Durability of SrTiO3–TiO2 eutectic composite as a photoanode for photoelectrochemical water splitting

The idea of employing sunlight – a virtually inexhaustible source of energy – to catalyze various chemical reactions or generate electrical current is intensively studied nowadays. Here, we describe a method for testing photoelectrochemical (PEC) stability developed using the example of photoanodes from an SrTiO3–TiO2 eutectic composite. Eutectic composite stability measurements were carried out in long-term cycles: 0.5, 1, 2, 5, 10, 20 and 50 h of constant electrode operation (total of 88.5 h). After each cycle, cyclic voltammetry, electrochemical impedance spectroscopy, reflectance, roughness, SEM/EDS microstructure analysis and the content of Sr and Ti ions in the applied electrolyte solution were examined. The initial value of the photocurrent density was 1.95 mA cm−2 at a potential of 1.5 V vs. Ag/AgCl in a pH 2 electrolyte environment and under 6 suns of illumination it increased almost four times, reaching 7.22 mA cm−2 after a total of 88.5 h of PEC stability cycles. Due to the better catalytic properties of TiO2, this phase degrades faster, causing an increase in the roughness of the electrode surface. At the same time, reflectance of the photoanode active layer dropped from around 35% to 15%. The investigated method of PEC material testing can be applied in areas beyond photoelectrochemical water splitting, such as chemistry, photovoltaics, sensing and others.


Introduction
Many techniques of obtaining hydrogen gas from water using sunlight exist today.One of the most promising is photoelectrochemical (PEC) splitting due to many advantages over other methods.These advantages include: (i) ability to carry out the process in ambient conditions (at room temperature and pressure); 1 (ii) the hydrogen and oxygen evolution reactions take place at different electrodes, which eliminates mixing of these gases; 2 (iii) preparation of photoelectrodes is usually very simple.Methods of PEC hydrogen production have been developed for over 50 years. 3At rst, the light absorption layer of the electrodes comprised bulk materials, in particular single crystals. 4Promising conversion efficiencies of H 2 and O 2 generation were obtained through various doping, which extended the useable range of the solar spectrum.The breakthrough came with employing dyesensitized mesoscopic materials in PEC cells, proposed by O'Regan and Gratzel. 5anostructures functionalized with organic dyes are the most commonly-investigated group of materials for solar-driven water splitting; but multijunction [6][7][8] and organic 9,10 electrodes are also showing promising results.Such nanomaterials exhibit relatively high conversion efficiencies due to their extended surface areas and good absorption properties.TiO 2 heterojunctions can be a good example of this type of composite.Zhang et al. obtained high PEC performance by extending the absorbance range spectrum through the localized surface plasmon resonance (LSPR) effect in the prepared electrodes with Au NPs on the surface of ower-like TiO 2 . 11Efficient sensitization of TiO 2 was achieved by sequential ionic layer absorption and reaction (SILAR), which allowed absorption to be extended from 3.21 eV (pure TiO 2 ) to 1.98-2.16eV (depending on the level of BiOI on TiO 2 ). 12 Again SILAR was used to obtain a heterojunction of TiO 2 nanotube arrays/Bi 2 S 3 quantum dots, which led to a minimization of the band gap to 1.3 eV, and thus to a 3.8 greater photocurrent density compared to bare TiO 2 . 13Almost 4 times higher photocurrent was obtained by creating a p-n junction of p-ZnO quantum dots and n-TiO 2 by a solvothermal method.Thanks to the ZnO/TiO 2 junction, the electrode showed greater electron-hole separation and thus increased the charge carrier lifetime. 14However, such materials suffer from high recombination rates of charge carriers and low stabilities.On the other hand, highly stable oxide compounds have wide band gaps, which enables absorption only in the UV range and a narrow visible range.A material which simultaneously exhibits high efficiency, durability, low cost and ease of manufacture is yet to emerge. 15Most currently developed materials exhibit only one of these properties.
Recently, we have shown the possibility of utilizing eutectic composites in PEC water splitting [16][17][18][19] taking advantage of their ability to combine various material phases into one solid composite.Additionally, signicant advantages of eutectic composites result from (i) high crystallinity (as single crystals), 20 (ii) sharp interfaces between the component phases, 21 (iii) the possibility of selecting different components enabling broadband absorption (UV-Vis), (iv) the use of phases that are difficult to achieve or otherwise unavailable, and (v) the ability to tune eutectic properties depending on requirements.The reaction, which leads to splitting of water molecules on semiconducting materials under solar irradiation, is associated with generation of charge carrier species: electrons and electron holes.The electron-hole pairs exhibit high redox potentials.This fact is exploited in photocatalytic processes such a PEC hydrogen production and photocatalytic CO 2 reduction or degradation of organic compounds. 22However, in reality only a fraction of charge carriers react with the active material instead of the target medium (electrolyte solution), leading to degradation of material and loss of photocatalytic properties.Solar-driven catalysis is oen carried out with an applied electrical bias, in an aggressive environment of an electrolyte solution which is either highly acidic or basic.
The stability of photoelectrodes in the PEC system is one of the crucial parameters because unstable materials affect the solar energy conversion efficiency.The stability of devices used for PEC water splitting is dened as the ability of the system to maintain the solar energy conversion efficiency at a steady level with predetermined tolerance over extended time periods.The fact that the water splitting reaction in PEC cells is carried out under solar light irradiation, together with an externally-applied bias, leads to a highly corrosive environment for the photoactive material.Thus, achieving high electrode stability, especially of the photoanode, is not easy and is being intensively studied.For example, Dias et al. 23 investigated a hematite photoanode prepared by spray pyrolysis over 1000 h of sunlight exposure and showed a stable photocurrent density of 0.94 mA cm −2 at 1.45 V vs. RHE.On the other hand, TiO 2 nanotube arrays with SrTiO 3 -TiO 2 hetero-nanoparticles obtained by hydrothermal reaction showed no drop of hydrogen production generation rate and a stable photocurrent of 1.91 mA cm −2 at 0.3 V vs. SCE (saturated calomel electrode) aer 25 h. 24Similar stability was reported by Bashiri et al. 25 when they investigated a SrTiO 3 @-TiO 2 @Fe 2 O 3 nanorod heterostructure and recorded stable hydrogen production during its 25 h operation time.Similarly, a heterostructured Cr-doped SrTiO 3 -TiO 2 photoanode retained a stable photocurrent density (4.05 mA cm −2 vs. 0.6 V SCE) over 2 h of continuous working. 26Additionally, it was shown that a SrTiO3-TiO 2 eutectic photoanode improved photocurrent response by 50% up to 8.5 mA cm −2 (at 1.5 V vs. NHE) aer 30 h of continuous operation under light.The increased photocurrent was a result of higher absorbance due to increased surface area of the electrode.To summarize these various studies on the PEC electrode material stability: it is clear that there is a lack of a robust method that would provide a fair comparison of the properties of different materials, preparation methods and determination of long-term stability in photocatalysis processes.
In this work, we propose a universal method for PEC electrode stability investigation which can enable easier comparison of various materials.The method is based on following changes in the photoanode active material, such as: (i) the level of generated photocurrent; (ii) reectance; (iii) concentration of photocorroded elements in the electrolyte; (iv) surface roughness aer a dened number of operating cycles of PEC operation.We investigated the durability of a SrTiO 3 -TiO 2 eutecticbased photoelectrode during 88.5 h of operation as a photoanode in PEC water splitting.The prepared photoanode revealed an improvement of photocurrent density, reaching its highest value of 7.22 mA cm −2 at 1.5 V vs. Ag/AgCl aer a PEC stability test.The effects of the ongoing stability test were examined by scanning electron microscopy, reectance measurements, prolometry, inductively coupled plasma mass spectrometry and PEC analysis.

Crystal growth and material characterization
The SrTiO 3 -TiO 2 eutectic composite rod was grown (Fig. 1a) from 23 mol% SrO and 77 mol% TiO 2 composition. 27The raw materials were prepared by mixing high-purity SrCO 3 (Alfa Aesar; 99.99%) and TiO 2 (Alfa Aesar; 99.99%) oxide powders.It was obtained by the micro-pulling-down method, m-PD, in a nitrogen atmosphere with 0.5 mm min −1 pulling rate.The grown rod was 120 mm long with a 3 mm diameter.The m-PD has been demonstrated to be a useful technique for quick material growth, and has been shown to enable fabrication of materials that nd applications in various elds: optoelectronics; 28,29 photonic crystals; 30-32 metamaterials; [33][34][35][36] THz materials; 37,38 and SERS detection. 39,40X-ray powder diffraction (XRD) qualitative phase analysis of the SrTiO 3 -TiO 2 eutectic composite was performed on the as-grown samples using a Rigaku SmartLab 3 kW diffractometer.The device was The ICP-MS measurements employed a calibration curve with external standards, which were prepared using 1% nitric acid and titanium and strontium standards.

Photoanode preparation
SrTiO 3 -TiO 2 rods were cut perpendicularly to the growth direction by a wire saw to form 1 mm thick slices.Slices were ground and polished to create c.a. 30 mm thick plates with a mirror-like surface.Aer that, the plates were glued to uorine-doped tin oxide (FTO) coated glass substrates using silver epoxy paste (Elpox Sc 515 AMEPOX) which served, at the same time, as the ohmic contact layer.Silver paste was cured at 150 °C for 2 h in air at the hotplate.Aerwards, the excess of silver was covered by epoxy resin to avoid unwanted redox reactions during PEC analysis.The nal electrode thickness, before the start of the stability test, was 31 mm with an active surface area of 0.04 cm 2 .The active surface area was measured using a Nikon Eclipse LV150 optical microscope.

PEC and stability measurements
The AgCl) electrode as the reference; and a platinum electrode as the counter electrode.Electrochemical impedance spectroscopy (EIS) was performed at 0 V in the frequency range of 1 MHz to 1 Hz using a 5 mV amplitude signal, both in dark and under illumination conditions.All PEC measurements were performed in electrolyte solution of 95% H 2 SO 4 , Na 2 SO 4 and deionized water with pH 2 measured by a Seven Multi Mettler Toledo pH-meter.The irradiation intensity of the xenon lamp was calibrated using a Solar Light PMA 2144 Class II pyranometer.

Crystal growth and eutectic composite characterization
In order to present a common method for judging the electrode durability and hydrogen production stability, we used the SrTiO 3 -TiO 2 eutectic composite that has been previously demonstrated to form an efficient PEC photoanode. 18The material for investigation was prepared as described previously and is illustrated in Fig. 1, including the as-grown rod with 0.5 mm min −1 pulling rate and the microstructure of TiO 2 precipitates embedded in the SrTiO 3 matrix 42 showed by SEM and energy dispersive spectroscopy (Fig. 1a-d).X-ray diffraction (Fig. 1e) conrmed the presence of a SrTiO 3 phase in the form of tausonite (04-002-1009) and a TiO 2 phase in the form of rutile (04-005-4625). 42,43Fig. 1f shows a working electrode prepared from the demonstrated SrTiO 3 -TiO 2 eutectic composite, which has been used for the stability test in this work.
In order to investigate the durability of the photoanode, we performed in parallel several measurements during the study to dene a new methodology.The new methodology is based on changes in the photoanode active material, such as: (i) the level of generated photocurrent; (ii) reectance; (iii) concentration of photocorroded elements in the electrolyte; (iv) surface roughness aer a dened number of cycles of subsequent operation in PEC.

Roughness and microstructure analysis
In order to investigate the photocorrosion of the electrode active material, we studied the concentration of the photocorroded elements in the electrolyte using ICP-MS.The total content of Ti and Sr ions in the electrolyte measured by ICP-MS aer each cycle was, on average, increasing with time.The rst measurable Ti content was recorded aer 5 h of cycles and then increased with subsequent cycles.The rst four recorded Sr contents seemed to be at the limit of the method since the value was not affected by the working time or the eutectic surface was still robust enough.Finally, aer 50 h of cycles, a higher concentration of Ti than Sr was observed in the electrolyte (Fig. 2a).Titanium ions can enter the electrolyte from both phases: titanium oxide and strontium titanate.
As the next step in the investigation of the photocorrosion of the electrode active material, we studied the surface roughness using a prolometer.The electrode surface quality can be expressed by two parameters: the arithmetic average roughness (Ra) and the root-mean-square roughness (Rq) as amplitude parameters of the deviation of the prole from the mean line.The parameters were estimated by scanning 1 mm 2 of the SrTiO 3 -TiO 2 electrode surface area.Results presented in Fig. 2b show an increase in the Ra and Rq parameters aer subsequent stability cycles, corresponding to an increase in roughness.The experimental results were tted using exponential functions.
In parallel to the above, we investigated the surface roughness with scanning electron microscopy (SEM).It is strongly visible in the SEM images, mostly in the TiO 2 phase which is suffering from the PEC process (Fig. 3b-e).The difference between the levels of individual phases deepens with progressive test time, starting from initial, optically polished active surface of the sample (Fig. 3a) to 2, 10, 20 and 50 h.Aer 50 h of PEC testing, the presence of the third phase can be observed on the sample surface (Fig. 3e).EDS map analysis revealed, in addition to the presence of Sr, Ti and O, Ag inclusions coming from the adhesive layer used in the tested SrTiO 3 -TiO 2 photoanode (Fig. 3f).The presence of silver on the top of the photoanode proves that the process of degradation of the working electrode has begun.

Reectance measurements of SrTiO 3 -TiO 2 photoanode
As the next step, we used measurements of the electrode surface reectance in order to study the surface roughness.Reectance spectra were recorded in the 400-1000 nm wavelength range and showed a decrease in the value of the measured light reection aer the PEC stability measurement cycles (Fig. 4a).This is consistent with the observed increase in the roughness of the electrode in relation to the length of its continuous operation.Irregularities on the electrode surface cause light scattering and thus reduction of reectance.The initial reectance value of about 35% decreased down to 15% aer 50 h of continuous operation.Fig. 4b and c show the initial mirror-like surface of SrTiO 3 -TiO 2 and its surface aer 5 h of stability test, respectively.

PEC and stability measurements
PEC analysis and stability tests were carried out with cycles of: 0.5, 1, 2, 5, 10, 20 and 50 h of continuous operation at the applied voltage of 1.23 V (vs.Ag/AgCl) in the whole Xe-lamp spectrum range.Firstly, the open-circuit potential (OCP) was recorded to check the ohmic connection within the entire cell.The initially stable OCP value decreased under the light, con-rming the photosensitivity and the n-type nature of the investigated electrode. 44,45Aer that, cyclic voltammetry (CV) analysis was carried out under modulated lighting (the light source was cyclically turned on and off) to investigate the possible presence of undesirable reactions (not related to photocatalysis).The current density values in the absence of  lighting (dark currents) started to increase at the voltage of 1.65 V (vs.Ag/AgCl), which means that water electrolysis took place at this voltage and photoelectrolysis should occur before this value.Fig. 5a shows examples of CV scans performed before stability testing aer 5 and 50 h of continuous operation.The initial photocurrent density increased from 1.95 to the maximum value of 7.22 mA cm −2 at 1.5 V vs. Ag/AgCl.This photocurrent density is still lower than the 8.5 mA cm −2 value reported previously for the SrTiO 3 -TiO 2 eutectic aer 30 h of operation .17However, the starting thickness of that eutectic active layer was only 18 mm, which is much lower when compared to this work, 31 mm.The thicker active layer itself provides worse electrochemical response (longer charge carrier diffusion distance and higher resistivity). 46Additional improvement of the photocurrent is expected for further optimization of the electrode thickness or progressed PEC operation.It should also be noted that following the initial increases in photocurrents, aer the next 2, 5 and 10 h cycles, the photocurrent values decreased to 4.15 mA cm −2 and then started to increase again.This can be explained by surface poisoning by residuals of etched TiO 2 and SrTiO 3 , which block the surface for effective PEC process.Such a blocked surface was not photoelectrochemically active because of darkening.Moreover, detached particles were present in the solution but there was not chemical bonding with the electrode surface allowing transfer of the photogenerated electron-hole.Additionally, EDS analysis detected the Ag signal at the electrode surface.The reduced thickness of the eutectic electrode caused by etching led to the formation of the gap between the electrode surface and epoxy resin and exposure of the silver from the ohmic contact.Similarly to the residuals of SrTiO 3 and TiO 2 , Ag particles blocked the electrode surface.Nevertheless, such a signicant change in the overall value of photocurrents can be explained primarily by the increase of the working electrode active surface, leading to the creation of more active sites for light absorption and electrolyte adsorption.In addition, the overall thickness of the photoactive material decreased as a result of the continuous electrode operation.The summary of changes in photocurrent density aer each PEC cycle is shown in Fig. 5b.

Electrochemical impedance spectroscopy (EIS)
In addition to PEC analysis, EIS measurements were performed when the light was turned on and off.The experimental results of EIS measured under illumination were tted using the equivalent circuit model shown in Fig. 6a.The most important components of the model were the R-C associated with the semiconductor SrTiO 3 -TiO 2 represented by the resistance (R sc ) and capacitance of the capacitor (C sc ) and R-CPE associated with the charge transfer between the semiconductor and the electrolyte (R ctcharge transfer resistance, Q dlcapacitance of the double layer). 47The t results are summarized in Table 1.
The total resistance to charge transfer decreases as the sample is illuminated.The main factor contributing to the decrease in total resistance was R ct .
The generated photo-holes migrate towards the semiconductor-electrolyte interface, while the photoelectrons move Table 1 The results of the experimental data fitting to the proposed model of the equivalent circuit  into the semiconductor.Photo-holes are involved in the oxidation of electrolyte ions.At the same time, no signicant changes in the resistance values R sc and R ct were observed (Fig. 6a and b).

Conclusions
In this paper, we proposed a method for testing the stability of photoanodes for PEC water splitting.This method includes the study of: (i) the level of saturated photocurrents; (ii) the amount of light reected from the surface of the active material; (iii) the concentration of Sr and Ti ions in the electrolyte; (iv) changes in surface roughness and (v) observation of the surface aer stability cycles of progressive operation of the photoelectrode in intense illumination conditions.The PEC stability measurements based on a SrTiO 3 -TiO 2 eutectic composite photoanode, were carried out over long time cycles starting from: 0.5, 1, 2, 5, 10, 20 and up to 50 h.The initial value of the photocurrent density of 1.95 mA cm −2 at the potential of 1.5 V vs. Ag/AgCl (in the electrolyte environment with pH = 2) increased almost four times to the value of 7.22 mA cm −2 aer a total of 88.5 h of PEC stability cycles.The roughness of the active surface of the tested electrode, measured by prolometry, increased.It was additionally conrmed by a decrease in reectance from about 35 to 15%.The higher concentration of Ti ions compared to Sr was measured using ICP-MS.The method of photoanode stability testing presented here can become a very practical tool for comparing the properties of electrodes made of different materials and determining their long-term usefulness in photocatalysis processes.The dissemination of this research method will contribute to accelerated stability testing, which is extremely necessary in the perspective of the development of this scientic eld.Furthermore, this method can be used in the areas of material durability analysis with long operating cycles in the elds of science such as fuel cells, catalysis, nanotechnology or photovoltaics.

Fig. 1
Fig. 1 Directionally solidified SrTiO 3 -TiO 2 eutectic composite as an active material in PEC water splitting.(a) As-grown rod obtained by the m-PD method with 0.5 mm min −1 pulling rate; (b), SEM image presenting the composite microstructure; (c) and (d) EDS maps showing the distribution of Ti and Sr in the composite; (e) powder X-ray diffraction demonstrating two phases in the eutectic: TiO 2 in the form of rutile and tausonite SrTiO 3 phase; (f) image of the electrode with the SrTiO 3 -TiO 2 eutectic composite as the photoelectrochemically active material.
PEC measurements under illumination and dark conditions employed a CHI -660D potentiostat (CHI Instruments) using: (i) open-circuit potential (OCP) to assess the stability of the system without irradiation and quality of the developed electrical contact; (ii) cyclic voltammetry (CV) to estimated intensity of photocurrent densities with an applied potential between −0.5-1.8V and a 10 mV s −1 scan rate; (iii) amperometric currenttime to test stability of electrodes in cycles (0.5, 1, 2, 5, 10, 20 and 50 h, total time of 88.5 h) under irradiation (600 mA cm −2 from a Solar Light 150 W xenon lamp).A threeelectrode conguration placed in electrolyte solution in a Teon/PTFE cell with a quartz window was used as a PEC analysis setup.A SrTiO 3 -TiO 2 eutectic sample with thickness of 31 mm served as a working electrode; a silver/silver chloride (Ag/

Fig. 2 Fig. 3
Fig. 2 Photocorrosion of the SrTiO 3 -TiO 2 electrode active material studied during 50 h of PEC water splitting operation: (a) amount of photocorroded Sr and Ti ions in the electrolyte used for PEC stability tests; (b) changes in the photoelectrode surface roughness characterised by measured Ra and Rq roughness parameters, together with fits.

Fig. 4
Fig. 4 Changes of the SrTiO 3 -TiO 2 electrode reflectance after successive cycles of PEC measurements.(a) Reflectance changes starting from the as-grown optically polished sample to the same sample after 50 h of operation; (b) and (c) microspectrophotometer images of the photoanode surface during reflectance measurements at the beginning and after 5 h of the PEC stability test.

Fig. 5
Fig. 5 The evolution of generated photocurrent during continuous PEC operation.(a) The cyclic voltammetry (CV) results for the initial electrode, and after 5 and 50 h of operation under switching on and off the light, (b) changes of the photocurrent density with the operation time, measured for all stability test cycles at 1.5 V at Ag/AgCl.

Fig. 6
Fig. 6 Electrochemical impedance spectroscopy: (a) Nyquist plots under illumination and equivalent circuit model (inset), (b) changes in semiconductor resistance (R sc ), (c) changes in semiconductor-electrolyte charge transfer resistance (R ct ) after successive times of the stability test.