We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Structure–Property Relationships for Biodegradability in Copolyesters
Original title: Structure–Property Relationships for Biodegradability in Copolyesters
Summary
Scientists tested 300 combinations of plastic-like materials (copolyesters) and found that mixing different building blocks makes plastics break down naturally far more often than using single ingredients alone—over 80% of the mixed versions were biodegradable, compared to just 50% of the single-ingredient versions. This matters because it points toward practical ways to design plastics, like packaging or bottles, that break down instead of lingering in the environment and breaking into microplastics that can end up in our water, food, and bodies.
With growing concerns about increasing plastic pollution, interest in biodegradable polymers, particularly polyesters, continues to grow. Copolymerization is an important molecular handle to tune properties to achieve material performance and biodegradation simultaneously. To better understand the structure–property relationships that govern biodegradability, a 300-member copolymer library was synthesized and tested using a high-throughput clear-zone biodegradation assay; testing shows over 80% of the copolymer library is biodegradable even though only 50% of the homopolymers from which they are derived are degradable. Repeat units with longer carbon backbones decreased biodegradability, though copolymers showed biodegradability at all average repeat unit lengths examined. For both homopolymers and copolymers, oxygen substitution of backbone carbons was established as a lever to improve biodegradability. A novel chemical similarity-informed embedding that considers polymer chemical structure and composition was developed and implemented for the simultaneous quantitative structure–property relationship modeling of homopolymers and copolymers. Random forest models could simultaneously capture homopolymer and copolymer behavior with 78% and 95% accuracy, respectively; however, models trained only on homopolymer data could not predict copolymer biodegradability. Unlike random forest models, linear models were not able to capture both homopolymer and copolymer biodegradability.