Computational Approach for Predicting Temperature-Dependent Polymer Solubilities

Billions of pounds of plastic waste are produced in the United States every year, with much of it being “multilayer” plastics, or plastics composed of multiple distinct polymers. A large portion of multilayer plastic waste in manufacturing facilities is clean, unused scraps that do not make it out of the doors of the facility. Currently, there are no widely-practiced techniques to separate multilayer plastics back into their respective polymer resins, which would allow plastic scraps to be recycled back into the production stream. To address this challenge, a new strategy to chemically recycle multilayer plastics, named the solvent-targeted recovery and precipitation (STRAP) process, is being developed by the Center for the Chemical Upcycling of Waste Plastics at UW-Madison. In the STRAP process, a series of solvent washes is used to selectively dissolve each polymer composing a given multilayer plastic, resulting in separate polymer resins that are chemically similar to the virgin resins. A computational solubility database was thus developed to identify appropriate solvents, dissolution temperatures, etc. for a multilayer plastic given its composition. This project aimed to add another commercial polymer to the solubility database: polyurethane. A short oligomer of a common polyurethane molecule was constructed and modeled in a molecular dynamics simulation to obtain information about its conformations. A sample set of conformers were input into the conductor-like screening model for real solvents (COSMO-RS) to formulate solubility predictions. The resulting data set involved the solubility values for our polyurethane in over 1000 common solvents. At low temperatures, our polyurethane was insoluble in most of the solvents in our database. However, the solubility of our polyurethane increased with increasing temperature - specifically, it became slightly more soluble in dimethyl sulfoxide, acetylacetone, and N,N-dimethylformamide. Recommended next steps for this project include the experimental verification of our results in laboratory solvents, and modeling differently-composed polyurethanes.