Solvation Dynamics of Electron Produced by Two-Photon Ionization of Liquid Polyols. II. Propanediols J. Bonin, I. Lampre, P. Pernot, and M. Mostafavi* Laboratoire de Chimie Physique/ELYSE, UniV Paris-Sud, UMR 8000, Ba ˆ t. 349, Orsay, France 91405, and CNRS, Orsay, France 91405 ReceiVed: December 4, 2006; In Final Form: March 15, 2007 Temporal evolution of transient absorption spectra of electrons produced by two-photon ionization of two isomers, propane-1,2-diol (12PD) and propane-1,3-diol (13PD), with 263 nm femtosecond laser pulses has been studied on picosecond time scale. The two-photon absorption coefficients of 12PD and 13PD at 263 nm were determined to be ) (2.0 ( 0.3) × 10 -11 and (2.4 ( 0.3) × 10 -11 mW -1 , respectively. Time-resolved absorption spectra ranging from 440 to 720 nm have been measured, showing a blue shift for the first tens of picoseconds for both solvents. However, the observed solvation dynamics of electron appears faster in 13PD than in 12PD. The transient signals of electron solvation have then been reconstructed with different models (stepwise mechanism or continuous relaxation model) using a Bayesian data analysis method. Results are discussed, compared with those previously obtained in ethylene glycol (J. Phys. Chem. A 2006, 110, 1705) and corroborate the interpretation, according to which the solvation of electrons is mainly governed by continuous solvent molecular motions. 1. Introduction Much research has been focused on understanding the mechanisms of thermalization, localization, and solvation of an excess electron in liquid solvents, particularly in water. Due to the development of ultrashort laser pulses, great strides have been made in the observation of the dynamics of this electron on subpicosecond to picosecond time scales. 1-18 The initial work of Migus et al. 1 and subsequent studies in water 3 and alco- hols 2,16,18 suggested the existence of infrared absorbing precur- sors of the solvated electron. However, other works on electron solvation favored a continuous relaxation, often called “continu- ous shift” model, in which the existence of infrared absorbing species is not required. 13,14,19 Combinations of both mechanisms, stepwise and continuous models, have also been proposed to account for the observed spectral evolution. 5,7,20,21 As described in our preceding paper, which is referred to henceforth as paper I, 22 this context instigates our study of the ultrafast dynamics of electron in liquid polyols. In paper I, the solvation dynamics of the electron in ethylene glycol (EG, ethane-1,2-diol) have been detailed. The electron was produced by two-photon ionization of the solvent with 263 nm femto- second laser pulses. The two-photon absorption coefficient of EG at 263 nm was determined to be ) (2.1 ( 0.2) × 10 -11 mW -1 . Transient absorption spectra were recorded in the visible domain between 430 and 720 nm for the first 50 ps. In that time range, the electron decay due to geminate recombination is negligible. At very short delay time after the pump pulse, the excess electron in EG presents a very broad absorption band in the visible and near-infrared domain with a maximum around 675 nm. The red part of the absorption band drops rapidly for the first 5 ps while the blue part increases slightly, leading to a blue shift of the absorption band maximum down to 590 nm. Then the absorbance on the red side of the spectrum follows its decrease while the absorbance on the blue side remains nearly constant. As a consequence, the maximum of the absorption band continues to shift toward shorter wavelengths and, 50 ps after the pump pulse, is around 570 nm, the position of the absorption band maximum of the equilibrated solvated electron in EG. Using a Bayesian data analysis method to obtain parametric identification and model comparison, we performed a fitting procedure of the spectro-kinetics signals with different models. Sequential stepwise models with two, three and four states or species were tested. It appeared that at least three species are necessary but sufficient to properly adjust the data. We also wondered whether our observations could be explained by continuous relaxation models. We did not postulate that the absorption profile of the electron shifts, as a whole, to the blue during the solvation, without changing its shape. But, we described the time-dependent absorption spectrum by a log- normal function with time-dependent parameters, the amplitude ǫ(t), the peak position E max (t), the full width at half-maximum Ω(t) and the asymmetry γ(t). The time dependence of each parameter was modeled by a mono-, bi- or triexponential function or by a stretched exponential function. We obtained a very good fit of the data using biexponential functions with the same characteristic times τ 1 = 1.7 ps and τ 2 = 25.5 ps for the parameters E max (t), Ω(t), γ(t) and a monoexponential function with τ = 1 ps for the amplitude ǫ(t). Stretched exponential functions provided a good adjustment of the signals, too. But, we found small values for the parameters < 0.5, corresponding to very broad distributions of exponential relaxation times, which make difficult the physical interpretation of the stretched exponential function. We also tried the “hybrid” model proposed by Pe ´pin et al. 5 in which two initial unrelaxed species, e wb,1 - and e sb,2 - , undergo continuous blue shift with spectra keeping the same shapes and the e wb,1 - species converts into e sb,2 - relaxing to the fully solvated electron. That model did not * Corresponding author. E-mail: mehran.mostafavi@lcp.u-psud.fr. Present address: Laboratoire d’Electrochimie Mole ´culaire, Universte ´ Paris 7 Denis Diderot, 2 Place Jussieu, 75251 Paris Cedex 05, France. 4902 J. Phys. Chem. A 2007, 111, 4902-4913 10.1021/jp068323q CCC: $37.00 © 2007 American Chemical Society Published on Web 05/19/2007