8.1.1 Perspective.......................................................
Calculations on isolated molecules, unencumbered by solvent, are undoubtedly
simpler conceptually, theoretically, and algorithmically, than in vacuo computa-
tions (although the vacuum is not what it used to be [ 1 ]). So we ask: how realistic
are vacuum (gas phase) calculations, and how important is it to take into account the
embrace of solvent molecules? Serious questions about the value of calculations
which ignore solvent are clearly justified in the case of biological molecules and
reactions, since these entities are immersed in water. A relatively early article on
molecular modelling and computer-aided drug design [ 2 ] elicited incisive critical
comments: “When a process as fundamental as the absorption of one dioxygen
molecule by hemoglobin involves 80 water molecules...what can we learn about
docking a drug in vacuo?” gives the flavor of the critique [ 3 ]; a response to this
conceded that neglecting solvation is an “apparent oversimplification”, but con-
tended that “gas-phase structures correlate surprisingly well with a number of
known physiological facts” [ 4 ]. Nearly 2 decades later a study of the 20 natural
amino acids examined in detail their calculated geometries in the gas phase and in
solution (using various continuum models – see below) and concluded that “the use
of gas-phase-optimized geometries can in fact be quite a reasonable alternative to
the use of the more computationally intensive continuum optimizations” [ 5 ].
Examination of the literature and judicious reflection lead to the conclusion that
for some purposes in vacuo (gas phase) computations are not only adequate but are
the appropriate ones, while for other purposes considering solvation is essential.
If the purpose of a calculation is to probe the inherent properties of amoleculeas
athing in itself, or of aphenomenoncentered on isolated molecules, then we do not
want the complication of solvent. For example, a theoretically oriented study of the
geometry and electronic structure of a novel hydrocarbon, e.g. pyramidane [ 6 ], or of
the relative importance of diatropic and paratropic ring currents [ 7 ], properly
examines unencumbered molecules. On the other hand, if we wish, say, to calculate
from first principles the pKa of acids in water, we must calculate the relevant free
energiesin water[ 8 ]. Noteworthy too is the fact that solvation, in contrast to gas
phase treatments, is somewhat akin to molecules in bulk, in crystals [ 9 ]. Here a
molecule is “solvated” by its neighbors in a lattice, although the participants have
a much more limited range of motion than in solution. Rates, equilibria, and
molecular conformations are all affected by solvation. Bachrach has written a
concise review of the computation of solvent effects with numerous apposite
references [ 10 ].
8.1.2 Ways of Treating Solvation......................................
There are two basic ways to treat solvation computationally: explicit and implicit.
Microsolvation, explicit solvation, places solvent molecules around the solute
522 8 Some “Special” Topics