Since the metallic Cu has demonstrated good performance for the NO3−RR[9], we select Cu as core to construct the yolk-shell structure. We note that the water dissociation process on Cu is an endothermic process with large energy requirement[20], which may be adverse to the NO3−RR. We reason that constructing a shell with good water dissociation ability can accelerate the sluggish proton transfer and reaction kinetics of NO3−RR on Cu. DFT calculations were then performed on Cu and Cu2Se to examine their water dissociation ability. (Figure 1a) On the Cu, the water dissociation is an endothermic process with the requirement energy of 0.14 eV, in consistent with other literatures.[20] In sharp contrast, the water dissociation on the Cu2Se turns to be exothermic with the release energy of 0.47 eV. This suggests that the water dissociation is more favorable on the Cu2Se than the Cu. Moreover, the adsorption strength of H* is higher on the Cu2Se than the Cu (Figure 1a), and the too strong adsorption strength makes the H* difficult to achieve the dimerization and generate H2 on the Cu2Se for suppressing the HER.[21, 22] We are grateful to find that the adsorption strength of NO3− reactants is also much higher on the Cu2Se than Cu, which can favor the adsorption of NO3− for promoting the NO3−RR. (Figure S1) The optimize surface configurations of water, hydrogen, and NO3− adsorbed on the Cu and Cu2Se are shown in Figure 1b-1c and S1-S2. With unique features of strong water dissociation ability and strong adsorption strength of both H* and NO3−, the Cu2Se may be the suitable one as shell for constructing the yolk-shell structure, thereby boosting the proton transfer and reaction kinetics of NO3−RR on the Cu core.
As a proof-of‐concept study, we constructed the yolk-shell structure comprising the metallic Cu core and Cu2Se shell (denoted YS-Cu@Cu2Se). The YS-Cu@Cu2Se was then synthesized via the electrochemical reduction of the yolk-shell structure comprising Cu2O core and Cu2Se shell (denoted YS- Cu2O@Cu2Se). A two-step route was developed to synthesize the YS-Cu2O@Cu2Se. (Figure 1d) First, uniform Cu2O nanocubes (denoted Cu2O-NB) in diameter of 400 nm (Figure S3 and Figure S4) were reacted with NaHSe aqueous solution to generate the core-shell structured Cu2O@Cu2Se (denoted CS-Cu2O@Cu2Se). Then, the CS-Cu2O@Cu2Se was treated with an ammonia aqueous solution to etch the Cu2O core, thereby creating voids between core and shell and thus generating the yolk-shell structured YS-Cu2O@Cu2Se. As the Cu2Se shell is not as high activity as Cu2O in the reaction with ammonia, the shell can be kept after the treatment. To optimize the NO3−RR performance, the void size of YS-Cu2O@Cu2Se is continuously regulated by altering the ammonia treatment time.
Transmission electron microscopy (TEM) images of CS-Cu2O@Cu2Se display a typic core-shell structure with a shell in thickness of ~19 nm. (Figure 2a) Corresponding high-resolution TEM (HRTEM) images signify that the shell has continuous lattice fringes with a distance of 0.20 nm, corresponding to (220) plane of Cu2Se. (Figure S5) Upon treatment of ammonia for 0.5 h, a void in size of 40 nm can be observed between the shell and core. (Figure 2b) With ammonia treatment time gradually increases from 0.5 to 1 and 1.5 h, the size of void enlarges to 60 and 80 nm. (Figure 2c and 2d) These samples after the treatment were denoted YS-Cu2O@Cu2Se-40, YS-Cu2O@Cu2Se-60, and YS-Cu2O@Cu2Se-80 (40, 60, and 80 stands for the void size). Scanning electron microscopy (SEM) images of sample before and after the treatment are shown in Figure S6, revealing that the treatment makes the smooth surface of shell turn to be rough. This may increase the surface area for exposing more active sites. Since the YS-Cu2O@Cu2Se-60 displays higher NO3−RR activity than other yolk-shell structure with different void sizes, our attention is focused on this sample. (Figure S7) Energy-dispersive X-ray spectrum (EDX) elemental mapping of YS-Cu2O@Cu2Se-60 further confirms the successful generation of a yolk-shell structure with a Cu2Se shell. (Figure 2e-2g) For comparison, we also completely remove the Cu2O core to obtain the hollow structured Cu2Se nanoboxes (denoted H-Cu2Se) by further extending the treatment time
(Details see experimental section). (Figure S8) X-ray diffraction (XRD) characterizations of H-Cu2Se manifest a set of peaks for the Cu2Se without other peaks for the Cu2O (Figure S9), signifying the complete removal of Cu2O core. As for the YS-Cu2O@Cu2Se-60 with both Cu2Se and Cu2O, the only presence of peaks for Cu2O may be due to the shielding effect of strong Cu2O peaks on weak Cu2Se ones. (Figure S10)
As it well reported, the Cu2O easily transforms to metallic Cu under a reductive potential[23, 24], whereas the Cu2Se can resist the reduction even at a highly negative potential of -2.4 V vs Ag/Ag+ (nearly -1.8 V vs RHE).[25] Inspired by this, we selected a moderate reductive potential (i.e., -0.4 V vs RHE) to electrochemically reduce the Cu2O core of YS-Cu2O@Cu2Se to Cu while remain the crystal phase of Cu2Se shell unchanged. In order to avoid the mutual interference of Cu2Se core and Cu2O shell, their electro-reduction behaviors were investigated separately. After the electro-reduction of Cu2O-NBs, strong XRD peaks for metallic Cu are observed instead of those for Cu2O phase (Figure S11), indicating that the Cu2O has completely transformed to the metallic Cu. Moreover, all peaks for Cu-O bonds are disappeared on the Raman spectra (Figure S12), and X-ray photoelectron spectroscopy (XPS) spectra display only peaks for metallic Cu (Figure S13). By combining the above XRD, Raman and XPS results, it is clear that the electro-reduction enables the complete transformation of Cu2O to metallic Cu. However, the electro-reduction does not change the crystal phase of H-Cu2Se, and the XRD peaks for Cu2Se still dominates after the reduction. (Figure S14) This can be further evidenced by the Raman and XPS spectra. (Figure S15 and S16) Based on the above results, we reason that the Cu2O core of YS-Cu2O@Cu2Se-60 transforms to metallic Cu yet the Cu2Se remains its crystal phase. The transformation of Cu2O-Cu can be confirmed by the disappearance of both XRD peaks for Cu2O after the reduction of YS-Cu2O@Cu2Se-60. (Figure S17) As for the Cu2Se shell, unchanged lattice fingers for the Cu2Se are observed, implying that the crystal phase of Cu2Se shell remains unchanged after the reduction. (Figure S18) Elemental mapping images of YS-Cu2O@Cu2Se-60 after the reduction manifest that the yolk-shell structure with a Cu2Se shell is well preserved. (Figure S19) Overall, after the electro-reduction, the YS-Cu2O@Cu2Se-60 successfully converts into the yolk-shell structured nanoreactor composed of a metallic Cu core and a Cu2Se shell (denoted YS-Cu@Cu2Se-60). In addition, according to the distinct electro-reduction behaviors of Cu2O and Cu2Se, the Cu2O-NB, CS-Cu2O@Cu2Se, and H-Cu2Se after the reduction are denoted Cu-NB, CS-Cu@Cu2Se and H-Cu2Se, respectively.
The electrochemical performance of Cu-NB, CS-Cu@Cu2Se, YS-Cu@Cu2Se-60, and H-Cu2Se for the NO3−RR was evaluated in 1 M KOH containing 0.1 M NO3−, and their linear sweeping voltammograms (LSV) curves are shown in Figure 3a. With the shell of Cu2Se, the CS-Cu@Cu2Se displays higher NO3−RR activity than the Cu-NBs without the shell, indicating that the shell facilitates the NO3−RR. After creating void space to generate the yolk-shell structure, a further large enhancement of NO3−RR activity is achieved on the YS-Cu@Cu2Se-60, reflecting that the yolk-shell structure also plays important roles for boosting the NO3−RR. For example, at an applied potential of 0.18 V vs RHE, the YS-Cu@Cu2Se-60 yields a current density of 9.6 mA cm−2, which is 30- and 2.8-fold larger than the Cu-NB, and CS-Cu@Cu2Se, respectively, indicating the much higher intrinsic NO3−RR activity of YS-Cu@Cu2Se-60. Impressively, no obvious NO3−RR activity is observed on the YS-Cu@Cu2Se-60 after the mass loading is halved, further confirming its good NO3−RR activity. (Figure S20) Moreover, the H-Cu2Se is almost inactive for the NO3−RR with negligible current densities, which can be attributed to the too strong adsorption strength of Cu2Se towards NO3− (as shown in above DFT calculations) for inhibiting the following reaction processes of NO3−RR. Thus, in the present YS-Cu@Cu2Se-60, the metallic Cu core are the genuine active sites for the NO3−RR, and the yolk-shell structure with a Cu2Se shell can significantly enhance the activity of Cu.
Liquid products from the NO3−RR were fully analyzed through the ultraviolet–visible (UV-Vis) spectra for determining FENH3. According to Figure S21-S22, the NH4+ cations are the only detected liquid products at different potentials. We then quantified the content of NH4+ in liquid products via both UV-Vis and 1H nuclear magnetic resonance (NMR) spectroscopy (Figure S23). We find that the results of UV-Vis and 1H NMR match well one with another (Figure S24), reflecting the reliability of our tested NH4+ content. The origins of NH4+ were studied by performing an isotope tracing experiment. After changing the 14NO3− to 15NO3− as reactants, the 1H NMR of obtained NO3−RR products display two peaks corresponding to 15NH4+ instead of three peaks to 14NH4+, verifying that the produced NH4+ is indeed originated from the NO3−RR other than other impurities.[26] (Figure 3b) Moreover, chromatograms of gas products indicate that the hydrogen (H2) is the only product at different potentials without the byproduct of N2. (Figure S25)
Combining results of UV-Vis and chromatograms, we calculated the FENH3 at different potentials. (Figure 3c) Obviously, the YS-Cu@Cu2Se has good ability for suppressing the competitive HER with high FENH3 values of more than 95% over a wide potential range from -0.05 V to -0.20 V vs RHE. Notably, the YS-Cu@Cu2Se delivers a maximum FENH3 up to 99.6% at the potential of -0.15 V vs RHE. Accordingly, at the potential of -0.15 V vs RHE, the YS-Cu@Cu2Se achieves a NH3 yield rate of 0.94 mmol cm−2 h−1, 2.54- and 1.45-fold larger than the Cu-NB, CS-Cu@Cu2Se, further corroborating the much higher NO3−RR activity of YS-Cu@Cu2Se-60. (Figure 3d) Impressively, the NH3 yield rate of YS-Cu@Cu2Se-60 (0.94 mmol cm−2 h−1) compares favorably with other reported non-precious electrocatalysts, such as Fe-PPy single-atom (0.16 mmol cm−2 h−1)[27], Cu50Ni50 alloys (0.79 mmol cm−2 h−1)[11], Fe single-atom (0.062 mmol cm−2 h−1)[10], Cu nanosheets (0.023 mmol cm−2 h−1)[28], Porous Cu70Ni30 (0.0026 mmol cm−2 h−1)[17], Copper powders (0.27 mmol cm−2 h−1)[29], and even comparable with some precious-metal-based electrocatalysts, such as Strained Ru Nanoclusters (1.17 mmol cm−2 h−1)[12]. (Figure 3e), demonstrating the great potentials of YS-Cu@Cu2Se-60 as electrocatalysts for practical electrosynthesis of NH3 under mild conditions.
Electrochemically active surface area (ECSA) provides a method to estimate the number of active sites on electrocatalysts.[21, 30, 31] (Figure S26) In general, the ECSA values can be estimated through the double-layer capacitance (Cdl). The Cdl value of YS-Cu@Cu2Se-60 (9.6 mF cm−2) is larger than the Cu-NB (7.6 mF cm−2) and CS-Cu@Cu2Se (9.0 mF cm−2). The larger Cdl value of YS-Cu@Cu2Se-60 implies that the void space of yolk-shell promotes active sites exposure. Moreover, LSV curves, normalized to Cdl values, reveal that the YS-Cu@Cu2Se-60 has higher intrinsic NO3−RR activity than the other two electrocatalysts. (Figure S27) In addition, yolk-shell structured electrocatalysts with a protective shell usually have enhanced stability than the counterpart without.[32] To this end, we tested the stability of YS-Cu@Cu2Se-60 by conducting 20-h consecutive NO3−RR. (Figure 3f) During the stability test, the nearly unchanged current density, together with the unchanged FENH3 (Figure S28), signifies the excellent stability of YS-Cu@Cu2Se-60 for NO3−RR. Moreover, the well-preserved yolk-shell structure after the stability test further confirms the stable nature of such yolk-shell structure. (Figure S29)
As discussed above, the Cu2Se shell may promote the water dissociation to provide protons for boosting the proton transfer and reaction kinetics of NO3−RR on Cu core. To verify it, we performed kinetic isotope effect (KIE) measurements to investigate the proton transfer kinetics of electrocatalysts. As it well reported, KIE values can be obtained according to the following equation: KIE = jD/jH, where jD and jH stand for the current densities recorded in protic solution and deuterium solution, respectively.[33] Note that the equation hypothesizes that the rate of electrochemical reactions is directly proportional to the tested current density.[34] Accordingly, the equation is applicable only under conditions that the Faradaic efficiency is unchanged in either protonic or deuterium electrolyte. As such, we also tested the FENH3 of Cu-NB and YS-Cu@Cu2Se-60 at the potential of -0.15 V vs RHE in a deuterium electrolyte (1 M NaOD in D2O). (Figure S30) The nearly unchanged FENH3 after altering the protonic to deuterium electrolyte suggest the availability of the above equation to calculate KIE values of these two electrocatalysts.
KIE values can offer information regarding with the proton transfer kinetics of electrochemical reactions and thus help to understand the rate-determining step (RDS). Generally, a KIE value > 1 signifies that the proton transfer process is existed in the RDS of reactions, however a KIE value ≈1 reflects no existence of proton transfer event in the RDS.[35, 36] A large current density decrease is observed on the Cu-NB after changing the electrolyte from 1 M NaOH/H2O to 1 M NaOD/D2O (Figure 4a), and the KIE is as large as 4.8, suggesting that the proton transfer process is involved in the RDS of NO3−RR on Cu-NB, and the reaction kinetics is greatly limited by the sluggish proton transfer. In strikingly contrast, the change of electrolyte only induces a slight current density decay on the YS-Cu@Cu2Se-60 with a KIE value of 1.3. (Figure 4b) The much smaller KIE value of YS-Cu@Cu2Se-60 than the Cu-NB suggests that the sluggish proton transfer of Cu can be indeed enhanced by the yolk-shell structure. The reaction kinetics of these two electrocatalysts were then investigated through determining Tafel slopes. The Tafel slope of YS-Cu@Cu2Se-60 is 107.3 mV dec−1, smaller than that of Cu-NB (123.9 mV dec−1) and CS-Cu@Cu2Se (119.0 mV dec−1), implying the rapider reaction kinetics of YS-Cu@Cu2Se.[37, 38] (Figure S31) Overall, protons provided by Cu2Se promote the sluggish proton transfer of Cu and thus enhances the reaction kinetics of NO3−RR.
To deepen our understanding on the tandem reaction process, operando Raman spectra were conducted on the Cu-NB and YS-Cu@Cu2Se-60 to monitor reaction intermediates during the NO3−RR. On scanning the potential of 0.2 V vs RHE over the Cu-NB, two Raman bands appear at 818 and 1047 cm−1 that are assigned to in-situ generated NO2* intermediates and NO3* reactants adsorbed on the surface of Cu-NB, respectively.[14] (Figure 4c) Further scanning to the potential of 0.1 V vs RHE manifests two new bands at 1534 and 1590 cm−1 corresponding to adsorbed *NOH intermediates and *NH3 products[14], respectively, and the intensities of these two bands gradually enhanced during the potential further negatively moves to -0.3 V vs RHE. As it reported, the appearance of *NOH other than the *NH2O intermediates implies that the NO3−RR undergoes the pathway: NO3− → *NO3 → *NO2 → *NO → *NOH → *N → *NH → *NH2 → *NH3, and the RDS of this pathway is the reduction of *NO to *NOH or the following reduction of *NOH to *N with the same energy barrier.[39] Notably, the Raman bands for the *NOH and *NH3 are already observed on the YS-Cu@Cu2Se-60 at a highly positive potential of 0.2 V vs RHE, 100 mV-positive than that of Cu-NB. (Figure 4d) This suggests that the energy barrier for accomplishing the transformation of *NO to *NOH and *NOH to *N is decreased on the YS-Cu@Cu2Se-60 with respect to the Cu-NB. In other words, the YS-Cu@Cu2Se-60 can reduce the energy barrier for the generation of *NOH and *N, which are in the RDS of NO3−RR. This can be further verified by the fact that the stronger intensities of *NOH and *NH3 on the YS-Cu@Cu2Se-60 than the Cu-NB at each tested potential. Recent reports indicate that the proton can markedly reduce the energy barrier for the hydrogen of nitrogen-containing intermediates during the NO3−RR, such as the hydrogenation of *NHO to *NH2O and *NH2 to *NH3.[12, 40] Furthermore, our KIE results indicates that the proton transfer is involved in the RDS of NO3−RR, and the Cu2Se shell of YS-Cu@Cu2Se-60 can provide enough protons to accelerate the proton transfer of NO3−RR, especially the RDS for reducing the energy barrier. This can be evidenced by the smaller apparent energy barrier for the NO3−RR on the YS-Cu@Cu2Se-60, compared with the Cu-NBs. (Figure 4e and Figure S32) Thus, we reason that the protons provided by the Cu2Se shell promotes the proton transfer of NO3−RR and the hydrogenation of key intermediates (i.e., *NO and *NOH), thereby reducing the energy barrier of RDS step in the NO3−RR. (Figure 4f)
On basis of above comprehensive knowledge of tandem reaction mechanism, we are in a good position to seek a possible explanation on the different NO3−RR activity induced by the geometric alteration of the yolk-shell structure. At first, our electrocatalytic tests are not in static electrolytes but under the stirring of 300 rad/min, and this can eliminate the mass transport limit during the NO3−RR. To verify if the existence of mass transport limit on the yolk-shell structures with different void sizes and Cu-NB, we recorded their LSV curves at different scan rates. All of them display unchanged LSV curves when scan rates are varied, indicating that their mass transport is very rapid without limit.[31, 41] (Figure S33) In other words, the NO3− and OH− concentration is the same on the Cu2Se shell of different yolk-shell structures, and thus it can assume the proton yield rate on the Cu2Se shell is similar at a specific applied potential. Accordingly, the local proton flux arriving to the Cu core varies with the diffusion distance, which is the void size of yolk-shell structure, and obeys the Fick’s first law: