Enhancing CO2 electroreduction to CH4 over Cu nanoparticles supported on N-doped carbon

The electroreduction of CO2 to CH4 has attracted extensive attention. However, it is still a challenge to achieve high current density and faradaic efficiency (FE) for producing CH4 because the reaction involves eight electrons and four protons. In this work, we designed Cu nanoparticles supported on N-doped carbon (Cu-np/NC). It was found that the catalyst exhibited outstanding performance for the electroreduction of CO2 to CH4. The FE toward CH4 could be as high as 73.4% with a high current density of 320 mA cm−2. In addition, the mass activity could reach up to 6.4 A mgCu−1. Both experimental and theoretical calculations illustrated that the pyrrolic N in NC could accelerate the hydrogenation of *CO to the *CHO intermediate, resulting in high current density and excellent selectivity for CH4. This work conducted the first exploration of the effect of N-doped species in composites on the electrocatalytic performance of CO2 reduction.


Introduction
The electrochemical CO 2 reduction reaction (CO 2 RR) is a promising approach to achieve carbon neutrality, 1-6 which can convert CO 2 into valuable chemicals and fuels using renewable energy. [7][8][9][10][11] Among all these products, CH 4 is a highly desired product because it holds the highest heating value of 55.5 MJ kg À1 . 12 However, CO 2 electroreduction to CH 4 still suffers from high overpotential and low activity and selectivity. [13][14][15][16] Designing highly efficient and robust electrocatalysts is crucial to solve this problem.
Cu-based catalysts have been proven to be the most promising electrocatalysts for producing hydrocarbon products from the CO 2 RR. In recent years, many methods have been applied to enhance the activity of the CO 2 RR over Cu-based catalysts, including alloying, 17 doping, 18,19 modifying with other compounds, 20,21 changing the shape and size, [22][23][24] and building an interface and defects. [25][26][27] However, it is still challenging to achieve high selectivity for CH 4 at a high current density. [28][29][30][31][32][33] This is because the generation of CH 4 involves eight electrons and four protons, which easily bifurcates to give broad product distributions. [34][35][36] According to a previous report, 37 the coadsorption of *CO and *H played an important role in the production of CH 4 , and the selectivity of CH 4 can be enhanced by a high surface *H coverage, due to sufficient *H supply for the hydrogenation of intermediates. 37 However, the high surface *H coverage could result in the undesired hydrogen evolution reaction (HER) and hinder the adsorption of intermediates. Thus, it is necessary to nd a method to break the linear scale relationship of the single catalyst sites, which will enhance the selectivity of CH 4 and decrease the yield of H 2 simultaneously.
To solve the above problem, here we proposed the idea to introduce additional catalytic sites, which can enhance the activation of H 2 O and hydrogenation of intermediates but not cause excessive production of H 2 . According to a previous report, N-doped carbon (NC) catalysts could promote the activation of H 2 O. [38][39][40] In addition, the adsorption of *H can be regulated by changing the type of N-doped species. [41][42][43] Thus we can assume that NC would be a potential platform for tuning the activity and selectivity of Cu-based catalysts. Although NC has been used to modify Cu-based catalysts, the products were ethanol and ethylene. 41 In addition, the role of the N-doped species in the composites is not clear.
Herein, we used NC with different contents of N-doped species as another component to modify Cu nanoparticles (Cu-np/NC). In this strategy, the selectivity of CH 4 over Cu-np could be enhanced by introducing NC with rich pyrrolic N species, and the faradaic efficiency (FE) of CH 4 could reach up to 73.4% with a current density of 320 mA cm À2 . Especially, the mass activity reached up to 6

Results and discussion
Cu-np and NC were prepared respectively according to previous reports. 44,45 The size of the obtained Cu-np was about 40 nm (Fig. S1 †). NC(x : y) was prepared using arginine and melamine, 45 and x : y represents the ratio of arginine and melamine in the preparation process. All of the NC(x : y) exhibited a nanosheet morphology (Fig. S2 †). Cu-np/NC(x : y) was prepared by blending Cu-np and NC(x : y) (see the Methods for experimental details).
The electrocatalytic performance of Cu-np/NC(1 : 4) was rst studied, and the other Cu-np/NC(x : y) catalysts will be studied in the following section. The electrocatalytic performance of the CO 2 RR was evaluated in a ow cell, as reported in our previous report. 34,41 The catalysts were sprayed on hydrophobic carbon paper as the cathode, and Ni foam was used as the anode. 34,41,46,47 1 M KOH solution was used as the electrolyte. The gaseous and liquid products were analyzed by gas chromatography (GC) and nuclear magnetic resonance (NMR) spectroscopy, respectively.
Over Cu-np, CO 2 could be reduced to various products, such as CO, CH 4 , C 2 H 4 and C 2 H 5 OH ( Fig. 1A and S3 †). However, the selectivity of a single product is low (less than 45%). It is noted that the selectivity of CH 4 was signicantly enhanced by adding NC(1 : 4) and increased with the content of NC(1 : 4) ( Fig. 1A and S4 †). The selectivity of CH 4 reached the highest when the content of NC(1 : 4) was 99 wt% (Cu-np/NC(1 : 4, 99 wt%)). The FE of CH 4 could reach up to 73.4%, which is much higher than that of Cu-np. When the content of NC(1 : 4) was 99.5 wt%, the selectivity of CH 4 decreased with increasing yield of H 2 , which may have originated from the insufficient Cu sites. For NC(1 : 4), only small amounts of CO were detected (Fig. S5 †). Consequently, it can be assumed that the Cu sites were the active sites and the selectivity of CH 4 was enhanced by the addition of NC(1 : 4). The partial current density of CH 4 over Cunp/NC(1 : 4, 99 wt%) was 234 mA cm À2 at À1.1 V vs. RHE, which is about 4.1 times higher than that over Cu-np (Fig. S6 †). In the meantime, the CH 4 -to-other ratio was enhanced from 0.25 on Cu-np to 20.2 on Cu-np/NC(1 : 4, 99 wt%) ( Fig. S7 †), indicating that the selectivity of CH 4 was enhanced and the other products were suppressed. Compared with the state-of-the-art catalysts, Cu-np/NC(1 : 4, 99 wt%) performed as one of the best catalysts in FE, current density and overpotential for CH 4 ( Fig. 1B and Table S1 †). In addition, based on such a low content (1 wt%) of Cu-np in Cu-np/NC(1 : 4, 99 wt%), the mass activity could be as high as 6.4 A mg Cu À1 . In the following discussion, the content of NC was xed at 99 wt% in Cu-np/NC(x : y).
The electrolysis of the CO 2 RR was also carried out using membrane electrode assembly-based reactors (Fig. S8 †). Cu-np/ NC(1 : 4) was used as the cathode catalyst for the CO 2 RR and an iridium oxide-based catalyst as the anode for the oxygen evolution reaction. The selectivity of CH 4 was 60% with a current density of 230 mA cm À2 at a cell voltage of À4 V ( Fig. 1C and S9 †). Furthermore, the stability of Cu-np/NC(1 : 4) was investigated at a cell voltage of À4 V. H 2 O was injected into the cathode ow channel to prevent salt accumulation in the gas diffusion layer (GDL) micropores. The selectivity of CH 4 and the current density had no obvious change for 50 h (Fig. 1D), indicating that Cu-np/NC(1 : 4) exhibited excellent stability.
The electrocatalytic performance of the CO 2 RR over other Cu-np/NC(x : y) was also evaluated in a ow cell. As shown in Fig. S10, † the selectivity of CH 4 over Cu-np/NC(x : y) varied with the x : y value. The highest FEs of CH 4 over Cu-np/NC(1 : 2) and Cu-np/NC(1 : 8) were 65.8% and 59.6% respectively. Thus, we can deduce that NC(x : y) played an important role in the selectivity of CH 4 . According to a previous report, 45 the intrinsic properties of NC were mainly attributed to the N-doped species. The correlation between CH 4 selectivity and the type of N-doped species was investigated (Fig. S11, S12 and Table S2 †). The FE of CH 4 increased with increasing pyrrolic N content, whereas no regularity can be found for pyridinic N and graphitic N. These results indicated that the pyrrolic N in the NC(x : y) may play a crucial role in the enhancement of selectivity of CH 4 .
In addition, the electrochemically active surface areas (ECSAs) and Nyquist plots of Cu-np/NC(x : y) were measured. The charge transfer resistance (Rct) for the different Cu-np/ NC(x : y) was similar (Fig. S13 †). Although the ECSAs of Cu-np/ NC(x : y) varied slightly with the different NC(x : y) (Fig. S14 †), the normalized partial current densities for CH 4 by ECSAs were similar to the geometric partial current density (Fig. S15 †). These results indicated that different CO 2 RR performances of Cu-np/NC(x : y) with different x : y values were not originated from the slight change of the R ct and ECSAs.
Cu-np/NC(1 : 4) was characterized by transmission electron microscopy (TEM), and we can observe that Cu-np was dispersed on NC(1 : 4), as shown in Fig. 2A and B. In addition, the lattice distance of Cu(111) was observed by high-resolution transmission electron microscopy (HR-TEM) (Fig. 2C), which is consistent with that in Cu-np. However, the characteristic peaks of Cu cannot be observed on Cu-np/NC(1 : 4) in XRD patterns, and this is because the content of Cu is too low (Fig. S16 †).
X-ray photoelectron spectroscopy (XPS) and operando X-ray absorption spectroscopy (XAS) were carried out to monitor the valence state and coordinate environment of Cu during the CO 2 RR by the method used in our previous study. 34,41,48 From XPS, we can observe that the Cu valence state in Cu-np and Cunp/NC(1 : 4) was similar, which is attributed to Cu 0 (Fig. S17 †). As shown in X-ray absorption near edge structure (XANES) spectroscopy (Fig. 2D), the pre-edge peaks of Cu-np and Cu-np/ NC(1 : 4) were close to Cu foil before reaction. When the potential (À0.7 V vs. RHE) was applied, the spectra of Cu-np and Cu-np/NC(1 : 4) were still similar to that of metallic Cu. According to extended X-ray absorption ne structure (EXAFS) spectroscopy ( Fig. 2E and S18 †), only a peak corresponding to the Cu-Cu bond was observed, indicating that the metallic Cu was the active site for Cu-np and Cu-np/NC(1 : 4) during the CO 2 RR. Furthermore, the Cu-Cu coordination number of Cu-np and Cu-np/NC(1 : 4) during the CO 2 RR were quantied by using the ARTEMIS programs of IFEFFIT (Fig. S19, S20 and Table  S3 †).The Cu-Cu coordination number and bond distance in Cunp/NC(1 : 4) were close to that in Cu-np during the CO 2 RR. These results indicate that the addition of NC did not change the coordination properties of Cu during the CO 2 RR.
DFT calculations were then carried out to gain insights into the effect of the N-doped species in Cu-np/NC on the selectivity of CH 4 . Cu(111) was used to represent Cu-np (Fig. S21 †), which is in accordance with the results of HR-TEM. Cu(111) was located on a layer of N-doped graphene (NG) to represent the model of Cu-np/NC (Fig. S22 †). From the results above, the pyrrolic N species played a crucial role in the selectivity of CH 4 . Then, the reaction energy diagrams of CO 2 reduction to CH 4 were rst characterized over Cu(111) and Cu(111) on pyrrolic Ndoped graphene (Cu(111)/pyrrolic N).
As shown in Fig. 4A, S23 and S24, † CO 2 was rst reduced to *CO through the *COOH intermediate, and then *CO was further reduced to CH 4 through the *CHO intermediate. On Cu(111), the hydrogenation of *CO to *CHO shows the highest energy barrier (0.71 eV), which is considered as the rate-limiting step for producing CH 4 . The hydrogenation of CO 2 to *COOH and hydrogenation of *CO to *CHO were promoted over Cu(111)/pyrrolic N. Although the hydrogenation of *CO to *CHO still shows the highest energy barrier over Cu(111)/ pyrrolic N, it was only 0.30 eV, which was much lower than that over Cu(111). These results suggested that the reduction of CO 2 to CH 4 over Cu(111) can be signicantly enhanced by combining with pyrrolic N doped NC. Furthermore, the reaction energy diagrams were characterized at À0.5 V applied potential (Fig. S25 †), and Cu(111)/pyrrolic N also is more favorable for producing CH 4 than Cu(111).
In addition, the hydrogenation of *CO to *CHO was also studied over Cu (111)/NC with different N-doped species (Fig. 3B and S26-S29 †), and H 2 O was used as the donor of hydrogen, because 1 M KOH solution was used as the electrolyte in the CO 2 RR. Compared with Cu(111), the formation of *CHO and *OH from *CO and *H 2 O can be enhanced over Cu(111)/ graphitic N, Cu(111)/pyrrolic N and Cu (111)/pyridinic N. It is noted that Cu(111)/pyrrolic N exhibited the lowest energy barrier (À0.16 eV), indicating that the pyrrolic N played the main role in the outstanding activity and selectivity of CH 4 , which was consistent with the experimental results. According to a previous report, 49 the adsorption of *CO played an important role in the selectivity of products. Thus, the adsorption of *CO on different models was studied. The adsorption of *CO on Cu(111), Cu(111)/graphitic N, Cu(111)/ pyrrolic N and Cu(111)/pyridinic N was comparable (Fig. 3C), indicating that the adsorption of *CO was not changed by adding NC. Thus we can assume that the outstanding performance for CH 4 over Cu-np/NC was attributed to the activation of H 2 O by pyrrolic N.
From the results of DFT calculations above, we can know that the pyrrolic N in NC can enhance the activation of H 2 O and accelerate the hydrogenation of intermediates. To explore the effect of H 2 O activation on the generation of CH 4 , the kinetic isotopic effect (KIE) of H/D over Cu-np/NC(x : y) catalysts was measured ( Fig. 4A and S30 †). The KIEs of H/D are dened as the ratio of CH 4 formation rates in H 2 O and D 2 O. It has been reported that the reaction is considered to be controlled by the primary isotope effect when the KIE value is greater than 2. 49,50 The KIE value was 2.9 over Cu-np, suggesting that the activation of H 2 O was involved in the rate-determining step. The KIE value over Cu-np/NC(x : y) decreased with the increase of the content of pyrrolic N in NC, suggesting that the dissociation of H 2 O can be enhanced by the pyrrolic N. For Cu-np/NC(1 : 4), the content of pyrrolic N reached the highest, and the KIE value was about 1.6, indicating that the dissociation of H 2 O was no longer involved in the rate-determining step, which is consistent with the results of DFT. Thus it can be deduced that the pyrrolic N in NC can accelerate H 2 O activation.
The role of H 2 O activation in the generation of CH 4 was further studied by investigating the effect of alkali metal (M) cations in a MOH electrolyte. It is known that the cation can combine with H 2 O to form a hydrated cation of M + (H 2 O)n, and the value of n was 13, 7 and 6 for Na + , K + and Cs + , respectively. 49 The radii of M + (H 2 O)n decrease in the order of Na + > K + > Cs + . 51 The smaller n and radii of M + (H 2 O)n enable a greater ability to dissociate H 2 O. 49,51 It can be known that the formation rate of  CH 4 was improved markedly over Cu-np/NC(1 : 4) by changing the cation from Na + to K + (Fig. 4B). The formation rate of CH 4 in CsOH was smaller than that in KOH. This may be because the generation of H 2 was also enhanced (Fig. S31 †). For Cu-np, the formation rate of CH 4 increased slightly when changing the cation from Na + to K + . Thus we can assume that NC can enhance CH 4 formation by promoting H 2 O activation through interaction with hydrated cations.
Furthermore, the reaction intermediates during the CO 2 RR were traced by in situ surface-enhanced Raman spectroscopy (SERS). 34,41 At open-circuit potential (OCP), no Cu x O species were observed on Cu-np and Cu-np/NC(1 : 4), which was consistent with the results of XPS and XAS. The peaks located at 1336 cm À1 and 1580 cm À1 were observed on NC and Cu-np/ NC(1 : 4), which were assigned to the D band and G band of graphene, respectively. [52][53][54] Weak peaks were also observed on Cu-np, which may be from the carbon paper. It is noted that a new Raman peak located at 526 cm À1 appeared on Cu-np and Cu-np/NC(1 : 4) at À0.3 V vs. RHE, which was attributed to the adsorption of preliminary intermediates (such as *CO 2 or *OCO À ) on the Cu surface. 55 These results indicated that the activation of CO 2 occurred on Cu sites. For Cu-np, a new Raman band located at 1895 cm À1 appeared at À0.3 V vs. RHE, which corresponded to the C^O stretching on Cu. 41 In contrast, no adsorption of *CO was observed on Cu-np/NC(1 : 4). This may be because the obtained *CO can be consumed quickly. Compared with Cu-np, Cu-np/NC(1 : 4) showed three other new peaks located at 567 cm À1 , 1430 cm À1 and 1660 cm À1 at À0.3 V vs. RHE, which may be attributed to the adsorption of *OH, *HCOH and *CHO on the Cu surface. 50,54,55 Thus we can deduce that the reaction *CO + *H 2 O / *CHO + *OH can be accelerated over Cu-np/NC and thus the selectivity of CH 4 can be enhanced, which was consistent with the experimental and calculation results (Fig. 5).

Conclusions
In summary, Cu-np/NC composites were designed to enhance the electroreduction of CO 2 to CH 4 . The selectivity of CH 4 reached 73.4%, with a current density of 320 mA cm À2 . Based on experimental and theoretical studies, the effect of the N-doped species on the electrocatalytic performance of CO 2 reduction was elucidated. The pyrrolic N in NC can enhance the activation of H 2 O, which could accelerate the hydrogenation of intermediates, resulting in excellent selectivity for CH 4 and high current density. We believe that this work opens a way for the design of efficient catalysts for the electroreduction of CO 2 to CH 4 .

Data availability
The data that support the ndings of this study are available within the article and its ESI. †

Conflicts of interest
There are no conicts to declare.