Biomedical Engineering Reference
In-Depth Information
dynamics codes which utilize CUDA-capable GPUs, reporting speed-up factors in
the order of 100 with respect to conventional CPUs, [51, 52, 53].
11.3.1 Generation and equilibration of the atomistic models of PLA
Our specific objective is to predict the water and polymer diffusion as a function of
the composition of the water/polymer mixture, characterized by the coordinates in
the state space
previously introduced. The diffusion coefficients
are then passed to the macroscale model.
To generate atomistic models of PLA we have applied Materials Studio 4.4 (Ac-
celrys, Inc.) and the COMPASS force field [44]. The repeating unit of PLA is avail-
able as standard model in Materials Studio. The polymeric chains with the desired
length are generated starting from the repeating units using the “Build Polymers”
tool of Materials Studio. Finally, we generate solvated amorphous bulk models con-
taining PLA and water using the “Amorphous Cell” tool of Materials Studio. The
polymeric chains are solvated in a periodic box with a total atom number varying
from 6,000 atoms (for quasi-dry systems) to 35,000 atoms (for highly solvated sys-
tems), as depicted in Fig. 11.1
The initial geometries of the bulk models are refined following the procedure
which consists in a series of MD calculations at different temperature and ensem-
bles in order to obtain chain redistribution within the periodic cell [10, 11, 18, 19].
Specifically, we perform a preliminary minimization followed by a sequence of nine
MD simulations (see Table 11.1). MD molecular dynamics simulations are carried
using the Discover module and the COMPASS force field implemented in Materi-
als Studio. Nonbonding interactions are computed using a cutoff for neighbor list at
ϒ
=(
ρ
w
,
ρ
1
,...,
ρ
N
)
Fig. 11.1.
Atomistic bulk model of quasi-dry PLA matrix (water molecules in blue color)
