Publications

Previous applications of DF theory required a single chain Monte Carlo simulation to be performed within a self-consistent loop. In the current work, a methodology is developed which permits the simulation to be taken out of the iterative loop. Consequently, the calculation of the self-consistent, medium-induced-potential, or field, is decoupled from the simulation. This approach permits different densities, different forms of U{sub M}(r), and different wall-polymer interactions to be investigated from a single Monte Carlo simulation. The increase in computational efficiency is immense.

Molecular dynamics (MD) simulations were performed on dense liquids of polyethylene chains of 24 and 66 united atom CH{sub 2} units. A series of models was studied ranging in atomistic detail from coarse-grained, freely-jointed, tangent site chains to realistic, overlapping site models subjected to bond angle restrictions and torsional potentials. These same models were also treated with the self-consistent, polymer reference interaction site model (PRISM) theory. The intramolecular and total structure factors, as well as, the intermolecular radial distribution functions g(r) and direct correlation functions C(r) were obtained from theory and simulation. Angular correlation functions were also simulation obtained from the MD simulations. Comparisons between theory and reveal that PRISM theory works well for computing the intermolecular structure of coarse-grained chain models, but systematically underpredicts the extent of intermolecular packing as more atomistic details are introduced into the model. A consequence of g(r) having insufficient structure is that the theory yields an isothermal compressibility that progressively becomes larger, relative to the simulations, as overlapping the PRISM sites and angular restrictions are introduced into the model. We found that theory could be considerably improved by adding a tail function to C(r) beyond the effective hard core diameter. The range of this tail function was determined by requiring the theory to yield the correct compressibility.

The ability to model the packing of polymers in melts and blends is important in many polymer applications. One significant application is the development of new polymer blends. It would be exceedingly helpful to the materials chemist if molecular modeling could be employed to predict the thermodynamics and phase behavior of hypothetical polymer alloys before embarking on a time consuming and expensive synthesis program. The well known Flory-Huggins theory has been remarkably successful in describing many aspects of polymer mixing from a qualitative point of view. This theory is known, however, to suffer from several deficiencies which can be traceable to the fact that: (1) it is a lattice model requiring both monomer components to have the same volume; and (2) a mean field or random mixing approximation is made which effectively ignores chain connectivity. Because of these limitations the Flory-Huggins theory does not include packing effects and cannot be used to make quantitative molecular engineering calculations. Recently Curro and Schweizer developed a new approach for treating polymer liquids and mixtures which the authors call PRISM theory. This is an extension to polymers of the Reference Interaction Site Model (RISM Theory) developed by Chandler and Andersen to describe the statistical mechanics of small molecule liquids. The PRISM theory is a continuous space description of a polymer liquid, which includes chain connectivity and nonrandom mixing effects in a computationally tractable manner. The primary output from PRISM calculations is the average structure or packing of the amorphous liquid given by the radial distribution function denoted as g(r). This radial distribution function is employed to deduce thermodynamic or structural properties of interest. Here, the authors describe the theoretical approach and demonstrate its application to polyethylene, isotactic polypropylene, syndiotactic polypropylene, and polyisobutylene liquids and blends.

This report summarizes the technical progress made in the past three years on CRADA No. 1078, Molecular Engineering of Polymer Alloys. The thrust of this CRADA was to start with the basic ideas of PRISM theory and develop it to the point where it could be applied to modeling of polymer alloys. In this program, BIOSYM, Sandia and the University of Illinois worked jointly to develop the theoretical techniques and numerical formalisms necessary to implement the theoretical ideas into commercial software aimed at molecular engineering of polymer alloys. This CRADA focused on developing the techniques required to make the transition from theory to practice. These techniques were then used by BIOSYM to incorporate PRISM theory and other new developments into their commercial software.

We have recently developed a new theoretical approach to the study of polymer liquids. The theory is based on the reference interaction site model'' (RISM theory) of Chandler and Andersen, which has been successful in describing the structure of small molecule liquids. We have recently extended our polymer RISM theory to the case of polymer blends. In the present investigation we have applied this theory to two special binary blends: (1) the athermal mixture where we isolate structural effects, and (2) the isotopic mixture in which structurally identical polymer chains interact with dissimilar attractive interactions. By studying these two special cases we are able to obtain insights into the molecular factors which control the miscibility in polymer mixtures. 18 refs., 2 figs.