Synthesis and characterization of the asymmetric Ru–N–In single-atom pairsThe Ru–N–In/NC catalyst was synthesized via an atomic interface regulation strategy as illustrated in Supplementary Fig. 138. Typically, nitrogen-doped carbon, ruthenium acetylacetonate (Ru(acac)3), and MIL-68(In) were ground thoroughly to give a uniform mixture, which was then subjected to calcination under an Ar atmosphere. The single-metal counterparts of RuN4/NC and InN4/NC were prepared using the same procedures but without MIL-68(In) or Ru(acac)3, respectively. Herein, the nitrogen-doped carbon supports for anchoring metals were obtained through polyaniline (PAN) pyrolysis at 800 °C (Supplementary Figs. 2–4). Inductively coupled plasma optical emission spectrometry (ICP-OES) determined that the contents of Ru and In were ~0.98 and 0.19 wt%, respectively, which were close to the nominal loadings (Supplementary Table 1).The transmission electron microscopy (TEM) image presents that the Ru–N–In/NC moiety possessed a hexagonal-like shape composed of randomly arranged nanosheets (Fig. 2a), similar to the morphology of NC (Supplementary Fig. 5), RuN4/NC (Supplementary Fig. 6), and InN4/NC (Supplementary Fig. 7). The selected area electron diffraction (SAED) pattern suggests the absence of metallic nanoparticles and low crystallinity of the carbon matrix with abundant carbon defects39. The C/N ratio derived from CHNS analysis was 6.52 for Ru–N–In/NC, lower than that of NC (6.73), demonstrating that Ru–N–In/NC had more carbon vacancies and unsaturated nitrogen sites (Supplementary Fig. 10)40. Energy dispersive spectroscopy (EDS) mappings confirmed that Ru, In, and N were uniformly distributed on the carbon matrix (Fig. 2b). Powder X-ray diffraction (XRD) patterns show all the samples had two diffraction peaks of (002) and (100) facets (Fig. 2c)41. The aberration-corrected high-angle annular dark-field scanning TEM (AC-HAADF-STEM) image exhibits that Ru and In atoms were well-dispersed as atomic pairs throughout the whole matrix, in which dimeric white bright spots (marked by yellow circles/boxes) were identified as Ru-N-In/NC moieties (Fig. 2d). The average distance of the two neighboring metal sites of Ru and In was determined as ~0.37 nm (Fig. 2e and Supplementary Figs. 6 and 7). The atomically dispersed Ru or In on RuN4/NC or InN4/NC was also confirmed (Supplementary Figs. 8 and 9). Subsequently, we compared the thermodynamic formation energies for six single/dimer-atom catalysts including RuN4/NC, InN4/NC, Ru2/NC, In2/NC, Ru–N–In/NC, etc., which revealed that the asymmetric Rn–N–In dual-atom structure had the lowest free energy of –2.793 eV (Supplementary Fig. 11). Additionally, N2 adsorption–desorption curves indicate that Ru-N-In/NC had a high surface area of 690.13 m2/g (Supplementary Fig. 12 and Supplementary Table 2). Fourier-transform infrared spectroscopy (FTIR) (Supplementary Fig. 13) and X-ray photoelectron spectroscopy (XPS) (Supplementary Figs. 14–16) also further proved the formation of Ru–In coordination.Fig. 2: The synthesis and structural characterization of the Ru–N–In/NC catalyst.a and b TEM images of the Ru–N–In/NC moiety and the corresponding EDS mappings for Ru, In, and N, respectively. c Powder XRD patterns of Ru–N–In/NC, RuN4/NC, InN4/NC, and NC. d AC-HAADF-STEM image of the Ru–N–In/NC sample, in which the single-atom Ru–In pairs are highlighted by the yellow circles/boxes. e Atomic scale identified the distance (nm) of isolated Ru and In dual single-atom sites in the selected area with yellow rectangles 1, 2, and 3 in figure (d).The Ru oxidation state in Ru–N–In/NC was resolved by X-ray absorption near edge spectroscopy (XANES) (Fig. 3a). The Ru K edge spectrum exhibits a pre-edge transition at 22112 eV, which falls between the peaks associated with RuCl3 (22110 eV) and RuO2 (22114 eV). This indicated an intermediate oxidation state between +3 and +4 for Ru in Ru–N–In/NC. The coordination structure of Ru single atom was further revealed by the extended X-ray absorption fine structure (EXAFS) analysis. Figure 3b compares Fourier-transformed EXAFS spectra for Ru–N–In/NC, RuN4/NC, RuO2, and Ru foil. Ru–N–In/NC exhibits first-shell scattering at 1.48 Å in R space (prior to phase correction), which is proximate to the value of 1.52 Å, found for RuO2. This was distinct from the case for Ru foil, the first shell scattering of which is located at 2.30 Å. From these observations, we assigned the primary feature at 1.48 Å to be Ru–N bonding42. The atomic dispersion of Ru was also evidenced by the wavelet transform (WT) plot, with the intensity at ≈4.5 and 9.5 Å−1 arising from Ru–O and Ru–Ru scattering, respectively (Fig. 3g). The quantitative structural parameter analysis based on the fitted EXAFS spectra suggested that each Ru atom coordinates with three N atoms in the first shell within the inter-plane of the tri-striazine framework, with an average bond length of 1.99 Å (Fig. 3c and Supplementary Tables 3 and 4). The atomic dispersion of Ru over the control RuN4/NC was also investigated, where each Ru atom was coordinated with four N atoms (Supplementary Figs. 17–20 and Supplementary Tables 3).Fig. 3: Chemical state and atomic structure of Ru and In in Ru–N–In/NC.Ru K-edge XAFS analysis of Ru–N–In/NC and RuN4/NC: a Normalized XANES, b Fourier-transform XAFS spectra, and c XAFS curves fitting in R space. In K-edge XAFS analysis of Ru–N–In/NC: d Normalized XANES, e Fourier transform XAFS spectra, and f EXAFS curves fitting in R space. g Wavelet transform contour plots of Ru–N–In/NC at Ru K-edge and In K-edge.The coordination environments and local structures of In centers were simultaneously determined by the In K-edge XANES curves of Ru–N–In/NC, In foil, and In2O3 (Fig. 3d). The absorption edge of Ru–N–In/NC was centered at a higher energy than that of In2O3, indicating that In atoms exist at a higher valence state than in In2O3, which is in agreement with the result of In 3d XPS (Supplementary Fig. 14). Evidenced by the Fourier-transformed EXAFS spectra, unlike the In–In (≈2.96 Å) and In–O (≈1.67 Å) peaks observed in In foil and In2O3, Ru–N–In/NC exhibits a dominant peak of first-shell scattering (assigned to In–N scattering) located at ≈1.56 Å (phase-uncorrected distance) (Fig. 3e). Figure 3f presents the fitting results of the extended In K-edge, which reveals that each In atom is coordinated with four N atoms with an average In-N bond length of ≈2.15 Å (Supplementary Figs. 21–24 and Supplementary Tables 3 and 4). Moreover, the WT plots of In K-edge-weighted EXAFS indicate the atomically dispersed In species on NC supports (Fig. 3g). In addition, combining the results of DFT and X-ray absorption spectra (XAS), we further consider Ru–N–Ru/In and In–N–In/Ru as the comprehensive models and fit the raw XAS data (Fig. 3c and f). The optimal results in Supplementary Table 4 demonstrate that Ru and In are dispersed as atomic pairs on the NC supports, with a higher possibility of the presence of Ru–N–In heterostructure due to the lowest free formation energy of –2.793 eV compared to other counterparts (Supplementary Fig. 11).The partial density of states (PDOS) results were compared for Ru–N–In/NC and RuN4/NC to obtain the orbital hybridization information induced by the electronic structures of asymmetric coordination. As illustrated in Supplementary Fig. 25, the energy of the In–(p) band matches well with that of the Ru–(d) band relative to the Fermi level, indicating that a strong d–p orbital hybridization interaction occurred between Ru and In atoms. Such a d–p hybridization interaction leads to lower orbital energy caused by enhanced electron delocalization and electron redistribution. In addition, charge density analysis provided more evidence for the Ru centers in Ru–N–In/NC moiety that exhibited significant electronic delocalization accompanied by the asymmetric electronic redistribution (Supplementary Fig. 26). Compared with RuN4/NC (–0.92 eV), the reduced Bader charge of the Ru in Ru–N–In/NC (–0.98 eV) indicated that partial electrons from the Ru centers were extracted after In introduction. The reduced electron density is also proved by the up-shifted binding energy (~0.3 eV) of Ru 3p in Ru–N–In/NC compared to that in RuN4/NC according to the XPS characterization (Supplementary Fig. 15).From the above discussion, the distinct atomic structures of N-bridged Ru, In dual-atom Ru–N–In/NC catalyst have been clearly unveiled. We hypothesize that the electronic chemical configuration derived by d(Ru)–p(In) hybridization as compared to RuN4/NC would lend them exquisite catalytic performance for acetylene hydrochlorination.Enhanced catalytic performance for acetylene hydrochlorinationThe catalytic performances of the Ru–N–In/NC catalysts were evaluated in a continuous flow fixed-bed microreactor (Supplementary Fig. 27). The optimal reaction temperature, Ru/In ratio, gas hourly space velocity (GHSV), and carrier N-content were determined as 180 °C, 5, 180 h−1, and 5.86 wt%, respectively (Supplementary Figs. 28–32). Compared with the lower acetylene conversions of RuN4/NC (~76.35%), InN4/NC (~18.15%), and NC (~15.69%), Ru–N–In/NC showed an excellent initial activity of ~99.51% with negligible deactivation (Fig. 4a). Based on Fig. 4a, it can be concluded that Ru was the main active center. The introduction of In atoms apparently played a crucial role in stabilizing Ru, as evidenced by the deactivation rate of RuN4/NC (~4.78%) compared to no discernible drop of acetylene conversion found for Ru–N–In/NC. Notably, the VCM selectivity for these catalysts was more than 99% except for InN4/NC (Supplementary Fig. 33). The main by-product, which may originate from the oxidation of HCl or further hydrochlorination of vinyl chloride, was identified as dichloroethane (1,1-dichloroethane and 1,2-dichloroethane) through a gas chromatograph (Supplementary Fig. 34)6,43. To investigate the intrinsic reactivity of the aforementioned catalysts, the kinetics study was evaluated by considering the Weisz–Prater criterion and the Mears criterion to avoid the interference of diffusion (Supplementary Texts 1 and 2)44. The lower apparent activation energy (Ea) (46.89 kJ/mol) (Supplementary Fig. 35) and Ru loss ratio (0.2%) (Supplementary Table 1) of Ru–N–In/NC compared to 65.02 and 79.31 kJ/mol for RuN4/NC and InN4/NC contribute to its higher catalytic activity. Further, the reaction orders of HCl and C2H2 were derived from kinetic studies, and a higher reaction order for HCl (0.67–0.92) was obtained in comparison with C2H2 (0.65–0.83) (Supplementary Fig. 36).Fig. 4: The catalytic performance for acetylene hydrochlorination.a C2H2 conversions of Ru–N–In/NC, RuN4/NC, InN4/NC and NC [Reaction conditions: T = 180 °C, P = ambient pressure, GHSV(C2H2) = 180 h−1, V(HCl)/V(C2H2) = 1.15]. The error bars indicate the standard deviations of three experimental measurements. b Comparison of TOF (molC2H2/molmetal/h) of Ru–N–In/NC with other Ru-based, non-noble metal and non-metal catalysts6,7,59,60,61,62,63,64,65,66,67. Note that all the obtained data are the same as our reaction conditions. The data are calculated at 180 °C, ~5% C2H2 conversion to eliminate the influence of internal and external diffusion. Each point was determined by an isolated test to eliminate the interference of catalyst deactivation. c Long-term catalytic performances of Ru–N–In/NC and RuN4/NC, and corresponding AC HAADF-STEM images for the used samples. [Reaction conditions: T = 180 °C, P = ambient pressure, Vcat = 1.2 mL, V(HCl)/V(C2H2) = 1.15, GHSV(C2H2) = 180 h−1.] d Comparison of long-term stability (TOS and deactivation rate) for Ru–N–In/NC with other recently reported catalysts.Subsequently, we compared the turnover frequency (TOF, molC2H2/molmetal/h) of Ru–N–In/NC with the recently reported catalysts (Supplementary Table 5). Intriguingly, the activity of Ru–N–In/NC was not only far exceeded by those of other Ru-based catalysts (Fig. 4b), but also higher than that of several Au- and Pt-based samples (Supplementary Fig. 37). Moreover, the successful synthesis and application strategy of Ru–N–In/NC is generally applicable to other noble metals (i.e., AuIn/NC, PdIn/NC, and PtIn/NC), which exhibit promising reactivities for acetylene hydrochlorination. Herein, the Ru–N–In/NC still has better activity compared to other catalysts (Supplementary Fig. 38). In addition, the activity of Ru–N–In/NC also outperforms that of RuIn/AC (activated carbon (AC) is the common support for commercial catalysts) (Supplementary Fig. 39) and commercial HgCl2/AC moieties (Supplementary Fig. 40).The long-term stability of Ru–N–In/NC and RuN4/NC was subsequently evaluated at the GHSV of 180 h-1, where Ru–N–In/NC can maintain stable catalysis with an admirable VCM productivity of ~0.98 kgvcm/kgcatal./h for more than 600 h (Fig. 4c). Such stable catalysis for Ru–N–In/NC in terms of time on stream (TOS, 600 h) and deactivation rate (0.001%/h) in the present study was apparently superior to other recently reported catalysis systems (Fig. 4d). However, the completed deactivation was observed for RuN4/NC within 120 h (Fig. 4c and Supplementary Fig. 41). From the perspective of practical application, the concomitant CO2 (~250 ppm) was pulsed to evaluate the tolerance of Ru–N–In/NC, in which the VCM productivity was attenuated to ~0.76 kgvcm/kgcatal./h due to the competitive adsorption between CO2 and the reaction gases, however, the stability was still kept at 600 h (Supplementary Fig. 42)45. We also synthesized and tested the stability of RuIn/AC, which resulted in decreased activity (~0.81 kgvcm/kgcatal./h) and poor stability (450 h) (Supplementary Fig. 43). The results of XRD (Supplementary Fig. 44) and XPS (Supplementary Fig. 45) indicated that the structure of post-hydrochlorination Ru-N-In/NC was perfectly intact. TEM (Supplementary Fig. 46) and AC-HAADF-STEM (Fig. 4c) images show that the used sample retained the original morphology of randomly stacked nanosheets and, most importantly, the atomically dispersed Ru–In pairs.Mechanisms of acetylene hydrochlorination on asymmetric Ru–N–In/NCPrior to the mechanism investigations, the adsorption of C2H2 or HCl on asymmetric Ru–N–In/NC was studied. Temperature-programmed-desorption (TPD) experiments, in which all the desorption peaks are located at ~205 °C (Fig. 5a), were carried out to determine the adsorption sites for acetylene molecules. Compared with RuN4/NC (2329.7), the higher desorption peak area for InN4/NC (6653.7) proved that In atoms were more favorable for capturing C2H2 (Supplementary Table 6). As shown in Fig. 5b and Supplementary Fig. 47, In-situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) for C2H2 adsorption also offers a similar conclusion, in which the νas(C2H2) of 3258 cm−1 band underwent a downward shift with reference to the gas-phase value at 3287 cm−1, indicating that the C2H2 molecule was subjected to strong chemisorption of In–CHCH due to its significant bond polarization32,46. The –C ≡ C– bond cleavage into –C = C– bond of C2H2 upon adsorption on In species, forms p-π-bonded HC = CH species47. The integrated crystal orbital Hamilton population (ICOHP) was carried out to analyze the adsorption of C2H2 on Ru–N–In/NC and RuN4/NC (Supplementary Fig. 48)48,49. The up ICOHP in Ru–N–In/NC decreases from –1.58 to –2.24 compared with RuN4/NC, and the down ICOHP decreases from –1.76 to –2.26, demonstrating the stronger adsorption strength of the C2H2 molecules on Ru–N–In/NC. Furthermore, the C2H2 adsorption energies (Eads (C2H2)) on Ru–N–In/NC, RuN4/NC, and InN4/NC were calculated as –0.94, –0.78 and –1.18 eV, which aligned with the trend found in TPD (Supplementary Figs. 49 and 50). The modest C2H2 adsorption strength allowed the optimal interaction of C2H2 with Ru–N–In/NC (R2 = 0.87), which is required to trigger the reaction (Supplementary Fig. 51) compared to the single atom counterparts.Fig. 5: The study of the catalytic mechanism for acetylene hydrochlorination.a C2H2-TPD curves of the Ru–N–In/NC, RuN4/NC, and InN4/NC samples. b In-situ DRIFTS of C2H2 adsorption profiles over the Ru–N–In/NC surfaces. c HCl-TPD curves of the Ru–N–In/NC, RuN4/NC, and InN4/NC. d Comparison of adsorption energy for HCl molecule on Ru and In centers. e The adsorption energy of C2H2/HCl versus TOF of the Ru–N–In/NC, RuN4/NC, and InN4/NC. The error bars indicate the standard deviations of three experimental measurements. f d-band center as a descriptor versus the TOF of the Ru–N–In/NC, RuN4/NC, and InN4/NC, in which the projected density of states analysis of Ru–N–In/NC was inserted. The error bars indicate the standard deviations of three experimental measurements. g DFT calculations of the acetylene hydrochlorination reaction pathways over the Ru–N–In/NC and RuN4/NC moieties. Cyan, orange, green, blue, gray, and white spheres represent Ru, In, Cl, N, C, and H atoms, respectively.The HCl-TPD curves show that the desorption temperature of the three samples is located at ~170 °C. A remarkable decline of the desorption peak area for Ru–N–In/NC (4467.3) was observed compared to RuN4/NC (5474.3), but still higher than that of InN4/NC (2638.7), indicating that the addition of In atoms weakened the adsorption of HCl on the Ru site in the case of Ru–N–In/NC (Fig. 5c and Supplementary Table 6). Further, we compared the adsorption energy of Ru and In over Ru–N–In/NC. Lower adsorption energy (–0.63 eV) over Ru sites was obtained, in which HCl was adsorbed via the interaction of Ru···Cl–H between Cl (Lewis base) and Ru (Lewis acid) (Fig. 5d and Supplementary Fig. 52). In contrast, stronger HCl adsorption over RuN4/NC (Eads (HCl) = –0.83 eV) was observed, demonstrating a significant over-chlorination potential for Ru centers (Supplementary Fig. 53). In summary, the dimeric Ru-In geometry with proper distance (~0.37 nm) enables the independent adsorption configuration for C2H2 and HCl and mediates the steric hindrance for that, accounting for the enhanced activities compared to RuN4/NC.Subsequently, we investigated the structure-activity relationship for the Ru–N–In/NC dual-atom catalyst. Generally, the catalyst with higher adsorption energy for the reactants has higher catalytic activity. However, as depicted in Fig. 5e, Ru–N–In/NC owns the medium adsorption strengths for reactants, giving rise to the highest level of reactivity50. Next, an attempt was undertaken to correlate the d-band center with TOF, which exhibited a remarkable linear relationship (R2 = 0.91) (Fig. 5f and Supplementary Fig. 54). Hence, it could be understood that the strong d(Ru)–p(In) interaction results in the up-shift of d-band center of Ru to Fermi level, forming active sites with unique electron configurations for the appropriate adsorption strengths of reactants on the atomic Ru and In sites that improve the intrinsic activity of Ru–N–In/NC51.Figure 5g schemes the Gibbs free energy diagrams of acetylene hydrochlorination over the RuN4/NC and Ru–N–In/NC catalysts based on the first principle, and this process follows the Langmuir-Hinshelwood co-adsorption mechanism. Differently, C2H2 and HCl are simultaneously adsorbed on the Ru atom of RuN4/NC, whereas C2H2 and HCl molecules are separately adsorbed on the In and Ru atoms of Ru–N–In/NC, eventually forming the co-adsorbed (C2H2 + HCl)ads state. Subsequently, the HCl molecule begins to migrate, and the H–Cl bond is continuously elongated and broken, causing the *H atom to be endothermically added to *C2H2 to form *CH2 = CH. Due to the diminished steric hindrance for the Ru–N–In/NC in comparison with RuN4/NC, this process (energy barrier = 0.54 eV) serves as the new rate-determined step (RDS) for Ru–N–In/NC moiety. In the following step, the Cl* atom approaches another C atom of *C2H2, and then coordinates to *CH2 = CH to generate *CH2CHCl over Ru centers. For the RuN4/NC moiety, this endothermal process can be regarded as the RDS, while the Ru–N–In/NC decreases the energy barrier of *Cl addition from 0.61 to 0.21 eV. Overall, the cooperation of In atoms changes the RDS of acetylene hydrochlorination by regulating the steric hindrance effect of atomic active sites, skipping the conventional barrier limitation of *Cl addition for the RuN4/NC surfaces. Eventually, the reaction proceeds at a lower free energy on Ru-N-In/NC with the pathway of (C2H2 + HCl)ads → *C2H2 + *HCl → *CH2 = CH → *CH2CHCl → CH2CHCl.Intrinsic understanding of the high stability of asymmetric Ru–N–In/NCTo unlock the intrinsic mechanism of the promising stability of Ru–N–In/NC, we carefully investigated the chlorination processes of Ru–N–In/NC and RuN4/NC. At the beginning, the dissociation energy of Ru was calculated as 1.86 and 1.27 eV for Ru–N–In/NC and RuN4/NC, respectively, suggesting a higher binding strength between single atom Ru and the substrate in the case of a dual-atom catalyst (Fig. 6a). Then, in-situ DRIFTS NH3 adsorption characterization was carried out to further determine Lewis acidity. The peak of 1591 cm−1 attributed to the NH3 adsorption on Lewis acid sites, in which the intensity of this peak for Ru–N–In/NC was remarkably faded, demonstrating that the Lewis acidity of Ru–N–In/NC was weakened compared to RuN4/NC (Supplementary Fig. 55), which could potentially prevent over-chlorination on the atomic Ru site. Further, scanning electron microscopy (SEM) and EDS images of the used Ru–N–In/NC and RuN4/NC samples indicate that the enrichment of Cl* on the catalyst surface is noticeably reduced from 14.41 to 4.08 wt% (Supplementary Fig. 56).Fig. 6: Stabilization of the Ru center by decreasing its chlorination degree.a Ru dissociation energy of Ru–N–In/NC and RuN4/NC moieties. Gray, blue, green, orange, and pink balls represent C, N, Cl, Ru, and In atoms, respectively. b The normalized Ru K-edge XAFS plots of the post-hydrochlorination samples of Ru–N–In/NC-3h, RuN4/NC-3h, and RuN4/NC-1h, in which the Ru foil, RuCl3, and RuO2 spectra serve as references. c Projected density of states of Ru p orbitals and Cl* d orbitals after Cl* adsorption over the interface of RuN4/NC and Ru–N–In/NC. σ and σ* represent the bonding and antibonding between \({d}_{z}^{2}\) orbital of Ru and p orbital of Cl, π1 and π1* represent the bonding and antibonding between dyz/dxz orbital of Ru and p orbital of Cl, π2 represents the bonding between the dx2–y2 orbital of Ru and p orbital of Cl. d The formation energy of various Ru chlorinated species and the corresponding illustration of the evolution pathway of Ru species over the Ru–N–In/NC and RuN4/NC structures. Gray, pink, purple, yellow, and green balls represent C, N, Ru, In, and Cl atoms, respectively. e Coke deposition of the post-hydrochlorination Ru-based catalysts, determined by the weight loss differences. The error bars indicate the standard deviations of three experimental measurements.Next, we utilized Ru K-edge XANES to track the chemical and geometric states of Ru after chlorination for three hours for Ru–N–In/NC and RuN4/NC. The pre-edge transition of Ru–N–In/NC-3h is between RuCl3 and RuO2 references, suggesting that the Ru valence states were almost the same (between +3 and +4) before and after the reaction. In contrast, the white-line intensity of chlorinated RuN4/NC decreased with the increasing exposure time, indicating that the Ru chemical state gradually reduced (Fig. 6b). The FT k3-weighted χ(k)-function and structural parameters analysis based on the fitting of EXAFS spectra showed that the coordination number of Ru–Cl over Ru–N–In/NC-3h was kept ~1, while the coordination number of Ru–Cl on the post-hydrochlorination RuN4/NC presented an increasing trend, from ~1.3 (RuN4/NC-1h) to ~2.6 (RuN4/NC-3h) (Supplementary Table 7), verifying that the asymmetric Ru–N–In/NC configuration inherently blocked the over-chlorination of Ru. AC-HAADF-STEM analysis of RuN4/NC-3h indicated that the Ru centers aggregated from single atoms to nanoparticles with sizes up to ~4 nm, whereas the single-atom pairs were still retained on Ru–N–In/NC-3h catalyst (Supplementary Fig. 57), and even on Ru–N–In/NC-600h catalyst (Fig. 4c). Further, the XPS spectra of Cl 2p for the post-hydrochlorination samples show an apparent increase in the surface Cl content than that of the fresh sample for RuN4/NC-3h (Supplementary Fig. 58 and Supplementary Table 8).To obtain insights into bonding information for Ru–Cl interactions, we conducted PDOS analysis for Ru d orbitals before and after Cl* adsorption. Generally, the Hδ+ and Clδ− species in HCl can be seen as Brønsted acid and Lewis base, respectively. Thus, the Ru atom (Lewis acid) can accept the Clδ− atom (Lewis base) of HCl to form Ru–Clδ− and promote the scission of the H–Cl bond52. As illustrated in Fig. 6c and Supplementary Fig. 59, upon the adsorption of Cl*, the interaction between the p orbital of Cl and the \({d}_{z}^{2}\) orbital of Ru gives the σ bonds on both Ru–N–In/NC and RuN4/NC. However, the π bonds are derived from the coupling between the p orbital of Cl and the dx2–y2 orbital of Ru for Ru–N–In/NC, while the involved orbitals are changed to p(Cl) and \({d}_{{yz}}/{d}_{{xz}}\) for RuN4/NC (Fig. 6c and Supplementary Fig. 60). Such distinct hybridization manner is attributed to the asymmetric geometry and unique electron configuration of atomic Ru enabled by the strong d(Ru)–p(In) interaction for Ru–N–In/NC53,54,55.To reveal the evolution pathways of the chlorination process on the atomic Ru of Ru-N-In/NC and RuN4/NC, the formation energies of chlorinated species and the related PDOS analysis were carried out. For RuN4/NC, the gradual addition of *Cl to Ru centers was observed, giving a stable three Cl-coordination configuration (denoted as [RuN4/NC]Cl3) eventually. Most importantly, the entire chlorination process from RuN4/NC to [RuN4/NC]Cl3 is highly exothermic (Fig. 6d). That is because once the original symmetrical confinement of RuN4/NC was broken with the adsorption of the first Cl*, the subsequent chlorination process became more energy favorable. In contrast, although the evolution of [Ru–N–In/NC]Cl was easily obtained, the addition of second Cl* turned to be highly endothermic, as evidenced by the emergence of several obvious antibondings, such as the σ* bond (p orbital of Cl and \({d}_{z}^{2}\) orbital of Ru) and π* bond (p orbital of Cl and dx2–y2 orbital of Ru) shown in the PDOS plots of the [Ru–N–In/NC]Cl2 atomic interface (Supplementary Fig. 61). Therefore, the over-chlorinated structures such as [Ru–N–In/NC]Cl2 and [Ru–N–In/NC]Cl3 (Eform > 0 eV) are unstable (Fig. 6d). Overall, the strong d(Ru)–p(In) interaction orbital couplings in Ru–N–In/NC promote the thermodynamic transition of the chlorination process from exothermal to endothermal compared to RuN4/NC, intrinsically avoiding the over-chlorination of Ru and ensuring its excellent stability. Furthermore, the progressive chlorination of Ru–N–In/NC, particularly in proximity to Ru atoms, results in a reduced electron density at the metal sites, as concluded from a shift to higher binding energies in the XPS spectra of Ru 3p (Supplementary Fig. 62). Unfortunately, the active Ru species of RuN4/NC in comparison with Ru–N–In/NC was completely deactivated to Ru0 under the reactive atmosphere (C2H2 + HCl), as demonstrated by Ru 3p XPS (Supplementary Figs. 63–65).Coke accumulation is an important indicator for measuring the stability of catalysts, thus, thermogravimetry (TG) was applied to determine coke deposition through the weight loss differences56,57. Fortunately, slight coke deposits (ca. 0.18%) were observed on the surface of Ru–N–In/NC, while the ample coke coverage (ca. 5.23%) was accumulated over the RuN4/NC surface (Fig. 6e). Thus, the possible explanation for the observed changes from “coking-prone” carbon to “coking-resistant” carbon upon indium introduction is that Ru–N–In/NC not only effectively adsorbed acetylene but also was favorable for the desorption of VCM13. Furthermore, Ru–N–In/NC can restrain excessive Cl* deposition and addition on the catalyst surface, thus alleviating the aggregation of Ru active sites and preventing the coupling of C–C bonds to form coke precursors. However, for RuN4/NC, excessive Cl* adsorption induced the aggregation of Ru, leading to the shift in the predominating deactivation mode, from agglomeration to coke deposition15. In addition, the leaching experiment and extraction operation interpreted that the Ru–N–In/NC catalyst also had superior recyclability and reusability (Supplementary Figs. 66 and 67). In conclusion, the successful design of Ru–N–In/NC catalyst with multiple excellent properties opens up the possibility of practical application of single-atom Ru catalysts.
