The intrinsic reasons for the enhanced activity and selectivity for CO2 reduction to C2+ products over Cu-nr-OR were further investigated. The electrochemical active surface areas (ECSA) of Cu-nr and Cu-nr-OR were estimated by measuring double layer capacitance. We can observe that the ECSA of Cu-nr-OR was similar to that of the Cu-nr (Supporting Information, Figure S8). Thus, the improved C2+ products generation resulted not mainly from the slight change of the ECSA. Furthermore, electrochemical impedance spectroscopy (EIS) was carried out to measure the charge transfer resistance (Rct) for Cu-nr and Cu-nr-OR. The Cu-nr-OR showed similar interfacial Rct with Cu-nr(Supporting Information, Figure S9).
Due to the selectivity of the C2+ products and activity are sensitive to the surface species of catalysts, the role of surface species in enhancing generation of C2+ products was explored by in situ SERS (Supporting Information, Figure S10). For the Cu-nr-O, two broad bands at 524 and 610 cm-1 were observed (Figure 3A), which were attributable to the Cu2O, indicating that Cu2O existed in Cu-nr-O. This was consistent with the results of XAS. We can observe that the corresponding Raman features of Cu2O disappeared at -0.2 V vs. RHE or below, indicating full reduction of Cu-nr-O to Cu. New bands appeared at -0.2 V vs. RHE or below during CO2RR, the bands will be analyzed according to the different potentials in the following.
At -0.2 V vs. RHE or below, a well-defined bands appeared at 1064 cm-1 on Cu-nr-OR and Cu-nr during CO2RR (Figure 3A and Supporting Information, Figure S11), which was attribution to CO32-,[12] becuase CO2 can dissolve in the KOH electrolyte, forming a neutral-pH carbonate mixture. The intensity of the band became weak with the potential decrease, which is because the CO2 can be reduced at negetive potential and the formation of CO32- become slow.
At -0.3 V vs. RHE or below, an additional band appeared at 524 cm-1 for Cu-nr-OR and Cu-nr during CO2RR. According to the previous report,[39] the band can be assingned CuOx or CuOxOHy species. When conducting in D2O instead of H2O, the band at 524 cm-1 displayed a negligible shift (Supporting Information, Figure S12), indicating that the band was attributed to CuOx species rather that the CuOxOHy. There are three possible reasons for formation of CuOx species (Figure 3C): (1) the CuOx species were from the original Cu2O; (2) from the reaction between reduced Cu and H2O; and (3) from the reaction between reduced Cu and CO2.
To address these concerns, 18O isotope labeling experiments were carried out to confirm the source of oxygen in CuOx. First, 18O enriched Cu-nr-O catalysts were synthesized by oxidation of Cu-nr in H218O electrolyte. The Cu218O was formed in the Cu-nr-O, the bands were 504 and 590 cm-1 (Figure 3B), which exhibited significant red-shift compared with the Cu216O.[27] This result indicates that the bands of CuOx species showed a significant red shift when the 16O was replaced by 18O, which can be an important indicator for exploring the oxygen source of CuOx species. For the 18O enriched Cu-nr-OR, the CuOx was also formed at -0.3 V vs. RHE, and the band was still at 524 cm-1, which displays a negligible shift compared with that of the Cu-nr-OR. The results suggest that the CuOx species were not from the original Cu218O. Thus it can be deduced that the CuOx was produced during the CO2RR. Due to both the H2O and CO2 can react with reduced Cu to form CuOx species, the oxygen source of CuOx species should be further studied. Furthermore, the Cu-nr-OR were tested in H218O during CO2RR, we can observe that the band of CuOx was still at 524 cm-1 (Supporting Information, Figure S13), which indicated that the CuOx was not from the reaction between Cu and H218O. Thus the last possible reason is reasonable, i.e., the CuOx was from the reaction between Cu and CO2. Because the CO2 reduction is slow at -0.3 V vs. RHE and the signal of CO cannot be observed in raman spectra, we can assume that the CuOx species were from the chemisorption of CO2 on Cu. To further verify this argument, the electrolysis experiment over Cu-nr-OR was tested under N2 atmosphere, no bands were observed at negative potential (Supporting Information, Figure S14), indicating that the CO2 played the key role for the formation of CuOx species.
From the above results, we can deduce that the CuOx existed during CO2RR was just the singal of chemisorption of CO2 on Cu, which was not the main factor for facilitating C2+ products during CO2RR. In order to further verify this conclusion, the commercial Cu nanoparticles (about 50 nm) were also studied by in situ SERS (Supporting Information, Figure S15). It was shown that the CuOx appeared at -0.3V vs. RHE or below (Supporting Information, Figure S16), which was similar to that of Cu-nr-OR, thus the CuOx species were not specific for the Cu oxide derived catalysts.
At -0.4 V vs. RHE or below, for both the Cu-nr and Cu-nr-OR, the presence of adsorbed *CO on Cu was demonstrated by the appearance of Raman peaks located at 276, 360, and 2000-2100 cm-1, which correspond to the restricted rotation of adsorbed *CO on Cu, Cu-CO stretching, and C≡O stretching, respectively (Figure 3A, and Supporting Information, Figure S11).[40] It is interesting to note that the band of C≡O stretching on Cu-nr-OR is different from that on Cu-nr (Figure 3D). A new peak appeared at about 2000 cm-2 on Cu-nr-OR compared with the Cu-nr. Specifically, the stretch mode of surface-adsorbed CO can serve as a molecular probe of surface structure due to its sensitivity to the structure of adsorption sites.[41-42] Thus, we can assume that new active sites were produced on Cu-nr-OR. It is reasonable to analyze the active sites using the surface-adsorbed CO at -0.4 V vs. RHE, becuase the C-C coupling step is slow at this potential. According to previous report,[43] different CO adsorption sites exhibit distinct catalytic behavior for the C-C coupling step, thus we can suppose that the enhanced generation of C2+ products over Cu-nr-OR was originated mainly from the formation of new active sites.
After the potential was removed, we could observe that the Cu2O was formed rapidly (Supporting Information, Figure S17), indicating that the reduced Cu can be oxidized in electrolyte. Cu2O can be formed via oxidation of Cu by the O2 in electrolyte.[27] However, the content of O2 is very low in the cathodic electrolyte. It is interesting to note that the Cu218O was formed when using H218O as electrolyte after the potential was removed (Supporting Information, Figure S18). Thus we can assume that the Cu2O was from the reaction between Cu and H2O. After the potential was removed, the reduced Cu was very active, which could react with the H2O. The results indicate that the in situ method has obvious advantage and is necessary to explore the real state of the catalyst in the reaction.
Due to the coordination environment of Cu can alter the adsorption of CO and the energy barrier of C-C coupling step, we used operando XAS to monitor the local structure of Cu-nr and Cu-nr-OR during CO2RR. For both catalysts, only peaks corresponding to metallic Cu were observed at negative potential during CO2RR (Figure 4A, B and Supporting Information, Figure S19-S20), indicating that the CuO or Cu2O was reduced to metallic Cu in CO2RR. Moreover, the quantified Cu-Cu coordination number of the Cu-nr-OR and Cu-nr were fit using the ARTEMIS programs of IFEFFIT during CO2RR (Supporting Information, Figure S21-S22 and Table S2). No obvious difference of Cu-Cu coordination number was observed for the Cu-nr-OR and Cu-nr during CO2RR, indicating that the enhancing of C2+ products over Cu-nr-OR was not mainly from the slight change of the coordinate environment.
The surface structure of the catalysts can be probed by electrosorption of hydroxide (OHads).[44] Qualitatively, for the Cu-nr, the intensity of (111) OHads feature was higher than that of (100) and (110), which suggests a high surface density of (111) on Cu-nr (Figure 4C). In contrast, for the Cu-nr-OR, the intensity of (111) OHads peak is reduced, which reflects that the proportions of (100) and (110) on Cu-nr-OR surface were higher that on Cu-nr. According to previous report,[32, 45] the (100) facet was favorable crystal orientation for the C-C coupling process. And the CO dimerization reaction (Figure 4D, and Supporting Information, Figures S23-25, Table S3) on Cu (111) and Cu (100) were investigated by DFT calculation, which is crucial for producing C2+ products.[46-49] The Cu (100) exhibited lower energy barrier for CO dimerization compared to Cu (111) (Figure 4E), suggesting that Cu (100) can enhance the formation of C2+ products. Thus, the increased Cu (100) facet can be considered as the main factor for enhancing C2+ products over Cu-nr-OR.
In summary, the surface species and structure over OD-Cu catalysts were systematically investigated by in situ SERS, operando XAS, and 18O isotope labeling experiments combined with theoritical calculation. It was showed that the Cu oxides indeed existed on the surface of catalysts at during CO2RR. However, they were formed by chemisorption of CO2 on Cu instead of the active sites of the catalyst. The redox cycling treatment could create active Cu (100) facet, and DFT calculations suggested that the Cu (100) active facets could decrease the energy barrier of C-C coupling step, and enhancing C2+ products. In addition, this work also shows that in situ techniques have obvious advantages and are sometimes necessary to explore the structure of the catalyst and surface species in CO2RR. We believe that the findings of this work provide useful knowledge for designing other efficient electrocatalysts for CO2 reduction.