A random-walk approach is developed to model the electron-transport dynamics in dye-sensitized TiO2 solar
cells within a multiple-trapping framework, and the predicted results are compared with those measured by
transient photocurrent. The illumination geometry and the wavelength of the probe light are used to create
certain initial spatial distributions of photoinjected electrons in the TiO2 films. Both have a dramatic effect
on the shape of the measured photocurrent transient. Cells are probed with light incident from either the
collecting (substrate) electrode side or the counter-electrode side. Excellent correspondence between simulated
and measured current transients is observed. When electrons are injected far from the collecting electrode,
their diffusion is found to be classical, corresponding to thermalized (nondispersive) transport. Nonthermalized
(dispersive) electron transport is shown to be important when electrons are injected near the collecting electrode,
which corresponds to the illumination condition under which the cell normally operates. For strongly absorbed
light incident from the collecting electrode side, it is estimated that about 80% of injected electrons are collected
before they are within 95% of complete thermalization. Failure to account for the presence of nonthermalized
electrons is shown to be a major limitation of previous theories of electron transport. The total density of trap
states is estimated to be relatively small, on the order of 1 trap per particle. The average detrapping time is
on the order of 10 ns. When electrons are generated far from the collecting electrode, they undergo an average
of about 106 trapping events before being collected. Analytical expressions are derived that relate the
experimentally measured collection time to other parameters affecting transport (e.g., trap density, light intensity,
film thickness, and free-electron mobility). Experimental evidence is presented for ambipolar diffusion.