The strength of the electronic coupling between the molecular adsorbate and the semiconductor depends on the electronic structure of the donor molecule, the spacers and the anchoring group, and the semiconductor. One of the most convenient ways to vary the coupling strength systematically is to control the bridge length between the chromophore and the binding group. This approach was used successfully in earlier studies of thermal interfacial ET on metal or semiconductor electrodes modified by self-assembled-monolayers (SAM) of alkanethiols (Chidsey, 1991; Becka and Miller, 1992; Finklea and Hanshew, 1992; Smalley et al., 1995; Sachs et al., 1997; Gu and Waldeck, 1998; Creager et al., 1999; Smalley et al., 2003). Electron transfer mediated by longer bridges has generally been found to vary exponentially with the length of the bridge (Smalley et al., 1995; Smalley et al., 2003):
where kET(r = 0) is the extrapolated rate constant for r = 0, and в is the exponential decay constant. This dependence is attributed to an exponential decay of coupling strength with distance, which can be understood by the exponential dependence of the electron-tunnelling probability on distance or, more quantitatively, by a superexchange model of transfer of electrons and holes through these bridges (Newton, 1991; Jordan and Paddon-Row, 1992).
The dependence at short bridge length is less well understood, and deviation from exponential dependence has been reported (Smalley et al., 1995; Khoshtariya et al., 2001; Sikes et al., 2001; Smalley et al., 2003). For photoinduced ET in dye-sensitised nanocrystalline thin films, only a few systematic studies of distance dependence have so far been reported. Distance-dependent electron injection processes to TiO2 and SnO2 were observed from a rhenium bipyridyl complex with n = 0-5 methylene groups inserted between the bipyridine ligand and each of the two carboxylate binding groups, as shown in Fig. 11.7b (Asbury et al., 2000; Anderson et al., 2003a). On SnO2 with n > 3 methylene bridges, the observed rate decreased exponentially with bridge length, with в = ~1.0 per methylene unit or ~0.8 A-1. The decay constant is similar to that observed for long methylene bridges (Smalley et al., 1995; Smalley et al., 2003). At shorter bridge lengths n = 2 and n = 1, similar injection rates (~1/40 ps) were observed. The observed trend probably reflects the non-exponential dependence of coupling strength with distance at short chain lengths, which has been predicted in computational studies (Newton, 1991; Curtiss et al., 1993; Liang and Newton, 1993). Other studies have reported weak distance dependence in ET through short bridges (Kilsa et al., 2003; Piotrowiak et al., 2003). In many studies, knowledge of the flexibility of the bridge and surface-binding modes is often lacking, which hinders a quantitative understanding of distance dependence.
In addition to the bridge, the anchoring group provides another control of electronic coupling strength in interfacial electron transfer. The most common anchoring groups for metal oxides are based on phosphonate, carboxylate, catechol and their derivatives (O’Regan and Gratzel, 1991; Heimer et al., 1996; Gillaizeau – Gauthier et al., 2001; Odobel et al., 2003). The identity, number and location of the anchoring groups have been shown to affect photocurrent, suggesting their influence on charge separation and recombination kinetics (Heimer et al., 2000; Hara et al., 2002; Odobel et al., 2003). Variation of the injection rate with the position of the anchoring group has been reported (Piotrowiak et al., 2003). It was shown that the electron-injection rate from an ReC1 complex (Fig. 11.7) to TiO2 is significantly slower through carboxylate than through phosphonate anchoring groups (She et al., 2005).
Among numerous adsorbate/semiconductor combinations, the best solar-to-electric power conversion efficiency has been achieved in solar cells based on N3-sensitised TiO2 nanocrystalline thin films (O’Regan and Gratzel, 1991; Nazeeruddin et al., 1993). As a result, ET dynamics in this system have been most extensively studied (Tachibana et al., 1996; Hannappel et al., 1997; Ellingson et al., 1998; Ellingson et al., 1999; Asbury et al., 1999a; Haque et al., 2000; Heimer et al., 2000; Tachibana et al., 2001; Asbury et al., 2001a; Asbury et al., 2001b; Benko et al., 2002; Kalloinen et al., 2002; Kuciauskas et al., 2002; Asbury et al., 2003). It is now generally accepted that the dynamics of photoinduced electron injection from N3 to TiO2 are biphasic, consisting of a distinct <100 fs ultrafast component and one or more slower components on the picosecond and longer time scales.
The cause of the ultrafast electron injection from N3 and the mechanism of the biphasic injection has been a subject of considerable interest. This is discussed briefly here, and at greater length in Chapter 8. To help understand the injection dynamics, it is useful to review the photophysics of N3. Like many Ru bipyridyl complexes, the optical absorption spectrum of N3 is dominated by singlet metal-to-ligand-charge- transfer (*MLCT) bands, which can be roughly considered as a transfer of electron density from the Ru T2g d orbitals to the n* orbital of the bipyridine. Recent DFT studies suggest that there is a strong mixing of the metal d orbital with SCN- groups and the MLCT bands consists of transitions to a dense manifold of excited states (Monat et al., 2002). Optical excitation of N3 with a visible photon excites the molecule to the *MLCT states, at ~1-2 eV above the lowest energy 3MLCT state, depending on the excitation photon energy (see Fig. 11.7). The excited molecules can undergo rapid electronic relaxation (internal conversion and intersystem crossing) within the excited state manifold to the 3MLCT states and concurrent vibrational energy redistribution and relaxation and solvation relaxation (Barbara and Jarzeba, 1990; Rosenthal et al., 1991; Damrauer et al., 1997; Castner and Maroncelli, 1998). The singlet-to-triplet intersystem crossing time has been determined to be ~75 fs for adsorbed N3 on TiO2 in acetonitrile (Benko et al., 2002; Kalloinen et al., 2002).
In the light of the rapid excited-state relaxation dynamics and the observed <100 fs electron injection component, the fast injection component must occur from non – equilibrated excited states. Distinct singlet and triplet electron-injection pathways have been observed for Ru polypyridyl complexes on SnO2 (Iwai et al., 2000) and TiO2 (Benko et al., 2002; Kalloinen et al., 2002). In a detailed study of electron injection dynamics in N3/TiO2 (Benko et al., 2002; Kalloinen et al., 2002), 55% of electron injection was shown to occur with ~50 fs injection time from singlet MLCT states by monitoring the decay of the stimulated emission of the singlet states and the rise of absorption by the triplet and oxidised adsorbate. The remaining 50% of injection, on the 1-100 ps timescale, was confirmed to occur from the triplet state by recording the transient absorption spectra (extending from 400-1000 nm) of the triplet state, oxidised form and injected electrons, as shown in Fig. 11.3.
Further evidence of injection from unrelaxed excited states comes from excitation wavelength-dependent studies. Injection from the unrelaxed excited state must compete with intramolecular relaxation within the excited-state manifold. By varying the excitation wavelength, the molecules can be excited to different energy levels, affecting the branching ratio between the intramolecular relaxation and the interfacial electron-transfer pathways. Indeed, excitation wavelength dependence of injection dynamics has been reported (Benko et al., 2002; Kalloinen et al., 2002; Kuciauskas et al., 2002; Asbury et al., 2003; Anderson et al., 2003b). Shown in Fig. 11.9 is a comparison of injection kinetics of Ru N3-sensitised TiO2 films in a 1:1 ethylene/ propylene carbonate mixture with excitation at 400, 530, and 630 nm (Asbury et al.,
2003) . In this study, the IR absorption of injected electrons in TiO2 was monitored. These kinetics traces, normalised to the same signal magnitude at 150 ps, consist of fast (<100 fs) and slow injection components (Asbury et al., 2003). The amplitude of the fast injection component decreases at longer excitation wavelengths. In this case,
Figure 11.9 Normalised comparison of electron absorption dynamics of N3-sensitised TiO2 films in ethylene/propylene carbonates (1:1) following excitation at different wavelengths. Inset: the same data plotted on a shorter time scale. The solid lines are fits using the two-state injection model. The fast component is well described by a <100 fs rise and the slow component is fitted by a stretched exponential function with a 50 ps time constant. Reproduced with permission from J. Phys. Chem. B 107, 7376 (2003) (Asbury et al., 2003). Copyright 2003 American Chemical Society.
the instrument response function (~150 fs) was not short enough to resolve the difference in the rate of fast components. With better time resolution, this difference can be directly resolved (Benko et al., 2002; Kalloinen et al., 2002). Excitation wavelength-dependent injection kinetics were also observed in N3-sensitised ZnO (Anderson et al., 2003b) and SnO2 (Ai et al., 2005; Guo et al., 2006). (Iwai et al., 2000; Benko et al., 2003a; Ai et al., 2005); ZnO (Asbury et al., 1999b; Furube et al., 2003; Anderson et al., 2003b) and Nb2O5 (Ai et al., 2004).