Fuel Cell Catalyst Layer Evaluation using a Gas Diffusion Electrode Half-Cell: Oxygen Reduction Reaction on Fe-N-C in Alkaline Media

Anion exchange membrane fuel cells (AEMFC) are a promising technology to allow the application of non-precious metal catalysts. While many of such catalysts have been identified in numerous recent fundamental research studies, reports evaluating these catalysts in realistic AEMFC catalyst layers together with stability assessments are rare. In the present work we show that fast and reliable evaluation and optimization of Fe-N-C-based oxygen reduction reaction (ORR) catalyst layers can be achieved using a gas diffusion electrode (GDE) half-cell approach. To set a benchmark in such measurements, a commercial Pajarito Powder Fe-N-C catalyst and commercial Aemion TM ionomer are used. It is demonstrated that the ORR performance can be increased significantly by fine-tuning of the ionomer activation time. Furthermore, the optimized Fe-N-C-based catalyst layer shows very high stability with no observable performance deterioration after 5000 cycles in the 0.6 (cid:177) 1.0 V vs. RHE potential window.


Introduction
After decades of optimization, proton exchange membrane fuel cells (PEMFC) with Pt-group metal (PGM) catalysts have reached commercialization level. [1] However, the price for those devices is still high. The membrane electrode assembly (MEA) with PGM-based catalyst layers accounts for approximately one third of the overall device costs. [2] Therefore, the search for efficient, durable and cheap electrocatalysts and their implementation in real devices are two of the main topics in fuel cell research. [3] Recently, catalysts synthesized from earth-abundant, inexpensive and easily exploited materials as iron, nitrogen, and carbon (Fe-N-C) exhibit promising oxygen reduction reaction (ORR) activities, which opens new opportunities to significantly reduce the overall fuel cell technology costs. [4] Still, for this kind of catalyst the acidic environment of conventional PEMFCs imposes stability challenges, mainly due to Fe dissolution and carbon corrosion. [5][6][7] When operated in an alkaline environment instead, the durability of non-PGM materials can be improved significantly. [8] Additionally, the intrinsic activity of those materials can be slightly increased at high pH values. [9] Therefore, anion exchange membrane fuel cells (AEMFC) could facilitate the widespread application of non-PGM catalysts and hence significantly reduce overall fuel cell device costs. Recently tremendous progress in AEMFC development has led to performances exceeding power densities of PEMFCs using non-PGM catalyst materials. [10] This progress mainly arises from optimized mass transport of water, OHand the reactant gases, which is influenced by (i) membrane and ionomer materials, (ii) catalyst layer design and (iii) operating conditions. For more details, the readers are referred to excellent reviews in the field. [10][11][12][13]  of ionomer, ionomer content and distribution and manufacturing condition) on the morphology and transport behaviour in the electrode and therefore the single cell performance. [17][18][19] Generally it has to be pointed out, that a fuel cell electrode is a complex structure including catalyst, ionomer and void space, where each one of these plays a crucial role for electrode function. [12] This can only be investigated insufficiently with TF-RDE experiments. Hence MEA experiments are widely used to study catalyst layer effects on the fuel cell performance.
But MEA experiments also introduce major drawbacks as they are (i) time and material intensive, (ii) do not allow direct insights into one single electrode (usually lack of a reference electrode) and (iii) are often difficult to compare due to deficiency of standardized test protocols and conditions for the evaluation of non-PGM catalysts. To bridge the gap between RDE and MEA experiments and to combine their advantages, recently different groups have conducted experiments with electrochemical half-cells using gas diffusion electrodes (GDE) as working electrodes. [20][21][22] Those GDE experiments allow fast and comparable testing at standardized operating conditions and provide dedicated insights into one single electrode with realistic catalyst layer parameters at relevant potential and current ranges. Therefore, this method can be an optimal supplement to drastically shorten the time from successful catalyst synthesis to an operating electrode for fuel cell applications.
In this communication, we present the first implementation of this novel GDE method for non-PGM catalysts in alkaline environment. As the ORR is still the limiting half-cell reaction [12], we focus on cathode catalyst layers and will address anode catalyst layer optimization in further work.

Catalyst ink and electrode preparation
The PGM-free high pH GDE was fabricated from an ink comprised of a total of 20 wt% solids

GDE half-cell measurements
For the electrochemical measurements the GDE setup, electrochemical components and protocols presented in a previous work [22] were used. For more detailed information, see also

Results and Discussion
As described above, GDE half-cell experiments can be used to investigate and optimize single catalyst layer parameters in a fast and dedicated way. One of those parameters is the preconditioning or activation of the ionomer used in the catalyst layer. As the ionomer is present in the iodide form, an exchange of those iodide ions with hydroxide ions is required to obtain a hydroxide conductive polymer. [ [28,29]. However, the impact of immersion time for activation of alkaline electrodes has never been compared. conductivity. Surprisingly a drastic decrease in mass transport properties (but not intrinsic activity, see low-current region in Figure 1B) can be observed after 72 h of pre-treatment in 1 M KOH. This is probably due to catalyst or ionomer degradation in aqueous alkaline environment.
It has to be noted that this activation data is very specific to the alkaline ionomer which is used for GDE manufacturing. But the example shows how GDE experiments can be used to drastically accelerate electrode evaluation and activation. Whereas one single cell experiment almost takes one day of experiments, due to sophisticated MEA fabrication in alkaline, assembling, heating up, cooling down and disassembling of the cell, all the experiments for this activation study with repetitions could be conducted in less than four working days.
Additionally, in half-cell experiments, factors such as anodic reaction limitations and carbonation of the alkaline ionomers [25,30] can be excluded and therefore, the effect of one single parameter can be investigated exclusively.
After the impact of ionomer activation was investigated, the durability of the GDEs in alkaline environment as well as a possible impact of carbonation should be assessed. The carbonation of ionomers can lead to significant problems in AEMFC, mainly due to changes in conductivity behaviour and therefore mass transport limitations, but also electrocatalytic effects. [25,30] Although carbonation of the ionomer mainly occurs when operating AEMFC with ambient air, it can already take place during the manufacturing process or sample storage as the exchange from OHto HCO3 -/CO3 2occurs very fast. Nevertheless, it has been found out, that at high current densities all HCO3 -/CO3 2are exchanged by OH -- [30,31] Therefore, the samples have first been tested towards their ORR activity up to 2 A cm -2 three times consecutively to check for any effect of ionomer carbonation on ORR activity determination. Subsequently they have been exposed to a load cycling protocol comprising 5000 triangular cycles between 0.6 and 1.0 V vs. RHE performed at a scan rate of 100 mV s 1 in Ar-saturated electrolyte as used in previous degradation studies for non-PGM catalysts. [8] In Figure  A catalyst system should always be compared to recent state-of-the-art developments to provide a comprehensive evaluation to the readers. Unfortunately, this is not always done appropriately, which can lead to difficulties in comparing results between different research groups. In In previous AEMFC experiments with the commercial Fe-N-C catalyst from Pajarito Powder (empty squares), Hossen et al. [26] have been able to achieve current densities up to approximately 700 mA cm -2 . In comparison with our present results with the same catalyst it can be seen that the onset of the polarization curve in the activation region is in a comparable operating conditions in AEMFC testing (60 °C, 1.4 bar) and higher catalyst loading (3.5 vs. 1.6 mgCat cm -2 ). However, when exceeding low current densities severe reaction limitations can be observed in the high current region. There can be multiple explanations for that: (i) catalytic limitations at the Pt/C anode, (ii) mass transportfactors plays a role in limiting the overall reaction. But in the conducted single cell experiment it is difficult to reveal which effect is the most severe, stressing again the advantage of the presented GDE half-cell method. The benchmark for Fe-N-C catalysts in alkaline environment was just recently set by Firouzjaie et al. [10] using a [8]. The better performance of their catalyst system in comparison with the one presented in the present work can be explained by: (i) their use of a catalyst tuned for alkaline environment, instead of a commercial catalyst developed especially for acidic environment; (ii) improved massmethods to drastically improve AEMFC performance through dedicated catalyst layer design. [10,12] This is not only crucial working with non-PGM catalysts, but also in AEMFCs using PGM catalysts, confirmed by the fact that also the benchmark PGM system in AEMFC has been developed by the same group. [32]

Conclusions and Outlook
AEMFCs are a promising technology to allow the application of non-PGM catalysts to fuel cells due to their improved stability in alkaline media. and distribution, and the morphology of the catalyst layer. Further work will be dedicated to understanding the impact of those parameters on both anode and cathode performance. Also the effect of carbonation of the ionomer will be addressed by using different CO2 saturation in the aqueous electrolyte. Figure 1: Comparison of polarization curves of alkaline Fe-N-C cathodes (Pajarito Powder, 1.5 1.7 mgcat cm -2 ) with Aemion TM ionomer and different pre-treatment procedures applied. Tested in 1.0 M KOH. A: Cyclic voltammograms (100 mV s -1 ) with Ar. B&C: Tafel plot and polarization curve for the oxygen reduction reaction with pure O2.