Highly selective catalytic oxidation of methane to methanol using Cu–Pd/anatase

Abstract

Direct conversion of methane into high value-added products is of great practical significance. The synergistic effect in catalysts with dual-active components show potential to increase the methanol yield and selectivity. In this work, Cu–Pd/anatase is in situ generated and exhibits a relatively high methanol yield rate of ∼31 800 μmol g_cat⁻¹ h⁻¹ and near-exclusive selectivity of liquid products (methanol). The reaction mechanism behind the heterogeneous catalysis process has been investigated. It is confirmed that copper ions hold the ability to produce hydrogen peroxide which can be further promoted by anatase. Chlorine ions can promote the stable adsorption of CO and the formation of *CH₃ intermediates, facilitating high activity and selectivity for methanol production. Pd and Cu cooperatively dissociate methane, which promotes the formation of key configuration metal-CH₃. The ˙CH₃ intermediate desorption will be facilitated on Cu–Pd/anatase through the manner of electron regulation, which is proved by the combination of density functional theory calculations and in situ infrared spectroscopy. Methanol is formed when a ˙CH₃ is desorbed from a copper site and combines with a hydroxyl radical.

---

Broader context

The selective oxidation of methane to methanol is essential for efficient energy utilization. However, the performance of selective methane conversion remains limited and requires further improvement. The main issues lie in the difficulty of activating methane and the subsequent tendency for the over-oxidation of methanol. Therefore, developing a strategy that directly participates in the methane activation and the desorption steps of key intermediates is expected to significantly increase the yield and selectivity of methanol. We design a Cu–Pd/anatase catalyst in which highly dispersed Cu species are generated through in situ reactions, and achieve a relatively high methanol yield rate of ∼31 800 μmol g_cat⁻¹ h⁻¹ with >99% selectivity in aqueous products under mild reaction conditions. We thoroughly investigate the synergistic effects in Cu–Pd/anatase by detailed characterizations and first-principles calculations. Thus, this manuscript expounds the key points of methanol generation and inspires new avenues in catalyst design for the selective conversion of methane to methanol.

Introduction

The direct conversion of methane into methanol is a desirable route for efficient energy utilization, with a wide potential for industrial applications.^1–4 Methane possesses a stable tetrahedral structure, therefore considerable energy is required to break the C–H bond, which may lead to subsequent peroxidation. Numerous efforts have been made throughout the years to prevent the over-oxidation of methanol, but this tends to occur more readily than methane activation.^5–7 Currently, the utilization of methane in the industry is as follows: the CH₄ is first converted into syngas via the water vapor reforming approach, and then the syngas is transformed into CH₃OH through the Fischer–Tropsch synthesis.^8,9 These procedures are complicated, energy-consuming, and require harsh reaction conditions. Consequently, direct CH₄ partial oxidation to C1 products under benign conditions with good activity and selectivity is particularly desired.

Using reductive gas to promote selective methane oxidation is a promising route for industry.^10–13 H₂O₂ could be in situ generated with the addition of CO that promotes the selectivity oxidation of methane,^12,14 and CO could participate in and enhance the reaction as a ligand.¹³ Although the yields of methane selective oxidation are well improved with the addition of the reductive gas, researchers have faced challenges in maintaining high selectivity while increasing the methanol yield. Thus, a strategy of incorporating copper ions into the solution to improve methanol selectivity has emerged. Hutchings et al. reported that the addition of Cu²⁺ would reduce the over-oxidation process of methanol because of a decrease in the concentration of hydroxyl radicals.¹⁵ Stephanopoulos et al. proposed that Cu ions improved the selectivity of methanol by suppressing the formation of formic acid.⁵ However, because these strategies participate in the reaction indirectly, the enhancement of selective methane conversion is limited. Therefore, developing a strategy that directly participates in the methane activation and the desorption steps of key intermediates is expected to significantly increase the yield and selectivity of methanol.

The dual active-component catalysts exhibit promising features in catalyzing methane to oxygenates. Copper is a classic element for methane selective oxidation. Methane monooxygenases (MMOs), a natural family of enzymes, possess the inherent capacity of selectively oxidizing the inert C–H bonds of methane via copper species.^16,17 Bioinspired catalysis has become a promising strategy, leading to the design of numerous catalysts containing Cu reactive species for both homogeneous and heterogeneous catalysis.^18–21 Pd-based catalysts show great capacity for oxidating CH₄ molecules because of the outstanding performance for hydrogen activation.^22–24 The previous researches showed the excellent ability of Pd species in methane activation.^25–28 At this point, dual-active components seem like an effective approach to combine the advantages of both species. Meanwhile, the potential synergistic effects among multiple components, which are essential for their higher efficiency and selectivity to methanol, need further investigation.

Herein, we design a Cu–Pd/anatase catalyst in which highly dispersed Cu species are generated through in situ reactions. A relatively high methanol yield rate of ∼31 800 μmol g_cat⁻¹ h⁻¹ with >99% selectivity in aqueous products is achieved under mild reaction conditions. To explore the reaction mechanisms, we investigate the synergetic effects between copper cations and Pd during the process of methane conversion to methanol. We report that copper cations not only generate H₂O₂ but also play an important role in the generation of methanol. The synergistic interaction between Cu and Pd species altered the reaction pathway of methane activation, promoting the desorption of methyl radical (˙CH₃), which is identified as a key step for the high yield and selectivity. The presence of chlorine species also facilitates the stable adsorption of CO and the generation of key *CH₃ intermediates. Accordingly, we report an effective strategy to enhance the selective oxidation of methane to methanol by modulating the multicomponent active sites, and the reaction mechanisms are then revealed.

Experimental

Catalysts preparation

Pd deposition onto the commercial anatase TiO₂ (Aladdin Shanghai, China) support was accomplished using the deposition–precipitation method at room temperature. 1 g L⁻¹ Pd(NO₃)₂ solution was dropwise added to an anatase suspension with a pH of 12. After being stirred for 12 hours, the resulting solid was filtered and dried at room temperature overnight. Subsequently, reduction of the solid was conducted using 10% hydrogen at 300 °C for 1 hour. The resulting catalysts, with Pd loadings of 0.1 wt%, 0.25 wt%, 0.5 wt%, 1.0 wt%, and 2.0 wt%, were denoted as 0.1 Pd/anatase, 0.25 Pd/anatase, 0.5 Pd/anatase, 1.0 Pd/anatase, and 2.0 Pd/anatase, respectively. The catalysts collected after the reactions were denoted as Pd/anatase-used.

Pd/anatase was suspended in CuCl₂ solution (30 mg CuCl₂·3H₂O dissolved in 20 mg water) and stirred for 1 h. Then the catalysts were filtered and dried at room temperature. The catalysts collected were denoted as Cu–Pd/anatase. Cu/anatase were the same except for replacing Pd/anatase with anatase. Pd/anatase was suspended in Cu(NO₃)₂ solution to prepare Cu–Pd/anatase-N with Cu(NO₃)₂ as a precursor using a similar method to the one used for Cu–Pd/anatase prepared with CuCl₂ as a precursor.

Results and discussions

Catalyst preparation and methane oxidation.

A Pd/anatase catalyst was first synthesized via the deposition–precipitation method, incorporating a Pd loading of 0.25 wt%. Evaluation of the as-prepared catalysts for CH₄ direct oxidation was conducted under 150 °C with 20 bar CH₄, 3 bar O₂, and 5 bar CO purge then charged at room temperature in aqueous phase with copper cations. X-ray diffraction (XRD) patterns show no large Pd nanoparticles or phase changes existing after calcination (Fig. S1, ESI†). Aberration-corrected high-angle annular dark-field scanning transmission electron microscope (AC-HAADF-STEM) was employed to elucidate the states of Pd and adsorbed Cu species. Fig. 1a and b show plentiful Pd₁ single atoms on Pd/anatase and Pd/anatase-used surface. Since the atomic number (Z) contrast image-forming concept is used in the AC-HAADF-STEM characterization technique, the intensity of metal atoms is roughly proportional to the Z^1.7.²⁹ The intensity ratio of Pd₁ to Cu₁ should theoretically be 2.2 in this case. In Fig. 1c and Fig. S2 (ESI†), the ratio of signal intensity is around 2.0, confirming the existence of Cu species near Pd atoms. The significant decrease of Cu content in the reaction in Table S1 (ESI†) means the precipitation of copper ions on the catalyst surface. The EDS mappings also show that Pd and Cu species are well-distributed on the anatase (Fig. 1d and Fig. S3, ESI†). Electron paramagnetic resonance (EPR) measurements (Fig. S4, ESI†) exhibit characteristic signals at g-values of ∼1.9997 assigned to oxygen vacancies.^30,31 To investigate the catalyst capability of oxygen activation, O₂ TPD was also conducted in Fig. S5 (ESI†). The desorption peaks at 303 and 341 °C are assigned to surface chemisorbed oxygen species.³⁰ Pd/anatase shows more surface oxygen adsorption at 663 °C, which suggests a greater capacity of oxygen activation owing to the enriched surface oxygen vacancies. The existence form of Pd and Cu elements needs to be further verified by XAS (Section S1, ESI†). The Pd species on the catalyst surface exist as both highly dispersed Pd atoms and as Pd clusters and the Cu species undergo charge transfer upon loading. XAS analysis confirms that Cu exists predominantly as Cu⁺, with no detectable Cu⁰.

Fig. 1 The AC-HAADF-STEM images of Pd/anatase (a) and Pd/anatase-used (b). Pd single atoms are highlighted with yellow circles. (c) Image intensity line profiles of areas #1-2 in (b), in which the brightest dots are assigned to Pd species and the second brightest dots are assigned to Cu species. (d) The EDS mapping for Pd/anatase-used, O 1s (e) Ti 2p (f) and Pd 3d (g) results of Pd/anatase and Pd/anatase-used.

X-ray photoelectron spectroscopy (XPS) was employed to assess the surface properties of the catalyst and reveal the interaction between copper ions and Pd during the reaction process. Fig. 1e manifests that the lattice oxygen in TiO₂ (O_L), terminal hydroxyl groups (O_OH), and surface adsorbed water (O_W) are all identified as peaks in the O 1s XPS spectra for Pd/anatase and Pd/anatase-used at around 529.8, 531.9, and 533.3 eV.³⁰ The O species fractions for O_L, O_OH, and O_W in Pd/anatase are 83.8%, 11.2%, and 5.0%, whereas these corresponding fractions in Pd/anatase-used are 71.1%, 21.3%, and 7.6%. On the (101) surface of anatase, molecular water is known to preferentially dissociate on oxygen vacancies and generate surface OH groups. The reduced proportion of lattice oxygen is almost the same as the increased proportion of the surface hydroxyl group, indicating that water dissociates at the oxygen vacancies generated by lattice oxygen desorption.³² In the Ti 2p XPS comparison (Fig. 1f), the Ti 2p_3/2 and Ti 2p_1/2 peaks at 458.64 and 464.31 eV in Pd/anatase are attributed to Ti⁴⁺–O bonds. In the Pd/anatase-used catalyst, the Ti 2p_3/2 peak shifts negatively to 458.57 eV, manifesting an increase in Ti³⁺ content. As shown in the Cu 2p XPS spectrum (Fig. S6, ESI†), a clear Cu⁺ peak is detected at 932.33 eV. Commonly, adsorbed Cu species would act as electron donors, transferring electrons to Pd and TiO₂ carriers. In the Pd 3d XPS graph (Fig. 1g), the signal for PdO species weakens after the reaction, while low-valence Pd species are present. There are two possible reasons for this: (1) detachment of PdO species, or (2) reduction of PdO. Considering that the experiments in Table S1 (ESI†) have confirmed that the phenomenon is not due to the leach of Pd, it is more likely that the PdO species have been reduced. Due to CO being a typical reducing gas, Pd is reduced by CO during the reaction, and Pd⁰ is more stable in a reductive environment. Because the Pd²⁺ signal is below the detection limit in the XPS analysis, AC-HAADF-STEM characterization was performed on Cu–Pd/anatase-used. The results in Fig. S2 (ESI†) indicate that there are numerous Pd single atom sites in the used catalyst, suggesting that Pd has not undergone significant aggregation.

The conversion of methane catalyzed by Pd/anatase with the assistance of Cu ions shows high activity and selectivity in Fig. 2. After raising the temperature, the methanol yield increases at the same time and reaches a maximal value of ∼31 800 μmol g_cat⁻¹ h⁻¹, and then decreases commonly due to methanol overoxidation at high temperatures (Fig. 2a). The total methanol selectivity among all products of CH₄ transformation is 86.7%, and the CH₃OH selectivity in the aqueous phase is >99% with no other liquid product detected in ¹H NMR spectra (Fig. S7, ESI†). Table S2 (ESI†) compares the selective methane oxidation performance of Cu–Pd/anatase with well-performed literature data. In order to facilitate comparison with other reported works, we adopted a reaction temperature of 150 °C. The comparison indicates that Cu–Pd/anatase exhibits a high methanol formation rate and selectivity. The methanol yield decreases on a longer time scale, which is due to the rapid consumption of CO (Fig. S8, ESI†). With the increase of Pd mass fraction, a slight rise in methanol yield in Fig. 2b. Pd single atom sites existed in large quantities at Pd loadings of 0.1 wt% and 0.25 wt%, while significant clusters formed at loadings of 0.5 wt% and above. A reasonable explanation for the superior performance of the 0.25 wt% loading compared to the 0.1 wt% is that the increased loading led to a higher number of Pd single atoms, while at 0.5 wt%, aggregation of Pd resulted in decreased methanol yield. This indicates that Pd single atoms are the primary Pd active sites.

Fig. 2 Catalytic performances for methane oxidation to methanol. (a) The yields of methanol obtained at different reaction temperatures. (b) Effect of Pd mass fraction over the Cu–Pd/anatase catalyst. (c) Stability of Cu–Pd/anatase catalysts in the methane oxidation condition. (d) Performances of Cu²⁺, Cu/anatase, and Cu–Pd/anatase. Standard reaction conditions for this work unless otherwise stated were: total pressure 28 bar (CH₄ : O₂ : CO = 20 : 3 : 5), 20 mL H₂O, 30 mg CuCl₂·2H₂O, react at 150 °C.

To illustrate the effect of Pd and Cu species in promoting the production of methanol, tests were carried out using a Pd/anatase-used catalyst (Fig. S9, ESI†). Without the addition of Cu ions, Pd/anatase-used produces more methanol than that of Pd/anatase catalysts, but still less than Pd/anatase with Cu ions, indicating that residual copper ions on the surface serve as active sites. After introducing Cu ions, the yield of the used catalyst is comparable to that of the anatase without Pd, demonstrating that many Pd active sites are blocked. To prove the coverage of Pd sites by copper residuals, the used catalyst was suspended in a hydrochloric acid solution at a pH of approximately 3 to dissolve surface copper species for catalyst regeneration. There are Pd species but lots of Cu species in the solution detected by ICP experiments (Table S1, ESI†). The results of the repeated tests (Fig. 2c) demonstrate that catalyst activity is restored to its original level. It has been certified that the deactivation of Pd/anatase is due to excessive copper species enveloping the Pd active site. Furthermore, it is also vindicated that there is indeed a significant interaction between Cu and adjacent Pd species.

Roles of copper cations and chlorine anions in methane oxidation to methanol

Since the methanol generation with the cooperation of copper cations in solution remains to be explored, experiments have been designed to investigate how the copper cations participate in the reactions.^5,12,15 The comparative experimental results are shown in Fig. 2d. Moderate methanol is produced when only copper ions are present, and the subsequent addition of anatase and Pd further benefits methanol generation. As depicted in Fig. S10 (ESI†), the level of Cu²⁺ concentration has a positive correlation with the rate of methanol production, while the Pd loading trend had a relatively weak influence on the methanol generation. This leads to the speculation that copper ions will be crucial in the generation of methanol. Copper species not only serve as active sites for the synthesis of H₂O₂ but are also one of the most widely used active species in methane oxidation to methanol process.^33–37 A plausible reaction route has been proposed that copper ions could participate in methanol production via facilitating the in situ generation of H₂O₂. As is shown in Table S3 (ESI†), H₂O₂ can be in situ generated in the presence of CO, O₂, H₂O, and copper cations, evidenced by DMPO–OH trapping experiments (Fig. S11, ESI†). The role of reactants in the case of only copper ions in solution has been explored (Table S4, ESI†). O₂ is the essential oxidant to produce methanol over Cu–Pd/anatase catalyst, especially the addition of CO would largely magnify its yield as well. Worth noticing, that the replacement of O₂ with H₂O₂ as the oxidant (Table S4, entry 6, ESI†) would intensely promote CH₃OH production indicating the possible reaction mechanism involving H₂O₂ or ˙OH radicals. This manifests that CO not only takes part in the production of H₂O₂ but also continues to participate in other reaction steps, which will be explored later.

The influence of anion has also been taken into consideration. The comparison between CuCl₂ and Cu(NO₃)₂ was used to explore the influence of chlorine on the reaction for the generation of H₂O₂ and partial oxidation of CH₄ to methanol. As is shown in Table S5 (ESI†), there is an apparent difference in CH₃OH yield between CuCl₂ and Cu(NO₃)₂ (entries 1 and 2). With the addition of Pd/anatase, both performances of CuCl₂ and Cu(NO₃)₂ have been improved (entries 3 and 4), but the yield distinction still exists between Cl⁻ and NO₃⁻. After adding NaCl to Cu(NO₃)₂ solution (entry 7), the CH₃OH yield upgrades to the level of CuCl₂. It has been confirmed that chlorine does promote the yield of CH₃OH in the presence of Cu species. The replacement of CO/O₂ with H₂O₂ as the oxidant (entries 9 and 10) would intensely promote CH₃OH production. The high selectivity of Cu(NO₃)₂ in liquid products (Fig. S12, ESI†) manifests that the anions do not change the reaction route of methane oxidation. From the perspective of H₂O₂ generation and methane conversion, the influences of chlorine species are explored. According to the literature chlorine can affect the generation of H₂O₂, the effect of anions on the generation of H₂O₂ has been explored in Table S3 (ESI†).³⁸ Compared with Cu(NO₃)₂, CuCl₂ does promote the generation of H₂O₂. And the CH₃OH yields have a positive correction with H₂O₂ yields both in the case of CuCl₂ and Cu(NO₃)₂. More detailed experiments were conducted on the relationship between chlorine concentration and H₂O₂ yield, as well as methanol yield (Tables S3 and S5, ESI†). Both the H₂O₂ yield and methanol yield increase with higher chlorine concentration. The impact of chlorine ions on CO concentration and selectivity of CH₄ to CH₃OH is detected in Table S6 (ESI†). The results indicate that the presence of chlorine promotes the conversion of CO to CO₂, which is consistent with the generation of H₂O₂ from CO. The adsorption of CO was characterized by physically adsorbing CO at 0 °C using the Autosorb iQ Automated Gas Sorption Analyzer, and the results are shown in Table S7 (ESI†). The results indicate that chlorine promotes the physical adsorption of CO on the catalyst surface. With the addition of chlorine ions, the selectivity of methane towards methanol increases. The role of chlorine will be further explored in subsequent investigations.

To unravel the involvement of copper cations in the reaction, the chemical state of copper species and intermediates was revealed by in situ DRIFTS of reaction gas adsorption (Fig. S13, ESI†). Gas CO is located at 2171 and 2120 cm⁻¹.³⁹ The peak at 2112 cm⁻¹ is assigned to CO–Cu⁺ species, indicating that Cu²⁺ ions are reduced to Cu⁺ ions by CO and they are Cu⁺ ions that participate in the following reactions.^40–42 The weak adsorption of methoxy (CH₃O*) in 2824 cm⁻¹ suggests the dissociation of the C–H bond in the CH₄ molecule.^43–45 Based on the results from experiments and in situ DRIFTS measurement, the conclusion has been drawn that copper ions are the critical reaction species, which could activate CH₄ and eventually generate methanol.

Direct oxidation of methane to methanol over Pd/anatase

To uncover the underlying factors affecting product formation, experiments have been designed to explore the influence of anatase and Pd atoms. Fig. S14 (ESI†) shows that anatase observably improves the yield of H₂O₂, consistent with the improvement of methanol yield. Different crystalline forms of TiO₂ are also taken into consideration, whose methanol yields are consistent with their H₂O₂ yields. DRIFTS of CO chemisorption measurements were conducted on CuCl₂ and Cu/anatase to verify the effect of the interaction between copper ions and anatase (Fig. 3a). On CuCl₂, besides the CO gas (2171 and 2120 cm⁻¹) and CO–Cu⁺ species (2129 cm⁻¹), there are two peaks at 1344 and 1241 cm⁻¹ assigning to bidentate formate and COOH* species respectively, manifesting that CO is activated on Cu species with surface hydroxyl group and adsorbed oxygen.^31,46,47 It was reported that CO reacts with water to produce H₂ and then reacts with O₂ to produce H₂O₂.¹² However, no H₂ was detected in the gas products, so CO directly reacted with H₂O and O₂ to produce H₂O₂. COOH* species has been reported to be the intermediates of the H₂O₂ generation, corresponding with the experiments in Table S3 (ESI†).⁴⁸ On anatase, there is only one adsorption peak of CO, and the intensity of other peak positions is very inconspicuous, indicating that TiO₂ cannot activate CO to generate other species. After loading CuCl₂ on anatase, there is an evident increase in the intensity of COOH* species and a new peak at 1300 cm⁻¹ assigned to bridged-CO₃*. Besides, the CO–Cu⁺ peak at 2119 cm⁻¹ on Cu/anatase shows a blue shift up to ∼10 cm⁻¹. It has been reported that when CO is adsorbed onto metal atomic sites, the metal atoms lose electrons, resulting in reduced charge transfer to the adsorbed CO. This leads to a blue shift in the stretching frequency of CO.⁴⁹ We used Ti 2p XPS to demonstrate the electron transfer from Cu species to the surrounding atoms (Fig. S15, ESI†). The Ti 2p peak position for anatase is at 458.50 eV, while the peak for Cu/anatase is at 485.44 eV. This indicates that electron transfer has occurred from Cu to Ti. Meanwhile, an obvious CO₂ peak emerged at 2315 cm⁻¹, which could be explained that Cu species interact with surface oxygen atoms, facilitating the reaction of surface-active oxygen species with CO. Therefore, copper ions transfer charge to anatase support, thus promoting CO activation, generating more COOH* species, which is a crucial intermediate product in the generation of H₂O₂, ergo enhancing H₂O₂ generation. We prepared Cu–Pd/anatase-N with Cu(NO₃)₂ as a precursor using a similar method to the one used for Cu–Pd/anatase prepared with CuCl₂ as a precursor. The effect of chlorine on copper species was characterized by XPS, and the results are shown in Fig. S15 (ESI†). In the XPS spectra, we detected a weak chlorine signal, indicating the presence of a small amount of chlorine species on the surface. With the introduction of chlorine species, the Cu 2p XPS peak shifted by 0.09 eV, indicating electron transfer to Cu. Based on previous experiments in Table S3 (ESI†), the presence of chlorine promotes the generation of H₂O₂. Therefore, a comparison was made in the CO DRIFTS without chlorine to investigate the phenomenon. Comparing CuCl₂/anatase and Cu(NO₃)₂/anatase in Fig. S16 (ESI†), CO reacts in both cases even without chlorine, and there is a clear COOH* peak. This indicates that the presence of chlorine ions does not alter the reaction pathway for the generation of H₂O₂ on copper species. The introduction of Pd does not increase the yield of H₂O₂, but significantly increases the yield of methanol. Therefore, experiments were designed to explore the effects of Pd introduction. In the CO DRIFTS experiment, strong CO₂ peaks appear at 2379 and 2311 cm⁻¹ on Pd/anatase, while the peak intensities of bidentate formate (1344 cm⁻¹) and COOH* (1241 cm⁻¹) are low. This indicates that the adsorption of CO on the Pd atoms is not conducive to the stabilization of COOH* species, but tends to combine with surface oxygen species to generate CO₂. As to Cu–Pd/anatase, besides the obvious CO₂ and CH₄ peaks, a sharp CO–Cu⁺ peak appeared at 2115 cm⁻¹ in the CO DRIFTS of Cu–Pd/anatase. Compared with Cu/anatase, a blue shift up to 4 cm⁻¹ indicates that a stronger electron-donating effect occurs on Cu–Pd/anatase. COOH* peak also appears, proving that some CO is still adsorbed on Cu⁺ to generate H₂O₂ although CO is oxidized on the Pd single atom, which is consistent with the experimental conclusion.

Fig. 3 In situ DRIFTS spectra of CuCl₂, anatase, Cu/anatase, Pd/anatase, Cu–Pd/anatase with the introduction of CO (a), CH₄ (b), CH₄ + O₂ + CO + H₂O (c). (d) EPR trapping experiment with DMPO as the radical scavenger in the presence of CuCl₂ and Cu–Pd/anatase. DMPO–CH₃, A_N = 15.6 G, A_Hβ = 22.4 G. (e) In situ DRIFTS of Cu/anatase, Cu/anatase-N, Cu–Pd/anatase, and Cu–Pd/anatase-N with the introduction of reaction gas. (f) The proposed reaction pathways.

To further illustrate the adsorption behavior of CO on different active sites, CO DRIFTS adsorption and desorption experiments were carried out at 150 °C (Fig. 4). When only anatase is present, CO gradually adsorbed on the surface of anatase, and a small amount of CO₂ and stable adsorption peaks of CO₃* and HCO₃* species in the range of 1600–1400 cm⁻¹ are generated. When the catalyst is transformed into Cu/anatase, the CO adsorption becomes weaker, and a weak CO₂ peak appears. In stark contrast, when Pd is present on anatase, strong CO₂ peaks appear at 1379 and 2311 cm⁻¹ immediately after CO introduction, and CO₃* and HCO₃* species adsorption peaks also appear at 1600–1400 cm⁻¹ simultaneously. As CO infeed continues, CO adsorption peaks gradually increase and stabilize at 14 minutes. After being purged with Ar, the CO peaks disappear rapidly within 4 minutes, while a large amount of stable adsorbed CO₂, CO₃*, and HCO₃* species remain on the catalyst surface. This indicates that CO tends to adsorb on the Pd single atom first, then continues to be converted into CO₂ and bicarbonate species. Compared with evanescent CO and CO₂, CO₃* and HCO₃* species are more stable. After introducing the Cu species, CO₂, gaseous CO, CO–Cu⁺, CO₃*, and HCO₃* peaks appeared on the surface of Cu–Pd/anatase. CO no longer only adsorbs on Pd, but also adsorbs on Cu⁺ in the form of CO–Cu⁺. During the Ar purging process, the intensity of most peaks gradually decreased. The suppression of the effect of Pd leads to a reduced CO₂ intensity. In summary, Pd atoms on Pd/anatase exhibit strong CO adsorption and activation ability, making CO hard to stabilize as COOH* species, but ready to transform into more stable CO₃* and HCO₃* species. The introduction of copper can weaken the CO adsorption ability of Pd atoms, possibly because the electrons of Cu have been transferred to Pd. As a result, Pd/anatase itself could not generate H₂O₂; the addition of copper ions helps the generation of H₂O₂, but the H₂O₂ yield on Pd/anatase is still lower than on anatase, which is consistent with the experimental phenomenon (Fig. S15 and Table S3, ESI†). This also suggests that the promotion of methanol production by Pd/anatase is not achieved through the promotion of hydrogen peroxide but rather through strong CO activation ability. To further illustrate the effect of chlorine on the adsorption behavior of CO, CO DRIFTS adsorption and desorption experiments were carried out on Cu/anatase-N (Fig. S17, ESI†). Cu/anatase-N exhibits similar properties to those with chlorine, with weak CO adsorption accompanied by a small amount of CO₂ generation. CO₃* and HCO₃* species adsorption peaks also appear in the range of 1600–1400 cm⁻¹. The CO adsorption peak is almost not observed after 6 minutes of Ar purging, while in Cu/anatase, the CO adsorption peak is almost invisible after 12 minutes. This demonstrates that chlorine ions can promote the stable adsorption of CO. Upon the introduction of Pd, there is a noticeable presence of CO₂ during the Ar purging process in the CO adsorption–desorption spectra of Cu–Pd/anatase-N. However, compared to Pd/anatase, the intensity of the CO₂ peak is weaker. This phenomenon is consistent with the experimental observations of Cu–Pd/anatase. This shows that the presence of Cu species hinders the oxidation of CO by Pd. Within 4 minutes of purging with Ar, the CO peak in Cu–Pd/anatase-N disappears simultaneously, indicating weak CO adsorption. This also indicates that the presence of chlorine promotes the stable adsorption of CO, thereby facilitating the further involvement of CO in the reaction. It is consistent with our experimental observations that chlorine promotes the conversion of CO to H₂O₂.

Fig. 4 In situ DRIFTS spectra using CO probe. anatase (a), Cu/anatase (b), Pd/anatase (c), Cu–Pd/anatase (d) are exposed to CO and purged with Ar at 150 °C.

Mechanistic investigations

To investigate the mechanism of methanol formation, in situ CH₄ and reaction gas DRIFTS are used. In the CH₄ DRIFTS (Fig. 3b), the new peaks at 3015 and 1305 cm⁻¹ are assigned to adsorbed CH₄. CH₄ adsorbed on CuCl₂ tends to generate CO₂; when CH₄ is adsorbed on anatase, a stretching vibration peak of C–O bond of bridged carbonate appears around 1160 cm⁻¹, which is a product of CH₄ oxidation on the surface of anatase.⁵⁰ In Cu/anatase, the CO₂ signal disappears while the bridged carbonate still exists, indicating that the addition of CuCl₂ species prevents methane from overoxidation via carbonate species to CO₂. While CO₂ can be observed on the surface of Pd/anatase, manifesting that CH₄ cannot stabilize in the form of intermediate products such as carbonate on Pd/anatase. As to Cu–Pd/anatase, the CH₄ oxidation by Pd is inhibited, and intermediate products of carbonate are observed, indicating a possible superior selectivity for methanol, consistent with experimental results.

In the in situ DRIFTS under reaction conditions (Fig. 3c), weak CO₂ peaks appear at 2311 and 2379 cm⁻¹ on the surface of CuCl₂. A key intermediate CH₃O* peak appears at 2822 cm⁻¹, and a bridged carbonate emerges at 1170 cm⁻¹. CH₃O* species and CO₂ still exist on the anatase, but the bridged carbonate intermediate disappears, indicating that although the intermediate product CH₃O* of methane oxidation exists, it will continue to be oxidized to CO₂ on anatase. With the assistance of Cu species, the reappearance of bridged carbonate intermediate on the Cu/anatase indicates the inhibition of overoxidation and improvement of the selectivity for methanol. Pd/anatase exhibits similar properties to anatase, which only has CH₃O* and CO₂ species. However, CO₂ exists on Cu–Pd/anatase, and signals of CH₃O* and carbonate species are diminished compared with Cu/anatase. This is a very interesting phenomenon in that the synergistic effect of Cu and Pd leads to the disappearance of the CH₃O* species. Radical trapping experiments by DMPO were also carried out. The obvious DMPO–CH₃ signal is observed under the condition of Cu–Pd/anatase (Fig. 3d), while only the DMPO–OH signal is detected under the condition of CuCl₂. It confirms that ˙CH₃ species are produced because of the desorption of *CH₃ on Cu–Pd/anatase.

CH₄ DIRFTS confirms that the addition of CuCl₂ species prevents methane from overoxidation via carbonate species to CO₂. Given the previous experiments where chlorine was shown to increase the conversion rate of methane to methanol, The role of chlorine needs clarification. In the absence of chlorine during CH₄ introduction, apart from adsorbed CH₄, a weak CO₂ peak is observed (Fig. S18, ESI†). This indicates that Cu species possesses the ability to oxidize methane. A small CO₂ peak is observed in the absence of chlorine during CH₄ introduction, indicating that Cu species can oxidize methane. Compared to Pd/anatase, Cu/anatase shows more carbonate species both with and without chlorine, indicating that copper can inhibit the oxidation of methane. Comparing Cu–Pd/anatase-N with Cu–Pd/anatase, the introduction of chlorine inhibits the generation of CO₂, indicating that the oxidation of methane is suppressed by chlorine ions. This is consistent with the experimentally measured conversion rate of CH₄ to methanol. In the in situ DRIFTS under reaction conditions (Fig. 3e), signals of CH₃O* are also present in Cu/anatase-N, and the CH₃O* signal disappears with the introduction of Pd. This indicates that the desorption of the key intermediate species ˙CH₃ is due to not chlorine ions but rather the critical synergistic mechanism between Cu and Pd. It has been confirmed that ˙CH₃ species are produced because of the desorption of *CH₃ on Cu–Pd/anatase. It can be noted that the CO adsorption peak of Cu–Pd/anatase-N is noticeably weaker in comparison to that of Cu/anatase-N. This phenomenon is consistent with the CO adsorption situation of Pd (Fig. 4c), which suggests oxidation of CO after adsorption. In summary, the role of chlorine is to promote stable CO adsorption, leading to the generation of more H₂O₂. It also helps the existence of intermediates, thereby enhancing the selectivity of methane to methanol.

In conclusion, two major factors that promote methanol production have been uncovered: (1) Cu and chlorine species can promote the existence of key *CH₃ intermediates in the process of the reaction; (2) The synergistic effect of Cu ions and Pd leads to the generation and desorption of ˙CH₃, which then combines with ˙OH to generate methanol. In order to prove the decisive role of ˙OH, 50 mg Na₂SO₃ was added to the solution as a hydroxyl scavenger.¹⁴ After the reaction, the methanol yield dropped to 1625 μmol g_cat⁻¹, indicating that the free radical mechanism for methanol production is certainly via the combination of ˙CH₃ and ˙OH.

CH₄ can undergo activation just on Cu ions, which laid the foundation for further optimization of methanol yield and selectivity. Although methane oxidation species appear both on Cu and Pd sites under the reaction conditions, the inhibition of their overoxidation strongly correlates to Cu species. The conclusion could be drawn that dissociated methane on metal sites in the form of *CH₃, then M–O bonds are broken and ˙CH₃ species are generated, which could combine with ˙OH to form methanol (Fig. 3f).

Main factors affecting the selective methanol generation

DFT calculations were carried out to comprehend the synergistic effect of copper and palladium as well as the mechanism of methane oxidation to methanol. Because (101) is the dominant surface of this catalyst based on HAADF-STEM studies (Fig. S19, ESI†), in the theoretical simulations of the Pd/anatase surface structure, TiO₂ (101) was selected as the surface to attach the Pd atom. (Fig. S20, ESI†). The experiments indicate that the synergistic effect between Cu and Pd can significantly promote methanol production even in the absence of chlorine. Here, we will initially focus on exploring the synergistic effect between Pd and Cu. Given that adsorbed Cu ions are adjacent to Pd in Fig. 1c, the distance between Pd and Cu atoms in the model corresponds well with the AC-HAADF-STEM results. The projected density of states (PDOS) diagram of Cu (Fig. 5a) shows a downshift in the center of the d band, which may increase the chemical adsorption bond length between the adsorbed species and the metal center. For the properties of Cu and Pd atoms, the differential charge density map shows electrons are transferred between Cu and Pd (Fig. S21, ESI†). It is evident that the Cu species exhibit electron-deficient characteristics, while there is charge accumulation at adjacent atoms. This indicates that charge transfer occurred from Cu to the neighboring O and Ti atoms. Mulliken charge analysis shows a decrease in the charge from +1.34 a.u. to +0.99 a.u. This further reveals a significant increase in the electron density near Pd in the presence of copper ions, indicating that Cu transfers electrons to Pd.

Fig. 5 (a) Projected density of states (PDOS) for the total d orbitals of Cu. (b) PDOS of 3d orbitals of Cu on Cu–Pd/anatase and 2p orbitals of C in methane, and their interaction between Cu and C within Cu–*CH₃. (c) Spin density of Cu–Pd/anatase with *CH₃ (blue and yellow represent electron accumulation and depletion, respectively). (d) Charge density differences of Cu–Pd/anatase with *CH₃ (blue stands for electrons and yellow for holes). (e) The energy change of different reaction routes.

Experiments and characterizations have indicated the formation of ˙CH₃ species, and Cu–Pd/anatase shows no significant presence of CH₃O* species. Therefore, investigations into methane adsorption were conducted. We investigate the densities of states (DOS) of CH₄ adsorption on Cu–Pd/anatase. The α-spin d orbital of Cu in Cu–Pd/anatase aligns well with the C σ orbital of CH₄, indicating favorable adsorption between Cu–Pd/anatase species and methane. Based on the previous DRIFTS experiments, it can be deduced that CO tends to adsorb on the Pd site on the catalyst surface, and reacts with O₂ to generate metal–O sites, which then combine with the adsorbed methane to form the CH₃O* intermediate species. DFT model calculations have shown that the Pd–O–Cu structure is more stable than Pd–O (Fig. S22, ESI†). Methane adsorption energies at Cu and O sites are compared in Fig. 5e. Methane adsorption at Cu sites releases 0.94 eV, while adsorption at O sites requires 0.02 eV. This is consistent with previous in situ DRIFTS results, indicating that methane adsorption directly occurs at the Cu sites of Cu–Pd/anatase. Previous studies have shown that Pd has an affinity for hydrogen.²³ The Pd sites carry unpaired electrons (Fig. S23, ESI†). When methane is adsorbed on copper sites, the hydrogen from methane can combine with the Pd, forming the *CH₃ species. To further explore the key factors contributing to the highly selective production of methanol, the desorption of ˙CH₃ is emphasized. When Cu–Pd/anatase adsorbs the *CH₃ species, the Cu–C bond length is 1.93 Å, and the H–C–H bond angle is 112.883°, which is higher than Pd–OCH₃ on Pd/anatase and Cu–OCH₃ on Cu/anatase (Fig. S24 and Table S8, ESI†).

In Fig. 5b, the partial coupling between Cu and C orbitals is observed when Cu–Pd/anatase adsorbs the *CH₃ species. The center of Cu d-band shifts downward to 2.70 eV, weakening the adsorption on the Cu site (Fig. 5a). On the other hand, the PDOS orbitals of C and O in Cu–OCH₃ and Pd–OCH₃ species show a more pronounced coupling, indicating a stronger interaction between C and O (Fig. S25, ESI†). The spin density of Cu–Pd/anatase adsorbed with methyl group exhibits an unpaired electron, preserving the radical nature of *CH₃ (Fig. 5c). In the electron difference density (Fig. 5d), it is evident that upon *CH₃ adsorption, Cu electrons transfer to the substrate, while Pd electrons transfer to the proton, as compared to Cu–Pd/anatase without *CH₃ adsorption (Fig. S21, ESI†). In order to verify the impact of chlorine ions on the intermediate steps of the reaction, DFT calculations were conducted to validate their roles. Mulliken charge analysis indicates electron transfer from chlorine to Cu upon the introduction of chlorine, consistent with the XPS conclusion (Table S9, ESI†). It is demonstrated that the addition of chlorine promotes stable CO adsorption. Furthermore, we calculated the adsorption energy of CO and presented the results in Table S10 (ESI†). The presence of chlorine reduces the adsorption energy of CO by 1.67 eV, making CO more prone to adsorption, which is consistent with experimental observations. The activation energy of methane on Cu–Pd/anatase with chlorine is −0.42 eV. Meanwhile, the energy required for methane dissociation is only 0.37 eV, enabling efficient conversion of methane to methyl radicals.

As to the high selectivity of methane oxidation to methanol, the reason could be as follows. There are two main reaction routes for the overoxidation of methane: (1) methane is transformed into CH₃OOH and might decompose subsequently; (2) methane is transformed into methanol and further oxidated to peroxide (such as HCHO/HCOOH/CO₂). As to the first route, H₂O₂ would decompose into *OOH or *OH on catalysts. Methane would combine with *OOH to generate CH₃OOH or *OH to generate CH₃OH.^4,15,34,51 Different catalysts vary in the deposition of H₂O₂, then affect the following products of methane oxidation. The absence of CH₃OOH in this work indicates that H₂O₂ might decompose into *OH more readily than *OOH on Cu–Pd/anatase. Qi et al. reported that CH₄ is more likely to generate CH₃OH instead of CH₃OOH in the existence of CO.⁵² A plausible explanation was proposed that CH₃OO* would react with CO to generate CH₃O* and CO₂. To sum up, the reasons for the absence of CH₃OOH are that: (1) H₂O₂ tends to decompose into *OH instead of *OOH; (2) the low concentration of in situ generated H₂O₂ is unfavorable for the generation of *OOH; (3) the presence of CO also impedes the existence of methyl peroxide species. To further validate this possibility, DFT calculations were supplemented in Table S11 (ESI†). The calculation results indicate that on Cu–Pd/anatase, the energy for the cleavage of H₂O₂ into 2 *OH is lower by 2.20 eV compared to the energy for *H + *OOH. This confirms the accuracy of our hypothesis, suggesting that H₂O₂ is more prone to cleave into *OH and participate in the reaction with ˙CH₃, achieving high methanol selectivity.

When it comes to the second route, the oxidation of methanol is also a classic problem for product selectivity. This work focuses on the reasons for the great difference in the selectivity of methane oxidation compared with those previously reported in other literature. As is well-known, if methanol could not decompose from the catalyst's surface in time, it would be oxidated in the following step. Therefore, timely desorption of the product is essential to hinder the over-oxidation. In this work, we modulate the interaction between Cu and Pd species and observe apparent ˙CH₃ radicals, which are not often observed in methane oxidation. The characterization and DFT calculation results indicate that the generation and desorption of ˙CH₃ species promotes selectivity to methanol. The low concentration of in situ generated H₂O₂ is not enough for further oxidation of methanol in the solution.

Conclusions

Herein, an efficient and highly selective reaction mechanism for the conversion of methane to methanol, in the presence of CO in an aqueous phase, has been proposed. Cu ions are essential to produce methanol, as they react with H₂O, O₂, and CO to in situ generate H₂O₂ and participate in the activation of methane. The synergy between Pd and Cu species activates methane, generating the key intermediate species methyl. chlorine ions can promote the stable adsorption of CO and the formation of *CH₃ intermediates, facilitating high activity and selectivity for methanol production. ˙CH₃ species detach from Cu–Pd/anatase and then combine with ˙OH to generate methanol. This system exhibits a relatively high methanol generation rate and selectivity in the liquid phase. This discovery may carve out new avenues for the selective conversion of methane to methanol in future catalyst design.

Author contributions

W. L., L. Y. and L. W. conceived and supervised the research. L. W. designed and performed the experiments and data analysis. L. W. and C. L. performed the DFT simulations. L. W. and J. J. wrote the paper. All authors discussed the results and commented on the paper.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the Major Science and Technology Projects of Anhui Province (202003a05020022), the Institute of Energy, Hefei Comprehensive National Science Center under Grant No. 21KZS219 and the Key Research and Development Projects in Anhui Province (2022107020013). This work was partially carried out at the Instruments Center for Physical Science, University of Science and Technology of China.

References

1. X. Meng, X. Cui, N. P. Rajan, L. Yu, D. Deng and X. Bao, Chem, 2019, 5, 2296–2325.
2. X. Zhang, Y. Wang, K. Chang, S. Yang, H. Liu, Q. Chen, Z. Xie and Q. Kuang, Appl. Catal., B, 2023, 320, 121961.
3. M. S. A. Sher Shah, C. Oh, H. Park, Y. J. Hwang, M. Ma and J. H. Park, Adv. Sci., 2020, 7, 2001946.
4. Y. Xu, D. Wu, P. Deng, J. Li, J. Luo, Q. Chen, W. Huang, C. M. Shim, C. Jia, Z. Liu, Y. Shen and X. Tian, Appl. Catal., B, 2022, 308, 121223.
5. J. Shan, M. Li, L. F. Allard, S. Lee and M. Flytzani-Stephanopoulos, Nature, 2017, 551, 605–608.
6. L. Sun, Y. Wang, C. Wang, Z. Xie, N. Guan and L. Li, Chem, 2021, 7, 1557–1568.
7. V. L. Sushkevich, D. Palagin, M. Ranocchiari and J. A. v Bokhoven, Science, 2017, 356, 523–527.
8. M. A. Newton, A. J. Knorpp, V. L. Sushkevich, D. Palagin and J. A. van Bokhoven, Chem. Soc. Rev., 2020, 49, 1449–1486.
9. M. Ravi, V. L. Sushkevich, A. J. Knorpp, M. A. Newton, D. Palagin, A. B. Pinar, M. Ranocchiari and J. A. van Bokhoven, Nat. Catal., 2019, 2, 485–494.
10. Z. Jin, L. Wang, E. Zuidema, K. Mondal, M. Zhang, J. Zhang, C. Wang, X. Meng, H. Yang, C. Mesters and F.-S. Xiao, Science, 2020, 367, 193–197.
11. S. Sogukkanli, T. Moteki and M. Ogura, Green Chem., 2021, 23, 2148–2154.
12. F. Gu, X. Qin, M. Li, Y. Xu, S. Hong, M. Ouyang, G. Giannakakis, S. Cao, M. Peng, J. Xie, M. Wang, D. Han, D. Xiao, X. Wang, Z. Wang and D. Ma, Angew. Chem., Int. Ed., 2022, 61, e202201540.
13. T. Moteki, N. Tominaga and M. Ogura, Appl. Catal., B, 2022, 300, 120742.
14. W. Wang, W. Zhou, Y. Tang, W. Cao, S. R. Docherty, F. Wu, K. Cheng, Q. Zhang, C. Copéret and Y. Wang, J. Am. Chem. Soc., 2023, 145, 12928–12934.
15. C. Hammond, M. M. Forde, M. H. Ab Rahim, A. Thetford, Q. He, R. L. Jenkins, N. Dimitratos, J. A. Lopez-Sanchez, N. F. Dummer, D. M. Murphy, A. F. Carley, S. H. Taylor, D. J. Willock, E. E. Stangland, J. Kang, H. Hagen, C. J. Kiely and G. J. Hutchings, Angew. Chem., Int. Ed., 2012, 51, 5129–5133.
16. R. L. Lieberman and A. C. Rosenzweig, Nature, 2005, 434, 177–182.
17. R. Balasubramanian, S. M. Smith, S. Rawat, L. A. Yatsunyk, T. L. Stemmler and A. C. Rosenzweig, Nature, 2010, 465, 115–119.
18. A. A. Latimer, A. R. Kulkarni, H. Aljama, J. H. Montoya, J. S. Yoo, C. Tsai, F. Abild-Pedersen, F. Studt and J. K. Nørskov, Nat. Mater., 2017, 16, 225–229.
19. S. Grundner, M. A. C. Markovits, G. Li, M. Tromp, E. A. Pidko, E. J. M. Hensen, A. Jentys, M. Sanchez-Sanchez and J. A. Lercher, Nat. Commun., 2015, 6, 7546.
20. K. Narsimhan, K. Iyoki, K. Dinh and Y. Román-Leshkov, ACS Cent. Sci., 2016, 2, 424–429.
21. H. V. Le, S. Parishan, A. Sagaltchik, C. Göbel, C. Schlesiger, W. Malzer, A. Trunschke, R. Schomäcker and A. Thomas, ACS Catal., 2017, 7, 1403–1412.
22. X. Wu, Q. Zhang, W. Li, B. Qiao, D. Ma and S. L. Wang, ACS Catal., 2021, 11, 14038–14046.
23. R. Zhang, H. Wang, S. Tang, C. Liu, F. Dong, H. Yue and B. Liang, ACS Catal., 2018, 8, 9280–9286.
24. L. Kuai, Z. Chen, S. Liu, E. Kan, N. Yu, Y. Ren, C. Fang, X. Li, Y. Li and B. Geng, Nat. Commun., 2020, 11, 48.
25. C. Li, B. Tang, A. T. Ogunbiyi, S. Tang, W. Li, Q. Lu and L. Yuan, Mol. Catal., 2022, 528, 112475.
26. B. Wu, T. Lin, M. Huang, S. Li, J. Li, X. Yu, R. Yang, F. Sun, Z. Jiang, Y. Sun and L. Zhong, Angew. Chem., Int. Ed., 2022, 61, e202204116.
27. M. H. Ab Rahim, M. M. Forde, R. L. Jenkins, C. Hammond, Q. He, N. Dimitratos, J. A. Lopez-Sanchez, A. F. Carley, S. H. Taylor, D. J. Willock, D. M. Murphy, C. J. Kiely and G. J. Hutchings, Angew. Chem., Int. Ed., 2013, 52, 1280–1284.
28. N. Agarwal, S. J. Freakley, R. U. McVicker, S. M. Althahban, N. Dimitratos, Q. He, D. J. Morgan, R. L. Jenkins, D. J. Willock, S. H. Taylor, C. J. Kiely and G. J. Hutchings, Science, 2017, 358, 223–227.
29. P. Deng, J. Duan, F. Liu, N. Yang, H. Ge, J. Gao, H. Qi, D. Feng, M. Yang, Y. Ren and Y. Qin, Angew. Chem., Int. Ed., 2023, 62, e202307853.
30. M. He, Y. Cao, J. Ji, K. Li and H. Huang, J. Catal., 2021, 396, 122–135.
31. J. Zhang, W. Liao, H. Zheng, Y. Zhang, L. Xia, B.-T. Teng, J.-Q. Lu, W. Huang and Z. Zhang, J. Catal., 2022, 405, 152–163.
32. M. He, J. Ji, B. Liu and H. Huang, Appl. Surf. Sci., 2019, 473, 934–942.
33. D. Palagin, V. L. Sushkevich and J. A. van Bokhoven, ACS Catal., 2019, 9, 10365–10374.
34. P. Xie, J. Ding, Z. Yao, T. Pu, P. Zhang, Z. Huang, C. Wang, J. Zhang, N. Zecher-Freeman, H. Zong, D. Yuan, S. Deng, R. Shahbazian-Yassar and C. Wang, Nat. Commun., 2022, 13, 1375.
35. A. Koishybay and D. F. Shantz, J. Am. Chem. Soc., 2020, 142, 11962–11966.
36. Z. Zuo, P. J. Ramírez, S. D. Senanayake, P. Liu and J. A. Rodriguez, J. Am. Chem. Soc., 2016, 138, 13810–13813.
37. H. M. Rhoda, D. Plessers, A. J. Heyer, M. L. Bols, R. A. Schoonheydt, B. F. Sels and E. I. Solomon, J. Am. Chem. Soc., 2021, 143, 7531–7540.
38. V. R. Choudhary and C. Samanta, J. Catal., 2006, 238, 28–38.
39. X. Du, Y. Huang, X. Pan, B. Han, Y. Su, Q. Jiang, M. Li, H. Tang, G. Li and B. Qiao, Nat. Commun., 2020, 11, 5811.
40. K. I. Hadjiivanov, M. M. Kantcheva and D. G. Klissurski, J. Chem. Soc., Faraday Trans., 1996, 92, 4595–4600.
41. Z. Liu, F. Yu, K. Pan, X. Zhou, R. Sun, J. Tian, Y. Wan, J. Dan and B. Dai, Chem. Eng. J., 2021, 422, 130099.
42. C. Zhang, L. Wang, U. J. Etim, Y. Song, O. M. Gazit and Z. Zhong, J. Catal., 2022, 413, 284–296.
43. C. Han, Y. Cao, W. Yu, Z. Huang, F. Dong, L. Ye, S. Yu and Y. Zhou, J. Am. Chem. Soc., 2023, 145, 8609–8620.
44. S. Kattel, B. Yan, Y. Yang, J. G. Chen and P. Liu, J. Am. Chem. Soc., 2016, 138, 12440–12450.
45. Z. Li, J. Wang, Y. Qu, H. Liu, C. Tang, S. Miao, Z. Feng, H. An and C. Li, ACS Catal., 2017, 7, 8544–8548.
46. S. Li, Y. Xu, H. Wang, B. Teng, Q. Liu, Q. Li, L. Xu, X. Liu and J. Lu, Angew. Chem., Int. Ed., 2022, 62, e202218167.
47. A. Solis-Garcia, J. F. Louvier-Hernandez, A. Almendarez-Camarillo and J. C. Fierro-Gonzalez, Appl. Catal., B, 2017, 218, 611–620.
48. P. Lu, Y. Zhao, J. Dong, B.-B. Mei, X.-L. Du, Z. Jiang, Y.-M. Liu, H.-Y. He, Y.-D. Wang, X.-Z. Duan, X.-G. Zhou and Y. Cao, ACS Catal., 2020, 10, 13993–14005.
49. L. DeRita, S. Dai, K. Lopez-Zepeda, N. Pham, G. W. Graham, X. Pan and P. Christopher, J. Am. Chem. Soc., 2017, 139, 14150–14165.
50. H. Wijnja and C. P. Schulthess, Spectrochim. Acta, Part A, 1999, 55, 861–872.
51. S. Bai, F. Liu, B. Huang, F. Li, H. Lin, T. Wu, M. Sun, J. Wu, Q. Shao, Y. Xu and X. Huang, Nat. Commun., 2020, 11, 954.
52. G. Qi, T. E. Davies, A. Nasrallah, M. A. Sainna, A. G. R. Howe, R. J. Lewis, M. Quesne, C. R. A. Catlow, D. J. Willock, Q. He, D. Bethell, M. J. Howard, B. A. Murrer, B. Harrison, C. J. Kiely, X. Zhao, F. Deng, J. Xu and G. J. Hutchings, Nat. Catal., 2022, 5, 45–54.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ee02671c

---