Chemistry Reference
In-Depth Information
Truhlar and coworkers have developed an elaborate algorithm for smoothing
the forces and the potential of the atoms switching regions in order to conserve
both energy and momentum [
21
]. Their adaptive partitioning (AP) method is able
to fix the conservation problem in a long time regime (400-1,000 ps), and they
suggest using their permuted AP method when dealing with equilibrium condi-
tions, a situation when energy or momentum drift could noticeably influence the
physical or chemical behavior of the system. The permuted AP method does
however include a number of multilevel computations at each time step (as
opposed to just one) in order to assess how the potential smoothing functions
are applied. This increases the number of computation by 2
N
for each
N
groups in
the buffer zone. By our estimates, in the squalane system ~4.4 carbon atoms
would be in the buffer zone on average, thus increasing the QM calculations by
>
21 times and making the AP method unattractive when thousands of trajectories
are needed.
In both of the algorithms just described, only small solvent molecules (e.g.,
H
2
O, NH
3
,
) are present in the simulations, and these molecules never straddle
a boundary. The use of switching functions when an atom or small molecule is in a
buffer region is convenient when the solvent molecule does not span from inside
the QM region, through the buffer, and into the MM region. In this situation, the
buffer region atoms are being given some MM character while still being bonded
to atoms in the QM region, which could lead to extraneous forces on all the QM
atoms in that molecule. This makes the inclusion of such algorithms to our
current model nontrivial. Additionally, it must be noted that in our simulations
we are not concerned with equilibrium conditions since the incident atom collides
at the surface with 1-5 eV of translational energy. Furthermore, our simulation
times are
...
10 ps, at which appreciable energy or momentum drift should be
very minimal.
Though we had success with dynamic partitioning with squalane, when we
began to study gas-liquid collision chemistry of room temperature ionic liquids,
dynamicpartitioningaswehaddoneinthe past became problematic. This is
because the liquid, 1-ethyl-3-methyl-imidazolium nitrate or [emim][NO
3
], is
composed of cations and anions, and each of these cannot meaningfully be
divided into parts in the QM and MM region. This would require some sort of
charge partitioning algorithm, likely requiring charges to switch back and forth
between regions as a function of time, resulting in extraneous forces. Because
the liquid is ionic, it has slower diffusion and much higher density than the hydro-
carbon liquids, preventing the surface from changing much over time. After
testing numerous trajectory paths, we found that it was rare that the incident
atom with an
E
T
<
5.0 eV had a direct interaction with more than the first two ion
pair layers. Therefore we decided to fix the QM region to include five ion pairs
(115 atoms) at or near the surface and the incident O(
3
P) atom. In comparison
to squalane, where there were often several well separated radical species,
O(
3
P)
þ
[emim][NO
3
] can be treated with a more localized QM description.
Search WWH ::
Custom Search