(587c) Simulation of Polymer Crystal Nucleation at Shallow Undercoolings

Understanding crystallization of polymers is important due to the ubiquity of semi-crystalline polymer products in our everyday lives and their industrial manufacturing. The earliest stages of this crystallization involve polymer crystal nucleation, which occurs at a small spatio-temporal scale. This scale makes experimental characterization of nucleation difficult. Thus, molecular dynamics simulations have become a popular way of studying and understanding polymer nucleation. These simulations are often carried out well below the melting point temperature (at a deep undercooling) in order to reduce their computational cost. On the other hand, industrial processes often involve shallow undercoolings. Thus, extending the range of temperatures over which polymer crystal nucleation can be studied, including temperatures close to the melting point, becomes necessary.

In order to mitigate the computational cost of molecular dynamics, especially at shallow undercoolings we implement an advanced path sampling method known as forward flux sampling (FFS). We use this method to simulate crystal nucleation in linear polyethene at temperatures that would be practically inaccessible to brute-force MD simulations. We calculate nucleation rates from these simulations and compare them across temperatures. The FFS method involves various parameters such as sampling frequency, interface number and positions. We vary these parameters to check the sensitivity of our results and choose the right ones for our system. Since FFS does not involve a biasing potential, we get realistic trajectories from our simulations. From these trajectories and the results of FFS, we calculate important properties of crystal nuclei such as size of the critical nucleus. We compare these results across the range of temperatures simulated. This work shows the feasibility of speeding up MD simulations of polymer crystal nucleation using FFS. It can be extended to simulate other cases where crystal nucleation might be too infrequent to observe using brute force MD.