Synthesis and structure characterization of L-CuxO catalysts.
The copper oxide-based nano-bipyramid electrocatalyst with high tip curvature (L-CuxO-HC) was synthesized by pulsed laser ablation in liquid (PLAL). Supplementary Fig. 1. illustrates the formation of nano-bipyramids: a copper target immersed in water is transiently heated by a pulsed laser and transformed into a vapour and/or plasma state, which is then quenched by the cool water to a solid state. The fast heating/quenching kinetics cause abundant defects of stacking faults. To understand the growth process of L-CuxO-HC, we tracked its structural evolution during the synthetic process from ex situ scanning electron microscopy and transmission electron microscopy (SEM and TEM; Supplementary Fig. 2 and 3) images. At first, the PLAL procedure yielded dispersed nanoparticles characterized as amorphous copper oxide (CuxO) according to X-ray diffraction (XRD; Supplementary Fig. 2). The formation of oxide nanoparticles could be ascribed to the reaction of laser-stimulated Cu nanoparticles with dissolved oxygen in water to form CuxO (10 min). Next, the CuxO nanoparticles self-assembled into loosely interconnected agglomerates and further merged to form a double vertebral body structure (30 min), which kept growing and ripening to develop into CuxO bundles eventually (60 min).
Transmission electron microscopy (TEM), high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy-dispersive X-ray spectroscopy (EDX) elemental mapping analyses confirm the bipyramid structure of L-CuxO-HC (Fig. 1a–c and Supplementary Figs. 4–7)). The spherical aberration correction transmission electron microscopy (SAC-STEM; Fig. 1d and e) images show that the surface consists of areas with different contrasts; electron energy loss spectroscopy (EELS) spectra of O K-edge and Cu L-kedge (Supplementary Fig. 8) reveals that the dark and bright domains correspond to CuO and Cu2O phases, respectively, which agrees with selected area electron diffraction (SAED) and X-ray diffraction (XRD) pattern results (Fig. 1h and Supplementary Fig. 9). The stacking faults induce atomic steps on the surface of L-CuxO-HC, which could be seen at the edge (Fig. 1f) and from the surface intensity profile (Fig. 1d and e). The intensity fluctuation along the blue frame in Fig. 1d reveals the surface steps of L-CuxO-HC (Fig. 1g). For comparison, we also prepared a suite of L-CuxO with medium and low tip curvatures (15° for L-CuxO-HC; 30° for L-CuxO-MC; 150° for L-CuxO-LC) (Supplementary Figs. 10–15). Both L- CuxO-MC and L-CuxO-LC possess a mixed CuxO phase, as revealed by the XRD patterns with the simultaneous existence of CuO and Cu2O signals.
Electrochemical CO2RR Performance.
To validate the enhancement of CO2RR by the electric field and interface engineering, we also synthesized Cu tip with a smooth interface by an anodic method35 and used commercial CuO and Cu for comparison of electrochemical performance (Supplementary Figs. 16 and 17). We evaluated the CO2RR performance of these catalysts in a flow cell in 1 M KOH electrolyte (versus the reversible hydrogen electrode (RHE) and no iR-compensation). We detected and analyzed the products under different potentials through the online gas chromatograph (GC) and nuclear magnetic resonance (NMR) (Supplementary Figs. 18–22).
The built-in electric field effect can be indicated by comparing L-CuxO with different tip curvatures. Figure 2a–c compare the CO2RR performance of different samples. For L-CuxO-HC (Fig. 2a), the partial current density of C2+ production reached 665.9 mA cm−2 with a corresponding FEC2+ of 81.2 ± 5% at −2.8 V. With an increasing tip curvature, the FEC2+ decreases to 73% and 57% for L-CuxO-MC and L-CuxO-LC, respectively (Fig. 2b). The field effect is more pronounced when comparing the C2H4 yield; 1.55 mmol−1 h−1 cm−2 was obtained on the L-CuxO-HC, which is 2-fold of L-CuxO-MC and 3-fold of L-CuxO-LC. L-CuxO-HC always maintains the highest C2+ current density over the entire voltage range compared with other samples (Fig. 2c), highlighting the electric field effect. In addition, the L-CuxO-HC also exhibits good durability and works steadily under a high current density of 600 mA cm–2 for 12 h with an average C2H4 selectivity of 55% (Fig. 2d and Supplementary Figs. 23–29).
The nanograins interface effect can be inferred by comparing the L-CuxO samples to other CuxO prepared from conventional methods, including C-Cu, C-CuO, and Cu tip (Fig. 2a–c). Specifically, all the L-CuxO with different curvatures have a much better FEC2+ than the Cu tip. For C-Cu and C-CuO, both show similar curvatures to L-CuxO-LC, but their FEC2+ are significantly smaller (Fig. 2b and c). The above comparisons suggest the rich nanograins interface is favourable for C2+ generation during CO2RR.
Built-in electric field mechanism investigations.
We first investigate the distribution of locally enhanced electric fields on L-CuxO with diverse curvatures of 15, 30, 60 and 90 nm−1 by FEM-based theoretical simulations performed on COMSOL multiphysics (Fig. 3a and Supplementary Fig. 30). We found that with decreasing tip curvature from 90° to 15°, the tip-concentrated electron density shows a 6-fold enhancement, resulting in a significantly enhanced electric field at the 15° tip (Fig. 3b).
We then experimentally evaluate the effect of tip curvatures in shaping the local environments. We performed the K+ absorbing test by measuring the concentration of adsorbed K+ on the electrodes (Fig. 3c). The results show that the three L-CuxOs have a higher K+ concentration than the quasi-planar C-Cu, and the adsorbed K+ concentration can be further enhanced with a sharper tip (L-CuxO-HC). This K+ absorbing test result is consistent with the simulation results. It has also been reported that OH− species located at the catalyst surface could benefit C–C coupling and C2+ production36,37. We thus use cyclic voltammetry (CV) to verify the OH‒ adsorption (OHad) features on L-CuxO. Figure 3d demonstrates the pronounced OHad peaks associated with Cu (100) facets on L-CuxO-HC. Noted that there is a peak shifting with tip angles. The OHad peaks shift to more negative potentials with decreasing tip curvatures, which suggests OH− could be adsorbed more easily on L-CuxO-HC, leading to a beneficial micro-environment for CO2-to-C2+ conversion (Note 1). Those results indicate that the local environments arising from the tip structure would concentrate the K+ cation and OH− species near the catalyst surface, both of which could promote the C2+ products.
Based on the above simulation and experimental results, we further broadened the CO2RR performance to pH-universal conditions for L-CuxO-HC. Fig. 3e, f and Supplementary Figs. 31 and 32 demonstrated the FE and C2+ partial current density under acidic, neutral, and alkaline electrolytes. Notably, L-CuxO-HC achieves 72.8% ± 4.3 FEC2+ in neutral electrolytes and maintains 56.9% ± 5.4 FEC2+ in acidic electrolytes. The partial current density of C2+ reaches 297.7 ± 24 mA cm-2 and 397.1 ± 19 mA cm-2 under acidic and neutral environments, respectively. Those good performances may be due to the good buffer capacity induced by the electric field effect of tip. The enhanced K+ adsorption capacity will kinetically reduce the proton coverage on the Helmholtz plane through the competitive adsorption behavior driven by the electrostatic field13, thus inhibiting the hydrogen evolution and contributing to the good performance in the pH-universal conditions (Fig. 3g, Supplementary Fig. 33 and Supplementary table 1 ).
Nanograins Cu+/Cu2+ interfaces mechanism investigations
In addition to the electric field effect, the unique heterogeneous structures of L-CuxOs also benefit the C2+ formation. To gain more accurate structural information about the chemical status and elemental composition of the samples, X-ray absorption spectroscopy at the Cu k-edge was used. As the X-ray absorption near-edge structure (XANES) result shows (Fig. 4a), the absorption edges of L-CuxO with different tip angles exhibit nearly identical profiles. These absorption edges lie between the spectrum features of C-CuO and C-Cu2O, indicating the existence of the complex state of Cu+ and Cu2+ in the three L-CuxO electrocatalysts.
The extended X-ray absorption fine structure (EXAFS) reveals the coordination numbers are about 3.6 (L-CuxO-HC), 3.4 (L-CuxO-MC) and 2.8 (L-CuxO-LC) (Fig. 4b and Supplementary table 2). This might result from the increased Cu2O phases that reduce the average Cu-O coordination numbers. Wavelet transform was further applied to investigate the coordination environment of the Cu species in samples. As shown in Fig. 4c, the intensity of L-CuxO-HC, L-CuxO-MC, and L-CuxO-LC are very close, confirming that the Cu-O bonds are dominant in all these samples, matching well with the EXAFS fitting results (Supplementary Fig. 34). These results agree with XANES and XRD data that the Cu has the complex oxidation among these catalysts.
To probe the dynamic evolution of surface adsorptions during CO2RR and to elucidate the mechanism of boosted C2+ selectivity, we then performed in situ attenuated total reflection Fourier transform infrared spectroscopy (ATR-FTIR) using L-CuxO-HC and C-CuO. For ATR-FTIR, the downward band in the resulting spectra indicated the formation of intermediates during CO2RR, while the upward band referred to the consumption/desorption of surface species. Starting from −0.4 V, the ATR-FTIR spectra for the L-CuxO-HC catalyst exhibit several new peaks compared to C-CuO (Fig. 4d). Specially, two peaks at ∼1034 and ∼1231 cm−1 associated with *COH and *CHO, respectively, are observed for L-CuxO-HC, which are the important intermediates for C2+ products, especially C2H438. L-CuxO-HC also exhibits additional peaks at ∼1182, ∼1126, and ∼1301 cm−1 attributed to the absorbed *OCCOH and *OC2H5, which are different from those of C-CuO39. This indicates that more key intermediates of *OCCOH and *OC2H5, favoring the production of C2+, are formed on L-CuxO-HC than on C-CuO catalysts. The ATR-FTIR results have provided experimental evidence for the defect and tip-assisted C−C coupling mechanism over the L-CuxO-HC catalyst.
To characterize the dynamic product profiles during the CO2RR, a differential electrochemical mass spectrometry (DEMS) was conducted (Supplementary Fig. 35). We performed a continuous 4-cycle DEMS measurement while scanning the potential between 0 V and −2.1 V; each cycle took about 420 s. Figure 4e compares the mass signal intensity of C2H4, CO and CH4 at m/z = 26, 28, and 15, respectively, as a function of cycle number and time. We choose m/z=26 for C2H4 to avoid interference from CO. With the increasing overpotential, CO is produced first and then the CO production rate decreases when the C2H4 emerges; the rate is further reduced with the appearance of CH4, which strongly proves CO is an essential intermediate in the production of C2H4 and CH4. In addition, the ratio of integrated mass signal intensity of CO relative to the sum area of CO and C2H4 production on L-CuxO-HC dropped down sharply with the increasing overpotential, which indicates that CO is consumed more rapidly on L-CuxO-HC than C-CuO during CO2RR, supporting that the tip effect is more significant than the C-CuO (Fig. 4f).
To further verify the role of the nanograins interfaces for C−C coupling over L-CuxO-HC, we used DFT to calculate the energy barriers for the C2H4 formation pathway. Here, Cu2O(110) and CuO(110) slabs are first constructed as the nanograins of L-CuxO-HC based on the STEM and XRD results (Fig. 4g and Supplementary Figs. 36–40). Then, the energy barrier of each reaction step on the Cu2O(110) slab, CuO(110) slab and Cu2O(110)/CuO(110) interfaces are calculated to evaluate the catalytic performance of different catalytic sites in L-CuxO-HC. The full reaction pathway (Supplementary Fig. 36) shows that the rate-determining step (RDS) of Cu2O(110)/CuO(110), Cu2O(110) and CuO(110) are the same, which is *CO + *CO→*CO + *COH, but it is smallest on Cu2O(110)/CuO(110) (0.78eV), comparing with Cu2O(110) (1.22 eV) and CuO(110) (1.4 eV). (Fig. 4g). Hence, C2H4 production happens more easily on the Cu2O(110)/CuO(110) interface, indicating that these interfaces are more catalytically active in L-CuxO-HC. The *CO–*COH dimerization pathway has been reported to be highly essential over other pathways, particularly for C2 products involving C-C coupling 40-43. Furthermore, DFT calculations reveal that the adsorption energy of the *OCCOH intermediate is 0.14 eV on Cu2O(110)/CuO(110), which is smaller than that on Cu2O(110) (0.15 eV) and CuO(110) (0.61 eV) (Fig. 4g). We further consider the pH effects on the reactions, which show similar trends with a lower energy barrier at the CuO/Cu2O interfaces (Supplementary Fig. 37). These results verify that the presence of abundant interfaces is beneficial to the adsorption of the post-dimerization intermediate (*OCCOH), thus reducing the energy barrier of C–C dimerization. Our calculation mutually agreed with the in situ FTIR and online DEMS analysis.
Electrochemical NITRR reduction performance and mechanism investigations.
The field effect and nanograins interfaces might also benefit other multielectron electrochemical reactions. As a proof-of-concept application, the electrocatalytic NITRR activity of catalysts was evaluated using L-CuxO-HC in an H-type cell (Supplementary Figs. 41–48). Figure 5a shows the LSV curves of L-CuxO-HC and glass carbon in 1 M KOH with and without 1 M KNO3. Glass carbon electrode shows similar currents in solutions with and without 1 M KNO3, mainly attributed to hydrogen evolution reaction (HER). In 1 M KNO3 solution, the current density of the glass carbon is slightly higher in KNO3-containing solution because of carbon defects, but the total current remains negligible22,44. Notably, when using the L-CuxO-HC catalyst, the onset potential is significantly reduced to +0.2 V and the current density reaches >1000 mA cm–2 at −1.2 V. Chrono-amperometry between −0.5 V and −1.0 V is performed; the NH3 detected by H nuclear magnetic resonance (NMR) and the FE are summarized in Fig. 5b. Meanwhile, no N2, NO, N2O and NO2 can be detected at all considered potentials according to the online electrochemical mass spectrometry (OEMS) result for the possible gaseous products (Supplementary Fig. 43). L-CuxO-HC reaches >80% FENH3 at the window from −0.8 V to −1.0 V and an NH3 yield rate of 81.8 mg–1 h–1 mgCu–1 at −1.0 V (Fig. 5c). We examine NH3 in nitrate-free electrolyte and confirm that all our NH3 products are generated by NITRR rather than by any contaminations (Fig. 5d). We also test isotope-labelled before and after NITRR coupled to future ensures the source of NH3 (Supplementary Fig. 49). As different wastewater may have a varied nitrate concentration, we further examine the L-CuxO-HC catalytic performance with different KNO3 concentrations (Fig. 5e and Supplementary Figs. 50 and 51). L-CuxO-HC preserves its high ammonia FE and activity under 0.1 M and 0.5 M KNO3 and delivers an NH3 partial current density of 353 and 573 mA cm–2 at −1.0 V, respectively. The FE(NH3) decreases with increasing KNO3 concentrations from 0.1 M to 1 M. Notably, the FENH3 could reach ~95% from −0.5 V to −1.0 V in 0.1 M KNO3. The higher selectivity in lower concentrations could be attributed to the competing adsorption between NO3– and NO2–. During NITRR, NO3− is first reduced to NO2− and then undergoes deoxygenation and hydrogenation to form NH3.When NO3− is in higher concentration, the first step will compete with the subsequential reduction, leading to a higher FENO2− in 1 M KNO3 (Supplementary Fig. 52). To probe the durability, a stability test was performed at −1.0 V for 10 consecutive electrolysis cycles; each with refreshed electrolyte and lasting 30 min. The current density keeps relatively steady at ∼750 mA with a slight decrease (Supplementary Fig. 53). The FENH3 and yield rate show negligible decay over the whole test, implying high stability of the catalyst (Fig. 5f).
We further used in situ FTIR to track intermediates adsorbed on the surface. In Fig. 5g, five obvious absorption bands appear in the spectra of L-CuxO-HC20,44. Firstly, with the potential increased, the upward absorption bands at 1354 cm−1 ascribe respectively to symmetric and asymmetric N-O stretching of NO3−, indicating consumption of NO3−; at the same time, the downward band at 1236 cm−1 is attributed to N-O antisymmetric stretching vibration of NO2−, indicating NO2− formation from NO3− reduction; with potential negatively moving to 0.2 V, close to the onset potential of the LSV curve, another intermediate observed around 1110 cm−1 was ascribed to N-O stretching vibration of hydroxylamine (NH2OH), which is a key intermediate for NH3 formation. In addition, with the negative shift of the working potential, the ratio of area of NO2− relative to the sum area of NO2− and NH2OH production on L-CuxO-HC dropped sharply, indicating that the tip effect could deeply enhance the hydrogenation of *NO2 to the final NH3 product.
We compare the FE, product partial current density and yield rate with other reported catalysts shown in Fig. 5h. It shows that L-CuxO-HC exhibits excellent performance in both CO2RR and NITRR, attributed to the unique morphology and abundant nanograins interfaces. To our knowledge, L-CuxO-HC is among the best-reported catalysts with bifunctional activities (Supplementary tables. 3 and 4).
Practical applications of the NITRR and CO2RR.
The chemical industry plays a key role in sustaining the world economy and underpinning future technologies2. Green chemistry aims to design chemical products and processes that reduce or eliminate the use or generation of hazardous substances3,5. Flue gases from fossil fuel consumption and nitrate waste from industrial water are two pollutants that cause the imbalance of carbon and nitrogen cycles. We further show that L-CuxO-HC, as a bifunctional catalyst, could be used for the cascade valorization of nitrate waste and flue gas. Specifically, an alkaline ammonia solution from NITRR could be used to capture CO2 from flue gases. The addition of acid will form ammonium fertilizer and potash fertilizer (Fig. 6a) while releasing CO2 for further conversion into C2+ chemicals. Thus, coupling the NITRR-CO2RR system reduces the chemical discharge that could be attractive to the environment and economics for wastewater and waste gas treatment.
Since industrial wastewater and polluted groundwater are mostly complex and the solution has a wide range of acidity and alkalinity, we assumed four different scenarios, including pure nitrate, acid (pH 3), neutral (pH 7), and alkaline (pH 14). After electrolysis at ~600 mA cm-2 for 1h, the ammonia yield rate reaches 1380.97, 1309.59, 1593.17 and 2454.97 mg L–1 h–1, and their CO2 capture values are ~22 mg mL–1, ~24.2 and 30.8 mg mL–1 and 41.8 mg mL–1, respectively. The CO2 captured by the initial NO3– electrolytes is negligible; only alkaline have the ability of CO2 capture, which also proves that the treated electrolyte after NITRR is the main absorbent of CO2 (Fig. 6b and c). Finally, the ammonia fertilizer and potash fertilizer are collected by rotary evaporation (Fig. 6d and Supplementary Figs. 54-56).