Precise arrangement of metal atoms at the interface by a thermal printing strategy

Significance

This work depicts a thermal printing strategy to fabricate single-atom sites at the interface of different supports, which endows them with high accessibility and excellent catalytic activity. This strategy depends on the controlled migration of movable nanoparticles between two contact substrates and the simultaneous emission of atoms from the nanoparticle surface at high temperatures.

Abstract

The kinetics and pathway of most catalyzed reactions depend on the existence of interface, which makes the precise construction of highly active single-atom sites at the reaction interface a desirable goal. Herein, we propose a thermal printing strategy that not only arranges metal atoms at the silica and carbon layer interface but also stabilizes them by strong coordination. Just like the typesetting of Chinese characters on paper, this method relies on the controlled migration of movable nanoparticles between two contact substrates and the simultaneous emission of atoms from the nanoparticle surface at high temperatures. Observed by in situ transmission electron microscopy, a single Fe₃O₄ nanoparticle migrates from the core of a SiO₂ sphere to the surface like a droplet at high temperatures, moves along the interface of SiO₂ and the coated carbon layer, and releases metal atoms until it disappears completely. These detached atoms are then in situ trapped by nitrogen and sulfur defects in the carbon layer to generate Fe single-atom sites, exhibiting excellent activity for oxygen reduction reaction. Also, sites' densities can be regulated by controlling the size of Fe₃O₄ nanoparticle between the two surfaces. More importantly, this strategy is applicable to synthesize Mn, Co, Pt, Pd, Au single-atom sites, which provide a general route to arrange single-atom sites at the interface of different supports for various applications.

Catalytic reactions usually take place on the surface of catalysts because the adsorption and desorption of reactants rely on the surface for their occurrence (1–4). The electronic and atomic structures of surface can be manipulated to regulate the adsorption and desorption behavior of reactants, thereby controlling the pathways and kinetics of reactions (5–9). For complex reactions involving multiple electrons and molecules, the combination of various functional surfaces to form an active interface can promote the reaction (10–16). For example, the electron transfer can be promoted on the interface composed of single atoms and clusters/nanoparticles, leading to a significant improvement in the oxygen reduction activity (17, 18). Therefore, integrating heterogeneous surfaces to form catalytic interfaces can effectively offer the optimized electronic distribution, synergistic effect, or tandem catalysis at the interface (19–22).

Recently, single-atom catalysts (SACs) with optimal atom efficiency and excellent catalytic activity have attracted wide attention, showing great potential to replace nanoparticles in catalyzing many important reactions in the energy field (23–35). However, single-atom sites are often insufficient for complex catalytic reactions involving the transfer of multiple electrons and reactant species because single atoms cannot provide a broad d-band for the adsorption of multiple reactant molecules (34). By placing single atoms as active sites on the interface, it is possible to utilize the different surface characteristics of the interface to simultaneously activate the adsorbed molecules and increase the frequency of effective collision, thereby enhancing the transfer of electrons and reactant species. Conventional synthesis methods (such as impregnation and thermal atomization) usually synthesize single-atom sites on the surface of a single support, where the adsorption of precursors on the surface is not regionally selective (36–43). Compared to a single surface, achieving the selective and precise arrangement of single atoms at the interface of different supports remains a significant challenge because it requires the enrichment of precursors at the interface. If we can successfully manipulate the selective enrichment, migration, and cleavage of precursors, the precise arrangement of single atoms at the interface might be achieved.

Results

The Proposed Thermal Printing Strategy.

We have developed a thermal printing strategy to arrange single atoms at the interface of two common supports, endowing them with high accessibility and excellent catalytic activity. In conventional ink printing, ink is initially ejected onto the roller and gradually sprayed onto the paper as the roller rotates (Fig. 1A). Similar to the typesetting of Chinese characters on paper, if we could precisely control the movement of each nanoparticle at the interface of different supports and gradually release atoms, the emitted metal atoms could be atomically arranged at the interface (Fig. 1B). The main challenge in adopting this strategy is to load a single nanoparticle onto the interface of different supports, thereby ensuring the same concentration of the arranged metal atoms. However, conventional methods encounter difficulties in effectively achieving the loading of a single nanoparticle at the surface/interface of the supports (SI Appendix, Fig. S1).

Fig. 1.

The proposed thermal printing strategy. (A) The ink printing process; (B) The proposed thermal printing strategy; (C and D) In situ TEM images of Fe₃O₄@SiO₂/CdS@PDA; (E and F) Aberration-corrected HAADF-STEM images corresponding to the selected area in (D).

To tackle this challenge, we develop a strategy involving the encapsulation of single nanoparticle (e.g., Fe₃O₄) within a porous SiO₂ sphere (SI Appendix, Figs. S2 and S3). This encapsulated nanoparticle is then loaded onto a CdS nanorod and coated with polydopamine, denoted as Fe₃O₄@SiO₂/CdS@PDA (Fig. 1C and SI Appendix, Figs. S4–S8). CdS nanorods play a dual role in this system. First, they contribute to the high dispersion of Fe₃O₄@SiO₂ spheres. Second, as the temperature increases, CdS nanorods volatilize and release S elements into the carbon layer, resulting in the formation of S, N co-doped carbon. Also, with the activation of high temperatures, Fe₃O₄ nanoparticle escapes from the interior to the surface through the porous channels of SiO₂ (SI Appendix, Fig. S9) (44). This escape mechanism enabled the presence of Fe₃O₄ nanoparticle at the interface of SiO₂ and carbon layer. Upon increasing the temperature to 1,273 K, Fe₃O₄ nanoparticle migrates along the interface of SiO₂ and coated carbon layer, and gradually releases atoms that are subsequently trapped in situ by the carbon layer. As revealed in the in situ transmission electron microscopy (in situ TEM) images (Fig. 1 C and D and SI Appendix, Fig. S10), CdS nanorods and embedded Fe₃O₄ nanoparticle disappear completely after heating treatment. Aberration-corrected high-angle annular dark field scanning transmission electron microscopy (aberration-corrected HAADF-STEM) images reveal that Fe species are atomically and uniformly arranged on the skeleton of coated carbon (Fig. 1 E and F), indicating that the migration of Fe atoms from the core of SiO₂ to the coated carbon layer is successfully manipulated. Compared to conventional impregnation and thermal atomization strategies (45, 46), this strategy can simultaneously ensure the uniformity and thermal stability of atomic sites due to the controlled enrichment and cleavage of precursors (SI Appendix, Fig. S11).

Structural and Morphological Characterizations.

After removal of SiO₂, a hollow carbon skeleton resembling grape tree structure can be obtained (Fig. 2 A and B and SI Appendix, Figs. S12 and S13). HAADF-STEM and selected-area electron diffraction images reveal that the well-bonded hollow carbon nanotubes and spheres consist of amorphous carbon with a thickness of 3.2 nm (Fig. 2B and SI Appendix, Fig. S13). Aberration-corrected HAADF-STEM images further show that Fe single atoms are well preserved in the carbon skeleton (Fig. 2 C and D). Besides, energy-dispersive X-ray spectroscopy (EDS) mappings and line scanning profiles suggest the uniform distribution of Fe, S, N, C species (Fig. 2E and SI Appendix, Fig. S14), demonstrating that Fe single atoms arranged on S, N co-doped carbon skeleton are successfully obtained (denoted as Fe SAs/S-NC). Inductively coupled plasma atomic emission spectroscopy (ICP-AES) was performed to reveal the Cd and S contents for Fe₃O₄@SiO₂/CdS@PDA and Fe SAs/S-NC. The results indicate that no signals for Cd are detected in Fe SAs/S-NC, corresponding to <0.001 wt % content below the detection limit of ICP equipment, while the content of S species is 1.16 wt% (SI Appendix, Table S1). X-ray photoelectron spectroscopy (XPS) spectra also reveal the presence of C, S, N, and Fe in Fe SAs/S-NC. The C 1s and S 2p XPS spectra proclaim the formation of C-N and C-S bonds, demonstrating the successful doping of N and S into the C lattice (Fig. 2F and SI Appendix, Fig. S15). The N 1s XPS spectrum exhibits three peaks of pyridinic N (398.15 eV), pyrrole N (399.2 eV), and graphitic N (401.2 eV), among which pyridinic N and pyrrole N can coordinate with metal single atoms (Fig. 2F) (47).

Fig. 2.

Structural and morphological characterizations. (A) The schematic sketch of etching SiO₂; (B) HAADF-STEM image of Fe SAs/S-NC and (C and D) aberration-corrected HAADF-STEM images corresponding to the selected area in (B); (E) EDS mappings of Fe SAs/S-NC; (F) S 2p and N 1s XPS spectra of Fe SAs/S-NC; (G) N₂ adsorption isotherm and the corresponding pore-size distribution; (H) X-ray absorption fine structure and (I) the corresponding Fourier transform EXAFS spectra.

N₂ adsorption–desorption isotherm (Fig. 2G) further indicates that the carbon skeleton of Fe SAs/S-NC possesses a large specific surface area (1,368.9 m² g⁻¹) and abundant mesopores, which can facilitate the accessibility of the supported single atoms. The combined use of CdS nanorods and SiO₂ spheres as double templates is responsible for the high surface area and the formation of a porous structure in Fe SAs/S-NC. X-ray diffraction (XRD) patterns (SI Appendix, Fig. S16) demonstrate that no observable signals of Fe metal and oxide crystals can be detected in Fe SAs/S-NC, ruling out the existence of large Fe aggregation. The chemical state of Fe in Fe SAs/S-NC lies between Fe⁰ and Fe³⁺, demonstrating its oxidation state (SI Appendix, Fig. S17). Furthermore, X-ray adsorption fine structure (XAFS) was employed to investigate the chemical state and coordination environment of Fe single atoms. As shown in Fig. 2H, the adsorption threshold of Fe atoms from Fe SAs/S-NC situates between the Fe foil and Fe₃O₄, showing partially positive valence states. The R space spectrum of the extended XAFS (EXAFS) (Fig. 2I) for Fe SAs/S-NC only exhibits a main peak around 1.45 Å, which is attributed to the Fe-N coordination, demonstrated by the EXAFS fitting curve (SI Appendix, Fig. S18). The first coordination number of the central Fe is about 4 and the average bond length of Fe-N is 2.17 Å (detailed fitting data can be seen in SI Appendix, Table S2).

The Evolution Mechanism from Metal Nanoparticle to Single Atoms.

To elucidate the evolution mechanism from metal nanoparticle to single atoms, we conducted in situ TEM experiments to trace the process under high-temperature pyrolysis (Fig. 3A and Movie S1). We identify that Fe₃O₄ nanoparticle experiences a thermal printing process similar to the principle of ink printing, where it could move like an ink droplet at the interface of SiO₂ and coated carbon layer and simultaneously emit metal atoms to arrange them on coated carbon layer (Fig. 3B). In detail, CdS nanorods gradually become unstable and evaporate as the temperature rises, as demonstrated by the changes in the in situ TEM images (Fig. 3A and SI Appendix, Fig. S19) and the corresponding thermogravimetric analysis curve (SI Appendix, Fig. S20). However, Fe₃O₄ nanoparticle with a diameter of ~8 nm remains well at the core of SiO₂ sphere even at a temperature of 1,173 K. Upon further increasing the temperature, Fe₃O₄ nanoparticle starts to flow like a droplet from the core of SiO₂ toward the surface (1,273 K, 0 s) and migrated along the interface because of the confinement of coated S, N co-doped carbon layer (1,273 K, 0 to 90 s). During migration, the nanoparticle is compressed and deformed, gradually emitting atoms from the surface, leading to size reduction. When the temperature is kept at 1,273 K, the particle size is reduced to 5 nm at 60 s and disappears completely at 90 s. The emitted atoms act like ink sputtering and are eventually arranged on carbon layer, similar to typesetting Chinese characters on paper. In situ EDS mappings further demonstrate the uniform distribution of Fe species on S, N co-doped carbon skeleton after pyrolysis at 1,273 K, indicating the successful arrangement of Fe single atoms on support by thermal printing strategy (SI Appendix, Fig. S21). The consistent Fe content observed through inductively coupled plasma mass spectroscopy suggests that the arrangement of Fe single atoms remains unaffected by the varying thickness of SiO₂ (SI Appendix, Fig. S22 and Table S3).

Fig. 3.

The evolution mechanism from metal nanoparticle to single atoms. (A) In situ TEM images of Fe₃O₄@SiO₂/CdS@PDA at different temperature and duration; (B) Schematic diagram for the atomic diffusion process; (C) Structure configurations and the binding energy of Fe₃O₄ and Fe single atoms anchored by amorphous SiO₂ and armchair S-NC. Color scheme: pink for Fe, blue for N, yellow for S, gray for C, green for Si, and white for O.

To further explain the thermal printing mechanism, we performed density functional theory (DFT) calculations (Fig. 3C). We adopt Fe₃O₄ nanoparticle as the initial state, which can emit Fe single atom from its surface because of the activation of high temperature. As the Fe₃O₄ nanoparticle migrates along the interface of the SiO₂ and S, N co-doped carbon, the emitted Fe single atom is considered to be trapped by both. To understand this process, we calculate the binding energy of the Fe single atom on SiO₂ (with Fe-O₄ coordination) and S, N co-doped carbon (with Fe-N₄ coordination) and compared them with that on Fe₃O₄. The results show that loading a Fe atom into SiO₂ (Fe SAs/SiO₂) requires to overcome an energy barrier of 13.2 eV, while trapping a Fe atom with S, N co-doped carbon requires much lower energy (SI Appendix, Fig. S23). In particular, Fe atom trapped by armchair S, N co-doped carbon (Fe SAs/armchair S-NC) is exothermic, suggesting that Fe single atoms emitted from Fe₃O₄ nanoparticle are more inclined to be anchored by this way.

The Generality of the Thermal Printing Method.

To demonstrate the generality of the thermal printing method, we synthesized Fe₃O₄ nanoparticles with varying sizes and encapsulated them within SiO₂ spheres (SI Appendix, Figs. S24–S26). As the size of Fe₃O₄ nanoparticles increases, the local Fe content also increases, and more Fe species are inclined to be in situ trapped by carbon support (Fig. 4A and SI Appendix, Table S4). Atomically precise printing of Fe₃O₄ nanoparticles with diameter of 6 nm, 8 nm, and 12 nm onto carbon supports result in Fe loadings of 0.1 wt%, 0.2 wt%, and 0.5 wt%, respectively. However, with Fe₃O₄ nanoparticles of 18 nm in diameter, the limited defect sites on the carbon support result in the formation of Fe clusters, with a higher Fe loading of 1 wt%. The density of Fe sites at different Fe loadings was calculated using statistical methods based on HAADF-STEM images, revealing a correlation between Fe site density and Fe loading (Fig. 4 B–E and SI Appendix, Fig. S27).

Fig. 4.

The generality of the thermal printing method. (A) The schematic sketch of the thermal printing process of nanoparticles with different sizes; (B–E) Aberration-corrected HAADF-STEM images of Fe SAs/S-NC printed by 6 nm, 8 nm, 12 nm, and 16 nm Fe₃O₄ nanoparticles, respectively; (F) EDS mappings and (G) aberration-corrected HAADF-STEM image of Mn SAs/S-NC; (H) EDS mappings and (I) aberration-corrected HAADF-STEM image of FeCo SAs/S-NC.

Furthermore, we conducted similar experiments by replacing Fe₃O₄ nanoparticles with MnO₂ and FeCoO_x nanoparticles (SI Appendix, Figs. S28 and S29), which are pyrolyzed at the same condition. The results demonstrate that Mn single atoms can be arranged on the carbon skeleton, as evidenced by the EDS mappings and the aberration-corrected HAADF-STEM image (Fig. 4 F and G). Additionally, Fe and Co single atoms can be simultaneously arranged on the carbon skeleton (Fig. 4 H and I). Using this strategy, we also arranged Pt/Pd/Au single atoms on N-doped carbon or TiO₂ and CeO₂ layer (SI Appendix, Figs. S30–S32). Further, other N-containing precursors [such as aniline, pyrrole and 3-aminophenol (3-AP)] derived carbon layer can also anchor single atoms at the interface, as shown in SI Appendix, Fig. S33.

Catalytic Performance Verification.

To demonstrate the practicability of the thermal printing strategy, we investigated their reactivity in catalyzing the cathodic reaction of anion exchange membrane fuel cells and Zn air battery. Oxygen reduction reaction (ORR) performance was evaluated using a typical three-electrode device in 0.1 M KOH solution. Linear sweep voltammetric (LSV) curves demonstrate that the ORR activity of Fe SAs/S-NC exhibits a volcanic-type relation between Fe loading, and Fe SAs/S-NC with 0.2 _wt% loading possesses the highest ORR activity (SI Appendix, Fig. S34). For comparison, we also fabricated S, N co-doped carbon skeleton (denoted as S-NC) and Fe single atoms anchored on N-doped carbon (denoted as Fe SAs/NC) by controlling the synthesis condition (SI Appendix, Figs. S35 and S36). Fe SAs/S-NC exhibits a high half-wave potential (E_1/2 = 0.91 V), and kinetic current density at 0.9 V (J_k = 8.06 mA/cm²), suppressing those of S-NC (E_1/2 = 0.75 V, J_k = 0.02 mA/cm²), Fe SAs/NC (E_1/2 = 0.84 V, J_k = 1.45 mA/cm²), Pt/C (E_1/2 = 0.86 V, J_k = 1.71 mA/cm²), and most of the Fe-based SACs (Fig. 5 A and B and SI Appendix, Table S5). Additionally, the Tafel slope of Fe SAs/S-NC is much lower than the counterparts, suggesting faster kinetics (Fig. 5C). The rotating ring disk electrode (RRDE) testing and Koutecky–Levich plots demonstrate the high electron transfer number of Fe SAs/S-NC (n > 3.93) (Fig. 5D and SI Appendix, Figs. S37–S39). Furthermore, Fe SAs/S-NC displays almost unchanged E_1/2 and electrochemical double-layer capacitance value (24.8 mF cm⁻² vs. 24.6 mF cm⁻²) after 10,000 continuous potential cycles (SI Appendix, Figs. S40–S43). TEM image, aberration-corrected HAADF-STEM image and EDS mappings after potential cycles also demonstrate the stability of the catalyst (SI Appendix, Fig. S44). Besides, Fe SAs/S-NC exhibits no observable reduction in current density after 200 s upon injecting methanol, suggesting the strong tolerance to methanol of Fe single atoms (SI Appendix, Fig. S45). These results all demonstrate the excellent activity and stability of Fe SAs/S-NC in catalyzing ORR. Furthermore, Fe-SAs/S-NC based cathode was used in Zn-air battery, which delivers a higher open-circuit voltage (1.55 V) and larger discharge specific capacity (794 mA h/g_Zn) than commercial Pt/C (SI Appendix, Figs. S46 and S47). It can also reach a maximum power density of 272 mW cm⁻² at 0.59 V, higher than those of commercial Pt/C (Fig. 5E) and most of the reported catalysts (SI Appendix, Table S5). Besides, Fe SAs/S-NC exhibits a low charge-discharge voltage gap after 1,400 continuous galvanostatic discharge/charge cycles, demonstrating its excellent stability (Fig. 5F).

Fig. 5.

Electrochemical ORR and Zn-air battery performances. (A) ORR LSV curves in O₂-saturated 0.1 M KOH solution; (B) E_1/2 and J_k at 0.9 V vs. RHE; (C) Tafel slop curves; (D) H₂O₂ yield and electron transfer number; (E) Polarization curves of Zn-air battery; (F) Galvanostatic discharge–charge cycling profiles at 5 mA cm⁻² of Fe SAs/S-NC+RuO₂ and Pt/C+RuO₂ electrodes; (G) Free-energy diagram for ORR process on Fe SAs/armchair S-NC and Fe SAs/NC moieties at the equilibrium potentials of U = 0 V in the alkaline. (Inset) corresponding schematic models of samples. Color scheme: pink for Fe, blue for N, yellow for S, gray for C, white for O, and red for H.

DFT calculations were further conducted to explore the structure–activity relationship of Fe single-atom sites obtained by thermal printing strategy. We constructed three possible models of FeN₄ structure with S coordination according to XAFS and XPS characterization results and found that the formation energy of Fe SAs/armchair S-NC is the lowest (SI Appendix, Fig. S23). We then calculated the Gibbs free energy changes on Fe SAs/armchair S-NC and Fe SAs/NC (Fig. 5G). From the calculated free energy diagram of the reaction path in Fig. 5G, three oxygen-involved intermediates (OOH*, O*, and OH*) are attached to Fe single atoms of Fe SAs/armchair S-NC and Fe SAs/NC, which are generated in succession according to the associative mechanism with incremental addition of molecular H₂O and electrons to form four OH⁻ in the end. The limiting potential steps of ORR for the Fe SAs/armchair S-NC and Fe SAs/NC are the reduction of OH* to H₂O (*OH + H⁺ + e⁻ → * + H₂O). However, the corresponding theoretical onset potential (U_L) for the Fe SAs/armchair S-NC is 0.74 V, which is much higher than that of the Fe SAs/NC (0.59 V). Therefore, the Fe SAs/armchair S-NC shows much lower theoretical overpotential (0.49 V) than that of the Fe SAs/NC (0.65 V) (SI Appendix, Table S6), indicating a superior ORR activity of Fe single atoms with the adjacent S coordination.

Discussion

In summary, our study presents a thermal printing strategy to arrange metal atoms at the interface of carbon and silica, resulting in enhanced ORR activity. By controlling the size of nanoparticle at the interface, we can effectively regulate the metal loading. This versatile strategy can be extended to arrange various single atoms (such as Fe, Mn, and Co,) at the interface, offering promising applications across diverse fields.

Materials and Methods

Chemicals and Materials.

All chemicals were used as received without any additional purification. NaOH, KOH, NH₄OH, iron chloride, manganese(II) chloride, cobalt(II) chloride, n-hexane, cyclohexane, tetraethyl orthosilicate (TEOS), hexadecyltrimethylammonium bromide (CTAB), N₂H₄·H₂O, sodium diethyldithiocarbamate trihydrate (Tris), ethanol, chloroauric acid (HAuCl₄·4H₂O), ethylenediamine, sodium oleate, sodium borohydride (NaBH₄), and Ce(NO₃)₃·6H₂O were purchased from Sinopharm Chemical Regent Co., Ltd. 1-octadecene (ODE, 90%), oleic acid (OA, 90%), potassium tetrachloroplatinate (K₂PtCl₄), Pd(NO₃)₂·2H₂O, myristyltrimethylammonium bromide (TTAB), hydroxypropyl cellulose (HPC), tetrabutyl titanate (TBOT), and 2-methylaminoethanol were purchased from Alfa Aesar. Dopamine HCl, Igepal CO-520, and Nafion (5 wt%) were purchased from Sigma-Aldrich. Deionized water was used throughout this study.

Preparation of Fe₃O₄ Nanoparticles.

Thermal decomposition of iron oleate was adopted to fabricate monodisperse, 8-nm Fe₃O₄ NCs (48). Fe₃O₄ nanoparticles were then dispersed in cyclohexane to form a stable colloidal solution with a concentration of 2 mg/mL.

Preparation of Fe₃O₄@SiO₂ Spheres.

Silica was formed to coat Fe₃O₄ nanoparticles via a reverse microemulsion method (49). Five hundred milligram of Igepal CO-520 was added into 10 mL cyclohexane and sonicated for 10 min. Then, 4 ml Fe₃O₄ solution (2 mg/mL in cyclohexane) was added to the above solution under ultrasound for 30 min at room temperature. Subsequently, 2 mL NH₄OH (25 to 28%) was added to the above solution, and 2 mL TEOS was added drop by drop. After stirring at room temperature for 10 h, the resulting Fe₃O₄@SiO₂ core-shell spheres were collected after adding ethanol and centrifuging. After washing with ethanol for three times, Fe₃O₄@SiO₂ core-shell nanoparticles were dispersed in ethanol for further use.

Preparation of CdS Nanorods.

Uniform CdS nanorods were prepared by a solvothermal method (50).

Preparation of Fe₃O₄@SiO₂/CdS@PDA.

200 mg CdS was dispersed in 200 mL Fe₃O₄@SiO₂ ethanol solution (1 mg/mL) and stirred at room temperature for 10 h. 300 mg Fe₃O₄@SiO₂/CdS was obtained after centrifugation at 11,000 rpm and drying at 343 K in a vacuum. Then, the presynthesized Fe₃O₄@SiO₂/CdS (300.0 mg) were dispersed by ultrasound in 100.0 mL freshly prepared Tris buffer solution (10 mM, pH 8.5). Then, a certain amount of dopamine HCl in 10.0 mL H₂O was added to the Tris buffer solution, and the solution was stirred at room temperature for 12 h. The desired Fe₃O₄@SiO₂/CdS@PDA was obtained by washing the resulting products with deionized water and ethanol three times, collecting by centrifugation, and drying at 343 K in a vacuum for 6 h. The adding quantity of dopamine-HCl was 600 mg, unless otherwise stated.

Preparation of Fe₃O₄@SiO₂/CdS@PANI.

200 mg CdS was dispersed in 200 mL Fe₃O₄@SiO₂ ethanol solution (1 mg/mL) and stirred at room temperature for 10 h. 300 mg Fe₃O₄@SiO₂/CdS was obtained after centrifugation at 11,000 rpm and drying at 343 K in a vacuum. Then, the presynthesized Fe₃O₄@SiO₂/CdS (300.0 mg) and 30 mg sodium dodecyl benzene sulfonate were dispersed in 200 mL deionized water by ultrasound for 30 min at room temperature. Then, 400.0 μL aniline, 6.0 mL 1 M HCl, and 600 mg ammonium persulphate were added to the above solution in sequence and stirred for 2 h at room temperature. After that, the suspension was collected by centrifugation and washed by deionized water and ethanol three times. After drying at 343 K in a vacuum for 6 h, the desired Fe₃O₄@SiO₂/CdS@PANI was obtained.

Preparation of Fe₃O₄@SiO₂/CdS@PPY.

Using 3.0 mL phytic acid instead of HCl and substituting aniline with pyrrole, Fe₃O₄@SiO₂/CdS@PPy was obtained in a similar way as Fe₃O₄@SiO₂/CdS@PANI.

Preparation of Fe₃O₄@SiO₂/CdS@3-AP.

150 mg 3-AP, formaldehyde solution (100 μL, 37 wt %), and Fe₃O₄@SiO₂/CdS (100 mg) were dispersed in 50 mL ethanol/water solution (v/v 1:1) with ethanediamine (100 μL, 5 mg/mL) as catalyst. After that, the solution was stirred at room temperature for 24 h. Finally, the obtained Fe₃O₄@SiO₂/CdS@3-AP was collected and purified with distilled water by centrifugation and dried at 343 K in a vacuum for 6 h.

Preparation of Fe SAs/S-NC.

In a typical procedure, the Fe₃O₄@SiO₂/CdS@PDA power was grinded and annealed at 1,273 K for 3 h under Ar atmosphere with a heating rate of 5 K min⁻¹. After cooling down to room temperature, SiO₂ spheres were etched away in a 1 M NaOH aqueous solution at 353 K for 24 h to obtain the hollow Fe SAs/S-NC (PDA). Fe SAs/S-NC (PANI), Fe SAs/S-NC (PPY) and Fe SAs/S-NC (3-AP) can be obtained by substituting Fe₃O₄@SiO₂/CdS@PDA with Fe₃O₄@SiO₂/CdS@PANI, Fe₃O₄@SiO₂/CdS@PPY and Fe₃O₄@SiO₂/CdS@3-AP.

Preparation of Fe SAs/NC.

The presynthesized Fe₃O₄@SiO₂ (100.0 mg) were dispersed by ultrasound in 100.0 mL freshly prepared Tris buffer solution (10 mM, pH 8.5). Then, 10.0 mL dopamine HCl (300.0 mg) solution was added to the Tris buffer solution and stirred for 12 h at room temperature. The desired Fe₃O₄@SiO₂@PDA was obtained by washing with deionized water and ethanol three times, collecting by centrifugation, and drying at 343 K in a vacuum for 6 h. The Fe₃O₄@SiO₂@PDA power was grinded and annealed at 1,273 K for 3 h under Ar atmosphere with a heating rate of 5 K min⁻¹. After cooling down to room temperature, SiO₂ spheres were etched away in a 1 M NaOH aqueous solution at 353 K for 24 h to obtain the hollow Fe SAs/NC.

Preparation of S-NC.

First, SiO₂ spheres were prepared by adding 0.68 mL TEOS to a mixture solvent composed of 14 mL ethanol, 1.6 mL distilled water, and 0.25 mL ammonium hydroxide (25 to 28%) in a 50-mL round bottom flask. After stirring at room temperature for 12 h, SiO₂ spheres were obtained by rotary evaporation of the solvent. 100 mg SiO₂ spheres and 200 mg CdS nanorods were dispersed in 200 mL ethanol and stirred at room temperature for 10 h. 300 mg SiO₂/CdS was obtained after centrifugation at 11,000 rpm and drying at 343 K in a vacuum. Then, the presynthesized SiO₂/CdS (300.0 mg) were dispersed by ultrasound in 100.0 mL freshly prepared Tris buffer solution (10 mM, pH 8.5). Then, 10.0 mL dopamine HCl (600.0 mg) solution was added to the Tris buffer solution and stirred for 12 h at room temperature. The desired SiO₂/CdS@PDA was obtained by washing with deionized water and ethanol three times, collecting by centrifugation, and drying at 343 K in a vacuum for 6 h. The SiO₂/CdS@PDA power was grinded and transferred into a ceramic boat and annealed at 1,273 K for 3 h under Ar atmosphere with a heating rate of 5 K min⁻¹. After cooling down to room temperature, SiO₂ spheres were etched away in a 1 M NaOH aqueous solution at 353 K for 24 h to obtain the S-NC.

Characterization.

XRD measurements were performed on a Rigaku Miniflex-600, which operated at 40 kV voltage and 15 mA current using a Cu Kα radiation (λ = 0.15406 nm) at a step width of 2° min⁻¹. Hitachi-7650 worked at 100 kV was used to obtain TEM images. The HAADF-STEM and EDS mappings were performed on a Talos F200X high-resolution transmission electron microscope. The in situ TEM study and aberration-corrected HAADF-STEM were performed on a JEM-ARM200F microscope. XPS results were recorded from Catalysis and Surface Science End station at the BL11U beamline of National Synchrotron Radiation Laboratory (NSRL) in Hefei, China. Thermogravimetric analyses were carried out by heating from room temperature to 1,273 K at the rate of 5 K min⁻¹ on a TA SDT Q600 thermal analyzer. ICP-AES was performed on Optima 7300 DV to detect the elemental contents in the solid samples. The Brunauer–Emmett–Teller results were tested on micromeritics ASAP 2020 HD88 PLUS, during which the samples were degassed at 573 K for 3 h, and the pore size distribution was calculated from the BJH method.

The X-ray absorption fine structure data were performed at 1W1B station in Beijing Synchrotron Radiation Facility (BSRF). We employed the ATHENA module within the IFEFFIT software packages to process the EXAFS data. These data were acquired by subtracting the postedge background from the overall absorption and then normalizing it based on the edge-jump step. Following this, χ(k) data in k-space were Fourier transformed into real (R) space using a Hanning window (dk = 1.0 Å-1) to distinguish the EXAFS contributions from various coordination shells.

Electrochemical Measurement for ORR.

The samples (Fe SAs/S-NC, S-NC, Fe SAs/NC, and 20 wt% Pt/C) catalyst powders were dispersed into 1 mL solution (960 μL ethanol and 40 μL Nafion), ultrasound at room temperature for 20 min. The loadings of Fe and Pt were 90 and 102 μg cm⁻², respectively. The electrochemical measurements were conducted on an electrochemical workstation (CHI 760E) using a three-electrode system. The working electrode is a glassy carbon RDE of 5 mm in diameter, coated with the catalyst ink. Graphite rod and Ag/AgCl were used as the counter electrode and reference electrode, respectively. The potential values were adjusted to match the reversible hydrogen electrode (RHE). The electrolyte was 0.1 M KOH, which should be saturated with N₂/O₂ prior to the measurement. We can obtain the background capacitance in N₂-saturated electrolyte. The net ORR curves were obtained from recording in O₂-saturated electrolyte and subtracting the background capacitance.

A sweep rate of 10 mV s⁻¹ was advocated to perform RDE measurement, with different rotating speeds (from 900 to 2,500 rpm). We use Koutecky–Levich equation to calculate the electron transfer number (n) and kinetic current density:

$$\frac{1}{\mathrm{J}}=\frac{1}{{\mathrm{J}}_{\mathrm{l}}}+\frac{1}{{\mathrm{J}}_{\mathrm{k}}}=\frac{1}{{\text{Bω}}^{\frac{1}{2}}}+\frac{1}{{\mathrm{J}}_{\mathrm{k}}},$$

$$\mathrm{B}=0.62\text{nF}{\mathrm{C}}_0{\mathrm{D}}_0^{\frac{2}{3}}{\mathrm{V}}^{-\frac{1}{6}},$$

where J is the current density measured from the ORR, J_k is the kinetic current density, J_l is the limiting current density, ω represents the angular velocity of the disk, C₀ is the bulk concentration of O₂ (1.2 × 10⁻⁶ mol cm⁻³), D₀ is the diffusion coefficient of O₂ in 0.1 M KOH (1.9 × 10⁻⁵ cm² s⁻¹), and V is the kinematic viscosity of the electrolyte (0.01 cm² s⁻¹). The accelerated durability tests of Fe SAs/S-NC were performed by applying potential cycling between 1.0 and 0.6 V vs. RHE for 10,000 cycles (O₂-saturated 0.1 M KOH, 50 mV s⁻¹, room temperature).

For the RRDE tests, the disk electrode was scanned negatively (10 mV s⁻¹). Meanwhile, a high potential (1.20 V vs. RHE) was applied on the ring electrode to electro-oxidize H₂O₂. The hydrogen peroxide yield [H₂O₂ (%)] and electron transfer number (n) were calculated based on the following equations:

$${\mathrm{H}}_2{\mathrm{O}}_2\left(\%\right)=200\times \frac{\frac{I_r}{N}}{{\mathrm{I}}_{\mathrm{d}}+\frac{{\mathrm{I}}_{\mathrm{r}}}{\mathrm{N}}},$$

$$\mathrm{n}=4\times \frac{\frac{{\mathrm{I}}_{\mathrm{r}}}{\mathrm{N}}}{{\mathrm{I}}_{\mathrm{d}}+\frac{{\mathrm{I}}_{\mathrm{r}}}{\mathrm{N}}},$$

where I_d is the disk current, I_r is the ring current, and N = 0.37 is the current collection efficiency of the platinum ring.

When conducting tests on Zn-air batteries, we assembled a custom-designed cell with catalysts loaded on carbon paper (with a loading density of 1 mg/cm²) serving as the air cathode, and Zn foil was used as anode. The electrolyte consisted of 6 M KOH, with the addition of 0.2 M zinc acetate to create a rechargeable zinc–air battery. Charge/discharge cycling was performed using a 5-min discharge period followed by a 5-min charging period.

DFT Calculations.

The DFT calculations were performed via Vienna ab initio simulation package (51). The ion–electron interaction was described with the projector-augmented plane-wave method (52). Exchange-correlation energy was expressed by Perdew–Burke–Ernzerhof functional with the generalized gradient approximation (53). To avoid the interlayer interaction, the vacuum layer set to be 15 Å. For geometry optimization, the cutoff energy was set to be 520 eV, and the Brillouin zone was sampled with 3×3×1 k-points. The systems were relaxed until the energy and force reaching the convergence threshold of 1 × 10⁻⁵ eV and 0.02 eV/Å, respectively. We describe the van der Waals interactions by utilizing the DFT-D₃ method (54). The binding energy (ΔE) of Fe atoms on the substrates can be given by:

$$\mathrm{\text{ΔE}}=\text{Etotal}-\text{EFe}-\text{Esubstrate},$$

where E_total, E_Fe, and E_substrate are the energy of Fe binding on the substrate, energy of Fe, and energy of substrate, respectively.

Generally, the ORR involves four proton-electron transfer steps on the active sites; thus, there are three different intermediate adsorbates: *OH, *O, and *OOH, the asterisk represents the adsorption site. In an acidic electrolyte, the 4e⁻ pathway of ORR can be expressed as:

$$\mathrm{\text{O}}2\left(\mathrm{\text{g}}\right)+{\mathrm{\text{H}}}^{+}+{\mathrm{\text{e}}}^{-}+\ast \to {\text{OOH}}^{*},$$ [a]

$${\text{OOH}}^{*}+{\mathrm{\text{H}}}^{+}+{\mathrm{\text{e}}}^{-}\to {\mathrm{\text{O}}}^{*}+{\mathrm{\text{H}}}_2\mathrm{\text{O}},$$ [b]

$${\mathrm{\text{O}}}^{*}+{\mathrm{\text{H}}}^{+}+{\mathrm{\text{e}}}^{-}\to {\text{OH}}^{*},$$ [c]

$${\text{OH}}^{*}+{\mathrm{\text{H}}}^{+}+{\mathrm{\text{e}}}^{-}\to \mathrm{\text{H}}2\mathrm{\text{O}}+\ast .$$ [d]

The reaction free energy of Eqs. a–d, (ΔG_a, ΔG_b, ΔG_c, ΔG_d) for ORR can be calculated using the following equations:

ΔG_a = G(*OOH) − G(*) − G(O₂) − G(H⁺ + e⁻)

= G(*OOH) − G(*) − (2G(H₂O)(l) − 2G(H₂) + 4 × 1.23) − 0.5G(H₂)

= G(*OOH) − G(*) − 2G(H₂O)(l) + 1.5G(H₂) − 4.92

= ΔG(*OOH) − 4.92

ΔG_b = G(H₂O) + G(O*) − G(*OOH) − G(H⁺ + e⁻)

= G(O*) + G(H₂O)(l) − 0.5G(H₂) − G(*OOH)

= (G(O*) + G(H₂) − G(H₂O)(l) − G(*)) − [G(*OOH) + 1.5 G(H₂) − 2G(H₂O)(l) − G(*)]

=ΔG(*O) − ΔG(*OOH)

ΔG_c = G(*OH) − G(H⁺ + e⁻) − G(O*)

=[G(*OH) + G(H₂O)(l) − 0.5G(H₂) − G(*)] − [G(O*) + G(H₂O)(l) − G(H₂) − G(*)]

=ΔG(*OH) − ΔG(*O)

ΔG_d = G(H₂O)(l) + G(*) − G(*OH) − G(H⁺ + e⁻)

= G(H₂O)(l) + G(*) − G(*OH) − 0.5G(H₂)

= −[G(*OH) − G(*) − G(H₂O)(l) + 0.5G(H₂)]

= −ΔG(*OH)

Therefore, we can conclude that ΔG(*OH) = −ΔG_d; ΔG(*O) = −ΔG_c − ΔG_d; ΔG(*OOH) = 4.92 + ΔG_a. In addition, we define that the rate-determining step in the free energy diagram is the minimum Gibbs free energy difference of the two adjacent intermediates.

In this model, we set up RHE as the reference electrode, which allows us to replace the chemical potential (μ) of the proton–electron pair with that of half an H₂ molecule: μ(H⁺) + μ(e⁻) = 1/2 μ(H₂), at conditions with U = 0 V and P_H2 = 1 bar. Given that the high-spin ground state of the oxygen molecule is poorly described in DFT calculations, the free energy of the O₂ molecule was derived according to G(O₂)(g) = 2G(H₂O)(l) − 2G(H₂)+ 4 × 1.23 (eV).

With this approach, the theoretical onset potential (U_onset) for ORR at standard conditions, which is used to assess the ORR performance, is defined as

$$U_{\text{onset}}=\frac{-\text{max}\left\{{\text{ΔG}}_{\text{a}},{\text{ΔG}}_{\text{b}},{\text{ΔG}}_{\text{c}},{\text{ΔG}}_{\text{d}}\right\}}{\text{e}}.$$

The Gibbs free energy change ΔG of ORR on the designed catalysts was evaluated by the formula

$$∆\mathrm{\text{G}}=∆\mathrm{\text{E}}+∆\text{ZPE}+∆{\int }_0^{\mathrm{\text{T}}}{\mathrm{C}}_{\text{p}}{\text{dT}-\mathrm{\text{T}} ∆\mathrm{\text{S}}+∆\mathrm{\text{G}}}_{\mathrm{\text{U}}}{+∆\mathrm{\text{G}}}_{\text{pH}},$$

where ΔE is the adsorption energy of ORR intermediates, and ΔZPE is their corresponding zero-point energy, ΔG_U is the free energy contribution induced by electrode potential U. ΔG_pH is the correction of the H⁺ free energy by the concentration, which can be evaluated as ΔG_pH = 2.303 × k_BT × pH (or 0.06 × pH), here the value of pH was assumed to be zero. Cp represents the constant-pressure heat capacity. The entropy and integration terms are computed using the vibrational energies of ORR intermediates.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

In-situ TEM images of the thermal printing process.

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information.