Nanoporous Au-Sn with solute strain for simultaneously enhanced selectivity and durability during electrochemical CO₂ reduction

1. Introduction

The level of carbon dioxide (CO₂) in the atmosphere has rapidly increased for decades due to huge consumption of fossil fuels, causing a serious concern for global warming [1,2]. Electrochemical carbon dioxide reduction (CDR) is a promising solution to circumvent this climate issue by converting CO₂ into useful chemicals in mild environments (room temperature and atmospheric pressure) [3,4]. However, as a multi-step reaction with sluggish kinetics, most of metallic catalysts for CDR suffer from low efficiency, poor selectivity, and rapid deactivation [5,6]. An effective strategy is to design highly active and robust electrocatalysts to strengthen the adsorption of intermediate by tuning electronic structure, such as alloying or strain effect [[7], [8], [9], [10]].

In comparison with single component catalysts, alloying provides a promising method to tailor the geometric and electronic environments of active sites [11]. Electronic structure changes typically observed on alloying involve a shift in the d-band position relative to the Fermi level, which can be used to tune the adsorption of intermediate on catalyst surface [12]. Monodisperse Au-Cu bimetallic nanoparticles with different compositions were found to have synergistic geometric and electronic effects during the CDR [13,14]. Cu-In and Cu-Sn bimetallic catalysts can generate CO with a Faradaic efficiency (FE) greater than 90 % by inhibiting the formation of adsorbed H* [15,16].

However, the catalytic performance enhancement in bimetallic catalysts may arise from two aspects: one is the new phase formed in the alloy; the other is surface strain induced from lattice mismatch [17,18]. Specifically, the contribution of surface strain to high activity of alloy catalysts should not be ignored, which is a powerful method to regulate the catalytic properties of heterogeneous catalysts by modifying the electronic structures [19,20]. For instance, the Cu/SnO₂ core-shell structure with thin oxide thickness shows highly selectivity to CO due to the synergy of uniaxial compress strain of SnO₂ and self-doping of Cu atoms [21]. The Cu-Ag bimetallic electrocatalysts, due to their surface strain in compressive state, exhibit an unusually high selectivity in multi-carbon production, compared to pure copper [22]. Thus far, the difficulty in distinguishing the surface strain from alloying effect impeded an unambiguous understanding of the intrinsic mechanism in electrochemical reduction of CO₂. Herein we design a nanoporous Au-Sn bimetallic catalyst (NPAS) with a trace amount of tin atoms as a solute in Au lattices. Uniform strain distributions on the surface of three-dimensional nanoporous skeleton provide an ideal platform to identify real strain effect on catalytic selectivity and durability. Nanoporous gold (NPG) catalyst without Sn solute was also fabricated for comparison.

2. Experimental

2.1. Fabrications of the NPAS and NPG catalysts

Ag₆₅Au₃₄Sn₁ and Ag₆₅Au₃₅ precursors were prepared by arc melting pure Ag(Shenyang Dongchuang Precious Metals Material CO., LTD, 99.99 %), Au (Shenyang Dongchuang Precious Metals Material CO., LTD, 99.99 %) and Sn (Sinopharm Chemical Reagent Co., LTD, 99.999 %) with a desired atomic ratio under argon atmosphere. After the verification of composition by energy dispersive X-ray spectroscopy, the alloy ingot was cut into thin plates with dimensions of 10 × 6×0.20 mm³ using a precision wafering machine. Then the thin plates were rolled at room temperature to get a ribbon with a thickness of ∼25 μm. In the following step, the ribbon was cut into the size of ∼ 5 × 5 mm² and corroded into 16 mol/L HNO₃ at -25 °C for 24 h. The NPAS and NPG catalysts were carefully rinsed with deionized water for several times to remove the residual chemicals.

2.2. Microstructural characterizations

X-ray diffraction (XRD) measurements were carried out on a Rigaku DMAX/2400 X-ray diffractometer with Cu Kα radiation. SEM observations were performed with a field-emission scanning electron microscope (JSM-7600 F). A JEOL JEM-2100 F transmission electron microscope with an acceleration voltage of 200 kV was employed to characterize the structure of nanoporous catalysts. The surface atomic structure of NPAS and NPG were characterized by using spherical-aberration-corrected TEM in a scanning TEM (STEM) mode with a high-angle annular dark-field (HAADF) detector. X-ray photoelectron spectra were obtained in AXIS-ULTRA DLD-600W XPS system with an Al Kα (mono) anode. The surface valence band spectra of nanoporous catalysts were collected by high-resolution XPS with the energy range of 0-20 eV, and each spectrum was collected over ∼1000 scans.

2.3. Electrochemical characterizations and product analysis

Nanoporous Au-Sn (NPAS) with minor Sn atoms was fabricated by corroding Ag₆₅Au₃₄Sn₁ alloy ribbon in HNO₃ solution at -25 °C for 24 h. Nanoporous gold (NPG) was also prepared by corroding Ag₆₅Au₃₅ in the same condition for comparison. All electrochemical measurements were performed with a CHI 760E electrochemical workstation with a three-electrode configuration. A Pt mesh was used as counter electrode. All potentials were measured versus an Hg/HgCl reference electrode (saturated KCl, TJ Aida R0232) and converted to RHE reference scale using E (versus RHE)=E (versus SCE)+0.0591×pH + 0.244 V. Electrolysis was performed in a gas-tight electrochemical cell with two compartments separated by a piece of anion exchange membrane (Nafion 212, DuPont). Each compartment cell contained 29 mL electrolyte and ∼15 mL headspace. Before electrolysis, the KHCO₃ electrolyte (Sinopharm Chemical Reagent Co., Ltd, ≥99.5 %) was purged with CO₂ bubbles (Xiangyun gas incorporations, 99.995 %) for at least 30 min. During the electrolysis, the KHCO₃ electrolyte in cathodic compartment was stirred at a rate of 1200 RPM using a magnetic stirrer. Cathodic compartment cell with working electrode was connected with the gas-sampling loop of a gas chromatograph (GC-9790II, FULI analyzing incorporations) for the continuous cycling. Gas-phase product was automatically sampled every 15 min for quantitative analysis. Argon was used as carrier gas during the measurement. The GC columns led directly to a thermal conductivity detector (TCD) to quantify H₂ and a flame ionization detector (FID) equipped with a methanizer to quantify CO. Gas concentration was determined by the stand curves calibrated from standard gas with different concentrations. The HCO₃^- concentration dependence was mixed in 0.1∼1.0 M NaClO₄ (Sinopharm Chemical Reagent Co., Ltd, ≥99.0 %) that was utilized to keep a constant ionic concentration in the low-concentration electrolytes.

The liquid product were quantified by NMR (Bruker avance III 500 MHz) spectroscopy, in which 0.5 mL electrolyte was diluted with 0.1 mL D₂O (deuterated water) and 0.05 μL dimethyl sulfoxide (DMSO, Sigma, 99.99 %) as an internal standard. The one-dimensional ¹H spectrum was measured with water suppression using a pre-saturation method.

Underpotential deposition of lead was tested in 0.1 M NaOH +1 mM Pb(NO₃)₂ electrolyte at a scan rate of 5 mV/s. The electrolyte was purged with pure N₂ for 30 min prior to measurements. The Hg/HgO (1 M KOH, TJ Aida R0501) electrode was selected as reference and converted to the potnetiall by E (versus RHE) =E (versus Hg/HgO) +0.0591×pH + 0.098 V. Cyclic voltammogram measurements of the NPG and NPAS in 0.1 M HClO₄ were conducted from 0.2 to 1.5 V with a scan rate of 5 mV/s to measure the reduction of AuOH.

3. Results and discussion

3.1. Microstructure of nanoporous catalysts

Representative scanning electron microscopy (SEM) images and transmission electron microscopy (TEM) images in Fig. 1 show the porous characters of the NPG and NPAS, respectively. All the samples exhibit bicontinuous skeletons and the interconnected channels. Statistical distributions of nanopores/ligaments are estimated for the NPAS (7.9 nm) and NPG (8.3 nm) based on SEM images (Fig. S1). The NPAS with a relatively smaller pore size is a result from sluggish Au surface diffusion perturbed by spontaneous releasing and burying of tin atoms into the ligament through chemical dealloying [23,24]. The concentration of tin element is estimated as 2.6 % based on energy dispersive X-ray spectra (EDX) analysis (Fig. S2). From the X-ray diffraction (XRD) pattern of the NPAS (Fig. S3) shows that all the crystal planes corresponding to face-centered-cubic (fcc) Au without any additional peaks, implying the formation of Au-Sn solid solution after chemical dealloying. However, the 220 and 331 peaks in the NPAS shift toward lower diffraction angles with respect to the NPG, verifying tensile strain in Au-Sn solid solution.

View original graphic| Download| PPT slide

Fig. 1.   Nanoporous structures of the NPG and NPAS. (a, b) SEM and TEM images of the NPG. (c, d) SEM and TEM images of the NPAS.

As shown in Fig. S4, chemical bonding of Sn element is characterized by X-ray photoelectron spectroscopy (XPS). The 3d_3/2 peak at 493.9 and 3d_5/2 peak at 485.5 eV in metallic state (Sn⁰) can be detected in the NPAS [21]. The paired shoulder peaks at 495.6 and 487.1 eV are related to tin oxide (SnO₂ and SnO) [25]. SnO₂ that appears on the surface of the NPAS is caused by spontaneous oxidization of Sn in the air. A small amount of tin oxide on the surface will be reduced into metallic Sn during electrochemical measurements. Cyclic voltammetry (CV) measurement by sweeping in 0.1 M KHCO₃ electrolyte reveals a cathodic peak at 0.21 V, which indicates the reduction of SnO₂ to SnO, together with an additional peak at 0.01 V corresponding to the further reduction of SnO to metallic Sn (Fig. S5). The two anodic peaks, on the other hand, stem from Sn oxidization to SnO (at 0.19 V) and further oxidization of SnO to SnO₂ (at 0.38 V). The reproducible CV curves can be obtained after 20 sweeping cycles, indicating the formation of a stable surface composition. Tin oxide nanoparticles are not observed on the surface of the NPAS by high-angle annular dark-field scanning TEM (Fig. S6).

3.2. CDR performance of nanoporous catalysts

The CDR performance of nanoporous catalysts is evaluated in CO₂-saturated 0.1 M KHCO₃ (pH≈6.8). The gas products are automatically conducted within 15 min for quantitative analysis. Fig. 2 shows the CO FE for the NPG and NPAS catalysts at different potentials. Compared to the NPG, the NPAS exhibits ∼10 % enhancement of CO FE at the potential range -0.6∼-0.9 V (vs. RHE), suggesting that the intrinsic activity of the NPAS is substantially higher than that of the NPG. Specifically, the maximum conversion efficiency for CO in the NPAS reaches 92 % at -0.85 V. In addition to the gas products, the liquid-phase products after 8 h of electrolysis are measured by means of ¹H NMR. Trace amounts of formate (FE < 0.2 %) are detected at several potentials (Fig. S7). It is found that, besides selectivity enhancement, NPAS has the advantage to produce high current density. The partial current density for CO production j_CO, normalized with respect to the geometrical surface area, as reported in Table S1, can be as high as 9.5 mA/cm² in 0.5 M KHCO₃, which is 35 % higher than that of the case when the NPG is applied. Moreover, the j_CO in the case of the NPAS is higher than that in the cases of polycrystalline Au, Au rods, Au nanoparticle, oxide-derived Au nanostructure and Au film [[26], [27], [28]]. In principle, the j_CO depends on the actual value of electrochemical active surface area (ECSA) or the catalytic ability of the active site. Underpotential deposition of lead (Pb UPD, Fig. S8) is employed to estimate the ECSA of nanoporous catalysts. The roughness factor for the the NPAS is obtained to be around 255, comparable to that in the NPG (247), implying almost equivalent active sites for both cases. Therefore, simultaneous enhancements of selectivity and current density observed in the NPAS probably result from intrinsic catalytic activity per unit site. The underlining mechanism will be discussed in the following section.

View original graphic| Download| PPT slide

Fig. 2.   Eletrocatalytic performances of the NPG and NPAS. (a, b) Faradaic efficiencies of CO and H₂ at various potentials in CO₂-saturated 0.1 M KHCO₃ for the NPG and NPAS, respectively. (c) Comparisons of FEs for CO and H₂ products between NPG and NPAS at various potentials. (d) The long-term stability tests for the NPG and NPAS at -0.85 V (vs. RHE). Total current density vs. time on (left axis) and CO Faradaic efficiency vs. time on (right axis).

Besides improved selectivity and current density, the NPAS also has shown robust durability. 8 -h electrolysis is performed at -0.85 V in a CO₂-saturated 0.1 M KHCO₃ solution, as shown in Fig. 2(d). Geometric current density (j_tot) for the case of NPG decreases significantly after 5 h, a clear indication of the degeneration of catalytic activities. SEM analysis reveals the coarsening of the ligaments and the pinning of nanopores after long term electrolysis (Fig. S9). Different from the NPG, the NPAS exhibits a weakly enhanced current within 8 h without abrupt decay. The CO FE of the NPAS remains at or above 90 % throughout the electrochemical process, suggesting its remarkable stability. Furthermore, no obvious nanoporous morphology change can be detected in the NPAS by SEM. Essentially, the coarsening of the metallic ligament occurs at the electrode/electrolyte interface as a result of surface diffusion of the atoms. The extraneous element (i.e., Pt, Pd, Sn) diluted into the lattices can deactivate this diffusion process by the intervention of defects and strain field within the ligaments, which might increase the free energy (ΔG^m) of surface Au atoms diffusion (Fig. S9). In our work, Sn solute into the NPAS solid solution can be also proposed to suppress the diffusion of Au atoms during electrochemical electrolysis. Adding small fractions of foreign elements in the precursor to inhibit structure coarsening and improve electrochemical stability have been demonstrated in the Au-Pt and Au-Pd alloys before [29,30].

3.3. The inhibited HER on the NPAS

As two competitive reactions, both the CDR and the HER need to consume protons. To understand the electrokinetics mechanism of nanoporous catalysts, the dependence of HER activity on HCO₃^- ions is investigated in electrolytes with a concentration varying from 0.1 to 1 M. The chemical environment will influence the kinetics of proton transfer, exhibiting a more sensitive effect on the HER than on the CDR [31]. To confirm the effect of proton transfer on the j_H2, both nanoporous catalysts are evaluated and the results are shown in Fig. 3(a). The NPAS shows significantly diminished j_H2 at a low concentration (0.1∼0.5 M) in comparison with the NPG. In this concentration regime, weak buffering of HCO₃^- ions induces protons depletion and large pH gradients within the pores [28]. While in the high concentration regime (0.5∼1 M), as strong buffering of HCO₃^- ions provides a relatively sufficient supply of protons, the j_H2 is less sensitive to ionic concentration. Since their pore sizes are comparable, Sn solute in the NPAS is believed to enable the inhibition of HCO₃^- migration within the hollow channels during the HER process.

View original graphic| Download| PPT slide

Fig. 3.   Electrokinetics of CDR and HER on the NPG and NPAS. (a) The HCO₃^- concentration effect of current density corresponding to the HER at -0.9 V and -1.0 V vs SCE. (b) Tafel plots of H₂ product at various scan rates. (c) Schematic illustration for surface migration of HCO₃^- at different conditions.

Besides the effect of ionic concentration, scanning rate is another parameter to control the kinetics of the HER. From the linear sweep voltammogram (LSV) curves collected in CO₂-saturated 0.5 M KHCO₃ (Fig. S10), it is apparent that the current density increases with scan rate, which is dependent on mass transport within the porous channels. Here the j_H2 is correlated with the overpotential η by Tafel equation:

η=a+blog|jH₂| (1)

$j_{ H _{2}}=j_{ total }×FE_{ H _{2}}$ (2)

where j_total is total current in LSV curves, and FE_H2 is Faradaic efficiency of H₂. As shown in Fig. 3(b), the j_H2 for the NPG monotonically increases with the overpotential, irrespective of the scan rate going from 5 to 100 mV/s. The NPAS displays a monotonous change of the j_H2 with respect to the overpotential at low scan rates (5-20 mV/s). At high scan rates (50-100 mV/s), a critical ovepotential at 0.25 V is observed, beyond which the j_H2 decreases. The NPAS catalyst exhibits a much steeper Tafel slope for the HER than the NPG. The distinction between these two catalysts becomes more noticeable with the increase of the scan rate, which indicates the inhibition of the HER by the NPAS at high scan rates.

The inhibited HER by the NPAS is connected with bicarbonate concentration and scanning rate, suggesting it is a diffusion-mediated mechanism. To understand the kinetics of the proton transfer, a simple scenario is proposed to illustrate the surface migration of HCO₃^- at different conditions, as shown in Fig. 3(c). During the HER, proton from the bicarbonate is transferred to the cathode and gets reduced to H₂. It is obvious that the surface migration of HCO₃^- would be prohibited within the nanopores in comparison to the plane surface, leading to a concentration gradient of HCO₃^- and rising of the pH within the pores [32,33]. Furthermore, due to the strong adsorption between Sn atoms and bicarbonate ions [25], surface migration rate of HCO₃^- in the nanopores of the NPAS is further deceased, resulting in the inhibited HER behavior.

3.4. Intrinsic mechanism of the enhanced selectivity

Catalytic selectivity is usually attributed to two aspects, the inhibited HER and the enhanced CDR. In our experiments, the difference between the NPAS and the NPG can be ascribed to two possible factors: the type of active materials including tin oxide and metallic tin; and the evolution of electronic structure mediated by size effect, internal facets and surface strain. On one hand, small amount of residual SnO₂ formed naturally in the air is completely reduced to metallic tin because the potential for the CDR is more negative than the reduction potential of tin oxide at -0.08 V (Fig. S5). On the other hand, there is no sign of elemental aggregation on the surface, as shown in Fig. S6. To explore the intrinsic mechanism of the high activity in the NPAS and evaluate the contribution of SnO_x on the surface, a SnO_x-free nanoporous Au electrode with identical porous structure was synthesized by performing one additional dealloying in 6 M HCl solution to remove the SnO_x from the NP Au-SnO_x surface. The CV measurement also confirms that there is no redox peak corresponding to tin oxide after additional dealloying. The CDR performance of SnO_x-free nanoporous Au electrode show almost the same as the NPAS electrode, indicating that small amount of residual oxide in the surface of NPAS has no effect on the catalytic selectivity (Fig. S11). According to previous investigations, metallic Sn catalysts preferably reduce CO₂ to formate in most cases [[34], [35], [36]]. The maximum FE for formate production is about 50 %∼70 % on ultrafine metallic Sn, SnO and SnO₂ nanoparticles [37]. As for Au-Sn alloy, phase composition especially surface compositions of the electrode topmost surface do affect electrocatalytic performance very much. Ismail et al. reported that the activity and selectivity in CDR for Au-Sn bimetallic nanoparticles strongly depend on the phase composition. Au₁Sn₂ catalysts with almost pure AuSn phase showed the lowest overpotential and superior catalytic performance among different bimetallic Au-Sn nanoparticls [38]. Todoroki et al. investigated electrochemical CO₂ reduction on 0.1 monolayer-thick-Co and Sn-deposited Au (110) surfaces and showed that alloying elements of Sn at the topmost surface increased H₂ selectivity with much lower overpotential [39]. However, in our experiment, the concentration of Sn element is only 2.6 % based on EDX analysis, which means Sn can be dissolved into the lattice of Au to form a solid solution from the Au-Sn phase diagram. Whether NPAS or NPG catalysts, gold element accounts for the most important component on the surface (>89 %). Moreover, the CO FE at -0.80 V in the NPAS is 86.5 %, which is 1700 times higher than the FE of formate (∼0.05 %) (Fig. S7). Significant improvement of CO selectivity observed in our work cannot be entirely attributed to trace amount of metallic Sn. And CO as a main production in the NPAS verifies that Au matrix rather than tin solute actually serves as the active sites for the CDR.

After considering the effect of active materials, the detailed structures such as internal facet might affect electronic structure of surface atoms, thereafter catalytic selectivity. Internal facets within the nanoporous Au can be quantitatively monitored by Pb UPD in an alkaline media [[40], [41], [42]]. As displayed in Fig. S12, the voltammetric profile of the NPAS exhibits three distinct peaks, a small peak at 0.33 V and another two intense peaks at 0.42 V and 0.58 V, respectively. The small peak corresponds to the step/kink sites, and the intense peak at 0.42 V arises from Au {111} facets. Broad peak at 0.58 V is the convolution of the two peaks. One peak at 0.59 V corresponds to the {110} facets, and the second peak at 0.52 V corresponds to the {100} facets (Fig. S13). Because the peak intensity is proportional to the surface area percentage of corresponding facets, we estimated the proportions of different facets on the internal surfaces of the NPAS and NPG. The fractions of step/kink sites, the {111}, {100}, and {110} facets in these two catalysts were in comparable levels. As a result, the facet effect of the electronic structures can be safely excluded for the enhancement in catalytic activity in the NPAS.

Except for alloying effect on catalytic selectivity, surface strain in electrocatalysts also provides a versatile path to tune the electronic structure and enhance the catalytic activity [20,[43], [44], [45]]. To quantitatively demonstrate the strain distributions in nanoporous structures, aberration-corrected HRTEM images are taken carefully to ensure that the atomic columns exhibit homogeneous contrast without sharp changes and then directly visualized by using the geometrical phase analysis (GPA) [46]. Based on the GPA technique, quantitative strains as shown in Fig. 4(c) and (f) are reliably extracted from linear scanning along the red lines (Fig. 4(a) and d), respectively. Tensile strain of ∼2.2 % is detected in the NPAS, while the strain in the NPG is only ∼0.5 %, which agrees well with the previous investigation [47]. Altogether, both XRD and GPA analysis can demonstrate the tensile strain in the NPAS catalysts as a result of trace amount of Sn atoms diluted into Au lattices.

View original graphic| Download| PPT slide

Fig. 4.   Strain distributions of the NPG and NPAS. HRTEM images of the NPG (a), and the NPAS (d). (b, e) Lattice strain maps corresponding to the ligaments in (a) and (d). (c, f) Quantitative strains extracted from the maps after linear scanning as indicated by red arrows in panels (a) and (d).

The GPA analysis of HRTEM indicates solute strain with a level of 2.2 % in the NPAS, which is three times higher than that in the NPG. To understand the strain-induced changes in electronic structure, XPS and surface valence band photoemission spectra are conducted to exam the electronic structures. The Au 4f peaks in the NPAS shows 0.3 eV shift towards lower binding energy relative to the NPG (Fig. S14). As shown in Fig. 5(a), the d-band center (with regard to Fermi level) of the NPAS exhibits an upward shift from -5.0 (NPG) to -4.76 eV. It is well accepted that the upward shift of the d-band center pushes the anti-banding states above the Fermi level, resulting in a stronger adsorbate bonding [48,49]. Intrinsically, the adsorption energy of key intermediate (*COOH) is strain-sensitive by considering the binding configuration [50]. Tensile strain in the NPAS tends to strengthen the bonding energy between the catalyst and the adsorbate. Correspondingly, the strong adsorption energy of *COOH intermediate on the NPAS will promote selective conversion of CO₂ towards CO, which is also demonstrated in Pd icosahedra with a lattice strain [51]. Besides surface valence band photoemission spectra, electrochemical analysis also verifies the increased adsorption energies of the reactive intermediates. We thus obtained the cyclic voltammograms at 5 mV s^-1 in N₂-saturated 0.1 M HClO₄ solution, as shown in Fig. 5(b). For the hydroxide peaks, the NPAS shows a negative shift in the reduction peak of AuOH during its cathodic scanning with respect to the NPG, indicating a stronger adsorption affinity of the OH^- surrogate ion on the NPAS surface [35,36,51].

View original graphic| Download| PPT slide

Fig. 5.   (a) Surface valence band photoemission spectra of the NPG and NPAS. The d-band centers are marked with white lines. (b) The magnified region from 1.3 to 0.85 V in the negative scan of cyclic voltammograms for the NPG and NPAS in N₂-saturated 0.1 M HClO₄ solution with a scan rate of 5 mV/s.

Different from preferable adsorption of CO₂ molecules on the top of metallic atoms, the *H prefers to adsorb on the hollow site between surface atoms during electrochemical process [16]. Therefore, lattice strain will adversely weaken the adsorption energy of *H onto the NPAS that ultimately leads to the inhibited HER, which is supported by the two kinetic experiments concerning the HCO₃^- concentration and the scan rate as discussed before.

Similar strategy with severely inhibited HER can be also extended toward some other bimetallic catalysts, such as Cu-In [15], Cu-Sn [16], and Ag-Sn alloys [52]. Especially, the strain effect on bimetallic solid solution with a wide diluted range can be tailored with a refinement step, which allows the adsorption behaviors of different intermediate species are finely tuned. That is, active sites for the target reaction and side reaction can be simultaneously tuned. Therefore, solute strain based on bimetallic catalysts offers a means to locally tailor electronic structure at the atomic scale in favor of superior activity and robust durability in a catalytic reaction that involves multiple protons and electrons.

4. Conclusion

In this work, the NPAS with Sn solid solution was designed for electrochemical carbon dioxide reduction. Lattice strain within the Au ligament at a level of ∼2.2 % was introduced by adding a small amount of Sn atoms. The NPAS has shown a maximum FE of 92 % for CO production at -0.85 V (vs. RHE) while maintaining the activity for at least 8 h. This investigation has demonstrated that exclusive selectivity of CO and the inhibited HER may arise from the strain-induced changes in electronic structure, which strengthen the adsorption of key intermediate *COOH. The idea to introduce lattice strain, as reported here in this study, opens a new opportunity to design bimetallic electrocatalysts with superior catalytic performance at low overpotential.

Acknowledgements

We acknowledge financial support from the National Natural Science Foundation of China (Nos. 51771078, 91545131 and 51371084) and China Postdoctoral Science Foundation (No. 2017M612455). Electronic structure characterizations (XPS and surface valence band photoemission spectra) were carried out in Analytical and Testing Center in HUST. Microstructure characterizations were carried out in State Key Laboratory of MPDMT.