Bernard Wenger
Dye-to-semiconductor electron transfer is the initial step in many processes where light is used for the storage of information (e. g. color photography) or converted into electricity as in dye-sensitized solar cells. In the latter, interfacial charge injection occurs on a timescale spanning from tens of femtoseconds (10-15 s) to several picoseconds (10-12 s), with an efficiency approaching 100 %. In standard electron transfer theory, the parameters controlling charge transfer dynamics are divided into two categories: a) electronic factors, depending essentially upon the overlap between the electronic wavefunctions of the donor and the acceptor, and b) nuclear factors, characterized by the reorganization energies of both the reactants and the surrounding solvent molecules. Because of the ultrafast injection rates observed for dye/semiconductors couples typically used in photovoltaic applications, the hypotheses leading to these models are likely to be invalid. In this work, we make use of time-resolved spectroscopic techniques to study the influence of some electronic and nuclear parameters influencing charge injection dynamics into wide bandgap semiconductors. As a preliminary study, we investigated reductive quenching of excited dyes by iodide at the surface of nanocrystalline TiO2. Our experiments show that the quantum yield of this reaction, that is in competition with electron injection, depends strongly upon the I- concentration, but also upon the preparation and aging of the samples. We deduce that a fraction of the dye molecules are not adsorbed directly onto the surface of the semiconductor, but rather aggregated in a second adsorption layer. In the light of these results we reconsidered the issue of charge injection from the standard N3 dye into TiO2. This reaction, largely studied due to its importance for dye-sensitized solar cells, has previously been reported to take place with a fast (< 100 fs) phase, followed by a slower (1-100 ps) component. Our measurements show that the slow part is actually related to the presence of weakly adsorbed dye molecules. When aggregation is minimized, we observe monophasic injection dynamics (τ < 20 fs). This result suggests the process is beyond the scope of vibration-mediated electron transfer models and is controlled by electron dephasing in the solid. Next, we applied two different approaches to investigate the influence of the distance over which charge transfer takes place. First, forward and back electron transfer kinetics were measured for a series of bridged-sensitizers containing p-phenylene spacer groups. For both reactions the rate decays exponentially with distance. However, the damping factor is much larger for the back transfer (β = 0.5 Å-1) than for injection (β = 0.19 Å-1), suggesting that nuclear reorganization plays an important role for the former reaction. In the second approach, a thin layer of Al2O3 of controlled thickness is inserted between the sensitizer and TiO2. In this case the distance parameter β is similar for both reactions. The weak estimated value (β = 0.15 Å-1) indicates that thin layers of aluminum oxide is much less insulating that pure, crystalline Al2O3, and that electron tunneling is mediated by empty states in the thin layer according to the superexchange mechanism. Finally, the influence of the density of acceptor states has been examined by using a ruthenium complex adsorbed on Nb2O5 films. Varying the excitation wavelength made it possible to probe the bottom of the conduction band where the density of states decays exponentially. Despite the spectral width of the pump laser pulses, a weak retardation (2-3 times) of electron injection was observed. In summary, our experiments show that electronic factors, in particular variations of electronic coupling with distance, mainly control electron injection dynamics at dye/semiconductor interfaces. By revealing the role of aggregation for standard dyes adsorbed on TiO2, these results have important consequences for the development of dye-sensitized solar cells. Finally, in the case of ultrafast monophasic kinetics, standard electron transfer models are not obeyed, and the dynamics is rather controlled by electron dephasing and relaxation in the continuum of acceptor states in the solid.
EPFL2006
