dye-sensitised mesoscopic solar cells

503

CHAPTER 8

DYE-SENSITISED MESOSCOPIC SOLAR CELLS

MICHAEL GRÄTZEL Laboratory for Photonics and Interfaces, Swiss Federal Institute of Technology,

CH–1015 Lausanne, Switzerland

[email protected]

and JAMES R. DURRANT

Department of Chemistry, Imperial College, London SW7 2AZ, UK

[email protected]

On the arid lands there will spring up industrial colonies without smoke and without smoke-

stacks; forests of glass tubes will extend over the plains and glass buildings will rise every-

where; inside of these will take place the photochemical processes that hitherto have been the

guarded secret of the plants, but that will have been mastered by human industry which will

know how to make them bear even more abundant fruit than nature, for nature is not in a hurry

and mankind is. And if in a distant future the supply of coal becomes completely exhausted,

civilisation will not be checked by that, for life and civilisation will continue as long as the Sun

shines!

Giacomo Ciamician, Eighth International Congress of Applied Chemistry, Washington and New York, September 1912.

8.1 Introduction

Photovoltaic devices are based on charge separation at an interface of two materials of different conduction mechanism. To date this field has been dominated by solid-state junction devices, usually made of silicon, and profiting from the experience and material availability resulting from the semiconductor industry. The dominance of the photovoltaic field by inorganic solid-state junction devices is now being challenged by the emergence of a range of new device concepts, including devices based on nanocrystalline and conducting polymer films. These devices offer the prospect of very low-cost fabrication and present a range of attractive features that will facilitate market entry. It is now possible to depart completely from the classical solid-state junction device by replacing the phase contacting the semiconductor by an electrolyte,

M. Grätzel and J. R. Durrant

504

liquid, gel or solid, thereby forming a photoelectrochemical cell. The phenomenal progress realised recently in the fabrication and characterisation of nanocrystalline materials has opened up vast new opportunities for these systems. Contrary to expect-ation, devices based on interpenetrating networks of mesoscopic semiconductors have shown strikingly high conversion efficiencies, competing with those of conventional devices. The prototype of this family of devices is the dye-sensitised solar cell (DSSC), which realises the optical absorption and charge-separation processes by the association of a sensitiser as light-absorbing material with a wide-bandgap semiconductor of nanocrystalline morphology (O’Regan and Grätzel, 1991). 8.2 Historical background

The history of the sensitisation of semiconductors to light of wavelength longer than that corresponding to the semiconductor bandgap has been presented elsewhere (McEvoy and Grätzel, 1994; Hagfeldt and Grätzel, 2000). It represents an interesting convergence of photography and photoelectrochemistry, both of which rely on photoinduced charge separation at a liquid–solid interface. The silver halides used in photography have bandgaps of the order of 2.7–3.2 eV, and are therefore insensitive to much of the visible spectrum, as are the metal oxide films now used in dye-sensitised solar cells.

The first panchromatic film able to render the image of a scene realistically into black and white followed on the work of Vogel in Berlin after 1873 (West, 1874), in which he associated dyes with the halide semiconductor grains. The first sensitisation of a photoelectrode followed shortly thereafter, using similar chemistry (Moser, 1887). However, the clear recognition of the parallels between the two procedures— realisation that the same dyes can in principle function in both systems (Namba and Hishiki, 1965) and verification that their operating mechanism is by injection of electrons from photoexcited dye molecules into the conduction band of the n-type semiconductor substrates (Gerischer and Tributsch, 1968)—dates only to the 1960s. In the years that followed it became recognised that the dye could function most efficiently if chemisorbed on the surface of the semiconductor (Tsubomura et al., 1976; Anderson et al., 1979; Dare-Edwards et al., 1980). The concept of using dispersed particles to provide a sufficient interface area then emerged (Dung et al., 1984), and was subsequently employed for photoelectrodes (Desilvestro et al., 1985).

Titanium dioxide (TiO2) quickly became the semiconductor of choice for the photoelectrode on account of its many advantages for sensitised photochemistry and photoelectron chemistry: it is a low-cost, widely available, non-toxic and bio-

Dye-Sensitised Mesoscopic Solar Cells

505

compatible material, and as such is even used in healthcare products as well as industrial applications such as paint pigmentation. Initial studies of the TiO2-based DSSC employed tris-bipyridyl ruthenium(II) dyes, which are paradigm sensitisers for photochemical studies, functionalised by the addition of carboxylate groups to attach the chromophore to the oxide substrate by chemisorption. Progress thereafter was incremental, a synergy of structure, substrate roughness and morphology, dye photophysics and electrolyte redox chemistry, until the announcement in 1991 (O’Regan and Grätzel, 1991) of a sensitised electrochemical photovoltaic device with an energy conversion efficiency of 7.1% under solar illumination. That evolution has continued progressively since then, with certified efficiencies now over 11%. 8.3 Mode of function of dye-sensitised solar cells

8.3.1 Device configuration

Figure 8.1 is a schematic of the components of a DSSC. At the heart of the system is a mesoporous oxide layer composed of nanometre-sized particles that have been sintered together to allow electronic conduction to take place. The material of choice has been TiO2 (anatase) although alternative wide-gap oxides such as ZnO, SnO2 and Nb2O5 have also been investigated (Hoyer and Weller, 1995; Ferrere et al., 1997; Rensmo et al., 1997; Sayama et al., 1998; Green et al., 2005). Attached to the surface of the nanocrystalline film is a monolayer of a sensitiser dye. Photoexcitation of the latter results in the injection of an electron into the conduction band of the oxide, generating the dye cation. The original state of the dye is subsequently restored by electron donation from the electrolyte; this step is often referred to as the regeneration reaction. The electrolyte usually comprises an iodide/triiodide redox couple dissolved in a liquid organic solvent, although attention is increasingly focusing on alternatives for the solvent, including ionic liquids, gelled electrolytes and polymer electrolytes (Stathatos et al., 2003; Wang et al., 2003b; Nogueira et al., 2004; Durr et al., 2005). The regeneration of the sensitiser by iodide intercepts the recapture of the injected electron by the oxidised dye. The iodide is in turn regenerated by the reduction of triiodide at the counter electrode, with the electrical circuit being completed via electron migration through the external load.

The high surface area of the mesoporous metal oxide film is critical to efficient device performance as it allows strong absorption of solar irradiation to be achieved by only a monolayer of adsorbed sensitiser dye. Whereas a dye monolayer absorbed on a flat interface exhibits only negligible light absorption (the optical absorption


506

cross-sectional areas for molecular dyes being typically 2–3 orders of magnitude smaller than their physical cross-sections), the use of a mesoporous film dramatically enhances the interfacial surface area over the geometric surface area, by up to a 1000-fold for a 10 µm thick film, leading to high visible-light absorbance from the many successive monolayers of adsorbed dye in the optical path. Another advantage of the use of a dye monolayer is that there is no requirement for exciton diffusion to the dye/metal oxide interface, and also the non-radiative quenching of excited states often associated with thicker molecular films is avoided. The high surface area of such mesoporous films does however have a significant downside, as it also enhances interfacial charge-recombination losses, a topic we return to in more detail below.

A recent alternative embodiment of the DSSC concept is the replacement of the redox electrolyte with a solid-state hole conductor, which may be either inorganic (Tennakone et al., 1995; O’Regan and Schwartz, 1998) or organic (Bach et al., 1998), thereby avoiding the use of a redox electrolyte. Such solid-state sensitised hetero-junctions can be regarded as functionally intermediate between redox electrolyte-based photoelectrochemical DSSCs and the organic bulk heterojunctions described in

Figure 8.1 Schematic of a liquid electrolyte dye-sensitised solar cell. Photoexcitation of the sensitiser dye is followed by electron injection into the conduction band of the mesoporous oxide semiconductor, and electron transport through the metal oxide film to the TCO-coated glass working electrode. The dye molecule is regenerated by the redox system, which is itself regenerated at the platinised counter electrode by electrons passed through the external circuit.

light

external circuit

electrons

I–/I3

– based

electrolyte

TCO-coated

glass

sensitiser

dye

I3

I-

platinised TCO-

coated glass

nanocrystalline

TiO2 film


507

Chapter 7. Devices efficiencies for such solid-state DSSCs are as yet limited to ~4% (Snaith et al., 2005), in contrast to efficiencies of over 11% achieved for the more widely studied redox electrolyte-based DSSCs (Chiba et al., 2006). 8.3.2 Device fabrication

DSSCs are typically fabricated on transparent conducting oxide (TCO) glass substrates, enabling light irradiance through this substrate under photovoltaic operation. The conductive coating typically used is fluorine-doped SnO2 (FTO), preferred over its indium-doped analogue (ITO) for reasons of lower cost and enhanced stability. Prior to deposition of the mesoporous TiO2 film, a dense TiO2 film may be deposited to act as a hole blocking layer, preventing recombination (shunt resistance) losses between electrons in the FTO and oxidised redox couple.

The TiO2 nanoparticles are typically fabricated by the aqueous hydrolysis of titanium alkoxide precursors, followed by autoclaving at temperatures up to 240 C to achieved the desired nanoparticle dimensions and crystallinity (anatase) (Barbe et al., 1997). The nanoparticles are deposited as a colloidal suspension by screen printing or by spreading with a doctor blade, followed by sintering at ~450 C to ensure good interparticle connectivity. The film porosity is controlled by the addition of an organic filler such as carbowax to the suspension prior to deposition; this filler is subsequently burnt off during the sintering step. Figure 8.2 shows a scanning electron micrograph of a typical mesoporous TiO2 film. Typical film thicknesses are 5–20 µm, with TiO2 mass 1–4 mg cm –2, film porosity 50–65%, average pore size 15 nm and particle diameter 15–20 nm.

Figure 8.2 Scanning electron micrograph of a typical mesoscopic TiO2 film employed in DSSCs. Note the bipyramidal shape of the particles, with (101) oriented facets exposed. The average particle size is 20 nm.


508

The classic sensitiser dye employed in DSSCs is a ruthenium (II) bipyridyl dye, cis-bis(isothiocyanato)-bis(2,2′-bipyridyl-4,4′-dicarboxylato)-ruthenium(II), which is usually referred to as ‘N3’, or in its partially deprotonated form (a di-tetra butylammonium salt) as ‘N719’ (Nazeeruddin et al., 1993). Figure 8.3 shows the structure of this dye. The incorporation of carboxylate groups allows ligation to the film surface via the formation of bidendate and ester linkages, whilst the (–NCS) groups enhance the absorption of visible light. Adsorption of the dye to the mesoporous film is achieved by simple immersion of the TiO2 film in a solution of dye, which results in the conformal adsorption of a dye monolayer to the film surface. The counter electrode is fabricated from FTO-coated glass, with the addition of a Pt catalyst to catalyse the reduction of the redox electrolyte at this electrode. Electrical contact between working and counter electrodes is achieved by the redox electrolyte, with capillary forces being sufficient to ensure the electrolyte efficiently penetrates the film pores. 8.3.3 Energetics of operation

Figure 8.4 illustrates the energetics of DSSC operation. In contrast to silicon and organic devices, the high concentration (~0.5 M) of mobile ions in the electrolyte effectively screens out any macroscopic electric fields. Charge separation is therefore primarily driven by the inherent energetics (oxidation/reduction potentials) of the different species at the TiO2/dye/electrolyte interface rather than by the presence of macroscopic electrostatic potential energy gradients. Similarly, charge-transport

N

N

COOH

COO-

N

N

COOH

COO-

Ru

N

SC

NC

S

Figure 8.3 Chemical structure of the N719 ruthenium complex used as a charge-transfer sensitiser in dye-sensitised solar cells. The N3 dye has the same structure, except that all four carboxylates are protonated.


509

processes are primarily diffusive, driven by concentration gradients in the device generated by the photoinduced charge separation. Photoinduced charge separation occurs at the dye-sensitised TiO2/electrolyte interface. Electron injection requires the dye excited state to be more reducing than the TiO2 conduction band, enabling transfer of an electron from photoexcited dye into the metal oxide. This energetic requirement is equivalent to the dye excited-state LUMO orbital having a lower electron affinity than the electrode conduction-band edge. Similarly, regeneration of the dye ground state by the redox couple requires the dye cation to be more oxidising than the iodide/triiodide redox couple. Charge separation in DSSCs can be regarded as a two-step redox cascade, resulting in the injection of electrons into the TiO2 electrode and the concomitant oxidation of the redox electrolyte. Figure 8.4 shows typical values for the interfacial energetics in DSSCs. Both charge-separation reactions are thermodynamically downhill and can be achieved with near unity quantum efficiency in efficient devices.

Following charge separation, charge collection from the device requires transport of the photogenerated charges to their respective electrodes. For an efficient DSSC under AM1.5 solar irradiance, these charge fluxes are of the order of 20 mA cm–2. The high ionic concentrations in the device effectively screen out any macroscopic

Red /Ox

maximum

voltage

S / S+

S*/ S+

diffusion

electrolyte

sensitiser

dye TiO2 conducting

glass U / V

vs. NHE

1.0

0.5

0

–0.5

e–

regeneration

injection

EF

CB

cathode

e–

hν

Figure 8.4 Energetics of operation of DSSCs. The primary free energy losses are associated with electron injection from the excited sensitiser into the TiO2 conduction band and regeneration of the dye by the redox couple. The voltage output of the device is approximately given by the splitting between the TiO2 Fermi level (dashed line) and the chemical potential of the redox electrolyte.


510

electric fields, thereby removing any significant drift component of these transport processes. The transport of both electrons and redox ions are therefore both primarily driven by diffusion processes resulting from concentration gradients. Under optimum conditions (e.g. good TiO2 nanoparticle interconnections and a low-viscosity electrolyte), these charge-transport processes (electrons towards the FTO working electrode, triiodide towards the counter electrode) can be efficiently driven with only modest concentration gradients, and therefore only small free energy losses (<50 meV). At the counter electrode, the triiodide is re-reduced back to iodide, the platinum catalyst enabling this reaction to proceed with minimal overpotential (again <50 meV). A similarly ohmic contact is achieved at the TiO2/FTO interface. It is apparent that the energetics at the TiO2/dye/electrolyte interface are of primary importance in determining the overall device energetics.

Power output from the DSSC requires not only efficient charge collection by the electrodes but the generation of a photovoltage, corresponding to a free energy difference between the working and counter electrodes. In the dark at equilibrium, the Fermi energy of the TiO2 electrode (corresponding to the free energy of electrons in this film after thermalisation) equilibrates with the mid-point potential of the redox couple, resulting in zero output voltage. Under these conditions, the TiO2 Fermi level lies deep within the bandgap of the semiconductor, and the film is effectively insulating, with a negligible electron density in the TiO2 conduction band. Photo-excitation results in electron injection into the TiO2 conduction band and concomitant hole injection into (oxidation of) the redox electrolyte. The high concentrations of oxidised and reduced redox couple present in the electrolyte in the dark mean that this photooxidation process does not result in a significant change in chemical potential of the electrolyte, which remains effectively fixed at its dark, resting value. In contrast, electron injection into the TiO2 conduction band results a dramatic increase in electron density (from the order of 1013 cm–3 to 1018 cm–3), raising the TiO2 Fermi level towards the conduction-band edge, as illustrated in Fig. 8.4, and allowing the film to become conducting. This shift of the TiO2 Fermi level under irradiation increases the free energy of injected electrons and is responsible for the generation of the photovoltage in the external circuit.

The mid-point potential of the redox couple is given by the Nernst equation, and is therefore dependent on the relative concentrations of iodide and iodine. The concentrations of these species required for efficient device function are in turn constrained by kinetic requirements of dye regeneration at the working electrode, and iodide regeneration at the counter electrode, as discussed below. Typical concen-trations of these species are in the range 0.1–0.7 M iodide and 10–200 mM iodine, constraining the mid-point potential of this electrolyte to ~0.4 V vs. NHE. It should


511

furthermore be noted that in the presence of excess iodide, the iodine is primarily present in the form I3

–, resulting in this electrolyte often being referred to as an iodide/triiodide redox couple.

Determining the energetics of the TiO2 conduction band is more complex. As with most oxides, the surface of TiO2 may be more or less protonated depending on the pH of the surrounding medium. The resultant changes in surface charge cause the surface potential to exhibit a nernstian dependence on effective pH, shifting by 60 mV per pH unit (Rothenberger et al., 1992). In bulk metal oxides, surface charge can result in significant bending of the conduction and valence bands adjacent to the surface. However, in the mesoporous TiO2 films employed in DSSCs, the nanoparticles are too small to support significant band bending. As a consequence, the whole conduction band of such mesoporous films shifts with the surface potential. At pH 1 in aqueous solutions, the conduction band edge for TiO2 has been reported at –0.5 V vs. NHE, shifting negatively as the pH is increased (Hengerer et al., 2000). DSSCs typically employ organic rather than aqueous electrolytes, complicating quantification of the effective pH. Nevertheless studies in organic solvents have demonstrated shifts of the conduction band of mesoporous TiO2 films of up to 1 V depending on the concentration of potential-determining ions (primarily small cations such as protons or lithium cations) in the electrolyte (Redmond and Fitzmaurice, 1993). For this reason, the concentration of such potential-determining ions in the electrolyte plays a key role in determining the energetics of the dye-sensitised interface, and thereby device performance. Additives added to the electrolyte to determine such energetics include Li+, guanadinium ions, N-methylbenzimidazol and t-butyl pyridine (which functions as a base). Further influence on these interfacial energetics can be achieved by variation of the extent of protonation of the sensitiser dye.

The choice of suitable sensitiser dye energetics is essential to achieve suitable matching to the metal oxide and redox couple. The excited-state oxidation potential (Um(S+/S*)) must be sufficiently negative to achieve efficient electron injection into the TiO2 conduction band, whilst the ground-state oxidation potential must be sufficiently positive to oxidise the redox couple. The redox properties of adsorbed sensitiser dyes may differ significantly from those measured in solution, mainly due to the high surface-charge densities and dipoles, present at this interface (Zaban et al., 1998). Notwithstanding this, typical dye redox energies empirically found to be compatible with efficient device function are Um (S

+/S) > 0.6 V vs. NHE.


512

8.3.4 Kinetics of operation

Figure 8.5 is a photochemical view of the function of a DSSC, illustrating the sequence of electron transfer and charge-transport processes which result in photo-voltaic device function. In addition to the forward electron transfer and transport processes, this figure also illustrates several competing loss pathways, shown as blue arrows. These loss pathways include decay of the dye excited state to ground, and charge recombination of injected electrons with dye cations and with the redox couple. Each charge-transfer step results in an increased spatial separation of electrons and holes, increasing the lifetime of the charge-separated state, but at the expense of reducing the free energy stored in this state. This functionality exhibits a close parallel to function of photosynthetic reaction centres. As in natural photosynthesis, kinetic competitions between the various forward and loss pathways are critical to determining the quantum efficiencies of charge separation and collection, and are therefore key factors determining energy conversion efficiency.

The theory of interfacial electron transfer, and the dependence of electron-transfer rate on interfacial energetics, are addressed in Chapter 4 of this book, and the

G/eV

0

1

Dye* electron injection

ps

decay to ground

~ns

recombination to

electrolyte, ms – s

hν

recombination to dye,

µs – ms

regeneration reaction

~1 µs

charge transport to

electrodes, ms – s

I3– / e

–TiO2

dye+ / e

–TiO2

Dye

Figure 8.5 State diagram representation of the kinetics of DSSC function. Forward processes of light absorption, electron injection, dye regeneration and charge transport are indicated by blue arrows. The competing loss pathways of excited-state decay to ground and electron recombination with dye cations and oxidised redox couple are shown in grey. The vertical scale corresponds to the free energy G stored in the charge-separated states. Note the free energy of injected electrons is determined by the Fermi level of the TiO2; the figure is drawn assuming a TiO2 Fermi level 0.6 V above the chemical potential of the redox electrolyte (corresponding to a cell voltage of ~0.6 V, close to the maximum power point of typical DSSCs).


513

dynamics of charge transport in mesoscopic electrodes are addressed in Chapter 11. In this chapter we will therefore confine ourselves only to consideration of the impact of these dynamics on the performance of DSSCs.

Efficient electron injection requires the rate of electron injection to exceed excited-state decay to ground. Typical rates of dye excited-state decay to ground are in the range 107–1010 s–1. The rate of electron injection depends on the electronic coupling between the dye excited-state LUMO orbital and accepting states in the TiO2, and on the relative energetics of these states. In model system studies of dye-sensitised metal oxide films, electron injection rates of >1012 s–1 have been reported for a range of sensitiser dyes, consistent with efficient electron injection (Anderson and Lian, 2005; Durrant et al., 2006). However it should be noted that fast electron injection dynamics require both strong electronic coupling of the dye LUMO orbital to the metal oxide conduction-band states, and a sufficient free energy difference to drive the reaction. As such, electron-injection dynamics are dependent on the energetics of the TiO2 conduction band, and therefore on the concentration of potential-determining ions (e.g. Li+) in the electrolyte. Omission of such ions from the electrolyte can result in an insufficient energetic driving force, reducing the quantum yield of charge injection, and thereby reducing device photocurrent.

It should be noted that the heavy metal ion in the ruthenium bipyridyl dyes widely employed in DSSCs results in ultrafast (up to 1013 s–1) intersystem crossing from the initially formed singlet excited state to a triplet state. This relaxation process reduces the excited-state free energy by ~400 meV. While several studies have reported sub-picosecond electron injection from the singlet excited state of these dyes, electron injection from the triplet state is significantly slower (1010–1011 s–1), consistent with the lower free energy of this excited state, but still fast compared with decay of the triplet state to ground (107–108 s–1) (Anderson and Lian, 2005; Durrant et al., 2006). Recent studies of complete devices have suggested that this relatively slow electron injection from the dye triplet state may be the dominating injection pathway in efficient DSSCs (Haque et al., 2005).

Efficient dye regeneration requires the rate of re-reduction of the dye cation by the redox couple to exceed that of charge recombination of injected electrons with these dye cations. This recombination reaction has been shown to be strongly dependent on the electron density in the TiO2 electrode (and therefore light intensity and cell voltage), accelerating by at least an order of magnitude between short-circuit and open-circuit conditions (Durrant et al., 2006). It is furthermore dependent on the spatial separation of the dye cation (HOMO) orbital from the metal oxide surface, with the rate constant decaying exponentially with distance, consistent with electron tunneling theory (Chapter 4). The regeneration reaction is dependent on the iodide


514

concentration, electrolyte viscosity and dye structure. For the N719 sensitiser dye, and employing a low-viscosity electrolyte such as acetonitrile, the regeneration reaction has a half-time of ~1 µs, sufficiently fast to compete effectively with the competing recombination reaction and ensuring that the regeneration reaction can be achieved with unity quantum efficiency (Green et al., 2005).

Efficient charge collection by the external circuit requires the time constant for electron transport within the TiO2 matrix to be faster than charge recombination of injected electrons with the redox couple. Electron transport is a diffusive process, strongly influenced by electron trapping in localised sub-bandgap states, resulting in the dynamics being strongly dependent on position of the TiO2 electron Fermi level: raising the Fermi level towards the conduction-band edge resulting in increased trap filling. Typical electron-transport times under solar irradiation are of the order of milliseconds (Peter and Wijayantha, 2000; Frank et al., 2004). If a low-viscosity solvent such as acetonitrile is used, transport of the oxidised redox couple to the counter electrode is not rate-limiting. However, the use of higher viscosity solvents more suitable for practical device applications (for reasons of device stability) can result in significantly lower ionic diffusion constants, with the resultant iodide/ triiodide concentration gradients causing significant free energy (series resistance) losses, and also potentially accelerating interfacial charge recombination.

Given the relatively slow timescale for charge transport in DSSCs compared with most other photovoltaic devices, and the extensive interfacial area available for charge recombination in the device (due to its mesoscopic structure), it is remarkable that the quantum efficiency of charge collection can approach unity. The key factor enabling this high efficiency is the slow rate constant for the interfacial charge recombination of injected electrons with the oxidised redox couple. This reaction is a multi-electron reaction, most simply being described by the equation

3

I 2 e 3 I− − −

+ → (8.1)

which must therefore proceed via one or more intermediates states. The mechanism of this reaction has been extensively studied, and while the details remain somewhat controversial, it is apparent that without a suitable catalyst such as platinum, one or more of these intermediates steps exhibits a significant activation barrier, resulting in a slow overall rate constant for this reaction. The low rate constant for this recombination reaction on TiO2, contrasting to the facile electrochemistry of this redox couple on the platinised counter electrode, is a key factor behind the remarkable efficiencies achieved to date for DSSCs. Nevertheless, eq. 8.1 is the primary recomb-ination pathway in DSSCs. The flux of this recombination pathway increases with


515

increasing electron density in the TiO2 electrode (and therefore with the TiO2 Fermi level or cell voltage). In the dark it responsible for the diode-like leakage current observed in current–voltage scans, while under illumination it is the primary factor limiting the voltage output of the device.

The kinetic competition between charge transport and recombination in DSSCs has been analysed in terms of an effective carrier diffusion length Ln, given by

effnL D τ= (8.2)

where Deff is the effective electron diffusion length, and τ the electron lifetime due to the charge-recombination reaction given by eq. 8.1 (Peter and Wijayantha, 2000). Deff increases with light intensity (due to the increased electron density in the TiO2 film) whilst τ shows a proportional decrease, resulting in Ln being largely independent of light intensity. Typical values for Ln are 5–20 µm, and even 100 µm near the optimum power point for cells with >10% conversion efficiency, consistent with the high carrier-collection efficiencies observed in efficient DSSCs.

It is important to emphasis that the energetics and kinetics of DSSC function are not independent considerations. The kinetics of the interfacial electron-transfer dynamics depend strongly on the energetics of the TiO2/dye/electrolyte interface and on the density of electrons in the TiO2 (i.e. the TiO2 Fermi level). Raising the energy of the TiO2 conduction band reduces recombination losses (as for a given TiO2 Fermi level, the electron density in the TiO2 film will be lower), and therefore may give a high cell output voltage, but at the expense of a lower free energy driving force for charge separation, which may result in a lower quantum efficiency for charge generation and therefore a lower output current. In practice, modulation of these energetics and kinetics to achieve optimum device performance remains one of the key challenges in DSSC research and development.

8.4 DSSC research and development

8.4.1 Panchromatic sensitisers

The ideal sensitiser for a single-junction photovoltaic cell converting standard global AM1.5 sunlight to electricity should absorb all light below a threshold wavelength of about 920 nm. In addition, it must also carry attachment groups such as carboxylate or phosphonate to firmly graft it to the semiconductor oxide surface. On excitation, it should inject electrons into the solid with a quantum yield of unity. The energy level


516

of the excited state should be well matched to the lower bound of the conduction band of the oxide to minimise energetic losses during the electron-transfer reaction. Its redox potential should be sufficiently high that it can be regenerated via electron donation from the redox electrolyte or the hole conductor. Finally, it should be stable enough to sustain about 100 million turnover cycles, corresponding to about twenty years of exposure to natural light. A single-junction device with such a sensitiser could reach a maximum conversion efficiency of 32% in global AM 1.5 sunlight.

Much of the research in DSSC dye chemistry is devoted to the identification and synthesis of sensitisers matching these requirements, while retaining stability in the photoelectrochemical environment. The attachment group of the dye ensures that it spontaneously assembles as a molecular layer on exposing the oxide film to a dye solution. This molecular dispersion ensures a high probability that, once a photon is absorbed, the excited state of the dye molecule will relax by electron injection into the semiconductor conduction band. However, the optical absorption of a single monolayer of dye is weak, a fact which originally was cited as ruling out the possibility of high-efficiency sensitised devices, as it was assumed that smooth substrate surfaces would be imperative in order to avoid the recombination loss mechanism associated with rough or polycrystalline structures in solid-state photovoltaics. This objection was invalidated by recognising that the injection process places electrons in the semiconductor lattice, spatially separated from the positive charge carriers by the dye molecules, which are insulating in the ground state and hence provide a barrier for charge recombination. By now, the use of nanocrystalline thin-film structures with a roughness factor of >1000 has become standard practice.

The best photovoltaic performance in terms of both conversion yield and long-term stability has so far been achieved with polypyridyl complexes of ruthenium and osmium. Sensitisers having the general structure ML2(X)2 where L stands for 2,2′-bipyridyl-4,4′-dicarboxylic acid, M is Ru or Os, and X represents a halide, cyanide, thiocyanate, acetyl acetonate, thiocarbamate or water substituent, are particularly promising. Thus, the ruthenium complex cis-RuL2(NCS)2, known as N3 dye, has become the paradigm heterogeneous charge-transfer sensitiser for mesoporous solar cells (Nazeeruddin et al., 1993). The absorption spectrum of fully protonated N3 has maxima at 518 and 380 nm, with extinction coefficients of 1.3 × 104 M–1 cm–1 and 1.33 × 104 M–1 cm–1, respectively. The complex emits at 750 nm, the excited-state lifetime being 60 ns. The optical transition has MLCT (metal-to-ligand charge-transfer) character: excitation of the dye involves transfer of an electron from the metal to the π* orbital of the surface-anchoring carboxylated bipyridyl ligand, from where it is released in a timescale of femtoseconds to picoseconds into the conduction band of TiO2, generating electric charges with unit quantum yield.


517

Discovered in 1993, the photovoltaic performance of N3 has been unmatched for eight years by virtually hundreds of other complexes that have been synthesised and tested. However, in 2001 the ‘black dye’ tri(cyanato)-2,2′2′′-terpyridyl-4,4′4′′-tricarboxylate)Ru(II) achieved 10.4% AM1.5 solar-to-power conversion efficiency in full sunlight (Nazeeruddin et al., 2001). Conversion efficiencies have meanwhile been improved further, the current record validated by an accredited laboratory being 11.1% (Chiba et al., 2006).

Figure 8.6 compares the spectral response of the photocurrent observed with the N3 dye and the black dye sensitisers. The incident photon-to-current conversion efficiency (IPCE) of the DSSC is plotted as a function of excitation wavelength. Both chromophores show very high IPCE values in the visible range. However, the response of the black dye extends 100 nm further into the IR than that of N3. The photocurrent onset is close to 920 nm, near the optimal threshold for single-junction converters. From there the IPCE rises gradually until at 700 nm it reaches a plateau of ~80%. If one accounts for reflection and absorption losses in the conducting glass, the conversion of incident photons to electric current is practically quantitative over the whole visible domain. From the overlap integral of the black dye curves in Fig. 8.6 with the AM 1.5 solar spectrum, the short-circuit photocurrents of the N3 and black dye-sensitised cells are predicted to be 16 mA cm–2 and 20.5 mA cm–2 respectively, in agreement with experimental observations. Employing antireflective coatings and optimising light-scatter effects to capture more photons in the near-IR regions can increase the photocurrents to a maximum of 18 mA cm–2 and 22 mA cm–2 for the two

Figure 8.6 Spectral response of the photocurrent for mesoscopic injection solar cells. Grey curve: bare TiO2 film, blue and black curves: same film sensitised by the N3 and black ruthenium dyes, respectively.


518

dyes under standard reporting conditions (SRC). Any higher values point to a spectral mismatch between the output of the simulator and AM1.5 sunlight. The overall maximum-power efficiency (ηmp) of the cell is calculated from the integral short-circuit photocurrent density (isc), the open-circuit photovoltage (Voc), the fill factor (ηfill) and the incident solar irradiance ( o

sE = 1000 W m–2) as

η mp = isc × Voc × ηfill /os

E (8.3)

8.4.2 Opportunities for performance improvement

The solar-to-electric power conversion efficiency of DSSC laboratory cells, validated by an accredited PV calibration laboratory, has reached 11.1% under standard reporting conditions, i.e. air mass 1.5 global sunlight at 1000 W m–2 intensity and 298 K temperature (Chiba et al., 2006), rendering the DSSC a credible alternative to conventional p-n junction photovoltaic devices. Figure 8.7 shows the photovoltaic performance data of such a champion cell. A couple of years ago, solid-state equivalents using organic hole conductors reached an efficiency of 4.2% (Schmidt-Mende et al., 2005b), and recently Snaith et al. (2007) have reported an efficiency of 5.1%, whereas nanocomposite films comprising only inorganic materials, such as TiO2 and CuInS2, have achieved efficiencies between 5% and 6% in the ETA cells discussed by Könenkamp in Chapter 6 (see also Fig. 1.2 and Nanu et al., 2005).

Figure 8.7 Photovoltaic performance of a state-of-the-art DSSC laboratory cell: i–V curve measured under AM 1.5 standard test conditions.

20

15

10

5

0

800 600 400 200 0

potential / mV

isc = 17.73 mA cm–2

Voc = 846 mV

FF = 0.745

ηmp = 11.18%

i / m

A c

m–2


519

The most important further gains to be expected in the near term will be in the short-circuit photocurrent. New ruthenium complexes showing increased optical cross-sections and capable of absorbing longer wavelengths are currently under development (Wang et al., 2005a). For example, Fig. 8.8 shows the structure of a new dye coded K–19, which exhibits enhanced absorption by virtue of its π-system.

Figure 8.8 Molecular structure of the K–19 dye.

Judicious molecular engineering of the ruthenium dye structure will allow further increase of light harvesting in the 700–900 nm region. Ruthenium complexes of quaterpyridyl derivatives show great promise in this respect. The goal is to obtain a DSSC having optical features similar to those of GaAs. A nearly vertical rise of the photocurrent close to the 920 nm absorption threshold would increase the short-circuit photocurrent from currently 20.5 mA cm–2 to about 28 mA cm–2, raising the overall efficiency to over the 15% mark without changing the currently used redox system. Ultimately the combination of two sensitisers, one being the red or black ruthenium complex and the other an organic dye showing strong absorption in the near-infrared region, may be required to achieve this desired spectral response. Such dye ‘cocktails’ are presently under intense investigation and the first results look promising (Reddy et

al., 2007). A road map to achieve this goal by the year 2009 has been elaborated and will serve to coordinate synthetic efforts of several groups on the international scale.

8.4.3 Tandem cells

A feature that makes the DSSC particularly attractive for tandem cell application is that its optical transmission and short-circuit photocurrent can be readily adjusted by changing the film thickness, pore size, the nature of the dye and the dye loading. This, along with the ease of forming layered structures, for example by producing the


520

mesoscopic oxide films using screen printing or doctor blading methods, renders the DSSC particularly well suited for the fabrication of tandem solar cell structures that capture the solar emission in an optimal fashion. Note that for a two-level tandem cell a conversion efficiency of up to 47% can be reached, the optimal bandgaps for the top and bottom cells being around 1.65 eV and 1 eV respectively.

Several publications have dealt with the use of stacked DSSC configurations in which two dyes absorbing different parts of the solar spectrum serve as sensitisers (Durr et al., 2004; Kubo et al., 2004). We recently demonstrated that a tandem device comprising a DSSC as a top cell for high-energy photons and a copper indium gallium selenide (CIGS) thin-film bottom cell, capturing the red and near-IR solar emission respectively, produces AM1.5 solar-to-electric conversion efficiencies greater than 15% (Liska et al., 2006).

Figure 8.9 shows the i–V curve and other characteristics obtained under AM 1.5 insolation for one of the first embodiments of this two-terminal tandem device. The performance of the tandem was clearly superior to that of the individual cells, despite the fact that the short-circuit currents of the two cells were not perfectly matched, as evidenced by the shoulder in the i–V curve. Likewise, no effort was made to minimise the optical losses. This leaves no doubt that further rapid efficiency gains reaching well beyond the 20% mark can be expected from the fructuous marriage of these two thin-film PV technologies (Liska et al., 2006). Combining a relatively low-cost thin-film CIGS substrate cell with a DSSC superstrate cell may be a cheaper method of achieving efficiencies above 15% than use of a high-efficiency CIGS cell alone.

Figure 8.9 Photocurrent density–voltage characteristics under AM 1.5 full sunlight (1000 mW cm–2) for a two-terminal tandem DSSC/CIGS cell. The DSSC top cell produced isc = 13.66 mA cm–2; Voc = 0.8 V; FF = 0.75 and η = 8.2%, while the CIGS bottom cell produced isc = 14.3 mA cm–2; Voc = 0.65 V and FF = 0.77 and η = 7.28%. The DSSC/CIGS tandem cell produced isc = 14.05 mA cm–2; Voc = 1.45 V; FF = 0.74 and η = 15.09%. The reverse-bias characteristics are shown as a black curve. Adapted from Liska et al. (2006).

0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

voltage / V

i / m

A c

m–2

16

14

12

10

8

6

4

2

0

DSSC/CIGS ηmp = 15.09%


521

8.4.4 Stability

Unlike amorphous silicon, which suffers from degradation due to the well-known Staebler–Wronski effect, the intrinsic stability of the DSSC has been confirmed by extensive accelerated light-soaking tests carried out over the last decade. One major issue that has been settled during this period is that the sensitisers employed in the current DSSC embodiments can sustain twenty years of outdoor service without significant degradation. However, as new and more advanced dye structures emerge, and in order to avoid repeating these lengthy tests every time the sensitiser is modified, kinetic criteria have been developed to allow prediction of long-term sensitiser performance.

s

excitation

injection

regeneration

product

decomposition

s+ s* k1

kinj

kreg

k2

hν

product

decomposition

Figure 8.10 The catalytic cycle of the sensitiser during DSSC operation.

Figure 8.10 illustrates the catalytic cycle that the sensitiser undergoes during cell operation. Critical for stability are side reactions that occur from the excited state (S*) or the oxidised state of the dye (S+), which would compete with electron injection from the excited dye into the conduction band of the mesoscopic oxide and with the regeneration of the sensitiser. These destructive channels are assumed to follow first or pseudo-first order kinetics and are assigned the rate constants k1 and k2. Introducing the two branching ratios P1 = kinj/(k1 + kinj) and P2 = kreg/(k2 + kreg), where kinj and kreg are the first order or pseudo-first order rate constants for the injection and regeneration process, respectively, the fraction of the sensitiser molecules that survives one cycle is given by the product P1 × P2. A simple calculation (Grätzel, 2006) shows that the sum of the branching ratios for the two bleeding channels should not exceed 1 × 10–8 in order for the lifetime of the sensitiser to reach at least twenty years. The turnover frequency of the dye in the working DSSC, averaged diurnally and seasonally, is about 0.16 s–1.


522

For most of the common sensitisers, the rate constant for electron injection from the excited state of the dye to the conduction band of the TiO2 particles is in the femtosecond range. Assuming kinj = 1011–1013 s–1, any destructive side reaction should have k1 < 102 s–1. Ruthenium sensitisers of the N719 or K–19 type readily satisfy this condition as the decomposition from the excited-state level occurs at a much lower rate than the 102 s–1 limit. Precise kinetic information has also been gathered for the second destructive channel involving the oxidised state of the sensitiser, the key parameter being the ratio k2/kreg of the rate constants for the degradation of the oxidised form of the sensitiser and its regeneration. The S+ state of the sensitiser can readily be produced by chemical or electrochemical oxidation and its lifetime can be independently determined by absorption spectroscopy. A typical value of k2 is around 10–4 s–1 while the regeneration rate constant is at least in the 105 s–1 range. Hence the branching ratio is well below the limit of 10–8, which can be tolerated to achieve the 100 million turnovers and a 20-year lifetime for the sensitiser.

Many long-term tests which have been performed with the N3-type ruthenium complexes have confirmed the extraordinary stability of these charge-transfer sensitisers. For example, a European consortium supported by the Joule program (Hinsch et al., 2001) confirmed cell photocurrent stability during 8,500 hours of light soaking at 2.5 Suns, corresponding to about 56 million turnovers of the dye without any significant degradation. These results corroborate the projections from the kinetic considerations made above. A more difficult task has been to achieve stability under prolonged stress at higher temperatures, i.e. 80–85 C. The introduction of hydro-phobic sensitisers has been particularly rewarding in allowing the DSSC to meet, for the first time, the specifications laid out for outdoor applications of silicon photo-voltaic cells (Wang et al., 2003a). In addition these dyes show enhanced extinction coefficients due to the extension of the π conjugation onto one of the bipy ligands by styrene moieties. Taking advantage of these properties and using a novel robust electrolyte formulation, a ≥8% efficient DSSC has been realised that shows strikingly stable performance under both prolonged thermal stress and light soaking (Wang et

al., 2005b). While impressive progress has been made in the development of stable, non-

volatile electrolyte formulations, the conversion yields obtained with these systems are presently in the 7–10% range, below the 11.1% reached with volatile solvents. Future research efforts will be dedicated to bridging the performance gap between these systems. The focus will be on hole conductors and solvent-free electrolytes such as ionic liquids. The latter are a particularly attractive choice for the first commercial modules, owing to their high stability, negligible vapour pressure and excellent compatibility with the environment.


523

8.4.5 Organic dyes

When considering organic dyes for use in DSSCs, porphyrins and phthalocyanines have attracted particular attention, the former because of the analogy with natural photosynthetic processes, the latter because of their photochemical and phototherapeutic applications. However, porphyrins cannot compete with the N3 or black dye sensitiser due to their lack of red light and near-IR absorption. Phthalocyanines, on the other hand, show intense absorption bands in this spectral region. However, problems with aggregation and the unsuitable energetic position of the LUMO level, which is too low for electron transfer to the TiO2 conduction band, have turned out to be intractable for the moment.

More recently, the groups of Arakawa and Uchida have made remarkable advances in the development of organic dyes for use in DSSCs (Horiuchi et al., 2004; Hara et al., 2005). Using coumarin or polyene-type sensitisers, strikingly high solar-to-electric power conversion efficiencies, reaching up to 9.2% in full sunlight (Ito et

al., 2006), have been achieved in liquid-electrolyte cells. The rational design of such sensitisers based on quantum mechanical considerations has made great progress during the last few years, and replacement of ruthenium sensitisers by such molecularly engineered dyes is likely to occur in the near future. 8.4.6 Quantum dots as sensitisers

Semiconductor quantum dots are another attractive option for panchromatic sensitisers. These are II–VI and III–V type semiconductors particles whose size is small enough to produce quantum confinement effects of the type discussed by Nozik in Chapter 3. The absorption spectrum of such quantum dots can be adjusted by changing the particle size. Thus, the bandgap of materials such as InAs and PbS can be adapted to attain the value of 1.35 eV, which is ideal for a single-junction solar quantum converter. During the last decade a wealth of information has been gathered on the physical properties of quantum-dot materials and research is being pursued very actively. One problem with this approach is the photocorrosion of the quantum dots that will almost certainly happen if the junction contact is a liquid redox electrolyte. However quantum dots are expected to display higher stability in solid-state heterojunction devices (Plass et al., 2002). The advantage of Q-dots over conventional dyes as sensitisers is their very high extinction coefficients, which allows thinner films of the mesoporous oxide to be used. This should reduce the dark current, increasing Voc and the overall efficiency of the cell.


524

An exciting discovery made in recent years is that multiple exciton generation (MEG) can be obtained from the absorption of a single photon by a quantum dot if the photon energy is at least two times higher than its bandgap (Nozik, 2003; Schaller and Klimov, 2004). The challenge is now to find ways to collect the excitons before they recombine. As recombination occurs on a picosecond time scale, the use of mesoporous oxides as electron collectors presents a promising strategy, because the electron transfer from the quantum dot to the conduction band of the oxide electrode occurs within femtoseconds (Plass et al., 2002). As Nozik points out in Chapter 3, this opens up research avenues that ultimately may lead to photon converters reaching external quantum efficiencies values of several hundred percent. A calculation based on Henry’s model (Henry, 1980) shows that the maximum conversion efficiency of a single-junction cell could be increased from 34% to 44% by exploiting MEG effects. 8.4.7 Mesoporous oxide film development

When the dye-sensitised nanocrystalline solar cell was first presented, perhaps the most puzzling phenomenon was the highly efficient charge transport through the nanocrystalline TiO2 layer. Mesoporous electrodes differ greatly from their compact analogs because (i) the inherent conductivity of the film is very low; (ii) the small size of the nanocrystalline particles does not support a built-in electrical field; and (iii) the electrolyte penetrates the porous film all the way to the back contact, making the semiconductor/electrolyte interface essentially three-dimensional. The mechanism of charge transport in mesoporous systems is under keen debate today and several interpretations based on the Montrol–Scher model for random displacement of charge carriers in disordered solids have been advanced (Nelson and Chandler, 2004). However, the ‘effective’ electron diffusion coefficient is expected to depend on a number of factors such as trap filling and space-charge compensation by ionic motion in the electrolyte. The theoretical and experimental effort will continue as there is a need for further in-depth analysis of this intriguing charge-percolation process. The factors controlling the rate of charge-carrier percolation across the nanocrystalline film are presently under intense scrutiny. Intensity-modulated impedance spectro-scopy (see Section 12.6) has proved to be an elegant and powerful tool (Bisquert, 2002; Kubo et al., 2002; Cass et al., 2003; Frank et al., 2004; Cass et al., 2005; Fabregat-Santiago et al., 2005) to address these and other important questions related to the characteristic time constants for charge-carrier transport and reaction dynamics in dye-sensitised nanocrystalline solar cells.


525

On the material science side, future research will be directed towards synthesising structures with a higher degree of order than a random assembly of nanoparticles. A desirable morphology of the films would have the mesoporous channels or nanorods aligned in parallel to each other and vertically with respect to the TCO glass current collector. This would facilitate pore diffusion, give easier access to the film surface, avoid grain boundaries and allow the junction to be formed under better control. One approach to fabricate such oxide structures is based on surfactant template-assisted preparation of TiO2 nanotubes, as described by Jiu et al. (2005). These and the hybrid nanorod–polymer composite cells developed by Alivisatos and co-workers (Huynh et

al., 2002) have confirmed the superior photovoltaic performance of such films as compared with random-particle networks. 8.4.8 Molecular engineering of the interface

The high contact area of the junction in nanocrystalline solar cells renders mandatory the grasp and control of interfacial effects for future improvement of cell perform-ance. The nature of the exposed surface planes of the oxide and the mode of interaction with the dye is the first important information to gather. As far as the adsorption of the N3 dye on TiO2 is concerned, this is now well understood. The prevalent orientation of the anatase surface planes is (101) and the sensitiser is adsorbed through two of the four carboxylate groups, at least one of them being anchored via a bidentate configuration bridging two adjacent titanium sites (Nazeeruddin et al., 2000). Molecular dynamic calculations employing a classical force field have been carried out to predict the equilibrium geometry of the adsorbed sensitiser state (Burnside et al., 1998; Shklover et al., 1998). More sophisticated first-principle density functional calculations have also been undertaken (Vittadini et al., 1998) to model the surface interactions of TiO2 with simple adsorbates as well as the surface reconstruction effects resulting from the adsorption. The latter approach is particularly promising and will provide an important tool for future theoretical investigations.

Synthetic efforts focus on the molecular engineering of sensitisers that enhance charge separation at the oxide solution interface. The structural features of the dye should match the requirements for current rectification: by analogy to the photofield effect in transistors, the gate for unidirectional electron flow from the electrolyte through the junction and into the oxide is opened by the photoexcitation of the sensitiser. The reverse charge flow, i.e. recapture of the electron by the electrolyte, could be impaired by judicious design of the sensitiser. The latter should form a


526

tightly packed insulating monolayer blocking the dark current. The gain in open-circuit voltage can be calculated from the diode equation Voc = (nRT/F) ln[(isc/io) –1] (8.4) where n is the ideality factor, whose value is between 1 and 2 for DSSCs, and io is the reverse saturation current.. Thus for each order of magnitude decrease in the dark current, the gain in Voc would be 59 mV at room temperature. Work in this direction is indispensable to raise the efficiency of the DSSC significantly over the 15% limit with the currently employed redox electrolytes. 8.5 Solid-state dye-sensitised cells

Research on solid-state DSSCs has gained considerable momentum in recent years as this embodiment is attractive for realising flexible PV cells in a roll-to-roll production (www.konarka.com). The most successful p-type organic conductor employed to date is spiro-OMeTAD, which has a work function of ~4.9 eV and hole mobility of 2 × 10–4 cm2 s–1 (see Fig. 8.11). First reported in 1998, conversion efficiencies of solid-state cells incorporating this hole conductor have increased dramatically over the last few years, from a fraction of a percent (Bach et al., 1998) to over 4% (Schmidt-Mende et al., 2005a). The main drawback of these cells has been fast interfacial electron-hole recombination, reducing the electron diffusion length to a few microns (Kruger et al., 2003) as compared with 20–100 µm for electrolyte-based DSSCs. This restricts the film thickness employed in these cells to only 2 µm, which is insufficient for the adsorbed sensitiser to harvest all incident sunlight, thus reducing the photocurrent. The dye monolayer can block this back reaction to some extent because it is electrically insulating (Snaith et al., 2005). Hence current efforts are directed towards molecular engineering of the interface to improve the compactness and order of the monolayer and prevent charge carriers from recombining.

Another difficulty that has been encountered has been the filling of the porous network with the hole conductor. This impediment may be overcome by developing oxide films having regular mesoporous channels aligned perpendicular to the current collector. On the other hand, the Voc values obtained with solid-state DSSCs are high, reaching nearly 1 V, owing to a better match of the hole conductor work function than that of the electrolyte with the redox potential of the sensitiser. The future of these solid hole-conductor systems thus looks very bright if the recombination and pore filling problems can be solved.


527

Glass

Light

e -

h+

e -

TiO2

Gold

OMeTAD Dye SnO2

N N

OCH3

H3CO

OCH3

OCH3

NN

OCH3

OCH3

OCH3

H3CO

Figure 8.11 Cross-sectional view of a solid-state DSSC containing the hole conductor spiro-OMeTAD, the structure of which is indicated on the right (left drawing courtesy B. O’Regan).

8.6 Pilot production of modules, outdoor field tests and

commercial DSSC development

Figure 8.12 shows two prototypes of the monolithic Z-type interconnected DSSC modules, fabricated by Aisin Seiki in Japan, using carbon as a back contact to cut costs. Comparative field tests of these modules and polycrystalline silicon (pc-Si) have been running for several years; Figure 8.13 shows a photograph of the test station. The test results revealed the advantages of the DSSC as compared with silicon modules under realistic outdoor conditions: for equal rating under standard test conditions (STC), the DSSC modules produced 20–30% more energy than pc-Si modules (Toyoda, 2006). The superior performance of the DSSC can be ascribed to the following factors: • The DSSC efficiency is practically temperature-independent in the range 25–65 C

while that of monocrystalline and pc-Si declines by ~20% over the same range.

• Outdoor measurements indicate that light capture by the DSSC is less sensitive to the angle of incidence, although this needs to be further assessed.

• The DSSC is more efficient than pc-Si in diffuse light or cloudy conditions.

N N

OCH3

H3CO

OCH3

OCH3

NN

OCH3

OCH3

OCH3

H3CO


528

While it is up to the commercial supplier to set the final price for such modules, it is clear that the DSSC shares the cost advantages of all thin-film devices. In addition it uses no high-vacuum, cost-intensive steps and only cheap and readily available materials. Although it might be thought that the ruthenium-based sensitiser adds high material cost, its contribution to cost is <€ 0.01 Wp

–1 because of the small amount used. Also purely organic sensitisers can now achieve practically the same efficiency as Ru complexes. Given these advantages at comparable conversion efficiency, module costs below €1 Wp

–1 are realistic targets even for production plants having well below GW capacity. The DSSC has thus become a viable contender for large-scale solar energy conversion systems on the basis of cost, efficiency, stability and availability, as well as environmental compatibility. Building-integrated DSSC panels have been installed in the walls of the Toyota ‘Dream House’ shown in Fig. 8.14, and the British company G24i(www.G24i.com) has recently announced the building of the first 20 MW DSSC manufacturing plant in Wales.

Figure 8.12 Production of DSSC prototypes by Aisin Seiki in Japan. Note the monolithic design of the PV modules and the use of carbon as interconnect and counter electrode. The red dye is related to N–719 while the black dye has the structure RuL′(NCS)3 where L′ = 2, 2′,2′′- terpyridyl. 4,4′,4′′-tricarboxylic acid. The hole conductor is a non-volatile electrolyte.

AISINAISINAISINAISIN

Å| Å| Å| Å| Å| Å| Å| Å| Å| Å|

Glass with TCO

Damp-proof film

Counter-electrode

Electrolyte

TCO

Photo-electrode

with Dye

seal

Glass with TCO

Light Frame

Search of new module design

from industrial point of view< performance, durability,

number of parts, production time,

cost >

Prototype production


529

Figure 8.13 Outdoor field tests of DSSC modules in Kariya City, Japan at latitude 35°10′N. Azimuthal angle 0°, facing due south, tilt angle 30°. Note the pc-Si modules in the second row.

Figure 8.14 The Toyota ‘Dream House’ featuring DSSC panels made by Aisin Seiki (http://www.toyota.co.jp/jp/news/04/Dec/nt04_1204.html)

DSC

made byAISIN -SEIKI

The Toyota Dream HouseThe Toyota Dream HouseThe Toyota Dream HouseThe Toyota Dream House


530

8.7 Outlook

Using a principle derived from natural photosynthesis, mesoscopic injection solar cells, and in particular the DSSC, have become a credible alternative to solid-state p-n junction devices. Conversion efficiencies over 11% and 15% have already been obtained with single-junction and tandem liquid-electrolyte cells, respectively, on the laboratory scale, but there is ample room for further amelioration. Future research will focus on improving the short-circuit current by extending the light response of the sensitisers in the near-IR spectral region. Substantial gains in the open-circuit voltage are expected from introducing ordered oxide mesostructures and controlling the interfacial charge recombination by judicious engineering on the molecular level. Hybrid cells based on solid-state inorganic and organic hole conductors are an attractive option, in particular for the flexible DSSC embodiment. Nanostructured ETA cells using purely inorganic components, discussed by Könenkamp in Chapter 6, will also be developed.

Mesoscopic dye-sensitised cells are well suited for a whole realm of applications, ranging from the low-power market to large-scale applications. Their excellent performance in diffuse light gives them a competitive edge over silicon in providing electric power for stand-alone electronic equipment both indoors and outdoors. DSSCs are already being applied in building-integrated PV and this will become a fertile field of future commercial development.

As the quotation at our chapter head shows, at the 1912 IUPAC conference in Washington almost one hundred years ago the famous Italian photochemist Professor Giacomo Ciamician predicted that mankind will unravel the secrets of photosynthesis and apply the principles used by plants to harvest solar energy in glass buildings. His visionary thoughts appear now close to becoming a reality. Acknowledgements We gratefully acknowledge the insights and support from the many co-workers with whom we have had the pleasure of working on these devices, and particularly thank Dr Brian O’Regan for his help in preparing this manuscript.


531

References

Anderson N. A. and Lian T. Q. (2005), ‘Ultrafast electron transfer at the molecule-

semiconductor nanoparticle interface’, Annu. Rev. Phys. Chem. 56, 491–519. Anderson S., Constable E. C., Dare-Edwards M. P., Goodenough J. B., Hamnett A.

and Seddon K. R. (1979), ‘Chemical modification of a titanium(IV) oxide electrode to give stable dye sensitization without a supersensitizer’, Nature 280, 571–573.

Bach U., Lupo D., Comte P., Moser J. E., Weissortel F. and Salbeck J. (1998), ‘Solid-state dye-sensitized mesoporous TiO2 solar cells with high photon-to-electron conversion efficiencies’, Nature 395, 583–585.

Barbe C. J., Arendse F., Comte P., Jirousek M., Lenzmann F. and Shklover V. (1997), ‘Nanocrystalline titanium oxide electrodes for photovoltaic applications’, J. Am.

Ceramic Soc. 80, 3157–3171. Bisquert J. (2002), ‘Theory of the impedance of electron diffusion and recombination

in a thin layer’, J. Phys. Chem. B 106, 325–333. Burnside S. D., Shklover V., Barbe C., Comte P., Arendse F. and Brooks K. (1998),

‘Self-organization of TiO2 nanoparticles in thin films’, Chem. of Materials 10, 2419–2425.

Cass M. J., Qiu F. L., Walker A. B., Fisher A. C., and Peter L. M. (2003), ‘Influence of grain morphology on electron transport in dye sensitized nanocrystalline solar cells’, J. Phys. Chem. B 107, 113–119.

Cass M. J., Walker A. B., Martinez D., and Peter L. M. (2005), ‘Grain morphology and trapping effects on electron transport in dye-sensitized nanocrystalline solar cells’, J. Phys. Chem. B 109, 5100–5107.

Chiba Y., Islam A., Watanabe Y., Komiya R., Koide N. and Han L. Y. (2006), ‘Dye-sensitized solar cells with conversion efficiency of 11.1%’, Jap. J. Appl. Phys.

Part 2—Letters and Express Letters 45, L638–L640. Dare-Edwards M. P., Goodenough J. B., Hamnett A., Seddon K. R. and Wright R. D.

(1980), ‘Sensitization of semiconducting electrodes with ruthenium-based dyes’, Faraday Discussions, 285–298.

Desilvestro J., Grätzel M., Kavan L., Moser J. and Augustynski J. (1985), ‘Highly efficient sensitization of titanium dioxide’, J. Am. Chem. Soc. 107, 2988–2990.

Dung D. H., Serpone N. and Grätzel M. (1984), ‘Integrated systems for water cleav-age by visible-light–sensitization of TiO2 particles by surface derivatization with ruthenium complexes’, Helvetica Chim. Acta 67, 1012–1018.

Durr M., Bamedi A., Yasuda A. and Nelles G. (2004), ‘Tandem dye-sensitized solar cell for improved power conversion efficiencies’, Appl. Phys. Lett. 84, 3397–3399.


532

Durr M., Schmid A., Obermaier M., Rosselli S., Yasuda A. and Nelles G. (2005), ‘Low-temperature fabrication of dye-sensitized solar cells by transfer of composite porous layers’, Nature Materials 4, 607–611.

Durrant J. R., Haque S. A. and Palomares E. (2006), ‘Photochemical energy conversion: from molecular dyads to solar cells’, Chem. Comm., 3279–3289.

Fabregat-Santiago F., Bisquert J., Garcia-Belmonte G., Boschloo G. and Hagfeldt A. (2005), ‘Influence of electrolyte in transport and recombination in dye-sensitized solar cells studied by impedance spectroscopy’, Solar Energy Mat. Solar Cells 87, 117–131.

Ferrere S., Zaban A. and Gregg B. A. (1997), ‘Dye sensitization of nanocrystalline tin oxide by perylene derivatives’, J. Phys. Chem. B 101, 4490–4493.

Frank A. J., Kopidakis N. and van de Lagemaat J. (2004), ‘Electrons in nano-structured TiO2 solar cells: transport, recombination and photovoltaic properties’, Coordination Chem. Rev. 248, 1165–1179.

Gerischer H. and Tributsch H. (1968), Ber. Bunsenges. Phys. Chem. 72, 437. Grätzel M. (2006), ‘Photovoltaic performance and long-term stability of dye-

sensitized mesosocopic solar cells’, Comptes Rend. Chimie 9, 578–583. Green A. N. M., Palomares E., Haque S. A., Kroon J. M. and Durrant J. R. (2005),

‘Charge transport versus recombination in dye-sensitized solar cells employing nanocrystalline TiO2 and SnO2 films’, J. Phys. Chem. B 109, 12525–12533.

Hagfeldt A. and Grätzel M. (2000), ‘Molecular photovoltaics’, Acc. Chem. Res. 33, 269–277.

Haque S. A., Palomares E., Cho B. M., Green A. N. M., Hirata N. and Klug D. R. (2005), ‘Charge separation versus recombination in dye-sensitized nanocrystalline solar cells: the minimization of kinetic redundancy’, J. Am. Chem. Soc. 127, 3456–3462.

Hara K., Sato T., Katoh R., Furube A., Yoshihara T. and Murai M. (2005), ‘Novel conjugated organic dyes for efficient dye-sensitized solar cells’, Adv. Functional

Mat. 15, 246–252. Hengerer R., Kavan L., Krtil P. and Grätzel M. (2000), ‘Orientation dependence of

charge-transfer processes on TiO2 (anatase) single crystals’, J. Electrochem. Soc. 147, 1467–1472.

Henry C. H. (1980), ‘Limiting efficiencies of ideal single and multiple energy-gap terrestrial solar cells’, J. Appl. Phys. 51, 4494–4500.

Hinsch A., Kroon J. M., Kern R., Uhlendorf I., Holzbock J. and Meyer A. (2001), ‘Long-term stability of dye-sensitised solar cells’, Progr. Photovoltaics 9, 425–438.


533

Horiuchi T., Miura H., Sumioka K. and Uchida S. (2004), ‘High efficiency of dye-sensitized solar cells based on metal-free indoline dyes’, J. Am. Chem. Soc. 126, 12218–12219.

Hoyer P. and Weller H. (1995), ‘Potential-dependent electron injection in nanoporous colloidal ZnO films’, J. Phys. Chem. 99, 14096–14100.

Huynh W. U., Dittmer J. J. and Alivisatos A. P. (2002), ‘Hybrid nanorod-polymer solar cells’, Science 295, 2425–2427.

Ito S., Zakeeruddin S. M., Humphry-Baker R., Liska P., Charvet R. and Comte P. (2006), ‘High-efficiency organic-dye-sensitized solar cells controlled by nano-crystalline TiO2 electrode thickness’, Adv. Mat. 18, 1202–1205.

Jiu J. T., Wang F. M., Isoda S. and Adachi M. (2005), ‘Highly efficient dye-sensitized solar cells based on single crystalline TiO2 nanorod film’, Chem. Lett. 34, 1506–1507.

Kruger J., Plass R., Grätzel M., Cameron P. J. and Peter L. M. (2003), ‘Charge transport and back reaction in solid-state dye-sensitized solar cells: a study using intensity-modulated photovoltage and photocurrent spectroscopy’, J. Phys. Chem. B 107, 7536–7539.

Kubo W., Kitamura T., Hanabusa K., Wada Y. and Yanagida S. (2002), ‘Quasi-solid-state dye-sensitized solar cells using room temperature molten salts and a low molecular weight gelator’, Chem. Comm., pp. 374–375.

Kubo W., Sakamoto A., Kitamura T., Wada Y. and Yanagida S. (2004), ‘Dye-sensitized solar cells: improvement of spectral response by tandem structure’, J. Photochem. Photobiol. A—Chemistry 164, 33–39.

Liska P., Thampi K. R., Grätzel M., Bremaud D., Rudmann D. and Upadhyaya H. M. (2006), ‘Nanocrystalline dye-sensitized solar cell/copper indium gallium selenide thin-film tandem showing greater than 15% conversion efficiency’, Appl. Phys.

Lett. 88, 203103–203106. McEvoy A. J. and Grätzel M. (1994), ‘Sensitization in photochemistry and photo-

voltaics’, Solar Energy Mat. Solar Cells 32, 221–227. Namba S. and Hishiki Y. (1965), J. Phys. Chem. 69, 774. Nanu M., Schoonman J. and Goossens A. (2005), ‘Solar-energy conversion in TiO2/

CuInS2 nanocomposites’, Adv. Functional Materials 15, 95–100. Nazeeruddin M. K., Amirnasr M., Comte P., Mackay J. R., McQuillan A. J. and

Houriet R. (2000), ‘Adsorption studies of counterions carried by the sensitizer cis-dithiocyanato(2,2'-bipyridyl-4,4'-dicarboxylate) ruthenium(II) on nanocrystaline TiO2 films’, Langmuir 16, 8525–8528.

Nazeeruddin M. K., Kay A., Rodicio I., Humphrybaker R., Muller E. and Liska P. (1993), ‘Conversion of light to electricity by cis-X2-bis(2,2'-bipyridyl-4,4'-


534

dicarboxylate)ruthenium(II) charge-transfer sensitizers (X = Cl–, Br–, I–, CN– and SCN–) on nanocrystalline TiO2 electrodes’, J. Am. Chem. Soc. 115, 6382–6390.

Nazeeruddin M. K., Pechy P., Renouard T., Zakeeruddin S. M., Humphry-Baker R. and Comte P. (2001), ‘Engineering of efficient panchromatic sensitizers for nano-crystalline TiO2-based solar cells’, J. Am. Chem. Soc. 123, 1613–1624.

Nelson J. and Chandler R. E. (2004), ‘Random walk models of charge transfer and transport in dye sensitized systems’, Coordination Chem. Rev. 248, 1181–1194.

Nogueira A. F., Longo C. and de Paoli M. A. (2004), ‘Polymers in dye sensitized solar cells: overview and perspectives’, Coordination Chem. Rev. 248, 1455–1468.

Nozik A. J. (2003), ‘Quantum dot solar cells’, in Next Generation Photovoltaics:

High Efficiency Through Full Spectrum Utilization, Marti A. and Luque A., eds, CRC Press, Boca Raton, pp. 196–222.

O’Regan B. and Grätzel M. (1991), ‘A low-cost, high-efficiency solar-cell based on dye-sensitized colloidal TiO2 films’, Nature 353, 737–740.

O’Regan B. and Schwartz D. T. (1998), ‘Large enhancement in photocurrent efficiency caused by UV illumination of the dye-sensitized heterojunction TiO2/ RuLL'NCS/CuSCN: initiation and potential mechanisms’, Chem. of Materials 10, 1501–1509.

Peter L. M. and Wijayantha K. G. U. (2000), ‘Electron transport and back reaction in dye sensitised nanocrystalline photovoltaic cells’, Electrochim. Acta 45, 4543–4551.

Plass R., Pelet S., Krueger J., Grätzel M. and Bach U. (2002), ‘Quantum dot sensitiz-ation of organic–inorganic hybrid solar cells’, J. Phys. Chem. B 106, 7578–7580.

Reddy P. Y., Giribabu L., Lyness C., Snaith H. J., Vijaykumar C., Chandrasekharam M., Lakshmikantam M., Yum J.-H., Kalyanasundaram K. Graetzel M. and Nazeeruddin M. K. (2007), ‘Efficient sensitization of nanocrystalline TiO2 films by a near-IR-absorbing unsymmetrical zinc phthalocyanine’, Angewandte Chem.

Int. Ed. 46, 373–376. Redmond G. and Fitzmaurice D. (1993), ‘Spectroscopic determination of flat-band

potentials for polycrystalline TiO2 electrodes in nonaqueous solvents’, J. Phys.

Chem. 97, 1426–1430. Rensmo H., Keis K., Lindstrom H., Sodergren S., Solbrand A. and Hagfeldt A.

(1997), ‘High light-to-energy conversion efficiencies for solar cells based on nanostructured ZnO electrodes’, J. Phys. Chem. B 101, 2598–2601.

Rothenberger G., Fitzmaurice D. and Grätzel M. (1992), ‘Optical electrochemistry. 3. Spectroscopy of conduction-band electrons in transparent metal-oxide semi-


535

conductor films—optical determination of the flat-band potential of colloidal titanium dioxide films’, J. Phys. Chem. 96, 5983–5986.

Sayama K., Sugihara H. and Arakawa H. (1998), ‘Photoelectrochemical properties of a porous Nb2O5 electrode sensitized by a ruthenium dye’, Chem. of Materials 10, 3825–3832.

Schaller R. D. and Klimov V. I. (2004), ‘High efficiency carrier multiplication in PbSe nanocrystals: implications for solar energy conversion’, Phys. Rev. Lett. 92, 86601–86604.

Schmidt-Mende L., Bach U., Humphry-Baker R., Horiuchi T., Miura H. and Ito S. (2005a), ‘Organic dye for highly efficient solid-state dye-sensitized solar cells’, Adv. Materials 17, 813–815.

Schmidt-Mende L., Kroeze J. E., Durrant J. R., Nazeeruddin M. K. and Grätzel M. (2005b), ‘Effect of hydrocarbon chain length of amphiphilic ruthenium dyes on solid-state dye-sensitized photovoltaics’, Nano Lett. 5, 1315–1320.

Shklover V., Ovchinnikov Y. E., Braginsky L. S., Zakeeruddin S. M. and Grätzel M. (1998), ‘Structure of organic/inorganic interface in assembled materials compris-ing molecular components. Crystal structure of the sensitizer bis[(4,4'-carboxy-2,2'-bipyridine)(thiocyanato)]ruthenium(II)’, Chem. of Materials 10, 2533–2541.

Snaith H. J., Zakeeruddin S. M., Schmidt-Mende L., Klein C. and Grätzel M. (2005), ‘Ion-coordinating sensitizer in solid-state hybrid solar cells’, Angewandte Chemie

Int. Ed. 44, 6413–6417. Snaith H. J., Moule A. J., Klein C., Meerholz K., Friend R. H. and Graetzel

M. (2007), ‘Efficiency enhancements in solid-state hybrid solar cells via reduced charge recombination and increased light capture’, Nano Letters 7, 3372–3376.

Stathatos E., Lianos R., Zakeeruddin S. M., Liska P. and Grätzel M. (2003), ‘A quasi-solid-state dye-sensitized solar cell based on a sol–gel nanocomposite electrolyte containing ionic liquid’, Chem. of Materials 15, 1825–1829.

Tennakone K., Kumara G. R. R. A., Kumarasinghe A. R., Wijayantha K. G. U. and Sirimanne P. M. (1995), ‘A dye-sensitized nanoporous solid-state photovoltaic cell’, Semiconductor Sci. Technol. 10, 1689–1693.

Toyoda T. (2006), ‘Outdoor tests of large area dye sensitized solar cell modules by Aisin Seiki, Inc.’, Re2009 Renewable Energy Congress, Makura, Japan, October 2006.

Tsubomura H., Matsumura M., Nomura Y. and Amamiya T. (1976), ‘Dye sensitized zinc oxide–aqueous electrolyte–platinum photocell’, Nature 261, 402–403.

Vittadini A., Selloni A., Rotzinger F. P. and Grätzel M. (1998), ‘Structure and energetics of water adsorbed at TiO2 anatase (101) and (001) surfaces’, Phys. Rev.

Lett. 81, 2954–2957.


536

Wang P., Klein C., Humphry-Baker R., Zakeeruddin S. M. and Grätzel M. (2005a), ‘A high molar extinction coefficient sensitizer for stable dye-sensitized solar cells’, J. Am. Chem. Soc. 127, 808–809.

Wang P., Klein C., Humphry-Baker R., Zakeeruddin S. M. and Grätzel M. (2005b), ‘Stable �8% efficient nanocrystalline dye-sensitized solar cell based on an elec-trolyte of low volatility’, Appl. Phys. Lett. 86, 1235081–1235083.

Wang P., Zakeeruddin S. M., Humphry-Baker R., Moser J. E. and Grätzel M. (2003a), ‘Molecular-scale interface engineering of TiO2 nanocrystals: improving the efficiency and stability of dye-sensitized solar cells’, Adv. Mat. 15, 2101–2103.

Wang P., Zakeeruddin S. M., Moser J. E., Nazeeruddin M. K., Sekiguchi T. and Grätzel M. (2003b), ‘A stable quasi-solid-state dye-sensitized solar cell with an amphiphilic ruthenium sensitizer and polymer gel electrolyte (Vol. 2, p. 402, 2003)’, Nature Materials 2, 402–407.

West W. (1874), Vogel Centennial Symp., Photogr. Sci. Eng. 18, 35. Zaban A., Ferrere S. and Gregg B. A. (1998), ‘Relative energetics at the semi-

conductor sensitizing dye electrolyte interface’, J. Phys. Chem. B 102, 452–460.

dye-sensitised mesoscopic solar cells

Documents