Title: Mechanical Properties of Glassy Polymer Blends and Thermosets

Presenter: David Rigby

Authors: D. Rigby, B. Leblanc, C.M. Freeman and P.W. Saxe

Affiliation: Materials Design, Inc., Angel Fire, NM, USA

Abstract: Knowledge of the elastic constants of glassy engineering polymers and thermosets is important in a variety of aerospace, automotive and related engineering applications. A number of attempts at using atomistic level simulations were made beginning in the mid 1980's 1 extending through the mid 1990's 2. However, widespread use of simulation to perform mechanical property calculation of these materials was impeded by a number of factors such as the limited availability of high accuracy force fields and by the fact that significant computational resource is required to effectively sample the distribution of molecular packing found in amorphous polymeric systems, either through averaging the computed elastic constants of many smaller models containing perhaps a few thousand atoms, or alternatively through use of much larger models. Furthermore, irrespective of the system size and number of independent models studied there was a lack of attention directed towards relating elastic constant data determined on collections of nanoscopic domains to precise predictions of the elastic constants of the corresponding bulk macroscopic material.

A significant contribution to this data analysis problem was made by Suter and Eichinger in their 2002 paper 3 addressing the question of how to effectively analyze elastic constants computed for multiple independent simulated models using the methods developed earlier by Hill and Walpole for composite materials.

In view of the improvements in force field quality, and development of simulation codes such as LAMMPS 4 designed for use on fast parallel computational hardware, we have recently attempted to reassess the question of the precision and accuracy with which it is now possible to perform relatively routine predictions of elastic constants of glassy amorphous polymers. This presentation will illustrate two aspects of the recent research, first addressing the question of precision and accuracy by exploring whether simulation is capable of predicting the relatively small changes in elastic moduli of the well-known polystyrene-poly(phenylene oxide) miscible blend system that occur as the blend composition is varied. In addition to comparing the predictions with experiment, we examine the amount of sampling of individual independent configurations required to obtain predictions with the necessary precision. The second study then continues by focusing on elastic constants of thermosetting epoxy resins, created using the MedeA® simulation environment 5, and based on reactants with differing chemical functionalities, which have been shown experimentally to possess quite different moduli. The extent to which these differences can be reproduced qualtitatively will be discussed.

References:

1 Theodorou, D.N. and Suter, U.W., Macromolecules 9, 139 (1986).

2 see, for example, Hutnik, M, Argon, A.S. and Suter, U.W., Macromolecules 24, 5956 (1991); Fan, C.F., Cagin, T., Chen, Z.M. and Smith, K.A. Macromolecules 27, 2383 (1994).

3 Suter, U.W. and Eichinger, B.E., Polymer 43, 575 (2002).

4 Plimpton, S., J. Comp. Phys. 117, 1 (1995); http://lammps.sandia.gov

5 MedeA®, Materials Design, Inc., Angel Fire, New Mexico, USA; http://www.materialsdesign.com