Plastic waste has become a major environmental crisis,
with the
majority of plastic being produced ending up in open landfills and
waterways every year. Solvent-based recycling approaches offer an
effective means of recovering high-quality plastic materials from
waste by the use of a solvent to selectively dissolve the plastic
waste and recover specific polymers. In this work, we report on the
properties of 9587 potential glycerol-based solvents that can be synthesized
from biomass-derived glycerol. We predict the density and dipole moment
using quantum mechanical calculations, while LogS, LogP, and melting
point are predicted using machine learning that outperforms other
prediction methods such as Hansen Solubility Parameters in Practice
(HSPiP). Additionally, we analyze the ability of the solvents to dissolve
common plastic materials [polyethylene (PE), poly(ethylene terephthalate)
(PET), poly(ether sulfone) (PES), polypropylene (PP), polystyrene
(PS), and poly(vinyl chloride) (PVC)] based on a comparison of their
Hansen solubility parameters (HSPs). Our results show that functionalization
of glycerol can significantly alter its properties, and based on the
HSPs and melting point, we recommend selective solvents for PE, PET,
and PVC, while for PES, PP, and PS, we suggest using a combination
of solvents in a solvent/antisolvent setup for solvent-based plastic
waste recycling. Finally, based on stricter solvent selection criteria,
we also propose a strategy that may help reduce the costs of sorting
waste plastic whereby the waste feedstock is first separated into
polar and nonpolar fractions.