CO2 Electroreduction towards CO with Near-unity Selectivity on NiCu-embedded ZIF-derived N-doped Carbon Nanoparticles

The electrocatalytic reduction of CO 2 towards CO is one of the most desirable routines to reduce atmospheric CO 2 concentration and maintain a global carbon balance. In this work, a novel porous NiCu-embedded ZIF-derived N-doped carbon nanoparticles (NiCu@NCNPs) catalyst has been identi�ed as an active, highly selective, stable, and cost-effective catalyst in CO 2 reduction. A CO selectivity as high as 100% has been achieved on NiCu@NCNPs which is the highest reported to date. The particle current density of CO on NiCu@NCNPs is around 15 mA cm –2 under the optimized potential at -0.9 V vs. RHE. The NiCu@NCNPs electrode also exhibits excellent stability during the �ve sequential CO 2 electroreduction experiments. The superior catalytic performance of NiCu@NCNPs in CO 2 RR can be related to its microstructure with high electrochemical surface area and low electron transfer resistance. Furthermore, a kinetic analysis has shown the formation of intermediate *COOH is the rate-determining step in CO 2 RR towards CO. According to the results of density functional theory (DFT) calculations, a low Gibbs free energy change ( ∆ G) for the rate-determining step leads to the enhanced catalytic performance of CO 2 RR on NiCu@NCNPs.


Introduction
The tremendous increase of anthropogenic activities such as fossil fuel burning since the industrial revolution have resulted in the accumulation of carbon dioxide in the atmosphere, which will give rise to serious environmental concerns (Davis et al. 2010;Porosoff et al. 2016).The electrochemical reduction of carbon dioxide will not only reduce the atmospheric CO 2 concentration maintaining a carbon-neutral energy cycle, but also produce fuels or other value-added chemicals (Pan and Yang 2020).In particular, the electrochemical carbon dioxide reduction reaction (CO 2 RR) to CO is a promising pathway because the product CO can be utilized as a feedstock in Fischer-Tropsch technology to produce high-quality fuels and high-end chemicals (Gong et  The fundamental bottleneck for the practical implementation of electrocatalytic CO 2 reduction is the activation of inert C = O bonds, which results in a relatively low conversion e ciency of CO 2 and low product selectivity (Sheng and Sun 2017).The e cient reduction of CO 2 to CO under electrocatalytic conditions remains challenging and enormous effort has been taken towards the design of a catalyst with high CO selectivity (Kattel et al. 2017).Previously, noble metals (like Ag (Lu et al. 2014), Au (Kaneco et al. 1998) and Pd (Huang et al. 2017)) have been reported with relatively high e ciency in converting CO 2 to CO.However, these metals all display disadvantages of scarcity, high price, and low stability limiting their large-scale application.Additionally, forementioned noble metal catalysts generally present a 50 ~ 60% CO selectivity from CO 2 RR which will bring di culty in the following separation procedures and increase the industrial operating cost.Therefore, this work is dedicated to designing low-cost catalysts for CO 2 RR and improving the activity and CO selectivity.
Heteroatom-doped carbon materials are known to be with high atomic e ciency and low coordination metal centers (Cui et al. 2018;Liu et al. 2020b;Wang et al. 2014), and therefore provide the potential to not only reduce the consumption of the heteroatom, but also improve the catalytic e ciency in CO  According to numerous prior investigations, bimetallic catalysts tend to outperform their parent metals on the same support due to the ligand effect and strain effect (Kitchin et al. 2004;Yu et al. 2012).The ligand effect results from the electronic interaction from the formation of heteroatomic bonds, whereas the variation in bond lengths between different metals causes the strain effect.For example, in previous study about CO 2 RR, PtCo/CeO 2 bimetallic catalyst has presented a higher activity and CO selectivity than Pt/CeO 2 and Co/CeO 2 (Porosoff and Chen 2013).With this in mind, in this work dual-metal-doped porous carbon catalysts will be applied for the electrochemical reduction of CO 2 .In order to reduce the cost of the catalyst, low-cost transition metals Ni and Cu are chosen to prepare NiCu-co-doped carbon nanoparticles (NiCu@NCNPs) and to turn the selectivity of CO in CO 2 RR.

Catalyst Synthesis
Preparation of NiCu-ZIF.0.2529g NiNO 3 •6H 2 O and 0.2101g Cu(NO 3 ) 2 •3H 2 O were ground in a mortar and pestle for 10 minutes to achieve complete mixing, and the resultant powder was dissolved in 100 mL of methanol solution to obtain a mixture (tagged as mixture A) which was then sonicated for complete dissolution.Then 9 g Zn(NO 3 ) 2 •6H 2 O and 20 g 2-methylimidazole were dissolved in 800 mL of methanol to obtain another mixture (marked as mixture B).Afterwards the sonicated mixture A was slowly and uniformly added into the stirring mixture B to obtain a new mixture C.After mixing, the mixture C was stirred at high speed for 4 h.The resulting mixture C was centrifuged for 3 minutes at 11000 rpm to collect the precipitation, and the precipitate was then wash with methanol.This process was repeated for ve times to wash the sample thoroughly, and the nal precipitate was uniformly coated on the wall of a cup and dried under vacuum at 80°C for 12 h.
Preparation of NiCu@NCNPs.500 mg as-prepared NiCu-ZIF was added into 250 mL of a mixture of water and methanol (water : methanol = 10 : 1), then 125 mg cetyltrimethylammonium bromide, 50 mg sodium hydroxide and 1 mL of ethyl orthosilicate were added to the mixture, which was then sonicated for 30 min and stirred for 1h.The solution was centrifuged at 10000 rpm for 1 min to collect the precipitate and the precipitate was washed with ethanol 5 times.The nal precipitate was layered evenly on the wall of a cup to dry it under vacuum at 80°C for 12 h, afterwards the product was grinded for 10 minutes and calcined in a tube furnace at 900°C under nitrogen for 3 hours, etched with HF (15 wt%) solution for 3 hours and washed with water three times.Control samples Ni@NCNPs and Cu@NCNPs with the same mole of metal loadings were also synthesized using similar experimental process.Similarly, Ni 3 Cu 1 @NCNPs was prepared by adding 0.7587g NiNO The material was prepared into electrodes as follows: 2mg of NiCu@NCNPs material and 30uL of Na on solution were added into1mL of ethanol solution.The resulting mixture was sonicated for 2 h to obtain the catalyst ink.Then 500 µL catalyst ink was steadily and evenly dripped onto the carbon cloth using a pipette gun.Both sides of the carbon cloth should be dripped.Finally, the carbon cloth was placed on a small sheet of tin foil and dried in an oven overnight with a catalyst loading of 1.0 mg cm -2 .

Physical Characterizations
A scanning electron microscope (SEM, FEI Scios HiVac, Hitachi, Japan) and a transmission electron microscope (TEM, FEI TF20, Netherlands) were used to observe the surface morphology of electrodes.
The phase of the sample was determined by X-ray diffraction (XRD, PAN-alytical, Netherlands) with working voltage at 45 kV and the working current at 40 mA.X-ray photoelectron spectroscopy (XPS, Thermo Scienti c K-Alpha, America) were obtained using a Thermo Scienti c K-Alpha spectrometer with monochromatized Al Kαradiation operated at a power of 1486.6 eV.

CO 2 Electroreduction
The reaction was carried out in a H-type reactor where the cathode chamber KHCO 3 electrolyte was bubbled with CO 2 for at least 30 minutes and air over the solution was supposed to be emptied entirely, as reported in previous work (Yu et al. 2021b).In the H-reactor reactor, the as-prepared electrode, an Ag/AgCl electrode and a platinum electrode were used as working electrode, reference electrode and counter electrode, respectively (Yu et al. 2021a).During the reaction, products were analyzed every 15 min.CO and H 2 are the two major products with no methanol, formate or methane detected.
The Faradaic e ciency (FE) and partial current density (PCD) of each product were calculated using the equations as follows: Here, z denotes the number of electrons obtained or lost in the reaction, n represents the amount of reaction product material, Q stands for the actual amount of electricity passing through the circuit and F is the electrochemical constant (96485 C/mol).
In this work, the applied voltage was converted to the corresponding reversible hydrogen electrode (RHE) with iRu compensation considered using E (vs.RHE) = E (vs.Ag/AgCl) + 0.059 × pH + 0.1976 -iRu.
3 Results And Discussion 3.1 Characterization results illustrating electrode morphology and structure SEM and TEM have been applied to explore the morphology and microstructure of NiCu@NCNPs as well as corresponding control samples Cu@NCNPs and Ni@NCNPs electrodes.The SEM and TEM results of the prepared NiCu@NCNPs electrode are shown in Figs.1a and 1b, which exhibit monodispersed spherical porous particles.According to literature, the surface micropores are formed by the evaporation of the zinc species during the high temperature pyrolysis step, thus leaving micropores in the carbon matrix, and are externally manifested as tiny micropores on the surface of the material (Yang et al. 2018a).The Ni@NCNPs and Cu@NCNPs obtained by the same preparation method yield similar results as shown in Figure S1, with an average particle size of 50 nm and clear spherical particles as NiCu@NCNPs.The corresponding energy dispersive X-ray spectroscopy (EDS) images demonstrate that the Ni and Cu elements are equally dispersed and precisely matched on the mapping.
The elemental composition and valence states of various elements were determined by XPS with surface sensitivity.Figure 2a shows the spectra of Cu 2p spectrum for Cu@NCNPs and NiCu@NCNPs.As illustrated in Fig. 2a, two deconvoluted peaks are observed at 932.7 eV and 934.9 eV for Cu@NCNPs which are attributed to Cu 2+ and Cu + /Cu 0 species, respectively.Similar peaks are observed on the electrode of NiCu@NCNPs in Fig. 2a demonstrating that doping another metal will not change the valence state of Cu species on the surface of the catalyst.

Electrochemical measurements comparing electrode catalytic performance
Electrochemical experiments have been performed on various electrodes to compare the catalytic performance in CO 2 RR.Faradaic e ciency (FE) and particle current density (PCD) are used to describe the product selectivity and activity from CO 2 RR.As described in the experimental section above, catalysts with various Ni/Cu mole ratios were prepared and they were all tested for the electroreduction of CO 2 .To illustrate the results more evidently and explicitly, NiCu@NCNPs and control samples Cu@NCNPs and Ni@NCNPs were chosen as examples in Fig. 3a to display the pro le of CO 2 electroreduction selectivity under different potentials.As can be seen from Fig. 3a, the value of FE CO on Cu@NCNPs at rst grows with a continuing increase in the applied potential and reaches the highest value (55.9 ± 3.5%) at the optimal potential between − 0.78 V and − 0.90 V vs RHE.It then drops afterward because of the signi cant enhancement in the competing hydrogen evolution reaction (HER), consistent with the observed increasing FE of H 2 when the potential applied is higher than the optimal potential, as shown in Fig. 4. A similar trend is observed about FE values from CO 2 RR on the Ni@NCNPs catalyst where the optimum FE CO of 78.2 ± 2.4% is achieved at -0.9 V vs RHE.Surprisingly, different with the other two catalyst, the electrode with NiCu@NCNPs in this work displays a reversal outstanding performance.A high CO Faradaic e ciency of 93.1 ± 1.2% can be obtained at a low potential of -0.65 V vs RHE for NiCu@NCNPs, even higher than the optimal values of FE CO on Cu@NCNPs and Ni@NCNPs, indicating that NiCu@NCNPs is more selective in CO production and consumes less energy in CO 2 RR.An optimal CO FE around 100% can be achieved on NiCu@NCNPs with trace production of H 2 after the applied potential reaches at -0.9 V vs RHE.It seems that − 0.9 V vs RHE is the optimal potential for all three samples.Additionally, the CO FE remains almost unchanged in the voltage range of -0.9 V to -1.14V vs RHE, which signi es that at the appropriate voltage range, NiCu@NCNPs can signi cantly suppress HER and produce almost pure CO from CO 2 RR.This indicates the probability that on NiCu@NCNPs, CO 2 can be completely converted to the required CO product in a proper voltage region without following separation procedures.
In order to nd out the ideal Ni/Cu mole ratio in carbon material for CO 2 RR, the electrocatalytic performance of electrodes with various Ni/Cu mole ratios was examined under the optimal potential at -0.9V vs. RHE.As shown in Fig. 3b, the CO Faradaic e ciencies of Ni@NCNPs, Ni 3 Cu 1 @NCNPs, NiCu@NCNPs, Ni 1 Cu 3 @NCNPs and Cu@NCNPs samples were compared to illustrate the optimal Ni/Cu mole ratio for CO 2 RR at -0.9 V vs RHE.It is obvious that dual-metal-doped catalysts possess higher selectivity than monometallic ones, and among the various dual-metal-doped catalysts, NiCu@NCNPs exhibit the highest CO Faraday e ciency (nearly 100%), indicating the optimal Ni/Cu mole ratio is 1.
Figure 3c compares the PCD values of CO on all three materials with the change in applied potential.As shown in Fig. 3c, the current density of CO on NiCu@NCNPs is 8 times the value on Cu@NCNPs and 4 times that on Ni@NCNPs at the optimized potential at -0.9 V vs. RHE.The results in Fig. 3c indicate that compared with the other two electrodes, NiCu@NCNPs has exhibited enhanced activity in CO 2 RR towards CO.
Besides of activity and product selectivity, the catalytic stability is also an important criterion in evaluating its catalytic performance in CO 2 RR.The stability of the NiCu@NCNPs electrode has also been evaluated using ve successive CO 2 RR experiments at the optimized applied potential.It is demonstrated in Fig. 3d that the CO selectivity of the material keeps almost the same during the ve cycles of CO 2 RR experiments, verifying that the NiCu@NCNPs electrode exhibits excellent stability during the electrochemical reduction of CO 2 .
In order to illustrate this point more clearly, XPS tests were performed for the NiCu@NCNPs sample before and after CO 2 RR. Figure S2 shows the spectra of N 1s spectrum for NiCu@NCNPs before and after the reaction separately.The N spectrum after the reaction exhibited 5 deconvoluted peaks assigned to pyridinic-N (398.4 eV), M-N (399.3 eV), pyrrolic-N (400.2 eV), graphitic-N (401.2 eV), and oxidized-N (404.1 eV) which stay same with the N spectrum of NiCu@NCNPs before reaction.Which indicates that the structure of the NiCu@NCNPs sample was not altered throughout the process of CO 2 RR.

The correlation between the structure of electrodes and corresponding catalytic performance
The excellent electrocatalytic effect of NiCu@NCNPs is directly related to its structure.Combined with various characterization results, conclusions can be easily drawn that NiCu@NCNPs is a crystalline material with an organized pore structure, allowing the active center of the material to be exposed at a high density and therefore being well suited for catalytic applications.Furthermore, the low dimensionality of NiCu@NCNPs enhances its external speci c surface area, further optimizing the electrocatalytic e ciency.In order to illustrate this point more clearly, cyclic voltammetry experiments have been performed to compare the electrochemically active surface areas (ECSA) on all three materials.Figure S3 shows cyclic voltammograms for three catalysts at various scan rates.To compare the ESCA of three materials, double layer capacitance (C dl ) measurements were carried out.As clearly illustrated in Fig. 5, NiCu@NCNPs shows the highest C dl value among all three catalysts.Speci cally, the C dl value of NiCu@NCNPs is at 2.858 mF cm − 2 , which is roughly 3.3 times that of Ni@NCNPs (0.878 mF cm − 2 ) and 8.2 times that of Cu@NCNPs (0.349 mF cm − 2 ).The ESCA value for each electrode can be calculated from the formula: and the Cs in the work is 20µF cm −2 .Thus, the calculated ECSA values on the three electrodes follow this trend: NiCu@NCNPs (142.5 cm − 2 ) > Ni@NCNPs (43.9 cm −2 ) > Cu@NCNPs (17.5 cm −2 ).Based on this result, the introduction of NiCu diatomic metals contributes to more surface-active sites, thus leading to the enhanced catalytic performance in CO 2 RR.
In order to further explore the potential mechanisms leading to the superiority of the electrode with NiCu@NCNPs over the other two materials, electrochemical impedance spectroscopy (EIS) has been employed to comprehensively understand the reaction kinetics of electrochemical CO 2 RR over the three catalysts.As shown in Fig. 6, the impedance arc in the Nyquist curve of NiCu@NCNPs is much smaller than that of the control samples, implying that NiCu@NCNPs presents the lowest electron transfer resistance during CO 2 reduction.The results in Fig. 6 lead to the conclusion that the NiCu@NCNPs electrode features the swiftest interfacial charge transfer process followed by Ni@NCNPs and Cu@NCNPs, which is consistent with their experimental results shown in Fig. 3.
As discussed above, on one hand, the introduction of NiCu diatomic metals contributes to the formation of more surface-active sites, which will facilitate the adsorption of CO 2 and corresponding reaction intermediates on the surface of the electrode.On the other hand, a lower impedance on the NiCu@NCNPs electrode leads to a faster electron transfer during the electrochemical reduction of CO 2 .Both points will result in a higher CO 2 reduction e ciency.

Mechanism of enhanced catalytic performance on NiCu@NCNPs
As shown in previous literature (Liu et al. 2019;Liu et al. 2020a;Rosen et al. 2015), the electrocatalytic reduction of CO 2 to CO involves three steps: CO 2 gas molecules are rst adsorbed and protonated to *COOH on active sites (*) of the catalyst surface.Afterwards *COOH is converted to *CO in the presence of electron, which will ultimately desorb from the surface as gaseous CO.This process can be described as follows: Tafel analysis and DFT calculations have been performed to further obtain a profound understanding of the reaction kinetics of electrochemical CO 2 RR over the three catalysts towards CO.It has been concluded from previous studies that a Tafel slope of 118 mV dec − 1 means that the rate-determining step in CO 2 reduction is the rst electron transfer step (Yang et al. 2018b).As shown in Fig. 7, the Tafel slope of NiCu@NCNPs is very close to 118 mV dec − 1 , so the *COOH generation step with the rst proton-coupled electron transfer is the rate-determining step.In contrast, the Tafel slopes for Ni@NCNPs and Cu@NCNPs On this basis, DFT calculations were carried out to theoretically explore the reaction mechanism of CO 2 electroreduction on these three electrodes in this work.Figures 8a-8c show the reaction pathways of CO 2 RR on Cu@NCNPs, Ni@NCNPs and NiCu@NCNPs, respectively.The optimized con gurations of important intermediates *COOH and *CO are also shown on each catalytic surface.
Figure 8d is the Gibbs free energy diagrams for CO 2 reduction to CO.As shown in Fig. 8d, the rst reaction step about CO 2 activation for *COOH production presents an uphill potential for each electrode, while the *CO production step and CO desorption step are thermodynamically downhill and can proceed spontaneously.Among all three steps, the rst step displays the highest Gibbs free energy difference (∆G).Therefore, the step involving intermediate *COOH formation is the rate-determining step which agrees well with the results from Tafel analysis.Figure 8d also demonstrates that for the rate-determining step about the formation of *COOH, NiCu@NCNPs presents the lowest ∆G of 1.19 eV, followed by Ni@NCNPs (2.0 eV) and Cu@NCNPs (2.04 eV), which will lead to the highest CO 2 RR activity and CO selectivity on NiCu@NCNPs.According to the results of Tafel analysis and DFT calculations, the lowest ΔG for the rate-determining step on NiCu@NCNPs among all three catalysts leads to its superior catalytic performance in CO 2 RR.

Conclusions
In summary, a porous, low-dimensionally organized NiCu@NCNPs material has been successfully synthesized and applied in the e cient electroreduction of CO 2 to CO.The diatomic NiCu@NCNPs catalyst exhibits superior catalytic activity and higher CO selectivity compared to Ni@NCNPs and Cu@NCNPs.A CO selectivity as high as 100% is achieved on NiCu@NCNPs which is the highest reported to date.The superior catalytic performance of NiCu@NCNPs in CO 2 RR could be related to its microstructure with higher electrochemical surface area and lower electron transfer resistance.The combination of Tafel analysis and DFT calculations indicate that the underlying mechanism of remarkable catalytic e ciency of NiCu@NCNPs in CO 2 RR is due to the lowest ∆G for the rate-determining step generating the *COOH intermediate.NiCu@NCNPs, Ni 1 Cu 3 @NCNPs Cu@NCNPs under -0.9 V vs. RHE.(c) Partial current densities of CO at different potentials for Cu@NCNPs, Ni@NCNPs, and (d) Catalytic stability test for NiCu@NCNPs at−0.9V (vs RHE) for 5 cycles.

2 RR(
Duan et al. 2017; Liu et al. 2018; Pan et al. 2020).As shown in previous work, metal − organic frameworks (MOFs), due to the advantages such as varied backbone, large speci c surface area, con gurable pore structure, and a diverse range of functional groups (Al-Rowaili et al. 2018; Shao et al. 2020), have been widely utilized as precursors to obtain various novel porous carbon materials with excellent catalytic activities.For example, using ZIF-8, one kind of MOFs composed of Zn and 2methylimidazole, as precursor (Jiang et al. 2018; Yan et al. 2018), Pan and his co-workers have synthesized a monoatomic nickel linked carbon material, and the resultant catalyst has demonstrated enhanced reactivity and selectivity (Pan et al. 2019).And therefore, in this work, ZIF-derived porous carbon catalysts will be utilized for the electrochemical reduction of CO 2 .
Figure 2b indicates that both NiCu@NCNPs and Ni@NCNPs possess + 2 valence Ni species (856.1 eV).The N1s peaks of all three samples are depicted in Fig. 2c, with ve deconvoluted peaks for N assigned to pyridinic-N (398.4 eV), M-N (399.3 eV), pyrrolic-N (400.2 eV), graphitic-N (401.2 eV), and oxidized-N (404.1 eV).The results observed in Fig. 2c are consistent with those in literature (Cheng et al. 2020; Hu et al. 2019), con rming the successful synthesis of heteroatom-embedded ZIF-derived N-doped carbon materials.In addition, XRD was adopted to determine the composition of materials and the physical phases.As shown in Fig.2d, two wide peaks of modest intensity are observed around 24° and 44° in the XRD patterns of all three samples, which are consistent with previous report(Li et al. 2021), con rming the M-N-C structure on all samples.However, no metal diffraction peak regarding Ni or Cu is found indicating the absence of metal particles.That means Ni and Cu coexist across the entire carbon matrix in the form of atoms.
→ C O (g) + * are 142 mV dec − 1 and 163 mV dec − 1 , respectively, which indicates that the reaction rates are in uenced not only by the rst proton-coupled electron transfer to generate the *COOH intermediate, but also by some other factors such as the mass transfer(Liu et al. 2017; Shi et al. 2020; Yang et al. 2018a).

(
a) SEM and (b) TEM images of NiCu@NCNPs as well as (c) corresponding EDS mappings of C, N, Ni, and Cu elements.

Figure 2 High
Figure 2