1.   Introduction

The electroreduction of CO₂ to fuels is desirable for constructing a carbon-cyclable energy system, particularly because renewable electricity is used for the electrolysis process [1]. However, CO₂, typically inert at room temperatures, requires a large overpotential to form the activated-state $\text{C}{\text{O}}_{\text{2}}^{\text{*}-}$, and the following reduction process suffers from slow kinetics [2]. Complicated proton-coupled multicharge transfer reactions and the competitive reduction of protons to hydrogen in an aqueous solution over most electrocatalysts will lead to various cathodic products [3]. Thus, developing electrocatalysts with high selectivity and high activity remains crucial.

Among various CO₂ electrochemical reduction (CO₂ER) products, formic acid (HCOOH) is particularly appealing, as it can serve as a fuel for combustion or in fuel cells, a hydrogen storage carrier, and a valuable raw chemical for industrial synthesis [4]. Regarding commercial viability, a techno-economic analysis revealed that HCOOH or formate was the most profitable product of CO₂ER [2]. A recent study reported that many metallic catalysts with high hydrogen evolution overpotential, including Cd, Pb, Hg, and Bi, showed high activity for CO₂ER [5]. Among these metallic catalysts, Bi stands out in terms of formate selectivity, and it is characterized by a relatively high elemental abundance and environmental friendliness [6–8]. Although Bi catalysts have shown high Faradaic efficiency (FE) for formate generation, the geometric productive rate of the Bi electrodes is still lower than the metrics required for practical application. Therefore, improving Bi catalyst activity without sacrificing selectivity is crucial.

Compared with bulk materials, nanostructured materials typically exhibit superior electrocatalytic performance because they facilitate the construction of highly porous electrodes with numerous active sites for CO₂ molecule adsorption [9–11]. Additionally, nanostructure metals with an integrity conductive network can improve the utilization rate of catalytic materials [12]. For example, porous Bi nanosheets have demonstrated high activity and formate selectivity in CO₂ER [6,13]. Beyond creating a nanoporous structure of the catalyst, researchers have tried to reduce the size of primary particles in the porous electrodes to increase the specific surface area of catalysts. A study on porous Ag demonstrated that as primary particle size decreased from 400 to 50 nm, the FE of CO increased from 75% to 82% under similar conditions [14]. Jia et al. [15] explored the influence of bismuth nanoparticle size on formic acid production, and the optimal catalyst yielded a formate FE higher than 90% within a potential window close to 400 mV. However, preparing nanostructure catalysts while maintaining the integrity of their conductive network is a rather complicated procedure.

Herein, nanostructured Bi (nBi) was prepared through facile electrochemical reduction of solid bismuth oxide (Bi₂O₃). Through the adjustment of the Bi₂O₃ particle size, nano Bi catalysts with different surface morphology and particle sizes were obtained. When used as an electrode catalyst for CO₂ER, Bi₂O₃-derived nBi yielded a formate FE approaching 98% at −0.78 V vs. reversible hydrogen electrode (RHE) and high stability over a 10 h electrolysis period. The straightforward preparation method and excellent electrocatalytic performance of nBi make the Bi₂O₃-derived nBi a desirable catalyst candidate for CO₂ER.

2.   Experimental

2.1   Preparation of nBi catalysts

The nBi catalysts were synthesized through electrochemical reduction of a bismuth oxide electrode. To prepare the oxide electrode, bismuth oxide was rolled into a membrane (~0.1 mm in thickness, 5.1 mg⋅cm⁻²) using 10wt% polytetrafluoroethylene (PTFE) as binder and isopropanol as solvent. Then, a rectangular piece (5 mm × 7 mm) of the membrane was pressed onto a titanium (Ti) mesh as a working electrode (WE) for constant-current reduction (5 mA) for 30 min in a 1.0 M K₂CO₃ solution against a graphite counter electrode (CE). The Ti mesh was used as the substrate, as it would not cause apparent currents for CO₂ER and hydrogen evolution [16]. Two types of nBi from bismuth oxide of different sources were prepared. One was derived from the as-received commercial bismuth oxide (purchased from Shanghai Reagent No. 2 Factory, China), with particles in the micrometer scale (denoted as mBO), and the prepared nBi was denoted as mBO-nBi. The other was derived from the self-made nanometer bismuth oxide (nBO) and was denoted as nBO-nBi. The nBO was synthesized via precipitation between bismuth nitrate pentahydrate and potassium hydroxide in aqueous solution. Specifically, 270 mg of Bi(NO₃)₃⋅5H₂O was added to 1.0 mol⋅L⁻¹ HNO₃, and then KOH was added under constant magnetic stirring. The resulting white precipitate was collected after water washing at least six times, and then, the nBO was dried in a vacuum oven at 60°C for further use. All chemicals were purchased from Sinopharm Chemical Reagent Company (China) unless otherwise specified.

2.2   Electrochemical reduction of CO₂

The electrochemical measurements were controlled by a CS350H electrochemical workstation (Wuhan, China) using a customized gas-tight H-cell. The anode and cathode chambers were separated by a Nafion 117 membrane (DuPont, the United States of America). The reference electrode was an Ag/AgCl (saturated KCl solution) electrode, and the CE was a Pt foil. A 0.5 M KHCO₃ electrolyte was utilized, and before CO₂ER tests, CO₂ gas was bubbled into the electrolyte for saturation. Potentials were calibrated to a RHE as E vs. RHE) = E vs. Ag/AgCl + 0.197 + 0.059 × pH.

Any gaseous products from the CO₂ER were collected and analyzed via gas chromatography (GC, Shandong Lunan Ruihong Chemical Instruments Co., Ltd., China). Liquid products were investigated via nuclear magnetic resonance (NMR, Bruker AVANCE III HD 400MHz, Germany) with calibration curves. Each quantitative analysis was performed three times under ambient pressure at 25°C, and an average result was reported.

The FEs for H₂, CO, and formate generation were calculated via the following equations:

where v (15 mL⋅min⁻¹) represents the flow rate of CO₂; ɑ represents the concentration of CO (or H₂) determined via GC; n_formate (mol) represents the molar amount of formate; N represents the electron transfer number for the generation of 1 mol formate, CO, or H₂; i (A) represents the reduction current during the CO₂ER process; t (s) represents the reduction time; Q_formate and Q_total (C) represent the formate charge or total charge during the electrolysis, respectively; F represents the Faraday constant 96485 C⋅mol⁻¹; p = 1.013 × 10⁵ Pa; T = 298.15 K; R = 8.314 J⋅K⁻¹⋅mol⁻¹.

2.3   Material characterizations

X-ray diffraction (XRD, Cu K_α radiation) analysis was conducted using the Bruker D8-advanced instrument (Bruker Corporation, Billerica, United States) or Rigaku Miniflex600 (Rigaku Corporation, Tokyo, Japan). The sample morphologies were imaged via scanning electron microscopy (SEM, SIRION 200 with a field-emission gun, the United States of America) and transmission electron microscopy (TEM, JEOL, and JEM-2100, Japan).

3.   Results and discussion

3.1   Solid cathode reduction of mBO and nBO to nBi

Fig. 1 illustrates the nBi electrode preparation procedure. The as-received commercial micrometer Bi₂O₃ (mBO) or the homemade nanosize Bi₂O₃ (nBO) was used as the precursor. It was made into a membrane electrode and electrochemically reduced to form the nBi membrane electrode. Fig. S1(a) depicts the morphologies of the mBO and shows the aggregation of many Bi₂O₃ micrometer curved sheets. The commercial mBO sample was composed of Bi₂O₃ (PDF#, 71-2274) and trace impurities (Bi₂O₄ or Bi₂O₂CO₃), according to the XRD analysis (Fig. 2(a)). The impurities were probably caused by long-time exposure to the chemical in the atmosphere. The XRD pattern of the mBO membrane electrode subjected to electrochemical reduction in 1 M K₂CO₃ (Fig. 2(a)) shows only peaks of Bi (PDF#, 85-1329) and Ti; the peaks of Ti likely originated from the current collector. This indicates the complete conversion of mBO into Bi metal.

Figure  1.  Schematic of the nBi catalyst preparation process.

下载: 全尺寸图片 幻灯片

Figure  2.  XRD patterns: (a) mBO and the obtained mBO-nBi electrode; (b) the homemade nanometer Bi₂O₃ (nBO) and the obtained nBO-nBi electrode.

下载: 全尺寸图片 幻灯片

The surface of the as-rolled mBO membrane appeared dense (Fig. S1(c)). After electroreduction, the resulting mBO-nBi electrode exhibited a nanoporous structure comprising nBi dendrites with a primary particle size of ~80 nm (Fig. 3(a) and (b)). These nBi dendrites were tightly interconnected to form a nanostructure conductive network, likely advantageous for facilitating the mass transfer of CO₂ in the electrolytes during CO₂ER [12,17–19].

Figure  3.  Typical SEM images of the (a, b) mBO-nBi electrode and (c, d) nBO-nBi electrode.

下载: 全尺寸图片 幻灯片

To explore the influence of Bi₂O₃ particle sizes, homemade Bi₂O₃ consisting of nanoparticles was used as a mBO substitute, and the prepared nBi electrode was denoted as the nBO-nBi electrode. Fig. 2(b) displays the XRD pattern of the homemade Bi₂O₃. All of the diffraction peaks can be indexed to Bi₂O₃ (PDF#, 76-1730). In contrast, the synthesized Bi₂O₃ (Fig. S1(b)) was composed of nanorods with a diameter of ~30 nm, confirming the successful synthesis of nano bismuth oxide (nBO). Fig. 2(b) further confirms that after the electrochemical reduction, nBO was entirely converted to metallic bismuth (PDF#, 85-1329). SEM reveals the morphology change from the nBO membrane to the nBO-nBi electrode. The nBO membrane was dense (Fig. S1(d)). In the nBO-nBi electrode, needle-shaped dendritic Bi was hierarchically aligned on the Ti mesh surface to form a highly porous Bi network (Fig. 3(c)). Compared with the mBO-nBi electrode, the nBO-nBi electrode featured a smaller Bi particle size (approximately 30–50 nm, Fig. 3(d)). The TEM image of nBO-nBi (Fig. 4(a)) confirmed that nBO-nBi was composed of nanoparticles 30–50 nm in diameter. The lattice fringe spacing of the high-resolution TEM (HRTEM) image in Fig. 4(b) was 0.33 nm, corresponding to the Bi (012) plane, which proves that the solid electrochemical reduction of nBO led to the generation of metallic Bi nanoparticles.

Figure  4.  (a) TEM image and (b) HRTEM image of nBO-nBi.

下载: 全尺寸图片 幻灯片

3.2   Electrocatalytic reduction of CO₂ with nBi catalysts

The electrochemical performances of the Bi₂O₃-derived nBi catalysts were investigated in the CO₂-saturated KHCO₃ electrolyte. Linear sweep voltammetry at 50 mV⋅s⁻¹ was used to evaluate the influence of the catalysts on the reduction current. At potentials more negative than −0.7 V, the electrolyte in its natural state (i.e., without the addition of CO₂ gas) began to decompose at mBiO-nBi and nBiO-nBi electrodes (Fig. 5(a)). However, upon the bubbling of the CO₂ into the electrolyte, both catalysts yielded significantly larger cathodic current under the same potential as that of the case without CO₂ addition. This suggests that nBi has a higher catalytic activity for CO₂ER than for the hydrogen evolution reaction (HER). In the CO₂-saturated KHCO₃ electrolyte, the reduction at the mBO-nBi electrode started at −0.65 V. In contrast, the onset potential at the nBO-nBi electrode was −0.62 V. Simultaneously, at −0.9 V, the cathodic reduction currents were −36 and −30 mA⋅cm⁻² at the nBO-nBi and mBO-nBi electrode, respectively. The earlier onset potential and the larger cathodic current at the nBO-nBi electrode compared with those at the mBO-nBi electrode reveals the higher electrocatalytic activity of the nBO-nBi catalyst. Notably, the obtained reduction current of approximately −70 mA⋅cm⁻² at the nBO-nBi electrode is nearly the highest reported current for CO₂ER based on an H-type cell [20–21].

Figure  5.  (a) Linear scan (50 mV⋅s⁻¹) curves of the mBO–nBi and nBO–nBi electrodes in 0.5 M KHCO₃ before and after the bubbling of CO₂ gas. FEs of different products from CO₂ reduction: (b) formate; (c) CO and H₂. (d) Partial current density for formate production (j_formate) at different nBi electrodes.

下载: 全尺寸图片 幻灯片

However, high activity does not necessarily mean high catalyst selectivity, especially considering that the CO₂ER is complicated and has various products. The product selectivity of the two Bi catalysts was analyzed via constant-potential electrolysis of the CO₂-saturated KHCO₃ solution, and the products of CO₂ER based on the mBO-nBi and nBO-nBi electrodes were analyzed. The gas and liquid products were detected through GC and NMR, respectively. Three products, H₂, CO, and formate were detected. After the quantitative determination of the electrolyzed products generated at different potentials, the potential-dependent FEs of the three products, FE_formate, FE_CO, and FE_hydrogen, were calculated (Fig. 5(b) and (c)). The FE_formate of the nBO-nBi electrode reached 81.7% at −0.63 V (Fig. 5(b)), higher than that of the mBO-nBi electrode (68%) under similar conditions. This study suggests that constructing an interlinked catalyst layer and reducing Bi particle size can improve the electrocatalytic activity of Bi for CO₂ reduction at a lower overpotential. The maximum FE achieved for formate generation was 97.2% (−0.74 V) and 97.6% (−0.78 V) for nBO-nBi and mBO-nBi, respectively, demonstrating the high formate selectivity of the nBi catalyst toward CO₂ER, consistent with the previous observation that the FE_formate over the metallic Bi catalyst exceeded 90% [22–23]. Both nBi catalysts in the present study effectively suppressed CO generation, as indicated by an FE_CO of <7% (Fig. 5(c)) in the considered potential range. The nBO-nBi catalyst yielded FE_formate higher than 80% at potentials of −0.63 to −0.98 V. The mBO-nBi catalyst yielded 80% FE_formate at potentials of −0.65 to −1.1 V. When the potential shifted negatively to −1.18 V, the formate selectivity at the nBO-nBi electrode decreased to 52%. In comparison, the ${\mathrm{F}\mathrm{E}}_{{\mathrm{H}}_{2}}$ increased to 46%. However, the FE_formate at the mBO-nBi electrode remained at 74% at −1.18 V.

The partial current of formate generation (j_formate, Fig. 5(d)) at the two nBi electrodes was calculated as the product of the FE_formate (Fig. 5(b)) and the total reduction current (Figs. S2–S3). The nBO-nBi electrode yielded greater j_formate than mBO-nBi at potentials more positive than −0.9 V. However, the nBO-nBi and mBO-nBi electrodes displayed similar j_formate values under potentials more negative than −0.98 V, and the mBO-nBi electrode even outperformed the nBO-nBi electrode at −1.18 V in terms of j_formate (~40 mA⋅cm⁻²), owing to the sharp decrease in FE_formate at the nBO-nBi electrode.

To analyze the reason for the difference in current densities delivered at different nBi electrodes during the CO₂ER, the electrochemical surface areas (ECSAs) of the two nBi electrodes were calculated according to double-layer charging capacities (Fig. 6). The double-layer charging capacity was measured via cyclic voltammetry at the double-layer potential domain (between −0.01 and −0.13 V). The double-layer charging currents for both electrodes were proportional to the scan rate, and the half-gap currents between the anodic and cathodic branches at −0.07 V were plotted against the scan rate (Fig. 6), forming perfect lines (Fig. 6(b) and (d)). The double-layer capacitance of the electrodes can be calculated via linear curve fitting, and it is linearly correlated with the ECSA of the electrode [24]. The double-layer capacitances of the nBO-nBi and mBO-nBi electrodes were approximately 4.45 and 1.98 mF⋅cm⁻², respectively, indicating that the nBO-nBi electrode had an ECSA more than twice that of the mBO-nBi electrode. Hence, the decrease in bismuth nanoparticle size could effectively increase the ECSA of the electrode. However, besides the size and ECSA effects, the electrode structure and porosity may influence nBi catalysts’ activity and selectivity toward CO₂ER [25]. The ECSA and current density are not in a directly proportional relationship, especially at more negative potentials [26]. For example, the ECSA ratio of the nBO-nBi electrode to the mBO-nBi electrode was ~2, but the corresponding CO₂ reduction current ratio was ~1.7 at −0.74 V and 1.1 at −1.08 V. This suggests a comprehensive influence of ECSA and electrode structure on catalyst performance. The nBi electrodes used in this study were porous, presenting a potential mass transfer issue in the electrode membranes. As a result of the smaller primary particles of nBO-nBi, the pores in nBO-nBi should be smaller and more tortuous. Therefore, while the large ECSA of nBO-nBi may provide more catalytic sites, the electrode may face more mass transfer difficulties, especially at larger polarization conditions. In particular, at high overpotentials, when the mass transfer became the rate-determining step, the performance of nBO-nBi might be inferior to that of the mBO-nBi electrode (Fig. 5(d)). In such a case, the local CO₂ concentration in the inner part of the electrode significantly decreased owing to the mass transfer difficulty of CO₂, unlike in the case of water; consequently, the contribution of CO₂ER to the overall electrochemical reaction at the electrode would decrease, and the HER would become increasingly greater with increasing polarization, resulting in a limited formate partial current and a high FE_hydrogen (Fig. 5(c)–(d)) [27]. In contrast, at relatively positive potentials, in which the electron transfer step essentially controls the reaction, the nBO-nBi electrode would display a higher formate partial current than the mBO-nBi electrode. Therefore, at relatively negative potentials (high overpotentials), the mBO-nBi electrode outperformed (Fig. 5) the nBO-nBi electrode, as the latter displayed a lower limiting diffusion current for CO₂ reduction to formate (Fig. 5(d)).

Figure  6.  Cyclic voltammograms (CVs) of the (a) mBO-nBi and (c) nBO-nBi electrodes obtained in a 0.5 M KHCO₃ electrolyte. Plots of current densities against the scan rates: (b) mBO-nBi and (d) nBO-nBi. For each CV, the current density was half the gap between anodic and cathodic branches at −0.07 V.

下载: 全尺寸图片 幻灯片

As previous works have reported the high catalytic activity of bismuth oxide for the reduction of CO₂ to formate [28–29], the electrocatalytic CO₂ER performance of the pristine nano bismuth oxide (nBO) electrode was tested for comparison (Fig. S4). The catalytic performance of the nBO was comparable to those reported in the literature [28–29] but lower than that of nBO-nBi: (i) at low overpotentials (−0.68 V) when the Bi₂O₃ has not been fully reduced (Fig. S5(a)), the nBO electrode delivered an FE_formate of merely 38.5%, only approximately half of the nBO-nBi electrode; (ii) at high overpotentials, the Bi₂O₃ was reduced to Bi (Fig. S5(a)), and the FE_formate significantly increased (89% at −0.78 V and 85% at −0.88 V), but it still slightly inferior to that of nBO-nBi (90.4% at −0.78 V and 91.2% at −0.88 V). The SEM images show that the nBO electrodes exhibited markedly different morphologies after CO₂ER at different potentials (Fig. S5(b)–(h)). Therefore, the solid electroreduction of bismuth oxide to nBi in advance would be beneficial for maintaining the consistency of the electrode for a high electrocatalytic CO₂ER performance.

3.3   Investigation of the CO₂ reduction mechanism

Tafel slopes were recorded to elucidate the reduction mechanism of the CO₂ER at nBi electrodes. Typically, the formate formation mechanism includes the following steps [22,30]:

Reaction (3) reflects the adsorption of CO₂ on the catalyst. The nBO-nBi electrode displayed a Tafel slope of 93.1 mV⋅dec⁻¹ (Fig. 7(a)). In contrast, the Tafel slope of the mBO-nBi electrode was relatively large (100.6 mV⋅dec⁻¹), indicating that increasing the overpotential allows the nBO-nBi electrode to more easily reach a high CO₂ reduction current. The comparison suggests that nBO-nBi exhibited better intrinsic catalytic activity than mBO-nBi. As reported in previous studies, a Tafel slope of 118 mV⋅dec⁻¹ suggests that the CO₂ reduction could be governed by the first electron transfer (reaction (4)) [31]. However, the Tafel slopes of both nBi electrodes were <118 mV⋅dec⁻¹, suggesting that CO₂ reduction to formate was subject to hybrid control by the first electron transfer and the following hydrogenation step [32]. Bicarbonate ion (${\mathrm{H}\mathrm{C}\mathrm{O}}_{3}^{-}$) has been considered the main source of protons according to its pKa (i.e., $-\text{lg}\,{K}_\text{a}$, and K_a is the acid equilibrium constant of a solution) value (10.33) compared with that of water (15.7) [33]; therefore, the influence of the concentration of ${\mathrm{H}\mathrm{C}\mathrm{O}}_{3}^{-}$ on j_formate was investigated. Fig. 7(b) suggests a linear relationship between j_formate and ${\mathrm{H}\mathrm{C}\mathrm{O}}_{3}^{-}$ concentration, with a slope of ~1.2, indicating a first-order dependence of ${\mathrm{H}\mathrm{C}\mathrm{O}}_{3}^{-}$ in the rate equation for formate generation. This further suggests that the protonation or the proton-coupled electron transfer reaction (reaction (5)) was involved in the kinetic control of CO₂ reduction to formate [12,34].

Figure  7.  (a) Tafel plots of formate partial current against overpotential at different nBi catalysts; (b) logarithmic plot of the formate partial current and the $\mathbf{\left[\mathbf{H}\mathbf{C}\mathbf{O}_3^-\right]}$ concentration at the nBO-nBi electrode (measured at −0.78 V); (c) Nyquist plots of impedance measured at −0.78 V at different nBi catalysts in a CO₂-saturated 0.5 M KHCO₃ electrolyte (100 kHz to 1 Hz, amplitude: 10 mV); (d) equivalent circuit used for fitting the data shown in Fig. 7(c).

下载: 全尺寸图片 幻灯片

Charge-transfer information was explored via electrochemical impedance spectroscopy (Fig. 7(c)), and kinetic parameters were extracted by fitting the impedance spectra with a widely used equivalent circuit (Fig. 7(d)). The charge-transfer resistance (R_ct) values of two electrodes were higher than the solution and ohmic resistance, indicating that the charge-transfer step at −0.78 V was dominant in the CO₂ER process (Table 1). The mBO-nBi electrode exhibited a higher R_ct than the nBO-nBi electrode, consistent with the smaller ECSA of mBO-nBi. However, as discussed above, the formate FE of mBO-nBi during CO₂ER was also lower than that of nBO-nBi, indicating that reducing the particle size of nBi would increase the formate selectivity of the catalyst. These findings align with the Tafel plots. Hence, constructing a three-dimensional connected catalyst layer with small size nBi could improve both reaction kinetics and product selectivity [35].

Table  1.  Data obtained after fitting Nyquist plots in Fig. 7(c) (R_s, R₂, and R_ct denote solution resistance, ohmic resistance, and charge-transfer resistance, respectively.)

Electrode	Rs / Ω	R2 / Ω	Rct / Ω
mBO-nBi	2.9	5.4	25.7
nBO-nBi	3.0	4.3	22.8

下载: 导出CSV 

| 显示表格

3.4   Long-term stability of nBi catalysts

Stability is another important criterion in developing electrode catalysts for practical applications. To evaluate the long-term stability of the prepared nBi catalysts, continuous electrolytic reduction of CO₂ was performed through the application of a potential of −0.78 V for 20 h. Some slight fluctuations of current density occurred during the electrolysis process because H₂ and CO were often generated during the CO₂ER process (Fig. 8(a) and (b)). The adsorption and evolution of gas generally lead to current fluctuation, and the higher the gas generation current, the larger the current fluctuation [7,28]. During the long-time CO₂ reduction, the total current of both electrodes revealed no pronounced attenuation, but the FE_formate of the mBO-nBi electrode decreased (Fig. 8(a)), with a retention of 78%. However, the nBO-nBi electrode performed stably (Fig. 8(b)), with almost no FE_formate change observed during the 20 h electrolysis.

Figure  8.  Total current and product efficiencies during electrolysis of CO₂ at −0.78 V at the (a) mBO-nBi and (b) nBO-nBi electrodes. XRD patterns of (c) mBO-nBi and (d) nBO-nBi after the 20 h electrolysis of CO₂ at −0.78 V; (e, f) SEM images of the electrodes characterized in (c) and (d), respectively.

下载: 全尺寸图片 幻灯片

After long-term electrolysis, the electrodes were analyzed via XRD and SEM. Before and after the electrolysis, none of the electrodes exhibited any change in XRD diffraction peaks, indicating they were phase-stable during the electrolysis (Fig. 8(c) and (d)). The SEM results indicate that the particle sizes of mBO-nBi (Fig. 8(e) vs. Fig. 3(a)) and nBO-nBi (Fig. 8(f) vs. Fig. 3(c)) increased after the 20 h electrolysis. The size increase was probably the main reason for the decrease in the formate selectivity of the mBO-nBi electrode. In contrast, while the Bi nanoparticles in the nBO-nBi electrode agglomerated and the size of the Bi particle increased, the FE_formate changed little, suggesting that there was a size range for the Bi to achieve a high FE_formate; for example, both the mBO-nBi (~80 nm) and nBO-nBi (30–50 nm) electrodes exhibited similar FE_formate values in the potential range of −0.65 to −1.0 V. However, the stability test results suggest that a further increase in the particle size of nBi beyond the aforementioned range would result in a rapid drop in FE_formate. Studies have often attributed the morphological changes of electrode catalysts after long-term electrolysis to the surface reorganization of the catalysts, probably induced by the surface adsorption/desorption, related to the strong interaction between the Bi metal and CO₂ or its derivations [12,26,36]. The long-term instability of the nBi catalysts needs to be investigated further. Nevertheless, during the 20 h CO₂ER test, the nBO-nBi electrode displayed high activity, durability, and selectivity. Moreover, the mBO-nBi and nBO-nBi electrodes exhibited comparable and even higher catalytic activity compared with other formate catalysts recently reported in the literature (Table S1) [37–40]. This demonstrates the application potential of the oxide-derived nBi catalysts for CO₂ER.

4.   Conclusion

In summary, a nanostructured Bi catalyst was prepared through the direct solid electroreduction of Bi₂O₃ in a K₂CO₃ solution. Both mBO-nBi and nBO-nBi electrodes, derived from micronized and nanosized bismuth oxide, respectively, exhibited a hierarchical structure and an integrated conductive network and demonstrated high CO₂ reduction activity and formate selectivity. They achieved high formate FEs (>80%) in a wide potential window, between −0.63 and −0.98 V for the nBO-nBi electrode and between −0.65 and −1.08 V for the mBO-nBi electrode. However, at lower polarization conditions, nBO-nBi exhibited a higher formate production rate and FE than mBO-nBi. This suggests the intrinsic high catalytic activity of the smaller nanoscale Bi particles toward CO₂ reduction to formate. This study provides a novel approach for fabricating interlinked nanostructured catalysts, which may prove helpful for designing high-efficiency catalysts in the future.