Bi@Sn Core–Shell Structure with Compressive Strain Boosts the Electroreduction of CO₂ into Formic Acid

Abstract

As a profitable product from CO₂ electroreduction, HCOOH holds economic viability only when the selectivity is higher than 90% with current density (j) over −200.0 mA cm⁻². Herein, Bi@Sn core–shell nanoparticles (Bi core and Sn shell, denoted as Bi@Sn NPs) are developed to boost the activity and selectivity of CO₂ electroreduction into HCOOH. In an H‐cell system with 0.5 m KHCO₃ as electrolyte, Bi@Sn NPs exhibit a Faradaic efficiency for HCOOH (FE_HCOOH) of 91% with partial j for HCOOH (j _HCOOH) of −31.0 mA cm⁻² at −1.1 V versus reversible hydrogen electrode. The potential application of Bi@Sn NPs is testified via chronopotentiometric measurements in the flow‐cell system with 2.0 m KHCO₃ electrolyte. Under this circumstance, Bi@Sn NPs achieve an FE_HCOOH of 92% with an energy efficiency of 56% at steady‐state j of −250.0 mA cm⁻². Theoretical studies indicate that the energy barrier of the potential‐limiting step for the formation of HCOOH is decreased owing to the compressive strain in the Sn shell, resulting in the enhanced catalytic performance.

Bi@Sn core–shell nanoparticles (Bi core and Sn shell) are developed to boost the activity and selectivity of CO₂ electroreduction into HCOOH. The lattice mismatch between the Bi core and Sn shell generates a compressive strain in the Sn shell. The compressive strain in the Sn shell decreases the energy barrier for the formation of HCOOH, resulting in the enhanced catalytic performance.

The excessive utilization of fossil fuels and accelerating emissions of CO₂ have led to the energy shortage and greenhouse effect.^{[ 1} , ² , ³ , ⁴ , ^{5 ]} CO₂ electroreduction into useful chemicals and fuels represents a promising way that not only meets the ever‐increasing energy demands but also mitigates environmental crisis caused by CO₂ emissions.^{[ 6} , ⁷ , ⁸ , ⁹ , ¹⁰ , ^{11 ]} As a value‐added product from CO₂ electroreduction, formic acid (HCOOH) is an important feedstock for pharmaceutical and chemical industry.^{[ 12 ]} Meanwhile, HCOOH is a liquid fuel for proton‐exchange membrane fuel cell.^{[ 13 ]} Additionally, HCOOH also serves as potential hydrogen carrier.^{[ 14 ]} Based on the gross‐margin model, HCOOH has been suggested to be one of the most economically viable products during CO₂ electroreduction process.^{[ 15 ]} To this end, the electroreduction of CO₂ into HCOOH is of great significance.

Currently, various metal‐based electrocatalysts such as Pd, Pb, Hg, Cd, Tl, In, and Sn have been explored to achieve the high activity and selectivity for electroreduction of CO₂ into HCOOH.^{[ 16} , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ^{22 ]} Among these catalysts, Sn‐based catalysts have drawn considerable attentions due to the superiorities of nontoxicity, earth abundance, and low cost.^{[ 23} , ²⁴ , ²⁵ , ^{26 ]} Up to now, several effective strategies have been exploited to improve the catalytic performance of Sn‐based catalysts. For instance, owing to the abundant grain boundaries, the ultrathin sub‐2 nm SnO₂ quantum wires composed by individual SnO₂ quantum dots achieved improved Faradaic efficiency for HCOOH (FE_HCOOH) of 87.3% with the current density (j) of −15.7 mA cm⁻² relative to SnO₂ nanoparticles (NPs).^{[ 27 ]} Besides, Sn quantum sheets confined in graphene exhibited high conductivity and fast charge‐transfer process, resulting in improved catalytic activity for HCOOH.^{[ 28 ]} Moreover, the mesoporous SnO₂ displayed a maximum FE_HCOOH of 75% and a j of −10.8 mA cm⁻² at −1.15 V versus reversible hydrogen electrode (vs RHE).^{[ 29 ]} The enhanced catalytic performance was attributed to the promoted CO₂ activation by the construction of oxygen vacancy.^{[ 29 ]} However, most previously reported Sn‐based catalysts still suffer from limited FE_HCOOH at high current density, prohibiting the practical application of Sn‐based electrocatalysts. Therefore, it is of great importance to develop efficient Sn‐based catalysts with high activity and selectivity for HCOOH toward CO₂ electroreduction.

Herein, we developed Bi@Sn core–shell nanoparticles (Bi core and Sn shell, denoted as Bi@Sn NPs) to boost the activity and selectivity for electroreduction of CO₂ into HCOOH. In H‐cell system with 0.5 m KHCO₃ as electrolyte, Bi@Sn NPs exhibited an FE_HCOOH of 91% with partial j for HCOOH (j _HCOOH) of −31.0 mA cm⁻² at −1.1 V versus RHE. The potential application of Bi@Sn NPs was testified via chronopotentiometric measurements in flow‐cell system with 2.0 m KHCO₃ electrolyte. Under this circumstance, Bi@Sn NPs achieved an FE_HCOOH of 92% with a steady‐state j of −250.0 mA cm⁻². Theoretical studies indicate that energy barrier of the potential‐limiting step for the formation of HCOOH was decreased owing to the compressive strain in Sn shell, resulting in the enhanced catalytic performance.

Bi@Sn NPs were synthesized via electroreduction of Bi₂Sn₂O₇ NPs in 0.5 m KHCO₃ at −0.8 V versus RHE for 1 h. Specifically, Bi₂Sn₂O₇ NPs were prepared via a solvothermal reaction (Figure S1, Supporting Information). Figure  1a shows the transmission electron microscopy (TEM) image of Bi@Sn NPs, which took a spherical morphology with an average diameter of 45 nm. As shown in Figure 1b, the high‐resolution TEM (HRTEM) image of an individual Bi@Sn NP exhibited a clear contrast between the Bi core and the Sn shell, indicating the core–shell structure of Bi@Sn NPs. To further characterize the exquisite core–shell structure of Bi@Sn NPs, we employed the high‐angle annular dark field scanning TEM (HAADF‐STEM). As shown in Figure 1c, the interplanar spacing for Sn (101) and (020) planes in Sn shell was 0.270 and 0.283 nm, respectively, less than the standard values of 0.279 and 0.291 nm in tetragonal Sn. Other regions of the Bi@Sn NP were also analyzed via HAADF‐STEM. As shown in Figure S2 in the Supporting Information, the HAADF‐STEM images of different regions exhibited the similar feature to that in Figure 1c. As shown in Figure S3 in the Supporting Information, the total distance of five groups of successive (020) plane was measured and then divided by five to obtain the interplanar spacing of (020) plane. The interplanar spacing of (020) plane for Sn shell in Bi@Sn NP was 0.283 nm. For comparison, the interplanar spacing of (020) plane for Sn NP was determined to be 0.291 nm, slightly larger than that of Sn shell in Bi@Sn NP. These results suggested that the lattice of Sn shell was compressed. In the Bi core, the lattice fringe with an interplanar spacing of 0.323 nm was ascribed to the (012) plane of rhombohedral Bi. Meanwhile, the thickness of Sn shells in Bi@Sn NPs ranged from 2.160 to 2.700 nm, revealing that the shell was composed of 8–10 layers of Sn atoms. The core–shell structure of Bi@Sn NPs was further confirmed by the energy dispersive X‐ray (EDX) elemental mapping. As shown in Figure 1g, the interior core of Bi (green) was surrounded by the outer shell of Sn (red). This result was also supported by the line‐scanning profiles across an individual Bi@Sn NP (Figure 1h). To investigate the phase composition of Bi@Sn NPs, we carried out X‐ray diffraction (XRD) measurement. As evidenced by XRD patterns in Figure 1i, the Bi@Sn NPs exhibited the diffraction peaks at 30.74°, 32.09°, and 44.98°, which were indexed to the (200), (101), and (211) planes of tetragonal Sn (JCPDS No. 89‐2761).^{[ 5 ]} The diffraction peaks of Sn (200) and Sn (101) in Bi@Sn NPs shifted to higher diffraction angles, further proving that the lattice of Sn shell was compressed (inset of Figure 1i; Table S1, Supporting Information). In addition, the diffraction peaks at 22.56°, 27.22°, 38.02°, 39.71°, 48.80°, 56.20°, 62.38°, and 64.69° were assigned to the (003), (012), (104), (110), (202), (024), (116), and (122) planes of rhombohedral Bi (JCPDS No. 85‐1330).^{[ 30 ]} Bi@Sn NPs exhibited a larger I _Bi(012)/I _Bi(104) value (4.18) than that (2.95) of standard rhombohedral Bi, suggesting that the Bi core exhibited preferred orientations of (012) facets. The strong diffraction peak located at 26.48° was attributed to the (002) plane of graphite due to the substrate of carbon paper. These results together revealed that Bi@Sn NPs consisted of metallic Sn and Bi. Cyclic voltammogram (CV) measurements confirmed that Bi core was totally covered by Sn shell (Figure S4, Supporting Information). For comparison, we also prepared Sn NPs with an average diameter of 48 nm by reducing SnCl₂ with NaBH₄ (Figure S5, Supporting Information).

Figure 1.

a) TEM image, b) HRTEM image, and c) HAADF‐STEM image of Bi@Sn NPs. d–g) HAADF‐STEM and EDX elemental mapping images of an individual Bi@Sn NP. h) Line‐scanning profiles of Sn and Bi along the yellow line in (d). i) XRD pattern of Bi@Sn NPs; the corresponding magnified XRD pattern of Bi@Sn NPs is given in the inset.

Bi@Sn NPs were applied as a heterogeneous catalyst toward CO₂ electroreduction in an H‐cell system. Linear sweep voltammetry (LSV) curves of Bi@Sn NPs were measured in both CO₂‐saturated and Ar‐saturated 0.5 m KHCO₃ electrolytes. As shown in Figure S6 in the Supporting Information, Bi@Sn NPs exhibited a j of 45.1 mA cm⁻² in CO₂‐saturated electrolyte at −1.2 V versus RHE, which was 1.6 times as high as that (27.2 mA cm⁻²) in Ar‐saturated electrolyte. In this regard, CO₂ electroreduction was more favorable than competing hydrogen evolution reaction (HER) over Bi@Sn NPs. In addition, based on the intercept of the linear region in Tafel plots, the exchange current density (j ₀) for HER over Bi@Sn NPs was calculated to be 16.1 µA cm⁻², which was lower than that (35.4 µA cm⁻²) over Sn NPs (Figure S7, Supporting Information). In this case, Bi@Sn NPs restrained the competing HER relative to Sn NPs. We applied in situ attenuated total reflection infrared (ATR‐IR) spectroscopy to monitor the process of CO₂ electroreduction over Bi@Sn NPs at different potentials in CO₂‐saturated 0.5 m KHCO₃ electrolyte (Figure S8, Supporting Information). The characteristic band at 1372 cm⁻¹ for the symmetric O–C–O stretching mode of HCOOH was observed, indicating the formation of HCOOH over Bi@Sn NPs.^{[ 31} , ^{32 ]}

To evaluate the catalytic performance of Bi@Sn NPs and Sn NPs toward CO₂ electroreduction, we conducted chronoamperometry measurements with a series of applied potentials. The gaseous products and liquid products were quantitatively analyzed via online gas chromatography and ¹H nuclear magnetic resonance (¹H NMR), respectively (Figure S9, Supporting Information). Figure  2a shows the FE_HCOOH toward CO₂ electroreduction over Bi@Sn NPs and Sn NPs. The FE_HCOOH of Bi@Sn NPs was higher than that of Sn NPs at all applied potentials. Specifically, at −1.1 V versus RHE, the FE_HCOOH of Bi@Sn NPs was 91%, which was 1.6 times as high as that (56%) of Sn NPs. Meanwhile, Bi@Sn NPs exhibited lower Faradaic efficiency for CO (FE_CO) and Faradaic efficiency for H₂ (FE_H2) than Sn NPs at all applied potentials, demonstrating that Bi@Sn NPs suppressed the formation of CO and H₂ (Figure S10, Supporting Information). As shown in Figure 2b, the j _HCOOH of Bi@Sn NPs was higher than that of Sn NPs at all applied potentials. Especially, when the applied potential was set at −1.2 V versus RHE, the j _HCOOH of Bi@Sn NPs reached −38.0 mA cm⁻², whereas the j _HCOOH of Sn NPs was −20.9 mA cm⁻². Figure S11 in the Supporting Information shows the yield rates for HCOOH of Bi@Sn NPs and Sn NPs at different applied potentials. At −1.2 V versus RHE, the yield rate for HCOOH of Bi@Sn NPs achieved 708.9 µmol cm⁻² h⁻¹, which was 1.8 times as high as that (390.6 µmol cm⁻² h⁻¹) of Sn NPs.

Figure 2.

a) FE_HCOOH, b) j _HCOOH, and c) Φ _HCOOH of Bi@Sn NPs and Sn NPs at different applied potentials. d) The j and FE_HCOOH of Bi@Sn NPs at −0.9 V versus RHE with 20 h potentiostatic test.

The efficiency from electrical energy to the chemical energy of target product is also an important parameter to evaluate the catalytic performance.^{[ 27} , ^{33 ]} Figure 2c shows the energy efficiency for HCOOH (Φ _HCOOH) of Bi@Sn NPs and Sn NPs at different applied potentials. The Φ _HCOOH of Bi@Sn NPs exceeded 50% in a wide potential range from −0.8 to −1.2 V versus RHE. Notably, at −0.9 V versus RHE, the Φ _HCOOH of Bi@Sn NPs achieved 59%, which was 1.6 times as high as that (36%) of Sn NPs. The catalytic stability for CO₂ electroreduction over Bi@Sn NPs was also evaluated via the chronoamperometry electrolysis. As shown in Figure 2d, Bi@Sn NPs exhibited negligible decay in both FE_HCOOH and j at −0.9 V versus RHE during 20 h electrolysis. The morphology, core–shell structure, and phase for Bi@Sn NPs were still preserved after the durability test (Figure S12, Supporting Information).

To rationalize the enhanced activity for HCOOH of Bi@Sn NPs relative to that of Sn NPs, we conducted the electrochemical surface areas (ECSAs) and electrochemical impedance spectroscopy (EIS) measurements. Derived from CVs measurements under different scan rates (Figure S13, Supporting Information), ECSAs of Bi@Sn NPs and Sn NPs were calculated via measuring double layer capacitance (C _dl). Figure S14 in the Supporting Information shows the linear relationship between the differences of charging current density and scan rates for Bi@Sn NPs and Sn NPs. The values of C _dl were fitted to be 2.6 and 2.3 mF cm⁻² for Bi@Sn NPs and Sn NPs, respectively. We normalized the j _HCOOH of Bi@Sn NPs and Sn NPs based on the value of C _dl. The normalized j _HCOOH of Bi@Sn NPs was larger than that of Sn NPs at all applied potentials (Figure S15, Supporting Information). As such, the difference in activity between Bi@Sn NPs and Sn NPs was independent of ECSA. Meanwhile, Bi@Sn NPs also exhibited higher mass activities than Sn NPs, following the same trend of specific activity (Figure S16, Supporting Information). Figure S17 in the Supporting Information shows the Nyquist plots of Bi@Sn NPs and Sn NPs. The charge transfer resistance (R _ct) (205.8 Ω) of Bi@Sn NPs was smaller than that (326.8 Ω) of Sn NPs. Accordingly, Bi@Sn NPs exhibited a faster Faradaic process than Sn NPs toward CO₂ electroreduction.

To gain insight into the intrinsic reason for the high catalytic performance of Bi@Sn NPs, we conducted density functional theory (DFT) calculation. Based on the modeling study, we found that there existed 8.5% compressive strain in Sn shell for Bi@Sn NPs owing to the lattice mismatch between Bi core and Sn shell (Figure S18, Supporting Information). Taking the strain effect into consideration, we investigated the Gibbs free energy of CO₂ reduction and the competing HER process on the Sn slab with 8.5% compressive strain (compressive Sn slab) and pristine Sn slab (Table S2, Supporting Information). Generally, *HCOO and *COOH were considered to be the intermediates for the formation of HCOOH and CO, respectively.^{[ 34 ]} Figure  3a shows Gibbs free energy of the HCOOH pathway. The Gibbs free energy change (ΔG) of the conversion from *HCOO to HCOOH was higher than that of the formation of *HCOO on both compressive Sn slab and pristine Sn slab. Therefore, the conversion from *HCOO to HCOOH served as potential‐limiting step on both compressive Sn slab and pristine Sn slab. Notably, the reaction barrier on compressive Sn slab was 0.54 eV, lower than that (0.75 eV) on pristine Sn slab. Accordingly, Sn with compressive strain facilitated the formation of HCOOH during CO₂ electroreduction process. We also calculated the Gibbs free energy for each steps involved in the CO₂ reduction into CO on both compressive Sn slab and pristine Sn slab. As shown in Figure 3b, the formation of *COOH served as potential‐limiting step on both compressive Sn slab and pristine Sn slab. The ΔG for the formation of *COOH on compressive Sn slab and pristine Sn slab were 1.35 and 1.19 eV, respectively. This result indicated that the introduction of compressive strain into Sn depressed the formation of CO. Furthermore, the ΔG for the formation of *H on compressive Sn slab was 0.58 eV, which was higher than that (0.36 eV) on pristine Sn slab, indicating that the competing HER was suppressed by introducing the compressive strain into Sn (Figure 3c).

Figure 3.

Gibbs free energy diagrams for a) CO₂ reduction to HCOOH, b) CO₂ reduction to CO, and c) H₂ evolution on both compressive Sn slab and pristine Sn slab. The blue, red, black, and pink spheres represent Sn, O, C, and H atoms, respectively. d) The difference in limiting potentials between CO₂ reduction into HCOOH and CO₂ reduction into CO on both compressive Sn slab and pristine Sn slab. e) The difference in limiting potentials between CO₂ reduction into HCOOH and H₂ evolution on both compressive Sn slab and pristine Sn slab.

To investigate the origin of the strain‐induced optimization of adsorption energies for reaction intermediates, we calculated the projected density of states (PDOS) of compressive Sn slab and pristine Sn slab (Figure S19, Supporting Information). The p‐band center (with regard to the Fermi level) of compressive Sn surface is −1.57 eV, which was 0.1 eV lower than that (−1.47 eV) of pure Sn surface. The downward shift of p‐band center decreases the antibonding states above the Fermi level, resulting in weaker adsorption bonding. Accordingly, the weakened adsorption of intermediates was further confirmed by the shortened bond length between adsorbed species and the reaction sites on compressive Sn surface (Table S3, Supporting Information). In addition, based on the analysis of DFT results, the conversion from *HCOO to HCOOH serves as the potential‐limiting step on both compressive Sn slab and pristine Sn slab. As such, the weakened adsorption strength of *HCOO contributes to the decreased energy barrier for the formation of HCOOH, thus promoting the catalytic performance. Furthermore, the conversion from CO₂ to *COOH and the formation of *H serve as the potential‐limiting step on both compressive Sn slab and pristine Sn slab. The weakened bonding strength of reactions intermediates (*COOH, *H) results in the suppression of CO and H₂ production.

The thermodynamic limiting potentials between target product and byproduct is an important way to understand the competing mechanism.^{[ 35} , ^{36 ]} The thermodynamic limiting potentials were donated as U _L(target product) − U _L(byproduct), where U _L = −ΔG/e, and the ΔG is the value of Gibbs free energy change for the potential‐limiting step. A more positive value of U _L(target product) − U _L(byproduct) corresponds to higher selectivity for target product. As shown in Figure 3d, the value of U _L(HCOOH) − U _L(CO) for compressive Sn slab was 0.82 V, which was higher than that (0.45 V) for pristine Sn slab. The theoretical analysis was in good agreement with the experimental results of the enhanced selectivity for HCOOH over Bi@Sn NPs relative to that over Sn NPs. In addition, compressive Sn slab also exhibited a more positive U _L(HCOOH) − U _L(H₂) value (0.04 V) than pristine Sn slab (−0.38 V) (Figure 3e). As such, the more positive values of U _L(HCOOH) − U _L(CO) and U _L(HCOOH) − U _L(H₂) for compressive Sn slab relative to pristine Sn slab, contributed to the enhanced selectivity for CO₂ electroreduction into HCOOH over Bi@Sn NPs.

Considering that HCOOH is a profitable production with great economic benefits, it is suggested that the economically compelling application of HCOOH required the minimum j of −200.0 mA cm⁻² with FE_HCOOH of 90%, and energy efficiencies exceeding 50% in a wide potential range.^{[ 15 ]} Such a large j was difficult to achieve using the traditional H‐cell system due to the mass‐transfer limitation of CO₂ in aqueous electrolyte. To this end, we conducted CO₂ electroreduction over Bi@Sn NPs on gas diffusion electrode (GDE) using a flow‐cell system in 2.0 m KHCO₃ (Figure  4a; Figure S20, Supporting Information). Chronopotentiometric measurements were conducted to evaluate the catalytic performance of Bi@Sn NPs. The FE_HCOOH exceeded 92% for Bi@Sn NPs at all applied j from −25.0 to −250.0 mA cm⁻² with corresponding potential ranging from −0.81 to −1.15 V versus RHE (Figure 4b,c). Meanwhile, the energy efficiency was higher than 55% for Bi@Sn NPs when j ranged from −25.0 to −250.0 mA cm⁻² (Figure 4d). Besides, Bi@Sn NPs showed 1% increase for potential and 2% decay for FE_HCOOH during the 8 h durability test at j of −200.0 mA cm⁻² (Figure 4e). We further compared the FE_HCOOH and j _HCOOH of Bi@Sn NPs with other reported Sn‐based catalysts (Table S4, Supporting Information). Possessing the high j _HCOOH and FE_HCOOH at low applied potential, Bi@Sn NPs represented one of the best Sn‐based catalysts for the electroreduction CO₂ into HCOOH up to now.

Figure 4.

a) Schematic illustration of the flow‐cell system. b) Plots of potential–time curves with iR‐correction, c) FE_HCOOH, and d) Φ _HCOOH of Bi@Sn NPs at the j ranging from −25.0 to −250.0 mA cm⁻² in 2.0 m KHCO₃. e) Durability tests of Bi@Sn NPs in the flow‐cell system at j of −200.0 mA cm⁻² for 8 h.

In conclusion, we developed Bi@Sn NPs with core–shell structure as efficient catalyst for CO₂ electroreduction into HCOOH. Bi@Sn NPs achieved an FE_HCOOH of 92% with j as high as −250.0 mA cm⁻² using flow‐cell system with 2.0 m KHCO₃ as electrolyte. Theoretical studies indicate that energy barrier of the potential‐limiting step for the formation of HCOOH was decreased owing to the compressive strain in Sn shell, resulting in the enhanced catalytic performance. This work not only developed promising catalysts toward CO₂ electroreduction into HCOOH, but also provided a strategy for the rational design of highly efficient electrocatalysts by regulating the lattice strain.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Xing Y., Kong X., Guo X., Liu Y., Li Q., Zhang Y., Sheng Y., Yang X., Geng Z., Zeng J., Bi@Sn Core–Shell Structure with Compressive Strain Boosts the Electroreduction of CO₂ into Formic Acid. Adv. Sci. 2020, 7, 1902989 10.1002/advs.201902989