Orientation‐Entropy‐Mediated Derivative Cu Sites for Selective CO₂ Electroreduction to Ethylene

ABSTRACT

Electrocatalytic conversion of CO₂ to ethylene (C₂H₄) provides a sustainable route for decarbonized chemical manufacturing. Cu‐based catalysts are uniquely capable of driving the C−C coupling essential for C₂H₄ formation, yet their practical implementation is limited by uncontrollable dynamic reconstruction that deteriorates both selectivity and stability. Here, we develop an orientation‐entropy‐mediated regulation strategy to programmatically direct the structural evolution of Cu₂O toward highly active sites for C₂H₄ electrosynthesis. Remarkably, the medium‐entropy Cu₂O catalyst achieves a high C₂H₄ Faradaic efficiency of ∼75% at an industrial‐level current density of 400 mA cm⁻² and operates stably for over 80 h. Operando spectroscopy combined with experimental analysis reveals that entropy‐regulated Cu sites establish an optimized kinetic balance between C−C coupling and competing *CO hydrogenation, thereby promoting the energetically preferred formation of the key *CO−*COH intermediate. Preliminary techno‐economic analysis projects an additional profit of $566 per ton of C₂H₄, and life‐cycle assessment demonstrates an overall ∼42% reduction in environmental impact compared to conventional production routes.

The data‐driven Cu₂O catalyst enables programmable structural reconstruction by an orientation‐entropy‐mediated regulation strategy, controlling the configurational degrees of freedom governing Cu evolution, producing derivative Cu sites whose reaction pathways can be tuned precisely.

1. Introduction

Ethylene (C₂H₄), the simplest olefin and a cornerstone platform chemical, underpins the manufacture of polymers, fibers, and numerous value‐added chemicals [1, 2, 3, 4, 5]. At present, nearly all industrial C₂H₄ is produced through high‐temperature steam cracking of fossil hydrocarbons (> 750°C), an energy‐intensive route that emits over 360 million metric tons of CO₂ annually [6, 7, 8]. By contrast, the electrochemical CO₂ reduction reaction (CO₂RR), driven by renewable electricity, offers a sustainable pathway to green C₂H₄ by CO₂ conversion rather than emission [9, 10, 11, 12, 13]. Nevertheless, selectively directing CO₂RR toward C₂H₄ remains challenging because the reaction pathway involves complex proton‐coupled electron‐transfer steps, while numerous competing reactions divert intermediates to C₁ products or hydrogen [14, 15]. Cu‐based catalysts represent the most promising candidates for C₂H₄ electrosynthesis owing to their moderate CO adsorption energy [16, 17, 18, 19], yet their inherently low controllability of dynamic structural reconstruction toward metallic Cu species under reductive potentials severely limits both C₂H₄ selectivity and long‐term stability.

Orientation‐entropy engineering, achieved by modulating surface configurational heterogeneity [20, 21], has recently emerged as a promising strategy to regulate electrocatalytic pathways by influencing intermediate adsorption and surface structural evolution. Here, orientation entropy is used to describe the statistical distribution of exposed crystallographic facets, reflecting the degree of surface configurational heterogeneity. By precisely tuning the orientation and entropy of active sites, this approach provides fine control over the dynamic reconstruction behavior of Cu‐based pre‐catalysts, thus facilitating the formation of catalytically competent Cu derivative species [22, 23]. Beyond activity enhancement, orientation‐entropy modulation reshapes the local coordination environment and electronic distribution of surface atoms, which steers critical intermediates toward favorable C−C coupling while suppressing side reactions [24, 25]. Additionally, this strategy enhances structural stability by constraining entropically driven atomic migrations, reducing the uncontrolled reconstruction and thus promoting long‐term operational stability [26, 27, 28]. Despite these advantages, the implementation of orientation‐entropy regulation engineering in electrochemical CO₂RR remains virtually unexplored.

Herein, we introduce a conceptually new descriptor of crystallographic orientation entropy that establishes a quantitative relationship between the intrinsic entropic characteristics of Cu‐based pre‐catalysts and the C₂H₄ selectivity of the derived Cu phase. Guided by machine‐learning screening of a self‐built database integrating representative Cu‐based catalysts, we identify an optimal pre‐catalyst prototype and develop a controllable synthesis strategy to construct data‐driven Cu₂O models with well‐defined low‐, medium‐, and high‐entropy states. Experimental studies demonstrate that the medium‐entropy Cu₂O generates a near‐octa‐coordinated Cu^0.4+ species under −1.5 V_RHE, achieving a maximum FE_C2H4 of ∼75% at 400 mA cm⁻² and stable operation for over 80 h. Operando spectroscopic analyses reveal that these entropy‐derived Cu sites establish an optimized kinetic balance between C−C coupling and hydrogenation pathways, which governs the efficiency and selectivity of C₂H₄ formation. Furthermore, preliminary techno‐economic analysis (TEA) indicates an additional revenue of $566 per ton relative to conventional production routes, while life‐cycle assessment (LCA) demonstrates an overall ∼42% reduction in environmental impact. This study not only establishes a generalizable machine‐learning framework for identifying ideal pre‐catalyst models for selective C₂H₄ electrocatalysts but also provides fundamental insights into the entropy‐governed reconstruction mechanism of Cu‐based CO₂RR catalysts.

2. Results and Discussion

The rational design of high‐performance catalysts remains a major challenge due to the time cost and experience dependence of conventional experimental screening. Machine learning (ML) and artificial intelligence (AI) offer a data‐driven paradigm to efficiently explore vast compositional and structural spaces, predict activity and selectivity, and guide catalyst optimization [12, 29, 30, 31]. By enabling accelerated iterative design cycles and reducing trial‐and‐error, these approaches pave the way for discovering catalysts with enhanced performance and sustainability [32, 33, 34]. To accelerate the discovery of highly selective Cu‐based catalysts for CO₂RR to C₂H₄, we implemented a machine learning‐assisted screening workflow, as illustrated in Figure 1a. First, research articles related to Cu‐based catalysts for CO₂RR over the past decade were collected from Web of Science. The textual content involving catalyst structures, reaction conditions, and performance was then extracted and processed using the large language model (GPT‐3.5 Turbo from OpenAI) with customized prompts, ultimately generating a database of 495 valid entries (see details in Methods). Spearman correlation analysis was subsequently used to examine the relationships between catalyst structural descriptors and performance metrics. This data‐driven screening approach guided the selection of some representative Cu‐based catalysts, such as Cu, Cu single atom (SA), Cu₂O, CuO, and CuO_x, for subsequent screening and experimental validation. Spearman correlation coefficient, denoted as ρ (rho), is a non‐parametric statistical measure used to assess the strength and direction of a monotonic relationship between two variables [35]. The +N indicates a positive monotonic relationship, meaning that the FE_C2H4 in CO₂RR is more favorable. Conversely, −N indicates a negative monotonic relationship. As shown in Figure 1b, among the surveyed copper‐based catalyst categories, Cu₂O exhibited the highest a Spearman correlation coefficient with FE_C2H4, highlighting its clear advantage for C₂H₄ production. We further explored a series of modification strategies, including doping, oxidation, carburization, and alloying, to optimize catalyst performance (Figure 1c). Although these strategies provided varying degrees of improvement for most materials, pristine Cu₂O consistently maintained the best performance with Spearman correlation coefficient of 0.107, indicating that its pure state is inherently optimal for CO₂RR to C₂H₄ production. Therefore, in the subsequent work, we deliberately synthesized Cu₂O as a model catalyst and investigated the effect of crystal‐plane orientation entropy on its catalytic activity.

FIGURE 1.

(a) Schematic overview of CO₂RR model catalyst database establishment. (b) Spearman correlation analysis between FE_C2H4 and different Cu‐based catalysts (e.g., Cu, Cu SA, Cu₂O, CuO, and CuO_x). (c) Spearman correlation coefficient between FE_C2H4 and Cu‐based catalysts combined with different modification strategies (oxide: coupling with other oxides to form a composite). The positive Spearman coefficient (rho) indicates that FE_C2H4 increases with the variable, whereas a negative coefficient reflects a negative relationship.

The Cu₂O samples were synthesized through a simple reduction process (Figure 2a). Specifically, copper acetate and PVP were dissolved in water and heated under stirring. NaOH and ascorbic acid solutions were then sequentially added, followed by reaction under controlled temperature, where Cu₂O with different crystal orientation entropies was obtained at different reaction temperatures (T1‐T3). Figure 2b presents the x‐ray diffraction (XRD) pattern for the Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE samples. The diffraction peaks located at 29.5°, 36.4°, 42.2°, and 61.3° can be indexed to (110), (111), (200), and (220) planes of Cu₂O phase (JCPDS No. 26‐4853), respectively, confirming the successful synthesis of Cu₂O phase. As the orientation entropy increases, the diffraction features gradually broaden and decrease in intensity, indicating reduced long‐range crystallographic coherence while preserving the local Cu−O coordination environment. This suggests a structural transition from a highly oriented polycrystalline framework toward a more orientation‐disordered nanocrystalline network [36]. The morphological features were investigated by field‐emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM). As shown in Figures S1–S6, the Cu₂O‐LOE shows a well‐defined tetrakaidecahedron structure with uniform size, whereas both Cu₂O‐MOE and Cu₂O‐HOE consist of irregularly shaped nanoparticles. Such particle morphology may be attributed to enhanced orientation entropy arising from their intrinsically polycrystalline nature. A high‐resolution TEM (HRTEM) image taken from the target Cu₂O‐MOE nanoparticles reveals well‐resolved lattice fringes with interplanar distances of 2.98, 2.43, 2.12, and 1.49 Å, corresponding to the (110), (111), (200), and (220) planes of Cu₂O (Figure 2c,d). The corresponding fast Fourier transformation (FFT) pattern confirms the coexistence of multiple crystal orientations, consistent with a polycrystalline structure. Moreover, the selected area electron diffraction (SAED) pattern exhibits concentric diffraction rings indexed to these planes, where each diffraction spot represents an exposed crystal facet and its size correlates with the facet area. Among the three samples, Cu₂O‐HOE displays the densest distribution of crystal spots within each ring, indicating the highest degree of disorder in its crystal plane orientation (Figure S7). For the Cu₂O‐LOE with the lowest orientation entropy, it presents only a few sharp diffraction spots corresponding to two dominant crystal planes (Figure S8). In contrast, the target Cu₂O‐MOE exhibits an intermediate spot density, reflecting a moderate level of orientation disorder and the medium‐entropy nature (Figure 2e). The energy dispersion spectrum (EDS) and element mappings in Figure 2f reveal the homogeneous dispersion of Co and O across the nanoparticles, with the component ratio of Co:O = 2:1, consistent with the stoichiometry of Cu₂O.

FIGURE 2.

(a) Schematic showing the controllable synthetic process of Cu₂O with different crystallographic orientation entropies. (b) XRD patterns of Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE. (c) TEM image of Cu₂O‐MOE. (d) HRTEM image and (e) SAED pattern of Cu₂O‐MOE. (f) EDS and element mappings of Cu₂O‐MOE. (g) Cu K‐edge XANES spectrum and (h) FT‐EXAFS spectrum of Cu₂O‐LOE, Cu₂O‐MOE, Cu₂O‐HOE, and standard samples.

X‐ray photoelectron spectroscopy (XPS) was employed to probe the surface chemical states of the Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE samples. The survey spectra in Figure S9 clearly confirm the coexistence of Cu and O elements in these three samples. In the high‐resolution Cu 2p spectra (Figure S10), the peaks at 933.1 and 953.0 eV can be assigned to the Cu 2p_3/2 and Cu 2p_1/2 of the Cu⁺, whereas the peaks centered at 935.6 and 955.5 eV correspond to Cu²⁺ species. Furthermore, the Cu Auger LMM spectra indicate that Cu predominantly existed in the Cu⁺ state across all samples (Figure S11). For the O1s spectra (Figure S12), two distinct peaks are observed at 530.2 and 531.9 eV, which are assigned to lattice oxygen (O_L) and defective oxygen (O_V) species, respectively [37]. To further elucidate the local electronic and coordination structure, x‐ray absorption spectroscopy (XAS) analyses were conducted. The Cu K‐edge x‐ray absorption near‐edge spectroscopy (XANES) provides direct evidence for subtle variations in the Cu valence states across the Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE samples (Figure 2g). The average valence of Cu at each sample was determined by linear combination fitting of the absorption edge, yielding +1.13 for Cu₂O‐LOE, +1.02 for Cu₂O‐MOE, and +0.95 for Cu₂O‐HOE. The slightly reduced valence may be attributed to the slight oxygen vacancies caused by entropy increase. Fourier transformed extended x‐ray absorption fine structure spectra (FT‐EXAFS) spectra of these three samples exhibit the prominent Cu−O and Cu−Cu coordination at 1.47 and 2.71 Å, respectively (Figure 2h). Wavelet transform (WT) analysis (Figure S13) further corroborates the coexistence of these two coordination shells. Notably, the Cu−O peak shows a slight downward shift with increasing entropy, indicating subtle variations in the local coordination environment. Quantitative EXAFS fitting (Figure S14 and Table S1) reveals that the average coordination number of Cu atoms in Cu₂O‐MOE was about 1.8, which is between those of Cu₂O‐LOE (1.9) and Cu₂O‐HOE (1.7), consistent with the presence of oxygen vacancies. Collectively, these spectroscopic results demonstrate that entropy‐mediated control of crystal orientation allows precise tuning of the Cu₂O valence state and coordination environment, which in turn governs catalyst reconstruction under reductive potential and promotes the formation of highly active Cu sites, and is expected to show considerable catalytic CO₂RR performance.

To assess the advantages of Cu₂O nanoparticles with different orientation entropies toward CO₂RR activity, electrochemical measurements were carried out in a standard flow‐cell electrochemical setup using 1 M KOH as the electrolyte. The gas and liquid products were analyzed by gas chromatography (GC) and nuclear magnetic resonance (NMR) spectroscopy, respectively. Linear sweep voltammetry (LSV) was first performed under CO₂ and Ar atmospheres (Figure S15). The Cu₂O‐MOE catalyst exhibits a markedly lower onset potential and higher current density under CO₂ compared to Ar, confirming its preferential activity for CO₂RR over the competing HER [38]. Subsequently, the Cu₂O‐MOE catalysts and the control Cu₂O‐LOE and Cu₂O‐HOE were evaluated at various potentials to investigate the FE and current density for each individual product, particularly toward C₂H₄ formation. Meanwhile, several key parameters in the Cu₂O synthesis, including the amount of PVP, the concentration of NaOH, and the concentration of ascorbic acid, have been investigated to achieve the best performance (Figures S16–S18). As indicated in Figure 3a, Figure S19, and Tables S2–S4, all three catalysts exhibit the volcano‐shaped trend of the FE for C₂H₄ (FE_C2H4) with an applied potential from 1.2 to 1.7 V_RHE. As the operating potential keeps decreasing, Cu₂O‐MOE reaches a maximum of ∼75% at −1.5 V_RHE and then decreases to ∼60% at −2.3 V_RHE with the other by‐products gradually rising. Remarkably, FE_H2 remains below 15% throughout the entire potential window, indicating that the competitive HER has been significantly suppressed [39]. On the contrary, the Cu₂O‐LOE and Cu₂O‐HOE catalysts exhibit lower C₂H₄ selectivity and more severe HER, reaching maximum FE_C2H4 values of ∼41% at −1.6 V_RHE for the former and ∼43% at −1.3 V_RHE for the latter, demonstrating relatively poor C−C coupling ability with limited active sites.

FIGURE 3.

(a) FEs of all products on Cu₂O‐MOE under different potentials. (b) C₂H₄ partial current densities under different potentials of Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE. (c) Tafel slope of Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE. (d) Comparison of CO₂RR performance for Cu₂O‐MOE and previously reported electrocatalysts. (e) Electrochemical stability of Cu₂O‐MOE. Error bars: Data are presented as mean ± standard deviation (SD) from three independent experiments (n = 3).

In addition to the product selectivity, the efficiency of C₂H₄ production was further evaluated by calculating the partial current densities for C₂H₄ product formation (j_C2H4) of these catalysts at various applied potentials (Figure 3b). Across the entire potential range, Cu₂O‐MOE consistently delivers the highest j_C2H4 among these three catalysts. Notably, it achieves a remarkable j_C2H4 of 304 mA cm⁻² at −1.5 V_RHE, demonstrating strong potential for industrially relevant CO₂RR operation. For comparison, the Cu₂O‐LOE and Cu₂O‐HOE catalysts exhibit significantly lower j_C2H4, further corroborating the superior CO₂RR activity for C₂H₄ production on Cu₂O with medium surface entropy. In addition, the charge transfer rate and electrochemical active area of catalysts were assessed by electrochemical impedance spectroscopy (EIS) and double‐layer capacitance (C _dl) values (Figure S20 and Figure S21a–d). As expected, Cu₂O‐MOE consistently exhibits the lowest charge‐transfer resistance (R _ct = 7.6 Ω) and highest C _dl values (C _dl = 3.97 mF cm⁻²) among all Cu₂O catalysts, reflecting its enhanced charge transport and abundant active sites [40]. In addition, the intrinsic activity of Cu₂O‐MOE catalysts was assessed by normalizing the electrode performance to both BET surface area and electrochemical surface area (ECSA) as well. As expected, Cu₂O‐MOE consistently exhibits the highest specific activity among all Cu₂O catalysts, regardless of whether j_C2H4 is normalized to BET surface area or ECSA (Figure S21e,f). To gain insight into the reaction kinetics, the Tafel analysis was conducted in Figure 3c. Tafel values of C₂H₄ product display 93, 119, and 208 mV dec⁻¹ for Cu₂O‐MOE, Cu₂O‐LOE, and Cu₂O‐HOE, respectively. The smallest Tafel slope of Cu₂O‐MOE confirms that the medium entropy effect endows faster electron transfer between the carbon intermediate and the catalytic sites, thus promoting the electrocatalytic CO₂RR. Moreover, Cu₂O‐MOE exhibits superior catalytic performance compared with most representative Cu‐based catalysts, including simple/complex oxide‐based catalysts and other Cu‐based catalysts, underscoring its great potential as a highly efficient Cu‐based CO₂RR electrocatalyst for C₂H₄ production (Figure 3d and Table S5). Beyond high activity, long‐term durability is another critical parameter for evaluating the practical applicability of electrocatalysts. In this regard, the Cu₂O‐MOE‐based flow cell delivers a total current density of approximately 300 mA cm⁻² at −1.3 V_RHE and maintains stable electrolysis with ∼50% FE_C2H4 after a 20 h period (Figure 3e). In order to effectively alleviate the adverse effects of salt precipitation on long‐term stability, intermittent operation at open‐circuit potential (OCP) for 10 min was adopted after every 5 h of electrolysis, ultimately extending stable electrolysis beyond 80 h. Salt precipitation has long been recognized as a critical bottleneck limiting the development of alkaline CO₂ electroreduction systems. The cause of salt precipitation is the recrystallization of alkali‐metal ions caused by local alkalinity, forming K₂CO₃ and KHCO₃. Therefore, we studied the relationship between pH and salt precipitation (Figure S22). The results indicate that relatively low KOH concentration can partially suppress salt precipitation in alkaline CO₂ electroreduction systems by reducing the formation rate of carbonate and bicarbonate species. However, this suppression occurs at the expense of C₂H₄ selectivity, which is likely associated with the diminished ability to sustain a locally high‐pH microenvironment, thereby disfavoring C−C coupling pathways while enhancing competition from the hydrogen evolution reaction.

In situ electrochemical surface‐enhanced Raman spectroscopy (SERS) was carried out to study the intermediate species. As shown in Figure 4a and Figure S23, the characteristic peaks at approximately 2050 cm⁻¹ of these three catalysts were assigned to the C≡O stretching vibration of *CO intermediates. As the potential decreased from −1.2 to −1.4 V_RHE, the *CO signal on Cu₂O‐MOE intensifies rapidly and then stabilizes at a markedly higher level than those of Cu₂O‐LOE and Cu₂O‐HOE, reflecting a higher *CO coverage on Cu₂O‐MOE surface that can promote the C−C coupling for C₂H₄ production [37]. Additional evidence from the in situ attenuated total reflection Fourier transform infrared spectroscopy (ATR‐FTIR) highlights the benefits of Cu₂O‐MOE in terms of intermediate stabilizing and C−C coupling, including more pronounced *CO, *COH, and *COCOH peaks (Figure 4b, Figure S24, and Table S6). Specifically, the characteristic peak at approximately 1832 cm⁻¹ corresponded to the bridge‐site adsorption configuration of *CO, which enables higher local CO surface coverage and shorter intermolecular distances, thereby facilitating the formation of key C−C coupling intermediates (e.g., *OCCO) and their subsequent hydrogenation toward ethylene formation [41, 42]. Notably, no signals corresponding to *CO_atop were detected, which may be attributed to the rapid conversion into more stable bridge‐bound CO or subsequent reaction intermediates [43]. The *CO coverage on Cu₂O‐MOE was higher than that on Cu₂O‐LOE and Cu₂O‐HOE, indicating more effective activation of adsorbed CO₂. Meanwhile, a peak at 1112 cm⁻¹ was assigned to the *COH species [37]. The higher *COH signals observed on the Cu₂O‐MOE indicated that the hydrogenation reaction of *CO was more favorable compared with Cu₂O‐LOE and Cu₂O‐HOE. Moreover, the characteristic peak located at 1558 cm⁻¹ was attributed to the *COCOH species, which served as the pre‐intermediate in the C−C coupling pathway toward C₂₊ products [44]. To further elucidate the difference in C−C coupling behavior among the Cu₂O‐MOE and control catalysts, we compared the relative adsorption intensity of the *COCOH intermediate. As shown in Figure 4c, the *COCOH‐related band on Cu₂O‐MOE exhibits a markedly higher intensity and a steeper potential‐dependent growth across the entire potential window relative to Cu₂O‐LOE and Cu₂O‐HOE, which is consistent with the trend of enhanced C₂H₄ selectivity shown in Figure 3a [45]. Notably, the *COCOH‐related band intensity of Cu₂O‐MOE begins to exhibit a clear advantage beyond −1.4 V and becomes increasingly pronounced with further increases in the applied potential. This trend suggests a higher propensity for C−C coupling on the Cu₂O‐MOE surface, indicating a strong potential for enhanced C₂H₄ product formation. The selective conversion of CO₂ to C₂H₄ requires not only efficient C−C coupling but also a well‐matched hydrogenation process. The relatively stronger *COOH peaks on Cu₂O‐MOE further support the more favorable hydrogenation reaction [46, 47].

FIGURE 4.

(a) The intensity of adsorbed *CO according to in situ SERS spectra. (b) In situ SEIRAS spectra of Cu₂O‐MOE. (c) The intensity of adsorbed *COCOH on Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE under different potentials. (d) In situ XANES spectra of Cu₂O‐MOE at the Cu K‐edge under different potentials. The inset is the magnified region of the spectra between 8978 and 8980 eV. (e) Valence states of Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE at different potentials. (f) Wavelet transforms EXAFS plots of Cu 2p for Cu₂O‐MOE. (g) In situ Cu EXAFS spectra for Cu₂O‐MOE under different potentials. (h) Coordination numbers of Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE at different potentials.

According to well‐established knowledge, the Cu₂O undergoes in situ reconstruction to metallic Cu upon exposure to sufficiently cathodic potentials during electrochemical CO₂RR. Then, the generated metallic Cu serves as the authentic active species for CO₂RR, where the unique electronic and coordination structures generated through reconstruction predominantly determine the intrinsic activity of catalysts. This mechanistic insight offers a rational strategy for designing highly selective catalysts by precisely manipulating the reconstruction process. To probe the evolution of the atomic and electronic structures in the reconstruction process of the catalyst, in situ XAS experiments were employed, providing a powerful tool to probe the dynamic process of Cu species during the successive CO₂RR process. The XANES spectra at the Cu K‐edge with potential decreasing from −1.2 to −1.7_RHE are depicted (Figure 4d and Figure S25). The absorption edge energy of Cu₂O‐HOE exhibits a significantly more negative shift than those of Cu₂O‐MOE and Cu₂O‐LOE at the same potential, which shows a trend of Cu₂O‐HOE < Cu₂O‐MOE < Cu₂O‐LOE, illustrating that high orientation entropy promotes deep reconstruction of Cu₂O to metallic Cu. Regarding this viewpoint, we tracked the changes in the composition and morphology of three electrocatalysts during the reaction process, further supporting the above conclusion (Figure S26). Furthermore, the average valence of Cu at each condition was estimated by regressing the shifts in absorption edge energy of Cu K‐edge through linear combination fitting of standard Cu foil (Cu⁰) and CuO (Cu²⁺) XANES spectra (Figure 4e and Figure S27). We focused on the potential window of −1.4 to −1.6 V_RHE, where all three catalysts exhibited their highest FE toward C₂H₄ production. The average oxidation states of Cu within this range were calculated to be 0.532, 0.407, and 0.293 for Cu₂O‐LOE, Cu₂O‐MOE, and Cu₂O‐HOE, respectively. These results reveal that the Cu^0.4+ derivative could optimally balance C−C and hydrogenation kinetics, thus exhibiting the highest intrinsic activity for C₂H₄ production. Fourier transforms (FT) and wavelet‐transforms (WT) of k ³‐weighted EXAFS spectra also revealed a gradual transformation, in which the Cu−O signal disappears and the Cu−Cu shell intensity increases with the applied potential (Figure 4f and Figure S28). This corresponds to the formation of metallic Cu phase and the changes in coordination environment. Therefore, the changes in local interatomic distance and coordination number (CN) were deducted by EXAFS curve‐fitting analysis. As shown in Figure 4g and Figures S29–S32, a remarkable CN difference in the Cu−Cu shell is observed among the three catalysts under the same potential. Interestingly, under the potential window (from −1.4 to −1.6 V_RHE) corresponding to their respective maximum FE_C2H4, all three catalysts exhibit Cu−Cu coordination numbers close to 8 (Figure 4h and Table S7). Overall, the medium‐entropy‐mediated Cu₂O undergoes reconstruction at −1.5 V to form metallic Cu catalysts enriched in Cu^0.4+ species with approximately 8‐fold coordination numbers, which exhibit the highest selectivity and stability toward C₂H₄ formation.

A preliminary TEA was conducted to assess the economic viability of CO₂RR system by using Cu₂O‐MOE model for C₂H₄ production in an alkaline electrolyte [48, 49, 50]. In this context, the costs include electricity, electrolyzer, catalyst and membrane (C&M), separation, chemical, balance of plant and installation (B&I), and other operations, while the revenue comes from C₂H₄ and O₂ production [8, 51]. As can be seen, the profitability of the process depends on the electricity cost (53.2%), which was commonly dominated by FE_C2H4 and operating current density (Figure 5a). In this context, we analyzed the variation in the relative contributions of different cost components at various applied voltages and calculated the corresponding total costs. The results indicate that, as the voltage increases, the proportion of electricity costs rises while the share of chemical costs gradually decreases (Figure 5b). Notably, the total cost reaches a minimum of 773 $ ton_C2H4 ⁻¹ at −1.5 V_RHE, suggesting an optimal operating voltage from an economic perspective (Figure S33). We further investigated the impact of FE and current density on the profitability of C₂H₄ electrosynthesis. We found that the profitability is highly sensitive to FE and current density of C₂H₄, since a decrease in FE or current density results in a significant increase in operational costs, thereby substantially augmenting the total cost. Benefiting from the high FE_C2H4 and j_C2H4, the catalytic system using Cu₂O‐MOE as the catalyst keeps profitability within the potential range of −1.4 to −1.7 V_RHE and even remains the most profitable (566 $ ton_C2H4 ⁻¹ at −1.5 V_RHE) among recently reported systems (Figure 5c and Table S8).

FIGURE 5.

(a) Total cost of the C₂H₄ electrosynthesis process under the base case scenario at conditions of the FE of 75%, current density of 400 mAcm⁻² and potential of −1.5V_RHE. (b) Sensitivity analysis of kinds of costs to different potentials. (c) The impact of different FEs and current densities on the profit of C₂H₄ electrosynthesis, assuming the potential remains constant at −1.5 V_RHE. (d) LCA of the traditional and electrochemical synthesis of C₂H₄.

Furthermore, the sustainability of this system was evaluated by performing a life‐cycle assessment (LCA) [52, 53, 54, 55]. The theoretical model was according to 1 t C₂H₄ production in 100 cm² flow reactor, with the calculated various environmental implications being listed. In LCA model, we compared the environmental impacts of electrochemical C₂H₄ production with those of the conventional industrial process in terms of atmospheric emissions, radiation, global warming potential, fossil resource depletion, and toxicity (Figure 5d). Across all these categories, the electrochemical route demonstrates remarkable environmental advantages. Specifically, for every ton of ethylene produced electrochemically, only 0.02 tons of harmful gaseous and particulate emissions (NO_x, SO₂, PM_2.5) are generated [56]. The associated ionizing radiation is equivalent to 0.21 MBq of Co‐60, while the global warming potential (GWP) corresponds to 1.48 tons of CO₂ emissions. The overall human and ecological toxicity (marine, freshwater, and terrestrial) is equivalent to 30.71 tons of 1,4‐dichlorobenzene (1,4‐DCB) released, and the fossil fuel consumption is approximately 0.423 tons of oil. Compared with the conventional process, these values represent an overall reduction of approximately 42% in total environmental impact, with reduction exceeding 85% in the fossil resource scarcity impact categories.

3. Conclusion

We employed a machine‐learning‐guided screening to achieve an ideal Cu₂O model for elucidating how crystallographic orientation entropy governs reconstruction dynamics. Guided by these insights, a series of Cu₂O catalysts with high, medium, and low orientation entropy were rationally synthesized. Notably, the medium‐entropy Cu₂O catalyst achieved a remarkable FE_C2H4 of ∼75% at −1.5 V_RHE and a current density of 400 mA cm⁻² during alkaline CO₂RR, demonstrating performance comparable to current state‐of‐the‐art systems. The near‐octa‐coordinated Cu^0.4+ species derived from entropy‐mediated reconstruction facilitated CO₂ activation and H₂O dissociation, which synergistically increased *CO surface coverage and hydrocarbonization, thereby promoting the *CO−*COH coupling pathway. Furthermore, preliminary TEA and LCA show the profitability of 566 $ and an overall reduction of ∼42% in total environmental impact of this electrocatalytic strategy for C₂H₄ per ton production. This study advances the fundamental understanding of in situ reconstruction for the CO₂RR to C₂H₄ and offers a promising entropy‐governed strategy for steering selective reaction pathways.

Author Contributions

Jing Ma: investigation, resources. Zhenwei Tong: conceptualization, methodology, writing – original draft. Shucong Zhang: conceptualization, methodology, writing – original draft. Xinshuo Shi: investigation, resources. Lei Shi: investigation, resources. Bingzheng Wu: investigation, methodology, resources. Dawei Shang: investigation, resources. Shengwei Zhang: investigation, resources. Shenlong Zhao: conceptualization, writing – review and editing, supervision. Tong Yu: investigation, resources. Piaoping Yang: methodology, software, writing – original draft, resources. Rihua Xiong: methodology, resources, writing – review and editing, supervision. Jiali Zhou: investigation, resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

The authors have cited additional references within the Supporting Information.Supporting File: anie72837‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.