We demonstrate how an iterative method for potential inversion from distribution functions developed for simple liquid systems can be generalized to polymer systems. It uses the differences in the potentials of mean force between the distribution functions generated from a guessed potential and the true distribution functions to improve the effective potential successively. The optimization algorithm is very powerful: convergence is reached for every trial function in few iterations. As an extensive test case we coarse-grained an atomistic all-atom model of polyisoprene (PI) using a 13:1 reduction of the degrees of freedom. This procedure was performed for PI solutions as well as for a PI melt. Comparisons of the obtained force fields are drawn. They prove that it is not possible to use a single force field for different concentration regimes.

We present the results of molecular dynamics simulations of very long model polymer chains analyzed by various experimentally relevant techniques. The segment motion of the chains is found to be in very good agreement with the reptation model. We also calculated the plateau modulus G 0 N . The predictions of the entanglement length Ne from G 0 N and from the mean square dispacement of the chain segments disagree by a factor of about 2.2(2), indicating an error in the prefactor in the standard formula for G 0 N . We show that recent neutron spin echo measurements were carried out for chain lengths which are too small to allow for a correct determination of Ne.How an entangled polymer chain moves in a dense melt of other chains has been a long standing problem. The most widely accepted picture is that the chains reptate like a snake [1,2]. For short times chains move isotropically until they feel the constraints of their neighboring chains. For intermediate times, the chain segments move along the path or tube created by the surrounding chains in a Rouse-like [3] motion. Only the ends explore new space. For the inner section of the chain, this Rouse-like motion on a contour which is also a random walk gives rise to the famous t 1/4 power law regime in the mean-square displacement of the beads g 1 (t). For very long time, the motion is diffusive, with a chain diffusion constant D that scales as N −2 for large N , where N is the chain length. For short chains, the motion is much simpler and can be approximately described by the Rouse model and D ∼ N −1 . To characterize the crossover between the Rouse and reptation regime, one can define an entanglement length N e . Within the reptation model, N e can be related to both the tube diameter d T and crossover time τ e from the early time Rouse regime where g 1 (t) ∼ t 1/2 to the t 1/4 regime as well as from the value of the plateau modulus G o N . Over the years there have been a number of experiments [4][5][6] and simulations [7][8][9][10][11] designed to test various aspects of the theory. Recent neutron spin echo (NSE) scattering experiments [6] which measure the dynamic structure factor S(k, t), suggest that N e as measured from d T is consistent with that determined from G 0 N . Our previous simulation results [7,9] suggested an inconsistency in that N e measured from g 1 (t) was about one half that measured from S(k, t), though both g 1 (t) and S(k, t) are single chain quantities and measure the same motion. However since the chains were only a few N e (N ≤ 200), our results were not conclusive. We have now extended our simulations to much longer chains, as long as N = 10000, and measured not only single chain quantities but also G 0 N . We find clear evidence for differences on the order two between N e measured from g 1 (t) or S(k, t) and that from G 0 N . The previous reported agreement in N e determined from NSE data for S(k, t) and G 0 N is shown to be largely due to finite chain length effects in d T as determined from S(k, t).To overcome the long time scale...

We report on self-consistent Polymer Reference Interaction Site Model (PRISM) calculations as well as molecular dynamics (MD) simulations for several types of polyolefins. For all polymer types one single set of potential parameters was used. In general we find good semi-quantitative agreement between PRISM and MD results. Further we compare both MD and PRISM results to experimental X-Ray scattering data and show that the potentials used give a good to excellent description of these data. From the quality of the PRISM calculations it is clear that PRISM can be used as an efficient tool in model development.

The non-linear stress-strain relation for crosslinked polymer networks is studied using molecular dynamics simulations. Previously we demonstrated the importance of trapped entanglements in determining the elastic and relaxational properties of networks. Here we present new results for the stress versus strain for both dry and swollen networks. Models which limit the fluctuations of the network strands like the tube model are shown to describe the stress for both elongation and compression. For swollen networks, the total modulus is found to decrease like (V 0 /V ) 2/3 and goes to the phantom model result only for short strand networks.