Coupling photocatalytic CO2 reduction and CH3OH oxidation for selective dimethoxymethane production

Catalyst synthesis and structural characterizationThe designed model catalysts were prepared by simply reducing WCl6 and AgNO3 with sodium borohydride (NaBH4) on the BTO substrate. The tungsten (W) ions were successfully introduced to the lattice of BTO, and the resulting Ag nanoparticles were in situ formed on the BTO surface during the reduction process (Fig. 2a). More details of the preparation process are supplied in the experimental section. Comparative catalysts, including BTO, Ag-BTO and W-BTO, were synthesized using the same method (Supplementary Figs. 1–11 and Supplementary Table 1). The crystal structures of all samples are shown in the X-ray diffraction (XRD) patterns of Fig. 2b. As expected, the BTO substrates in the composite samples were identified as pure anatase TiO2 (PDF#21-1272) after Li-EDA treatment. In this case, we expected high UV-visible light absorption and increased stability on the newly obtained BTO during the reaction24. The Ag nanoparticles were primarily in their metallic phase (PDF#04-0783). No prominent peak corresponding to W was evident, indicating that the W atoms were successfully doped into the BTO. The blocky Ag.W-BTO was ~40–50 nm wide, as shown in the transmission electron microscopy (TEM) image of Fig. 2c. In addition, the interplanar d spacing was 0.35 nm and 0.235 nm, shown in the Ag.W-BTO high-resolution TEM (HRTEM) image and belonging to the (101) and (001) anatase phase TiO2, respectively (Fig. 2d)24. The Ag nanoparticles were ~2–10 nm in diameter, with a clear lattice and a distance spacing of 0.24 nm, which can match Ag (111) (Fig. 2e). A fast Fourier-transform (FTT) image (inset) shows the (111), (200), (220), and (311) facets of Ag, which coincide with XRD patterns25,26. High-angle annular dark-field scanning TEM (HAADF-STEM) revealed Ag nanoparticles 2–10 nm in diameter (yellow circle, Fig. 2f), and energy dispersive spectrometry (EDS) was used to map the elemental distribution of Ti, O, W, and Ag. Due to the large amount of AgNO3 added during the preparation progress (checked by ICP-OES 38.53 wt.%; Supplementary Table 2), the Ag nanoparticles showed some aggregation of Ag.W-BTO-150 (Supplementary Fig. 5).Fig. 2: Morphology and structural characterizations of Ag.W-BTO.a Proposed structure of Ag.W-BTO and reaction. b XRD of all catalysts. c–e TEM and HRTEM images of Ag.W-BTO. f HAADF-STEM image of Ag.W-BTO and EDS elemental mappings of Ti, O, W, and Ag. (a. u.) represents arbitrary units.The chemical states of the catalysts were determined by X-ray photoelectron spectroscopy (XPS). In the Ti 2p spectra (Fig. 3a), four characteristic peaks appear near 464.30, 463.31, 458.64, and 457.82 eV, corresponding to Ti 2p1/2 Ti4+, Ti 2p1/2 Ti3+, Ti 2p3/2 Ti4+, and Ti 2p3/2 Ti3+, respectively27,28. The presence of peaks corresponding to Ti3+ confirms that the synthesized catalyst contains oxygen vacancies (Ov). Compared with BTO and Ag-BTO, all W-BTO and Ag.W-BTO peaks displayed a positive shift. This phenomenon is primarily caused by W species with a lower electron cloud density and a strong electron affinity that can absorb electrons from BTO, resulting in the formation of a stable structure29. The Ag peaks centered at 372.60 eV and 366.70 eV can be ascribed to metallic Ag (Fig. 3b). The higher shift of the Ag binding energy in Ag.W-BTO is attributed to strong heterogeneous interaction and electron transfer between Ag and W-BTO substrate30. In the W 4f spectrum (Fig. 3c), W6+ 4f5/2, W5+ 4f5/2, W6+ 4f7/2, and W5+ 4f7/2 can be observed at 38.38, 36.72, 35.02, and 33.24 eV, respectively31. To prevent photo-generated carriers from recombining under light irradiation, W6+, as a donor just below the conduction band, can be gradually transformed to unsaturated W5+, which works in W6+/W5+ pairs to improve performance (Supplementary Fig. 29c)32. The W of Ag.W-BTO also shifted to a positive binding energy, indicating more electron transmission from W to O after Ag modification. This is attributed to the strong electron acceptor properties of O33,34. All the highest peaks of the O 1s spectra can be ascribed to Ti−O. The other peaks correspond to Ov, an −OH/O−W group, and H2O (absorbed on the surface), respectively (Supplementary Fig. 12a)35. No boron species remained in samples during the synthesis process (Supplementary Fig. 12b). The presence of Ti3+ species and Ov were also confirmed by the Electron paramagnetic resonance (EPR) spectra (Supplementary Fig. 13). Both BTO and Ag.W-BTO show distinctive EPR signals of paramagnetic Ti3+ (g = 1.96) and Ov (g = 2.003) while TiO2 (P25) showed a negligible EPR signal36,37,38. This result implies that BTO with Ov was successfully synthesized after reduction with Li-EDA. The peak intensities of Ti3+ and Ov in BTO are higher than those of Ag.W-BTO, meaning the doping of W may slightly replace the Ti3+ or cover the Ov in Ag.W-BTO sample.Fig. 3: XPS and X-ray absorption fine structure (XAFS) characterizations.a–c Ti 2p, Ag 3d, and W 4 f XPS of BTO, W-BTO, Ag-BTO, and Ag.W-BTO, respectively. (a. u.) represents arbitrary units. d–f Normalized Ti K-edge, Ag K-edge, W L3-edge, and their difference X-ray absorption near edge structure (XANES) of catalysts with standard references. g–i Extended XAFS (EXAFS) k3 χ(k) Fourier-transform ( | X (R)|) spectra of Ti, Ag, and W in R-space of catalysts with standard references, respectively. R (Å) represents radial distance in Angstroms.The electronic states of Ti, Ag, and W were explored by X-ray absorption near-edge spectroscopy (XANES). The state of Ti for all prepared samples shows features similar to those of anatase TiO2 (Fig. 3d). Compared with BTO, Ag-BTO, and W-BTO, the higher Ti K-edge absorption peak intensity in Ag.W-BTO is due to the collective electronic modulations of extrinsic Ag and W species. Meanwhile, the Ag in Ag-BTO and Ag.W-BTO can be traced to metallic Ag, as shown in the XRD and XPS results (Fig. 3e). However, the Ag pre-edge adsorption position of Ag.W-BTO had a positive shift compared with that of Ag-BTO, which is caused by high interaction among Ag and W-BTO support. Strong interface interactions are anticipated to promote beneficial electron transfer in the photocatalytic process. From the W L3-edge absorption spectra (Fig. 3f), the white line peak positions of W in W-BTO and Ag.W-BTO are higher than those of W foil and WO3 attributed to the strong electron adsorption properties of O in W-BTO and Ag.W-BTO, which agrees with the discussions of XPS spectra. The distinctive coordination environments of Ti, Ag, and W were investigated by Fourier-transformed (FT) extended X-ray absorption fine structure (EXAFS) spectra and fitting data (Supplementary Fig. 14 and Supplementary Tables 3−5). In the Ti FT-EXAFS spectra, Ti−O bonds can be seen at ~1.63, 1.65, and 1.62 Å for Ag-BTO, W-BTO, and Ag.W-BTO, respectively (Fig. 3g and Supplementary Fig. 14a). The Ti−O bonds of these samples are extended compared to those of pure BTO (located at ~1.52 Å) due to the introduction of W and Ag species that can stretch the Ti−O bond length and synergistically tune the local coordination environment of Ti. The dominant peaks of Ag−Ag shown in Ag-BTO and Ag.W-BTO are nearly no difference which is consistent with that of standard Ag foil (Fig. 3h and Supplementary Fig. 14b). For k3 χ(k)-FT of the W L3-edge EXAFS spectra (Fig. 3i and Supplementary Fig. 14c), the first coordination shell (W−O peak at 0.7–1.9 Å) was composed primarily of single-scattering O atoms, accompanied by multiple-scattering contributions from the second shell (peaks at 2–3 Å)39,40. The interatomic distances of W−O in W-BTO (~1.47 Å) and Ag.W-BTO (~1.29 Å) are different from those of standard WO3 (1.36 Å) as W atoms are doped into the BTO lattice rather than the isolated oxide species. Overall, the electronic interaction of Ag and W dual-active sites is expected to be capable of guaranteeing the photo-redox synergic reaction for higher efficiency.The adsorption abilities of light in all catalysts were measured by UV-visible light spectra (Supplementary Fig. 15a)41. The absorption edge of Ag.W-BTO exhibited a slight redshift after W implantation and Ag deposition, which can enable stronger UV-visible light absorption and improve hot-carrier generation and catalytic performance. The band gaps of the photocatalysts were calculated using Kubelka–Munk and Tauc equations (more details are provided in the Supporting Information)42. The BTO and W-BTO values were 2.91 and 2.86 eV43, respectively (Supplementary Fig. 15b). All the BTO-related samples showed a narrower band structure than pure TiO2 (P25, 3.2 eV), which was attributed to the existence of Ti3+ defect sites with multiple internal energy band gaps after Li-EDA treatment22,44. The valence band (VB) edge potentials and Femi levels of these catalysts were determined using Ultraviolet photoelectron spectroscopy (UPS) spectra43. In details, the positions of the secondary electron cutoff (Ecut off) and the valence band maximum (EVBM) positions are determined using linear extrapolation of UPS45. As shown in Supplementary Fig. 15c, the Ecut off values of BTO and W-BTO are 17.21 and 17.49 eV, respectively, and the EVBM values of these two samples are 1.79 and 2.15 eV, respectively. The work function (ϕ) can be calculated by subtracting Ecut off from the energy of the incident UV light (hγ) after measuring the width of the emitted electrons from the onset of the secondary electrons up to the Fermi edge, according to the Formula (1).$$\phi=h\gamma – {E}_{{{{{\rm{cut\; off}}}}}}$$
(1)
Here, the energy of Helium line as a hγ is 21.22 eV. According to formula (1), the ϕ of BTO and W-BTO are calculated to be 4.01 and 3.73 eV, respectively. Consequently, the Femi levels of BTO and W-BTO are –4.01 and –3.73 eV, respectively. The VB of BTO and W-BTO in Vacuum (EVAB) were calculated using the Formula (2):$${E}_{{{{{\rm{VAB}}}}}}=- ({E}_{{{{{\rm{VBM}}}}}}+\phi )\,$$
(2)
Resulting in –5.80 and –5.88 eV, respectively. All values corresponding to vacuum should be replaced with normal hydrogen electrode (NHE), resulting in a difference of –4.44 eV46. Therefore, band gap structures of BTO and Ag.W-BTO are shown in Supplementary Fig. 15d. The VB of W-BTO (1.44 V) exhibited higher positivity than BTO (1.36 V), imparting that doping W species theoretically facilitates MOR performance. Due that the Femi level (Ef) of metallic Ag nanoparticles (Ef = –4.26 eV) is more negative than that of BTO (Ef = –4.01 eV) and W-BTO (Ef = –3.73 eV) (Supplementary Fig. 16 a, b). Take Ag.W-BTO as an example, based on the strong interfacial interaction by Mott-Schottky junction, the electrons flow from W-BTO to Ag induced by the difference in Ef between Ag.W-BTO until the system reaches equilibrium, resulting in band bending and Schottky barrier formation at the interface. The suitable Schottky barrier facilitates the migration of photogenerated electrons, which also proves that CO2RR are more likely to occur on Ag species of Ag.W-BTO during DMM synthesis process.N2 adsorption/desorption analyses were carried out to determine the effect of doping W and depositing Ag on the pore structures (Supplementary Fig. 17)47. The presence of mesoporous structures is implied by type IV isotherms with H3 hysteresis loops in all prepared catalysts48. The pore-structure distribution curves in Supplementary Fig. 17f are evidence of the dominant mesoporous and microporous structures of all samples. The specific surface areas (SSAs) of BTO, W-BTO, Ag-BTO, and Ag.W-BTO were 28.62, 44.68, 35.26, and 52.12 m2 g−1, respectively (additional parameters are shown in Supplementary Table 6). The increased SSAs of Ag.W-BTO is likely attributed to the porous structures (Supplementary Fig. 10b).Photocatalytic performance by coupling CO2RR and MORA homemade photocatalytic reaction system (Supplementary Fig. 18) was used to verify the photocatalytic performance of the samples used as photocatalysts for photo-redox CO2RR and MOR systems. Specifically, 30 mL of CH3OH was added to the reactor, which was purged with CO2 gas for 1 h without any use of sacrificial reagents in the dark. After irradiation with UV-visible light (320–780 nm) for 9 h, gas production (H2, O2, CO, CH4) and liquid production (CH2O and DMM) were quantitively detected by gas chromatography (GC). The corresponding productivity and selectivity of each product were calculated using Eqs. (3) – (9) as detailed in the Method section. The CO production rate on BTO reached 217.71 µmol g-1, which is ~100 times greater than that on Ag.W-BTO (Fig. 4a, b). In contrast, Ag.W-BTO produced the largest amount of DMM (5702.49 µmol g-1), but scant DMM products were seen on BTO (Fig. 4c). In addition, CH4, as by-product, were produced with a minimal amount of Ag.W-BTO, confirming that it is possible to make extensive use of reaction intermediates at an Ag and W dual-redox site in an Ag.W-BTO sample. The amount of CO2RR products (CO, CH4, and CH2O) on Ag-BTO was significantly greater than on W-BTO, which indicates that Ag prefers to be the active reduction site. Overall, W-BTO, Ag-BTO, Ag.W-BTO-50, and Ag.W-BTO-150 achieved DMM yields of 97.69, 488.47, 3337.96, and 551.11 µmol g-1, respectively (Fig. 4c and Supplementary Fig. 19–24). Compared with Ag-BTO and W-BTO, the greater photo-efficiency of the larger DMM yield on Ag.W-BTO was achieved because the Ag and W species individually acted as active sites to promote reductive and oxidative reactions, respectively. Ag agglomerated into large nanoparticles that covered many oxidation sites on the Ag.W-BTO-150 when excess Ag was added, resulting in less DMM and inferior selectivity (Supplementary Fig. 5). Due to producing the large amount of DMM on the Ag.W-BTO sample, the production of DMM changes was detected as time increased. As shown in Supplementary Fig. 25, the yield of DMM also exhibited a growth trend as time increased. DMM selectivity on Ag.W-BTO reached 92.08%, which is greater than that of other samples (Fig. 4d). The Apparent quantum yield (AQY) for BTO and Ag.W-BTO were measured using a range of monochromatic light band-pass filters. As depicted in Supplementary Fig. 26, the AQY of this two samples trend aligns with the UV-visible absorption spectrum, especially the AQY values of Ag.W-BTO reaching 2.15% and 1.01% at 395 nm and 420 nm, respectively. Furthermore, DMM yield and selectivity did not decrease after four catalysis cycles on Ag.W-BTO, indicating that Ag.W-BTO has excellent stability (Supplementary Fig. 27). The TEM, SEM images, XRD patterns and XPS spectrums after the cycling test were also recorded (Supplementary Fig. 28 and 29). Ag.W-BTO presents excellent sustainability of morphology and crystalline nature, which results in the higher stability of activity and selectivity for DMM synthesis. The valence states of Ti, Ag, W and O also showed nearly no difference before and after stability. The high stability is due to the fact that W is doped into the lattice of BTO with lattice distortion49.Fig. 4: Photocatalysis CO2RR coupling with MOR performance.a Gas production. b Enlarged H2, CO and CH4 production. c Liquid production of prepared samples and d selectivity of DMM after 9 h of reaction time.Furthermore, a series of controlled experiments were performed to examine whether our photocatalytic DMM generation integrated the advantages of CO2RR and MOR (Supplementary Fig. 30). Ag.W-BTO at standard conditions produced DMM at a rate of 5702.49 µmol g-1, which is ~7 times greater than without CO2 (866.59 µmol g-1) and ~43 times greater than without CH3OH (132.03 µmol g-1). To explore the influence of pH on DMM synthesis, the pH of CH3OH before and after CO2 exposure were tested to be 6.256 and 5.202, respectively (Supplementary Fig. 31), which indicates that the introduction of CO2 can indeed decrease the reactant pH, which is essential for the initiation of the coupled reaction. Besides, the pH of the CH3OH reactant was adjusted to ~5 by using HCl (CH3OH + Ar, pH = 5) to avoid the CO2 effect. It showed that the synthesis amount of DMM in CH3OH + Ar with pH  = 5 condition is 533.20 µmol g-1 after a 9 h reaction, while that of DMM in CH3OH + Ar and CH3OH + CO2 without adjusting pH is about 866.59 and 5702.49 µmol g-1, which indicates that lower pH is not the main factor affecting the production of DMM in CH3OH (Supplementary Fig. 32). Moreover, to more precisely investigate the impact of *CO intermediates on the reaction, we checked DMM production by substituting CO2 with CO on Ag.W-BTO under the same reaction conditions. The marked increase in DMM production (7008.35 µmol g-1) observed when using CO and CH3OH as reactants, compared to the utilization of CO2 and CH3OH (5702.49 µmol g-1), strongly suggests the active participation of *CO intermediates in the synthesis process of DMM (Supplementary Fig. 33). Additionally, commercial TiO2 (P25) was substituted for BTO in Ag.W-BTO using the same NaBH4 reduction method to assess their catalytic activity. TiO2 was chosen for its minimal oxygen vacancies (Ov), allowing for comparison with Ag.W-BTO to assess whether the Ov are active sites in the reaction. DMM yields on TiO2 and Ag.W-TiO2 were found to be ~0 and 1688.35 µmol g-1, respectively (Supplementary Fig. 34). The presence of DMM on Ag.W-TiO2 (without Ov) suggests that Ov may not be the primary reactive sites. However, the lower DMM yield on Ag.W-TiO2 compared to Ag.W-BTO can be attributed to the weaker absorption of visible light, low specific surface area, and high electron and hole recombination rate of the TiO2 substrate22,36. Humid CO2 flowed instead of CO2 as a controlled experiment to explore the moisture effect. The produced amount of DMM is about 2307.10 µmol g-1. The reason why the amount decreases compared with normal conditions (CO2 + CH3OH) may be due to H2O oxidation in the reaction system, which competes with MOR. The high overpotential of H2O oxidation to O2 (1.23 V vs. RHE) than MOR (0.58 V vs. RHE) may result in slower reaction kinetics (Supplementary Fig. 35)17. In addition, when the frequency of the incident photons matches the oscillation frequency of the surface free electrons in Ag nanoparticles, the localized surface plasmon resonance (LSPR) effect occurs50. As for exploring the LSPR effect of deposition of Ag on the Ag.W-BTO sample, the surface temperatures of the Ag.W-BTO were assessed using an infrared thermal imager. Analysis of the temperature versus time curves for the Ag.W-BTO revealed a gradual increase from about 21–31 °C in 10 min and stabilization of about 31 °C within 30 min (Supplementary Fig. 36). This stabilization phenomenon is attributed to the equilibrium reached between the heat dissipation of the sample and its surrounding medium46.To further explore the reaction mechanism and identify the carbon and oxygen sources origin of the produced DMM, the result of CO2 and CH3OH isotope-labeling experiments were determined by Gas chromatography-mass spectrometry (GC-MS). In the non-labeled DMM, the highest intensity abundance signal (base ion peak) is m/z = 45 ([CH3OCH2]+). The ion fragment of m/z = 75 ([CH2OCH2OCH3]+) is more stable than molecular ion fragment (m/z = 76, [CH3OCH2OCH3]+) after ionized and fragmented by a mass spectrometer. The relative abundance intensity of m/z = 75 (the second highest intensity) is much higher than m/z = 76. Therefore, the second highest intensity ion fragment of m/z = 75 can determine the DMM51. As depicted in Fig. 5a, the second highest intensity ion peak of DMM derived from non-labeled CH3OH + CO2 was m/z = 75 (Fig. 5a-II), while that of the DMM derived from labeled 13CH3OH + CO2 reached m/z = 77 (Fig. 5a-I), evidently proving that the two carbon sources in DMM originate from CH3OH and the one carbon originates from CO2. It is also observed that the m/z values of other fragments derived from 13CH3OH + CO2 (Fig. 5a-I) such as [13CH3]+ (m/z = 16) and [13CH3O]+ (m/z = 32), and m/z = 46 [13CH3O12CH2]+ are elevated by one unit higher than those of DMM derived from non-labeled CH3OH and CO2 (m/z = 15 [12CH3]+, m/z = 31 [12CH3O]+, and m/z = 45 ([12CH3O12CH2]+) (Fig. 5a-II), suggesting that the two terminal carbons of DMM ([13CH3O]212CH2) are derived from 13CH3OH and the central carbon comes from 12CO2. In Fig. 5b, the ion fragment peak of [CH3]+ (m/z = 15) and [CH3O]+ (m/z = 31) are unchanged comparing CH3OH + 13CO2-derived DMM (Fig. 5b-I) and CH3OH + 12CO2-derived DMM (Fig. 5b-II). But, the 13CO2-derived DMM shows a base ion peak of [CH3O13CH2]+ (m/z = 46) and the secondary intensity ion fragment peak [CH2O13CH2OCH3]+ (m/z = 76) (Fig. 5b-I), which is only one m/z higher than that of the 12CO2 labeled case (m/z = 45 [CH3OCH2]+ and m/z = 75 [CH2OCH2OCH3]+) (Fig. 5b-II), which means middle position carbon in the DMM molecule originates from a CO2 source. In 13CH3OH + 13CO2 labeled experiments, it is clearly showed the base ion peak is m/z = 47 ([13CH3O13CH2]+) and the second intensity fragment is m/z = 78 ([13CH2O13CH2O13CH3]+) (Fig. 5c-I), respectively, which is two and three units higher than those of non-labeled m/z = 45 ([CH3OCH2]+) and m/z = 75 ([CH2OCH2OCH3]+) (Fig. 5c-II), respectively. In addition, comparing the result of isotope labeled 13CH3OH + 12CO2 experiments (Fig. 5a-I), the fragments of m/z = 47 ([13CH3O13CH2]+) and m/z = 78 ([13CH2O13CH2O13CH3]+) showed one unit higher than m/z = 46 ([13CH3O12CH2]+) and m/z = 77 ([13CH2O12CH2O13CH3]+), respectively. This result also confirmed that the middle carbon of DMM comes from CO2. Furthermore, comparing the result of isotope labeled 12CH3OH + 13CO2 experiments (Fig. 5b-I), the fragments of m/z = 47 ([13CH3O13CH2]+) and m/z = 78 ([13CH2O13CH2O13CH3]+) showed one and two unit higher than those of m/z = 46 ([12CH3O13CH2]+) and m/z = 76 ([12CH2O13CH2O12CH3]+), respectively, which exhibited that the two terminal carbons of DMM comes from CH3OH. In the only 13CH3OH labeled isotope experiment (Supplementary Fig. 37), the m/z = 78 peak corresponds to the ion of the [13CH2O13CH2O13CH3]+ structure (Supplementary Fig. 37-I), exhibiting a 3-mass unit increase compared to the non-labeled [12CH2O12CH2O12CH3]+ ion (Supplementary Fig. 37-II). Comparing the GC-MS results of the 13CH3OH + CO2 (Fig. 5a-I) and pure 13CH3OH isotope (Supplementary Fig. 37-I) experiments further confirmed not only two CH3OH oxidations but also CO2RR involvement in our DMM formation pathway. The 18O-labelled CH318OH with CO2 isotope experiment (CH318OH + CO2) was conducted to trace the O atoms in the generated DMM (Fig.5d). The base ion peak and second intensity peak of 18O-labled DMM are m/z = 47 ([CH318OCH2]+) and m/z = 79 ([CH218OCH218OCH3]+) (Fig. 5d-I), respectively, which exhibited two and four units higher than those of non-labeled DMM (m/z = 45 [CH3OCH2]+ and m/z = 75 [CH2OCH2OCH3]+) (Fig. 5d-II), verifying the all the O atoms in DMM are produced by the reaction of CH3OH.Fig. 5: The results of isotope labeling experiments.a GC-MS results of I) isotope labeled 13CH3OH + CO2 and II) isotope non-labeled 12CH3OH + CO2. b GC-MS results of I) isotope labeled CH3OH + 13CO2 and II) isotope non-labeled CH3OH + 12CO2. c GC-MS results of I) isotope labeled 13CH3OH + 13CO2 and II) isotope non-labeled 12CH3OH + 12CO2. d GC-MS results of I) isotope labeled CH318OH + CO2 and II) isotope non-labeled CH316OH + CO2. (a. u.) represents arbitrary units.Photoinduced electrons and holes separation and transportationThe separation and transportation of electrons and holes were determined by photoluminescence (PL) measurements of the BTO, W-BTO, Ag-BTO, and Ag.W-BTO catalysts (Supplementary Fig. 38)52,53. The PL measurements, which assess internal charge transfer in solid samples, are excited under the monochromatic wavelength (350 nm) that may not entirely mimic the actual reaction conditions. All measured samples displayed the same spectral peaks near 498 nm, while the peak intensity of Ag.W-BTO significantly declined, indicating that the Ag.W-BTO catalyst had a lower charge-carrier recombination rate. This is because Ag species tend to readily provide electrons for a CO2RR, while the W species prefer to provide holes to promote MOR. Ag.W-BTO displayed the strongest photocurrent response under light irradiation (Supplementary Fig. 39a). In addition, Ag.W-BTO showed the smallest arc radius of electrochemical impedance spectra (Supplementary Fig. 39b), indicating that Ag.W-BTO had the lowest charge-transfer resistance and highest photo-charge separation rate, which favor accelerated reaction kinetics and enhanced efficiency54.Additionally, fs-TA measurements were conducted under 350 nm laser-flash photolysis to further elucidate electron transfer behavior at room temperature, with the checking conditions exhibiting disparities compared to actual reaction conditions such as temperature, pressure, and reactant composition. As depicted in Fig. 6a–d, the fs-TA spectra of BTO, Ag-BTO, W-BTO, and Ag.W-BTO included both broad negative and positive signals. The negative absorption signals can be ascribed to stimulated emission and bleaching of the ground state, and the positive absorption signals correspond to the presence of excited-state absorption by electrons55. The negative signals of Ag-BTO and W-BTO were stronger than those of BTO due to the instantaneous generation of charge carriers from the ground state and their direct excitation to the emission state by fast-electron transfer (Fig. 6a–c)56,57. As shown in Fig. 6d, the negative peak center (near 410 nm) of Ag.W-BTO was nearly three times higher than that of BTO, which can be attributed to the promotion of the utilization of electrons and holes to prevent charge carriers from recombination by Ag and W species, respectively. The dynamical decay traces and fitting of the photo-excited charge carriers for BTO, Ag-BTO, W-BTO, and Ag.W-BTO were probed at 388 nm (Fig. 6e and Supplementary Table 7), and a tri-exponential decay function was applied to fit the kinetics traces. The BTO, Ag-BTO, and W-BTO only show delay times of τ1 and τ2, which can be ascribed to electron transfer to the shallow trapping state (TS) and deep TS, respectively. However, Ag.W-BTO has three different trapping signals (τ1 = 0.10 ps, τ2 = 6.71 ps, and τ3 = 172.05 ps), in which τ1 and τ2 are assigned to electron transfer to a shallow Ag species and shallow TS, respectively58, and τ3 is due to the recombination of the shallow-trapped electrons with holes59. The final charge transfer process is shown in Fig. 6f. The electron transitions to an excited state on the conductive band (CB), leaving holes at the VB under irradiation. Subsequently, partial electrons quickly transfer to Ag sites (τ1) and shallow TS (τ2); the remaining electrons are combined with unconsumed holes (τ3). The longer delay time of τ3 indicates a lower speed for the recombination of charge electrons and holes during the photo-redox process.Fig. 6: Fs-TA spectra.a BTO, (b) Ag-BTO, (c) W-BTO, and (d) Ag.W-BTO at 350 nm. e The normalized TA kinetics of BTO, Ag-BTO, W-BTO, and Ag.W-BTO, respectively. f A scheme for the charge-transfer process of Ag.W-BTO. ΔA(mOD) represents the change absorbance in milli-optical density.Study of reaction intermediates and mechanismIn-situ DRIFTS was used to identify the reaction intermediates and mechanisms in DMM synthesis. KBr served as a background (Supplementary Fig. 40) was removed before all measurements. CO2 and CH3OH were purged into the system for 1 h in the dark. All peaks were marked with dotted lines, with black cases representing negligible changes in absorption bands and colorful lines emphasizing gradual increments of the bands. Due to the continuous filling of CO2 gas with CH3OH, a high stretching vibration bond at 2300 cm-1 belonging to CO2 adsorption is evident (Supplementary Fig. 41)59. The band located in 1155 cm-1 corresponds to the spreading −CH3O group of CH3OH, which confirms that CH3OH is successfully adsorbed onto the Ag.W-BTO surface (Fig. 7b and c)60. The signals centered at 1377, 1455, 2860, and 3030 cm-1 were caused by the deformation of methyl groups (Fig. 7a and b). The absorption bands at 1492 and 2930 cm-1 can be attributed to the stretching signals of C−H bonds60,61. The bands at 3640 and 3727 cm-1 correspond to the isolated −OH group (provided from CH3OH) and the adsorbed −OH group on the surface of TiO2, respectively61. All these bands confirm the successful absorption of CO2 and CH3OH. After UV-visible light irradiation, new intermediates emerged. Three characteristic bands appeared at 1260, 1716, and 2060 cm-1, corresponding to *CH2O, *COOH, and *CO intermediates, respectively60,62. Due to the small amount of sample added, the formation of the *CH3O intermediate for CH3OH as a reactant of a −CH3O group was not readily observed during the reaction. However, upon doubling the sample amount, a subtle increment in the −CH3O peak became evident as time increased, which confirmed that it produced the *CH3O (Fig. 7c). Furthermore, to elucidate the preference of CO2RR and MOR in generating *CH2O over *CH3O intermediates on Ag and W species, respectively, controlled in-situ DRIFTS experiments were conducted by flowing CO2 with H2O (CO2 + H2O) and CH3OH with Ar (CH3OH + Ar) on Ag-BTO and W-BTO catalysts, respectively. During the in-situ DRIFTS reaction of flowing CO2 and H2O on the surface of Ag-BTO, the emergence of CO2RR intermediates such as *CO, *COOH, and notably *CH2O which are located about 2060, 1716, and 1260 cm-1 was observed, suggesting Ag with sensitivity to CO2RR and its tendency to produce *CH2O intermediates (Supplementary Fig. 42a). Conversely, in the case of W-BTO, the production of these three intermediates, especially *CH2O, did not increase with reaction time, indicating a lack of sensitivity towards *CH2O generation (Supplementary Fig. 42b). This suggests that Ag is more inclined to produce *CH2O intermediates. In the in-situ DRIFTS spectrum of flowing CH3OH and Ar on the surface of the W-BTO catalyst, a significant increase in the peak at 1155 cm-1 corresponding to the *CH3O intermediate was observed (Supplementary Fig. 43a). However, no notable enhancement in the intensity of the *CH3O peak was detected on the Ag-BTO surface when CH3OH and Ar were introduced (Supplementary Fig. 43b). This indicates that W species are more effective in promoting the oxidation of CH3OH to *CH3O intermediates. In addition, in-situ DRIFTS experiments were conducted by flowing CO and CH3OH to explore whether CO participate in the reaction. The bands of *CO (2060 cm-1) and *CH3O (1155 cm-1) intermediates are produced immediately after the adsorption of CO and CH3OH for 1 h in the dark (Supplementary Fig. 44). Subsequently, a band corresponding to the *CH2O intermediate (around 1260 cm-1) reduced by *CO, appeared after UV-visible light irradiation which is consistent with our performance result. The peak intensity of these intermediates (*CO, *CH2O, and *CH3O) exhibits a smaller increase during the reaction, possibly due to the consumption of these intermediates during the synthesis of DMM at a higher reaction rate.Fig. 7: In-situ DRIFTS spectroscopy characterization of CO2 + CH3OH on Ag.W-BTO.a, b In-situ DRIFTS spectra at detailed reaction times (0, 10, 20, 30, 60, 120, 180, 240 min) with enlarged view of shaded area. c In-situ DRIFTS spectra from 1200 – 1100 upon doubling Ag.W-BTO amount. (a. u.) represents arbitrary units.Although W and Ag tend to serve as oxidation and reduction sites, respectively, their roles have been confirmed by in-situ DRIFTS, band gap structures, and controlled experiments. The electronic structures and adsorption characteristics concerning CO2 and CH3OH were investigated through DFT calculations to further delve into the mechanisms and active sites for this reaction. The optimized structures of Ag.W-BTO were established based on TEM, XPS, and XANES analyses (Supplementary Fig. S45). The adsorption energies of CO2 and CH3OH on distinct active sites (W, Ag, Ti, and Ov) were analyzed to confirm the reduction and oxidation sites on the Ag.W-BTO sample (Supplementary Fig. 46 and 47). As depicted in Fig. 8a, all the active sites (W, Ag, Ti, and Ov) displayed negative values for CO2 adsorption energies, indicating a higher tendency for CO2 adsorption across these sites. The most negative adsorption energy for CO2 (–2.413 eV) signifies a preference for CO2 adsorption on the Ag sites within the Ag.W-BTO sample. This highlights the heightened propensity of the Ag sites within Ag.W-BTO for CO2RR. The most negative adsorption energy (–1.375 eV) observed for W species during CH3OH adsorption suggests their higher inclination towards undergoing MOR. In addition, the Charge density difference and Bader charge analyses were conducted to explore the reason for the superior performance of Ag.W-BTO. The Ag-BTO and Ag.W-BTO were chosen as an analysis model due to their better performance in CO2RR (Fig. 8b). The adsorbed CO2 species on Ag.W-BTO gained a notably higher quantity of electrons (1.49 e-) than on Ag-BTO (1.46 e-). These results confirmed more electrons can participate in the CO2RR process in Ag.W-BTO. It further validates that the combined effect of Ag and W facilitates the separation and transfer of electrons and holes during the reaction process on Ag.W-BTO, which is consistent with findings from fs-TA and PL data.Fig. 8: DFT calculations.a Calculated CO2 and CH3OH adsorption energy on the different active sites (Ag, W, Ti, and Ov) of Ag.W-BTO, respectively. b Charge density difference and Bader charge analysis of CO2 adsorbed on the Ag-BTO and Ag.W-BTO surface. c Free energy diagram of DMM production via different pathway. d Geometries of reaction intermediates involved in MOR coupling with CO2RR pathways.To further explore the formation pathway of DMM in the coexistence of CO2 and CH3OH, the free energy profile and reaction pathway of the DMM synthesis were compared on the Ag.W-BTO in different routes (MOR pathway and MOR coupling with CO2RR (MOR + CO2RR) pathway, Fig. 8c). The optimized intermediate structures corresponding to each reaction steps are displayed in Fig. 8d and Supplementary Fig. 48. The DMM formation on Ag.W-BTO through two pathways is triggered by the thermodynamical spontaneity from CH3OH to *CH3OH with an energy barrier of −0.98 eV. Then, *CH3OH is further oxidated to *CH3O with an energy barrier of −0.65 eV. Thereafter, it was found that the process from *CH3O to *CH3O + *CO2 with an energy barrier of –0.31 eV is more thermodynamically favorable than the oxidation of *CH3O to*CH2O with an energy barrier of −0.09 eV, which suggests that *CH3O tends to couple with *CO2 intermediates rather than to form individual *CH2O intermediates in CH3OH and CO2 coexist system. In addition, the process of the next step (*CH2O + 2*CH3O) in MOR pathway free energy barrier (0.98 eV) requires higher than *CH3O + *COOH (0.53 eV) in MOR + CO2RR route. That means direct oxidation of CH3OH to DMM is unfavorable. Moreover, three steps require energy input in the MOR + CO2RR pathway, including the conversion of *CO2 to *COOH intermediates, the reduction of *CO to *CH2O, and the coupling of *CH2O with *CH3O. Particularly, the conversion of CO2 to *COOH intermediates consumes the highest energy of 0.53 eV and is considered the rate-determining step during the DMM formation process.According to the in-situ DRIFTS analysis, systematic isotope labeling experiments, and DFT calculations, the most probable reaction paths for the catalytic system are:Overall$${2{{{{\rm{CH}}}}}}_{3}{{{{\rm{OH}}}}}+{{{{{\rm{CO}}}}}}_{2}\to {({{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}})}_{2}{{{{{\rm{CH}}}}}}_{2}({{{{\rm{DMM}}}}})+{{{{{\rm{O}}}}}}_{2}$$In details$${{{{\rm{Ag}}}}}.{{{{\rm{W}}}}}\!-\!{{{{\rm{BTO}}}}}{\to }^{hv}{{{{{\rm{h}}}}}}^{+}+{{{{{\rm{e}}}}}}^{-}$$MOR:$${{{{{{\rm{CH}}}}}}_{3}{{{{\rm{OH}}}}}}\, _{{{{{\rm{ads}}}}}}\, ({{{{\rm{W}}}}})+{{{{\rm{h}}}}}^{+} \to {{\bullet }}{{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}}+{{{{\rm{H}}}}}^{+}$$$${{{{{{\rm{CH}}}}}}_{3}{{{{\rm{OH}}}}}}\, _{{{{{\rm{ads}}}}}}\, ({{{{\rm{W}}}}})\to {{\bullet }}{{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}}+{{{{{\rm{H}}}}}}^{+}-{{{{{\rm{h}}}}}}^{+}$$CO2RR:$${{{{{{\rm{CO}}}}}}_{2}}\, _{{{{{\rm{ads}}}}}}\, ({{{{\rm{Ag}}}}})+{{{{{\rm{e}}}}}}^{-}+{{{{{\rm{H}}}}}}^{+}\to {{\bullet }}{{{{\rm{COOH}}}}}$$$${{\bullet }}{{{{\rm{COOH}}}}}+{{{{{\rm{e}}}}}}^{-}+{{{{{\rm{H}}}}}}^{+}\to {{\bullet }}{{{{\rm{CO}}}}}+{{{{{\rm{H}}}}}}_{2}{{{{\rm{O}}}}}$$$${{{{{\rm{H}}}}}}_{2}{{{{\rm{O}}}}}\to {2{{{{\rm{H}}}}}}^{+}+{2{{{{\rm{e}}}}}}^{-}+{{\bullet }}{{{{\rm{O}}}}}$$$${{\bullet }}{{{{\rm{CO}}}}}+{2{{{{\rm{e}}}}}}^{-}+{2{{{{\rm{H}}}}}}^{+}\to {{\bullet }}{{{{{\rm{C}}}}}{{{{\rm{H}}}}}}_{2}{{{{\rm{O}}}}}$$Total of CO2RR$${{{{{{\rm{CO}}}}}}_{2}}\, _{{{{{\rm{ads}}}}}}\, ({{{{\rm{Ag}}}}})+{2{{{{\rm{e}}}}}}^{-}+{2{{{{\rm{H}}}}}}^{+}\to {{\bullet }}{{{{{\rm{CH}}}}}}_{2}{{{{\rm{O}}}}}+{{\bullet }}{{{{\rm{O}}}}}$$$${{{{{{\rm{CO}}}}}}_{2}}\, _{{{{{\rm{ads}}}}}}\, ({{{{\rm{Ag}}}}})\to \underline{{{\bullet }}{{{{{\rm{CH}}}}}}_{2}{{{{\rm{O}}}}}+{{\bullet }}{{{{\rm{O}}}}}-{2{{{{\rm{e}}}}}}^{-}-{2{{{{\rm{H}}}}}}^{+}}$$Coupling process:$${2{{{{\rm{CH}}}}}}_{3}{{{{\rm{OH}}}}}+{{{{{\rm{CO}}}}}}_{2}\to {({{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}})}_{2}{{{{{\rm{CH}}}}}}_{2}+{{{{{\rm{O}}}}}}_{2}$$$$ 2(\underline{{{\bullet }}{{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}}+{{{{{\rm{H}}}}}}^{+}-{{{{{\rm{h}}}}}}^{+}})+(\underline{{{\bullet }}{{{{{\rm{CH}}}}}}_{2}{{{{\rm{O}}}}}+{{\bullet }}{{{{\rm{O}}}}}-{2{{{{\rm{e}}}}}}^{-}-{2{{{{\rm{H}}}}}}^{+}})\\ \to (2{{\bullet }}{{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}}+{2{{{{\rm{H}}}}}}^{+}-{2{{{{\rm{h}}}}}}^{+})+({{\bullet }}{{{{{\rm{CH}}}}}}_{2}{{{{\rm{O}}}}}+{{\bullet }}{{{{\rm{O}}}}}-{2{{{{\rm{e}}}}}}^{-}-{2{{{{\rm{H}}}}}}^{+})\\ \to 2{{\bullet }}{{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}}+{{\bullet }}{{{{{\rm{CH}}}}}}_{2}{{{{\rm{O}}}}}+{{\bullet }}{{{{\rm{O}}}}}+{2{{{{\rm{H}}}}}}^{+}-{2{{{{\rm{H}}}}}}^{+}\to {({{{{{\rm{CH}}}}}}_{3}{{{{\rm{O}}}}})}_{2}{{{{{\rm{CH}}}}}}_{2}({{{{\rm{DMM}}}}})+{{{{{\rm{O}}}}}}_{2}$$CH3OH and CO2 are adsorbed on the surface of Ag.W-BTO in the dark, with more significant adsorption evidence on the W and Ag active sites, respectively. During UV-visible light irradiation, electrons (e−) in the VB of Ag.W-BTO are excited to the CB, while the holes (h+) remain in the VB. Some e− transfer to Ag sites by Schottky-junction for CO2RR, leaving h+ to oxidize CH3OH on the W sites. CH3OH is oxidized to form *CH3O and H+ intermediates after binding to h+ on the W sites. Simultaneously, CO2 molecules obtain e− and combine with H+ to produce COOH* on the active Ag sites. Furthermore, *COOH is reduced to *CO and then changed to *CH2O by a multi-step proton-coupled electron transfer (PCET) process. Finally, DMM is produced by coupling *CH2O with *CH3O.