Back to Table of Contents
46140-AC6
Modeling Spectroscopic Measurements of Solvation and Solvent Friction
Mark Maroncelli, Pennsylvania State University
Two unrelated computational
projects were proposed for this 2-year grant.
The first project involved molecular dynamics simulations of solvation,
solvation dynamics, and solvatochromism in gas-expanded liquids (GXLs). This work has now been completed and 2 papers
are currently in press reporting the work.
GXLs are conventional solvents such as acetonitrile “expanded”, i.e.
mixed with up to 90 mole percent, with a near critical gas, most often CO2. Such solvents are of interest both as greener
replacements for conventional solvents and for specialized purposes such as
nanoparticle fabrication. Our research
sought to understand how the solvation energies and dynamics present in a
conventional solvent are altered by the incorporation of gaseous molecules and
how electronic spectra of probe solutes report on these properties. Experimental studies in our group and others
showed what appeared to be extensive preferential solvation of probe
chromophores by the liquid component of the GXL. Although preferential solvation is expected,
application of commonly used methods for interpreting the experimental data
yielded unreasonably large values for the excess fraction of the liquid
component. Classical molecular dynamics
simulations of a number of solute + GXL combinations, together with a simple
model of the spectroscopy provided a clear understanding of how shifts of
electronic spectra must be interpreted in order to derive quantitative
information about local environment. The
gross exaggeration of the extent of preferential solvation by the naïve
analysis methods results from both the collective nature of the electrostatic
component of solvation (also relevant in conventional liquid mixtures) and from
the density variations with composition unique to GXLs. The dynamics of solvation were also studied
in detail for one particular solute and polar solvent + CO2 system
for which we could obtain experimental data.
Simulation – experiment comparisons showed that the dynamics consists of
two distinct components. At early times,
solvation is accomplished mainly via reorientation of the more polar component
present in the first solvation shell of the solute. This portion of the response is essentially
complete in 1 ps. The dynamics at longer
times (tens of picoseconds at high gas-component concentrations), consists of a
solvent sorting process in which polar solvent molecules exchange for CO2
in the first solvation shell of the solute in order to increase the degree of
preferential solvation of the excited state.
This biphasic dynamics is qualitatively similar to that which occurs in
conventional liquid mixtures, but the timescales are shorter because of the
greater fluidity of GXLs. Overall, these
simulations have provided a satisfying molecular level interpretation of
structure and dynamics of solvation in GXLs.
The second project involves modeling
the isomerization dynamics of malononitriles.
Molecules such as DMABMN, shown at the right, have been previously
studied for two reasons. First the “push-pull”
(electron donor acceptor) character of these molecules provides high nonlinear
polarizabilities of interest for optical device applications. The second characteristic, of primary
interest here, is the fact that they possess short, medium-sensitive
fluorescence quantum yields and lifetimes.
By virtue of this latter property, malononitriles have found use as
probes of “local viscosity” or “free volume” in studies of polymers and various
biological environments. This
sensitivity to environment results from some large-amplitude intramolecular
motion being required for electronic deactivation, but very little is known
about the precise nature of this motion.
The electronic structure calculations we have performed to date indicate
that the dominant motion involved is the double-bond torsion q, which leads to a conical intersection between S0
and S1. S1
torsional potentials developed using varying levels of theory (CIS, TDB3LYP,
MCSCF, MRMP2, RICC2 with basis sets of 6-31G(d) quality or higher) show
qualitatively similar behavior but with important quantitative
differences. Molecular dynamics
simulations using these torsional potentials show that the time scale of
internal conversion is critically dependent on rather subtle details of the
surface. In particular, experimental
time-resolved emission spectra are only consistent with potentials having a
broad (>40°) and barrierless region
prior to an abrupt drop leading to conical intersection. Thus, such comparisons between experimental
data and molecular dynamics simulations provide a means of assessing the
accuracy of ab initio excited-state
potentials. Work on both the ab initio calculations (including
solvent) and molecular dynamics simulations with semi-empirical potentials is
still in progress. Our first goal is to
use comparisons between experiment and simulation in simple solvents to develop
an accurate classical representation of the excited-state potential. Further simulations using this potential in
various simple and complex environments will then be used to understand in
detail what aspects of structure and dynamics of its surroundings DMABMN
reports on. We anticipate this work will
provide new perspectives on the molecular nature of friction as well as a much
better foundation for using such solutes as environmental probes.
Back to top