Abstract Scope |
Accurately describing reactive events over length and time scales relevant to composite materials remains a grand challenge of computational materials science. REACTER is a general protocol for modeling chemical reactions using classical force fields, and is implemented in the popular molecular dynamics software LAMMPS. REACTER has a growing user base and has been used as a model-building tool for a variety of materials, including thermoplastics, thermosets, glassy materials and composites. This work seeks to increase the accuracy with which REACTER models polymerization reactions in confined settings, by utilizing a committor function that specifies the probability of a reaction occurring on the basis of the local atomic configuration. The committor function is a useful mathematical tool for modeling rare events, but is difficult to compute for chemical reactions in a general way. This work describes a method for approximating the committor function using a machine learning approach, specifically a deep neural network trained with data from DFT-based dynamics simulations. This machine-learned model is coupled to the existing REACTER protocol, as implemented in the LAMMPS MD package, and used to make on-the-fly predictions of reaction probabilities without the more extensive user input previously required. The new method is demonstrated using the polymerization of polystyrene and a benzoxazine as case studies. The cationic ring-opening polymerization of bis-benzoxazine, a promising resin for use in advanced high-temperature application composites, is characterized under nanoscale confinement and compared to its polymerization behavior and resulting morphology in non-confined bulk systems. |