Efficient and Safe Membrane‐Free Flow Electrolyzer for Formate Synthesis and Direct Fuel Cell Integration

ABSTRACT

Membrane‐free electrocatalysis represents a promising alternative to conventional systems, yet the potential H₂/O₂ intermixing stands as the foremost barrier to practical implementation. Here, we report an efficient and safe membrane‐free flow electrolyzer that kinetically matches methanol oxidation and CO₂ reduction for symmetric formate production. To suppress H₂/O₂ generation, we developed low‐cost, separate catalysts that can operate compatibly in a single electrolyte. When stabilized by lattice‐matched FeOOH, Ni₅(II)O(OH)₈, a new material synthesized for the first time, addresses the common issue of Ni‐based catalysts being oxidized to NiOOH. This feature effectively suppresses competing O₂ evolution, enabling ∼100% methanol‐to‐formate conversion within an expanded potential window. In parallel, SO₄ ²⁻ incorporated Bi₂O₂CO₃ shields methanol from the electrolyte while promoting highly exclusive CO₂ reduction to formate at evaluated current densities. Thus, the system achieves >195% overall formate Faradaic efficiency over a record‐wide current density range (2.0 to 424.6 mA cm⁻²) with minor H₂/O₂ production rates (e.g., H₂: 1.6 mL h⁻¹, O₂: 0.5 mL h⁻¹, at 200 mA cm⁻²), demonstrating excellent production efficiency and safety under fluctuating renewable energy input. The produced formate‐rich solution can be further utilized in a high‐performance fuel cell. This work establishes a low‐cost and safe CO₂‐to‐power loop route for sustainable energy conversion.

Efficient formate production was achieved by pairing methanol oxidation and CO₂ reduction in a membrane‐free flow cell. With designing selective and compatible catalysts, high Faradaic efficiency (>195%) was achieved at wide current densities, showing the robustness driven by fluctuating renewable electricity. The produced electrolyte can be utilized for fuel cells, exemplifying a sustainable loop for chemical production and energy conversion.

1. Introduction

The growing penetration of renewable energy sources is accelerating the electrification of chemical and fuel production for industrial decarbonization [1, 2, 3], yet developing electrochemical technologies that are feasible for large‐scale deployment is still highly challenging [4, 5, 6]. At present, the dominant membrane‐based electrolyzers suffer from long‐standing issues, most notably high cost and irreversible degradation of ion exchange membranes (IEMs), which have become one of the primary barriers to improving overall efficiency and scalability [7, 8, 9, 10, 11].

Membrane‐free electrolysis offers a promising alternative by eliminating the costly and fragile IEMs [12, 13]. However, integrating anodic and cathodic reactions in a shared electrolyte remains a major challenge due to multiple interrelated issues, including mismatched reaction kinetics, pH incompatibility, energy‐intensive product separation, and, critically, parasitic H₂/O₂ intermixing problem that becomes even more severe under fluctuating renewable energy input [14, 15, 16, 17]. Given the broad explosive range of hydrogen in oxygen (ca. 4–75 vol%), suppressing the H₂/O₂ co‐generation is the primary bottleneck without membrane separation. To address this issue, strategies such as coupling organic molecule oxidation with hydrogen evolution reaction (HER) have been explored to unilaterally avoid O₂ evolution reaction (OER) [18, 19]. However, these often result in diverse products that complicate downstream processing [20, 21]. Specialized reactors like microfluidic or sequential flow configurations have been also proposed to improve gas management, which however inevitably increase the system resistance and fabrication complexity [22, 23]. Thus, achieving efficient and safe single‐compartment membrane‐free electrolysis remains an unresolved challenge.

Electrochemical synthesis of formic acid or formate provides a sustainable and less energy‐intensive approach compared to traditional methods given its broad utility as a basic chemical feedstock [24, 25]. They can also serve as a liquid‐phase hydrogen carrier that rapidly releases hydrogen at ambient conditions, addressing the key issues of low storage density and poor safety faced by gaseous hydrogen systems. Both the methanol (CH₃OH, US$0.2–0.3 kg⁻¹) oxidation reaction (MOR) and CO₂ reduction reaction (CO₂RR) can efficiently produce formate (e.g., HCOOK, US$0.9–1.2 kg⁻¹) in alkaline media, enabling a coupled process with a theoretical Faradaic efficiency (FE) of up to 200% [26, 27]. However, prior studies have predominantly relied on membrane‐based flow cells (Figure 1a), where the typical volcano‐type relationship between selectivity and current densities of both reactions unavoidably leads to compromised efficiency and severe H₂/O₂ crossover (Figure 1b) [28, 29, 30]. These challenges become far more complex in a membrane‐free system, as the shared environment in a single electrolyte exacerbates reaction interference, product separation difficulties, and gas mixing issues. It is worth mentioning that a membrane‐free CO₂ electrolyzer integrating CO₂ reduction with glycerol oxidation recently demonstrated significant economic advantages but without considering the safety issues of H₂/O₂ mixing [31]. Therefore, we hypothesize that a practically effective membrane‐free flow system (Figure 1c) must achieve highly selective operation (approaching ∼100% FE) for both reactions across overlapping current‐density windows (Figure 1d), and the window should be sufficiently wide to accommodate the large fluctuation in renewable power [32, 33]. To realize this, the design of selective and highly active catalysts tailored to each half‐reaction is imperative.

FIGURE 1.

(a) Schematic of the conventional membrane‐based flow electrolyzer that couples MOR and CO₂RR. (b) Illustration of compromised overall performance resulting from the common volcano‐type performance. (c) Schematic of the membrane‐free flow electrolyzer employed in this study. (d) Both MOR and CO₂RR are required to achieve maximum FE_formate (∼100%) across overlapping current densities for a practically effective membrane‐free electrolyzer.

Herein, we demonstrate paired and high‐yield formate synthesis in a membrane‐free flow electrolyzer. Both the anode and cathode catalysts are rationally designed from low‐cost Ni‐ and Bi‐based materials and are further engineered for optimal compatibility within the coupled system. The anode features a Ni₅(II)O(OH)₈/FeOOH heterostructure with stabilized Ni(II) sites, facilitating the CH₃OH‐to‐formate conversion with ∼100% FE across the current densities from 2.8 to 690.1 mA cm⁻². The cathode employs SO₄ ²⁻‐incorporated Bi₂O₂CO₃, also achieving high selectivity over a broad current densities range (1.8 to 677.4 mA cm⁻²) with minimal interference from CH₃OH‐containing electrolyte. Together, both catalysts enable a single‐cell system with overall FEs exceeding 195% in an unprecedented wide current density window (2.0 to 424.6 mA cm⁻²) with highly durable operation for up to 820 h, demonstrating the robustness powered by commercial solar cells. The produced electrolyte was directly utilized in formate fuel cells, exemplifying a sustainable loop for chemical production and energy conversion.

2. Results and Discussion

Ni‐based catalysts are known for their high CH₃OH‐to‐formate selectivity but are prone to oxidation into NiOOH, an active phase for OER. This poses a critical challenge to maintaining formate selectivity as well as the operational safety. In this work, we prepared the pristine Ni₅(II)O(OH)₈ grown on a nickel foam (denoted as Ni‐NF) via a facile hydrothermal method, representing the first reported artificial synthesis to our knowledge. Scanning electron microscopy (SEM, Figure S1) and transmission electron microscopy (TEM, Figure S2) reveal the nanofiber morphology with polycrystalline nature. X‐ray diffraction (XRD) pattern (Figure 2a) displays characteristic peaks consistent with natural Ni₅O(OH)₈ ore (NO.99‐000‐4007). X‐ray photoelectron spectroscopy (XPS, Figure S3) confirms Ni(II) with two peaks at 873.6 and 855.9 eV [34, 35]. However, the detailed atomic structure of Ni₅O(OH)₈ is still unclear due to the lack of database. After extensive structural modeling and simulations, we tentatively assign the structure to a monoclinic phase (Figure S4, Tables S1, and S2) by systematically varying the occupancies of Ni, H, and O atoms to best match the experimental XRD patterns. Nevertheless, the synthesis of suitable single crystals remains essential but highly challenging for further precise structural determination.

FIGURE 2.

(a) XRD patterns of Ni‐NF and NiFe‐NF. (b) SEM image of NiFe‐NF. (c, d) HAADF‐STEM image of NiFe‐NF and corresponding EDX mapping over a single nanosheet. (e) Atomic‐resolution HAADF‐STEM image of NiFe‐NF, showing the lattice‐matched interface. (f) High‐resolution Ni 2p spectra of Ni‐NF and NiFe‐NF. (g) The normalized Ni K‐edge XANES spectra of Ni‐NF and NiFe‐NF. (h) Fourier transform of the Ni K‐edge EXAFS for Ni‐NF and NiFe‐NF. (i) Wavelet transform of Ni K‐edge for NiFe‐NF.

The Ni₅O(OH)₈/FeOOH heterostructure on nickel foam (NiFe‐NF) was then prepared by additionally introducing FeCl₃ precursor. The formation of FeOOH is evidenced by the presence of additional XRD peaks at 16.9°, 26.8°, 35.2°, 43.2°, and 56.0° (Figure 2a). It is noticed that several XRD peaks of Ni₅O(OH)₈ and FeOOH are nearly overlapped, suggesting the potential lattice matching. SEM image (Figure 2b) further reveals the nanosheet morphology of NiFe‐NF with lateral sizes of 300∼500 nm. High‐angle annular dark‐field scanning transmission electron microscopy (HAADF‐STEM, Figure 2c) and corresponding energy‐dispersive x‐ray spectroscopy (EDX, Figure 2d) disclose that NiFe‐NF possesses a FeOOH@Ni₅O(OH)₈ core@shell structure with a shell thickness of approximately 2 nm. Atomic‐resolution HAADF‐STEM image (Figures 2e and S5) shows the lattice distances of 0.26 and 0.23 nm, assigned to Ni₅O(OH)₈ and FeOOH, respectively. The lattice‐matched interface observed between the shell and core suggests the possibility of an epitaxial growth. XPS (Figure S6 and Table S3) and inductively coupled plasma optical emission spectrometry (ICP‐OES, Table S4) determine a Ni:Fe ratio of ∼2.5. High‐resolution Ni 2p spectra (Figure 2f) show a positive shift of Ni(II) by 0.3 eV in NiFe‐NF relative to Ni‐NF, indicating the electron transfer from Ni₅O(OH)₈ to FeOOH.

The normalized Ni K‐edge x‐ray absorption near‐edge structure (XANES) spectroscopy (Figure 2g) shows the absorption edge position of Ni‐NF is close to that of Ni(OH)₂, suggesting an average valence near +2. The positive energy shift observed for NiFe‐NF indicates an increased oxidation state, further confirming the electron transfer from Ni₅O(OH)₈ to FeOOH. Fourier transform of extended x‐ray absorption fine structure (EXAFS) discerns the contracted Ni─O (1.58 Å) and Ni─Ni (2.54 Å) distances in NiFe‐NF relative to those (1.61 Å, 2.73 Å) in Ni‐NF (Figures 2h and S7). Despite the possible tensile strain in a core@shell heterostructure, the observed contraction could be attributed to the reduced ionic radius of Ni resulting from the electron transfer from Ni₅O(OH)₈ to FeOOH. Fitting results (Tables S5 and S6) further demonstrate the close coordination number of Ni─O in NiFe‐NF (7.2 ± 0.89) and Ni‐NF (6.8 ± 0.85). However, the Ni─Ni coordination number (10.8 ± 2.82) in NiFe‐NF is increased compared to the pristine Ni‐NF (6.9 ± 1.46), probably due to the detection of comparable Ni─Fe coordination at the interface. Notably, despite the increased fitted coordination, the lower FT‐EXAFS peak intensity of NiFe‐NF can be attributed to its higher structural disorder (i.e., a larger Debye‐Waller factor), which dampens the signal amplitude. The Morlet wavelet transform contour plots (Figures 2i and S8) visualize the shortened Ni─Ni distances in NiFe‐NF, further evidencing the electronic modulation by FeOOH incorporation.

The electrochemical MOR performance of NiFe‐NF and Ni‐NF was evaluated in 1.0 M KOH containing 1.0 M CH₃OH in a three‐electrode system (Figure S9). Both NiFe‐NF and Ni‐NF exhibit increased current densities in linear sweep voltammetry (LSV) curves (Figure 3a), indicative of their intrinsic MOR activities. In addition, NiFe‐NF delivers a lower overpotential and higher current densities, suggesting the superior MOR performance. A lower Tafel slope (31.2 mV dec⁻¹, Figure S10) of NiFe‐NF signifies the improved reaction kinetics relative to Ni‐NF (49.5 mV dec⁻¹). When normalized to the electrochemically active surface area (ECSA, Figure S11), NiFe‐NF still shows higher current densities, suggesting the intrinsic enhancement due to the optimized electronic interaction rather than surface area effects. An anodic peak at about 1.48 V (vs. reversible hydrogen electrode, RHE) is particularly noticed for Ni‐NF, suggesting the oxidation of Ni₅O(OH)₈ during MOR. The oxidation can be responsible for the intensified OER at high current densities, similar to previously reported Ni‐based catalysts. In contrast, no such peak is detectable for NiFe‐NF even at expanded current scales, highlighting the excellent resistance of Ni(II) sites against oxidation. The absence of cathodic peak in reverse scans further validates the suppressed surface redox activity of NiFe‐NF, contrasting with the obvious cathodic feature observed for Ni‐NF (Figure S12). Note that NiFe‐NF still exhibits a distinct oxidation peak in the electrolyte without CH₃OH (Figure S13), indicating that the presence of CH₃OH plays a key role in suppressing the composition oxidation. Electrochemical impedance spectroscopy (EIS, Figure S14) reveals a lower interfacial charge‐transfer resistance for NiFe‐NF relative to Ni‐NF, further demonstrating the improved electron transfer kinetics.

FIGURE 3.

(a) LSV curves of NiFe‐NF and Ni‐NF at a scan rate of 50 mV s⁻¹ in 1.0 M KOH containing 1.0 M CH₃OH electrolyte. (b) FE_formate and (c) j _formate of NiFe‐NF and Ni‐NF at different potentials. (d) Stability evaluation of NiFe‐NF at 200 mA cm⁻². In situ Raman spectra of (e) Ni‐NF and (f) NiFe‐NF at various potentials. (g) OCP variation of NiFe‐NF and Ni‐NF before and after CH₃OH removal. (h) In situ IRRAS spectra of NiFe‐NF during MOR at different potentials (IR spectrum at 1.0 V_RHE used as the background).

Chronoamperometric measurements were conducted at different potentials to evaluate the MOR product selectivity. At 1.5 V_RHE, Ni‐NF (Figure 3b) achieves the highest FE_formate of 92.2%, which however decreases as the potential moves positively. Meanwhile, more bubbles can be observed, indicating the signified OER process. As a comparison, NiFe‐NF exhibits nearly 100% FE_formate across a wide potential range of 1.35∼1.8 V_RHE, corresponding to the current densities of 2.8∼690.1 mA cm⁻². Only formate and CH₃OH are detected in the ¹³C NMR spectra of the liquid products (Figure S15), which can exclude the CO₂ production as a deep oxidation product of CH₃OH. Accordingly, NiFe‐NF presents enhanced partial current densities of formate (j _formate) than Ni‐NF at varied potentials (Figure 3c). In addition, it can operate stably at 200 mA cm⁻² with ∼100% FE_formate for 200 h (Figure 3d), demonstrating the excellent durability at industrial‐scale current densities.

The structural evolution of NiFe‐NF and Ni‐NF was further investigated to elucidate the enhancement mechanism with FeOOH incorporation. Initially, Ni‐NF appears light green, which turns black after 10 min at 1.6 V_RHE, a clear sign of oxidation (Figure S16). In contrast, NiFe‐NF maintains its brown color after electrolysis at 1.8 V_RHE for 1 h, visually suggesting the compositional stability at higher potentials. However, when electrolyzed in 1.0 M KOH without CH₃OH at 1.6 V_RHE for 10 min, NiFe‐NF also turns black, indicating that the stability observed under MOR conditions is attributable to the presence of CH₃OH. As shown in the in‐situ Raman spectroscopy, both Ni‐NF (Figures 3e and S17a) and NiFe‐NF (Figures 3f and S17b) show three peaks at 212, 457, and 635 cm⁻¹ at the open circuit potential (OCP), attributable to the characteristic modes of Ni₅O(OH)₈. NiFe‐NF exhibits additional peaks at 683 cm^{− 1} assigned to FeOOH [36, 37]. As the potential increases, the characteristic peaks associated with Ni₅O(OH)₈ in Ni‐NF disappear, while the peaks at 473 and 555 cm⁻¹ assigned to NiOOH are intensified, indicating the surface oxidation [38, 39]. In contrast, NiFe‐NF shows negligible peak change, validating its compositional stability during MOR.

In the Raman spectroscopy, the peaks assigned to adsorbed CH₃OH and formate are also detected at 1018/1470 and 1339 cm⁻¹, respectively [40, 41]. The higher intensities for NiFe‐NF (Figure S18) demonstrate the enhanced CH₃OH adsorption capability, beneficial for promoting the CH₃OH‐to‐formate conversion. Furthermore, both NiFe‐NF and Ni‐NF are initially immersed in a 1.0 M KOH electrolyte containing 1.0 M CH₃OH, during which OH⁻ and CH₃OH are competitively adsorbed on the catalyst surface (Figure 3g). Upon replacing the electrolyte with 1.0 M KOH, NiFe‐NF exhibits a higher increase in OCP (65.1 mV) relative to Ni‐NF (29.6 mV), confirming the enhanced CH₃OH adsorption and thus more CH₃OH molecules released from NiFe‐NF. Thus, we can reasonably deduce that the enhanced CH₃OH adsorption acts as a physical shield, preventing OH⁻ from attacking the electron‐deficient Ni sites. This effectively directs the reaction pathway toward MOR rather than NiOOH formation, which is beneficial for suppressing OER at high potentials.

To highlight the distinctive function of the newly developed Ni₅O(OH)₈, we synthesized a control sample—Ni(OH)₂, a prototypical nickel‐based MOR catalyst, also incorporated with FeOOH on nickel foam (denoted as Ni(OH)₂‐FeOOH‐NF, Figure S19). EDX mapping confirms the formation of Ni(OH)₂/FeOOH heterostructure with uniform distribution of Ni, Fe, and O elements over a single nanosheet (Figure S20). However, LSV measurements reveal a broad anodic peak (Figure S21), indicative of significant oxidation during MOR. As an MOR catalyst, Ni(OH)₂‐FeOOH‐NF shows an initial FE_formate of 92.1% at 1.4 V_RHE, rapidly dropping to 64.8% at 1.8 V_RHE (Figure S22), attributable to enhanced OER on the oxidized surface. These results clearly demonstrate that Ni₅O(OH)₈ possesses a uniquely robust resistance to oxidation and maintains high formate selectivity, setting it apart from conventional Ni‐based catalysts.

In situ infrared reflection absorption spectroscopy (IRRAS) was conducted to reveal the reaction mechanism of MOR on NiFe‐NF (Figure 3h) and Ni‐NF (Figure S23). Both catalysts exhibit the peaks at 1571, 1382, and 1350 cm⁻¹, which can be attributed to the ʋ_as (O─C─O), δ (C─H), and ʋ_s (O─C─O) vibrations of formate [42, 43], respectively. The intensity of those peaks for NiFe‐NF is higher relative to Ni‐NF, demonstrating the enhanced CH₃OH‐to‐formate activity. Another peak at 1408 cm⁻¹ for Ni‐NF can be assigned to the ʋ_as (C─O) of carbonate species formed after CO₂ dissolved in KOH [44, 45]. In contrast, it is not detected on NiFe‐NF, suggesting the highly selective conversion to formate rather than CO₂. The cyclic voltammetry (CV) curve in 1.0 M KOH containing 1.0 M formate (Figure S24) further reveals negligible formate oxidation on NiFe‐NF, confirming the selective oxidation of CH₃OH to formate without further oxidation.

Bi‐based catalysts have been demonstrated with high CO₂‐to‐formate selectivity, which are further required to shield CH₃OH from the electrolyte in this membrane‐free system. In this work, a facile, room‐temperature wet chemical method was developed to synthesize Bi₂O₂CO₃ by using commercially available BiCl₃, Bi(NO₃)₃·5H₂O, and Bi₂(SO₄)₃, denoted as BOC‐Cl⁻, BOC‐NO₃ ⁻, and BOC‐SO₄ ²⁻, respectively. This method can be scaled up to the kilogram level, showing potential for large‐scale synthesis (Figure S25). All the XRD patterns (Figure 4a) are indexed to Bi₂O₂CO₃. SEM images reveal the nanosheet morphology with lateral dimensions increased from BOC‐Cl⁻ (ca. 50∼200 nm, Figure S26a) and BOC‐NO₃ ⁻ (ca. 50∼200 nm, Figure S26b) to BOC‐SO₄ ²⁻ (ca. 200∼500 nm, Figure 4b). EDX mapping shows the uniform distribution of Bi and O, with Cl, N, and S elements for BOC‐Cl⁻ (Figure S26c), BOC‐NO₃ ⁻ (Figure S26d), and BOC‐SO₄ ²⁻ (Figure 4c), respectively. XPS analysis of thoroughly washed samples reveals the presence of Cl⁻ and NO₃ ⁻ species in BOC‐Cl⁻ (Figure S27a) and BOC‐NO₃ ⁻ (Figure S27b), respectively. Considering the overlap of S 2p for SO₄ ²⁻ and Bi 4f_5/2 for Bi³⁺, Fourier transform infrared (FTIR) spectroscopy (Figure S28) was further measured, confirming the existence of SO₄ ²⁻ in BOC‐SO₄ ²⁻. Moreover, the gradually positive shift of Bi 4f (Figure 4d) in BOC‐Cl⁻, BOC‐NO₃ ⁻, and BOC‐SO₄ ²⁻ suggests the effective electronic modulation by those anions despite their minor amounts (Table S7).

FIGURE 4.

(a) XRD patterns of BOC‐Cl⁻, BOC‐NO₃ ⁻ and BOC‐SO₄ ²⁻. (b) SEM image of BOC‐SO₄ ²⁻. (c) EDX mapping of BOC‐SO₄ ²⁻. (d) High‐resolution Bi 4f spectra of BOC‐Cl⁻, BOC‐NO₃ ⁻, and BOC‐SO₄ ²⁻. (e) HRTEM image of BOC‐SO₄ ²⁻ after CO₂RR. (f) LSV curves of BOC‐SO₄ ²⁻ in 1.0 M KOH and 1.0 M KOH containing 1.0 M CH₃OH. (g) CH₃OH adsorption energies on Bi (012) modified by Cl⁻, NO₃ ⁻, and SO₄ ²⁻. (h) FE_formate of BOC‐SO₄ ²⁻ in 1.0 M KOH containing 1.0 M CH₃OH electrolyte. (i) In situ ATR‐IR spectra of BOC‐SO₄ ²⁻ in the CH₃OH‐containing electrolyte.

The electrochemical CO₂RR performance of BOC‐Cl⁻, BOC‐NO₃ ⁻, and BOC‐SO₄ ²⁻ was evaluated in a three‐electrode flow cell (Figure S29). LSV curves (Figure S30) show increased current densities when shifting Ar to CO₂ feeding, indicating the intrinsic CO₂RR behaviors. BOC‐SO₄ ²⁻ exhibits higher current than BOC‐Cl⁻ and BOC‐NO₃ ⁻, suggesting higher CO₂RR activity. The main products of three catalysts include H₂, CO, and formate, with a total FE of close to 100%. Specifically, BOC‐Cl⁻ and BOC‐NO₃ ⁻ (Figure S31) achieve FE_formate exceeding 90.8% and 96.7% at −0.2∼‐1.2 V_RHE, respectively. However, increasing current decay is observed as the potential shifts negatively, indicating their poor stability. In contrast, BOC‐SO₄ ²⁻ exhibits higher FE_formate of >97.0% at −0.2∼‐1.4 V_RHE, corresponding to the current densities of 1.8∼677.4 mA cm⁻². Besides, the current density is more stable. At 200 mA cm⁻², the cell voltage is maintained over 200 h, with a FE_formate of >96.2% (Figure S32), demonstrating the superior stability with SO₄ ²⁻ incorporation.

The catalysts after testing were further investigated to probe the origin of distinct CO₂RR stability. XRD patterns (Figure S33) and high‐resolution TEM (HRTEM, Figures 4e and S34) disclose the reduction of catalysts to metallic Bi during CO₂RR, which partially accounts for the initial current decay. In addition, both spent BOC‐Cl⁻ and BOC‐NO₃ ⁻ catalysts show structural collapse, evidenced by the formation of discrete particles (Figure S35a,b). Differently, BOC‐SO₄ ²⁻ maintains structural integrity after reconstruction (Figure S35c,d), which can be attributed to the strong binding to Bi sites via robust bidentate or bridging modes and thus prevents the agglomeration of Bi atoms [46]. This porous structure preserves a highly active surface area, guarantees CO₂ diffusion channels, and suppresses the competing HER, which supports the more stable current output.

To couple with MOR in the membrane‐free system, the influence of CH₃OH on CO₂RR performance was further investigated using 1.0 M KOH containing 1.0 M CH₃OH as the electrolyte. LSV curves show the decreased current densities for BOC‐Cl⁻ and BOC‐NO₃ ⁻ (Figure S36), ascribed to the reduced active sites with CH₃OH adsorption. In contrast, the current density of BOC‐SO₄ ²⁻ is nearly unchanged (Figure 4f). Density functional theory (DFT) calculations (Figure 4g) reveal that BOC‐SO₄ ²⁻ has a higher CH₃OH adsorption energy (−0.23 eV) than BOC‐Cl⁻ (−0.38 eV) and BOC‐NO₃ ⁻ (−0.34 eV), indicating better CH₃OH shielding ability. In addition, the FE_formate of BOC‐SO₄ ²⁻ (Figure 4h) remains high (>96.5%) at different potentials in the CH₃OH‐containing electrolyte. In contrast, BOC‐Cl⁻ and BOC‐NO₃ ⁻ exhibit decreased FE_formate, especially at negative potentials (Figure S37). Furthermore, in‐situ ATR‐IR spectra of BOC‐SO₄ ²⁻ (Figures 4i and S38) in the CH₃OH‐containing electrolyte reveal the peak at 1367 cm⁻¹ assigned to *OCHO and another negative peak at 1445 cm⁻¹ corresponding to *CO₃ ²⁻, indicating the same reaction mechanism in the methanol‐free electrolyte (Figure S39). The negligible CH₃OH effect on the formate production of BOC‐SO₄ ²⁻ suggests its robustness in the membrane‐free system.

We assembled a membrane‐free flow electrolyzer that couples MOR and CO₂RR (Figure S40) to maximize the formate production efficiency, by employing NiFe‐NF and BOC‐SO₄ ²⁻ as the anodic and cathodic catalysts, respectively. An electrolyte comprising 1.0 M KOH containing 1.0 M CH₃OH was utilized to support both reactions. This configuration achieves a total FE_formate of 195%∼200% at the cell voltages of 1.7∼4.1 V (Figure 5a), corresponding to the current densities of 2.0∼424.6 mA cm⁻². It represents a record‐wide range of current density in the membrane‐free flow cell compared to previously reported studies that primarily utilized membrane‐based counterparts (Figure 5b). No gaseous products are observed in the outlet electrolyte (Movie S1), indicating the safe operation with negligible H₂ and O₂ evolution. At a high cell voltage of 5.0 V, the total FE_formate remains as 182.5%, with a current density of 601.6 mA cm⁻². The current density range for FE_formate exceeding 180% (2.0∼601.6 mA cm⁻²) still outperforms those reported non‐noble metal‐based catalysts (Figure S41 and Table S8). The integrated configuration powered by a solar cell can produce approximately 12 mmol of formate in 1 h (Figure S42), validating the potential of accommodating fluctuating renewable energy sources.

FIGURE 5.

(a) FE_formate of the membrane‐free flow electrolyzer. (b) Comparison of the current density range for FE_formate >195% with the currently reported electrocatalysts in membrane‐based electrolyzers. (c) Formate yielding rates in membrane‐free and AEM‐based flow electrolyzers. (d) Stability evaluation at 200 mA cm⁻². (e) Polarization and power density curves of DFFC. (f) Schematic of the CO₂‐to‐power route. (g) TEA for producing 1 ton formate via this design.

The membrane‐free configuration also shows higher formate production rates (Figure 5c) compared to the conventional anion exchange membrane (AEM)‐based flow electrolyzer. This enhancement is primarily attributed to reduced ohmic resistance and thus improved current densities (Figure S43). For example, the formate production rate reaches 6.46 mmol h⁻¹ cm⁻² at 3.2 V in the membrane‐free flow cell, representing 1.54 times that in the AEM‐based one. We further reduced the inter‐electrode distance from 8.0 to 3.0 mm, to minimize the ohmic resistance of system. At 200 mA cm⁻², the cell voltage was decreased to 2.52 V (Figure S44) compared to the membrane‐free (2.90 V) and AEM‐based cells (3.13 V) with an 8.0 mm gap, which is significantly lower than the most reported membrane‐based systems (Table S9). Furthermore, the design can stably run for 820 h at 200 mA cm⁻² (Figure 5d) with a preserved cell voltage of 2.95 V and FE_formate exceeding 190%. Meanwhile, only trace gases of H₂ and O₂ are generated (approximately 1.6 and 0.5 mL h⁻¹, respectively), indicating negligible risk of H₂/O₂ mixing. Note that the electrolyte (2 L, 1.0 M KOH containing 1.0 M CH₃OH) was periodically renewed (∼160 h), corresponding to the CH₃OH conversion efficiency of 13%∼17% (Table S10).

In industrial formic acid production, distillation and purification steps are highly energy‐intensive, accounting for 30%∼50% of total energy input due to the close boiling points of formic acid (ca. 101°C) and water (ca. 100°C). Thus, we further explored the utilization of easily transportable formate salts in direct formate fuel cells (DFFCs) as a potential alternative to hydrogen‐oxygen fuel cells. PdCu alloy nanoparticles (Figure S45a) supported on nickel foam (denoted as PdCu‐NF) were prepared via an electrodeposition method. XRD pattern (Figure S45b) reveals the PdCu alloy with diffraction peaks positioned between Pd and Cu. HRTEM image (Figure S45c) shows the lattice d‐spacing of 0.218 nm, intermediate between Pd (111) and Cu (111). EDX mapping (Figure S45d–f) reveals uniform distribution of Pd and Cu elements, verifying the alloy phase. A concentrated alkaline formate solution (∼3.0 M) was obtained by simply distilling the electrolyte produced after electrolysis at 200 mA cm⁻² for 100 h, to remove CH₃OH and partial water (Figure S46). Using PdCu‐NF as the anodic catalyst and Pt/C as the cathodic catalyst, the DFFC achieves a high peak power density of 73.5 mW cm⁻² (Figure 5e), comparable to that of a freshly prepared 3.0 M formate electrolyte (Figure S47). It can continuously power a fan (Movie S2), verifying the feasibility of the produced electrolyte for power generation (Figure 5f).

The techno‐economic analysis (TEA) was preliminarily performed based on the electrochemical performance of the membrane‐free system with 3.0 mm inter‐electrode distance to assess the economic feasibility of the system [47]. The costs considered include capital, operational, and material expenses (Figure 5g and Table S11), while the revenue is derived from formate production. Under a specific operating condition (a current density of 200 mA cm⁻², a cell voltage of 2.52 V, and a FE_formate of 97.5% at the anode and cathode), the total costs and total revenues for producing the formate are US$817.69 ton⁻¹ and US$1000 ton⁻¹, respectively, yielding a net profit of US$182.31 ton⁻¹ with this design.

3. Conclusion

In summary, we demonstrate the efficient and reliable formate synthesis in a membrane‐free flow electrolyzer. The Ni₅(II)O(OH)₈/FeOOH developed as the anodic catalyst shows excellent stability against potential oxidation and thus maintains nearly 100% selectivity for CH₃OH‐to‐formate conversion at varied potentials. Concurrently, the cathodic SO₄ ²⁻‐incorporated Bi₂O₂CO₃ achieves highly selective CO₂‐to‐formate conversion with negligible CH₃OH interference. Benefiting from the well‐matched pH, reaction kinetics, and product type, the integrated membrane‐free system attains a total FE_formate exceeding 195% across a record‐wide current density range (2.0∼424.6 mA cm⁻²), enabling efficient and feasible operation under fluctuating working conditions. The produced electrolyte can be utilized as fuels in the downstream DFFCs. This system holds great potential for scaling up electrocatalysis by eliminating IEMs and maximizing production efficiency.

Author Contributions

Yicheng Li: methodology, conceptualization, investigation, validation, data curation, writing ‐ original draft, writing – review and editing. Ernest Pahuyo Delmo: methodology, formal analysis, writing – original draft, data curation. Xingqiu Li: formal analysis, investigation, methodology. Ya Liu: methodology, formal analysis. Sheng Dai: methodology, formal analysis, data curation. Xuan Tang: methodology. Zhuoying Zhu: methodology, investigation, software. Daobin Liu: investigation, methodology, software. Pengfei Tian: formal analysis. Ming Zhao: investigation, methodology. Yu Zhang: conceptualization, methodology, formal analysis, supervision, resources, project administration, writing – review and editing, funding acquisition, investigation, validation. Fu‐Zhen Xuan: conceptualization, project administration, investigation, formal analysis. Minhua Shao: formal analysis, writing – review and editing, methodology, conceptualization, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: anie73173‐sup‐0001‐SuppMat.docx.

Supporting File 2: anie73173‐sup‐0002‐Movie1.mp4.

Supporting File 3: anie73173‐sup‐0003‐Movie2.mp4.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.