Asymmetric Cu(I)─W Dual‐Atomic Sites Enable C─C Coupling for Selective Photocatalytic CO₂ Reduction to C₂H₄

Abstract

Solar‐driven CO₂ reduction into value‐added C₂₊ chemical fuels, such as C₂H₄, is promising in meeting the carbon‐neutral future, yet the performance is usually hindered by the high energy barrier of the C─C coupling process. Here, an efficient and stabilized Cu(I) single atoms‐modified W₁₈O₄₉ nanowires (Cu₁/W₁₈O₄₉) photocatalyst with asymmetric Cu─W dual sites is reported for selective photocatalytic CO₂ reduction to C₂H₄. The interconversion between W(V) and W(VI) in W₁₈O₄₉ ensures the stability of Cu(I) during the photocatalytic process. Under light irradiation, the optimal Cu₁/W₁₈O₄₉ (3.6‐Cu₁/W₁₈O₄₉) catalyst exhibits concurrent high activity and selectivity toward C₂H₄ production, reaching a corresponding yield rate of 4.9 µmol g⁻¹ h⁻¹ and selectivity as high as 72.8%, respectively. Combined in situ spectroscopies and computational calculations reveal that Cu(I) single atoms stabilize the *CO intermediate, and the asymmetric Cu─W dual sites effectively reduce the energy barrier for the C─C coupling of two neighboring CO intermediates, enabling the highly selective C₂H₄ generation from CO₂ photoreduction. This work demonstrates leveraging stabilized atomically‐dispersed Cu(I) in asymmetric dual‐sites for selective CO₂‐to‐C₂H₄ conversion and can provide new insight into photocatalytic CO₂ reduction to other targeted C₂₊ products through rational construction of active sites for C─C coupling.

An efficient Cu(I) single atoms‐modified W₁₈O₄₉ (Cu₁/W₁₈O₄₉) photocatalyst is rationally designed for photocatalytic CO₂ reduction to C₂H₄. Compared to W₁₈O₄₉ that produces CO only, Cu(I) single atoms stabilize *CO intermediates and the asymmetric Cu─W dual sites significantly reduce the energy barrier of C─C coupling, thus leading to the highly selective C₂H₄ generation from CO₂ photoreduction in Cu₁/W₁₈O₄₉.

1. Introduction

Solar‐driven conversion of CO₂ and water into high‐value‐added chemical fuels represents a promising strategy to mitigate the global greenhouse effect and fossil energy crisis.^{[ 1} , ² , ³ , ⁴ , ⁵ , ^{6 ]} As photocatalytic CO₂ reduction comprises multiple proton‐coupled electron transfer (PCET) processes, a variety of carbon‐containing products, including CO, CH₃OH, CH₄, and even advanced hydrocarbons, are thus obtained. Due to the favorable kinetics, C₁ compounds such as CO and CH₄ dominate the products of photocatalytic CO₂ reduction in most cases.^{[ 7} , ⁸ , ⁹ , ^{10 ]} By contrast, C₂₊ compounds generated by the C─C coupling are considered to be more intriguing because of their high energy density and rich chemical reactivity. Of particular attention has been paid to the C₂H₄ product, which is a highly important chemical raw material that is widely used in rubber industry, medicine, and agriculture.^{[ 11} , ¹² , ^{13 ]} However, the high energy barrier of the C─C coupling make it rather difficult to achieve C₂H₄ production with concurrent high activity and selectivity from photocatalytic CO₂ reduction.^{[ 14} , ¹⁵ , ^{16 ]}

Among CO₂ reduction catalysts, Cu‐based catalysts are of great interest for their excellent ability to produce C₂H₄ and other C₂₊ products.^{[ 17} , ¹⁸ , ¹⁹ , ^{20 ]} In electrocatalysis, Cu species, especially in the form of residual Cu(I), are considered as the most thermodynamically‐favorable catalysts for C─C coupling,^{[ 21} , ²² , ^{23 ]} where appropriate binding strength to the *CO intermediate facilitates their subsequent coupling process.^{[ 24} , ²⁵ , ²⁶ , ^{27 ]} For photocatalytic process, however, the valence state of Cu(I) species (Cu₂O) is variable, which could be reduced to metallic Cu⁰ by photogenerated electrons or oxidized to Cu(II) by photogenerated holes, thus leading to the unwanted deactivation. Consequently, it is ideal to develop and maintain stable Cu(I) active sites during photocatalytic CO₂ reduction. Moreover, the production of C₂H₄ necessitates the coupling of adsorbed CO molecules at two neighboring sites to generate the key *CO─CO intermediate.^{[ 28} , ^{29 ]} Nevertheless, it becomes quite difficult to perform C─C coupling in conventional semiconductor photocatalysis, where almost identical charge distribution between symmetric double Cu sites inevitably results in strong dipole‐dipole repulsion, thereby improving the energy barrier for C─C coupling. Alternatively, design of photocatalyst systems with asymmetric dual sites that are able to attenuate the dipole repulsion is more appealing.^{[ 30} , ^{31 ]} Therefore, it is of paramount importance to develop stable and asymmetric Cu(I)‐containing dual sites during photocatalytic CO₂ reduction, which could accelerate C─C coupling process toward C₂H₄ formation, and meanwhile this is rather challenging at the atomic level.

The unique crystal structure enables W₁₈O₄₉ to possess a high concentration of lattice defects and distortions, thereby providing the increased doping capacity of heteroatom atoms (e.g., Cu atoms).^{[ 32} , ^{33 ]} Furthermore, according to the standard electrode potential, the abundant W⁵⁺ ions in W₁₈O₄₉ can reduce Cu²⁺ ions to Cu⁺ ions (Equation 1).^{[ 34 ]} As a consequence, the interconversion between W(V) and W(VI) in W₁₈O₄₉ could ensure the stability of Cu(I) during the photocatalytic process.

$${\mathrm{W}}^{5+}+\mathrm{C}{\mathrm{u}}^{2+}\to {\mathrm{W}}^{6+}+\mathrm{C}{\mathrm{u}}^{+}$$ (1)

Guided by the two features above, the defect‐rich W₁₈O₄₉ ultrathin nanowires were chosen as an ideal support for construction of stable and asymmetric Cu(I)‐containing dual sites. To this end, through rational incorporation of Cu(I) single atoms in W₁₈O₄₉ nanowires (Cu₁/W₁₈O₄₉), we have reported here the realization of photocatalytic CO₂ reduction to C₂H₄ with concurrent high activity and selectivity. The interconversion between W(V) and W(VI) in W₁₈O₄₉ ensures the stability of Cu(I) during the photocatalytic process, and the resulting asymmetric Cu─W dual sites were thus constructed to stabilize *CO intermediates and promote C─C coupling. At an optimal concentration, Cu₁/W₁₈O₄₉ catalyst exhibits a high activity (4.9 µmol g⁻¹ h⁻¹) and high selectivity (72.8%) for CO₂ photoreduction toward C₂H₄ production. In situ near‐ambient pressure X‐ray photoelectron spectroscopy (NAP‐XPS) and diffuse reflectance infrared Fourier‐transform spectroscopy (DRIFTS) measurements were carried out to elucidate the transfer direction of photoinduced charge, decipher the catalytically‐active sites and reveal the reaction path for photocatalytic CO₂ reduction to C₂H₄. The detailed experimental characterizations and density functional theory (DFT) calculations unravel that both Cu(I) and W(V) atoms are enriched in electrons as likely active sites for photocatalytic CO₂ reduction toward C₂H₄ evolution, and the asymmetric Cu─W dual sites effectively reduce the energy barrier for the coupling of two neighboring CO intermediates.

2. Results and Discussion

2.1. Synthesis and Characterizations

The Cu‐modified W₁₈O₄₉ catalysts with different Cu loadings (x wt.%) were synthesized by a one‐step hydrothermal method (see details in the Supporting Information), and the resulting samples were designated as x‐Cu₁/W₁₈O₄₉ (x = 1.4−7.0). The actual concentration of Cu in various Cu₁/W₁₈O₄₉ samples was measured by inductively coupled plasma mass spectrometry (ICP‐MS) and found to be in the range of 1.29−5.01 wt.% (Table S1, Supporting Information). As shown in the X‐ray diffraction (XRD) patterns (Figure S1, Supporting Information), all samples are well indexed to monoclinic W₁₈O₄₉ (JCPDS No. 36–101) phase, and no diffraction peaks of Cu‐containing species (i.e., Cu and CuO_x) are observed. Additionally, as Cu loading increases, the spacing of characteristic W₁₈O₄₉ (020) crystal plane becomes widened with a decreased signal‐to‐noise ratio, indicating the successful doping of Cu into the lattice of W₁₈O₄₉ host material.^{[ 35 ]} The morphological information of the samples was revealed by scanning electron microscopy (SEM, Figure S2, Supporting Information) and transmission electron microscopy (TEM, Figure S3, Supporting Information). Analogously to pristine W₁₈O₄₉, Cu‐modified W₁₈O₄₉ samples consist of homogeneous nanowires with lengths of 200−500 nm and diameters of 5−10 nm, and no aggregated nanoparticles are observed on the surfaces. To further reveal the distribution and configuration of Cu species in Cu₁/W₁₈O₄₉, high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) with atomic resolution was carried out. From the low‐resolution HAADF‐STEM images of 3.6‐Cu₁/W₁₈O₄₉ sample (Figure 1a,b), well‐defined nanowires with a lattice spacing of 0.38 nm are observed, corresponding well to the (010) crystal plane spacing of monoclinic W₁₈O₄₉, indicating that the nanowires grow along the [010] direction. As shown in Figure 1c, no clusters or nanoparticles of Cu species are found in the HAADF‐STEM image at atomic resolution, verifying that Cu species are highly dispersed at the atomic level in W₁₈O₄₉ nanowires. Additionally, nanometer‐resolution energy‐dispersive X‐ray (EDX) elemental mapping (Figure 1d) displays that Cu, W, and O are uniformly distributed in the nanowire, further confirming the atomically‐dispersed Cu in the W₁₈O₄₉ structure.

Figure 1.

Morphological characterizations. a,b) Low‐magnification HAADF‐STEM images of 3.6‐Cu₁/W₁₈O₄₉. c) Atomic resolution HAADF‐STEM image of 3.6‐Cu₁/W₁₈O₄₉. d) EDX mapping images of Cu (red), W (green), and O (cyan) in 3.6‐Cu₁/W₁₈O₄₉ nanowire. The scale bars are 5 nm.

To elucidate the oxidation state of Cu species and its interaction with the W₁₈O₄₉ support, high‐resolution X‐ray photoelectron spectra (XPS, Figure 2a) were carried out. As presented in Figure 2a, the primary Cu 2p _3/2 peak was observed to be located at a binding energy of 932.0 eV, corresponding well to the monovalent Cu(I) ions.^{[ 36} , ^{37 ]} Additionally, the existence of Cu(II) is excluded by the absence of its characteristic satellite peaks. As shown in the Cu LMM Auger spectrum (Figure 2b), only a characteristic peak for Cu(I) at the kinetic energy of 915.8 eV is detected, while the characteristic peak for Cu⁰ at 918.3 eV is absent.^{[ 38} , ^{39 ]} The results above indicate that the Cu²⁺ ions added to the precursor solution were reduced to Cu(I) anchored on the W₁₈O₄₉ nanowires. According to the standard electrode potential, W(V) can reduce Cu²⁺ ions to Cu⁺ ions (Equation 1), but it cannot be determined whether W(V) can reduce Cu²⁺ to Cu(I) sites. However, it can be seen from the XPS spectrum of W 4f (Figure S4, Supporting Information) that the introduction of Cu single atoms significantly reduces the content of W⁵⁺ from 26.12% to 23.55%, indicating that W(V) can reduce Cu²⁺ to Cu(I) sites. This result also indicates that existence of a fast electron transfer channel between Cu(I) sites and W atoms. At the same time, due to the decrease of W⁵⁺ content, a decrease in the visible light absorption of Cu₁/W₁₈O₄₉ is seen compared to pristine W₁₈O₄₉ (Figure S5, Supporting Information). As a consequence, the Cu species in Cu₁/W₁₈O₄₉ can be slightly reduced by electrons from W(V) of W₁₈O₄₉ and thus maintain its oxidation state of +1.

Figure 2.

Structural characterizations of Cu₁/W₁₈O₄₉. a) Cu 2p XPS spectra and b) Cu LMM Auger spectrum of Cu₁/W₁₈O₄₉. c) Normalized Cu K‐edge XANES spectra, d) FT‐EXAFS spectra. e) WT‐EXAFS of Cu K‐edge for Cu₁/W₁₈O₄₉, Cu foil, Cu₂O, and CuO references.

To further determine the chemical environment and coordination structure of atomically‐dispersed Cu species in Cu₁/W₁₈O₄₉, Cu K‐edge X‐ray absorption near‐edge structure (XANES) and extended X‐ray absorption fine structure (EXAFS) spectra were performed. As shown in Figure 2c, in contrast to CuO and Cu foil, the normalized Cu K‐edge XANES spectrum of 3.6‐Cu₁/W₁₈O₄₉ exhibits an absorption edge more close to Cu₂O reference. From the linear fitting for the peak position of the first derivative in XANES curves (Figure S6, Supporting Information), it is verified that the valence state of Cu species in Cu₁/W₁₈O₄₉ is located ≈+1, and this finding is in good line with the XPS results (Figure 2a,b). rom the Fourier transform EXAFS (FT‐EXAFS) spectra (Figure 2d), 3.6‐Cu₁/W₁₈O₄₉ displays a dominant peak belonging to the Cu─O bond at ≈1.5 Å. Compared to Cu‐based references, either the Cu─Cu bond (2.25 Å for Cu foil) or Cu─O─Cu bond (2.45 Å for CuO and 2.72 Å for Cu₂O) is absent in Cu₁/W₁₈O₄₉, suggesting an isolated distribution of Cu atoms in W₁₈O₄₉ nanowires. The accurate coordination structure of Cu₁/W₁₈O₄₉ was obtained through least‐squares fitting of the K‐edge EXAFS data (Figure S7 and Table S2, Supporting Information). Notably, the fitting results reveal a Cu─O coordination number of 3.89 in 3.6‐Cu₁/W₁₈O₄₉ nanowires, indicating that Cu atoms are predominantly anchored on the surface of W₁₈O₄₉ and coordinated by oxygen atoms. This result is in good accordance with the electron spin resonance (ESR, Figure S8, Supporting Information) test, where the concentration of oxygen vacancies (g = 2.003) gradually decreases with an increase in Cu doping ratio, probably due to the occupation of oxygen vacancies by a significant amount of atomically‐dispersed Cu sites.^{[ 40 ]} Apart from Cu─O bonds, 3.6‐Cu₁/W₁₈O₄₉ also shows a considerable portion of Cu─W path with a coordination number of 0.88 from the fitted second shell layer. The wavelet transform (WT) analysis was performed to obtain more intuitive data on atom distance (Figure 2e). The WT intensity maxima of Cu₁/W₁₈O₄₉ is observed ≈4.0 Å⁻¹, which overlaps with the Cu─O bond of Cu₂O reference. In contrast, the intensity maximum of Cu foil at 6.7 Å⁻¹ ascribed to the Cu─Cu bond is not observed in Cu₁/W₁₈O₄₉, further confirming the presence of atomically isolated Cu species in Cu₁/W₁₈O₄₉.

2.2. Photocatalytic CO₂ Reduction Evaluation

The photocatalytic CO₂ reduction performance of the as‐prepared samples were evaluated in a solid‐gas system using a 300 W Xe lamp as the light source without any sacrificial agent (Figure S9, Supporting Information). The gaseous products were monitored and quantified by gas chromatography (GC, Figure S10, Supporting Information). In the case of pristine W₁₈O₄₉, only CO product was obtained with a yield rate of 7.1 µmol g⁻¹ h⁻¹ (Figure 3a). Interestingly, the introduction of atomically‐dispersed Cu sites improves the photocatalytic activity and selectivity, along with the production of CH₄ and C₂H₄. This suggests that Cu single atoms could stabilize the adsorption of CO and facilitate its subsequent PCET process to generate hydrocarbons. For 1.4‐Cu₁/W₁₈O₄₉ sample, apart from CO production, it exhibited CH₄ and C₂H₄ production rates of 5.7 and 0.33 µmol g⁻¹ h⁻¹, respectively, with a low selectivity of 6.7% for C₂H₄ product. At an optimal doping concentration of Cu single atoms, C₂H₄ production reaches the maximum over 3.6‐Cu₁/W₁₈O₄₉ photocatalyst. As shown in Figure 3b, 3.6‐Cu₁/W₁₈O₄₉ photocatalyst exhibited a near‐linear increase of gas phase products with prolonged light irradiation time, and could deliver a yield rate of 4.9 µmol g⁻¹ h⁻¹ for C₂H₄ product, along with considerable production of CO (5.1 µmol g⁻¹ h⁻¹) and CH₄ (2.2 µmol g⁻¹ h⁻¹). In addition, O₂ as the oxidation product was detected by GC and found to increase linearly with the reaction time (Figure S11, Supporting Information). Accordingly, a high selectivity of 82.5% for hydrocarbons (CH₄ + C₂H₄) and 72.8% for C₂H₄ is thus reached over 3.6‐Cu₁/W₁₈O₄₉ photocatalyst. Notably, in this work, 3.6‐Cu₁/W₁₈O₄₉ photocatalyst exhibits a satisfactory concurrent high activity and high selectivity of C₂H₄ production from solar‐driven CO₂ reduction, outperforming most of the reported Cu(I)‐based photocatalysts (Table S3, Supporting Information) and being comparable to those state‐of‐the‐art photocatalysts systems toward C₂₊ products (Table S4, Supporting Information). Further increase of Cu doping concentrations, however, leads to the decreased activity and selectivity of C₂H₄ product over 5.6‐Cu₁/W₁₈O₄₉ (3.7 µmol g⁻¹ h⁻¹, 64.4%) and 7.0‐Cu₁/W₁₈O₄₉ (1.53 µmol g⁻¹ h⁻¹, 25.3%) photocatalysts, which could be probably due to their inferior separation capacity of photogenerated carriers (Figure S12, Supporting Information).

Figure 3.

Photocatalytic CO₂ reduction performance evaluation. a) The yield rate and selectivity of products over different catalysts. b) The yield rate products as a function of reaction time over 3.6‐Cu₁/W₁₈O₄₉. c) The measured wavelength‐dependent gas yield rates with monochromatic incident light irradiation for 4 h. d) The control experiments of photocatalytic CO₂ reduction performances under different conditions, where N.D. denotes product not detected. e) MS spectrum of ¹³C₂H₄ (m/z = 30) production from photocatalytic ¹³CO₂ reduction over 3.6‐Cu₁/W₁₈O₄₉. Inset: the corresponding GC spectrum. f) Cycling measurements for CO₂ photoreduction.

The wavelength‐dependent experiments were carried out with monochromatic incident light irradiation. As shown in Figure 3c, the yield rate of C₂H₄ product from photocatalytic CO₂ reduction corresponds well to the light absorption of 3.6‐Cu₁/W₁₈O₄₉ photocatalyst, giving direct proof that CO₂ reduction is driven by solar energy. To exclude the possible carbon contamination, a series of controlled experiments were conducted. From Figure 3d, it demonstrates that almost no gas‐phase products were detected in the absence of light irradiation, CO₂ feeding gas, or Cu₁/W₁₈O₄₉ photocatalyst, thus verifying the C₂H₄ product is derived from light‐driven photocatalytic CO₂ reduction process on Cu₁/W₁₈O₄₉. Furthermore, isotope labeling experiments using ¹³CO₂ feeding gas were conducted to verify the origin of carbon in the product. As depicted in Figure 3e, along with the GC spectrum, peaks of ¹³C₂H₄ (m/z = 30), ¹³CO (m/z = 29), and ¹³CH₄ (m/z = 17) were detected in the mass spectrometry (MS) spectrum, confirming that the products resulted from CO₂ reduction rather than possible carbon contamination. Apart from activity and selectivity, stability is another important factor in assessing the performance of photocatalysts. To demonstrate the stability of Cu(I), the Auger spectra of Cu₁/W₁₈O₄₉ after photocatalytic CO₂ reaction was also measured (Figure S13, Supporting Information). As expected, the oxidation state of Cu in Cu₁/W₁₈O₄₉ remained +1 after photocatalytic CO₂ reaction, confirming the effective stabilization of Cu(I). Moreover, after four consecutive photocatalytic cycles, 3.6‐Cu₁/W₁₈O₄₉ photocatalyst maintained nearly constant activity (Figure 3f) and unaltered structure (Figures S14–S16, Supporting Information), demonstrating its good stability in both performance and structure.

2.3. Active Sites Identification

To understand the high activity and selectivity of C₂H₄ production over 3.6‐Cu₁/W₁₈O₄₉ photocatalyst for CO₂ reduction, a comprehensive study was conducted to decipher the possible active sites, especially the C─C coupling sites. To this end, in situ NAP‐XPS measurements were performed to illustrate the transfer direction of photoinduced charges and identify the probable active site during photocatalysis. As shown in Figure 4a, in clear contrast to dark condition, a dramatic shift of the Cu(I) peak (952.0 eV) to lower binding energy (951.2 eV) can be observed upon light irradiation, indicating that Cu(I) single atoms are enriched with photoinduced‐electrons. More importantly, when CO₂ gas is introduced into the system, the peak of Cu(I) shifts back toward higher binding energy (951.8 eV), which indicates that Cu(I) ions lose partial electrons by donating them to the surface‐adsorbed CO₂ molecules.^{[ 41} , ^{42 ]} In addition, after removing the light irradiation, Cu(I) single atoms recover to their original state at 952.0 eV and in such way Cu(I) single atoms maintain stability after the reaction. According to previous studies,^{[ 41} , ⁴³ , ^{44 ]} the electron transfer processes under illumination involve two main stages: a rapid accumulation of photogenerated electrons and a slower chemical reduction process. The former typically occurs within seconds and importantly, it is reversible compared to the latter. In our test, we clearly observe the sample returning to its original state after the illumination ends, indicating that only the reversible first stage of photogenerated electron accumulation on Cu(I) occurs under illumination, and there is no chemical reduction of Cu(I) during this process. This observation further supports the stability of Cu(I) single atoms during photocatalytic reactions. Analogously, a similar phenomenon is observed in the in‐situ NAP‐XPS spectra of W 4f in 3.6‐Cu₁/W₁₈O₄₉ (Figure 4b). Compared to dark condition, light irradiation enables the W atoms to capture photogenerated electrons to give birth to W(V) ions with a lower valence state. Generally, the Cu₁/W₁₈O₄₉ catalyst contains two types of W atoms, that is, one close to the Cu single atoms to form Cu─W dual sites and one far from the Cu single atoms. The electron transfer in the light of the specific two W atoms can be further determined by the product distribution of photocatalytic CO₂ reduction. Specifically, W atoms far from the Cu single atoms can only generate CO, while the W sites near the Cu single atoms can form asymmetric Cu─W dual sites to produce C₂H₄. The simultaneous production of CO and C₂H₄ in 3.6‐Cu₁/W₁₈O₄₉ indicates that both W atoms are enriched in electrons for CO₂ reduction. Upon further introduction of CO₂ gas, W(V) ions are partially re‐oxidized to W(VI) ions by transferring electrons to surface‐adsorbed CO₂ molecules, and eventually only a portion of W atoms recover to their original state when the light irradiation is off. As a result, the stability of Cu(I) can be ensured by the following reasons: i) The abundant W(V) in W₁₈O₄₉, which is more easily oxidized, can reduce Cu(II) to Cu(I) to avoid its oxidation; ii) Under light irradiation, the more easily reduced W(VI) around Cu(I) inhibits the excessive aggregation of photo‐generated electrons, thus avoiding the reduction of Cu(I) to Cu(0). Therefore, the coexistence of W(V) and W(VI) ensures the stability of Cu(I) sites during photocatalytic process. In summary, in‐situ NAP‐XPS measurement results indicate that both Cu(I) and W(V) atoms are probable active sites for photocatalytic CO₂ reduction toward C₂H₄.

Figure 4.

Active sites identification. a) Cu 2p and b) W 4f in situ NAP‐XPS spectra obtained over 3.6‐Cu₁/W₁₈O₄₉ photocatalyst. In‐situ DRIFTS spectra obtained during CO desorption on c) W₁₈O₄₉ and d) Cu₁/W₁₈O₄₉ under Ar purging at room temperature and ambient pressure.

To further determine the active sites of C─C coupling, we conducted in‐situ DRIFTS experiments using CO as a probe molecule, which is a key intermediate of CO₂ reduction to C₂H₄. In the test, the catalyst was initially exposed to CO gas adsorption until saturation and subsequently treated with Ar gas to desorb the physically‐adsorbed CO while retaining the chemically‐adsorbed CO. As shown in the DRIFT spectra of adsorbed CO on pristine W₁₈O₄₉ (Figure 4c) and 3.6‐Cu₁/W₁₈O₄₉ (Figure 4d) catalysts reveal two bands centered at 2170 and 2120 cm⁻¹. Generally, the peak located at 2170 cm⁻¹ represents physically adsorbed CO (mode‐I),^{[ 45} , ^{46 ]} where the adsorption peak in the range of 2150−2100 cm⁻¹ indicates more complex CO adsorption behavior (mode‐II).^{[ 47 ]} It is worth noting that the peak intensity of mode‐I is higher than that of mode‐II over pristine W₁₈O₄₉, while the peak intensity of mode‐II is higher than that of mode‐I after the introduction of Cu(I) single atoms in 3.6‐Cu₁/W₁₈O₄₉, confirming that Cu(I) single atoms indeed enhance the adsorption of CO. After Ar purging, both absorption peaks gradually weaken, and the CO absorption peak of W₁₈O₄₉ disappeared completely after 30 min, indicative of only weak physically‐adsorbed CO on W₁₈O₄₉. Intriguingly, following a long purge process, the mode‐I adsorption on 3.6‐Cu₁/W₁₈O₄₉ was completely eradicated, while mode‐II adsorption remains and can be further divided into two individual peaks located at 2120 and 2107 cm⁻¹, respectively. According to previous studies, the peak at 2107 cm⁻¹ is the characteristic peak of the linear adsorption of CO on Cu(I) sites.^{[ 48} , ⁴⁹ , ^{50 ]} Due to the absence of other oxidation states of Cu species in Cu₁/W₁₈O₄₉ as CO adsorption sites, the peak at 2120 cm⁻¹ is more likely attributed to CO chemisorption on W atoms. The CO adsorption free energy further supports this observation (Table S5, Supporting Information), as introducing Cu atoms reduce the CO adsorption energy on W atoms from −0.14 eV in W₁₈O₄₉ to −0.52 eV in Cu₁/W₁₈O₄₉. It is thus concluded that Cu(I) sites not only have a stronger adsorption strength of CO, but they stabilize the adsorbed CO on W atoms. This result also indicates that the W sites in the Cu─W dual sites can also have a stronger adsorption of *CO. The simultaneous enhancement of *CO adsorption strength in the Cu─W dual sites is conducive to the generation of two adjacent *CO, thereby accelerating the subsequent C─C coupling process. In addition, the disparity in the adsorption of CO on W₁₈O₄₉ and 3.6‐Cu₁/W₁₈O₄₉ was probed using CO temperature‐programmed desorption (CO‐TPD) analysis (Figure S17, Supporting Information). Pristine W₁₈O₄₉ exhibited two desorption peaks, that is, the weak (200−230 °C) and strong CO adsorption (350−380 °C), in the range of 100−600 °C. However, after the introduction of Cu(I) single atoms, both of the desorption peaks shifted toward higher temperatures, indicating that Cu(I) single atoms enhance the adsorption of CO.

To demonstrate the unique role of Cu(I) sites for CO adsorption, Cu(II)/WO₃ sample that consists entirely of Cu(II) ions was prepared by calcining 3.6‐Cu₁/W₁₈O₄₉ in air (see details in the Supporting Information). The corresponding characterizations (XRD and XPS, Figures S18 and S19, Supporting Information) showed that after calcination, W₁₈O₄₉ has been converted to WO₃ and Cu(I) has been oxidized to Cu(II) ions. Different from 3.6‐Cu₁/W₁₈O₄₉, Cu(II)/WO₃ exhibited analogous CO adsorption behavior to pristine W₁₈O₄₉, where complete CO desorption takes place after Ar gas purge (Figure S20, Supporting Information). As a consequence, Cu(II)/WO₃ displayed no production of C₂H₄ in the photocatalytic CO₂ reduction test (Figure S21, Supporting Information), where only CO (10.1 µmol g⁻¹ h⁻¹) and a small amount of CH₄ (1.4 µmol g⁻¹ h⁻¹) were produced. It is worth noting that the intensity and mode of CO significantly influence the energy barrier of C─C coupling and the selectivity of subsequent reduction products. Simultaneous adsorption of *CO molecules on adjacent Cu(I) and W(V) sites, that is asymmetric Cu─W dual sites, is an important process for the coupling of two CO molecules over the 3.6‐Cu₁/W₁₈O₄₉ photocatalyst. Consistent with the fitted XAFS and in‐situ NAP‐XPS results, the DRIFT spectra for CO desorption provide further evidence supporting the existence of Cu─W dual sites.

2.4. Reaction Mechanism Investigation

The band structure of Cu₁/W₁₈O₄₉ was first studied by Mott‐Schottky test (Figure S22a, Supporting Information). The flat band potential of 3.6‐Cu₁/W₁₈O₄₉ was −0.32 V (vs NHE), and its conduction band (CB) position was estimated to be −0.62 V (vs NHE). For n‐type semiconductors, the CB position is ≈0.3 V lower than the flat‐band potential, so the CB position of 3.6‐Cu₁/W₁₈O₄₉ is located at ≈−0.62 V (vs NHE), which is more negative than the reduction potential of CO₂/CO (−0.51 V vs NHE), CO₂/CH₄ (−0.24 V vs NHE) and CO₂/C₂H₄ (−0.33 V vs NHE). Apparently, it is indicated that the CB position of 3.6‐Cu₁/W₁₈O₄₉ is thermodynamically favorable for photocatalytic CO₂ reduction to carbon‐containing products of CO, CH₄ and C₂H₄ (Figure S22b, Supporting Information).

To gain a deeper understanding of the CO₂ photoreduction process, specifically the C─C coupling pathway, we conducted in‐situ DRIFTS measurements. As shown in Figure 5a, a series of CO₂ adsorption and reaction intermediates were monitored over the 3.6‐Cu₁/W₁₈O₄₉ photocatalyst under light irradiation. The peak observed at 1472 cm⁻¹ can be attributed to the asymmetric stretching vibrations of the HCO₃ ⁻ group, while the peaks at 1457 and 1558 cm⁻¹ are associated with the monodentate carbonate (m‐CO₃ ²⁻) group.^{[ 51} , ^{52 ]} In addition, the peaks observed at 1462, 1540, 2107, and 2120 cm⁻¹ are attributed to the vibrations of *CHO, *COOH, and *CO groups, respectively, which are important intermediates in the production of CH₄ and CO products.^{[ 53} , ^{54 ]} Moreover, from the enlarged view in the range of 2200−2000 cm⁻¹ (Figure S23, Supporting Information), two distinct peaks assigned to *CO intermediates can be clearly observed. Notably, the peaks at 1567 and 1517 cm⁻¹ are assigned to asymmetric stretching vibrations and symmetric stretching vibrations of the *CO─*CO intermediate, respectively, which is key intermediate for the C─C coupling.^{[ 55} , ^{56 ]} Importantly, we also observed the *CHO─CO intermediate (1576 cm⁻¹) in the in‐situ DRIFT spectra that originates from the subsequent hydrogenation process of the *CO─*CO intermediate.^{[ 57 ]} In addition, broad peaks were observed at 3000−3300 cm⁻¹ (Figure 5b), which are ascribed to C─H stretching vibration,^{[ 58 ]} suggesting the generation of hydrocarbons (i.e., CH₄ and C₂H₄) in this work. By contrast, no clear signals of C─C coupling intermediates were observed over pristine W₁₈O₄₉ (Figure S24, Supporting Information), verifying that atomically‐dispersed Cu(I) sites play a critical role in the C─C coupling process. Based on the in‐situ DRIFTS detection of *CO and *CHO─CO intermediates over 3.6‐Cu₁/W₁₈O₄₉ photocatalyst, we propose a possible pathway for the reduction of CO₂ to C₂H₄ as follows:

Figure 5.

Reaction mechanism investigation. a) In situ DRIFTS detections on the Cu₁/W₁₈O₄₉ photocatalyst in a humid CO₂ atmosphere under light irradiation b) enlarged region for CH_x detection. c) Free energy changes of C─C coupling at different active sites. The corresponding structural model consists of Cu (blue), W (dark green), O (red) and C (gray) atoms. d) The free energy diagram of CO₂ reduction on Cu₁/W₁₈O₄₉ and W₁₈O₄₉. The blue line shows the more favorable way, while the gray line shows the less favorable way over W₁₈O₄₉ catalyst. e) Transition state energy barriers for C─C coupling on W₁₈O₄₉ and Cu₁/W₁₈O₄₉.

$$\ast +\mathrm{C}{\mathrm{O}}_2+{\mathrm{e}}^{-}+{\mathrm{H}}^{+}\to \ast \mathrm{COOH}$$ (2)

$$\ast \mathrm{COOH}+{\mathrm{e}}^{-}+{\mathrm{H}}^{+}\to \ast \mathrm{CO}+{\mathrm{H}}_2\mathrm{O}$$ (3)

$$\ast \mathrm{CO}+{\mathrm{e}}^{-}+{\mathrm{H}}^{+}\to \ast \mathrm{CHO}$$ (4)

$$\ast \mathrm{CHO}+5{\mathrm{e}}^{-}+5{\mathrm{H}}^{+}\to \mathrm{C}{\mathrm{H}}_4+{\mathrm{H}}_2\mathrm{O}$$ (5)

$$\ast \mathrm{CO}+\ast \mathrm{CO}\to \ast \mathrm{CO}-\ast \mathrm{CO}$$ (6)

$$\ast \mathrm{CO}-\ast \mathrm{CO}+{\mathrm{e}}^{-}+{\mathrm{H}}^{+}\to \ast \mathrm{CHO}-\ast \mathrm{CO}$$ (7)

$$\ast \mathrm{CHO}-\ast \mathrm{CO}+7{\mathrm{e}}^{-}+7{\mathrm{H}}^{+}\to {\mathrm{C}}_2{\mathrm{H}}_4+\ast +2{\mathrm{H}}_2\mathrm{O}$$ (8)

where * denotes the catalytically active sites To further reveal the underlying mechanism of photocatalytic CO₂ reduction to C₂H₄, DFT calculations were performed. Given the results of XRD, TEM and XAFS measurements, four oxygen‐coordinated Cu single atoms on the (100) crystal plane of W₁₈O₄₉ were constructed and refined (Figure S25, Supporting Information). As a prerequisite for *CO intermediates, the formation barrier of *COOH was first calculated (Figure S26, Supporting Information). Compared to pristine W₁₈O₄₉, it can be observed that the introduction of Cu single atoms significantly reduces the formation barrier of *COOH intermediate. As demonstrated in in‐situ DRIFTS, the coupling of two *CO intermediates is the key step of the reaction toward C₂H₄ formation. To investigate this process, we studied the adsorption capacity of the key *CO intermediates over the 3.6‐Cu₁/W₁₈O₄₉ and pristine W₁₈O₄₉ photocatalysts. First, to determine the possible active sites of the reaction, the Gibbs free energy at different sites of *CO adsorption—a key intermediate of C─C coupling, was investigated (Figure 5c). The adsorption of *CO on Cu sites (Cu─*CO) and W sites (W─*CO) release energy of 1.15 and 0.14 eV, respectively; this indicates more favorable CO adsorption on Cu(I) sites, and is consistent with CO desorption DRIFTS results. Interestingly, W₁₈O₄₉ and Cu₁/W₁₈O₄₉ exhibit completely different thermodynamic processes when a second CO is adsorbed on the neighboring sites. The adsorption of another *CO on the adjacent W atom in pristine W₁₈O₄₉ (W─*CO, W─*CO) requires overcoming a high energy barrier (0.75 eV). In contrast, the energy barrier for the adsorption of another *CO on the adjacent W atom with Cu atom (Cu─*CO, W─*CO) is only −0.02 eV, which is nearly a spontaneous process. Consequently, it is thermodynamically unfavorable to produce two adjacent *CO intermediates at W sites in pristine W₁₈O₄₉, however, the introduction of Cu(I) sites modulate the adsorption of *CO intermediates on W sites and favor the generation of two adjacent *CO on the asymmetric Cu─W dual sites. This result also indicates that the W sites in the Cu─W dual sites can also have a stronger adsorption of *CO. The formation of double *CO adsorption on Cu₁/W₁₈O₄₉ surface can increase the probability of C─C coupling and improve the corresponding reaction kinetically. The generation of *CO─*CO from CO* (2*CO→*CO─CO) is usually considered a crucial step in the formation of C₂₊ compounds.^{[ 59} , ^{60 ]} Notably, the formation energy of *CO─*CO on Cu₁/W₁₈O₄₉ (0.50 eV) is lower compared to the pristine W₁₈O₄₉ (0.57 eV), indicating that the asymmetric Cu─W dual sites lower the thermodynamic energy barrier of *CO coupling and promote the formation of C─C bonds.

The full‐path Gibbs free energy changes (Figure 5d) were then calculated for the generation of C₂H₄ on W₁₈O₄₉ and Cu₁/W₁₈O₄₉, respectively. With corresponding structural models (Figures S27 and S28, Supporting Information), it is clear that *CO intermediate on W₁₈O₄₉ prefers to desorb as CO product (0.34 eV), rather than combine together to form *COCO (1.32 eV). By contrast, the overall free energy of C₂H₄ generation process at the asymmetric Cu─W dual sites is a downhill path, demonstrating the thermodynamic advantage of the Cu─W dual sites toward C₂H₄ formation. To further elucidate the role of Cu─W dual sites in the C─C coupling process, DFT calculations on transition state barriers were carried out. As shown in Figure 5e, Cu₁/W₁₈O₄₉ exhibits a lower transition state barrier (0.66 eV) for C─C coupling process compared to pristine W₁₈O₄₉ (0.83 eV), indicating the facilitating role of the Cu─W asymmetric dual sites in this crucial rate‐determining step. In brief, DFT calculations unravel that CO adsorption is more favorable at the asymmetric Cu─W dual sites, which lower the energy barrier of C─C coupling and promote the photoreduction of CO₂ to C₂H₄.

3. Conclusion

We have successfully designed and prepared stabilized Cu(I)‐containing asymmetric Cu─W dual sites, that is, Cu(I) single atoms modified W₁₈O₄₉ nanowires photocatalyst, for selective photocatalytic CO₂ reduction to C₂H₄. Impressively, under light irradiation, the optimal Cu₁/W₁₈O₄₉ photocatalyst exhibits C₂H₄ production with concurrent high activity and selectivity. Through interconversion of W(V) and W(VI) in W₁₈O₄₉, Cu(I) single atoms maintain their stability during photocatalytic process. Compared to pristine W₁₈O₄₉, the introduction of Cu(I) single atoms stabilize *CO intermediates and improve the adsorption of *CO on adjacent W atoms in Cu₁/W₁₈O₄₉. As a consequence, the as‐formed asymmetric Cu─W dual sites significantly reduce the energy barrier of C─C coupling, thus leading to the high selectivity of C₂H₄ product from photocatalytic CO₂ reduction over Cu₁/W₁₈O₄₉. This work provides new insight into design of rational active sites at the atomic level for CO₂ photoreduction to C₂₊ products, and may shed some light on future selective artificial photosynthesis.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Mao Y., Zhang M., Zhai G., Si S., Liu D., Song K., Liu Y., Wang Z., Zheng Z., Wang P., Dai Y., Cheng H., Huang B., Asymmetric Cu(I)─W Dual‐Atomic Sites Enable C─C Coupling for Selective Photocatalytic CO₂ Reduction to C₂H₄ . Adv. Sci. 2024, 11, 2401933. 10.1002/advs.202401933

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.