Modulating electronic structure of Co-N₅S₁ sites in Co single atom catalysts via phosphorus incorporation and nanoclusters to promote oxygen electrocatalytic activity

INTRODUCTION

The development of advanced catalysts equipped with green renewable energy conversion and storage technologies, such as zinc-air batteries (ZABs), lithium-sulfur batteries, lithium-air batteries, and fuel cells, presents exciting new opportunities for harnessing sustainable energy moving forward^[1,2]. For these devices, a superior bifunctional oxygen electrocatalyst is a cornerstone of the cathodes. These are indispensable to expedite the practical reaction rates in the oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) process. Noble metal-based composites (such as Pt/C, Pd/C, Ru/C, Ir/C, etc.) with cutting-edge performance are predominantly utilized to enhance the ORR/OER process. However, due to their catalytic selectivity, sustainability, and scarcity, the further adoption of noble metal-based catalysts in the commercial market is severely constrained. This situation considerably encourages the exploration of potential substitutes. Recently, innumerable attention has been centralized on developing transition metal-based catalysts possessing practically equivalent electrocatalytic activity, excellent economic affordability, and structural robustness. Specially, single-atomic catalysts (SACs) featuring the four-coordinated MN₄ moiety (M=Fe, Mn, Ni, Co, Cu, etc.), with an ultrahigh atom utilization rate and tunable electronic properties, are among the most popular electrocatalytic materials in investigation^[3,4]. Ji et al. developed a single-atom iron catalyst on two-dimensional nitrogen-doped carbon support with defects (Fe₁/DNC) using a microenvironment engineering approach^[5]. Fe₁/DNC catalysts demonstrated remarkable ORR activity in an extremely wide pH range (alkaline, acidic, and neutral) because of the synergistic enhancement impact of defect-induced electrical characteristics and FeN₄(OH) moieties. In addition, Co-N₄-C-^[6-8] and Mn-N₄-C^[9,10]-based electrocatalysts also achieved fine electrocatalytic activity in the former research. Nevertheless, in the wake of further experimental studies and theoretical calculations, the symmetric microenvironment of MN₄ in SACs may lead to suboptimal adsorption/desorption processes in electrocatalytic reactions, limiting the reaction kinetics of the catalyst^[11-13].

Hitherto, introducing a second active site^[14-16], nanoclusters (NCs)^[17,18], and heteroatoms^[19-21] in MN₄ catalysts were corroborated to the major ways to break the symmetry of the planar structure, optimizing both electrocatalytic activity and stability of metal-nitrogen-carbon (M-N-C) composites. On the one hand, plentiful studies have manifested that the import of additional active metal atoms largely enhanced the site density of the catalysts, realizing a multi-functional synergism and excellent catalysis^[22,23]. For example, Li et al. developed a Fe-Co bimetallic monoatomic catalyst with asymmetric configurations [Fe-Co(DSA)@3DNC]^[24]. For the seesaw interaction of the bimetallic atom, the d-band center of Fe-Co(DSA)@3DNC kept a downward shift than the symmetric Fe-Co model, which was conducive to promoting the dissociation of oxygen intermediates and boosting the ORR/OER bifunctional activity of the catalyst. Additionally, Rao et al. proposed a high-entropy single-atom catalyst with several different single-atom centers dispersed on an N-doped carbon skeleton, and the quinary FeCoNiCuMn catalyst exhibited more excellent activity and durability than the benchmark Pt/C catalyst in electrochemical terms^[25]. Nevertheless, the structure-activity relationship and underlying reaction mechanism in multi-site systems remain in the infancy stages of understanding, presenting considerable challenges for conquering technical bottlenecks. On the other hand, tremendous investigation efforts were devoted to analyzing the NCs or heteroatoms (S, P, B, etc.) doping on the M-N-C catalysts and clarifying the relationship between the microenvironment and catalytic performance^[26,27]. Among these studies, phosphorus presented a distinct doping effect on M-N-C systems, promoting catalytic activity more effectively than other heteroatoms due to the stronger electron-donation ability^[28,29]. Notably, phosphorus doping favors enhancing electro-oxidation resistance while mitigating carbon corrosion throughout electrocatalytic processes. Zhou et al. have successfully incorporated phosphorus atoms into the Fe-N-C to construct the second coordination, and the P/Fe-N-C catalyst demonstrated extraordinarily bifunctional catalytic activity with an exceptionally narrow potential gap (△E = 0.63 V), outperforming the Fe-N-C counterpart^[30]. Furthermore, extensive experimental investigation and density functional theory (DFT) calculations revealed that metal NCs can act as effective promoters for M-N-C system catalysts, significantly enhancing their catalytic efficiency toward the ORR^[31,32]. The interaction between the M atom in the M-N₄ structure and the surrounding NCs induced a downward shift in the d-band center, which weakened the binding affinity for *OH intermediates and optimized the catalytic efficiency for ORR by facilitating faster intermediate dissociation and electron transfer^[33]. Hence, the controlled anchoring of NCs and the implementation of asymmetric hetero-atomic coordination onto a porous carbon framework represent an encouraging strategy forging significant improvements in the electrocatalytic efficiency and stability of M-N-C catalysts for oxygen reactions.

Toward that end, through a straightforward pyrolysis-secondary coordination process using amberlite@Co²⁺ as the precursor, we designed a catalyst with cobalt NCs-decorated asymmetric CoN₅S₁ single-atom sites anchored in N, S, and P-codoped carbon nanosheets, labeled Co_SA+NC/NPSC. The nitrogen-rich melamine (MA) was introduced in situ into the amberlite@Co²⁺ precursor, where nitrogen invaders partially substituted for the original Co-S moiety, prompting the generation of asymmetric Co-N₅S₁ sites. Simultaneously, triphenylphosphine was utilized as a phosphorus source to modulate the electron configuration of the Co SA active centers, thereby elevating the electrocatalytic efficiency and stability of the Co_SA+NC/NPSC catalyst. Notably, by virtue of the distinct atomic structural configuration, the optimized Co_SA+NC/NPSC delivered markedly improved activities toward OER (E_j=10 = 1.58 V) and a narrow bifunctional potential gap (ΔE = 0.753 V), which was superior to the counterparts featuring symmetrical Co-S coordination or phosphorus-free doping. According to the theoretical simulation of the ORR and OER processes, the Co_SA+NC/NPSC catalyst demonstrated the lowest energy barrier, which is unified with the accelerated reaction kinetics in the experiment. When integrated into an air electrode, the Co_SA+NC/NPSC catalyst-based flexible ZAB demonstrated exceptional charge-discharge cycle stability (over 105 h) and impressive initial round-trip efficiency of 72.42% at 0 °C and 5 mA cm^-2.

EXPERIMENTAL

RESULTS AND DISCUSSION

The synthesis of a nitrogen, sulfur, and phosphorus co-doped nanocarbon catalyst, featuring abundant isolated cobalt single-atom active sites and cobalt NCs, facilitated by an excess nitrogen-source-assisted pyrolysis strategy (MA), was depicted in Figure 1A. Initially, Co²⁺ ions were firmly encapsulated within the amberlite 732 resin through ion exchange, yielding an amberlite@Co²⁺ hybrid structure. Subsequently, the amberlite@Co²⁺ precursor was thoroughly mixed with PMMA, PPh₃, and ethanol solution to incorporate soft templates and phosphorus sources. The resulting amberlite@Co²⁺/PMMA/PPh₃ composite was then uniformly ground with MA to achieve a homogeneous mixture. This mixture was subjected to thermal treatment in a corundum crucible at 900 °C for 2 h under an argon atmosphere. During the pyrolysis, as the temperature approached the decomposition points of PMMA (270 °C) and MA (300 °C), the polymers decomposed, releasing gaseous products and forming layered C₃N₄, thereby generating a porous substrate. At higher temperatures (> 500 °C), the amberlite@Co²⁺/PMMA/PPh₃ composite underwent carbonization, generating two-dimensional graphene-like nanosheets enriched with nitrogen, phosphorus, sulfur, and cobalt species. As shown in Figure 1B, the Co_SA+NP/NPSC catalyst featured aggregated nanocrystals dispersed across the carbon matrix. Notably, MA served as an efficient nitrogen source, facilitating the formation of Co-N coordination within the carbon nanosheets, disrupting the symmetric Co-S coordination environment, and leading to the formation of asymmetric CoN₅S₁ sites. Following the annealing process, the final Co_SA+NC/NPSC electrocatalyst was obtained by treating the product with dilute hydrochloric acid (0.3 M) to remove any residual metallic nanoparticles. To elucidate the synergistic effects of the coordination environment and heteroatom doping on the active sites, control samples without phosphorus doping (Co_SA+NC/NSC) and with symmetric S coordination (Co_SA+NC/SC) were also synthesized for comparative analysis. As illustrated in Supplementary Figure 1, the Co_SA+NC/NSC exhibited a well-defined two-dimensional nanosheet structure, suggesting a higher surface area contrasted to the stacked morphology of Co_SA+NC/SC. The morphological characteristics of the catalysts were initially investigated through SEM [Figure 1C] and TEM. As illustrated in Figure 1D, the TEM image of the synthesized Co_SA+NC/NPSC catalyst revealed a two-dimensional nanosheet structure with thin layers, devoid of any discernible metal nanoparticles. Further structural details of Co_SA+NC/NPSC were obtained via high resolution transmission electron microscopy (HRTEM) [Figure 1E]. Supplementary Figure 2A displayed the enlarged details in the circled area of Figure 1E, which showed distinct lattice fringes with interplanar spacings of 0.33 nm, indicative of the presence of cobalt NCs anchored within the carbon nanosheet, corresponding to the (111) plane of these NCs [Supplementary Figure 2B]. Furthermore, Co_SA+NC/NPSC exhibited no evident diffraction bright spots attributable to cobalt NCs, aligning with the ring-like selected area electron diffraction (SAED) patterns shown in Supplementary Figure 2C, indicating the existence of only a handful of cobalt NCs. EDS mapping in Figure 1F confirmed the homogeneous distribution of Co, C, N, S, and P across the entire Co_SA+NC/NPSC catalyst. ICP-AES analysis data was listed in Supplementary Table 1, which further presented that the actual content of Co atoms in Co_SA+NC/NPSC was approximately 1.64 wt%. XRD curves were characterized to record the transformation of the crystallization state from Co_SA+NP/NPSC to Co_SA+NC/NPSC. As represented in Supplementary Figure 3A, the XRD pattern of Co_SA+NP/NPSC demonstrated diffraction peaks at 44.8°, 51.6°, and 76.0°, which correspond to the different planes of Co nanoparticles (JCPDS No.15-0806). Following acid treatment, the diffraction peaks are attributed to the Co crystal planes disappear entirely, while the two broad peaks observed at 25.4° and 43.8° correspond to the different planes of graphitic carbon (JCPDS No.41-1487), respectively, consistent with the SEM and TEM results [Supplementary Figure 3B]. Similarly, the XRD patterns of Co_SA+NP/NSC [Supplementary Figure 3C] and Co_SA+NP/SC [Supplementary Figure 3D] also confirm the existence of graphitic carbon substracts, with no other impurities detected. To further elucidate the pore structure of the catalysts, nitrogen adsorption-desorption isotherms were recorded. The isotherms for Co_SA+NC/NPSC, Co_SA+NC/NSC, and Co_SA+NC/SC, shown in Supplementary Figure 4A-C, demonstrated typical type IV isotherms, indicating the presence of a substantial number of mesopores and micropores. A comparison of the Brunauer-Emmett Teller (BET) surface areas revealed that importing the nitrogen source significantly enhanced the surface area of Co_SA+NC/NSC relative to Co_SA+NC/SC, facilitating the dispersion of more active sites [Supplementary Figure 4D-F].

Figure 1. (A) The synthesis flowchart of Co_SA+NC/NPSC; SEM pictures of (B) Co_SA+NP/NPSC and (C) Co_SA+NC/NPSC; (D) TEM image; (E) HRTEM image; and (F) EDS mapping of Co_SA+NC/NPSC. SEM: Scanning electron microscope; TEM: transmission electron microscope; HRTEM: high resolution transmission electron microscopy; EDS: energy-dispersive spectrometer.

XANES and EXAFS spectroscopy were favorable descriptors to elucidate the detailed layout and chemical state of the Co atom active centers comprehensively. As shown in Figure 2A, the near-edge absorption energy of Co_SA+NC/NPSC was situated near that of CoO, indicating the presence of a positively charged Co species, primarily in the Co²⁺ oxidation state. The coexistence of metallic Co-Co, Co-N, and Co-S scattering pathways suggested the coordinating atoms around the Co atom are N, S and Co atoms, respectively [Figure 2B]. To further explore the coordination environment, wavelet transform (WT)-EXAFS analysis was employed, offering enhanced resolution by distinguishing backscattering atoms, even in cases of substantial overlap in R-space, while simultaneously providing both radial distance and k-space resolution. Figure 2C presents the WT contour plots for the three k²-weighted χ(k) signals, derived using Morlet wavelets for optimal resolution. The intensity maxima, corresponding to distinct coordinates (k, R), were primarily related to the path length R and the atomic number Z. Notably, Co foil and CoS₂ exhibited intensity maxima at 7.3 Å^-1 (Co-Co contribution) and 5.8 Å^-1 (Co-S contribution), respectively. For Co_SA+NC/NPSC, the WT analysis of the Co K-edge EXAFS predominantly showed an intensity maximum arising from Co-N scattering at lower k-space (3.8 Å^-1). Additionally, a substantial overlap was noted among the Co-N, Co-S, and Co-Co regions, indicating that these coordination environments converge within a unified region in the wavelet maps. The bond lengths and coordination numbers of the catalyst were calculated by least-squares EXAFS fitting analysis. The K-space and R-space fitting curves indicated a strong correlation with the experimental spectra [Figure 2D-G, Supplementary Figure 5, and Supplementary Table 2]. Based on these results, the coordination around the central Co atom revealed an average coordination number of six, comprising five nitrogen atoms and one sulfur atom. As summarized in Supplementary Table 2, the EXAFS fitting parameters for the Co-Co and Co-S pathways in Co_SA+NC/NPSC indicated bond lengths of 2.46 Å and 2.32 Å, respectively, which were marginally shorter than those observed for Co foil (2.49 Å) and CoS₂ (2.31 Å). These findings collectively affirmed that the Co_SA+NC/NPSC catalyst predominantly comprises CoN₅S₁ moieties, accompanied by a trace amount of cobalt NCs.

Figure 2. (A) XANES spectra for the Co foil, CoO and Co_SA+NC/NPSC; (B) FT-EXAFS spectra of Co K-edge in Co_SA+NC/NPSC; (C) k²-weighted WT-EXAFS signals of Co foil, CoS₂ and Co_SA+NC/NPSC; EXAFS analysis of Co_SA+NC/NPSC at (D) k and (E) R spaces, respectively; Fitting results of Co K-edge FT-EXAFS analysis in (F) k and (G) R spaces for Co_SA+NC/NPSC. XANES: X-ray absorption near edge structure; EXAFS: extended X-ray absorption fine structure; FT-EXAFS: fourier transform- extended X-ray absorption fine structure; WT-EXAFS: wavelet transform- extended X-ray absorption fine structure.

The ORR activity of the Co_SA+NC/NPSC catalyst was rigorously evaluated using a RDE in 0.1 M KOH solution, with the Co_SA+NC/NSC and Co_SA+NC/SC catalysts used for comparative analysis. As illustrated in Figure 3A, the linear sweep voltammetry (LSV) curves confirmed that the Co_SA+NC/NPSC catalyst exhibited superior ORR performance relative to its Co_SA+NC/NSC and Co_SA+NC/SC counterparts, as evidenced by a significantly reduced overpotential. Notably, the half-wave potential (E_1/2) of Co_SA+NC/NPSC was measured at 0.827 V, markedly outperforming that of Co_SA+NC/NSC (0.803 V) and Co_SA+NC/SC (0.695 V). The ORR kinetics were further examined by evaluating the Tafel slope, as depicted in Figure 3B. The Co_SA+NC/NPSC catalyst exhibited a Tafel slope of 45.7 mV dec^-1, lower than those of Co_SA+NC/NSC (50.5 mV dec^-1) and Co_SA+NC/SC (89.1 mV dec^-1), signifying faster ORR kinetics for Co_SA+NC/NPSC. Beyond its competitive catalytic activity, the catalyst favored a highly selective four-electron (4e^-) ORR pathway, a crucial factor for minimizing the generation of deleterious H₂O₂ byproducts, which could otherwise undermine catalyst stability. The catalytic selectivity was evaluated via Koutecky-Levich (K-L) plots [Supplementary Figure 6], with the K-L curves for Co_SA+NC/NPSC across a range of rotation speeds (400-2025 rpm) confirming a predominant 4e^- ORR pathway (n = 4.03), which was superior to its counterparts. The enhanced selectivity likely stems from stronger Co-O interactions in Co_SA+NC/NPSC, facilitating efficient O-O bond cleavage along the 4e^- ORR route. This exceptional selectivity also contributed to the catalyst’s promising long-term stability, as demonstrated by the chronoamperometric response of Co_SA+NC/NPSC at 0.7 V versus reversible hydrogen electrode, where it retained 90.81% of its initial current after 24,000 s of continuous operation, affirming robust electrochemical stability in alkaline media [Figure 3C]. The OER efficacy of the Co_SA+NC/NPSC catalyst was assessed in 1.0 M KOH to evaluate its bifunctional catalytic performance in advanced ZABs. As shown in Figure 3D and E, the LSV curve of Co_SA+NC/NPSC exhibited the lowest overpotential of 350 mV at a current density of 10 mA cm^-2, presenting a notable negative shift compared to Co_SA+NC/NSC (41 mV), Co_SA+NC/SC (69 mV) and the IrO₂/C benchmark (11 mV). The Co_SA+NC/NPSC catalyst further demonstrated superior OER kinetics, as reflected by the lowest Tafel slope of 33.3 mV dec^-1, in contrast to 82.6, 72.3, and 98.0 mV dec^-1 for Co_SA+NC/NSC, Co_SA+NC/SC and IrO₂/C, respectively [Figure 3F]. This substantial enhancement in OER activity was likely due to the improved interaction with reaction intermediates, as theoretical studies suggested that the OER performance of Co_SA+NC/NSC and Co_SA+NC/SC was limited by weaker intermediate binding. Furthermore, the durability of Co_SA+NC/NPSC in OER electrocatalysis was confirmed by its stable chronopotentiometric response, with no discernible degradation after 60 h of continuous polarization at 10 mA cm^-2 [Figure 3G]. In comparison, the potentials for Co_SA+NC/NSC and Co_SA+NC/SC increased to 1.673 V and 1.645 V, respectively, after 24 h under the same conditions. The bifunctional catalytic activity was quantified using the potential difference (ΔE) between the E_1/2 and the OER potential (E_j=10), where a smaller ΔE indicated superior bifunctional performance. The Co_SA+NC/NPSC catalyst exhibited the smallest ΔE of 0.753 V, surpassing the other catalysts [Figure 3H]. These results conclusively demonstrated that the synergistic effects of phosphorus doping and asymmetric atomic coordination in Co_SA+NC/NPSC significantly enhanced both ORR and OER activities. Furthermore, the electrochemically active surface area (ECSA), a critical indicator of intrinsic electrocatalytic activity, was assessed by measuring the electric double-layer capacitance (C_dl) in the non-Faradaic region, where C_dl was proportional to ECSA. As shown in Supplementary Figure 7, Co_SA+NC/NPSC displayed the highest C_dl value of 20.08 mF cm^-2, suggesting that the catalyst exposed more active sites, thus reinforcing its remarkable electrocatalytic ORR activity.

Figure 3. (A) ORR curves and (B) derived Tafel slopes Co_SA+NC/SC, Co_SA+NC/NSC and Co_SA+NC/NPSC; (C) Chronoamperometric response of Co_SA+NC/SC, Co_SA+NC/NSC and Co_SA+NC/NPSC catalyst at 0.7 V; (D) OER polarization curves; (E) overpotential and (F) the corresponding Tafel slopes of Co_SA+NC/SC, Co_SA+NC/NSC, and Co_SA+NC/NPSC catalysts; (G) The chronopotentiometry curves of Co_SA+NC/SC, Co_SA+NC/NSC, and Co_SA+NC/NPSC were taken at 10 mA cm^-2; (H) The potential gap (ΔE = E_j=10 - E_1/2) of different catalysts. ORR: Oxygen reduction reaction; OER: oxygen evolution reaction.

Given the enhanced bidirectional catalytic performance of the Co_SA+NC/NPSC catalyst, a flexible ZAB integrated with Co_SA+NC/NPSC (cathode), Zn foil (anode), and a PAM-based organic hydrogel electrolyte with exceptional low-temperature adaptability was fabricated (denoted as ZAB-Co_SA+NC/NPSC). Prior to assembly, the anti-freezing electrolyte was immersed in a solution of 6.0 M KOH and 0.2 M Zn(OAc)₂ for 48 h. Comparative ZABs were also constructed by substituting Co_SA+NC/NPSC with Co_SA+NC/NSC and Co_SA+NC/SC, referred to as ZAB-Co_SA+NC/NSC and ZAB-Co_SA+NC/SC, respectively. The electrochemical performance of these ZABs was first evaluated at ambient temperature. As shown in Figure 4A, ZAB-Co_SA+NC/NPSC exhibited a relatively high open-circuit voltage (OCV) of 1.40 V, approaching the theoretical OCV of 1.65 V for ZABs, thereby surpassing the other counterparts. Moreover, as depicted in Figure 4B, the maximum power density of ZAB-Co_SA+NC/NPSC reached 62.18 mW cm^-2, significantly outperforming ZAB-Co_SA+NC/NSC (33.87 mW cm^-2) and Co_SA+NC/SC (38.85 mW cm^-2). Figure 4C revealed that ZAB-Co_SA+NC/NPSC also delivered an excellent discharge plateau of 1.23 V and a specific capacity of 793.91 mAh g^-1 at 5 mA cm^-2, corresponding to 96.82% of the theoretical discharge capacity of ZABs. More practically, the combination of superior bifunctional catalytic activity and an anti-freeze hydrogel electrolyte endowed the flexible ZAB with remarkable adaptability at sub-zero temperatures. Following an initial activation process and a 2 h exposure to a low-temperature chamber, ZAB-Co_SA+NC/NPSC demonstrated extended charge-discharge cycling (over 105 h) with a narrow charge-discharge potential gap (ξ = 0.78 V) at 0 °C and 5 mA cm^-2, outperforming the comparative ZABs [Figure 4D]. Detailed charge-discharge curves at various time intervals were magnified in Figure 4E-G. Notably, the Co_SA+NC/NPSC-based ZAB, owing to its superior OER and ORR catalytic activities, exhibited an initial round-trip efficiency (ε) of 72.42% and a charge-discharge potential gap of 0.475 V in region I, far surpassing ZAB-Co_SA+NC/NSC (ε = 67.98%, ξ = 0.564 V) and ZAB-Co_SA+NC/SC (ε = 59.80%, ξ = 0.788 V) [Figure 4E]. Impressively, ZAB-Co_SA+NC/NPSC maintained a high efficiency of 66.8% after 30 cycles, with only a slight increase in the voltage gap of 117 mV. In contrast, the round-trip efficiency of ZAB-Co_SA+NC/NSC declined significantly to 49.21% after 30 cycles at 5 mA cm^-2 and 0 °C, highlighting its inferior low-temperature tolerance and structural stability [Figure 4F]. Furthermore, in region III, the Co_SA+NC/NPSC-based ZAB retained its advantage with an efficiency of 64.1% and a potential gap of 0.656 V, while the potential gap of ZAB-Co_SA+NC/SC increased to 1.474 V, more than twice that of ZAB- Co_SA+NC/NPSC [Figure 4G]. The discharge curves with a distinct step-like shape, shown in Figure 4H, indicated that ZAB-Co_SA+NC/NPSC provided the highest discharge plateaus among the comparative ZABs across various current densities, especially at higher current densities, underscoring its potential for practical applications. To further investigate the underlying kinetic processes, electrochemical impedance spectroscopy (EIS) was employed to evaluate the interfacial resistance (R_s) and charge transfer resistance (R_ct) of ZAB-Co_SA+NC/NPSC. The frequency range was controlled between 100 kHz and 1 Hz, with the ZABs in an open-circuit state throughout the experiment. As presented in Supplementary Figure 8A, the EIS fitting curves indicated that ZAB-Co_SA+NC/NPSC exhibited the lowest R_s and R_ct values, correlating with excellent rechargeability. Conversely, the increased internal resistance in ZAB-Co_SA+NC/NSC and ZAB-Co_SA+NC/SC resulted in greater ohmic polarization and reduced ion conductivity, thereby compromising their overall battery performance. To gain deeper insights into the structure-activity relationship, the microstructure of the Co_SA+NC/NPSC electrode was examined via TEM after several charge-discharge cycles. Supplementary Figure 8B demonstrated that Co_SA+NC/NPSC retained its original nanosheet morphology even after repeated cycling, consistent with the experimental data shown in Figure 4D and H. In conclusion, the exceptional rechargeability and low-temperature reversibility of Co_SA+NC/NPSC can be attributed to two key factors. First, its unique microstructure facilitated the full exposure of abundant active sites, significantly reducing ion diffusion distances and minimizing interfacial impedance, thereby mitigating the negative effects of sluggish ion diffusion kinetics at low temperatures. Second, the synergistic effects of asymmetric Co single-atom active sites, cobalt NCs, and heteroatom doping in Co_SA+NC/NPSC reduced the activation energy barrier for oxygen electrocatalysis, allowing the catalyst to maintain superior performance under low-temperature conditions.

Figure 4. (A) Open-circuit voltage; (B) discharge power density plots and (C) the discharge specific capacity curves of flexible ZABs with Co_SA+NC/SC, Co_SA+NC/NSC and Co_SA+NC/NPSC cathode catalysts; (D) Cycling stability of the flexible ZABs at 0 °C and 5 mA cm^-2); The enlarged (E) region I, (F) region II and (G) region III of (D); (H) discharge curves of Co_SA+NC/SC, Co_SA+NC/NSC and Co_SA+NC/NPSC-based apparatus at different current densities. ZABs: Zinc-air batteries.

To elucidate the synergistic effects of heteroatom doping and cobalt NCs on the properties of asymmetric CoN₅S₁ sites, and their promoted bifunctional oxygen electrocatalytic activity, DFT calculations were performed. Based on the findings from X-ray absorption fine structure analysis, four potential structural models of Co_SA+NC/NPSC are illustrated in Supplementary Figure 9. The optimized atomic configuration was identified according to the criteria of minimal formation energy and a stable oxygen intermediate structure [Figure 5A]. For the purpose of comparative analysis, theoretical models were developed for both symmetric single-atomic cobalt moieties and phosphorus-free doped CoN₅S₁ sites, as illustrated in Supplementary Figure 10. By integrating the previously computed electron transfer numbers, it was determined that the ORR pathway primarily follows a 4-electron mechanism, progressing through the intermediates *O₂, *OOH, *O, and *OH on the optimized Co active center [Supplementary Figures 11 and 12]. Importantly, Supplementary Figure 13 revealed a pronounced stretching of the O-O bond (1.36 Å) in Co_SA+NC/NPSC when compared to the Co_SA+NC/NSC catalyst (1.06 Å), signifying an apparently enhanced activation of the O₂ molecule at the phosphorus-doped CoN₅S₁ site^[39]. The free energies of the adsorbed intermediates in the ORR process and the inserted adsorption atomic structures on the Co_SA+NC/NPSC model were depicted in Figure 5B. Without applying additional voltage (0 V vs. RHE), all optimized theoretical models presented a downhill energy pathway in the ORR process, indicating their exothermic and spontaneous characteristics [Supplementary Figure 14]. At U = 1.23 V, the Co_SA+NC/NPSC catalyst demonstrated the lowest free energy difference (△G), effectively corroborating the findings illustrated in Supplementary Figure 13. Moreover, the third elementary step of the ORR across the Co_SA+NC/SC, Co_SA+NC/NSC, and Co_SA+NC/NPSC systems revealed the utmost △G, suggesting that the bottleneck reaction step (BRS) involved the transformation of *O to *OH (*O + H* + e^- → *OH) [Figure 5C]. Integrated with Co_SA+NC/SC (2.19 eV) and Co_SA+NC/NSC (1.29 eV), Co_SA+NC/NPSC demonstrated a notably narrow △G of the BRS, revealing an easier *O hydrogenation occurred on Co_SA+NC/NPSC catalyst. These findings verified that the boosted ORR activity of Co_SA+NC/NPSC arose from the asymmetric coordination and incorporation of heteroatom doping, which promoted the protonation of *O and significantly improved the surface adsorption capacity for *OH at the CoN₅S₁ site. Importantly, from the perspective view of the integral ORR process, Co_SA+NC/NPSC showed a significantly decreased overpotential of 1.25 V than Co_SA+NC/SC (2.19 V) and Co_SA+NC/NSC (1.88 V), further underscoring its promoted efficiency in catalyzing ORR. Analogously, ΔG calculations were performed for single-atomic Co sites in the context of OER. As depicted in Figure 5D, the final step (*OOH → * + O₂ +H⁺ + e^-) constituted the BRS for OER at U = 1.23 V. Importantly, the ΔG values, arranged in ascending order, were Co_SA+NC/NPSC,Co_SA+NC/SC, and Co_SA+NC/NSC, aligning with the experimental data presented in Figure 3D. As outlined in the d-band center theory developed by Hammer and Nørskov, the relative position of the d-band center with respect to the Fermi level acted as a key “indicator” for assessing the binding affinity of single-atomic active centers toward reaction intermediates^[40,41]. In Figure 5E, the ε_d value for Co_SA+NC/NPSC presented a downshift of 0.12 eV and 0.03 eV relative to Co_SA+NC/SC and Co_SA+NC/NSC counterparts, respectively. This suggested a more balanced binding strength between active sites and the intermediates, contributing to the promoted bifunctional ORR/OER efficiency. The electronic transmission served as a crucial descriptor of electrocatalytic performance. Consequently, the electronic characteristics of the Co_SA+NC/NPSC catalyst were deeply explored by charge density difference analyses. As depicted in Figure 5F, the Bader charge for Co_SA+NC/NPSC was determined to be 0.76 e, exceeding that of its respective counterparts. This observation suggested that the import of phosphorus atoms, along with the asymmetric coordination environment surrounding the single-atomic Co sites, facilitated enhanced electron redistribution.

Figure 5. (A) Schematic presentation of side view and top view atomic model of Co_SA+NC/NPSC; (B) Pathways for the Co_SA+NC/NPSC at 0 V with four oxygen intermediate models inserted; (C) ORR and (D) OER free energy ladder diagrams at 1.23 V vs. RHE; (E) The theoretical d band center and (F) differential charge density of Co_SA+NC/SC, Co_SA+NC/NSC and Co_SA+NC/NPSC (yellow and cyan represented the electron accumulation and electron depletion). ORR: Oxygen reduction reaction; OER: oxygen evolution reaction; RHE: reversible hydrogen electrode.

CONCLUSIONS

In summary, we have developed a bifunctional oxygen electrocatalyst featuring single Co atoms integrated within an unsymmetrical Co-N₅S₁ moiety, along with NC complexes embedded in amberlite-derived N,P,S-codoped carbon frameworks, synthesized through a pyrolysis-secondary coordination approach. The deliberate design of active sites, incorporation of NCs, and heteroatom doping endowed the Co_SA+NC/NPSC catalyst with exceptional OER activity, outperforming benchmark IrO₂, and ORR/OER bifunctional performance in alkaline environments. The experimentally and theoretically analytical methodologies employed in this work elucidated the superior bifunctional efficiency, attributed to the optimized atomic configuration and the enhanced density-of-states distribution at the Co-N₅S₁ active centers. Additionally, the flexible ZAB assembled with Co_SA+NC/NPSC air cathodes demonstrated ultralong charge-discharge cycle stability, maintaining performance for over 105 h at 0 °C and 5 mA cm^-2. The proposed approach for the structural regulation of SA holds significant potential to advance research into other cutting-edge electrocatalytic processes, such as the hydrogen evolution reaction, nitrogen fixation, and CO₂ reduction reaction.