Activation of α‐Fe2O3 for Photoelectrochemical Water Splitting Strongly Enhanced by Low Temperature Annealing in Low Oxygen Containing Ambient

Abstract Photoelectrochemical (PEC) water splitting is a promising method for the conversion of solar energy into chemical energy stored in the form of hydrogen. Nanostructured hematite (α‐Fe2O3) is one of the most attractive materials for a highly efficient charge carrier generation and collection due to its large specific surface area and the short minority carrier diffusion length. In the present work, the PEC water splitting performance of nanostructured α‐Fe2O3 is investigated which was prepared by anodization followed by annealing in a low oxygen ambient (0.03 % O2 in Ar). It was found that low oxygen annealing can activate a significant PEC response of α‐Fe2O3 even at a low temperature of 400 °C and provide an excellent PEC performance compared with classic air annealing. The photocurrent of the α‐Fe2O3 annealed in the low oxygen at 1.5 V vs. RHE results as 0.5 mA cm−2, being 20 times higher than that of annealing in air. The obtained results show that the α‐Fe2O3 annealed in low oxygen contains beneficial defects and promotes the transport of holes; it can be attributed to the improvement of conductivity due to the introduction of suitable oxygen vacancies in the α‐Fe2O3. Additionally, we demonstrate the photocurrent of α‐Fe2O3 annealed in low oxygen ambient can be further enhanced by Zn‐Co LDH, which is a co‐catalyst of oxygen evolution reaction. This indicates low oxygen annealing generates a promising method to obtain an excellent PEC water splitting performance from α‐Fe2O3 photoanodes.


Introduction
With an increasing globald emand for clean and sustainable energys ystems, hydrogen as ar enewablee nergy source is a candidate to replace fossil fuels. Photoelectrochemical (PEC) waters plitting is ap romising method of converting solar energyi nto chemical energy stored in the form of hydrogen. [1] Since Fujishima and Honda first demonstrated electrochemical photolysis in 1972, [2] aw ide range of metal oxide semiconductors have been investigated as photoanode for PEC water splitting. [3][4][5][6] Among them,hematite (a-Fe 2 O 3 )isone of the most attractive materials due to its favorable band gap (1.9-2.2 eV)f or utilizing visible light,h igh chemical stability in many electrolytes, abundance,a nd low cost. [7,8] However, the PEC performance of a-Fe 2 O 3 is limited by severalf actors such as a short excited state lifetime (< 10 ps), [9] as hort hole diffusion length( 2-4 nm), [10] al ow carrier mobility (10 À2 to 10 À1 cm 2 Vs À1 ), [11,12] and ap oor electrical conductivity (10 À14 Scm À1 ). [13] These detrimental features lead to ap oor collectiono fp hotogenerated holes and their fast recombination with electronsi nt he a-Fe 2 O 3 electrodes. In order to enhance the PEC performanceo fa-Fe 2 O 3 ,awide range of approaches have been explored,s uch as nanostructuring, [14] elemental doping (e.g.,b yS n, Ti,S i), [15][16][17] and decoration by various oxygen evolution reaction (OER) co-catalysts (e.g.,I rO 2 ,C o-Pi (Co-phosphate), Zn-Co layered double hydroxides (LDH)). [18][19][20] Nanostructuring of a-Fe 2 O 3 provides ah ighly improved efficiency in chargec arrier generation and collection due to the enhancement of the specific surface area and the drastic shortening of the minority carrierd iffusion length. [14] Nanostructured a-Fe 2 O 3 has been synthesized using av ariety of techniques including sol-gel processing, [21] electrodeposition, [22] spray pyrolysis, [23] hydrothermal synthesis, [20,24] magnetron sputtering, [25,26] and electrochemical anodization. [27,28] Among them, anodization is considered as ap romising methodf or the fabrication of nanostructured a-Fe 2 O 3 from the viewpoint of low cost and large scale production. [28] However, it is well establishedt hat various precursor iron oxides are formed by anodizing and thus as uitablea nnealing procedure is needed to obtain a-Fe 2 O 3 .A nnealing conditions such as temperature and atmosphere affect the PEC performance of a-Fe 2 O 3 layers. [15,[28][29][30][31][32] Ling et al. [29] reported that photoresponse of a-Fe 2 O 3 nanowires fabricated on an FTO (fluorine doped tin oxide) glass by hydrothermal synthesis was activated by annealing in an oxygen deficient atmosphere achieved in an evacuated furnacer efilled by pure N 2 .I tw as demonstrated that the introductiono fo xygen vacancies considerably increasedt he electricalc onductivity.H owever,u nder these reducing condition precursor oxidesc an easily form magnetite (Fe 3 O 4 )w hich is highly detrimental for photolysis as it is either metal-like or behaves like anarrow band gap (< 1eV) semiconductor. [27,33] In other words, it is desired to introduce oxygen vacancies into a-Fe 2 O 3 while suppressing the formation of Fe 3 O 4 .I nt he presentw ork, we investigate the effect of annealing in ad efined gas environment containing av ery low concentration of oxygen (Ar + 0.03 %O 2 atmosphere)o nt he PEC performance of nanostructured a-Fe 2 O 3 prepared by anodization of iron foils. We demonstrate that controlled annealing in low oxygen ambient can drastically improve the PEC performance of a-Fe 2 O 3 photoanodes even at al ow temperature of 400 8C. Additionally,i no rder to furthere nhancet he PEC properties of a-Fe 2 O 3 photoanodes annealed in the low oxygen ambient, we conducts urface modification with Zn-Co LDH, whichi sr eported to be ah ighly efficient co-catalyst for OER, [20] and investigate the combineda nnealing-modification effect.

Experimental Section
Preparation of a-Fe 2 O 3 layers a-Fe 2 O 3 photoanodes were prepared by anodization of Fe foils followed by ambient controlled annealing. For this Fe foils with a thickness of 0.1 mm (99.99 %, Alfa Aesar) were anodized at 50 Vf or 5min at 20 8Ci nasolution of ethylene glycol (EG, ! 99.5 %, Carl Roth) containing 0.2 m NH 4 F( ! 98 %, Carl Roth) and 3vol %H 2 O. A two-electrode system, in which the Fe foil and aP ts heet served as the working and counter electrodes, respectively,w as used. After the anodization the samples were rinsed with water and dried in a nitrogen stream. The anodized layers were then annealed at 400 8C for 40 min in air,p ure Ar,a nd 0.03 %O 2 -Ar ambient using at ube furnace (Linn High Therm, FRH-40/250/1500). The pure Ar and 0.03 %O 2 -Ar ambient was provided by ac ontinuous flow through the furnace using commercial gas cylinders (Ar ! 99.999 %a nd VARIGON S, respectively,L inde) prior to and during the annealing process. The samples were labeled based on the annealing atmosphere (i.e.,A IR, AR, LO (Low Oxygen)).

Zn-Co LDHtreatments
Zn-Co LDH nanosheets were synthesized as mentioned in reference [20] 44 mg of zinc nitrate hexahydrate (0.15 mmol), 87 mg of cobalt nitrate hexahydrate (0.3 mmol), and 144 mg of urea (2.4 mmol) were dissolved in 10 mL deionized water,a nd then 40 mL of ethylene glycol was added to the above solution. The resulting solution was treated under microwave irradiation in aX H-MC-1 microwave reactor at 750 Wf or 10 min with 30 so n/off interval and then cooled naturally.T he product was filtered, washed thoroughly with water and absolute ethanol, and dried at 60 8C overnight. The obtained Zn-Co LDH powder was dispersed in deionized water (0.1 mg mL À1 ). After the Zn-Co LDH solution was sonicated for 10 min, a-Fe 2 O 3 annealed in the low oxygen ambient (LO) was immersed in the Zn-Co LDH solution for 10 min. The sample was washed by water and dried in an itrogen stream (labeled as LO/LDH).

Layer characterization
The morphology and structure of the layers were investigated using as canning electron microscope (SEM, Hitachi, S-4800). X-ray diffraction (XRD) was performed with an X'pert Philips MPD (equipped with aP analytical X'celerator detector) with ag raphite monochromatic Cu Ka radiation (l = 1.54056 ). The oxidation state of the layers was characterized by X-ray photoelectron spectroscopy (XPS, PHI 5600), and the peak positions were calibrated on the C1 sp eak at 284.8 eV. 57 Fe conversion electron Mçssbauer spectroscopy (CEMS) was used to monitor the physicochemical features of the thin layers. More specifically,ah omemade CEMS2010 spectrometer was employed operating in ab ackscattering geometry;i t is equipped with ap roportional continuous gas flow counter filled with aP enning mixture consisting of 90 %H ea nd 10 %C H 4 .A sa source of g-rays,a5 0mCi 57 Co(Rh) radioactive emitter was used inside the spectrometer.T he CEMS spectra were collected at room temperature for one month and then processed and fitted with the MossWinn software package. Electron Paramagnetic Response (EPR) spectra were recorded on aJ EOL continuous wave spectrometer JES-FA200 equipped with an X-band Gunn diode oscillator bridge, ac ylindric mode cavity and aN 2 cryostat. The samples were measured in the solid state under argon atmosphere in quartz glass EPR tubes at 95 Kw ith as imilar loading of % 20 mg. The spectra shown were measured using the following parameters: Temperature 95 K, microwave frequency n = 8.959 GHz, modulation width 0.1-0.01 mT,m icrowave power 1.0 mW,m odulation frequency 100 kHz and at ime constant of 0.1 s. Analysis and simulation of the data was carried out using the software "eview" and "esim" written by E. Bill (MPI for Chemical Energy Conversion, Mülheim an der Ruhr).

Photoelectrochemical measurements
The PEC performance of the a-Fe 2 O 3 photoanodes was measured in at hree-electrode PEC cell, where aP tc ounter electrode and a Ag/AgCl (3 m KCl) reference electrode in a1 .0 m KOH electrolyte were used. The photocurrent-potential (J-V)p roperties were studied by scanning the potential from À0.5 to 0.7 Va tas can rate of 2mVs À1 under periodic illumination of AM 1.5 G( 100 mW cm À2 ) light. The potentials versus Ag/AgCl (3 m KCl) were converted to where E RHE is the converted potential versus RHE, E Ag=AgCl is the experimentally measured potential, and E 0 Ag=AgCl ¼ 0:209 Va t2 5 8C for aA g/AgCl electrode in 3 m KCl. Electrochemical impedance spectroscopy (EIS) and intensity modulated photocurrent spectroscopy (IMPS) measurements were carried out using aZ ahner IM6 (Zahner Elektrik) with at unable light source TLS03. The EIS measurements were carried out in the frequency range from 100 kHz to 10 mHz at 1.3 Vv s. RHE with ap erturbation amplitude of 10 mV and a3 69 nm light source. The Mott-Schottky measurements were conducted at af requency of 10 kHz under dark conditions. The donor density (N d )was calculated by the following Equation (2).
where e 0 is the electron charge (1.60 10 À19 C), e is the dielectric constant of a-Fe 2 O 3 (80), [17] e 0 is the permittivity vacuum (8.85 10 À12 Fm À1 ), and C is the capacitance derived from the electrochemical impedance at each potential (V). The IMPS responses were recorded in the range of 1.0-1.7 Vv s. RHE with 0.1 Vs teps under 452 nm light illumination. The light intensity was modulated by 10 %between 10 kHz and 0.1 Hz.

Results and Discussion
In order to produce nanostructured a-Fe 2 O 3 ,i ron foils were anodized in an EG electrolyte containing 0.2 mol L À1 NH 4 Fa nd 3vol % H 2 Oa t5 0Vfor5mina t2 08C. The surface and cross-sectional SEM images after anodization are shown in Figure 1(a). The morphology of the layer exhibited an anoporous structure with al ayer thickness of approximately 1 mm.S ubsequently,t he anodized layers were annealed at 400 8Cf or 40 min in air,0 .03 %O 2 -Ar,a nd pure Ar ambient. The SEM images of these layers are shown in Figure 1(bd). After annealing in the air ambient (labeled "AIR"), the nanoporous structure obtained during the anodization was fully maintained. However,t he wall thickness of the layer annealed in the low oxygen ambient (labeled "LO") and in the pure Ar ambient (labeled "AR") increased during the annealing. On the other hand, the cross-sectional SEM image for AIR showed ad ouble layer structure consisting of an anoporous layer with at hickness of 500 nm on the top and ac ompact inner layer with at hickness of 1.3 mm.
Since the compact inner layer did not exist after anodization, it was formed by thermal annealing. In general nanostructuring of a semiconductor provides as ignificantly improved efficiencyi nt he charge carrier generation and collection due to the enhancement of the specific surface area and the drastic shortening of the minority carrier diffusionl ength. [14] Therefore, the compact inner layer due to thermal annealing is considered to be detrimental to the PEC performance. Furthermore, literature generally describes that a thermal oxidative annealing leads to oxide layers that consist of a gradient of wustite (FeO), magnetite (Fe 3 O 4 ), and a-Fe 2 O 3 . [30] Since FeO and Fe 3 O 4 are either metal-like or behave like an arrow band gap (< 1eV) semiconductor,t hey are not desired for photolysis. [27,33] As imilar double layer structure was also observed for the LO sample, but the thickness of the compact inner layer was slightly thinner due to the suppression of the thermal oxidation reaction in the low oxygen ambient. For the layer formed after the Ar treatment (AR), the double layer structure was not observed because the thermal oxidation reaction is essentially suppressed in the oxygen deficient atmosphere.
The PEC water splitting performance of these layers was measured in 1.0 m KOH electrolyte. Photocurrent-potential (J-V)c urves with chopped light illumination (AM 1.5G, 100 mW cm À2 )a re shown in Figure 2. The photocurrent for AIR sample annealed at 400 8C shows aq uite small value. In the case of air annealing, 400 8Ci s considered at oo low temperature to activate the photoresponse of the layer,b ecause al ayer annealed at 500 8Cu nder the same condition shows ab etter photoresponse ( Figure S1). However,t he LO sample exhibits an excellent photoresponse even at al ow annealing temperature of 400 8C. The photocurrent for LO at 1.5 Vv s. RHE is 0.5 mA cm À2 ,w hich is 20 times higher than that for AIR sample (0.025 mA cm À2 ). Additionally,t he photoelectrochemical stability for these samples was also confirmed in 1.0 m KOH electrolyte at 1.3 Vv s. RHE under illumination. Over the entire time the LO sample exhibits an early constant and drastically higher photocurrent compared with AIR sample as shown in Figure S2. On the other hand, the AR sample, which is annealed in even lower These results indicate that the photoresponse of nanoporous iron oxide layers can be highly activated using low oxygen concentration annealing and this can be achieved even at comparably very low temperatures, whereas annealing in too low oxygen concentration cause degradation of PEC performance. Annealing at lower temperature is of practical significance because it can reduce the thickness of the compact inner layer which is detrimental to the PEC performance, and the energy cost of the annealing process. As described above, since variations of PEC performance were identified depending on the annealing ambient, the mechanism from which the differences of these properties are derived is discussed below.
The crystal structure of the layers after annealing was characterized by XRD and the resulting XRD patterns are shown in Figure 3. Clearly,peaks corresponding to a-Fe 2 O 3 and/or Fe 3 O 4 can be identified for all the samples. Since the anodized layers before annealing are of an amorphous nature, [34] it is evident that these iron oxides can be crystallized even at al ow temperature of 400 8C. Whereas peaks attributed to both a-Fe 2 O 3 and Fe 3 O 4 appeared for LO and AIR samples, only peaks of Fe 3 O 4 are confirmed for the AR sample.
In annealing in pure Ar ambient, a-Fe 2 O 3 is reduced to Fe 3 O 4 due to al ow oxygen activity in the annealing atmosphere. As mentioned above, a-Fe 2 O 3 has excellent PEC properties but Fe 3 O 4 is not suitable for photolysis. Therefore, the AR sample showed no photoresponse in PEC water splitting measurements as shown in Figure 2. Although the annealing in low oxygen activity ambient was applied in order to introduce oxygen vacancies in a-Fe 2 O 3 electrodes, it was found that too low oxygen activity leads the reduction of a-Fe 2 O 3 to Fe 3 O 4 and degrades its PEC performance. To elucidate the difference between air and low oxygen annealing, peaks with higher intensity measured for LO were compared with AIR. From this comparison, one can deduce that low oxygen annealing results in ahigh crystallinity of the iron oxide.
In order to confirm the state of oxygen vacancies, the layers after annealing in air (AIR) and low oxygen ambient (LO) were analyzed by XPS. Figure 4s hows the high-resolution Fe 2p and O1 ss pectra for AIR and LO, together with their difference spectrum ("LO" minus "AIR"). Clearly,F e2 p 3/2 peaks around 711eV, Fe 2p 1/2 peaks around 724 eV,a nd satellite peaks of Fe 3 + around 719 eV are identified for both samples. These values have been reported as typical binding energies for Fe 2 O 3 . [29,[35][36][37][38] However,as atellite peak of Fe 2 + around 716 eV that is typically attributed to oxygen vacancies [29,38] could not be found. Additionally,t he O1 ss pectra also show no   difference between LO and AIR samples, which may be attributed to aconcentration below the XPS detection limit.
To study introduction of oxygen vacancies into the a-Fe 2 O 3 annealed in low oxygen ambient, the samples were further analyzed by 57 Fe conversion electron Mçssbauer spectroscopy (CEMS). It is well known that the CEMS option provides as elective characterization of iron containing phases (including amorphous or nanocrystalline) within the depth of layers up to 300 nm. The measured 57 Fe Mçssbauer spectra of AIR and LO samples, recorded at room temperature, are shown in Figure 5, and the values of the Mçssbauer hyperfine parameters, derived from the spectral fitting, are listed in Ta ble 1. For both samples, the room-temperature 57 Fe Mçssbauer spectra can be well fitted with only one spectra component;n oo ther spectra components were observed within the experimental error of the Mçssbauer technique. [39] In addition, the value of the isomer shift (d)l ies in the interval typical for Fe 3 + in a high-spin state (i.e., S = 5/2), and the value of the magnetic hyperfine field (B hf )i sn early identical for both samples. Thus, a-Fe 2 O 3 is the only iron-containing phase of which the samples are com-posed. Although, the both a-Fe 2 O 3 and Fe 3 O 4 phase were identified from the XRD patterns (Figure 3), the Fe 3 O 4 phase was not recognized from 57 Fe Mçssbauer spectra. This suggests that Fe 3 O 4 is only present as acompact inner layer.A ss how in Figure 1(b,c), the thickness of nanoporous top layers was approximately 500 nm. Considering that the detection depth of CEMS is within % 300 nm, and iron oxide species formed in the oxygen gradient, [30] it is plausible that the nanoporous top layer for both AIR and LO samples are mainly composed of a-Fe 2 O 3 while the compact inner layer consists of Fe 3 O 4 .T his is also consistent with the peak positions measured by XPS being that of typical Fe 2 O 3 (Figure 4). Comparing Figure 1(b),(c), the thickness of the compact inner layer,w hich mainly consists of Fe 3 O 4 ,f or LO sample is approximately 15 %t hinner than AIR sample. This thinner Fe 3 O 4 layer,w hich is not suitable for photolysis, for LO could be one of the reason for the improved PEC properties. However,s ince the PEC performance for LO sample is significantly improved comparing with AIR sample as shown in Figure 2, it is difficultt oa scribe it only to the reduction of thickness.
On the other hand, ac lear difference between AIR and LO can be seen when the relative intensities of the individual Mçssbauer resonant lines are compared. For the AIR sample, the relative spectral ratio of the Mçssbauer resonant lines in asextet is 3:1.88:1:1:1.88:3 while for the LO sample, a3 :3.13:1:1:3.13:3 ratio is observed. In the ideal a-Fe 2 O 3 ,a3:2:1:1:2:3r atio is expected;a ny deviation from this ratio can be related to texture effects, that is, preferential orientation of magnetic moments and, hence, crystallites. The texture effect can be quantified by calculating the angle q,w hich is defined as the average angle between the magnetic moments and direction of the g-ray beam from the Mçssbauer source. The angle q can be determined directly by using Equation (3).
where x is the relative spectral intensity of the second (fifth) Mçssbauer sextet resonant line, I 2,5 /I 3,4 .S maller values of q imply that magnetic moments are preferentially oriented along the g-ray beam, which is perpendicular to the layer surface. In contrast, q values larger than 578 indicate preferential orientation of the magnetic moments along the layer surface. As clearly seen from Ta ble 1, the LO sample shows as ignificant texture effect with preferential orientation of the magnetic moments parallel to the film surface while for the AIR sample, the texture effect is negligible. It has been reported that the conductivity of a-Fe 2 O 3 depends on its crystal orientation, and the conductivity along the (001) basal plane is four orders of magnitude larger than the conductivity along the [001] direction. [40] CEMS measurements show that the LO sample has ap referential orientation of the magnetic moments corresponding to ac rystal structure that provides excellent conductivity.T his difference of relative intensity ratio of a-Fe 2 O 3 peaks for LO and AIR cannot be recognized in the XRD patterns shown in Figure 3, as the peak attributed to the (110) plane of a-Fe 2 O 3 ,  Moreover,t he values of the quadrupole splitting (DE Q )a re characteristic of a-Fe 2 O 3 above the Morin transition temperature, that is, in aw eakly ferromagnetic state, when Fe 3 + magnetic moments are slightly canted from the basal plane, not producing ap erfect antiparallel alignment of the magnetic moments located in the adjacent crystal layers. The DE Q value for ideal a-Fe 2 O 3 should show À0.20 mm s À1 above the Morin transition temperature. The DE Q value derived for the LO sample is smaller than the ideal value, implying ap ossible occurrence ordering of vacancies affecting the distribution and orientation of the electric field gradient, whereas the DE Q value for AIR sample exhibits nearly ideal values. This means that while the AIR sample closely resembles features of ideal a-Fe 2 O 3 ,t he LO sample is composed of a-Fe 2 O 3 crystallites with defined defects. These defects are believed to be oxygen vacancies, given that LO sample were annealed in al ow content oxygen atmosphere. In order to characterize the existence of oxygen vacancies, EPR spectroscopic measurements were carried out. The EPR spectrum of air annealed sample shows only one species with g iso = 2.31.
In contrast the LO sample shows ac haracteristic signature with g 1 = 4.20, g 2 = 2.10, and g 3 = 1.80. These results clearly show the different nature of oxygen vacancies in the LO sample. [41] The results From the XPS measurements (Figure 4), the satellite peak of Fe 2 + was not identified and no evidence for presence of oxygen vacancies was obtained, which is consider to be due to the detection depth and limit of the XPS. While the a-Fe 2 O 3 layers are annealed in the low oxygen ambient, the activity of oxygen at the a-Fe 2 O 3 surface is equal to that in the atmosphere, and decreases toward the inner layer.T his is also supported by the fact that the nanoporous top layer and compact inner layer are mainly composed of a-Fe 2 O 3 and Fe 3 O 4 ,r espectively,a sd escribed above. Therefore, this means that oxygen vacancies are more easily generated in the inner region of the a-Fe 2 O 3 .S ince the detection depth of XPS is generally several nm, it is considered that the concentration of oxygen vacancies at the surface was low,a nd could not be detected by XPS measurements. On the other hand, since detection depth of CEMS is within % 300 nm, the higher concentration of oxygen vacancies formed in the inner region could be detected by CEMS. Thus, CEMS is the key method to detect oxygen vacancies deeper in a-Fe 2 O 3 layers. It is well known that oxygen vacancies can act as electron donor,a nd improve the electrical conductivity of a-Fe 2 O 3 via ap olaron hopping mechanism. [40,42] In order to investigate the donor densities of a-Fe 2 O 3 layers, Mott-Schottky measurements were carried out in 1.0 m KOH electrolyte under dark conditions. The Mott-Schottky plots are shown in Figure S4. The slopes of Mott-Schottky plots for both LO and AIR samples are positive, which indicates that they are n-type semiconductor with electrons as majority carriers. The donor densities estimated form the slopes of Mott-Schottky plots are shown in the inset. The LO sample shows ah igher donor density more than that of AIR sample. These results support the introduction of oxygen vacancies to a-Fe 2 O 3 during the annealing in the low oxygen atmosphere, which can improve the electrical conductivity. Therefore, from the above results, the main reason of the improvement of PEC performance for LO sample is the enhanced electrical conductivity due to the introduction of oxygen vacancies.
This is fully in line with EIS measurements for AIR and LO that were carried out in a1 .0 m KOH electrolyte at 1.3 Vv s. RHE under illumination (wavelength of 369 nm). The EIS results in the form of Nyquist plots are shown in Figure 6. The LO sample clearly shows two semicircles, which indicate the validity of the equivalent circuit depicted in the inset of Figure 6. The equivalent circuit consists of as eries resistance, R s ,b ulk resistance of a-Fe 2 O 3 , R 1 ,c harge transfer resistance at the a-Fe 2 O 3 /electrolyte interface, R 2 ,s pace charge capacitance of the bulk a-Fe 2 O 3 , C 1 ,a nd space charge capacitance at the a-Fe 2 O 3 /electrolyte interface, C 2 . [43,44] On the other hand, the AIR sample exhibits only ap art of large semicircle;t his means that the AIR sample has the high bulk resistance of a-Fe 2 O 3 , R 1 ,a nd/or charge transfer resistance at the a-Fe 2 O 3 /electrolyte interface, R 2 . The CEMS results described above, which suggest the introduction of oxygen vacancies and enhanced electrical conductivity for LO sample, fully support the higher bulk resistance, R 1 ,ofthe AIR compared with the LO sample.
In order to elucidate the effects of low oxygen annealing on the kinetics of hole transfer and electron-hole recombination, IMPS measurements were carried out under intensity modulated visible light illumination (wavelength of 452 nm). The theoretical background to IMPS measurements was in-depth described by Peter et al. [45][46][47][48] The variations of IMPS responses in a1 .0 m KOH electrolyte were measured as af unction of applied potential. All IMPS responses for AIR and LO are shown in Figure 7. All results for the LO sample represent two semicircles in the complex plane plots, whereas for the AIR sample all experimental plots gathered at the origin and exhibited no semicircle. Typically,w hen illumination is switched on to a a-Fe 2 O 3 electrode, instantaneous photocurrent is observed, and then under the continued illumination, the photocurrent exponentially decays due to the hole build-up and recombination of the holes and electrons until it reaches steady-state photocurrent ( Figure S5 (a)). Figure S5 (b) shows the IMPS response for LO at 1.4 Vv s. RHE. The maximum real photocurrent at high frequency, j HF ,a nd the minimum real photocurrent at low frequency, j LF ,c orrespond to the instantaneous photocurrent when the light is irradiated and the steady-state photocurrent under illumination, respectively.H ere, the instantaneous photocurrent corresponds to ah ole current, which is not associated with charge transfer across the interface between electrode/electrolyte. Therefore, the results that all plots converged at the origin indicate that  Finally,ino rder to explore the effect of atypical OER co-catalyst on samples prepared by 0.03 %O 2 -Ar annealing (LO), we treated same samples with Zn-Co LDH (labeled LO/LDH). The Zn-Co LDH was produced by as imple microwave synthesis. For decoration Feoxide layers were immersed in as olution in which Zn-CO LDH was dispersed for 10 min as described elsewhere. [20] The results of PEC water splitting measurements for decorated samples and layer characterization are shown in Figure S6. The surface SEM image exhibits that the morphology is similar to LO sample before treatment (Figure 1(c)), and no precipitate due to the Zn-Co LDH treatment is found. However,s amples decorated with Zn-Co LDH (LO/ LDH) exhibits as expected an even better PEC water splitting performance ( Figure S6 (a)). It provides al ower onset potential and a higher photocurrent than that of the LO samples. From the EIS results in the form of Nyquist plots ( Figure S6 (c)) and fitting results using the equivalent circuit model depicted in the inset image (Figure S6 (d)), LO and LO/LDH samples show similar R 1 value (bulk resistance of a-Fe 2 O 3 ), while LO/LDH samples exhibit as lightly smaller R 2 value (charge transfer resistance at the a-Fe 2 O 3 /electrolyte interface). In other words, charge transfer from the a-Fe 2 O 3 to the electrolyte is enhanced by Zn-Co LDH treatment. Therefore, Zn-Co LDH treatment does not affect the bulk a-Fe 2 O 3 but provides, as expected, effective catalysis of the OER on the electrode surface.
These results imply that a-Fe 2 O 3 annealed in low oxygen ambient which have been achieved to show ah igh PEC performance can be further improved by suitable OER co-catalyst. Therefore, the combination of low oxygen annealing and OER co-catalyst is af ea-sible concept to obtain a-Fe 2 O 3 photoanodes with an excellent PEC water splitting performance.

Conclusions
In the present study,w ei nvestigated the photoelectrochemical behavior of a-Fe 2 O 3 prepared by anodization of iron foils and particularly the effect of annealing the electrode in al ow oxygen content environment that is a0 .03 %O 2 -Ar ambient. The a-Fe 2 O 3 layer annealed at 400 8Cinl ow oxygen ambient provides asignificantly enhanced PEC performance compared with conventional air annealing. The photocurrent of the former was 0.5 mA cm À2 at 1.5 Vv s. RHE, which was 20 times higher than that of the latter.I ta lso means that the photoresponse of a-Fe 2 O 3 can be activated even at al ow temperature of 400 8C, which is of high practical significance. CEMS measurements show that a-Fe 2 O 3 annealed in low oxygen ambient contains beneficial defects assigned to oriented oxygen vacancies introduced during the low oxygen annealing. Furthermore, IMPS measurements indicate that the transport of photogenerated holes in the a-Fe 2 O 3 annealed in low oxygen ambient is strongly promoted. The PEC performance of a-Fe 2 O 3 annealed in low oxygen atmosphere can further be improved by decoration with suitable OER co-catalyst such as Zn-Co LDH. EIS measurements reveal that the treatment by Zn-Co LDH enhances the charge transfer between the a-Fe 2 O 3 surface and electrolyte. From above results, it is evident that the combination of anodization, low oxygen annealing, and OER co-catalysts is av ery effective strategy to enhance the PEC water splitting performance of a-Fe 2 O 3 photoanodes.