Atomistic Molecular Dynamics Simulation of Polydisperse Linear Polyethylene Melts

Abstract

Well-relaxed atomistic configurations of polydisperse, linear polyethylene (PE) melts, obtained with the end-bridging Monte Carlo algorithm, have been subjected to detailed molecular dynamics simulations in both the canonical (NVE) and microcanonical (NVT) ensembles. Three different systems have been investigated, characterized by mean molecular lengths C24, C78, and C156, and by the same polydispersity index I of about 1.09. Results are presented for the static and (mainly) dynamic properties of these melts at P = 1 atm and T = 450 K. The diffusion coefficient D, determined for various chain lengths, N, is in very good agreement with experimentally measured values. The friction coefficient ζ D is extracted from D by invoking the Rouse model; it is seen to increase from a relatively small value characteristic of short alkanes to a chain-length-independent plateau, reached in a region of N = 60−80. The friction coefficient ζτ is also obtained by analyzing the decay of the time autocorrelation function for the normal modes X p at various chain lengths; the values thus extracted are consistent with those obtained from D for N above 40. Although the decay of the autocorrelation function of the end-to-end vector is very well described by the Rouse model, individual Rouse modes show some deviation from theoretical predictions. Even for chains sufficiently long to be in the asymptotic ζ regime, only the first two normal modes fully conform to Rouse theory in terms of their squared amplitudes and correlation times. Zero-shear viscosities computed from ζ D values by means of the Rouse model are in excellent agreement with available experimental data for N = 90.

Zero-shear stress relaxation and long time dynamics of a linear polyethylene melt: A test of Rouse theory

“…Simulations of the NVT ensemble were performed using a weak coupling scheme to a temperature bath of 450 K with a coupling constant of 0.1 ps. 16 The density was set to 0.7614 g/cm 3 , close to the normal pressure density at 450 K. 17 The average pressure from the simulation was pϷ54 atm. For a correct sampling of the pressure tensor autocorrelation it is important to have a fully equilibrated melt in which there is no average stress present.…”

“…19 Quite successful in obtaining a well-equilibrated melt is the end-bridging Monte Carlo algorithm, as suggested by Pant and Theodorou, which, however, yields polydisperse melts because of the connectivity-altering moves. 20,3 Van der Vegt et al have investigated the influence of three different ways of generating initial configurations on the solubilities of small molecules in amorphous polymer melts. 21 In one of the methods a dilute system is slowly compressed, during which process only the repulsive parts of the nonbonded Lennard-Jones potentials are taken into account.…”

“…Several other studies have been reported on simulations of melts of polymers of comparable size. [1][2][3][4][5][6][7][8][9] They all suggest that individual properties of these melts can well be described by the Rouse model. 10 In this paper we will test whether the Rouse model can be used to predict the time correlation functions of a comprehensive set of physical processes by using one single set of parameters.…”

“…Similar results were found by Harmandaris et al for chains in polydisperse blends. 3 Mondello and coworkers investigated a number of chain lengths between C 10 and C 66 . 5,6 By use of the Green-Kubo ͑GK͒ relation they were able to calculate the zero-shear viscosity 0 for chains up to C 16 by integrating the zero-shear stress relaxation curves obtained from equilibrium molecular dynamics simulations ͑EMD͒.…”

Chain dynamics of poly(ethylene-alt-propylene) melts by means of coarse-grained simulations based on atomistic molecular dynamics

“…Such simulations have been performed only for medium length chains where large-scale dynamics and rheological properties seem to be well described in terms of the simple Rouse model. [5][6][7][8][9] In order to simulate dynamical and rheological behavior of long chains, one has to resort to coarse-grained models. Many different coarse-grained models have been reported in the literature, 10 being perhaps the most popular one that of bead-spring type, where the polymer chain is represented as a sequence of beads connected by springs.…”