Electronic regulation of Ni single atom by conned Ni nanoparticles for fast and energy-ecient CO 2 electroreduction

Electrocatalytic CO 2 to CO conversion is approaching the industrial benchmark. Currently, Au electrodes show the best performance, whereas non-precious metal catalysts exhibit inferior activity. Here we show a densely populated Ni single-atom on nanoparticle catalyst (NiSA/NP) via direct solid-sate pyrolysis, where Ni nanoparticles donate electrons to Ni(i)-N-C sites via carbon nanotubes network, achieves a high CO current of 352 mA cm −2 at -0.55 V vs RHE in an alkaline ow cell. When coupled with a NiFe-based oxygen evolution anode into a zero-gap membrane electrolyser, it delivers an industrial-relevant CO current of 310 mA cm −2 at a low cell voltage of -2.3 V, corresponding to an overall energy eciency of 57%. The superior CO 2 electroreduction performance is attributed to the enhanced adsorption of key intermediate COOH* on electron-rich Ni single atom, together with the dense active sites.


Introduction
Electrochemical CO 2 reduction reaction (CO 2 RR) emerges as a carbon-neutral approach to recycle CO 2 and store intermittent renewable energies. [1][2][3][4] However, the chemical inertness of CO 2 molecule and its multiple reaction pathways with up to 16 possible products cause electrocatalytic conversion processes energy-ine cient. 5,6 To achieve scalability for the technology, the reaction must occur at industrial-level currents at low overpotentials for ideally a single product. Currently, electrolyzers for CO 2 conversion to CO are reaching the pilot scale, 7,8 of which Au and Ag are the most promising catalysts. 9 Based on the membrane electrode assembly (MEA) device, certain Au IrO 2 full cell has achieved a CO current density of 300 mA cm −2 at a cell voltage of ~2.65 V. 10 According to the industrial benchmarks for CO 2 -to-CO electrolysis proposed in 2021, there remains a big gap to achieve higher current densities ( 400 mA cm −2 ) at the lower cell voltages ( 2.5 V), by using non-precious metal catalysts. 7 Single-atom catalysts (SACs) with metal-N-C (Fe, Co, Ni, etc.) structures have demonstrated promising performance for CO 2 conversion. 11 Gu et al. have developed an atomically dispersed Fe 3+ -N-C site that can deliver a CO partial current density of 94 mA cm −2 at an overpotential of 340 mV. 12 Certain selfsupported Ni-N-C membrane catalysts can produce CO with a high current of 308.4 mA cm −2 and 88% Faradaic e ciency (FE) in a ow cell. 13 Despite these breakthroughs, their single-atom nature with homogeneously dispersed active sites are only e cient in catalyzing single-molecule conversions. 14 For complex reactions that involve multiple intermediates, they generally show either too strong or too weak binding to one or more intermediates due to the scaling relationship. 15 For example, Ni-N-C sites typically exhibit very weak binding to COOH* while Fe-N-C sites exhibit too strong binding to CO*, both of which hinder e cient CO production. 16 Besides, the very low metal loading of SACs also severely limits their activity for practical applications.
Here, we show a cooperative Ni single-atom-on-nanoparticle catalyst (NiSA/NP) for energy-e cient CO 2 electroreduction. The design of this hybrid SAC enables: (i) electron density regulation of surface Ni-N-C by the encapsulated Ni nanoparticles for faster intermediate adsorption and desorption; (ii) in situ growth of high loading Ni single atoms on carbon nanotubes (CNTs) fed by Ni nanoparticles; (iii) a simple and scalable method for the mass synthesis of Ni SACs. The NiSA/NP catalyst achieves state-of-the-art CO 2to-CO performance in both the H-cell and zero-gap MEA full cells.

Results
Catalyst design and DFT prediction. Most of Ni-N-C catalysts have been revealed to show very weak binding to COOH* intermediate upon CO 2 -to-CO coversion. 13,16 Therefore, injecting foreign electrons into Ni single atom to tune its d band structure is desirable for enhancing the adsorption of COOH*. Based on this principle, previous studies have found that Ni(i)-N-C shows signi cantly enhanced CO 2 RR activity compared to Ni(ii)-N-C. 17 Note that metal nanoparticles can be easily formed with single atoms because of the aggregation effect, and the formation of nanoparticles sometimes is inevitable if a high loading of single atom is pursued. In this case, the electron-rich structure of metal nanoparticles can potentially tune the electronic con guration of nearby metal-N-C, 18 and enhance its electrocatalytic performance. Given that Ni is a strongly interacting metal with carbon, 19 we propose a Ni-N-C on Ni nanoparticles design, and rstly use DFT calculations to predict the electron distribution and charge transfer between Ni nanoparticles and a layer of carbon via the deformation electronic density calculation and Bader analysis. Here face-centered cubic Ni(111) is used as the targeted model because of its lowest formation energies among the different phases. 20 As shown in Fig. 1a,b, the rst layer of Ni(111) can easily bind with the adjacent carbon, which leads to a total transfer of 2.45 e − to the carbon surface.
We next shift our attention to study the effects of extra electrons on the Ni-N-C structures for CO 2 RR. Here, two representative structures including coordinatively saturated NiN 4 (Fig. S1) and unsaturated NiN 3 V are considered ( Fig. S2), 13,21,22 where V stands for vacancy and Ni single atom is the active site. 23 COOH* and CO* are intermediates for electrochemical CO 2 -to-CO conversion. 24 On neutral Ni-N-C catalysts, as shown in Fig. 1c,d, the rate-limiting steps are CO 2 (g)→COOH* on both NiN 4  Given that hydrogen evolution reaction (HER) is a dominant side reaction that is competitive with the CO 2 RR, we calculate the difference between thermodynamic limiting potentials for CO 2 RR and HER (denoted as UL(CO 2 )-UL(H 2 )), which can re ect the selectivity (Fig. S4). 25,26 Figure 1f shows the UL(CO 2 )-UL(H 2 ) on negatively charged NiN 4 and NiN 3 V, in comparison with the neutral cases. Clearly, negatively charged NiN 4 and NiN 3 V show more positive values of UL(CO 2 )-UL(H 2 ), corresponding to their higher selectivity for CO production.
Catalysts synthesis and characterizations. We pursued the nanoparticle co-catalysis effect experimentally by one-step solid-state pyrolysis with excessive Ni sources. The schematic diagram of the catalyst design and formation mechanism is shown in Fig. S5a. First, the melamine was decomposed into carbon nitride at ~400°C. 27 With the increase of annealing temperature, the Ni nanoparticles catalyzed the carbon nitrite to generate the carbon nanotubes (CNTs), 28 meanwhile Ni single atoms were trapped by the nitrogen-doped CNTs via thermal emitting. 29,30 The NiSA/NP was obtained after the heat treatment followed by the acid washing to remove the Ni nanoparticles outside of CNTs. Figure  The unpaired electron in catalysts was directly observed by EPR spectroscopy (Fig. 2e), where a g-value of 2.17 is obtained for both samples and could be assigned to the unpaired electron in the 3d x 2 − y 2 orbital of Ni(i). 32,33 The EPR signal intensity of NiSA/NP is above 2 times higher than NiSA, corresponding to the electron-rich structure due to the presence of Ni nanoparticles, consistent with the Bader analysis. We further examined the d band structure of Ni using ultraviolet photoelectron spectroscopy (UPS, Fig. 2f and Fig. S8). No obvious metallic Ni peak can be observed across the Fermi level, which can be attributed to the fully con ned Ni nanoparticles inside of CNTs that exceeded the detection limit of UPS (2-3 nm), in line with XPS results (Fig. S7a). Therefore, the shift of near Fermi level signals (0-1 eV) can be attributed to the change of electronic structure of Ni single atoms on the surface of CNTs. The position of NiSA increases in binding energy compared to NiSA/NP (away from the Fermi level). According to d band theory, the downshifted d band of NiSA (grey curve) corresponds to a weakened adsorption strength, 34 which is in good accordance with DFT calculation.
The Ni K-edge X-ray absorption near-edge structure (XANES) and extended X-ray absorption ne structure (EXAFS) measurements were carried out to identify the electronic structure and coordination environment of catalysts. Ni foil and NiPc were used as references for metallic Ni and Ni 2+ , respectively. The pre-edge curve of NiSA is in the middle of Ni foil and NiPc (Fig. 2g), suggesting a valence state of Ni close to +1. 35 Ni(i)-N-C has been revealed as a highly active site for CO 2 -to-CO conversion. 32 More importantly, the preedge position of NiSA/NP is between the Ni foil ad NiSA, corresponding to an even lower average valence state than Ni + . The Fourier-transformed k 3 -weighted spectra of the samples exhibit two major peaks at 1.45 Å and 2.1 Å, which are assigned to Ni-N and Ni-Ni coordination shells, respectively (Fig. 2h). The signi cantly decreased Ni-Ni peak intensity in NiSA compared with NiSA/NP corresponds to the removal of Ni nanoparticles. Note that the Ni-Ni peak can be easily observed even there are only residual Ni NPs in the catalysts. 31 Based on the EXAFS tting of Ni-N shell on NiSA, the coordination number is determined to be 2.7 which is close to the coordinative unsaturated NiN 3 V structure ( Fig. 2i and Table S2).
Electrochemical CO 2 conversion. The electrochemical testing was rstly conducted in an H-cell using 0.5 M KHCO 3 electrolyte and the corresponding calibration curve is shown in Fig. S9. Figure (Table S3). The Faradic e ciency of CO (FE CO ) is shown in Fig. S10, where NiSA/NP can achieve an ultra-high FE CO above 99% at -0.8 V. Besides, the onsite potential of NiSA/NP is at around -0.5 V vs RHE, which is approaching to the values predicted from DFT calculation on 2 e − negatively charged NiN 3 V model. The electrochemically active surface area (ECSA) is evaluated from double-layer capacitance (C dl ) to elucidate the origin of the enhanced catalytic activity on NiSA/NP (Fig. S11). 36 The NiSA/NP shows slightly higher C dl compared with NiSA (1.4 mF vs 1.2 mF), suggesting that the improved catalytic activity is primarily attributed to the intrinsic reactivity of each site, rather than a surface area effect (Fig. 3b). Stability testing is then performed at -0.75 V vs RHE, NiSA/NP shows robust durability for CO 2 RR during 30 hours of continuous electrolysis, maintaining constant current densities and 99% of the initial FE for CO production (Fig. 3c). This can be attributed to the strong immobilization and close proximity between catalysts and CNTs, which reduce aggregation and bypass undesired side reactions to suppress catalyst deactivation. 37 To assess the performance of catalysts at industrial-relevant devices and current densities, we further evaluate the CO 2 RR in ow cells using gas diffusion electrodes (Fig. S12). Figure 4a displays the CO FEs and partial current densities plotted against the iR-corrected potentials in 1 M KOH ow cells. The NiSA/NP shows a low onsite potential at around -0.25 V vs RHE. Then, the CO partial currents quickly increase to above 200 mA cm −2 at -0.4 V with FE CO of 96%, and further reach 346 mA cm −2 at -0.5 V with FE CO of 98%. As a comparison, the highest current obtained with NiSA is 263 mA cm −2 at -0.65 V (Fig.   S13). The superior CO 2 RR performance of NiSA/NP in the ow cell can be attributed to (i) the signi cantly enhanced CO 2 mass transport based on gas diffusion electrode and (ii) the catalytic promotion effect in alkaline electrolytes. 43 The full cell performance was then tested based on the zero-gap MEA device, which is regarded as a promising prototype for practical application because of the very low cell resistance and robust structure. anode. It is worth noting that the Ir-based anode can cost more than 60% of the total price of MEA device, 7 so that the development of the low-cost substitution is highly desirable. As shown in Fig. 4c, the NiSA/NP NiFe cell delivers a j CO of 163 mA cm −2 with FE CO of 92% at -2.0 V without iR compensation.
With further increase of cell voltage to -2.3 V, the j CO can reach up to 310 mA cm −2 with FE CO of ~99%, corresponding to an outstanding full cell energy e ciency of 57% based on the calculation: EE = ΔE 0 / ΔE Applied × FE CO , where EE is the energy e ciency, ΔE 0 is the equilibrium full cell voltage (E 0 CO -E 0 OER = -0.11 V -1.23 V = -1.34 V), and ΔE Applied is the applied full cell voltage without iR-correction. Note that certain Ag-based MEA devices can achieve current densities above 600 mA cm −2 but at a much higher cell voltage of -3.5 V with a low energy e ciency of ~35%. 38 In comparison, our NiSA/NP NiFe cell shows the highest energy e ciency and the lowest cell voltage ever reported for CO production at an industrially relevant current of ~300 mA cm −2 ( Fig. 4d and Table S4).

Discussion
In conclusion, a cooperative Ni SAC via nanoparticles coupling was rstly designed by DFT modelling, which showed electron enrichment of single-Ni-sites with enhanced binding to the key intermediate COOH*. The promotion effect of Ni nanoparticles was pursued experimentally by the solid-state pyrolysis with excessive Ni sources. Extensive structural characterizations revealed a low valence of Ni single atom with abundant unpaired d-electrons in the structure. As such, the greatly enhanced CO 2 RR activity on NiSA/NP is understood as a result of its tuned electronic con guration. Together with the populated Ni single atom (4.6 wt%), a state-of-the-art performance is achieved in the MEA full cell, which delivered a high CO current of 310 mA cm −2 at -2.3 V with an overall energy e ciency of 57%. We believe this Ni SACs that rivalled nanostructured Au and Ag could accelerate the industrial implementation of energye cient CO 2 RR. The proof-of-concept also opens a new avenue for the design of cooperative SACs, which can potentially break the linear scaling relationship for a range of technologically important catalytic reactions.

Methods
Synthesis of NiSA/NP catalyst. 0.3 g of Ni(NO 3 ) 2 ·6H 2 O and 3 g of melamine were manually ground for 10 min to get the uniform light-green solid precursor. It was then annealing in the tube furnace in Ar atmosphere at 1000°C for 2h with a heating rate of 5°C min −1 . The nal NiSA/NP catalyst can be obtained by washing the sintered powder with 3 M HCl and H 2 O to remove the impurities.
Synthesis of NiSA catalyst. The NiSA catalyst was prepared by the NH 4 Cl treatment method. 32 Namely, 0.1 g of NiSA/NP and 3 g of NH 4 Cl were manually ground for 10 min to get the uniform black precursor. It was then annealing in the tube furnace in Ar atmosphere at 1000°C for 2h with a heating rate of 5°C min −1 . The nal NiSA catalyst can be obtained by washing the sintered powder with 3 M HCl at 80°C and H 2 O to remove the impurities.
Characterizations. Scanning electron microscope (SEM) images were collected with a QUANTA 450.
Transmission electron microscopy (TEM), high-resolution TEM (HRTEM), high angle annular dark-eld scanning TEM (HAADF-STEM) were carried out on JEOL JEM-ARM200f microscope at 200 kV. XRD was performed on a PANalytical X'Pert X-ray diffraction system (45 kV, 40 mA, Cu Kα radiation). XPS results were recorded by Thermo ESCALAB250Xi. XAFS spectra at the Ni K-edge were collected in the Australia Synchrotron center in uorescence mode. EPR spectra were conducted on a benchtop spectrometer Bruker EMXnano with x band.
Electrochemical measurements. CO 2 electrolysis in H-cells was performed in a gas-tight H-cell with two compartments separated by a cation exchange membrane (Na on®117). A Pt plate was used as the counter electrode, a saturated calomel electrode (SCE) was used as the reference electrode, and CO 2saturated 0.5 M KHCO 3 was used as the electrolyte, respectively. To prepare the working electrodes, 10 mg of catalyst and 100 µL of 5% Na on solution were introduced into 100 µL of water and 300 µL of ethanol solution and sonicated for 1 h. A 6.25 µL of the catalyst ink was coated onto a carbon ber paper substrate and dried in the air, giving an effective area of 0.25 cm − 2 with catalyst loading of 0.5 mg cm − 2 . In H-cells, LSV and potentiostatic data were corrected with an iR compensation of 80%. CO 2 electrolysis in ow cells. The windows for electrolysis were set to 1 cm × 1 cm. Each chamber has an inlet and an outlet for electrolyte, and an Ag/AgCl reference electrode was placed in the catholyte chamber. The catalyst ink was prepared by mixing 10 mg of catalyst, 3 mL of ethanol and 100 µL of a Na on per uorinated resin solution. Then, catalysts were air-brushed onto 3 × 3 cm −2 38 BC gas diffusion layer (FuelCellStore) electrodes and used as the cathode. Commercial IrO 2 sprayed on Titanium mesh was used as a counter electrode for oxygen evolution reaction (OER). An anion exchange membrane (FAA-3-50, FuelCellStore) was used to separate the cathode and anode chambers. 1 M KOH solution was used as the electrolytes. The catholyte and anolyte were cycled at a ow rate of 10 mL min −1 by using a peristaltic pump. LSV and potentiostatic data were corrected with an iR compensation of 80%. CO 2 electrolysis in zero-gap MEA device. The windows for electrolysis were set to 1 cm − 2 . Cathode and anode chambers have an inlet and an outlet for gas and electrolyte, respectively. Both the cathode and anode catalyst ink were prepared by mixing 10 mg of catalyst, 3 mL of ethanol and 100 µL of a Na on per uorinated resin solution. Then, the cathode catalyst was air-brushed onto 3 × 3 cm − 2 38 BC gas diffusion layer (FuelCellStore) electrodes, and the anode catalyst was air-brushed onto 3 × 3 cm − 2 35 AA carbon paper (Ion Power) electrodes. The anode NiFe catalysts were prepared by our previously reported method. 44,45 An Sustainion® X37-50 was used as the membrane. The cathode was feed with humidi ed CO 2 gas and the anode was feed with 1 M KOH electrolytes. The applied cell voltages were recorded without iR correction.