Ultrathin Hematite‐Hercynite Films for Future Unassisted Solar Water Splitting

Photoelectrochemical (PEC) water splitting requires stable, efficient, and cost‐effective photoelectrodes to enable future large‐scale solar hydrogen production. Ultrathin hematite‐hercynite photoanodes that meet all these criteria in an excellent way is presented here. Hematite‐hercynite photoelectrodes are synthesized in a self‐forming manner by thermal oxidation of iron–aluminum alloy films and characterized with regard to water splitting applications. Photoanodes fabricated from 17 wt.% Al at 493 °C for 8 h and 685 °C for 5 min exhibit, for instance, a photocurrent density of 1.24 and 1.53 mA cm−2 at 1.23 V versus RHE, respectively, as well as superior light absorption in the visible range of the solar spectrum. The PEC performance improvement in comparison to pure hematite thin film electrodes is first achieved by adjusting the aluminum concentration with an optimum range of 12–17 wt.% and second by optimizing the annealing conditions. The resulting photocurrent densities are about a factor of three higher than those obtained from electrodes synthesized from pure iron thin films using the same synthesis conditions. Finally, it is shown that ultrathin hematite‐hercynite photoelectrodes enable even unassisted solar water splitting in a NaOH (1 m) electrolyte with a maximum solar‐to‐hydrogen conversion efficiency of 0.78%.


Introduction
Solar energy conversion, for instance, to electricity or to hydrogen, is one promising strategy to cope with the global energy demands and dwindling resources of fossil fuels.Photoelectrochemical (PEC) water splitting is, in this regard, one intriguing and promising strategy for renewable and large-scale solar hydrogen production.[3] In addition to high light absorption, the DOI: 10.1002/admt.202300655semiconductor materials used must have high conversion efficiency and must be stable within the PEC reactions and generally in contact with the electrolyte.Heretofore, there is no single material that can fulfill all these requirements for market-compatible hydrogen production.For instance, impressive solar-tohydrogen conversion efficiencies were achieved using PEC devices comprising multijunction III−V semiconductors [4] Electrodes made from such materials can generate sufficient photovoltage, but they suffer from stability issues under operation, and their production is relatively complex, associated with environmental issues and related costs.7] Herein, we present a new strategy based on a hematite interface that overcomes the current limitations of hematite-based photoanodes.In general, hematite is an earth-abundant n-type semiconductor with a bandgap of 1.9-2.2eV and highly stable in aqueous solutions that have a pH value higher than three. [6,8,9]espite these promising properties, hematite suffers from major drawbacks, as discussed and summarized in Refs.[6,8].][11] Unintentionally doped hematite also suffers from an overall low charge carrier concentration and can even be considered insulating [12] Therefore, oxygen vacancy introduction and various synthesis and elemental doping strategies were explored to improve the conductivity.14] Thermal oxidation of iron represents the most straightforward method for hematite photoanode synthesis, but this method mainly yields hematite with poor PEC performance. [9,15]However, contrary to the abovementioned strategies, our approach relies not just on doping but on the implementation of a new phase called hercynite by oxidizing iron-aluminum alloys.The idea to implement iron-aluminum alloys goes back to 1931 when these materials were introduced and studied for applications in mechanics and civil engineering [16] Hauttmann reported in this context a remarkable improvement of the heat resistance of steel by adding 4-9 at.% of Al, [16,17] which can be assigned to a significantly lowered oxidation rate of iron due to the pronounced oxygen affinity of aluminum and the formation of ternary Fe─Al─O phases.The thermal oxidation for aluminum concentrations higher than 0.5 at.% in iron at temperatures below 570 °C yielded three surface oxide phases comprising a top layer of hematite (-Fe 2 O 3 ), followed by an underneath magnetite (Fe 3 O 4 ) layer and a bottom layer of hercynite (FeAl 2 O 4 ) adjacent to the non-oxidized metallic alloy.From this study, it can be assumed that hematite should still be the electrode surface for photoelectrodes made from iron-aluminum alloys.Nevertheless, due to the introduction of hercynite in the electrode volume, changes in the electrical conductivity and the optical properties and, consequently, changing the entire PEC performance must be expected.However, respective studies are missing, which motivated us to explore this material system.
As shown and discussed in the following, we observed a significant improvement in the PEC performance indeed, resulting from improved charge carrier concentration, as well as improved optical absorption.The PEC performance of hematite photoanodes is generally strongly dependent on the oxidation parameters. [9,18]A relatively high oxidation temperature and an extended oxidation time appear both beneficial for fabricating efficient hematite photoanodes. [9,18]For example, Sivula et al. [18] reported a photocurrent of 0.56 mA cm −2 at 1.23 V that was achieved from hematite synthesized at 400 °C for 10 h followed by annealing at 800 °C for 20 min to trigger tin diffusion from th fluorine-doped tin oxide (FTO) substrate into hematite for doping purposes.However, annealing at 800 °C must be considered undesirable because of the instability of many glass substrates, and the overall limitation to material systems that can withstand high temperatures.Relatively lower oxidation temperatures, such as ≈630 °C, are generally favorable but require, for instance, 6 h of oxidation as reported elsewhere [9] However, we aim to fabricate efficient photoanodes at lower oxidation temperatures or in a shorter time by using iron-aluminum alloys.For the electrode synthesis, we used thermal oxidation of ≈100 nm thick iron-aluminum alloy thin films in ambient air following the so-called direct annealing method as described in ref. [9].In our study, either relatively high annealing temperatures for very short annealing time or a low annealing temperature for a longer annealing time were examined and correlated with the electrode PEC performance.The highest PEC performance was achieved by photoanodes fabricated from ironaluminum films (17 wt.% Al) using 493 °C for 8 h and 685 °C for 5 min showing, respectively, a photocurrent density of 1.24 and 1.53 mA cm −2 at 1.23 V versus RHE, as well as an oxygen evolution reaction onset of 0.7 and 0.6 V versus RHE.These photocurrent densities are about a factor of three higher than those obtained from pure hematite photoanodes synthesized and tested under the same conditions, which is attributed to the enhanced light absorption and the increased charge carrier concentration.We observed in addition even unassisted solar water splitting in a 1 m NaOH electrolyte with a low, but noticeable, maximum solar-to-hydrogen (STH) conversion efficiency of 0.78%.

Photoanode Synthesis and Characterization
Iron-aluminum alloy thin films with a thickness of (100 ± 10) nm were first deposited on fluorine-doped tin oxide (FTO) covered glass substrates using thermal evaporation.The aluminum concentration was varied from 0 wt.% for pure hematite photoelectrodes, up to 17 wt.%targeting iron-rich alloy phases (Fe 3 Al).Higher aluminum concentrations were not studied due to the expected formation of aluminum-rich phases based on the Fe─Al binary phase diagram, [19] and associated increased risk for insulating aluminum-oxide formation.The phase composition, the aluminum concentration, and the elemental distribution within the deposited thin films were characterized prior to the oxidation process, as discussed in detail in the "Supporting Information".The as-deposited thin films were thermally oxidized in ambient air at various oxidation conditions.After thermal oxidation, the elucidation of the phase composition and the overall chemical composition is required.X-ray diffraction (XRD) patterns of pure hematite and aluminum-modified photoanodes fabricated from pure iron and iron-aluminum thin films, respectively, at different annealing conditions, are shown in Figure 1 (Figure S1, Supporting Information for full XRD scan).In the case of pure hematite photoanodes, as shown in Figure 1a, XRD patterns show the expected main peaks corresponding to (104) and (110) planes, as well as weaker peaks corresponding to (012), ( 113), (024), ( 116), (214), and (300) planes (Figure S1, Supporting Information).This pattern is well-aligned with the respective standard powder pattern (Figure S2, Supporting Information).The same pattern was measured for all oxidation conditions and no other oxide phases were detected, confirming the full transformation of iron into hematite.In the case of aluminum-modified photoanodes, all of the aforementioned weak peaks from hematite were, however, not detected (Figure S1, Supporting Information).As shown in Figure 1b,c, a strong preferential crystal orientation of the [104] axis at the expense of [110] normal to the FTO substrate surface plane is observable.This orientation is not favored because of the strong anisotropic conductivity of hematite, where the conductivity within (001) planes (i.e., [110] direction) is four times higher than, for instance, along a [001] direction as discussed by Iordanova et al. [20] This unfavored crystal orientation could hence yield hematite with poor electrical conductivity.Like iron, low-content aluminum alloys (<9 wt.%) were fully transformed into hematite.Thus, those photoanodes will be called aluminum-modified in the following.Figure 1b shows this full transformation into hematite at all studied annealing conditions, except for annealing at 685 °C for 5 min, where magnetite might also be present in a minute concentration, exemplarily for Fe─Al (9 wt.%).In the case of Fe─Al (17 wt.%), as shown in Figure 1c, an oxidation time of one h at 685 °C suffices to fully transform the alloy into hematite, similar to Fe─Al (9 wt.%).However, weak peaks of hematite, as well as a relatively sharp and strong peak at around 44.18°, are observed for the other oxidation conditions (see Figure 1).This distinct peak at around 44.18°, detected at a diffraction angle higher than the metal peak that solely exists prior to oxidation (44.06°), could be assigned to the hercynite phase (FeAl 2 O 4 ) among the other iron oxide phases (Figure S2, Supporting Information).The low intensity of hematite peaks in- dicates its low concentration or film thickness compared to the hercynite phase.The two additional peaks at 30.6°and 32°that most likely correspond to the wüstite phase (Figure S2, Supporting Information) were detected from alloys oxidized at 685 °C for 5 min.
According to literature, as discussed earlier, the oxidation of a Fe─Al alloy at temperatures below 570 °C forms an outer layer of hematite, magnetite in the middle, and hercynite underneath next to the metal [17] The hercynite phase forms at the expense of magnetite and provides an improved barrier to the diffusion of iron and hence improves also the oxidation resistance (heat resistance) [17] Thermodynamically, hematite does not coexist next to the metal phase without magnetite or hercynite in between.This theory is also supported by the presence of the wüstite phase observed from the oxidation at 685 °C for 5 min.Wüstite forms from iron at oxidation temperatures above 570 °C and could exist underneath hematite only with a magnetite phase in-between, as discussed in previous reports [21] Therefore, we could conclude that the sharp peak at 44.18°seems to correspond to the hercynite phase and denote the corresponding samples in the following hematite-hercynite photoelectrodes.and denote the corresponding samples in the following hematite-hercynite photoelectrodes.
Surface analysis of samples oxidized for 1 h at 685 and 493 °C (Figure S3, Supporting Information) by X-ray photoelectron spectroscopy (XPS) could indeed confirm the presence of the assumed hematite top layer, which is also true for all studied aluminum concentrations (here up to 17 wt.%).Aluminum was detected basically only for the higher synthesis temperatures as aluminum oxide at the surface, but not to a significant concentration.To investigate the oxide phase formation for the abovementioned set of samples, a depth profile across 180 nm was created using an Ar sputter gun and XPS targeting the oxygen 1s peak, as shown in Figure 2. The measurement before sputtering was not used for the analysis to avoid signals originating from ambient surface contaminations, such as oxygen-carbon and oxygen-hydrogen bonds.The oxygen 1s peak can be synthesized by assuming two oxide phases (peaks 1 and 2) for both annealing conditions, as shown in Figure 2a,b.According to the literature, oxygen has a slightly higher binding energy in aluminum oxide compared to the hematite phase.Iron-aluminum oxides (e.g., hercynite) have, on the contrary, significantly higher oxygen binding energies than other phases [22] As shown in Figure 2c, the binding energy of peak 1 is at 529.5 ± 0.1 eV and does not change substantially with the depth for both samples.This binding energy fits to values reported for hematite that was fabricated from pure iron oxidation [9] The peak 2 position is, however, dependent on the used synthesis temperature and the depth.For the higher oxidation temperature of 685 °C, the peak 2 binding energy is 530.1 eV and only shifts slightly with increased sputtering time to ≈530.5 eV, which thus results in a relatively constant peak ratio (cf. Figure 2d).The observed shift could also be due to slight variations within the oxide phase.Nevertheless, these values are still aligned with the values reported for aluminum oxide, which is also expected to form at this temperature and in agreement with the above-discussed XRD results.For the photoanode fabricated at 493 °C, the initial peak two position is 530.5 eV, similar to the higher temperature but shifts significantly from 60 to 180 nm depth from 531.1 to 531.2 eV.The significant shift to higher binding energies indicates a new phase of iron-aluminum-oxide, which appears to be aligned with the reported value for the hercynite phase [22] The ratio of peak 2 to peak 1 increases with the depth of the photoanode, which indicates the predominance of hercynite with depth and lower concentration of hematite can be still detected.Further in-depth material analysis of the film composition with respect to the hercynite phase and its interface to hematite, for instance by high-resolution transmission electron microscopy, are subject of future studies.
Aside from the film composition, the overall PEC performance of the electrodes depends also strongly on the optical behavior of the photoelectrodes.Figure 3 shows, in this regard, the light absorption of the fabricated photoanodes in the range 350- 800 nm as a percentage of the incident light (so-called absorptance) using a backside configuration for more accurate results, as described in the Experimental Section.It is worth mentioning here that all the fabricated photoanodes showed an absorptance slightly higher than 100% in the range below 500 nm, as shown in Figure 3.This referred to the reflection difference between the photoanodes and the FTO substrate, where the photoanodes showed lower reflection than the substrate used for the spectrometer calibration (Figure S4, Supporting Information).
Pure hematite photoanodes showed, regardless of their annealing conditions, a comparable optical absorption due to their full and single-phase transformation, as discussed earlier (Figure 3).The average thickness of the hematite thin films was also calculated to be 223 ± 10 nm using light absorbance.However, the thickness of the resulting hematite top layer for the hematite-hercynite photoelectrodes cannot be measured in this way due to the presence of hercynite.
The effect of the hercynite phase and aluminum addition is, regardless of the oxidation temperature and time, in particular, visible for a wavelength higher than 600 nm, which is beyond the bandgap of hematite (≈2.2 eV or 564 nm).The absorptance increased with increasing aluminum concentration, and the highest absorption was achieved by annealing sam- ples with higher aluminum concentrations (12 and 17 wt.%)at 685 °C for 5 min and 493 for 6 h, as shown in Figure 3b,c, respectively.
The improved light absorption can be mainly attributed to the formation of hercynite, which belongs to the spinel structure family, thus exhibiting a bandgap change in dependence on the cation distribution.Several bandgap values were therefore reported for hercynite, for instance, 1.78 [23] and 3.14 eV [24] To evaluate the contribution from hercynite further, incident photon-to-current efficiency (IPCE) measurements, for photoanodes oxidized at 685°C for 5 min, were employed to measure the photocurrent density as a function of incident photon wavelength (Figure S5, Supporting Information).For an incident wavelength of 600 nm, no photocurrent was detected from pure hematite samples (0% Al), as expected by assuming a bandgap of about 2.2 eV.A photocurrent was measured accordingly from a hematite-hercynite sample (17% Al), indicating an additional photoelectric element in addition to hematite.
As the PEC performance was characterized by illumination from the front (hematite side) rather than the backside, the optical behavior of the photoanodes was furthermore evaluated using the frontside configuration, which led to comparable results (Figure S6, Supporting Information).

Photoelectrochemical Characterization
Photoanodes with higher aluminum concentration and, therefore, higher light absorption are expected to exhibit an improved PEC performance supported by the aforementioned IPCE measurements.To study the PEC performance, linear sweep voltammetry (LSV) under dark and under AM 1.5G (100 mW cm −2 ) illumination conditions were realized.
First, the PEC performance of the photoanodes fabricated at 685°C for 1 h is discussed in dependence on the aluminum concentration, as shown in Figure 4a.As shown in Figure 4a, the best performance with a maximum photocurrent density of 0.55 mA cm −2 at 1.23 V versus RHE was notably achieved from pure hematite.This means that the PEC performance was in this case negatively affected by aluminum addition.This effect, as also discussed later, can be attributed to the insulator formation aluminum oxide and to an unfavored hematite crystal orientation which, as already mentioned, leads to hematite with poor electrical conductivity.A similar effect explained by the segregation and accumulation of insulating alumina at the electrode surface was reported by Kleiman-Shwarsctein et al. for hematite fabricated by electrodeposition followed by calcination at 600 °C for four hours [25] Their highest electrode performance was achieved by hematite doping with 0.46 at.% (0.22 wt.%) aluminum, while the highest aluminum concentration they studied was 10 at.% (5.09 wt.%), which showed the lowest performance.
By reducing the oxidation time, segregation effects might be suppressed.Pure hematite oxidized at 685°C for only 5 min showed an overall comparable but slightly reduced photocurrent density of 0.47 mA cm −2 at 1.23 V versus RHE (Figure 4b).While the PEC performance was not significantly affected by aluminum addition of up to 5 wt.%Al, a drastic improvement was achieved for higher aluminum concentrations reaching a maximum photocurrent of 1.53 mA cm −2 at 1.23 V versus RHE in the case of 17 wt.%Al addition, which is about a factor of 3 higher than the value recorded for pure hematite.This observation is in alignment with the optical measurements, as shown before in Figure 3b.Hematite-hercynite photoanodes (17 wt.% Al) also exhibited the lowest oxygen evolution reaction onset at around 0.6 V versus RHE.It is worth mentioning that the performance of a hematite-hercynite photoanode with 12 wt.%Al is only slightly lower than those achieved with 17 wt.% of Al.Based on the optical behavior (cf. Figure 3c), a reduced oxidation temperature of 493 °C (6 h) appears promising but resulted in a lower photocurrent of only 0.36 mA cm −2 at 1.23 V versus RHE for pure hematite and adding up to 5 wt.% exhibited again negative effects on the PEC performance, as shown in Figure 4c.Nevertheless, a drastic improvement was again achieved for hematite-hercynite photoanodes with a maximum photocurrent density of 1.04 and 1.07 mA cm −2 at 1.23 V versus RHE for 12 and 17 wt.%aluminum concentration, respectively.Although convincing PEC performance was obtained, oxidation times were varied to investigate their effects (Figure 4d).An increase of the oxidation time to 12 h improved the photocurrent of pure hematite slightly to 0.41 mA cm −2 at 1.23 V versus RHE, which was equally true for samples with 17 wt.%Al.Nevertheless, 8 h of oxidation was already sufficient and even resulted in the maximum photocurrent of 1.24 mA cm −2 at 1.23 V versus RHE.
Reducing the temperature down to 393 °C was also explored but resulted, unfortunately, in a higher oxygen evolution reaction onset of ≈1 V versus RHE (Figure S7, Supporting Information) and, consequently, in a decreased PEC performance.Furthermore, the surface morphology of hematite photoanodes showed a minor dependence on the aluminum concentration, as well as the annealing conditions.Oxidation of pure iron thin films resulted in a granular morphology with grain diameters ranging from about 120 down to 30 nm, whereas oxidation of aluminum containing iron alloy thin films yielded overall comparable morphologies but with larger grains and sometimes with a slightly porous appearance (Figures S8 and S9, Supporting Information).Due to the smaller grain sizes, the surface roughness of the pure hematite appears to be higher than that of the photoanodes fabricated from aluminum containing iron alloys.However, the overall surface roughness of the fabricated photoanodes is still comparable to some extent and has only a minor impact on the PEC performance.Therefore, the influence of surface roughness on the PEC performance is here excluded.
The measured PEC performance is intrinsically linked to the electrical conductivity of the photoelectrodes and aligned with the charge carrier concentration that was determined by means of Mott-Schottky (MS) measurements (Experimental Section).First, the n-type conductivity was confirmed for all types of photoanodes, which is indicated by the positive slopes in Figure 5.For pure hematite electrodes, higher and longer oxidation temperature and time, respectively, provides an improved PEC performance.Consequently, oxidation of pure iron films at 685 °C for 1 h results in a charge carrier concentration of 9.1 × 10 19 cm −3 , 685 °C for 5 min results in 0.9 × 10 19 cm −3 , and 493°C (12 h) in 0.1 × 10 19 cm −3 .[28] For hematite-hercynite samples (17 wt.% Al), the lowest charge carrier concentration of 0.02 × 10 19 cm −3 was achieved by the annealing at 685°C for 1 h (Figure 5b), which is well aligned with the observed poor PEC performance (cf. Figure 4a).However, a significantly increased charge carrier concentration of 2.0 and 10.0 × 10 19 cm −3 was accordingly achieved by annealing at 493 °C for 12 h and 685°C for 5 min, respectively, which are notably higher than the values obtained for pure hematite and aligned well with the observed PEC performance.As mentioned earlier, tin diffusion from the FTO substrates requires high annealing temperature, e.g.800 °C, which is significantly higher than the annealing temperatures used here.However, even if tin would diffuse into the photoanodes already at lower temperature to a noticeable extent, the effect is assumed to be comparable due to the use of the same temperatures and oxidation conditions.The flat band potential was within 0.15-0.22V versus RHE, which is significantly lower than those achieved for pure hematite photoelectrodes.Although the photoanodes fabricated at 685 °C for 1 h showed the lowest flat band potential, their PEC performance is still poor due to the comparatively low charge carrier concentration.
In addition to PEC performance, long-term stability in the electrolyte under PEC operation is also a crucial aspect.The stability of the hematite-hercynite photoelectrodes is expected to be similar to pure hematite due to the self-formation of a pure hematite top layer surface, as discussed in the previous section.We previously reported superior stability studied for up to 1000 h for pure hematite photoelectrodes [9] To evaluate the long-term stability, hematite-hercynite photoanodes (12 wt.% Al, 685 °C 5 min −1 ) were tested using the same electrolyte and illumination at a significantly high anodic potential of 1.5 V versus RHE for more than 64 h (Figure S11, Supporting Information).The PEC performance of the photoanode was only slightly reduced after 64 h of the stability test (Figure S11b, Supporting Information)  N d ) is calculated from the slope of a linear fit, whereas flat band potentials (U fb ) from the x-axis intercept (see "Experimental Section").As indicated in the graphs, some data were multiplied by a factor of 10 or 100 for scaling or plotting reasons without affecting the results.

Toward Unassisted Solar Water Splitting
An important parameter for the future use of photoelectrodes in the field of solar water splitting is the achieved solar-to-hydrogen (STH) efficiency.The STH efficiency was evaluated from the short-circuit photocurrent density using chronoamperometric measurements in a 1 m NaOH electrolyte for hematite-hercynite photoanodes (17 wt.% Al) that showed so far the highest PEC performance.The measurement was performed using the common two-electrode configuration with platinum wire counter electrode. [29]The Faradaic efficiency used for the STH evaluation is assumed to be 100%, based on previously reported results for pure hematite photoelectrodes synthesized in the same manner, [9] i.e., the photocurrent resulted from oxygen evolution reaction without any parasitic side reactions.
The STH efficiency was first calculated from the maximum photocurrent density showing a maximum STH of 0.54 and 0.78% for hematite-hercynite photoanodes fabricated at 493 °C for 12 h and 685 °C for 5 min, respectively (Figure 6a).Due to the sluggish oxygen evolution reaction, the photocurrent is expected to degrade, being limited by the electrolyte rather than the photoelectrode.However, the STH at the steady photocurrent is 0.17% and 0.19% for the photoanodes described above.The measured photocurrent indicates an unassisted oxygen evolution reaction.Nevertheless, the evolved gases must be still analyzed to confirm the water splitting reaction, hence a subject for future work.To demonstrate the limitation due to the electrolyte, a proper scavenger such as H 2 O 2 for the photogenerated holes was used in a mixture of 1 m NaOH and 0.5 m H 2 O 2 with the same volume ratio as shown in Figure 6b.A maximum STH of 1.10 and 1.23% is obtained from photoanodes fabricated at 493 °C for 12 h and 685 °C for 5 min, respectively.Although the STH efficiency cannot be calculated when a sacrificial electrolyte is used, it provides the maximum STH that can be achieved by the photoelectrode so far without the aforementioned electrolyte limitation.As discussed in several reports, [30][31][32][33][34][35] the charge transfer of the photogenerated holes can be enhanced, and consequently an enhancement of the photocurrent, by implementing a suitable co-catalyst.However, to the best of our knowledge, this is the first time an unassisted solar water splitting is presented using such a simple self-formed photoelectrode based on an earth-abundant material system as hematite-hercynite.

Conclusion
We presented a fabrication method for the self-formation of ultrathin hematite-hercynite photoanodes for PEC water splitting applications.We showed that thermal evaporation of 100 ± 10 nm thick iron-aluminum alloys followed by controlled thermal oxidation in ambient air yield hematite-hercynite photoelectrodes with enhanced light absorption and enhanced PEC efficiency.The fabricated photoelectrodes were thoroughly characterized with respect to the phase and chemical composition as well as with respect to the optical and PEC performance.The highest PEC performance was achieved for electrodes based on 17 wt.%Al and that were annealed at 685 °C for 5 min.These electrodes showed a high photocurrent density of up to 1.53 mA cm −2 at 1.23 V versus RHE and an oxygen evolution reaction onset as low as 0.6 V versus RHE.For lower temperatures of 493 °C, 17 wt.%Al provided after thermal oxidation for 8 h a photocurrent density of 1.24 mA cm −2 at 1.23 V versus RHE, as well as an oxygen evolution reaction onset of 0.7 V versus RHE.This means that the photocurrent densities are about a factor of three higher than those obtained from pure hematite thin film photoelectrodes.Comparable PEC performance was achieved from 12 wt.%Al, which increases the tolerance for optimum aluminum concentration (12-17 wt.%).
The PEC performance was also well-aligned with the measured charge carrier density.We showed furthermore that unassisted solar water splitting in 1 m NaOH electrolyte with a maximum solar-to-hydrogen (STH) conversion efficiency of 0.78% can be demonstrated using a self-formed earth-abundant photoelectrode fabricated using hematite-hercynite (e.g., 17 wt.%Al) photoelectrodes.Although the performance must still be further improved, we showed that hercynite is an intriguing material for photoelectrodes also with respect to the future of unassisted solar water splitting.

Experimental Section
Hematite Photoanodes Fabrication: Fluorine-doped tin oxide (FTO) coated glass was used as substrate (TEC 8, Ossila Ltd).The substrates were cleaned with a 5% solution of Tickopur R33 in an ultrasonic bath for 5 min and rinsed with deionized water.The substrates were thereafter cleaned with ultrasound support with acetone and isopropanol for 5 min in each, rinsed with deionized water, and finally dried with gaseous nitrogen.
Thin films of iron and iron-aluminum alloys were deposited from iron granules (99.98%) and an aluminum wire (99.99%) onto the freshly cleaned FTO-glass substrates by thermal evaporation (LEYBOLD AG, L560) equipped with a quartz crystal microbalance to monitor the film thickness increase.Iron and aluminum were placed in a tungsten boat (Umicore Thin Films AG) in the required weight fraction and melted to ensure alloying prior to the deposition on FTO.
The thermal oxidation (annealing) in ambient air of the deposited thin films was achieved using a horizontal quartz tube furnace with open tube ends (Linn High Therm).The length and inner diameter of the quartz tube were ≈48 cm and ≈2.1 cm, respectively.The samples were placed on a ceramic boat and inserted into the center of the quartz tube.The furnace temperature was monitored and controlled by two thermocouples that measure the temperature of the quartz tube itself and the temperature of the sample.The latter one was placed directly above the sample to monitor the actual oxidation temperature during the oxidation process with high accuracy ± 1 °C (Figure S12, Supporting Information).The oxidation process was realized using so-called direct annealing, as described in (Ref.[9]).In brief, the samples were inserted into the preheated furnace only after the furnace reached the desired oxidation temperature.After the annealing time, the samples were directly taken out from the furnace and quenched on a piece of copper.
Characterization Methods: The morphology of the hematite photoanodes was studied using a scanning electron microscope (SEM, Hitachi S-4800) at 10 kV accelerating voltage.SEM/EDX (10.2 kV accelerating voltage) was used for measuring the aluminum concentration in the fabricated alloy thin films as well as its distribution (Supporting Information).The achieved aluminum film concentration compared to the mixing ratio in the evaporation boat is summarized in Table S1 (Supporting Information).The crystal structure of the fabricated samples before and after the oxidation was studied by X-ray diffraction (XRD) in a Siemens D5000 using a step size of 0.02°and 15 s per step and a copper anode with K  = 0.1540 nm.X-ray photoelectron spectroscopy (XPS) was realized using a SPECS SAGE HR 150 system equipped with a 1D delayline detector using a monochromatic Al K  radiation (1486.7 eV) at an emission angle of 0°a nd takeoff angle of 90°.Sputtering was performed using an IQE12/38 ion source with argon plasma at an argon gas pressure of 2.2 × 10 −3 mbar and under application of 3 kV accelerating voltage.
The light absorption of the fabricated photoanodes was determined by measuring the total transmission as well as the total reflection using a Cary 5000 UV-vis spectrometer.An internal integrating sphere was used for both transmission and reflection measurements to collect the scattered light.To measure the light absorption by the photoanodes accurately, the measurements were mainly performed from the backside of the photoanodes (backside configuration).The optical behavior of the photoanodes was also characterized from the frontside (frontside configuration) as described further in the Supporting Information.
Photo-and Electrochemical Characterization: Linear sweep voltammetry (LSV) was employed to evaluate the PEC performance in dark and under illumination conditions.A three-electrode cell (one compartment) consisting of the hematite sample (working electrode), a platinum wire (counter electrode, 1 mm), and an Ag/AgCl (3 m KCL) reference electrode were used for the photoanode characterization.The hematite working electrode was pressed against a nitrile rubber O-ring situated on the back of the cell, allowing an accurate electrolyte contact area of 0.5 cm 2 and an irradiation area of 0.38 cm 2 .A solar simulator (LOT Quantum Design GmbH), additionally equipped with a KG-3 heat absorbing filter (SCHOTT AG), was used for illumination with a power density of 100 mW cm −2 (1 Sun).The working electrodes were illuminated from the frontside of the cell through a quartz window with a distance of 17 mm between the sample and the quartz window.The light power density was calibrated using a standard calibrated solar cell and a self-made calibration cell consisting of a quartz window glass and 1.7 cm optical path of electrolyte (similar to the measurement cell).An electrolyte of 1 m NaOH (pH 13.6) was used, and all

Figure 1 .
Figure 1.X-ray diffraction (XRD) patterns of 100 nm Fe─Al alloys a) 0% b) 9%, and c) 17 wt.%before (grey patterns) and after oxidation at 493 °C for 1 h (black patterns), 493 °C for 12 h (magenta patterns), 685 °C for 5 min (orange patterns), and 685 °C for 1 h (red patterns).Peaks are marked as indicated in the graph.Standard powder XRD patterns of oxide phases are shown in Figure S2 (Supporting Information).

Figure 2 .
Figure 2. High-resolution oxygen 1s spectra of photoanodes fabricated from 100 nm Fe─Al alloys 17 wt.%by annealing for 1 h at a) 685 °C (solid triangle) and b) 493 °C (solid circle).XPS spectra were collected after Ar sputtering for 10, 30, 60, 120, and 180 min.The orange and blue curves represent the fit peak 1 and peak 2, respectively.c) peak position and d) peak ratio of fitted oxygen 1s peaks as a function of sputtering time for both samples.The FWHM of peak 1 and peak 2 is 1.3 ± 0.1 and 2.5 ± 0.1 eV, respectively.The sputtering rate is ≈1 nm min −1 .XPS survey and high-resolution C 1s, O 1s, Al 2p, and Fe 2p spectra of both samples before sputtering are shown in Figure S3 (Supporting Information).

Figure 3 .
Figure 3. Absorptance of photoanodes fabricated from 100 nm Fe─Al alloys with 0, 5, 9, 12, and 17 wt.%aluminum as indicated by the markers by oxidation at a) 685 °C for 1 h, b) 685 °C for 5 min, and c) 493 °C for 6.The total transmittance and total reflection of the same set of samples are shown in Figure S4 (Supporting Information).Scale bars of insets: 5 mm.The reddish-brown appearance is a typical appearance of hematite, while the dark appearance indicates the formation of hercynite.

Figure 4 .
Figure 4. Linear sweep voltammetry (LSV) in 1 m NaOH (pH 13.6) under AM 1.5G (100 mW cm −2 ) illumination of hematite and hematite-hercynite photoanodes fabricated from 100 nm thick Fe─Al alloy films with 0, 2, 5, 9, 12, and 17 wt.%aluminum addition by a) oxidizing at 685 °C for 1 h, b) oxidizing at 685 °C for 5 min.c) oxidizing at 493 °C for 6 h.The markers along the colored curves indicate the aluminum concentration, as shown in the Figures.The insets show the photocurrent density at 1.23 V versus RHE as a function of the aluminum concentration.d) oxidizing a photoelectrode with 17 wt.%Al at 493 °C in dependence on the oxidation time: 1 to 12 h; The inset shows the photocurrent density at 1.23 V versus RHE as a function of oxidation time for a pure hematite electrode and the hematite-hercynite electrode.LSV in dependence on the oxidation time for pure hematite and LSV under dark conditions of the same set of samples is shown in Figures S7 and S10 (Supporting Information), respectively.

Figure 5 .
Figure 5. Mott-Schottky plots recorded at 10 kHz frequency in 1 m NaOH (pH 13.6) under dark conditions of hematite photoanodes fabricated from 100 nm Fe a) and Fe─Al (17 wt.%) alloy b) by annealing 685 °C for 1 h (solid square), 685 °C for 5 min (solid triangle), and 493 °C for 12 h (solid circle).The charge carrier concentration (N d) is calculated from the slope of a linear fit, whereas flat band potentials (U fb ) from the x-axis intercept (see "Experimental Section").As indicated in the graphs, some data were multiplied by a factor of 10 or 100 for scaling or plotting reasons without affecting the results.

Figure 6 .
Figure 6.Chronoamperometry (CA) measurement in two-electrode configuration at zero applied potential under AM 1.5G (100 mW cm −2 ) illumination of hematite-hercynite photoanodes (17 wt.% Al) by oxidation at 493 °C for 12 h (black curves), and 685 °C for 5 min (red curves) in a) 1 m NaOH (pH 13.6), and b) 0.5 m H 2 O 2 as a hole scavenger added to the electrolyte with the same volume ration 1:1.The STH efficiencies in (a) are calculated from the photocurrent at the maximum point (immediately after the light on), as well as at the steady level (before the light off).