Metal oxide photocatalysis is based on the use of metal oxides (for example titanium dioxide) as light-activated catalysts in the destruction of organic and inorganic materials and in organic chemistry synthesis. In this article, we will be looking at the use of thee types of materials in the degradation of organic matter, which has applicability in environmental remediation (aqueous and air-borne) and self-cleaning surfaces. The technique is already widely used in commercial applications, but is still hampered by one significant limitation. These materials generally absorb primarily ultra-violet light, and research in recent years has been concentrating on developing visible-light active materials, with an emphasis on nano-particulate materials to maximize surface area. This article discusses the background to metal oxide photocatalysis, using titanium dioxide as the exemplar material, and looks at strategies being researched to enhance the photocatalytic efficiency.
Titanium dioxide is a white powder, with titanium in oxidation state IV. Its d-electron configuration is therefore d0, and the white colour is explained by the lack of d-d or metal centred transitions. It exists in several polymorphs – two of interest here: anatase and rutile. As it is a semiconductor, its HOMO is termed a valence band and LUMO is termed a conduction band. Light absorption effectively results in a ligand to metal charge transfer, electrons from oxygen are transferred to the vacant titanium d-orbitals. For anatase (3.2 eV) and rutile (3.0 eV), this transition is in the UVA region, resulting in a sharp absorption band at 390 – 400 nm.
Looking more closely at the electronic processes, promotion of an electron to the conduction band, on irradiation by UV light, results in a ‘hole’ in the valence band – essentially a detriment of the electron density that was localised on that orbital, and usually assigned a positive charge to symbolize the loss of negative electron (of course negative and positive are just arbitrary notations). The hole is powerfully oxidizing – the orbital very much wants to retrieve electron density just lost after light irradiation. It can retrieve this simply by the electron in the conduction band recombining with the valence band – recombination is a sum of radiative (i.e. emission may be observed) and non-radiative processes. Based on the energy gap law, the fact that rutile energy levels are closer mean that the non-radiative process is more efficient, and hence recombination is more efficient. This is an important observation which we will return to shortly.
Alternative pathways to recombination are possible, and as you can guess, these result in the use of these materials as photocatalysts. The hole has the potential to oxidise water that may be on the surface of the material resulting in the formation of hydoxyl radicals. Hydroxyl radicals are themselves very powerful oxidisers, and can easily oxidise any organic species that happens to be nearby, ultimately to carbon dioxide and water. Meanwhile, upstairs in the conduction band, the electron has no hole to recombine with, since it has oxidised surface bound water. It quickly looks for an alternative to reduce, and rapidly reduces oxygen to form the superoxide anion. This can subsequently react with water to form, again, the hydroxyl radical. The processes are summarized below.
At the level of the material’s surface, the requirements for efficient photocatalysis can be deduced from the electronic reactions – there should be surface bound water to allow for efficient oxidation; and the water should be aerated to provide oxygen to the solution. Additionally, the degradation of the pollutant by the catalyst requires for the pollutant to be adsorbed or very close to the surface of the material, and hence the greater the surface area of the material, the more pollutant can adsorb. Nanoparticulate materials are therefore preferred as they vastly increase the surface area (see DSSC post).
Pilkington self-cleaning glass is an example of use of this technology in a commercial application. A thin film of nanoparticulate titanium dioxide is coated onto panes of glass (it is so thin that it is transparent). The glass is in the normal course of events, acquiring dirt. The titanium dioxide on the glass, once exposed to sunlight, produces hydroxyl radicals which degrade any surface adsorbed dirt. Once washed down with rain, this decomposed dirt is removed and the glass is ready for another cycle. The same process is observed for any organic species – they react with the hydroxyl radical to ultimately form carbon dioxide and water.
Given that the materials work readily, it is a good time to detail the limitations. the primary limitation is that the materials absorb only UV light, so the activation by sunlight is completed by the 5% of sunlight that is in the UV region. A large amount of research has looked into ways to enhance the visible light activity of the materials. Another limitation is the fact that recombination is an efficient, competitive process, and given that this is a less efficient process with anatase, it is generally accepted that anatase is a preferred photocatalyst to rutile. Below, we will discuss approaches taken to both increase the visible light absorption capability and increase the efficiency of subsequent reactivity over the recombination process.
Moving to Visible Light Absorption Capability
Given the requirement for UV light activation of TiO2, researchers became interested in tuning the materials so that they would become activated by visible light (e.g. room light) for applications for indoor use or by solar light for outdoor use. Various approaches were considered, and in 2001, a Japanese chemist named Asahi working out of Toyota labs, published a paper in the journal Science on nitrogen doped titanium dioxide materials. Nitrogen doping produced what is commonly called yellow TiO2 (because of, unsurprisingly, its yellow colour!) which showed effective UV and visible light activity. While there is some debate around how the activity is increased, the N-doped TiO2 is shown to have a much greater absorbance in the visible region (extending from a sharp cut off at about 390 nm to a broad cut off at above 500 nm). This subsequently increased the amount of visible light activity the material could absorb, and hence meant that visible light-activated photocatalysis was achievable.
There has been some discussion in the literature on the mechanism on enhancement of nitrogen doping, and the mechanism described here is one put forward by Nakoto (2004) and Irie (2003), and counters Asahi’s original explanation that the N-doping narrowed the gap between the valence band and conduction band of titania. these researchers proposed that the introduction of nitrogen introduced new occupied (i.e. electron rich) orbitals in between the valence band (which are comprised primarily of O-2p orbitals) and conduction band (which are comprised primarily of Ti-3d orbitals). These N-2p orbitals acted as a step up for the electrons in the O-2p orbital, which once populated had now a much smaller jump to make to be promoted into the conduction band.Once this process occurs, electrons from the original valence band can migrate into the mid-band gap energy level, leaving a hole in the valence band, which reacts as described before.
Increasing efficiency by incorporation of metal nanoparticles
Given that charge separation requires a great deal of effort, a second theme of research (as well as increasing visible light activity) is to facilitate charge separation. One clever way of doing this is to incorporate noble metal nanoparticles such as silver or gold into the titanium dioxide material. As an example, incorporation of a small amount of silver (1 – 5%) results in increased efficiency in photocatalysis. Silver has a “Fermi level” or electron accepting region at an energy just below the conduction band. Therefore, after light absorption and charge separation, the electron in the conduction band can be effectively trapped by the silver, while the hole oxidises water and forms hydroxyl radicals, without the threat of recombination. Various researchers, including our own work, have shown that there is an optimum amount or “Goldilock’s zone” of silver to add – just enough is needed so that there are silver sites dispersed through the material to rapidly trap electrons, but that too much silver may cover the titanium dioxide and prevent light absorption. In addition, too much silver may mean that the silver acts as a recombination site itself – essentially it will form a bridge between an electron and a hole.
The emission of titanium dioxide (and of similar studies with zinc oxide) can be interpreted as a measure of the recombination efficiency. Studies examining the emission of these metal oxides have demonstrated that the emission intensity reduces on increasing amounts of silver – indicating that the silver is trapping electrons and reducing electron-hole recombination, as indicated in the diagram below.
A similar strategy to that described above, an a rapidly evolving area, is the idea of incorporating different semiconductors which have different conduction band energy levels. The strategy is as before, trap the electron so the hole has more time to react. A simple example is the anatase-rutile heterojunction. Rutile has a smaller band gap (by about 0.2 eV) to anatase, although their valence band levels are at similar energies. Therefore, in an analogous fashion to the situation with silver, above, charge separation in anatase, followed by electron injection into the rutile conduction band means that there is a hole in the valence band of anatase that can freely oxidise water. It is no coincidence that the industry standard photocatalyst, Degussa P25, has a 75:25 ratio of anatase:rutile (it also has a very small particle size).
Semiconductor photocatalysis is the utilisation of photogenerated strongly oxidising hydroxyl radicals, which can be applied to a wide range of scenarios, including organic degradation (for pollution remediation) and in organic synthesis. Light induced charge separation, followed by generation of hydroxyl radicals is in the normal course of event reliant on UV light, given the energy gap (band gap) of titanium dioxide. Strategies to enhance the photocatalytic activity include doping to reduce the energy required for charge separation and incorporation of nanoparticles to lengthen the period of charge separation. The size of the materials is also a factor, as for degradation of materials, the pollutant needs to be very near to or adsorbed onto the surface of the semiconductor, and nanoparticulate materials mean that a greater surface area can be exploited.
Asahi, R., Morikawa, T., Ohwaki, T., Aoki, K. and Taga, Y., Visible-light photocatalysis in nitrogen-doped titanium oxides, Science, 2001, 294, 269 – 271. Asahi’s paper describing his results on N-TiO2. the work shows irradiation by UV-only and visible-only light, showing the enhancement by N-TiO2 with visible light source.
Bahnemann, D., Photocatalytic water treatment: solar energy applications, Solar Energy, 2004, 77, 445–459. Prof Bahnemann is one of Europe’s most active researchers in this field, and this very readable paper shows how the technology can and is used in solar decontamination technology.
Nakamura R, Tanaka T, and Nakato Y., Mechanism for visible light responses in anodic photocurrents at N-doped TiO2 film electrodes, J. Phys Chem. B., 2004, 108, 10617 – 10620. (See also Irie, H et al, J. Phys Chem. B., 2003, 107, 5483 – 5486). Papers explaining the origin of the hypothesis for the mid-gap energy levels introduced by nitrogen doping.
Seery, M. K., George, R., Floris, P. and Pillai, S. C., Silver doped titanium dioxide nanomaterials for enhanced visible light photocatalysis, J. Photochem. Photobiol A: Chemistry, 2007, 189(2-3), 258 – 263 and Georgekutty, R., Seery, M. K. and Pillai, S. C., A Highly Efficient Ag-ZnO Photocatalyst: Synthesis, Properties and Mechanism, J. Phys. Chem. C, 2008, 112(35), 13563 – 13570. these papers detail the incorporation of silver into titanium and zinc oxides respectively, including some consideration of mechanism.
Increasingly, solar energy has a vital role to play in providing energy to cater for an ever increasing demand. This article looks at what dye-sensitised solar cells are and their current technological status, as well as what needs to be done to make them big hitters in the energy game. It summarises some recent reviews on the topic, interested students are pointed to the source material and other references at the end of the article. [Aug 2009]
A FEW YEARS AGO when talking about dye-sensitised solar cells in lectures, a student asked me what the problem with them was. Gratzel had published his seminal paper in 1991, and now, over 15 years later, from an outside observer’s perspective, there wasn’t much progress in terms of developing a commercial applicable devide. It’s a good question, and one worth asking periodically of any innovation. We hear that drugs often take 10 – 15 years from bench to patient, so one might reasonably ask with DSSC – “What’s the delay?“.
Here’s a short podcast (with audio) introducing this section outlining what this article will cover:
Now that we are near the end of the first decade of the 21sr century, future energy demands make for sobering reading. The world currently uses about 13 terawatts (TW) of energy, and it is predicted by 2050, an additional 10 TW will be required. Not only that, the additional energy required will be have to be carbon neutral. Not only that, we are heading for peak oil. Considering all of this, what is the role of solar energy?
Kamat (2007) has summarised the role of solar energy very clearly with data that… ahem… blows other alternative/renewable energy sources out of the… ahem… water. Of the 10 TW additional energy required, building 1 GW nuclear power station every day for the next 50 years would meet the demand. (Nuclear isn’t strictly speaking renewable.) Hydroelectric could provide about 0.5 TW, tides and oceans could chip in 2 TW and wind power could blow 2 – 4 TW our way. Solar energy striking the earth amounts to 120,000 TW,
yet only 0.01 – 0.04% of current energy usage is derived from solar sources. (Approximately 13% of current energy needs being supplied by renewable sources.) In theory, one hour’s solar irradiation is enough to supply a year’s global energy demands.
How would increasing the role of solar energy manifest itself on the ground? Solar flux at ground level is approximately 340 W/m2 in the world’s sunniest areas. Assuming 10% efficiency, each metre-squared of solar cell could generate 34 W. Plugging in the numbers, you’d be looking at a mere 4 x 10^11 m2, or about 618 km square to address the increase in world energy needs by 2050. While large, it’s not unrealistic. (Dublin county has an area of approximately 115 km square). The map below shows how, based on average sunlight irradiance measured over 1993-1994, how placement of solar energy “stations” at various locations around the planet would provide a substantial amount of solar-derived energy.
Required land area to supply an average of 18 TW (by Matthias Loster, 2006, reproduced with permission, full details on source website)
Dye-sensitised solar cells are as a concept, ingeniously simple. The idea was first conveived in the late 1970′s but since a Swiss photochemistr, Michael Gratzel, published a Nature paper in 1991 reporting 7% efficiency, the interest in the systemhas grown enormously. The outline of a DSSC is shown below and discussed in detail in section 4. Light harvesters gather in energy from a solar/light source and pass on the energy to an electrical circuit which does work (how do you think this compares with nature’s way of generating energy from sunlight?!).
Light harvesting dyes absorb solar radiation incident on them. This results in excitation of these molecules, who pass the energy obtained by means of transferring electrons onto a nanocrystalline TiO2 substrate onto which they are adsorbed. The electrons, now in the conduction band of titanium dioxide, conduct around a circuit and do work. At some counter electrode, a redox couple is utilised (usually iodide-triiodide) to regenerate the dye so the process can occur all over again. Assuming the dye is efficient at harvesting light, the transfer of electrons to titanium dioxide is efficient, the conduction of electrons in the circuit displays good potential and the dye can regenerate multiple-million times before being degraded, the concept works well with reasonable efficiency (~7 – 11%).
Slideshow: Electronic Pathways in Dye-sensitised Solar Cells:
The chemistry involved is sandwiched between two sheets of conducting glass, coated with a conductive layer (e.g. ITO) which is transparent. One plate of glass (working electrode) is coated with titanium dioxide nanoparticles that have the dye adsorbed onto the surface and the other (counter electrode) is coated with a catalyst (platinum or carbon). The plates contain electrolyte solution between them with the redox couple to regenerate the dye.
Here’s a video of a DSSC in action:
4. Developing the idea further
The efficiency of dyes can be measured by consdering how many absorbed photons result in electron injection and how many of these injected electrons are collected to be used in the electrical circuit. This is expressed according to Equation (1), where IPCE(λ) is the incident photon to current efficiency – a measure of how many photons translate into electrical current.
IPCE(λ) = LHE(λ) × Φ(inj) × η(c)
The light harvesting efficiency (LHE) is the fraction of photons absorbed by the dye at a particular wavelength. The electron injection efficiency (Φ(inj)) is a measure of how many absorbed photons result in an injected electron into the semi-conductor and the charge collection efficiency is a measure of how many of these injected electrons are collected for electrical use. The equation essentially maps out each of the processes in the cell and considers their efficiency. All of the processes in the DSSC are kinetic – their efficiency is determined by how fast they occur relative to competing processes. We’ll consider below the various components of the dye and identify where any efficiencies could be improved upon in future developments. This section is based mainly on Hupp (2008 and subsequent more recent articles).
4.1 Dye Characteristics
The light harvesting dye is clearly a crucial component of the cell design and needs to fulfil several criteria; adsorption onto metal surface, overlap effectively with solar spectrum, inject electrons efficiently into metal oxide and be stable for many million cycles.
Adsorbtion of the dye onto the metal oxide surface is facilitated by incorporating a substituent that will adsorb readily. The most efficient studied are ruthenium dyes with carboxyl-substituted ligands – these carboxyl substituents adsorb onto the dyes surface.
The spectral overlap with the solar spectrum should be maximised so that as much of the sun’s energy as possible is utilised in exciting the dye, and promoting a high density of electrons into the excited state. In practice, dyes absorb in the visible and near infrared region (about 400 – 700 nm), capturing about half the available power and a third of the available photons from solar source. The ruthenium complexes which are currently “best in show” do have a limitation in that their exctinction coefficients are comparatively low (1 – 2 x 10e4 M-1 cm-1), requiring several hundred monolayers which in turn requires the metal oxide support to have a very high surface area. To achieve efficiencies of >15%, DSSCs will need to absorb bout 80% of light between 350 – 900 nm. Therefore using materials that have a higher absorption capacity may be a useful future strategy. Research here has included using osmium in place of ruthenium, which extended the absorption further into the red and enhanced the response of the cell to light relative to the ruthenium analogue. The 1MLCT to 3MLCT transition in osmium is much more intense than in ruthenium. Organic dyes have also been used successfully as attested by the very many articles and school projects on using fruit berries as the dye in these cells. A range of dyes are shown below, and while they vary in chemical structure, you should note a common factor between them and between these and the ruthenium dye shown above. Organic dyes have much larger extinction coefficients (5 – 20 x 10^5 M-1 cm-1), albeit across a narrow range than the ruthenium counterparts. (It seems that dyes will either absorb moderately well across a broad range or very well across a narrow range!)
In order for electronic transfer to be energetically favourable, the excited state energy of the dye should be higher in energy than the conduction band of the semiconductor. As well as this, the kinetics of electron injection into titanium dioxide should be faster than recombination of the dye (by luminescence or non-radiateive decay). This isn’t a problem with the N3 dye, above which injects on a femtosecond timescale and decays at a much more leisurely sub-picosecond timescale. (You might consider in your studies how this data would be determined experimentally). Electrons are injected from both the 1MLCT (on a sub-ps timescale) and the 3MLCT, which is formed within 100 fs by intersystem crossing (due to the heavy atom effect of ruthenium). Recent research (Durrant, J. Am. Chem. Soc., 2009, 131, 4808) has questioned whether similar difference in rates are observed in real DSSC (as opposed to model systems), based on results showing that in these real systems, the electron injection slowed down to ps timescale, allowing recombination to be competitive. Nevertheless, it is considered that this process is efficient, although caution is required in ensuring that efficiency isn’t reduced by other design factors on the cell.
Considering the discussion above regarding extending the dye’s absorption further into the red, this could be achieved by lowering the LUMO of the dye, although researchers have been reluctant to do this as since the LUMO-CB transition is so fast, the energy levels must be very well matched. (Goldilocks Principle: Not too high, not too low, just right).
Very recent research at Stanford University has coupled luminescent chromophores which absorb high energy photons and pass their energy on to the sensitising dye (Nature Photonics, 2009, 3, 406 – 411) – I’ll put more on this in a future article.
4.2 Metal Oxide Support
Following sucessful injection into metal oxide, the next phase is for the elctron to percolate through the oxide layer onto the working electrode. TiO2 is the most common substrate used. As a chemical, it is relatively inert, cheap and can be synthesised via the sol-gel process (offering flexibility and scalability). Unlike silicon solar cels, very high purity is not required. It exists in various forms, mainly anatase, rutile and brookite. Anatase has a high band gap (3.2 eV compared to rutile’s 3.0 eV) which gives several advantages. It absorbs very little of the solar spectrum, meaning it is transparent to incoming light source (so that the dyes rather than the metal oxide is activated by light). In addition, the larger band gap than rutile means that recombination is slower (ref the energy gap law) – some reports have determined a 30% lower efficiency with rutile.As with the dye, there are several factors to consider to maximise efficiency.
Gratzel’s innovation in his 1991 paper was among other things to use nanocrystalline TiO2. The nanomaterial’s much greater surface area meant that many more molecules of dye could adsorb onto the surface and pushed efficiency to the best reported at the time: 7%. Nanocrystalline surfaces in these cells are reported to have a surface: geometrical area of ca. 1200. Best performers actually have two distinct layers of metal oxide: a 12 micron thick transparent layer of 10 – 20 nm sized particles covered with a 4 micron thick layer of much larger (400 nm) particles. The larger particles scatter photons back into the film.
One significant limitation of this model is the very slow time taken for electrons to percolate through the material and onto the transparent conducting electrode; which is of the order of 100′s of microseconds, in comparison with conduction band-dye recombination which is of the order of a couple of microseconds. (It is however faster than the other decay process: conduction band-redox couple, which is in the millisecond range). Because of this, researchers are looking at improving the dynamics of electron percolation through the film. One strategy is to use nanotubes/rods rather than particles. These have the disadvantage of having lower surface area, but an advantage of very much improved electron transport because they provide an physical electron pathway from the particle to the electrode. Pagliaro (2009) gives a nice summary of this and this and other work on ZnO nanorods will be summarised in a future article.
4.3 Electrolyte and Regeneration
The electrolye contins the redox couple which regenerates the oxidised dye whcih were formed byinjection of electron from the dye to the titanium dioxide layer, leaving D+. The redox couple has a difficult job – it must be very efficient at reducing the dye cation back to the original state for another cycle, but not intercept or capture one of the electrons being injected in the first place! The latter process, which would result in inefficiency involves transfer of an electron from the conduction band of titanium dioxide to the redox couple (to the triiodide ion I3-). Of the range of redox couples studied, iodide-triiodide has proved to be in the most efficient cells, mainly because the transfer of electron from titania to trioidide is, as mentioned above, very slow (in the millisecond time range). One of the problems in trying to look for improved systems is that the redox chemistry of this redox couple is not well understood, so it is difficult to plan effective alternative dyes. One interesting approach is to use solid state redox couples, which allow for higher concentrations of the redox couple and extending the applicability of the device (because of removal of the liquid electrolyte).
5. Summary and Review
Dye-sensitized solar cells offer enormous potential as an alternative renewable energy provider. The principle of operation is for a light harverster to absorb light efficiently and pass on the energy to a metal oxide surface, which links in to a circuit generating current. The dye is regenerated at the counter-electrode via a redox couple. In reviewing dye sensitised solar cells, you should consider the two main themes covered in this article:
- Describe the principle of operation outlining the pathway the electron follow from light absorption to dye regeneration. Using the diagram below, represent these processes on an energy level diagram and for each of the stages outline the counter process available resulting in efficiency. By using time constants/scales, indicate whether these processes are likely to significantly impact on the efficiency of the cell.
- For each of the processes, outline improvement strategies that are being considered in current research in the area, indicating the rationale for such approaches.
This diagram below, from Hupp et al., Chem. Eur. J., illustrates effectively the rates of competing processes:
- Meeting the clean energy demand: nanostructure architechtures for solar energy conversion, P. V. Kamat, J. Phys. Chem. C, 2007, 111, 2834 – 2860. Excellent article on the wide variety of roles nanotechnology may play in providing future energy needs. This paper was used to provide much of the context. Prof. Kamat’s website is also worth looking at. Google “Light Energy Conversion” and this is the web-page that comes up, with good reason.
- Advancing beyond current generation dye-sensitized solar cells, Hamann, T.W., Jensen, R. A., Martinson, A. B. F., Van Ryswyk, H and Hupp, J. T., Energy Environ. Sci., 2008, 1, 66 – 78. An excellent paper on the properties of DSSC and how they may be technically advanced in future developments. Section 4 is substantially based on this article. Link
- Dye-sensitized solar cells: a safe bet for the future, Gonçalves, L. M., de Zea Bermudez, V., Aguilar Ribeiro, H. and Magalhães Mendes, A, Energy Environ. Sci., 2008, 1, 655 – 667. A concise overview on the device aspect of DSSC, considering current and future implementation requirements.
- Nanochemistry aspects of titania in dye-sensitized solar cells, Pagliaro, M., Palmisano, G., Ciriminna, R. and Loddo, V., Energy Environ. Sci., 2009, 2, 838 – 844. Good overview of the photoanode requirements and research developments.
Published August 2009. Subsequent additions/amendments/corrections will be logged in comments.