Photosynthesis of CH₃OH via oxygen-atom-grafting from CO₂ to CH₄ enabled by AuPd/GaN

Abstract

The direct co-conversion of methane and carbon dioxide into valuable chemicals has been a longstanding scientific pursuit for carbon neutrality and combating climate change. Herein, we present a photo-driven chemical process that reforms these two major greenhouse gases together to generate green methanol and CO, two high-valued industrial chemicals. Isotopic labeling and control experiments indicate an oxygen-atom-graft occurs, wherein CO₂ transfers one O into the C–H bond of CH₄ via photo-activated interfacial catalysis with AuPd nanoparticles supported on GaN. The photoexcited AuPd/GaN interface effectively orchestrates the CH₄ oxidation and the CO₂ reduction producing 13.66 mmol g⁻¹ of CH₃OH yield over 10 h. This design provides a solid scientific basis for the photo-driven oxygen-atom-grafting process to be further extended to visible light region.

Introduction

Global warming due to excess greenhouse emissions is sparking unprecedented attention worldwide. CO₂ and CH₄, accounting for over 90% of total global emissions in 2021, have been deemed as primary anthropogenic greenhouse gases¹. Concurrently, both gases, characterized by high-bond energy, have been recognized as C1 feedstocks for bulk chemicals and fuel productions, aligning with the vision of achieving net-zero emission (Fig. 1a)^2,3. In recent years, significant strides have been made in reducing CO₂^4,5,6,7 with H₂ or H₂O as hydrogen sources, as well as oxidizing CH₄^8,9,10 with O_2, H₂O, or H₂O₂ into high-valued products and fuels independently (Fig. 1b, c). However, a redox-neutral direct co-reforming of these greenhouse gases into valuable chemicals remains both an intricate scientific challenge and highly desirable. The merits of this proposed negative emission method include not having to use oxidants and reductants, thereby minimizing the risk of undesired side reactions such as overoxidation or over-hydrogenation. Toward this objective, while dry reforming technology^11,12,13 and direct co-conversions^14,15,16 undoubtedly offer a promising reaction model for the synthesis of syngas or acetic acid, the dry reforming process primarily generates syngas, which requires subsequent chemical transformations such as the Fischer–Tropsch (FT) process into valued chemicals (Fig. 1d). Moreover, even with cutting-edge dry reforming technology, significant challenges persist in identifying coke-resistant materials that can maintain optimal performance over the long term for syngas synthesis. The same issues are also encountered in the co-conversions of CH₄ and CO₂ for direct synthesis of acetic acid.

Fig. 1: Investigating the challenges and approaches for transformations of CO₂ and CH₄.

a The challenges in attaining a net-zero-carbon footprint. Previously established pathways of converting b CO₂ and c CH₄, facilitated by separate reduction and oxidation processes. d Co-conversion process of CO₂ and CH₄ for the production of valuable chemicals. e This work presents an oxygen-atom-graft process from CO₂ to CH₄ via photo-interfacial catalysis.

The escalating global energy demands are driving the speedy advancement of the novel energy industry. With the increasing maturity of CH₃OH-related energy storage and release technologies (i.e., fuel cells and combustion engines), green CH₃OH has become an important liquid energy storage and transportation carrier, thanks to its high energy density (726 kJ mol⁻¹)^17,18. The growing demand for CH₃OH provides an unprecedented opportunity to explore more sustainable and atom-efficient methodologies^19,20. Conceptually, a much more appealing approach involves the co-conversion of CH₄ and CO₂ directly into CH₃OH without the input of other chemicals. However, making this approach a powerful tool for producing CH₃OH, it is imperative to break an unfavorable uphill energy barrier with high reaction temperature. Such harsh conditions would exacerbate side reactions and coking. In this context, photocatalytic systems, especially heterogeneous photoactive surfaces where photoexcited electron-hole pairs function as active sites, are potentially powerful platforms for co-converting CO₂ and CH₄. However, despite tremendous advances achieved in light-driven CH₄ reforming conversion with CO₂, most research remains extensively focused on syngas formation. Furthermore, a co-conversion approach may be facilitated by interfacial photocatalysts comprising metals and semiconductors, ranging from early ZrO₂/MgO/Ga₂O₃, TiO₂, and C₃N₄ to recently developed LaNiO₃^{21,22,23,24,25}. Unfortunately, few reports associated with targeting photosynthesis of CH₃OH have been presented so far, mainly due to the intrinsic difficulties in pre-absorbing and subsequently activating both CH₄ and CO₂²⁶. We hypothesized that the key to overcoming such a challenge is to identify a heterogeneous photocatalyst that can activate CH₄ and CO₂ on its surface, and simultaneously break the C–H and C=O bonds under mild conditions. This process would form CH₃OH upon their recombination, resulting in the graft of an oxygen atom from CO₂ to CH₄.

Since the first example of GaN-enabled CH₄ photoconversion in 2014, our research group has been exploring the C–H functionalization catalyzed by the GaN surface. The studied systems have involved the deposition of a range of cocatalysts, namely Pd, Cu, Pt, and RhO_x, facilitating the production of chemicals such as toluene, benzene, and cyclohexane^27,28,29. In these well-examined cases, GaN typically serves two roles: (1) acting as a photosensitive solid ligand to activate the metal co-catalyst; and (2) directly initiating C–H bond cleavage on the photoexcited surface. In addition, it is well documented that CO₂ can also be polarized and undergo extensive photo-reductive conversions on GaN powder or nanowires^30,31. We postulated that combining metal nanoparticles as cocatalysts with GaN as photo-promotor would forge an efficient metal/GaN interface that will enable direct CH₃OH photosynthesis from CH₄ and CO₂. Herein, we report the direct and highly selective synthesis of CH₃OH from CH₄ and CO₂ under light illumination. The CH₃OH productivity achieved in this study reaches up to 1405 μmol g⁻¹ h⁻¹, thus overperforming existing co-reformings of CH₄ and CO₂ for syngas synthesis reported in the literature by a factor of 3.8.

Results

To commence our study, we carried out the photocatalytic reaction using various heterogeneous photo-catalysts in a handcrafted chamber illuminated by full-spectrum light (Fig. 2 and Table S1). The control experiments revealed that the combination of CH₄ and CO₂ at room temperature did not yield any detectable CH₃OH in the absence of light or catalysts. Utilizing commercial GaN powder as a catalyst led to a higher methanol rate of 41 μmol g⁻¹ h⁻¹ than other semiconductors traditionally used for photo-CH₄ dehydrogenation and CO₂ reduction. This finding underscores GaN’s exceptional potential for efficient photocatalytic conversion of CH₄ and CO₂ into CH₃OH.

Fig. 2: Methanol production through co-conversion of CH₄ and CO₂ using an array of semiconductors and metal/GaN-cr interfacial catalysts.

The rates of methanol production were assessed under initial reaction conditions using 1 mg of catalysts with irradiation from a 300 W full-spectrum Xenon lamp at room temperature and ambient pressure.

To further boost the reaction efficiency, air-stable single-metal particles, including Ir, Ru, Pt, Rh, Pd, and Au, were in situ deposited on the GaN support via a universal chemical-reduction method (Fig. S1), forming an active interface denoted as metal/GaN-cr. Among these, monometallic interfaces featuring Au and Pd nanoparticles provided a notable increase in photocatalytic activity. In the field of both traditional thermal and sustainable-energy-driven reactions, bimetallic nanoparticles can remarkably augment catalytic performance in terms of conversion and selectivity, a phenomenon described as a 1 + 1 > 2 synthetic effect^32,33. Motivated by this encouraging principle to improve reactivity, we experimented with 0.2 wt% metal loading of both physical mixture of Au with Pd and their alloy in varying weight ratios under standard conditions. Further, increasing metal loading to 0.5 wt% and even 1 wt% did not increase the CH₃OH yield (Table S2). Thus, 0.2 wt% loading of alloyed AuPd/GaN interface with a 1-to-1 weight ratio eventually offers us the best CH₃OH production rate among all tested samples.

With this AuPd/GaN-cr catalyst in hand, we first characterized the nano-sized distribution of metal particles. Transmission electron microscopy (TEM) images of AuPd/GaN-cr demonstrate a suboptimal metal dispersion in bulky GaN support, a consequence of the in situ chemical reduction preparation (Figs. S2 and S3). To improve this apparent drawback of metal aggregation, we employed a soft colloid-immobilization process. This method enabled the assembly of the AuPd alloy with GaN, resulting in evenly distributed metal nanoparticles coupled with intact and highly active GaN (denoted as AuPd/GaN-ci)—as opposed to the etching by base or acid (Fig. 3a). As can be seen from a typical high-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM) image and its size population analysis of AuPd/GaN-ci (Figs. 3b–e and S4), the average diameter of 3.59 nm of metal colloid particles has been evenly immobilized on the GaN surface. The well-established bimetallic composition of supported AuPd alloy was also confirmed by the following energy dispersive X-ray (EDX) spectroscopy analysis. More importantly, despite successive load-wash-dry catalyst processing treatments, the composition and crystalline structure of the GaN host remained largely unchanged, as reflected by the X-ray diffraction (XRD) results (Fig. S5). This implies the great prospect of GaN powder as active support in photocatalysis industrial settings. The TEM and high-resolution TEM (HRTEM) results clearly illustrate the well-dispersion and formation of active metal/GaN interface, respectively (Fig. 3f, g). In addition, the enhanced UV–vis photo-adsorption property (Fig. S6) in UV and visible range of AuPd/GaN-ci might arise from interband absorption and plasmonic effect on metallic AuPd nanoparticles’ surface, which was further confirmed in the results of Au 4f and Pd 3d X-ray photoelectron spectroscopy (XPS) (Fig. S7).

Fig. 3: Construction and characterization of AuPd/GaN-ci.

a Schematic diagram of preparation for AuPd/GaN-ci. b HAADF-STEM, c–e elemental mapping, f TEM, and g HRTEM images of AuPd/GaN-ci. The inset of b and c shows the corresponding nanoparticle size distribution and selected area of EDX spectrum for AuPd/GaN-ci, respectively.

Taking such an interfacial AuPd/GaN-ci as the best-in-class photocatalyst, we proceeded to evaluate its photocatalytic redox behavior relative to AuPd/GaN-cr under initial reaction conditions shown in Fig. 2. Notably, the CH₃OH production rate over AuPd/GaN-ci can increase to a high value that is 2.6 times greater than that attained by AuPd/GaN-cr (Fig. 4a and Table S3). Detailed analysis further confirmed that CH₃OH and CO are the predominant products in oxygenates and gas phases, respectively (Fig. S8 and Table S4). The distinctive disparity in reactivity between AuPd/GaN-ci and bare GaN, especially in CH₃OH production estimated to be 34 times higher, affirms the critical role of AuPd/GaN interface in the photocatalytic co-conversion of CH₄ with CO₂ to CH₃OH (Fig. 4b). Such inference is further consolidated by a giant leap in CH₃OH yield from unsupported AuPd particles to AuPd/GaN-ci (Fig. 4b and Table S5). These findings inspired us to embark on trials with additional interfacial catalysts comprising alloy AuPd and various representative semiconductors. However, it did not lead to further enhancement of CH₃OH yield, despite the wide-band structures of Ga₂O₃, ZnO, and TiO₂ being known for direct methane activation (Fig. 4c and Table S6). These results suggest that the interfacial AuPd/GaN surface possesses a unique catalytic efficacy in converting CH₄ and CO₂ into CH₃OH.

Fig. 4: Catalysts and conditions optimization of photoinduced conversion of CH₄ with CO₂.

a Methanol and CO yield rates over Au/GaN-ci, AuPd/GaN-ci, and Pd/GaN-ci under standard conditions. b The importance of AuPd/GaN-ci interface in photoconversion of CH₄ and CO₂. c The effect of different wide-band semiconductors on reactivity. d The methanol production of AuPd/GaN-ci under different ratios of CH₄ and CO₂ gas mixture, and e different pressures. f Performance comparison of AuPd/GaN-ci with the reported photocatalytic systems in the literature. (The references are listed in the Table S9) Standard reaction conditions: 2.5 mmol of CH₄, 2.5 mmol of CO₂, 1 mg of catalyst, full-spectrum Xenon lamp, 2 h.

Considering the intrinsic nature of this novel redox transformation, involving CH₄ and CO₂ simultaneously, we investigated the impact of varying CH₄/CO₂ ratio on CH₃OH production while maintaining total gas pressure of 1 bar (Fig. 4d and Table S7). Our data analysis revealed that an increase in the proportion of either CH₄ or CO₂ resulted in lower CH₃OH yield and selectivity, demonstrating that a 1:1 ratio of CH₄/CO₂ is most favorable for our designed photocatalytic system. Moreover, when submitting this photoconversion to low pressure of 0.2 atm, it still achieved a CH₃OH productivity of 435 μmol g⁻¹ h⁻¹, a rate that aligns with the record reported in prior studies for CH₃OH (Fig. 4e and Table S8). More importantly, the yield rate of 1405 μmol g⁻¹ h⁻¹ obtained under ambient pressure exceeded those of other conventional photocatalytic syngas-based dry-reforming systems (Fig. 4f and Table S9). However, such active AuPd/GaN surface begins to give a reduced CH₃OH yield after extensive reuse (Table S10), likely due to either unavoidable oxidation of metal cocatalyst with O species from CO₂ (Figs. S9 and S10) or slow aggregation of metal (Fig. S11). In spite of this, the robustness and applicability of the photocatalytic system toward CH₃OH production under batch reactor was demonstrated by reusing over 10 h with a total CH₃OH yield of 13.66 mmol g⁻¹ and a turnover number (TON) of 941, with no noticeable carbon depositions (Table S10 and Fig. S12).

With such an improved CH₃OH productivity, we move forward to understand the reaction mechanism under light-driven conditions, which appears to be very different from the well-known process of dry-reforming of CH₄ and CO₂. Overwhelming research on CH₃OH synthesis finds that CO₂, CO, and CH₄ can each serve as the carbon source, reacting with appropriate reductants or oxidants^2,34. In our analysis of gas products, CO was almost the exclusive product with negligible amounts of H₂ and other hydrocarbons (Figs. S13 and S14 and Table S4). This observation effectively eliminates the significant impact of CH₄ dehydrogenation as a side reaction³⁵. Given the previous indirect pathway from CH₄ (CO₂) to CH₃OH via intermediates (H₂, CO, O₂) (Fig. 5a inset and Table S11), we examined the reactivity of AuPd/GaN-ci in control experiments wherein identical ratios of H₂ + CO₂, H₂ + CO, and CH₄ + CO as reactants were used. As depicted in Fig. 5a, the control systems all showed much lower CH₃OH yield. These experiments ruled out the potential pathway in which CO₂ or CO, acting as the main carbon source, reacts with H₂ – which could be possibly generated from CH₄—to form CH₃OH. Furthermore, replacing CH₄ + CO₂ with CH₄ + O₂ in a 10:1 ratio, where the number of oxygen atoms exceeds those theoretically generated from CO₂ to CO (quantified as 6.6 μmol), did not prompt CH₄ conversion to CH₃OH. The result ruled out the CH₄-oxidation with O₂, possibly derived in situ from CO₂ cleavage, as the primary mechanism for CH₃OH production. On all accounts, it is induced that this unique photocatalytic CH₃OH synthesis is mostly likely to undergo a direct transformation of CH₄ and CO₂ through surface catalysis rather than an indirect route mediated by intermediate such as CO, O₂, or H₂.

Fig. 5: Mechanistic and kinetic study profiles for photoconversion of CH₄ and CO₂ over AuPd/GaN-ci.

a Generation rates of methanol from photoconversion of 2.5 mmol H₂ + 2.5 mmol CO₂, 2.5 mmol H₂ + 2.5 mmol CO, 2.5 mmol CH₄ + 2.5 mmol CO, and 2.5 mmol CH₄ + 0.25 mmol O₂ over AuPd/GaN-ci. b GC–MS spectra of CO produced from photoconversion of ¹²CH₄ + ¹²C¹⁶O₂ and ¹²CH₄ + ¹³C¹⁶O₂. c GC–MS spectra of CH₃OH produced from photoconversion of ¹²CH₄ + ¹²C¹⁶O₂, ¹³CH₄ + ¹²C¹⁶O₂, and ¹²CH₄ + ¹²C¹⁸O₂. d Kinetic isotope effects for photoconversion of CH₄ with CO₂. e The generation rate of methanol from CH₄ photoconversion with CO₂ over AuPd/GaN-ci under photo-driven and thermal catalytic conditions. The inset of e shows the Arrhenius plot of methanol photosynthesis over AuPd/GaN-ci.

In the next study, the ¹³C-labeled experiment was implemented to further ascertain the carbon source of products, sketching the specific functions of the two gases in the overall reaction. When using ¹³CO₂ instead of ¹²CO₂, the characteristic peak of CO at m/z = 28 shifted up to m/z = 29 (Fig. 5b). In conjunction with the reductive control experiment discussed earlier, it is demonstrated that the majority of formed CO comes from CO₂. In the presence of ¹³CO₂, there is no detectable ¹³C labeled CH₃OH (m/z = 33) by either NMR or GC–MS analysis, indicating that CO₂ was not the carbon source of CH₃OH (Fig. S15). Conversely, in the experiment with ¹³CH₄, ¹³CH₃OH signals were detected by both MS and NMR, thereby confirming the incorporation of CH₄ as the primary carbon source and CO₂ as the oxygen source (Figs. 5c and S16). To further trace the oxygen source in the methanol product, we conducted an O¹⁸ labeling isotopic experiment, where the generation of O¹⁸ labeled CH₃OH coincides with the use of O¹⁸ labeled CO₂ as the reactant. This evidence directly reveals the contribution of CO₂ to the oxygen in CH₃OH (Fig. 5c). Complementary to this synthesis approach, the determination of a high kinetic isotope effect (KIE) value of 3.79 and 4.88 for the CH₃OH formation over AuPd/GaN-ci and GaN samples reveals that the C–H cleavage in CH₄ is a critical component of the rate-determining step (Figs. 5d and S17). Based on all these findings, we proposed a distinctive transformation involving oxygen-atom-grafting (OAG) from CO₂ to CH₄, wherein CH₄ mainly contributes to the carbon for CH₃OH and CO₂ to the oxygen for CH₃OH in the reaction products.

To understand the role of the AuPd/GaN-ci interface in the catalytic transformation, we collected and analyzed the UV-vis diffraction and Photoluminescence (PL) spectra of various samples, including GaN, Au/GaN-ci, AuPd/GaN-ci and Pd/GaN-ci (Figs. S6 and S18). As UV–vis comparative results showed, introducing mono- or bimetallic metal catalysts did not significantly improve light adsorption in pristine GaN. Furthermore, the only PL characteristic peak observed across all GaN samples with or without metal demonstrates that the photoabsorption at the metal/GaN interface mainly originates from GaN semiconductor, rather than the metal particles. This consistency in PL peaks and UV-vis results suggests a negligible contribution of light-excited electrons from the metal alloy, which might otherwise participate in the reactant’s activation^36,37. On the other hand, the diminished absorption of GaN at wavelengths above 400 nm also explains the reduced reactivity under visible light conditions (Table S12). Meanwhile, thermographic photograph measurements demonstrated that this surface photocatalysis reaction occurred at 25 °C (Fig. S19). This result confirmed the absence of a concomitant photothermal effect upon light irradiation. All these analyses identify the synergy between photoexcited GaN and alloyed AuPd, forming an active photoelectron-transfer interface that potentially enables energy and electron transfer with CH₄ and CO₂. Such synergistic photo-active interface thus endows the OAG process with a lower activation energy (60 kJ mol⁻¹) than GaN (72 kJ mol⁻¹) alone (Fig. 5e inset and S20), as well as a higher apparent quantum efficiency (Table S13). Moreover, the negligible CH₃OH yield under thermally driven conditions implies the AuPd/GaN-ci interface’s effectiveness in overcoming the thermodynamic challenges of traditional thermo-catalysis in OAG conversion (Fig. 5e and Table S14). The successful application of interfacial AuPd/GaN-ci in the OAG process introduces a new perspective on using photoexcited metal/semiconductor interfaces as a versatile toolbox in advancing sustainable chemical conversions.

The further experimental findings that the molar ratio of CH₃OH to CO is markedly increased on the combined AuPd/GaN-ci surface compared to the single GaN surface point toward the substantial synergistic effect of the AuPd/GaN-ci surface on activating both CH₄ and CO₂ (Fig. 6a inset). The interaction of CH₄ and CO₂ with the GaN surface has been recognized to effectively activate inert molecules for conversions^19,28,30. To further elucidate this synergistic effect, density functional theory (DFT) calculations were thus performed to understand preliminary interaction behaviors that are critical for catalytic processes on GaN and AuPd/GaN-ci surfaces. The analysis commenced with the optimized configurations of the two surfaces interacting with CO₂. Our DFT results show that the originally linear CO₂ molecule adopts a bent structure with elongated C=O bonds upon adsorption on the polar GaN surface (Fig. 6b inset). This bending is essential for CO₂ activation, with the O–C–O angle notably decreasing on the synergistic AuPd/GaN-ci surface. That significant variation in adsorption mode, compared to GaN alone, likely results from the apparent steric hindrance at the metal/GaN interface, as evidenced by the nonbonding interaction between the metal and oxygen in CO₂ in both the adsorption model and corresponding differential electron density diagram (Tables S15 and S16). Consequently, the AuPd/GaN-ci surface exhibits more negative adsorption energy for CO₂, indicating stronger affinity and enhanced CO₂ activation. As a comprehensive result of such interaction, the evident electron transfer, particularly between CO₂ and both metal and GaN surfaces, implies that adsorbed CO₂ on the AuPd/GaN-ci surface is more likely to dissociate and release CO upon charge transfer (Fig. S21).

Fig. 6: Calculation and experiment study profiles of pre-adsorption and activation processes for CH₄ and CO₂ on the surface of GaN and AuPd/GaN-ci samples.

a The rates of CH₄ and CO₂ consumption and (inset of a) produced CH₃OH-to-CO mole ratio for GaN and AuPd/GaN-ci. b The calculated adsorption energies (E_ads) for adsorbed CO₂ and c CH₄ on GaN and AuPd/GaN-ci surfaces along with the corresponding optimized stereograms. Color code: C: dark gray; N: light gray; O: red; H: light pink; Ga: green; Au: orange; Pd: purple; d the C–H bond length of CH₄ adsorbed on the comparative catalysts’ surface with the differential charge density stereograms (left, inset of d) for GaN and schematic of C–H bond activation in CH₄ for AuPd/GaN-ci (right, inset of d). The cyan and yellow regions in the electron density cloud diagram indicate electron loss and gain.

Based on the O¹⁸ labeling CO₂ experiment discussed above, it is plausible that the remaining oxygen atom from CO₂ is then positioned to potentially graft with activated CH₄ on the AuPd/GaN-ci surface to form CH₃OH. For the CH₄ adsorption structures, CH₄ preferentially locates closer to the synergistic AuPd/GaN-ci surface, resulting in lower adsorption energy for CH₄ (Fig. 6c). This observation suggests a pronounced pre-adsorption characteristic of CH₄ with low-polarizability over AuPd/GaN-ci. Even so, AuPd/GaN-ci surface with a considerable negative adsorption energy of CO₂ reveals a higher affinity and easier activation for CO₂ than CH₄, which might explain the higher consumption rate of CO₂ (Fig. 6a). Consistent with previous cases, our data show significant CH₄ polarization on the GaN surface, evidenced by the disparity in the electron density cloud and the elongation of the C–H bond (Fig. 6d inset left and Tables S17 and S18). Importantly, the AuPd/GaN-ci surface further intensifies the polarity of the C–H bond in CH₄, leading to even greater bond stretching compared to GaN alone (Fig. 6d inset right). These results support the hypothesis that the AuPd/GaN-ci surface amplifies the direct activation of CH₄ compared to the single GaN surface. Considering all of the presented experimental and DFT data above, the synergistic surface of GaN with metal on AuPd/GaN-ci can facilitate strong enough interactions with both inert CH₄ and CO₂ at the same time. That multifunctional capability makes it possible to drive the surface-bound intermediates pathway rather than the gas phase for the oxygen-atom of CO₂ grafting with CH₄, leading to the formation of CH₃OH and CO.

Lastly, it is well known for its enhanced visible-light absorption capabilities when noble metals are incorporated into semiconductors^38,39. For this reason, we decided to explore the potential of our synthesis for broader applicability in the visible light region, conducting this OAG reaction over GaN and interfacial AuPd/GaN-ci catalyst under the irradiation of a Xenon lamp with 435 nm wavelength of cut-off filter. Notably, the UV–vis absorption results confirm AuPd/GaN-ci’s better visible-light absorption compared to GaN (Fig. S6). As anticipated, the AuPd/GaN-ci gave a reappeared methanol yield rate in the OAG process while the unmodified GaN surface did not show reactivity (Table S19). Overall, this compelling observation presents the feasibility of our approach in CH₃OH photosynthesis from CH₄ and CO₂ across a wide solar spectrum.

In summary, we have discovered a direct and synergistic photocatalytic method for CH₃OH synthesis by co-reforming two major greenhouse gases of CH₄ and CO₂ under mild conditions with very high selectivity. This method employs an interfacial metal/semiconductor as a photo-redox catalyst at room temperature. The photoexcited interface between AuPd nanoparticles and GaN support ensures simultaneous C–H activation of CH₄ and CO₂ reduction to produce methanol via an oxygen-atom-graft from CO₂ to CH₄. The scalability of this transformation was demonstrated in a batch reactor by recycling the catalyst multiple times. Adapting the herein OAG process to continuous flow operations, such co-conversions of CH₄ and CO₂ could revolutionize industrial CH₃OH production, while also making a positive impact on climate change mitigation.

Methods

Chemicals and materials

Gallium (III) nitride (Thermo scientific, 99.99%), chloroplatinic acid hexahydrate (Sigma-Aldrich, ACS reagent), gold (III) chloride trihydrate (Sigma-Aldrich, ACS reagent), rhodium (III) chloride hydrate (Sigma-Aldrich, ACS reagent), iridium (III) chloride trihydrate (Acros organics), NaPdCl₄ (Sigma-Aldrich, 98%), zinc oxide (Sigma-Aldrich, nanopowder, <100 nm particle size), titanium (IV) oxide (nanopowder, ~21 nm particle size, 99.5%), gallium (III) oxide (Sigma-Aldrich, 99.99%), methane (Air Liquide, 99.99%), carbon dioxide (Linde), ¹³C labeled methane (Sigma-Aldrich, 99 atom % ¹³C), ¹³C labeled carbon dioxide (Sigma-Aldrich, 99 atom % ¹³C) and ¹⁸O labeled carbon dioxide (Sigma-Aldrich, 97 atom % ¹⁸O).

Synthesis of different metal/GaN-cr by in situ chemical reduction

The cocatalyst precursors used include H₂PtCl₆·6H₂O (0.531 mg), Na₂PdCl₄ (0.553 mg), HAuCl₄·3H₂O (0.400 mg), RuCl₃·xH₂O (0.410 mg), IrCl₃ ·3H₂O (0.367 mg) and RhCl₃ ·3H₂O (0.407 mg). For a typical preparation of 0.2 wt % metal/GaN-cr photocatalyst, 100 mg of GaN was dispersed into 25 mL of an aqueous metal solution by ultrasonication. Then the mixture was stirred for 0.5 h, followed by the dropwise addition of an aqueous NaBH₄ solution (0.15 M, 38.6 µL). After another 0.5 h of stirring, the samples were filtered and washed three times with water, then dried at 100 °C under vacuum overnight.

Synthesis of monometallic and bimetallic Au(Pd)/GaN-ci by a soft colloid-immobilization process

The typical bimetallic AuPd/GaN-ci sample was synthesized by following a well-established method for metal colloid preparation^40,41. In total, 200 μL of HAuCl₄ ·3H₂O (2.54 mM) and 277 μL of Na₂PdCl₄ (3.40 mM) aqueous solutions were dissolved in 25 mL of deionized water. Polyvinylpyrrolidone (PVP, 0.2 mg), with an average molecular weight of 1,300,000, was added as a stabilizer to give the desired metal-to-PVP ratio (typically set as 1:1). After 5 min of stirring, a freshly prepared 0.15 M sodium borohydride (NaBH₄) solution (38.6 µL) was introduced and maintained a NaBH₄-to-metal molar ratio of 5. The bimetallic AuPd colloid solution was produced after stirring for 30 min. Subsequently, 100 mg of GaN was added to immobilize these bimetallic metal nanoparticles over another 30 minutes. The as-prepared AuPd/GaN-ci catalyst was filtered, thoroughly washed with distilled water, and left to dry under vacuum at 100 °C overnight. As controlled samples, single Au/GaN-ci and Pd/GaN-ci were prepared using the same method, but with the use of 400 μL of HAuCl₄ ·3H₂O (2.54 mM) and 554 μL of Na₂PdCl₄ (3.40 mM) aqueous solution as metal precursors, respectively.

Material characterizations: The bright field transmission electron microscopy (TEM) and HAADF-STEM observations were carried out on FEI Tecnai G2 F20 S/TEM at an accelerating voltage of 200 kV. Powder X-ray diffraction pattern (XRD) measurements were collected on a Bruker D8 Advance X-ray diffractometer with a Cu-Kα radiation source (λ = 1.5418 Å) and operated with an increment of 0.02° and a counting time of 0.24 s under the voltage of 40 kV and 40 mA. X-ray photoelectron spectroscopy (XPS) measurements were conducted on a Thermo ESCALAB 250 spectrometer equipped with a monochromated Al Kα radiation (hv = 1486.6 eV). The energy of the spectrometer was calibrated by C 1s peak at 284.8 eV. Ultraviolet–visible (UV–vis) measurement was performed by Agilent Cary 5000 series UV–vis–NIR spectrometer. Photoluminescence (PL) measurement was performed with a 213-nm He–Cd laser (Kimmon Koha) as the excitation source.

The photocatalytic reaction of CH₄ with CO₂ in a quartz chamber

Firstly, 1 mg of the catalysts was evenly distributed on the bottom of a 120 mL quartz chamber reactor. To remove water and potential gases adsorbed on the catalyst surface, the reactor was degassed at 250 °C. Then, we introduced the CH₄ (2.5 mmol) and CO₂ (2.5 mmol) gases into the reactor at a 1-to-1 ratio, following pump-vacuum cycles three times. The reactor was placed in a double-layered beaker with a water bath and irradiated under a 300 W Xenon lamp with a PE300BF light bulb (Excelitas) for 2 h (Fig. S22). After the photocatalytic reaction, gas products were identified and quantified by GC-FID and GC-TCD analysis. Following that, 1 mL of deionized water was injected into the reactor to absorb oxygenated products. The resulting oxygenated products were then analyzed by ¹H-NMR for CH₃OH, CH₃OOH and HCOOH, and a colorimetric method was applied for HCHO quantification.

Quantification of oxygenated products

The oxygenated products, including CH₃OH, CH₃OOH, and HCOOH were analyzed and quantified via a proton nuclear magnetic resonance (NMR) spectroscopy spectrometer (Bruker AVIIIHD 500 MHz). Specifically, 0.4 mL of as-formed liquid that absorbed the oxygenated products was filtered and then mixed with 0.1 mL deuterium oxide containing 0.0167 μL of DMSO as an internal standard. The calibration curves for CH₃OH and HCOOH were established by plotting the concentration of standard chemicals against the ¹H-NMR area ratio of products to DMSO. For the quantification of HCHO, a colorimetric method was used. Here 0.3 mL of the as-formed liquid was diluted with 2 mL of deionized water. Subsequently, 0.3 mL of reagent solution was added to this diluted mixture. The final test liquid was placed in a 35 °C water bath for 1 h. The typical reagent solution was prepared by dissolving 15 g of ammonium acetate, 0.3 mL of acetic acid and 0.2 mL of pentane-2,4-dione in 100 mL of deionized water.

Gas products analysis

Quantification of CO was obtained by gas chromatograph (GC, Clarus 590) equipped with a methanizer and flame ionization detector. We used GC (Agilent 6890 N) with a thermal conductivity detector (GC-TCD) to quantify the H₂ amount. Then we used GC (Agilent 6890 N) with flame ionization detection (GC-FID) to identify and quantify hydrocarbon products (ethane).

Density function theory experiment

The photocatalytic reaction mechanism of CH₄ and CO₂ on controlled GaN and AuPd/GaN samples was investigated using the Vienna Ab-initio Simulation Package (VASP)^42,43. The Perdew–Burke–Ernzerh of generalized gradient approximation functional was employed for the exchange-correlation potential⁴⁴. The projected augmented wave potential was applied to describe the ion–electron interaction⁴⁵. A k-mesh of 5 × 3 × 1 was adopted to the primitive cell of GaN (100)-wurtzite configured as a 2 × 2 supercell. The van der Waals dispersion forces were considered using the zero damping DFT-D3 method of Grimme⁴⁶. The cutoff energy for the plane-wave basis was set to be 520 eV. The AuPd/GaN-ci catalyst’s simulation involves a metal cluster of three atoms placed on a GaN surface, with the Au to Pd ratio closely reflecting the trend observed in the EDX spectrum result (Fig. 3c). The structure was optimized until the forces on each ion were smaller than 0.01 eV Å⁻¹, and the convergence criterion for the energy was 1.0 × 10⁻⁶ eV. The adsorption energies of the substrate molecule adsorbed on the surface were determined using the following equation: E_ads = E_GaN-X − (E_GaN + E_X), where E_X, E_GaN, and E_GaN-X represent the total energies of adsorbate X, GaN, and adsorbate X absorbed on the GaN, respectively.