Bi‐Functional Iron‐Only Electrodes for Efficient Water Splitting with Enhanced Stability through In Situ Electrochemical Regeneration

Scalable and robust electrocatalysts are required for the implementation of water splitting technologies as a globally applicable means of producing affordable renewable hydrogen. It is demonstrated that iron‐only electrode materials prove to be active for catalyzing both proton reduction and water oxidation in alkaline electrolyte solution with superior activity to that of previously established bi‐functional catalysts containing less abundant elements. The reported bi‐functionality of the iron electrodes is reversible upon switching of the applied bias through electrochemical interconversion of catalytic species at the electrode surface. Cycling of the applied bias results in in‐situ electrochemical regeneration of the catalytic surfaces and thereby extends the catalyst stability and lifetime of the water electrolyzer. Full water splitting at a current density of I = 10 mA cm−2 is achieved at a bias of ≈2 V, which is stable over at least 3 d (72 one hour switching cycles). Thus, potential‐switching is established as a possible strategy of stabilizing electrode materials against degradation in symmetrical water splitting systems.


Introduction
Sources of power based on limited fossil fuel reserves are depleting, and becoming increasingly socially and politically unacceptable due to their environmental and socio-economic impact. [1][2][3][4][5] Renewable energy sources including wind, geothermal, hydroelectric and solar, which are amongst the most abundant resources on Earth, offer solutions to these issues. Indeed, most countries could cover their total electricity demand using these resources. [ 6,7 ] Recent statistics, however, show that only 18.1% of global energy was supplied by electricity, with the rest of demand being met using chemical fuels. [ 8 ] The globalization of water splitting is a promising route to the sustainable production of the green fuel, hydrogen, and an ultimate goal of the storage of renewable energy. The world's most abundant energy source, solar energy, suffers from the issue of intermittency. Two highly debated approaches in the solar fuels fi eld in acidic solutions (0.5 M H 2 SO 4 ), [ 35 ] however, the applicability of FeP in water splitting is currently limited as it is only active in acidic to neutral pH, [35][36][37] where nonprecious metal OER catalysts do not typically operate. [ 30 ] Further, the scalability of phosphide materials is limited by toxicity and availability. Examples of iron oxyhydroxides operating as OER catalysts in alkaline solution are well known and operate in an overpotential range of 0.35-0.5 V, [38][39][40] but they have not been reported as HER catalysts. Chlorine oxidized stainless steels were also shown to be OER catalysts at alkaline (1.5 mA cm −2 at η = 260 mV) and neutral pH (1 mA cm −2 at η = 462 mV). [ 41,42 ] A regenerating ability to enhance stability is also unknown for iron-based water splitting catalysts.
Herein we demonstrate that electrode materials containing only iron as metallic component are bi-functionally active for both water splitting half reactions in alkaline media. Thus, a single and ultra-low cost iron electrode can be used in a symmetrical water electrolyzer and the stability of the iron-only materials can be improved through in-situ electrochemical regeneration of their catalytic surface under reversal of the applied bias. This provides an unprecedented means to repair bi-functional electrodes and provides an elegant solution to stabilize electrode materials that are widely associated with corrosion in aqueous solution. In addition, the use of a bi-functional electrode and this approach to regenerate active catalyst species can help alleviate the instability issues of electrodes in alkaline electrolyzers under conditions of fl uctuating currents and gas pressures, which are particularly relevant in the context of a PV-electrolyzer setup. This strategy is specifi c to bi-functional materials that can catalyze both water splitting half reactions, as otherwise electrolysis would halt during potential cycling.

Synthesis and Characterization of Electrode Materials
Iron-only catalytic thin fi lm coatings comprised of iron oxide nanorods (FeO x ), metallic iron nanoparticulates (Fe), and iron sulfi de nanocurls (FeS x ) were prepared by simple and scalable aqueous syntheses. [ 43 ] FeO x was synthesized by a chemical bath deposition (CBD) containing iron(III) chloride and urea, whereas FeS x was deposited from a CBD containing iron(II) chloride, urea, and thioacetamide, both at 90 °C. The Fe fi lm was prepared by electrochemical deposition from a solution containing iron(II) ammonium sulfate. Full synthetic procedures can be found in the Experimental Section, along with characterization of fi lm thickness and loading ( Figure 1 , Table S1 in the Supporting Information). Scanning electron microscopy (SEM) images reveal that the three materials have substantially different morphologies ( Figure 1 ). FeO x consists of nanorods, whereas Fe is an agglomeration of spherical nanocrystals and FeS x comprises of petal-like nanocurls.
The FeO x electrodes were determined to be amorphous by powder X-ray diffraction (XRD; Figure S2a in the Supporting Information). The surface composition of the fi lms was also analyzed by X-ray photoelectron spectroscopy (XPS; Figure S3 in the Supporting Information), where peaks at around 710 and 723 eV corresponding to the Fe2p 3/2 and Fe2p 1/2 levels in iron oxide species were found (spin-orbit Δ = 13.1 eV for Fe2p). Deconvolution of the primary Fe2p 3/2 peak of FeO x shows two major features at 710.0 and 711.0 eV in agreement with an iron(III) oxide-hydroxide phase ( Figure 2 ). [ 44 ] The metallic Fe fi lms possess a crystalline α-Fe structure with an intense peak at 44.7° 2θ corresponding to the 110 facet and a weaker feature at 82.4° 2θ (211) in the XRD pattern ( Figure S2b, Supporting Information). The surface composition, however, consists of iron(0) contained in an iron(III) oxide matrix. The fi tted Fe2p 3/2 XPS spectrum shows a major feature at 709.8 eV for iron(III) oxide and a minor feature at 710.0 eV for the oxide-hydroxide, in addition to a peak at 706.4 eV for metallic iron(0), [ 44 ] which comprises only 5.6% of the surface iron states (Figure 2 ). This mixed iron(0)-iron(III) oxide surface structure of the electrodeposited Fe fi lm seems to act as a  protection layer to the bulk α-Fe phase, as the material is stably stored in ambient air over many months. For comparison, a commercial iron foil (Alfa-Aesar; Puratronic; 99.995%) rusts considerably in the same timeframe.
The XRD pattern for the FeS x fi lm shows a strong crystalline feature at 16.8° 2θ, which is in good agreement with the 100 facet of FeS ( Figure S2c, Supporting Information). However, the Fe:S ratio of 1:0.7 obtained by energy dispersive X-ray (EDX) analysis of the FeS x fi lm suggests a sulfur-defi cient iron(II) sulfi de phase FeS 1x or a mixed sulfi de-oxide phase FeS/FeO x . Figure 2 shows the deconvoluted Fe2p 3/2 XPS spectrum of FeS x , where oxide peaks at 709.9 and 710.8 eV similar to FeO x are observed. Additionally, a peak at 707.8 eV is present, which agrees with the Fe(II)-S environment in FeS. [ 45 ] The S2p spectrum deconvolutes to two spin-orbit doublets (spin-orbit Δ = 1.2 eV for S2p), the major one at 161.4 and 162.6 eV and the other at 163.5 and 164.7 eV ( Figure S4, Supporting Information). The former doublet corresponds to monosulfi de (S 2− ), while the latter to a less reduced disulfi de species (S 2 2− ). [ 45 ] The surface composition of FeS x is therefore likely to be FeS in an iron(III) oxide-hydroxide matrix, but with some (28.0%) of the surface sulfi de sites partially oxidized.

Bi-Functional Electrocatalytic Performance
This series of iron-containing electrodes was tested for catalytic HER and OER activity using linear sweep voltammetry (LSV) under various conditions at room temperature. Aqueous acidic conditions (pH 1) are unsuitable for these materials, as none are stable with respect to dissolution, despite a previous report of iron sulfi de activity for HER in 0.5 M H 2 SO 4 . [ 46 ] Neutral conditions (pH 7, Figure S5 in the Supporting Information) revealed initial HER activity for Fe and FeS x , but only a stable current for FeS x (Fe loses > 80% of its initial activity within 1 h, Figure S6 in the Supporting Information). For OER catalysis, Fe is unsuitable under neutral conditions as it becomes unstable to dissolution under anodic potential. FeO x is stable, but requires a very high onset overpotential ( η ≈ 700 mV) for OER catalysis. FeS x is fi rst oxidized where it then has a similarly high overpotential. Hence, acidic and neutral conditions are not suitable for an efficient symmetrical water splitting system using these iron-based electrodes as both anode and cathode.
The best performance was obtained when an alkaline electrolyte solution (0.1 M KOH, pH 13) was employed. Under these conditions, FeO x displays a reasonably low overpotential ( η at 10 mA cm −2 ) of 400 mV ( Figure 3 a) for OER catalysis, which is comparable to other non-noble metal materials as well as iron-oxyhydroxides. [ 39 ] Strikingly, Fe and FeS x show a similar activity to FeO x at an anodic potential. FeS x undergoes a visible transformation from a black to a yellow fi lm, corresponding to a bulk oxidation of the fi lm. In contrast, the visual appearance of the Fe fi lm remains unchanged suggesting that any oxidation here occurs only at the surface.
In the fi rst cathodic LSV scan, Fe and FeS x exhibit similar activities, whereas FeO x requires a much larger η for HER (Figure 3 b). Given the comparable activity of all three materials for the OER and both Fe and FeS x for the HER in alkaline solution, it is unclear from the fi rst scan LSV as to whether the larger overpotential for HER using FeO x is the result of inactivity or a more diffi cult transformation into the catalytically active species. The latter possibility is confi rmed by taking successive LSV scans and observing the marked anodic shift of the HER wave with successive scans (Figure 3 b and Figure S7 in the Supporting Information).
The potential-time ( E-t ) traces of a galvanostatic measurement show that all three materials are active for both the HER and the OER with a remarkably similar bias potentials of −0.36 (at −10 mA cm −2 ) and 1.68 V (at +10 mA cm −2 ) versus reversible hydrogen electrode (RHE), respectively, once stabilized ( Figure 4 ). This current density was selected due to its increasing use as an activity benchmark, relevant to achieving 10% effi ciency in solar-to-fuels devices. [ 30 ] To investigate the reversibility of this bi-functional property, an alternating E-t trace was recorded, where the current density is periodically switched between + 10 and −10 mA cm −2 ( Figure 5 ). Several conclusions can be drawn from the results of these experiments. First, the highly similar catalytic potential of all three materials in both the HER and OER suggests that very similar surface iron species are present during catalysis independent of the initial chemical composition of the electrode. The electrocatalytic response varies according to the ease of conversion  of the surface species to the active catalysts on the short timescale of the fi rst LSV scan, whereas the most active catalyst is revealed almost instantly in the Et trace upon applying a suffi ciently high current density due to the rapid interconversion of species. Although the nanostructured FeO x and FeS x appear to have a higher surface area than Fe, due to the low conductivity of iron oxides, it is likely that their electrochemically active surface area is much lower than would be expected from the SEM images. Second, η is constant after several alternating cycles between HER and OER. These materials are hence bifunctional and their catalytic phases reversibly interconvertible.
The behavior of the iron-only electrodes is similar to the bi-functional material H 2 -CoCat/O 2 -CoCat (in phosphate buffer) [ 47 ] and the more recently reported Ni analog (in borate buffer), [ 48 ] but with superior water splitting performance, operating with η = 360 mV (at −10 mA cm −2 ) and 450 mV (at −10 mA cm −2 ) for HER and OER, respectively. CoCat showed HER activity of −2 mA cm −2 at η = 385 mV and OER activity of +1 mA cm −2 at η = 545 mV. [ 47 ] NiCat showed HER activity of −1.52 mA cm −2 at η = 452 mV and OER activity of +0.6 mA cm −2 at η = 618 mV. [ 48 ] In these materials, the potential controlled transformation between different catalytic species and morphologies was shown to be responsible for the bi-functional activity. H 2 -CoCat is reported to contain metallic cobalt active sites in an oxo-hydroxy-phosphate matrix, whereas O 2 -CoCat, more widely known as CoP i is a cobalt oxide. [ 27,47 ] The analogous NiCat is similar replacing cobalt and phosphate with nickel and borate. [ 48 ] Unlike these materials, however, here the two catalytic phases are not likely to exchange through solution via a dissolution-deposition mechanism, but instead may convert in the solid state. This is supported by the lack of major morphological change before and after catalysis as observed by SEM ( Figure S8, Supporting Information) and further while CoCat and NiCat are metastable with respect to dissolution under open-circuit potential, the iron-based fi lms reported herein are stable in their electrolyte solution. Finally, CoCat and NiCat operate at or close to pH neutral conditions and highcurrent water electrolysis may subsequently suffer from mass transport limitations imposed by the lack of H + and OH − ions. The bi-functional Fe catalysts avoid these limitations, as they are able to function under strongly alkaline conditions.
The bi-functional and reversible activity of FeO x is also shown to be highly stable over 3 d with only a minor increase in the applied bias for full water splitting during a galvanostatic experiment (≈0.02 V, Figure 6 ). This as yet unreported bi-functional property of iron-only materials under alkaline conditions reveals the broad potential to use iron-based materials as catalyst on both anode and cathode in symmetrical water splitting systems. Operating in a reasonable overpotential region [ 30 ] for both the HER and OER half reactions, full water splitting using iron-only materials is possible.

In Situ Electrochemical Regeneration of Bi-Functional Water Splitting Electrodes
In situ electrochemical regeneration is observed for the prepared FeO x electrode when potential switching is used. Although these electrodes are stable over the 12 h timescale (Figure 4 ), over a longer 3 d period degradation of the catalysts is observed when used separately and continuously as HER or OER catalyst as opposed to when potential switching occurs ( Figure 6 ). The HER catalyst gradually decays over 3 d (bias increase of 110 mV), likely due to corrosion or passivation of the surface Fe(0) sites by the electrolyte. The OER catalyst is relatively stable over the same timeframe. The resulting increase in bias for full water splitting with separate FeO x anode and cathode was therefore 0.11 V. In contrast, when the anode and cathode were reversed every 1 h by potential switching, the increase in bias for full water splitting over 3 d was negligible (0.02 V; Figure 6 ).
The reversible bi-functional activity even extends to a commercially available iron foil (Alfa-Aesar; Puratronic; 99.995%). Fe foil proved to be bi-functionally active although with an inferior water splitting bias of ≈2.2 V at I = 10 mA cm −2 , likely due to the smooth morphology as compared to the nanostructured fi lms prepared herein ( Figure 1 ). Like the Fe fi lm, its surface oxidizes at an anodic potential to form an active OER catalyst, which can be reduced back to the metallic HER catalyst under cathodic treatment. The foil, however, shows signifi cantly lower stability for HER than the in-house prepared iron-only fi lms ( Figure 7 a) with an increasingly negative potential over time, again due to either corrosion or passivation in the alkaline electrolyte solution. After only 12 h electrolysis with separate electrodes for HER and OER, the water splitting bias had increased by a further 0.2 V, whereas the bias of that with potential switching remained almost unchanged (Figure 7 b).
The reversible potential-controlled interconversion of catalytically active HER and OER surfaces hence stabilizes the materials against a loss of activity. This feature can be used to extend the stability and operability time of iron-based materials with a bi-functional property since regeneration of the catalytic surface occurs every time the potential of the anode and cathode are reversed, without the need to halt water splitting.

Mechanism of Bi-Functional Activity and In Situ Electrochemical Regeneration
In agreement with the Pourbaix diagram for iron species in aqueous solution, [ 49 ] metallic iron is likely to be present at sufficiently cathodic potential and is the likely site for HER catalysis. Likewise, at an anodic potential an iron(III) oxide will be present and is known to function as OER catalyst. Figure 8 shows the proposed mechanism of the electrochemically reversible bi-functional activity of iron-only electrodes. Under suffi ciently anodic bias, the material surface converts into an iron(III) oxide-(hydroxide), referred to as FeO x which acts as catalyst for OER in the strongly alkaline solution. At a cathodic bias, surface sites are reduced to Fe(0), which is the active HER catalyst. Regardless of the initial identity of the iron-based material employed, the surface converts under bias into the species involved in this cycle, and thus shows very similar activity. In this study, Fe, FeO x , and FeS x were employed, but this is also likely to apply to other iron-based species.
Adv. Energy Mater. 2016, 6, 1502095 www.MaterialsViews.com www.advenergymat.de  LSV and XPS were used to confi rm this proposed mechanism of bi-functional activity. Figure S9 (Supporting Information) shows the cathodic LSV scans of FeO x and FeS x before and after cathodization at −10 mA cm −2 for 12 h compared to an as-deposited Fe fi lm electrode. In the case of FeO x , the activation process clearly converts the active surface into a different catalytic species with almost identical activity to the asdeposited Fe fi lm, strongly supporting a metallic iron species as the active sites for hydrogen evolution.
The model of potential-operated switch between Fe(0) (cathodic) and Fe(III) oxide (anodic) phases with concurrent switch from HER to OER catalysis has been further confi rmed through XPS analysis of the electrodes after an HER step and after an OER step ( Figure 9 , and Figures S10 and S11 in the Supporting Information). The Fe2p XPS spectrum of an FeO x electrode after 3 h HER activity shows the appearance of a new peak at 706.2 eV corresponding to Fe(0) ( Figure 9 ). [ 44 ] XRD of the same sample does not show the presence of crystalline features of α-Fe indicating that the oxide-derived Fe(0) catalyst is amorphous or a thin surface layer. [ 29,50 ] Indeed, the oxide peaks fi tted at 709.8 and 710.2 eV and their relative ratio confi rm that the Fe(0) sites are held within an Fe(III) oxide matrix and this activated HER catalyst is very similar in surface composition to the as-prepared Fe fi lm electrode (vide supra, Figure 2 ). The spectrum of FeO x after 3 h OER activity is unchanged relative to the as-prepared material, indicating that FeO x is itself the active oxygen evolving catalyst. Similarly, the Fe(0) peak seen in the Fe fi lm after OER has decreased in relative magnitude (3.7%) to that before (5.6%), supporting increased iron oxide content in the OER active phase ( Figure S11a, Supporting Information). The spectrum of the Fe fi lm after HER is largely unchanged with only slightly increased iron content (5.9%), supporting iron(0) active sites in an oxide matrix as the hydrogen evolution catalyst.
FeS x was also found to be cathodically activated for HER ( Figure S9, Supporting Information). The active HER catalyst could either be Fe(0) or indeed iron sulfi de itself or a mixture of both species supported in an oxide matrix. The activation process in each case would be the reduction of surface iron oxide or the reduction of oxidized sulfur (disulfi de) sites to FeS respectively. No Fe(0) feature was observed for the FeS x fi lm after HER catalysis in the XPS spectrum ( Figure S11b, Supporting Information). However, quantifi cation of sulfur by both XPS and EDX indicates a loss of S during HER, suggesting that the active phase of the FeS x fi lm under HER is unstable to air after removal from the electrolyte solution and so could not be analyzed accurately by XPS. It is clear, however, that FeS x converts into an FeO x material at anodic potentials. The fi tted Fe2p XPS spectrum of FeS x after OER catalysis shows no Fe(II)-S peak at 707.8 eV and is almost identical to as-prepared FeO x with iron(III) oxide peaks at 709.9 and 711.0 eV ( Figure S11b, Supporting Information). Full loss of sulfur also occurs in the S2p spectrum ( Figure S12, Supporting Information). This sulfi de-derived FeO x can then enter into the same potentialcontrolled Fe(0)-Fe(III) cycling of surface catalysts as the FeO x and Fe electrodes.
The stabilization of the reported iron-only materials induced by potential switching is also explained by the potential-controlled interconversion of catalytically active surfaces between Fe and FeO x . This occurs primarily by renewal of the Fe(0) catalytic HER surface upon cycling, which is subject to chemical corrosion by the aqueous solution and could also prevent overoxidation/dissolution of the iron oxide OER catalyst.

Iron-Only Electrodes in Water Electrolysis
Water splitting was performed with the iron-only electrodes (two-electrode confi guration) as both anode and cathode in an alkaline electrolyzer with the electrochemical bias potential ( U ) recorded at a current density I = 10 mA cm −2 ( Figure 10 ). The fi lms proved robust under rapid gas evolution and operated with a stable potential of ≈2 V. Analysis of the headspace gas of the electrolyzer using gas chromatography showed that, in all cases, the Faradaic effi ciency for H 2 and O 2 evolution in the cathodic and anodic compartments respectively, was quantitative. For example, water electrolysis with an air-tight FeO x /FeO x cell at 10 mA cm −2 resulted in the passage of a charge of 18 C after 30 min with the production of 98.1 µmol H 2 and 44.7 µmol O 2 , which corresponds to a Faradaic effi ciency of 105.2 ± 8.0% and 95.8 ± 8.8% for H 2 and O 2 , respectively. Figure 10 shows that the all-iron cells outperform an allplatinum analog, both in terms of cell bias and stability. A precious metal system comprising of a Pt cathode and RuO x anode displays the highest activity with a bias potential of ≈1.56 V at I = 10 mA cm −2 , corresponding to a cell voltage effi ciency of 95%. [ 51 ] However, the all-iron equivalent competes strongly at 74% in the context of the superior abundance (at least eight orders of magnitude greater than any precious metal) and global availability.

Electrolysis of Seawater
Discussion of a global energy solution requires careful consideration of the economic and geographical availability of resources required for all components of the system. In this work so far, Adv. Energy Mater. 2016, 6, 1502095 www.MaterialsViews.com www.advenergymat.de pure water has been used with 0.1 M KOH being added as electrolyte. However, in many cases it would be preferable to use seawater without the need for energy intensive and costly desalination, conserving the limited supply of fresh water. [ 52 ] To this end, the water splitting ability of the iron-only cell was tested in an artifi cial seawater based electrolyte, where NaCl (0.6 M ) was added to the KOH (0.1 M ) solution. Headspace gas analysis (H 2 and O 2 ) after 30 min electrolysis of seawater revealed quantitative water splitting with the generation of H 2 :O 2 in 2:1 (and no observable Cl 2 production as competing side-reaction). A charge of 18 C passed through the FeO x /FeO x cell over 30 min with the production of 99.6 µmol H 2 and 53.6 µmol O 2 , which corresponds to Faradaic effi ciencies of 104.9 ± 8.0% and 106.6 ± 8.0%, respectively. Furthermore, the electrochemical response of FeO x in LSV scans in an electrolyte solution containing NaCl remained unchanged at anodic potentials and no additional oxidative signals or a signifi cant shift in the OER wave were observed ( Figure S13, Supporting Information).

Conclusions
Readily scalable iron-only materials are bi-functional electrocatalysts for both water splitting half reactions in alkaline media with activity greater than previously reported bi-functional Co and Ni catalysts. This dual activity is caused by a facile interconversion of an oxide-supported Fe(0) phase active for HER under a cathodic bias into an iron oxide-hydroxide (FeO x ) phase active for OER under an anodic bias. This interconversion of catalytic species at the electrode surface is reversible and allows for regeneration of the catalytic surface upon switching of anode and cathode to enhance stability and lifetime of the electrolyzer system. In situ electrochemical regeneration through cycling of the applied potential with bi-functional electrodes is a novel concept to enhance the stability of a bi-functional electrocatalyst, which is otherwise limited by corrosion processes in aqueous solution. This approach can also help alleviate instability issues more generally associated with electrodes in alkaline electrolyzers under conditions of fl uctuating currents and gas pressures by regeneration of a bi-functional electrode and is particularly relevant in the context of intermittent renewable energy sources such as solar and wind. This strategy is specifi c to bi-functional materials that can catalyze both water splitting half reactions, as otherwise electrolysis would halt during potential cycling. Thus, our fi ndings of an iron-based symmetrical water-splitting system with in-situ electrochemical regeneration represents a step toward the identifi cation of economical, ultra-abundant water splitting catalysts for use in electrolysis and solar-to-fuel devices.

Experimental Section
Materials and Methods : Reagents used throughout this work were obtained from commercial suppliers and used as received. Laboratory grade reagents were used in all synthetic procedures and electrochemical grade reagents were used for analytical measurements. Millipore water (18.2 MΩ cm at 25 °C) was used throughout this work in both synthetic and electrochemical procedures. Solutions were made using analytical grade reagents and titrated to the desired pH, as determined by a pH electrode (Mettler Toledo; SevenEasy) using stock solutions of H 2 SO 4 (Fisher Scientifi c Ltd.; > 95%) and KOH (Sigma-Aldrich; 99.99%).
Electrode Preparation : Briefl y outlined below are the procedures used for synthesis and characterization of the thin fi lm electrode materials. First, FTO coated glass (SnO 2 /F; Sigma-Aldrich; 7 Ω sq −1 sheet resistance) sheets (300 × 300 × 2 mm 3 ) were cut into 1 × 3 cm 2 slides for use as conductive electrode substrates. Prior to fi lm deposition, they were fi rst cleaned by heating them in a solution containing hydrogen peroxide (H 2 O 2 ; 100 vol.; Breckland Scientifi c Supplies), ammonia (NH 3 ; 35% solution, Fisher Scientifi c), and Millipore water in a ratio 1:1:5 at 80 °C for 2 h. The cleaned FTO substrates were rinsed with Millipore water and allowed to dry at room temperature. A 1 × 1 cm 2 area at each end of the slide was covered in a chemical-and heat-resistant epoxy tape (0.14 mm thick; 3M), leaving a central 1 × 1 cm 2 active electrode area for fi lm deposition. The below depositions were optimized for highest-performance with the lowest loading of material. No comment is made as to the effect of thickness on stability or other properties of the materials.
Deposition of FeO x Thin Films : Hydrated iron(III) oxide (FeOOH) was deposited onto the FTO substrate via CBD from an aqueous solution containing 0.15 M iron (III) chloride hexahydrate (FeCl 3 ·6H 2 O; Sigma-Aldrich; ≥ 99%) and 6.25 wt% urea (CON 2 H 4 ; Breckland Scientifi c Supplies; 99%) at 90 °C for 1 h in a low-temperature furnace (Thermofi sher Heratherm). Urea is thermally hydrolyzed in aqueous solution at 90 °C to produce in situ basic conditions, which causes the precipitation of iron oxide species from iron-containing solutions. Following CBD treatment, the electrodes were allowed to cool in situ before rinsing with Millipore water. The resultant yellow fi lms were left to dry in air at room temperature. These fi lms, designated as FeO x , could be converted into various less hydrated iron(III) oxide phases by annealing under air at increasing temperatures in a furnace (Carbolite; ELF11/14B). Throughout this work, FeO x refers to an annealing temperature of 200 °C for 30 min (2 °C min −1 ramp rate).
Deposition of FeS x Thin Films : Iron(II) sulfi de was deposited onto the FTO substrate via a single-step CBD method from an aqueous solution containing 50 × 10 −3 M iron(II) chloride tetrahydrate (FeCl 2 ·4H 2 O; Sigma-Aldrich; 98%), 0.1 M thioacetamide (CSN 2 H 4 ; Sigma-Aldrich; 98%), and 0.1 M urea at 90 °C for 6 h. Sulfi de is released by thermal hydrolysis of thioacetamide, causing nucleation of iron sulfi de onto the substrate. The resultant fi lms, which were black with metallic sheen, were allowed to cool in situ before rinsing with deionized water and methanol, and dried in vacuo. They are designated as FeS x .
Deposition of Fe Thin Films : Metallic iron was electrodeposited onto the FTO substrate using an adapted version of the conventional Adv. Energy Mater. 2016, 6, 1502095 www.MaterialsViews.com www.advenergymat.de ferrous sulfate bath. [ 43 ] Electrodeposition was effected from an aqueous solution containing 10 × 10 −3 M iron (II) ammonium sulfate hexahydrate (FeSO 4 (NH 4 ) 2 SO 4 ·6H 2 O; Sigma-Aldrich; 99%) as source of iron and 1 M ammonium sulfate ((NH 4 ) 2 SO 4 ; Sigma-Aldrich; ≥ 99.0%) as supporting electrolyte at room temperature. A current density of −10 mA cm −2 was applied for 80 s. Coevolution of H 2 from the aqueous electrolyte solution happens rapidly once iron deposition has begun. The resultant electrode, designated Fe, was a metallic gray fi lm. After deposition, the fi lm was rinsed with Millipore water and methanol. For comparison, a pure metallic iron foil (Alfa-Aesar; Puratronic; 99.995%; 0.1 mm thickness) was also used as an electrode material.
Quantifi cation of Fe Content : The amount of iron contained in each type of fi lm was determined using UV-vis spectroscopy. Spectra were recorded on a UV-vis spectrophotometer (Varian Cary 50) in dual-beam absorbance mode. The spectra were background subtracted to exclude any absorbance from the solvent. Disposable plastic cuvettes (Fisher Scientifi c Ltd.; 10 mm path length) were used. The fi lms were dissolved in concentrated hydrochloric acid (HCl; 20 mL; 11.65 M ; Fisher Scientifi c) by submersion of the electrodes, followed by ultra-sonication for 2 h. The iron-containing species, [FeCl 6 ] 3− is formed under these conditions which has two strong absorptions in its UV-vis spectrum ( Figure S14 Figure S14, Supporting Information). The same spectra and calibration is obtained using both Fe(II) and Fe(III) salts, confi rming the universal applicability of this method to all iron-containing fi lms, regardless of the initial oxidation state of iron, since in this concentration range, all Fe(II) is aerially oxidized to Fe(III) in conc. HCl. The average Fe content of each fi lm type was determined using this method and the values are given in Table S1 (Supporting Information).
Deposition of Pt Thin Films : Metallic platinum was electrodeposited onto the FTO substrate using an aqueous solution containing 10 × 10 −3 M potassium tetrachloroplatinate (II) (K 2 PtCl 4 ; Precious Metals Online) as the source of platinum and 0.3 M sodium sulfate (Na 2 SO 4 ; Sigma-Aldrich; ≥ 99.0%) as supporting electrolyte at room temperature. The deposition was effected by running cyclic voltammetry from 0 to −1 V vs. Ag/AgCl/KCl(sat.) for two cycles at 200 mV s −1 after purging of the solution with N 2 for 15 min. The resultant electrode, designated Pt, was a metallic gray fi lm. After deposition, the fi lm was rinsed with Millipore water and methanol. XRD, SEM, and LSV data for this Pt electrode are available in the Supporting Information (Figures S15-S17).
Deposition of RuO x Thin Films : Hydrated ruthenium(IV) oxide was electrodeposited onto the FTO substrate using procedures adapted from the literature. [ 53,54 ] An aqueous solution containing 5 × 10 −3 M RuCl 3 · x H 2 O (Precious Metals Online) and 0.1 M KCl (Sigma-Aldrich; ≥ 99.0%) and 10 × 10 −3 M HCl as the supporting electrolyte at 50 °C was used. The deposition was effected by running cyclic voltammetry from −0.2 to 1 V versus Ag/AgCl/KCl(sat.) for 100 cycles at 50 mV s −1 . The resultant electrode, designated RuO x , appeared as a brown fi lm and was rinsed with Millipore water and methanol. RuO x could be converted into various less hydrated ruthenium(IV) oxide phases by annealing under air at increasing temperatures. Throughout the remainder of this work RuO x , a gray-black fi lm, refers to an annealing temperature of 300 °C for 3 h (2 °C min −1 ramp rate). XRD, SEM, and LSV data for this RuO x electrode are available in the Supporting Information (Figures S15-S17).
Physical Characterization : The surface morphologies and thicknesses (Table S1, Supporting Information) of the various thin fi lm electrodes were characterized using SEM (Phillips XL-30-SFGE). A freshly cut and untouched edge was used for cross-sectional views and elemental analysis was also undertaken at the same time using an Xsight light element EDX spectrometer (Phillips XL-30-SFGE). Phase analysis of the various thin fi lms was carried out using XRD (X'Pert PRO, PANalytical BV). A slow scan rate of 0.5° min −1 was used to resolve the peaks of the thin fi lm against the background of the crystalline FTO substrate. Structures were referenced against those registered in the International Crystal Structures Database (ICSD) using X'Pert HighScore software (PANalytical). XPS spectra were obtained at the National EPSRC XPS User's Service (NEXUS) at Newcastle University, UK, an EPSRC Mid-Range Facility. Analysis was performed using a K-Alpha (Thermo Scientifi c, East Grinstead, UK) spectrometer utilizing a monochromatic Al Kα X-ray source (1486.6 eV, 400 µm spot size, 36 W). Survey spectra were collected with a pass energy of 200 eV and 3 sweeps, while high-resolution spectra were collected at a pass energy of 40 eV with 10 sweeps. Measurements were taken at 3 points on each sample surface to ensure consistency and charge neutralization was used throughout the analysis. Spectral analysis and peak fi tting was carried out using CasaXPS software. The C1s (adventitious) peak was used as calibration standard.
Electrochemical Methods : All electrochemical measurements (voltammetric and transient) were taken at room temperature (20-25 °C) using an Ivium ComptactStat potentiostat (Ivium Technologies BV) controlled by IviumSoft software. Various thin fi lm coated FTO-glass slides were prepared, as described above, and used as working electrodes. A fresh working electrode was used for each experiment. An aqueous Ag/AgCl/KCl(sat.) reference electrode (E(Ag/AgCl/KCl(sat.) = 0.197 V vs NHE) and a platinum mesh counter electrode were employed when required. The aqueous media contained either 0.1 M H 2 SO 4 at pH 1, 0.1 M Na 2 SO 4 (Sigma Aldrich; >99.0%) and triethanolamine (TEOA; Sigma-Aldrich; > 99.0%) at pH 7 or 0.1 M KOH at pH 13. Potentials are reported against the RHE using the Nernstian relation: E (V vs RHE) = E (V vs Ag/AgCl/KCl sat. ) + 0.197 + (0.059 × pH). Uncompen sated cells resistance, or Ohmic drop, equal to iR , where i is the current fl owing and R the cell resistance, was compensated for by inclusion into the potential calculation.
Electrical Impedance Spectroscopy (EIS) was used to determine the quantity R using the EIS function in the Ivium CompactStat. Impedance values were determined at the open circuit potential for each measurement. For the single compartment cell, the value of uncompensated resistance was typically in the range 20-30 Ω, whereas that in the larger two-compartment cell was in the range 80-120 Ω in the 0.1 M KOH electrolyte solution.
Electrochemical methods including LSV (1 mV s −1 scan rate) and chronopotentiometry were carried out using a single-compartment cell and a conventional three-electrode setup. Reaction vessels were purged with inert gas, (N 2 ; BOC; 99.998%) for at least 15 min prior to experiments to remove atmospheric gases, namely O 2 , which is potentially electroactive in the cathodic regime. Full water splitting electrolysis cells were run using a two-electrode setup with the cathode employed as the working electrode and the anode as the counterreference electrode. A two-compartment cell fi tted with a Nafi on 117 membrane was used to separate the headspaces of the cathodic and anodic compartments.
Product Quantifi cation : The gaseous products (H 2 and O 2 ) were quantifi ed by headspace gas analysis using a gas chromatograph (Agilent 7890A Series GC) with a 5 Å molecular sieve column (held at 45 °C) and a thermal conductivity detector. For H 2 and O 2 detection, N 2 and He were used as the carrier gas, respectively. The electrochemical cell was purged with 2% CH 4 in N 2 (BOC; 99.9995%; CH 4 is internal GC standard) for at least 15 min prior to experiment. ≈20 µL aliquots of headspace gas were removed using a gas-tight syringe (Hamilton; GASTIGHT). Measurements were taken in triplicate and a minimum error of 8.0% was applied.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author. Additional data related to this publication is available at the University of Cambridge data repository (https://www.repository. cam.ac.uk/handle/1810/252751).