Back electron transfer from TiO2 nanoparticles to Fe-III(CN)(6)(3-): Origin of non-single-exponential and particle size independent dynamics

Back-electron-transfer (ET) dynamics in Fe-II(CN)(6)(4-)-sensitized colloid al TiO2 nanoparticles are studied using ultrafast pump probe spectroscopy. Excitation of the adsorbate-to-nanoparticle charge-transfer band at 400 nm leads to direct injection of electrons from Fe-II(CN)(6)(4-) to TiO2. The k inetics of back electron transfer from TiO2 to the Fe-III(CN)(6)(3-) are me asured by monitoring the bleach recovery of the charge-transfer band in the 430-600 nm region. The measured back-ET kinetics are non-single-exponentia l, and a multiexponential fit requires at least four components on the <1 n s time Scale. The kinetics are independent of pump power, indicating a gemi nate recombination process. Recombination kinetics are very similar in two samples of 5 and 11 nm (A-type) particles prepared from dried-nanoparticle powder, but they are noticeably different from those in samples of 3 and 9 nm (B-type) nanoparticles prepared directly from colloids without drying. T his result indicates that the back-ET kinetics in this system are more infl uenced by the surface properties of the nanoparticles than their sizes. Two models with different distributions of trapped electrons are used to descr ibe the back-ET kinetics. Model I assumes a homogeneous distribution of ele ctrons on the surface of the entire particle. This model predicts a large p article size dependence and cannot fit the observed kinetics. Model II assu mes a more localized distribution of injected electrons and takes account o f relaxation from shallow to deep trap states during the recombination proc ess. This model can fit the back-ET kinetics with three fitting parameters. According to this model, the injected electrons are trapped near the adsor bate, which accounts for the size independent back-ET kinetics. This model also predicts that trapped electrons at longer distance and/or larger trap energy recombine slower. A distribution of distance and trap energy as well as relaxation between trap states give rise to multiexponential back-ET ki netics.