Bubble Formation in the Electrolyte Triggers Voltage Instability in CO2 Electrolyzers.

The electrochemical reduction of CO2 is promising for mitigating anthropogenic greenhouse gas emissions; however, voltage instabilities currently inhibit reaching high current densities that are prerequisite for commercialization. Here, for the first time, we elucidate that product gaseous bubble accumulation on the electrode/electrolyte interface is the direct cause of the voltage instability in CO2 electrolyzers. Although bubble formation in water electrolyzers has been extensively studied, we identified that voltage instability caused by bubble formation is unique to CO2 electrolyzers. The appearance of syngas bubbles within the electrolyte at the gas diffusion electrode (GDE)-electrolyte chamber interface (i.e. ∼10% bubble coverage of the GDE surface) was accompanied by voltage oscillations of 60 mV. The presence of syngas in the electrolyte chamber physically inhibited two-phase reaction interfaces, thereby resulting in unstable cell performance. The strategic incorporation of our insights on bubble growth behavior and voltage instability is vital for designing commercially relevant CO2 electrolyzers.

Introduction

Global CO2 emissions from fossil fuel consumption continue to grow despite significant advancements in clean energy technologies (Peters et al., 2020). The demand for electricity far exceeds the rate of clean energy technology implementation, consequently leading to a net increase in fossil-fuel-sourced CO2 emissions (York, 2012). Clean energy technologies that lessen societal dependence on fossil fuels are needed to accelerate the necessary transition from a carbon-based energy infrastructure to a clean-energy-based infrastructure.

The electrochemical conversion of CO2 to useful chemical products (e.g. CO and HCOOH) is a promising means to realize this transition to a carbon neutral energy infrastructure (Bushuyev et al., 2018, Smith et al., 2019). By utilizing fuels generated via CO2 electrolyzers that are powered by renewable energy, carbon neutral transportation can be achieved (Smith et al., 2019). In recent years, there has been a significant growth in research activities focused on developing next-generation catalysts and membranes for enhanced selectivity toward carbon-based chemical products for CO2 reduction (Dinh et al., 2018, Gallo et al., 2019, García de Arquer et al., 2020, Li et al., 2020, Verma et al., 2018, Vermaas and Smith, 2016, Zheng et al., 2019). Although these pioneering works have illustrated the promise of CO2 electrolysis, there are major challenges associated with commercially relevant operating conditions, such as achieving high current densities (i.e. i> 200 mA/cm2), which have remained largely overlooked (Burdyny and Smith, 2019, Smith et al., 2019, Weekes et al., 2018).

The operation of CO2 electrolyzers at elevated current densities has been accompanied by unfavorable conditions, such as increased electrolyte pH near the catalyst surface (Burdyny and Smith, 2019, Ma et al., 2020, Singh et al., 2017) and unstable cathodic overvoltage (Nwabara et al., 2019, Verma et al., 2018). In particular, previous studies have shown that unstable cathodic overvoltages directly limit the high current density operation of CO2 electrolyzers (Gabardo et al., 2019, Kuhl et al., 2012, Sen et al., 2019), yet the direct cause of this instability remains unclear. One proposed cause is salt precipitation within the liquid electrolyte adjacent to the membrane and gas diffusion electrode (GDE) (Dufek et al., 2012, Schulz et al., 2006). Recently, Nwabara et al. (Nwabara et al., 2019) demonstrated that the increase in cell voltage due to salt precipitation is partially recoverable with a flowing electrolyte. Another proposed cause of this voltage instability is the formation of gaseous species within the electrolyte, which physically blocks the reaction sites (Kuhl et al., 2012, Nwabara et al., 2019, Sen et al., 2019). When the GDE is not sufficiently porous for product gas diffusion, this product gas may accumulate within the electrolyte chamber (Nwabara et al., 2019). Although bubble-induced electrochemical losses have been identified by previous works (Angulo et al., 2020), there is a complete absence in the literature of any direct evidence of this gas accumulation and its effect on the cell voltage. Understanding the complex two-phase transport within the electrolyte chamber is a necessary first step in designing next-generation CO2 electrolyzers capable of operating at commercially relevant conditions.

Here, we directly visualized gas bubble formation within the electrolyte chamber via in operando synchrotron X-ray imaging with high spatial (6.5 um/pixel) and temporal (1 s per frame) resolution imaging (Lee et al., 2020). We performed the experiments at the Biomedical Imaging and Therapy Wiggler Insertion Device beamline at the Canadian Light Source (CLS) in Saskatoon, Canada (Wysokinski et al., 2013). The CO2 electrolyzer was custom designed to constrain the X-ray attenuation to the active area of the cell. A flow cell configuration with a liquid electrolyte chamber was used in this work. Although liquid electrolyte chambers have been used in similar cell configurations by other authors (Delacourt et al., 2008, Gabardo et al., 2019, Liu et al., 2019), a major distinguishing feature of our unique design is that our electrolyte chamber was packed with high-density polyethylene (HDPE) meshes to provide structural support to the proton exchange membrane (PEM) and the GDE for high-resolution synchrotron X-ray imaging (see Figure 1). Stabilizing the membrane is based on our previous imaging work of electrochemical cell designs (Banerjee et al., 2018, Chevalier et al., 2017, Ge et al., 2019). We supplied deionized water at 10 mL/min to the anode flow field, aqueous KHCO3 (0.5 M) at 6 mL/min to the electrolyte chamber, and dry CO2 gas at 100 mL/min to the cathode flow field. The GDE was coated with a silver catalyst, which was chosen for its high selectivity toward CO generation (Martín et al., 2015). For details, the readers are referred to the Methods section.

Figure 1

Schematic of the CO2 Electrolyzer Cell Used in This Work

The electrolyte chamber was packed with high-density polyethylene (HDPE) meshes to provide structural support to the proton exchange membrane (PEM) and the silver-based gas diffusion electrode (GDE). Gas bubbles (white) are shown within the electrolyte chamber. The red-dashed box indicates the region of interest that was captured via synchrotron X-ray imaging.

Results and Discussion

Characteristics of Bubble Formation within the Electrolyte Chamber

The results of imaging the electrolyte chamber are presented in Figure 2. We assumed two gases that were formed within the electrolyte chamber: (1) CO2, which preferentially formed near the PEM-electrolyte chamber interface and (2) a mixture of CO and H2 (i.e. syngas), which preferentially formed near the electrolyte chamber-GDE interface. The presence of these gases within the electrolyte chamber can be validated via performing gas chromatography (GC) on the gaseous products exiting the electrolyte chamber. At the PEM-electrolyte chamber interface (the upper region of each image in Figure 2), the following reaction drove the production of CO2 gas:

Figure 2

Processed Images of the Electrolyte Chamber

Images are taken at 1 frame per second during steady-state operation of the CO2 electrolyzer at current densities of (A) 10 mA/cm2, (B) 20 mA/cm2, (C) 40 mA/cm2, and (D) 60 mA/cm2. The images correspond to the spatial distribution of gas saturation (averaged over the last 100 frames of each experiment) within the electrolyte chamber. The upper region of each image corresponds to the interface between the electrolyte chamber and the PEM, whereas the lower region corresponds to the interface between the electrolyte chamber and the GDE. As a result, the gas clusters in the upper region corresponded to CO2 gas, whereas the gas clusters in the lower region corresponded to syngas.

At the GDE-electrolyte chamber interface (the lower region of each image in Figure 2), the CO2 reduction reaction (CO2RR) and hydrogen evolution reaction (HER) drove the formation of syngas. Surprisingly, we observed a contrasting trend between the syngas accumulation behavior near the GDE-electrolyte chamber interface and CO2 accumulation behavior near the PEM-electrolyte chamber interface. Specifically, syngas was not observed in the electrolyte chamber until a critical current density of i = 40 mA/cm2 was reached (Figure 2C), in contrast to the appearance of CO2 gas in the electrolyte chamber, which was visible when i = 10 mA/cm2 (Figure 2A).

To obtain further insight into the product gas accumulation behavior, we divided the electrolyte chamber into three regions of equal thickness: PEM interface region, bulk region, and GDE interface region (Figure 3A). As current density was increased, the gas saturation in the PEM interface region decreased, whereas the gas saturation near the GDE interface region sharply increased (Figure 3B). Specifically, the gas saturation in the PEM interface region decreased from 0.255 (at i = 20 mA/cm2) to 0.193 (at i = 60 mA/cm2), whereas the saturation in the GDE interface region increased from 0.009 (at i = 20 mA/cm2) to 0.217 (at i = 60 mA/cm2). The decrease in gas accumulation in the PEM interface region at higher current densities was attributed to more frequent bubble detachment from the PEM interface region (Figure 4A). We observed less frequent bubble detachment at lower current densities (∼0.07 s−1 at i = 20 mA/cm2, Figure 4A and see also Video S2) compared with the frequency of bubble detachment in the PEM interface region at high current densities (∼0.27 s−1 at i = 60 mA/cm2, Figure 4A and see also Video S4). More frequent removal of gas from the PEM interface region led to lower average gas saturation at the PEM interface region.

Figure 3

Average Gas Saturation in Three Distinct Regions of the Electrolyte Chamber

(A) Three regions in the electrolyte chamber were defined to calculate the local average gas saturation. The electrolyte chamber was divided into regions of equal thickness: the PEM Interface Region refers to the upper region immediately adjacent to the PEM, the Bulk Region refers to the region in the center of the electrolyte chamber, and the GDE Interface Region refers to the lower region immediately adjacent to the GDE.

(B) The resulting average gas saturation in each region. The error bars show one standard deviation of 100 images that were averaged for each data point. We observed a decrease in average gas saturation in the PEM interface region and a sharp increase in the GDE interface region with increasing current density.

Figure 4

Evolution in Local Gas Saturations

Gas evolution within (A) the PEM interface region and (B) the GDE interface region. The gas evolution behavior in the PEM interface region exhibited sequence of growth and detachment, whereas that in the GDE interface region exhibited a smooth and relatively slower growth.

Video S1. Imaging Results Showing the Evolution in Gas Distribution in the Electrolyte Chamber When i = 10 mA/cm2, Related to Figure 2A

The average of the final 100 s is shown in Figure 2A.

Video S2. Imaging Results Showing the Evolution in Gas Distribution in the Electrolyte Chamber When i = 20 mA/cm2, Related to Figure 2B

The average of the final 100 s is shown in Figure 2B.

Video S3. Imaging Results Showing the Evolution in Gas Distribution in the Electrolyte Chamber When i = 40 mA/cm2, Related to Figure 2C

The averaged of the final 100 s is shown in Figure 2C.

Video S4. Imaging Results Showing the Evolution in Gas Distribution in the Electrolyte Chamber When i = 60 mA/cm2, Related to Figure 2D

The average of the final 100 s is shown in Figure 2D.

The sharp increase in the gas saturation (increase of 0.208) at the GDE interface region with increasing current density was attributed to the porous nature of the GDE surface. Product syngas entered the electrolyte chamber and the GDE (depending on the porous structure of the GDE (Nwabara et al., 2019)). Thus, we hypothesized that at low current densities (i.e. i ≤ 20 mA/cm2), the majority of the generated syngas diffused into the GDE, and a further increase in current density promoted bubble formation in the electrolyte chamber. We further supported our hypothesis by comparing the trends in syngas evolution in the GDE interface region (Figure 4B) with that in the CO2 evolution in the PEM interface region (Figure 4A). The gas generated within the electrolyte layer in the GDE interface region exhibited a unique behavior compared with the gas generated in the PEM interface region (Figures 4A and 4B and see also Videos S1, S2, S3, and S4). For the CO2 gas that accumulated in the PEM interface region, we observed a sequence of bubble growth and detachment (Figure 4A). On the other hand, the accumulation of syngas in the GDE interface region exhibited a relatively smooth profile (Figure 4B), implying that syngas entered not only the electrolyte chamber but also the porous GDE. The majority of the syngas exited via the porous GDE; therefore, the volumetric growth rate of syngas bubbles in the electrolyte chamber was relatively slow compared with the CO2 bubbles in the PEM interface region.

Voltage Instability Caused by Bubble Formation

We observed a strong correlation between the bubble accumulation behavior at the electrolyte-GDE interface and the stability of the cell voltage. Specifically, unstable cell voltages were observed with the simultaneous accumulation of syngas in the GDE region of the electrolyte at a critical current density (i.e. i = 40 mA/cm2) (Figure 5A). The standard deviations in the cell voltage were 0.00, 0.00, 0.01, 0.06, 0.22, and 0.38 V for current densities of 5, 10, 20, 40, 60, and 80 mA/cm2 (shown in Figure 5A). CO2 bubbles were detected in the PEM interface region when i ≤ 20 mA/cm2, but this gas accumulation did not influence the voltage stability. The dynamic CO2 bubble generation in the PEM region may have influenced the overall two-phase transport behavior within the electrolyte chamber; however, the presence of syngas bubbles in the GDE region exhibited the most significant impact on unstable cell voltages.

Figure 5

Cell Voltage Instability Caused by Gas Accumulation in the GDE Interface Region

(A) Transient response in cell voltage with increasing current density, indicated by the arrow. The cell voltage was measured every second, and we present the last 300 s of each current density operation.

(B and C) Schematics of the catalyst layer reaction sites (B) in the absence of gas and (C) with the presence of gas. The presence of gas physically impeded the two-phase reaction interface (i.e. interface between the liquid electrolyte with dissolved CO2 and catalyst particles), resulting in a temporary decrease in cell voltage.

The strong correlation between bubble formation on the GDE-electrolyte chamber interface and the cell voltage stability was attributed to the physical break of the two-phase reaction interface (Figures 5B and 5C). The reaction sites for syngas generation exist at the interface between the catalyst particles and the liquid electrolyte with dissolved CO2 as the main reagent (the interface is highlighted in green in Figure 5B) (Burdyny and Smith, 2019). However, as syngas bubbles formed on the GDE-electrolyte chamber interface, the two-phase reaction boundaries became physically separated in the presence of syngas bubbles, temporarily reducing the total active reaction site area (Figure 5C). The relatively slow growth and detachment of syngas bubbles led to undesired reaction site blockage, which manifested as an unstable cell voltage.

Conclusion

Here, we directly observed unique bubble accumulation behavior within the liquid electrolyte chamber of a flow-cell-based CO2 electrolyzer, and we explained how this behavior leads to undesired cell voltage instabilities. CO2 and syngas bubbles accumulated in the PEM interface region and the GDE interface region, respectively, and these regions exhibited contrasting growth and accumulation behavior. Specifically, CO2 bubbles exhibited a two-stage (i.e. growth and detachment) behavior with a frequency << 1 s−1 from the onset current density, whereas significantly less frequent syngas bubble growth and detachment events were observed, only appearing when i ≥ 40 mA/cm2. The appearance of these syngas bubbles was accompanied by an undesired fluctuation in cell voltage, where the standard deviation suddenly escalated up to 60 mV at i = 40 mA/cm2. We attributed this instability in cell voltage to the temporary blockage of the two-phase reaction interface between the catalyst and the CO2-dissolved electrolyte. Our results inform the importance of mitigating bubble coverage of the GDE-electrolyte chamber interface, which is vital for achieving commercially relevant operating conditions of CO2 electrolyzers. Therefore, we recommend future work to mitigate bubble accumulation in the electrolyte chamber: (1) design new porous GDEs or electrolyte chambers that promote product transport into the GDE and (2) tailor the liquid electrolyte and gaseous CO2 flow rates to maximize the two-phase reaction interface areas.

Limitations of the Study

This work presents experimental evidence of the severe effects of bubble formation within the electrolyte chamber on cell voltage, and further investigations may be undertaken to extend our results to different experimental set-ups. Specifically, the addition of HDPE meshes into the electrolyte chamber was primarily for preventing the movement of the GDE and PEM during imaging, but these meshes may potentially encourage gas accumulation near the GDE interface region. Additionally, the type of GDE used in this work, although commercial, may have promoted bubble formation in the electrolyte chamber rather than product transport toward the cathode flow field due to its porous microstructure. Additional experiments with varied GDE structures are crucial for elucidating optimal GDE structures for enhanced transport. Lastly, the CO2 gas flow rate and electrolyte flow rate were held constant throughout the experiment; varying these parameters is also necessary for identifying the dependence of the bubble effects on the prescribed flow conditions.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.