Conversion of methane to C₂ liquid oxygenates by Ru atom arrays

Abstract

The efficient conversion of methane to valuable C₂ liquid oxygenates still remains silent. Here we show a new catalytic system of C-C coupling towards the C₂ liquid oxygenates with high selectivity and productivity by the selective anchoring zero-valent ruthenium atoms from individual ones to two and three atoms in the electron-rich 18-carbon cavity of graphdiyne. Theoretical calculations demonstrate that Ru₃-GDY supplies neighboring robust active sites to promote efficient C-C coupling to C₂ liquid oxygenates, due to the p-d coupling resonance that guarantees the distributed charge among Ru₃ sites resulting in active region to accelerate the methane oxidation. Our results show Ru atoms in GDY enable the highly efficient and selective activation of CH₄ to the key ·CH₃ and ·CH₂OH intermediates, which allows the selective C-C coupling to gain C₂ liquid oxygenates and shows the high selectivity (67%) and yields (7.25 mmol g_cat⁻¹ h⁻¹) towards C₂ liquid oxygenates.

Atomic-level catalyst design faces challenges in methane activation and selective C–C coupling. Here, Ru atoms anchored on graphdiyne enhance methane oxidation, achieving high selectivity and yield for valuable C₂ liquid oxygenates.

Introduction

In recent years, a great focus and research “hot” spot of scientists and engineers has been the development of new structures and components of catalytic systems^1–3. It can be used efficiently for highly selective C-C coupling reactions to convert CH₄ into high-value-added C₂₊ chemicals (e.g., ethanol, acetic acid, etc.) under mild and environmentally friendly conditions^4–7. However, as we expected, to date there are only a few successful examples of converting CH₄ to C₂ acetic acid with the assistance of mediative agents (e.g., coupling agent), the selectivity and yield are far from what we expected. Due to the complexity of the catalytic conversion process on the reaction process, not only the catalyst is required to have a high active site, but also the distribution of the active site and the distance between the sites are strictly required^8,9. These integrated advantages have the potential to enable CH₄ activation to produce highly selective key intermediates such as (e.g., ·CH₃ and ·CH₂OH).

An idealized model is that zero-valent metal atoms can be independently anchored to the support materials, and strong reversible incomplete charge transfer is formed and dynamic charge non-integer mutual transfer can be maintained between metal atoms and the support materials during the reaction process^10,11. Such a model has the ability to control all metal atoms through self-organization and assembly to form a powerful catalytic system, which is considered to be an ideal system for creating catalytic materials and efficient energy conversion models^12,13. Furthermore, considering the various competitive reactions during the CH₄ oxidation process, it is natural to modulate the properties (e.g., the types of the metal atoms, the anchoring position and relative distance in-between, etc.) of metal atoms in atomic catalysts to guide the formation of targeted intermediates and enrich the coverage of intermediate species on the catalyst surface, so that C-C coupling can proceed smoothly¹⁴. The challenge for our research is how multiple zero-valent metals are anchored and how precisely the anchored metal atoms create active centers. There is no doubt that traditional support materials cannot incorporate defects with atomic precision or to achieve the stable anchoring of metal atoms on the surface of traditional supports due to high surface energy resulting in aggregation of metal atom^15–17.

Due to the numerous alkyne bonds and the limiting effect of geometric patterns, the sp- and sp²-cohybridized graphdiyne (GDY) provide the innate conditions for atom by atom anchoring to form active sites and active centers¹⁸. The triangular electron-rich18-carbon cavity, and the specific interactions between GDY and metal atoms holds great potential for chelating multiple metallic atoms¹⁹. Another very important feature of GDY is that the distinctive incomplete charge transfer and chemical bond conversion offers efficient way for tuning the adsorption/desorption behavior of intermediates. GDY behaves like vast electronic pools, dynamically modulating the electronic structure of metal atoms to furnish optimal active sites for diverse reactions^13,20. These inherent characteristics of GDY establish an ideal platform for fabricating multi-atomic catalysts tailored for intricate reactions.

In this work, we achieved the controlled and selective construction of new active sites by atomically precise anchoring Ru atoms from individual ones to two and three atoms in the electron-rich 18-carbon cavity of GDY. Density functional theory (DFT) calculations have unraveled the unique electronic structures of Ru₃-GDY significantly lowers the orbital repulsion through the p-d coupling resonance, leading to the construction of active Ru₃ region with neighboring active sites to facilitate the C-C coupling towards the C₂ liquid oxygenates with enhanced selectivity and productivity. In-situ characterizations show that such well-defined atom catalysts enable the highly efficient activation of methane (CH₄) and the selective generation of the key ·CH₃ and ·CH₂OH intermediates, which allows the selective C-C coupling to gain C₂ liquid oxygenates in aqueous solution at mild temperature. Ru₃-GDY has showcased an impressive C₂ liquid product yield of up to 7.25 mmol g_cat⁻¹ h⁻¹ with a remarkable selectivity of 67% when the reactants are only methane and H₂O₂, in which ethanol and ethylene glycol account for 87% of the C₂ liquid oxygenates, outperforming the Ru single-atom and dimer catalysts.

Results

Fabrication and characterization of Ru atom arrays

The hierarchical fabrication of Ru atom arrays on GDY proceeds through synergistic π-d orbital interactions and spatial confinement effects (Fig. 1). We developed a self-limiting assembly strategy, including (i) the electron-rich hexagonal cavities in GDY (18-C pores) selectively coordinate with Ru precursors via π → d orbital electron transfer, and (ii) triangular alkyne offers steric constraints to direct the anchoring of Ru atoms on GDY (Supplementary Fig. 1-4)^21,22. Crucially, the GDY substrate establishes an electronic fencing mechanism through π-backdonation (Ru-GDY), simultaneously enriching electron density at Ru sites while suppressing atomic migration^23–26. Subsequent calcination treatment (400 °C, Ar) yields Ru_x-GDY catalysts (x = 1–3) with controlled nuclearity, as verified by TEM analysis and elemental mapping (Supplementary Fig. 5 and Table S1). High-resolution transmission electron microscopy (HRTEM) images reveal the crystallinity of GDY (lattice spacing of 0.46 nm; ABC stacking; Supplementary Fig. 3)^27,28 were well preserved for the Ru₁-GDY (Fig. 2a), Ru₂-GDY (Fig. 2b), and Ru₃-GDY (Fig. 2c) samples. AFM analysis show there is almost no changes in thickness (Supplementary Fig. 6), which might be due to the atomic anchoring. The hydrophobic properties of the Ru_x-GDY (Supplementary Fig. 7), benefiting to the adsorption of methane gases. Aberration-corrected high-angle annular dark-field scanning TEM (AC-HAADF-STEM, Fig. 2d-i and Supplementary Fig. 8) revealed numerous ordered patterned isolated, paired, and trimeric bright spots on Ru₁-GDY, Ru₂-GDY and Ru₃-GDY, without the metal aggregations. The intensity surface maps (Fig. 2g-i) can visually distinguish the Ru single-atom, dimers, and trimers, self-organized on the surface of GDY, indicating that the successful of our innovative strategy for selectively and controllably engineering the anchoring position and relative distance between isolated single metal atoms (Supplementary Fig. 9). The weakening of the Raman peak for alkyne bonds at 1961 and 2173 cm⁻¹ might be due to the interaction between Ru atoms and alkyne bonds of GDY (Supplementary Fig. 10)²⁹.

Fig. 1. Schematic illustration for the fabrication of Ru atom arrays.

The self-organizing of Ru metal precursors the formation process of metallic Ru atom arrays on the surface of GDY.

Fig. 2. Morphological characterizations.

HRTEM and SAED images of Ru₁-GDY (a), Ru₂-GDY (b) and Ru₃-GDY (c). Low and high-magnification AC-HADDF-STEM images of Ru₁-GDY (d, g), Ru₂-GDY (e, h) and Ru₃-GDY (f, i), 3D intensity surface plot and intensity range for the dashed white regions of the images of g-i.

Extended X-ray absorption fine structure (EXAFS; Fig. 3a) analysis shows that there is only one strong peak for Ru-C at 1.55 Å for Ru₁-GDY; while two peaks of Ru-C and Ru-Ru interactions were observed for both Ru₂-GDY and Ru₃-GDY, in consistent with the wavelet transform (WT) of Ru L-edge EXAFS contour maps (Fig. 3b)³⁰. Quantitative EXAFS curve fitting (Supplementary Fig. 11 and 12, Table S2) reveals that Ru atoms in Ru_x-GDY samples coordinate with ~ six carbon atoms, with an average distance of about 2.19 Å. For Ru-Ru interactions, the Ru₂-GDY and Ru₃-GDYexhibit average coordination numbers (CNs) of 1.3 and 2.0. These findings confirm the successful anchoring of Ru₁, Ru₂ and Ru₃ atomic units in one alkyne-rich holes of GDY (Supplementary Fig. 13). To reveal the origins of superior catalytic performances of Ru₃-GDY, we have performed theoretical calculations to reveal the interactions between Ru SA sites. First, we have demonstrated the electronic distributions of d orbitals in Ru₁-GDY (Fig. 3c). Notably, the Ru site has shown interactions with GDY support through p-d orbital coupling, where the Ru acts as the main active site to stabilize the intermediates. As the second Ru site is introduced nearby, we notice evident electronic modulations induced by the interactions between Ru-Ru sites. The overall Ru₂-GDY has become more electron-rich, where the p-d coupling orbitals become dominant with more active electrons distributed near Ru sites. Owing to the strong d-d orbital repulsion effect in Ru₂-GDY with uneven electronic distributions in the pore of GDY, the two Ru atoms slightly deviate from the GDY surfaces while keeping the Ru-Ru bond length unchanged. However, for Ru₃-GDY, the third Ru sites have played a significant role in balancing both lattice and electronic structures. The three Ru sites are evenly distributed in the pore of GDY, forming a highly symmetrical trigonal bonding with GDY. Accordingly, the electronic distributions demonstrate a p-d coupling resonance with nested trigonal orbital coupling, which largely alleviates the d-d orbital repulsion forces to supply stable and robust Ru active sites at close distances. Based on these results, we have further compared the electronic structures through the projected partial density of states (PDOS) for Ru sites in different catalysts (Fig. 3d). From Ru₁-GDY to Ru₃-GDY, we notice the band offset of Ru-4d orbitals, where the distances from Ru-4d to E_F gradually increase from-0.28 eV in Ru₁-GDY to −1.15 eV in Ru₃-GDY. Although the d-band center in Ru₃-GDY is downshifted, the evidently broadened orbitals are due to the much stronger p-d coupling resonance among the three Ru sites, leading to the construction of highly electroactive regions to promote catalysis. The slightly lowered d-band center potentially contributes to optimizing the binding strengths of intermediates, which facilitates the conversion with lower energy barriers. With the band offset, the broadened Ru-4d orbitals become more matching with s,p orbitals of the initial reactants and final products. For the free CH₄ molecules, the orbitals are located near E_V-4.98 eV (E_V denotes 0 eV), which are closer to the d-band center of Ru-4d in Ru₃-GDY to initiate the electron charge transfer during the adsorption of reactants. In contrast, the orbitals of free CH₄ are located at the peak splitting of Ru-4d orbitals, leading to much weaker orbital overlapping for the subsequent catalysis. The orbital shifting is also correlated with the charge, where more electron-rich features also result in the downshifting and broadened d orbitals. For the charge of Ru sites and d-band centers, we notice a consistent trend, where the downshifting of Ru-4d orbitals is accompanied by the reduced charges (Supplementary Fig. 13). These results are supportive of the experimental characterizations that Ru₃-GDY shows the lowest valence states. Meanwhile, the correlation of the d-band center displays a converse trend of the key reactants *CH₄ and *OH. The enhanced adsorption energies are potentially attributed to the improved p-d overlapping with broadened Ru-4d orbitals in Ru₃-GDY. The π states of free C₂H₅OH are located close to the E_F, which matches well with the Ru-4d orbitals due to the band offset, benefiting the conversions to C₂ products. The corresponding self-consistent screened onsite (SC-SO) potential calculations of Ru-4d are also revealed in Ru_x-GDY (Fig. 3e). From Ru₁-GDY and Ru₂-GDY, there is an evident increase of the SC-SO potential, which is attributed to the presence of strong d-d orbital repulsive energy between Ru-Ru in Ru₂-GDY. In general, such d-d orbital repulsive energy will contiuously increase for Ru₃-GDY. However, the SC-SO potential in Ru₃-GDY is reduced than the original trend, which is ascribed to the p-d coupling resonance that lowers the repulsion among three neighboring Ru sites, leading to the stabilization of Ru₃ sites with robust catalytic activity. The reduced charge of Ru sites is consistent with X-ray absorption near-edge structure spectroscopy (XANES; Fig. 3f) and X-ray photoelectron spectroscopy (XPS; Supplementary Fig. 15) results show the valent states of Ru atoms in Ru_x-GDY samples close to the zerovalent metallic, with peak intensities trend of Ru₃-GDY (0.06) < Ru₂-GDY (0.14) < Ru₁-GDY (0.26) (Fig. 3g). which are beneficial for enhancing the methane oxidation ability The lower valence states of Ru sites improve the oxidation process³¹. Compared to Ru₂-GDY_, the smaller charge difference among Ru sites further confirms the even charge distributions in the active Ru₃ region realized by the p-d coupling resonance after electronic redistributions, supplying more active sites in close distances to promote C-C coupling for C₂ liquid oxygenates. The electron density difference (EDD; Fig. 3h) plot of Ru₃-GDY also shows that the Ru₃ region shows the co-existence of electron depletion and accumulation induced between Ru and GDY with efficient electron exchange and transfer.

Fig. 3. Coordination Structure and Electronic Properties of Ru Atoms in Ru_x-GDY (x = 1, 2, 3).

a Fourier-transformed magnitudes of the experimental L-edge EXAFS signals of Ru catalysts along with reference samples (dashed lines). The Fourier transforms are not corrected for phase shift. b Wavelet transforms of the Ru L-edge EXAFS signals of Ru_x-GDY (x = 1, 2, 3), Ru₂O₃, and Ru foil. c The contour plots of electronic distributions for d orbitals in the Ru₁-GDY, and p-d orbitals in Ru₂-GDY and Ru₃-GDY. Blue isosurface = bonding orbitals, and green isosurface = anti-bonding orbitals. d The PDOS comparisons of Ru-4d orbitals in Ru_X-GDY. e The self-consistent screened onsite (SC-SO) potential calculations of Ru_x-GDY. f XANES spectra at the Ru L-edge of the Ru_x-GDY (x = 1, 2, 3), Ru foil, RuO₂, and RuCl₃; the inset shows the magnified image of XANES spectra. g Relationship between Ru k-edge absorption energy (E₀) and oxidation state for Ru_x-GDY, Ru foil, RuCl₃, and RuO₂. h The electron density difference in Ru₃-GDY.

Selective methane oxidation performances of Ru_x-GDY

The methane oxidation reaction was conducted in an autoclave reactor. All Ru_x-GDY exhibited high activity towards methane conversion, while there no any products were observed by using pure GDY as the catalysts (Fig. 4a, Supplementary Fig. 16 and 17, Supplementary Table 3), which implies that Ru atoms are the active center for methane oxidation. Under the same reaction conditions, Ru₁-GDY showed only C₁ liquid oxygenates with the yield of 2.2 mmol g_cat⁻¹ h^–1, without any C₂ products. The C₂ liquid oxygenates yield significantly increased to 0.81 and 3.87 mmol g_cat⁻¹ h⁻¹ for Ru₂-GDY and Ru₃-GDY, respectively. Ru₃-GDY possesses the highest C₂ selectivity of 65.87% than Ru₁-GDY (0%) and Ru₂-GDY (34.13%). Notably, among these C₂ liquid oxygenates, the proportion of higher-value ethanol and ethylene glycol in these obtained C₂ liquid oxygenates reached up to 82.17% for Ru₃-GDY, much higher than previously reported catalysts^5,32–36. Isotopic labeling experiments show that the ¹³C-labeled products could only be observed in the case of ¹³CH₄ + H₂O₂ (Fig. 4b and Supplementary Fig. 18). Furthermore, the oxygen generated from the decomposition of H₂O₂ does not influence the oxidation of CH₄ (Supplementary Fig. 19). This confirms that the oxygenates are produced through the direct conversion of H₂O₂ and CH₄, rather than from environmental contaminants or the catalyst itself. All catalysts can effectively prevent further over-oxidation, as evidenced by the undetectable levels of CO₂ in the GC (Fig. 4c, Supplementary Fig. 20-22). No byproducts (e.g., formaldehyde and carbon monoxide) in methane oxidation reaction were detected.

Fig. 4. CH₄ Oxidation Performance of Ru_x-GDY (x = 1, 2, 3).

a CH₄ oxidation properties on different catalysts. b ¹³C NMR spectrum of liquid products over Ru₃-GDY. c Schematic illustration of the CH₄ oxidation on Ru_x-GDY (x = 1, 2, 3). d Relationship between product distribution and CH₄ pressure in methane oxidation catalyzed by Ru₃-GDY. e Comparison of C₂ product yields in CH₄ oxidation reaction. f Evolution of product distribution and selectivity over time by Ru₃-GDY. g Evolution of product distribution and selectivity of Ru₃-GDY with cycling times.

The yield and selectivity of each product is highly dependent on the reaction temperatures (Supplementary Fig. 23 and Table S4). For example, with the temperatures increased from 20 °C to 60 °C, the oxygenate yield increased from 0.88 mmol g_cat⁻¹ h⁻¹ to 5.88 mmol g_cat⁻¹ h⁻¹; while the selectivity of C₂ liquid oxygenates decreased from 77.27% to 65.87%, and the selectivity for CH₃CH₂OH and CH₂OHCH₂OH in C₂ liquid oxygenates decreased from 100% to 82.17%. At the temperatures above 60 °C, the product yield started to decline, with increase in the selectivity of HCOOH (from 30.39% at 60 °C to 39.49% at 90 °C) and CH₃COOH (from 11.88% at 60 °C to 42.65% at 90 °C). These suggest that l the activation energy for acetic acid formation is higher than that for ethanol and ethylene glycol. The apparent activation energy barriers (E_a) for Ru₃-GDY of C₂ liquid oxygenates formation were calculated as ca.39.21 kJ mol⁻¹ and ca. 49.44 kJ mol⁻¹ of C₁ liquid oxygenates (Supplementary Fig. 24). Ru₁-GDY exhibited larger E_a (60.28 kJ mol⁻¹) (Supplementary Fig. 25 and 26, Table S5). Increasing the pressure of methane can significantly enhance the yield of liquid oxygenates, as this may increase the concentration of reaction intermediates near active sites. For Ru₃-GDY, the yield of liquid oxygenates increased from 1.92 mmol at 0.5 MPa methane to 10.81 mmol at 3.5 MPa methane (C₂ liquid oxygenates yield: 7.25 mmol g_cat⁻¹ h⁻¹, 3.5 MPa CH₄), with the selectivity of C₂ liquid oxygenates consistently about 60% and reaching up to 67.06% (Fig. 4d and Table S6). Supplementary Fig. 27 shows that an apparent 0.97 and 0.80 order dependence of the formation rates of C₁ and C₂ liquid oxygenates on CH₄ pressure was measured, lower than the stoichiometry (second order), implying that Ru trimers sites may be relatively saturated by CH₄ adsorption. These results outperform those reported for state-of-the-art catalysts in the literature (Fig. 4e and Table S7). These results further support that the change of the patterned of the metal center leads to different product selectivity.

To achieve the highest yield and selectivity of C₂ liquid oxygenates under mild conditions, we investigated the evolution of product distribution and selectivity over time at 60 °C and 1.5 MPa pressure (Fig. 4f and Table S8). Extending the reaction time led to a gradual increase in liquid oxygenates yield (C₂ liquid oxygenates yield: 7.33 mmol g_cat⁻¹ for 6 h) while maintaining the selectivity towards C₂ products. Particularly, the selectivity of ethanol gradually decreased with extended time (35.31% to 19.66%), while the selectivity of acetic acid increased (11.88% to 16.38%), indicating that ethanol undergoes further oxidation to form acetic acid. This observation aligns with the oxidation results of ethanol under the same conditions (Supplementary Fig. 28). Methanol appeared to be more readily oxidized on Ru₃-GDY (Supplementary Fig. 29), confirming the higher proportion of HCOOH in C₁ liquid oxygenates. The Ru₃-GDY catalyst demonstrated reusability. Post-reaction, the catalyst could be separated from the reaction mixture by filtration, dried, and reused in subsequent runs. Even after the 5th consecutive cycle, the yield of C₂ liquid oxygenates remained at 3.22 mmol with a sustained selectivity of 66.49% (Fig. 4g and Table S9). Furthermore, ACHADDF-STEM images of post-reaction Ru₁-GDY and Ru₃-GDY catalysts along with XAFS (Supplementary Fig. 30) clearly confirmed that the Ru species retained their initial shape with no significant leaching or aggregation post-reaction. These data validate the high activity, excellent selectivity of C₂ liquid oxygenates, and recyclability of Ru₃-GDY.

Insights into the origin of the catalytic performances

Electron paramagnetic resonance (EPR) and in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) were employed to identify intermediates during the CH₄ oxidation reaction over Ru_x-GDY at 60 °C. EPR results (Supplementary Fig. 31) show the formation of *OH and *OOH free radicals for Ru₁-GDY or Ru₃-GDY. The cooccurrence of CH₃OH and CH₃OOH for Ru₁-GDY suggests a direct involvement of both OH and *OOH species in product formation at the Ru single-atom sites. No peroxides were observed for Ru₂-GDY and Ru₃-GDY, indicating that Ru multi-atom sites are more inclined to convert *OOH to *OH¹⁵. The adsorptions of *OH and *OOH play a significant role in the oxidation process, which is investigated on different binding sites (Supplementary Fig. 32). For Ru₁-GDY, the adsorption of *OH becomes very weak beside the Ru sites, indicating that the absence of close active Ru sites hinders the oxidation process. As an evident contrast, the adsorptions of *OH on both Ru and GDY sites are highly preferred in Ru₃-GDY, which offers sufficient active sites to stabilize both *OH and intermediates to promote the oxidation process. For the adsorption of *OOH, we notice that all the bindings are strong, whereas the *OOH is easy to dissociate, indicating that *OOH is challenging to stabilize on Ru_x-GDY. The *OOH mainly stays near the catalyst surface to supply a high oxidation environment to promote the oxidation. DRIFTS of Ru₁-GDY exhibited characteristic peaks attributed to *CH₃ (2957, 2822, 1072 cm⁻¹) and *COOH (1559, 1508, 1341, 1262 cm⁻¹) (Fig. 5a)^37–39. The intensity of the COOH peaks increased with the increase of the reaction time, with no changes in *CH₃ peaks. This is in consistent with the high formic acid selectivity for Ru₁-GDY. in addition to *CH₃ and *COOH, Ru₂-GDY (Fig. 5b) and Ru₃-GDY (Fig. 5c) show obvious *CH₂OH (1042 cm⁻¹) signals, signifying the unique stability conferred by Ru diatomic or triatomic species to this intermediate. Besides, the positions of *CH₃/*COOH/*CH₂OH peaks in DRIFTS progressively shifted towards higher wavenumbers, accompanied by an increase in corresponding peak intensities, as the Ru atoms increased from 1 to 3. These demonstrated the enhanced adsorption capabilities for intermediates with an increasing number of Ru in the active sites. These findings align well with the results obtained from O₂ temperature-programmed desorption (O₂-TPD, Supplementary Fig. 33). Notably, as the reaction time extended, intense diffraction peaks corresponding to intermediates appeared on the surface of Ru₃-GDY (Supplementary Fig. 34), indicative of its strong adsorption and stabilization of intermediates, beneficial to the high catalytic performance. No infrared peaks associated with CO were observed from Ru₂-GDY and Ru₃-GDY during the methane oxidation (Supplementary Fig. 35). The addition of CH₃OH into the reaction system resulted in the increase in HCOOH, CH₃COOH and CH₂OHCH₂OH, indicating that CH₃OH can be re-adsorbed on Ru₃-GDY surface to produce HCOOH, CH₃COOH and CH₂OHCH₂OH (Fig. 5d and Supplementary Fig. 29). The introduction of HCOOH resulted in minimal changes in the yield of CH₃COOH, while the yields of CH₂OHCH₂OH and CH₃CH₂OH both increased, confirming that CH₃COOH does not form from the coupling of *CH₃ and *COOH (Supplementary Fig. 36). The increase in the yields of CH₂OHCH₂OH and CH₃CH₂OH may be attributed to the electron-rich nature of ^-COOH, which elevates the electron density at Ru sites, thereby enhancing the conversion ability of CH₄³¹. In addition, there no ethane (formed by the coupling of *CH₃ with *CH₃) or any reaction products associated with ethane oxidation were detected, suggesting that the generation of CH₃COOH in Ru₂-GDY and Ru₃-GDY originates from the oxidation of CH₃CH₂OH. We finally put our efforts into the reaction trends of the methane oxidation process (Fig. 5e). The pre-adsorptions of *OH and *OOH form a highly oxidative environment near the Ru_x-GDY surface, which further enhances the weak adsorptions of *CH₄. The initial dehydrogenation process for *CH₃ formation meets an energy barrier in Ru₁-GDY and Ru₂-GDY, while Ru₃-GDY displays an energetically favorable trend. The large barrier for C-C coupling between *CH₃ and *CH₂OH in Ru₁-GDY significantly suppresses the formation of C₂ liquid oxygenates. Although such a C-C coupling barrier is reduced in Ru₂-GDY is slightly reduced, the stronger reaction trends to C₁ liquid oxygenates still dominate the final products. Due to the nearby active Ru sites, the barriers for C-C coupling in Ru₃-GDY are promoted with much smaller energy barriers. The subsequent oxidation of CH₃CH₂OH only shows a subtle barrier at the conversion to *CH₃CHOH, guaranteeing the efficient formation of C₂ liquid oxygenates. From the reaction kinetics, we have calculated the transition states (TS) of two key reaction steps during the C₂ formation as shown in Supplementary Fig. S37. The conversion of *CH₃ + *CH₂OH → *CH₃CH₂OH is the key C-C coupling step, and the *CH₃CH₂OH → *CH₃CHOH is the rate-determining step (RDS) for the C₂ liquid oxygenates. For the C-C coupling reaction, Ru₁-GDY shows the highest TS barrier of 2.12 eV, which is attributed to the limited active sites to stabilize the two key intermediates simultaneously, resulting in poor selectivity of C₂ products. In comparison, the C-C coupling barriers are significantly reduced for Ru₂-GDY and Ru₃-GDY due to the increasing active sites, which supply much faster reaction kinetics. Moreover, the TS barrier of the RDS in Ru₃-GDY is much lower (0.55 eV) than that of Ru₂-GDY (1.37 eV), guaranteeing the high selectivity towards the C₂ liquid oxygenates. Meanwhile, the formation of HCOOH from the oxidation of CH₃OH shows a large barrier, limiting the selectivity to the C₂ liquid oxygenates in Ru₃-GDY.

Fig. 5. Investigation of CH₄ Conversion Reaction Mechanism.

In situ DRIFTS Spectra on Ru₁-GDY (a), Ru₂-GDY (b) and Ru₃-GDY (c) under 1 atm of CH₄ at 60 °C. (d) Yield of oxygenated products in controlled experiments by introducing different reactants (CH₃OH/HCOOH/CH₃CH₂OH) on Ru₃-GDY for comparison. Quantification of CH₃OH/HCOOH/CH₃CH₂OH and CH₂OHCH₂OH in unrevealed products upon addition of CH₃OH/HCOOH/CH₃CH₂OH (Reaction conditions: 5 mg catalyst, 10 ml H₂O, 1 ml H₂O₂, 1.5 MPa CH₄, reaction at 60 °C for 1 h). e The reaction trends of methane oxidation in Ru_x-GDY.

Discussion

We have successfully developed well-defined zero-valent Ru units, with Ru₃-GDY demonstrating the ability to selectively convert CH₄ into C₂ liquid oxygenates in the absence of CO, without generating CO₂. The abundance of alkyne bonds on GDY surface induces the ordered arrangement of metal atoms, while the uneven charge distribution and non-integer charge transfer effects between the metal atoms facilitate the formation of nearly zero-valent Ru single, dimer, and trimers species. In-situ characterization results show that Ru₁-GDY can only oxidize methane to C₁ liquid oxygenates and the C₂ products appeared with the number of the Ru atoms in the active site increase to 2 and 3. DFT calculations have indicated that the band offset and electronic modulations induced by the p-d coupling resonance among three Ru sites in Ru₃-GDY, which not only optimizes the orbital repulsion energy but also constructs the high catalytic active region to boost up the C-C couplings based on the stabilization of intermediates on neighboring active sites. And the specific interactions between Ru and GDY and Ru atoms in-between could change the reaction route to a new way that can selectively activate CH₄ selectively into the key *CH₃ and *CH₂OH intermediates for easier C-C coupling to gain C₂ liquid oxygenates in aqueous solution at mild temperature. The C₂ liquid oxygenates yield of Ru₃-GDY reaches a remarkable 7.25 mmol g_cat⁻¹ h⁻¹, with a selectivity of 67%. Ethanol and ethylene glycol constitute 87% of the C₂ liquid oxygenates, and CH₃COOH only contributes 13%. Our findings not only provide a straightforward approach for the synthesis of well-defined multi-atom catalysts but also explore a unique CH₄ oxidation coupling mechanism that differs from conventional CO insertion pathways.

Methods

Fabrication of Ru atom arrays

Triruthenium dodecacarbonyl (10 mg) and 150 mg of GDY powder were added to 100 mL of ethanol, and the mixed solution underwent ultrasound treatment for 10 minutes. The mixture was then stirred at room temperature for 6 h followed by stirring at 60 °C for 2 h until complete evaporation of ethanol. Subsequently, the resulting product was dried at 60 °C for 12 h, then transferred to a quartz boat in a tube furnace. It was heated to 400 °C in a N₂ atmosphere with a heating rate of 5 °C/min and maintained at 400 °C for 1 h. After cooling, the Ru₃-GDY powder was obtained.

Ru dimers and single-atom catalyst (Ru SAC) was prepared according to the above-mentioned conditions using Benzeneruthenium (II) chloride dimer (11.7 mg) and Bis(cyclopentadienyl)ruthenium (II) (10.8 mg) as precursors, respectively.

Catalyst characterization

Scanning electron microscopy (SEM) measurements were performed on a FEI Apreo SEM. Transmission electron microscope (TEM), aberration-correction high-angle annular dark-field scanning TEM (AC-HAADF-STEM) and Energy dispersive spectrometer (EDS) elemental mapping images were recorded on FEI Spectra aberration-corrected TEM at 300 kV and a Talos F200X TEM at 200 kV. The metal loadings of the catalysts were detected by inductively coupled plasma atomic emission spectrometer (ICP–AES, Thermo ICAP 6300). IR spectra were recorded with a Bruker Invenio S spectrometer with a resolution of 4 cm⁻¹, and each spectrum is an average of 64 scans. Raman spectra was measured by using HORIBA Raman spectrometer at 473 nm laser excitation wavelength. Contact angle measurements were performed on an SL200KB (USA KINO) optical contact angle measuring apparatus at room temperature. The X–ray photoelectron spectra (XPS) were recorded on an ESCALab–250 X photoelectron spectrometer using an Al Kα source (1486.6 eV). Electron paramagnetic resonance measurements were performed at room temperature using a Bruker EMX-10/12 EPR spectrometer operated in the X-band frequency using the following parameters: microwave frequency of 9.8 GZ, microwave power of 20 mW, modulation frequency of 100 kHz, and a 10 dB attenuator. O₂–TPD measurements were carried out using the AutoChem II 2920 with a flowing 5% O₂/He stream (50 ml/min) at –50 °C. The samples were pretreated with Ar at 150 °C for 1 h to remove the adsorbed gaseous impurities before the TPD test. The X–ray absorption spectroscopy experiments at the Ru L edge were conducted on the 1W1B beamline of the Beijing Synchrotron Radiation Facility (BSRF). Ru foil and Ru₂O₃ were used as reference samples. All spectra were recorded in transmission mode at room temperature. We used IFEFFIT software to calibrate the energy scale, correct the background signal, and normalize the intensity.

The sample for in situ IR spectroscopy was loaded into a diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) cell (Harrick Scientific Products, Praying MantisTM). The cell was connected to a flow system that allows recording of spectra while gases pass through and around the sample. The samples in IR cell were pretreated at 200 °C for 1 h and then cooled down to 60 °C in flowing Ar (30 ml/min). The infrared spectrum at this moment serves as the background. Subsequently, CH₄ and H₂O₂ were introduced into the gas stream while the time-resolved spectra were collected.

Catalytic activity measurements

Methane conversion was tested in a 50 ml stainless-steel autoclave. In a standard procedure, 5.0 mg of catalyst and 6.6 mL deionized water were added sequentially to a polytetrafluoroethylene (PTFE) liner and sonicated for 30 s to ensure uniform catalyst dispersion. The liner was then transferred to a high-pressure reactor fitted with a PTFE gasket, followed by addition of 3.4 mL of 30% H₂O₂ solution. Residual air was eliminated by purging the system with argon (10 cycles), after which the reaction mixture was magnetically stirred at 1000 rpm (optimized for mass transfer) while heating to the target temperature (60 °C at 10 °C min⁻¹ ramp rate). CH₄ was then introduced to initiate the reaction, with strict avoidance of non-catalyst metal contact throughout the process. Upon reaction completion, the autoclave was immediately cooled to < 10 °C in an ice bath to preserve volatile products. Gas-phase products were analyzed directly by GC, while liquid products were filtered for subsequent offline characterization. Carbon source tracing was performed in separate experiments using ≥99% ¹³C-enriched CH₄.

The reusability test of the catalyst was evaluated as follows. After the first reaction cycle, the catalyst (5 mg) was washed with deionized water and centrifuged several times, and then the sample was dried overnight at room temperature in a vacuum desiccator. This dried sample was subsequently evaluated under the same reaction conditions to obtain the performance of second cycle of the catalyst. The remaining cycles were carried out in the same way; more than 5 reaction cycles were repeated.

Product analysis

Liquid-state NMR experiments were measured on a Bruker Avance III 600 MHz spectrometer equipped with pulsed field gradient and ultralow-temperature probes; this is a highly accurate approach with high detection sensitivity and high reproducibility. The assignment of the C_1/2 oxygenated liquid products (CH₃OH, CH₃OOH, HCOOH, CH₃CH₂OH, CH₃OH, CH₂OHCH₂OH, CH₃COOH) was identified by ¹H NMR spectroscopy. Chemical shifts are expressed in ppm using D₂O (99%D) as a solvent. Sample analysis was conducted using a coaxial NMR tube. Normally, 0.2 mL of the reaction liquor was combined with 0.1 mL of D₂O to create a solution for NMR measurement, which was then added to the outer tube of the coaxial NMR tube. The inner tube contained a reference solution containing DMSO in D₂O (2.5 μL DMSO, 1 mL D₂O). The signal of protons from the solvent H₂O is much higher than that from the products. Therefore, all ¹H NMR spectra were recorded using a pre-saturation solvent suppression technique to suppress the dominant H₂O signal.

The gas produced after the catalytic reaction was analyzed using a FULI GC9790 II gas chromatograph. The hydrocarbon products (CH₄, C₂H₆, C₂H₄, C₃H₈, C₃H₆) were analyzed using a Flame Ionization Detector (FID) equipped with an HP-AL/S column. Oxygen (O₂), carbon monoxide (CO), and carbon dioxide (CO₂) were analyzed using a Thermal Conductivity Detector (TCD) equipped with a TDX-1 column.

Based on the acquired one-dimensional nuclear magnetic resonance (NMR) spectrum of hydrogen, the mass of the measured substance was calculated by the ratio of the integrated areas of the internal standard and the measured substance. The specific calculation formula is given in Eq. (1), where all nuclear magnetic resonance analyses utilized DMSO as an internal standard.

$$\mathrm{m}\left(x\right)=\frac{\mathrm{MW}\left(x\right)\mathrm{nH}\left(x\right)\mathrm{m}\left(\mathrm{std}\right)\mathrm{A}\left(x\right)}{\mathrm{MW}\left(std\right)\mathrm{nH}\left(x\right)\mathrm{p}\left(x\right)\mathrm{A}\left(std\right)}$$ 1

In the formula, m(x) and m(std) represent the masses of the measured substance and the internal standard, respectively; MW(x) and MW(std) represent the relative molecular weights of the measured substance and the internal standard, respectively; nH(x) and nH(std) represent the number of hydrogen atoms in the selected measured substance and the DMSO hydrogen spectrum signal used for integration of the measured substance and internal standard; P(x) and P(std) represent the purity of the measured substance and the internal standard, respectively; A(x) and A(std) denote the integrated areas of the signal peaks for the measured substance and the internal standard, respectively.

The method for calculating product selectivity is as follows:

$$\mathrm{oxygenate}\mathrm{selectivity}=\frac{Amountofliquidoxygenate}{Totalamountofliquidandgasproducts}\times 100\%$$ 2

Calculation setup

In this work, we have performed theoretical calculations through the DFT + U based on the embedded CASTEP packages⁴⁰. In particular, we have utilized the generalized gradient approximation (GGA) and Perdew-Burke-Ernzerhof (PBE) functionals to supply accurate exchange-correlation interactions^41–43. The electronic properties have been described based on the developed two-way crossover linear response method^44–47, which is a generalized searching method to identify the optimal parameters for describing the chemical bonding information within various materials systems. The geometry optimization has been performed through the Broyden-Fletcher-Goldfarb-Shannon algorithm with a k-point mesh of 3 × 3 × 3⁴⁸. The cutoff energy has been set with ultrafine quality at 380 eV with the ultrasoft pseudopotentials. For all the geometry optimizations, we have applied the following convergency criteria: 1) Hellmann-Feynman forces should not exceed 0.001 eV/Å, 2) the total energy difference should not exceed 5 × 10⁻⁶eV/atom, and the atomic displacement needs to be smaller than 5 × 10⁻⁴ Å.

Author contributions

Y.X. and Y.L. conceived the idea and supervised the work. F. B. and Y. X. designed the experiments, synthesized the catalysts, performed the activity tests, analyzed the characterization results, and wrote the draft. M.S. helped in the theoretical calculations. B.H. conceived and conducted the theoretical calculations and wrote the content related to the calculations. J.Y. and S.Z. helped with the synthesis of the catalyst. L.Q. helped with the morphological characterizations. Y. L. revised the draft. All authors contributed to the manuscript.

Peer review

Peer review information

Nature Communications thanks Yang Liu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data are available in the main text or the Supplementary materials. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-62785-9.