Efficient and stable acidic CO2 electrolysis to formic acid by a reservoir structure design Authors: Li-Ping Chi, Zhuang-Zhuang Niu, Yu-Cai Zhang, Xiao-Long Zhang, Jie Liao, Zhi-Zheng Wu, Peng-Cheng Yu, Ming-Hui Fan, Kai-Bin Tang, Min-Rui Gao Published: 2023 ABSTRACT SignificanceWe show that vertically grown bismuth nanosheets on a gas-diffusion layer can create numerous cavities as “electrolyte reservoirs,” which confine in situ–generated hydroxide and potassium ions and concurrently limit inward proton diffusion, thereby creating locally alkaline environments. The concept of electrolyte reservoir has been implemented in efficient CO2-to-HCOOH conversion in acidic media (pH ≤ 2) on sheet-like bismuth (Bi) catalyst, which enables HCOOH production at partial current density of 471 mA cm–2, FE of 96.3%, and SPCE of 79%. Electrosynthesis of carbon-based fuels and feedstocks via carbon dioxide reduction (CO2R) offers a sustainable route to store intermittent renewable electricity while reducing CO2 emissions (1, 2). Operating CO2R in alkaline or neutral electrolytes has enabled remarkable advances in Faradaic efficiency (FE) and current density (3–7); however, the strong local alkaline environments lead to carbonate formation through the reaction of input CO2 with OH− (2OH− + CO2→CO32− + H2O), which imposes a limited single pass carbon efficiency (SPCE; fraction of carbon in the CO2 feed converted to desired products) that is unacceptably low (8). For example, as to the 2-electron CO2-to-HCOO− conversion, the maximum SPCE is limited to 50% because every HCOO− generates at cathode will accompany with one CO2 consumed to form carbonate (8, 9). The regeneration of CO2 from carbonate in alkaline CO2 electrolysis is possible, but this process requires more than 50% of input energy, raising a substantial energy penalty (10). Moreover, carbonate cross-over to the anode and the consequent CO2 release incur further cost profile improvement for the CO2 separation from the anodic O2 stream (11, 12). Switching CO2R from alkaline or neutral electrolytes to acidic environment provides a fundamentally different avenue to address the CO2 utilization limit issue (13–16). In the acidic environment, no OH− is formed when hydronium (H3O+) acts as the proton source for CO2R; thus, no CO2 is consumed to generate carbonate (13). Despite carbonate can be produced when H2O offers protons for CO2R, the protons in the bulk electrolyte will convert the as-formed carbonate back to CO2 within the diffusion layer (17). Therefore, operating CO2R in acid could in principle minimize carbonate formation and consequently mitigate CO2 cross-over (18). However, the rate of the competing hydrogen evolution reaction (HER) in acidic electrolytes is kinetically much faster than that of CO2R, which has resulted in near 100% H2 selectivity over catalysts those are efficient for CO2R in alkaline and neutral media (19–21). To suppress HER in acid, a local proton-depletion microenvironment surrounding the cathode is highly desirable. The local high pH condition can be realized by performing CO2R at large current densities (e.g., >1 A cm−2) (10); as such, the consumption of local protons by in situ–generated OH− will surpass the influx of protons from the bulk, leaving locally concentrated OH−. Nevertheless, engendering ampere-scale current densities requires very high voltages, which not only leads to low energy efficiencies, but also causes overheating in the stack of CO2 electrolyzers, as well as issue with water management (22, 23). Toward practical implementation, CO2R is ideally operated at 200–500 mA cm−2 (23); unfortunately, at these moderate current densities, the mass transport of protons from the diffusion layer is faster than their local consumption; thus, creating a locally alkaline environment becomes difficult (13). Very recently, Zhao and coworkers demonstrated efficient CO2R in acid over current densities of 100–400 mA cm−2 by introducing a heterogeneous catalyst adlayer that can block proton transport from the bulk solution (24). This design, although interesting, would potentially increase electrode resistance and the stack cost because the use of insulating polymer nanoparticles and a perfluorinated sulfonic-acid ionomer to construct the adlayer (5 μm thick). We pursued direct catalyst design that can constrain proton flux and create proton-depleted local environments without using additives. For conventional particulate catalyst, the mass transfer resistance of electrolytes is small, which results in fast replenishment of protons surrounding the catalyst and consequent a proton-rich environment (Fig. 1A). We sought a cathode with plenty of cavities created by vertically growing sheet-like catalyst onto gas-diffusion layer (GDL), which would construct electrolyte reservoirs that prevent both the outward OH− diffusion and the inward proton diffusion, thus yielding local high-alkalinity regions (Fig. 1B). Herein, the concept of electrolyte reservoir has been implemented in efficient CO2-to-HCOOH (a product can be synthesized from CO2R only in acid) (25) conversion in acidic media (pH ≤ 2) on sheet-like bismuth (Bi) catalyst, which enables HCOOH production at partial current density of 471 mA cm–2, FE of 96.3%, and SPCE of 79%. Computed local pH environments in acidic medium on particulate and reservoir-like catalysts. (A and B) Schematics of interfacial reactions and proton transport near the surfaces of particulate (A) and reservoir-structured (B) catalysts. (C and D) Concentration profiles of OH– (C) and H+ (D) species on the surface of different catalysts. Arrows indicate the direction of species migration (Scale bars, 200 nm). (E) Local concentration of various species on the surface of different modeled catalysts. (F) pH distribution of different catalysts in the diffusion layer. The simulations were carried out at –1.7 V vs. SHE with a bulk pH of 1.69. Results and Discussion Computational Simulations. We started with carrying out COMSOL simulations based on a two-dimensional reaction–diffusion model to explore the impacts of reservoir structure on the local pH and cation environments (see Materials and Methods for details). We built cavities by vertically oriented sheets with different lengths to represent electrolyte reservoirs. Our simulation results reveal that the in situ–generated OH− migrated to the bulk solution and protons (H+) diffused to the catalyst surface (arrows in Fig. 1 C and D), a result of concentration gradient and negative electric field. Increasing the cavities of the reservoir structures leads to a substantial increase of local alkalinity (Fig. 1C), while the H+ ions nearby the electrode surface were greatly depleted (Fig. 1D). The locally improved pH environment is caused by the reservoir structure, which can confine in situ–produced OH− ions and prevent inward H+ diffusion from the bulk solution. Once H+ ions in the electrolyte reservoir were consumed, the in situ produced OH− ions will accumulate in the reservoir, creating high local alkalinity. This is in stark contrast to particulate catalyst that gives an open environment, in which the locally consumed H+ ions can be rapidly replenished by the protons in the bulk acidic electrolyte; consequently, a locally alkaline condition is barely achieved (Fig. 1 C and D). Increasing cavities in our simulation can locally enrich OH− ion concentration in the reservoir from 0.047 to 0.7 mol L−1, whereas H+ ion concentration experiences about 15-fold decrease (Fig. 1E). Meanwhile, compared with particulate catalyst, a 1.4-fold increased surface-adsorbed K+ ion concentration was observed in the reservoir owing to the electrostatic field and the confined structure (Fig. 1E and SI Appendix, Fig. S1). In Fig. 1F, the pH as a function of distance from the cathode was simulated for the above four models within a diffusion layer of 150 μm and a bulk pH of 1.69. We found that the local pH of the reservoir catalyst can reach 14, but the local alkalinity for the particulate catalyst decreased to pH ~12.7. A similar trend holds for both OH− and K+ ions (SI Appendix, Fig. S2). Conversely, for H+ ions, obvious proton-depleted local environments, plotted on SI Appendix, Fig. S2B, were observed when the model changed from particulate to reservoir-like catalysts. Additionally, our simulations demonstrated that the local pH and concentrations of OH− and K+ ions nearby the reservoir are positively correlated to the bulk pH (SI Appendix, Figs. S3–S5). For example, a bulk pH of 2 leads to local pH of 14.2; this value decreases to 13.5 when the bulk pH is 1.47. All these results, taken together, suggest the effectiveness of the reservoir design as a strategy to create high local alkalinity and K+-enriched environment, thus inhibiting the formidable HER in acid. Catalyst Synthesis and Characterization. To test the predictions experimentally, we choose to study the CO2-to-HCOOH conversion given that HCOOH can solely be electrosynthesized from CO2R in acid. We first prepared Bi nanoparticles (Bi NPs) by the reduction of Bi(NO3)3·5H2O using NaBH4 (SI Appendix, Fig. S6), which we then dispersed in isopropanol and spray-coated onto the GDL. Further immersing the Bi NPs-decorated GDL into 1 M KHCO3 was found to result in vertically grown nanosheets (Fig. 2A). The transformation process of Bi NPs to nanosheets is very fast, which constructs uniform and continuous electrode surface with favorable cavities at 60 s (SI Appendix, Fig. S7), but prolonged immersion (e.g., 120 s) led to rough surface and reduced cavities. X-ray diffraction (XRD) patterns in Fig. 2B show the phase transformation of hexagonal Bi to tetragonal Bi2O2CO3 (JCPDS 41-1488). This fast structural phase transformation was further monitored by in situ Fourier-transform infrared spectroscopy (FT-IR), which arises a band at 1,393 cm−1 once the immersion occurs, corresponding to ν(CO32−) of Bi2O2CO3 (Fig. 2C) (26, 27). The intensity of this band reaches maximum at about 60 s, indicative of the complete phase transformation. Prior studies have reported the transformation of Bi to Bi2O2CO3, but they often require very long reaction time (28). In this scenario (Fig. 2D), Bi NPs were first oxidized to [Bi2O2]2+ polycation layers in aqueous KHCO3 by dissolved oxygen (SI Appendix, Fig. S8). Due to the large interlayer distance between [Bi2O2]2+ layers, there are open channels for the insertion of CO32− groups from the solution (29). Then, the marked resemblance of Bi and O atomic arrangements in [Bi2O2]2+ layers and in Bi2O2CO3 allowed this phase transformation reaction to proceed extremely fast (30). Formation and physical characterization of the reservoir-like Bi2O2CO3 precursor. (A) SEM images acquired at different immersion times show the formation of abundant cavity structure [Scale bars, 200 nm (Up) and 100 nm (Down)]. (B and C) XRD patterns (B) and time-resolved in situ FT-IR spectra (C) of Bi NP upon immersion in 1 M KHCO3 for different times. (D) Schematic of the fast phase transformation from Bi to Bi2O2CO3. (E and F) HAADF-STEM images of Bi2O2CO3 (60 s). The Inset in (E) shows the corresponding FFT pattern (Scale bars, 1 nm). (G and H) AFM image (G) and corresponding height profile (H) of Bi2O2CO3 (60 s) (Scale bar, 100 nm). Atomic-resolution high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) revealed the high-crystalline structure of the as-obtained Bi2O2CO3 (60 s) with continuous lattice fringes (Fig. 2 E, Inset). A lateral view of the Bi2O2CO3 nanosheets along the thickness direction exhibited well-defined multilayered structure, where the interlayer distance was measured to be 0.684 nm (Fig. 2F). Using atomic force microscopy (AFM) measurements, we determined the thickness of the nanosheet to be ~4.85 nm (Fig. 2 G and H). CO2R Performances in Acidic Electrolytes. To verify the predicted CO2R enhancement by the reservoir structure, we chose to focus on Bi2O2CO3 nanosheets synthesized at 60 s, which have richer cavities as potential electrolyte reservoirs. Before CO2R measurements, Bi2O2CO3 on GDL was experienced an electrochemical reduction process (see Materials and Methods for details), which was readily reduced back to metallic Bi, as confirmed by in situ Raman spectra, XRD and HAADF-STEM (SI Appendix, Fig. S9). Close inspection of the electrode by scanning electron microscopy (SEM) image revealed that the cavities on GDL were well-preserved after reduction but exhibiting increased surface roughness (SI Appendix, Fig. S10). Hereafter, we abbreviate this sample as Bi RS (Bi reservoir structure), which was determined to be have normalized roughness factor of 11.1, about double of that for Bi NP (SI Appendix, Fig. S11). CO2 electroreduction activity of Bi RS and NP was examined in CO2-saturated 0.1 M H2SO4 with 0.5 M K2SO4 (electrolyte pH 1.69). Solution-phase and gas-phase products were quantified by NMR spectroscopy and online gas chromatography (SI Appendix, Figs. S12), respectively. Under this acidic condition, chemical shift of HCOOH at ~8.18 ppm was detected instead of that at ~8.36 ppm for formate in neutral electrolytes (SI Appendix, Figs. S13 and S14) (31, 32). Fig. 3A shows that Bi RS exhibited much greater FE to HCOOH than that of Bi NP at all potentials examined. At −1.7 V vs. the standard hydrogen electrode (SHE, unless otherwise stated, all potentials reported are vs. SHE), a maximum FE of 95.5% toward HCOOH was achieved over the Bi RS catalyst, whereas the H2 generation was substantially suppressed below 3% FE (Fig. 3A and SI Appendix, Fig. S15). Bi RS still attained a large HCOOH FE of >90% at −2.2 V, which afforded a CO2 partial current density toward HCOOH of 457 mA cm−2 and a HCOOH production rate of 8.5 mmol cm−2 h−1 (Fig. 3B and SI Appendix, Fig. S16). Conventional Bi NP, by contrast, exhibited a considerable decrease of the HCOOH selectivity and partial current density (Fig. 3 A and B), but a promoted HER (SI Appendix, Fig. S15). These results indicate the role of the reservoir structure in creating a local proton-depletion microenvironment and thereby suppressing H2 generation. CO2R performances in acidic media. (A and B) The potential-dependent Faradaic efficiency (A) and partial current density (B) of HCOOH on Bi RS and Bi NP catalysts in 0.1 M H2SO4 with 0.5 M K2SO4 (pH 1.69). (C and D) HCOOH Faradaic efficiency (C) and partial current density (D) of Bi RS and Bi NP catalysts at 200 mA cm–2 in acidic electrolytes with different pHs. (E) Stability test of Bi RS (red) and Bi NP (purple) catalysts at 200 mA cm–2 in acidic electrolytes with different pHs. (F) Comparison of HCOOH Faradaic efficiency, partial current density, production rate, and operating lifetime on different catalysts reported for acidic CO2R. The error bars represent the SD of three independent measurements. We investigated the impact of electrolytes with different pHs on the CO2-to-HCOOH conversion (Fig. 3 C and D and SI Appendix, Figs. S17–S21). In CO2-saturated 0.05 M H2SO4 with 0.5 M K2SO4 (pH 2), the HCOOH selectivity and partial current density were similar for the Bi RS and NP catalysts at a current density of 200 mA cm−2. The observations demonstrated that Bi RS and NP have similar CO2-to-HCOOH catalytic ability when proton flux was limited in higher pH electrolytes. The advantages of the reservoir design become prominent when the electrolyte pH decreases. For example, in CO2-saturated 0.15 M H2SO4 with 0.5 M K2SO4 (pH 1.47), the Bi RS catalyst exhibited >90% FE for HCOOH production and partial current density of 182 mA cm−2 (Fig. 3 C and D); however, on the Bi NP catalyst, the two performance metrics were markedly decreased to 41% and 83 mA cm−2, respectively. The vital role of reservoir structure in establishing a local proton-depletion microenvironment for suppressing the HER was further confirmed by long-term stability measurements. At present, CO2R stability in acidic electrolytes remains a huge challenge because the kinetically more favorable HER substantially outperforms CO2R (33, 34). Although locally alkaline condition can be obtained at high-rate CO2 electrolysis, the constructed microenvironments often show problems with lability; therefore, HER can quickly dominate the reaction again after mere a short time of operation. Fig. 3E shows the FE toward HCOOH as a function of testing time, under acidic electrolytes with different pHs, for the Bi RS and NP catalysts. The HCOOH FE on Bi NP catalyst undergoes a significant drop within several hours in all acidic electrolytes examined. This degradation is expected, due to the fast proton transport from the bulk solution to particulate catalytic surface, thus overwhelmingly benefiting the HER. Bi RS catalyst also shows drop in HCOOH selectivity, but at a substantially slower rate in comparison to the Bi NP counterpart (Fig. 3E and SI Appendix, Fig. S22). We surmise that the much better stability of the Bi RS catalyst could be due to the reservoir structure that, to a large degree, can prevent both the outward OH− diffusion and the inward proton diffusion and thus build stable local high-alkalinity regions. After long-term stability tests (e.g., operating at 200 mA cm−2 in 0.05 M H2SO4 with 0.5 M K2SO4 for 45 h), the Bi RS catalyst largely maintained the reservoir structure (SI Appendix, Fig. S23). Moreover, the bulk pH remained almost unchanged after the stability tests, suggesting a stable bulk electrolyte environment (SI Appendix, Fig. S24). We also examined Bi dissolution after acidic CO2 electrolysis using the inductively coupled plasma atomic emission spectroscopy (ICP-AES), showing that the dissolved Bi from Bi RS is negligible (SI Appendix, Fig. S25), consistent with previous observations that Bi exhibits good stability in acid (35, 36). We also compared the performance metrics of our Bi RS catalyst with those previously reported in terms of HCOOH FE, partial current density, production rate, and operating lifetime. Under acidic testing conditions, the results obtained on the Bi RS catalyst surpass prior catalysts with different structures achieved by other approaches (Fig. 3F and SI Appendix, Table S4) (9, 25, 37–41), demonstrating the effectiveness of the reservoir design in acidic CO2R. Next, we examined experimentally whether high local alkalinity was realized at the Bi RS catalyst using multiple characterization techniques. We first recorded the linear sweeping voltammetry (LSV) curves of Bi RS (Fig. 4A) and Bi NP (Fig. 4B) catalysts in N2-saturated acidic electrolytes with different pHs, where HER was the sole reaction. Both the two catalysts exhibited clear plateaux of current density between –1.0 and –1.5 V, which were the result of proton diffusion limitation and increased in magnitude with decreasing bulk pH from 2 to 1.47 (14). At more negative potentials, water reduction dominates the reaction instead of proton reduction. Compared with the Bi NP, our Bi RS catalyst yielded smaller plateau currents at all pHs studied. For example, in an electrolyte with pH 1.47, the diffusion-limiting current density was 50.4 mA cm–2 on Bi RS, much smaller than that of 64.7 mA cm−2 obtained on the Bi NP catalyst. These results indicate a lower local concentration of protons surrounding the Bi RS surface owing to its reservoir structure that confines in situ–generated OH− and constrains inward H+ diffusion. Mechanistic investigation. (A and B) LSV curves for Bi RS (A) and Bi NP (B) catalysts in N2-saturated acidic electrolytes with different pHs. (C) Surface pH values measured from in situ Raman spectra with respect to the applied current density on Bi RS and Bi NP catalysts. (D and E) In situ ATR-SEIRAS spectra on Bi RS (D) and Bi NP (E) catalysts as a function of the applied potential. We further performed in situ Raman spectroscopy measurements on Bi RS and Bi NP catalysts to probe their surface local pH during CO2R by a method described previously (see Materials and Methods for details; SI Appendix, Figs. S26 and S27) (42–45). Fig. 4C shows that in acidic electrolytes with different pHs, the surface pH of the Bi RS catalyst raises as the current density increases owing to more produced OH− ions (SI Appendix, Fig. S28) (46). Although similar trends were also observed over the Bi NP catalyst, the increase of surface pH was rather limited, likely caused by fast proton influx toward this open catalytic surface. In an electrolyte with pH 1.47, the surface of the Bi RS catalyst can reach an alkalinity of pH 10.2 at 100 mA cm−2 (Fig. 4C). By stark contrast, the Bi NP catalyst exhibited a much smaller pH that is out of pH range of CO32−/HCO3− equilibrium (pH < 8.8) even at 120 mA cm−2. Seeking to understand the enhanced CO2-to-HCOOH conversion over Bi RS catalyst, in situ attenuated total reflection surface-enhanced infrared absorption spectroscopy (ATR-SEIRAS) measurements were also conducted to monitor the possible reaction intermediates. Scanning the potentials from −1.0 to −1.45 V on Bi RS (Fig. 4D) and Bi NP (Fig. 4E) electrodes both resulted in several bands at 1,640 cm−1 (vibration of adsorbed water, δH2O), 1,553 cm−1 (νCO32−), and 1,338 cm−1 (νHCO3–) (47, 48). Intriguingly, a band appeared at 2,343 cm−1 (antisymmetric stretching vibration of dissolved CO2) (49) was detected on the Bi RS catalyst, which becomes sharper at more negative potentials, suggesting the consumption of adsorbed CO2 and thus a faster CO2R kinetics (50). Furthermore, both Bi RS and Bi NP exhibited a band at 1,395 cm−1, which can be attributed to the vibration of O-C-O (νO-C-O) of the bidentate *OCHO, a key intermediate that has been proposed during CO2-to-HCOOH conversion (51, 52). Compared with the band appeared at −1.0 V for Bi RS, a much higher potential >−1.2 V was required to form this band on the Bi NP catalyst, revealing that the reservoir structure thermodynamically favors *OCHO formation. Altogether, these observations establish that the reservoir design creates local high-alkalinity regions, which suppress the competing HER and promote CO2-to-HCOOH conversion in acid. Full-cell HCOOH Electrosynthesis. Last, we sought to perform the electrosynthesis of HCOOH from CO2 in a continuous-flow electrolyzer with a slim catholyte chamber, as depicted in Fig. 5A (SI Appendix, Fig. S29). The Bi RS catalyst was incorporated on the GDL as the cathode, with O2 evolution from acidic water as the corresponding anodic reaction (see Materials and Methods for details). At cell currents of 0.5–1.5 A, the FE toward HCOOH all surpassed 90%, with a maximum of 95.8% at 1.25 A (Fig. 5B). With this FE, a partial current toward HCOOH of 1.2 A was achieved at a moderate cell voltage of −4.56 V (Fig. 5C). The Bi RS catalyst enables a full-cell energy efficiency for conversion to HCOOH of ~40% at −2.96 V (Fig. 5C), exceeding prior catalysts reported in acids (16, 33). Continuous-flow electrosynthesis of HCOOH in acidic electrolyte. (A) Expanded view of the slim continuous-flow electrolyzer configurations. (B) Current-dependent Faradaic efficiencies for products in 0.05 M H2SO4 with 0.5 M K2SO4 (pH 2). (C) Full-cell voltage, partial current, and energy efficiency for HCOOH in 0.05 M H2SO4 with 0.5 M K2SO4 (pH 2). (D) Comparison of Faradaic efficiency of HCOOH in acidic electrolyte or formate in neutral and alkaline electrolytes at a current of 0.3 A. (E) The corresponding SPCE calculated from (D). (F) Stability test in 0.05 M H2SO4 with 0.5 M K2SO4 (pH 2) at an applied current of 0.5 A. (G) The 1H NMR spectrum of pure HCOOH aqueous solution. The inset shows the photograph of 360 mL pure HCOOH aqueous solution (0.91 M). The error bars represent the SD of three independent measurements. By reducing the flow rate of CO2 from 30 to 1 standard cubic centimeters per minute (sccm), the FE toward HCOOH in acid gradually surpassed that in neutral (pH 8.4) and alkaline (pH 13.8) electrolytes (Fig. 5D), caused by the greatly consumed input CO2 through reaction with OH− in neutral and alkaline conditions. SPCE analyses in Fig. 5E demonstrate that both neutral and alkaline electrolytes delivered SPCE below the theoretical limit of 50%. By contrast, a high SPCE of 79% for the HCOOH was achieved at 1 sccm in acidic electrolyte. The SPCE observed over Bi RS surpasses previously developed catalysts for CO2-to-HCOOH conversion in acidic CO2R electrolyzers (16, 25, 33). In Fig. 5F, we conducted a long-term continuous test under a fixed cell current of 0.5 A in the slim continuous-flow electrolyzer. It exhibited a stable HCOOH selectivity of >90% in acidic medium over 50 h, representing the electrosynthesis of HCOOH with an amount of 446 mmol. Post-mortem SEM analyses revealed that the reservoir structure largely maintained (SI Appendix, Fig. S30). By distillation of the collected catholyte at 101 °C (SI Appendix, Fig. S31), we harvested 360 mL pure HCOOH aqueous solution with a concentration of 4.2 weight % (0.91 M) (Fig. 5G). Our 1H and 13C NMR spectroscopy measurements confirmed the ultrahigh purity of the synthesized HCOOH aqueous solution (Fig. 5G and SI Appendix, Fig. S32). Discussion In conclusion, we demonstrate direct construction of cavities created by the vertically grown bismuth nanosheets on the GDL. These cavities could serve as ideal electrolyte reservoir into which in situ–generated OH− and K+ ions can be confined and concentrated. Meanwhile, protons from the bulk solution cannot fill out the reservoir until water is consumed by CO2R. This design results in an electrode with locally high alkaline environment, which considerably suppresses the HER in acidic media and enables efficient electrosynthesis of HCOOH from CO2 with high production rate, selectivity, SPCE, and operating lifetime. We envisage that the electrolyte-reservoir concept for creation of locally high-alkalinity regions will be generally applicable to other selective CO2R catalysts so that a variety of value-added chemicals can be synthesized through CO2 electrolysis in acid. Materials and Methods Synthesis of Bi Nanoparticles. The Bi nanoparticles were prepared through a facile reduction method. First, 0.8 mmol Bi(NO3)3·5H2O and 0.8 mmol polyvinylpyrrolidone (PVP) were dissolved into 80 mL absolute ethylene glycol saturated by Ar for 30 min. Then, 12 mL ethanol solution of NaBH4 (8 mmol) was added dropwise to the precursor solution in 10 min under ice bath condition. After stirring for 5 min, the obtained Bi nanoparticles were immediately washed with excess deionized water (DIW) and absolute ethanol at least three times and then dried at ambient temperature in an oven under vacuum. Synthesis of B2O2CO3 Nanosheets. The Bi2O2CO3 nanosheets were rapid transformed from Bi nanoparticles on the GDL substrate. In detail, 10 mg Bi nanoparticles were dispersed in 1 mL isopropanol with ultrasound. The resultant ink was uniformed sprayed on the GDL (YLS 30 T, purchased from Fuel Cell Store) of 3 × 3 cm2 in area by using an airbrush, yielding the prepared electrode with a catalyst loading of ~1.0 mg cm−2. The obtained Bi nanoparticles on GDL were immersed into 1 M KHCO3 for different time. The GDL was immediately taken out at wanted time, rinsed by DIW, and dried under Ar stream. The as-prepared Bi2O2CO3 nanosheets on the GDL were then directly used for further characterization and electrochemical measurements. Material Characterizations. XRD was collected by a Japan Rigaku DMax-γA X-ray diffractometer with Cu Kα radiation (λ = 1.54178 Å). The morphologic characteristics of the samples were captured by SEM (GeminiSEM 450 and Zeiss Supra 40). The TEM and HAADF-STEM images were performed on Talos F200X. ICP-AES data were obtained by an Optima 7300 DV instrument. The AFM characterization was carried out with an atomic force microscope (Dimension Icon). Raman spectra were measured on a Raman microscope (LabRam HR Evolution) excited with a 532 excitation laser. FT-IR spectra were collected on a Thermo Scientific Nicolet iS50. Electrochemical Measurements. The performance of CO2 electroreduction was tested in a flow cell with a three-electrode setup. The catalysts on the GDL were used as the working electrode, and the exposed area is 1 cm2. Iridium oxide loaded titanium mesh (IrOX-Ti) was employed as the counter electrode. An Ag/AgCl electrode in saturated KCl was used as the reference electrode. The anolyte and catholyte chambers were separated with a cation exchange membrane (Nafion 115, DuPont). The anolyte is 0.5 M H2SO4, while the catholyte is 0.5 M K2SO4 with different concentrations of H2SO4 (0.05/0.1/0.15 M). The pH of 0.05, 0.1, and 0.15 M H2SO4 with 0.5 M K2SO4 were 2.00, 1.69 and 1.47, respectively, monitored by pH-meter (FiveEasy Plus). Both catholyte and anolyte flow rates were maintained constant at 20 mL min–1 controlled by a peristaltic pump (BT100-2 J, Longer Pump). Gaseous CO2 (99.999%) was connected to the inlet of the gas chamber at 30 sccm by a mass flow controller (C100L, Sierra). The outlet CO2 flow rate was monitored by a mass flowmeter (AST10-HLC, Asert Instruments) before flowing to gas chromatography. Before the electrochemical measurements, all the catalysts were electrochemically reduced in CO2-saturated 0.05 M H2SO4 with 0.5 M K2SO4 at 50 mA cm–2 for 10 min. The double-layer capacitances for different catalysts were tested in N2-saturated 0.05 M H2SO4 with 0.5 M K2SO4. The linear sweep voltammetry (LSV) experiments were scanned in N2-saturated 0.5 M K2SO4 electrolyte with different concentrations of H2SO4 (0.05/0.1/0.15 M) at the scan rate of 50 mV s–1. The potential of the working electrode vs. SHE was calculated with iR correction according to:[1]Evs SHE=EvsAgAgCl+0.197-iRcorr, where the solution resistance R was determined by electrochemical impedance spectroscopy (EIS) at –0.6 V vs. SHE over a frequency range from 100 KHz to 10 mHz at the amplitude of the sinusoidal voltage of 5 mV. The iR correction (iRcorr) values of different current densities were summarized in SI Appendix, Table S3, where 85% positive feedback compensation was applied to the measured iR. The potential vs. RHE was calculated according to:[2]E(vs RHE)=E(vs SHE)+0.0591 V × pH. Full-cell HCOOH electrosynthesis was performed in a continuous-flow electrolyzer with a slim catholyte chamber of 3 mm for flowing acids of 0.05 M H2SO4 with 0.5 M K2SO4. The anolyte is 0.5 M H2SO4. The catalysts on the GDL were used as the working electrode, and the exposed area is 5 cm2. The IrO2 (Premetek Co.) was ultrasonically dispersed in 1 mL isopropanol with 50 μL Nafion, and then the ink was uniformed sprayed on Nafion 115 membrane with a load of ~1.5 mg cm–2. The anode (5 cm2) was prepared by hot-pressing an IrO2/Nafion onto a porous platinum loaded titanium mesh (Pt-Ti). The full-cell voltages were obtained without iR correction. The SPCE was measured in the continuous-flow electrolyzer with an exposed area of 1 cm2 (only for SPCE tests). In the neutral media, both anolyte and catholyte were 1 M KHCO3 (pH 8.4). In the alkaline media, both anolyte and catholyte were 1 M KOH (pH 13.8). CO2R Product Analysis. The gas products were quantified by gas chromatography (GC, Trace-1300, ThermoFisher) equipped with a thermal conductivity detector (TCD) for analyzing H2 and a flame ionization detector for analyzing CO. The Faradaic efficiency for gas products (FEx) was calculated by the following formula:[3]FEx% =nx×Cx×V×FI×100%, where F is the Faraday constant (96,485 C mol–1), I is the total current, nx is electrons transferred for reduction to product x, Cx is the mole fraction of the product x, and V is the outlet gas flow rate. The HCOOH products were quantified by 1H NMR (Bruker 400 MHz spectrometer) with water suppression, in which 400 µL of electrolyte was mixed with 200 µL of D2O containing 50 ppm (m/m) dimethyl sulphoxide (DMSO) as the internal standard. The concentration of HCOOH was determined by the standard curve of HCOOH to DMSO peak area ratio. The Faradaic efficiency of the HCOOH (FEHCOOH) can be calculated by the following equation:[4]FEHCOOH%=2×F×nHCOOHI×t×100%, where nHCOOH is the moles of HCOOH product, and t is the CO2 electrolysis time. The full-cell energy efficiency (EE) of the HCOOH product was calculated using[5]EE%=(1.23-EHCOOH)×FEHCOOHEcell, where EHCOOH is the theoretical equilibrium potential (–0.2 V vs. RHE) for the HCOOH formation (53), and Ecell is the full-cell voltage without iR correction. SPCE toward HCOOH was determined using the following equation:[6]SPCE=IHCOOH×60s/2×FV×1 min/Vm, where IHCOOH is the partial current (A) of HCOOH, and Vm is the unit molar volume of gas at 25 °C (24.5 L mol–1). The production rate for HCOOH was calculated using the following equation:[7]Production rate = Q×FEHCOOHF×2×t×S, where Q is the total charge passed, and S is the geometric area of the electrode (1 cm2). In Situ FT-IR. In situ FT-IR experiments for rapid transformation of Bi to Bi2O2CO3 were directly performed on the silicon prism. The ink of Bi nanoparticles and isopropanol (preparation method was the same as above) was uniformed sprayed onto the silicon prism with a load of ~1 mg cm–2. After the addition of 5 mL 1 M KHCO3, the spectra were set to be automatically sampled at 1-s intervals. Preparation of Au Film and Working Electrodes for SEIRAS. First, the silicon prism was washed with aqua regia solution. Then, the prism was polished with a 0.05-μm Al2O3 slurry, followed by ultrasound in acetone and DIW repeatedly. After drying under Ar stream, the silicon prism was soaked in the NH4F solution. To chemically deposit gold film, the silicon prism was immersed in a mixture of gold-plated solution and a 2 wt% HF aqueous solution under 55 °C water bath for 7 min. After deposition, the gold film on the silicon prism was cleaned with DIW and dried naturally. The ink of Bi nanoparticles and isopropanol (preparation method as above) was uniformed sprayed onto the obtained gold film with a load of ~1 mg cm–2. Bi2O2CO3 nanosheets were obtained by immersing Bi nanoparticles on the gold film into 1 M KHCO3 for 60 s. Then, the film was rinsed by DIW and dried under Ar stream. In Situ SEIRAS Experiments. In situ SEIRAS tests were conducted in a spectroelectrochemical cell with a three-electrode configuration. The working electrodes were the above-obtained electrodes, the counter electrode was a graphite rod, and the reference electrode was a saturated Ag/AgCl electrode. The cell was integrated into a NICOLET iS50 FTIR spectrometer equipped with a liquid nitrogen-cooled MCT detector. After electrochemical reduction in 0.005 M H2SO4 with 0.05 M K2SO4 (pH 3), the spectra were collected at different potentials while CO2 is fed to the electrolyte. Each spectrum consists of 32 single beams with a resolution of 4 cm−1. Supplementary Material Appendix 01 (PDF) Click here for additional data file.