Diagnosing Stagnant Gas Bubbles in a Polymer Electrolyte Membrane Water Electrolyser Using Acoustic Emission

The use of acoustic emission as a low-cost, non-destructive, and operando diagnostic tool has been demonstrated for a range of electrochemical energy conversion and storage devices, including polymer electrolyte membrane water electrolysers (PEMWEs) and fuel cells. In this work, an abrupt change in acoustic regime is observed during operation of a PEMWE as the current density is increased from 0.5 to 1.0 A cm−2. This regime change is marked by a sudden drop in the number of acoustic hits, while hit duration, amplitude, and energy increase significantly. It is found that the change in acoustic regime coincides with a significant extension of the stagnant bubble region in the flow channels of the PEMWE, observed with high-speed optical imaging. These results demonstrate that acoustic emission can be used effectively as an operando diagnostic tool to monitor bubble formation (two-phase flow conditions) in PEMWEs, facilitating rapid testing or prototyping, and contributing to operational safety.


Introduction
PEMWEs represent one of the most promising approaches to the production of green hydrogen and large-scale energy grid stabilization. The technology is likely to replace the widely commercially used alkaline electrolysis due to advantages including lower voltage at equal current density, lower gas crossover, compact build, and the possibility of high-pressure operation [1]. While currently more expensive than the alternative alkaline technology, the capital cost of a typical PEMWE system is dropping [2] and plants rated up to 6 MW are in operation [3].
Water flooding has been shown to be a major mass transport limitation occurring in polymer electrolyte membrane fuel cells (PEMFCs) at high current densities, particularly at high humidity when water condenses at the cathode forming droplets which coalesce. This leads to water blocking the flow channels and occupying the gas diffusion layer, causing a consequent increase in pressure drop and decrease in performance [4 6]. Similarly to water accumulation in a PEMFC, which eventually leads to flooding, product gas can accumulate in polymer electrolyte membrane water electrolysers (PEMWEs) leading to bubbles blocking the flow channels. This occurs if the gas production from the catalyst sites exceeds the gas removal capacity of the flow channels, which is mainly determined by the cross-sectional area and the flow rate of water through the channels. The effects of bubble blockage on performance, pressure drop, and life-time of a PEMWE have not yet been investigated, but it is expected that prolonged bubble blockage results in local water starvation, causing a non-uniform current distribution over the active area and a decrease in performance [7].
Acoustic emission (AE) is a non-destructive, operando diagnostic tool traditionally used in civil engineering, e.g. for monitoring crack propagation in steel [8] or the stability of bridges [9]. It uses a piezoelectric sensor to detect mechanical perturbations emitted by an object and has been applied to a range of electrochemical energy storage devices. It has been used to limitation occurring at high current densities at high current densities, particularly , particularly es at the cathode forming es at the cathode forming droplets which coalesce. This leads to droplets which coalesce. This leads to water blocking the flow channels and oc water blocking the flow channels and occupying the ga cupying the gas diffusion layer, s diffusion layer, increase in pressure drop and increase in pressure drop and decrease in performance decrease in performance in a PEMFC in a PEMFC, , which which eventually lead eventually lead monitor particle fracture and morphological changes in battery electrodes during charge and discharge [10 12], has been found to be sensitive to Li-ion intercalation and formation of the solid electrolyte interphase [13,14] in Li ion batteries, and has also been applied to PEMFCs [15 17].
Two-phase systems, such as the water-gas mixture in the flow channels of the PEMWE analysed in this work, are also readily analysed using acoustic emission. This includes the calculation of bubble size distribution [18], recognition of different flow patterns by analysing acoustic emission data with neural networks [19], and observing the formation and collapse of single bubbles [20]. Hence, acoustic emission is a valuable alternative diagnostic tool to other techniques for the investigation of two-phase dynamics [21 23].
In previous work, the authors demonstrated the ability of acoustic emission to detect changes in the number and size of bubbles passing through the flow channel of a PEMWE. This enabled the prediction of the change from bubbly to slug flow and showed that acoustic emission is a valuable operando tool for PEMWE diagnosis [24]. Here, we demonstrate that the acoustic emission signal changes dramatically when, rather than normal two-phase flow, stagnant bubbles are located within the vicinity of the acoustic emission sensor. This feature can be used to detect and locate bubble blockage in PEMWEs, for operando monitoring or design optimization. Booster; Gamry Instruments, USA). s acquisition). The transparent end-plates allowed for direct optical access to the flow-field [25,26].

Results and Discussion
The bubble evolution as a function of the current density has been captured with high-speed imaging experiments, which are shown as a function of increasing current density ( Figure 3).  (Figure 3 (b)). At 1.0 A cm -2 (Figure 3 (c)), bubble blockage covers more than a quarter of the channel length.
The acoustic emission parameters are strongly influenced by the current density ( Figure 4). As Further, the average hit amplitude (Figure 4 (b)) increases steeply by around 50 % between 0.5 A cm -2 and 1.0 A cm -2 , the same range within which the hit rate drops. The average hit duration increases from less than 0.1 ms to the cut off value of 1.0 s mentioned above (Figure 4 (c)). For current densities above 1.0 A cm -2 , a constant signal is detected, indicating permanent contact The limit of detection is an artefact of the data acquisition, which cuts off any hit longer than 1.0 s, which means that from 1.0 A cm cuts off any hit longer than 1.0 s, which means that from 1.0 A cm -2 onwards the AE signal onwards the AE signal continuously exceeds the noise threshold, with no individual acoustic hits discernible. continuously exceeds the noise threshold, with no individual acoustic hits discernible.
significant decrease significant decrease of acoustic hits of acoustic hits highlights highlights a a dramatic change of two dramatic change of two flow channels flow channels. T . The relationship between the he relationship between the number of acoustic hits and the number of bubbles passing thro passing through the flow channels ugh the flow channels between a bubble and the end-plate. Finally, an increase in hit duration and amplitude causes an increase in acoustic energy (Figure 4 (d)). All these changes occur in a step-like manner between 0.5 A cm -2 and 1.0 A cm -2 .
The decreasing number of hits, while hit amplitude and contact time between bubble and endplate increase, all suggest that the signal change is caused by the extension of the stagnant bubble region towards the sensor location in the current density range between 0.5 A cm -2 and 1.0 A cm -2 ( Figure 4). This is supported by the extension of the stagnant bubble region ( Figure   3) observed via high-speed imaging, a major part of which occurs between 0.6 A cm -2 and 1.0 A cm -2 .

Conclusion
Acoustic emission has been demonstrated as a useful technique for operando diagnosis of flow channel containing stagnant bubbles was found to increase with current density, flow channel containing stagnant bubbles was found to increase with current density, expected that acoustic emission can be used to detect local bubble blockage and insufficient water supply in specific areas of a PEMWE.
The use of this operando diagnostic tool has successfully been applied to a PEMWE, but could be extended to other applications. The accumulation of gas within a system or plant can cause inefficiencies or pose a hazard in many areas of chemical production and transport. Moreover, it has been shown that the change of two-phase flow regime influences the pressure drop [27,28]. Hence, the technique presented in this work could be deployed to screen various flowfield configurations or monitor safe limits of operation, replacing less cost-effective or accessible diagnostic tools such as neutron or X-ray imaging [29 31].  hit is initiated when the signal exceeds the noise threshold and ends when the signal falls back below the threshold. The hit amplitude is the intensity of the most prominent peak within the hit, and its energy is the integrated area of the hit (adapted from [32]).