Chemistry Reference
In-Depth Information
the first place. If all the necessary experimental data were in hand, there would
be little practical need to carry out simulations. This means
predictive methods
will be needed to estimate parameters, and fitting to experimental data is not
likely to be a viable strategy (at least for awhile).
Faced with these challenges, a handful of research groups forged ahead
anyway and began simulating ionic liquids. All used a classical force field with
the same basic features as the Huggins and Mayer force field of Eq. [1],
although intramolecular terms were treated in different manners. It is a bit dif-
ficult to say in a historical context what the first simulation of an ionic liquid
was, given the ambiguity with which an ionic liquid is defined. Hawlicka and
Dlugoborski
27
conducted a molecular dynamics study of tetramethylammo-
nium chloride dissolved in water in 1997 with the intent of studying hydro-
phobic phenonema in aqueous solutions, certainly a relevant topic for ionic
liquids. In early 2000, Oberbrodhage
28
conducted a molecular dynamics simu-
lation study of tetrabutylammonium iodide dissolved in formamide as well as
at the interface between formamide and hexane. Oberbrodhage was interested
in the use of tetrabutylammonium iodide as a phase-transfer catalyst. The
melting points of tetramethylammonium chloride and tetrabutylammonium
iodide are too high to fall under our definition of an ionic liquid, but the
quaternary ammonium cation, when paired with other anions, is used exten-
sively as an ionic liquid.
29
These two groups were probably unaware of the
then growing interest in ionic liquids when they published their studies.
In 2001, Hanke, Price and Lynden-Bell
30
were the first to conduct an
atomistic simulation of compounds that can be called ionic liquids under our
definition. They used molecular dynamics to model the crystalline state of
1,3-dimethylimidazolium chloride ([C
1
mim][Cl]), 1,3-dimethylimidazolium
hexafluorophosphate ([C
1
mim][PF
6
]), 1-ethyl-3-methylimidazolium chloride
([C
2
mim][Cl]), and 1-ethyl-3-methylimidazolium hexafluorophosphate ([C
2
mim][PF
6
]). They also modeled the liquid state of [C
1
mim][Cl] and [C
1
mim]
[PF
6
], both of which are relatively high melting substances. Because of this
(and the need to speed dynamics and thus limit computation times), the liquid
simulations were carried out at temperatures between 400 and 500K. The form
of the potential function they used was
1=2
q
i
q
j
r
ij
þð
ð
B
ii
þ
B
jj
Þ
r
ij
ð
C
ii
C
jj
Þ
1
=
2
U
ij
ð
r
ij
Þ¼
A
ii
A
jj
Þ
exp
½
2
r
ij
2
which has the same form as Eq. [1], except the dipole-quadrupole term is
omitted. This is also the functional form used by Williams and Cox to model
azohydrocarbons.
31
Bond lengths were kept fixed, as were all bond angles
except those between the N-C-H atoms in the methyl groups. Partial charges
q
i
were located at each atomic center, with values determined from the atomic
multipole moments derived from a distributed multipole analysis
32
of the
