Chemistry Reference
In-Depth Information
According to the workflow presented here to develop the CG force fields using
the IBI method, one obtains potentials for bonded and non-bonded interactions at
the same time on the basis of the same atomistic simulation; thus there is no clear
separation between the optimization procedures for bonded and non-bonded inter-
action potentials. One can achieve this separation by deriving CG bond length, bond
angle, and torsional distributions from the atomically detailed conformations sam-
pled by a single (chain) molecule in vacuum, if the conformational sampling of the
molecule in vacuum and in the bulk (or solution) phase does not differ substantially
[
48
]. The IBI method has the advantage that detailed structural information is
included into the CG model, and it has been used successfully for molecular liquids
[
18
], polymer melts [
15
,
37
,
73
], dendrimers [
20
], polymer solutions [
18
], polymer
blends [
74
], and ionic liquids [
19
].
However, there can be limits to this approach because it is not always clear
whether the chosen CG mapping scheme can converge to an optimal fit. For liquid
mixtures or solutions, the situation is more complex because several RDFs that
mutually affect each other need to be simultaneously reproduced. In addition, for
dilute solutions, where we have a low concentration of solute, the solute-solute
RDFs converge very slowly in the CG simulations. In this case, the PMF between
the solute molecules can be obtained using free-energy calculation methods such as
umbrella sampling or constraint dynamics. Recently, these methods have been used
in an iterative optimization approach to study self-assembling dipeptides at the CG
scale [
75
,
76
]. The PMF between solute molecules in a solvent box,
V
AA
PMF
ð
r
Þ
,is
calculated by all-atom simulation from
n
distance constraint simulations:
ð
d
s
r
2
k
B
T
s
V
PMF
ð
r
Þ¼
h
f
c
i
s
þ
þ
C
;
(7)
r
m
where
f
c
is the constraint force, and
r
m
is the maximum distance between the centers
of mass of the two molecules. This potential was successfully employed to simulate
the aggregation process of a hydrophobic dipeptide in solution with an implicit
solvent representation in a CG model [
76
]. Since the so-obtained PMF incorporates
the thermally averaged contributions from solute and solvent degrees of freedom, it
cannot be directly used as CG potential if the CG model has an explicit solvent
representation. To determine the solvent contribution that needs to be removed
from
V
AA
PMF
, the PMF calculations with the CG potential are run, while the direct
solute-solute interactions are excluded. The effective solute-solute potential can
then be obtained by subtracting
V
CG
PMF
;
excl
ð
r
Þ
from the all-atom PMF,
V
AA
PMF
[
76
].
This subtraction procedure removes the solvent contribution from the PMF, and is
similar to iteration steps in the IBI method. Recently, the method has also been used
to develop a CG model for PS [
77
]. To derive the non-bonded interactions, PS
oligomer pairs were simulated in vacuum with a detailed atomistic model. The
effective non-bonded potentials obtained in this procedure include the effects of
multibody correlations related to the chain connectivity (Brini et al., 2010, unpub-
lished results).
ð
r
Þ
ð
r
Þ
Search WWH ::
Custom Search