Thermodynamic Properties for Applications in Chemical Industry via Classical Force Fields - Multiscale Molecular Methods in Applied Chemistry - page 213

angle , adjacent bonds stretch to reduce the interactions between the atoms

forming the bond. Only few force fields include such cross terms, because it was
found that they are rarely important. Cross terms are usually a function of two
interactions like bond-bond, bond-angle, bond-torsion, or angle-torsion, but terms
containing more than two interactions can also be used. Cross terms are important
to cover vibrational spectra, but do not significantly affect structural or thermody-
namic properties [ 56 ]. Force fields can be classified depending on whether or not
they include cross terms. Various forms of cross terms can be found in [ 19 ] that are
not further discussed here.

y

2.2.6 1-4 Interactions

Van der Waals interactions were mentioned as intermolecular interactions. How-
ever, in many force fields Van der Waals and electrostatic interactions are also used
to describe the intramolecular interactions between different sites of the same
molecule that are separated by three (1-4 potential) or more bonds. Usually, the
intramolecular 1-4 potential is scaled for both the LJ and coulombic contributions
by an empirical factor, depending on the force field.

3 Force Field Parameterization

In the past, force fields were parameterized based only on experimental data;
nowadays, most modern force fields include substantial quantum chemical infor-
mation. According to the nature of the data used for parameterization, force fields
can be classified as ab initio, semi-empirical, and empirical. Simple potentials, e.g.,
for argon, which require few parameters, can be fitted exclusively to macroscopic
experimental data; however, more complex force fields have numerous parameters
and thus depend heavily on ab initio data. This chapter gives an introduction to the
present state-of-the-art in this field. Attention is given to the way modeling and
simulation on the scale of molecular force fields interact with other scales, which is
mainly by parameter inheritance. Parameters are determined both bottom-up from
quantum chemistry and top-down from experimental data.
In principle, every quantity that can be predicted from force field calculations
can also be used for its parameterization. The choice of the properties taken as
optimization target may depend on the intended application. However, if the target
properties are suitably chosen, force field models often show powerful predictive
capabilities. This is due to the fact that they reasonably separate the different types
of intermolecular interactions and are thus able to account for the interplay of
interaction energy and structure of the fluid, which is generally a weak point in
phenomenological approaches. Force fields for applications in the chemical indus-
try should be developed, including data on the liquid density as well as on entropic
properties, namely phase equilibria [ 57 ]. This is in line with more than 100 years