TiO2 working electrodes used in DSSC are composed of 20 nm size crystallites. These electrodes are essentially transparent since visible light is not scattered for titania particles of sizes on the order of the few tens of nanometers. In fact, the incident photons that are not absorbed by the dye sensitized electrode are either lost through the counter electrode or partially absorbed by the electrolyte solution. From a photo-chemical point of view, this implies that part of the reagent (light) is wasted. The first attempt to collect these escaping photons were based on the use of polydisperse packings of sub-micron size spheres as highly diffusive reflecting layers (see figure 4). The intensity of the scattering effect depends on the size and refractive index of the particles as well as on the refractive index of the medium surrounding them. Hence, particle sizes between 300 nm and 1000 nm made of transition metal oxides with high refractive index, such as TiO2 (rutile n=2.8 or anatase, n=2.4) or ZrO2 (n=2.1) can be used as efficient scatterers. The introduction of a reflecting layer in DSSC to scatter photons and re-inject them into the electrode was proposed in a theoretical work by Usami (Usami, 1997). In other attempt, Ferber and Luther simulated the scattering process for a mixture of small and large particles, concluding that an enhancement in photon absorption was produced (Ferber & Luther, 1998). Subsequent simulations using different approaches were made (Rothenberger et al., 1999), for which a 6% of increase of the photon flux was predicted.
The integration of scattering centre particles in DSSC can be experimentally done under different architectures. In one approach, they are jointly included with the TiO2 nanocrystallites that form the electrode (Tachibana et al., 2002). In another, they are deposited as a second layer on top of the dye sensitized electrode in a well-known configuration referred to as "double-layer" (Hore et al., 2006). The latter configuration is normally the most employed. Other ways of integrating scattering layers made of submicron size disordered particles have also been reported (Wang et al., 2004), (Zhang et al., 2007). In general, the scattering layer is deposited using methods similar to those used for the TiO2 electrode deposition, such as doctor blade, screen printing, etc. A porous network connection between both layers is needed to allow the dye load and a proper diffusion of the charge carriers.
Improvements of efficiency around 20% in average are reported using the arrangements before described for nc-TiO2 layers with thicknesses between 5 and 7 microns. In fact, the efficiency record attained for a DSSC corresponds to a cell that incorporates a highly scattering layer (Chiba et al., 2006). The boost in efficiency is mainly a consequence of the increase of JSC up to 20%. On the other hand, IPCE values can be incremented between 20% and 50% depending on the spectral region considered. The highest improvement is obtained in the red part of the spectrum, where the dye extinction coefficient is small. Additionally, this light-scattering layer has been shown to act not only as a photon-trapping system but also to be an equally active layer in photovoltaic generation (Zhang et al., 2007).
Fig. 4. Left: Scanning electron microcospy image of a slab made of polydisperse submicrometer ТЮ2 particles that can be used as diffuse scattering layer in DSSC. Right: Scheme representing a DSSC with a "double layer" architecture.
The disadvantage of employing diffuse light scattering layers or mixed light scattering particles in DSSC is the loss of transparency of the cell. Unfortunately, the cells turn opaque, leaving them useless as window modules or any other application where transparency, one of the added values of these cells, was required. Also, the thickness of the electrode largely increases, particularly in the case of the double layer configuration, which might cause an increase of the resistance of the cell and a reduction of the voltage.