains in Co 4 N, and only the lattice distance between adjacent cobalt atoms is expanded, leading to the negative shifts.It is worth noting that XRD patterns of Co 4 N and cobalt oxynitride (CoON) closely correspond with the crystal structure of metallic Co as reported in the recent literature, [12] suggesting the negatively shifted peaks stem from Co 4 N and/or partially CoON, which is well consistent with the above HR-TEM results.N 2 adsorption-desorption isotherms are obtained to investigate the relationship between the structural effect of aerogel and the porosity and specific surface area (Figure S15, Supporting Information).8a-d] However, the twodimensional GO layers typically undergo self-stacking that considerably reduces structural porosity and active surface, which is detrimental to electrocatalytic activities.Therefore, pretreated CNT has been adopted as a spacer to minimize the disadvantage.CNT also acts as "charge bridges" for faster transfers of electrons attributed to its 1D structure and intrinsically high electrical conductivity. [13]As a result, Co 4 N@CoON/PCGN and PCGN demonstrates much larger surface area (295.3 and 419.3 m 2 g −1 ) compared to Co 4 N@CoON/PGN and PGN (188.5 and 294.9 m 2 g −1 ).The pore size distribution results (insets of Figure S15a,b, Supporting Information) present the same trend, where a higher average pore volume (0.451 and 0.494 cm 3 g −1 ) for Co 4 N@CoON/PCGN and PCGN than those of Co 4 N@CoON/PGN and PGN (0.302 and 0.355 cm 3 g −1 ), respectively.The improved specific surface area and porosity clearly demonstrate that the role of CNT is successfully applied to our catalyst design strategy, which is crucial for enhanced electrocatalytic performance.Raman spectroscopy is conducted to investigate the degree of structural defects (Figure S15c,d, Supporting Information).All samples display two high-intensity peaks corresponding to G-band (≈1575 cm −1 ) and D-band (≈1342 cm −1 ).A higher I D /I G ratio generally indicates more defect level of heteroatom-doped carbon materials. [14]14a,15] The additional peaks at the region of lower Raman shifts indicate the formation of Co-O species possibly due to CoO x N y . [16]he sharp 2D peaks observed in Co 4 N@CoON/PCGN and PCGN, compared to their CNT-free counter samples, are majorly attributed to the high graphitization degree of the multiwalled CNT. [17]The amount of Co involved in the samples is investigated by thermogravimetric analysis (TGA) and inductively coupled plasma optical emission spectrometry (ICP-OES) (Figure S15e,f, Supporting Information).The amount of cobalt in both Co 4 N@CoON/PCGN and Co 4 N@CoON/PGN presented 29 wt%, while ICP-OES results showed 22.98 wt% of Co remained in Co 4 N@CoON/PCGN after leaching (see details in Supporting Information).9a,14a,18] X-ray photoelectron spectroscopy (XPS) is performed to investigate the chemical states and electronic environments of elements in the samples.XPS survey spectra of all samples are shown in Figure S16 (Supporting Information).As expected, clear cobalt peaks are observed in Co-involved samples including Co 4 N@CoON/PCGN, Co 4 N@CoON/PGN, and Co@CoO x /PCG, while a slight intensity Co peak is observed in Co-N x /PCGN.Interestingly, compared to Cofree catalysts (PCGN and PGN), higher atomic ratios of nitrogen and oxygen are revealed in both Co 4 N@CoON/PCGN and Co 4 N@CoON/PGN (Table S1, Supporting Information), which is mainly ascribed to the in-plane pores providing abundant edge defects that can host additional N-doping and the formation of Co 4 N@CoON particles, and possibly the Co-N x -C moieties.It has been reported that the formation of interfacial Co-O-C and Co-N-C bonds is most likely due to the coupling of cobalt oxide species and N-doped GO sheets, which can be applied to Co 4 N@CoON/PCGN. [19]he high-resolution N 1s, Co 2p 3/2 , and O 1s peaks have been deconvoluted (Figure 2) to explore the chemistry of each element in detail.Regarding N 1s (Figure 2a), it is notable that a deconvoluted peak at 399.2 eV is observed in Co 4 N@CoON/PCGN compared to the Co-free PCGN.14a,18a] The peak also observed in Co-N x /PCGN indicates Co-N x -C is left after the severe acidleaching, corroborating the existence of Co-N x -C moieties in Co 4 N@CoON/PCGN (Figure S17a,b, Supporting Information).Co 2p 3/2 peaks of Co 4 N@CoON/PCGN and Co@CoO x /PCG are compared (Figure 2b) to investigate the effects of in situ nitridation.Above all, Co 4 N@CoON/PCGN shows a positive peak shift of 0.4 eV in comparison to Co@CoO x /PCG, which results from strong electron interaction between cobalt and surrounding elements such as N and C. [11,20] Co (II) and Co (III) observed in both samples arise from the Co species such as Co 4 N, CoON, CoO, and Co 3 O 4 as confirmed in XRD results above.Interestingly, the Co-N x peak at 778.5 eV is observed only in Co 4 N@CoON/PCGN, elucidating the oxidized layer (CoON) is thin and accordingly the centered metallic Co 4 N is possible to be detected.In contrast, the metallic Co detection is hindered by the thick oxide layer in Co@CoO x /PCG, which demonstrates that the in situ nitridation confines further oxidation of cobalt particles and possibly hinders particle agglomeration.Additionally, the slight positive shift (≈0.5 eV) of the Co-N x peak (778.5 eV) in Co 4 N@CoON/PCGN compared to a pure metallic cobalt binding energy (778 eV) further confirms the formation of Co 4 N. [14a] Meanwhile, O 1s spectra of Co 4 N@CoON/PCGN and Co@CoO x /PCG are compared (Figure 2c).The peak at 529.8 eV is attributed to the cobalt oxide lattice bonds (O-Co), and the peak at 531.3 eV corresponds to O-defects resulting from oxygen vacancies (OV) due to low oxygen coordination. [21]The remarkably larger peak area ratio for O-defects to the O-Co in Co 4 N@CoON/PCGN (56:11.3)than Co@CoO x /PCG (48.5:26.7)21c,22] With the higher oxygen vacancies formed due to the introduction of nitrogen during the high temperature synthesis, it has been confirmed that the valence of cobalt is decreased from Co (III) to lower valence states of Co (II) and Co-Nx. [23]While the area ratio of Co (III) and Co (II) from Co@CoOx/PCG is 43.8:56.2, the result from Co4N@CoON/PCGN demonstrates 35.3:64.7 of the area ratio of Co (III) and the lower valance states, which further corroborates the increased oxygen defects in Co4N@CoON/PCGN (Figure 2b).3c,24] As presented in Figure S17c, Supporting Information, the lattice oxygen (O-Co) peak is vanished in Co-N x /PCGN, demonstrating all of oxidized cobalt particles have been successfully leached out.To further investigate the electronic state and local atomic structure of the Co 4 N@CoON particle, element-selective X-ray absorption fine structure (XAFS) analysis is performed at the Co K-edge.Co K-edge XANES spectra of Co 4 N@CoON/PCGN is compared with reference materials including Co foil, CoO, and Co 3 O 4 (Figure 2d).25a,26]

Bifunctional Oxygen Electrocatalysis Performance
activities are observed to investigate electrocatalytic performance.A systematic optimization of the oxygen catalytic activities is preconducted to understand the unique properties of each precursor (Figure S19, Supporting Information).Regarding ORR activity, Co 4 N@CoON/PCGN demonstrates most positive half-wave (E 1/2 ) potential (0.855 V) and the highest limiting current (j L ), compared to the other catalysts (Figure 3a, Figure S20a and Table S2, Supporting Information).In comparison to Co 4 N@CoON/PGN (PGN), the significantly improved ORR performance of Co 4 N@CoON/PCGN (PCGN) proves the positive effect of CNT as the charge high-way and the spacer which enables boosted electron transfer and enhanced exposure of active sites, respectively. [13,27]Compared to the Co-free samples, Co 4 N@CoON particle in Co-involved samples demonstrates a remarkably enhanced ORR performance (Figure S20a, Supporting Information).This is majorly attributed to the Co-N x -C sites formed through the strong interaction between the Co 4 N@CoON particle and the surrounding porous carbon aerogel (PCGN).14a,18a,28] This is further confirmed by observing ORR activity of the acid-leached sample, where Co-N x /PCGN presented competitive ORR activity with Co 4 N@CoON/PCGN, verifying the slight residue of Co is majorly in the form of Co-N x -C moieties.The ORR kinetics of the developed materials was further investigated by measuring Koutecky-Levich (K-L) slopes (Figure S21, Supporting Information).The number of electrons transferred during ORR on Co 4 N@CoON/PCGN (3.9) is the highest among the samples, which is close to the ideal fourelectron reduction pathway.Co-N x /PCGN reaches at 3.8, demonstrating the excellent kinetics of Co-N x -C sites during ORR.
The bifunctional capability of the catalysts has been confirmed by evaluating OER performance (Figure 3b and Figure S20b and Table S2, Supporting Information).The OER overpotential at 10 mA cm −2 (E j = 10 ) is adopted as a standard for performance comparison.With the lowest overpotential (365 mV), Co 4 N@CoON/PCGN demonstrates its excellent OER activity, compared to the reference samples (Figure 3b).Co-N x /PCGN exhibits similar OER overpotentials to PCGN (439 mV and 427 mV, respectively), which corroborates that the Co 4 N@CoON nanopar-ticle is the selective active site for OER.Furthermore, comparing the improved OER activity of Co 4 N@CoON/PCGN (365 mV) with Co@CoO x /PCG (408 mV) clarifies the significant contribution of the in situ nitridation toward OER performance.As confirmed in the material characterizations, Co 4 N@CoON particle holds abundant oxygen vacancies derived from the cointerposed N and O, which could tune the adsorption energy of OH − and its reaction intermediates.Generally, cobalt based oxides such as Co 3 O 4 , CoOOH, and CoO x S y are known to be good catalysts for facilitating OER. [11,29]22a,30] The abundant oxygen vacancies enhance defect states near the Fermi level, promoting electron injection/extraction from O 2 ensuring rapid exchange kinetics of O 2− /OH − , leading to optimized oxygen adsorption properties and improved the oxygen electrocatalysis . [31]Regarding the porous structure, much larger overpotential is observed on PGN and Co 4 N@CoON/PGN in comparison to PCGN and Co 4 N@CoON/PCGN, which is attributed to the enlarged surface and porosity derived from the contribution of CNT within the 3D structured aerogels.In addition, the electrochemical active surface area (ECSA) of the electrocatalysts is estimated based on the electrochemical double-layer capacitances (C dl ) obtained from the scan-rate dependent CV measurements (Figure 3c and Figure S22, Supporting Information).Co 4 N@CoON/PCGN presents 26.3 mF cm −2 , much enhanced ECSA than other catalysts.This shows quite different trend from physical specific surface area and porosity obtained by BET analysis.The variation demonstrates that the entry of cobalt and the in situ nitridation synergistically establish active sites such as the O-defectsrich Co 4 N@CoON particle and the defective Co-N x -C as well as the additional defects at the in-plane pores.Interestingly, Co-N x /PCGN exhibits higher ECSA (29.1 mF cm −2 ) probably due to the more active sites stemming from the additionally generated pores and higher mass loading of the remaining catalyst after etching the Co 4 N@CoON particles.The practicality of Co 4 N@CoON/PCGN is revealed by comparing its performance with the best-known precious metal-based ORR/OER electrocatalysts, Pt/C and RuO 2 , respectively, where their physical mixture is adopted as a bifunctional benchmark catalyst (Figure 3d,e).Regarding ORR, despite the slightly more positive onset potential in Pt/C + RuO 2 , E 1/2 of Co 4 N@CoON/PCGN demonstrates a positive shift of 12 mV, suggesting better ORR kinetics.Tafel slops are obtained to investigate the ORR kinetics (Figure 3f).Pt/C + RuO 2 shows a slope of 81 mV decade −1 that is larger than 40 mV decade −1 by Co 4 N@CoON/PCGN, highlighting the remarkable ORR kinetics of Co 4 N@CoON/PCGN.With respect to OER, Co 4 N@CoON/PCGN demonstrates competitive activity to Pt/C + RuO 2 (Figure 3e).While onset potential of Co 4 N@CoON/PCGN is higher than the precious benchmark, the overpotential measured at 10 mA cm −2 (E j = 10 ) for Co 4 N@CoON/PCGN (365 mV) is considerably reduced compared to Pt/C + RuO 2 (390 mV).Moreover, the gap between their OER currents in higher potential region gradually increases, indicating better OER kinetics of Co 4 N@CoON/PCGN.The enhanced OER activity at high potential makes a catalyst more practically viable in Zn-air battery systems.The lower Tafel slope of Co 4 N@CoON/PCGN (63 mV decade −1 ) than that of Pt/C + RuO 2 (181 mV decade −1 ) further confirms the better OER kinetics of Co 4 N@CoON/PCGN (Figure 3f).To elucidate bifunctional capability, total ORR/OER overpotential gaps between E 1/2 and E j = 10 are compared (ΔE over = E j = 10 − E 1/2, Figure 3 g and Table S2, Supporting Information).Co 4 N@CoON/PCGN demonstrates the lowest ΔE over (0.74 V) among the developed catalysts, which is comparable to Pt/C + RuO 2 (0.777 V).This result verifies the synergistic effects of the in situ nitridation and the highly porous carbon aerogel, which provides rich O-vacancies and Co-N x -C moieties, as well as enlarged active surface with expanded porosity, leading to the most favorable ORR and OER performances in Co 4 N@CoON/PCGN.Electrochemical stability is also critical criterion for evaluating bifunctionality of the catalysts.It is investigated by conducting chronopotentiometry (CP) in the OER potential region at a fixed anodic current density of 5 mA cm −2 (Figure 3 h and Figure S23a, Supporting Information).Compared to Pt/C + RuO 2 exhibiting considerable degradation of electrocatalytic durability before 5000 s, Co 4 N@CoON/PCGN demonstrates significant OER stability throughout 30 000 s with negligible overpotential increase, which corroborates the excellent viability of Co 4 N@CoON/PCGN as a practically usable electrocatalyst.The bifunctionality of the catalysts was examined after the durability test (Figure 3i and Figure S23, and Table S3, Supporting Information).Co 4 N@CoON/PCGN presents only 9% of overpotential ascent, while over 64% of increase is observed with Pt/C + RuO 2 .As the high OER potential region majorly degrades catalysts, the catalysts coated nickel foam electrodes have been prepared and utilized as cathodes within the half-cell testing.The CP polarizations are investigated for over 50 h, and the initial and final OER activities are obtained by testing the individual Co 4 N@CoON/PCGN and Pt/C + RuO 2 electrodes.Co 4 N@CoON/PCGN exhibits only 20 mV of potential increase while Pt/C + RuO 2 shows 85 mV of potential growth, indicating severe durability degradation (Figure S24b, Supporting Information).Accordingly, compared to the initial OER activities, Co 4 N@CoON/PCGN retains 97% of current retention, whereas Pt/C + RuO 2 loses 44% of its peak current density (Figure S24a,c, Supporting Information).To further clarify the excellent OER stability of Co 4 N@CoON/PCGN, we adopt accelerated degradation testing (ADT), repeated cyclic voltammetry (CV) cycling test in ORR/OER potential region to investigate oxygen reaction degradation (Figure S24d, Supporting Information).As a result, 2000 CV cycles have been applied to the activated electrode, and the final current retention turns out to be 95%, which is well consistent with the stability results above.To explore catalyst surface structure change after the harsh and long-term durability test, high-resolution XPS results have been obtained from the tested Co 4 N@CoON/PCGN electrode.The Co 2p deconvolution result exhibits consistent Co (II), Co (III), and Co-N x peaks (Figure S24e, Supporting Information), where, however, the area of Co-N x peak is reduced likely because the protection oxidized shell would be further oxidized leading to thicker oxynitride layers and less core detection.Similarly, the O 1s deconvolution result demonstrates a higher peak area ratio (74.5:8.4 for Odefects: lattice-O), which elucidates the increased thickness of the shell oxynitride (Figure S24f, Supporting Information).The corresponding STEM with elemental mapping results demonstrates that the oxide shell has been enlarged, while the core particle is well-retained (Figure S24g,h, Supporting Information).
Based on the various physicochemical and electrochemical characterization results, it can be concluded that the excellent durability of Co 4 N@CoON/PCGN is attributed to synergistic effects by (i) the protective and robust cobalt oxynitride layer; (ii) the selective active sites consisting of the abundant O-vacancies and the Co-N x -C moieties derived from the in situ nitridation, leading to rapid exchange kinetics of O 2− /OH − ; (iii) the enlarged active surface and porosity as well as the enhanced charge and mass transportations via the contribution of CNT-involved graphene aerogel.The synergy of the three specific contributions leads to stable electrocatalysis within the harsh OER potential region, efficient selective bifunctional activities, and sustained suppression of OER overpotentials, resulting in a relatively less oxidative environment and thus longer electrochemical stability.As observed in the various characterization results, the Co-N x -C moiety is regarded as the key ORR active site in Co 4 N@CoON/PCGN.18a,28] In this perspective, to investigate the existence of Co-N x -C moiety in Co 4 N@CoON/PCGN, ORR polarization curves of Co-N x /PCGN in acidic media (0.5 m H 2 SO 4 ) have been obtained and compared with Co-free counter samples such as PGN and PCGN, while Co 4 N@CoON/PCGN is excepted since the Co 4 N@CoON particle is not stable in acidic condition.As expected, Co-N x /PCGN demonstrates remarkably improved ORR activity and kinetics (Figure S25 and Table S4, Supporting Information) with the highest E 1/2 and K-L slop, which successfully verifies the emergence of Co-N x -C sites in Co 4 N@CoON/PCGN.

Theoretical Investigations on Active Sites for OER and ORR
DFT calculations are performed to provide further understandings of the respective active sites on the synthesized catalysts toward OER/ORR.According to the experimental characterization and previous literatures, [32] the heterostructure of Co 4 N(111), the most stable surface, covered by four different types of thin CoO (Fm 3m) layers, CoO(100), CoO(111), two CoO(100), and two CoO(111), are selected as for OER active sites to give deeper insight the and thickness of oxide layer (Figures S26 and S27, Supporting Information).To evaluate effect of N in Co 4 N, CoO//Co models are also proposed for comparison and included for further trend analysis between adsorption energies and OER activities.
The theoretical OER activity for the five models are predicted by Gibbs free energy diagram using widely accepted four-electron mechanism (see the Computational methods for details).OER activity of the heterostructures is drastically enhanced compared to that of Co 4 N (Figure 4a and Figure S28, Supporting Information).The rate-determining step (RDS) is determined as ΔG 3 (O* → OOH*) for all models and the trend of OER activity is stated from the most active to the least with the liming potential (U L ) in parenthesis as follow: CoO(111 To understand how the electronic interactions between layers influences OER activity, electronic structures of Co, adhesion energies, and adsorption energies are explored (Figure 4b,c, and Figure S29, Supporting Information).Since the number of charge (q) transferred from Co in Co 4 N increases to respective CoO surfaces, the adhesion energy became stronger.Also, the more electron is transferred from Co 4 N, the further d-band center of Co in CoO was downshifted from fermi level, which weakens adsorption energy. [33]he resultant tuned adsorption energy demonstrates the lowest overpotential with CoO(111)//Co 4 N (red), as obtained in the volcano plot (Figure 4g).OER scaling relationships between adsorption energies of each adsorbates as a function to that of OH species is investigated (Figure 4h).The adsorption energies between OOH (ΔG(OOH*)) and OH (ΔG(OH*)) demonstrates strong linear correlation with a slope close to 1 as expected based on the bond order conservation principle [34] and an y-interception of 3.32 eV confirms that the same linear scaling relation observed in metal oxide systems have applied to this model. [35]Interestingly, the adsorption energies of OOH for CoO(111)//Co 4 N model drastically increases compared to other species, breaking the scaling relations, which decreases the reaction energy of step 3 (O* → OOH*) leading to the highest OER activity.For ORR, the probable active sites with various nitrogen and carbon ratio are screened by tuning the number of doped N and C defects (Figures S30 and S31, Supporting Information).ORR activity of the various Co-N x -C models is predicted and the trend of ORR is listed by the decreasing order of activity whose limiting potential in parenthesis as follow: 4d and Figure S32, Supporting Information).The RDS was OOH* adsorption except for CoN 1 C whose RDS is OH desorption.Particularly, the trend of overpotentials is consistent to the integrated crystal orbital Hamilton population (ICOHP) [36] of OOH adsorption (Figure 4e,f).As the N:C ratio (amount of N relative to carbon) decreases, the adsorption site, Co, tends to lose less electrons due to the smaller electronegativity of C than N.The more negative ICOHP indicates stronger OOH adsorption. [37]As RDS of most models are OOH adsorption, the volcano plot is acquired as a function of ΔG 1 (O 2 → OOH*) (Figure 4g).According to the scaling relations, the slope of OOH* is 0.77 as a function of OH*, indicating OOH adsorption energies became relatively stronger for the models with stronger OH adsorption (Figure 4i).Therefore, the RDS of the model with strong OH adsorption energy, CoN 1 C, shifts to step 4 (OH desorption).Consequently, twodimensional contour maps are plotted based on the overpotentials as a function of each major reaction steps (Figure S33, Supporting Information).With respect to OER, most models are positioned linearly except for two dots (blue: CoO(100)//Co 4 N and orange: 2CoO(100)//Co 4 N) whose OH adsorption energies are weaker relative to other adsorbates.The best OER performance colored in red is placed at the middle of the contour plot implying the optimum adsorption strength to draw the best performance among the models have discovered.Also, the second peak colored in orange appears at the top of the map indicates that the reaction energy of step 3 (O* → OOH*) determines the activity regardless of OH adsorption energy.Regarding ORR, the models are arranged linearly and the peak colored in red are positioned on the left side of the map meaning weak OOH adsorption energy drastically hampers the ORR activity and a model with stronger OOH adsorption and weaker OH adsorption energies could be suggested as the optimal ORR catalyst.The overpotentials of all models are summarized in Figure S34 (Supporting Information), and mechanism for OER and ORR are shown with each model with the best performance in Figure S35 (Supporting Information).

Electrocatalytic Performance of Rechargeable Zn-Air Battery
The efficient bifunctionality of Co 4 N@CoON/PCGN confirmed by the three-electrode half-cell testing is further investigated in a practical rechargeable Zn-air battery (Figure 5).The mixture of the precious metal benchmark catalysts, Pt/C + RuO 2 , is tested as the state-of-the-art reference material.Galvano-dynamic charge/discharge polarization curves with corresponding power density were measured (Figure 5b).Although the open circuit voltage for Co 4 N@CoON/PCGN (1.47 V) is slightly lower than that for Pt/C + RuO 2 (1.51 V), the voltage gap, overvoltage, between charge and discharge profiles of Co 4 N@CoON/PCGN are smaller than Pt/C + RuO 2 when increasing the current density.Particularly, the gap at a current density of 100 mA cm −2 is 336 mV larger in Pt/C + RuO 2 .At higher current, the degree of overvoltage increases for Pt/C + RuO 2 is considerable, whereas Co 4 N@CoON/PCGN proves its high-current capability.Furthermore, the power density plots demonstrate almost twice higher power with Co 4 N@CoON/PCGN (153 mW cm −2 ), compared to Pt/C + RuO 2 (87 mW cm −2 ).Nyquist plots of electrochemical impedance spectroscopy (EIS) are obtained at various discharge voltages to explore the kinetics and degree of charge transfers in Zn-air batteries (Figure 5c).The solution resistance (R s ) of Co 4 N@CoON/PCGN is only 2.25 Ω, while 3.86 Ω is obtained with its counter catalyst, elucidating better kinetics.As the battery voltage decreases (deeper discharge), more electrons transferred at the interface between the catalyst layer and the electrolyte, leading to higher current density and reduced charge transfer resistance (R ct ), a diameter of the semicircles.At this viewpoint, Co 4 N@CoON/PCGN presents drastic reduction of R ct as the discharge voltage decreases from 1.2 to 1.1 V.However, the change of R ct is slight in the case of Pt/C + RuO 2 , implying higher resistance would be loaded in the high current region.This indicates that the cathode containing Co 4 N@CoON/PCGN facilitates better kinetics and stable operation at high current density, which is well consistent with the galvanodynamic polarization results.To further investigate discharge-current relationship, various current densities of 2, 5, 10, 20, and 50 mA cm −2 are applied with time durations of 2 h at each current (Figure 5d).As already observed in the i-v polarizations, Pt/C + RuO 2 shows superior battery discharge capacity at the low currents.However, when 20 mA cm −2 are applied to the battery, Co 4 N@CoON/PCGN outperforms Pt/C + RuO 2 , while the severe voltage drop is observed in Pt/C + RuO 2 at 50 mA cm −2 .This verifies the superb high-current capability of Co 4 N@CoON/PCGN, which is significantly important for further development of commercially available Zn-air battery.In addition, when 2 mA cm −2 is applied again after the high-current discharge, Co 4 N@CoON/PCGN exhibits 99% of current retention, whereas Pt/C + RuO 2 presents five times higher degradation.Full-discharge of the Zn-air single cell galvano-statically discharged at 15 mA cm −2 (Figure 5e) exhibits superior discharge-ability of Co 4 N@CoON/PCGN for longer than 13 h outperforming Pt/C + RuO 2 , demonstrating its superb battery capacity and energy density within Zn-air battery.Lastly, Zn-air battery cycling performance is of great significance to practical applications.Accordingly, the galvanostatic charge/discharge cycling curves at 5 mA cm −2 are obtained with Co 4 N@CoON/PCGN and Pt/C + RuO 2 (Figure 5g,h).Regarding the bifunctional catalytic activities, Pt/C + RuO 2 rapidly loses its initial discharge voltages only within a few cycles, where the loss is ≈110 mV, which is likely due to Pt/C has been severely degraded when it is exposed to high potential during battery charge.Furthermore, Pt/C + RuO 2 fails cycling only after 326 h (81 cycles) majorly because of catalyst layer flooding and leaking of electrolyte.In contrast, Co 4 N@CoON/PCGN demonstrates remarkably stable discharge/charge voltage profile for almost 1400 h without noticeable voltage fading, which is one of the most longevous Zn-air battery cycle-life results reported in the literature.The highest round-trip efficiency is 66% at 5.0 mA cm −2 .The slight decay with a round-trip efficiency of 64.5% after 1100 h elucidates excellent longevous cycling stability.Furthermore, another operation of the Zn-air battery at high current of 20 mA cm −2 (Figure S36, Supporting Information) exhibits competitively low discharge/charge voltage gap of 0.89 V and long battery life of 425 cycles, demonstrating its viability during high current operation.The performance of rechargeable Zn-air batteries with the Co 4 N@CoON/PCGN catalyst is outstanding in comparison to the current advanced Zn-air battery catalysts (Figure 5f and Table S5, Supporting Information).Co 4 N@CoON/PCGN demonstrates superior performance regarding low voltage gap and long-term cycle life, where the results clarify the significantly efficient bifunctional activity and long-term rechargeability of Co 4 N@CoON/PCGN as a cathode electrocatalyst in the rechargeable Zn-air battery system.The crystal planes distribution of Co 4 N@CoON/PCGN cathode has been identified by twodimensional synchrotron-based X-ray diffraction (2D-SXRD) to investigate possible crystal structure change after the Zn-air battery cycling (Figure 5i,j).The results with the corresponding 1D-SXRD patterns (Figure 5k) clarify that the strong crystalline structure majorly consists of cobalt (Co 4 N/CoO x N y ) and nickel (metallic nickel) based materials.Some new tiny peaks are added cycling between 10 and 35 degree (within the two boxes shown in Figure 5j) after 1350 h of Zn-air battery cycling, which corresponds to oxidized nickel and cobalt species.These results in-dicate that there are no noticeable crystal structure transformation and this is the reason for the excellent battery durability.The crystal planes distribution of Co 4 N@CoON/PCGN cathode before and after the rechargeable Zn-air battery cycling is also visualized by two-dimensional synchrotron-based X-ray diffraction (2D-SXRD) patterns to investigate possible structure changes (Figure 5i,j). [38]The patterns exhibit ( 111), ( 200), ( 220), (311), and (220) planes which are attributable to both cubic structured Co 4 N (possibly CoON) and Ni filler.The corresponding 1D-SXRD patterns (Figure 5k) clarify that the highly crystalline structure which fits well with the strong patterns observed.Interestingly, there are some minor patterns are observed in the cycled cathode as displayed in the white boxes (Figure 5j).Some tiny peaks are evolved between 10 and 35 degrees, corresponding to oxidized nickel and cobalt species generated after cycling for 1350 h in Zn-air battery.Although there are slight oxidized peaks evolved, the pristine structure is well retained after the long cycling, which further demonstrates that Co 4 N@CoON/PCGN is robust and durable toward the significantly repeated battery discharges and charges.

Conclusion
In summary, the design strategy for a "raisin-bread" hybrid catalyst is proposed to enable rechargeable Zn-air battery with longlasting lifespan.This strategy combines the in situ nitridation and the self-assembly of porous carbon aerogel, and also creates defective active sites consisting of O-vacancy-rich CoON robust protection covering Co 4 N core and Co-N x -C moieties toward OER and ORR, respectively.The CNT-implanted 3D carbon scaffold enhances the charge/mass transport and expand active area.Particularly, the strong coupling between the Co 4 N@CoON nanoparticle and the neighboring porous carbon aerogel permits rigorous interaction between the selective active sites.Therefore, the robust cobalt oxynitride shell, the defective sites, and carbon architecture synergistically provide efficient bifunctionality as well as prolonged stability in Zn-air battery.Based on the DFT calculations, it is confirmed that the thin CoO layer covering Co 4 N tunes the adsorption energies of reaction intermediates, thereby significantly improving OER activity.Meanwhile, the excellent ORR activity of Co 4 N@CoON/PCGN is attributed to Co-N x -C sites.Consequently, the Co 4 N@CoON/PCGN-introduced Zn-air battery exhibits excellent performance, including a power density of 153 mW cm -2 , rate-capability (99% of voltage retention after discharge at 50 mA cm −2 ), and longevous regenerative cycle life for 1350 h with a low voltage gap of 0.65 V at 5 mA cm −2 of current density.This research provides a guideline for the rational design of highly efficient selectively bifunctional hybrid electrocatalysts and indeed the optimization of the respective ORR and OER activities as well as long-term durability for commercially available rechargeable Zn-air battery systems.

Figure 1 .
Figure 1.a) Schematic illustration of synthetic procedure of Co 4 N@CoON/PCGN, b,c) bright-field TEM images of Co 4 N@CoON/PCGN obtained at low and high magnifications (inset: SEM image), d-g) dark-field STEM-EDS elemental mapping images exhibiting Co 4 N@CoON nanoparticles adjacent to in-plane pores (indicated by red arrows), h) dark-field STEM-EELS (electron energy loss spectroscopy) elemental mapping images of a single Co 4 N@CoON nanoparticle.i) High-resolution TEM image focused on a single Co 4 N@CoON nanoparticle with corresponding fast Fourier transform (FFT) image (inset), and corresponding j) lattice spacings of the selected regions from (i).

Figure 3 .
Figure 3. a) Oxygen reduction reaction (ORR) and b) oxygen evolution reaction (OER) polarization curves, c) linear plots of current density variations Δj (|j anodic − j cathodic |) (taken at 1.05 V versus RHE) as a function of the scan rates of of Co 4 N@CoON/PCGN, Co 4 N@CoON/PGN, Co@CoOx/PCG, and Co-Nx/PCGN, comparison of d) ORR and e) OER activity profiles of Co 4 N@CoON/PCGN to commercial precious metal benchmark catalyst, Pt/C + RuO 2 , with corresponding f) Tafel plots.The polarization curves were obtained at 1600 rpm in O 2 -and N 2 -saturated 0.1 m KOH electrolytes for ORR and OER, respectively, at a scan rate of 10 mV s −1 .g) Potential gaps between E 1/2 for ORR and E j = 10 for OER for all electrocatalysts (∆E = E j = 10 -E 1/2 ), h) stability test conducted by chronopotentiometry (CP) (v-t) at a fixed applied current density of 5 mA cm −2 , i) comparison of overpotentials of Co 4 N@CoON/PCGN and Pt/C + RuO 2 before (initial) and after (final) the stability test.

Figure 4 .
Figure 4. a) Gibbs free energy diagram for OER at 1.23 V, b) pDOS for d-band of Co in CoO layer and c) OH adsorption energy (bar) and d-band center of Co in CoO layer (half circle) for CoO//Co 4 N models.d) Gibbs free energy diagram for ORR at 1.23 V, e) COHP between Co and OOH* and f) integrated crystal orbital Hamilton population (ICOHP) value between Co and OOH* (half square) and overpotential (full square) relative to N:C ratio for CoN x C models.Scaling relations between adsorbates for g) OER and h) ORR models.i) Limiting potential as a function major rate determining step (RDS) for both OER and ORR models.(The insets in a) and d) present the model with best OER and ORR performances which are CoO(111)//Co 4 N and CoN 2 C-o, respectively.)The colors used in the plots are listed: Co 4 N (black), CoO(100)/Co 4 N (blue), CoO(111)/Co 4 N (Red), 2CoO(100)/Co 4 N (green), 2CoO(111)/Co 4 N (yellow) for OER models, and CoN 1 C (blue), CoN 2 C-o (red), CoN 2 C-s (green), CoN 3 C (orange), CoN 4 C (purple), CoN 4 C-d 1 (pink), and CoN 4 C-d 2 (yellow) for ORR models.

Figure 5 .
Figure 5. Electrochemical tests of electrically rechargeable Zn-air battery.a) Schematic illustration of the rechargeable Zn-air battery operation, b) galvanodynamic discharge/charge polarization curves with power density plots, c) Nyquist plots obtained at various discharge voltages, d) rate-capability at 2, 5, 10, 20, and 50 mA cm −2 , e) prolonged discharge curves at 15 mA cm −2 of Co 4 N@CoON/PCGN in comparison to Pt/C + RuO 2 in a form of air cathode, f) Zn-air battery performance comparison of Co 4 N@CoON/PCGN with recently reported non-precious metal-based electrocatalysts for the electrically rechargeable Zn-air battery application, g,h) long-term galvanostatic cycling of Zn-air single cell at 5 mA cm −2 of Co 4 N@CoON/PCGN and Pt/C + RuO 2 , and two-dimensional synchrotron-based X-ray diffraction (2D-SXRD) images from Co 4 N@CoON/PCGN cathode i) before and j) after Zn-air battery cycling, and k) corresponding one-dimensional synchrotron-based X-ray diffraction (1D-SXRD) patterns.