Regulating the *OCCHO intermediate pathway towards highly selective photocatalytic CO₂ reduction to CH₃CHO over locally crystallized carbon nitride

Abstract

Photocatalytic conversion of CO₂ to CH₃CHO is of increasing interest but confronts the significant challenges of forming C–C bonds and keeping the C=O bond intact throughout the process. Here, we report the selective photocatalytic hydrogenation of CO₂ to CH₃CHO using a modified polymeric carbon nitride (PCN) under mild conditions. The locally crystallized PCN offers a photocatalytic activity of 1814.7 μmol h⁻¹ g⁻¹ with a high selectivity of 98.3% for CH₃CHO production and a quantum efficiency of 22.4% at 385 nm, outperforming all the state-of-art CO₂ photocatalysts. The promoted formation of the *OCCHO intermediate on the locally crystallized PCN is disclosed as the key factor leading to the highly selective CH₃CHO generation. The locally crystallized PCN favors spontaneous C–C coupling towards *OCCHO formation rather than *CHO protonation, thus preventing HCHO formation. This work provides a new strategy for designing carbon nitrides for highly selective and sustainable conversion of CO₂ to CH₃CHO.

---

Broader context

Acetaldehyde is an indispensable chemical in the synthesis of pharmaceuticals, agrochemicals and fragrances. Photocatalytic conversion of CO₂ to acetaldehyde has emerged as a promising method for sustainable acetaldehyde production. However, the current carbon-based products of photocatalytic CO₂ conversion are mostly dominated by CO, HCOOH and CH₄. Specifically, a ten-electron transfer process and multiple hydrogenation steps are involved in the conversion of CO₂ to acetaldehyde, resulting in unexpected products (e.g., CO, HCHO, HCOOH) and limited conversion efficiency. This calls for a systematic mechanism investigation on the whole reaction at the atomic level and rational design of highly selective photocatalysts. Herein, we design a high-performance polymeric carbon nitride (PCN) catalyst with a locally crystallized structure through an amino-2-propanol (AP)-assisted hydrothermal pretreatment. The energy barrier from *CHO to the *OCCHO intermediate is significantly reduced on the locally crystallized PCN, whereas the hydrogenation of *CHO to *CH₂O is efficiently suppressed. The formation of the *OCCHO intermediate is beneficial to the subsequent proton-coupled electron transfer process to form CH₃CHO rather than HCHO. Eventually, a high CH₃CHO generation rate and high selectivity are achieved on the elaborately crafted locally crystallized PCN.

Introduction

Acetaldehyde, as an indispensable intermediate in the manufacture of high-value-added chemicals, is widely applied in the fields of pharmaceuticals, agrochemicals and fragrances.¹ Ethylene oxidation catalyzed by palladium(II) and copper chloride in strong acidic solutions via the Wacker process is a common method for acetaldehyde production; however, it involves intensive energy consumption and acid contamination, and it is restricted by the petroleum-based ethylene resource.^2,3 Partial oxidation of ethanol at high temperature via an endothermic reaction or partial dehydrogenation of ethanol is an alternative route.^4,5 However, this method possesses the drawback of massive consumption of agriculture crops. Thus, there is substantial interest in developing sustainable and atomic economic approaches for acetaldehyde production.

Using solar energy radiation instead of thermal activation to photo-reduce CO₂ to C₂ hydrocarbon-based products under mild operating conditions appears to be a promising strategy for the sustainable production of acetaldehyde, accompanied by the alleviation of ever-increasing CO₂ emission.^6,7 The current carbon-based products of photocatalytic CO₂ conversion are dominated by CO^8–10 and HCOOH¹¹via a two-electron reduction reaction. Other typical C₁ products include HCHO,^12,13 CH₄,^14–16 and CH₃OH,^17–19 whereas liquid multi-carbon (C₂₊) compounds,^6,20–22 specifically CH₃CHO as a product, have been rarely reported.^23,24 Recently, several works have reported the photocatalytic conversion of CO₂ to CH₃CHO by using transition-metal semiconductors with a d⁰ or d¹⁰ electronic configuration, including TiO₂,^25–28 lnTaO₄²⁹ or SnS₂.³⁰ The ternary metal materials of Zn_xCd_1−xS³¹ and ZnFe₂O₄³² and copper-containing species^24,33 have also been demonstrated to yield CH₃CHO. However, they still suffer from low productivity (10–500 μmol h⁻¹ g⁻¹) and uncontrollable selectivity.

The remaining great challenge for photocatalytic reduction of CO₂ to CH₃CHO is the trade-off between catalytic activity and selectivity.³⁴ This is because the complicated ten-electron transfer process involves multiple hydrogenation steps from CO₂ to CH₃CHO, during which various competitive unexpected products (e.g., CO, HCHO, HCOOH) can be generated through different pathways.^7,35,36 Moreover, currently, the mechanism for this complicated reaction is yet to be disclosed; to date, only the CO₂ adsorption and the initiate CO₂ activation process have been investigated.³⁰ This calls for a systematic mechanism investigation on the whole reaction at an atomic level and rational design of highly selective photocatalysts.

Here, we elaborately craft a polymeric carbon nitride (PCN) through an amino-2-propanol (AP)-assisted hydrothermal pretreatment. The modified PCN presents a unique bubble-like structure with amorphous domains and well-crystallized lattice matrices that appear alternately in the framework. It is found that the energy barrier from *CHO to the *OCCHO intermediate is significantly reduced on the locally crystallized PCN, whereas the hydrogenation of *CHO to *CH₂O is suppressed. Thus, the formation of the *OCCHO intermediate is beneficial to the subsequent proton-coupled electron transfer process to form CH₃CHO rather than HCHO. Moreover, the optimized PCN can prevent the excessive protonation of CH₃CHO to form CH₃CH₂OH, eventually affording highly selective CH₃CHO generation. Furthermore, isotopic tests prove that the components of carbon and oxygen in CH₃CHO originate from CO₂, and hydrogen in CH₃CHO is extracted from H₂O. For the first time, the reaction mechanism for the whole reaction of CO₂ reduction to CH₃CHO is systematically investigated and our strategy for fabricating dyadic PCN could be easily applied for the rational design of conjugated polymer semiconductors for a variety of energy conversion reactions.

Results and discussion

Usually, dicyandiamide (DCDA) can form a supramolecular precursor for polymeric carbon nitride (PCN) under the hydrothermal pretreatment strategy (Fig. 1a).³⁷ Previously, the addition of amino-2-propanol (AP) was found to improve the polymerization degree of the monomers and polymers for the transformation of oleochemicals.³⁸ Here, we attempted to improve the crystallinity and photocatalytic performance of PCN by introducing AP into the precursor during the hydrothermal pretreatment. The precursor with different amounts of AP addition was calcined at 550 °C in air to fabricate carbon the nitride products, denoted as HCN-A_x. The photocatalytic tests (vide infra) show that HCN-A₃ had the best performance; therefore, we used HCN-A to denote HCN-A₃ for the following description, unless otherwise specified. For comparison, CN and HCN were synthesized from the direct calcination of DCDA with and without hydrothermal pretreatment, respectively. The SEM image (Fig. 1b) reflects that the bubble-shaped structure of HCN-A is obviously different from that of the bulk particles of CN and shows the two-dimensional layered framework of HCN (ESI,† Fig. S1). The high-magnification SEM image (Fig. 1c) and transmission electron microscopy (TEM) image (Fig. 1d) manifest the formation of a hollow framework in the bubble-shaped HCN-A. The entity size was enlarged as the AP content increased (Fig. S2 and S3, ESI†). The selected area electron diffraction (SAED) image (inset in Fig. 1d) of HCN-A presents a bright diffraction ring indexed to the (002) plane, verifying a certain degree of highly crystalline configuration (Fig. S4, ESI†). In contrast, the HCN shows an amorphous structure (Fig. S5, ESI†), confirming that the crystallization in HCN-A is caused by the addition of AP. A series of characterizations were conducted (as shown in Fig. S6–S12, ESI†) to reveal the role of AP in the morphology and structure of the HCN-A. Briefly, the AP-modified precursor presents a higher thermal decomposition temperature, implying stronger thermal stability. The NMR peaks of HCN-A shift towards a high field when AP is added, suggesting that increased electron cloud density surrounds the carbon atom near the H-bonds and thus improves the degree of crystallinity.³⁹

Fig. 1 (a) Schematic of the synthesis of the HCN-A photocatalysts. (b and c) SEM images of HCN-A. (d and e) TEM images; inset is the corresponding selected area electron diffraction (SAED) pattern. (f) HR-TEM image of FCN-V. (g) High-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM) images and C, N element mappings; (h) high-resolution X-ray total scattering spectra with the PDFgui refinement, with a comparison of the differential distance within the structures of HCN and HCN-A via the differential correlation function of D(r). (i) C K-edge and (j) N K-edge XANES spectra of HCN and HCN-A.

The hollow bubble-shaped HCN-A (Fig. 1e) presents a well-defined wall with an average thickness of 32 nm. The HCN-A exhibits a clear lattice fringe of 0.32 nm that corresponds to the (002) plane of carbon nitride (Fig. 1f and Fig. S13, ESI†), further demonstrating the well-crystallized lattice matrices.⁴⁰ Interestingly, amorphous domains are observed between the well-crystallized lattice matrices (marked by white dotted lines). These results reveal the formation of an interface dyadic homo-junction that features alternately appearing amorphous domains and well-crystallized lattice matrices, which may rearrange the reaction pathway and the product selectivity.^41,42 High-angle annular darkfield scanning transmission electron microscopy (HAADF-STEM) further discloses the hollow bubble structures (Fig. 1g), and the corresponding energy dispersive X-ray spectroscopy (EDS) mapping images (Fig. 1g) reveal the uniform distributions of C and N elements in HCN-A. The specific surface area of HCN-A is determined to be 81.6 m² g⁻¹ and the pore volume is 0.28 m³ g⁻¹, which is slightly smaller than that of HCN (Fig. S14 and Table S1, ESI†).

As shown in Fig. S15 and S16 (ESI†), the (002) peak of HCN-A shifts towards a lower angle compared with that of HCN, suggesting enhanced distance of the interplanar space.⁴³ The FTIR spectra prove that the HCN-A maintains the backbone unit of the heptazine structure of CN (Fig. S17, ESI†). The chemical environment variation was further investigated by high-resolution X-ray total scattering spectra with the PDFgui refinement (Fig. 1i and Fig. S18, ESI†).

The partial double bond length of C=N–C is reduced from 1.52 Å (HCN) to 1.47 Å (HCN-A). The N–H bond length is 1.11 Å for HCN-A, which is shorter than that of HCN (1.15 Å). The reduced bond lengths of both C–N and N–H manifest the promoted crystallinity.⁴⁴ The detailed N1s XPS spectrum is shown in Fig. S19a (ESI†), from which the C–N=C, N–(C)₃ and C–N–H groups are found in HCN-A (Tables S2–S4, ESI†). The solid ¹³C NMR spectra (Fig. S19b, ESI†) reveal the presence of C–(N)₃ and NH₂–C(N)₂ in the tri-s-triazine motifs. Moreover, the C K-edge synchrotron-based X-ray absorption near-edge structure spectroscopy (XANES) results (Fig. 1i) demonstrate two characteristic dipole transition 2p π* resonance signals of C=C bonds at the defect sites, and signals of C–N–C bonds at 284.9 and 287.8 eV are observed for HCN and HCN-A. In the N K-edge region (Fig. 1j), two 2p π*characteristic resonances are observed at 399.1/399.2 eV and 402.0/402.1 eV in HCN-A and HCN, respectively, which correspond to the N–C=N (N₁) coordination structure in a heptazine unit and graphitic-type N–(C)₃ (N₂) bridging among the three heptazine units. Notably, a slight negative shift (about 0.17 eV) in the absorption edge is observed in HCN-A (Fig. 1i), suggesting the existence of a strong interaction of C–N–C and providing evidence of the elevated crystalline structure.⁴⁵ The intensity of both the N₁ and N₂ peaks in HCN-A is higher than that of the peaks in HCN, further signifying the well-ordered structure.⁴⁶

Collectively, based on the abovementioned results, we demonstrate that a unique hollow bubble-shape carbon nitride catalyst with alternately appeared amorphous domains and well-crystallized lattice matrices was successfully prepared. To unveil the electronic structure, the densities of states (DOS) by virtue of first principles calculations based on density functional theory (DFT) were determined (Fig. S20 and S21, ESI†). For HCN, the CB is mainly composed of the C 2p and N 2p orbitals, while the N 2p orbitals contribute to the VB; this is in agreement with the literature.⁴⁶ The calculated band gap of HCN-A is smaller than that of HCN. An additional midgap state is presented in HCN-A, which could facilitate carrier separation and lead to charge enrichment on the active sites of pyridinic N of the tri-s-triazine structure units.¹⁴ In addition, the optical absorption properties and band alignment positions are demonstrated in Fig. S22–S27 (ESI†), signifying that the conduction band (CB) of HCN-A is negative enough for efficient CH₃CHO generation.²³ The charge dynamics were investigated by femtosecond transient absorption measurements (fs-TA, Fig. S28 and S29, ESI†). The longer relaxation time in HCN-A indicates more efficient charge separation and a higher concentration of charge carriers to participate in the catalytic reaction.⁴⁷

The photocatalytic CO₂ conversion performance of CN, HCN and HCN-A was investigated (for experimental details, see Fig. S30–S33, ESI†). As shown in Fig. 2a, HCN-A presents a CH₃CHO generation rate of 1083.5 μmol h⁻¹ g⁻¹ for the first 2 h, which was 45.6 fold higher than that of CN (22.3 μmol h⁻¹ g⁻¹) and 7.4 fold higher than that of HCN (145.3 μmol h⁻¹ g⁻¹) (the gas chromatograph (GC) signals for the products are shown in Fig. S34, ESI†). The inset image in Fig. 2a presents that the HCN exhibits a HCHO evolution rate of 125.5 μmol h⁻¹ g⁻¹. In contrast, HCN-A exhibits a much lower HCHO production rate of 34.7 μmol h⁻¹ g⁻¹ (Fig. S35, ESI†). As shown in Fig. 2b and Table S5 (ESI†), the photocatalytic activity and selectivity of HCN-A could be regulated by varying the amount of AP added. Specifically, the selectivity for CH₃CHO increases from 51.6% in HCN to 95.5% in HCN-A₃ (1083.5 μmol h⁻¹ g⁻¹). As such, the HCN-A with locally crystallized domains achieves high selectivity for CH₃CHO production, which can be ascribed to the AP-assisted nucleation during the hydrothermal treatment. Notably, the typical gas products for the CO₂ reduction, such as CO and CH₄, were not detected in this reaction system. The addition of recently routinely exploited co-catalysts, including the Fe, Co, Ni, Ru and Cu species, to HCN-A did not lead to any improvement in the photocatalytic activity or selectivity (Fig. S36, ESI†). Specifically, the addition of Co species in HCN-A results in a main product of gaseous CO (Fig. S37, ESI†). Control experiments in Fig. 2c revealed that the HCN-A catalyst, light source, CO₂ gas, and H₂O were indispensable for this reaction.

Fig. 2 (a) Time dependence of the CH₃CHO product evolution rates of CN, HCN, and HCN-A with 3 mL solvent; inset shows the HCHO generation rates over CN, HCN, and HCN-A. (b) The average generation rates of the CO₂ reduction products and the selectivities of CH₃CHO using CN, HCN, HCN-A₁, HCN-A₂, HCN-A₃, and HCN-A₄ with 3 mL solvent. (c) The average generation rates of CO₂ reduction products under multiple experiments of control conditions by using HCN-A₃ with 3 mL solvent. (d) The generation rates of CO₂ reduction products upon altering the volume of the reaction solution in this system. (e) Time dependence of the CO₂ reduction product evolution rates of CN, HCN, and HCN-A with 30 mL solvent. (f) Quantum efficiencies of HCN and HCN-A for the CH₃CHO product (irradiated by LED light sources with different monochromatic wavelengths of 365 nm, 385 nm, 420 nm, 450 nm, 485 nm, 535 nm, 595 nm and 600 nm), together with the DRS spectra of HCN and HCN-A with 30 mL solvent.

As shown in Fig. 2a, the product yield reaches a plateau after 2 h and stops increasing.^29,48 To address this problem, we further investigated the effect of the reaction solution volume on the reaction activity and selectivity, as shown in Fig. 2d. Clearly, as the volume increased to 30 mL, the generation yield of CH₃CHO on HCN-A further increased to 1814.7 μmol h⁻¹ g⁻¹ with a selectivity of 98.3%, compared to 230.7 μmol h⁻¹ g⁻¹ on HCN with a selectivity of 54.7%. The increase in the CH₃CHO product yield is due to the dilution of the initially formed CH₃CHO, the accumulation of which was found to inhibit the right-shifting of the reaction equilibrium (Fig. S38, ESI†). Further increasing the volume to 40 mL resulted in a slight decrease in the CH₃CHO yield, which is due to the decreased mass concentration of the catalyst, as demonstrated in Fig. S39 (ESI†).

Furthermore, when the volume is 30 mL, the CH₃CHO product yield progressively increases during the measurement (8 h), as shown in Fig. 2e. In this case, increasing the solvent volume is an efficient strategy for boosting both the selectivity and yield for CH₃CHO production. Actually, we have screened a series of other semiconductor photocatalysts under identical reaction conditions for comparison (Fig. S40, ESI†). Most of the widely used catalysts were almost inactive for the conversion of CO₂ to CH₃CHO, and only trace amounts of CH₃CHO were found in the cases of Znln₂S₄ and CdS.

Increasing the light intensity from 50 to 150 mW cm⁻² led to enhancement of the product yield (Fig. S41, ESI†), demonstrating that the successful conversion of CO₂ to CH₃CHO is driven by the light source. Along with its outstanding activity and high selectivity, HCN-A exhibited prominent quantum efficiencies (QEs) of 22.4% and 13.3% at 385 nm and 420 nm, respectively (the calculation process is listed in Note S1, Tables S6 and S7, ESI†); these are clearly higher than those for HCN of 4.5% and 2.5%, respectively, as shown in Fig. 2f (the relationship between the catalyst dosage and performance activity is shown in Fig. S42, ESI†). The QE value of HCN-A decreases as the light wavelength increases, which coincides well with the trend in the diffuse reflection spectra. Compared to HCN, the CH₃CHO generation by HCN-A could be extended to longer wavelengths (595–630 nm) than that by HCN (Table S7, ESI†). The QE of HCN-A was determined to be 0.86% at λ = 595 nm and 0.19% at λ = 630 nm, signifying improved light-responsive activity over a broadened visible light region. More importantly, the HCN-A catalyst showed high stability after 10 cycle tests, as shown in Fig. S43 (ESI†). No obvious change was found in the structure or morphology after the stability tests, as confirmed by the XRD patterns, FTIR spectra (Fig. S44, ESI†) and TEM images (Fig. S45, ESI†). HCN-A exhibits excellent photocatalytic activity and selectivity towards CH₃CHO generation, outperforming the state-of-art photocatalysts, as summarized in Table S8 (ESI†).

To further confirm the CH₃CHO formation, GC-MS and ¹³CO₂ isotope labeling experiments were first conducted. The photographs of the employed devices for the measurements are shown in Fig. S46 and S47 (ESI†). The GC spectra confirm the dissolved CO₂ molecule and the CH₃CHO product (Fig. 3c) with either CO₂ or ¹³CO₂ as the feedstock. The dissolved CO₂ molecule weight increased when CO₂ was replaced with ¹³CO₂ isotope, as shown in Fig. 3d. The molecular weight of the generated CH₃CHO product was measured to be m/z = 43 (the molecular weight for the anion should minus 1), corresponding to the formation of CH₃CHO (Fig. 3e).^30,49 The ¹³CH₃ and ¹³CHO peaks located at m/z = 46 and 47 evidently prove that the carbon source of CH₃CHO is derived from the CO₂ feedstock, which is in accord with previous results.⁵⁰ We further employed deuterium oxide (D₂O) and ¹⁸O-labeled CO₂ in the reaction to trace the origin of the H and O elements in the produced CH₃CHO. The product peak at m/z = 47 was indexed to CD₃CDO (Fig. 3f) when using D₂O in the reaction, indicating that the proton in CH₃CHO comes from H₂O. This result further manifests that the formation of CH₃CHO is a multi-step hydrogenation process and that the proton is provided by H₂O.^51–53 Meanwhile, the molecule weight increased to m/z = 45 when C¹⁸O₂ was used, suggesting that the oxygen atom in CH₃CHO is derived from the CO₂ feedstock. The applied TEOA was oxidized to N-ethyl-2-methoxy-N-(2-methoxyethyl)-ethanamine, as verified by GC-MS (Fig. S48 and S49, ESI†).⁴⁰ All these results explicitly reveal that HCN-A photocatalysts can catalyze the stepwise conversion of CO₂ into CH₃CHO products via C–C coupling and a multiple proton-coupled electron transfer process. We note that this the first time that the origin of CH₃CHO through photocatalytic CO₂ conversion has been revealed with the assistance of isotope labeling experiments by using ¹³CO₂, CO₂, D₂O and C¹⁸O₂ as the feedstocks. Furthermore, the ¹H NMR spectra of the product confirm the formation of CH₃CHO (Fig. S50, ESI†). To the best of our knowledge, this is also the first time that a PCN-based catalyst was designed and applied for highly selective and efficient conversion of CO₂ to CH₃CHO.

Fig. 3 GC-MS spectra of CH₃CHO generated over HCN-A in the photocatalytic CO₂ conversion. (a) GC spectra of dissolved CO₂ molecules and the CH₃CHO product with CO₂ and ¹³CO₂. (b) MS spectra of CO₂ and ¹³CO₂. (c) MS spectra of CH₃CHO with the use of CO₂ + H₂O, ¹³CO₂ + H₂O. (d) MS spectra of CH₃CHO with the use of ¹³CO₂ + D₂O, C¹⁸O₂ + H₂O.

To understand the underlying reasons for the enhanced photocatalytic performance, we investigated the fundamental steps of the CO₂ conversion. The surface adsorption property of CO₂ is a prerequisite for conversion of CO₂ molecules. We therefore first measured the CO₂ adsorption capacity of the samples. As shown in Fig. S51 (ESI†), both HCN-A and HCN present fundamentally higher CO₂ adsorption than the pristine CN, which is due to the enhanced surface area. No significant change in the CO₂ adsorption capacity was found between HCN-A and HCN; therefore, the advanced photocatalytic performance of HCN-A compared to HCN cannot be ascribed to the CO₂ adsorption property.

It is worth noting that currently, the mechanism for the conversion of CO₂ to CH₃CHO has been seldom disclosed. To date, only the CO₂ adsorption and the initiate CO₂ activation process have been investigated.³⁰ In this work, we aimed to disclose the whole reaction mechanism. The photocatalysis intermediate on HCN-A was verified by in situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS, Fig. 4a–c) and compared with HCN (Fig. 4d–f) under visible light illumination. The peaks attributed to the monodentate bicarbonates (1532, 1513 cm⁻¹), bidentate bicarbonates (1420, 1375, 1320 cm⁻¹) and hydroxyl groups of carboxylic acid (1154, 1465 cm⁻¹) were recorded with increasing irradiation time (Fig. 4a).⁵⁴ The characteristic peak of *CO at 2078 cm⁻¹ obviously appears under illumination (Fig. 4b).⁵⁴ The peak intensities of HCN-A are notably higher than those of HCN (Fig. 4d), signifying the accelerated CO₂ transformation process to *CO in HCN-A. The asymmetric C–H stretching vibration (2942 and 2968 cm⁻¹), symmetric C–H stretching vibration (2878 cm⁻¹), and CH₃ bending vibration (1380–1408 cm⁻¹) were observed in the HCN-A reaction system, which can be assigned to the –CH₃ group of CH₃CHO.⁵⁵ More importantly, the vibration of the *CHO group at 1097 cm⁻¹ and the asymmetric stretching vibrations of C–C–O at 1038 cm⁻¹ and for *OCCHO at 917 cm⁻¹ were investigated. The *CHO formed from the hydrogenation of the *CO group is the prerequisite to generate the *OCCHO intermediate (eqn (1)), and the *OCCHO intermediate is generally regarded as the crucial intermediate for the conversion of CO₂ to the multicarbon (C₂₊) products.^14,53 The formation of *OCCHO through the C–C coupling of *CHO with *CO⁵⁶ is depicted in eqn (2).

Fig. 4 In situ diffuse reflectance infrared Fourier transform spectroscopy of HCN-A (a–c) and HCN (d–f). The IR spectrum was collected with interval times of 0 min, 3 min, 4 min, 5 min, 10 min, 15 min and 20 min for HCN-A and at 5 min intervals for HCN under illumination.

The possible reduction paths are as follows (for details, see Experimental section 1.6 in the ESI†):

*CO + H⁺ + e⁻ → *CHO (1)

*CHO + *CO + e⁻ → *OCCHO (2)

*CHO + H⁺ + e⁻ → *CH₂O → HCHO (3)

Additionally, the C=O stretching vibration (1717, 1661, 1630 cm⁻¹) and C–H out-of-plane bending vibration (1102 cm⁻¹) are present in the spectrum of HCN-A, together verifying the formation of CH₃CHO.⁵⁷ These peaks were also observed when using HCN, but the intensity was obviously lower (Fig. 4d–f); this suggests that CH₃CHO can also be generated by HCN, although the activity is substantially limited. This is in agreement with the photocatalytic performance observed experimentally. For HCN, two new peaks at 2017 cm⁻¹ and 2087 cm⁻¹ appear in addition to the *CHO band (1098 cm⁻¹), which can be attributed to the stretching vibration of the surface-absorbed *CH₂O⁵⁸ (eqn (3)), as shown in Fig. 4e. This indicates that the HCHO intermediate can be easily generated on HCN, leading to HCHO formation.^49,51–53 By contrast, this was not found in HCN-A; this may be due to the fact that *CHO in HCN-A is capable of rapid C–C coupling to form *OCCHO (eqn (2)) rather than being competitively protonated to form *CH₂O (eqn (3)). The successful formation of *OCCHO is the crucial step that results in high selectivity for the CH₃CHO product. These results conclusively demonstrate that the PCN-based samples with AP modification exhibit suppressed *CHO protonation while simultaneously boosting the C–C coupling and multiple proton-coupled electron transfer of CO₂ to form the CH₃CHO species. In situ EPR measurements were further applied to detect the radical active species during the CO₂ reduction reaction (Fig. S52, ESI†). Under illumination, the EPR signal of the characteristic DMPO–CO₂˙⁻ is clearly present in the reaction system with HCN-A as the photocatalyst, indicating that the CO₂ molecules can be activated to CO₂˙⁻ radicals. The CO₂ to CO₂˙⁻ process is then followed by the generation of the important *CO intermediate, which is the prerequisite for the conversion of CO₂ to the final CH₃CHO product. After bubbling CO₂ into the HCN-A catalyst suspension, the charge dynamics (fs-TA) were investigated (Fig. S53, ESI†). Notable decay of the trapped electrons in HCN-A was observed. This indicates that the reactant CO₂ would preferentially react with the trapped electrons and then be activated to CO₂˙⁻ radical.

To further reveal the reaction mechanisms, Gibbs free energy calculations (Fig. S54, ESI†) based on density functional theory (DFT) were performed; as illustrated in Fig. 5a, the energy barriers for the *CHO intermediate formation are 0.28 eV and −0.46 eV on HCN and HCN-A, respectively, suggesting easier formation of *CHO on HCN-A. The Gibbs free energy of CHO* formation is obviously smaller than the desorption energy of the CO molecules (Fig. 5b) on HCN-A. This means that the formation of the CHO* intermediate is an exothermic and spontaneous process (eqn (1)), while the desorption of CO* is endothermic and has a large activation energy barrier on HCN-A. This signifies that HCN-A favors the hydrogenation of CO* to generate CHO* rather than the desorption of CO* molecules from the catalyst surface. Therefore, there is no CO evolution on HCN-A. Meanwhile, a likely intermediate of *OCCHO in the C–C coupling pathway via a favorable thermodynamic process could be spontaneously achieved on HCN-A, which serves as a crucial intermediate for the formation of C₂ species⁵⁶ (eqn (2)). For the C–C coupling on the HCN-A surface, the formation energy of *OCCHO when *CO species reacts with the carbon atom of *CHO is −0.18 eV. In contrast, the *CHO is difficult to protonate on HCN-A to form *CH₂O (formaldehyde), which accounts for the high selectivity of the visible light-driven CO₂ conversion to CH₃CHO. Contrastingly, the formation of *OCCHO on HCN (Fig. 5a) still needs to overcome a certain energy barrier, whereas the Gibbs energy of hydrogenation of *CHO to form *CH₂O is close to the energy barrier for the C–C coupling to form *OCCHO, leading to a high possibility of HCHO formation. As such, the HCN could simultaneously yield HCHO and CH₃CHO with lower product selectivity, which is in accord with the experimental results (Fig. 2b).

Fig. 5 Calculated free energy diagram for the reduction of CO₂ to CH₃CHO on the HCN and HCN-A catalysts (a–d). (e) The proposed reaction mechanism for the photocatalytic CO₂ reduction conversion to CH₃CHO. The grey, red, and green color spheres denote carbon, oxygen, and hydrogen atoms, respectively.

It is worth noting that the formed CH₃CHO may also be further hydrogenated into CH₃CH₂OH via the formation of an *OCH₂CH₃ intermediate.⁴⁹ The formation of *OCH₂CH₃ is an uphill process, with an energy of 1.13 V vs. RHE in HCN-A; this is in accordance with previously reported results of CO₂ reduction.^49,52 This endothermic nature of the process limits further hydrogenation of CH₃CHO to CH₃CH₂OH (eqn (4) and (5)), which is supported by the experimental observation that no CH₃CH₂OH is released from the reaction system of HCN-A (Fig. 5d).

CH₃CHO + H⁺ + e⁻ → *OCH₂CH₃ (4)

*OCH₂CH₃ + H⁺ + e⁻ → CH₃CH₂OH (5)

The primary product CH₃CHO is eventually generated through the multi-hydrogenation-coupled electron transfer process with 98.3% product selectivity under visible-light irradiation on HCN-A. This result substantially indicates that the locally crystallized HCN-A catalyst favors *OCCHO formation via C–C coupling rather than the hydrogenation of *CHO to form *CH₂O. Meanwhile, HCN-A also possesses a high CH₃CH₂OH thermodynamic energy barrier, thus prohibiting CH₃CH₂OH formation and accounting for the excellent selectivity and activity of the conversion of CO₂ to CH₃CHO.

Thus, a reasonable mechanism for the reduction of CO₂ on HCN-A is proposed (Fig. 5e). Under visible light illumination, the AP-induced locally crystallized HCN-A enables the facile formation of C₂ intermediates and exhibits enhanced photocatalytic activity for the conversion of CO₂ to CH₃CHO. HCN-A could efficiently activate the CO₂ molecules to CO₂˙⁻ radicals and generate the *CO intermediate, which is further hydrogenated to form the *CHO intermediate. Compared to the amorphous HCN, the locally crystallized HCN-A is prone to have stronger bonding with the *OCCHO group, which regulates the endoergic C–C coupling step to a spontaneous process and changes the reaction pathway to form CH₃CHO instead of HCHO. The proton is rapidly extracted from the dissociation of H₂O. Eventually, CH₃CHO can be obtained via the C–C coupling and multi-step hydrogenation process with high selectivity and activity.

Conclusions

In summary, we have developed a locally crystallized HCN-A catalyst via hydrothermal pre-treatment with an amine-assisted nucleation strategy to enable the formation of alternately appearing amorphous domains and a well-crystallized lattice matrix framework. The HCN-A breaks the trade-off between the efficiency and the selectivity for photocatalytic CO₂ reduction to CH₃CHO. It furnishes CH₃CHO product with a high-yielding rate of 1814.7 μmol h⁻¹ g⁻¹ under mild conditions, accompanied with a QE of 13.3% at λ = 420 nm. The CH₃CHO selectivity on HCN-A reaches 98.3%, which is significantly higher than the value of 54.7% on HCN. This locally crystallized HCN-A exhibits significantly improved selectivity and activity compared to reported CO₂ reduction photocatalysts. We reveal the C–C coupling and multi-step hydrogenation mechanisms for the photocatalytic conversion of CO₂ to CH₃CHO on the locally crystallized HCN-A, and *OCCHO is determined to be the key reaction intermediate. Compared to the endoergic C–C coupling step on HCN, the locally crystallized HCN-A enables a spontaneous exoergic reaction to form the *OCCHO intermediates. The HCN-A shows suppressed hydrogenation of *CHO to *CH₂O, leading the reaction pathway towards CH₃CHO instead of HCHO. Our work not only demonstrates an efficient strategy for grafting locally crystallized PCN for highly selective CH₃CHO photosynthesis in a sustainable way, but also proposes a plausible mechanism for the whole reaction. The research findings shed light on the highly selective photosynthesis of liquid multi-carbon products by CO₂ reduction.

Author contributions

Q. L. Conceptualization, methodology, investigation, writing-original draft, visualization; H. C. Methodology, investigation, visualization; T. C. and T. B. L. Performed high-resolution X-ray total scattering spectra, PDFgui refinement, synchrotron-based X-ray absorption structure spectroscopy; M. X. Analysed GC-MS. F. W. Project administration; conceptualization and supervision; writing – review & editing; funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was funded by the National Natural Science Foundation of China (61904167), the Natural Science Foundation of Guangdong Province (2019A1515012081), the Science and Technology Program of Guangzhou (202002030017), the GDAS' Project of Science and Technology Development, and the China Postdoctoral Science Foundation (2020M672638).

Notes and references

1. M. Eckert, G. Fleischmann, R. Jira, H. M. Bolt and K. Golka, Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH, Weinheim, 2006.
2. R. Jira, Angew. Chem., Int. Ed., 2009, 48, 9034–9037.
3. J. A. Keith and P. M. Henry, Angew. Chem., Int. Ed., 2009, 48, 9038–9049.
4. P. Liu and E. J. M. Hensen, J. Am. Chem. Soc., 2013, 135, 14032–14035.
5. J. Mielby, J. O. Abildstrøm, F. Wang, T. Kasama, C. Weidenthaler and S. Kegnæs, Angew. Chem., Int. Ed., 2014, 126, 12721–12724.
6. S. Xie, W. Ma, X. Wu, H. Zhang, Q. Zhang, Y. Wang and Y. Wang, Energy Environ. Sci., 2021, 14, 37–89.
7. J. Albero, Y. Peng and H. García, ACS Catal., 2020, 10, 5734–5749.
8. W. Yang, H.-J. Wang, R.-R. Liu, J.-W. Wang, C. Zhang, C. Li, D.-C. Zhong and T.-B. Lu, Angew. Chem., Int. Ed., 2021, 60, 409–414.
9. Q. Liu, L. Wei, Q. Xi, Y. Lei and F. Wang, Chem. Eng. J., 2020, 383, 123792.
10. Y. Li, M. Wen, Y. Wang, G. Tian, C. Wang and J. Zhao, Angew. Chem., Int. Ed., 2021, 60, 910–916.
11. Q. Wang, J. Warnan, S. Rodríguez-Jiménez, J. J. Leung, S. Kalathil, V. Andrei, K. Domen and E. Reisner, Nat. Energy, 2020, 5, 703–710.
12. G. Qin, Y. Zhang, X. Ke, X. Tong, Z. Sun, M. Liang and S. Xue, Appl. Catal., B, 2013, 129, 599–605.
13. Y. Wang, Y. Tian, L. Yan and Z. Su, J. Phys. Chem. C, 2018, 122, 7712–7719.
14. X. Li, Y. Sun, J. Xu, Y. Shao, J. Wu, X. Xu, Y. Pan, H. Ju, J. Zhu and Y. Xie, Nat. Energy, 2019, 4, 690–699.
15. H. Wu, X. Y. Kong, X. Wen, S.-P. Chai, E. C. Lovell, J. Tang and Y. H. Ng, Angew. Chem., Int. Ed., 2021, 60, 8455–8459.
16. J. Xu, Z. Ju, W. Zhang, Y. Pan, J. Zhu, J. Mao, X. Zheng, H. Fu, M. Yuan and H. Chen, Angew. Chem., Int. Ed., 2021, 133, 8787–8791.
17. Y. Wang, X. Liu, X. Han, R. Godin, J. Chen, W. Zhou, C. Jiang, J. F. Thompson, K. B. Mustafa, S. A. Shevlin, J. R. Durrant, Z. Guo and J. Tang, Nat. Commun., 2020, 11, 2531.
18. Y. A. Wu, I. McNulty, C. Liu, K. C. Lau, Q. Liu, A. P. Paulikas, C.-J. Sun, Z. Cai, J. R. Guest, Y. Ren, V. Stamenkovic, L. A. Curtiss, Y. Liu and T. Rajh, Nat. Energy, 2019, 4, 957–968.
19. J. Li, W. Pan, Q. Liu, Z. Chen, Z. Chen, X. Feng and H. Chen, J. Am. Chem. Soc., 2021, 143, 6551–6559.
20. Y. Liu, B. Huang, Y. Dai, X. Zhang, X. Qin, M. Jiang and M.-H. Whangbo, Catal. Commun., 2009, 11, 210–213.
21. S. Zhu, X. Li, X. Jiao, W. Shao, L. Li, X. Zu, J. Hu, J. Zhu, W. Yan, C. Wang, Y. Sun and Y. Xie, Nano Lett., 2021, 21, 2324–2331.
22. X. Li, J. Yu, M. Jaroniec and X. Chen, Chem. Rev., 2019, 119, 3962–4179.
23. X. Zhao, L. Du, B. You and Y. Sun, Catal. Sci. Technol., 2020, 10, 2711–2720.
24. I. Shown, H.-C. Hsu, Y.-C. Chang, C.-H. Lin, P. K. Roy, A. Ganguly, C.-H. Wang, J.-K. Chang, C.-I. Wu, L.-C. Chen and K.-H. Chen, Nano Lett., 2014, 14, 6097–6103.
25. X. Qian, W. Yang, S. Gao, J. Xiao, S. Basu, A. Yoshimura, Y. Shi, V. Meunier and Q. Li, ACS Appl. Mater. Interfaces, 2020, 12, 55982–55993.
26. O. Ola, M. Maroto-Valer, D. Liu, S. Mackintosh, C.-W. Lee and J. C. S. Wu, Appl. Catal., B, 2012, 126, 172–179.
27. O. Ola and M. M. Maroto-Valer, Chem. Eng. J., 2016, 283, 1244–1253.
28. C.-W. Lee, R. Antoniou Kourounioti, J. C. S. Wu, E. Murchie, M. Maroto-Valer, O. E. Jensen, C.-W. Huang and A. Ruban, J. CO2 Util., 2014, 5, 33–40.
29. P.-Y. Liou, S.-C. Chen, J. C. Wu, D. Liu, S. Mackintosh, M. Maroto-Valer and R. Linforth, Energy Environ. Sci., 2011, 4, 1487–1494.
30. I. Shown, S. Samireddi, Y.-C. Chang, R. Putikam, P.-H. Chang, A. Sabbah, F.-Y. Fu, W.-F. Chen, C.-I. Wu, T.-Y. Yu, P.-W. Chung, M. C. Lin, L.-C. Chen and K.-H. Chen, Nat. Commun., 2018, 9, 169.
31. L. Tang, L. Kuai, Y. Li, H. Li, Y. Zhou and Z. Zou, Nanotechnology, 2018, 29, 064003.
32. J. Xiao, W. Yang, S. Gao, C. Sun and Q. Li, J. Mater. Sci. Technol., 2018, 34, 2331–2336.
33. M. L. Ovcharov, A. M. Mishura, N. D. Shcherban, S. M. Filonenko and V. M. Granchak, Sol. Energy, 2016, 139, 452–457.
34. X. Zhi, A. Vasileff, Y. Zheng, Y. Jiao and S.-Z. Qiao, Energy Environ. Sci., 2021, 14, 3912–3930.
35. W. Ma, S. Xie, T. Liu, Q. Fan, J. Ye, F. Sun, Z. Jiang, Q. Zhang, J. Cheng and Y. Wang, Nat. Catal., 2020, 3, 478–487.
36. D.-H. Nam, P. De Luna, A. Rosas-Hernández, A. Thevenon, F. Li, T. Agapie, J. C. Peters, O. Shekhah, M. Eddaoudi and E. H. Sargent, Nat. Mater., 2020, 19, 266–276.
37. Q. Liu, Z. Chen, W. Tao, H. Zhu, L. Zhong, F. Wang, R. Zou, Y. Lei, C. Liu and X. Peng, J. Mater. Chem. A, 2020, 8, 11761–11772.
38. G. Lligadas, J. C. Ronda, M. Galià and V. Cádiz, J. Polym. Sci., Part A: Polym. Chem., 2013, 51, 2111–2124.
39. X. Xiao, Y. Gao, L. Zhang, J. Zhang, Q. Zhang, Q. Li, H. Bao, J. Zhou, S. Miao, N. Chen, J. Wang, B. Jiang, C. Tian and H. Fu, Adv. Mater., 2020, 32, 2003082.
40. Q. Liu, H. Cheng, T. Chen, T. W. B. Lo, J. Ma, A. Ling and F. Wang, Appl. Catal., B, 2021, 295, 120289.
41. W.-J. Jiang, S. Niu, T. Tang, Q.-H. Zhang, X.-Z. Liu, Y. Zhang, Y.-Y. Chen, J.-H. Li, L. Gu, L.-J. Wan and J.-S. Hu, Angew. Chem., Int. Ed., 2017, 56, 6572–6577.
42. S. Kusama, T. Saito, H. Hashiba, A. Sakai and S. Yotsuhashi, ACS Catal., 2017, 7, 8382–8385.
43. C. Hu, F. Chen, Y. Wang, N. Tian, T. Ma, Y. Zhang and H. Huang, Adv. Mater., 2021, 33, 2101751.
44. R. R. Lieten, C. Fleischmann, S. Peters, N. M. Santos, L. M. Amorim, Y. Shimura, N. Uchida, T. Maeda, S. Nikitenko, T. Conard, J.-P. Locquet, K. Temst and A. Vantomme, ECS J. Solid State Sci. Technol., 2014, 3, P403–P408.
45. H. Wang, X. Sun, D. Li, X. Zhang, S. Chen, W. Shao, Y. Tian and Y. Xie, J. Am. Chem. Soc., 2017, 139, 2468–2473.
46. D. Zhao, C.-L. Dong, B. Wang, C. Chen, Y.-C. Huang, Z. Diao, S. Li, L. Guo and S. Shen, Adv. Mater., 2019, 31, 1903545.
47. Q. Liu, C. Chen, K. Yuan, C. D. Sewell, Z. Zhang, X. Fang and Z. Lin, Nano Energy, 2020, 77, 105104.
48. Z. Jin, N. Murakami, T. Tsubota and T. Ohno, Appl. Catal., B, 2014, 150-151, 479–485.
49. E. Bertheussen, A. Verdaguer-Casadevall, D. Ravasio, J. H. Montoya, D. B. Trimarco, C. Roy, S. Meier, J. Wendland, J. K. Nørskov, I. E. L. Stephens and I. Chorkendorff, Angew. Chem., Int. Ed., 2016, 55, 1450–1454.
50. A. Ott, J.-E. Germond and A. Chaintreau, J. Agric. Food Chem., 2000, 48, 1512–1517.
51. A. J. Garza, A. T. Bell and M. Head-Gordon, ACS Catal., 2018, 8, 1490–1499.
52. L. Wang, D. C. Higgins, Y. Ji, C. G. Morales-Guio, K. Chan, C. Hahn and T. F. Jaramillo, Proc. Natl. Acad. Sci. U. S. A., 2020, 117, 12572–12575.
53. X. Chang, A. Malkani, X. Yang and B. Xu, J. Am. Chem. Soc., 2020, 142, 2975–2983.
54. X. Ye, C. Yang, X. Pan, J. Ma, Y. Zhang, Y. Ren, X. Liu, L. Li and Y. Huang, J. Am. Chem. Soc., 2020, 142, 19001–19005.
55. Z. Yu and S. S. C. Chuang, J. Catal., 2007, 246, 118–126.
56. J. H. Montoya, A. A. Peterson and J. K. Nørskov, ChemCatChem, 2013, 5, 737–742.
57. X.-Y. Ma, C. Ding, H. Li, K. Jiang, S. Duan and W.-B. Cai, J. Phys. Chem. Lett., 2020, 11, 8727–8734.
58. J. T. Yates and R. R. Cavanagh, J. Catal., 1982, 74, 97–109.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ee02073k

‡ These authors contributed equally.

---