Role of Fe in the hydrogen oxidation reaction in a NiFe-based catalyst: an in situ Mössbauer spectroscopic investigation

Nickel-based catalysts reach a high activity for the hydrogen oxidation reaction (HOR) in anion exchange membrane fuel cells. While incorporation of iron significantly decreases the HOR overpotential on NiFe-based catalysts, the reason for the enhanced activity remains only partially understood. For the first time, in situ 57Fe Mössbauer spectroscopy is used to gain insights into the iron-related composition at different potentials. The aim is to evaluate which changes occur on iron at potentials relevant for the HOR on the active Ni sites. It is found that different pre-conditionings at low potentials stabilize the iron at a low oxidation state as compared to the as-prepared catalyst powder. It is likely that the lower average oxidation state enables a higher exchange current density and a more efficient OH adsorption, which make the Volmer step much faster in the HOR. Insights from in situ Mössbauer spectroscopy enlighten the role of iron in the nickel-iron catalyst, paving the way for developing improved Ni-based catalysts for HOR catalysis.


Introduction
For a sustainable future, it is required to move towards zero-emission technologies for energy supply. The anion-exchange membrane fuel cell (AEMFC) technology enables the conversion of chemical energy stored in hydrogen into electric energy with no emissions other than pure water. AEMFCs offer the possibility to use platinum group metal (PGM)-free catalysts for both reactions, the hydrogen oxidation reaction (HOR) [1][2][3][4] and the oxygen reduction reaction (ORR) [5][6][7][8][9][10]. Recent comprehensive reviews and studies further highlight the HOR challenges in alkaline pH [11][12][13][14]. For instance, Davydova et al critically reviewed the electrocatalysts for the HOR in alkaline electrolytes and underlined the major technological barrier to the development of completely PGM-free AEMFCs [11]. However, replacing PGM catalysts at the anode is highly challenging due to the two orders of magnitude lower kinetics of HOR in alkaline pH compared to acidic pH, even on PGM-based catalysts [15]. To make AEMFC a feasible and practical technology, PGM-free catalysts with reasonable activity and stability should be used. Ni-based catalysts are currently considered the only PGM-free catalyst to catalyze the challenging HOR hitherto [3,11,[16][17][18][19][20][21][22][23].
In our previous works [11,[17][18][19], we have shown that by alloying the nickel with an oxyphilic metal, the hydrogen and hydroxide binding energies can be optimized to enable higher HOR activity. Particularly for Ni 3 Fe, a higher OH surface coverage was found [17]. It is assumed that Ni is the active site for the HOR, while iron promotes its redox characteristics by optimized hydrogen and hydroxide binding energies. It is worth mentioning that bare Fe has negligible HOR activity in alkaline electrolytes and, therefore, likely acts as a promoter of the reaction. While these findings were promising, it was observed that the catalyst is only active at small overpotentials due to the thermodynamic stability window of nickel [19]. As a consequence, at ca. 0.3 V above E 0 , the activity dropped down. The question arises to what extent this is associated with changes in the composition associated with iron. To gain a better understanding of structural and electronic changes, Mössbauer spectroscopy can be a useful technique. 57 Fe Mössbauer spectroscopy is particularly well suited, as iron speciations and possible electronic changes can be distinguished [24]. For enriched electrodes, measurements under in situ or even operando conditions are possible [25][26][27][28][29][30]. In situ refers to the effect of the applied potential, whereas operando spectroscopy addresses the catalyst under catalytically relevant conditions. While in situ and operando were applied for ORR catalysts [25][26][27] and oxygen evolution reaction catalysts [31], to the best of our knowledge, no in situ Mössbauer spectroscopic measurements were yet applied to NiFe catalysts for the HOR. In this work, we performed Mössbauer spectroscopy on an 57 Fe enriched Ni 3 57 Fe/C catalyst to follow changes from the pre-catalyst to its active state as well as in situ measurements at various potentials to gain an understanding of the potential-induced changes related to iron and possibly correlate the findings to the observed decrease of the HOR activity at high overpotentials. Ni 3 57 Fe/C catalyst was prepared by chemical reduction at 0 • C using sodium borohydride as a reducing agent. For that, 100 mg of VXC MAX carbon black (Cabot) was dispersed in 15 ml of isopropanol as a hydrophilizer (HPLC Plus GC, 99.9%, Sigma Aldrich) in an ultrasound bath (XUBA3, Grant Instruments) for 1 h. 404 mg (or 1.7 mmol) of NiCl 2 × 6H 2 O (99.3%, Alfa Aesar) was dissolved in 15 ml of de-aerated double distilled water (DDW). The calculated amount of NiCl 2 was based on the intended Ni/C = 1/1 wt. ratio. 275 mg (or 4.92 mmol) of 57 Fe metal powder ( 57 Fe > 95% purity, purchased from Tzamal, Israel, produced by Electrochimpribor, Russia) was dissolved in 12 ml of 2 M HCl (or 24 mmol) solution and then diluted to 250 ml by de-aerated DDW. An aliquot of 28 ml of 57 FeCl 2 solution (or 0.57 mmol) was added to the NiCl 2 solution, and then mixed with the carbon suspension and cooled down in an ice bath. 266 mg (or 7.03 mmol) of NaBH 4 (>98%, Sigma Aldrich) was dissolved in 50 ml solution of 0.01 M KOH (AR, Bio-Lab) and was added dropwise to the precursor flask. The precipitates were separated using a centrifuge (Eppendorf 5804), rinsed with DDW, and dried in a vacuum oven (1407-2, MRC) at 80 • C for 24 h. The final samples were stored in a desiccator under vacuum.

Sample characterization 2.2.1. Ex situ electrochemical characterization
Catalyst ink preparation for rotating disk electrode (RDE). Ten milligrams of the catalyst was suspended in a mixture of 2500 µl isopropanol, 250 µl milli-Q H 2 O, and 250 µl Nafion solution (prepared by mixing 95 µl commercial Nafion solution (10 wt%, Sigma-Aldrich) and 905 µl of isopropanol). The suspension was treated in an ultrasound bath for 40 min with ice. Then, 10 µl of the resulting catalyst ink (four times 2.5 µl each) was dropped on the RDE to reach a loading of 0.17 mg cm −2 . The electrode was dried in air for about 1 h and mounted on the rotating shaft of the RDE rotator. In order to make the catalyst layer more hydrophilic, the electrode was placed into a hot electrolyte (T = 80 • C) for several seconds prior to any measurements.
RDE measurements. Electrochemical measurements were carried out in 0.1 M KOH electrolyte at 25 • C, either with Ar saturation (cyclic voltammetry, 1 mV s −1 ) or H 2 saturation (HOR, 1 mV s −1 , 1600 rpm). A glassy carbon electrode (A geom . = 0.196 cm 2 , Pine) embedded in a Teflon tip was used as the working electrode (WE) and was coated with the catalyst ink. Hg/HgO was used as reference electrode, and a Pt wire was the counter electrode (CE). The potential range was set to −915/-500 mV vs. Hg/HgO (or 0/400 mV vs. RHE). All potentials in this work are given relative to the reversible hydrogen electrode (RHE). As gases, hydrogen (99.999%) and argon (99.999%) were used for the HOR activity and background cyclic voltammograms, respectively.

In situ Mössbauer spectro-electrochemical characterization
Catalyst ink and in situ electrode preparation. In order to prepare the WE and CE, respectively, 20 mg of Ni 3 57 Fe/C catalyst and Black Pearls 2000 (BP2000, Cabot Corporation) were mixed with 568 µl distilled water, 333 µl isopropanol, and 112 µl 5% Nafion solution (QuinTech). After treatment with a vortex for a few seconds, the ink solution was sonicated for 1 h, cooled with ice, and finally treated with a vortexer for a few seconds to obtain a homogeneous suspension. Both the WE and CE were prepared on commercial carbon paper (3 cm × 11 cm, TP-060, QUINTECH) for the in situ Mössbauer spectroscopy and large electrode cyclic voltammetry test. Catalyst ink and BP2000 ink were deposited on an active area of 5 cm 2 at the lower part of the carbon paper layer by layer, reaching a loading of 4 mg cm −2 .
Measurements in the in situ cell. The design of the in situ Mössbauer spectroelectrochemical cell is described in detail in our previous work [27]. The cell was connected to a Versastat 3 F potentiostat (Princeton Applied Research). Cyclic voltammetry and the in situ measurements were carried out in N 2 saturated 0.1 M KOH electrolyte in a potential range of 0-0.4 V (RHE) at 25 • C with a scan rate of 1 mV s −1 . Based on the arrangement, no rotation of the in situ electrode was possible. While the Ni 3 57 Fe/C catalyst layer and BP served as WE and CE, a Hg/HgO was used as the reference electrode. In situ 57 Fe Mössbauer spectra were recorded in transmission mode at potentials of 0.4 V, 0.2 V, 0.1 V, 0.0 V, and −0.2 V (RHE). 57 Fe Mössbauer spectroscopy 57 Fe Mössbauer measurements were performed using a 57 Co/Rh-source at room temperature (298 K). An α-iron foil was used for calibration of the energy/velocity scale. The catalyst powder (59.5 mg) was filled into a polytetrafluorethylene sample holder (2 cm 2 ) and closed with Tesa tape on both sites, and mounted in front of the detector. For the in situ measurements approx. twenty milligrams of catalyst powder prepared as catalyst layer were present within the beam. The in situ cell was placed between source and detector and the thickness of the layer of aqueous electrolyte was minimized to avoid undesired absorption of the γ-rays. Mössbauer spectral simulations were performed with the MossA program and Recoil assuming a Lorentzian line shape [32].

Results
As discussed in the introduction, it is supposed that iron acts as a promotor that enhances the HOR on Ni active sites in the Ni 3 Fe/C catalyst surface [17]. The initial catalyst powder (named pre-catalyst in this work) has a complex chemical and phase composition, as described in the supporting information (figures S1-S5) and visible from the Mössbauer spectrum in figure S1 and briefly summarized here: Similar to our previous work, a core-shell structure with an nickel-iron core and a shell rich in oxygen and iron is obtained. The shell has a thickness of 1-2 nm. X-ray photoelectron spectroscopy (XPS) analysis indicate that only 8% of the nickel are present in form of Ni-O-Fe bonds or Ni-Fe alloy, the main part can be identified as nickel oxides and hydroxide. The XRD pattern mainly shows metallic states. The discrepancy to XPS might be caused by the different sample arrangements and penetration depth during the measurements. Moreover, amorphous layers, e.g. on the shell of the catalyst might only contribute minor to the signal intensities in XRD. While these data summarize the initial state of the pre-catalyst, it is expected that under relevant conditions (e.g. electrolyte pH, potential, or presence of hydrogen) the composition will undergo certain changes-at least at the catalyst surface. Thus, it is interesting to understand the compositional changes that appear from the Ni 3 57 Fe/C pre-catalyst powder to the catalyst in the as-prepared electrode and after contact with the electrolyte, as well as under certain potential values and to reveal any correlation between the HOR catalytic activity and the compositional changes associated with iron.
In the applied potential range (−0.2 to +0.4 V vs.RHE), the carbon does not undergo significant changes, at least within a relatively short period of time. From further chemical dissolution experiments, it was shown that boron oxide formed during the oxidation of the borides tends to leach under alkaline conditions. From the cylic voltammograms in inert atmosphere in figure 2(a) (black curve), one could estimate the degree of the surface coverage by α-Ni(OH) 2 and based on the HOR polarization curves one can see that below 0.25 V vs. RHE the nickel surface is fully oxidized (figure 2(a), red curve). Complementary information on the chemical state of Ni under certain polarization was also reported previously by in situ Raman spectroscopy [18]. In this work, we aim to follow the changes of Fe-containing phases of the catalyst under various conditions. Figure 1 shows the Mössbauer spectra of the pre-catalyst powder, the as-prepared electrode, and the electrode in contact with the electrolyte. In addition, the spectrum of the dried electrode E1 after all measurements is provided, as well as a comparison of the Mössbauer parameters (figure 1(e)) and of the absorption areas of all sites in figure 1(f). The spectra are dominated by two doublets that can be assigned as ferric (D1) and ferrous (D2) high spin species on the basis of their Mössbauer parameters. While the isomer shift of D1 remains almost unaffected (δ iso = 0.31-0.35 mm s −1 ), the quadrupole splitting decreases from 0.79 to 0.63 mm s −1 for electrode in KOH condition. The larger quadrupole splitting in combination with the isomer shift is indicative of ferrihydrite (Fe 3+ 10 O 14 (OH) 2 ) [33], or β-FeOOH [34], the smaller quadrupole splitting is associated with γ-FeOOH [33,34]. In all three structural configurations, iron is in a short range order of Fe 3+ (O) 6 or Fe 3+ O 3 (OH) 3 octahedrals [33]. It is suggested that the latter is formed from the pre-catalyst upon permanent contact of the electrode with KOH, as given in equation (1), The parameters of the second doublet (D2, δ iso = 1.0-1.2 mm s −1 , ∆E Q = 2.1-2.5 mm s −1 ) are related to ferrous iron hydroxide Fe(OH) 2 [34]. In the spectrum of the pre-catalyst also the presence of small amounts of sextet species can be found, which are associated with iron oxides based on the isomer shift value of 0.16-0.18 mm s −1 . In total, they account for ca. 18% of the absorption area. Iron is oxidized and present either in a bi-or trivalent state in the pre-catalyst. This observation of iron oxidation state agrees well with ex situ Raman results for a Ni 3 Fe catalyst prepared by solvothermal reduction [18]. It is believed that this is due to the high oxygen affinity of iron and that induced by the low particle size of the catalyst such species dominate the Mössbauer signature. Moreover, the results indicate that at the beginning of the experiments, only a small quantity of iron interacts with nickel (i.e. only small amount of alloy particles), in agreement with the XPS data.
Before further discussion, the assignment of different sextets that appear under ex situ (figure 1) and in situ conditions (see below) is explained. The sextets are characterized by the isomer shift value which is related to the electron density on iron and by the internal magnetic field. For bulk materials, these two parameters often enable an unambiguous assignment of iron environment. However, specifically for nanometer-sized particles the magnetic field collapses causing smaller values of the magnetic field or even a complete vanishing of the sextet site but appearance of superparamagnetic signatures. At the same time, the isomer shift is not affected by the particle size. Based on this, the obtained sextets were grouped on the basis of the observed isomer shift values, rather than on the basis of the assigned intrinsic magnetic fields. Depending on the size of the magnetic field, however, the sextets are distinguished with labels a, b, c …. A plot that provides the magnetic field as a function of isomer shift is given in figure S6. Sext1 has an isomer shift value of ca. 0.0 mm s −1 , assigned to be of metallic character as e.g. Fe-Ni alloy by comparing the Mössbauer parameters from the literature. Sext2, Sext3 and Sext4 are all related to oxidized Fe environments. Variations in the Mössbauer parameters might be caused by changes in the degree of oxidation or the particle size.
Induced by the electrolyte contact, the areas associated with the two doublets only change minor ( figure 1(f)). The sextet 3b in the as-prepared electrode and in KOH condition are similar. The average oxidation state remains almost unchanged with 2.94 (pre-catalyst) and 2.90 (in KOH). A short description of how the average oxidation states were determined and the resulting values for all electrodes can be found in the supporting information, table S1. Mössbauer fit parameters of the Ni 3 57 Fe/C catalyst and the Ni 3 57 Fe/C electrodes under in situ conditions or post mortem (p.m., Mössbauer recorded on the used electrode under dry state.) can be found in table S2. Table S3 lists the different iron species with structural similarities that were considered for the assignments. In the next step, it should be clarified how the iron signature is altered upon applied potential. Figure S7 compares the CV (in Ar) of the RDE with the CV obtained for the large area electrode (in N 2 ). While an upshift is observed for the large area in situ electrode, the overall shape seems relatively equal. Variations (e.g. capacitance) are attributed to the big difference in morphology and catalyst loading. Figure 2(a) shows the RDE data obtained in Ar (black curve) and H 2 -saturated (red curve) 0.1 M KOH. Under Ar, a broad peak I associated with the electrochemical oxidation of metallic nickel to α-Ni(OH) 2 appears in the potential range between 0.0 and 0.2 V vs. RHE. This peak is correlated with the peak in the forward scan of the HOR polarization curve (red): The first CV peak (black curve) is further followed by another broad peak of electrooxidation between 0.2 and 0.4 V vs. RHE. As in this potential regime the catalyst gets deactivated again (red curve), it is assumed that the peak is not related to the electrooxidation of metallic nickel, since there is no HOR activity of the electrocatalyst in that potential range. According to an in situ Raman study of a Ni 3 Fe catalyst, the α-Ni(OH) 2 layer that fully covers the surface by 0.4 V will be mostly reduced to Ni 0 at a potential of 0 V [18].
The increase in the current density under H 2 -saturated electrolyte (red curve) is an indication of intermediates adsorption on the catalyst surface. A peak current density of 0.4 mA cm −2 can be found at 0.15 V and the HOR performance for U > 0.3 V will decrease (red curve). It should be mentioned that at the potential more anodic than 0.4 V, the Ni surface gets fully oxidized, and the catalyst loses the activity for the HOR. On the basis of this, the potentials selected for the first set of in situ experiments were 0.0 V (E2), 0.2 V (E3), and 0.4 V (E1), always performed on fresh electrodes as indicated in parenthesis. The setup that was used has already been described previously [25][26][27]. The grey dashed lines in figure 2(a) stand for the relevant potentials that were applied under in situ Mössbauer measurements. In figure 2(a), the second series of experiments is indicated with green dashed lines. In this series, after applying the 0.2 V condition, the potential was reduced to −0.2 V and then swept back to 0.1 V (both conditions on electrode E3); the results will be discussed later. In figure 2(b) a representative image of the catalyst is provided. It is indicated that the catalyst particles have a core-shell structure, where the core is enriched by homogeneously distributed Ni and Fe, and it is surrounded by O. The average homogeneity of the element distribution is provided by EDS mapping and an average particle size of ca.10 nm can be seen from HR-SEM imaging in figures S2 and S3. Presumably, related to the iron speciation (figure 1) this is associated with the presence of a metallic core (Ni, NiFe alloy) and the ferri-hydride layer (D1). Just to point out: based on 57 Fe Mössbauer spectroscopy, only conclusions on iron-related species can be made. Whereas no direct information on the nickel is available.
The Mössbauer spectra obtained on the first series of experiments (0.0 V, 0.2 V, and 0.4 V) are shown in figure 3. While the spectrum at 0.4 V looks almost identic to the in KOH condition (figure 1(c)), especially, the iron oxide related Sext3b remains unchanged. We note that an additional decrease in the quadrupole splitting of the first doublet from 0.63 mms −1 (under KOH condition) to 0.48 mm s −1 (0.49 mm s −1 ) appears for a potential of 0.2 V (and 0.0 V), therefore it is marked with an asterix D1 * . At these potentials of 0.2 V and 0.0 V, the D1 * site keeps the same (δ iso = 0.3-0.34 mm s −1 and ∆E Q = 0.48-0.49 mm s −1 ). This D1 * is attributed to the formation of an iron site similar to those in a nickel-iron layered double hydroxide (NiFe LDH), with a molar ratio of 3:1 Ni:Fe, by comparison to literature [30]. Such sites are commonly known to catalyze the oxygen evolution reaction (OER) [36][37][38][39][40] and a direct neighborhood of iron and nickel octahedral coordination is given. Corrigan et al [41] studied the redox processes of iron/nickel hydroxide by in situ Mössbauer spectroscopy in 1 M KOH condition. It was found that the changes in the electronic structure observed for iron are caused by the redox process of host lattice Ni in NiFe-LDH. After performing a reduction at 0 V vs. Hg/HgO, (ca. 0.97 V vs. RHE) an iron site associated with an isomer shift of 0.32 mm s −1 and quadruple splitting of 0.44 mm s −1 was formed, which is very similar to our D1 site at E ⩽ 0.2 V.
Sext2 and Sext3 found at 0.2 V and 0 V are mainly related to oxidized iron species, while Sext1c is found under the 0.2 V condition and is related to Fe-Ni alloy. In figure 4 the absorption areas are compared to the in KOH condition (the change in color code in figure 4(a) is related to the change from D1 to D1 * ).
To gain further insights, the electrode initially polarized at 0.2 V was subjected to the additional in situ measurements, mentioned above, first at −0.2 V, then at 0.1 V. The potential of −0.2 V is typically used to activate Ni-based electrocatalysts prior to HOR, however in [18] it was shown, that the type of catalyst studied in this work does not get activated upon dwelling at reduction (−0.2 V) potential.  (compare table S2) [42,43]. At the same time, the hydrogen evolution reaction becomes apparent by the formation of gas bubbles in combination with a current of −7 mA in figure 5(d). When the potential is swept back to 0.1 V the absorption areas in the Mössbauer spectra remain the same. Thus, the reduced state formed at −0.2 V is stable in inert electrolyte condition, even at potentials where during initial measurements of fresh electrodes a larger average oxidation state was found.

Discussion
In order to consider the overall set of data, the effect of applied potential on the isomer shift and quadrupole splitting of D1 is shown in figure 6(a). Moreover, the average oxidation state is plotted as a function of potential and also indicates the 'history' of each electrode. While upon applied potential, the iron speciation associated with D2 remains constant (changes within the error margin, not shown), and only the absorption area varies; in the case of D1, a change from ferri-hydrite to ferric oxyhydroxide to a signature similar to NiFeLDH (NiFeOOH) is indicated by the change in quadrupole splitting. Analyzing the average oxidation state, we can conclude that between the potential associated with the in KOH condition and 0.4 V the average oxidation state of iron remains close to 3 (see figure 6(b), all E1 electrodes). Then it decreases to 2.7 (E3 at 0.2 V,E2 at 0.0 V) and even to an average value of 1.6 for the electrode subjected to the most negative potentials applied in this work (E3 -0.2V , and E3 0.1V ). Also, tentatively smaller average oxidation states are found when the electrode was subjected to a reductive potential prior to its applied condition (E2 p.m ., E3 −0.2V , E3 0.1V ).
The pre-conditioning has a positive effect on the Ni-based catalysts in the context of HOR [16]. In literature [2,44], the activation carried out in the potential interval from −0.2 V to 0.4 V had the aim to reduce nickel oxide to metallic nickel, thus improving the HOR of the Ni-based catalysts. In our previous work [18], there is a positive effect on the HOR activity for an Ni 3 Fe catalyst prepared by a solvothermal reduction and subjected to a pre-conditioning in the potential range −0.2-0.4 V. However, pre-treatment of   In (b) the arrows indicate the 'history' of the measurements, '∆' refers to the electrode subjected to cycling followed by OCP condition during Mössbauer spectroscopy and '□' to post-mortem electrodes (p.m.) after in situ testing. See Tudatalib repository for the original and fitted data associated with this figure [35]. the Ni 3 Fe catalyst (prepared by the same method as our Ni 3 57 Fe/C) was carried out in the potential range of 0.0-0.4 V. Interestingly, the calculated HOR exchange current density was improved. Thus, it is typical that nickel activation by a pre-conditioning in the potential range of 0-0.4 V has a positive effect on the HOR exchange current density.
The electrode E3 was polarized at a constant potential of −0.2 V, associated with pre-activation resulted in the lowest observed average oxidation state of iron in this work (1.55).
On the basis of the results of our study, together with the current literature data, we can hypothesize that reduced iron species are favorable prior to catalysis to improve the interaction of iron and nickel to enable higher exchange current densities. As already indicated by DFT calculations of hydrogen binding energy [17], the nickel sites are active for HOR. As mentioned before, the nickel is not probed by Mössbauer spectroscopy. However, based on the observation made in this work and previous findings, we assume that the direct interaction between Ni and Fe octahedrals helps the promotion of hydrogen oxidation by enabling a charge transfer between Ni and Fe in similarity to the water oxidation over NiFe-LDH electrodes [38]. While in the case of NiFe-LDH for OER a high valent state of +4 enables the performance enhancement, in the opposite, a low valency is desirable for the HOR, due to the different roles of iron (active site contributor vs. promotor) [31].
The formed Ni 3 57 Fe/C catalyst has Ni-and Fe-neighboring octahedral sites. The Mössbauer results for the initial catalyst indicate that the iron sites are mainly in the trivalent and divalent states. However, nickel may be in a different form. By performing pre-activation (−0.2 V), active sites with direct neighboring of Ni and Fe oxidic octahedrals will be formed similar to structural arrangements in NiFe-LDH, and a site with similar Mössbauer parameters was also found under in situ Mössbauer spectroscopic conditions at 0.1 V and 0.2 V. The oxidation state of iron will remain at a relatively low value, which may stabilize the nickel site and then enhance the performance of HOR through a synergistic mechanism between iron and nickel. A possible explanation might be that by this treatment, Fe sites are formed that are favorable for OH adsorption and can thus facilitate the Volmer step of the HOR.

Conclusion
In this work, in situ Mössbauer spectroscopy was performed to investigate the structural and electronic changes in iron in Ni 3 57 Fe/C alloy catalysts for the HOR. It is shown that depending on the applied potential, one of the doublets changes its local environment from ferrihydrite to Fe oxyhydroxide to NiFe-LDH. Besides the overall composition, the data enabled the extraction of the average oxidation state. It was found that below 0.2 V the average oxidation state of iron is reduced from close to +3 to a valency of about +1.5 at −0.2 V. This oxidation state remains stable under inert conditions even if the potential is increased again. Together with the results from previous work, this characteristic is an indication of the mechanistic role iron has as a promotor in the HOR in the bimetallic catalyst. In similarity to NiFe-LDH the close interaction between Ni and Fe is indicated. Induced by the reduction of iron sites, an arrangement of neighboring Ni and Fe oxidic octahedral sites seems to facilitate faster electron transfer, and in turn, faster HOR. This new understanding may pave the way for developing improved Ni-based catalysts for future AEMFCs.

Data availability statement
The data that support the findings of this study are openly available at the following URL/DOI: https:// tudatalib.ulb.tu-darmstadt.de/handle/tudatalib/3801.