The transferability of force fields in coarse-grained (CG) modeling of polymers,
particularly its dependence on extraction conditions (temperature and density),
presents a significant challenge. This study investigates the transferability of CG
potentials for polyamide-6,6 using two mapping schemes; one bead per monomer
and two beads per monomer. The CG potentials were tuned by iteratively
matching the CG distributions to their corresponding atomistic distributions. The
potentials were derived at high (600 K) and low (300 K) temperatures, and each
was evaluated based on its ability to reproduce atomistic distributions at 440 K,
i.e., a temperature midway between the two initial temperatures. We further
examined the effects of bead size (i.e., the number of beads per monomer) and
torsional degrees of freedom on the transferability of the CG potentials.
For the one-bead-per-monomer mapping scheme, increasing temperature led to a
broadening of structural distributions, including bond lengths, bond angles,
torsions, and nonbonded interactions. In terms of transferability, the force field
derived at 600 K showed acceptable performance when tested against atomistic
distributions at 440 K, although the density increased by approximately 8.5%. In
contrast, the force field derived at 300 K did not maintain a stable melt structure
for the polymer at 440 K.
The two-bead-per-monomer mapping scheme, on the other hand, resulted in a
significant improvement in force-field transferability. In this model, both force
fields (generated at 600 K and 300 K) were able to reproduce the structural
distributions and density at 440 K with higher accuracy, compared to the onebead-per-monomer model. Overall, the findings suggest that increasing the
resolution of the CG model improves the transferability of CG potentials.
However, the inclusion of torsional degrees of freedom in the CG potential does
not significantly affect the nonbonded distributions or the overall transferability of
the forc