Bipolar Membrane with Porous Anion Exchange Layer for Efficient and Long‐Term Stable Electrochemical Reduction of CO2 to CO

Bipolar membranes in forward bias have the potential to address several challenges of alkaline zero‐gap CO2 electrolyzers. However, the inevitable gas evolution of CO2 at the membrane junction typically leads to delamination and failure of the membrane after a few hours, limiting its applicability in electrolyzers so far. In this work, a bipolar membrane with a perforated anion exchange layer is presented that allows the CO2 gas to flow back to the cathode and thus preventing accumulation of gas at the junction. This configuration for the first time enables stable operation of a forward bias bipolar membrane showing a degradation rate of 0.5 mV h−1 in a 200 h constant current hold at 100 mA cm−2 and 3.1 V. Highest reported energy efficiency of EECO = 39% and Faradaic efficiency of FECO = 96% at 200 mA cm−2 among forward bipolar membranes prove that the perforated membrane does not compromise efficiency. A maximum CO2 single‐pass conversion of 52% at 0.75 mLCO2 min−1 cm−2 and a low specific energy consumption of 665 kJ mol−1 CO2 shows that this concept is superior not only to other bipolar membranes in the literature, but potentially also to conventional anion exchange membrane‐based electrolyzers.


Introduction
The electrochemical reduction of CO 2 has gained much research interest in recent years, as this technology promises to enable a fossil-free and scalable production of various feedstock chemicals for the chemical industry, using electricity from renewable DOI: 10.1002/aenm.202301614energy sources, water, and CO 2 .[9] For low-temperature CO 2 electrolyzers producing CO (<100 °C), zero-gap cells with an anion exchange membrane (AEM) and a gas-fed cathode (see Figure 1a) can be considered as the current state of the art.This cell architecture provides an alkaline environment to the cathode, enabling high selectivity for the CO 2 reduction reaction over the hydrogen evolution reaction , the zero-gap architecture reduces ohmic losses, while the gas-fed cathode enables high reduction rates. [9]On the other hand, this setup suffers from multiple inherent problems.The alkaline environment promotes the carbonation reaction, which consumes CO 2 and produces (bi)carbonate ions. [10]The conductivity of (bi)carbonate ions is significantly lower than that of hydroxide ions, which increases the overall cell voltage. [5]Furthermore, with the (bi)carbonates CO 2 is transported to the anode side, limiting the single-pass conversion. [11]dditionally, when cations, e.g.K + or Cs + , from the anolyte migrate to the cathode side, they may form low soluble (bi)carbonate salts, that can block the CO 2 from reaching the catalysts active sites. [9,12,13]This typically leads to rapid performance decrease in cells that use electrolytes with salt concentration above 0.1 m. [14] If alkaline electrolytes are used on the anode side, the hydroxide ions are consumed by the oxygen evolution reaction.When (bi)carbonate ions from the cathode side arrive at the anode, the pH of the electrolyte is neutralized over time. [13,15]To maintain the high pH needed for the employment of non-noble metal catalysts on the anode side (e.g., Ni or Co oxides) the hydroxide ions have to be recovered in an additional process, which increases the overall energy demand. [16]While the pH decrease can be mitigated by using neutral electrolytes from the beginning, the formation of the (bi)carbonate ions and in consequence the CO 2 crossover is inherent to AEM-based setups. [15]In neutral electrolytes the (bi)carbonate ions can be oxidized or react with protons at the anode releasing CO 2 to the anode gas stream (see Figure 1a). [9]The recovery of the CO 2 from the anode gas stream increases the complexity, cost as well as the energy demand of the electrolyzer setup. [1,16]sing cation exchange membranes (CEM, Figure 1b) can solve these issues by preventing the transport of anions to the anode side via Donnan exclusion.Furthermore, the (bi)carbonate ions react with protons in the acidic environment releasing gaseous CO 2 . [17]However, due to the high proton concentration in acidic environments, the hydrogen evolution reaction is strongly favored over the CO 2 reduction reaction at the cathode, leading to very low Faradaic efficiencies for CO (FE CO ). [18]Various approaches are currently being investigated to shift the selectivity toward the CO 2 reduction reaction in acidic media, like tuning the local reaction environment [17][18][19][20] or by introducing a buffer layer to shield the cathode catalyst surface from a high proton concentration. [21][24] Additionally, the water dissociation at the membrane interface drastically increases the cell potential, lowering the energy efficiency. [7]polar membranes in forward bias are orientated the opposite way, with the anion exchange membrane facing the cathode (see Figure 1c), promising to combine the beneficial properties of both: the AEM ensures an alkaline environment to the cathode for a highly selective CO 2 reduction reaction, while the cation exchange layer facing the anode blocks the crossover of the (bi)carbonate anions.For this reason, several groups study bipolar membranes in forward bias. [16,23,25,26]However, despite successfully reducing the crossover of CO 2, [26] the energy efficiency and CO 2 reduction rates are rather low compared to AEM-based setups.The lower energy efficiency can be attributed to higher cell resistances due to typically thicker membranes and the high pH gradient between both half cells. [15]Furthermore, the bipolar membranes show a substantially lower durability (several hours versus hundreds of hours), which is mainly attributed to delamination and blistering of the bipolar membrane. [26,27]As can be seen in Figure 1c, CO 2 and water are generated at the AEM|CEM interface, when protons from the anode side react with anions from the cathode side according to the following reactions: [15] H + + OH − → H 2 O ( 1 ) These reactions lead to rapid CO 2 accumulation at the interface, failure of the membrane and with this to a performance decrease, limiting the lifetime of the electrolyzer to below 10 h. [23,25,27]o mitigate the blistering and subsequent delamination of the bipolar membrane, we propose using a porous anion exchange membrane, which is meant to allow the recirculation of the evolved CO 2 to the cathode catalyst layer as depicted in Figure 1d.In this work, we present a perforation-based approach to fabricate a porous AEM for bipolar membranes, enabling more durable cell operation in forward bias.

Results and Discussion
While many approaches exist to create a porous membrane like phase inversion [28] or electrospinning, [29] we found the simple approach to perforate a monolithic membrane to be the most effective.Thus, to fabricate the bipolar membrane with a gas permeable anion-conductive layer, we used a needle roller (see Figure S1, Supporting Information) to perforate a 50 μm thick anion exchange membrane (Ionomr, Aemion+ AP3-HNN9-50-X), as described in detail in the experimental section.The diameter of the individual needles is ≈300 μm and they are arranged in a distance of 1.6 mm horizontally and 1 mm vertically, yielding ≈250 holes in one perforation step on a 4 cm 2 cell area (see Figure S2a, Supporting Information).To achieve a higher hole density, further membranes were fabricated with 3 perforation runs (1x horizontally, 1x vertically, 1x diagonally), yielding an uneven distribution of ≈750 holes (see Figure S2b, Supporting Information).The uneven distribution of the holes can be seen in the top view electron micrograph of the 3x-perforated membrane in Figure 2a.In the top-right corner are three holes in close proximity less than 200 μm apart, while the holes on the left are more evenly dispersed with distances of more than 500 μm.
To quantify the gas permeability of the perforated membranes, differential pressure measurements were conducted in the electrolysis cell fixture forcing humidified CO 2 at various flow rates through the membrane from the cathode to the anode compartment (Figure 2b).The non-perforated membrane was gas-tight up to 1 bar, and could not be evaluated with this method.The measurement setup accounted for a pressure loss of ≈10 mbar at 50 mL min −1 CO 2 .Adding a 1x-perforated membrane significantly increases the differential pressure reaching 740 mbar pressure difference at the same flow rate.Perforating the membrane three times increases the gas permeability, only showing a pressure increase of 32 mbar at a CO 2 flow rate of 50 mL min −1 .
To get a better understanding of the required gas permeability of the anion exchange layer (AEL), we calculated the highest possible rate of CO 2 released at the membrane interface during cell operation, assuming 100% ion conduction by HCO 3 − (see Figure S3, Supporting Information).A current density of 400 mA cm −2 would generate a CO 2 flow of ≈6 mL cm −2 min −1 at the AEL|CEL interface.Considering the results of the gas permeability measurements and the effective cell area of 4 cm 2 , the evolution of 24 mL min −1 CO 2 would translate to a pressure increase of ≈400 mbar for the 1x-perforated membrane and to ≈30 mbar for the 3x-perforated membrane.Although, the conditions in the electrolysis cell might be different (compression with CEM, liquid anolyte and higher temperature), this can be seen as a first indicator for the forces that are at work at the membrane interface.
To obtain bipolar membranes the (perforated) anion exchange membranes were placed on top of a Nafion 212 membrane that was priorly coated with an IrO 2 catalyst layer on the outer side.The bipolar membranes were then assembled between a cathode gas diffusion electrode (0.8 mg cm −2 Ag loading, with 7.5 wt.% PiperION A5, Versogen, as binder) and a titanium felt as porous transport layer (PTL) on the anode.
The electrochemical performance was evaluated in a custommade cell fixture.All measurements were conducted at a cell temperature of 40 °C with a CO 2 flow rate of 50 mL min −1 that was humidified at 30 °C, if not stated otherwise.The anode was  [25] and Hansen et al. [33] Measurements were conducted at 40 °C cell temperature.The CO 2 feed (50 mL min −1 ) was humidified at 30 °C.The anode was supplied with DI-water at a flow rate of 20 mL min −1 .Each current step was held for 10 min.
supplied with DI-water at a flow rate of 20 mL min −1 .As it was shown by multiple groups, that K + and Cs + can significantly boost CO 2 reduction rates on silver-based catalyst, [30][31][32] the cathode compartment was infused with 0.1 m CsOH in a mixture of 1:3 vol isopropanol/H 2 O before the electrochemical measurements, similar to the procedure proposed by Endrődi et al. [32] (see Experimental Section).
Figure 3 shows the electrochemical performance of four different cells: One with a commercial bipolar membrane (fumasep FBM-PK, from Fumatech), one with a bipolar membrane fabricated with a non-perforated AEM and two cells with bipolar membranes using perforated AEMs of different porosity.The non-perforated BPM shows the lowest performance: it could only be operated up to 50 mA cm −2 , while at 100 mA cm −2 the voltage was rapidly increasing.The FE CO reaches a maximum at 25 mA cm −2 with 82%.The commercial reference (fumasep FBM-PK) performed better than the non-perforated BPM, reaching 300 mA cm −2 at a cell voltage of 4.1 V with a maximum in the FE CO of 75% at 150 mA cm −2 .
Both cells with perforated BPMs show significantly improved cell performance compared to the non-perforated BPM and the commercial BPM.The perforation increases the maximum current density eightfold compared to the non-perforated membrane (400 mA cm −2 vs 50 mA cm −2 ).The Faradaic efficiencies FE CO of both cells surpass that of the commercial and the non-perforated BPM cells at all current steps above 25 mA cm −2 , yielding a maximum FE CO of 99% at 150 mA cm −2 for the 1x-perforated BPM and 96% at 200 mA cm −2 for the 3x-perforated BPM.At 12.5 and 25 mA cm −2 , however, the FE CO is lower than that of the commercial and non-perforated BPM.This is potentially due to the porous membrane not being able to fully shield the cathode from protons from the anode side directly after the CsOH infusion, as the holes in the membrane might be flooded by the activation procedure.The 1x-perforated BPM shows the highest energy efficiency for CO EE CO with 43% at 100 mA cm −2 (see Figure S4, Supporting Information).All fabricated membranes show similar high frequency resistances (HFRs) in the range of 0.85 Ω cm 2 -1.26 Ω cm 2 , which is substantially lower than the HFR of the commercial BPM (>2.2 Ω cm 2 , see Figure 3c).Hence, the membrane resistance is the major cause for the lower cell voltage of our perforated BPMs compared to the commercial BPM (≈1 Ω cm 2 lower resistance leads to, e.g., ≈200 mV lower cell potential at 200 mA cm −2 , see Figure 3a).
Figure 3d shows the full cell potential in relation to the CO partial current density i CO of all BPMs measured in this work in comparison to other forward bias BPM measurements found in literature.Up to 3 V, the cell measurements with the fumasep FBM-PK membrane presented in this work show a similar curve shape to the measurements by Pribyl-Kranewitter et al. [25] (curves with round symbols).However, our cell achieved a higher maximum i CO of 145 mA cm −2 because Pribyl-Kranewitter et al. stopped the measurements at 100 mA cm −2 .The measurements of Hansen et al. [33] mark the highest reported i CO for a bipolar membrane zero-gap electrolyzer, producing CO with 260 mA cm −2 at a full cell potential of 4.8 V.With the half-porous BPM concept presented in this work, we could further lower the cell voltage and achieved a partial current density i CO of 256 mA cm −2 at 3.6 V.
To demonstrate the applicability of the proposed bipolar membrane long-term measurements were performed, which normally led to significant degradation of the bipolar membranes within a few hours as observed by various groups. [23,25]Therefore, we operated the half-porous BPMs and the commercial reference membrane at a constant current of 100 mA cm −2 (see Figure 4).All cells were activated at the beginning of the experiments with a 0.1 m CsOH solution as described above and in the experimental section, in order to benefit of the co-catalytic effect of the Cs + at the cathode side, while operating with pure water to mitigate the formation of salt precipitates.
The electrochemical performance in the long-term measurements is comparable to the polarization data presented in Figure 3 (see Figure S5, Supporting Information).The custom BPM with a non-perforated AEL could not be operated at 100 mA cm −2 without severe blistering and subsequent rupturing of the anion exchange membrane (see Figure S6, Supporting Information).The fumasep FBM-PK, operated ≈17 h before the cell potential started to rise rapidly from 3.1 to 3.4 V.A second cathode activation could at first recover the cell voltage but it drastically rises over the following 15 h until it reaches the cutoff voltage of 4.5 V.The following cathode activation procedure could not recover the cell performance and the cell, thus, failed after a total of ≈37 h.
Contrary to that, our 1x-perforated BPM could be operated for 200 h without cell failure.During the measurement, the activation procedure was repeated twice, after ≈28 h and 140 h.After the first cathode activation, the cell voltage stayed below 3.2 V and was recovered by the consecutive activation procedures to ≈3.0 V.The Faradaic efficiency FE CO reaches a plateau of 88% until the second cathode activation.After that, the FE CO stays between 75 and 80% until the end of test (200 h).The 3x-perforated BPM shows the highest FE CO of the three cells with ≈90% at the beginning of the experiment.However, over the first 24 h the CO Faradaic efficiency constantly decreases to ≈80%.In this case, the second cathode activation step immediately leads to a slight recovery of the FE CO to 86%.Over the following 43 h, the FE CO decreases again to 70%, while the cell voltage rises to 3.5 V.After the third cathode activation (after 67 h), both, cell voltage and FE CO recover and show more stable values until the end of the experiment (82% FE CO at a cell voltage below 3.1 V).The 3x-perforated BPM shows no visible damages after disassembly (Figure S7, Supporting Information), in contrast to the 1xperforated BPM, which featured two lager holes in the AEL and two bumps in the GDL at the same spots (Figure S8, Supporting Information).This is indicating that 3x perforation is necessary for continuous gas removal.Furthermore, the isopropanol contained in the activation solution might have negative effects on the mechanical properties of the AEL and could be a possible reason for the sudden degradation of the FE CO of the 1x-perforated BPM after the second activation and the observed local blistering.
Post-operation electron micrographs of a 3x-perforated AEM show a similar hole distribution and hole shape as seen in Figure 2a (see Figure S9, Supporting Information).This indicates that the membrane porosity is maintained during operation.In the cross-section electron micrographs of the 1x-perforated cell after the long-term measurement (see Figure S10, Supporting Information) small delaminated regions (≈300 μm) can be found, while the major part of the membrane remains intact.
No salt precipitates were visible after the long-term measurements upon visual inspection.Furthermore, the constant potential and high FE CO indicate that the BPM did not suffer from the typical precipitation and gas blocking as AEM cells with alkaline or (bi)carbonate-based electrolytes do.For industrial application electrolyzer operating times of at least 50 000 h are seen necessary (based on life time goals for water electrolyzers). [34]A voltage increase margin of 500 mV over the life-time of 50 000 h would allow for a degradation rate of 0.01 mV h −1 .The degradation rate of the 1x-perforated BPM as captured by the cell voltage increase over the whole experiment (calculated from 3 -200 h) is 0.5 mV h −1 and ≈1.3 -2 mV h −1 between the individual activation procedures.While the degradation rate measured in this experiment is still more than one order of magnitude higher than the required degradation rate, the voltage could mostly be recovered by repeating the activation procedure.
All three cells show small erratic fluctuations in cell voltage during the measurement.These voltage instabilities are most probably caused by condensate droplets from the gas humidification that enter the cathode compartment (see Figure S11, Supporting Information).As cesium (bi)carbonate salts are soluble in water, the water droplets flushing through the cell can wash away the Cs + cations and accelerate the need for the reactivation of the cell. [32]o evaluate the single-pass conversion of CO 2 , we varied the CO 2 feed rates between 3 -50 mL min −1 at a constant current of 100 mA cm −2 .Figure 5a shows the single-pass conversion and the Faradaic efficiency of the 1x-perforated BPM along with the theoretical maximum single-pass conversion (see Figure S12, Supporting Information for results of the 3x-perforated BPM).As 100 mA cm −2 corresponds to the conversion of 2.98 mL min −1 CO 2 , a high single-pass conversion is only possible close to this flow rate.Thus, the cell reaches the highest single-pass conversion of 52% at a CO 2 flow rate of 3 mL min −1 .However, at these low flow rates the Faradaic efficiency FE CO is as low as ≈55%, while reaching values above 85% at higher flow rates.This shows the trade-off between single-pass conversion and Faradaic efficiency, which is another motivation to go to higher CO 2 flow rates.To find this optimum, several studies investigated the separation cost and concluded that the single-pass conversion should only be optimized, if the cell voltage increase stays below 0.2 V and the CO Faradaic efficiency does not decrease significantly. [1,35]Hence, for our cell the 5 mL min −1 flow rate might be the optimum with a FE CO close to 90% and a singlepass conversion of 49%.
Blommaert et al. compared the energy demand of bipolar membranes in reverse bias to the energy consumption of AEMbased electrolyzers that are operated with an alkaline anolyte. [16]s discussed above, AEM-based electrolyzers using alkaline electrolytes, e.g.KOH solutions, consume hydroxide ions at the anode.In continuous operation these hydroxide ions need to be  [16] and **calculated for a cell voltage of 2.8 V as measured under the same operating conditions as the BPM cells using a 0.1 m KHCO 3 solution as anolyte (see Figure S13, Supporting Information).regenerated externally to maintain the high pH at the anode.Furthermore, as CO 2 crosses over to the anode side in the form of (bi)carbonates, the CO 2 must be recovered from the anode gas stream.Those processes increase the overall energy demand of the electrolysis system.The energy demand that Blommaert et al. calculated for those processes in AEM-based electrolyzers using KOH as an anolyte is displayed in the first column of Figure 5b.The additional recovery steps increase the energy demand more than threefold compared to the pure electrical energy demand for the CO 2 reduction to ≈1792 kJ mol −1 CO.
To mitigate the need for the OH − recovery in AEM-based electrolyzers, they can be directly operated with neutral electrolytes at the cost of increased cell voltages. [14]However, this significantly reduces the energy demand to ≈905 kJ mol −1 CO as it removes the need for the OH − recovery (2.8 V cell voltage at 100 mA cm −2 as measured with 0.1 m KHCO 3 anolyte under the same operating conditions as in the BPM measurements, see Figure S13, Supporting Information).Hansen et al. published comparable cell performance at 100 mA cm −2 with a KHCO 3 -based anolyte and 40 °C. [33]In their study, Blommaert et al. achieved a FE CO of 30% with a reverse bias BPM at 100 mA cm −2 , yielding an energy demand of ≈2400 kJ mol −1 CO but highlighted the high potential of bipolar membranes, if the FE CO could be increased toward 90%. [16]The electrolysis energy demand of our perforated BPM presented is significantly lower (665 kJ mol −1 CO at FE CO = 90% and 3.1 V, without the energy demand of the activation solution).According to a techno-economic analysis by Moore et al. for CO 2 electrolysis producing ethylene the mitigation of (bi)carbonate crossover allows for a cell voltage increase of 0.4 to 0.8 V, when capital expenditure is accounted for. [35]Thus, the here presented perforated BPM is an attractive alternative to conventional AEM-based electrolyzers only showing 0.3 V higher cell voltage at 100 mA cm −2 compared to AEM-based electrolyzer using KHCO 3 .
While cation exchange membranes block CO 2 crossover via (bi)carbonates by Donnan exclusion and the release of gaseous CO 2 through local acidification, there are other processes that can still lead to CO 2 permeation through the CEL.The CO 2 released at the AEL|CEL interface can dissolve in the liquid water or the solid polymer phase of the CEL and permeate to the anode side. [36]However, diffusion through cation exchange membranes is typically very low compared to the carbonate crossover rates in AEM-based CO 2 electrolyzers as shown by O'Brien et al. (the (bi)carbonate crossover in AEM cells causes a 25-times higher CO 2 anode outlet flow compared to cells with a Nafion membrane). [27]Nonetheless, for the optimization of the proposed setup, a more detailed analysis of gas permeation should be conducted in future work to identify the ideal CEL material and thickness.

Conclusion
In this study, we propose a strategy to enhance the long-term performance of forward bias bipolar membranes in zero-gap CO 2 electrolyzers for CO production.The key aspect of this strategy is a porous anion exchange layer that facilitates the recirculation of CO 2 from the AEL|CEL interface to the cathode, effectively reducing blistering of the bipolar membrane.Additionally, we adopted the cathode activation procedure by Endrődi et al. [32] to enable high CO partial current densities, while utilizing pure water as anolyte.
With the combination of those two strategies, we were able to operate the bipolar membrane at 300 mA cm −2 and 3.6 V with a Faradaic efficiency FE CO of 85% yielding a partial current density i CO,max of 256 mA cm −2 .Compared to literature this is with an EE CO of 43% the highest energy efficiency achieved at 100 mA cm −2 for a zero-gap bipolar membrane setup, showing that the perforation has no negative effect on performance.
Furthermore, we show with long-term measurements of 200 h that the perforation approach is able to mitigate gas accumulation and thus membrane failure.While the BPM with a nonperforated AEM was not able to operate at 100 mA cm −2 , the half-porous BPMs showed only slow voltage degradation (1.3-2 mV h −1 ), which was mostly recovered by repeating the activation procedure.The formation of salt precipitates, which is commonly observed in electrolyzers using alkaline or (bi)carbonatebased electrolytes, was mitigated by the employment of pure water as an anolyte.The overall voltage increase rate of 0.5 mV h −1 is still more than one order of magnitude too high for the desired lifetime of 50 000 h and 500 mV voltage increase.However, with further optimization of the concept and a better understanding, especially of the cathode activation, it should be possible to further reduce the voltage increase.
A maximum single-pass conversion of 52% at 0.75 mL min −1 cm −2 CO 2 and a low specific energy consumption of 665 kJ mol −1 CO 2 shows that this concept is superior not only to other bipolar membranes in literature, but potentially also to conventional anion exchange membrane-based electrolyzers.The comparison assumes no CO 2 must be recaptured from the anode, as the cation exchange membrane prevents carbonate crossover, which should be addressed in a future study.Further ideas to improve the perforated BPM could be a) optimizing the AEL porosity and thickness or b) improving the AEL|CEL junction and integrating a reinforcement for improved stability.Furthermore, the effects of process parameters, like cell temperature, cathode back pressure, and water management need to be investigated in more detail.Both an improved design and understanding have the potential to improve the half-porous BPM concept even further and make forward bias bipolar membranes suitable for industrial scale CO 2 electrolysis.
Electrode Fabrication: For the fabrication of anode catalyst layers, IrO 2 nanoparticle dispersions were prepared according to the following procedure: IrO 2 nanopowder (500 mg), H 2 O (454 mg), and Nafion D2020 dispersion (328 mg) are mixed with a Thinky Mixer at 1500 rpm for 10 min.IPA (515 mg) is added, followed by an additional mixing step at 1500 rpm for 15 min.Then additional IPA (1030 mg) was added and a mixing step at 2000 rpm for 7 min was performed.The anode ink was casted onto a PTFE sheet with the help of a 120 μm Meyer-rod.Figure S14 (Supporting Information) shows the X-ray fluorescence (XRF) mapping of the casted IrO 2 catalyst layer, showing an average loading of 1.1 mg cm −2 .Only areas with uniform loading were used for electrode fabrication.Squares of 4 cm 2 were cut from the PTFE sheet with a customized punching tool.The catalyst layer was then decal transferred onto a Nafion 212 CEM at a temperature of 155 °C and a pressure of 1.4 kN.
Porous transport electrodes were fabricated for the measurements with commercial BPM, as the membrane should not be tried.For that, a titanium PTL was coated with an IrO 2 catalyst layer using an atomized ultrasonic spray coater (Sono-Cell SNR300, Sonaer).The ink consisted of Premion IrO 2 (200 mg), Nafion D520 (5%, 81.633 mg), water (10.063g), and IPA (10.063 g).The ink was sonicated with an ultrasonic horn for 30 min before spray coating.During spraying the flow rate was set to 0.2 mL min −1 and the ultrasonic nozzle to 60 kHz with a power of 6 W. The heating plate was set to 70 °C.
The Ag-based cathode gas diffusion electrodes were also fabricated by ultrasonic spray coating.The ink was prepared by mixing Ag nanoparticles (0.302 g), H 2 O (9.886 g), IPA (7.755 g), and PipierION A5 ionomer solution (0.488 g).The ink was sonicated for 30 min with an ultrasonic horn before spray coating.The ink was spray coated onto Freudenberg H23C6 gas diffusion layer at a flow rate of 0.2 mL min −1 with the heating plate being set to 50 °C.XRF mapping of the silver GDE is displayed in Figure S15 (Supporting Information).
Fabrication of Bipolar Membranes: For the fabrication of the porous AELs, anion exchange membranes (Aemion+ AP3-HNN9-50-X) were ionexchanged for 24 h in 1 m NaCl to the Cl − -form and subsequently to the OH − -form for 24 h in 1 m KOH.After the ion exchange, the AEM was perforated with a needle roller (Micro Skin Beauty System 540-0.3,Biotulin).For that, the wet membrane was placed on a polypropylene foil.After the perforation, the AEM was carefully placed on top of a wetted Nafion 212 membrane (coated with the IrO 2 catalyst layer on the other side).Excess KOH solution on the AEM from the ion exchange was carefully removed with a fuzz-free paper tissue, before stacking the membranes.If air bubbles were trapped between the membrane layers, the procedure was repeated.The commercial bipolar membrane (fumasep FBM-PK, Fumatech), originally designed for reverse-bias applications, arrived already in hydrated form and was washed in DI-water for ≈24 h prior to testing.
Measurement of the Gas Permeability: To assess the CO 2 gas permeability of the perforated AEMs, they were placed in the electrolysis cell fixture between a GDL (without catalyst layer) and a titanium PTL.The gasket thickness on the cathode side was reduced by 50 μm (from 200 to 150 μm) to obtain a similar compression as with the 50 μm thick Nafion membrane in the electrochemical experiments.The cathode outlet and the anode inlet were closed and a mass flow meter was attached to the anode outlet.Different CO 2 flow rates (5, 10, 25, 50, 100, and 200 mL min −1 ) were applied and the pressure increase on the cathode side was recorded with a digital pressure sensor (Omega, PXM409-001 BGUS BH).The schematics of the measurement setup can be seen in Figure S16 (Supporting Information).The highest possible rate of CO 2 formation at the AEL|CEL interface ( VCO2 ) was calculated according to the following equation: VCO2 = I z F V m .Where I is the applied current.The charge transfer number z equals 1, assuming 100% ion conduction by HCO 3 − .F represents the Faraday constant (96 485.33 C mol −1 ) and V m the molar volume of CO 2 at standard conditions.
Electrochemical Characterization: A serpentine flow field was used on the cathode side and a parallel flow field on the anode side (both made from titanium grade 2, plated with gold).PTFE gaskets of 200 μm thickness on the cathode side and two 250 μm thick PTFE gaskets on the anode side were used as spacers.The 8 screws of the cell fixture were tightened one after another diagonally with a torque of 3 Nm and 6 Nm subsequently.
The CO 2 flow rate at the cathode was set to 50 mL min −1 with a mass flow controller (EL-FLOW Prestige FG-201CV, Bronkhorst).A bubbler was heated to 30 ± 1 °C and used to humidify the CO 2 gas stream before entering the cell.DI-water served as the anolyte, which was circulated by a peristaltic pump (REGLO ICC Digital, Ismatec) and preheated with a thermostat.The cell temperature was set to 40 °C, with two heating cartridges inserted into the flow fields.
All electrochemical measurements were conducted using a BioLogic VSP-300 potentiostat.Prior to the electrochemical measurements a cathode activation procedure was conducted as proposed by Endrődi et al.. [32] For that, all cells were first operated at a constant current of 12.5 mA cm −2 for 10 min.The measurement was paused and the CO 2 gas flow was interrupted, to feed a 0.1 m CsOH solution (in IPA/H 2 O, 1:3 vol , 0.5 mL) to the cathode compartment.Another 12.5 mA cm −2 constant current step was performed for 30 min.The current was applied ≈5-10 s before restarting the CO 2 flow to flush out the activation solution from the cathode compartment.
For the polarization curves the current steps of 12.5, 25, 50, 100, 150, 200, 300, and 400 mA cm −2 were applied for 10 min each or until the cell voltage increased above 5 V.After 9 min a gas sample was taken from the cathode outlet and analyzed with an on-line gas chromatograph (GC, Agilent micro GC 990, equipped with a CP-COX column using Argon as carrier gas).Additionally, a galvanostatic electrochemical impedance spectroscopy measurement with an amplitude equal to 1/20 of the applied current in a frequency range of 800 kHz to 1 kHz was performed to determine the high-frequency resistance of the cell.Activation and polarization were conducted twice with each cell.For the long-term tests the conditioning procedure was performed analogously.
The Faradaic efficiency was calculated as follows, FE CO = Where z is the number of electrons needed to form CO from CO 2 , R is the ideal gas constant, p is the pressure, x CO is the mole fraction of CO as measured by GC, and T is the temperature.The cathode product gas flow rate Vout was measured with a mass flow meter (EL-FLOW Select F-111B, Bronkhorst) after drying.A fixed flow of N 2 was additionally added to the cathode product gas as an internal standard.
The single-pass conversion of CO 2 (SC) is calculated by dividing the volume flow of carbon monoxide in the product gas stream by the CO 2 feed flow rate: SC = VCO,out VCO2,in 100%.
The Energy efficiency for CO is calculated through EE CO = E 0 E cell FE CO where E 0 is the equilibrium potential of the reaction to generate CO and oxygen from CO 2 and E Cell is the voltage applied to the electrolysis cell.
Scanning Electron Microscopy: Electron micrographs were recorded with a TESCAN MAIA3 scanning electron microscope.The accelerating voltage was set to 5 kV and a secondary electron detector was used.
Calculation of Electrolysis Energy Demand: The electrical energy consumed by the electrolysis per produced mole of CO (J EL ) was calculated as follows, . With z = 2 (charge transfer number for CO 2 to CO), the cell voltage (E Cell ) and the FE CO for the specific working point of the electrolyzer.

Figure 1 .
Figure 1.Simplified schematics of different membrane concepts for zero-gap CO 2 electrolyzers.a) state-of-the-art concept with anion exchange membrane (AEM), b) cell arrangement with a cation exchange membrane (CEM), c) state-of-the-art bipolar membrane in forward bias orientation, and d) bipolar membrane with a porous anion exchange layer.

Figure 2 .
Figure 2. Perforated anion exchange membranes.a) top view electron micrograph of a three times perforated membrane showing an uneven distribution of the holes and b) differential pressure measurement of the perforated anion exchange membranes (AEM), sandwiched in the electrolysis cell fixture between a gas diffusion electrode and a porous titanium transport layer.The effective membrane area is 4 cm 2 .

Figure 3 .
Figure 3. Electrochemical characterization of various bipolar membranes.a) cell voltage E Cell , b) Faradaic efficiency for CO FE CO , c) high frequency resistance HFR in relation to the applied current density of four different bipolar membrane setups, and d) cell voltage in relation to the partial current density of CO i CO compared to other forward-bias BPM studies by Pribyl-Kranewitter et al.[25] and Hansen et al.[33] Measurements were conducted at 40 °C cell temperature.The CO 2 feed (50 mL min −1 ) was humidified at 30 °C.The anode was supplied with DI-water at a flow rate of 20 mL min −1 .Each current step was held for 10 min.

Figure 4 .
Figure 4. Long-term measurements of a commercial fumasep FBM-PK membrane and two bipolar membranes with perforated anion exchange layers.A constant current of 100 mA cm −2 was applied, with 50 mL min −1 CO 2 , a bubbler temperature of 30 °C, 20 mL min −1 DI-water on the anode side, and a cell temperature of 40 °C.Electrochemical impedance spectroscopy and gas chromatography measurements were conducted every 10 min.The asterisks mark the execution of the cathode activation procedure with a 0.1 m CsOH solution.

Figure 5 .
Figure 5. Single-pass conversion and energy consumption.a) Faradaic efficiency and CO 2 single-pass conversion at different CO 2 flow rates measured with a 1x-perforated membrane.The error bars represent the standard deviation of at least three consecutive measurements.b) energy demand by various cell designs with different membrane concepts.*values adapted from Blommaert et al.[16] and **calculated for a cell voltage of 2.8 V as measured under the same operating conditions as the BPM cells using a 0.1 m KHCO 3 solution as anolyte (see FigureS13, Supporting Information).