Nanoscale View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
PAPER
Cite this: Nanoscale, 2014, 6, 7585
View Journal | View Issue
Photoelectrochemical water oxidation by screen printed ZnO nanoparticle films: effect of pH on catalytic activity and stability† Monika Fekete,ab Wiebke Riedel,c Antonio F. Pattia and Leone Spiccia*ab Nanostructured ZnO films are promising photoanode materials in photoelectrochemical water splitting. While such ZnO photoanodes have achieved high activity and good light conversion efficiency in the UV spectral region, their application in water splitting devices has been hampered by the susceptibility of ZnO towards photocorrosion in aqueous electrolytes. We report a systematic investigation aimed at optimising the electrolyte solution to improve the long-term stability of ZnO photoanodes. A stability diagram, based on the band edge positions of ZnO and the pH-dependent photodegradation potentials of ZnO (relative to the decomposition of water), indicates that the optimum pH operating conditions for ZnO photoanodes lie between pH 9–12.5. To verify this prediction experimentally, the activity and longterm stability of uniform screen-printed nano-ZnO films was tested in a wide range of buffered and nonbuffered electrolytes (pH 6–13.5). The ZnO films were more active in buffered, than in non-buffered electrolytes, and the highest activities were observed close to the pKa of the phosphate and borate buffers used. Under zero applied potential, these screen-printed films achieved the highest reported photocurrents to date (0.42 mA cm2 at pH 6 and 0.67 mA cm2 at pH 10.5) for any pristine or modified ZnO-based water oxidation catalyst. The films were subjected to 12 h of controlled potential electrolysis, in selected electrolytes, under AM 1.5G simulated sunlight. The results are in good agreement with calculations based on thermodynamic data for ZnO. Films tested at pH 6 and 7 (representing typically
Received 10th April 2014 Accepted 9th May 2014
used operating conditions) degraded rapidly, whereas they exhibited the highest stability when tested in a pH 10.5 borate buffer. In this case, 75% of the initial photoactivity was preserved after 12 hours,
DOI: 10.1039/c4nr01935k
indicating that the lifetime of the electrode could be increased by over an order of magnitude compared
www.rsc.org/nanoscale
to standard testing conditions.
Introduction Photocatalytic water splitting over semiconductor electrodes has been the subject of intense research since the 1972 discovery of the Honda–Fujishima effect on titanium dioxide surfaces.1 In the past four decades, a wide range of semiconductors have been evaluated for their potential as efficient photocatalysts for hydrogen production by direct solar water splitting vide infra.2–11
a
School of Chemistry, Monash University, Victoria 3800, Australia. E-mail: leone. [email protected]; Fax: +61 3 9905 4597; Tel: +61 3 9905 4526
b
ARC Centre of Excellence for Electromaterials Science (ACES), Monash University, Victoria 3800, Australia
c Institut f¨ ur Heterogene Materialsysteme, Helmholtz-Zentrum f¨ ur Materialien und Energie, Hahn-Meitner-Platz 1, 14109 Berlin, Germany
† Electronic supplementary information (ESI) available: XRD pattern of a fresh ZnO lm (Fig. S1); LSV testing of 1-, 2- and 3-layer screen-printed ZnO lms (Fig. S2); dependence of photocurrent on light intensity (Fig. S3); IPCE of the screen-printed ZnO lms (Fig. S4); SEM image of ZnO lm aer CPE at pH 12 (Fig. S5); and ICP-TOF-MS analysis of the Zn-content of lms (Table S1). See DOI: 10.1039/c4nr01935k
This journal is © The Royal Society of Chemistry 2014
In general, if sufficient photovoltage is generated at the interface of an n-type semiconductor and the aqueous electrolyte, photo-oxidation of water by minority carriers (holes) can occur at the anode. Similarly, photo-reduction of water can be achieved by electrons generated at a p-type semiconductor photocathode.12 For these reactions to proceed, the band alignment of the semiconductor electrodes is crucial. Photooxidation at the anode becomes possible if the upper edge of the valence band of an n-type semiconductor lies at a potential more positive than that required for water oxidation. Analogously, photo-reduction proceeds when the lower edge of the conduction band of a p-type cathode is more negative than the potential required for hydrogen evolution by water reduction.13 Examples of materials studied as potential photocathodes include p-GaP,14 p-InP,15 p-GaInP2,16 p-Si,11a CdTe,10,17 p-WS2 (ref. 18) and Cu2O.19 Semiconductors, such as TiO2,1 WO3,2 GaN,20 SnO2,21 Fe2O3,3 SrTiO3,5 BaTiO3,5 ZnO22 or ZnS23 are promising photoanode candidates because they absorb high energy photons and are, therefore, capable of generating enough photovoltage (a minimum of 1.23 V) to drive unassisted water oxidation or both water oxidation and reduction (e.g.,
Nanoscale, 2014, 6, 7585–7593 | 7585
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
TiO2, TaON).24 The disadvantage of wide-bandgap semiconductors is that their light absorption is mostly limited to the UV region of the solar spectrum.11a Recently, advances have been made towards extending the absorption of these materials into the UV, either by incorporating dopants to narrow the bandgap25 or by the addition of hydrophobic dyes onto the electrode surface.11b Apart from the fundamental limitation of light absorption, photocorrosion of semiconductor electrode materials in aqueous electrolytes is a major issue that has hampered their practical use.24 At the illuminated surface of an n-type semiconductor anode surface, two main competing reactions take place; the decomposition of water by the photogenerated holes and, less desirably, the corrosion of the electrode itself due to destabilization caused by hole trapping on the semiconductor surface.13,26 Based on thermodynamic and kinetic criteria, the possibilities of avoiding photodecomposition of the anode material can be determined. Gerischer and co-workers published a series of studies discussing these criteria, focusing partly on ZnO (amongst other materials, such as TiO2 or CdS).13,26,27 ZnO has attracted much attention in the past decades as a favourable photoanode material towards achieving efficient photoelectrochemical water oxidation, due to its good electron mobility, excellent optical properties, wide availability and relatively low toxicity.22b,28 More recently, nanostructuring of ZnO lms has been shown to improve photocurrent yields due to their large surface area relative to their illuminated geometric area, decreased reectivity and improved charge transfer to the conducting electrode substrate.22a ZnO nanorods, nanosheets, nanowires and various nanoparticles have been synthesised in order to increase the efficiency of water oxidation.22,25,28a,29 Many recently developed ZnO-based photo-electrocatalysts for water oxidation exhibit improved light absorption properties with high photocurrent densities. Relatively little data is available, however, on the photostability and long-term performance of these electrodes. Yang et al. reported that the activity of Ndoped ZnO nanorods, prepared by a hydrothermal method, when tested in a 0.5 M NaClO4 electrolyte (pH 7),22b dropped by 25% within the rst 20 minutes, but then increased up to 90% of the original level aer 90 minutes. The authors suggested that the chemical instability of ZnO could be reduced by covering the ZnO nanowire core with, for example, a thin TiO2 shell. Liu et al.29e tested this approach by coating ZnO nanowire arrays with a thin, passivating TiO2 layer. Testing in highly alkaline electrolytes revealed a 10% drop in photoactivity for the uncoated ZnO NRAs aer 1 hour, whereas the activity of the passivated ZnO NRAs remained unchanged within the same timeframe. According to the authors, this was the rst report to use ZnO-based water oxidation catalysts in highly alkaline electrolytes. Indeed, most studies on ZnO photoelectrocatalysts focus on their operation in close to neutral media. Qiu et al. tested undoped and N-doped ZnO nano-tetrapods in a pH 7.0 Na2SO4 electrolyte solution,29a and found that the initial activity of the N-doped lms was higher compared to the undoped lms. When operated continuously over one hour, however, the activity of the undoped nanorods remained close to 100%, while the N-doped nanorods retained 90% of their
7586 | Nanoscale, 2014, 6, 7585–7593
Paper
activity within the same timeframe. Hassan et al. claimed ‘exceptional stability’ for (undoped) ZnO nano-tetrapods prepared by a thermal evaporation method when tested in a 0.5 M Na2SO4 solution.29b Inspection of their controlled potential electrolysis (CPE) data, collected over a 20 minute period, conrms that the less active catalyst lm indeed retains close to 100% activity, while the performance of the more active sample drops by approximately 20% over the same testing period. This result, in fact, seems to correspond well with the data reported by Yang et al.22b for studies at pH 7. While the optimum pH and electrolyte composition for maximum oxygen evolution has been extensively studied for “dark” electrocatalysts,30 these conditions have not been well established for semiconductor photocatalysts. Phosphate and borate buffers have been shown to enhance the rate of oxygen evolution when used with various water oxidation catalysts, for example, manganese, cobalt or nickel oxides, mainly by improving proton conduction between the anode and cathode.31 By preventing the accumulation of protons near the anode surface (which could destabilise the electrode material), these buffers signicantly increase the lifetime of the catalysts. Mohamad and co-workers studied the pH conditions for the sol–gel synthesis of ZnO nanoparticles, concluding those synthesized in the pH 8 to 11 region possessed optimal optical properties.32 In this study, thermodynamic considerations are used in an attempt to optimise the operating conditions for water splitting cells employing ZnO-based photoanodes. The pH range where the ZnO anode is theoretically stable against photodecomposition was rst predicted with the help of an E-pH stability diagram comparing decomposition potentials for the aqueous electrolyte and the ZnO electrodes. To test the prediction experimentally, reproducible ZnO nanoparticle lms were screen-printed and their photocatalytic performance was tested in a range of electrolyte solutions. The effects of the phosphate and borate buffers on photocurrent generation and on the longterm stability of ZnO under operational conditions are discussed. A particularly promising result was the observation of photocurrent even in the absence of an applied bias in an optimised, buffered electrolyte.
Background Stability of ZnO photo-electrocatalysts: thermodynamic considerations. The principles of the photodegradation of semiconductor electrodes in aqueous media had been established as early as the 1970's. For example, a series of studies by Gerischer and co-workers describe the thermodynamic criteria that need to be met to protect electrodes from photocorrosion during operation in water splitting cells.13,26,33 Briey, a semiconductor is protected against photodecomposition by electrons if its cathodic photodecomposition potential (EDCe) lies negative to the conduction band and against photodecomposition by holes if its anodic photodecomposition potential (EDCh+) lies positive to the valence band edge. This “ideal” situation is shown in Fig. 1A. Zinc oxide is stable against photodecomposition by electrons but is
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Nanoscale
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
number of electrons involved in eqn (1)–(3) and F is the Faraday constant. According to Zhang, the decomposition potentials under acidic (1) and highly alkaline (3) conditions are at 0.89 V and 0.43 V vs. SHE, respectively.35 The EDCh+ corresponding to (2) was calculated in this study as 1.08 V vs. SHE. Based on these calculations and experimental EIS data, a proposed stability diagram across the pH spectrum was constructed.
Experimental Screen printing of the ZnO lms (A) Band diagram for an ideally stable n-type semiconductor at pH ¼ 0. (B) Band diagram of ZnO at pH ¼ 0. EFB, marked with *, was calculated based on measurements presented in this study. Fig. 1
intrinsically unstable to photodecomposition by holes, as shown in Fig. 1B. Decomposition can be avoided, however, when water is oxidised at a potential negative to EDCh+. Towards this end, the relative positions of the potentials corresponding to the competing reactions need to be determined over the accessible pH range. The Nernstian shi (59 mV/pH) in the redox potential of the 1/2 O2/H2O couple with increasing solution pH is well known.34 The anodic photodecomposition potential of semiconducting oxides is generally independent of pH, but this does not hold true for zinc oxide.24 The decomposition potential is determined by the dissolved products formed when the electrode photocorrodes.35 A constant decomposition potential over the entire pH range would only be possible if the same product formed at any given solution pH. The Pourbaix-diagram of zinc oxide, however, indicates that at acidic to neutral pH, the predominant dissolved zinc species is Zn2+, at highly alkaline pH ZnO22, and HZnO2 between pH 9.2–13.2.36 Accordingly, the equations describing the electrode decomposition are: ZnO + 2p+ ¼ Zn2+ + 1/2O2
(1)
ZnO + 3OH + 2p+ ¼ HZnO2 + 1/2O2 + H2O
(2)
ZnO + 4OH + 2p+ ¼ ZnO22 + 1/2O2 + 2H2O
(3)
The solubility of ZnO is the lowest at pH 9.3 (107–108 M), and increases exponentially both towards more acidic and alkaline pH values.35 (Naturally, soluble species can co-exist in same solution, their relative concentrations determined by equilibria, as indicated in the Pourbaix diagram.) The decomposition potential EDCh+ for each reaction can be calculated from the change in the Gibbs free energy upon the breakdown of the electrode material: DG ¼ RT ln K
(4)
DG ¼ nFEDCh+
(5)
EDCh þ ¼
RT ln K nF
(6)
where K is the equilibrium constant for reactions (1)–(3), R is the ideal gas constant, T is the absolute temperature, n is the
This journal is © The Royal Society of Chemistry 2014
Screen printing, a well-established method for the preparation of highly uniform TiO2 nanoparticle lms in DSC fabrication,37 was adapted to deposit commercial zinc oxide nano-powders. The ZnO nanoparticles (Sigma Aldrich, d < 200 nm) were printed onto uorine-doped tin oxide (FTO) coated glass slides (10 U per square, DyeSol) using a modied screen-printing method adapted from a published procedure.30a In a typical preparation, ZnO nanoparticle powder (1.0 g) was mixed with 5.0 ml ethylcellulose binder solution (10 wt% in ethyl alcohol, Sigma Aldrich 433837) and 4.0 ml terpineol. The paste was screenprinted onto FTO glass substrates in 1, 2 and 3 layers (resulting in a thickness of 0.6 mm, 1.7 mm and 2.4 mm, respectively). The lms were sintered in air on a hotplate according to a heating program developed for TiO2 and indium-doped tin oxide (ITO) lms in dye sensitized solar cell (DSC) fabrication.38 The photocatalytic activity of the ZnO lms was tested by linear sweep voltammetry (LSV) in a 0.6 M Na2SO4 solution (pH 6). Based on this initial study, the 2-layer lms were chosen for subsequent characterization and photocatalytic testing. Phosphate (0.6 M, pH 6.5, 7, 7.5, 8.5 and 12) and borate (0.6 M, pH 8.5, 9, 9.5, 10, 10.5 and 11) pH buffer solutions as well as Na2SO4 and NaClO4 solutions (0.6 M) were made freshly with Millipore quality water just prior to electrochemical testing. Film characterisation and testing of photocatalytic activity Scanning Electron Microscopy (SEM) was performed with a JEOL JSM-7001F FEGSEM instrument tted with a Bruker 10 mm2 Si dri detector analysis system, operated at 30 kV. The lms on the FTO glass substrate were previously sputter-coated with platinum (1 nm). SEM characterisation was carried out the Monash Centre for Electron Microscopy (MCEM). X-ray Diffraction (XRD) patterns were obtained using a Philips instrument with powder XRD capability. The printed lms were gently scraped off the supporting glass slides and sprinkled onto a quartz plate thinly coated with Vaseline. XRD scans were taken using Cu Ka radiation, with a 1 divergence slit, 0.2 receiving slit and carbon monochromators at a 0.5 min1 scan rate (2F range 25 –80 ). Testing of catalytic activity of the electrodeposited ZnO lms was performed using a Bio-Logic VSP Modular 5-channel potentiostat, at room temperature (20 C), in a Zahner-Elektrik PECC-1 electrochemical cell. The at quartz window of the electrochemical cell helped to ensure well reproducible conditions during photoelectrochemical performance testing.
Nanoscale, 2014, 6, 7585–7593 | 7587
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
Photocurrent generation was systematically tested in electrolytes with pH ranging from 6 to 13.5, with particular focus on the pH 7.0–10.5 region. Illumination was provided by an Oriel 150 W solar simulator equipped with an ozone-free Xe lamp and an AM1.5G lter to match the solar spectrum. A fresh lm was used for each measurement. The electrolytes used were: 0.6 M Na2SO4, 0.6 M phosphate buffers (at pH 6.5, 7.0, 7.5, 8.0, 8.5, 12), 0.6 M borate buffers (at pH 8.5, 9.0, 9.5, 10, 10.5 and 11), a 0.6 M NaClO4 solution adjusted to pH 12 with NaOH and a 0.6 M NaOH solution (pH 13.5). Linear sweep voltammetry (LSV) testing was carried out at a 5 mV s1 scan rate, 0.2 to 1.5 V vs. Ag/AgCl, in the positive or oxidative direction, under chopped illumination (5 s on, 5 s off) at 1 Sun light intensity (100 mW cm2). The overpotential for O2 evolution was calculated as the difference between the applied potential (Eapp) vs. Ag/AgCl and the E0 of oxygen evolution at the appropriate pH (vs. Ag/AgCl), based on the Nernst equation, referenced against Ag/AgCl: h(V) ¼ Eapp(V) E0(V) ¼ Eapp(V) (1.23 V 0.2V 0.059 pH). Incident Photon to Charge Carrier Conversion Efficiency (IPCE) was measured at pH 10.5 (in a borate buffer) following a previously described procedure,39 using a 150 W Oriel solar simulator (with a horizontal lightbeam) equipped with an AM1.5G lter and bandpass lters. The dependence of photocurrent on light intensity was measured by recording linear sweep voltammograms of each lm between 0.5–1.5 V vs. Ag/AgCl, at a 5 mV s1 scan rate. The light intensity was adjusted to 0, 0.5, 1, 2 and 3 Suns using neutral density lters. Long-term performance was tested by Controlled Potential Electrolysis (CPE), at 120 mV overpotential corresponding to 670 mV, 620 mV, 490 mV, 410 mV, 320 mV and 230 mV vs. Ag/ AgCl, at pH 6.0 (0.6 M Na2SO4), 7.0 (phosphate buffer), 9.0 (borate buffer), 10.5 (borate buffer), 12.0 (phosphate buffer and 0.6 M NaClO4/NaOH) and 13.5 (0.6 M NaOH), respectively; under 1 Sun illumination over 12 hours. Additionally, CPE was carried out at 0.2 V, 0 V and 0.66 V (h ¼ 250 mV) vs. Ag/AgCl under 1 Sun illumination in a pH 10.5 borate buffer for 12 hours. Electrochemical Impedance Spectroscopy measurements (Mott–Schottky plots) were recorded using a Bio-Logic VSP Modular 5-channel potentiostat operated in Staircase PotentioElectrochemical Impedance Spectroscopy (SPEIS) mode. Mott– Schottky plots were recorded 1 V to +1 V vs. Ag/AgCl, using a RC equivalent circuit tting model. Inductively Coupled Plasma-Time of Flight-Mass Spectrometry (ICP-TOF-MS) analysis was carried out using a GBC Optimass 9500 instrument. Raw count rates from the analyses were externally standardised by means of a calibration curve constructed using Zn2+ stock solutions (containing Zn2+ ions in 0, 10, 25, 50 and 100 ppb concentrations). Indium was used as an internal standard. The pristine and post-12 hour operation ZnO lms (of 0.49 cm2 surface area each) were dissolved in 10 ml 70% nitric acid and diluted to 50 ml stock solutions prior to the ICP-MS analysis. Electrolyte solutions aer 12 hours of lm operation were also analysed.
7588 | Nanoscale, 2014, 6, 7585–7593
Paper
Results and discussion Flatband potential and stability diagram In aqueous electrolytes, the atband potential, which is dened by the relative positions of the band edges, is strongly inuenced by the solution pH. At each pH, the atband potential can be experimentally determined using the Mott–Schottky relationship: 2 kT CSC 2 ¼ E EFB (10) e330 ND e where e is the electron charge, 3 is the dielectric constant of ZnO, 30 is the permittivity of vacuum, ND is the donor density, E is the applied potential at the electrode, EFB is the atband potential of the ZnO electrode and k is the Boltzmann constant. Gerischer et al. found that, due to the formation of an additional electric double layer by the interaction of the ZnO surface with H+ and OH ions present in the solution, the pH dependence of the atband potential over a large pH range follows a Nerstian behaviour, i.e., a shi of 59 mV/pH.33 Therefore, the relative positions of the band edges and the water oxidation and reduction potentials remain unchanged over a large part of the pH scale. The 59 mV/pH shi was conrmed experimentally within the examined pH 6–10.5 range (marked by crosses in Fig. 2). Anodic and cathodic decomposition potentials were taken from Zhang.35 Where data was not available, the potentials were calculated based on the change in Gibbs free energy upon formation of the photodecomposition products and equilibria between the dissolved species (dened by the Pourbaix diagram.) ZnOH+, which forms in a small amount around pH 9, was not included in this calculation. The proposed ZnO stability diagram is shown in Fig. 2. Fig. 2 suggests that long-term operation of a ZnO-based water oxidation catalyst without extensive photodegradation should be possible in the pH 9–12.5 range. At pH 6 and 7, dissolution of the lm is expected even if water oxidation reaction occurs without applied overpotential (i.e., at the thermodynamic limit). If an overpotential is required, the blue line indicating oxygen evolution from water shis to more positive potentials, which narrows the window for optimum operation even further. These considerations are valid when no other species are present in the electrolyte that could either form a precipitate with ZnO or complex with the dissolved zinc species. When operating in buffered electrolytes, such possibilities need to be considered. Based on these theoretical considerations a working hypothesis was developed and tested in practice by examining the pH-dependent activity and long-term stability of ZnO lms during photo-electrochemical water oxidation.
Film deposition and characterization The crystal phase of the ZnO lms was conrmed to be the commonly occurring wurtzite phase by powder-XRD diffraction
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Paper
Nanoscale
Fig. 2 Stability diagram of ZnO. Black crosses mark the EFB values calculated based on Mott–Schottky measurements. Vertical dashed lines represent the equilibrium pH between the three main dissolved Zn-species. The light blue line shows the thermodynamic O2 evolution potential, while the light red line represents the anodic photodecomposition potential of ZnO by photogenerated holes (EDCH+). The photoanode is stable in the pH region where EDCH+ is more positive than the O2 evolution potential, approximately between pH 9–12.5. The dark blue line shows the thermodynamic hydrogen evolution potential, which lies positive to the cathodic decomposition potential (EDCe, marked with the dark red line) of ZnO across the entire pH range.
(see Fig. S1†). SEM studies showed that the porous catalyst lms are composed of small spherical (40–50 nm) and larger, “bricklike” particles, approximately 50 100–200 nm in size (see below). The uniformity of the two-layered lms was tested by ICP-MS analysis (as described in the Experimental section). The lms were found to contain the same amount of Zn (31 mg) within a 2% error (based on analysis of three fresh lms). Testing of the activity of the ZnO lms by LSV in a 0.6 M Na2SO4 solution revealed that the photocurrent was very similar for the 1- and 2-layered lms, (see Fig. S2†) which decreased slightly on addition of a third layer. Based on these results, the more robust two-layer lms were used in further studies. A dark current above 1.2 V applied bias (vs. Ag/AgCl) was observed in each case, which could be due to the degradation of ZnO, as predicted by the stability diagram, as well as water oxidation.
Testing of photocatalytic activity as a function of pH LSV scans recorded at pH 7.0 (phosphate buffer), pH 9.0 (borate buffer) and pH 13.5 (0.6 M NaOH) are shown in Fig. 3. The onset of dark current was found at 1.0 V, 1.1 V and 0.7 V, (vs. Ag/AgCl), at pH 7.0, 10.5, and 13.5, respectively. These values correspond to the upper stability limit of ZnO, and could be arising in part from lm degradation. Under illumination, signicant photocurrents were observed even at 0.2 V vs. Ag/AgCl, which equals 0 V applied potential vs. SHE (zero bias condition). At pH 7, the photocurrent increased with the applied bias up to 0.2 V vs. Ag/AgCl, where it reached a plateau of 0.75 mA cm2. At pH 10.5 and 13.5, the photocurrent showed little change throughout the measured voltage range and maximised at 0.70 and 0.55 mA cm2, respectively. These
This journal is © The Royal Society of Chemistry 2014
relatively high photocurrents at close to 0 V applied potential set the screen-printed ZnO apart from other recently reported highperforming nano-ZnO -based water oxidation catalysts, where the photocurrent was typically found to be highly dependent on the applied bias.22,29f,g The photocurrent density for the screen-printed ZnO photoanodes scales almost linearly with illumination power (up to 2 Suns, with near-saturation at 3 Suns) at low applied potentials. At the higher applied potentials (>1.2 V), the dark (degradation) current contributes to the measured current, and the difference between the currents recorded with and without illumination diminishes (Fig. S3†). The incident photon-to electron conversion (IPCE, shown in Fig. S4†) was found to be relatively high at 370 nm (41%) and 390 nm (10%), but effectively zero beyond 410 nm, as typically found for ZnO photoanodes.22b,29g The dependence of photocurrent generation on pH is shown in the inset of Fig. 2. For the non-buffered Na2SO4 electrolyte at pH 6.0, the photocurrent (0.28 mA cm2) was much lower than that measured for the buffered electrolytes (0.70 mA cm2 was measured at pH 6.5 in a phosphate buffer). In the phosphate and borate buffers, the photocurrents uctuate with local maxima close to their pKa. The highest photocurrent (0.82 mA cm2) was measured at pH 7.5 in a phosphate buffer (pKa2 ¼ 7.2) and at pH 9.5 in a borate buffer (pKa2 ¼ 9.2), where buffering was the most effective. The variation in photocurrent within the buffering range was relatively small, although the photocurrent dropped quite signicantly at pH 11, where the boric acid/sodium borate buffer is less effective. Higher current was measured again at pH 12 in a phosphate buffer, within the Na2HPO4/Na3PO4 buffering range (pKa3 ¼ 12.7). When an unbuffered NaClO4/NaOH electrolyte was used at the same pH,
Nanoscale, 2014, 6, 7585–7593 | 7589
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
Paper
LSV testing of the screen-printed ZnO photoanodes under 1 Sun chopped illumination, at pH 7 (blue line), pH 10.5 (red line) and 13.5 (black line). Dashed vertical lines represent the thermodynamic O2 evolution potential at the corresponding pH. The inset shows photocurrents at 0 mV overpotential at each pH, measured in various buffered and non-buffered electrolytes. Fig. 3
the current dropped slightly from 0.69 to 0.56 mA cm2. A similar photocurrent was measured at pH 13.5, in a 0.6 M NaOH solution. Thus, little variation in photocurrent with pH was observed when the electrolyte provided sufficient buffering capacity. More pronounced differences in the catalytic performance of the ZnO lms with pH were observed when the cells were subjected to long-term constant potential electrolysis (CPE).
Long-term stability Fig. 4 shows that in the pH 6.0 electrolyte, the activity, initially quite high (0.78 mA cm2) decreased to 0 mA cm2 within just 90 minutes. Even in the buffered pH 7.0 electrolyte, complete degradation was observed aer 2 hours. At pH 9, stability was
Fig. 4 Long-term stability over 12 h, under 1 Sun light intensity, (h ¼ 120 mV) of the ZnO screen-printed films at various pH. (pH 6 : 0.6 M Na2SO4; pH 7 and 12 : 0.6 M phosphate buffers; pH 9 and 10.5 : 0.6 M borate buffers; pH 13.5 : 0.6 M NaOH).
7590 | Nanoscale, 2014, 6, 7585–7593
much improved, 58% of the initial photocurrent was still preserved aer 12 h. An even better stability was observed at pH 10.5, where approximately 75% of the photocurrent was retained aer 12 h. At pH 12, stability was decreased; the current dropped to 50% aer 4 hours and was almost completely deactivated by the end of the 12 hour electrolysis. At 13.5, the photocurrent dropped by 50% in less than 90 minutes and then stabilised at close to 25% of the initial photocurrent. SEM images taken of the catalyst lms before and aer CPE testing, are shown in Fig. 5. The lm tested at pH 6.0 was completely dissolved off the FTO glass substrate. At pH 7.0, the lm had also almost completely dissolved, and the FTO crystals
Fig. 5 SEM images of (A) a fresh screen-printed ZnO film, (B–F) after 1 h CPE testing under 1 Sun illumination. (B) pH 7 (phosphate buffer); (C) pH 9 (borate buffer); (D) pH 10.5 (borate buffer); (E) pH 12 (phosphate buffer); (F) pH 13.5 (0.5 M NaOH).
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Paper
Nanoscale
Fig. 6 Photocurrent degradation at various applied potentials in a pH 10.5 borate buffer (under 1 Sun illumination).
can be seen underneath the remains of the lm in Fig. 5B. According to Awad and Kamel,40 phosphate ions (especially when present as a mixture of Na2HPO4 and NaH2PO4), can result in the formation of complex phosphates on the surface. A highly polymerised, semi-glassy interface has been proposed to be formed via a dissolution–precipitation process:41 Zn2+ + 2H2PO4 4 Zn(H2PO4)2
(7)
3Zn(H2PO4)2 4 Zn3(PO4)2$4H2O + 4H3PO4
(8)
The resulting interface is likely to adversely affect the catalytic activity of the ZnO lm and possibly accelerate the breakdown of the lm. The breakdown of the passive lm was postulated to involve diffusion of Zn2+ ions through the passive layer. The formation of H2PO4 near the lm surface could also promote the dissolution process.
At pH 9.0, the nanoparticles are preserved, but are covered with very small particles, possibly indicating deposition of a borate species on the catalyst surface (Fig. 5C). The solubility zinc species is the lowest at pH 9.3 (between 107 and 108 M), and, incidentally, the solubility of borates (mostly in the form of H4BO7 and B4O72) is also the lowest between pH 7–9.5 (ca. 102 to 103 M).36 Insoluble zinc borate species could passivate the ZnO surface, signicantly decreasing the surface area available for contact between the catalyst and the electrolyte, leading to loss of activity. At pH 10.5, there is less degradation and no borate deposition was observed, correlating well with the good stability of photocurrent over 12 h (Fig. 5D). At pH 12, the loss in activity was accompanied by extensive degradation of the ZnO lm, accelerated by the presence of PO43 ions (see Fig. 5E). The breakdown and coverage of the catalysts with a zinc-phosphate passive layer is evidenced by the formation of nodules with high phosphorous content on the surface.42 The long-term behaviour of the ZnO catalyst lm was also tested in a 0.6 M NaClO4/NaOH solution at pH 12 to examine lm degradation in the absence of phosphate ions. In this case, the photocurrent was somewhat more stable (20% retained aer 12 h), and the particles showed less degradation than in the phosphate buffer (shown in Fig. S5†). At pH 13.5, the formation of larger, spherical grains was observed (Fig. 5F). Aurian-Blajeni and Tomkiewicz proposed that these passive lms are composite layers of the electrolyte and the oxide lm,43 oen found to retain properties typical of an amorphous semiconductor lm.44 The passivation processes at highly alkaline pH were summarised by Dirkse.45 The Zn(OH)2 lm on the surface of ZnO lm dissolves as Zn(OH)3 or Zn(OH)42 and, once the electrolyte layer covering the
Comparison of operating conditions, initial activity and long-term stability of ZnO-based photoanodes with those reported in recent literature
Table 1
Photoanode
Electrolyte
Screen-printed ZnO NPs Screen-printed ZnO NPs Screen-printed ZnO NPs
0.6 M Na2SO4 (pH 6) 0.6 M Na2SO4 (pH 6) 0.6 M Borate buffer (pH 10.5) 0.6 M Borate buffer (pH 10.5) 0.5 M NaClO4 (pH 7.4) with PB 0.5 M NaClO4 (pH 7) with PB 0.1 M KOH (pH 13) 0.5 M Na2SO4 (pH 7) 0.5 M Na2SO4
Screen-printed ZnO NPs Nano-ZnO N-doped ZnO NRA TiO2-coated ZnO nanowires Long, H-annealed ZnO nanowires Hydrogenated, hierarchical ZnO NRAs N-doped ZnO nano-tetrapods ZnO nano-tetrapods N-and Al-co-doped ZnO thin lms a
0.5M Na2SO4 (pH 7) with PB 0.5 M Na2SO4 0.5 M Na2SO4
J initial at h ¼ 120 mV (mA cm2), by LSV
PCa retained aer 1 h CPE (%)
Max. CPE duration shown (h)
0.42 0.42 0.67
26b 26b 100b
12 At 3 h 12
0 0 75
Present study Present study Present study
0.67
100b
At 3 h
96
Present study
0
—
—
PC retained aer max. CPE testing (%)
—
0.01
85c
1.4
90
0.2 0.1 0.2
100d — —
1.0 — —
100 — —
0.05
90c
1.0
90
0.30 —
80e —
0.02 0
— —
Ref.
Wolcott et al.22a Yang et al.22b Liu et al.29e Cooper et al.29f Yao et al.29g Qiu et al.29a Hassan et al.29b Shet et al.25
PC ¼ photocurrent. b h ¼ 120 mV. c Eapp ¼ 0.75 V vs. SHE. d Based on LSV aer 1 h. e Eapp ¼ 1.0 V vs. SHE.
This journal is © The Royal Society of Chemistry 2014
Nanoscale, 2014, 6, 7585–7593 | 7591
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
electrode becomes supersaturated, re-deposits again as amorphous Zn(OH)2. XRD analysis aer the 12 hour CPE at pH 13.5 conrmed that the crystalline wurtzite phase had become more disordered and much less crystalline (see Fig. S1†). The less crystalline material was found to generate a stable photocurrent over the next 10 hours of CPE, but it was less active than the crystalline phase. The formation of larger particles also reduces the surface area and decreases the number of active surface sites, and most likely contributes to the lower activity observed aer 2 hours. Long-term stability was also investigated as a function of applied potential at pH 10.5. As shown in Fig. 6, the photocurrent versus time prole followed the same pattern over 12 hours of testing at 0.2, 0 and 0.66 V vs. Ag/AgCl. The current density increased when a small bias was applied (0 V vs. Ag/AgCl ¼ 0.2 V vs. SHE), but did not increase further at higher potentials. The loss of activity was the same (%) in all three experiments. The good stability of the ZnO lms at pH 10.5 was also conrmed by ICP-MS analysis of the electrolyte and the used lms. The amount of zinc that leached into solution was 75% of the initial photoactivity was retained aer 12 hours of catalyst operation. This result was corroborated by ICP-MS analysis of the Zn-content of the electrolyte solutions aer 12 hours, which showed a minimum zinc “leakage” from the lm. These ndings could be used to improve the lifetime of other high-performing ZnO-based catalysts. The screen-printed ZnO anode materials are promising candidates for integration with dark co-catalysts, such as electrodeposited or screen-printed water oxidising metal oxides. Sensitization of the ZnO lms with light-harvesting dyes could also be a promising approach for future applications, and with their activity further enhanced, ZnO lms could represent an attractive option for commercial water splitting.
Acknowledgements The authors would like to thank the Monash Institute of Graduate Research and the Monash Faculty of Science for scholarship support and a Postgraduate Publication Award (MF), the Australian Research Council (through the ARC Centre of Excellence for Electromaterials Science to LS) for nancial support and the Monash Centre for Electron Microscopy (MCEM) for instrument access and expertise. Mr R. Mackie is acknowledged for his help with XRD data acquisition and Dr M Raveggi for his help with ICP-MS measurements.
References 1 A. Fujishima and K. Honda, Nature, 1972, 238, 37–38. 2 G. Hodes, D. Cahen and J. Manassen, Nature, 1976, 260, 312– 313. 3 A. Kay, I. Cesar and M. Graetzel, J. Am. Chem. Soc., 2006, 128, 15714–15721. 4 K. Ohashi, J. McCann and J. O. Bockris, Nature, 1977, 266, 610–611. 5 I. E. Paulauskas, J. E. Katz, G. E. Jellison, N. S. Lewis and L. A. Boatner, Thin Solid Films, 2008, 516, 8175–8178. 6 H. Kato, K. Asakura and A. Kudo, J. Am. Chem. Soc., 2003, 125, 3082–3089. 7 J. Sun, C. Liu and P. Yang, J. Am. Chem. Soc., 2011, 133, 19306–19309. 8 A. Heller, Science, 1984, 223, 1141–1148. 9 S. W. Boettcher, E. L. Warren, M. C. Putnam, E. A. Santori, D. Turner-Evans, M. D. Kelzenberg, M. G. Walter, J. R. McKone, B. S. Brunschwig, H. A. Atwater and N. S. Lewis, J. Am. Chem. Soc., 2011, 133, 1216–1219. 10 X. Mathew, A. Bansal, J. A. Turner, R. Dhere, N. R. Mathews and P. J. Sebastian, J. New Mater. Electrochem. Syst., 2002, 5, 149–154. 11 (a) M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473; (b) W. J. Youngblood, S.-H. A. Lee, K. Maeda and T. E. Mallouk, Acc. Chem. Res., 2009, 42, 1966–1973; (c) C.-H. Liao, C.-W. Huang and J. C. S. Wu, Catalysts, 2012, 2, 490–516; (d) X. Chen, S. Shen, L. Guo and S. S. Mao, Chem. Rev., 2010, 110, 6503–6570; (e) A. J. Bard and M. A. Fox, Acc. Chem. Res., 1995, 28, 141–145.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Paper
12 R. Memming, Semiconductor Electrochemistry, Wiley-VCH, Rellingen, Germany, 2001. 13 H. Gerischer, J. Electroanal. Chem. Interfacial Electrochem., 1977, 82, 133–143. 14 B. Kaiser, D. Fertig, J. Ziegler, J. Klett, S. Hoch and W. Jaegermann, ChemPhysChem, 2012, 13, 3053–3060. 15 A. Heller, E. Aharon-Shalom, W. A. Bonner and B. Miller, J. Am. Chem. Soc., 1982, 104, 6942–6948. 16 H. Wang, T. Deutsch and J. A. Turner, J. Electrochem. Soc., 2008, 155, F91–F96. 17 K. Ohashi, K. Uosaki and J. O. M. Bockris, Int. J. Energy Res., 1977, 1, 25–30. 18 J. A. Baglio, G. S. Calabrese, D. J. Harrison, E. Kamieniecki, A. J. Ricco, M. S. Wrighton and G. D. Zoski, J. Am. Chem. Soc., 1983, 105, 2246–2256. 19 A. Paracchino, N. Mathews, T. Hisatomi, M. Stek, S. D. Tilley and M. Graetzel, Energy Environ. Sci., 2012, 5, 8673–8681. 20 (a) S.-W. Ryu, Y. Zhang, B. Leung, C. Yerino and J. Han, Semicond. Sci. Technol., 2012, 27, 015014; (b) I. Waki, D. Cohen, R. Lal, U. Mishra, S. P. DenBaars and S. Nakamura, Appl. Phys. Lett., 2007, 91, 093519. 21 M. Tian, W. Peng, B. Shang, W. Tao and Y. Xiao, Int. J. Hydrogen Energy, 2012, 37, 12827–12832. 22 (a) A. Wolcott, W. A. Smith, T. R. Kuykendall, Y. Zhao and J. Z. Zhang, Adv. Funct. Mater., 2009, 19, 1849–1856; (b) X. Yang, A. Wolcott, G. Wang, A. Sobo, R. C. Fitzmorris, F. Qian, J. Z. Zhang and Y. Li, Nano Lett., 2009, 9, 2331–2336. 23 (a) H. M. Chen, C. K. Chen, R.-S. Liu, C.-C. Wu, W.-S. Chang, K.-H. Chen, T.-S. Chan, J.-F. Lee and D. P. Tsai, Adv. Energy Mater., 2011, 1, 742–747; (b) P. P. Hankare, P. A. Chate and D. J. Sathe, J. Alloys Compd., 2009, 487, 367–369. 24 S. Chen and L.-W. Wang, Chem. Mater., 2012, 24, 3659–3666. 25 S. Shet, K.-S. Ahn, T. Deutsch, H. Wang, N. Ravindra, Y. Yan, J. Turner and M. Al-Jassim, J. Mater. Res., 2010, 25, 69–75. 26 H. Gerischer, J. Electrochem. Soc., 1966, 113, 1174–1181, discussion 1181–1172. 27 H. Gerischer, Electrochim. Acta, 1989, 34, 1005–1009. 28 (a) L.-W. Ji, S.-M. Peng, J.-S. Wu, W.-S. Shih, C.-Z. Wu and I. T. Tang, J. Phys. Chem. Solids, 2009, 70, 1359–1362; (b) Y. Zhang, M. K. Ram, E. K. Stefanakos and D. Y. Goswami, J. Nanomater., 2012, 624520, 624522. 29 (a) Y. Qiu, K. Yan, H. Deng and S. Yang, Nano Lett., 2012, 12, 407–413; (b) N. K. Hassan, M. R. Hashim and N. K. Allam, Chem. Phys. Lett., 2012, 549, 62–66; (c) I. Mora-Sero, F. Fabregat-Santiago, B. Denier, J. Bisquert, R. Tena-Zaera, J. Elias and C. Levy-Clement, Appl. Phys. Lett., 2006, 89, 203117–3; (d) S.-J. Lee, S. K. Park, C. R. Park, J. Y. Lee, J. Park and Y. R. Do, J. Phys. Chem. C, 2007, 111, 11793–
This journal is © The Royal Society of Chemistry 2014
Nanoscale
30
31
32 33 34
35 36 37
38
39 40 41
42 43 44 45
11801; (e) M. Liu, C.-Y. Nam, C. T. Black, J. Kamcev and L. Zhang, J. Phys. Chem. C, 2013, 117, 13396–13402; (f) J. K. Cooper, Y. Ling, C. Longo, Y. Li and J. Z. Zhang, J. Phys. Chem. C, 2012, 116, 17360–17368; (g) C. Yao, B. Wei, H. Ma, H. Li, L. Meng, X. Zhang and Q. Gong, J. Power Sources, 2013, 237, 295–299. (a) M. Fekete, R. K. Hocking, S. L. Y. Chang, C. Italiano, A. F. Patti, F. Arena and L. Spiccia, Energy Environ. Sci., 2013, 6, 2222–2232; (b) I. Zaharieva, P. Chernev, M. Risch, K. Klingan, M. Kohlhoff, A. Fischer and H. Dau, Energy Environ. Sci., 2012, 5, 7081–7089; (c) T. Takashima, K. Hashimoto and R. Nakamura, J. Am. Chem. Soc., 2012, 134, 1519–1527. (a) Y. Surendranath, M. Dinca and D. G. Nocera, J. Am. Chem. Soc., 2009, 131, 2615–2620; (b) A. J. Esswein, Y. Surendranath, S. Y. Reece and D. G. Nocera, Energy Environ. Sci., 2011, 4, 499–504; (c) V. B. R. Boppana and F. Jiao, Chem. Commun, 2011, 47, 8973–8975. S. S. Alias, A. B. Ismail and A. A. Mohamad, J. Alloys Compd., 2010, 499, 231–237. H. Gerischer, D. M. Kolb and J. K. Sass, Adv. Phys., 1978, 27, 437–498. A. M. Bond, Broadening electrochemical horizons: principles and illustration of voltammetric and related techniques, Oxford University Press, Oxford, New York, 2002. X. G. Zhang, Corrosion and Electrochemistry of Zinc, Plenum Press, New York, 1996. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon, 1966. S. Ito, P. Chen, P. Comte, M. K. Nazeeruddin, P. Liska, P. Pechy and M. Graetzel, Prog. Photovoltaics, 2007, 15, 603–612. T. Daeneke, A. J. Mozer, Y. Uemura, S. Makuta, M. Fekete, Y. Tachibana, N. Koumura, U. Bach and L. Spiccia, J. Am. Chem. Soc., 2012, 134, 16925–16928. M. Fekete, L. Wiebke, S. Gledhill, J. Chen, A. F. Patti and L. Spiccia, Eur. J. Inorg. Chem., 2014, 4, 750–759. S. A. Awad and K. M. Kamel, J. Electroanal. Chem. Interfacial Electrochem., 1970, 24, 217–225. P. C. P. De, O. A. H. Derosa and M. C. Giordano, J. Electroanal. Chem. Interfacial Electrochem., 1978, 86, 335– 348. P. C. P. De, M. C. Giordano, B. A. Lopez and M. E. Manzur, Corros. Sci., 1988, 28, 769–785. B. Aurian-Blajeni and M. Tomkiewicz, J. Electrochem. Soc., 1985, 132, 1511–1515. P. Scholl, X. Shan, D. Bonham and G. A. Prentice, J. Electrochem. Soc., 1991, 138, 895–899. T. P. Dirkse, J. Appl. Electrochem., 1971, 1, 27–33.
Nanoscale, 2014, 6, 7585–7593 | 7593
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
PAPER
Cite this: Nanoscale, 2014, 6, 7585
View Journal | View Issue
Photoelectrochemical water oxidation by screen printed ZnO nanoparticle films: effect of pH on catalytic activity and stability† Monika Fekete,ab Wiebke Riedel,c Antonio F. Pattia and Leone Spiccia*ab Nanostructured ZnO films are promising photoanode materials in photoelectrochemical water splitting. While such ZnO photoanodes have achieved high activity and good light conversion efficiency in the UV spectral region, their application in water splitting devices has been hampered by the susceptibility of ZnO towards photocorrosion in aqueous electrolytes. We report a systematic investigation aimed at optimising the electrolyte solution to improve the long-term stability of ZnO photoanodes. A stability diagram, based on the band edge positions of ZnO and the pH-dependent photodegradation potentials of ZnO (relative to the decomposition of water), indicates that the optimum pH operating conditions for ZnO photoanodes lie between pH 9–12.5. To verify this prediction experimentally, the activity and longterm stability of uniform screen-printed nano-ZnO films was tested in a wide range of buffered and nonbuffered electrolytes (pH 6–13.5). The ZnO films were more active in buffered, than in non-buffered electrolytes, and the highest activities were observed close to the pKa of the phosphate and borate buffers used. Under zero applied potential, these screen-printed films achieved the highest reported photocurrents to date (0.42 mA cm2 at pH 6 and 0.67 mA cm2 at pH 10.5) for any pristine or modified ZnO-based water oxidation catalyst. The films were subjected to 12 h of controlled potential electrolysis, in selected electrolytes, under AM 1.5G simulated sunlight. The results are in good agreement with calculations based on thermodynamic data for ZnO. Films tested at pH 6 and 7 (representing typically
Received 10th April 2014 Accepted 9th May 2014
used operating conditions) degraded rapidly, whereas they exhibited the highest stability when tested in a pH 10.5 borate buffer. In this case, 75% of the initial photoactivity was preserved after 12 hours,
DOI: 10.1039/c4nr01935k
indicating that the lifetime of the electrode could be increased by over an order of magnitude compared
www.rsc.org/nanoscale
to standard testing conditions.
Introduction Photocatalytic water splitting over semiconductor electrodes has been the subject of intense research since the 1972 discovery of the Honda–Fujishima effect on titanium dioxide surfaces.1 In the past four decades, a wide range of semiconductors have been evaluated for their potential as efficient photocatalysts for hydrogen production by direct solar water splitting vide infra.2–11
a
School of Chemistry, Monash University, Victoria 3800, Australia. E-mail: leone. [email protected]; Fax: +61 3 9905 4597; Tel: +61 3 9905 4526
b
ARC Centre of Excellence for Electromaterials Science (ACES), Monash University, Victoria 3800, Australia
c Institut f¨ ur Heterogene Materialsysteme, Helmholtz-Zentrum f¨ ur Materialien und Energie, Hahn-Meitner-Platz 1, 14109 Berlin, Germany
† Electronic supplementary information (ESI) available: XRD pattern of a fresh ZnO lm (Fig. S1); LSV testing of 1-, 2- and 3-layer screen-printed ZnO lms (Fig. S2); dependence of photocurrent on light intensity (Fig. S3); IPCE of the screen-printed ZnO lms (Fig. S4); SEM image of ZnO lm aer CPE at pH 12 (Fig. S5); and ICP-TOF-MS analysis of the Zn-content of lms (Table S1). See DOI: 10.1039/c4nr01935k
This journal is © The Royal Society of Chemistry 2014
In general, if sufficient photovoltage is generated at the interface of an n-type semiconductor and the aqueous electrolyte, photo-oxidation of water by minority carriers (holes) can occur at the anode. Similarly, photo-reduction of water can be achieved by electrons generated at a p-type semiconductor photocathode.12 For these reactions to proceed, the band alignment of the semiconductor electrodes is crucial. Photooxidation at the anode becomes possible if the upper edge of the valence band of an n-type semiconductor lies at a potential more positive than that required for water oxidation. Analogously, photo-reduction proceeds when the lower edge of the conduction band of a p-type cathode is more negative than the potential required for hydrogen evolution by water reduction.13 Examples of materials studied as potential photocathodes include p-GaP,14 p-InP,15 p-GaInP2,16 p-Si,11a CdTe,10,17 p-WS2 (ref. 18) and Cu2O.19 Semiconductors, such as TiO2,1 WO3,2 GaN,20 SnO2,21 Fe2O3,3 SrTiO3,5 BaTiO3,5 ZnO22 or ZnS23 are promising photoanode candidates because they absorb high energy photons and are, therefore, capable of generating enough photovoltage (a minimum of 1.23 V) to drive unassisted water oxidation or both water oxidation and reduction (e.g.,
Nanoscale, 2014, 6, 7585–7593 | 7585
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
TiO2, TaON).24 The disadvantage of wide-bandgap semiconductors is that their light absorption is mostly limited to the UV region of the solar spectrum.11a Recently, advances have been made towards extending the absorption of these materials into the UV, either by incorporating dopants to narrow the bandgap25 or by the addition of hydrophobic dyes onto the electrode surface.11b Apart from the fundamental limitation of light absorption, photocorrosion of semiconductor electrode materials in aqueous electrolytes is a major issue that has hampered their practical use.24 At the illuminated surface of an n-type semiconductor anode surface, two main competing reactions take place; the decomposition of water by the photogenerated holes and, less desirably, the corrosion of the electrode itself due to destabilization caused by hole trapping on the semiconductor surface.13,26 Based on thermodynamic and kinetic criteria, the possibilities of avoiding photodecomposition of the anode material can be determined. Gerischer and co-workers published a series of studies discussing these criteria, focusing partly on ZnO (amongst other materials, such as TiO2 or CdS).13,26,27 ZnO has attracted much attention in the past decades as a favourable photoanode material towards achieving efficient photoelectrochemical water oxidation, due to its good electron mobility, excellent optical properties, wide availability and relatively low toxicity.22b,28 More recently, nanostructuring of ZnO lms has been shown to improve photocurrent yields due to their large surface area relative to their illuminated geometric area, decreased reectivity and improved charge transfer to the conducting electrode substrate.22a ZnO nanorods, nanosheets, nanowires and various nanoparticles have been synthesised in order to increase the efficiency of water oxidation.22,25,28a,29 Many recently developed ZnO-based photo-electrocatalysts for water oxidation exhibit improved light absorption properties with high photocurrent densities. Relatively little data is available, however, on the photostability and long-term performance of these electrodes. Yang et al. reported that the activity of Ndoped ZnO nanorods, prepared by a hydrothermal method, when tested in a 0.5 M NaClO4 electrolyte (pH 7),22b dropped by 25% within the rst 20 minutes, but then increased up to 90% of the original level aer 90 minutes. The authors suggested that the chemical instability of ZnO could be reduced by covering the ZnO nanowire core with, for example, a thin TiO2 shell. Liu et al.29e tested this approach by coating ZnO nanowire arrays with a thin, passivating TiO2 layer. Testing in highly alkaline electrolytes revealed a 10% drop in photoactivity for the uncoated ZnO NRAs aer 1 hour, whereas the activity of the passivated ZnO NRAs remained unchanged within the same timeframe. According to the authors, this was the rst report to use ZnO-based water oxidation catalysts in highly alkaline electrolytes. Indeed, most studies on ZnO photoelectrocatalysts focus on their operation in close to neutral media. Qiu et al. tested undoped and N-doped ZnO nano-tetrapods in a pH 7.0 Na2SO4 electrolyte solution,29a and found that the initial activity of the N-doped lms was higher compared to the undoped lms. When operated continuously over one hour, however, the activity of the undoped nanorods remained close to 100%, while the N-doped nanorods retained 90% of their
7586 | Nanoscale, 2014, 6, 7585–7593
Paper
activity within the same timeframe. Hassan et al. claimed ‘exceptional stability’ for (undoped) ZnO nano-tetrapods prepared by a thermal evaporation method when tested in a 0.5 M Na2SO4 solution.29b Inspection of their controlled potential electrolysis (CPE) data, collected over a 20 minute period, conrms that the less active catalyst lm indeed retains close to 100% activity, while the performance of the more active sample drops by approximately 20% over the same testing period. This result, in fact, seems to correspond well with the data reported by Yang et al.22b for studies at pH 7. While the optimum pH and electrolyte composition for maximum oxygen evolution has been extensively studied for “dark” electrocatalysts,30 these conditions have not been well established for semiconductor photocatalysts. Phosphate and borate buffers have been shown to enhance the rate of oxygen evolution when used with various water oxidation catalysts, for example, manganese, cobalt or nickel oxides, mainly by improving proton conduction between the anode and cathode.31 By preventing the accumulation of protons near the anode surface (which could destabilise the electrode material), these buffers signicantly increase the lifetime of the catalysts. Mohamad and co-workers studied the pH conditions for the sol–gel synthesis of ZnO nanoparticles, concluding those synthesized in the pH 8 to 11 region possessed optimal optical properties.32 In this study, thermodynamic considerations are used in an attempt to optimise the operating conditions for water splitting cells employing ZnO-based photoanodes. The pH range where the ZnO anode is theoretically stable against photodecomposition was rst predicted with the help of an E-pH stability diagram comparing decomposition potentials for the aqueous electrolyte and the ZnO electrodes. To test the prediction experimentally, reproducible ZnO nanoparticle lms were screen-printed and their photocatalytic performance was tested in a range of electrolyte solutions. The effects of the phosphate and borate buffers on photocurrent generation and on the longterm stability of ZnO under operational conditions are discussed. A particularly promising result was the observation of photocurrent even in the absence of an applied bias in an optimised, buffered electrolyte.
Background Stability of ZnO photo-electrocatalysts: thermodynamic considerations. The principles of the photodegradation of semiconductor electrodes in aqueous media had been established as early as the 1970's. For example, a series of studies by Gerischer and co-workers describe the thermodynamic criteria that need to be met to protect electrodes from photocorrosion during operation in water splitting cells.13,26,33 Briey, a semiconductor is protected against photodecomposition by electrons if its cathodic photodecomposition potential (EDCe) lies negative to the conduction band and against photodecomposition by holes if its anodic photodecomposition potential (EDCh+) lies positive to the valence band edge. This “ideal” situation is shown in Fig. 1A. Zinc oxide is stable against photodecomposition by electrons but is
This journal is © The Royal Society of Chemistry 2014
View Article Online
Paper
Nanoscale
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
number of electrons involved in eqn (1)–(3) and F is the Faraday constant. According to Zhang, the decomposition potentials under acidic (1) and highly alkaline (3) conditions are at 0.89 V and 0.43 V vs. SHE, respectively.35 The EDCh+ corresponding to (2) was calculated in this study as 1.08 V vs. SHE. Based on these calculations and experimental EIS data, a proposed stability diagram across the pH spectrum was constructed.
Experimental Screen printing of the ZnO lms (A) Band diagram for an ideally stable n-type semiconductor at pH ¼ 0. (B) Band diagram of ZnO at pH ¼ 0. EFB, marked with *, was calculated based on measurements presented in this study. Fig. 1
intrinsically unstable to photodecomposition by holes, as shown in Fig. 1B. Decomposition can be avoided, however, when water is oxidised at a potential negative to EDCh+. Towards this end, the relative positions of the potentials corresponding to the competing reactions need to be determined over the accessible pH range. The Nernstian shi (59 mV/pH) in the redox potential of the 1/2 O2/H2O couple with increasing solution pH is well known.34 The anodic photodecomposition potential of semiconducting oxides is generally independent of pH, but this does not hold true for zinc oxide.24 The decomposition potential is determined by the dissolved products formed when the electrode photocorrodes.35 A constant decomposition potential over the entire pH range would only be possible if the same product formed at any given solution pH. The Pourbaix-diagram of zinc oxide, however, indicates that at acidic to neutral pH, the predominant dissolved zinc species is Zn2+, at highly alkaline pH ZnO22, and HZnO2 between pH 9.2–13.2.36 Accordingly, the equations describing the electrode decomposition are: ZnO + 2p+ ¼ Zn2+ + 1/2O2
(1)
ZnO + 3OH + 2p+ ¼ HZnO2 + 1/2O2 + H2O
(2)
ZnO + 4OH + 2p+ ¼ ZnO22 + 1/2O2 + 2H2O
(3)
The solubility of ZnO is the lowest at pH 9.3 (107–108 M), and increases exponentially both towards more acidic and alkaline pH values.35 (Naturally, soluble species can co-exist in same solution, their relative concentrations determined by equilibria, as indicated in the Pourbaix diagram.) The decomposition potential EDCh+ for each reaction can be calculated from the change in the Gibbs free energy upon the breakdown of the electrode material: DG ¼ RT ln K
(4)
DG ¼ nFEDCh+
(5)
EDCh þ ¼
RT ln K nF
(6)
where K is the equilibrium constant for reactions (1)–(3), R is the ideal gas constant, T is the absolute temperature, n is the
This journal is © The Royal Society of Chemistry 2014
Screen printing, a well-established method for the preparation of highly uniform TiO2 nanoparticle lms in DSC fabrication,37 was adapted to deposit commercial zinc oxide nano-powders. The ZnO nanoparticles (Sigma Aldrich, d < 200 nm) were printed onto uorine-doped tin oxide (FTO) coated glass slides (10 U per square, DyeSol) using a modied screen-printing method adapted from a published procedure.30a In a typical preparation, ZnO nanoparticle powder (1.0 g) was mixed with 5.0 ml ethylcellulose binder solution (10 wt% in ethyl alcohol, Sigma Aldrich 433837) and 4.0 ml terpineol. The paste was screenprinted onto FTO glass substrates in 1, 2 and 3 layers (resulting in a thickness of 0.6 mm, 1.7 mm and 2.4 mm, respectively). The lms were sintered in air on a hotplate according to a heating program developed for TiO2 and indium-doped tin oxide (ITO) lms in dye sensitized solar cell (DSC) fabrication.38 The photocatalytic activity of the ZnO lms was tested by linear sweep voltammetry (LSV) in a 0.6 M Na2SO4 solution (pH 6). Based on this initial study, the 2-layer lms were chosen for subsequent characterization and photocatalytic testing. Phosphate (0.6 M, pH 6.5, 7, 7.5, 8.5 and 12) and borate (0.6 M, pH 8.5, 9, 9.5, 10, 10.5 and 11) pH buffer solutions as well as Na2SO4 and NaClO4 solutions (0.6 M) were made freshly with Millipore quality water just prior to electrochemical testing. Film characterisation and testing of photocatalytic activity Scanning Electron Microscopy (SEM) was performed with a JEOL JSM-7001F FEGSEM instrument tted with a Bruker 10 mm2 Si dri detector analysis system, operated at 30 kV. The lms on the FTO glass substrate were previously sputter-coated with platinum (1 nm). SEM characterisation was carried out the Monash Centre for Electron Microscopy (MCEM). X-ray Diffraction (XRD) patterns were obtained using a Philips instrument with powder XRD capability. The printed lms were gently scraped off the supporting glass slides and sprinkled onto a quartz plate thinly coated with Vaseline. XRD scans were taken using Cu Ka radiation, with a 1 divergence slit, 0.2 receiving slit and carbon monochromators at a 0.5 min1 scan rate (2F range 25 –80 ). Testing of catalytic activity of the electrodeposited ZnO lms was performed using a Bio-Logic VSP Modular 5-channel potentiostat, at room temperature (20 C), in a Zahner-Elektrik PECC-1 electrochemical cell. The at quartz window of the electrochemical cell helped to ensure well reproducible conditions during photoelectrochemical performance testing.
Nanoscale, 2014, 6, 7585–7593 | 7587
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
Photocurrent generation was systematically tested in electrolytes with pH ranging from 6 to 13.5, with particular focus on the pH 7.0–10.5 region. Illumination was provided by an Oriel 150 W solar simulator equipped with an ozone-free Xe lamp and an AM1.5G lter to match the solar spectrum. A fresh lm was used for each measurement. The electrolytes used were: 0.6 M Na2SO4, 0.6 M phosphate buffers (at pH 6.5, 7.0, 7.5, 8.0, 8.5, 12), 0.6 M borate buffers (at pH 8.5, 9.0, 9.5, 10, 10.5 and 11), a 0.6 M NaClO4 solution adjusted to pH 12 with NaOH and a 0.6 M NaOH solution (pH 13.5). Linear sweep voltammetry (LSV) testing was carried out at a 5 mV s1 scan rate, 0.2 to 1.5 V vs. Ag/AgCl, in the positive or oxidative direction, under chopped illumination (5 s on, 5 s off) at 1 Sun light intensity (100 mW cm2). The overpotential for O2 evolution was calculated as the difference between the applied potential (Eapp) vs. Ag/AgCl and the E0 of oxygen evolution at the appropriate pH (vs. Ag/AgCl), based on the Nernst equation, referenced against Ag/AgCl: h(V) ¼ Eapp(V) E0(V) ¼ Eapp(V) (1.23 V 0.2V 0.059 pH). Incident Photon to Charge Carrier Conversion Efficiency (IPCE) was measured at pH 10.5 (in a borate buffer) following a previously described procedure,39 using a 150 W Oriel solar simulator (with a horizontal lightbeam) equipped with an AM1.5G lter and bandpass lters. The dependence of photocurrent on light intensity was measured by recording linear sweep voltammograms of each lm between 0.5–1.5 V vs. Ag/AgCl, at a 5 mV s1 scan rate. The light intensity was adjusted to 0, 0.5, 1, 2 and 3 Suns using neutral density lters. Long-term performance was tested by Controlled Potential Electrolysis (CPE), at 120 mV overpotential corresponding to 670 mV, 620 mV, 490 mV, 410 mV, 320 mV and 230 mV vs. Ag/ AgCl, at pH 6.0 (0.6 M Na2SO4), 7.0 (phosphate buffer), 9.0 (borate buffer), 10.5 (borate buffer), 12.0 (phosphate buffer and 0.6 M NaClO4/NaOH) and 13.5 (0.6 M NaOH), respectively; under 1 Sun illumination over 12 hours. Additionally, CPE was carried out at 0.2 V, 0 V and 0.66 V (h ¼ 250 mV) vs. Ag/AgCl under 1 Sun illumination in a pH 10.5 borate buffer for 12 hours. Electrochemical Impedance Spectroscopy measurements (Mott–Schottky plots) were recorded using a Bio-Logic VSP Modular 5-channel potentiostat operated in Staircase PotentioElectrochemical Impedance Spectroscopy (SPEIS) mode. Mott– Schottky plots were recorded 1 V to +1 V vs. Ag/AgCl, using a RC equivalent circuit tting model. Inductively Coupled Plasma-Time of Flight-Mass Spectrometry (ICP-TOF-MS) analysis was carried out using a GBC Optimass 9500 instrument. Raw count rates from the analyses were externally standardised by means of a calibration curve constructed using Zn2+ stock solutions (containing Zn2+ ions in 0, 10, 25, 50 and 100 ppb concentrations). Indium was used as an internal standard. The pristine and post-12 hour operation ZnO lms (of 0.49 cm2 surface area each) were dissolved in 10 ml 70% nitric acid and diluted to 50 ml stock solutions prior to the ICP-MS analysis. Electrolyte solutions aer 12 hours of lm operation were also analysed.
7588 | Nanoscale, 2014, 6, 7585–7593
Paper
Results and discussion Flatband potential and stability diagram In aqueous electrolytes, the atband potential, which is dened by the relative positions of the band edges, is strongly inuenced by the solution pH. At each pH, the atband potential can be experimentally determined using the Mott–Schottky relationship: 2 kT CSC 2 ¼ E EFB (10) e330 ND e where e is the electron charge, 3 is the dielectric constant of ZnO, 30 is the permittivity of vacuum, ND is the donor density, E is the applied potential at the electrode, EFB is the atband potential of the ZnO electrode and k is the Boltzmann constant. Gerischer et al. found that, due to the formation of an additional electric double layer by the interaction of the ZnO surface with H+ and OH ions present in the solution, the pH dependence of the atband potential over a large pH range follows a Nerstian behaviour, i.e., a shi of 59 mV/pH.33 Therefore, the relative positions of the band edges and the water oxidation and reduction potentials remain unchanged over a large part of the pH scale. The 59 mV/pH shi was conrmed experimentally within the examined pH 6–10.5 range (marked by crosses in Fig. 2). Anodic and cathodic decomposition potentials were taken from Zhang.35 Where data was not available, the potentials were calculated based on the change in Gibbs free energy upon formation of the photodecomposition products and equilibria between the dissolved species (dened by the Pourbaix diagram.) ZnOH+, which forms in a small amount around pH 9, was not included in this calculation. The proposed ZnO stability diagram is shown in Fig. 2. Fig. 2 suggests that long-term operation of a ZnO-based water oxidation catalyst without extensive photodegradation should be possible in the pH 9–12.5 range. At pH 6 and 7, dissolution of the lm is expected even if water oxidation reaction occurs without applied overpotential (i.e., at the thermodynamic limit). If an overpotential is required, the blue line indicating oxygen evolution from water shis to more positive potentials, which narrows the window for optimum operation even further. These considerations are valid when no other species are present in the electrolyte that could either form a precipitate with ZnO or complex with the dissolved zinc species. When operating in buffered electrolytes, such possibilities need to be considered. Based on these theoretical considerations a working hypothesis was developed and tested in practice by examining the pH-dependent activity and long-term stability of ZnO lms during photo-electrochemical water oxidation.
Film deposition and characterization The crystal phase of the ZnO lms was conrmed to be the commonly occurring wurtzite phase by powder-XRD diffraction
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Paper
Nanoscale
Fig. 2 Stability diagram of ZnO. Black crosses mark the EFB values calculated based on Mott–Schottky measurements. Vertical dashed lines represent the equilibrium pH between the three main dissolved Zn-species. The light blue line shows the thermodynamic O2 evolution potential, while the light red line represents the anodic photodecomposition potential of ZnO by photogenerated holes (EDCH+). The photoanode is stable in the pH region where EDCH+ is more positive than the O2 evolution potential, approximately between pH 9–12.5. The dark blue line shows the thermodynamic hydrogen evolution potential, which lies positive to the cathodic decomposition potential (EDCe, marked with the dark red line) of ZnO across the entire pH range.
(see Fig. S1†). SEM studies showed that the porous catalyst lms are composed of small spherical (40–50 nm) and larger, “bricklike” particles, approximately 50 100–200 nm in size (see below). The uniformity of the two-layered lms was tested by ICP-MS analysis (as described in the Experimental section). The lms were found to contain the same amount of Zn (31 mg) within a 2% error (based on analysis of three fresh lms). Testing of the activity of the ZnO lms by LSV in a 0.6 M Na2SO4 solution revealed that the photocurrent was very similar for the 1- and 2-layered lms, (see Fig. S2†) which decreased slightly on addition of a third layer. Based on these results, the more robust two-layer lms were used in further studies. A dark current above 1.2 V applied bias (vs. Ag/AgCl) was observed in each case, which could be due to the degradation of ZnO, as predicted by the stability diagram, as well as water oxidation.
Testing of photocatalytic activity as a function of pH LSV scans recorded at pH 7.0 (phosphate buffer), pH 9.0 (borate buffer) and pH 13.5 (0.6 M NaOH) are shown in Fig. 3. The onset of dark current was found at 1.0 V, 1.1 V and 0.7 V, (vs. Ag/AgCl), at pH 7.0, 10.5, and 13.5, respectively. These values correspond to the upper stability limit of ZnO, and could be arising in part from lm degradation. Under illumination, signicant photocurrents were observed even at 0.2 V vs. Ag/AgCl, which equals 0 V applied potential vs. SHE (zero bias condition). At pH 7, the photocurrent increased with the applied bias up to 0.2 V vs. Ag/AgCl, where it reached a plateau of 0.75 mA cm2. At pH 10.5 and 13.5, the photocurrent showed little change throughout the measured voltage range and maximised at 0.70 and 0.55 mA cm2, respectively. These
This journal is © The Royal Society of Chemistry 2014
relatively high photocurrents at close to 0 V applied potential set the screen-printed ZnO apart from other recently reported highperforming nano-ZnO -based water oxidation catalysts, where the photocurrent was typically found to be highly dependent on the applied bias.22,29f,g The photocurrent density for the screen-printed ZnO photoanodes scales almost linearly with illumination power (up to 2 Suns, with near-saturation at 3 Suns) at low applied potentials. At the higher applied potentials (>1.2 V), the dark (degradation) current contributes to the measured current, and the difference between the currents recorded with and without illumination diminishes (Fig. S3†). The incident photon-to electron conversion (IPCE, shown in Fig. S4†) was found to be relatively high at 370 nm (41%) and 390 nm (10%), but effectively zero beyond 410 nm, as typically found for ZnO photoanodes.22b,29g The dependence of photocurrent generation on pH is shown in the inset of Fig. 2. For the non-buffered Na2SO4 electrolyte at pH 6.0, the photocurrent (0.28 mA cm2) was much lower than that measured for the buffered electrolytes (0.70 mA cm2 was measured at pH 6.5 in a phosphate buffer). In the phosphate and borate buffers, the photocurrents uctuate with local maxima close to their pKa. The highest photocurrent (0.82 mA cm2) was measured at pH 7.5 in a phosphate buffer (pKa2 ¼ 7.2) and at pH 9.5 in a borate buffer (pKa2 ¼ 9.2), where buffering was the most effective. The variation in photocurrent within the buffering range was relatively small, although the photocurrent dropped quite signicantly at pH 11, where the boric acid/sodium borate buffer is less effective. Higher current was measured again at pH 12 in a phosphate buffer, within the Na2HPO4/Na3PO4 buffering range (pKa3 ¼ 12.7). When an unbuffered NaClO4/NaOH electrolyte was used at the same pH,
Nanoscale, 2014, 6, 7585–7593 | 7589
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
Paper
LSV testing of the screen-printed ZnO photoanodes under 1 Sun chopped illumination, at pH 7 (blue line), pH 10.5 (red line) and 13.5 (black line). Dashed vertical lines represent the thermodynamic O2 evolution potential at the corresponding pH. The inset shows photocurrents at 0 mV overpotential at each pH, measured in various buffered and non-buffered electrolytes. Fig. 3
the current dropped slightly from 0.69 to 0.56 mA cm2. A similar photocurrent was measured at pH 13.5, in a 0.6 M NaOH solution. Thus, little variation in photocurrent with pH was observed when the electrolyte provided sufficient buffering capacity. More pronounced differences in the catalytic performance of the ZnO lms with pH were observed when the cells were subjected to long-term constant potential electrolysis (CPE).
Long-term stability Fig. 4 shows that in the pH 6.0 electrolyte, the activity, initially quite high (0.78 mA cm2) decreased to 0 mA cm2 within just 90 minutes. Even in the buffered pH 7.0 electrolyte, complete degradation was observed aer 2 hours. At pH 9, stability was
Fig. 4 Long-term stability over 12 h, under 1 Sun light intensity, (h ¼ 120 mV) of the ZnO screen-printed films at various pH. (pH 6 : 0.6 M Na2SO4; pH 7 and 12 : 0.6 M phosphate buffers; pH 9 and 10.5 : 0.6 M borate buffers; pH 13.5 : 0.6 M NaOH).
7590 | Nanoscale, 2014, 6, 7585–7593
much improved, 58% of the initial photocurrent was still preserved aer 12 h. An even better stability was observed at pH 10.5, where approximately 75% of the photocurrent was retained aer 12 h. At pH 12, stability was decreased; the current dropped to 50% aer 4 hours and was almost completely deactivated by the end of the 12 hour electrolysis. At 13.5, the photocurrent dropped by 50% in less than 90 minutes and then stabilised at close to 25% of the initial photocurrent. SEM images taken of the catalyst lms before and aer CPE testing, are shown in Fig. 5. The lm tested at pH 6.0 was completely dissolved off the FTO glass substrate. At pH 7.0, the lm had also almost completely dissolved, and the FTO crystals
Fig. 5 SEM images of (A) a fresh screen-printed ZnO film, (B–F) after 1 h CPE testing under 1 Sun illumination. (B) pH 7 (phosphate buffer); (C) pH 9 (borate buffer); (D) pH 10.5 (borate buffer); (E) pH 12 (phosphate buffer); (F) pH 13.5 (0.5 M NaOH).
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Paper
Nanoscale
Fig. 6 Photocurrent degradation at various applied potentials in a pH 10.5 borate buffer (under 1 Sun illumination).
can be seen underneath the remains of the lm in Fig. 5B. According to Awad and Kamel,40 phosphate ions (especially when present as a mixture of Na2HPO4 and NaH2PO4), can result in the formation of complex phosphates on the surface. A highly polymerised, semi-glassy interface has been proposed to be formed via a dissolution–precipitation process:41 Zn2+ + 2H2PO4 4 Zn(H2PO4)2
(7)
3Zn(H2PO4)2 4 Zn3(PO4)2$4H2O + 4H3PO4
(8)
The resulting interface is likely to adversely affect the catalytic activity of the ZnO lm and possibly accelerate the breakdown of the lm. The breakdown of the passive lm was postulated to involve diffusion of Zn2+ ions through the passive layer. The formation of H2PO4 near the lm surface could also promote the dissolution process.
At pH 9.0, the nanoparticles are preserved, but are covered with very small particles, possibly indicating deposition of a borate species on the catalyst surface (Fig. 5C). The solubility zinc species is the lowest at pH 9.3 (between 107 and 108 M), and, incidentally, the solubility of borates (mostly in the form of H4BO7 and B4O72) is also the lowest between pH 7–9.5 (ca. 102 to 103 M).36 Insoluble zinc borate species could passivate the ZnO surface, signicantly decreasing the surface area available for contact between the catalyst and the electrolyte, leading to loss of activity. At pH 10.5, there is less degradation and no borate deposition was observed, correlating well with the good stability of photocurrent over 12 h (Fig. 5D). At pH 12, the loss in activity was accompanied by extensive degradation of the ZnO lm, accelerated by the presence of PO43 ions (see Fig. 5E). The breakdown and coverage of the catalysts with a zinc-phosphate passive layer is evidenced by the formation of nodules with high phosphorous content on the surface.42 The long-term behaviour of the ZnO catalyst lm was also tested in a 0.6 M NaClO4/NaOH solution at pH 12 to examine lm degradation in the absence of phosphate ions. In this case, the photocurrent was somewhat more stable (20% retained aer 12 h), and the particles showed less degradation than in the phosphate buffer (shown in Fig. S5†). At pH 13.5, the formation of larger, spherical grains was observed (Fig. 5F). Aurian-Blajeni and Tomkiewicz proposed that these passive lms are composite layers of the electrolyte and the oxide lm,43 oen found to retain properties typical of an amorphous semiconductor lm.44 The passivation processes at highly alkaline pH were summarised by Dirkse.45 The Zn(OH)2 lm on the surface of ZnO lm dissolves as Zn(OH)3 or Zn(OH)42 and, once the electrolyte layer covering the
Comparison of operating conditions, initial activity and long-term stability of ZnO-based photoanodes with those reported in recent literature
Table 1
Photoanode
Electrolyte
Screen-printed ZnO NPs Screen-printed ZnO NPs Screen-printed ZnO NPs
0.6 M Na2SO4 (pH 6) 0.6 M Na2SO4 (pH 6) 0.6 M Borate buffer (pH 10.5) 0.6 M Borate buffer (pH 10.5) 0.5 M NaClO4 (pH 7.4) with PB 0.5 M NaClO4 (pH 7) with PB 0.1 M KOH (pH 13) 0.5 M Na2SO4 (pH 7) 0.5 M Na2SO4
Screen-printed ZnO NPs Nano-ZnO N-doped ZnO NRA TiO2-coated ZnO nanowires Long, H-annealed ZnO nanowires Hydrogenated, hierarchical ZnO NRAs N-doped ZnO nano-tetrapods ZnO nano-tetrapods N-and Al-co-doped ZnO thin lms a
0.5M Na2SO4 (pH 7) with PB 0.5 M Na2SO4 0.5 M Na2SO4
J initial at h ¼ 120 mV (mA cm2), by LSV
PCa retained aer 1 h CPE (%)
Max. CPE duration shown (h)
0.42 0.42 0.67
26b 26b 100b
12 At 3 h 12
0 0 75
Present study Present study Present study
0.67
100b
At 3 h
96
Present study
0
—
—
PC retained aer max. CPE testing (%)
—
0.01
85c
1.4
90
0.2 0.1 0.2
100d — —
1.0 — —
100 — —
0.05
90c
1.0
90
0.30 —
80e —
0.02 0
— —
Ref.
Wolcott et al.22a Yang et al.22b Liu et al.29e Cooper et al.29f Yao et al.29g Qiu et al.29a Hassan et al.29b Shet et al.25
PC ¼ photocurrent. b h ¼ 120 mV. c Eapp ¼ 0.75 V vs. SHE. d Based on LSV aer 1 h. e Eapp ¼ 1.0 V vs. SHE.
This journal is © The Royal Society of Chemistry 2014
Nanoscale, 2014, 6, 7585–7593 | 7591
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Nanoscale
electrode becomes supersaturated, re-deposits again as amorphous Zn(OH)2. XRD analysis aer the 12 hour CPE at pH 13.5 conrmed that the crystalline wurtzite phase had become more disordered and much less crystalline (see Fig. S1†). The less crystalline material was found to generate a stable photocurrent over the next 10 hours of CPE, but it was less active than the crystalline phase. The formation of larger particles also reduces the surface area and decreases the number of active surface sites, and most likely contributes to the lower activity observed aer 2 hours. Long-term stability was also investigated as a function of applied potential at pH 10.5. As shown in Fig. 6, the photocurrent versus time prole followed the same pattern over 12 hours of testing at 0.2, 0 and 0.66 V vs. Ag/AgCl. The current density increased when a small bias was applied (0 V vs. Ag/AgCl ¼ 0.2 V vs. SHE), but did not increase further at higher potentials. The loss of activity was the same (%) in all three experiments. The good stability of the ZnO lms at pH 10.5 was also conrmed by ICP-MS analysis of the electrolyte and the used lms. The amount of zinc that leached into solution was 75% of the initial photoactivity was retained aer 12 hours of catalyst operation. This result was corroborated by ICP-MS analysis of the Zn-content of the electrolyte solutions aer 12 hours, which showed a minimum zinc “leakage” from the lm. These ndings could be used to improve the lifetime of other high-performing ZnO-based catalysts. The screen-printed ZnO anode materials are promising candidates for integration with dark co-catalysts, such as electrodeposited or screen-printed water oxidising metal oxides. Sensitization of the ZnO lms with light-harvesting dyes could also be a promising approach for future applications, and with their activity further enhanced, ZnO lms could represent an attractive option for commercial water splitting.
Acknowledgements The authors would like to thank the Monash Institute of Graduate Research and the Monash Faculty of Science for scholarship support and a Postgraduate Publication Award (MF), the Australian Research Council (through the ARC Centre of Excellence for Electromaterials Science to LS) for nancial support and the Monash Centre for Electron Microscopy (MCEM) for instrument access and expertise. Mr R. Mackie is acknowledged for his help with XRD data acquisition and Dr M Raveggi for his help with ICP-MS measurements.
References 1 A. Fujishima and K. Honda, Nature, 1972, 238, 37–38. 2 G. Hodes, D. Cahen and J. Manassen, Nature, 1976, 260, 312– 313. 3 A. Kay, I. Cesar and M. Graetzel, J. Am. Chem. Soc., 2006, 128, 15714–15721. 4 K. Ohashi, J. McCann and J. O. Bockris, Nature, 1977, 266, 610–611. 5 I. E. Paulauskas, J. E. Katz, G. E. Jellison, N. S. Lewis and L. A. Boatner, Thin Solid Films, 2008, 516, 8175–8178. 6 H. Kato, K. Asakura and A. Kudo, J. Am. Chem. Soc., 2003, 125, 3082–3089. 7 J. Sun, C. Liu and P. Yang, J. Am. Chem. Soc., 2011, 133, 19306–19309. 8 A. Heller, Science, 1984, 223, 1141–1148. 9 S. W. Boettcher, E. L. Warren, M. C. Putnam, E. A. Santori, D. Turner-Evans, M. D. Kelzenberg, M. G. Walter, J. R. McKone, B. S. Brunschwig, H. A. Atwater and N. S. Lewis, J. Am. Chem. Soc., 2011, 133, 1216–1219. 10 X. Mathew, A. Bansal, J. A. Turner, R. Dhere, N. R. Mathews and P. J. Sebastian, J. New Mater. Electrochem. Syst., 2002, 5, 149–154. 11 (a) M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473; (b) W. J. Youngblood, S.-H. A. Lee, K. Maeda and T. E. Mallouk, Acc. Chem. Res., 2009, 42, 1966–1973; (c) C.-H. Liao, C.-W. Huang and J. C. S. Wu, Catalysts, 2012, 2, 490–516; (d) X. Chen, S. Shen, L. Guo and S. S. Mao, Chem. Rev., 2010, 110, 6503–6570; (e) A. J. Bard and M. A. Fox, Acc. Chem. Res., 1995, 28, 141–145.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 12 May 2014. Downloaded by University of Central Florida on 28/10/2014 15:12:05.
Paper
12 R. Memming, Semiconductor Electrochemistry, Wiley-VCH, Rellingen, Germany, 2001. 13 H. Gerischer, J. Electroanal. Chem. Interfacial Electrochem., 1977, 82, 133–143. 14 B. Kaiser, D. Fertig, J. Ziegler, J. Klett, S. Hoch and W. Jaegermann, ChemPhysChem, 2012, 13, 3053–3060. 15 A. Heller, E. Aharon-Shalom, W. A. Bonner and B. Miller, J. Am. Chem. Soc., 1982, 104, 6942–6948. 16 H. Wang, T. Deutsch and J. A. Turner, J. Electrochem. Soc., 2008, 155, F91–F96. 17 K. Ohashi, K. Uosaki and J. O. M. Bockris, Int. J. Energy Res., 1977, 1, 25–30. 18 J. A. Baglio, G. S. Calabrese, D. J. Harrison, E. Kamieniecki, A. J. Ricco, M. S. Wrighton and G. D. Zoski, J. Am. Chem. Soc., 1983, 105, 2246–2256. 19 A. Paracchino, N. Mathews, T. Hisatomi, M. Stek, S. D. Tilley and M. Graetzel, Energy Environ. Sci., 2012, 5, 8673–8681. 20 (a) S.-W. Ryu, Y. Zhang, B. Leung, C. Yerino and J. Han, Semicond. Sci. Technol., 2012, 27, 015014; (b) I. Waki, D. Cohen, R. Lal, U. Mishra, S. P. DenBaars and S. Nakamura, Appl. Phys. Lett., 2007, 91, 093519. 21 M. Tian, W. Peng, B. Shang, W. Tao and Y. Xiao, Int. J. Hydrogen Energy, 2012, 37, 12827–12832. 22 (a) A. Wolcott, W. A. Smith, T. R. Kuykendall, Y. Zhao and J. Z. Zhang, Adv. Funct. Mater., 2009, 19, 1849–1856; (b) X. Yang, A. Wolcott, G. Wang, A. Sobo, R. C. Fitzmorris, F. Qian, J. Z. Zhang and Y. Li, Nano Lett., 2009, 9, 2331–2336. 23 (a) H. M. Chen, C. K. Chen, R.-S. Liu, C.-C. Wu, W.-S. Chang, K.-H. Chen, T.-S. Chan, J.-F. Lee and D. P. Tsai, Adv. Energy Mater., 2011, 1, 742–747; (b) P. P. Hankare, P. A. Chate and D. J. Sathe, J. Alloys Compd., 2009, 487, 367–369. 24 S. Chen and L.-W. Wang, Chem. Mater., 2012, 24, 3659–3666. 25 S. Shet, K.-S. Ahn, T. Deutsch, H. Wang, N. Ravindra, Y. Yan, J. Turner and M. Al-Jassim, J. Mater. Res., 2010, 25, 69–75. 26 H. Gerischer, J. Electrochem. Soc., 1966, 113, 1174–1181, discussion 1181–1172. 27 H. Gerischer, Electrochim. Acta, 1989, 34, 1005–1009. 28 (a) L.-W. Ji, S.-M. Peng, J.-S. Wu, W.-S. Shih, C.-Z. Wu and I. T. Tang, J. Phys. Chem. Solids, 2009, 70, 1359–1362; (b) Y. Zhang, M. K. Ram, E. K. Stefanakos and D. Y. Goswami, J. Nanomater., 2012, 624520, 624522. 29 (a) Y. Qiu, K. Yan, H. Deng and S. Yang, Nano Lett., 2012, 12, 407–413; (b) N. K. Hassan, M. R. Hashim and N. K. Allam, Chem. Phys. Lett., 2012, 549, 62–66; (c) I. Mora-Sero, F. Fabregat-Santiago, B. Denier, J. Bisquert, R. Tena-Zaera, J. Elias and C. Levy-Clement, Appl. Phys. Lett., 2006, 89, 203117–3; (d) S.-J. Lee, S. K. Park, C. R. Park, J. Y. Lee, J. Park and Y. R. Do, J. Phys. Chem. C, 2007, 111, 11793–
This journal is © The Royal Society of Chemistry 2014
Nanoscale
30
31
32 33 34
35 36 37
38
39 40 41
42 43 44 45
11801; (e) M. Liu, C.-Y. Nam, C. T. Black, J. Kamcev and L. Zhang, J. Phys. Chem. C, 2013, 117, 13396–13402; (f) J. K. Cooper, Y. Ling, C. Longo, Y. Li and J. Z. Zhang, J. Phys. Chem. C, 2012, 116, 17360–17368; (g) C. Yao, B. Wei, H. Ma, H. Li, L. Meng, X. Zhang and Q. Gong, J. Power Sources, 2013, 237, 295–299. (a) M. Fekete, R. K. Hocking, S. L. Y. Chang, C. Italiano, A. F. Patti, F. Arena and L. Spiccia, Energy Environ. Sci., 2013, 6, 2222–2232; (b) I. Zaharieva, P. Chernev, M. Risch, K. Klingan, M. Kohlhoff, A. Fischer and H. Dau, Energy Environ. Sci., 2012, 5, 7081–7089; (c) T. Takashima, K. Hashimoto and R. Nakamura, J. Am. Chem. Soc., 2012, 134, 1519–1527. (a) Y. Surendranath, M. Dinca and D. G. Nocera, J. Am. Chem. Soc., 2009, 131, 2615–2620; (b) A. J. Esswein, Y. Surendranath, S. Y. Reece and D. G. Nocera, Energy Environ. Sci., 2011, 4, 499–504; (c) V. B. R. Boppana and F. Jiao, Chem. Commun, 2011, 47, 8973–8975. S. S. Alias, A. B. Ismail and A. A. Mohamad, J. Alloys Compd., 2010, 499, 231–237. H. Gerischer, D. M. Kolb and J. K. Sass, Adv. Phys., 1978, 27, 437–498. A. M. Bond, Broadening electrochemical horizons: principles and illustration of voltammetric and related techniques, Oxford University Press, Oxford, New York, 2002. X. G. Zhang, Corrosion and Electrochemistry of Zinc, Plenum Press, New York, 1996. M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous Solutions, Pergamon, 1966. S. Ito, P. Chen, P. Comte, M. K. Nazeeruddin, P. Liska, P. Pechy and M. Graetzel, Prog. Photovoltaics, 2007, 15, 603–612. T. Daeneke, A. J. Mozer, Y. Uemura, S. Makuta, M. Fekete, Y. Tachibana, N. Koumura, U. Bach and L. Spiccia, J. Am. Chem. Soc., 2012, 134, 16925–16928. M. Fekete, L. Wiebke, S. Gledhill, J. Chen, A. F. Patti and L. Spiccia, Eur. J. Inorg. Chem., 2014, 4, 750–759. S. A. Awad and K. M. Kamel, J. Electroanal. Chem. Interfacial Electrochem., 1970, 24, 217–225. P. C. P. De, O. A. H. Derosa and M. C. Giordano, J. Electroanal. Chem. Interfacial Electrochem., 1978, 86, 335– 348. P. C. P. De, M. C. Giordano, B. A. Lopez and M. E. Manzur, Corros. Sci., 1988, 28, 769–785. B. Aurian-Blajeni and M. Tomkiewicz, J. Electrochem. Soc., 1985, 132, 1511–1515. P. Scholl, X. Shan, D. Bonham and G. A. Prentice, J. Electrochem. Soc., 1991, 138, 895–899. T. P. Dirkse, J. Appl. Electrochem., 1971, 1, 27–33.
Nanoscale, 2014, 6, 7585–7593 | 7593
