Introduction

Single-atom catalysts (SACs) with isolated metal atoms on supporting materials differ remarkably from traditional nanoparticle catalysts in catalytic activity1. Atomic-level dispersed SACs not only maximize metal utilization but also allow precise control of the chemical and electronic environment of each single metal site, thereby significantly enhancing catalytic performance2,3. Thus, SACs have been highly promising candidates for a range of catalytic applications, including catalytic water treatment and disinfection as well as green synthesis (e.g., NO3− and CO2 reduction reactions)4,5,6,7,8. However, the practical applicability of SACs is largely contingent upon the precise and controlled large-scale synthesis strategies tailored to different catalytic reactions. One major bottleneck is the absence of a universal principle for scalable SAC synthesis to achieve repeatable, well-defined and tailored coordination using a standardized method9,10. Additionally, the mechanisms governing SAC formation and structural evolution remain poorly understood. The natural tendency of metal atoms to aggregate in the synthesis, driven by the Gibbs-Thomson effect11, further challenges preservation of the atomically dispersed metal sites in deliberate synthesis.

Conventional “top-down” or “bottom-up” synthesis strategies have been explored to address these challenges12. Top-down methods typically involve disrupting metal-metal bonds in nanoparticles or bulk metals through physical or chemical processes (e.g., etching or mechanical milling), controlled release of metal atoms to be subsequently incorporated into the defects or vacancies of a supporting matrix13,14. However, control of defects with on-demand population and geometric features is difficult, resulting in suboptimal regulation of both metal atom dispersion and coordination environments. Conversely, bottom-up methods, where catalysts are assembled at the molecular or atomic scale followed by thermal conversion, have recently been extended to construct libraries of single-atom sites2,9,10,15,16. However, these approaches are often constrained by limited structural tunability and poor morphology regulation (Fig. 1a). Thus, templating strategies (e.g., SiO2, polystyrene, and MgO) have been introduced to provide a scaffold for enhanced atomic dispersion and morphology control17,18,19,20. For example, Wang and co-workers used different types of SiO2 hard template to fabricate a family of SACs (Ni, Co, Mn, Zn, Cu, Sc and Fe) with different mesoporous structures21. Although these template-assisted techniques achieved the SAC synthesis, challenges still remain in precisely controlling the local coordination environment of metal atoms due to the chemical inertness of the templates (Fig. 1a). Moreover, these used templates are typically non-recoverable, requiring complex and destructive removal processes that increase the costs of material synthesis and induce potential environmental risks22. Therefore, developing a universal and cost-saving synthesis strategy to manage large-scale SAC production with simultaneous control of morphology and coordination structure is highly desirable.

Fig. 1: Synthesis of Fe1CNCl.
figure 1

a Synthesis scheme of Fe1CNCl. b TEM and c HAADF-STEM images of Fe1CNCl and d the corresponding EDS mappings of C, N, Fe and Cl (The scale bar is 100 nm). e Fe K-edge WT-EXAFS plots of Fe foil and Fe1CNCl. f FT-EXAFS analysis of Fe1CNCl in R space.

Full size image

In this work, we present a versatile template-assisted strategy for massive production of a well-defined SAC library. Unlike traditional methods relying on expensive precursors/templates or unsustainable procedures, we use low-cost and recyclable NaCl as a template to simultaneously control both morphology and local coordination structure of SACs. Specifically, NaCl lattice confines metal atom migration during high-temperature pyrolysis, effectively preventing aggregation and promoting formation of a 3D honeycomb-like morphology in symmetric coordination. In addition, its ion dissociation upon melting facilitates formation of axial metal–Cl bond (Fig. 1a) for asymmetric coordination. Such pyrolysis-regulated local coordination evolution was experimentally and theoretically proved via temperature-dependent formation of a group of Fe SACs. Based on this strategy, we successfully synthesized a library of 25 SACs (denoted as M1CNCl, M is corresponding metal elements) with an average mass yield of 30.5%, and multi-atom five-metal high-entropy SACs (HESACs). These designated SACs exhibited remarkable performances in catalytically activating peroxymonosulfate (PMS) to mediate a selective nonradical pathway for diverse organics oxidation in water. Additionally, the SACs also demonstrated exceptional high activity in electrocatalysis, including NO3− reduction, CO2 reduction, and oxygen reduction reactions. This study not only presents a general synthetic strategy for the large-scale production of SACs with a well-defined coordination structure and 3D porous morphology but also highlights their practical potential in ensuring water safety and energy conversion.

Results

Synthesis and morphology of Fe1CNCl

We employed a simple NaCl-templating method to successfully synthesize a series of SACs anchored on a three-dimensional (3D) nitrogen-doped porous carbon network. Taking Fe element as an illustrative example, the synthesis process and mechanism are illustrated in Fig. 1a. Initially, FeCl2·4H2O (Fe source), dicyandiamide (nitrogen source), glucose (carbon precursor), and sodium chloride (NaCl, hard template) were dissolved in water to form a homogeneous precursor solution. The solution was then freeze-dried to obtain a solid powder. During the freeze-drying process, cubic NaCl crystals precipitated from the solution and created a 3D hard template as water sublimated, confining the metal precursor Fe and other organic molecules (e.g., glucose and dicyandiamide) between the NaCl crystal lattices. The resulting powder mixture was then annealed at 900 °C under an argon atmosphere. During this process, glucose underwent carbonization and NaCl melted at the high temperature, which resulted in a graphitic network within the molten NaCl and evolution into a 3D Fe single-atom-dispersed nanosheet/NaCl composite. Notably, the melting-induced expansion of NaCl crystals promoted the interpenetration of carbon nanosheets, particularly at the grain boundaries of NaCl particles. Additionally, the controlled carbonization between NaCl particles ensured a high mass yield and uniformity in the carbon layers. After acid washing, the NaCl template was removed, yielding a 3D honeycomb-like structure (denoted as Fe1CNCl).

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images both reveal the uniformly distributed open pores (~0.5–1 μm in diameter) and the absence of large metal particles (Fig. 1b and Supplementary Fig. 1), which is confirmed by powder X-ray diffraction (XRD) (Supplementary Fig. 2). High-resolution high-angle annular dark field scanning TEM (HAADF-STEM) images display evenly dispersed bright white spots across the carbon substrate (Fig. 1c), which correspond to atomically dispersed Fe sites. Energy-dispersive X-ray spectroscopy (EDS) mapping further confirms the uniform distribution of C, N, Fe, and Cl elements in Fe1CNCl nanosheets (Fig. 1d). Furthermore, unlike conventional templating methods, where the template is often non-recoverable or requires complex recovery processes, our method allows for the efficient recovery of the NaCl template with a recovery rate of up to 90.2% (Supplementary Fig. 3). This high recovery rate demonstrates the sustainability and cost-effectiveness of the NaCl-assisted synthesis process.

The chemical environments in Fe1CNCl are examined using X-ray photoelectron spectroscopy (XPS) and X-ray absorption spectroscopy (XAS) techniques. As shown in Supplementary Fig. 4, the XAS absorption edge of Fe1CNCl shows a close location to Fe2O3, indicating that the average valence state of Fe species in Fe1CNCl is Fe3+. The N K-edge X-ray absorption near-edge structure (XANES) spectra of the Fe1CNCl catalysts display three well-resolved peaks, corresponding to pyridinic, pyrrolic, and graphitic N species (Supplementary Fig. 5). This finding is further supported by the high-resolution N 1 s XPS spectra, which exhibit a Fe–N peak at 399.0 eV, indicative of FeNx species, suggesting a distinct nitrogen environment compared to the catalyst without Fe (denoted as CN) (Supplementary Fig. 6)23,24. Notably, Cl 2p spectra also confirm the Fe–Cl bond at around 198 eV in Fe1CNCl (Supplementary Fig. 7)23,25, indicating that both N and Cl atoms are coordinated with Fe. This arrangement diverges from the conventional pyrolysis-derived Fe–N4 structures26,27, demonstrating that our synthesis method effectively tailors the Fe coordination microenvironment.

To further elucidate the coordination environment, we conducted extended X-ray absorption fine structure (EXAFS) analysis. The wavelet transform (WT) signal of Fe1CNCl was markedly different from the intensive WT signal of Fe foil (Fig. 1e). Consistently, the characteristic Fe–Fe coordination path at 2.2 Å in the Fourier transform (FT) EXAFS spectrum of Fe foil was absent in Fe1CNCl (Supplementary Fig. 8). All these findings collectively suggested the absence of Fe–Fe bonds in Fe1CNCl, confirming the atomic dispersion of Fe atoms in the Fe1CNCl structure. To elucidate the local coordination structure of Fe1CNCl, we further conducted a quantitative EXAFS fitting analysis (Supplementary Fig. 8). As shown in Fig. 1f and Supplementary Table 1, the best-fit model for Fe1CNCl included two coordination paths: Fe–N and Fe–Cl. Moreover, the fitting results revealed coordination numbers of 4.3 for Fe–N and 1.0 for Fe–Cl, with bond distances of 1.91 and 2.26 Å in Fe1CNCl, respectively. Therefore, all the analyses demonstrated the successful synthesis of atomically dispersed Cl1–Fe–N4 active sites in Fe1CNCl.

Temperature-dependent coordination evolution of Fe SACs on the template phase

To reveal the role of the NaCl template in facilitating Fe1CNCl formation, we performed comparative experiments without adding NaCl (denoted as Fe1CN-woNaCl). In the absence of NaCl, a compact and minimal porous structure is observed (Supplementary Fig. 9), underscoring the importance of the NaCl template in directing the synthesis of 3D honeycomb-like morphology. Accordingly, we further investigated the temperature-dependent morphology and local structure evolution during SAC synthesis using NaCl as the template. As shown in Fig. 2a, with the pyrolysis temperature increasing from 300 to 700 °C, the resulting products (denoted as Fe1CN-300, Fe1CN-500 and Fe1CN-700) exhibit pronounced changes in morphology. The morphology of Fe1CN-300 remains amorphous due to incomplete carbonization, and the NaCl template is insufficient for structural guidance at 300 °C. However, with the increasing temperature to 500 and 700 °C, a distinct sheet-like morphology emerges for Fe1CN-500 and Fe1CN-700 with a higher carbonization degree, revealing that the higher temperatures enable the NaCl template to more effectively direct a stable and porous carbon framework. Notably, no bigger nanoparticle aggregate is observed under any pyrolysis temperature, which is consistent with XRD patterns (Supplementary Fig. 10). Meanwhile, HAADF-STEM images exhibit the single-atom dispersion of Fe in all samples (Fig. 2b), indicating that the NaCl template directs the universal formation of single-atom Fe sites. Likewise, the NaCl-induced morphology-guiding effect was also observed in Co and Ni systems (Supplementary Fig. 11).

Fig. 2: Fe SAC Formation mechanism investigation.
figure 2

Series of a TEM and b HAADF-STEM images of Fe SACs acquired at different temperatures. c EXAFS analysis of Fe SACs acquired at different temperatures. The y-axis values for each sample were individually normalized by the intensities of their respective main peaks for comparing the location of main peaks. d Fe K-edge WT-EXAFS plots of Fe SACs acquired at different temperatures. e Density functional theory calculations for formation energy of Fe SACs as a function of the temperature. The colored areas show the region of most stable structures under the lines with the lowest formation energy. The corresponding experimental temperature range is marked by dashed lines. The black arrow indicates the template activation temperature of 801 °C, beyond which Cl-containing species begin to form.

Full size image

We further characterized the coordination structures formed at various temperatures, focusing on the transition between different Fe–N coordination modes across the temperature range. As shown in Supplementary Fig. 12, the first derivatives XANES of Fe K-edge energy profiles of Fe1CN-300, Fe1CN-500 and Fe1CN-700 show that the oxidation state of atomic Fe center gradually increases with the temperature, which can be attributed to the differing local coordination environment of the Fe single-atom sites. In addition, the FT-EXAFS results demonstrate that atomically dispersed Fe in Fe1CN-300, Fe1CN-500, and Fe1CN-700 exhibits a prominent peak at ~1.60–1.47 Å, which is assigned to the Fe−N scattering path. Notably, no peak corresponding to the Fe−Fe scattering path is observed (Fig. 2c), confirming that all Fe atoms in these three samples remain in isolated dispersion after pyrolysis. Furthermore, the quantitative EXAFS curve fitting reveals that, in contrast to Fe1CNCl (synthesized at 900 oC) exhibiting an Fe–Cl scattering path, only Fe–N scattering paths can be fitted in these three samples. The corresponding Fe–N coordination numbers for Fe1CN-300, Fe1CN-500 and Fe1CN-700 are 6.0, 3.9, and 3.6, respectively (Supplementary Fig. 13 and Table 1). Thus, EXAFS analyses reveal pyrolysis temperature-dependent Fe coordination structures. Specifically, at a low temperature of 300 °C, the Fe center predominantly forms the Fe–N6 coordination. As the temperature increases to the range of 500–700 °C, the coordination number decreases, transitioning to an Fe–N4 structure. Further, when the temperature reaches 900 °C, the coordination structure evolves into asymmetric Cl1–Fe–N4 with the axial incorporation of Cl1–Fe bond. Such a temperature-dependent structural transformation can further be clearly corroborated by the WT analysis (Fig. 2d), where the shifts in the intensity maxima align with the changes in the coordination environment of the single-atom Fe sites. The above results indicate that a heat treatment at 900 °C is essential for the formation of Cl1–Fe–N4 configuration. Since NaCl melts at above 801 °C and dissociates into free-moving ions, we speculate that this temperature-induced Fe–Cl formation may be related to the molten state of NaCl. More specifically, at temperatures above 900 °C, molten NaCl increases chloride ion mobility, thereby promoting their interaction with dispersed iron atoms25.

Density functional theory (DFT) calculations were conducted to examine the dependence of the Fe coordination evolution pathways on pyrolysis temperature. The stability of each structure is represented by its formation energy, incorporating temperature effects through the entropy corrections in the vibrational analysis for gas-phase molecules (Supplementary Fig. 14). As shown in Fig. 2e, at a relatively lower temperature of 573 K (300 °C), Fe tends to form a Fe–N6 coordination. However, when the temperature increases to 773 K (500 °C), the coordination N number decreases to 4 (Fe–N4). This Fe–N4 remains most stable up to 1074 K (801 °C), where the NaCl template begins to melt. Beyond this temperature, molten NaCl facilitates ionic interactions with the carbon substrate and drives the formation of Cl1–Fe–N4 (with Cl in the axial position) at 1173 K (900 °C). These theoretical computations correspond well with the experimental results of coordination structure evolution under different temperatures (from 300 to 900 °C, Supplementary Fig. 13 and Table 1). Consequently, NaCl not only serves as a hard template to create the 3D honeycomb-like structure morphology but also promotes the formation of axial Fe–Cl coordination at high temperatures above its melting point of 801 °C.

The generality of NaCl templating synthesis

Building on the successful synthesis of Fe-based SACs, we extended the NaCl-template strategy to a wide range of metal elements to confirm its versatility and broad applicability. By simply replacing FeCl2·4H2O with other metal chlorides or nitrates, we produced a comprehensive library of SACs (M1CNCl, where M represents different metal elements in Fig. 3a). This library encompasses 25 metals, spanning main group (Mg, Al, K, Ca, Ga, Ge, In, Sn, Sb, Cs, Ba, Pb, Bi), transition (Cr, Mn, Fe, Co, Ni, Cu, Zn, Cd), and rare earth metals (Y, La, Pr, Tb). In every case, each SAC exhibits a honeycomb-like porous structure under the guidance of NaCl.

Fig. 3: Synthesis of different SACs.
figure 3

a Schematic diagram of a SAC library. The orange, blue and purple boxes represent the main group, transition and rare earth metals, respectively. b HAADF-STEM images of 24 SACs. The scale bar is 5 nm. c The FT-EXAFS analysis for M1CNCl fits in R space. d Photographs of batches containing 10 g of Mn, Fe, Co, Ni, Cu and 5 g of Ca, Cr, Zn, Ga, La, Tb SACs.

Full size image

XRD analyses show no reflection corresponding to bulk metal or metal oxides (Supplementary Figs. 15–38), but the characteristic diffraction peaks of carbon support, confirming the atomic-scale metal dispersion. HAADF-STEM images provide a direct evidence of atomic dispersion (Fig. 3b), showing bright, well-dispersed atomic sites on the carbon support without visible clusters or nanoparticles. EDS mapping further confirms the uniform distribution of metal atoms along with C, N and Cl (Supplementary Figs. 15–38).

To gain deeper insights into the coordination environment, we conducted EXAFS analysis to investigate representative samples, containing Mn, Co, Ni, Cu, Zn, Ga, and Ge (Supplementary Figs. 20–26). Their respective FT-EXAFS spectra reveal the absence of characteristic metal-metal bonds typically observed in their corresponding metal foils, corroborating the atomic dispersion as revealed by HAADF-STEM. Additionally, EXAFS fitting of the R-space reveals that different single atoms exhibit the same coordination mode of Cl1–M–N4 (Fig. 3c and Supplementary Table 2), confirming a universal guiding principle of axially engineering for SACs enabled by the molten NaCl template.

To further assess the efficiency of this synthesis method, we analyzed the mass yield across various metal SACs (Supplementary Fig. 39). Mn1CNCl exhibited the lowest yield at 18.3%, while Sb1CNCl had the highest yield at 50.9%. Despite this variation, the overall average mass yield for the 25 metals was 30.5%, demonstrating the effectiveness and reliability of the method across a broad range of metals. The high product yield is attributed to the confined catalytic graphitization of precursors between the NaCl grain boundaries, which significantly reduces the precursor (and decomposed intermediate) volatilization and SAC decomposition losses. Leveraging this high yield, we successfully produced gram-scale SACs (Fig. 3d), highlighting the scalability of the method for practical production and applications.

Moreover, we successfully extended this method to synthesize multi-metal HESACs, containing Mn, Fe, Co, Cu, and Zn on the carbon support, overcoming the aggregations not only among identical metals but also across different metallic species (Supplementary Fig. 40). The HAADF-STEM image and EDS element mapping exhibit uniform dispersion of all five metals across the carbon support, and TEM image confirms the similar 3D honeycomb structure (Fig. 4a–c). The positive oxidation states of the metal sites are confirmed by corresponding XANES (Supplementary Fig. 40). The absence of the metal–metal path between 2.0 and 2.5 Å in FT-EXAFS (Fig. 4d) and the lack of corresponding WT-EXAFS signal at r > 2.0 Å in WT-EXAFS confirm that all metal sites remain atomically isolated (Supplementary Fig. 41). The FT-EXAFS fitting results indicate that all five metals have a moiety of Cl1–M–N4 (Fig. 4d and Supplementary Table 3). Therefore, this work also provides a universal synthetic strategy to fabricate various well-defined HESACs.

Fig. 4: Synthesis of MnFeCoCuZn-HESACs.
figure 4

a HAADF-STEM, b TEM image of HESACs and c corresponding EDS mappings of N, Cl, Mn, Fe, Co, Cu and Zn (The scale bar is 400 nm). d FT-EXAFS spectra of the metals in HESACs. e HAADF-STEM images of Fe SACs synthesized via an alternate component (The scale bar is 2 nm). The inset images correspond to the following replacement components: sucrose, citric acid, melamine, urea, potassium chloride, and commercial table salt, respectively. Gray sphere: C; purple sphere: O, blue sphere: N; orange sphere: K; yellow sphere: Cl.

Full size image

In addition, this strategy can be adapted using alternative carbon precursors (e.g., citric acid and sucrose) and nitrogen sources (e.g., melamine and urea), as well as different salt templates (e.g., potassium chloride and commercial table salt), to achieve single-atom dispersion on carbon (Fig. 4e and Supplementary Figs. 42–47). Moreover, it also exhibits compatibility with secondary coordination sphere modulation, as demonstrated in our recent work on multidimensional engineering of Co sites with second-shell S atoms4, further confirming its versatility in constructing diverse and well-defined M1CNCl catalysts.

Catalytic applications

The catalytic performance of these newly developed SACs was firstly evaluated in the Fenton-like reaction for catalytic degradation of a pharmaceutical pollutant-sulfamethoxazole (SMX) with PMS (Supplementary Fig. 48–51). The metal-free catalyst (CN) exhibited limited efficacy in PMS activation (Supplementary Fig. 52), achieving only 34% of SMX removal within 10 min. Figure 5a shows the rate constants (kobs) of different SACs. Among them, Fe1CNCl exhibited the highest catalytic activity, achieving a complete SMX removal in just 3 min, with a kobs at 35.6 times higher than CN. Notably, Fe1CNCl demonstrated substantially higher activity than its Cl-free counterpart Fe1CN (Supplementary Fig. 53). This exceptional catalytic performance can be attributed to the unique axial coordination of Fe, which modulates its electronic properties and enhances the reactivity toward PMS activation28. Considering that kobs can be affected by reaction conditions, a normalized kinetic model (k-n), accounting for catalyst loading, oxidant and pollutant concentrations, was employed to provide a fair comparison across different heterogeneous Fenton-like systems. Under this metric, Fe1CNCl outperformed most state-of-the-art catalysts in PMS-based oxidation reported to date, including carbon-supported metal nanoparticles and other SACs (Fig. 5b and Supplementary Table 4)29,30,31,32,33,34,35,36,37. Based on controlled quenching experiments, chemical probes, electron paramagnetic resonance (EPR) and open-circuit potential (OCP) tests (Supplementary Figs. 54–58), the exceptional catalytic activity of the Fe1CNCl/PMS system can be attributed to the synergistic mechanism of high-valent iron-oxo species (HVI) and electron transfer processes mediated by the Fe1CNCl − PMS* complex. These factors together facilitate efficient pollutant degradation via nonradical oxidation.

Fig. 5: Catalytic applications of catalysts.
figure 5

akobs of SMX removal by M1CNCl catalysts. b Comparison of k-n for the state-of-the-art heterogeneous Fenton-like catalysts. c Removal efficiency of pharmaceutical and phenolic pollutants (0.02 mM) in the Fe1CNCl/PMS system. Experimental conditions: [SMX] = 0.02 mM, [PMS] = 0.5 mM, [catalyst] = 0.05 g L–1. d Schematics of SACs for other catalytic reduction reactions. Error bars represent the standard deviations by repeating the experiment twice.

Full size image

We further evaluated the practical effectiveness of the Fe1CNCl/PMS system for pollutant removal considering background factors in real wastewater. The system exhibited robust resistance to interference from co-existing ions, including representative cations (Ca2+, Mg2+) and anions (Cl−, NO3−, H2PO4−) (Supplementary Fig. 59). Furthermore, the system maintained high catalytic activity over a broad pH range (3–9), achieving 100% SMX degradation within 10 min (Supplementary Fig. 60). Only trace amounts of Fe leaching were detected after the reaction, confirming a negligible contribution to PMS activation and minimal risk of secondary contamination (Supplementary Fig. 61). Post-reaction characterizations further confirmed that the morphology and structure of Fe1CNCl were well preserved (Supplementary Fig. 62). Moreover, SMX removal efficiency was evaluated in various water matrices, including ultrapure water, tap water, river water (Torrens River), simulated urine and seawater (Glenelg beach), and demonstrated consistently high removal values (>89%) in all the cases (Supplementary Fig. 63). These findings highlight the versatility and stability of the Fe1CNCl catalyst in real-world remediation tasks.

With the growing concerns over emerging contaminants in aquatic environments, we extended the assessment of Fe1CNCl/PMS against a broader range of pollutants. As a result, this nonradical oxidative system can effectively remove 39 diverse pollutants (Fig. 5c, Supplementary Fig. 64 and Table 5), including thirteen phenolic compounds, eleven sulfonamide antibiotics, five quinolone antibiotics, five analgesics, and other five kinds of pharmaceutical contaminants, achieving over 95% removal within 10 min, demonstrating its broad applicability and high efficiency in tackling complex wastewater contaminants. Additionally, Fe1CNCl demonstrated universal activation capability toward different oxidants, including persulfate (PDS), periodate (PI) and peracetic acid (PAA) (Supplementary Fig. 65). These findings collectively highlight the adaptability and promise of Fe1CNCl for complex water treatment scenarios.

Additionally, we explored M1CNCl in electrocatalytic processes for chemical synthesis and energy conversion (Fig. 5d), such as nitrate reduction reaction to ammonia (NO3−RR, Supplementary Figs. 66–68), CO2 reduction reaction to CO (CO2RR, Supplementary Figs. 69–70) and oxygen reduction reaction (ORR, Supplementary Fig. S71). Fe1CNCl achieved higher current densities and significantly improved ammonia yields across all tested potentials compared to CN, delivering an ~3.3-fold improvement over the metal-free counterpart at −1.6 V vs. Ag/AgCl (Supplementary Figs. 66–68). Additionally, CO2RR performance for the M1CNCl catalysts was directly evaluated via electrolyzing at specified currents (Supplementary Figs. 69–70). Ni1CNCl exhibited a high CO2-to-CO conversion efficiency in a broad current density of 10–200 mA cm–1, with a maximum FE of 97.4% at 200 mA cm−1, comparable to or surpassing reported Ni SACs (Supplementary Table 6). In the ORR measurements (Supplementary Fig. 71), Ni1/Fe1/Co1CNCl exhibit an enhanced four-electron pathway to convert oxygen into water. These findings demonstrate that NaCl-assisted M1CNCl catalysts offer versatile approaches for tackling key environmental and sustainable issues, underlining their broad applicability.

Discussion

In summary, we developed a facile and scalable NaCl-templating method for massive production of a comprehensive library of SACs with precise coordination structures. Ex-situ characterizations, combined with DFT calculations, revealed a temperature-driven metal coordination transformation during the annealing process. A low-cost and recyclable NaCl template acted as both a morphology-directing and coordination-controlling agent. This process led to the formation of a 3D honeycomb-like structure, with lower temperatures favoring in-plane M–Nx (x = 4 or 6) coordination, while higher temperatures around 900 °C led to axial M–Cl coordination through ion dissociation from the molten salt. This approach has enabled the successful synthesis of 25 distinct SACs as well as HESACs containing five different metal atoms. Moreover, the outstanding catalytic activity of Fe1CNCl was demonstrated in PMS-based pollutant oxidation in a selective nonradical manner. The broad applicability of these SACs was also validated in various electrocatalytic reduction reactions for chemical synthesis and energy conversion. This NaCl-assisted synthesis method not only offers a robust platform for the precise design of SAC materials and mass production but also underscores its potential for diverse environmental and sustainable catalysis.

Methods

Preparation of M1CNCl catalysts

A mixture consisting of 3 g of NaCl, 300 mg of glucose, 0.1 mmol of metal chlorides or nitrates, and 400 mg of dicyandiamide was dissolved in 30 mL of deionized water. The solution was freeze-dried under vacuum to obtain a solid powder, which was then pyrolyzed for 2 h at 900 °C in an argon atmosphere, with a heating rate of 5 °C/min. After cooling to room temperature, the resulting black powder was treated with 0.5 M H2SO4. The treated powder was dried at 60 °C and was referred to as M1CNCl. The synthesis process can be proportionally scaled up to produce larger quantities of the catalyst. For the synthesis of the nitrogen-doped carbon catalyst (CN), the same procedure was followed, excluding the addition of metal salts. Detailed methods for other comparison samples are provided in the Supplementary Information.

Characterizations

X-ray powder diffraction (XRD) patterns were acquired using a Rigaku MiniFlex 600 X-ray diffractometer. SEM images were captured with a FEI Quanta 450 FEG Environmental Scanning Electron Microscope. TEM images were obtained using an FEI Tecnai G2 Spirit TEM. High angle annular dark-field scanning TEM (HAADF-STEM) images along with EDS analysis were performed on an FEI Titan Themis 80-200 operating at 200 kV. XAS analysis was conducted under ambient conditions. Metal foils and metal oxides were regarded as the standard references. Both XANES and EXAFS spectra were obtained at the Australian Synchrotron in Melbourne. All the spectra data were processed and analyzed using the QANT software program and ATHENA/ARTEMIS module of IFEFFIT software package38,39.

Fenton-like reaction

In a temperature-controlled water bath set at 25 °C, a catalyst concentration of 0.05 g/L and 0.5 mM PMS were added to 50 mL of a 0.02 mM SMX solution. The pH was modified using sulfuric acid (H2SO4) and sodium hydroxide (NaOH). Samples were periodically collected using a syringe and immediately filtered through a 0.22 μm polyether sulfone membrane. Each 1 mL sample was mixed with 20 μL of sodium thiosulfate solution before being transferred into vials for high-performance liquid chromatography (HPLC) analysis. The detection of pollutants was performed using an ultrahigh-performance liquid chromatography (UHPLC) system from Thermo Scientific, equipped with a C-18 column and a UV detection system.