Solvation Effects on Chemical Shifts by Embedded Cluster Integral Equation Theory

The accurate computational prediction
of nuclear magnetic resonance
(NMR) parameters like chemical shifts represents a challenge if the
species studied is immersed in strongly polarizing environments such
as water. Common approaches to treating a solvent in the form of,
e.g., the polarizable continuum model (PCM) ignore strong directional
interactions such as H-bonds to the solvent which can have substantial
impact on magnetic shieldings. We here present a computational methodology
that accounts for atomic-level solvent effects on NMR parameters by
extending the embedded cluster reference interaction site model (EC-RISM)
integral equation theory to the prediction of chemical shifts of N-methylacetamide (NMA) in aqueous solution. We examine
the influence of various so-called closure approximations of the underlying
three-dimensional RISM theory as well as the impact of basis set size
and different treatment of electrostatic solute–solvent interactions.
We find considerable and systematic improvement over reference PCM
and gas phase calculations. A smaller basis set in combination with
a simple point charge model already yields good performance which
can be further improved by employing exact electrostatic quantum-mechanical
solute–solvent interaction energies. A larger basis set benefits
more significantly from exact over point charge electrostatics, which
can be related to differences of the solvent’s charge distribution.