The NSMMS & CRASTE Symposium

Optimizing energetic binder formulations for additive manufacturing using physics-based modeling and machine learning

Speaker:
David Nicholson, Principal Scientist I, Schrödinger

Abstract:
Plasticizers are critical components of energetic binder formulations, lending processability and flexibility to materials that would be otherwise inadequate. The identification of a good pairing between polymer and plasticizer is a formulation design problem that is challenging to solve via brute-force experimentation. Computational approaches, including ML and physics-based modeling, provide a more direct pathway to the desired material characteristics. This type of approach is especially valuable in designing materials for novel applications where identification of suitable materials is less mature and the design space is broad. In this study, we started from a design space consisting of 10 acrylate-terminated polymers and 10 energetic plasticizers and identified an optimal two-component formulation for additive manufacturing applications based on criteria for compatibility and thermomechanical properties. We utilized molecular dynamics (MD) simulations to perform an initial screening for solubility parameter differences to eliminate over half of the plasticizer-polymer pairs. For the remaining pairs, as well as the pure components, we used additional MD simulations to characterize low-temperature modulus and glass transition temperature. These simulation results were subsequently used to train formulation machine learning models for these two properties, and further utilized to identify a set of top-performing formulations using optimization. The properties of top formulations were verified using MD simulations.