Impulsive solvent heating probed by picosecond x-ray diffraction.

The time-resolved diffraction signal from a laser-excited solution has three principal components: the solute-only term, the solute-solvent cross term, and the solvent-only term. The last term is very sensitive to the thermodynamic state of the bulk solvent, which may change during a chemical reaction due to energy transfer from light-absorbing solute molecules to the surrounding solvent molecules and the following relaxation to equilibrium with the environment around the scattering volume. The volume expansion coefficient alpha for a liquid is typically approximately 1 x 10(-3) K(-1), which is about 1000 times greater than for a solid. Hence solvent scattering is a very sensitive on-line thermometer. The decomposition of the scattered x-ray signal has so far been aided by molecular dynamics (MD) simulations, a method capable of simulating the solvent response as well as the solute term and solute/solvent cross terms for the data analysis. Here we present an experimental procedure, applicable to most hydrogen containing solvents, that directly measures the solvent response to a transient temperature rise. The overtone modes of OH stretching and CH3 asymmetric stretching in liquid methanol were excited by near-infrared femtosecond laser pulses at 1.5 and 1.7 microm and the ensuing hydrodynamics, induced by the transfer of heat from a subset of excited CH3OH* to the bulk and the subsequent thermal expansion, were probed by 100 ps x-ray pulses from a synchrotron. The time-resolved data allowed us to extract two key differentials: the change in the solvent diffraction from a temperature change at constant density, seen at a very short time delay approximately 100 ps, and a term from a change in density at constant temperature. The latter term becomes relevant at later times approximately 1 mus when the bulk of liquid expands to accommodate its new temperature at ambient pressure. These two terms are the principal building blocks in the hydrodynamic equation of state, and they are needed in a self-consistent reconstruction of the solvent response during a chemical reaction. We compare the experimental solvent terms with those from MD simulations. The use of experimentally determined solvent differentials greatly improved the quality of global fits when applied to the time-resolved data for C2H4I2 dissolved in methanol.