Synthesis and Characterization of Ru–Cl–N SAC. Ruthenium chloride (RuCl3) molecules as the ruthenium and chlorine resource were initially incorporated into the cavity of ZIF-67 via solvothermal reaction (Scheme 1), followed by pyrolysis (according to the TG-DSC curve, Supplementary Fig. S1) and acid etching of cobalt nanoparticles and unstable Ru species. From the scanning electron microscope (SEM) and transmission electron microscope (TEM) images as shown in Fig. 1a, b, the pyrolyzed product Ru–Cl–N SAC inherits the polyhedral morphology of the as-synthesized CoRu-ZIF-67 (Supplementary Fig. S2). Besides, numerous nanopores could be observed in Ru–Cl–N SAC, which would facilitate the diffusion process of electrolyte and gas bubbles generated during the electrochemical reaction50. Selected area electron diffraction (SAED) image (Fig. 1c) derived from the marked circle in Fig. 1b indicates the formation of graphitic carbon with typical diffraction rings of (002) and (200) crystal lattice51. No Co-based nanoparticles, which was obviously present in the pyrolysis of Ru–Cl–N SAC product (defined as Co NP/ Ru–Cl–N SAC, Supplementary Fig. S3), were found during the TEM observation. Two broad peaks observed at 25.3 and 43.8º in the X-ray diffraction (XRD) pattern (Fig. 1d) can be assigned to the (100) and (002) plane of carbon substrate52, 53, which coincided with the TEM result. Raman spectrum (Fig. 1e) of Ru–Cl–N SAC shows D and G band located at 1181 and 1428 cm-1, respectively, with an ID/IG intensity ratio of 0.99, indicating the presence of rich defects existed in the carbon matrix35. The specific surface area and pore volume (Fig. 1f) are determined to be 253.02 m2 g-1 and 0.22 cm3 g-1, respectively. According to the pore size distribution analysis, most of the nanopores are centered at 1.9 and 3.3 nm. The TEM and XRD results implied that Ru may exist in the form of single atom state. To conform this point, aberration-corrected high-angle-annular-dark-field scanning transmission electron microscopy (AC-HAADF-STEM) was employed. A large number of bright dots depicted in Fig. 1g, which represent single atom, can be clearly seen. The elemental mapping displayed in Fig. 1h further verified the uniform distribution of Ru, Cl and N on the surface of carbon substrate.
To throw insight to the local coordination environment, chemical states and structure features within Ru–Cl–N SAC, highly sensitive X-ray adsorption fine spectroscopy (XAFS) experiments including the extended X-ray adsorption fine spectroscopy (EXAFS) were carried out29, 54, 55. Fig. 3a shows the Ru K-edge X-ray adsorption near-edge structure (XANES) spectrum of Ru–Cl–N SAC, Ru foil, RuCl3 and RuO2. The adsorption location of Ru–Cl–N SAC was slightly higher than that of Ru foil and lower than those of RuCl3 and RuO2, indicative of a chemical state between 0 and +3 (ref. 35). Impressively, the valley in the range of 22125-22148 eV originated from Ru-N/O was slightly blue-shifted, which may be attributed to the bonding with Cl. The coordination environment of Ru–Cl–N was further revealed by Fourier-transform (FT) k2-weighted EXAFS spectrum of Ru K edge. As shown in Fig.3b, the dominant coordination peak located at 1.97 Å for Ru–Cl–N SAC was assigned to Ru-Cl bond of the atomic Ru coordinated with Cl, which was almost overlapped with the peak of RuCl3. While the shoulder peak at 1.5 Å could be ascribed to Ru-N coordination of the atomic Ru bonded with N in the carbon substrate56. Besides, Ru nanoparticles were also observed with a peak located at 2.44 Å in reference to Ru foil. Wavelet transform (WT) was then performed to investigate the Ru K-edge EXAFS oscillations of Ru–Cl–N SAC30. The two dominant peaks were observed at around 6.24 and 8.55 Å-1 that can be assigned to atomic Ru-N bond and Ru-Cl bond, respectively. Combined with the AC-HAADF-STEM result, one can safely confirmed that the atomic Ru in Ru–Cl–N SAC was coordinated with Cl and N embedded in carbon substrate. Fitting the FT EXAFS spectrum of Ru–Cl–N SAC in R space revealed an average coordination number (CN) of 2.6 for Ru-Cl and 2.0 for Ru-N (Supplementary Fig. S4 and Table S1), leading to a chemical formula of RuCl2N2. To the best of our knowledge, this is the first report on Cl, N-coordinated Ru single atom electrocatalyst. Notably, when ruthenium chloride was replaced by ruthenium acetylacetonate, RuN4 SAC with Ru coordinated with four N atoms was obtained (Supplementary Fig. S5 and S6). On the other hand, the role of Co was critical in preventing the aggregation of Ru atoms, e.g., when we used ZIF-8 as the template, an isostructural polymer of ZIF-67 with Zn2+ as the metal node instead of Co2+, Ru nanoparticles was synthesized (Supplementary Fig. S7 and S8).
The electronic structure of Ru–Cl–N SAC was further investigated by X-ray photoelectron spectroscopy (XPS). The survey XPS pattern of Ru–Cl–N SAC displays the presence of Ru, Cl, N and C (Supplementary Fig. S9). Compared to the XPS survey of Co NP/Ru–Cl–N (Supplementary Fig. S10), the absence of Co signal in Ru–Cl–N SAC indicated the completely removing of Co NP by acid etching. The high-resolution spectrum of Ru 3p3/2 for Ru–Cl–N SAC can be deconvoluted into two peaks at 462.2 and 464.5 eV, which could be ascribed to metallic Ru and oxidized Ru species57, respectively. Meanwhile the Ru 2p1/2 peaks at 484.2 and 487.2 eV were assigned to the same species. The single peak for Cl 2p spectrum was observed at 199.4 eV, corresponding to Ru-Cl bond49. Besides, the N 1s spectrum can be deconvoluted into several characteristic peaks. The XPS peaks located at 398.4, 399.1, 400.7, 401.4 and 403.1 eV could be ascribed to pyridinic N, Ru-N, graphitic N, oxidized N and adsorbed N species56, respectively. The XPS results further confirmed the coordination environment of Ru that bonded with Cl and N.
Electrocatalytic Performance of Ru–Cl–N SAC. The multifunctional electrocatalytic performance of Ru–Cl–N SAC for HER, OER and ORR was evaluated. For comparison, RuN4 SAC, NC and commercial Pt/C or IrO2 were also tested as reference samples. To get the accurate potentials, the reference electrode was calibrated against RHE in H2-saturated KOH electrolyte (Supplementary Fig. S11 and S12)57. HER and OER measurements were performed in a standard three-electrode electrochemical workstation in 1 M N2-saturated KOH solution. As shown in Fig. 3a, the overall HER activity of Ru–Cl–N SAC surpassed all the samples. The overpotential (η10) to reach a current density of 10 mA cm-2, which represents the reasonable descriptor of the catalytic activity58, was merely 12 mV for Ru–Cl–N SAC, much lower than that of Pt/C (29 mV), RuN4 SAC (131 mV) and NC (248 mV). The reaction kinetics was reflected by the Tafel plot, which was derived from the linear scan voltammetry (LSV) curve8. The Tafel slope of Ru–Cl–N SAC (Fig. 3b) was fitted to be 23.9 mV dec-1, indicative a fast Volmer–Tafel pathway24. While the larger Tafel slope for Pt/C (47.8 mV dec-1), RuN4 SAC (130.5 mV dec-1) and NC (287.9 mV dec-1) revealed a sluggish Volmer–Heyrovsky route. Moreover, the turnover frequency (TOF) was estimated to evaluate the inherent electrocatalytic efficiency by means of underpotential deposition (UPD) of copper (Supplementary Fig. S13)38, 59. The calculated TOF for Ru–Cl–N SAC was 0.66 H2 s-1 at overpotential of 25 mV, which was two times larger than that of Pt/C (0.32 H2 s-1), and comparable to many recently-reported HER catalysts (Supplementary Fig. S14 and Table S2). The charge transfer kinetics was evaluated by the electrochemical impedance spectrum (EIS) measurement. From the Nyquist plot displayed in Fig. 3c, the charge transfer resistance (Rct) value was recorded as 8.72, 22.79, 26.90 and 197.3 Ω for Ru–Cl–N SAC, Pt/C, RuN4 SAC and NC, respectively, demonstrating that Ru–Cl–N SAC underwent favorable charge transfer process during the HER reaction60. To investigate the intrinsic activity of the as-synthesized catalyst, double-layer capacitance (Cdl) was obtained by cycling voltammetry (CV) measurement (Supplementary Fig. S15), which was proportional to the electrochemical active surface area (ECSA).8 The results showed that Ru–Cl–N SAC possessed the largest Cdl (Fig. 3d), unveiling more active reaction sites in Ru–Cl–N SAC. It should be noted that the HER performance of Ru–Cl–N SAC was superior to the benchmark Pt/C, and even better than many other recently-reported Ru-based HER catalysts (Supplementary Table S3). Furthermore, the Ru–Cl–N SAC exhibited dramatical stability. From the long-term current-time curve (Fig. 3e), the activity of Ru–Cl–N SAC could maintain stable for at least 120 h.
The OER activity of Ru–Cl–N SAC was then carried out in the same condition. From the LSV curves depicted in Fig. 3f, the η10 value for Ru–Cl–N SAC (233 mV) was much lower than that of RuN4 SAC (362 mV) and IrO2 (347 mV). The corresponding Tafel slope of Ru–Cl–N SAC disclosed a favorable reaction kinetics compared to other reference samples (Supplementary Fig. S16). Besides, the TOF was estimated to be 0.271 O2 s-1 at overpotential of 300 mV, which is very competitive compared to many other reported OER catalysts (Supplementary Fig. S17 and Table S4). Furthermore, the Rct value (Supplementary Fig. S18) for Ru–Cl–N SAC (21.68 Ω) measured at 10 mA cm-2 disclosed a faster electron transfer process, compared to RuN4 SAC (129.8 Ω), IrO2 (37.37 Ω) and NC (245.2 Ω). The intrinsic activity was then evaluated by Cdl (Supplementary Fig. S19). The largest Cdl value of Ru–Cl–N SAC among all the tested samples demonstrated the atomically dispersion of Ru–Cl–N SAC boosts more active reaction sites. The above results proved that the OER activity of Ru–Cl–N SAC outperformed the commercial IrO2 and most reported Ru-based electrocatalysts in alkaline medium (Supplementary Table S5). The stability test suggested the Ru–Cl–N SAC could endure the long-term chronoamperometry for more than 120 h (Fig. 3e).
The ORR performance of Ru–Cl–N SAC was characterized by rotating ring-disk electrode (RRDE) method in 0.1 M KOH. Supplementary Fig. S20 compared the CV curves of Ru–Cl–N SAC operated in O2-saturated and Ar-saturated solution, displaying distinct ORR cathodic peak and demonstrating the occurrence of oxygen electroreduction at the electrode interface61. Fig. 3g displayed the ORR polarized curves of all the catalysts with a rotating speed of 1600 rpm. Apparently, the Ru–Cl–N SAC exhibited remarkable ORR activity with the highest onset potential (Eonset = 1.10 V), strongest limited current density (jL = 6.93 mA cm-2) and largest half-wave potential (E1/2 = 0.90 V). Notably, the E1/2 of Ru–Cl–N SAC was 70 mV larger than that of the benchmark Pt/C catalyst, which was comparable or even better than many other reported ORR catalysts (Supplementary Table S6). Furthermore, the Tafel slope of Ru–Cl–N SAC was also smaller than other reference samples (Supplementary Fig. S21). The calculated number of electrons transferred per O2 molecule on the Ru–Cl–N SAC electrode was close to 4.0 (Supplementary Fig. S22), indicating the ORR process followed the most efficient electron pathway62. The H2O2 generation yield was recorded lower than 1.9%, confirming the high efficiency and selectivity of Ru–Cl–N SAC toward ORR. The stability measurement disclosed that the Ru–Cl–N SAC could continuously work for at least 20 h without apparent decay (Fig. 3h), much superior to the commercial Pt/C. Moreover, the Ru–Cl–N SAC exhibited robust tolerance of poisoning by methanol. Taking overall consideration of the performance of the present Ru–Cl–N SAC for HER, OER and ORR, this newly-synthesized multifunctional catalyst exhibited outstanding activities with favorable η10 value (for HER and OER) and E1/2 (for ORR) among most recently-reported trifunctional electrocatalysts as shown in Fig. 3i and Supplementary Table S7.
Theoretical analysis of electrocatalytic mechanism for Ru–Cl–N SAC. To unveil the intrinsic active species for the newly-synthesized catalyst of Ru–Cl–N SAC, control experiments as well as DFT calculations were further performed. Firstly, LSV curves of Co NP/Ru–Cl–N SAC displayed slightly inferior catalytic activities for HER, OER and ORR compared with Ru–Cl–N SAC (Supplementary Fig. S23). DFT calculation indicated that the formation energy of O2 molecule on the (002) crystal surface of Co (Supplementary Fig. S24-S26) is too high and thus the OER process is indeed difficult due to the large overpotential (4.86 eV). On the other hand, the adsorption free energy of H on the Co (002) plane is unfavorable for HER process (Supplementary Fig. S27). Therefore, Co NP was unable to promote the electrocatalytic reaction and deserved to be removed. Secondly, to uncover the role of the residue Ru NP within Ru–Cl–N SAC, KSCN and EDTA were used as poisoning agents to discover the contribution of Ru NP and single atomic Ru to the electrocatalytic reaction. It has been established that EDTA mainly coordinates with single atomic Ru, while KSCN could cover both Ru NP and single atom24. The polarized curves (Supplementary Fig. S28) displayed similar poisoning effect of EDTA and KSCN for Ru–Cl–N SAC, elucidating that the single atom Ru species was the real intrinsic active sites.
To clearly show the effect of the coordination atom on the electrocatalytic activity, the RuCl2N2/C configuration of a single Ru atom bonding with two Cl atoms and two N atoms in the carbon substrate was constructed, according to the XAFS and TEM results, to simulate the catalytic reactions of HER, OER and ORR. Moreover, a RuN4/C model comprising a single Ru atom bonded with four N atoms was built for comparative analysis (Fig. 4a, b). The Hirshfeld charge partitioning analysis revealed that the introduction of the Cl atom effectively improves the catalytic activity of Ru62. The amount of charge transfer is about 0.35 e (e is the unit of positive charge) from Cl to Ru atom (Fig. 4c and S29). To gain a deeper insight into the interaction of electronic structure, Fig. 4d presents the partial density of states (PDOS) of RuCl2N2/C and RuN4/C56. The large overlap between the Ru 3d states and Cl 2p states at the energy range from -4.5 to 2.0 eV confirmed the strong interaction between Cl and Ru atom. Compared with the Ru 3d state peak of RuN4/C, the Ru 3d state peak of RuCl2N2/C near the Fermi level splits a part to above the Fermi level, so that Ru is more conducive to transfer of electrons to promote the catalytic reaction. The HER activity on RuCl2N2/C and RuN4/C at different sites was also calculated as shown in Fig. 4e. The optimal value of ΔGH* is zero, which balances the adsorption and desorption capacity and represents high catalytic activity8. The most stable free energy of H adsorption is 0.10 eV on RuCl2N2/C and -0.80 eV on RuN4/C, respectively. The optimal adsorption sites of the oxidized intermediates were tested on the RuCl2N2/C and RuN4/C systems. The results showed that oxygen-containing intermediates tend to be adsorbed on Ru sites, indicating that Ru is the active site. The adsorption free energies of oxygen-containing species uncovered the origin of catalytic activities50. The OER and ORR catalytic performance on RuCl2N2/C and RuN4/C was then evaluated by the change in free energies of each reaction step. It can be seen from Fig. 4f that RuCl2N2/C and RuN4/C have the same rate-limiting steps. Due to the change of the coordination environment from N to Cl atom, the overpotentials of OER and ORR are reduced from 1.74 and 1.78 V to 1.07 and 1.06 V, respectively. In addition, the overpotentials of OER and ORR are basically the same on RuCl2N2/C (1.07 V vs. 1.06 V), which indicated that RuCl2N2/C has more potential as a dual-function catalyst for OER and ORR. On the whole, the DFT calculation results indicated that the catalytic activities for HER, OER and ORR on RuCl2N2/C are favorable than those on RuN4/C. The present work validated the pivotal role of Cl atom within Ru–Cl–N SAC in boosting the electrocatalytic activity.
Overall water splitting and Zn-air battery. Inspired by the outstanding HER and OER activities of Ru–Cl–N SAC, the overall water splitting performance was further investigated8, 50. As shown in Fig.5a, the homemade alkaline electrolyzer with two electrode construction was fabricated by using Ru–Cl–N SAC supported on hydrophilic carbon paper both as cathode and anode. The Ru–Cl–N SACǁRu–Cl–N SAC cell only required a voltage of 1.49 V to deliver a current density of 10 mA cm-2, much lower than the cells (1.59 V) assembled by commercial Pt/C and IrO2 (Fig. 5b), which is also superior to many recently-reported bifunctional electrocatalysts (Supplementary Table S8). The chronoamperometry measurement disclosed that the Ru–Cl–N SACǁRu–Cl–N SAC cell could keep stable at 10 mA cm-2 for at least 100 h without obvious degradation (inset of Fig. 5b) and numerous gaseous bubbles could be observed on both electrodes (Supplementary Fig. S30).
Benefiting from the excellent ORR and OER behavior of Ru–Cl–N SAC with low potential difference (∆E = E1/2 – EOERη10) of 0.56 V, a rechargeable liquid Zn-air battery was assembled by using Ru–Cl–N SAC as cathode in 6 M KOH solution containing 0.2 M zinc acetate (Fig.5c)4, 62, and the mixture of Pt/C and IrO2 with a mass ratio of 1:1 was also tested for comparison. Supplementary Fig. S31 revealed that the Ru–Cl–N SAC-based battery possessed a stable open-circuit voltage (OCV) of 1.455 V, slightly higher than that of Pt/C-IrO2 counterpart (1.452 V). Furthermore, the battery assembled by Ru–Cl–N SAC delivered a maximum power density of 205 mW cm-2 at 325 mA cm-2 (Fig. 5d), much higher than that of Pt/C-IrO2 equipped battery (87 mW cm-2 at 437 mA cm-2). Besides, the specific capacity of Ru–Cl–N SAC-based battery hold a value of 804.26 mAh g-1, corresponding to a gravimetric energy density of 981.20 Wh kg-1 at 20 mA cm-2 (Fig. 5e), which is much better than those of Pt/C-IrO2 equipped battery (736.61 mAh g-1 and 832.37 Wh kg-1) and comparable to many other reported Zn-air batteries (Supplementary Table S9). An LED light array could be continuously lit by single Ru–Cl–N SAC-assembled battery for more than 20 h without brightness decay (Fig. 5f and Supplementary Video S1). The cycling stability of the air cathode was then evaluated as displayed in Fig. 5g. Unlike Pt/C-IrO2 pair, where dramatical voltage degradation occurred after tens of cycles, tiny voltage decay was observed in Ru–Cl–N SAC-assembled battery after 1080 cycles (360 h) with a round-trip efficiency of 64%, indicating the robust durability of Ru–Cl–N SAC. In addition, the multifunctional performance of Ru–Cl–N SAC enabled the integration of Zn-air battery and overall water electrolyzer cell. As displayed in Fig. 5h and Supplementary Video S2, a water splitting cell could be powered by two Zn-air batteries in series fabricated by using Ru–Cl–N SAC as the “all-in-one” catalytic active material.