Construction of a Cu@hollow TS-1 nanoreactor based on a hierarchical full-spectrum solar light utilization strategy for photothermal synergistic artificial photosynthesis

Abstract

The artificial photosynthesis technology has been recognized as a promising solution for CO₂ utilization. Photothermal catalysis has been proposed as a novel strategy to promote the efficiency of artificial photosynthesis by coupling both photochemistry and thermochemistry. However, strategies for maximizing the use of solar spectra with different frequencies in photothermal catalysis are urgently needed. Here, a hierarchical full-spectrum solar light utilization strategy is proposed. Based on this strategy, a Cu@hollow titanium silicalite-1 zeolite (TS-1) nanoreactor with spatially separated photo/thermal catalytic sites is designed to realize high-efficiency photothermal catalytic artificial photosynthesis. The space–time yield of alcohol products over the optimal catalyst reached 64.4 μmol g⁻¹ h⁻¹, with the selectivity of CH₃CH₂OH of 69.5%. This rationally designed hierarchical utilization strategy for solar light can be summarized as follows: (1) high-energy ultraviolet light is utilized to drive the initial and difficult CO₂ activation step on the TS-1 shell; (2) visible light can induce the localized surface plasmon resonance effect on plasmonic Cu to generate hot electrons for H₂O dissociation and subsequent reaction steps; and (3) low-energy near-infrared light is converted into heat by the simulated greenhouse effect by cavities to accelerate the carrier dynamics. This work provides some scientific and experimental bases for research on novel, highly efficient photothermal catalysts for artificial photosynthesis.

INTRODUCTION

The urgent need to reduce CO₂ emissions while maintaining fuel supply for the development of human society remains one of the greatest challenges. Inspired by natural photosynthesis, CO₂ reduction by H₂O to solar fuels such as CH₄, methanol, and other hydrocarbons, which is commonly known as artificial photosynthesis (CO₂ + H₂O → CH₄ + CH₃OH + …), is a promising way to realize the capture and conversion of CO₂ simultaneously under the ambient conditions.^1,2 Nevertheless, the efficiency of artificial photosynthesis is affected due to certain limitations, such as limited light absorption and sluggish reaction kinetics.^3,4

Recently, photothermal catalysis, which utilizes both light and heat from the sun, has been developed.⁵ Photothermal catalysis extends the utilization of the full solar spectrum and increases the reaction rate compared to a pure photocatalytic system, and reduces the high-energy input compared to a conventional thermal catalysis system.^6,7 Therefore, this synergistic effect by coupling both photochemistry and thermochemistry leads to significantly enhanced activity in CO₂ reduction systems, especially in CO₂ hydrogenation reactions, such as the Sabatier reaction (CO₂ + 4H₂ ↔ CH₄ + 2H₂O),^8,9 the reverse water–gas shift reaction (CO₂ + H₂ ↔ CO + H₂O),^10–12 alcohol synthesis (CO₂ + 3H₂ ↔ CH₃OH + H₂O)^13–15, and the dry reforming reaction (CO₂ + CH₄ ↔ 2CO + 2H₂).^16–18 In contrast, very limited catalytic systems have been reported in terms of photothermal artificial photosynthesis. Wang et al.'s¹⁹ group discovered semiconductors with abundant oxygen vacancies as promising catalyst candidates for photothermal artificial photosynthesis to CH₄ and methanol. Our group fabricated a series of catalysts, including AuCu alloy-modified ultrathin porous g-C₃N₄ nanosheets and ionic liquid-assisted Cu₂O/g-C₃N₄, for photothermal artificial photosynthesis to ethanol with high selectivity.^20,21 These studies demonstrate that production of solar fuels from artificial photosynthesis can be more efficient under photothermal conditions, which is inspiring for future studies.

In order to design an effective photothermal catalyst, it is necessary to develop nanoscale materials and architectures that meet several requirements, including strong light absorption across the full solar spectrum, enhanced charge carrier separation efficiency, and high photothermal conversion efficiency.^22,23 To this end, most of the current photothermal catalysts are made up of semiconductors and plasmonic metals.^24,25 However, there is still a lack of general guidelines for catalyst design in this emerging research field, to improve the photothermal catalytic performance. On the one hand, semiconductor-based materials with a narrow band gap, elemental doping, and a heterojunction structure are widely used, such as V₂O_5−x, Rh_1−xO, UO_2+x, Zn_1+xO, TiO_2−x, WO_3−x, and BaTiO_3−xH_x.²³ These approaches enable the broadening of the light-absorption range to visible (Vis) and near-infrared (NIR) light but may lead to loss of ultraviolet (UV)-excited high-energy electrons, resulting in weakened redox ability.^26,27 On the other hand, plasmonic metals (e.g., noble metals like Ru, Rh, Au, and Pt, or nonnoble metals like Fe, Co, Ni, and Cu) may generate hot electrons due to the localized surface plasmon resonance (LSPR) effect under a specific excitation wavelength to facilitate the catalytic reaction.^28–30 Nevertheless, the lifetime (fs) of hot electrons is considerably short compared to the chemical reactions' time level (ps).²³ Therefore, LSPR hot electrons contribute very little to the initial activation of CO₂ molecules under photothermal conditions, which is a thermodynamically stable and kinetically sluggish reaction step.⁷

Based on the above considerations, it is proposed that controlling the hierarchical utilization of solar light with different wavelengths is critical for high-efficiency photothermal catalytic artificial photosynthesis. Utilization of high-energy UV light for the first and the most difficult step in the adsorption and activation of CO₂ has been proven feasible. For example, we preliminarily demonstrated that the monodisperse tetrahedral-coordinated Ti in titanium silicalite-1 zeolite (TS-1), which can be considered as a “single-site heterogeneous catalyst,” shows enhanced CO₂ activation capability under UV light irradiation.³¹ For Vis light, it can be utilized for driving other intermediate steps. Inspired by previous work on photothermal catalytic CO₂ hydrogenation and the water–gas shift reaction,^32,33 in which an active role of LSPR hot electrons in the dissociation of H₂ and H₂O has been reported, it is proposed that Vis light can be used to provide *H intermediates for subsequent hydrogenation steps. Besides, the low-energy NIR light can be converted into heat to improve photocatalytic performance. For example, Cai et al.'s³⁴ group designed a core–shell-structured catalyst composed of Ni cores and SiO₂ shells inspired by the greenhouse effect. It is found that this core–shell architecture can reduce the heat conduction and infrared (IR) light radiation to the environment, leading to the generation of high temperature within the catalyst.^35,36

On the basis of the above discussion, here, we propose an innovative hierarchical full-spectrum solar light utilization strategy to achieve efficient photothermal synergistic artificial photosynthesis: (1) high-energy UV light should be intensively used to activate CO₂; (2) Vis light can induce the LSPR effect on plasmonic metal to generate hot electrons for subsequent reaction steps with lower energy requirements; and (3) low-energy NIR light can promote the reaction by providing extra energy via photothermal conversion. Using this strategy, we designed a nanoreactor with plasmonic Cu nanoparticles encapsulated in hollow TS-1 (Cu@H-TS-1). Under photothermal conditions, the optimal catalyst shows enhanced performance in artificial photosynthesis, with notably enhanced yields of alcohol products and selectivity of ethanol. A series of control experiments were performed to determine the role of light of different frequencies in the nanoreactors, including photothermal conversion capability, photoexcitation, and the LSPR effect. Structural and spectroscopic characterizations as well as in situ characterizations and theoretical calculations were performed to investigate the reaction mechanisms on Cu and Ti sites in this rationally designed nanoreactor. Therefore, the plausibility and feasibility of the above-proposed full-spectrum solar utilization strategy were explored in detail. These results pave the way for the development of catalysts with full-spectrum utilization capability for artificial photosynthesis and other photothermal reactions.

EXPERIMENTAL SECTION

The experimental details including chemicals, synthesis of different catalysts, catalyst characterization, catalytic performance test, theoretical calculations, and simulations are provided in the Supporting Information.

RESULTS AND DISCUSSION Materials synthesis and structural properties

xCu@H-TS-1 nanoreactors were synthesized using a top-down method (x represents the theoretical percentage of the Cu content, and the actual contents from inductively coupled plasma optical emission spectrometry results are listed in Table S1; H represents the hollow structure),^37,38 which involves several steps as illustrated in Figure 1A. Taking 0.5Cu@H-TS-1 as an example, the hollow cavities can be clearly observed (Figure 1B). Cu nanoparticles with an average size of 8.58 nm and a lattice fringe space of 0.208 nm, attributed to the (111) plane of Cu (JCPDS No. 65-9026), are embedded within the TS-1 shell (Figure 1C,D). Elemental mapping images (Figure 1F) reveal that the Ti and Cu elements are homogeneously distributed throughout the whole sample. The structural properties of the TS-1 shell and loadings of Cu nanoparticles were regulated by controlling the synthetic parameters, as shown in Figures S1–S4. These results indicate the successful fabrication of Cu@H-TS-1 nanoreactors using the top-down method.

Figure 1. Preparation and morphological properties of Cu@H-TS-1 nanoreactors. (A) Schematic diagram for the synthesis of Cu@H-TS-1 nanoreactors. (B, C) Transmission electron microscopy, (D) high-resolution transmission electron microscopy, (E) SEM in the secondary electron imaging mode, and (F) energy-dispersive spectroscopy mapping images of 0.5Cu@H-TS-1 nanoreactors; and (G) XRD patterns of as-prepared samples. The inset shows the enlarged XRD patterns in the 42°–44° range.

As shown in Figure 1G, all the xCu@H-TS-1 samples show the characteristic diffraction peaks of the mobil-type five zeolite  structure,³⁹ indicating the formation of a TS-1 shell with high crystallinity. The relative crystallinities decrease with increased Cu loading (Table S1), probably due to the inhibition effect of excess ethane diamine on the crystallization process. No diffraction peaks of the Cu species can be observed in the X-ray powder diffraction (XRD) patterns of the samples with low Cu loading, due to either low loading or high dispersion of Cu on the inner walls, while weak peaks of Cu(111) at 43.3° are observed for 1Cu@H-TS-1 and 1.5Cu@H-TS-1. The N₂ adsorption–desorption isotherms and pore size distribution curves of xCu@H-TS-1 samples are shown in Figure S5, and the textural properties are summarized in Table S2. The benchmark TS-1 shows a type I isotherm, indicating that it is absolutely microporous. The adsorption–desorption isotherms of xCu@H-TS-1 samples show a hysteresis loop, which is a hybrid of types I and IV, indicating a hierarchically porous structure with both micropores and mesopores. Compared with H-TS-1, very slight decrease of BET surface areas can be observed for 0.5Cu@H-TS-1, indicating that the loading of Cu in the cavities led to minimal loss of porosity. These results demonstrate the successful fabrication of cavities, which could promote mass transport in the reaction.

Spectroscopic characterizations

The chemical states of Cu and Ti in xCu@H-TS-1 nanoreactors were investigated. As shown in the Fourier-transform infrared (FT-IR) spectra (Figure S6), all the samples show peaks at 964 cm⁻¹, which can be assigned to framework Ti in the TS-1 shell. The gradually decreased intensities also indicate the inhibition effect of excess ethane diamine on the crystallization process (more details can be found in the supplementary discussion in the Supporting Information). The X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) spectra of the samples were obtained. The wavelet transform-EXAFS spectra are shown in Figures S7 and S8, and the detailed structural parameters based on EXAFS fittings are summarized in Tables S3 and S4. The Ti K-edge XANES spectra of all the samples are very similar to that of benchmark TS-1 (Figure 2A), indicating that Ti species in all the samples are identical, in the form of four-coordinated Ti, as shown in the EXAFS spectra (Figure 2B). The Cu K-edge XANES spectra in Figure 2D show that the near-edge absorption energy and the white-line intensity of Cu@H-TS-1 are higher than those of the Cu foil and lower than those of Cu₂O. The Cu K-edge EXAFS spectra of Cu@H-TS-1 (Figure 2E) show prominent peaks that are assigned to metallic Cu−Cu bonding. The minor peaks that are assigned to Cu−O can also be observed, indicating the existence of some atomically dispersed Cu stabilized by O atoms in TS-1 zeolite. With increased Cu loading, the coordination numbers (C.N.) for Cu−O decrease and the C.N. for Cu−Cu increase (Table S4), consistent with the increased sizes of Cu in these samples (Figure S4). Importantly, the peak at ∼2.42 Å, which is attributed to Cu−Ti coordination, cannot be observed.⁴⁰ X-ray photoelectron spectroscopy (XPS) spectra in Figure 2C,F further indicate that Cu species mainly exist as metallic Cu (peaks at 953.6 and 933.8 eV), and Ti species mainly exist as framework Ti (peaks at 460.0 eV), while some extra-framework Ti peaks (at 458.8 eV) appear in samples with excess Cu loadings caused by excess ethane diamine. These results demonstrate that the highly active framework Ti remains largely preserved after the construction of nanoreactors, and Cu exists predominantly as large clusters. No obvious interactions such as the Schottky junction occurred between Ti and Cu, indicating that the two sites are spatially separated in the nanoreactors and might play distinct roles in the reactions.

Figure 2. Electronic structure of Cu@H-TS-1 samples. (A) Ti K-edge XANES spectra. (B) Fourier transforms of EXAFS spectra at the Ti K-edge. (C) XPS spectra of Ti 2p peaks. (D) Cu K-edge XANES spectra. (E) Fourier transforms of EXAFS spectra at Cu K-edge. (F) XPS spectra of Cu 2p peaks.

Photothermal catalytic performance in artificial photosynthesis

The photothermal-induced artificial photosynthesis performance of xCu@H-TS-1 was investigated. First, a heating temperature of 393 K was chosen for our research, based on the performances of the 0.5Cu@H-TS-1 catalyst at different temperatures (Figure S10). A reaction temperature that is too high or too low may lead to a decrease in photothermal catalytic activity.⁷ Subsequently, the catalytic performances of the as-prepared materials were compared under photocatalytic, thermal catalytic, and photothermal catalytic conditions. As shown in Figure 3A, all the samples show significantly higher space–time yield (STY) of alcohol products (calculated based on the total number of reduced carbons in alcohol products, STY_{total C}) under photothermal conditions (heating temperature of 393 K, full-spectrum light irradiation), indicating the synergetic effect between photochemistry and thermochemistry. Notably, all the samples are almost inactive under heated-only conditions, suggesting the important role of photocatalysis, such as driving the initial and the most difficult CO₂ activation step. The effect of Cu loadings on catalytic activities was investigated. As shown in Figure 3B, within a certain range, Cu loading has little effect on STY but can significantly affect the product distribution. When Cu loading was increased from 0.1 to 1.5 mol%, the selectivity of CH₃CH₂OH significantly increased from 59.6% to 69.5%. These results suggest that Cu sites in the Cu@H-TS-1 nanoreactors would more likely promote the hydrogenation and C−C coupling steps, other than CO₂ activation. Meanwhile, this difference in activities is not due to other factors, like the catalyst particle size, because the reactants and products' molecules are all smaller than the pore channels of the TS-1 shell and can diffuse freely into and out of the nanoreactor. When Cu loading exceeded 1%, although the optical properties were promoted by the formation of some heterojunction structures between TS-1 and extra-framework Ti species (discussed in detail in the following sections), the activities still decreased, which can be due to the poor crystallinities and loss of highly active four-coordinated framework Ti (relative crystallinities are presented in Table S1 and FT-IR results are shown in Figure S6).

Figure 3. Performance of Cu@H-TS-1 in photothermal artificial photosynthesis. (A) STY of alcohol products over TS-1 and Cu@H-TS-1 catalysts under thermal catalytic, photocatalytic, and photothermal catalytic conditions, a heating temperature of 393 K, and 300 W UV enhanced Xe lamp irradiation. (B) Product distributions over TS-1 and Cu@H-TS-1 catalysts under photothermal conditions. (C) Product distributions and STY over catalysts with different components and architectures under photothermal conditions. (D) STY over Cu@H-TS-1 under photothermal conditions in five cycling tests. (E) Comparison of Cu@H-TS-1 with previously reported photothermal catalysts.

The photothermal catalytic performances of samples with different architectures were investigated. As shown in Figure 3C, Cu@H-TS-1 shows both superior activity and CH₃CH₂OH selectivity to benchmark TS-1 and Cu@TS-1, indicating the potential advantages of cavities' structures. CO₂-TPD was implemented, as shown in Figure S11A, which demonstrates that H-TS-1 has higher CO₂ adsorption capacity than TS-1 without cavities. In addition, thermogravimetric analysis was performed to determine CO₂ adsorption abilities and the calculated amounts of adsorbed CO₂ per milligram of catalyst, as shown in Figure S11B, which shows similar trends. Notably, after loading of Cu, CO₂ adsorption capacities decreased, which is due to the weak CO₂ adsorption ability on metallic Cu, which requires energy expenditure, and loading of Cu might cover some adsorption sites, such as surface hydroxyl groups. However, it can be seen that 0.5Cu@H-TS-1 still has higher CO₂ adsorption capacity than 0.5Cu@TS-1. These results indicate that the nanoreactors preferentially promote the adsorption of CO₂ molecules as well as the intermediates, therefore promoting both STY and C−C coupling probability, which is consistent with findings in the literature.^41,42 The size of cavities was also regulated by changing the etching time (Figure S3), and their influences on activities were investigated, as shown in Figure S12. The results demonstrate that larger cavities are beneficial to activity, while excessive etching time would lead to a decrease in activities.

The reusability of Cu@H-TS-1 catalysts was tested, implying that there was no significant decrease in alcohol production after five consecutive cycles (Figure 3D). Scanning electron microscopy (SEM) images and XRD patterns (Figures S13 and S14) show that the Cu@H-TS-1 nanoreactor was still unchanged after the reaction, confirming the superior stability, which is attributed to the intrinsic stability of TS-1 zeolites as well as the confinement effect of nanocavities. The performances of previously reported photocatalysts and photothermal catalysts in artificial photosynthesis to C1 and C2 alcohol products are summarized in Table S5. A number of catalysts have been extensively studied in pure photocatalytic CO₂ reduction with H₂O. In contrast, very limited catalytic systems have been reported on photothermal artificial photosynthesis. In addition, due to the unsatisfactory efficiency and low C−C coupling capability, the current yields of alcohol products are still not suitable for practical applications. Therefore, state-of-the-art studies on CH₃CH₂OH production are listed in Figure 3E. The comparison reveals that the Cu@H-TS-1 nanoreactor is superior to the majority of previously reported photocatalytic systems and representative photothermal photocatalysts, and can be comparable to those in some studies with the addition of sacrificial agents (more details such as yields and reaction conditions are listed in Table S5). Thus, our findings may provide guidelines for future development of solar fuel production under mild conditions.

Full-spectrum solar light utilization strategy Photothermal conversion capabilities under illumination at different wavelengths

The optical absorption and solar energy conversion patterns of the samples were explored. UV–vis diffuse reflectance spectroscopy spectra (Figure 4A) of both benchmark TS-1 and Cu@H-TS-1 samples show strong absorption peaks at around 215 nm, attributed to the charge-transfer excited state generated by Ti sites in TS-1.⁴³ The absorption bond, which begins from 400 nm and is further enhanced at 600 nm, is attributed to the LSPR peak of plasmonic Cu nanoparticles, demonstrating the strong LSPR response throughout the entire Vis region.⁴⁴ Besides, the absorption range is extended to NIR light, indicating the enhanced solar light utilization ability of Cu@H-TS-1 nanoreactors.

Figure 4. Photothermal conversion capability of Cu@H-TS-1 nanoreactors. (A) UV–vis absorption spectra of as-synthesized photocatalysts. Photothermal IR images of (B) TS-1 and (C) Cu@H-TS-1 during the reaction under illumination of various wavelengths. Photocurrent responses of Cu@H-TS-1 under (D) illumination of light with different wavelengths and (E) different heating temperature conditions. (F) PL spectra of Cu@H-TS-1 under different heating temperature conditions. (G) STY and AQY of Cu@H-TS-1 under illumination of light with different wavelengths.

Heat can be generated through the photothermal conversion process under solar illumination, which can contribute to improved photothermal catalytic performance. Therefore, the photothermal conversion ability of Cu@H-TS-1 nanoreactors was investigated by recording the temperature evolution. The initial temperature of the reaction solution system was 297.0 K. As shown in Figure 4B,C, after only 40 min of illumination, the temperature of the reaction system of Cu@H-TS-1 could reach 322.1 K under full-spectrum light irradiation, which is higher than that of benchmark TS-1 (316.5 K), suggesting good photothermal conversion ability of Cu@H-TS-1. Furthermore, the temperature evolution of the reaction system was measured under different light irradiations. The temperature of Cu@H-TS-1 increased to 301.2, 304.2, and 316.8 K under UV, Vis, and NIR light, respectively (Figure 4C). Notably, the temperature of TS-1 could only increase to 308.8 K under NIR light (Figure 4B). The temperature changes of the as-synthesized sample powders were also recorded under different light irradiations directly exposed to air (Figure S15–S17), which show a similar trend. These results suggest that heat is mainly generated by NIR photons, especially by the hollow structure of nanoreactors, which is similar to greenhouse effects.³⁴ It has been recognized that the low-energy NIR light hardly participates in the reactions directly, and very limited studies on NIR light-induced catalytic CO₂ reduction have been reported.⁴⁵ Therefore, we believe that it is reasonable to convert NIR light into heat to improve photocatalytic reactions. In addition, limited heat is generated under Vis light, suggesting that the hot electrons generated via LSPR by plasmonic Cu might be utilized in the reactions rather than be lost in the form of heat release by nonradiative electronic relaxation processes.²³ Moreover, the wavelength-dependent photocurrent test (Figure 4D) shows that significant photocurrent responses are observed for Cu@H-TS-1 under UV light irradiation, but no current is generated under vis and NIR light irradiations. These results indicate that TS-1 shell effectively captured and transformed a significant portion of high-frequency UV photons into high-energy photogenerated electrons, rather than dissipating them as heat through thermalization.

The effect of the increased temperature of the reaction solution on carrier behavior was further investigated. The photocurrent density curves and fitted electrochemical impedance spectra plots of Cu@H-TS-1 under different temperatures are shown in Figures 4E and S18, respectively. As the temperature increases, the photocurrent density increases and the impedance decreases, implying faster carrier dynamics. Meanwhile, the temperature-dependent photoluminescence (PL) spectra of Cu@H-TS-1 (Figure 4F) show lower peak intensity under higher temperature, which again indicates enhanced carrier dynamics.⁴⁶

Finally, control experiments were further conducted to validate the utilization patterns of light with different frequencies. As shown in Figure 4G, Cu@H-TS-1 shows distinct performance as well as the corresponding apparent quantum yield (AQY) under different light irradiation. STY under UV irradiation is much higher than that under Vis-NIR, and no product is obtained under NIR light. However, both STY and selectivity improved under UV–Vis light and can be increased even further with additional NIR light. Based on the above results, it can be speculated that UV light is essential for driving the reaction, and Vis light is mainly used for regulating the products' distribution, while NIR light contributes to the activity but is not essential.

Spatially separated photo/thermal catalytic sites

As was concluded from X-ray absorption spectroscopy and XPS results (Figure 2), no interactions such as the Schottky junction occurred between Cu nanoparticles and Ti sites. The photoelectric chemical properties of xCu@H-TS-1 were investigated. In the PL spectra (Figure S19), the trends of the PL peak intensities were irregular for samples with Cu loading below 1%, probably due to the complex mechanism of fluorescence emission caused by the complex compositions of catalysts. However, a discernible decrease was detected for 1Cu@H-TS-1 and 1.5Cu@H-TS-1 due to the formation of some heterojunction structures between TS-1 and extra-framework Ti species, leading to enhanced photogenerated carrier separation (more details are provided in Figure S19). The accelerated decay time-resolved PL spectra (Figure S20) and the calculated photoelectron lifetimes of each catalyst (Table S6) also suggest similar photoelectron lifetimes of xCu@H-TS-1 with Cu loading below 0.5 mol%. To gain deeper insights into the LSPR effect of Cu as well as the transfer behaviors of LSPR hot electrons, light-induced electric field distributions around Cu nanoparticles and TS-1 were simulated using the finite difference time domain (FDTD) method. As shown in Figure 5A, the intensity of the electric field around a supported Cu nanoparticle with a diameter of 8 nm was enhanced under light irradiation from 400 to 800 nm, consistent with the strong LSPR adsorption of Cu in the entire Vis spectra (Figure 4A). However, the electric fields were strongly localized around Cu nanoparticles other than at the interface of Cu and TS-1, unlike previously reported studies, in which the highest electromagnetic field intensity was found at the interface of plasmonic metal and semiconductor support.^47,48 Moreover, the electric field vector results in Figure S21 indicate that although the direction of the electric field started to change within the LSPR excitation wavelength, the direction had never been toward the interfaces. These results clearly indicate the spatially separated existence status of Cu nanoparticles and Ti sites.

Figure 5. Carrier transfer behavior and reaction mechanisms on spatially separated photo/thermal catalytic sites. (A) Spatial distribution of the LSPR-induced enhanced electric field intensity from FDTD simulations for the Cu/TS-1 model at excitation wavelengths of 300–800 nm. E denotes the vector of the electric field. (B–D) In situ DRIFTS spectra of the photo-induced artificial photosynthesis on the Cu@H-TS-1 catalyst under different light irradiations. Gibbs free-energy diagrams of CO2 reduction on (E) Cu sites and (F) Ti sites. (G) Illustration of the reaction mechanism on spatially separated photo/thermal catalytic sites.

It has been recognized that LSPR hot electrons hardly play a direct role in the sluggish CO₂ activation steps, due to the mismatch of both the time and the energy scale.^7,23 Therefore, some investigators loaded plasmonic metals on semiconductor supports to construct Schottky junctions, in which the LSPR hot electrons can effectively transfer to other sites or conversely, leading to a prolonged lifetime for participation in reactions, such as the state-of-the-art studies carried out by Xiong's groups.^49,50 However, in our work, the electrically insulating nature of TS-1 zeolite may prevent the transfer of electrons.⁵¹ We therefore suspected that this feature instead has distinct advantages in the CO₂ reduction reaction, in which photoelectrons and hot-electrons can be utilized in different reaction steps.

To verify this conjecture, in situ diffuse reflectance infrared Fourier-transform spectroscopy (in situ DRIFTS) was first performed to probe the reaction pathway, as shown in Figures 5B–D and S22–S26, and the positions of the main peaks and corresponding groups are shown in the diagram and summarized in Table S7. In the absence of light (Figure S22), only the adsorption peaks for H₂O and CO₂ can be observed. The peak intensities remained unchanged on altering the heating temperature, and no other new peaks were observed, indicating that these adsorbed molecules did not undergo a chemical reaction without light illumination. In order to explore the differences in the reaction pathway under different light irradiations, in situ DRIFTS were conducted by using different light sources. In Figure 5B,C, after the input of CO₂ and H₂O, the corresponding adsorption peaks appear, including *CO₂, *COOH, *OCO, *CO, *H₂O, and so forth. The increase of the intensities of these peaks suggest the gradual generation and accumulation of these intermediates. Notably, no peaks are observed at 2900 cm⁻¹, which could readily be assigned to the C−H vibrations in formate (*HCOO) created through a CO₂ hydrogenation pathway (*CO₂ + *H).^33,52 Therefore, these results suggest that the alcohols were generated through a carboxyl pathway other than the direct CO₂ hydrogenation pathway. Besides, the peaks at 1362 and 1489 cm⁻¹ are attributed to *OCCO which is a critical C−C coupling intermediate. The peaks at 1339 cm⁻¹ can be attributed to C−H in CH₃CH₂OH, indicating the generation of CH₃CH₂OH.⁵³

After switching to Vis light, the locations and species of the peaks did not change. However, as shown in the enlarged image in Figure 5D, the *H₂O peaks at 1636 cm⁻¹ first increase significantly, indicating the rapid adsorption of H₂O. Then, a decrease in the peak intensities can be observed, indicating the consumption of *H₂O during the subsequent reduction of CO₂.⁵⁴ These results indicate that the adsorption and activation of water are promoted under Vis light irradiation. Combined with optical adsorption properties and the LSPR effect of Cu under Vis light, it can be concluded that H₂O molecules are largely split by LSPR hot electrons on Cu to supply *H sources for hydrogenation steps. No other new peaks were observed, indicating that the hot electrons created by Cu nanoparticles did not change the main reaction path or produce any new intermediates but simply promoted the original carboxyl reaction mechanism by overcoming the major barriers of water activation and dissociation. This result is similar to the finding of Zhao et al.'s³³ group, who reported that plasmonic Cu particles show enhanced H₂O dissociation capability in the photo-driven water–gas shift reaction.

Finally, density-functional theory calculations were performed to gain a full understanding of the reaction mechanism. The Cu(111) surface model and four-coordinated Ti in TS-1 were established as catalytic sites, and the potential mechanism for CO₂ reduction was studied on each site. As shown in Figure 5E, it is obvious that the initial CO₂ adsorption and activation step (Step II) and the first hydrogenation step (Step III) hardly occurred on Cu sites. However, as shown in Figure 5F, CO₂ is preferentially adsorbed and activated on Ti sites (Step II, −0.54 eV). These results confirm that plasmonic Cu show limited CO₂ adsorption and activation ability, consistent with the results in the literature,⁵⁵ while the UV light-excited photoelectrons generated by TS-1 show enhanced CO₂ activation capability, consistent with our previous work.³¹ Besides, it is found that Cu may also act as subsequent reaction sites, via the coupling of two *CHO (Figure 5E, Step VI), consistent with the results that the selectivity of CH₃CH₂OH is enhanced with loading of Cu (Figure 3B).

Based on these results, the reaction mechanisms on spatially separated photo/thermal catalytic sites are proposed, as illustrated in Figure 5G. CO₂ is primarily adsorbed and activated by UV light on Ti sites. H₂O is dissociated by LSPR hot electrons on Cu sites, to provide abundant *H sources for subsequent hydrogenation steps. In addition, Cu can provide additional sites for C−C coupling, thus improving CH₃CH₂OH selectivity.

CONCLUSION

In summary, Cu@H-TS-1 nanoreactors were successfully synthesized using a top-down method, and they showed enhanced performance for artificial photosynthesis under photothermal conditions, compared to those under only heating or light irradiation conditions. STY of alcohol products over the optimal catalyst reached 64.4 μmol g⁻¹ h⁻¹, and the selectivity of CH₃CH₂OH reached 69.5%. The enhanced performance is owing to the rationally designed hierarchical utilization strategy for solar light with different wavelengths: (1) high-energy UV light is utilized to drive the initial and difficult CO₂ activation step on the TS-1 shell; (2) Vis light can induce the LSPR effect on plasmonic Cu to generate hot electrons for H₂O dissociation and subsequent reaction steps; and (3) low-energy NIR light is converted into heat by the simulated greenhouse effect in cavities to accelerate the carrier dynamics. Remarkably, the interactions between Cu and the TS-1 shell are weakened in this ingeniously designed Cu@H-TS-1 nanoreactor, leading to the formation of spatially separated photo/thermal catalytic sites. Therefore, photoelectrons and hot electrons can be retained on Ti sites and Cu sites, respectively, to drive different reaction steps. These results validate the plausibility and feasibility of the above-proposed full-spectrum solar utilization strategy, and might provide some scientific and experimental bases for research on novel, highly efficient photothermal catalytic artificial photosynthesis catalysts.

ACKNOWLEDGMENTS

This work was supported by the National Natural Science Foundation of China (Grant Nos. 21908052 and 22108200); the Key Program of the Natural Science Foundation of Hebei Province (Grant No. B2020209017); the Project of Science and Technology Innovation Team, Tangshan (Grant No. 20130203D); the Natural Science Foundation of Zhejiang Province (Grant No. LQ22B060013); and the Science and Technology Project of Hebei Education Department (Grant No. QN2021113).

CONFLICT OF INTEREST STATEMENT

The authors declare that there are no conflicts of interests.