Liquid Dynamics

Photoexcitation of a dye molecule in solution changes its electronic distribution almost instantaneously (~1 fs timescale). However, the solvent molecules surrounding the dye molecule can not respond this rapidly. Immediately after excitation, the solvent remains in whatever configuration optimally solvated the ground state of the dye molecule. With time, the solvent rearranges to accommodate the excited state charge distribution in a new configuration that minimizes the potential energy. This process takes place on a timescale of 50 fs to several ps.

Our studies probe the low frequency, collective modes of the solvent during this process of solvation. The changes in the solvent spectrum of chloroform, benzene, toluene, and pyridine are monitored after photoexciting dissolved TBNC (2,11,20,29-tetra-tert-butyl-2,3-naphthalocyanine). We find that there is a transient change in the solvent spectra that persists for the order of 1 ps, and that there is an additional modification to the far-IR optical constants (absorption coefficient and index of refraction) only when the excitation pulse is present. This is an analog of the so-called coherent artifact in ultrafast spectroscopy, and arises from instantaneous electronic processes rather than molecular ones.

We have modeled the solvent as a mixture of damped, driven collective harmonic oscillators (often referred to as multimode Brownian oscillators, or MBOs). The majority of the collective oscillators are not excited, and therefore are described with parameters obtained from the absorption and refractive index of the neat solvent. However, a small fraction of oscillators have a different resonant frequency and/or damping coefficient because of their proximity to the photoexcited dye molecule. Furthermore, there is a contribution from a collective oscillator whose "population" tracks the excitation laser pulse.

It is possible to generate a simulated data set by numerically propagating a THz pulse through this time-evolving ensemble of collective oscillators. The simulated data set is compared to the measured one, and the parameters describing the collective oscillators are iteratively adjusted until good agreement with experiment is achieved. An example of a measured and calculated 2D data set is shown in the FDTD Section. Based on the results of this type of fit, we can obtain the time-resolved, frequency dependent absorption coefficient and index of refraction.

For illustration, the absorption coefficient for chloroform as a function of time after photoexcitation of dissolved TBNC is shown here. At negative pump-probe delay times, the absorption coefficient is simply that for the unexcited solvent. At "time = 0", when the excitation pulse arrives, the absorption coefficient increases rapidly and fairly uniformly across the spectrum. However, by subtracting the unexcited solvent spectrum from those at later times, it becomes clear that the absorption change is not completely uniform. Information regarding the solvation process is contained in the time-evolution of the solvent spectrum.