Ordered Hollow Sphere Array Nanoreactors for Direct Electroreduction of Diluted CO₂ Into Ethylene

ABSTRACT

Direct electroreduction of industrially diluted CO₂ into C₂H₄ represents a promising route for sustainable energy conversion, yet the selectivity and current density of this process are fundamentally constrained by the insufficient reactant concentration and generation of *CO intermediates. Herein, we addressed this issue by designing an ordered hollow sphere array nanoreactor with Cu nanoparticle and Ni single atom as the tandem catalytic sites (Cu‐NP/Ni‐NC OHSpA) using a unique reverse templating protocol. Mechanistic studies demonstrate that *CO can be generated on Ni site and undergoes spillover to the neighboring Cu particle. Significantly, the ordered hollow sphere array serves as effective CO reservoir to enhance the *CO coverage and thereby promote the C‐C coupling to form C₂H₄. As a consequence, Cu‐NP/Ni‐NC OHSpA exhibits excellent electrocatalytic performance toward the conversion of diluted CO₂ (i.e., 15% CO₂/Ar), affording a C₂H₄ Faradaic efficiency (FE_C2H4) of 57.5% with a partial current density (J_C2H4) of 110 mA cm⁻².

An ordered hollow sphere array nanoreactor integrating both tandem catalysis and molecule enrichment functions is able to break the production and enrichment limitations of CO intermediates, thereby affording an outstanding efficiency in industrial dilute CO₂ electroreduction to ethylene.

1. Introduction

Electrochemical conversion of carbon dioxide into value‐added chemicals and fuels offers a promising approach to mitigate the current energy and environmental issues caused by excessive CO₂ emissions [1, 2]. In the electrochemical CO₂ reduction reaction, CO₂ can be converted into single‐carbon (C₁) products (e.g., carbon monoxide (CO), methane (CH₄) and formate (HCOO⁻)) and multicarbon (C₂₊) products (e.g., ethylene (C₂H₄), ethanol (C₂H₅OH)) [3, 4]. Compared with the C₁ products, the production of C₂₊ products has attracted more interest due to their higher energy density and economic value. For example, C₂H₄ with a market price of ca. 2.0 US$/kg is an important industry feedstock to produce polyethylene and polyvinyl chloride [5]. Currently, pure CO₂ has been usually utilized in CO₂ reduction reaction in pursuit of high selectivity of C₂H₄ product [6, 7]. However, the capture, enrichment, and purification processes with high energy consumption and extra investment costs are essential for acquiring high‐purity CO₂ since the low‐concentration nature of anthropogenic CO₂ (e.g., CO₂ concentrations are below 15% in industrial waste gases) [8, 9, 10]. Thus, the direct electroreduction of industrially diluted CO₂ into C₂H₄ has emerged as an attractive strategy as it can avoid the high energy consumption from the separation and concentration of anthropogenic CO₂.

It has been commonly hypothesized that the formed *CO serves as the intermediate that determines the selectivity of C₂H₄. During the CO₂‐to‐C₂H₄ conversion process, CO₂ molecule is converted to *CO via the COOH* intermediate and the as‐derived *CO can be desorbed to form gaseous CO product or dimerized to generate C₂ products [11, 12]. The major obstacle to obtain C₂H₄ product from diluted CO₂ comes from the sluggish kinetic of C–C coupling process since the generation of *CO intermediate is inhibited and the competitive hydrogen evolution reaction (HER) will become dominant under the conditions with insufficient reaction substrates (i.e., CO₂) [13, 14]. Given that the surface coverage of *CO correlates closely with the C–C coupling process, an efficient way to accelerate C₂H₄ evolution is to promote the formation and lift the local concentration of CO via engineering the active sites and architectures of catalysts to ensure a sufficient coverage of *CO for dimerization. For example, designing a tandem catalyst with dual active sites for CO generation and C‐C coupling respectively is able to spatially decouple the CO₂ to C₂H₄ process, thus effectively improving the yield of C₂ products [15, 16]. An alternative approach to increase the local concentration of CO is to create a nanoconfinement space where the close contact and interaction between CO adsorbates can increase the incidence of dimerization. In this case, hollow architectures serve as the ideal platform for the CO₂‐to‐C₂H₄ conversion, which may facilitate the enrichment and transformation of CO₂ molecules as well as the capture of CO intermediates, thus promoting the C–C coupling [17, 18, 19]. Nevertheless, conventional hollow structures inevitably possess open defects or large pores that connect the inner cavity to the external electrolyte, offering rapid escape pathways for CO intermediates. Even if the shell is perfectly dense, CO produced on the outer surface is released directly without confinement, and consequently such unsophisticated architectures fail to accumulate sufficient *CO for efficient C–C coupling [15, 20, 21]. Enlightened by these premises, we reason that a precise control over the hollow structure and the internal active sites to strengthen the CO production and C‐C coupling processes is potentially able to achieve a high yield of C₂H₄ in electrocatalytic diluted CO₂ reduction (EDCR).

Herein, we report the design and fabrication of an ordered hollow sphere array nanoreactor (OHSpA) that integrates tandem catalysis and local enrichment of CO₂, and CO intermediates for efficient electroreduction of low‐concentration CO₂ to C₂H₄. We started with the synthesis principle of OHSpA. Computational screening suggests that a catalyst with both NiN₄C and metallic Cu sites should potentially facilitate the formation and dimerization of *CO intermediate while the ordered hollow sphere array architecture could promote the enrichment of intermediate. With this design principle, we developed a unique metal‐organic framework‐mediated reverse templating protocol to anchor Cu nanoparticles and atomically dispersed Ni within the OHSpA to construct a tandem catalytic nanoreactor (Cu‐NP/Ni‐NC OHSpA). Remarkably, Cu‐NP/Ni‐NC OHSpA exhibited outstanding electrocatalytic performance toward the reduction of diluted CO₂ (i.e., 15% CO₂/Ar), achieving a C₂H₄ Faradaic efficiency up to 57.5% with a maximum partial current density of as high as 110 mA cm⁻². Mechanistic studies unraveled that the formed *CO underwent spillover from the Ni site to the neighboring Cu particle, followed by a C‐C coupling process and deep reduction to generate C₂H₄. Moreover, the unique ordered hollow sphere array structure led to an increased surface coverage of CO adsorbates for more efficient carbon dimerization.

2. Results and Discussion

2.1. Catalyst Design

We first carried out density functional theory (DFT) calculations to screen the optimal catalytically active sites for CO₂‐to‐C₂ conversion. In this work, 12 kinds of transition metal‐based models with potential for the catalytic reduction of CO₂ were considered (Figures 1a and S1) [22, 23]. Previous reports have confirmed that the formation and dimerization of *CO intermediate are the key steps during the CO₂‐to‐C₂ conversion process [24, 25]. Thus, we sought to calculate the activity trends of CO₂‐to‐CO conversion over these models, where the hydrogen evolution reaction (HER) was considered as a competing process, especially in diluted CO₂ atmosphere (Figures S2–S6). Accordingly, the difference between thermodynamic limit potential (ΔU_L) for CO₂‐to‐CO and HER (i.e., U_L(CO₂‐CO)‐U_L(H₂)) was selected as a reasonable descriptor to screen the active sites with higher activity toward CO₂‐to‐CO conversion than HER [26]. As shown in the scaling relation plot (Figure 1b), the NiN₄C site exhibits the most positive value in ΔU_L (U_L(CO₂‐CO)‐U_L(H₂) = ‐0.06 V) and the relative low limiting potential for CO₂‐to‐CO (U_L(CO₂‐CO) = ‐1.05 V), indicating that CO₂ prefers to be selectively converted to *CO over the single‐atom Ni sites. In the case of the subsequent C–C coupling process, a moderate *CO binding energy is crucial as over‐weakening or over‐strengthening binding energies toward *CO will lead to the desorption or deep hydrogenation, respectively to form the C₁ products (e.g., CO and CH₄). Therefore, we established scaling relations by using the *CO adsorption energy and thermodynamic limiting potential for coupling as the descriptors to construct the activity trend for the evolution of C₂ products (Figures S7–S9). As shown in the reaction phase diagram (Figures 1c and S10), the Cu (111) model is an appropriate candidate for *CO to C₂ products, displaying a moderate *CO adsorption energy (‐0.52 eV) and the lowest limiting energy for C‐C coupling among the investigated candidates. Moreover, the NiN₄C site shows the weakest *CO adsorption strength (‐0.17 eV), suggesting that the formation of *CO on Ni site can be readily desorbed and potentially be captured by another active site (i.e., the CO spillover process) [27]. Overall, our computational screening suggests that a catalyst with both NiN₄C and metallic Cu sites should potentially facilitate the formation and dimerization of *CO intermediate, and therefore could effectively promote the reduction of diluted CO₂ to form C₂₊ products.

FIGURE 1.

Screening of the optimal active sites and architecture. (a) Distribution of favorable products in CO₂ reactions for different structures. Schematic illustration of the main products of electrocatalytic reduction of CO₂ over 12 kinds of candidate catalysts. (b) U_L(CO₂‐CO) versus [U_L(CO₂‐CO)‐U_L(H₂)] over 12 kinds of catalyst models. (c) Reaction phase diagram for electrocatalytic reduction of CO₂ to C₂₊. The red and blue dashed lines represent the reaction free energies for the C‐C coupling and the protonation process of *CO, respectively. (d) Concentration (in mol m^‐3, color scale) distribution of carbon monoxide and C₂H₄ on the cavity restriction structure. FEM simulated CO and C₂H₄ intermediate concentration distribution over solid (model M1), hollow (model M2), ordered hollow array (model M3) and ordered solid sphere array (model M4) structures. Simulation results of the concentration of (e) CO and (f) C₂H₄ over different models.

Considering that the CO₂‐to‐C₂ conversion is carried out in diluted CO₂ atmosphere, an elaborate design of catalyst architecture is essential to concentrating the substrates and intermediates while improving their diffusion kinetics. Hollow structure is able to concentrate carbon intermediates through the spatial confinement effects [28, 29, 30]. We thus applied finite‐element method (FEM) simulations to explore the prospects of spatial confinement‐enhanced C₂ selectivity. Four sphere models with solid, hollow, hollow array and solid sphere structures were used to simulate the mass transport of C₁ and C₂ species around an individual particle (Figure S11a). The simulation results indicate that the hollow structure can restrict the outflow of locally produced CO species, leading to an increment of the surface coverage of intermediates necessary for C₂ production (Figures 1d and S11b,c). Significantly, such physical constraints effect will be significantly amplified by arranging hollow structures in an array. As shown in Figures 1e,f and S11, the model with hollow array structure is able to aggregate CO to the concentration of 1.61, 1.91 and 1.69 times that of the solid sphere, solid and hollow models, respectively, ultimately leading to 1.66, 2.01, 1.73 times enhancement in C₂H₄ generation rate. Altogether, these simulations validate the virtue of nanoreactors with hollow array structures in promoting the production of C₂ products.

2.2. Synthesis and Characterization of Materials

Under the guidance of the simulation results, we proposed a ZIF‐8 single‐crystal mediated reverse template strategy for the synthesis of ordered hollow sphere array nanoreactors (OHSpA) (Figure 2a). Briefly, an opal consisting of ordered closely packed polystyrene (PS) microspheres was used as an initial template to construct a three‐dimensional ordered macroporous silica (3DOM‐SiO₂) template via the hydrolysis of tetraethyl orthosilicate followed by the removal of PS with high‐temperature pyrolysis (Figure S12). Subsequently, Ni²⁺, Zn²⁺ and 2‐MeIM were impregnated into the interstices of 3DOM‐SiO₂ and crystallized to form Ni‐ZIF‐8, which was subjected to pyrolysis to form N‐doped carbon and Ni species. The Ni single‐atom decorated carbon OHSpA (Ni‐SA OHSpA) was then obtained after removing the SiO₂ template using a HF solution. Finally, Cu nanoparticles were implanted on Ni‐NC OHSpA via an impregnation‐reduction strategy to yield the Cu‐NP/Ni‐NC OHSpA.

FIGURE 2.

Preparation and characterization of catalysts. (a) Schematic illustration of the synthetic procedure for the Cu‐NP/Ni‐NC OHSpA. (b‐c) SEM and (d) TEM image of Ni‐NC OHSpA. (e‐f) TEM, (g) EDS mapping images and (h‐i) AC‐HAADF‐STEM of Cu‐NP/Ni‐NC OHSpA. (j) SEM image of Cu‐NP/Ni‐NC OHSpA with different sizes of sphere.

Scanning electron microscope (SEM) images clearly illustrate the highly uniform macroporous structure of the as‐prepared 3DOM‐SiO₂ template with an aperture of 120 nm in diameter, which matches with the size of the original PS sphere array (Figures S13,S14). Such highly interconnected spatial network of 3DOM‐SiO₂ provides a venue for the confined growth of Ni‐ZIF‐8 crystals (Figure S15). As shown in the SEM images of Ni‐NC OHSpA, numerous uniform carbon nanospheres are assembled to form the 3D ordered close packed face‐centered cubic (fcc) array (Figure 2b). The representative SEM images and the corresponding schematic illustrations of the (111) and (100) orientations reveal that Ni‐NC OHSpA inherits the long‐range and highly ordered array structure of PS template but with a slight contraction of sphere diameter from 120 to 100 nm (Figures 2c and S16,S17). Transmission electron microscopy (TEM) images clearly elucidate the hollow sphere structure of Ni‐NC OHSpA, where each sphere is fully interconnected (Figures 2d and S18). The formation of unique hollow structure in Ni‐NC OHSpA may be attributed to the interface shrinkage effect [31, 32], namely the 3DOM‐SiO₂ template with rigidity nature is able to stretch the internal “soft” ZIF outward during the carbonization process (Figure S19). Notably, during the carbonization process, the pore structure of the material changes from microporous to hierarchically porous, and the external specific surface area increases gradually (Figure S20 and Table S1).

SEM images of Cu‐NP/Ni‐NC OHSpA (Figure S21) clearly reveal that the ordered sphere array structure remains intact after the implantation of Cu nanoparticles. The powder X‐ray diffraction (XRD) pattern of Cu‐NP/Ni‐NC OHSpA exhibits three diffraction peaks at 43.3°, 50.5°, and 74.2°, corresponding to the (111), (200) and (220) planes of metallic Cu (Figure S22). TEM and high‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM) images show that metal nanoparticles with an average diameter of 10 nm are uniformly loaded on the ordered hollow sphere array (Figures 2e,f and S23). EDS mapping images reveal the distinct distributions of Ni and Cu elements over the hollow sphere framework, where Ni element is highly dispersed over the carbon matrix without any agglomeration of metal elements (Figure 2g). The characteristic distributions of Cu elements over Cu‐NP/Ni‐NC OHSpA are also witnessed by the corresponding EDS line scan analysis (Figure S24). The contents of Ni and Cu in Cu‐NP/Ni‐NC OHSpA were determined as ca. 0.78% and 18.27%, respectively (Table S2). To further clarify the spatial distributions of Ni and Cu species, aberration corrected high‐angle annular dark‐field scanning transmission electron microscopy (AC‐HAADF‐STEM) was adopted. For the Ni‐NC OHSpA, Ni atoms are atomically dispersed on the hollow sphere array without obvious Ni clusters or nanoparticles (Figure S25). For the Cu‐NP/Ni‐NC OHSpA, Cu nanoparticles can be clearly observed accompany with numerous isolated Ni atoms (Figures 2h and S26). Specifically, the lattice fringes of the nanoparticles show the spacing of ca. 2.1 Å, corresponding to the (111) plane of metallic Cu (Figure 2i).

Given that the reverse template strategy allows OHSpA to inherit the architecture of the PS sphere array, we attempted to control the size of hollow sphere in Cu‐NP/Ni‐NC OHSpA by tuning the PS sphere sizes (Figure S27 and Table S3). Importantly, a series of Cu‐NP/Ni‐NC OHSpA with the sphere diameters of 80, 120, 140, 160, and 200 nm could be prepared (Figures 2j and S28), highlighting the general applicability of the proposed strategy for the design of hollow sphere with periodic structure.

The atomic structures and chemical properties of the as‐synthesized materials were investigated. X‐ray absorption near‐edge structure (XANES) and extended x‐ray absorption fine structure (EXAFS) were conducted to reveal the coordination environment and electronic structure of Cu‐NP/Ni‐NC OHSpA. In the Ni K‐edge XANES spectra, the near edge absorption energy of Cu‐NP/Ni‐NC OHSpA is located between those of Ni foil and NiO, indicating the oxidation state of Ni species with a valence state between 0 and +2 (Figure 3a). The oxidation state of Ni species in Cu‐NP/Ni‐NC OHSpA is also clarified by the Ni 2p x‐ray photoelectron spectroscopy (XPS) spectrum, which shows a prominent peak assigned to Ni 2p_3/2 at 855.4 eV (Figure S29). The Fourier transform (FT) extended EXAFS spectrum of Cu‐NP/Ni‐NC OHSpA shows a distinct peak at 1.7 Å, which is close to that of the Ni‐N coordination in nickel phthalocyanine (NiPc) (Figure 3b). Additionally, no Ni‐Ni signals at 2.15 Å are observed in the FT extended EXAFS spectrum, suggesting the potential existence of atomically dispersed Ni species in Cu‐NP/Ni‐NC OHSpA rather than Ni clusters or nanoparticles. Wavelet transform (WT) analysis of Ni K‐edge EXAFS exhibits the maximum intensity at 4.10 Å⁻¹ assigned to the Ni‐N interactions, which is analogous to that of NiPc but distinct from Ni foil and NiO (Figures 3g and S30). According to the fitting EXAFS coordination information, the coordination number of Ni‐N is calculated to be ca. 4 (Table S4). For the Cu K‐edge spectra, the oscillations of post‐edge in the XANES spectrum and K‐space of Cu‐NP/Ni‐NC OHSpA share the same trend with that of Cu foil (Figures 3c and S31). In addition, the FT extended Cu K‐edge EXAFS spectrum of Cu‐NP/Ni‐NC OHSpA displays a typical peak at ca. 2.23 Å, which is assigned to the Cu‐Cu scattering in metal particles (Figures 3d and S32). Three‐dimensional contour WT spectra reveal that the maximum value of the Cu K‐edge EXAFS oscillation WT is 6.67 Å^‐1, attributing to the Cu‐Cu coordination (Figure 3g). The metallic state of Cu species in Cu‐NP/Ni‐NC OHSpA is also observed by the Cu 2p XPS spectrum, which exhibits the characteristic peaks of Cu⁰ at 932.9 and 952.6 eV (Figure S33) [33]. Moreover, the Cu LMM Auger spectrum also demonstrates that the Cu species in Cu‐NP/Ni‐NC OHSpA are in the form of metallic Cu particle rather than Cu₂O (Figure 3f) [34]. Based on the above results, a plausible atomic structure of Cu‐NP/Ni‐NC OHSpA is proposed (insert of Figure 3e), in which Ni is embedded in the substrate as a single atom (i.e., Ni‐N₄) and Cu is loaded on the carrier in metal particle form (Figure 3e).

FIGURE 3.

Structural representation. (a) Ni K‐edge XANES spectra and (b) FT k³‐weighted Ni K‐edge EXAFS spectra. (c) Cu K‐edge XANES spectra and (d) FT k³‐weighted Cu K‐edge EXAFS spectra. (e) Fitting curves of FT Cu K‐edge and Ni K‐edge EXAFS of Cu‐NP/Ni‐NC OHSpA (insert: schematic atomic interface model of Cu‐NP/Ni‐NC OHSpA). (f) Cu LMM Auger spectrum. (g) WT k²‐weighted EXAFS contour plots of Cu‐NP/Ni‐NC OHSpA and reference samples.

2.3. Electrocatalytic Reduction of Diluted CO₂

The electrocatalytic performance of Cu‐NP/Ni‐NC OHSpA was evaluated in diluted CO₂ atmosphere (i.e., 15 % CO₂/Ar) using a gas diffusion electrode (GDE)‐based flow cell with 1 M KHCO₃ as the electrolyte. As shown in the linear sweep voltammetry (LSV) curves, Cu‐NP/Ni‐NC OHSpA shows significantly higher current density in 15% CO₂/Ar compared to that in pure Ar atmosphere, suggesting the excellent performance toward the reduction of CO₂ (Figure 4a). The catalytic products were quantitatively analyzed by gas chromatography (GC) and nuclear magnetic resonance hydrogen spectroscopy (NMR) (Figures S34–36). As shown in Figures 4b and S37, C₂H₄, CH₄, CO and H₂ were detected as the products in CO₂ reduction, in which the total Faradaic efficiency (FE) of these products was calculated to be ca. 100%. Significantly, Cu‐NP/Ni‐NC OHSpA exhibits a remarkable selectivity toward C₂H₄ in 15% CO₂/Ar atmosphere, giving an FE_C2H4 of as high as 57.5% at ‐1.1 V versus RHE and a C₂H₄ partial current density (J_C2H4) of 110 mA cm⁻². Notably, both the FE_C2H4 and J_C2H4 of Cu‐NP/Ni‐NC OHSpA exceed those of all other reported catalysts under diluted CO₂ conditions and even outperform that of most state‐of‐the‐art catalysts in pure CO₂ atmosphere (Figure 4c and Table S5).

FIGURE 4.

Electrochemical reduction of diluted CO₂. (a) The LSV curves and (b) product distribution and C₂H₄ partial current density of Cu‐NP/Ni‐NC OHSpA measured in 15% CO₂/Ar atmosphere. (c) Comparison of the Faradaic efficiency and partial current density of C₂H₄ over the previously reported electrocatalysts under diluted CO₂ conditions (values sourced from the literatures where they were reported as such (see Table S6)). (d) Long‐term stability tests at ‐1.1 V versus RHE for 500 h over Cu‐NP/Ni‐NC OHSpA (inset: schematic diagram of a flow electrolytic cell; positions indicated by orange arrows represent those of replacement of electrolyte). (e) LSV curves (inset: SEM images of Cu‐NP/Ni‐NC MiNC, Cu‐NP/Ni‐NC MeNC and Cu‐NP/Ni‐NC OHSpA) and (f) FE_C2H4 measured in 15% CO₂/Ar atmosphere. (g) Electrochemical activity and (h) the corresponding digital image of the PV‐EC reaction system under outdoor sunlight irradiation in Guangzhou city, China (23°04’N, 113°40’E), at 10:00 to 18:00, Jun., 2025. (i) LSV curves of EDCR//BOR two‐electrode system (inset: schematic diagram of the coupled system and the digital image of FDCA product). The error bars represent the standard deviations of at least three independent measurements of the same sample.

Given the superior catalytic performance of Cu‐NP/Ni‐NC OHSpA, the electrocatalytic CO₂ reduction was then conducted in simulated exhaust gas atmosphere (i.e., 15% CO₂ + 1% SO₂, NO_x, O₂). Interestingly, the electro‐catalytic performance of Cu‐NP/Ni‐NC OHSpA can be well maintained after the introduction of O₂, SO_x and NO_x, delivering a current density and FE_C2H4 of 95 mA cm⁻² and 53.0%, respectively (Figure S38). The slight inhibition in current density under simulated exhaust gas conditions is caused by the coexistence of impurities (i.e., SO_x, NO_x, O₂) in the feed gas, which may poison the active sites and introduce competing side reactions (Figure S39). The exceptional performance highlights the application potential of the Cu‐NP/Ni‐NC OHSpA based electrocatalytic system for treating exhaust gas with nitroxides and sulfides.

The stability test of Cu‐NP/Ni‐NC OHSpA was carried out in the flow cell at the optimal potential for FE_C2H4 (i.e., ‐1.1 V). As shown in Figure 4d, no significant reductions in current density and FE_C2H4 are observed after 500 h of continuous electrolysis, suggesting excellent durability of Cu‐NP/Ni‐NC OHSpA. The extraordinary stability duration represents the highest value among all state‐of‐the‐art catalysts in electrolytic CO₂‐to‐C₂H₄ conversion reported to date. In addition, SEM and TEM images confirm that the ordered hollow sphere array structure of Cu‐NP/Ni‐NC OHSpA can be well retained after electrolysis (Figure S40). XRD pattern and XPS spectra reveal that the chemical states of Cu and Ni in Cu‐NP/Ni‐NC OHSpA remain unchanged after electrocatalysis, pointing out its excellent structural stability (Figure S41).

Control experiments demonstrate that OHSpA without metal species exhibits modest CO₂ reduction performance with tiny CO as the sole carbon‐based product (FE_CO 10%) (Figure S42a,b). In contrast, the introduction of Ni species (i.e., Ni‐NC OHSpA) significantly improves the selectivity of CO from 10.3% to 80.8%, but no C₂ products are detected (Figure S42c). For the Cu‐NP OHSpA with Cu species as the active site, C₂H₄ is detected with an FE_C2H4 of 33.8%, which is remarkably lower than that of Cu‐NP/Ni‐NC OHSpA (FE_C2H4 of 57.5%) (Figure S42d). These results unambiguously suggest that Ni and Cu nanoparticles should serve as the active sites for CO evolution and C‐C coupling reaction, respectively, and their synergistic effect significantly promotes the generation of C₂H₄ (Figure S42e,f).

For comparison, Ni single atom and Cu nanoparticles embedded in mesoporous nitrogen‐doped carbon (Cu‐NP/Ni‐NC MeNC) and microporous nitrogen‐doped carbon (Cu‐NP/Ni‐NC MiNC) were synthesized using conventional ZIF‐8 rhombic dodecahedron as the precursor (Figure S43). Both Cu‐NP/Ni‐NC MeNC and Cu‐NP/Ni‐NC MiNC display much lower current densities and FE_C2H4 compared to Cu‐NP/Ni‐NC OHSpA (Figure 4e,f), suggesting the advantages of ordered hollow sphere array structure in accelerating the evolution of C₂ products. To further elucidate the influence of the ordered hollow sphere array structure on the competing hydrogen evolution reaction, we designed and synthesized two comparative materials with solid sphere array structure (i.e., Ni‐NC OSSpA and Cu‐NP/Ni‐NC OSSpA) (Figure S44). The results reveal that the solid sphere catalysts exhibit higher FE_H2 than their hollow counterparts, demonstrating that the ordered hollow sphere array structure can suppress the HER side reaction in electrocatalytic dilute CO₂ reduction (Figures S45,S46).

The structure‐activity relationship between the cavity size of Cu‐NP/Ni‐NC OHSpA and C₂H₄ selectivity was investigated. As shown in Figure S47, the FE of C₂H₄ increases gradually from 32.2% to 57.5% as the cavity size of Cu‐NP/Ni‐NC OHSpA decreases from 200 to 100 nm. This trend can be rationalized by the enhanced nanoconfinement effect as the hollow interior becomes smaller. When the cavity size of Cu‐NP/Ni‐NC OHSpA is further reduced to 80 nm, however, the FE_C2H4 unexpectedly drops instead of continuing to rise. This decline is primarily due to unavoidable partial structural disordering arising from the use of a smaller PS template (Figures S47,S48).

Subsequently, we attempted to integrate a bias‐free photovoltaic‐electrochemical (PV‐EC) system toward the conversion of diluted CO₂ under outdoor condition, in which a commercial crystalline Si solar cell and a voltage stabilizer were applied to supply a photovoltage of ca. 4.0 V. Significantly, the bias‐free PV‐EC system can be operated smoothly under the irradiation of natural sunlight, affording a C₂H₄ yield of 227.8 umol cm⁻² h⁻¹ with a FE_C2H4 of 57.5% (Figure 4g,h).

To enhance product economic value and reduce energy consumption, the biomass oxidation reaction (BOR) was strategically employed as an alternative to replace the oxygen evolution reaction (OER) process in the anode to construct the EDCR//BOR paired system (insert of Figure 4i). In situ Raman spectroscopy and DFT calculations demonstrate that the energy input for BOR on Cu surface (0.56 eV) is significantly lower than that for OER (2.06 eV), pointing out the kinetic feasibility of the EDCR//BOR paired systems (Figures S49,S50). As a proof of concept, Cu‐NP/Ni‐NC OHSpA was applied as a bifunctional catalyst for both CO₂ reduction and biomass oxidation in cathode and anode, respectively. Remarkably, all the EDCR//BOR systems exhibit significantly higher current densities and require lower anode potentials compared to the EDCR//OER system upon the addition of biomass, among which the EDCR//5‐hydroxy methyl furfural oxidation reaction (EDCR//HMFOR) paired system shows the optimal performance (Figures S51,S52). In detail, the electrolytic cell requires a voltage of only 2.40 V for EDCR//HMFOR at a current density of 10 mA cm⁻², which is 650 mV lower than that for EDCR//OER (3.05 V), suggesting that the construction of paired system can decrease the energy consumption (Figure 4i). More importantly, C₂H₄ and value‐added product 2,5‐furandicarboxylic acid (FDCA) are simultaneously obtained at cathode and anode, with an FE of 55.4% and 95.4%, respectively. In addition, XRD, XPS, SEM and TEM characterizations confirm that Cu‐NP/Ni‐NC OHSpA retains its intact ordered hollow spherical structure after the HMFOR reaction, while the Cu species are transformed into the CuO phase. Despite the structural reconstruction of the active sites in Cu‐NP/Ni‐NC OHSpA, the EDCR//HMFOR paired system can still be operated stably for 275 h (Figures S53–56).

2.4. Mechanism Insights

In situ attenuated total reflection‐surface enhanced infrared absorption spectroscopy (ATR‐SEIRAS) was first employed to ascertain the intermediates on Cu‐NP/Ni‐NC OHSpA during the EDCR process. As shown in Figures 5a and S57, several characteristic absorption bands appear upon applying the working potential. Among them, the peaks at 1285, 1390, 1980, and 2080 cm⁻¹ are attributed to the stretching vibrations of *COOH(C = O), *COOH(O‐H), bridge‐bonded *CO (CO_B) and atop‐bonded *CO (CO_At), respectively [35, 36, 37]. These features confirm the occurrence of CO₂‐to‐CO reaction pathway on Cu‐NP/Ni‐NC OHSpA via the formation of *COOH intermediate (*CO₂→*COOH→*CO). Additionally, the characteristic peaks at 1190 and 1157 cm⁻¹ can be assigned to the stretching vibrations of *COCOH and *COH, respectively, which are the key intermediates for C‐C coupling process for the CO₂‐to‐C₂H₄ conversion. With an increase in the working potential, the signals of *CO and *COH gradually decay accompany with enhanced *COCOH signal, suggesting a C‐C coupling process between *CO and *COH to form the *COCOH intermediate.

FIGURE 5.

Mechanistic investigations. (a) In situ ATR‐SEIRAS spectra and (b) In situ Raman spectra of Cu‐NP/Ni‐NC OHSpA at different potentials in 15% CO₂/Ar atmosphere. (c) CO‐DRIFTS and (d) CO‐TPD spectra. (e) CO and (f) N₂ adsorption isotherms. DRT analysis of (g) Cu‐NP/Ni‐NC OHSpA, (h) Cu‐NP/Ni‐NC MeNC, (i) Cu‐NP/Ni‐NC MiNC.

In situ Raman spectroscopy was also employed to further verify the roles of Ni and Cu species on the formation of reaction intermediates on Cu‐NP/Ni‐NC OHSpA. In the in situ Raman spectra of Ni‐NC OHSpA, a distinct stretching vibration peak of *COOH and a weak *CO signal appear when the work potential is applied (Figure S58a,b). This result indicates that CO evolution can be proceeded on Ni sites via *COOH intermediate and the formed CO is easily desorbed from the active sites. Significantly, obvious *CO Raman signals at 356.8, 1983.3, 2051.7 cm^‐1 are observed in the in situ Raman spectra of Cu‐NP/Ni‐NC OHSpA, suggesting that the introduction of Cu can strengthen the adsorption of *CO (Figure 5b) [38, 39]. Moreover, a Raman peak at 1159.2 cm^‐1 can be ascribed to the stretching vibration of *COCOH [40], suggesting that the C‐C coupling process may be occurred on Cu species. Based on the evidence of in situ spectra and the results of control experiments (Figure S58c,d), a rational reduction pathway of CO₂ to C₂H₄ over Cu‐NP/Ni‐NC OHSpA is proposed. Initially, diluted CO₂ is adsorbed and reduced to CO on Ni sites of Cu‐NP/Ni‐NC OHSpA via the formation of *COOH intermediate, and a spillover effect of CO molecules is occurred from Ni sites to Cu surface. Subsequently, the successive protonation and C‐C coupling process for adjacent *CO take place on Cu surface to transfer *CO into the *COCOH (*CO+*CO → *CO+*COH → *CO‐COH), which is further reduced to C₂H₄ products.

Considering that the adsorption and coupling of CO are the key step in C₂H₄ evolution, the adsorption behavior of CO on Cu‐NP/Ni‐NC OHSpA was studied by CO adsorption diffuse reflectance infrared Fourier transform spectroscopy (CO‐DRIFTS). As shown in Figures 5c and S59, an obvious *CO signal appears in the CO‐DRIFTS spectra of Cu‐NP/Ni‐NC OHSpA after purging CO, and this signal is still retained after N₂ purging. In comparison, no detectable *CO signal is observed in the CO‐DRIFTS spectra of Ni‐NC OHSpA, highlighting the significant role of Cu in strengthening the adsorption of CO. Such a strong CO affinity of Cu‐NP/Ni‐NC OHSpA is also confirmed by CO temperature‐programmed desorption (CO‐TPD) experiments (Figure 5d), in which Cu‐NP/Ni‐NC OHSpA exhibits a higher CO desorption temperature (300°C) and a larger desorption peak compared with Ni‐NC OHSpA. CO stripping voltammetry was further employed in CO atmosphere to reveal the CO binding behavior with Cu‐NP/Ni‐NC OHSpA [41, 42]. As shown in Figure S60, the CO stripping curve of Cu‐NP/Ni‐NC OHSpA exhibits a considerable oxidation peak at 0.55 V, which is originated from the electrochemical oxidation of surface‐adsorbed CO. Significantly, Ni‐NC OHSpA displays more negative and weaker CO oxidation peak (0.52 V) compared with Cu‐NP/Ni‐NC OHSpA, suggesting that the introduction of Cu species is able to enhance the chemical binding strength toward CO and thus potentially facilitate the C‐C coupling process.

Given that the CO₂‐to‐C₂H₄ conversion was conducted in diluted CO₂ atmosphere, the physical adsorption process of CO should be equally indispensable to improve the *CO coverage [43]. In this regard, CO adsorption experiments were performed (Figure 5e). Obviously, CO adsorption isotherms show that Cu‐NP/Ni‐NC OHSpA has the highest CO uptake compared to Cu‐NP/Ni MeNC and Cu‐NP/Ni MiNC. Moreover, CO uptake of Cu‐NP/Ni‐NC OHSpA is also higher than that of Cu‐NP/Ni OMNC with ordered macroporous structure. These results demonstrate the virtue of ordered hollow sphere array in enriching CO molecules surrounding the active sites.

The porous structure of Cu‐NP/Ni‐NC OHSpA was further investigated by nitrogen adsorption/experiments. As shown in Figures 5f and S61, the N₂ adsorption isotherm of Cu‐NP/Ni‐NC OHSpA exhibits a distinct hysteresis loop with high adsorption quantity in the high‐pressure region (0.9‐1.0 P/P₀), confirming both mesoporous and macroporous feature [44]. The BET surface area of Cu‐NP/Ni‐NC OHSpA is ca. 682.1 m² g⁻¹, which is remarkably larger than that of Cu‐NP/Ni‐NC MiNC (197.0 m² g^‐1) and Cu‐NP/Ni‐NC MeNC (328.5 m² g^‐1). In addition, the averages size of mesopores in Cu‐NP/Ni‐NC OHSpA (ca. 48 nm) is significantly larger than that of Cu‐NP/Ni‐NC MiNC (ca. 0.68 nm) (inserts of Figure S61), suggesting that the “rigid” SiO₂ shell is able to induce the formation of mesoporous structure.

The double‐layer capacitance (C_dl) was calculated from the cyclic voltammetry (CV) curves in the non‐Faradaic region to estimate the active site density and electrochemical active surface area (ECSA) of Cu‐NP/Ni‐NC OHSpA. As shown in Figure S62, the C_dl of Cu‐NP/Ni‐NC OHSpA is estimated to be 12.22 µF cm^‐2, which is significantly higher than that of Cu‐NP/Ni‐NC OSSpA (7.03 µF cm^‐2), Cu‐NP/Ni‐NC MeNC (2.45 µF cm^‐2) and Cu‐NP/Ni‐NC MiNC (1.85 µF cm^‐2). This result indicates that the ordered hollow sphere array structure of Cu‐NP/Ni‐NC OHSpA should increase the external active sites for electrocatalytic process. Thus, the unique hierarchically porous feature of Cu‐NP/Ni‐NC OHSpA can not only enrich the *CO intermediate via pore confinement effect, but also improve the accessibility of active sites (i.e., Ni‐N₄ and Cu particles) for the formation of *CO and C‐C coupling process.

Considering that the insufficient reactants (i.e., CO₂) and intermediate (i.e., *CO) caused by the low CO₂ concentration, a superior diffusion efficiency is required to meet the reaction demand. In this case, in situ electrochemical impedance spectroscopy (in situ EIS) tests were performed to investigate the mass transport capability of Cu‐NP/Ni‐NC OHSpA during the EDCR process. As shown in Figures S63,S64, Cu‐NP/Ni‐NC OHSpA displays significant response in the high‐frequency region of Bode plot compared with Cu‐NP/Ni‐NC MeNC and Cu‐NP/Ni‐NC MiNC, suggesting that the hierarchically ordered pore array can promote the intralayer electronic transfer [45, 46, 47]. In addition, the low‐frequency region of Cu‐NP/Ni‐NC OHSpA assigned to the reaction response is much smaller than that of Cu‐NP/Ni‐NC MiNC, indicating a faster kinetic for EDCR over Cu‐NP/Ni‐NC OHSpA. We subsequently decoupled the interfacial processes through distributed relaxation time (DRT) analysis to explore the diffusion process. As shown in Figure 5g–i, the quantitative analysis reveals that the charge transfer resistance of Cu‐NP/Ni‐NC OHSpA (integrated area of 483) is significantly lower than Cu‐NP/Ni‐NC MiNC (integrated area of 2520) and Cu‐NP/Ni‐NC MeNC (integrated area of 912), suggesting that the ordered hollow sphere array is able to significantly reduce the mass transport and diffusion resistance during the EDCR process.

The interface reaction behavior on Cu‐NP/Ni‐NC OHSpA was investigated by DFT calculations. To reveal the intrinsic role of Ni single atom and Cu particle in CO₂‐to‐C₂H₄ conversion, two theoretical models (i.e., NiN₄C and Cu (111)) were constructed based on the characterization results (Figure S65). For the *CO formation process, the protonation of *CO₂ to *COOH on both NiN₄C and Cu (111) is highly endothermic, suggesting that the activation of CO₂ should be a critical step in *CO formation (Figure S66). Climbing image nudged elastic band (CI‐NEB) method was then utilized to reveal the CO₂ activation process. As shown in Figure 6a and S67, the adsorbed CO₂ on NiN₄C is activated and protonated into *COOH overcoming an energy barrier of 0.4 eV with the formation of a transition state (TS). In contrast, the energy barrier for CO₂ activation on Cu (111) is as high as 1.01 eV, suggesting that the CO₂‐to‐CO process tends to be occurred on Ni sites rather than the Cu surface.

FIGURE 6.

Results of DFT calculation and proposed mechanism. Free energy diagrams for the conversion of (a) CO₂‐to‐CO and (b) *CO‐to‐C₂H₄ over Cu (111) and NiN₄C. (c) The adsorption energies of *CO and *COCOH on Cu (111) and NiN₄C. (d) Variable adsorption configurations of *CO on the Cu/NiN₄C surface (snapshot images from AIMD simulation of 5 ps at 298.15 K). (e) Free energy diagrams for the formation of *COCOH on Cu (111) with different CO coverages. (f) The charge density difference of models with different CO coverages. (g) The pCOHP of C‐C bonds in COCOH/Cu (top) and 2CO‐COCOH/Cu (bottom). The bonding and antibonding states are shown on the right and the left of the vertical zero line, respectively. The horizontal line indicates the Fermi level. (h) Proposed reaction mechanism for EDCR over Cu‐NP/Ni‐NC OHSpA.

Regarding the subsequent *CO reduction process on Cu (111), the C‐C coupling of *CO is an endothermic process and has been regarded as the rate‐determining step (RDS), while the succedent reduction processes is exothermic. In addition, the ethylene pathway exhibits the most thermodynamically favorable profile than ethanol and acetic acid pathways (Figures S68,S69). Notably, the energy input for the C‐C coupling process on Cu (111) is 0.68 eV, which is remarkably lower than that on NiN₄C (1.44 eV). Specifically, *CO is protonated and coupled to *COCOH on Cu (111), overcoming an energy barrier of 1.17 eV via TS‐1 (Figures 6b and S70,S71). On the contrary, the energy barrier for the formation of *COCOH on NiN₄C is calculated to be as high as 1.87 eV. These results elucidate the inherent property of the Cu (111) as an efficient active center for *CO dimerization.

Notably, the CO adsorption/desorption processes can be smoothly proceeded on NiN₄C and Cu (111), respectively due to their low energy requirements (Figure 6c). The inherent CO sorption behavior on Cu‐NP/Ni‐NC OHSpA should enable the spillover of CO from NiN₄C to Cu (111). Ab initio molecular dynamics (AIMD) simulations were thus performed to verify the CO spillover behavior on Cu‐NP/Ni‐NC OHSpA, in which the variable adsorption configurations of CO molecule are analyzed via the selected snapshot images (Figure 6d, Movie 1). Interestingly, the *CO can be readily desorbed from the NiN₄C site after ca. 100 fs of AIMD simulation, and the released CO molecule is captured by the adjacent Cu particle during the subsequent AIMD process. The CO spillover process is further evidenced by the dynamic evolution of C‐metal bonds, showing lengthened Ni‐C bond from 1.88 to 4.56 Å accompanied with shortened Cu‐C bond from 4.90 to 2.06 Å (Figure S72).

Generally, the CO spillover effect from NiN₄C to Cu (111) should increase the *CO coverage on Cu (111) surface, which may facilitate the subsequent C‐C coupling process to form the *COCOH intermediate, namely the RDS [48]. Significantly, the results of DFT calculation demonstrate that the energy input for the C‐C coupling process decreases dramatically from 0.68 to 0.34 eV when increasing the number of *CO on Cu (111) from 2 to 4, pointing out the indispensable role of CO coverage in promoting the C‐C coupling process (Figure 6e). In addition, the distance between two adjacent *CO is shrunken from 4.10 to 3.40 Å after the introduction of surrounding CO*, suggesting that increasing *CO coverage is able to enhance the spatial proximity of adsorbed species (Figure S73). We subsequently investigated the effect of *CO spillover on the C‐C coupling from an electronic structure perspective. Charge density difference (CDD) analysis shows that the introduction of adjacent CO* significantly increases the electron transformation from Cu (111) to CO*, thus stabilizing the *CO intermediate and potentially facilitating their dimerization (Figures 6f and S74,S75). Furthermore, we performed crystal orbital Hamilton population (COHP) calculations to elucidate the effect of *CO coverage on the orbital interactions within *COCOH (Figures 6g and S76). Clearly, the COCOH/Cu (111) exhibits noticeable bonding states in C‐C via the orbital interactions of 2px‐2s. Additionally, the relatively negative integrated COHP (ICOHP) value for 2CO‐COCOH/Cu (111) with two *CO surrounding the *COCOH (‐12.87 eV) is more negative than that of COCOH/Cu (111) without adjacent *CO (‐11.19 eV), suggesting that an increase in *CO coverage on Cu (111) can strengthen the orbital interaction between two *CO and promote the C‐C coupling process.

We further evaluated the CO reduction performance of Cu‐NP/Ni‐NC OHSpA under different CO concentrations. As shown in Figure S77, both the total current density and FE of C₂H₄ increase with the rise of CO concentrations in the feed gas, indicating that the *CO coverage should play a critical role in promoting C₂H₄ formation.

Considering that the enhanced C₂H₄ performance may arise from electronic structure modification of Cu nanoparticles, we thus examined the electronic interaction between Ni and Cu sites. As shown in Figures S78,S79, an electronic effect does exist between Ni single atoms and Cu nanoparticles in Cu‐NP/Ni‐NC OHSpA. To assess whether this electronic modification is the primary cause of the high C₂H₄ selectivity, we also designed a control experiment using a physically mixed catalyst composed of separately synthesized Ni‐NC OHSpA and Cu‐NP OHSpA. Interestingly, this mixture delivered a C₂H₄ FE of 50.4% at ‐1.1 V versus RHE in 15% CO₂/Ar, far exceeding that of Ni‐free Cu‐NP OHSpA (33.8%) and approaching that of Cu‐NP/Ni‐NC OHSpA (57.5%). These results strongly indicate that tandem catalysis is the primary contributor to the enhanced C₂H₄ selectivity (Figure S80).

Collectively, a plausible electrocatalytic reaction mechanism of diluted CO₂‐to‐C₂H₄ over Cu‐NP/Ni‐NC OHSpA is proposed (Figure 6h). First, diluted CO₂ is converted to CO on single atomic Ni sites by a proton‐assisted electron transfer process. Then, the CO species undergo a spillover process from Ni sites to the neighboring Cu particle, followed by a C‐C coupling process and deep reduction to generate C₂H₄. Remarkably, the unique ordered hollow sphere array structure plays important roles in efficient conversion of diluted CO₂ to C₂H₄ by: (i) improving the *CO coverage to facilitate the C‐C coupling process; and (ii) promoting the diffusion kinetics of both reactant (i.e., diluted CO₂) and product (i.e., C₂H₄).

3. Conclusion

In summary, this work presents a novel reverse templating strategy for the construction of ordered hollow sphere array nanoreactors for electrocatalytic reduction of diluted CO₂. The unique architecture serves as a highly efficient reservoir for both CO₂ and *CO intermediates, thereby significantly enhancing the C₂H₄ selectivity in the reduction of diluted CO₂. In situ spectroscopy and density functional theory calculations indicate that the tandem catalytically active centers (i.e., single atom Ni and Cu particle) accelerate the formation kinetics of CO and enable the spillover of *CO intermediates. Additionally, the ordered hollow structure restricts the outflow of locally generated CO species, thereby elevating the surface coverage of *CO intermediates required for C‐C coupling to produce C₂H₄. Beyond developing a high‐Faradaic‐efficiency catalyst for the upgrading of diluted CO₂ to C₂₊ products, this work highlights a design principle of ordered porous materials for advanced catalysis application.

Author Contributions

Datong Chen: methodology, formal analysis, data curation, investigation, writing – original draft, visualization. Fengliang Wang: conceptualization, funding acquisition, investigation, visualization, project administration, formal analysis, supervision, resources, writing – review and editing, writing – original draft. Wenyuan Lyu: visualization, formal analysis, investigation. Liyu Chen: formal analysis, funding acquisition. Kui Shen: formal analysis, funding acquisition. Yingwei Li: supervision, writing – review and editing, funding acquisition, project administration, formal analysis.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: anie73260‐sup‐0001‐SuppMat.docx.

Supporting File 2: anie73260‐sup‐0002‐MovieS1.mp4.

Data Availability Statement

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