Spatial configuration of Fe–Co dual-sites boosting catalytic intermediates coupling toward oxygen evolution reaction

Significance

Water electrolysis is a main strategy for sustainable hydrogen production. The main obstacle limiting the efficiency of electrolytic water is the sluggish OER of four electron transfer steps. Dual-site catalysts are widely considered to be more fantastic than single-site catalysts due to the cooperation of pair sites. However, how to obtain the optimized OER catalytic performance by regulating the spatial configuration of dual sites is still unclear. Here, we report two kinds of catalysts with different spatial structures; we further explain the influence of spatial configuration on the dual-sties cooperation for boosting OER performance through DFT calculations and isotope-DEMS and in situ spectroscopies. This work provides insights into the subsequent research of OER catalysis through spatial structure of dual-sites catalysts.

Abstract

Oxygen evolution reaction (OER) is the pivotal obstacle of water splitting for hydrogen production. Dual-sites catalysts (DSCs) are considered exceeding single-site catalysts due to the preternatural synergetic effects of two metals in OER. However, appointing the specific spatial configuration of dual-sites toward more efficient catalysis still remains a challenge. Herein, we constructed two configurations of Fe–Co dual-sites: stereo Fe–Co sites (stereo-Fe-Co DSC) and planar Fe–Co sites (planar-Fe-Co DSC). Remarkably, the planar-Fe-Co DSC has excellent OER performance superior to stereo-Fe-Co DSC. DFT calculations and experiments including isotope differential electrochemical mass spectrometry, in situ infrared spectroscopy, and in situ Raman reveal the *O intermediates can be directly coupled to form *O–O* rather than *OOH by both the DSCs, which could overcome the limitation of four electron transfer steps in OER. Especially, the proper Fe–Co distance and steric direction of the planar-Fe-Co benefit the cooperation of dual sites to dehydrogenate intermediates into *O–O* than stereo-Fe-Co in the rate-determining step. This work provides valuable insights and support for further research and development of OER dual-site catalysts.

Hydrogen is an ideal energy source pursued by people because of its zero carbon, high calorific value, and clean use (1, 2). Nowadays, electrolysis water is considered as a renewable approach to generate hydrogen (3, 4). However, the main obstacle affecting the efficiency of electrolytic water is that anodic oxygen evolution reaction (OER) has four electron transfer steps (5–7), namely, the kinetics is slow. At present, iridium dioxide (IrO₂) and ruthenium dioxide (RuO₂) are regarded as effective OER catalysts (8–10). However, the high cost, scarcity, and instability of these noble metal catalysts under alkaline conditions seriously hinder their practical application (11–14). Therefore, it is very desirable to find an efficient and economical electrocatalyst to promote OER.

Dual-site catalysts (DSCs), which refer to two adjacent active sites employing synergistic effects to promote the overall reaction kinetics (15), recently become an extension of the single atom catalysts (SACs) toward more efficient catalysis (16–18). For example, when the dual-active sites are involved in the OER process, the coupling of *O–O* instead of OOH* intermediates could be produced (19, 20), leading to the replacement of the four-electron process in traditional adsorbate evolution mechanism (AEM). It thereby can effectively break the limitation of the theoretical overpotential (370 mV) of AEM (21, 22). The DSCs are significantly dependent on the spatial structure such as metal-site location and the distance of metal–metal due to electron orbital interaction. Accordingly, the electronic state of each active site can be optimized, and on this basis, the binding behaviors between the site and catalytic intermediates could be regulated (23, 24). Although many DSCs have been used widely for efficient catalysis recently (25–30), the insight into the spatial configuration of the two metallic active sites in DSCs to boost OER catalysis still remains obscurity and should be further clarified.

Herein, we prepared atomically dispersed Fe sites embedded in CoOOH substrate, in which the monatomic Fe cooperates with adjacent Co forming DSCs. The spatial structure of Co–Fe dual sites can be regulated by the monatomic Fe location: stereo Fe–Co sites (named stereo-Fe-Co DSC) and planar Fe–Co sites (named planar-Fe-Co DSC). Compared with stereo-configuration DSC, the planar-Fe-Co DSC showed more excellent OER catalytic performance under alkaline conditions, with an over-potential of 190 mV at 10 mA cm⁻² and 229 mV at 100 mA cm⁻², besides, the durability of more than 160 h. Experiments including isotope-labeling and in situ spectroscopy as well as density functional theory (DFT) calculations demonstrate the prepared catalysts following a pathway of dual-site mechanism (also namely oxide path mechanism) rather than traditional AEM of OER. The planar-Fe-Co DSC is more conducive to the intermediates dehydrogenation in the rate-determining step than stereo-Fe-Co DSC to form *O–O* species. Eventually, the proper spatial structure of monatomic Fe sites could promote the cooperation of isolated Fe atoms and adjacent Co atoms, which is the origin of much enhanced electrocatalytic performance.

Results and Discussion

The DSC substrate CoOOH was synthesized via a method of electrodeposition on nickel foam (NF) followed by electroactivation (SI Appendix). The characterization shows the obtained CoOOH vertically arranged on NF with hexagonal γ-phase and in nanosheet morphography (SI Appendix, Fig. S1). After the incorporation of Fe through the electrodeposition, two DSCs of stereo-configuration or planar-configuration are prepared (Fig. 1A). Fig. 1B shows that the morphology of planar-configuration DSC exhibits no perceptible change compared to original lamellar CoOOH. The high-resolution transmission electron microscopy (HRTEM) image (Fig. 1C) demonstrates it without any discernible metallic nanoparticles, and the lattice spacings of 0.141 nm and 0.240 nm are corresponded to the (110) and (101) crystallographic planes of γ-CoOOH (SI Appendix, Fig. S2), respectively. The Fe atom could not be distinguished in it because of its close atomic number with Co atom. The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) shows no obvious metal clusters and some bright spots were scattered on the CoOOH nanosheet surface (Fig. 1D). The existence of the Fe element is demonstrated by energy-dispersive X-ray spectroscopy (EDX) mappings (Fig. 1E), in which Fe elements uniformly dispersed on the CoOOH nano-slice. To further confirm the loading of Fe single-atoms on CoOOH nanosheets, we selected the red region for scanning the electron energy loss spectroscopy (EELS) mapping through spherical aberration-corrected transmission electron microscope (ACTEM) (Fig. 1 F and G). One of the isolated red pixels surrounded by the white dash square is marked in Fig. 1G, where the serial numbers match with that of extracted EELS spectra in Fig. 1H. Thereinto, the signal peak around 710 eV, corresponding to the Fe L edge, only appeared in the yellow circle area, which confirms that the material was indeed loaded with a small amount of Fe single-atoms with great dispersibility. Subsequently, the inductively coupled plasma optical emission spectrometry (ICP-OES) result of planar-Fe-Co DSC verifies that the Fe concentration is 5.28 wt% approximately. On the other hand, the experimental details of stereo-Fe-Co DSC are also studied (SI Appendix, Figs. S3–S5). The ICP-OES result of stereo-Fe-Co DSC verifies that the Fe concentration is 5.15 wt% approximately.

Fig. 1.

Schematic of the synthetic process and characterization. (A) Schematic representation of the synthetic procedure for stereo-Fe-Co DSC and planar-Fe-Co DSC. (B) TEM images of planar-Fe-Co DSC. (C) HRTEM images of planar-Fe-Co DSC. Inset shows the selected area electron diffraction (SAED) pattern of planar-Fe-Co DSC. (D) HAADF-STEM images of planar-Fe-Co DSC. (E) EDX maps of planar-Fe-Co DSC. (F) ACTEM image of planar-Fe-Co DSC. (G) EELS-mapping of the area in the red square in (F). (H) EELS spectra acquired from the corresponding pixels in the white square in (G).

The atomic-site structures of Co and Fe in the stereo-Fe-Co DSC and planar-Fe-Co DSC were investigated by X-ray absorption fine spectroscopy (XAFS). Fig. 2A shows the X-ray absorption near-edge structure (XANES) of Co element in DSC. It shows the absorption edge of Co of planar-Fe-Co DSC is closer to the high energy region compared with Co foil and standard Co₃O₄, and further valence fitting displays its oxidation state of +3.11 (SI Appendix, Fig. S6). That indicates the planar-Fe-Co DSC catalyst has strong electronic metal-support interaction (EMSI), which increases the positive charge of the substrate Co induced by the supported Fe atoms. It is also proved from the side that the introduction of Fe single atom definitely affects the electronic density of Co atom. However, the absorption edge of Co of stereo-Fe-Co DSC locates to lower energy than that of CoOOH, and further valence fitting displays its oxidation state of +2.98 (SI Appendix, Fig. S6). The Fourier transform (FT) Co K-edge extended X-ray absorption fine structure (EXAFS) shows that CoOOH, stereo-Fe-Co DSC, and planar-Fe-Co DSC have similar double peaks at 1.44 and 2.42 Å, corresponding to Co–O and Co–O–Co (Fig. 2B) (31), respectively. By fitting the experimental EXAFS spectra, the first-shell coordination number of Co–O for stereo-Fe-Co DSC and planar-Fe-Co DSC is five. Im addition, the second-shell coordination number of Co–O–Co for stereo-Fe-Co DSC and planar-Fe-Co DSC is six (SI Appendix, Table S1).

Fig. 2.

(A) Co K-edge XANES spectra of Co foil, Co₃O₄, CoOOH, stereo-Fe-Co DSC and planar-Fe-Co DSC. The illustration shows a partial enlarged view. (B) FT-EXAFS characterization of Co foil, Co₃O₄, CoOOH, stereo-Fe-Co DSC, and planar-Fe-Co DSC. (C) Fe K-edge XANES spectra of Fe foil, Fe₂O₃, stereo-Fe-Co DSC, and planar-Fe-Co DSC. The illustration shows a partial enlarged view. (D) FT-EXAFS characterization of Fe foil, Fe₂O₃, stereo-Fe-Co DSC, and planar-Fe-Co DSC. (E) Fe K-edge FT-EXAFS fitting curves of planar-Fe-Co DSC and the corresponding atomic structure (Inset). (F) Fe K-edge FT-EXAFS fitting curves of stereo-Fe-Co DSC and the corresponding atomic structure (Inset). Structural characterizations of CoOOH, stereo-Fe-Co DSC, and planar-Fe-Co DSC. (G) Co 2p XPS. (H) O 1s XPS. (I) Fe 2p XPS.

We further identified atomic-site structures of stereo-Fe-Co DSC and planar-Fe-Co DSC by XANES simulations of the Fe K-edge as shown in Fig. 2C. The inset of Fig. 2C and valence fitting (SI Appendix, Fig. S7) display their oxidation states of +3.08 and +3.14 for stereo-Fe-Co DSC and planar-Fe-Co DSC, respectively. In Fig. 2D of Fe K-edge FT-EXAFS, the absence of Fe–Fe bonding at approximately 2.20 Å confirmed the atomic dispersion of single Fe atoms in both samples, precluding the formation of Fe-based clusters or nanoparticles. The stereo-Fe-Co DSC and planar-Fe-Co DSC demonstrate the presence of Fe–O bonds at 1.41 and 1.50 Å, respectively, representing the first shell coordination. In addition, the Fe–O–Co bond present at 2.48 and 2.58 Å that proved the interaction between Fe atoms and Co atoms in both the samples and represented the second shell coordination. Meanwhile, EXAFS fitting curves of planar-Fe-Co DSC show that the Fe atom is located on the same plane with oxygen and cobalt (Fig. 2E), and Fe is directly coordinated with five oxygen atoms (SI Appendix, Table S2). Significantly, the Fe–O–Co peak at 2.58 Å from the second shell coordination confirms the incorporation of Fe atoms into the lattice of CoOOH in planar-Fe-Co DSC. In contrast, the Fe K-edge FT-EXAFS fitting curves of stereo-Fe-Co DSC demonstrate the Fe atom is anchored above the surface with the Fe–O–Co distance of 2.48 Å, and the Fe atom is coordinated with five oxygen atoms, which can be considered as the adsorption of Fe on CoOOH surface (Fig. 2F).

In addition, we further characterized the atomic-site structures of DSC through the X-ray photoelectron spectroscopy (XPS) results. The Co 2p XPS result of planar-Fe-Co DSC has a high-energy shift of 0.4 eV compared with the stereo-Fe-Co DSC catalysts (Fig. 2G). It shows that the Co valence of planar-Fe-Co DSC is higher than the stereo-Fe-Co DSC, which is consistent with the EXAFS results (Fig. 2A). The O 1s XPS results show that the planar-Fe-Co DSC and stereo-Fe-Co DSC have a low-energy shift of 0.5 eV compared with the CoOOH catalysts (Fig. 2H), which means that the loaded single atom of Fe takes up part of the O and forms the Fe–O bond (32), leading to the shift of O_M-O peak. The Fe 2p XPS results of planar-Fe-Co DSC has a high-energy shift of 0.4 eV compared with the stereo-Fe-Co DSC catalysts (Fig. 2I). It shows that the Fe valence of planar-Fe-Co DSC is higher than the stereo-Fe-Co DSC, which is consistent with the EXAFS results (Fig. 2C).

In order to investigate the different spatial configurations affecting OER performance, we compared the OER performance of stereo-Fe-Co DSC and planar-Fe-Co DSC in 1.0 M KOH electrolyte using a conventional three-electrode system (Fig. 3A). After testing the LSV performance, the planar-Fe-Co DSC exhibits apparently the highest OER performance. As shown in Fig. 3B, at the current density of 10 and 100 mA cm⁻², the over-potential of planar-Fe-Co DSC is only 190 and 229 mV, respectively, which is lower than the stereo-Fe-Co DSC (210 and 256 mV), Meanwhile, this OER performance is quite better than that of cobalt-based catalysts reported recently (SI Appendix, Table S3), and as well higher than that of some dual-site catalyst, such as (Fe, Co)OOH (230 mV at 10 mA cm⁻²) (33), Fe–Ni DSC (270 mV at 10 mA cm⁻²) (17), and Ni–Fe DSC (245 mV at 20 mA cm⁻²) (34).

Fig. 3.

Electrocatalytic performances toward OER. (A) Polarization curves of NF, CoOOH, stereo-Fe-Co DSC, planar-Fe-Co DSC, and RuO₂. The measurements were conducted in 1.0 M KOH. (B) OER activities of the catalysts expressed as overpotentials required for 10 and 100 mA cm^–2. (C) Tafel slopes of CoOOH, stereo-Fe-Co DSC, planar-Fe-Co DSC, and RuO₂. (D) Electrochemical impedance spectroscopy of the CoOOH, stereo-Fe-Co DSC, and planar-Fe-Co DSC catalysts. (E) Charging current density differences of planar-Fe-Co DSC. (F) OER polarization curves with the current density normalized by ECSA of electrocatalysts. (G) Chronopotentiometry curves of stereo-Fe-Co DSC and planar-Fe-Co DSC at 10 mA cm⁻² for 160 h. (H) In situ Raman spectra of planar-Fe-Co DSC at different applied potentials. (I) In situ Raman spectra of stereo-Fe-Co DSC at different applied potentials.

Clearly, planar-Fe-Co DSC could be more conductive to the reaction process of OER than stereo-Fe-Co DSC. In order to further evaluate the performance of the catalyst, we evaluate the OER kinetics by calculating the Tafel slope, as shown in Fig. 3C, the Tafel slopes of the stereo-Fe-Co DSC and the planar-Fe-Co DSC are lower than that of the CoOOH catalyst, which indicates that the introduction of Fe atom can indeed enhanced its intrinsic catalytic activity for the OER (35). This result also shows that the planar-Fe-Co DSC catalyst has the minimum Tafel slope (33.6 mV·dec⁻¹), which is lower than the those of the stereo-Fe-Co DSC (39.3 mV·dec⁻¹), CoOOH (68.1 mV·dec⁻¹), and RuO₂ (86 mV·dec⁻¹), that means the planar-configuration Fe–Co dual-sites can induce and optimize the electronic structure of Co atom and improve the electron transfer rate. The electrochemical impedance spectroscopy (EIS) can be regarded as an important result of judging OER kinetics. The arc of EIS spectrum is positively correlated with the charge transfer resistance (R_ct) of the electrocatalyst. As shown in Fig. 3D, it can be concluded that the charge transfer resistance decreases, and the electron transfer ability increases after loading Fe atoms. The high activity of planar-Fe-Co DSC is also confirmed by the calculation of the turnover frequency (TOF) of 59.22 s⁻¹, which is higher than that of stereo-Fe-Co DSC (9.838 s⁻¹) and CoOOH (0.0592 s⁻¹). In addition, we also evaluated the double-layer capacitance of the planar-Fe-Co DSC catalyst, which is proportional to the ECSA value (SI Appendix, Fig. S11). According to the cyclic voltammograms at different scanning rates, the planar-Fe-Co DSC shows a higher C_dl value (1.41 mF cm⁻²) than the stereo-Fe-Co DSC (1.35 mF cm⁻²) and CoOOH (1.18 mF cm⁻²) (Fig. 3E, SI Appendix, Figs. S12 and S13). Then, normalize the LSV curve with C_dl value to eliminate the influence of ECSA change on geometric current density. The results show that the planar-Fe-Co DSC catalyst has high intrinsic activity. When the anodic current density was normalized by calculated ECSA, planar-Fe-Co DSC still maintained the highest value, indicating outstanding intrinsic activity (Fig. 3F).

The stability of the structure of the catalyst described is the stability test chrono-potentiometric (CP) method. When we apply a fixed voltage to keep the current density at 10 mA cm⁻² for 160 h, the loss of current density can also be ignored of the stereo-Fe-Co DSC and planar-Fe-Co DSC (Fig. 3G). In addition, the polarization curve of the planar-Fe-Co DSC after 2,000 cv cycle is close to the polarization curve after initial cycle (SI Appendix, Fig. S14). These results show that the planar-Fe-Co DSC catalyst has excellent OER performance and high stability. We also conducted in situ Raman analysis of stereo-Fe-Co DSC and planar-Fe-Co DSC to analyze the OER potential-resolved intermediate variation for structural stability. We conducted in situ Raman testing on planar-Fe-Co DSC catalyst (Fig. 3H), and from OCP to 1.5 V, the peak of Fe–CoOOH at 472 cm⁻¹ (E_g bending vibration of O–Co–O) and 540 cm⁻¹ (A_1g stretching vibration of O–Co–O) have no shifted (36–38), which proves that the material has good structural stability. Similarly, the stereo-Fe-Co DSC catalyst also show the same properties (Fig. 3I). This result consistent with the excellent OER performance and high stability.

In OER, several mechanisms have been extensively studied, such as AEM, lattice oxygen mechanism (LOM), and oxide path mechanism (OPM) (21, 39, 40). The AEM occurs at the single metal catalytic site and the alkaline OER process involves four electrochemical reaction steps (21): HO*→O*→OOH*→O₂. However, this reaction mechanism is limited by the theoretical minimum over potential of 370 mV. The LOM is that lattice oxygen participates in the formation of O₂ through O–O direct coupling, which can break the limit of AEM theoretical overpotential, but lattice oxygen will precipitate and lead to the instability of the catalyst. The OPM means that dual sites participate in the reaction cooperatively, and O₂ is formed through O–O direct coupling, and there is no lattice oxygen involved.

In order to further investigate the reaction mechanism of catalyst, we conducted in situ differential electrochemical mass spectrometry (DEMS) testing with isotope labeling measurements (SI Appendix, Fig. S17). We first marked the stereo-Fe-Co DSC and planar-Fe-Co DSC catalyst in the KOH solution with H₂¹⁸O as the water source, and then carried out the DEMS test in the KOH solution with H₂¹⁶O as the water source (Fig. 4 A and B). For the OER through the AEM (SI Appendix, Figs. S18 and S19), ³²O₂ and ³⁴O₂ signals will be generated. There will be the absence of ³⁶O₂ signal, but for the OER through the OPM or LOM (SI Appendix, Figs. S20 and S21), ³²O₂, ³⁴O₂, and ³⁶O₂ signals will be generated. It is obvious that we have detected the ³⁶O₂ signal through DMES test, so the OERs of the stereo-Fe-Co DSC and planar-Fe-Co DSC catalysts are carried out with OPM or LOM. In addition, peroxo-like (*O₂²⁻) species are considered as reaction intermediates of LOM pathway (41). In addition, the *O₂²⁻ species can be directly probed during the OER process. It has been reported that *O₂²⁻ can be bound by tetramethylammonium (TMA⁺) (22, 41, 42), leading to reduced kinetics of OER through the LOM pathway, so we conducted a control experiment (with or without TMA⁺) (Fig. 4 C and D), and no change of polarization curve was detected, which proves that there is no production of *O₂²⁻ intermediate during the catalytic process. This result proves that the catalysts stereo-Fe-Co DSC and planar-Fe-Co DSC do not follow the LOM. Therefore, the OERs of the stereo-Fe-Co DSC and planar-Fe-Co DSC catalysts are carried out with OPM. Furthermore, we conducted surface-enhanced infrared spectroscopy in the attenuated total reflection (ATR-SEIRAS) analysis (Fig. 4 E and F). After the applied potential was increased from OCP to 1.3 V, the sharp band at 1,240 cm⁻¹ can be assigned to the Si–O–Si vibration from the ATR crystal (43). Significantly, both the planar-Fe-Co DSC catalyst and the stereo-Fe-Co DSC catalyst have observed a peak at 1,160 and 1,170 cm⁻¹ representing *O–O* intermediate (19, 44), which represent the oxygen bridges intermediate at the dual-site (20). As the voltage increases, the peaks at 1,470 and 1,468 cm⁻¹ gradually appear, representing O–O* intermediate (45, 46), which are the intermediate before releasing O₂. This result further confirms that the catalyst follows the OPM path during OER process in alkaline solution.

Fig. 4.

(A) DEMS signals of O₂ products for planar-Fe-Co DSC. (B) DEMS signals of O₂ products for stereo-Fe-Co DSC in the electrolyte using H₂O as the solvent during five times of LSV in the potential range of 1.20 to 1.65V vs. RHE, with a 5 mV s⁻¹ scan rate. (C) Polarization curves of planar-Fe-Co DSC with or without TMA⁺. (D) Polarization curves of stereo-Fe-Co DSC with or without TMA⁺. In situ FTIRs spectra recorded in the potential range of OCP to 1.6 V vs. RHE for (E) planar-Fe-Co DSC and (F) stereo-Fe-Co DSC.

To further investigate the reaction mechanism of planar-Fe-Co DSC and stereo-Fe-Co DSC catalysts at dual sites, particularly elucidating the impact of spatial configuration on the intrinsic catalytic activity for OER, a series of DFT calculations was carried out. Fig. 5A shows the OER network of planar-Fe-Co DSC. Free energy calculation determined that hydroxylated Fe is thermodynamic favorable than hydroxylated its neighboring Co site (0.80 vs. 1.24 eV). Hence, the following OER networks were proposed based on the active center composed of a hydroxylated Fe and its neighboring innocent Co site. It should be noted that dragging out one lattice oxygen atom to design O–O coupling step is impossible since the corresponding intermediates would step back to the configuration where lattice oxygen atom was still in its initial location in clean CoOOH surface after optimization. This result is in good agreement with our TMA⁺ testing (Fig. 4 C and D). Based on both TMA⁺ testing and theoretical calculation, LOM is excluded, thus, reaction network–based AEM and OPM were proposed. Classic AEM is suggested to undergo four proton coupling electron transfer steps via *OH → *O → *OOH → O₂ on the active site as shown in the topmost branch in Fig. 5A. Interestingly, there is a competing reaction in the second electrochemical step, where forming dual *OH is more favorable than forming Fe–*O directly in the framework of AEM. The free energy difference of these two parallel reactions is 0.31 eV. Subsequently, the third electrochemical step occurs on Fe–*OH overwhelmingly, instead of on Co–*OH. Here, the free energy difference of these two parallel reactions is 0.53 eV. The fourth electrochemical step could undergo via*O–O* or *OOH*OH intermediates. However, forming the later one needs to climb a step of 1.69 eV. However, forming *O–O* needs to climb a step of 1.00 eV only. Finally, *O–O* desorbs from the active center via a thermochemical step. This reaction network determined that OER on planar-Fe-Co DSC is mainly followed OPM pathway, instead of LOM or AEM. The results show that the rate-determining step (RDS) of OPM is the step from *OH*OH to *O*OH with a barrier of 1.46 eV, which is 0.26 eV lower than that of AEM (*O → *OOH).

Fig. 5.

Scheme of OER mechanism and Gibbs free energy. (A) Planar-Fe-Co DSC and (B) stereo-Fe-Co DSC (blue: Co; yellow: Fe; red: O; white: H). (C) The Gibbs free energy diagram of the OER process on CoOOH, planar-Fe-Co DSC and stereo-Fe-Co DSC. (D) PDOS of surface Co atoms in planar-Fe-Co DSC. Insets show the electron density map of Co atoms (on the same terrace with Fe). (E) PDOS of surface Fe atoms in planar-Fe-Co DSC. Insets show the electron density map of Fe atoms.

In order to determine the configuration of the stereo-Fe-Co DSC, the theoretical calculation results show the model of SI Appendix, Fig. S27C is most stable (details are shown in SI Appendix, Figs. S26–S28), which agree with the EXAFS curve-fitting analysis well (Fig. 2F). Hence, in stereo-Fe-Co DSC model, Fe is coordinated with two lattice oxygen as well as three hydroxyl groups. The OER paths on stereo-Fe-Co DSC model were shown in Fig. 5B. Unlike any other form of planar-Fe-Co DSC, the Fe atom located spatially higher than the Co sites in CoOOH substate. Meanwhile, the hydroxyl groups connected Fe and Co atoms are not lattice oxygen featured, which could not be involved in OER via LOM, as determined by TMA⁺ testing (Fig. 4 C and D) and theoretical calculations. In this model, *O–O* intermediate is formed on single Fe atom instead of Fe–Co or Co–Co dual metal sites. Although the investigated two models underwent OER via OPM path, there are significant difference for their active centers (dual Fe–Co and single Fe for planar-Fe-Co DSC and stereo-Fe-Co DSC, respectively). It is noted that the *O–O* intermediate in some typical single-site catalysts has been suggested to take into account when predicting the activities of catalysts (47).

The OER free energy profiles of three models via OPM path are shown in Fig. 5C. We also found that the rate-determining step of planar-Fe-Co DSC catalyst is the third electrochemical step (from *OH*OH to *O*OH), while the rate-determining step of stereo-Fe-Co DSC is the fourth electrochemical step (from *O*OH to*O–O*). The rate-determining step height of planar-Fe-Co DSC is 1.46 eV, which is lower than its counterparts of stereo-Fe-Co DSC (1.66 eV), indicating that planar-Fe-Co configuration benefited for dehydrogenation of intermediate to form *O–O* than stereo-Fe-Co configuration, which is in good agreement with our experimental results.

To further reveal the excellent structure of catalytic performance planar-Fe-Co DSC, the projected density of stats (PDOS) of these models were calculated as shown in SI Appendix, Fig. S29. According to the surface configuration exposed, one terrace which contains doped Fe atom, while another terrace contains Co atoms exclusively, was chosen to represent Fe–Co dual site and Co–Co dual site in the same configuration, respectively. PDOS of Co atoms on CoOOH surface exhibit a symmetry for spin-up and spin-down, indicating their low-spin configuration character. When latticed-Fe is introduced, the surface Co atoms far away from latticed-Fe exhibit a significant asymmetry for spin-up and spin-down channels. Clearly, spin-up d_z2 is fully occupied. This agrees the O K-edge XANES result Co³⁺ configuration transformation from t_2g⁶ e_g⁰ to t_2g⁵ e_g¹ quite well (SI Appendix, Fig. S8). For the Co atom which is close to latticed-Fe_, its spin-polarization feature is less significant. Comparing the PDOS as well as their contribution to the highest occupied orbitals (HOMO) of Fe–Co dual sites and Co–Co dual sites, we can find that d_z2 of Fe and d_zx/d_zy of neighboring Co contributes to HOMO synchronously (Fig. 5 D and E). Meanwhile, on terraces without latticed-Fe, Co–Co dual sites exposed their d_z2 orbitals in the vicinity of Fermi level. It is the asynchronization of d_z2 in Fe–Co dual site, leading to the selection of first electrochemical step on Fe site. Furthermore, this characteristic might facilitate the *O*O→*O₂→O₂(g) thermochemical step, with breaking Co–O bond followed by Fe–O bond, successively. This is consistent with the *O–O* and O–O* reaction intermediates detected in the in situ infrared experiments.

The active center of stereo-Fe-Co DSC is remarkably distinguished from the active center in planar-Fe-Co DSC. However, the coordination numbers vary between five and six during the OER loop for both stereo-Fe-Co DSC and planar-Fe-Co DSC. The main d-orbitals involved in the evolution of intermediates are d_zx/d_xy for stereo-Fe-Co DSC (SI Appendix, Fig. S30) and d_z2 for planar-Fe-Co DSC, respectively, due to their different spatial distributions. Once the reaction is triggered, center Fe atoms remain six-coordinated and form octahedral geometry. In stereo-Fe-Co DSC, the distance of Fe–Co dual sites is 2.357 Å, which is shorter than it is in planar-Fe-Co DSC (2.914 Å). The shorter Fe–Co distance together with their relative direction determined that side by side *OH*OH is less stable than one shared OH on the Fe–Co bridge, accompanied by one Fe–OH exclusively. The bridging OH groups separated Fe and neighboring Co atoms, making Fe atom as the sole active metal site.

Conclusions

In summary, by different electrodeposition methods, two catalysts stereo-Fe-Co DSC and planar-Fe-Co DSC were successfully synthesized. The different spatial relationship between Fe single-atom and Co atom in CoOOH substrate were characterized by XPS and XAFS. In electrolysis water tests, the planar-Fe-Co DSC catalyst showed excellent OER catalytic performance with the overpotential of 190 mV (at 10 mA cm⁻²) as well as excellent stability of more than 160 h, which indicates that it may be a very promising candidate for economically effective non-noble metal electrocatalysts. In addition, benefited from dual active sites, the stereo-Fe-Co DSC and planar-Fe-Co DSC followed the OPM reaction mechanism rather than traditional AEM path when being used as OER anodes, which was confirmed by a serious of in-situ characterization and theoretical calculation. The change of active site space configuration from stereo-configuration to planar-configuration is regarded as facilitating the dehydrogenation of the catalytic intermediates, boosting their coupling for oxygen releasing. Our work presents a unique aspect of dual-atom catalysts regulation. This new regulation strategy may further facilitate optimization of dual-atom catalysts toward high performance in the future application.

Methods

Synthesis of CoOOH.

The Co(OH)₂ electrodeposited on nickel foam (NF) was synthesized according to a previous report (48). Then, the obtained Co(OH)₂ on nickel foam was used as the working electrode, Hg/HgO as the reference electrode, platinum plate electrode as the counter electrode, and a 1.0 M KOH solution as the electrolyte. The electrochemical method of LSV technique was employed with parameters set from 0.20 to 0.85 V at a deposition rate of 5 mV s⁻¹. After activation, CoOOH on NF was cleaned three times with deionized water.

Synthesis of Stereo-Fe-Co DSC.

Fe atom was electrodeposited onto the CoOOH using a three-electrode system consisting of the CoOOH as the working electrode, Ag/AgCl (saturated KCl) as the reference electrode, platinum plate electrode as the counter electrode, and a 1.0 M KOH solution containing 100 µM FeCl₃ as the electrolyte. The electrochemical method used LSV technique with parameters was set from 1.1 to 1.8 V at a deposition rate of 5 mV s⁻¹. After electrodeposition, it was cleaned with deionized water for three times and baked it at 100 °C in vacuum for 1 h.

Synthesis of Planar-Fe-Co DSC.

The synthesis of planar-Fe-Co DSC is consistent with the synthesis method of stereo-Fe-Co DSC, except that after depositing iron single atoms for five times, the electrochemical activation process is repeated, and then another deposition of iron single atoms is repeated for five times. After electrodeposition, the obtained samples were washed with deionized water and used for subsequent electrochemical measurements. The formula for converting Hg/HgO electrodes to reversible hydrogen electrodes (RHE) for all potential measurements is E (V vs. RHE) = E (V vs. Hg/HgO) + 0.098 V+ 0.0592 pH V. The formula for converting Ag/AgCl (saturated KCl) electrodes into reversible hydrogen electrodes (RHE) for all potential measurements mentioned in this work is E (V vs. RHE) = E (V vs. Ag/AgCl) + 0.197 V + 0.0592 pH V.

Electrochemical Testing.

The electrocatalytic performance of the sample was measured using an electrochemical workstation (CHI660E, Shanghai CH Instruments). Electrocatalytic measurements are conducted in a standard three-electrode system at room temperature. The foam nickel electrode loaded with the obtained catalyst was used as the working electrode. The platinum electrode was used as the counter electrode. The Hg/HgO electrode was used as the reference electrode. The electrolyte is 1.0 M potassium hydroxide. The electrochemical method is linear scanning voltammetry. The set parameter scanning range is 1.10 to 1.80 V, the scanning rate is 5 mV·s⁻¹, and the iR compensation is 95% to obtain the OER polarization curve. The corrected potential was calculated via the equation: E_iR-corrected = E (V vs. RHE) – iR. The potential range of ECSA measurements was set as 0.70 to 0.80 V.

The Tafel slope (b) was determined by fitting polarization curves to the Tafel equation [1]:

$$\eta =a+b \mathrm{l}\mathrm{o}\mathrm{g} \left|j\right|,$$ [1]

where η is the overpotential for OER and j is the current density at the given overpotential.

The electrochemical impedance spectroscopy (EIS) measurements were conducted at OCP. The amplitude of the sinusoidal wave was 10 mV. The frequency scan range was 100 kHz to 0.01 Hz.

The electrochemically active surface area (ECSA) of the electrocatalysts was calculated using Eq (2):

$$\mathit{ECSA}=\frac{C_{\mathrm{dl}}}{C_{\mathrm{s}}},$$ [2]

where C_dl is the electric double-layer capacitance calculated from the non-Faradaic region, and C_s is the specific capacitance of a flat, smooth electrode surface, whose value was numerically taken as 40 μF cm⁻² (49). For the calculation of C_dl, cyclic voltammograms of the electrode material were recorded in the non-Faradaic region at various scan rates, and a graph of scan rates vs. current densities (Δj = j_anodic − j_cathodic) was plotted to determine the slope of the graph through a linear fitting. The value of the slope was numerically equal to twice the C_dl value; therefore, C_dl was half the slope value.

TOF Calculation.

The TOF values were estimated according to the following equation [3]:

$$\mathit{TOF}=\frac{j\times A}{\alpha \times N\times F}$$ [3]

where j (mA cm⁻²) is the current density at a given overpotential (250 mV). A (cm²) is the geometric area of the loading Ni foam. α represents that the number of transferred electrons in OER is 4. N (mol) is the number of catalytic active sites. F is the Faraday constant (96485.3 C mol⁻¹).

Operando DEMS with Isotope Labeling.

The operando DEMS system (QAS 100 Linglu Instruments) is similar to the system reported by Li et al. (19). The device consists of two interconnected vacuum chambers, including a mass spectrometer chamber with high vacuum and a second chamber with mild vacuum. The second chamber is directly connected to an electrochemical cell operating under ambient pressure. The pressure difference allows the oxygen generated in situ to be sucked down into the vacuum chamber for mass spectrometry analysis, rather than released upward into the air. The working electrode is connected by the catalyst and polytetrafluoroethylene membrane through copper tape. The hydrophobic polytetrafluoroethylene membrane permits gas flow while rejecting liquid. A cold trap cooled by dry ice needs to be installed between the electrochemical cell and the vacuum chamber to capture water vapor and avoid potential damage to the mass spectrometer. The electrochemical cell is a typical three-electrode system with a volume of approximately 3 mL. The small volume is suitable for the isotope experiments. For isotope labeling studies, 2 mL KOH solution (1 M) was prepared using H₂¹⁸O as the solvent. The catalysts including CoOOH, stereo-Fe-Co DSC, and planar-Fe-Co DSC were subjected to five LSV cycles in the potential range of 1.20 to 1.65 V vs. RHE at a scan rate of 5 mV s⁻¹, while the mass signals of the gaseous products ³²O₂, ³⁴O₂, and ³⁶O₂ were recorded. Then, five consecutive LSV cycles (1.20 to 1.65 V vs. RHE at 5 mV s⁻¹) were applied for labeling the catalyst surface with ¹⁸O. The catalysts were washed with abundant water (H₂¹⁶O) to remove H₂¹⁸O molecules physically attached to the catalyst layer, while the ¹⁸O-containing species that chemically bonded on the surface remained. The catalysts with isotope-labeled surface then operated in a normal electrolyte 1 M KOH with H₂¹⁶O as the solvent. Meanwhile, the gaseous products including ³²O₂, ³⁴O₂, and ³⁶O₂ were monitored by the mass spectrometer. Before the electrochemical measurements, all the electrolytes were purged with high-purity Ar to remove the dissolved oxygen.

Experiments of In Situ Raman Spectra.

Ex situ/in situ Raman spectrum analysis was carried out at the control potential of CHI660E electrochemical workstation. A standard three-electrode system was used, with planar-Fe-Co DSC and stereo-Fe-Co DSC as the working electrode, platinum plate as the counter electrode, Ag/AgCl (saturated KCl) electrode as the reference electrode, electrolyte as 1.0 M KOH solution, and control potential as 0.20 to 0.70 V. In situ Raman spectra were operated in a commercial electrolytic cell with a quartz window, and the experimental scheme is shown in SI Appendix, Fig. S31. The Raman shifts were calibrated to 520.5 cm⁻¹ with a silicon slice, being irradiated by a 785 nm excitation light source. To obtain the steady-state data, each spectrum was collected after 60 s electrochemical reaction at constant potentials. The exposure time per scan was set as 5 s under the static mode with 10 times accumulation.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.