Chemistry Reference
In-Depth Information
reproduces the atmospheric pressure melting point quantitatively, it predicts a
significantly higher melting temperature as pressure (and hence density)
increases.
The results for alkali halides demonstrate that a simple force field like
that given in Eq. [1] can reproduce many of the experimental properties of a
molten salt. It appears to capture much of the physics of these systems
correctly, which is cause for optimism. Contrarily, it is difficult to obtain
both thermodynamic
and
transport properties correctly with the same model;
differences of 10-20% between computed and experimental transport proper-
ties are typical for alkali halide force fields that nominally get thermodynamic
properties correct. Therefore, we should expect similar or even larger differ-
ences between experimental properties of ionic liquids and those obtained
from simulations if this same general class of force field is used. Additionally,
even if a particular property is modeled accurately at a given state point, one
should not expect that property will be modeled accurately at another state
point. That is,
transferability
of the force field to other conditions can be
problematic and is a concern. Finally, note that the ionic liquids shown in
Figure 3 are much more complicated than simple alkali halides. The alkali
halide force fields were parameterized against known crystal data. Such data
are often lacking for ionic liquids, and there exist many more kinds of ionic
liquids than alkali halides. Moreover, people have been simulating alkali
halides for over 30 years, and they still cannot predict everything about
them with simulations. We should therefore expect it will be
much harder
to develop force fields for ionic liquids, and our expectations of their accuracy
should be modest.
EARLIEST IONIC LIQUID SIMULATIONS
As should be apparent by now, simulating ionic liquids raises many new
challenges that are not present with molten alkali halides. First, the cations
and anions are no longer spheres but are instead multiatom molecular species.
This means that the functional form of Eq. [1] is inadequate for treating these
systems; additional intramolecular terms need to be developed to model
bonded interactions accurately. Second, the ions are large and contain many
intramolecular degrees of freedom; great care is needed to ensure proper sam-
pling is obtained, making it likely that the computational costs will be high.
Third, part of the inaccuracy of the alkali halide force fields is due to the pri-
mitive ''fixed-charge'' method for handling electrostatic interactions; a more
sophisticated polarizable force field may be necessary to model these systems.
Finally, unlike alkali halides, there is a dearth of experimental data with which
to parameterize the force field (especially when the earliest ionic liquid
simulation studies were conducted). As suggested in the introductory remarks,
however, that is one of the main reasons for wanting to model ionic liquids in
