In the field fo polymer simulation, the explicit atom (EA) model (all the hydrogens are taken into account explicitly) is relative accurate, but need a lot time to calculate. The united atom (UA) model (CH2 and CH3 groups are considered as a single unit) can reduce the calculation time. Can I build UA model based on the EA model? Detailedly, CH2 and CH3 groups are considered as a single unit, and only their pair interactions and chagres are readjustmented. Whereas the parameters of bond, bend and torsion are kept the same as in EA model. Is this correct?
Generally this is not the best solution, because interactions of hydrogens are not included in torsions of EA model (angle and bond parameters should not be affected that much). With charges it also gets complicated for polar and h-bond forming compounds (which is probably not the case with you). There are two main alternatives. 1) find a suitable UA model, UA branch of OPLS force-field for example. A lot of UA models of normal alkanes were developed. 2) develop your own coarse-grained model, where large groups of atoms (even maybe several carbons per coarse-grained particle, or the whole aromatic ring as an example) interact with the potentials of mean force, derived via (Boltzmann) inversion of pair correlation functions or mean forces, produced beforehand with EA model. For references you can read papers of Kurt Kremer group. The second way is time-consuming and unavoidably includes EA modelling, fitted to the experimental data (if you have some), and recursive coarse-grained modelling as well. Few months of work at least. But final model will run in about 3...20 times faster, and dynamics itself (i mean diffusivities, relaxation rates etc.) will be in about 100 times faster due to lower friction.
- Badges
- Science topic