Host, Suppressor and Promoter – the Roles of Ni and Fe on Oxygen Evolution Reaction Activity and Stability of NiFe Alloy Thin Films in Alkaline Media

Understanding the oxygen evolution reaction (OER) activity and stability of the NiFe-based materials is important for achieving low-cost and highly efficient electrocatalysts for practical water splitting. Here, we report the roles of Ni and Fe on the OER activity and stability of metallic NiFe and pure Ni thin films in alkaline electrolyte. Our results support that the Ni(OH) 2 /NiOOH does not contribute to the OER directly, but it serves as an ideal host for Fe incorporation, which is essential for obtaining high OER activity. Furthermore, the availability of Fe in the electrolyte is found to be important and necessary for both NiFe and pure Ni thin films to maintain enhanced OER performance, while the presence of Ni is detrimental to the OER kinetics. The impacts of Fe and Ni species present in KOH on the OER activity are consistent with the dissolution/re-deposition mechanism we propose. Stability studies show that as long as Fe (both co-electrodeposited Fe and incorporated Fe) is present in the electrocatalyst film, the OER activity will degrade under prolonged continuous operation. Satisfactory stability can, however, be achieved with intermittent OER operation in which the catalyst cycles between a degraded and recovered state. Accordingly, two important ranges, i.e., the recovery range and the degradation range, are proposed. Prolonged continuous OER operation (i.e. in the degradation range) generates NiOOH in the electrocatalyst which likely enhances the formation of FeOOH and eventually leads to the OER deactivation. If the electrode works in the recovery range for a certain period, i.e., at a sufficiently low reduction potential, where the Ni 3+ is reduced to Ni 2+ , the OER activity can be maintained (and even improved).


Introduction
The generation of hydrogen through water splitting is considered a promising route for storing energy from intermittent renewable sources such as solar, wind, and tidal energy. 1,2 Since one of the main energy losses in water splitting is the kinetic overpotential of the oxygen evolution reaction (OER), 3,4 OER electrocatalysts have been extensively investigated in order to improve the efficiency of water splitting. Noble metal oxides, such as RuO2 and IrO2, have been proven to exhibit high catalytic activity toward the OER. 3,5 However, the scarcity and high cost of these noble metals largely hinder their large-scale application. Thus, the development of earthabundant electrocatalysts with low cost and high efficiency is critical.
Mixed Ni-Fe based materials have been reported as the most active earth-abundant electrocatalysts for OER in alkaline water electrolysis. 6,7 Although it is widely accepted that the interaction between Ni and Fe is significant for enhancing the OER activity, 8 their roles for the OER activity are still debated. 9,10 Various forms of NiFe-based materials, such as alloys, 11,12 oxides, 13 and oxyhydroxides, 14,15 have been studied as OER electrocatalysts for alkaline water splitting. Among them, the alloys have attracted attention because of the simplicity of their preparation by electrodeposition and strong adhesion to the substrate. 13,16 However, the asprepared structure of NiFe alloys is not the active phase for OER catalysis and therefore, an electrochemical activation step is usually required to convert the metallic/oxide phase to (oxy)hydroxide in order to achieve better OER activity. 10,16,17 Unlike NiFe (oxy)hydroxides, which have been thoroughly investigated as OER electrocatalysts, few studies have been conducted on NiFe-based alloys. 10,17,18 The electrochemical behavior and the surface structural evolution of NiFe-based alloys in alkaline media under OER operating conditions remain unclear and still need further investigations. 16 A concern about NiFe-based materials is their stability. Although significant effort has been devoted to developing highly active NiFe-based OER electrocatalysts, few studies have focused on the topic of stability, [19][20][21][22][23] which is also important for practical water splitting. Recent studies show that degradation is a common issue among the NiFe-based electrocatalysts under prolonged OER conditions, but the causes are still not fully understood. 19,20,22,24 For example, Speck et al. found that the Fe in the FeNiOx films (anode), which were synthesized via electrodeposition or near-infrared driven decomposition, leached into the KOH electrolyte and was deposited on the Ni-cathode during the OER, which is likely related to the OER degradation especially at high current densities. 21 Obata et al. also observed the loss of Fe from 4 electrodeposited NiFeOx films (on an Au-coated FTO substrate) into the KOH electrolyte during the OER, which resulted in gradual OER deactivation with time. 22 Additionally, the existing stability studies on NiFe-based OER electrocatalysts have mainly concentrated on NiFe-(oxy)hydroxides. Since NiFe alloys would experience a surface change from a metallic/oxide to a hydroxide/(oxy)hydroxide during the OER, which is different from pure Ni-Fe (oxy)hydroxide because this is an active phase towards OER itself, 25 further studies are needed to clarify the details about the OER stability of NiFe alloys. Specifically, what electrochemical/chemical conditions can cause degradation, how the structure changes during the degradation, and if and how the degradation can be restored/maintained.
Recently, the dissolution/re-deposition of metallic species has been proposed to explain the dynamic metal ion exchange between the catalyst surface and the electrolyte during the OER process. [26][27][28][29] Anantharaj et al. observed a change in faceting of crystallographic planes from (001) to (101) and (202) planes for nickel hydroxide after cyclic voltammetry (CV) treatment and they ascribed it to the re-deposition of nickel ions during OER. 30 Ye et al. reported that Ni(OH)2/NiOOH film suffered from chemical dissolution, evidenced by the presence of Ni(II) species in NaOH electrolyte observed with in situ UV-vis spectroscopy. 31 Chung et al. reported that the surface Fe, regardless of its phase/oxidation state, is not stable in KOH and always dissolves in aqueous KOH during the OER operation, which leads to reduced activity of the electrocatalyst. 26 They also found that to maintain the OER activity, certain amount of Fe in KOH is necessary to continuously compensate for the Fe dissolved from the electrocatalyst surface via a Fe dissolution/re-deposition process. Therefore, they suggested that the dynamically stable Fe is the active site for OER and keeping the balance between the dissolution rate and re-deposition rate is significant to maintain the OER activity and stability of Fe-MOxHy (M=Ni, Co, Fe). Kuai et al. recently studied the stability of the NiFe-hydroxide and suggested that the decay in catalytic activity under prolonged continuous OER is caused by phase segregation due to the Fe dissolution/re-deposition process. 29 Therefore, operation at the reducing potential for Ni could mitigate the phase segregation and avoid the decay in OER activity. However, the details of the dissolution/re-deposition process remain unclear, e.g., how does the Fe and/or the Ni concentration in the alkaline electrolyte affect the dissolution and redeposition rate, and how do the operating conditions affect the OER activity and stability.
In this contribution, we provide new insights into the roles of Fe and Ni in both metallic NiFe (and pure Ni thin) films and KOH electrolytes toward OER activity and stability. We first study the surface structural evolution and its influence on OER activity of the thin films in the 5 presence and absence of Fe in the KOH. We further perform a series of experiments by varying the concentration of the Ni and Fe species in the KOH electrolyte to understand the roles of Ni and Fe on the OER activity of the NiFe and pure Ni thin films. Based on the achieved results, we propose a dissolution/re-deposition mechanism for the Ni and Fe species present in KOH.
Using continuous and intermittent OER operation, we show that two potential ranges can be distinguished for the OER stability, apparently divided by the Ni(OH)2/NiOOH redox potential.
Lastly, a simple but reasonable hypothesis is proposed to explain the OER degradation and preservation with respect to the existing form of Fe in the film which is likely linked with the Ni valence state that affects the stability.

Preparation of KOH electrolytes
All electrolytes used in this work for catalytic study are 1.0 M KOH. The unpurified KOH was prepared by using the KOH pellets listed in section 2.1. For this KOH, the Fe and Ni concentrations are 1.38 and 0.31 μmol L -1 , respectively, determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Table S1). The purified KOH was obtained by purifying the unpurified KOH according to the procedure described by Trotochaud et al. 32 and the resulting Fe and Ni concentrations were determined as 0.45 and 16.43 μmol L -1 , respectively. In addition, two groups of KOH solutions containing various Ni and Fe concentrations were prepared by adding a certain amount of nickel nitrate and iron(III) nitrate solutions, respectively. 6 Typically, a 500 mL electrochemical bath for electrodeposition consisting of 0.024 M NiSO4·6H2O, 0.006 M FeSO4·7H2O and 0.030 M Na3C6H5O7 was used. The pH of the bath was adjusted to 10.0 using 20% ammonia solution. The electrodeposition was done in a threeelectrode configuration. A 6 × 5 cm 2 nickel foil (thickness: 0.5 mm, Alfa Aesar, 99.94%) and

Electrodeposition of NiFe alloy films on FTO
an Ag/AgCl electrode (METTLER TOLEDO) filled with 3.0 M KCl were employed as the counter electrode and the reference electrode, respectively, while 1 × 3 cm 2 FTO substrates were used as working electrodes. All the FTO substrates were masked off with Kapton tape to expose a 1 × 1 cm 2 area for deposition.
Electrodeposition of NiFe films was performed at a current density of -12 mA cm -2 for 20 s at room temperature (~25°C) using Biologic potentiostat (SP-150). After the deposition, the obtained NiFe films were rinsed several times with DI water and then dried with nitrogen gas.
The electrodeposition of pure Ni films was prepared on the same sized FTO substrate in a similar electrolyte but without the addition of iron(II) sulfate heptahydrate. The electrodeposition was conducted at -12 mA cm -2 for 40 s.

Material characterization
The structure of the as-prepared electrocatalysts on FTO was characterized by grazing incidence X-ray diffraction (GIXRD) using a Bruker AXS D8 Advance X-ray diffractometer with Cu Kα radiation (λ = 0.15406 nm). The angle of incidence for the GIXRD measurement was controlled to be 0.25° and the step size was 0.02 with step time of 2 s. The surface morphology of the films was characterized by field emission scanning electron microscopy (FESEM) using a LEO GEMINI 1530 instrument from ZEISS at 50 k magnification and 5 kV acceleration voltage.
The surface composition of the film surface was determined by X-ray Photoelectron Spectroscopy (XPS) in the CISSY-set-up at the BESSY II synchrotron in Berlin, details of which can be found elsewhere. 33 For all measurements performed in this study, Mg Kα excitation with an energy of 1253.6 eV was used. XPS fine spectra were recorded with a pass energy of 20 eV, with the exception of the Fe 2p, which was recorded with Epass = 50 eV to get a better signal-to-noise ratio. The corresponding change in the Fe sensitivity was taken into account for the quantitative analysis. The Raman spectra were recorded using a custom-made system from S&I components (Germany) equipped with a frequency-doubled Nd:YAG laser emitting light at a wavelength of λ = 532 nm. During the measurement, the exposure time was set to be 60 s and 10 exposures were accumulated per spectrum. The thicknesses of the NiFe 7 and pure Ni films on FTO after deposition were determined with a stylus profilometer (Dektak XT, Bruker).

Electrochemical characterization
The electrochemical characterization was conducted in an H-cell with the two chambers separated by a Zirfon membrane (Provided by AGFA) in order to avoid the crossover of H2 and O2. The working and reference electrodes were placed in one chamber and the counter electrode in the other. An Hg/HgO (Fisherbrand™ accumet™) electrode filled with 1.0 M KOH was employed as the reference electrode while a 3 × 4 cm 2 graphite plate (Alfa Aesar) was used as the counter electrode. All electrochemical measurements were carried out using Biologic SP-150 potentiostat at room temperature, and in all measurements the stirring rate was controlled to be 600 rpm.

Morphology and structure of as-prepared NiFe and pure Ni films
Electrodeposition was used to prepare electrocatalytic NiFe and pure Ni thin films for the OER.
The films show strong adhesion to the FTO substrates ( Figure S1) and provide a suitable 8 platform to investigate the OER activity and stability. The SEM images (Figure 1a) show that the NiFe films consist of uniform small particles of ~40 nm diameter and that the particles homogeneously cover the FTO substrate, while the Ni film in Figure 1b is also very uniform and its surface consists of smaller particles (diameter ca. 10 nm). The phase structure of the freshly prepared thin films was characterized using grazing incidence X-ray diffraction (GIXRD) and Raman spectroscopy. As shown in Figure 1c, the diffraction peaks marked with the hash symbol "#" are indexed to tin oxide, SnO2, with crystalline tetragonal structure (ICDD, PDF No. 00-041-1445). The appearance of diffraction peak positioned at 44.7° (marked with the asterisk symbol "*") for NiFe film is attributed to the (111) plane of crystalline NiFe. 34,35 Similarly, there is only one characteristic peak located at 44.7° for the pure Ni film, indicating that the film is phase-pure (ICDD, PDF No. 00-001-1258). 36 The Raman spectra of the NiFe and pure Ni films were collected in the wavenumber range of 100 cm -1 to 900 cm -1 (Figure 1d).
The appearance of a characteristic signal at 533 cm -1 for the Ni film can be ascribed to the stretching mode of the Ni-O bond and is generally considered to be evidence for oxide film formation on metallic Ni-alloys. 37,38 The shift of the wavenumber of the Ni-O bond to 552 cm -1 for the NiFe film is likely caused by the presence of co-deposited Fe due to the co-deposited Fe. The GIXRD and Raman results indicate that the surface of the NiFe and pure Ni films have been oxidized to some extent although the bulk of the films remains in the metallic phase.

CV activation and surface evolution
It has been reported that the surface of pure Ni is very sensitive to alkaline electrolyte and can be easily oxidized to Ni(OH)2 even without applied potential. 39,40 Under OER operation, the Ni 0 and Fe 0 in NiFe will be oxidized into high-valence Ni/Fe oxides and (oxy)hydroxides. 18 Unlike NiFe (oxy)hydroxides, which have been proven to be OER-active, metallic NiFe will first experience a surface phase evolution in alkaline media under OER conditions. Furthermore, Fe is important for achieving enhanced OER activity for Ni-based electrocatalysts, 41 Figure 2 shows the CV curves of the NiFe and pure Ni films before and after 200 CVs. Both in unpurified or purified KOH, the size of the Ni redox feature for both films is very small in the first cycles, indicating that only a small fraction of the films is oxidized/reduced initially. 43 This is in accordance with the GIXRD and Raman results in Figure 1, in which only the metallic and oxide phases were detected. After 200 CVs, the height of the redox feature in all the KOH electrolytes has significantly increased, suggesting that more metallic Ni was oxidized to Ni 2+ /Ni 3+ , which can be described as: 44 Ni 0 ↔ Ni 2+ + 2e -(1) It is noted that the oxidation peak of the films in unpurified KOH is located at a ~60 mV more anodic potential, i.e., 1.45 VRHE, than for films in purified KOH (1.39 VRHE). This can be explained by the partial charge transfer effect due to the incorporation of Fe from unpurified KOH. 32 This is supported by the smaller size of redox feature in unpurified KOH compared to that in purified KOH, because the presence of Fe partially suppressed the oxidization of the Ni. 45 The CV results of pure Ni films also exhibit the same trends in terms of Ni redox feature size and position ( Figure 2b), with about 20-30 mV less anodic peak positions as for NiFe. The appearance of the feature at 1.54 VRHE in purified KOH (black curve) is ascribed to the formation of Ni 4+ . 32 Regarding the OER activity, the improved OER activity of NiFe and pure Ni films in unpurified KOH after the CV cycling could be due to the incorporation of Fe. 46 Surprisingly, the OER activity of pure Ni reaches that of NiFe after 200 CVs (overpotential of 290 mV at 8 mA cm -2 ).
This result indicates that an enhanced OER activity can be obtained by modifying the surface of the pure Ni film with Fe-containing KOH. It is noticeable that the increase in Ni(OH)2/NiOOH content after 200 CVs does not directly contribute to the OER because the overpotential of the pure Ni sample in purified KOH does not decrease (although after the CV treatment the size of the redox feature increases significantly). This is consistent with the incorporation of Fe being the main cause of the improved activity. 43,47 Even so, the increased Ni(OH)2/NiOOH content still plays an important role in improving the OER activity. As shown in Figure S2, after one hour of OER (@10 mA cm -2 ) or OCP operation in unpurified KOH, the size of redox feature has only slightly increased, as did the OER activity. However, after cycling for one hour (=200 CV cycles) in unpurified KOH, the size of the redox feature has increased significantly (i.e. more Ni(OH)2/NiOOH generated) and the OER activity has greatly improved.
These results suggest that the Ni(OH)2/NiOOH mixed phase is a better host for Fe incorporation than the nickel oxide/metallic Ni structure. In addition, the OER activity of NiFe in purified KOH decreases with CV cycling, while that of the pure Ni is poor and remains almost unchanged. Pure NiOx has already been proven to be a very poor OER electrocatalyst. 48 The poor OER activity of pure Ni in purified KOH proves the importance of Fe for the OER. The deactivation of NiFe in purified KOH after 200 CVs indicates that even for a NiFe film, the 11 presence of additional Fe ions in the electrolyte is essential in order to maintain the OER activity.
These results are consistent with the recent report by Halaoui et al. where they found the Fe presence in electrolyte to be critical for the superior OER activity of NiFeOxHy. 28 In some articles, 42 conditions. 50 In the present work we found that the surface phase of NiFe film after the CV treatment can be stable in air for at least four hours ( Figure S3). Therefore, the Raman spectra collected immediately after electrochemical tests should reflect the true surface condition of the films in KOH.
As shown in Figure 3, at the beginning of the cyclic voltammetry (i.e. after the 1 st CV), two characteristic modes positioned at 480 and 557 cm -1 are observed for NiFe and pure Ni films due to the formation of NiOOH. 39,40 These two modes are ascribed to the depolarized Eg mode and the polarized A1g stretch vibration of the NiO2 framework, respectively. 51 Also, the peaks in the Raman spectra of the films treated in the purified KOH solution are generally narrower and sharper than those of the corresponding films treated in the unpurified KOH, indicating fewer crystalline defects and consistent with a smaller number of incorporated impurities. It is noticeable that the intensities of both I480 and I557 increase with increasing number of CV cycles.
Note that the I557/I480 ratio in this work is calculated based on the peak intensities of I557 and I480. Since these two modes are associated with the NiOOH phase, the increase in intensity must be related to the formation of NiOOH deeper in the films. 40 This result is indeed consistent with the CV results in Figure 2, where the size of Ni redox peak, which is due to the formation of NiOOH (Eq. (2)), increases with the number of CV cycles. It is also observed that I480 increases faster with CV cycling compared to I557, indicating a gradual structure change of the film. After 200 CVs the I557/I480 ratio for NiFe and pure Ni films in unpurified KOH is higher than that for 12 the films treated in purified KOH. It has been reported that the I557/I480 ratio has a close relationship with the Fe content in the film, 39,48 i.e., the increase in Fe content results in the increasing structural disorder of the film. 52,53 Normally the higher the Fe content, the higher value of the I557/I480 ratio. The I557/I480 ratios as a function of CV cycle number are shown in Figure 3e and 3f. In unpurified KOH, the I557/I480 value decreases within the first 100 CV cycles for both NiFe and pure Ni films, indicating that more NiOOH has formed on the film surface. The I557/I480 value stabilizes after 100 CV cycles, suggesting that the Fe content on the catalyst surface has reached an equilibrium. However, in purified KOH, the I557/I480 ratio continues to decrease with CV cycling, 13 indicating less Fe incorporation and more NiOOH formation on the surface. The GIXRD data for both NiFe and pure Ni films after 200 CV cycles was also collected. However, metallic phase still remains as the main structure of both ( Figure S4), indicating that the CV treatment only transforms the surface of the films while the bulk remains metallic. 54 This is different from the NiFe-oxy(hydroxide) film, where the whole film is highly accessible to the electrolytes due to their layered-double hydroxide (LDH) structure which allows the intercalation of anions. 16 The comparison of the unpurified and purified KOH results suggests that Fe incorporated from KOH onto the catalyst surface during CV cycling not only affects the structure of the films (more disordered) but also the OER activity. Ni(OH)2/NiOOH during the CV cycling. The XPS spectra of Fe 2p3/2 (Figure 4b) show that for 14 freshly prepared NiFe, the surface of the film contains a FeO (709.8 eV) phase. 57 However, after 200 CV cycles it disappears and a new peak at 711.2 eV appears, which can be attributed to the presence of Fe 3+ in the form of Fe2O3 and/or Fe-oxy(hydroxide). 58 The characteristic binding energy at 715.8 eV belongs to the Sn 3p3/2 signal of the FTO substrate ( Figure S12). 59 The XPS spectra of O 1s can be deconvoluted into four peaks at 528.7, 529.8, 530.5, and 531.3 eV. The peak at 528.7 eV is located at lower binding energies than normally found in NiO, but Hu et al. found a similar peak for a selenide-derived NiFe oxide that exhibited a layered structure. 60 The 529.8 eV peak can be ascribed to lattice oxygen in polycrystalline NiO, further confirming the existence of NiO. 61 The peak features positioned at 530.5 and 531.3 eV can be attributed to oxide and hydroxy groups of the NiFe alloy, respectively. 62 From the Ni and Fe spectra ( Fig. 4a and 4b), we calculated that the Fe content (molar ratio of Fe/(Fe+Ni)) in unpurified KOH before CV cycling was 9.7 at.% and increased to 18.7 at.% after 200 CV cycles.
These results further prove that the NiFe film absorbs Fe from the unpurified KOH electrolyte during CV cycling.

Impact of concentration of Fe and Ni species and dissolution/redeposition mechanism
So far, we have studied the NiFe and pure Ni films in unpurified and purified KOH electrolytes.
The  low Ni concentration has almost no effect on the OER activity. However, as the Ni concentration is increased to 10 μmol L -1 , the OER activity drops significantly, even though it still improves during cycling (solid vs. dashed curves in Figure 5a). Our results prove that the presence of Ni in KOH indeed reduces the OER activity, probably by occupying the Fe sites.
As the Ni concentration increases to 15 μmol L -1 , the OER activity reduces further and remains unchanged after 200 CVs, indicating that the dissolution/re-deposition between the Ni and Fe has reached a dynamic equilibrium. From the OCP-OES results (Table S1, Supporting Information), we know that the Fe concentration in unpurified KOH is 1.38 μmol L -1 . Since the Ni concentration needed to achieve a dynamic Ni/Fe equilibrium is ~10 higher, we deduce that the dissolution/re-deposition rate of Fe is around 10-fold that of Ni. Further increasing the Ni concentration (to 35 μmol L -1 ) resulted in a worse OER activity (with little difference before and after cycling), suggesting that a high Ni concentration increases the deposition rate of Ni on the Fe sites, which suppresses the OER activity. This again indicates that the dissolution/redeposition rate of Ni is much lower than that of Fe, which is consistent with some reports that the corrosion rate of Ni is low in KOH under OER operation. 29 16 Figure 5.  Further quantification of the results in Figure 5 indicates that the concentration ratio of Ni/Fe (or Fe/Ni) is a very good descriptor for the OER activity. As shown in Figure 6, the data from both the purified and unpurified electrolytes essentially falls on the same curve when plotted as 18 a function of the concentration ratio, despite an order of magnitude difference in the absolute Fe and Ni concentrations for Ni/Fe ratios less than 2. Both the potential at 1 mA cm -2 and the Tafel slope exhibit the same general trend. The current density 1 mA cm -2 was chosen, because it was low enough that accurate potential could be determined for all measured CV curves. The highest performance is achieved with Ni/Fe ratio less than 2, and in those cases the OER performance appears nearly independent of the ratio. This matches the aforementioned observations of Chung et al. that increasing the Fe concentration of the electrolyte beyond a certain value had no additional benefit (they did not report Ni concentrations). 26 With increasing Ni/Fe ratio, the performance gradually becomes worse until there is a large deterioration at a value of around 30. This (and the results in Figure 5) indicates that the Fe dissolution/redeposition rate is much faster than the corresponding rate of Ni, and the enhancing effect seems to saturate when the Fe concentration reaches about half of the Ni concentration. Conversely, the effect of Fe species can also be suppressed by Ni species, but only when the Ni concentration is much higher (over ~7-fold) than the Fe concentration. Based on above results and the discussions, we propose a dissolution/re-deposition mechanism for the NiFe and pure Ni films in alkaline conditions (Figure 7). First, the Raman and GIXRD

Stability under continuous and intermittent OER operations
The  Figure S8). It is noticeable that the size of the redox feature for both cases (unpurified and purified KOH) before and after 48 h OER operation remains almost unchanged, indicating that the total Ni(OH)2/NiOOH content did not change during this period. These results imply that the increase in Fe content during the continuous OER operation does not improve the OER activity.
In addition, the GIXRD data ( Figure S6) collected after 48 h continuous OER for NiFe in unpurified KOH show that the bulk of the film still remains present as a metallic phase.
As a comparison, the stability of the pure Ni film in unpurified KOH and purified KOH electrolytes was also measured by the same protocol as NiFe film (Figures 8e-f). Similarly to NiFe film in unpurified KOH, the current density of the pure Ni film decayed over time. The corresponding EIS and CV exhibit the same trend as observed for NiFe film in unpurified KOH.
The similarity between the NiFe and Ni films is not unexpected, since the surface of the pure 21 Ni film is also expected to contain Fe after prolonged cycling in the unpurified electrolyte. In contrast, the behavior in purified KOH is very different, with no degradation occurring for the pure Ni films. Note that the OER activity of these films is poor due to the absence of Fe in both the film and the electrolyte. 64 Surprisingly, the OER activity slightly increases after the 48 h stability measurement. Table S2 demonstrates that in purified KOH the Cdl decreases from 2.4 to 1.7 mF cm -2 , with the Rct value simultaneously decreasing from 2339 Ω cm 2 to 377 Ω cm 2 .
These results imply that the slight OER improvement for the pure Ni film in purified KOH after 48 h was not due to an increase in the active surface area (it actually decreases). Figure 8f shows that the Ni 2+ /Ni 3+ redox peak shifts to slightly more anodic potentials after cycling. All these observations are consistent with the presence of trace Fe impurities in the purified KOH, which can enhance the OER activity but also block part of the Ni surface sites. 65 The Raman spectra for NiFe and pure Ni films before and after the 48 h continuous operation in unpurified and purified KOH were also measured and compared ( Figure S7). The I557/I480 ratio increases slightly after the stability tests in unpurified KOH ( Figures S7a and S7c), from 0.92 to 0.99 for NiFe and 0.96 to 0.99 for pure Ni. This small increase hints at increased structural disorder, likely caused by an increased surface Fe content (see the XPS results in Figure S8). In purified KOH electrolyte, however, the I557/I480 ratio of NiFe remained almost unchanged after 48 h continuous OER operation ( Figure S7b). This is because the Fe has already leached out before the start of the 48h measurement, and is not re-deposited in the purified KOH electrolyte. The slight decrease of the I557/I480 ratio from 0.71 to 0.65 can be ascribed to the increased NiOOH content. The difference between the NiFe and pure Ni in purified KOH is that the pure Ni film does not contain any Fe originally; therefore, the NiOOH content increases with the continuous OER operation. Based on the stability results above, it seems that there are two important potential ranges which determine the OER stability of the NiFe, i.e. the ranges cathodic and anodic of the Ni reduction potential (called here "recovery range" and "degradation range", respectively, as shown in Figure 9e). When the NiFe-based electrocatalysts are exposed to OER conditions for a period of time, it is important to periodically change the applied potential to the recovery range, where the catalyst can recover its lost OER activity. The border between these two ranges is determined by the Ni redox profile of the electrocatalysts, i.e. the applied potential should be low enough to reduce the NiOOH to Ni(OH)2.
To verify the mechanism proposed above, the NiFe film was measured under accelerated stress test (AST) conditions, i.e., 10 min at the potential of 1.725 VRHE, then 10 min at the potential of 1.575 VRHE, and then the OCP for another 10 min. The total duration time for the measurement was 96 h. This AST protocol has been employed to evaluate the stability of electrocatalysts in proton exchange membrane (PEM) water electrolyzers. 67,68 It is useful also for testing NiFe-based electrocatalysts, because its operating conditions are closer to real-life operation in PEC and PV-EC devices than simple chronopotentiometry or chronoamperometry. 24 As shown in Figure S10a and Table S3, the NiFe film does not show any OER performance deactivation during the 96 h measurement period. Actually, its OER activity improves a little after the stability tests, which is further confirmed by EIS and CV results in Figure S10b and S10c. We notice that during the stability measurements the OCP value was located within the range 1.125-1.325 VRHE, which corresponds to the recovery range. Thus, the key for maintaining the OER performance seems to be intermittent OCP operation. This explains the long-term stability of the film and further supports the proposed mechanism.

Explanations to the OER degradation and preservation
To understand the surface change of the NiFe film, we also collected and compared the XPS spectra after the 48 h continuous and intermittent OER operations ( Figure 10). The XPS spectra of Ni 2p3/2 for both cases can be fitted with two characteristic peaks at 855.1 and 856.2 eV, which are associated with the formation of Ni(OH)2 and NiOOH, respectively. 69 Such results are consistent with the Raman spectra in Figure S6 in which the NiOOH phase is present. It is noticeable that the surface NiOOH content (25.1%) after the continuous operation is much higher than that after intermittent operation (7.7%). These results confirm the hypothesis that a high content of NiOOH is unfavorable to maintaining the OER activity. However, it must be pointed out that the high content of NiOOH is unlikely to be the direct cause for the OER degradation because the pure Ni film did not show any OER deactivation in purified KOH after 48 h continuous OER operation (Figure 8d). OER operating conditions. 25 The characteristic Fe 2p3/2 peak located at 711.2 eV indicates the generation of iron oxides and/or (oxy)hydroxides. 70 Due to the partial overlap with Sn 3p3/2 (715.8 eV) and the complexity of iron phases, it is difficult to assign the signal at 711.2 eV to any single phase of iron. 71 Still, the surface Fe content after the prolonged OER operations can be calculated. As shown in Figure S8, the relative Fe content (i.e., Fe/(Fe+Ni)) after 48 72 We note that the OER measurements on the NiFe-based electrocatalysts in these references (i.e. 21,22,26,72 ) were conducted in purified KOH. Therefore, the loss of Fe is likely due to the Fe dissolution and the high content of Ni species in purified KOH, which occupy the Fe sites, as explained by the proposed mechanism in Figure 7. This eventually results in the deactivation of the OER activity. In contrast, we observe an increase in Fe because the Fe content exceeds the Ni content in our unpurified electrolyte (Table S1). be ascribed to the oxide and hydroxyl groups of the NiFe alloy, respectively. 62 Additionally, the appearance of the bond at 532.7 eV after continuous OER operation can be attributed to chemically adsorbed water molecules, 10 which could originate from the intercalation of H2O molecules into the layers of γ-NiOOH. 44 The peak at 528.7 eV for the intermittent conditions is ascribed to the lattice oxygen (O 2-) in the layered structure. 60 These results further indicate that continuous operation enhances the formation of NiOOH.
The CV curves of the NiFe film recorded for both continuous and intermittent operation are also compared. As shown in Figure 11, the area of the Ni 2+ /Ni 3+ redox peaks for continuous OER operation did not change significantly after operation, but the anodic peak slightly shifted towards more positive potentials. For intermittent operation the area of the redox peak increased greatly and both the cathodic and anodic peaks shifted to more anodic potential, implying that more Ni(OH)2/NiOOH was generated and that the NiFe film was further activated under the intermittent OER operation. Therefore, the potential shifts between the "degradation" and "recovery" ranges during the 48 h intermittent operation are similar to the CV cycling, which further improves the OER activity. Since we previously showed that the increase in 26 Ni(OH)2/NiOOH content alone does not have a direct relationship with the OER activity, the improvement after intermittent operation is caused by the increased Fe content. However, it is noticeable from XPS that the surface Fe content after the continuous OER operation also increased but its OER performance degraded. This result contradicts several previous observations that the Fe content in the NiFe-based electrocatalysts decreases. [20][21][22] Therefore, we propose that during OER operation Fe is converted likely to FeOOH, and its accumulation in continuous operation reduces the OER activity, while under intermittent operation the FeOOH is removed periodically, so the catalyst activity remains more stable. 29 In addition, the SEM images ( Figure S9) collected before and after the 48 h continuous OER operation (applied potential 1.725 VRHE) show that the morphology of the NiFe film remained almost intact, indicating the good stability of the film. However, these conclusions conflict with our results which indicate that, in fact, the reduction 27 of Ni(III) to Ni(II) is the key to maintaining the OER activity. Furthermore, they did not mention the KOH purity, which is important to the stability, as illustrated by pure Ni film in purified KOH not showing activity degradation after prolonged OER operation, while in unpurified KOH its activity was degraded (Figure 8d). Kaui et al. recently found that the formation of FeOOH caused by the Fe segregation is responsible for the OER degradation of the mixed Ni-Fe hydroxide under prolonged continuous OER operation. 29 They proposed that the phase segregation can be alleviated by holding the electrode at a reduction state intermittently.
Although we were not able to identify the FeOOH phase in our current study, our results suggest that the increase in NiOOH content under prolonged continuous OER is likely related to the formation of FeOOH. Obviously, the reduction of the NiOOH to Ni(OH)2 by CV cycling or intermittent operations are conducive to maintaining (and even improving) the OER activity.

Conclusions
In summary, we have reported the effects of Fe on the surface structure evolution and the OER activity of metallic NiFe (and pure Ni) films in alkaline electrolyte. proposed to explain the OER degradation and recovery. Compared to the prolonged intermittent OER operation, the prolonged continuous OER operation generates more NiOOH that likely enhances the formation of FeOOH, eventually leading to the OER deactivation. Importantly for the application in solar driven devices, the electrode works in the recovery range for a period of the daily cycle, i.e., at night the electrode operates at a sufficiently low reduction potential to reduce Ni 3+ to Ni 2+ , the OER activity can be maintained (and even improved).