The amount of energy from sunlight, striking the surface of the Earth in one hour, is just short of the energy consumed on the planet in one year. A major challenge, therefore, is to be able to convert a source of practically limitless energy (i.e. the sun) into chemical energy (i.e. hydrogen) with very little or no carbon footprint. There are many ways of splitting water; thermochemically, biophotolytically, mechano-catalytically, plasmolytically, magnetolytically and radiolytically. The most promising route though, is photoelectrolytically.
The process is summarised as follows: H2O + hν + photocatalyst → 1/2 O2 + H2
This reaction is catalysed by many inorganic semiconductors, the first of which was TiO2, developed in 1971 by Fujishima and Honda. To date, over 130 materials have been discovered that can catalyse the splitting of water according to the above equation. Figure 1 shows a plot of quantum efficiency (QE) against wavelength for a wide range of materials. For any prospective photocatalyst to be considered commercially viable it has to display a QE of greater than 10%, in the visible region of the electromagnetic spectrum. To date no materials have been discovered that satisfy these criteria (as seen in the area circled red).
Figure 1: The plot shows the experimentally determined quantum efficiency (QE), against the wavelength of incident photons (used to perform the experiments).
To develop an efficient photocatalyst, it is vital to achieve a fundamental understanding of the chemical and physical processes that are occurring. Plants and some bacteria release oxygen through the process of photosynthesis, which is an ecient method of transforming solar energy into energy rich materials (starch or sugar). The process involves the photolysis of water to liberate O2 and has evolved over billions of years, with a peak illumination photochemical conversion yield of 6%. This highlights the scale of the current challenge, if an \articial photosynthetic” photocatalyst is to be discovered with an efficiency of greater than 10%. Figure 2(a) shows the main processes of the photocatalytic reaction in a material. They are (i) the absorption of photons to form (ii) electron-hole pairs, (iii) charge separation, (iv) migration to surface reaction sites and (v) suppression of charge recombination. To understand the processes in Figure 2(a) it is necessary to look at the electronic structure of the material, shown in Figure 2(b). Semiconductors have a band structure in which the conduction band (CB) is separated from the valence band (VB) by a band-gap of a width specic to each individual material.
Figure 2: (a) Schematic of the processes occuring in a typical single absorption photocatalyst. Note: The smaller the particle size, the lower the rate of recombination (b) Electronic structure of a typical single photon absorption photocatalyst.
Band levels of various oxides are shown in Figure 3. The most important factors for a potential photocatalyst are the magnitude of the band gap, and the respective energy levels of the CB and VB. The bottom level of the CB has to be more negative than the redox potential of H+/H2 (O V at NHE), while the top of the VB has to be more positive than the redox potential of O2/H2O (1.23 eV). Therefore, the theoretical minimum band gap for water splitting is 1.23 eV. The maximum band gap for efficient photocatalysis in visible light is 3.0 eV, which is the visible region cut-off of the solar spectrum.
Figure 3: Band edge positions of several oxide semiconductors in contact aqueous electrolyte (pH = 2). *The band gap has recently been revised from 3.75 eV to ∼2.8 eV.
It is immediately apparent from Figure 3, that the majority of oxides have VB edges at an energetically unfavourable distance (too positive) from the oxidation potential of reduction reaction of H2O/O2. This means that the VB maximum (VBM) for most oxide materials are too \far away” for it to occur efficiently. The VBM is dominated by the anion character of the material, how would changing the anion, change the band position? Metal nitrides and sulphides have higher energy VBMs (red and yellow lines respectively), making them more compatible with the oxidation potential of H2O/O2. The major drawback is that these materials suffer from extensive photocorrosion, as the anion is typically oxidised instead of water.
Work on BiVO4
The electronic structure of BiVO4 has been studied by x-ray photoelectron, x-ray absorption, and x-ray emission spectroscopies (XPS, XAS & XES), in comparison with density functional theory (DFT) calculations. Our results confirm both the direct band gap of 2.48 eV and that the Bi 6s electrons hybridize with O 2p to form antibonding “lone pair” states at the top of the valence band. This result confirms the validity of using the RLP model to describe BiVO4, specifically the Bi 6s-O 2p antibonding state derived VBM. As a result, we can assign the fundamental Eg at 2.4 eV (Fig. 4) as due to transitions between the hybrid Bi 6s-O 2p orbital and unoccupied V 3d states with significant O 2p and Bi 6s contributions. Meanwhile, the peak in the diffuse spectrum ∼0.5 eV above the onset edge is likely associated with occupied O 2p states solely, thus experimentally confirming the original hypothesis of Kudo et al. (Kudo et al., J. Am. Chem. Soc. 121 11459 (1999)), used to explain the difference between tetragonal and monoclinic BiVO4. Experimental confirmation of the hybrid O 2p–Bi 6s character of the VBM, adds further credence to the claims by Walsh et al. (A. Walsh et al., Chem. Mater. 21 547 (2009)) of the symmetric light masses for both holes and electrons within this compound.
Figure 4: (a) The diffuse reflectance spectrum of monoclinic BiVO4. A linear extrapolation to the baseline was used to determine the on- set of the conduction band i.e., Eg. (b) The O K-edge XES (solid line) and core-hole corrected TEY-mode O K-edge (XAS dashed line) spectra of BiVO4. The derivatives (solid and dashed lines) are plotted underneath, along with the energetic separation between their edges (i.e., Eg).
The nature of electron lone pairs in BiVO4
D. J. Payne, M. D. M. Robinson, R. G. Egdell, A. Walsh, J. McNulty, K. E. Smith, L. F. J. Piper, Applied Physics Letters 98 212110 (2011).
In a thin film semiconductor electrode, a space-charge layer (electron depletion layer) forms at the semiconductor – electrolyte interface. Charge-carrier separation occurs as a result of the internal electric eld formed at the depletion layer. In nanocrystalline materials, the interfacial kinetics are considered more important than the internal electric eld because individual particles are too small to form a depletion layer, and therefore charge separation of the photoexcited electron-hole pairs is dominated by diusion rather than electric eld assisted drift. A phenomenon observed in nanomaterials is an increase in the magnitude of the band gap with decreasing particle size (due to the quantum size eect). As the size of the semiconductor falls below a critical radius, charge carriers begin to behave quantum mechanically, and the charge connement leads to a series of discrete electronic states. As a result, there is an increase in the eective band gap and a shift of the band edges. Thus by varying the size, it is possible to enhance the redox potential of the VB holes and the CB electrons. For oxide materials, an increase in the CB edge is advantageous, as it can be seen in Figure 3, the CB edge generally lies below the H2/H2O reduction potential. Thereby, combining anion doping, which shifts the VB edge upwards and closer to the H2O/O2 oxidation potential, together with nanosize-induced increases in the CB edge, above the H2/H2O reduction potential, band gap engineering of these materials can lead to increased photocatalytic performance.