Konrad
Szaciłowski
*,
Wojciech
Macyk
and
Grażyna
Stochel
Uniwersytet Jagielloński, Wydział Chemii, ul. Romana Ingardena 3, 30-060 Kraków, Poland. E-mail: szacilow@chemia.uj.edu.pl; Fax: +4812 634 0515
First published on 9th October 2006
The nanocomposite comprising of two simple components – titanium dioxide and Prussian blue – have been synthesized and used for photoelectrode construction. Switching of the photocurrent direction in semiconducting systems upon changes of the electrode potential has been observed. The nanocomposite was characterized by optical spectroscopy and electrochemistry. The structure of the surface complex was modeled using simple quantum chemical models. The behaviour of the photoelectrode was simulated by an adequate electronic circuit. Possible applications of the composite material have been presented.
Recent developments in nanoelectronics3–5 and information processing in chemical systems6–13 have turned our attention towards the application of semiconducting nanoconposites as materials for the construction of optoelectronic devices. Currently used computing devices are based on monolithic semiconductor structures fabricated on the surface of silicon wafers.14–18 All of these devices use binary logic for information processing and utilize electric signals for communication. All of the information is encoded in series of zeros and ones, represented as low and high potential values. Logic gates are basic elements which process information: they function as switches whose output (0 or 1) depends upon the input conditions. Further development (i.e. miniaturization) of semiconductor devices is limited by technological and physical factors.19 Application of chemical structures for information processing seems to bypass, at least temporarily, the limitations of classical silicon technology and sustains the development of information technologies.6,7,9,13,20,21 Furthermore, light seems to be the most efficient medium for communication between chemical logic systems,10,22 and the utilization of light in chemical information processing is very common. Moreover, binary logic can be easily implemented into chemical systems as was shown in numerous reports starting from a breakthrough paper by de Silva et al. in 1993.23 Since that time an untold variety of chemical or biochemical systems has been used for the mimicking of binary information processing and especially of logic gate functions. Surprisingly, semiconducting materials, which at first glance are easily associated with electronics and information processing, were not in the limelight of chemical logic research. Very recently it has been discovered that surface-modified titanium dioxide and other wide bandgap semiconductors may exhibit properties of chemical switches upon simple chemical treatment.24,25 These logic systems utilized light pulses and potential variations as input channels and photocurrent pulses as an output channel. Apart from the facile preparation of logic devices, their integration with silicon-based electronic devices should not present any difficulties.
This paper describes the photoelectrochemical properties of nanocomposites comprised of titanium dioxide particles embedded in a Prussian blue polymeric matrix. The selection of such materials was conditioned by the chemical and photochemical stability of titanium dioxide, its affinity towards cyanometallates and high quantum yields of photocurrent generation. Prussian blue, in turn, was selected due to its photostability and electrochemical activity.26–30
TiO2 (Degussa P25, ca. 70% anatase, 30% rutile; 50 m2 g−1) was used as received. The composite was prepared via deposition of the Prussian blue polymer onto the TiO2 particles suspended in water. The Prussian blue was deposited on the surfaces of the semiconductors using sequential impregnation in 0.01 M aqueous solutions containing the [Fe(CN)6]4− and [Fe(H2O)6]3+ ions, respectively, or by the reaction between [Fe(CN)6]3− and [Fe(H2O)6]3+ (both 0.005 M, 10 ml) with 30% H2O2 (0.5 ml) in the presence of the semiconductor suspension. The latter procedure was also applied to deposit the Prussian blue (PB) layer on silica aerogel (SiO2) and SnO2. After modification, all materials were centrifuged, washed five times with distilled water and air-dried at room temperature. After preparation the modified semiconductors were stored in the dark at 4 °C.
The electric circuit shown later in Fig. 10 was numerically modelled using Spice/XSpice simulator CircuitMaker 6.2 (Protel Technology Inc., USA). The R1, R2, R2′ and R3 resistances were set equal 10 kΩ, RITO + Re + Rload were equal 5 kΩ and the inner source resistance Rs = 500 Ω.43,44 The capacitances of C1 and C3 capacitors were experimentally adjusted to 100 μF and 10 μF, respectively. The capacitance C2 was varied within the range 1 nF–700 μF. The current source was a square wave generator of the following characteristics: initial potential 0 V, peak amplitude 5 V, pulse period 1 s, pulse width 500 ms, pulse rise time 1 ns, pulse fall time 1 ns.
Various cyanoferrate(II) anions are easily chemisorbed onto the surface of titanium dioxide.25,49–58 The binding of the [Fe(CN)6]4− complex onto the TiO2 surface involves the formation of FeII(CN)5–CN–TiIV surface species, characterized by a broad MMCT absorption band in the visible range. Reaction of [Fe(CN)6]4−-modified titanium dioxide with [Fe(H2O)6]3+ cations results in the formation of Prussian blue-like two-dimensional species on the surface of the titania crystals. Subsequent reaction with hexacyanoferrate(II) results in further growth on the Prussian blue thin film on the surface of titanium dioxide as schematically depicted in Fig. 1. This approach enables, at least theoretically, extremely precise control over film growth, but due to surface defects it is difficult to predict the real structure of the Prussian blue deposits. An analogous method of sequential adsorption was shown to be efficient for ITO surfaces28 and Langmuir–Blodgett films.59 Formation of the Prussian blue deposit could be easily seen as a dark blue coloration of the initially white semiconductor suspension. The structure of these films should be almost identical with those obtained during the photochemical deposition of Prussian blue on titania suspensions.45
![]() | ||
Fig. 1 Schematic two-dimensional representation of the TiO2 surface: (a) neat semiconductor equilibrated with water; (b) upon reaction with [Fe(CN)6]4−; (c) upon sequential deposition of Fe3+ ions; and (d) upon second deposition of [Fe(CN)6]4−. Octahedra and spheres depict the [Fe(CN)6]4− anions and Fe3+ cations, respectively. |
The other studied technique involved the slow deposition of bulk Prussian blue in the presence of a titanium dioxide suspension. Addition of a dilute hydrogen peroxide solution to the solution containing an equimolar mixture of Fe3+ and [Fe(CN)6]3− and suspended titanium dioxide resulted in the slow precipitation of Prussian blue according to eqn (1).60
2![]() ![]() ![]() ![]() ![]() | (1) |
![]() | ||
Fig. 2 Scanning electron micrographs of (a) pure titanium dioxide, and (b) the TiO2–Prussian blue composite. |
The polymerization of Prussian blue in the presence of a TiO2 suspension results in material of a very different morphology [Fig. 2(b)]. Prussian blue fills some of the gaps between the TiO2 aggregates, gluing them together. The aggregates seem to be larger, with rather random shapes, as can be expected for clusters of spherical aggregates.
![]() | ||
Fig. 3 (a) Diffuse reflectance spectra of neat TiO2 (—), Prussian blue deposited on SiO2 aerogel (⋯), TiO2–Prussian blue nanocomposite (![]() ![]() |
The recorded diffuse reflectance spectra were transformed to (αhν)0.5vs.hν function, which is used for indirect semiconductors. UV part of the spectrum exhibits a steep slope of semiconductor absorption, and the absorption edge of the spectrum is located at 370 nm, which corresponds to an energy bandgap of ca. 3.35 eV. This is a significant change as compared to neat titanium dioxide (388 nm, 3.22 eV). This change may originate from two factors. First of all, the electronic interaction of Prussian blue with the titanium dioxide particles may significantly shift the energies of the valence band (VB) and conduction band (CB) edges contributing to the increase of the bandgap. Moreover, as the studied TiO2 material comprises small particle aggregates, partial deagglomeration and insulation of individual TiO2 particles should result in an increase of the bandgap due to the quantum size effect.61–64 Broadening of the bandgap of TiO2–PB as compared with neat TiO2 must result from the electronic interaction of both components since the particle size of ca. 20 nm does not rationalize any quantum size effect.65,66 This is in agreement with the observed bathochromic shift of the PB absorption band from 760 to 800 nm measured for SiO2–PB and TiO2–PB, respectively (Fig. 3). This conclusion is supported by comparison of the spectra recorded for the SnO2–PB nanocomposite and SnO2 (data not shown). In the case of the SnO2-based material, neither the bathochromic shift of the 760 nm band nor the bandgap broadening is observed [Fig. 3(a)].
K2FeII[FeII(CN)6] ⇄ KFeII[FeIII(CN)6] + K+ + e− | (2) |
KFeII[FeIII(CN)6] ⇄ FeIII[FeIII(CN)6] + K+ + e− | (3) |
![]() | ||
Fig. 4 Cyclic voltammogram of the titanium dioxide–Prussian blue nanocomposite deposited on an ITO transparent electrode. |
Neat Prussian blue shows very small separation between the cathodic and anodic current peaks of both electrochemical waves. These separations are much higher in the case of the nanocomposite. It may indicate more difficult diffusion of potassium cations through the nanocomposite and/or partial involvement of the semiconducting particles in neutralization of the polymer negative charge. Apart from the above observations the behaviour of the nanocomposite is similar to that of neat Prussian blue. Spectral and electrochemical measurements indicate a weak, but well documented, interaction between the polymeric matrix and the semiconductor nanoparticles.
![]() | ||
Fig. 5 Energy diagram of (a) neat titanium dioxide, (b) titanium dioxide modified with hexacyanoferrate(II), and (c) titanium dioxide–Prussian blue nanocomposite. Potentials vs. Ag/AgCl. |
Photoelectrodes made of neat titanium dioxide generate pulses of anodic photocurrent upon chopped illumination within their absorption spectrum at a wide range of potentials. Excitation with light of energy higher than the bandgap energy results in generation of excitons, which can annihilate (via a radiative or non-radiative pathway) or dissociate yielding electrons in the conduction band and holes in the valence band. At very negative potentials the cathodic photocurrent is generated due to efficient reduction of molecular oxygen present in the electrolyte solution.24,79
The photoelectrochemical behaviour of the TiO2–Prussian blue nanocomposite is much more complex and depends strongly on photoelectrode potential. The fully oxidized nanocomposite, with a matrix containing the FeIIIFeIII species, behaves like a neat titanium dioxide and generates an anodic photocurrent upon illumination. The only difference consists of the photocurrent onset occurring at wavelengths shorter than for neat titanium dioxide, which is explained by the increase of the bandgap energy (vide supra). Apart from the above effect there is apparently no interaction between the Prussian blue matrix and the semiconducting nanoparticles, altering the electronic structure of the semiconductor. The reduction of Prussian yellow to Prussian blue (the FeIIFeIII species) changes the optical properties of the nanocomposite dramatically, but it does not affect the photoelectrochemical properties of the material at all and only the anodic photocurrent is recorded upon illumination. However, the situation changes dramatically upon even partial reduction of Prussian blue to Prussian white (the FeIIFeII species; Everitt's salt) at potentials ⩽300 mV. Completely different photoelectrochemical characteristics of the reduced nanocomposite can be observed (Fig. 6). Pulsed irradiation results only in cathodic photocurrent generation. Moreover, a bathochromic shift of the photocurrent onset is observed. This photosensitization to visible light is less pronounced as in penta- or hexa-cyanoferrate-modified titanium dioxide, but the origin of the process is the same.24,25,49–51,55,56,80 It results from the MMCT transition between the FeII and TiIV centers in the composite. A full picture of the photocurrent dependence on the electrode potential and incident light wavelength can be presented as a three-dimensional photocurrent map (Fig. 7).
![]() | ||
Fig. 6 Photocurrent action spectra of the TiO2–Prussian blue nanocomposite recorded at various photoelectrode potentials in the presence of oxygen. Potentials vs. Ag/AgCl. |
![]() | ||
Fig. 7 Photocurrent generated at the TiO2–Prussian blue nanocomposite as a function of the electrode potential and incident light wavelength. |
![]() | ||
Fig. 8 Frontier molecular orbitals for the [Fe(H2O)5–N![]() ![]() |
It was found that the HOMO orbital of the FeIIFeIITiIV limiting case is localized on four equatorial cyanide ligands and an iron atom within the hexacyanoferrate moiety [Fig. 8(a)]. The LUMO orbital of the same test molecule is localized on the bridging cyanide and the TiIV center [Fig. 8(b)]. Therefore the excitation with sub-bandgap energy light results in electron transfer to the conduction band of the semiconductor and photocurrent generation.
Oxidation of the test system results in complete reorganization of the frontier orbitals. The HOMO orbital of the oxidized model molecule is localized mainly on the four equatorial cyanide ligands with a small contribution from the d-orbitals belonging to the terminal metal centers [Fig. 8(c)]. The LUMO orbital has a dx2−y2 character with a small contribution from the p-orbitals of the four equatorial nitrogen atoms. This electronic configuration does not allow photoinduced electron transfer towards the titanium center and actual photoelectrochemical investigation showed no evidence for any photosensitization in the visible range of the electromagnetic spectrum.
![]() | ||
Fig. 9 Kinetic profiles of the photocurrent generated upon pulsed illumination of the TiO2–Prussian blue nanocomposite: (a) anodic photocurrent at 400 mV; (b) anodic photocurrent with cathodic spikes at 300 mV; and (c) cathodic photocurrent recorded at 100 mV. All potentials are referenced to the Ag/AgCl electrode. |
Complete reduction of Prussian blue to Prussian white results in inversion of the photocurrent direction. It also brings about a dramatic change in the photocurrent kinetics. The second limiting case shows the immediate increase of the photocurrent upon shutter opening and subsequent slow decrease to a steady-state photocurrent. When the shutter is closed, the photocurrent immediately changes its direction to anodic and then gradually decays to the background value. This kinetic traces show two sharp spikes: cathodic upon shutter opening and anodic upon shutter closure [Fig. 9(c)].
Incomplete reduction of Prussian blue to Prussian white results in a mixed type of kinetic traces. The photocurrent develops in the way characteristic for the anodic photocurrent, but every illumination switching on or off results in the generation of a sharp spike, similar to those recorded in a cathodic regime [Fig. 9(b)].
The complex transient photocurrent profiles imply that the semiconducting photoelectrode composed of the TiO2–Prussian blue nanocomposite is a complex electric system (Fig. 10). The photoelectrode can be treated as a series of junctions. The nanocomposite–ITO electrode Schottky junction81,82 can be mimicked by the R1C1 loop circuit.83–87 There are, however, more complex equivalent circuits,86,88–95 but a simple RC circuit reproduces the behaviour of the photoelectrode sufficiently well. In the same way one can describe the nanocomposite–electrolyte junction – the Helmholtz double layer.96,97 This junction is mimicked by the R3C3 loop. The model is completed with the RITO resistor standing for the resistance of the supporting ITO electrode, the Re resistor for the resistance of the bulk electrolyte and the Rload resistor for the resistance of the load. The I0 represents the photocurrent square wave pulses generated at the semiconductor particle upon light pulses and Rs is the resistance of inner part of the semiconductor particle associated with the majority carrier diffusion. More difficult is the description of the nanocomposite layer. The most accurate approach should involve an infinite network of resistors and capacitors,98–102 but a simple R2′C2 circuit also enables reliable simulation of the photocurrent transients.103 This approach would give, however, symmetric cathodic and anodic spikes. The real cathodic photocurrent shows a much higher amplitude than the anodic spikes. This effect can be easily simulated using a diode with a parallel resistor. In this configuration the cathodic photocurrent intensity is higher than the anodic one. The application of a diode in the equivalent circuit is justified as electron transfer between the titanium dioxide and the Prussian white is strongly favoured in one direction (Fe→Ti) and disfavoured in the other one (Ti→Fe).
![]() | ||
Fig. 10 Equivalent circuit for the ITO–nanocomposite–electrolyte junction. |
In order to simplify the computation, R1 = R2 = R2′ = R3 as well as C1 and C3 were kept constant during the whole simulation and C2 was varied over a wide range. This choice is justified by literature examples of the strong influence of the applied voltage on the TiO2 electrode capacitance.44 The Iph current flowing through the Rload upon square-wave stimulation (I0 source) was taken as a model for the photocurrent generated by the photoelectrode upon pulsed irradiation.
When C2 < C1 the current generated by the circuit has typical spiked characteristics [Fig. 9(c)] due to rapid charging of the C1 and C3 capacitors. When the I0 current vanishes the C1 and C3 capacitors discharge through the load resistor yielding the reversed current pulse. With increasing C2 capacitance the spike becomes smoother as the charging curve of the C2 capacitor contributes more and more significantly to the overall current. At C2 ≈ 3 × C1 the current–time profile changes to the purely charge–discharge curve of the C2 capacitor and the spikes due to the charging of the C1 and C3 capacitors become invisible. Values of the C1 and C3 capacitors are responsible for the amplitude and shape of the spikes generated at low C2 capacitance. At very low C1 and C3 only the current originating from the charging and discharging of the C2 capacitor can be observed. Increasing the C1 and C3 values results in the appearance of spikes and the subsequent increase of the spike amplitude and prolongation of its decay time. It is consistent with the increasing time constants of R1C1 and R3C3. Furthermore, variation in the R2 resistance changes the ratio of the anodic and cathodic spikes. The results of the numerical simulation of the photocurrent generated at the titanium dioxide–Prussian blue nanocomposite are presented in Fig. 11.
![]() | ||
Fig. 11 Simulated transient photocurrent profiles generated in the model circuit from Fig. 10 for various C2 values: spiked transients for the cathodic photocurrent (solid bold line) and charge–discharge transients characteristic for the anodic photocurrent (dashed bold line). See text for details. |
This simple model illustrates the most important features of photocurrent generated by semiconducting photoelectrodes. The photocurrent kinetics can be switched between two limiting regimes just by changing the C2 capacitance which corresponds to the capacitance of the semiconductor particle.
Apart from these purely electronic effects, the electrostatic influence of the surface complex on the electrons inside the semiconductor particle seems to be very important (Fig. 12).63 The oxidized form of the Prussian blue polymer (containing only FeIII centers) interacts only weakly with the surface and is a relatively good electron acceptor. Therefore it should bend the valence and conduction bands downwards which results in the formation of an accumulation layer [Fig. 12(b)] and only anodic photocurrents can be generated. Upon electrochemical reduction of this material the surface concentration of FeII species increases, which results in the formation of a depletion layer, thus decreasing the efficiency of the anodic photocurrent [Fig. 12(c)]. Upon further reduction the semiconductor photoelectrode can generate only a cathodic photocurrent, irrespective of the irradiation wavelength [Fig. 12(d)].
![]() | ||
Fig. 12 Simplified electronic structures of semiconductor particles: (a) neat semiconductor; (b) semiconductor modified with the oxidized form of the complex, an accumulation layer is formed due to interaction with the surface complex; (c) semiconductor modified with the partially reduced form of the complex, a depletion layer is formed due to interaction with the surface complex; (d) an inversion layer is formed upon complete reduction of the surface species. In the case of small particles vertical shifts of the VB and CB instead of band bending should be considered. |
The changes of the redox state of the surface of the semiconductor particles also influence the electric properties of the ITO–semiconductor and semiconductor–electrolyte Schottky junctions. The height of the Schottky barrier depends upon the charge and electron transfer properties of the surface complex. Cyanoferrate(II) moieties of high negative charge increase the energy of the Schottky barrier thus increasing the resistance and capacitance of the junction. This must result in a significant change in the photocurrent kinetics: a transition from charging–discharging to spiked characteristics is observed during switching from the anodic to cathodic regimes of photoelectrochemical activity, which is consistent with observations in an equivalent electric circuit (cf.Fig. 10 and Fig. 11). Moreover, it was found that the resistance and capacitance of the semiconductor particles strongly depend upon the electrode potential.44 This further justifies the strong potential dependence of the kinetic behaviour of the nanocomposite electrodes.
![]() | ||
Fig. 13 Electronic equivalent circuit of the nanocomposite photoelectrode working as a two-channel optoelectronic demultiplexer. |
Thus, a signal applied to the data input is always transmitted by one of the buffers and directed into one of the output channels. It is easy to imagine the application of similar systems in telecommunications. Nowadays, information transmitted through optical fibers must be converted into electric signals and directed to the destination point. The PB@TiO2 system can serve as a simple two-channel demultiplexer. This is an interesting example of a simple chemical system working like a complex electronic circuit consisting of three logic elements.
Footnote |
† Electronic supplementary information (ESI) available: colour version of Fig. 8. See DOI: 10.1039/b606402g |
This journal is © The Royal Society of Chemistry 2006 |