Because of their diverse physical and mechanical properties, polymers—such as polyethylene—have become essential materials in a variety of defense systems as well as medical, industrial, and environmental applications. Typically, polymers are composed of both crystalline and amorphous regions, and their properties are strongly influenced by the size and shape of the crystalline regions. Understanding how and why these regions form will help us control these properties, leading to new applications and improvements to existing materials.
However, the atomistic processes that lead to polymer crystallization are not well formulated, and experiments have not given us the level of detail we need. Using LLNL’s terascale computational capabilities, we are able to further our understanding of these processes on an atomistic level.
By increasing our understanding of polymer crystallization, we are gaining fundamental knowledge about the nature of crystal formation and, more specifically, the aging of technologically important materials in a variety of defense systems. The polymer materials in defense systems such as nuclear weapons must perform within predictable margins, and we are turning increasingly to simulations to ensure that these materials will perform as required.
In our polymer crystallization investigations, we have focused on examining high-explosive aging phenomena relevant to the Stockpile Stewardship Program’s Enhanced Surveillance Campaign. Our goal is to obtain, through accurate models, a lifetime assessment of physical and mechanical properties of polymers relevant to their use in national defense systems. In a broader sense, the results of our work can also be applied in atomic-scale control of the physical and mechanical properties of polymers.
|Figure 1. In the largest molecular dynamics simulations of polymer crystallization to date, our models show the evolution of the spinodal-assisted crystallization process for the polar (orange atoms) and nonpolar (grey atoms) polymers. The crystalline regions are explicitly shown, while the amorphous domains are shown as the white space in each panel.|
Under some circumstances, polymers nucleate almost instantly, in contrast to classical nucleation, where a crystal seed is formed over time. Using the LLNL 23-TFLOP/s Thunder supercomputer and other computational resources, we conducted the largest molecular dynamics simulations of polymer crystallization to date to test recent experimental findings and theoretical predictions that, when rapidly cooled, polymeric materials undergo instant crystallization (known as spinodally assisted nucleation). Understanding spinodally assisted nucleation at the atomistic level is the first step to elucidating the pretransition state of polymer crystallization.
We considered two classes of polymers: polar polymers (polyvinylidene fluoride, or PVDF) and nonpolar polymers (polyethylene, or PE). Both classes are made up of united atoms in chains, similar to a necklace of beads. We put these chains in a simulation cell and equilibrated them at several temperatures before we observed the almost instant chain ordering.
|Figure 2. The first panel (a) shows a single ordered polymer domain with a morphology for the polar polymer model. Individual polymers are colored to illustrate polymer chains entering the amorphous region from an ordered region and adjoining two separate adjacent ordered domains (adjacent domains are not shown). The second panel (b) shows a representative ensemble of ordered domains.|
To address questions about the atomistic mechanism responsible for this instant ordering, our simulations used the length scale of 10 to 20 nm—the range at which spinodally assisted nucleation occurs. Using simulation cells twice the size of our length scale, our results provided atomistic information of polymer nucleation with unprecedented resolutions. (For example, the number of beads in one nonpolar melt was five million, and the size of the simulation cell was approximately 50 nm.)
Confirming earlier experimental findings, our simulations revealed microphase separation of bulk amorphous polymer ensembles into many ordered domains (Figure 1). These domains coalesce and grow, initially forming a small number of crystalline regions that grow to occupy most of the simulation volume at the end of the runs. Figure 2a shows an enlarged ordered domain isolated from the sample, which reveals an interface between oriented and unoriented domains. Ordered crystalline domains are circled in Figure 2b to demonstrate the complex nature of the polymer morphology.
As the first atomistic observation of the debated prenucleation, this work provides unambiguous insight into the physics of the early stages of polymer ordering leading to crystallization. Furthermore, with the unprecedented resolutions and details of our simulations, it was possible to compare and confirm our findings with laboratory experiments. Our results pertaining to crystallite size and structure as well as the critical length of polymer segments provide crucial information needed to form a fundamental understanding of the structure–property relationships of semicrystalline polymers.
R. H. Gee et al., “Atomistic Simulations of Spinodal Phase Separation Preceding Polymer Crystallization,” Nat. Mater. 5, 39 (2006).
R. H. Gee et al., “Ultrafast Crystallization of Polar Polymer Melts,” J. Chem. Phys. 118, 3827 (2003).