Identifying and alleviating the durability challenges in membrane-electrode-assembly devices for high-rate CO electrolysis

doi:10.21203/rs.3.rs-2491212/v1

Download PDF

Article

Identifying and alleviating the durability challenges in membrane-electrode-assembly devices for high-rate CO electrolysis

https://doi.org/10.21203/rs.3.rs-2491212/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 28 Sep, 2023

Read the published version in Nature Catalysis →

Version 1

posted

You are reading this latest preprint version

CO electrolysis (COE) has emerged as an important alternative technology to couple with other sustainable techniques for transitioning towards a carbon-neutral future. A large challenge for the deployment of high-rate COE is the limited durability of the membrane-electrode-assembly (MEA). In this work, by utilizing an operando wide-angle X-ray scattering (WAXS) technique and monitoring the change of electrolyte, we identified several degradation mechanisms of the MEA during high-rate COE. Cathodic gas diffusion electrode (GDE) flooding and metal contaminants (crossover from anode) are two major issues causing excessive HER, which can be partly alleviated by increasing the PTFE content in the GDEs and using an alkaline stable Ni-based anode. When analyzing MEA’s long-term stability, the dynamic evolution of anolyte became the major issue: the pH would continuously drop due to cathodic acetate formation and anodic ethanol oxidation. By compensating for this issue, we maintained a Faradaic Efficiency (FE) of C₂₊ products to be over 70% during a 136-h testing period. Our findings provide clear guidelines to circumvent the durability challenges for high-rate CO/CO₂ electrolysis.

Physical sciences/Chemistry/Electrochemistry/Electrocatalysis

Physical sciences/Nanoscience and technology/Techniques and instrumentation/Characterization and analytical techniques

WAXS

CO electrolysis

membrane-electrode-assembly

durability

Electrochemical conversion of carbon monoxide (CO) to valuable fuels and chemical feedstocks offers a promising route for realizing carbon recycling and a sustainable future^1–3. Compared to CO₂ electrolysis (CO₂E), directly utilizing CO as a reactant helps to prevent CO₂ loss through anion exchange membranes (via carbonate/bicarbonate) and allows for a high conversion efficiency to the C₂₊ products^4–6. Benefiting from these unique features, many techno-economic assessments have pointed out that the COE has high flexibility as a module process to couple with other advanced techniques such as being the downstream process of solid oxide CO₂ electrolysis cells (SOEC-CO₂) or being the upstream process of fermentation^7,8.

Generally, CO has a low solubility in aqueous electrolytes, which causes a low mass-transfer limiting current density (< 10 mA‧cm^− 2) in a standard electrochemical H-cell^9,10. Owing to the development of gas-diffusion electrode (GDE) akin to the fuel cells community, a much higher current density (> 200 mA‧cm^− 2) for COE^11,12 can be achieved by employing GDEs in flow cells and membrane-electrode assembly (MEA) electrolyzers. In these cases, research tends to focus on the higher activity and selectivity, i.e., high partial current density (J_C2+) for C₂₊ products such as acetate, ethylene, propanol, etc., by the design of cathode catalysts or interfacial engineering^13–15. Many excellent activities have been recently reported, for example, the J_C2+ can even reach > 1 A‧cm^− 2 for a short period¹⁶. However, only a few of these electrolyzers were able to operate for > 50 hours^13–28, as shown in Fig. S1. These results indicate the durability challenges of COE to simultaneously maintain high activity and selectivity during long-term operation.

Alkaline electrolyte such as the concentrated KOH has proven to be an essential condition for realizing high-performance COE^29,30. Such highly concentrated electrolytes can not only suppress the competitive hydrogen evolution reaction (HER)^31,32 and enhance the reaction kinetics for CO reduction, but they can also reduce the ohmic resistance related to ion transport. However, durability challenges including the gradual predomination of HER, the continuous increase of cell voltage, etc. have always been accompanying long-term (> 10 h) electrolysis experiments. Some similar degradation behaviors can be learned from communities of other types of electrolysis^33,34. However, the characteristic features of COE such as the high formation rate of liquid products and high sensitivity to interfacial environments entail the MEA-based COE has unique degradation mechanisms, which are non-relatable to other electrolysis-based fields. Thus, there is currently a lack of understanding in identifying the degradation behavior of COE during long-term operation.

Herein, by combining the operando wide-angle X-ray scattering (WAXS) measurement with detailed electrochemical characterizations and monitoring the dynamic change of electrolyte, we explored several degradation mechanisms of the MEA during high-rate and long-term COE.

Identifying MEA degradation behavior

To precisely investigate the degradation behavior inside the MEA, a customized synchrotron cell (Fig. 1) was used for performing COE experiments. An industrially relevant current density of 200 mA‧cm^− 2 was used to operate the cell for 3 h with the anolyte of 0.1 M CsOH, meanwhile operando WAXS was measured. A detailed introduction of the MEA configuration and operating conditions are given in the experimental procedures (Methods). During COE measurement, a high transmission and focused synchrotron X-ray beam (5 µm × 20 µm, vertical × horizontal) periodically scanned the MEA from the IrO₂ anode through the membrane and all the way to the cathode flow field. Each scan takes 210 s to acquire 100 WAXS patterns across the MEA region. Therefore, each operando experiment produces substantial amounts of WAXS data in terms of both time and space, allowing for the tracking of real-time behavior of the entire scanned MEA. From the intensity and background changes of characteristic signals in the WAXS mapping, the change in electrolyte content and the formation of new crystalline phases can be identified³⁵. Moreover, the gas products during the COE and liquid products after COE were measured by in-line gas chromatography (GC) and ex-situ high-performance liquid chromatography (HPLC), respectively. Detailed discussions of operando X-ray measurements and data analyses are given in Methods and Supporting Information (Note 1).

Sputtered Cu on a commercial GDE (i.e., 39BB) was used as a baseline cathode. Its morphology and surface chemical state are given in Fig. S2. The cross-section SEM image and corresponding EDS mapping of a typical MEA configuration are shown in Fig. S3. The MEA consisting of an AEM with a compressed thickness of ~ 35 µm, is sandwiched by the Cu cathode and a commercial IrO₂ anode. Figure 2 demonstrates the correlated cell voltage, Faradaic Efficiency (FE) of gas products, and the processed operando WAXS mapping during a COE measurement. In the WAXS mapping, the characteristic peaks of specific materials help to identify the layered configuration of the MEA. For example, the position of the Cu layer can be monitored by integrating WAXS patterns in the q-range of 2.95–3.05 Å⁻¹, which corresponds to the main peak of fcc-structured copper. Figure 2b shows that the sputtered Cu is located both on top of and in the microporous layer (MPL). Moreover, changes in the GDE electrolyte content can be identified by noting changes in background scattering and is shown in Fig. 2c.³⁶ A typical WAXS pattern within the Cu layer region after reacting for ~ 30 min is shown in Fig. S4.

During COE, the cell voltage (Fig. 2a) quickly increases in the first 30 min at a rate of ~ 5.5 mV min^− 1, and then slowly increases at a rate of ~ 0.7 mV min^− 1 in the subsequent 150 min. This trend is in accordance with the increased water content in the GDE, where both macro- and micro-porous layers are flooded by water (Fig. 2c). A more direct display is shown in Fig. S5a by the WAXS mapping in the region of the MPL. It shows the intensity of electrolyte signals increases by ~ 31% in the initial 30 min and then only by ~ 14% in the next 150 min. This gradual increase of water content in the GDE should be one of the reasons for the FE shift towards HER (Fig. 2e). It needs to be mentioned that the low FE of gas products during the first 15 min of COE is attributed to electrode conditioning and some dead volume in the tubing connected to the GC for gas analysis.

The WAXS mapping of the integrated Ir peak is presented in Fig. 2d. It shows that Ir appears to deposit on the Cu cathode during COE. Fig. S5b shows the evolution of WAXS patterns at the Cu layer facing toward the AEM side, which demonstrates that the Ir facet distribution is inconsistent between various redepositions indicating the complex nature of this process. In addition, the XPS spectrum of Ir 4f for the post-tested Cu cathode (Fig. S6) further verifies the existence of Ir contaminant. The Pourbaix diagram of the IrO₂ indicates that IrO₂ can be unstable in the alkaline condition under oxidative potentials³⁷. Thus, we speculate that the Ir-based anode would slowly dissolve in anolyte with pH ~ 13 under oxidative potentials. Additionally, the underlying carbon on which Ir is supported is also not stable under oxidative potentials, which may accelerate Ir issues³⁴. Once dissolved, it is then believed this Ir will crossover the AEM and subsequently deposit on the cathode under reductive potential. As a result, the Ir contaminant, being a good HER catalyst, causes a significant HER increase at the cathode.

Alleviating the MEA degradation

To further decouple the influence of GDE flooding and Ir contamination on performance loss, we operated two more operando measurements by changing the cathode GDEs and the thickness of the membrane, as shown in Fig. 3a. A higher PTFE content has been reported to lessen the water flooding of the GDE during CO₂ reduction^38–40. In view of this, two kinds of customized GDEs: Sigracet 39BB (with 5 wt% PTFE) and 39DC (with 20 wt% PTFE) were compared. The surface morphology and chemical state of sputtered Cu on different GDEs are consistent (Fig. S2 & S7). On the other hand, employing a thicker membrane would increase the water/ion transfer pathway and thus mitigate flooding and Ir ions crossing over the membrane⁴¹. In Fig. 3b, the 39DC-2Mem indicates the use of two layers of membrane and 39DC GDE during the COE. Detailed information about cell voltage, FEs of gas and liquid products, and the processed WAXS mappings are provided in Fig. S8 & S9 and Table S1. Below, we highlight the most noticeable improvements by changing those MEA components.

Figure 3b compares the FE changes of the gas products between different experiments. Expectedly, the 39DC with a higher PTFE content shows a lower FE of HER than that of 39BB, demonstrating an improvement of ~ 10%. Meanwhile, the FE of COR (including CH₄ and C₂H₄) shows an opposite trend. The 39DC-2Mem shows a stable FE towards both HER and COR. Operando WAXS mapping and signals in Fig. 3c and Fig. S10 show the electrolyte content change in the MPL at the cathode. The 39DC shows a lower increase rate of electrolyte content compared to 39BB during the first 60 min, agreeing with the observed HER trends in 39DC and 39BB. This variation trend in electrolyte content indicates that increasing the PTFE content helps to alleviate the electrolyte flooding and thus prevents HER from increasing. The 39DC-2Mem further shows a slowdown of the increased rate of electrolyte content in the MPL during the entire electrolysis period, indicating that a thicker membrane reduces the water crossover from the anode to the cathode.

As we are interested in Ir contaminations deposited just on the Cu layer, this allowed us to analyze the WAXS at this fixed location as a function of time. Figure 3d shows the WAXS mapping of Cu layer facing towards the AEM for the 3 samples (39BB, 39DC, and 39DC-2Mem) with the grey and red dotted lines showing the diffraction peaks in q-space for Cu and Ir respectively. The 39DC-2Mem exhibits a much smaller Ir contamination content than those of 39BB and 39DC. The XPS of post-measurement cathode GDEs reveals the surface Ir/Cu ratios (Fig. S6) to be ~ 0.07 for 39DC-2Mem, ~ 0.12 for 39DC, and ~ 0.11 for 39BB. A similar Ir/Cu ratio for 39DC and 39BB can explain their comparable FE increasing rate of HER (e.g., ~ 5.6%‧h^− 1 for 39 BB and ~ 5.3%‧h^− 1 for 39DC). For the 39DC-2Mem, the thicker AEM, which intrinsically resists the crossover of cations, can largely reduce the Ir crossover and thus alleviate the COR degradation issue observed in 39DC and 39BB. However, increasing the membrane thickness effectively just delays the Ir crossover, thus a long-term permanent solution is needed.

As nickel-based catalysts (e.g., Fe-doped NiO_x) are normally stable materials and the benchmark catalyst in alkaline water electrolysis, they can be used as alternative anodes^34,42. Practically, a Ni catalyst would convert to a stable NiOOH/NiO₂ without the undesired metal dissolution under the oxidative potential according to the Pourbaix diagram (Fig. S11)⁴³. Based on this argument, we used a commercial nickel foam (NF) as an anode and achieved a stable 3 h COE performance for both 39DC and 39BB with only one membrane, as shown in Fig. S12. 39DC shows a stable FE of HER below 8% during this 3 h testing period, which is much lower in contrast to that of the IrO₂ anode. This result again indicates that Ir contamination is a dominating parameter towards the increase in HER. Survey XPS of post-electrolysis Cu cathode (Fig. S13) detected no Ni contamination, further verifying the stability of NF as an anode. Moreover, the stabilized FE of HER for 39BB when coupled with the NF anode is ~ 13%, which is ~ 5% higher than the 39DC. This indicates that even excluding effects due to Ir crossover, a higher PTFE content still can reduce the level of HER by reducing the degree of flooding in the GDE.

In brief, we conclude that increasing the hydrophobicity of MPL and employing a thicker AEM can help to circumvent the water penetration but cannot completely prevent electrolyte flooding. The Ir contamination from the IrO₂ anode degradation under alkaline conditions is the dominating reason that cause a significant increase in HER, which cannot be easily prevented. Considering that the Ir-based catalysts are still used as an anode in almost half of the works on CO/CO₂ electrolysis⁴², we strongly advocate researchers to pay attention to the reacting conditions. If testing in an alkaline solution, an alkaline stable anode such as Ni-based catalysts should be employed.

Anolyte influenced by anodic oxidation

According to the above results, we expected to achieve long-term COE stability by employing sputtered Cu on 39DC as the cathode and NF as the anode. Figure 4a demonstrated the cell voltage, pH value of the anolyte, and FEs of gas products during the electrolysis process. However, an interesting phenomenon appeared to catastrophically degrade the COE selectivity during the measurement. As Fig. 4a shows, the FE of COR and HER were relatively stable up until 9 h and thereafter started to vary much more rapidly. By monitoring the pH of the anolyte, we found that the bulk pH dropped continuously from 13 to 5.4 during the COE process. Considering the Ni dissolution would be largely accelerated when the pH < 12 (i.e., after 9 h), the sudden increase of cell voltage could be due to the Ni dissolution process. Afterward, the Ni²⁺ crossover would contaminate the cathode where it could reduce in act as a H₂ catalyst (analogous to Ir), leading to a further drop in the COR/HER ratio. Figure 4b and Fig. S14 give the galvanostatic electrochemical impedance spectroscopy (GEIS, measured per one hour), the equivalent circuit model, and their fitting results. The R_ct and R_s are in response to the charge transfer resistance and ohmic resistance of the testing system, respectively. The fitted R_ct shows a large increase and a peak value after 9 h, which should be the combined influence of the Ni dissolution process and reaction kinetics retardation due to the pH change. The R_s did not change much until the cell shorted out after 13.5 h. The digital image of the post-tested anode in Fig. S15a verifies that half of the NF anode was dissolved. For the post-tested cathode (Fig. S15b, c), the morphology of sputtered Cu on the 39DC did not show any significant change, while the energy dispersive X-ray spectroscopy (EDS) reveals the formation of Ni (~ 11 at%) on the Cu cathode, verifying that the dissolved Ni anode deposited on the cathode.

In addition, another interesting phenomenon is that the distribution of liquid products changes during the COE process, as shown in Fig. 4c. The FE of all products is calculated based on the entire electrolysis period. The liquid products are primarily detected from the anolyte since the cathode is a zero-gap device, and there is nowhere to extract any liquid products except from a water wash of the outgoing cathode gas stream. During the first 1 h, ethanol is the main liquid product for COR, showing a maximum FE of ~ 27%. The FEs of acetate, propanol, and acetaldehyde are ~ 22%, ~ 10%, and ~ 2%, respectively. After 13.5 h, the FEs of acetate and acetaldehyde increase to ~ 47% and ~ 6%, while the FEs of ethanol and propanol decrease to ~ 4% and ~ 2%. This selectivity change of liquid products from ethanol to acetate and acetaldehyde during the COE implies the occurrence of ethanol oxidation at the anode due to the substantially more cathodic redox potentials compared to water oxidation^14,44. Similar cases were also reported in the alkaline direct ethanol fuel cell (DEFC) field where the ethanol would be partially oxidized to acetate during operation^45,46.

More importantly, when considering the production and consumption of hydroxides of the available cathodic and anodic reactions to the respective products (Table 1), we found that these reactions can explain the observed pH drop (Fig. 4a) during the electrolysis. Generally, a complete overall reaction should result in a constant concentration of charged species. This means that a produced OH¯ at the cathode side should diffuse through the AEM and participate in the anodic reaction, thus the bulk pH of the anolyte should not change. However, some anodic oxidations would consume extra OH¯, thereby breaking the expected pH balance. For example, the formation of acetate or propanoate from ethanol or propanol would consume one more OH¯ at the anode than the produced OH¯ at the cathode. This net loss of one OH¯ should be one reason for the pH drop of the anolyte. On the other hand, the formation of acetate via COR can also cause an extra loss of OH¯. Thus, we demonstrate that the pH drop during the electrolysis is a combined effect of acetate formation from anodic oxidation and CO reduction. After COE for 13.5 h, the pH value of the anolyte was maintained at ~ 5.4 due to the buffer effect of acetate salts (pKa of acetate is ~ 4.7).

Table 1

Cathodic and anodic reactions during CO electrolysis and their net OH⁻ balance relationship.
Products	Cathodic reaction (OH⁻ production)	Anodic reaction (OH⁻ consumption)	Net OH⁻
Hydrogen	2𝐻₂𝑂+2𝑒⁻ → 𝐻₂ + 2𝑂𝐻⁻	2𝑂𝐻⁻ → 1/2𝑂₂+𝐻₂𝑂+2𝑒⁻	0
Acetate	2𝐶𝑂+3𝐻₂𝑂+4𝑒⁻ → 𝐶𝐻₃𝐶𝑂𝑂⁻+3𝑂𝐻⁻	4𝑂𝐻⁻ → 𝑂₂ + 2𝐻₂𝑂+4𝑒⁻	-1
Methane	𝐶𝑂+5𝐻₂𝑂+6𝑒⁻ → 𝐶𝐻₄+6𝑂𝐻⁻	6𝑂𝐻⁻ → 3/2𝑂₂ + 3𝐻₂𝑂+6𝑒⁻	0
Ethylene	2𝐶𝑂+6𝐻₂𝑂+8𝑒⁻ → 𝐶₂𝐻₄+8𝑂𝐻⁻	8𝑂𝐻⁻ → 2𝑂₂ + 4𝐻₂𝑂+8𝑒⁻	0
Ethanol	2𝐶𝑂+7𝐻₂𝑂+8𝑒⁻ → 𝐶₂𝐻₅𝑂𝐻+8𝑂𝐻⁻	8𝑂𝐻⁻ → 2𝑂₂ + 4𝐻₂𝑂+8𝑒⁻	0
		𝐶₂𝐻₅𝑂𝐻+8𝑂𝐻⁻ → 𝐶𝐻₃𝐶HO + 3/2𝑂₂+5𝐻₂𝑂+8𝑒⁻	0
		𝐶₂𝐻₅𝑂𝐻+9𝑂𝐻⁻ → 𝐶𝐻₃𝐶𝑂𝑂⁻+ 𝑂₂+6𝐻₂𝑂+8𝑒⁻	-1
Propanol	3𝐶𝑂+10𝐻₂𝑂+12𝑒⁻ → 𝐶₃𝐻₇𝑂𝐻+12𝑂𝐻⁻	12𝑂𝐻⁻ → 3𝑂₂ + 6𝐻₂𝑂+12𝑒⁻	0
Propanol	3𝐶𝑂+10𝐻₂𝑂+12𝑒⁻ → 𝐶₃𝐻₇𝑂𝐻+12𝑂𝐻⁻	𝐶₃𝐻₇𝑂𝐻+13𝑂𝐻⁻ → 𝐶₂𝐻₅𝐶𝑂𝑂⁻+ 2𝑂₂+8𝐻₂𝑂+12𝑒⁻	-1

To decouple the influence of the anodic oxidation and the CO reduction, a control experiment of adding an extra 1 M ethanol in the anolyte and feeding Ar gas at the cathode was performed (Fig. S16). These results confirmed our hypothesis related to Fig. 4 in that both the drop in pH and production of species such as acetate, acetaldehyde and even formaldehyde can be at least partially attributed to anodic ethanol oxidation.

Additionally, we computationally simulated the pH change when considering a consistent formation of acetate/acetic acid during the reaction (regardless of via ethanol oxidation or CO reduction), as shown in Fig. 4d (details are discussed in the Methods). The simulated curve shows a similar pH change tendency to that of the experimental curve but a sharper pH drop, confirming our assumptions. In summary, the results indicate that the formation of acetate during the entire COE process plays a bifunctional role to change the pH of anolyte: (1) when the OH¯ is rich (i.e., at high pH), the formation of acetate acidifies the solution and leads to the pH drop, similar to a titration process; and (2) when the pH starts to near the pKa of acetate/acetic acid, then the acetate converts to acetic acid, effectively buffering the pH to ~ 5.4, which is slightly above the pKa of 4.7.

Selectivity changes during long-term COE

The above results have shown the importance of maintaining the pH of the anolyte for achieving long-term stability of COE. Accordingly, several methods were employed to achieve this, such as increasing the anolyte volume and refreshing the anolyte. By artificially eliminating anolyte pH issues, this allowed us to analyze if any other obvious degradation mechanisms were occurring.

One approach to temporarily mitigate pH drop is to increase the anolyte volume. Figure 5a demonstrates the use of a 200 mL anolyte (0.1 M CsOH) for a stability test (previously 50 mL of anolyte was used) and refreshing anolyte per 18 ~ 24 h. As expected, the pH value remains higher than 12 during each refreshing periods, though it still slowly drops due to acetate formation. As a result, a high FE of > ~ 70% for COR was maintained at the end of this 136-h long-term COE test. During the electrolysis, we observe an increase in cell voltage along with a drop in pH value and vice versa. This potential variation can be attributed to two factors: the decreasing anolyte pH causing additional ohmic resistance loss and the Nernstian shift of equilibrium potential due to pH variations (deconvoluting these two issues is beyond the scope of this work). Moreover, a similar phenomenon was observed where the selectivity of ethanol decreases with a simultaneous increase in acetate formation. We attribute anodic oxidation (ethanol converted to acetate) discussed in the previous section as a partial contribution to this. Additionally, the dynamic evolution of the microenvironment nearby the Cu catalyst may also potentially contribute to the selectivity change. Similar performance and phenomenon were achieved in another 110-h experiment (Fig. S17) with pH value maintained between 12 and 13, indicating the repeatability of our tactics for long-term CO electrolysis.

To decouple the selectivity change at the cathode from the anodic oxidation, the liquid products collected both at the cathode (by passing the outlet cathode gas through a water trap) and anolyte were separately analyzed, and their production yields are shown in Fig. 5b, c. Approximately, 95% of liquid products were found in the anolyte, suggesting a significant diffusion of liquid products via the AEM. This phenomenon was also observed in previous studies.⁴⁷ The liquid production yields achieved at the anode show the combination of liquid products crossing over and their anodic oxidation. On the cathode side, the production yields are a function of the concentration of liquid products in the micro-environment nearby the catalyst. Considering that acetate would be the only product influenced by the electro-migration, we expected that the cathodic production yields of all the other liquid products (such as ethanol, propanol, etc.) can indirectly reflect the intrinsic selectivity change of the catalyst during COR. Particularly, we observed that the ethanol production yield at the cathode gradually decreases before anolyte refreshing, suggesting that the intrinsic selectivity of the Cu catalyst-based cathode indeed changes during long-term COE. By comparing the change in the rate of production yield at the cathode or anode, for example, the rate of ethanol production yield decreases by ~ 55% at the cathode and ~ 75% at the anode during the entire reacting period. Such a discrepancy of ~ 20% illustrates the large impact of anodic oxidation on the electrolysis process.

To explore the selectivity change during the stability test, we further analyzed the impedance results that were measured along with the long-term measurement. As the pH value would affect both the reaction kinetics and the ohmic resistance, we compared the initial GEIS at each refreshing period (when the pH of anolyte is ~ 13). Their Nyquist plots and fitting results are exhibited in Fig. S18 and Table S2. It shows that the fitted ohmic resistances (R_s) are slightly increased from 0.41 to 0.50 Ω; the fitted charge transfer resistances (R_ct) are gradually dropped from 1.25 to 0.83 Ω; the fitted double layer capacities (C_dl) are boosted ~ 3 times from 0.09 to 0.29 mF cm^− 2 during the entire reacting period. The gradual decrease of R_ct agrees with the increasing trend of FE_H2, indicating that the HER becomes dominant. The survey XPS of post-measurement cathode (Fig. S19a) shows no Ni signal on the Cu surface, indicating the metal contamination is not the primary reason for increasing HER. On the other hand, the dramatic increase of C_dl, which was reported to reflect the flooding degree of GDE by our previous work,⁴⁸ suggests that the cathodic GDE flooding can be the main reason for the selectivity change from COR to HER. The digital image of the AEM (Fig. S19b) further shows a change of color from transparent (pristine) to yellow (post) at electrode-contact region, reflecting the membrane degradation during measurement, which can be the reason for slight increase of R_s.

In addition, the collected liquid products proportion at the cathode or anode can give some guidelines for future development. It shows that a higher acetate proportion of ~ 90% can be achieved at the anode side due to the selectivity change, helping to gain a concentrated acetate salt solution in the anolyte. This result matches with the recent studies by Feng and co-authors where they achieved an acetate proportion of ~ 99% in the anolyte by using highly concentrated KOH and coupling with the anodic oxidation¹⁴. At the cathode side, a high ethanol proportion of ~ 60% was achieved by eliminating the influence of anodic oxidation. These results indicate that the actual ethanol selectivity at the cathode is much higher than that one would assume by just an anode-based analysis. Furthermore, if the ethanol can be more easily extracted from the cathode, such as by operating at elevated temperatures to increase its vapor pressure (boiling point of ethanol is 78°C), this could lead to much higher extracted yields without having to use a flow cell device and the intrinsic added ohmic loss that comes with the electrolyte layer.

In this work, we identified several degradation mechanisms of the MEA during the high-rate and long-term COE by utilizing an operando wide-angle X-ray scattering (WAXS) technique and monitoring the pH and change of species within the electrolyte. We found that although the cathodic GDEs flooding leads to a relatively higher FE of HER in the initial stage, the Ir contaminant (crossover from anode) is the dominating reason that causes a significant HER increase. By increasing the PTFE content in the GDEs and using an alkaline stable Ni anode, we can circumvent these two issues and achieve 3-h stability. However, the composition change of anolyte, due to the accumulation of liquid products and anodic oxidation, appears to be another big challenge during the long-term test. A pH drop was observed in the anolyte due to the acetate formation, leading to the corrosion of the anode catalyst. In view of this, we tried to maintain the pH of the anolyte by either increasing anolyte volume or refreshing the anolyte; as a result, we achieved the FE of COR (versus HER) always larger than 70% during a 136-h testing period.

Our work is expected to provide guidelines to circumvent the durability challenges in the high-rate CO/CO₂ electrolysis field. Based on our understanding, we strongly advocate researchers to select the appropriate anode catalysts according to the reaction conditions. Ni-based catalysts are recommended to be used in alkaline conditions. Additionally, in general, the anodic oxidation reaction should be carefully monitored since COE can create large changes in the anolyte composition, leading to unexpected anode degradation and misleading performance analysis. For example, the FE of acetate would be overestimated, and the FE of ethanol would be underestimated when ethanol oxidation to acetate occurs. In addition, developing advanced operando measurements is still urgently required to provide knowledge for achieving high-rate and selective performance during long-term electrolysis.

Anolyte preparation The anolyte used during CO electrolysis experiments was 0.1 M solution of CsOH (unless stated otherwise). The CsOH (purity of 99.9%) was purchased from Sigma Aldrich and used without any further purification. Millipore-quality water (18.2 MΩ at 25^◦C) was used in all experiments.

MEA configuration The cathode was a 150 nm layer of Cu (6N) sputtered on Sigracet 39 BB or Sigracet 39 DC GDL (Fuel Cell Store), while, a commercial IrO₂-based GDE (purchased from Dioxide Materials) or nickel foam (SynLectro™) was used as an anode. In between the cathode and anode, Sustanion AEM (X37-RT from Dioxide Materials) was used. For each experiment, a fresh cathode, anode, and AEM was used.

Operando X-ray measurements A customized synchrotron cell (Fig. S20a) was used for the operando X-ray studies. A full description of synchrotron cell is provided in our previous work³⁵. All wide-angle X-ray scattering (WAXS) measurements were carried out at the ID31 beamline of the European Synchrotron Radiation Facility (ESRF) in Grenoble, France. The high-energy monochromatized beam (68keV) was focused on the cell (5×20 µm² v×h) in grazing incidence mode (beam parallel to the MEA layers). The scattered signal was collected using a 2D Dectris Pilatus CdTe 2M detector positioned 780 mm behind the cell. A standard CeO₂ powder (NIST) was used for the calibration of the sample-detector distance and detector tilt. Calibration and data integration to the 1D pattern were performed using the pyFAI software package⁴⁹.

Electrochemical cell assembly and measurements A commercial CO₂ electrolyzer purchased from Dioxide Materials was used for the long-term stability test (Fig. S20b). In a typical experiment, a dry CO (5N) was fed to the cathode GDE at a constant flow rate of 20 sccm by means of Vöegtlin red-y smart series mass flow controller. At the anode, 0.1 M of CsOH anolyte was circulated using a digital diaphragm pump (KNF NF 1.5TTDCB-4) which was set to 35% of its maximum current range (flowrate of ca. 10 ml/min). The outlet cathode gas stream was passed through a gas wash (containing 20 ml pure water) to extract the liquid products before it went to gas chromatography (GC). The flow rate of the cathodic outlet streams of the electrolyzer was measured by the internal standard method (a constant flow rate of the inert N₂ gas was sent into GC) or the volumetric flow meter (GFM Pro, Thermo Scientific). The flow diagram of the electrolysis setup is exhibited in Fig. S21. At the end of each experiment, liquid samples were collected both at the anode and cathode. All electrochemical tests were carried out in a galvanostatic mode using a Biologic potentiostat (SP-300 or SP-240) connected to an EC lab software interface. The reported cell potentials are without any iR correction. For typical long-term testing, the galvanostatic electrochemical impedance spectroscopy (GEIS) was measured every one hour during the electrolysis period.

Gas and liquid products measurements Gas products at the cathode were measured by Agilent 8890 GC equipped with a thermal conductivity detector (for H₂, N₂, CO₂, and O₂ detection) and a flame-ionization detector (for CO, CH₄, C₂H₄, and C₃H₆ detection). Ar (5N, Air Linde) was used as a carrier gas in the GC. The liquid products were measured using an Agilent liquid chromatograph (1260 Infinity II) with 0.05 M H₂SO₄ as an internal eluent. In the experiment, the volume of anolyte (0.1 M of CsOH) was ~ 40 mL except as otherwise noted. The special feature of MEA let the produced liquid products at the cathode to crossover to the anode during COE, thus liquid products were normally collected from both the cathode wash and anolyte for the FE calculation.

Simulation model of pH change by acetate formation In this sample model, we assumed a rapid response of concentration and pH change without mass transfer limitation. The initial pH of the electrolyte is 13 and the volume of the electrolyte is 50 mL. The K_a constant for acetic acid is 1.58 × 10^− 5. The total formation rate of acetate and acetic acid is ~ 4.2 × 10^− 4 mol h^− 1, which is estimated a total of ~ 5% of e¯ participated in acetate formation during the reaction (either ethanol oxidation to acetate or CO reduction to acetate). The acetate and acetic acid are considered the main formation products when pH > 7 or < 7, respectively.

Data availability

The authors declare that the data supporting the findings of this study are available within the paper and its Supplementary Information files. Raw X-ray data generated at the European Synchrotron Radiation Facility (ESRF) large-scale facility are available at https://data.esrf.fr/doi/10.15151/ESRF-ES-703258873 from 2025. Alternatively, this data can be available from the corresponding author upon request.

Acknowledgements

The research leading to these results has received funding from the ECOEthylene project from Innovation Fund Denmark (Grant# 8057-00018B), and the Villum Center for the Science of Sustainable Fuels and Chemicals grant 9455. The authors would like to thank the European Synchroton Radiation Facility (ESRF) for providing the high-energy X-ray beam and ID 31 beamline staff for experimental support.

Competing interests

The authors declare no competing interests.

Additional information

Supplementary information The online version contains supplementary material available at XXX.

Davis, S. J. et al. Net-zero emissions energy systems. Science 360, eaas9793, doi:doi:10.1126/science.aas9793 (2018).
Jouny, M., Hutchings, G. S. & Jiao, F. Carbon monoxide electroreduction as an emerging platform for carbon utilization. Nature Catal. 2, 1062–1070, doi:10.1038/s41929-019-0388-2 (2019).
Molitor, B. et al. Carbon recovery by fermentation of CO-rich off gases – Turning steel mills into biorefineries. Bioresour 215, 386–396, doi:https://doi.org/10.1016/j.biortech.2016.03.094 (2016).
Ozden, A. et al. Carbon-efficient carbon dioxide electrolysers. Nat. Sustain. 5, 563–573, doi:10.1038/s41893-022-00879-8 (2022).
Ma, M. et al. Local reaction environment for selective electroreduction of carbon monoxide. Energy Environ. Sci. 15, 2470–2478, doi:10.1039/D1EE03838A (2022).
Xu, Q. et al. Enriching Surface-Accessible CO₂ in the Zero-Gap Anion-Exchange-Membrane-Based CO₂ Electrolyzer. Angew. Chem. Int. Ed., e202214383, doi:https://doi.org/10.1002/anie.202214383 (2022).
Haas, T., Krause, R., Weber, R., Demler, M. & Schmid, G. Technical photosynthesis involving CO₂ electrolysis and fermentation. Nature Catal. 1, 32–39, doi:10.1038/s41929-017-0005-1 (2018).
Song, Y., Zhang, X., Xie, K., Wang, G. & Bao, X. High-Temperature CO₂ Electrolysis in Solid Oxide Electrolysis Cells: Developments, Challenges, and Prospects. Adv. Mater. 31, 1902033, doi:https://doi.org/10.1002/adma.201902033 (2019).
Li, C. W., Ciston, J. & Kanan, M. W. Electroreduction of carbon monoxide to liquid fuel on oxide-derived nanocrystalline copper. Nature 508, 504–507, doi:10.1038/nature13249 (2014).
Wang, L. et al. Electrochemically converting carbon monoxide to liquid fuels by directing selectivity with electrode surface area. Nature Catal. 2, 702–708, doi:10.1038/s41929-019-0301-z (2019).
Wakerley, D. et al. Gas diffusion electrodes, reactor designs and key metrics of low-temperature CO₂ electrolysers. Nature Energy 7, 130–143, doi:10.1038/s41560-021-00973-9 (2022).
Rabiee, H. et al. Gas diffusion electrodes (GDEs) for electrochemical reduction of carbon dioxide, carbon monoxide, and dinitrogen to value-added products: a review. Energy Environ. Sci. 14, 1959–2008, doi:10.1039/D0EE03756G (2021).
Jouny, M., Luc, W. & Jiao, F. High-rate electroreduction of carbon monoxide to multi-carbon products. Nature Catal. 1, 748–755, doi:10.1038/s41929-018-0133-2 (2018).
Overa, S. et al. Enhancing acetate selectivity by coupling anodic oxidation to carbon monoxide electroreduction. Nature Catal. 5, 738–745, doi:10.1038/s41929-022-00828-w (2022).
Ji, Y. et al. Selective CO-to-acetate electroreduction via intermediate adsorption tuning on ordered Cu–Pd sites. Nature Catal. 5, 251–258, doi:10.1038/s41929-022-00757-8 (2022).
Li, J. et al. Constraining CO coverage on copper promotes high-efficiency ethylene electroproduction. Nature Catal. 2, 1124–1131, doi:10.1038/s41929-019-0380-x (2019).
Li, J. et al. Copper adparticle enabled selective electrosynthesis of n-propanol. Nature Commun. 9, 4614, doi:10.1038/s41467-018-07032-0 (2018).
Zhuang, T.-T. et al. Copper nanocavities confine intermediates for efficient electrosynthesis of C3 alcohol fuels from carbon monoxide. Nature Catal. 1, 946–951, doi:10.1038/s41929-018-0168-4 (2018).
Li, J. et al. Effectively Increased Efficiency for Electroreduction of Carbon Monoxide Using Supported Polycrystalline Copper Powder Electrocatalysts. ACS Catal. 9, 4709–4718, doi:10.1021/acscatal.9b00099 (2019).
Sullivan, I. et al. A Hybrid Catalyst-Bonded Membrane Device for Electrochemical Carbon Monoxide Reduction at Different Relative Humidities. ACS Sustain. Chem. Eng. 7, 16964–16970, doi:10.1021/acssuschemeng.9b04959 (2019).
Ripatti, D. S., Veltman, T. R. & Kanan, M. W. Carbon Monoxide Gas Diffusion Electrolysis that Produces Concentrated C2 Products with High Single-Pass Conversion. Joule 3, 240–256, doi:https://doi.org/10.1016/j.joule.2018.10.007 (2019).
Luc, W. et al. Two-dimensional copper nanosheets for electrochemical reduction of carbon monoxide to acetate. Nature Catal. 2, 423–430, doi:10.1038/s41929-019-0269-8 (2019).
Pang, Y. et al. Efficient electrocatalytic conversion of carbon monoxide to propanol using fragmented copper. Nature Catal. 2, 251–258, doi:10.1038/s41929-019-0225-7 (2019).
Zhu, P. et al. Direct and continuous generation of pure acetic acid solutions via electrocatalytic carbon monoxide reduction. PNAS 118, e2010868118, doi:doi:10.1073/pnas.2010868118 (2021).
Yadegari, H. et al. Glycerol Oxidation Pairs with Carbon Monoxide Reduction for Low-Voltage Generation of C2 and C3 Product Streams. ACS Energy Lett. 6, 3538–3544, doi:10.1021/acsenergylett.1c01639 (2021).
Zhou, Y., Ganganahalli, R., Verma, S., Tan, H. R. & Yeo, B. S. Production of C3–C6 Acetate Esters via CO Electroreduction in a Membrane Electrode Assembly Cell. Angew. Chem. Int. Ed. 61, e202202859, doi:https://doi.org/10.1002/anie.202202859 (2022).
Yan, Y. et al. Interface regulation promoting carbon monoxide gas diffusion electrolysis towards C2 + products. Chem. Commun. 58, 3645–3648, doi:10.1039/D2CC00343K (2022).
Wang, X. et al. Efficient electrosynthesis of n-propanol from carbon monoxide using a Ag–Ru–Cu catalyst. Nature Energy 7, 170–176, doi:10.1038/s41560-021-00967-7 (2022).
Ji, Y., Guan, A. & Zheng, G. Copper-based catalysts for electrochemical carbon monoxide reduction. Cell Rep. Phys. Sci. 3, 101072, doi:https://doi.org/10.1016/j.xcrp.2022.101072 (2022).
Kastlunger, G. et al. Using pH Dependence to Understand Mechanisms in Electrochemical CO Reduction. ACS Catal. 12, 4344–4357, doi:10.1021/acscatal.1c05520 (2022).
Wang, L. et al. Electrochemical Carbon Monoxide Reduction on Polycrystalline Copper: Effects of Potential, Pressure, and pH on Selectivity toward Multicarbon and Oxygenated Products. ACS Catal. 8, 7445–7454, doi:10.1021/acscatal.8b01200 (2018).
Heenen, H. H. et al. The mechanism for acetate formation in electrochemical CO(2) reduction on Cu: selectivity with potential, pH, and nanostructuring. Energy Environ. Sci. 15, 3978–3990, doi:10.1039/D2EE01485H (2022).
Li, D. et al. Durability of anion exchange membrane water electrolyzers. Energy Environ. Sci. 14, 3393–3419, doi:10.1039/D0EE04086J (2021).
Xu, Q. et al. Anion Exchange Membrane Water Electrolyzer: Electrode Design, Lab-Scaled Testing System and Performance Evaluation. EnergyChem 4, 100087, doi:https://doi.org/10.1016/j.enchem.2022.100087 (2022).
Asger, M. et al. Versatile high energy X-ray transparent electrolysis cell for operando measurements. ChemRxiv., doi:10.26434/chemrxiv-2022-7v75p-v2 (2022).
Asger, M. et al. In Operando investigations of oscillatory water and carbonate effects in MEA-based CO₂ electrolysis devices.. ChemRxiv. 10.26434/chemrxiv-2022-54ccs (2022).
Wang, Z., Guo, X., Montoya, J. & Nørskov, J. K. Predicting aqueous stability of solid with computed Pourbaix diagram using SCAN functional. npj Computational Materials 6, 160, doi:10.1038/s41524-020-00430-3 (2020).
Xing, Z., Hu, L., Ripatti, D. S., Hu, X. & Feng, X. Enhancing carbon dioxide gas-diffusion electrolysis by creating a hydrophobic catalyst microenvironment. Nature Commun. 12, 136, doi:10.1038/s41467-020-20397-5 (2021).
Wu, Y. et al. Mitigating Electrolyte Flooding for Electrochemical CO₂ Reduction via Infiltration of Hydrophobic Particles in a Gas Diffusion Layer. ACS Energy Lett. 7, 2884–2892, doi:10.1021/acsenergylett.2c01555 (2022).
Yang, K., Kas, R., Smith, W. A. & Burdyny, T. Role of the Carbon-Based Gas Diffusion Layer on Flooding in a Gas Diffusion Electrode Cell for Electrochemical CO₂ Reduction. ACS Energy Lett. 6, 33–40, doi:10.1021/acsenergylett.0c02184 (2021).
Garg, S., Giron Rodriguez, C. A., Rufford, T. E., Varcoe, J. R. & Seger, B. How membrane characteristics influence the performance of CO₂ and CO electrolysis. Energy Environ. Sci. 15, 4440–4469, doi:10.1039/D2EE01818G (2022).
Vass, Á., Kormányos, A., Kószó, Z., Endrődi, B. & Janáky, C. Anode Catalysts in CO₂ Electrolysis: Challenges and Untapped Opportunities. ACS Catal. 12, 1037–1051, doi:10.1021/acscatal.1c04978 (2022).
Xu, Q. et al. Fluorination-enabled Reconstruction of NiFe Electrocatalysts for Efficient Water Oxidation. Nano Lett. 21, 492–499, doi:10.1021/acs.nanolett.0c03950 (2021).
Lai, S. C. S. et al. Effects of electrolyte pH and composition on the ethanol electro-oxidation reaction. Catalysis Today 154, 92–104, doi:https://doi.org/10.1016/j.cattod.2010.01.060 (2010).
Monyoncho, E. A., Woo, T. K. & Baranova, E. A. in Electrochemistry: Volume 15 Vol. 15 1–57 (The Royal Society of Chemistry, 2019).
Guo, J., Chen, R., Zhu, F.-C., Sun, S.-G. & Villullas, H. M. New understandings of ethanol oxidation reaction mechanism on Pd/C and Pd₂Ru/C catalysts in alkaline direct ethanol fuel cells. Appl. Catal. B 224, 602–611, doi:https://doi.org/10.1016/j.apcatb.2017.10.037 (2018).
Hasa, B. et al. Benchmarking anion-exchange membranes for electrocatalytic carbon monoxide reduction. Chem Catalysis, 100450, doi:https://doi.org/10.1016/j.checat.2022.10.026 (2022).
Larrazábal, G. O., Okatenko, V., Chorkendorff, I., Buonsanti, R. & Seger, B. Investigation of Ethylene and Propylene Production from CO2 Reduction over Copper Nanocubes in an MEA-Type Electrolyzer. ACS Applied Materials & Interfaces 14, 7779–7787, doi:10.1021/acsami.1c18856 (2022).
Ashiotis, G. et al. The fast azimuthal integration Python library: pyFAI. J Appl Crystallogr 48, 510–519, doi:10.1107/S1600576715004306 (2015).

There is NO Competing Interest.

SupportingInformation.docx
Supporting Information

Download PDF

Journal Publication

published 28 Sep, 2023

Read the published version in Nature Catalysis →

Version 1

posted

You are reading this latest preprint version

Identifying and alleviating the durability challenges in membrane-electrode-assembly devices for high-rate CO electrolysis

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

Identifying MEA degradation behavior

Alleviating the MEA degradation

Anolyte influenced by anodic oxidation

Selectivity changes during long-term COE

Conclusions

Methods

Declarations

Data availability

Acknowledgements

Competing interests

Additional information

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1