Synthesis and characterization of CuSnx
The Cu-Sn bimetallic catalysts were synthesized by a rapid and facile method that based on simultaneous galvanic replacement reactions (Figure S2a). Driven by the difference in the redox potential, this galvanic replacement process is a powerful method for fabricating dendritic structures with controllable chemical composition.35 Once a piece of cleaned Zn foil was added into a mixed solution containing Cu2+, Sn2+ and HCl, the surface of silver gray Zn foil turned into black immediately, indicating the reaction took place rapidly (Figure S2b). The Zn foil almost disappeared and the color of the solution became light after 1 hour, indicating the deposition of Cu and Sn as well as the dissolution of Zn2+. The obtained solid products were emersed in diluted HCl solution and then washed with deionized water repeatedly to remove the superfluous Zn. The X-ray diffraction (XRD) patterns of CuSnx catalysts with x ≤ 0.10 showed similar diffraction peaks with bare Cu catalyst while no crystalline Sn species were detected (Figure S3a). As seen in Fig. 1a, with the increasing content of Sn, all the diffraction peaks shifted to lower 2θ angle and the peaks became widened, indicating the enlarged lattice spacing and decreased crystallinity and grain size.29 These results are likely due to the Sn atoms with larger atomic radius than Cu atoms are highly dispersed in the lattice of Cu, which will be further discussed below.36 Furthermore, the XRD pattern of CuSn0.25 catalyst showed that the intermetallic phase corresponding to Cu41Sn11 emerged when the Sn content further increased to 25% (Figure S3b)
The morphologies of the CuSnx bimetallic catalysts were first characterized by scanning electron microscope (SEM). For comparison, bare Cu sample was also prepared by the same method without Sn2+ addition. The bare Cu counterpart exhibited irregular rod-like morphology (Figure S4a). Interestingly, with only small amount of Sn addition, the microstructure of CuSn0.01 catalysts changed drastically and exhibited typical dendritic morphology (Fig. 1b), indicating that Sn2+ ions may facilitate the formation of dendritic morphology in the replacement process. With the increasing Sn content, the dendrite surfaces became coarsening and small grains emerge (Figure S4). The dendrite structure is beneficial to CO2RR process by lowering the charge transfer resistance and enhancing the mass transfer.34, 37 Representative high-resolution transmission electron microscope (HR-TEM) images showed good crystalline of CuSn0.01 catalyst (Figure S5b). The zoomed-in edge of CuSn0.01 catalyst exhibited distorted lattice fringe, which is likely due to the heteroatomic doping of Sn in Cu lattice (Fig. 1c). The elemental mapping of CuSn0.01 catalyst showed the predominant Cu and highly scattered Sn elements were uniformly distributed in the catalyst (Fig. 1d). On the other hand, the TEM images of CuSn0.10 catalyst showed that the coarsened dendrite surface become assembly of small particles and a distinct amorphous layer of about 5 nm formed on the catalyst surface(Figure S6a-c). The elemental mapping of CuSn0.10 catalyst showed a uniform distribution of both Cu and Sn elements (Figure S6d-g). The double layer capacities of the CuSnx catalysts were then measured to determine the electrochemical surface area (ECSA). Compared to bare Cu, the introduction of Sn increased the ECSAs of the CuSnx catalysts (Figure S7). The CuSnx catalysts exhibited similar ECSA except for CuSn0.25, which may be due to the severe coarsening of CuSn0.25.
The X-ray photoelectron spectroscopy (XPS) measurement was then carried out to investigate the surface properties of CuSnx catalysts. The Cu 2p1/2 and Cu 2p3/2 peaks of the CuSnx catalysts located at 952.4 eV and 932.5 eV, respectively, which can be assigned to either Cu0 or Cu1+ (Fig. 1e). The Cu LMM Auger electron spectra suggested that Cu1+ was the main Cu species in bare Cu catalysts due to the spontaneous oxidation under ambient condition. However, with the increasing Sn content in CuSnx, the valance of Cu gradually shifted to metallic Cu0, which is likely due to the stronger O affinity of Sn than Cu (Fig. 1f). The doublet peaks of Sn 3d (Fig. 1g) that can be assigned to the 3d3/2 and 3d5/2 peaks centered at 494.6 and 486.3 eV was in between the peaks of Sn2+ and Sn4+, indicating the oxidation state of Snδ+ in the CuSnx catalysts. The peak shift for both Cu and Sn demonstrated the electronic interaction between incorporated Sn and Cu matrix in the CuSnx catalysts. Furthermore, a small peak at about 485.0 eV that assigned to metallic Sn0 in Sn 3d5/2 peak (493.2 eV for Sn 3d3/2 peak) appeared in CuSn0.10 and CuSn0.25 catalysts. The composition of the CuSnx catalysts was characterized by both energy dispersive spectroscopy (EDS) and XPS. Given the nanoscale of the CuSnx catalysts, the Cu/Sn ratios determined by EDS reflects the bulk composition of the catalysts, which were close to the designed values (Fig. 1h). However, the surface and subsurface Cu/Sn ratio in depth of a few nanometers determined by XPS measurements were quite different from the EDS results. The Cu/Sn ratio in the CuSnx catalysts decreases in the order: surface (XPS) > subsurface (XPS after Ar sputtering) > > bulk (EDS). These results indicate that the Sn species enriched in the surface of the dendrite CuSnx catalysts, which is crucial for the catalytic reaction.33
CO2RR performance
The CO2RR performance of the dendrite CuSnx catalysts were evaluated in a H-cell divided by a Nafion membrane. The total FEs of all products on different CuSnx catalysts were in the range of 93–101% (Figure S8). The FEs of different reduction products are plotted against potentials to show the change of product distributions with CuSnx catalysts (Fig. 2 and Figure S9). The competing HER was largely suppressed for all the CuSnx catalysts at the tested potentials apart from only a few exceptions at -0.7 VRHE (Figure S9a). As reported by many previous studies, the introduction of Sn greatly improved the selectivity of Cu-based catalysts and CO became the major reduction product at low and moderate overpotentials.32 Among all the CuSnx catalysts, CuSn0.01 exhibited the highest FE of 96.4% for CO with a partial current density of 6.50 mA/cm2 at -0.8 VRHE (Fig. 2a and Figure S10b). The FEs of CO decreased gradually with the increasing Sn content and the maximum FEs of CO for CuSn0.025, CuSn0.05, CuSn0.10, and CuSn0.25 were 80.6%, 71.7%, 59.8%, and 53.7%, respectively. As another 2e transfer products, the FEs of formate increased with the increasing Sn content in CuSnx catalysts, which is due to the surface-enriched Sn and the high formate-producing activity of Sn in CO2RR (Fig. 2b). As mentioned before, the suppressed HER and enhanced 2e products (CO and formate) have been extensively observed and discussed by many studies on bimetallic Cu-Sn catalysts. Sn metal possess a weaker H affinity than Cu metal and SnO2 modification was reported to lower the H2 chemisorption of CuO.32, 38 Therefore, it is commonly accepted that the surface decorated Sn atoms on Cu disfavor the adsorption of *H and thereby suppress the HER.37, 39 Meanwhile, the formation of the important *COOH intermediate is facilitated, while the adsorption of CO is either weakened or unaffected by the combination of Cu and Sn.30, 31, 39 Therefore, the enhanced production of 2e products is usually reported in previous studies about Cu-Sn bimetallic catalysts. Similarly, A recent study showed that Cu99Sn1 catalyst with only 1% of Sn exhibited a high CO selectivity over 90% while Cu70Sn30 catalyst exhibited a high formate selectivity at low and moderate overpotentials.33
We then continued the CO2RR tests at more negative potentials where deep reduction products are generated. The bare Cu catalyst exhibited the highest FEs and current densities of C2H4 in the whole potential range (Fig. 2c and e). Clearly, the formation of C2H4 on Cu catalysts was suppressed by Sn modification. With the increasing Sn content in CuSnx catalysts, not only the FEs of C2H4 decreased, but also the onset potentials for C2H4 formation lowered. The onset potentials of C2H5OH on the CuSnx catalysts were also lower than that on the bare Cu catalyst. However, the FEs of C2H5OH on CuSn0.01, CuSn0.025 and CuSn0.05 catalysts were higher than that on Cu catalyst at potentials below − 1.2 VRHE (Fig. 2d and Figure S11). The highest FE of C2H5OH reached 25.93% on CuSn0.025 at -1.4 VRHE with a large partial current density of 15.05 mA/cm2 (Fig. 2d and f). Previous studies of Cu-Sn bimetallic catalysts for CO2RR always exhibited high FEs for CO or formate at potentials higher than − 1.1 VRHE (Table S1).30, 33 Herein, it was found that C2H4 activity was suppressed while C2H5OH production was boosted at relatively large overpotentials on our CuSnx catalysts.
Bifurcation between C2H4 and C2H5OH
To gain further insight into the selectivity trend of C2H4 and C2H5OH on CuSnx catalysts, the FE ratios of C2H5OH/C2H4 of different CuSnx catalysts were compared at the potentials ranging from − 1.1 VRHE to -1.4 VRHE (Figure S12). Obviously, the FE ratios of C2H5OH/C2H4 of the CuSnx catalysts were higher than that of the bare Cu catalyst and showed good linearity against the surface Sn/Cu ratio (CuSn0.25 catalyst was not included in Fig. 3a since the deviation is relatively large, which is likely due to its low FEs for both C2H4 and C2H5OH). To understand the unusual C2 products selectivity, we first sought mechanistic insight into the bifurcation between C2H4 and C2H5OH. Although the pathways for the formation of C2H4 and C2H5OH have not reached a consensus, it has been evidenced by experimental and theoretical results that C2H4 and C2H5OH products branch after C-C coupling.4, 16 Koper and co-workers have previously shown that H2C-CHO* is the selectivity-determining intermediate (SDI) that bifurcating C2H4 and C2H5OH.14 The hydrogenation of Cα in H2C-CHO* leads to H3C-CHO*, which is further reduced to C2H5OH. On the contrary, the hydrogenation of Cβ forms H2C-CH2O*, which leads the formation of C2H4.10, 17 Based on these previous studies, we carried out theoretical calculation of the Gibbs free energy of these two pathways. Given the feature of incorporated Sn in CuSnx and the high C2 selectivity of Cu(100) facet, slab models of Sn substituted Cu(100) facets were built and Sn-Cu(100) slab was selected for free energy calculation (Figure S13). As seen in Fig. 3b and c, on Cu(100), the energy barriers for C2H4 and C2H5OH pathways were 1.54 eV and 1.77 eV respectively, suggesting the C2H4 formation is thermodynamically more favorable. In contrast, the energy barrier of C2H5OH pathway was 0.22 eV lower than that of C2H4 pathway on Sn-Cu(100). The calculation results revealed that the C2H5OH pathway on Cu sites was enhanced by Sn modification, which is in accordance with the FEs trends.
In the whole potential range, the introduction of Sn into Cu significantly improved the production of CO, which is believed to be the key intermediate for C-C coupling. The enhancement of local CO concentration could facilitate C2+ production via tandem effect on bimetallic catalysts.40 However, the reversed selectivity tendencies of C2H4 and C2H5OH in this work ruled out the tandem effect in our CuSnx catalysts. The surface oxophilicity of catalysts is of great importance for the processes that involves oxygen relevant species, such as *O, *OH, and adsorbed oxygenic intermediates.28, 41, 42 The binding strength of O bonded SDI of C2H4 and C2H5OH could be influenced by the oxophilicity of the binding sites.23, 26 To this end, we first tried to understand how the surface oxophilicity changed with the component of the CuSnx catalysts. To accomplish that, cyclic voltammetry (CV) tests in 0.1 M NaOH solution were performed to determine the binding strength of adsorbed hydroxide ions (OHad) on the CuSnx catalysts, which reflects their oxophilicity.42, 43 As depicted in Fig. 4a, the CV curve of bare Cu catalyst showed Cu0-Cu1+ redox peaks at around 328.6 and 577.7 mVRHE.44 In addition, a peak associated with the adsorption of OHad appeared at around 360 mVRHE (OHad-I).44 The zoomed-in OHad-I peaks on bare Cu catalyst exhibited typical OHad peaks obtained on polycrystalline Cu, where a strong OHad peak corresponding to Cu(100) is at 359.1 mVRHE, followed by two weak OHad peaks assigned to Cu(110) and Cu(111) (Fig. 4b).45, 46 The introduction of Sn altered the adsorption behavior of OHad-I on Cu sites. On the CuSnx catalyst, there was only a strong OHad-I peak that shifted to more negative potentials, indicating the enhanced adsorption strength of OHad (Fig. 4d and Figure S14).42 The adsorption strength of OHad-I reached the maximum on CuSn0.025 catalyst, on which the onset and vertex of OHad-I peak negatively shifted to 343.1 and 352.4 mVRHE, indicating the strongest oxophilicity of Cu sites on CuSn0.025 (Fig. 4d).45 With the increasing Sn content in the CuSnx catalysts, the intensity of OHad-I peak decreased and an additional broad peak at around 150 mVRHE (OHad-II) appeared on CuSn0.05 and CuSn0.10 catalysts (Figure S14b-d). Given the stronger O affinity of Sn than Cu, the intensified OHad-II peak with the increasing Sn content should be assigned to the strong adsorption of OHad sites of Sn. For CuSn0.25 catalyst, only a broad peak at 100 ~ 200 mVRHE could be observed, indicating that strong adsorption of OHad on Sn sites prevailed when the surface is abundant in Sn (Figure S4d and h). These results demonstrate that the introduction of Sn on Cu surface enhances the oxophilicity of the CuSnx catalysts in two ways: 1) enhanced oxophilicity of Cu sites; 2) creation of new stronger oxophilic Sn sites.
Based on the above results, we tried to elucidate the importance of surface oxophilicity to C2H5OH production in CO2RR. The FEs of C2H5OH at -1.3 and − 1.4 VRHE on the CuSnx catalysts were plotted against the vertex potentials of OHad-I peaks in the CV curves, which reflects the surface oxophilicity of the CuSnx catalysts (Fig. 4e). The great linearity strongly supported the positive correlation between the oxophilicity of the Cu sites and C2H5OH production (Bare Cu and CuSn0.25 catalysts are not included in Fig. 4e because the multiple OHad-I peaks in bare Cu catalyst and the indiscernible OHad-I peak in CuSn0.25 catalyst). We then sought theoretical calculation for understanding the effects of oxophilicity on the SDI. As seen in Fig. 4f, the stable SDI bonded to Cu sites on both Cu(100) and Sn-Cu(100) and the calculated adsorption energy were − 1.77 eV and − 1.85 eV respectively, indicating the enhanced oxophilicity of Cu sites could stabilize the adsorption of SDI. As discussed above, the next proton and electron transfer step of the SDI bifurcates the C2H4 and C2H5OH pathways. The charge analysis of adsorbed SDI showed that the Cα carried less positive charge and the Cβ also carried less negative charge on Sn-Cu(100), which both contribute the next protonation on Cβ to be easier on Sn-Cu(100). Therefore, we believe that the enhanced oxophilicity of Cu sites in CuSnx catalysts stabilize the SDI and affect its protonation to facilitate the C2H5OH pathway in CO2RR.
To further verify the effect of surface oxophilicity on CO2RR, we combined Cu with Pb due to its modest oxophilicity between Cu and Sn and suitable redox property (Figure S15). The CuPb0.025 catalyst was prepared by the same method and their CO2RR performance were tested at -0.7 to -1.3 VRHE. The CuPb0.025 catalyst exhibited a higher FE of C2H5OH than bare Cu catalyst. Although its FEs of C2H5OH were not as high as CuSn0.025, the FEs of C2H4 for CuPb0.025 was comparable or even superior to bare Cu catalyst and the maximum FEs of C2+ for CuPb0.025 reached 70.78% at -1.2 VRHE. We then synthesized CuAg0.025 bimetallic catalyst by the similar method as Ag possesses lower oxophilicity than Cu (Figure S16). Both the FEs of C2H4 and C2H5OH on CuAg0.025 catalyst was lower than that on bare Cu catalyst. These results strongly support the surface oxophilicity regulation strategy for improving C2H5OH production in CO2RR. Interestingly, there was no n-C3H7OH product that could be detected on CuAg0.025 catalyst at any potentials, while a new doublet assigned to H in acetaldehyde emerged in the 1H NMR patterns (Figure S16f). Although the mechanism of the acetaldehyde formation is rarely reported, it is believed that acetaldehyde formed on the C2H5OH pathway.47 Such interesting selectivity behaviors of CuPb0.025 and CuAg0.025 catalysts imply that the effect of surface oxophilicity cannot be accomplished alone without other effects in bimetallic catalysts, such as electronic and strain effects.11, 21 Therefore, it is believed that this strategy can cooperate with other ways that facilitate the C-C coupling, such as tandem effect and Cu1-Cu0 synergistic effect,6, 40 to further increase the production of C2+ oxygenic products in CO2RR.