MoS₂-confined Rh-Zn atomic pair boosts photo-driven methane carbonylation to acetic acid

Abstract

Direct carbonylation of CH₄ to CH₃COOH provides a promising pathway for upgrading of natural gas to transportable liquid chemicals, in which high-efficiency CH₄ activation and controllable C–C coupling are both critical but challenging. Herein, we report that highly efficient photo-driven carbonylation of CH₄ with CO and O₂ to CH₃COOH is achieved over MoS₂-confined Rh-Zn atomic-pair in conjunction with TiO₂. It delivers a high CH₃COOH productivity of 152.0 μmol g_cat.⁻¹ h⁻¹ and turnover frequency of 62.0 h⁻¹ with a superior selectivity of 96.5%, outperforming previous photocatalytic CH₄ carbonylation processes. Mechanistic investigations disclose the key effect of Rh-Zn synergy in combination with photo-excited electrons from TiO₂ for CH₃COOH formation. The active OH species produced from O₂ photoreduction on the Zn site through proton-coupled electron transfer promotes CH₄ dissociation to CH₃ species, which then facilely couples with adsorbed CO on the adjacent Rh site forming the key CH₃CO intermediate for CH₃COOH formation.

Introduction

Conversion of methane (CH₄) to high value-added multi-carbon (C₂₊) oxygenates, such as acetic acid (CH₃COOH), is a significant pathway for the utilization of natural gas^1,2,3, which typically goes through the indirect and energy-intensive syngas route with severe carbon emission and poor product selectivity^4,5,6,7. Direct conversion of CH₄ to CH₃COOH under mild conditions is highly attractive and environmental-friendly. However, it suffers from not only the difficulty in the C–H activation of the inert CH₄, but also the trade-off in the formation of key CH₃ and CO surface intermediates with opposite carbon valance states. Introducing carbon monoxide (CO) as a co-reactant can bypass the generation of CO* intermediate from the complex and sluggish dehydrogenation and oxygenation of CH₄. Hence, combined with utilization of O₂ as an inexpensive oxidant, oxidative carbonylation of CH₄ with CO offers an economical and sustainable route for producing CH₃COOH.

The oxidative carbonylation of CH₄ with CO and O₂ to CH₃COOH involves three critical steps, including (i) activation and dissociation of O₂ to generate active oxygen species, (ii) activation and dissociation of CH₄ to CH₃ species, and then (iii) C–C coupling between adsorbed CO and CH₃. A variety of catalysts with supported transition metal (Rh, Pd, Ir, Pt, Au, Fe) single atoms or nanoparticles were developed for the oxidative carbonylation of CH₄ with CO under mild conditions^{8,9,10,11,12,13,14,15,16,17,18,19,20,21}. However, these processes are still limited by the insufficient productivity or selectivity of CH₃COOH, which are mostly due to the difficulty in controlling the complex reaction network with a single type of active phase. Achieving both high activity and selectivity for the conversion of CH₄, CO and O₂ to CH₃COOH requires precise regulation of above three key steps while suppressing the competing CO oxidation and product overoxidation to CO₂, which remains a great challenge.

Herein, we report a high-efficiency photo-driven carbonylation of CH₄ with CO and O₂ to CH₃COOH (CCOC₂) process over a heterostructure catalyst with Rh-Zn atomic pair dual sites confined in MoS₂ lattice integrated with TiO₂ (RhZn-MoS₂/TiO₂). It exhibits a high CH₄ turnover frequency (TOF) of 62.0 h⁻¹ and CH₃COOH productivity of 152.0 μmol g_cat.⁻¹ h⁻¹ with a superior selectivity of 96.5%, significantly surpassing those of previously-reported photocatalytic CH₄ conversion to CH₃COOH processes without additional energy input. Systematic experimental and computational studies propose a synergy at the MoS₂-confined Rh-Zn dual sites in combination with photoexcitation of the RhZn-MoS₂/TiO₂ for the selective production of CH₃COOH, which enables much higher productivity and selectivity of CH₃COOH than those over the catalysts with only Rh or Zn confined in the MoS₂ parts. The photoexcited electrons from TiO₂ prominently promotes the O₂ reduction at the confined Zn-Mo bridge site via a proton-coupled electron transfer (PCET) mechanism to generate Mo=O and Zn–OH species, in which the highly reactive Zn–OH species can activate CH₄ and enable dissociation of CH₄ to CH₃ species and then form the Zn–OH₂ species. The preferentially adsorbed CO on the adjacent Rh site could then facilely couple with the generated CH₃ forming CH₃CO species. Then, after the desorption of the generated H₂O from the Zn site and the PCET-based hydrogenation of the neighboring Mo=O to Mo–OH, the CH₃CO on the Rh site could be facilely transferred to the adjacent Zn site coupled with the adsorption of CO on the vacated Rh site, followed by the formation of CH₃COOH via the combination of the CH₃CO and OH species (Fig. 1a). In this regard, this photo-driven CH₄ carbonylation process is analogous to the CH₄ conversion process with O₂ by methane monooxygenase (MMO) in the aid of reduced nicotinamide adenine dinucleotide (NADH) which supplies protons and electrons²² (Fig. 1b). The construction of Rh-Zn atomic-pair dual sites separates the catalytic sites for the C–H activation and C–C coupling while establishes a synergy between them, thereby breaking the trade-off between the activity and selectivity in the CCOC₂ process.

Fig. 1: Schematic illustration of catalytic process for CH₄ conversion over the RhZn-MoS₂/TiO₂ and MMO catalyst.

The catalytic cycle for the photo-driven CH₄ carbonylation with CO and O₂ to CH₃COOH over the RhZn-MoS₂/TiO₂ (a), in comparison with the biocatalytic CH₄ conversion with O₂ by the MMO (Protein Data Bank 1FYZ) in the aid of NADH (b).

Results

Photocatalytic performance of CH₄ carbonylation

The photo-driven CCOC₂ process was carried out in a windowed autoclave with water (H₂O) as solvent under UV-visible light irradiation. To investigate the distribution and origin of reaction products, a series of controlled experiments were conducted by using CH₄, CO and O₂ (CH₄ + CO + O₂), isotopically labeled ¹³CH₄, CO and O₂ (¹³CH₄ + CO + O₂), isotopically labeled ¹³CO, CH₄ and O₂ (CH₄ + ¹³CO + O₂) and isotopically labeled ¹⁸O₂, CH₄ and CO (CH₄ + CO + ¹⁸O₂) as the feed gas, respectively, over the RhZn-MoS₂/TiO₂ catalyst. ¹H and ¹³C nuclear magnetic resonance (NMR) spectra and gas chromatography-mass spectrometry (GC-MS) analysis of the products show that the CH₄ was mainly converted to CH₃COOH with a small amount of CH₃OH and HOCH₂OH (Fig. 2a and Supplementary Fig. 1a). Only ¹³C-labeled products (¹³CH₃COOH, ¹³CH₃OH and HO¹³CH₂OH) were produced in the case of ¹³CH₄ + CO + O₂, and no liquid C_1-2 oxygenated products was detected in the control experiments using pure Ar, CO + O₂, or CH₄ + CO as the feed gas (Supplementary Fig. 1c). Moreover, to track the source of specific carbon atoms in the product, isotopically labeled experiment was conducted by using CH₄ + ¹³CO + O₂ as the reactant and an obvious signal of CH₃¹³COOH can be observed in ¹H NMR spectra (Fig. 2a). These results suggest that the carbon atom in the CH₃ part and CO part of CH₃COOH comes from the CH₄ and CO, respectively, rather than from possible C-containing contaminants in the environment or catalyst itself. Furthermore, isotopically labeled experiment using CH₄ + CO + ¹⁸O₂ as the reactant showed that obvious peaks at m/z = 62 and m/z = 47 assigned to CH₃CO¹⁸OH and CO¹⁸OH fragment were detected by GC-MS analysis, indicating that the oxygen atom in the OH part of CH₃COOH comes from O₂ (Supplementary Fig. 1b). In addition, no obvious signal of ¹³CO₂ was observed in the GC-MS of gaseous products and ¹³C NMR spectra of liquid products when using ¹³CH₄ + CO + O₂ as the feed gas (Supplementary Fig. 2), suggesting that the overoxidation of CH₄ is efficiently suppressed.

Fig. 2: Photocatalytic performance of CH₄ carbonylation with O₂ and CO.

a ¹H NMR spectra of products when using ¹³CH₄ + CO + O₂, CH₄ + ¹³CO + O₂ and CH₄ + CO + O₂ as the feed gas over the RhZn-MoS₂/TiO₂ catalyst, respectively. b Comparison of catalytic activity for different catalysts. c The effect of power density on the photocatalytic activity over RhZn-MoS₂/TiO₂ catalysts. d The catalytic activity over RhZn-MoS₂/TiO₂ under visible light irradiation (wavelength > 420 nm), UV-visible light irradiation (wavelength = 200–800 nm) and thermal catalysis at 65 °C with a similar reaction temperature to that of induced by the UV-visible light irradiation, respectively. Data in b and d are all presented as mean ± s.d. Each reaction was repeated for three times to obtain the error bars (n = 3). e Two-dimensional contour snapshot of CH₃COOH selectivity under different partial pressure of CH₄, O₂ and CO. f The productivity and selectivity of CH₃COOH in comparison with the previously reported photocatalytic systems for CH₄ conversion to CH₃COOH (see Supplementary Table 2 for more details). Reaction tests in (a-e) were conducted by using 5 bar CH₄ (or ¹³CH₄), 1 bar O₂ (or ¹⁸O₂), 11 bar CO (or ¹³CO), 5 mL H₂O and 25 mg catalyst with a stirring rate of 1000 revolutions per minute (rpm) for 5 h under a 300 W Xe lamp irradiation (wavelength = 200–800 nm) with a light intensity of 1500 mW cm⁻².

The CH₃COOH productivities over different catalysts under UV-visible light irradiation are summarized in Fig. 2b. Compared with the undoped MoS₂/TiO₂ and pure TiO₂ catalysts, Rh-doped or Zn-doped MoS₂ in conjunction with TiO₂ (Rh-MoS₂/TiO₂ and Zn-MoS₂/TiO₂) notably enhances the production of CH₃COOH, which indicates the important effect of Zn-doping and Rh-doping in promoting the CCOC₂ process. More importantly, co-doping Rh and Zn atoms into the MoS₂ part (RhZn-MoS₂/TiO₂) with optimized mass ratio of 2.7 and 2.8 w.t.%, respectively, further, significantly increase the CH₃COOH productivity and selectivity to 152.0 µmol g_cat.⁻¹ h⁻¹ and 96.5%, respectively, compared with those of 56.3 µmol g_cat.⁻¹ h⁻¹ and 87.7% over the Rh-MoS₂/TiO₂ catalyst, and 33.6 µmol g_cat.⁻¹ h⁻¹ and 59.5% over the Zn-MoS₂/TiO₂ catalyst (Fig. 2b and Supplementary Fig. 3 and 4). These results demonstrate the existence of a synergy effect between the MoS₂-confined Rh and Zn atoms in promoting the CCOC₂ process. In addition, with the increase of UV-visible light (λ = 200–800 nm) intensity from 300 to 1500 mW cm⁻², a notable enchantment on the CH₃COOH productivity and selectivity from 40.5 µmol g_cat.⁻¹ h⁻¹ and 86.0% to 152.0 µmol g_cat.⁻¹ h⁻¹ and 96.5% was observed over the RhZn-MoS₂/TiO₂ catalyst with an optimized TiO₂ mass ratio of 50 w.t.% (Fig. 2c and Supplementary Fig. 5a). However, the thermal-catalytic and visible light driven CCOC₂ process by RhZn-MoS₂/TiO₂ presents much lower CH₃COOH productivity and selectivity (Fig. 2d), implying that the UV-visible light can significantly promote the CH₄ carbonylation to CH₃COOH. Moreover, replacing TiO₂ with SiO₂ (RhZn-MoS₂/SiO₂) leads to a notably low CH₃COOH productivity of 28.4 µmol g_cat.⁻¹ h⁻¹ with a lower selectivity of 73.3% (Supplementary Fig. 5b). These results suggests that the photoexcited charge carriers from TiO₂ can prominently facilitate the production of CH₃COOH over RhZn-MoS₂/TiO₂ heterostructure under UV-visible light irradiation. Besides, the partial pressure ratio of CH₄, O₂ and CO also has a great influence on the CH₃COOH productivity and selectivity. Under a total pressure of 17 bar with CH₄/O₂/CO partial pressure ratio of 5/1/11, the CH₃COOH productivity reaches up to 152.0 µmol g_cat.⁻¹ h⁻¹ with a selectivity of 96.5% (Fig. 2e and Supplementary Fig. 6), which is significantly superior to the previously reported photo-driven CH₄ carbonylation process without external energy input (Fig. 2f). Moreover, the RhZn-MoS₂/TiO₂ catalyst offers significantly better selectivity and comparable productivity than that of photo-driven CCOC₂ process when operated at similar reaction temperatures above 110 °C (Supplementary Table 1). The rising of CH₄ pressure to 30 bar further promotes the productivity of CH₃COOH up to 426.0 µmol g_cat.⁻¹ h⁻¹ with a TOF of around 192.0 h⁻¹ (Supplementary Table 2). The RhZn-MoS₂/TiO₂ catalyst also displays good reusability without obvious deactivation in six consecutive reaction cycles (Supplementary Fig. 7). The morphology and valence states of the used RhZn-MoS₂/TiO₂ catalyst are well maintained in contrast with the fresh catalyst (Supplementary Fig. 8), indicating the high stability of this catalyst.

Morphology and electronic structure of RhZn-MoS₂/TiO₂

To reveal the active sites of RhZn-MoS₂/TiO₂ catalyst for the CCOC₂ process, comprehensive characterizations were performed to study the structural and electronic properties of the heterostructure. X-ray diffraction (XRD) analysis shows the obvious crystallographic structure and lattice parameters of 2H-MoS₂ and anatase and rutile mixed phase TiO₂ in the RhZn-MoS₂/TiO₂ catalyst (Supplementary Fig. 9)²³. In addition, the Raman spectra of RhZn-MoS₂/TiO₂ displays two characteristic peaks at 378 cm⁻¹ and 402 cm⁻¹ assigned to in-plane (E¹_2g) and out-of-plane (A_1g) vibration peaks of the 2H phase MoS₂, respectively, which further demonstrates the existence of 2H-MoS₂ in RhZn-MoS₂/TiO₂ (Supplementary Fig. 10). The HRTEM presents that the TiO₂ nanoparticles with a diameter of 20-30 nm are coated by 2-4 layers of MoS₂ nanosheets (Fig. 3a and Supplementary Fig. 11). The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image presents a distinct interface between the MoS₂ and TiO₂ (Fig. 3b, c and Supplementary Fig. 12), which suggests the existence of Ti–S bonds at the interface of MoS₂ and TiO₂ based on the intensity profile analysis (Fig. 3d)²⁴. This interface could work as the transport channels to promote the separation and transfer of the photogenerated electrons and holes, which have an important impact on the photocatalytic activity.

Fig. 3: The Structural characterization of RhZn-MoS₂/TiO₂ catalyst.

High-resolution TEM images (a) and high-resolution HAADF-STEM images (b) of RhZn-MoS₂/TiO₂. c-f The wide-field HAADF-STEM, the geometrical configuration and atomic intensity profile analysis of RhZn-MoS₂/TiO₂. g Simulated image of the monolayer RhZn-MoS₂. h The measured and simulated atomic intensity profile along the corresponding dotted boxes in f and g, respectively. i, j Normalized XAFS spectra of the Zn K-edge and Rh K-edge for obtained RhZn-MoS₂/TiO₂ and referenced Zn foil, ZnS, Rh foil and Rh₂O₃. k The EXAFS signal in R-space for the Zn foil, ZnS, Rh foil, Rh₂O₃ and RhZn-MoS₂/TiO₂ catalysts. l Wavelet transform of Zn K-edge and Rh K-edge of the RhZn-MoS₂/TiO₂ catalysts, respectively.

Beyond that, the high magnification HAADF-STEM image combined with the extended X-ray absorption fine structure (EXAFS) and wavelet transform reveal that the Zn and Rh were atomically confined in the MoS₂ nanosheets via replacing the Mo atoms and the confined Zn and Rh were coordinated with S atoms without the formation of Rh–Rh and Zn–Zn bonds (Fig. 3e, k, l and Supplementary Fig. 13)²⁵. The intensities of spots are counted through a cross-line scan (Fig. 3e, f), and the relative statistical intensities suggest the existence of adjacent Zn-Rh pairs confined in MoS₂ lattice, which is further confirmed by the simulated image of the atomic model and the line intensity profiles (Fig. 3g, h and Supplementary Fig. 14). The detailed analysis of the X-ray adsorption near-edge structure (XANES) spectra show that the Zn atoms are electron-deficient while the Rh atoms are electron-rich in the RhZn-MoS₂/TiO₂ compared with those in the Zn-MoS₂/TiO₂ and Rh-MoS₂/TiO₂, respectively (Fig. 3i, j), which could be due to the interaction between the adjacent Zn and Rh atoms in the RhZn-MoS₂/TiO₂ catalysts, leading to charge transfer from the less electronegative Zn to the more electronegative Rh. Based on the above analysis, we deduce that the enhanced photocatalytic CH₄ conversion could be originated from the Zn-Rh pair confined in MoS₂ in combination with the Ti–S bonds at the MoS₂-TiO₂ interface.

Furthermore, the light absorption properties, charge separation and transport efficiency were investigated to understand the photo-effect on the CCOC₂ process over the RhZn-MoS₂/TiO₂ catalyst. Compared with the TiO₂ and RhZn-MoS₂, the RhZn-MoS₂/TiO₂ catalyst presented a stronger light absorption on the UV-Vis diffuse reflectance spectroscopy (UV-Vis DRS) measurement, indicating a better ability of broad-spectrum light trapping from UV to infrared region (Fig. 4a). The increased optical absorption can significantly enhance the photothermal catalytic effect, which render an enhanced CH₄ conversion. Besides, the photoluminescence (PL) intensity and average lifetime (ave. τ) derived from the time-resolved PL (TRPL) spectroscopy decreased in the following sequence: RhZn-MoS₂/TiO₂ < RhZn-MoS₂ < TiO₂, which indicates that RhZn-MoS₂/TiO₂ possesses a better ability to inhibit the recombination of photogenerated electrons and holes (Fig. 4b and Supplementary Fig. 15a)²⁶. In addition, the RhZn-MoS₂/TiO₂ presents a higher photocurrent density than the TiO₂ and RhZn-MoS₂, which suggests that the RhZn-MoS₂/TiO₂ has a stronger photo response ability and greater transfer efficiency of the photogenerated carriers (Fig. 4c)²⁶. Based on the above results, we propose that the enhancement in the light absorption, separation and transfer of photogenerated excitons over the RhZn-MoS₂/TiO₂ effectively promote photocatalytic CCOC₂ process^26,27.

Fig. 4: The optical properties of TiO₂, RhZn-MoS₂ and RhZn-MoS₂/TiO₂ catalyst.

a UV-vis diffuse reflectance spectra. b Time-resolved photoluminescence spectra. c Transient photocurrent responses_. d–f Quasi in-situ high-resolution Zn 2p, Rh 3d and Mo 3d XPS spectra with and without UV-visible light irradiation. g, h Normalized XAFS spectra of the Zn K-edge and Mo K-edge in RhZn-MoS₂/TiO₂ catalysts with and without UV-visible light irradiation. i Schematic energy band diagrams and charge transfer route of the heterojunction between TiO₂ and RhZn-MoS₂ (E_vac = Vacuum energy, CB = Conduction band, VB = Valence band, E_F = Fermi level).

Quasi-in-situ X-ray photoelectron spectroscopy (XPS) and in-situ X-ray absorption fine structure (XAFS) were further investigated to analyze the photo-generated electron transfer direction by measuring the variation of the valence state of metal sites before and after light irradiation of the RhZn-MoS₂/TiO₂ catalyst. Quasi-in-situ XPS results display the obvious shifts towards lower binding energy for Zn 2p, Rh 3d and Mo 3d but towards higher binding energy for Ti 2p after UV-visible light, which indicates that the Zn, Rh and Mo atoms receive electrons from TiO₂ (Fig. 4d–f and Supplementary Fig. 12e). This is in accordance with the results from in-situ XAFS spectroscopy, where the Zn and Mo K-edge shifts to a lower energy level but the Ti K-edge shifts to a higher energy level under light irradiation (Fig. 4g, h and Supplementary Fig. 12f)²⁸. Furthermore, we have conducted UV-vis diffuse reflectance spectrum (UV-vis DRS), ultraviolet photoelectron spectroscopy (UPS) and femtosecond transient absorption to clarify the band structure, contact type and the separation of photoinduced carriers between of TiO₂ and RhZn-MoS₂. The UV-vis DRS and UPS analysis indicate that RhZn-MoS₂ has a smaller work function but a higher Fermi level than that of TiO₂ (Fig. 4i and Supplementary Fig. 16). This demonstrates that the electrons could transfer from RhZn-MoS₂ to TiO₂ to attain the Fermi level equilibrium, thus leading to the generation of an electric field pointing from RhZn-MoS₂ to TiO₂ at the interface. The built-in electric field induces the transfer of photogenerated electrons from the conduction band of TiO₂ to the valence band of RhZn-MoS₂, thus resulting in the photogenerated electrons and holes left on the RhZn-MoS₂ and TiO₂, respectively. Such a direct Z-scheme heterojunction between RhZn-MoS₂ and TiO₂ is not only beneficial for the electron mobility by shortening the charge transfer time and distance (Supplementary Fig. 15b), but also favorable for enhancing O₂ activation by using the photogenerated electrons, thus promoting the photocatalytic CH₄ conversion to CH₃COOH.

Mechanism analysis for the photocatalytic CH₄ conversion

To understand the reaction mechanism for the photocatalytic CCOC₂ process over the RhZn-MoS₂/TiO₂ catalyst, we constructed models of MoS₂ with Rh and Zn atoms co-confined into the MoS₂ lattice substituting the Mo atoms and performed density functional theory (DFT) studies. Calculation results show that introducing Rh and Zn atoms into the MoS₂ lattice notably decreases the average formation energy of sulfur vacancies (SVs) by 1.17 and 1.57 eV for the edge and in-plane SVs, respectively (Supplementary Fig. 17), which is favorable for the generation of SVs during the H₂ reduction treatment of the RhZn-MoS₂ catalyst at 300 °C. Among the models with different Rh-Zn co-doping configurations, the SVs confining an adjacent Rh-Zn pair is relatively more stable by over 1.42 eV than those confining separated Rh and Zn atoms (Supplementary Fig. 18), which suggests that the formation of SVs at the Rh-Zn pair-doped sites is energetically more favorable. Then, the adsorption-free energies (∆G_ads) of CO and O₂ on different sites of edge SVs confining a Rh-Zn pair were investigated, which shows that CO tends to be adsorbed on the Rh top sites with ∆G_ads of –1.10 eV while O₂ can be facilely co-adsorbed on the Zn-Mo bridge site with ∆G_ads of –1.01 eV (Supplementary Fig. 19). However, as shown in Fig. 5a and Supplementary Fig. 20, direct dissociation of the adsorbed O₂ (O₂*) to form Zn=O* and Mo=O* species for CH₄ activation (the symbol ‘=’ denotes top site adsorption configuration for O*) is endergonic by 0.92 eV and thus is thermodynamically unfavorable. In contrast, combined with protons in the water solvent and photoexcited electrons from the TiO₂, O₂* can be facilely hydrogenated to OOH* with a free energy change of –0.66 eV through a proton-coupled electron transfer (PCET) mechanism, followed by readily dissociation of OOH* to Zn–OH* and Mo=O* species with a free energy gain of –0.99 eV (Fig. 5a). In addition, a CO-assisted O₂* dissociation path without involving the PCET process was also considered, in which the O₂* reacts with another CO to generate Zn–O*–Mo (the symbol ‘–’ denotes bridge site adsorption configuration for O*) and CO₂ with an energy barrier of 0.57 eV and free energy change of –2.74 eV (Fig. 5a). Electronic structure analysis shows that the Zn–OH* species possesses notably higher electronic density of O 2p states near the Fermi level compared with those of the Zn–O*–Mo and Mo=O* species, suggesting that the Zn–OH* species has a higher activity for the following CH₄ activation (Fig. 5b).

Fig. 5: DFT studies on the reaction mechanisms of CH₄ conversion.

a Comparison of the formation energy for different O species. b Projected density of states (PDOS) of the O sites in the Mo=O*, Zn–O*–Mo and Zn–OH* structure. c Formation pathways for CH₃OH and CH₃COOH at Zn–O*–Mo and Zn–OH* site, respectively. Insets show the atomic structures of reaction intermediates.

After the PCET-assisted formation of Zn–OH* or the CO-assisted formation of Zn–O*–Mo species at the edge SVs-confined Rh-Zn pair sites, the CH₄ activation and C–C coupling steps were further studied and compared to understand the formation of CH₃COOH. At the Zn–OH* site, the breakage of the first C–H bond of CH₄ forming Zn–OH₂* and CH₃ followed by the direct coupling of the generated CH₃ with the pre-adsorbed CO on the adjacent Rh site (Rh–CO*) to form the key CH₃CO* intermediate, requires an overall activation energy of 0.70 eV, which is notably lower than the CH₄ activation energies of 0.90 eV and 1.30 eV at the Zn–O*–Mo and Mo=O* sites, respectively (Fig. 5c). Then, after the desorption of the generated H₂O* from the Zn site and the PCET-based hydrogenation of the neighboring Mo=O* to Mo–OH*, the CH₃CO* on the Rh site could be facilely transferred to the adjacent Zn site coupled with the adsorption of CO on the vacated Rh site, followed by the formation of CH₃COOH via the combination of the CH₃CO* and OH* with a low activation energy of 0.13 eV. Such a PCET-involved reaction mechanism with one oxygen of O₂ entering H₂O and the other entering CH₃COOH (Supplementary Fig. 21), was further verified by isotope-labeling experiments using CH₄ + ¹⁸O₂ + H₂O as the reactants, in which H₂¹⁸O was observed in the products under the photocatalytic condition while no H₂¹⁸O was observed without light irradiation (Fig. 6a).

Fig. 6: Experimental investigations of reaction mechanisms.

a GC-MS spectra of H₂O generated from CH₄ conversion with and without UV-visible light irradiation. b Impact of electrons and holes scavengers on photocatalytic CH₄ conversion. Reaction tests in a were conducted by using 5 bar CH₄, 1 bar O₂ (or ¹⁸O₂), 5 mL H₂O, and 25 mg catalyst with a stirring rate of 1000 rpm for 5 h.

Furthermore, we have conducted the in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) experiments to identify the role of Rh and Zn sites during the reaction process (Supplementary Fig. 22). In-situ DRIFTS using CO as a probe molecule was first performed on various catalysts under ambient conditions (Supplementary Fig. 22a). On the MoS₂/TiO₂ and Zn-MoS₂/TiO₂, similar peaks were observed at 2051 cm⁻¹ corresponding to adsorbed CO on the Mo site (Mo-CO*) rather than on the Zn site. In contrast, on the Rh-MoS₂/TiO₂ and RhZn-MoS₂/TiO₂ catalysts, larger peaks appear at 2083 cm⁻¹ beside the attenuated peak of Mo-CO*, which can be attributed to the adsorbed CO on the Rh site (Rh-CO*)¹⁵. These results suggest that CO prefers to be adsorbed at the Rh sites, which is in accordance with the DFT calculations that Rh sites possess the lowest CO adsorption free energy in contrast to the Zn and Mo sites (Supplementary Fig. 19).

Additionally, in-situ DRIFTS was performed to capture the reaction intermediates by controlling the reaction gas for CH₄ conversion over the RhZn-MoS₂/TiO₂ catalyst (Supplementary Fig. 22b and c). Firstly, the RhZn-MoS₂/TiO₂ catalyst is treated by 11 bar CO for 30 min, which leads to an absorption peak at 2083 cm⁻¹ corresponding to the vibrations of Rh-CO* being detected. As the feed gas is switched to 1 bar O₂ + H₂O for 1 min (Supplementary Fig. 22b), notable increase in the Mo=O* and Mo–O*–Mo signals owing to dissociation of O₂ is observed under light irradiation compared with the reference case without introducing O₂ (Supplementary Fig. 22c). It is worth noting that the signal of Zn-O species is hard to be observed because the infrared signals of Zn-O species are generally below 600 cm⁻¹. When RhZn-MoS₂/TiO₂ catalyst is further treated with 5 bar CH₄ for 10 min, the peak intensity of Mo=O* and Mo-O*-Mo gradually decreases but the intensity of the CH₃* peak at 1436 cm⁻¹ increases, indicating the reaction of CH₄ with the surface oxygen species. Meanwhile, the C=O vibration of CH₃CO*(1743 cm⁻¹) and the stretching vibration of the C-H bond in CH₃CO* (2974 and 2872 cm⁻¹) can be clearly observed, suggesting that the generated CH₃ species can facilely couple with the adsorbed CO on the adjacent Rh site to form the key intermediate CH₃CO* for CH₃COOH formation.

To interpret the product selectivity, the formation path of CH₃OH was also explored and compared with that of the CH₃COOH. At the Mo=O* site adjacent to Zn–OH*, the activation of CH₄ for CH₃OH formation is kinetically unfavorable due to its notably high activation energy of 1.30 eV as mentioned above (Fig. 5c). In contrast, at the Zn–O*–Mo site, dissociation of CH₄ to Zn–CH₃* and Mo–OH* requires a lower activation energy of 0.90 eV, followed by the formation of CH₃OH via the combination of CH₃* and OH* with an energy barrier of 1.02 eV or the formation of CH₃CO* intermediate via the combination of CH₃* with a pre-adsorbed CO* on the adjacent Rh site with an energy barrier of 1.13 eV (Fig. 5c). This indicates that the formation of CH₃OH at the Zn–O*–Mo site is kinetically more favorable with a relatively lower energy barrier than the formation of CH₃COOH.

The effect of TiO₂ on the reaction mechanism was also considered by constructing a heterojunction model with RhZn-MoS₂ integrated with TiO₂ (RhZn-MoS₂/TiO₂) (Supplementary Fig. 23). The energetics of several key steps including the adsorption of CO and O₂ and the activation of CH₄ were calculated using DFT (Supplementary Fig. 24 and 25), which shows that the adsorption energies of CO and O₂ and the CH₄ activation energy on the sites of RhZn-MoS₂/TiO₂ have no significant difference from those on the similar sites of the RhZn-MoS₂ (Supplementary Table 3 and 4). Moreover, the Zn, Rh, and Mo sites in the RhZn-MoS₂/TiO₂ possess the same activity trend compared with those in the RhZn-MoS₂ for the key adsorption and activation steps. On both structures, CO tends to be adsorbed on the Rh top sites and O₂ can be facilely co-adsorbed on the Zn-Mo bridge site, and the Zn–OH* sites exhibit higher activity than the Zn-O*-Mo sites for the C-H bond activation of CH₄. These calculation results indicate that TiO₂ in the heterojunction has little effect on the intrinsic activity of the lattice-confined catalytic sites in the MoS₂.

Control experiments were also conducted by using AgNO₃ as electron scavenger to interrupt the PCET process or using Na₂SO₃ as hole scavenger to prevent exciton recombination and prolong the lifetime of excited electrons, in which CH₃COOH productivity was significantly decreased and increased, respectively (Fig. 6b). These results indicate the crucial role of photoexcited electrons for the CH₃COOH production via the CH₃CO* intermediate, which are supportive to the above mechanism analysis. In contrast to the edge SVs-confined Rh-Zn pair site, the in-plane SVs-confined Rh-Zn pair site tends to be occupied by oxygen due to the much stronger adsorption of O₂ than CO and the generated O* species are less active for CH₄ activation owing to the higher energy barriers of over 1.13 eV compared with that of 0.70 eV at the edge (Supplementary Fig. 26). Therefore, we propose that the photocatalytic promotion effect on the CH₃COOH production is due to the lower energy barriers for the formation of active OH species from photoreduction of O₂ on the Zn site and the C–C coupling on the adjacent Rh site for the formation of CH₃CO* intermediate.

Discussion

In summary, an efficient RhZn-MoS₂/TiO₂ heterostructure catalyst is synthesized for the photo-driven CCOC₂ process, which delivers a record-high CH₃COOH selectivity of 96.5%, productivity up to 152.0 µmol g_cat.⁻¹ h⁻¹ and a TOF of 62.0 h⁻¹, outperforming previous photocatalytic CH₄ conversion to CH₃COOH processes. The Rh-Zn atomic-pair dual sites separates the catalytic sites for the C–H activation and C–C coupling while establishing a synergy between them, thereby breaking the trade-off between the activity and selectivity in the CCOC₂ process and leading to an exceptional performance for CH₃COOH production. Detailed in-situ characterization combined with theoretical investigations reveal that the photoexcited electrons from TiO₂ significantly promote the O₂ dissociation at the confined Zn site to generate highly reactive Zn–OH* species via a PCET mechanism, which can efficiently activate the C–H bond of CH₄ to CH₃ species, followed by facile C–C coupling between the CH₃ and adsorbed CO on the adjacent Rh site forming CH₃CO* species for CH₃COOH production. This work provides a reference to design an efficient catalyst for photocatalytic CH₄ carbonylation to CH₃COOH.

Methods

Catalyst preparation

The RhZn-MoS₂ catalysts were prepared via a one-pot solvothermal synthesis method. Typically, 900 mg of (NH₄)₆Mo₇O₂₄·4H₂O, 58.6 mg of (NH₄)₃RhCl₆, and 118.0 mg Zn(NO₃)₂·6H₂O were dissolved into 20 mL deionized water to form a homogeneous solution. Then, the homogeneous solution and 10 mL of CS₂ were added together in a 40 mL stainless steel autoclave in a glovebox under argon. The autoclave was then transferred to a muffle furnace and heated at 400 °C for 4 h. Subsequently, the formed RhZn-MoS₂ was collected and treated by 100 mL 6 M KOH solution at a constant temperature of 60 °C with stirring for 3 h, then washed several times with water and ethanol, and finally dried at 60 °C overnight. Finally, the RhZn-MoS₂ catalyst was transferred into a quartz tube reactor and reduced with H₂ (>99.9%) at 300 °C for 2 h to create sulfur vacancies. The mass contents of atomic Zn and Rh in RhZn-MoS₂ samples were measured by inductively coupled plasma optical emission spectrometry (ICP-OES).

The MoS₂ nanosheets were prepared via the direct chemical reaction of 900 mg (NH₄)₆Mo₇O₂₄ and 10 mL CS₂ under H₂O-assisted (20 mL) environment at 400 °C. In detail, CS₂ and H₂O can react to produce H₂S and CO₂ (CS₂ + H₂O → CO₂ + H₂S), and then the MoO₄^2- generated from the dissolution of (NH₄)₆Mo₇O₂₄ can be reduced by H₂S to form MoS₂ nanosheets (MoO₄^2- + H₂S → MoS₂ + H₂O).

The RhZn-MoS₂/TiO₂ catalysts were prepared through a solid phase synthesis method. Typically, the H₂ pretreated RhZn-MoS₂ (50 mg) and TiO₂ nanoparticles (50 mg) were fully mixed and then the mixed sample was transferred into a quartz tube reactor and treated with argon at 200 °C for 2 h.

Evaluation of photocatalytic methane conversion performance

The photocatalytic CH₄ carbonylation reaction was performed in a 50 mL autoclave equipped with a top quartz window to allow light irradiation. First, the autoclave was charged with 25 mg catalyst and 5 mL distilled water and then sealed. Before the reaction, the autoclave was flushed with argon for several times and then pressurized to 5 bar CH₄, 1 bar O₂ and 11 bar CO. The reaction mixture was stirred at an optimized speed of 1000 rpm to promote mass transfer under a 300 W Xe lamp irradiation (wavelength = 200–800 nm) with a light intensity of 1500 mW cm⁻² for a fixed reaction time (typically 5 h). During the reaction, a thermocouple was inserted into the solution to detect the temperature of the liquid solution. After the reaction, the autoclave was cooled in an ice bath to room temperature. The gaseous products were analyzed by gas chromatograph (GC, Shimadzu) and the liquid products were collected by filtration and then analyzed by 600 M ¹H and ¹³C NMR.

The reusability test of the catalyst

The reusability test of the catalyst was carried out as follows. After the first reaction cycle, the catalyst (25 mg) was separated from the reaction solution by a high-speed rotary dryer under vacuum conditions at room temperature. The dried sample was subsequently evaluated under the same reaction conditions to obtain the performance of the second cycle of the catalyst. The remaining cycles were carried out in the same way and six reaction cycles were repeated.

Products analysis

Liquid-State Nuclear Magnetic Resonance (NMR) experiments were performed on Bruker Avance III 600 MHz equipped with pulsed field gradient and ultra-low temperature probes, which is a highly accurate approach with high detection sensitivity and high reproducibility. The assignment of oxygenated products (CH₃COOH, CH₃OH, and HCOOH) was identified by ¹H NMR and ¹³C NMR spectroscopy. The quantitative analysis was based on the linear relationship between CH₃COOH, CH₃OH or HCOOH concentration and relative area versus 2 mM dimethyl sulfoxide (DMSO) as internal standard (Supplementary Fig. 27). Typically, 850 μL liquid product was firstly mixed with 50 μL of D₂O and 100 μL 2 mM DMSO, and then 600 μL of the mixture was sealed into the NMR tube for testing. The productivity (µmol g_cat.⁻¹ h⁻¹), CH₃COOH selectivity in oxygenated products, and the turnover frequency (TOF) were calculated as the following Eqs. (1), (2) and (3). The calculation of CH₃COOH selectivity in this system is based on the generated products from CH₄ conversion.

$${{\rm{Productivity}}} \, \left(\frac{{\mu mol}}{{g}_{{cat}.}{{\cdot h}}}\right)=\frac{{{\rm{amount}}}\; {{\rm{of}}}\; {{\rm{C}}}{{{\rm{H}}}}_{3}{{\rm{COOH}}} \, ({\mu mol})}{{{\rm{mass}}}\; {{\rm{of}}}\; {{\rm{catalysts}}}\left(g\right) {{\times }}{reaction\; time}(h)}$$

(1)

$${{\rm{C}}}{{{\rm{H}}}}_{3}{{\rm{COOH}}}\; {{\rm{selectivity}}} \, (\%)=\frac{{{\rm{amount}}}\; {{\rm{of}}}\; {{\rm{C}}}{{{\rm{H}}}}_{3}{{\rm{COOH}}} \, ({\mu mol})}{{{\rm{total}}}\; {{\rm{amount}}}\; {{\rm{of}}}\; {{\rm{oxygenated}}}\; {{\rm{products}}}\, ({\mu mol})}$$

(2)

$${{\rm{TOF}}}=\frac{{{\rm{total}}}\; {{\rm{amount}}}\; {{\rm{of}}}\; {{\rm{oxygenated}}}\; {{\rm{products}}} \, ({\mu mol})}{{{\rm{the}}}\; {{\rm{number}}}\; {{\rm{of}}}\; {{\rm{active}}}\; {{\rm{sites}}} \, ({\mu mol}){{\times }}{reaction\; time} \, (h)}$$

(3)

Gaseous products were quantified by gas chromatography (GC-2014, Shimadzu), which was equipped with flame ionization detector (FID) coupled with a methanizer (UMTR unit). The generated CO₂ was quantified by FID detector with a methane reformer (Supplementary Fig. 28).

Isotope labeling experiment

For isotopic reaction, ¹³C enriched CH₄ (¹³CH₄, Sigma-Aldrich, ≥99%), ¹³C enriched CO (¹³CO, Sigma-Aldrich, ≥99%) and ¹⁸O enriched O₂ (¹⁸O₂, Sigma-Aldrich, ≥97%) were used to trace the source of C and O in the products, respectively. The products were analyzed by 600 M ¹H NMR, ¹³C NMR spectra and GC-MS. The detailed reaction conditions for isotope labeling experiment were shown in the notes of Fig. 2.

Catalyst characterizations

Transmission electron microscopy (TEM) measurements were performed on a Phillips Analytical FEI Tecnai 20 electron microscope operated at an acceleration voltage of 200 kV. The high-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) images were collected on a JEOL JEM-2300F field-emission transmission electron microscope operating at 200 kV accelerating voltage.

Quasi in-situ XPS measurements were carried out on an Omicron XPS System using Al Kα X-rays as the excitation source at a voltage of 15 kV. Before the collection of spectrums, the catalyst was transferred into the pretreatment chamber and illuminated for 10 min by a 300 W Xe lamp (wavelength = 200–800 nm, light intensity of 1500 mW cm⁻²).

Zn K-edge XAFS spectra, Rh K-edge XAFS spectra, and Mo K-edge XAFS spectra for RhZn-MoS₂/TiO₂ were measured at the BL14W1 beamline of the Shanghai Synchrotron Radiation Facility (SSRF). For the in-situ XAFS test, the spectrum was collected when the RhZn-MoS₂/TiO₂ catalyst was exposed to a 300 W Xe lamp (wavelength = 200–800 nm, light intensity of 1500 mW cm⁻²).

Quantitative identification of active sites

To quantitatively identify the coordination unsaturated Rh-Zn pair at the Mo edge of MoS₂ that serve as the active sites for CH₄ conversion, in-situ O₂ pulse adsorption method^29,30,31 in combination with in-situ DRIFTS characterizations of the Mo=O and Mo–O–Mo species were applied to determine the density of sulfur vacancies (SVs) at the edge of the prepared MoS₂-based catalysts. The total density of SVs was calculated on the basis of specific adsorption capacity of O₂, where one O₂ molecule corresponds to one SV. The relative ratio of the integrated areas from DRIFTS was used to distinguish the proportion of edge and in-plane SVs, so the density of SVs at the edge of catalysts can be approximately obtained. Three edge SVs correspond to one active site for CH₄ carbonylation to CH₃COOH production based on the results of DFT calculations (Supplementary Fig. 29 and 30).

Computational details

DFT calculations were performed using the Vienna Ab-initio Simulation Package (VASP) with the projector augmented-wave (PAW) pseudopotential for describing the ionic cores^32,33,34. All calculations were based on the generalized gradient approximation (GGA) method with the Perdew-Burke-Ernzerhof (PBE) functional for the exchange-correlation term with a plane-wave cutoff energy of 400 eV^35,36,37. DFT-D3 method of Grimme was used to calculate the van der Waals correction^38,39. The spin-polarize calculation was carried out in all cases. A nanoribbon model of RhZn-MoS₂ with six repeating units along the ribbon direction was built to simulate the edges, with the Mo edge saturated with S monomers^40,41. A heterojunction model was built with the nanoribbon RhZn-MoS₂ integrated with one oxygen vacancy anatase TiO₂ (101), corresponding to a 20.66 × 22.69 × 25.79 ų large supercell. The vacuum thicknesses were set larger than 15 Å. A Monkhorst-Pack k-point sampling of 1 × 1 × 1 was selected. In structural optimizations, the residual forces between atoms were converged to below 0.05 eV Å⁻¹. Formation energies of SVs at the edge are calculated by using H₂ and H₂S as reference states. The transition states were searched using the fixed bond-length method as implemented in the Atomic Simulation Environment (ASE)⁴² and VASP. All transition states were confirmed to have only one imaginary frequency. The free energies (G) of the adsorption intermediates on the surface were calculated as E_total + ZPE + H_vib − TS_vib, where E_total is the DFT-calculated total energy, ZPE is the zero-point energy, T is the temperature and H_vib and S_vib are the enthalpy and entropy parts.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.