Title: Experiment and Theory on the Complex Dynamics of Water
Speaker: Dr. John Asbury, Fayer Research Group, Stanford University
Venue: 112 Chem Annex
Date: Wednesday, February 2, 2005

Hydrogen bonds between water molecules form networks that evolve in time. These networks are dynamic, with constantly changing bond lengths which determine the strength of bonding intereactions. Fluctuations in these networks (on the order of 10 fs to 1 ps) play an important role in solvation processes due to solute molecules having to fit their way into this structure in order to dissolve. Also, hydrogen bond dynamics also play an important role in protein folding.

The presence of a hydrogen bond lowers the intermolecular potential and results in redshifting of the vibrational frequencies. This can be monitored via spectral diffusion using vibrational echo experiments. The pulse sequence can be described using the runners’ analogy due to Erwin Hahn, 1952. Three pulses are employed, analogous to guns. The first shot starts the race, with runners running at their own speeds. The second shot stops the race and the third gets everyone running backward with the same average speed as they started out with. But since fast runners would also run back faster than slow runners, etc., the effect of having a large spread in the initial speeds is cancelled, leaving only differences that they had picked up along the way. Similarly for molecules in a pulse echo experiment, such a sequence rephases inhomogeneities in speed but not fluctuations in speed. This method eliminates the large linear absorptive part of the signal, resulting in increased sensitivity.

Hydrogen bond dynamics of HOD was studied in D2O. A vibrational echo correlation spectroscopy setup with heterodyne detection was employed, with monitoring of both the ν1->2 and ν0->1 transition frequencies, corresponding to stimulation emission and absorption respectively. Very short pulses of 40 fs were required to illuminate the broad O-H stretching peak. The frequency autocorrelation function (FFCF) was then calculated and fit with simulations to a theoretical model using diagrammatic perturbation theory.”;s:4:”body”;s:3119:”The experimental results were compared with simulation results from Jim Skinner (UWM) using three different models of water, all of which were rigid: TIP4P (four-point transferable intermolecule potential), SPC/E (extended simple point charge) [pdf]and SPC-FQ (SPC with fluctuating charge). Comparisons of the frequency autocorrelation function from experiment and simulation initially showed an overestimate of the FFCF by all three models. However, a major experimental effect that was not accounted for in the theoretical fit was a frequency-dependent linewidth, which could not be adequately captured using a tri-exponential fit as well the Gaussian approximation employed in diagrammatic perturbation theory.

The workaround was to restrict the data analysis to the region where the signal peak was roughly symmetric. SPC-FQ was then seen to be able to reproduce the dynamical linewidth well, but still overestimating. TIP4P and SPC/E predicted autocorrelation functiona that were way too fast. This seems to indicate hydrogen bond equilibriation on the picosecond time scale as well as the importance of dynamic polarization in capturing those dynamics. However, some residual effects due to damped many-body motion were also observed experimentally that were not captured.

In order to study these many-body correlations, ultrafast transient absorption spectroscopy was then performed, also with heterodyne detection. Two dimensional time-frequency spectra showed the formation of a redshifted photoproduct as the time delay between pulses was increased up to ~6 ps. Analysis of the data with singular value decomposition was necessary to resolve the spectrum into the ν1->2, ν0->1 and photoproduct bands respectively. A wavelength-independent vibrational relaxation time of τ = 1.42 ± 0.05 ps was obtained, in contrast with previous findings. This could be explained by previous investigators not accounting for the photoproduct absorption peak. An equilibrium temperature increase of 2°C was calculated.

The presence of an isosbestic point in the spectra indicated exchange between to populations, namely vibrational population of the OH mode vs. population of the OD mode. This in turn was a precise measure of whether or not the OD bond was particpating in hydrogen bonding. The data could not be explained with the conventional hydrogen-bond mode “bath” model, necessitating the use of a kinetic model where other intermolecular and intramolecular modes had to be accounted for explicitly.

It is interesting that the SPC-FQ model was able to account for the experimental data, given the rigidity of the model. Also interesting in the accurate reproduction of the frequency-dependent dielectric constant from 0.3 wavenumbers to 3300 wavenumbers, even though the model was not specifically optimized for that purpose.