Photoelectrochemistry of polycrystalline SrTiO3 and BaTiO3 semiconductor electrodes

Abstract

This thesis deals with the investigations on the photoelectrochemistry and photoelectrolysis of water, making use of polycrystalline semiconductor electrodes of BaTiO? and SrTiO? with a view to understand the effect of bulk and surface contributions. The introductory chapter incorporates the general principles of semiconductor electrochemistry. In the light of the most popular Schottky-barrier model, the formation of depletion layers and the basic electrode processes at the semiconductor/electrolyte interface are presented. Since the flat-band potential is a critical parameter, its significance and methods of determination are explained, followed by the photocurrent-voltage relations and the parameters obtainable from Gartner relations. The wavelength dependence of photocurrent can be used to find out the band tailing, indicative of the mid-band gap states. Energy distribution in the solar spectrum and the efficiency relations are discussed with respect to the band gap energies of semiconductors. This is followed by a brief presentation of the limitations of the Schottky-barrier model, particularly in presence of larger concentrations of surface states. Characteristics of different semiconductor materials utilised in the fabrication of photoelectrochemical cells (PECs) are presented. The materials tested so far suffer from either stability problems, including photocorrosion, or they have larger band gap energies to effectively match with the solar spectrum, though stable. The perovskite titanates belong to the latter category which produce H? and O? during the photoelectrolysis of water, without any external bias. Attempts have been made to extend their response to the visible part of the spectrum through doping with transition metal ions or by stabilising the mid-band gap states. However, the resulting electrodes require higher bias potentials for measurable photoelectrolysis. Since the present work involves the titanate perovskites, a brief discussion is taken up on their crystal chemistry, electronic energy level schemes, conduction mechanism around room temperature, and crystal structure-photoelectrochemical property correlations. In the context of ceramic processing, the defect chemistry is important; similarly, the frozen-in lattice defects have tremendous influence on the electrical transport properties of these solids. The polycrystalline specimens are much cheaper than the single crystals, particularly for practical devices. Hence the possible complexities arising from the grain boundaries are discussed. The scope of the present work is presented at the end of this chapter. The general methods of preparation, chemical analyses, processing of ceramic electrodes and the instrumental techniques employed during the present investigations are given in Chapter II. Chemical methods standardised for the preparation of high-purity powders of MTiO? (M - Ba, Sr, Ca or Pb), their solid solutions as well as those with the corresponding zirconates and the donor-doped phases through the oxalate precursor route are presented here. The methods of chemical analyses of the various constituents have been standardised through wet chemical methods, including AAS. The physical methods adopted for characterisation include X-ray powder diffraction, EPR, SEM and XPS as well as the electrochemical techniques. The basic concepts in interpreting the EPR results of powder specimens are discussed. The steps involved in sintering the ceramics and their microstructural analysis are given. A procedure is standardised to obtain reliable ohmic contacts through electroless-Ni deposition followed by thickening with Ni-electroplating so that good solderability is possible on one side of the electrodes. Instrumental techniques used for photoelectrochemical measurements, involving cyclic voltammetry, current-voltage and capacitance measurements and the cell arrangements under illumination and gas collection are presented. Chapter III deals with the electrical properties and the conduction mechanism on polycrystalline BaTiO? and SrTiO? ceramics. The electrical resistivities of BaTiO? and SrTiO? vary as a function of donor dopants and differing partial pressures of oxygen during the high-temperature annealing and the cooling rates. Both the materials acquire low resistivity when annealed in atmospheres of pO? < 10?¹ atm above 650 °C, in the absence of any dopants. BaTiO? doped with donor impurities of limited concentration range has lower resistivity, although the specimens are sintered in air and cooled at moderate to faster rates. These samples show positive temperature coefficient of resistance (PTCR) around the Curie temperature (Tc). SrTiO? with the same concentration of donors is of high resistivity when sintered in air and acquires lower resistivity only when annealed in reducing atmospheres. Doping with acceptor impurities does not produce any measurable p-type conductance, whereas these impurities enhance the room temperature resistivities. Diffuse reflectance spectra of low-resistivity BaTiO? and SrTiO? indicate the presence of Ti(III) as well as the oxygen vacancy-related defect centers. EPR spectra of donor-doped BaTiO? show two signals with g = 1.963 of medium intensity and g = 1.997 of low intensity. The former is identified to arise from Ti³?–V? defect complex whereas the latter is shown to originate from singly ionised barium vacancies, VBa?. Above Tc, the g = 1.997 signal acquires high intensity, which indicates the “activation” of the corresponding defect centre. This causes the diminution of the charge carrier concentration leading to the enhancement of resistivity, thus leading to the charge redistribution equilibrium VBa? + e? ? VBa? across the structural phase transition. 'Activation' of acceptor states also may arise from the acceptor impurities; thus, changes in electronic states such as Mn³? + e? ? Mn²? and Fe³? + e? ? Fe²?–V?(e?) are observed with the EPR spectra. The higher resistivity of donor-doped SrTiO?, on sintering in air, is attributed to the 'activated' acceptor states. The charge redistribution at the acceptor centers accompanying the structural phase transition is explained on the basis of vibronic interactions where the electronic states couple with the vibronic modes of the lattice via Frohlich mechanism. The vibronic contribution to phase transition dynamics has been known in literature for ferroelectrics. In the case of semiconducting titanates, the effects of lattice instability on extrinsic acceptor states are prominent so that the itinerant electrons may redistribute differently in various lattice symmetry configurations, with changing location of acceptor energy states with respect to valence and conduction band edges. In such cases, the 'activated' states correspond to excited vibrational states of the small polaron-lattice complex in the ground electronic state of the polaron. The photoelectrochemical properties of polycrystalline BaTiO? and SrTiO? electrodes are incorporated in the fourth chapter. New features in the cyclic voltammograms, related to the surface states, are observed for the mechanically polished electrodes in the dark. These include the maxima in the anodic as well as the cathodic directions of the sweep and the abrupt rise in anodic current above ~0.5 V (SCE). These features in cyclic voltammograms are not observed when the electrodes are etched in dilute acids for extended periods of time. The nature of the surface states is investigated by alternating etching and polishing of the same electrode, by varying. The applied potential limits, change of pH and concentration of alkali, noting the effect of dissolved O? and O?, modifying the electrode surface by applying Ti(III) and Ti(IV) solutions and drying, and lastly, by reannealing the electrodes in oxidising as well as reducing atmospheres. The results show that the anodic maxima originate from the oxygen vacancies at the electrode surface where the strongly chemisorbed surface hydroxyls are present. Under the applied anodic potential and the accompanying upward band bending, electrons are transferred to the conduction band so that OH° radicals are formed. Interaction of these radicals with OH? ions in the electrolyte produces peroxohydroxyls, HOO?, which decompose to O? and H?O under anodic potentials ? 0.5 V (SCE). The accompanying electron tunneling from the surface states to the conduction band, through the depletion layer, causes sudden rise in anodic current. Tunneling through the depletion layer is supported by the facts that illumination does not change the potential at which the enhancement in anodic current is noticed and that this potential is somewhat increased with decreasing donor density at the surface region, accompanied by decreasing photocurrent. The surface states which give rise to the cathodic peaks are related to dissolved O?, since they vanish after the removal of O? in the electrolyte and also by limiting the upper voltage. They originate from surface hydroxyls attached to Ti?? ions which facilitate the transfer of electrons to dissolved O? under cathodic potentials. The results show that these surface states themselves may act as electron traps leading to the formation of Ti³?–OH? sites at the surface. The latter is involved in the H?-evolution reaction at more cathodic potentials because the redox potentials of Ti³?/Ti?? and H?/H? couples overlap. The easy formation of Ti³?–OH? around the electron-rich Ti³?–V? defect sites accounts for the enhanced intensity of the cathodic peaks on the etched electrode surfaces and that these maxima shift to the cathodic direction with increasing applied potential limits. I–V measurements are carried out at nearly steady-state conditions during band gap illumination and using 1 M NaOH as the electrolyte for three types of polycrystalline electrodes, viz. BaTiO? : 0.3% La, BaTiO?–H? and SrTiO?–H?; the latter are annealed in low oxygen partial pressures at 1000 °C. The Von potentials, where the photocurrents start appearing and are approximated to Vfb, are determined from the Ip vs Vapp curves, following Gartner relations. These plots are not linear throughout the range of applied potentials and differ considerably with the surface treatments. Influence of the bulk electrode properties on the Ip–V relations are investigated which include the effects of electrical conductivity and carrier concentrations, average grain size, ceramic processing conditions such as cooling rate after sintering and the effect of acceptor impurities, mostly transition metal ions replacing Ti(IV). Capacitance-voltage relations are measured for the semiconductor/electrolyte interface and from the Mott-Schottky plots, the values of Vfb and Nd at the space charge layer are determined. The bulk Nd values are slightly lower than those obtained from the capacitance data. The changes in Vfb with the pH of the electrolyte and the presence of selected anions with differing strength of surface adsorption have been investigated. Dependence of photocurrent on wavelength has shown that the response of BaTiO? : La and BaTiO?–H? tails into the visible region, although the magnitude of photocurrents in the visible is much lower than those under band gap illumination. The former is unaffected by chemical etching as well as mechanical polishing. SrTiO? shows low photoresponse to visible light only on polishing and is destroyed by chemical etching. The visible photoresponse arises from the mid-band gap states. The plots of (?h?)² vs h? show that these mid-band gap states extend to ~0.8 eV above the valence band edge for BaTiO?. The variations in photocurrents with applied voltages and intensities of illumination showed that the visible light response is not a surface phenomenon but a bulk process occurring throughout the depletion layer. The hole diffusion length (Lp) and the donor concentration in the depletion layer are obtained from the plots of ln(1–?) vs Vapp. The polished electrodes show lower Lp and enhanced Nd values as compared to the corresponding etched electrodes under band gap illumination. This has been explained on the basis of hole trapping at the defect centers within the depletion layer, which, in turn, leads to accumulation of positive charges and thus enhancing the electrons in the conduction band. It also leads to the tailing of the band gap absorption arising from the mid-band gap energy states. The Lp values obtained from the visible light response for BaTiO? are greater than those of SrTiO?. The diminished response of BaTiO? as compared to that of SrTiO? during band gap illumination is also explainable in terms of these energy states. The difference in the photoelectrochemical characteristics in the visible region for BaTiO? as compared to SrTiO? can be explained on the basis of the difference in lattice defects that generate the mid-band gap energy states. The flat-band potential, Vfb, of BaTiO? is around –0.90 V on band gap illumination whereas the apparent flat-band potential for the visible photocurrent, Vfb* = –0.57 V, indicates the presence of mid-band gap states of 0.33 V (= Vfb – Vfb*) below the Fermi level. Donor-doped as well as H?-reduced tetragonal BaTiO? has the Fermi levels located ~0.3 eV (Ec – Ef) below the conduction band edge. Since Vfb = –0.90 V, Ec may be located around –1.20 V. The mid-band gap states located around 0.3 eV below Ef cannot arise from V??, since these energy states should be of acceptor type, located below mid-way in the band gap (3.15 eV). Hence the defect states located ~0.3 eV below Ef may arise from Ti³?–V? centers, whose EPR signal is detected both in BaTiO? : La and BaTiO?–H?. Photoexcited electrons from the valence band move to the Ti³?–V? sub-band from where they tunnel into the conduction band and flow into the external circuit. The photogenerated holes travel uphill, finding their way into neutral V?? states that overlap with the valence band. These holes will be available at the interface in response to the space charge layer. In the cubic SrTiO?, mobility of photogenerated holes into the states is restricted, since they no longer overlap with the valence band. Mechanical polishing may introduce additional mid-band gap states through surface damage or lattice distortions around the surface region. This leads to the photoresponse in the visible spectrum for polished SrTiO? and the response vanishes on chemical etching. Since differences are noticed in the anomalous response to visible light for BaTiO? and SrTiO?, it is proposed that the nature of the defect-related mid-band gap states may not be the same in these two cases. Since the solid solutions of MTiO? perovskites exhibit crystal symmetry changes by replacing the M²? ions without altering the nature of the defects, the photochemistry of the titanate solid solutions is investigated, which forms the subject matter of Chapter V. Characterisation of the titanate solid solutions has been carried out to establish the phase singularity and the crystal symmetry changes with composition, in Ba–Sr, Ca–Sr and Pb–Sr titanates. In electrical conductivity, reflectance spectra and effective dielectric constant are studied. The conductivity of donor-doped samples, sintered in air, showed that the room temperature values of the tetragonal phases are much larger than those of the cubic phases. The I–V curves under band gap illumination of (Ba,Sr)TiO? do not show enhanced response with the dielectric constants whose maximum shifts to room temperature with Sr-content. Since the depletion layer width is related to dielectric constant, this shows the limitation in applying the Schottky-barrier model for the semiconductor/electrolyte system. For the cubic phases, visible photocurrent is noticed only after mechanical polishing, whereas for the tetragonal phases, the same is noticeable without any treatment and mechanical polishing or chemical etching has only marginal influence on visible photoresponse. In the case of (Sr,Pb)TiO?, specimens containing >35 mole % PbTiO? could not be annealed in atmospheres of lower oxygen partial pressures due to the formation of metallic lead and therefore the measurements are restricted to the cubic phase. The influence of the high bulk-dielectric-constant in the band gap photoresponse is not observed in this case as well. For the tetragonal phase of (Sr,Ca)TiO?, with Sr = 55–85%, the photochemical behaviour in the visible region is the same as that of the tetragonal (Ba,Sr)TiO?. The electrical conductivity as well as the photoresponse of Ba(Ti,Zr)O? decreases with Zr-content. The difference in the phase change with increasing Zr is evident in the action spectrum of these electrodes. The low stability of Zr(III) as compared to Ti(III) is one of the major factors. The results indicate that the difference in photoelectrochemical characteristics for the tetragonal as compared to the cubic phases of the titanate solid solutions arise from the nature of the lattice defects which give rise to the mid-band gap states. The results support the explanation presented in Chapter IV. The lower efficiency of the tetragonal phases during band gap illumination arises from the recombinations at these defect centers. The decrease in photocurrent with Zr-content is in accordance with the general conclusion that the photoelectrochemical properties of MTiO? are related to TiO? octahedra. The Sixth Chapter deals with the kinetics of evolution of gaseous products, H? and O?, from the electrochemical cells having SrTiO? or BaTiO? photoanode with the aim of enhancing the efficiency of product generation. The volumes of H? and O? evolved at different time intervals from the two-compartment PEC, containing Pt-cathode and titanate photoanode that is variously surface treated and illuminated with near band gap radiation are measured. The rates of evolution of both the gases are less for the polished electrodes as compared to the electrodes etched in dilute acids for extended periods of time. The rates of evolution of the products from PEC with donor-doped BaTiO? are lower than those from BaTiO? annealed in low oxygen partial pressures. With increase in applied potential, the rate of evolution of the products increases and remains linear up to ~2 V (SCE) and at higher values, the rate of evolution decreases, possibly due to irreversible changes in surface composition. The rate of evolution increases with the concentration of alkali, in the range of 1–20 M NaOH. Enhanced photoresponse is noticed for the titanate electrodes when exposed to dilute HNO? + HF for extended periods up to 120 minutes. The steep rise in photocurrent at potentials just above Vfb may indicate the elimination of surface recombination centers. The effects of concentration, temperature and time show critical values and that the Vfb potential shifts cathodically with the time of exposure to the reagent. The individual compounds such as NaF, Na?TiF?, Na?TiF? and TiOF? either present in the electrolyte or when applied on the electrode surface do not enhance the photocurrent. Similar is the effect of hole scavengers such as acetate and EDTA ions in the electrolyte. The spectral characteristics of HNO? + HF-treated electrodes show red shift as compared to the untreated ones in the band gap region. The fact that the intensity of light alters the photocurrents of the treated samples shows that the characteristics of the space charge layer are modified. There is increased efficiency with respect to the gaseous products evolved at a given applied potential. Chemical analyses showed that the fluorine content in the bulk of the electrode is below the detection limit of fluoride-ion-selective electrode and that Ti/M cation ratio is lower than unity in the solution. Mott-Schottky plots of the treated electrodes show only marginal variations in Nd-values. X-ray photoelectron spectra (XPS) of the variously treated electrode surfaces have been studied with reference to unreduced TiO? and SrTiO? both in binding energies (BE) and the relative intensities. For different oxides of Ti, the O(1s) binding energy remains unchanged, whereas Ti(2p?/?) shifts to lower BE. In the case of TiH and Ti-metal, the shifts in BE are maximum. Although both Ti(IV) and Ti(III) valence states can be expected in reduced TiO?, SrTiO?, Ti?O? etc., the same is not reflected by way of multiple bands in Ti(2p?/?) spectra. Therefore the continuous shift is correlated with the electron enrichment arising from oxygen vacancies and lower oxidation states of Ti. The intensity ratio, IO/ITi, decreases with the extent of reduction and the values are lower for the mechanically polished samples as compared to those of the as-prepared ones. The surface compositions of 1N HNO?-treated (60 min) electrodes show heterogeneity by way of the presence of surface hydroxyls, multiple bands for Ti(2p?/?) and variations in NSr/NTi and NO/NTi ratios. In the case of electrode surfaces treated with 1N HNO? + 0.05N HF, the concentration of F? ion is found inversely related to that of hydroxyl ions. The surface compositions show heterogeneity including NSr/NTi ratios. The sputter-etching experiments show that the penetration of F is ~50 Å below the electrode surface. By comparing with the XPS results on different oxides of Ti, it is evident that the surfaces of treated SrTiO? and BaTiO? electrodes cannot be considered of uniform concentrations with respect to the lattice constituents. There exist regions on the electrode surface where oxidised as well as reduced characteristics are individually dominant and hence, with differences in charge carrier concentrations. The enhanced photoresponse arises from the combined effect of elimination of undesired recombination centers in the space charge layer and the presence of surface regions differing in charge carrier density. The XPS results point towards the partial reduction of Ti(IV) on the titanate electrode, giving rise to catalytic effects similar to that reported for photocathodes. The possible consequence of such heterogeneous regions at the surface include the local enhancement in light intensity and also the local differences in Schottky-barrier heights. The effect of fluoride ions substituting at the oxygen sites may eliminate the oxygen vacancies, yet maintaining the electron compensation so that conductivity is not decreased. The present results also explain the reason for higher photon efficiency of titanates even without the metal catalysts and also the differences in photoresponse with various methods of surface treatment.