(204g) Machine Learning Aided Ionic Liquid Design for Polyethylene Terephthalate Glycolysis

Every year, more than 350 million tons of plastic waste are generated globally, but only a small percentage (9%) is recycled. The rest of this non-degradable waste ends up in landfills and causes water and land pollution. To address this pressing issue, there is an urgent need to upcycle plastic waste into value-added bioproducts. In this study, we address the multi-dimensional challenges of ionic liquid (IL)-based polyethylene terephthalate (PET) glycolysis by harnessing the power of ML integrated with a process simulation model. This approach allows for simultaneous catalyst design and reaction condition optimization that eliminates the need to individually tune each reaction parameter. Through optimization towards tailored performance indicators, our integrated model enables a smooth translation from lab to process for an industrial oriented design. Our model features a graph neural network that predicts reaction yield based on different reaction parameters and IL structures. This information is then incorporated into a process model that simulates real-world production scenarios to estimate cost and carbon emissions associated with the glycolysis process. Guided by the combined economic and environmental performance indicators, we can identify the ideal catalyst and reaction conditions for a more efficient, economic, and eco-friendly glycolysis process. Based on computational results, we have identified several promising sets of ILs and reaction conditions for validation. These ILs demonstrate exceptional PET glycolysis performance in terms of yield, cost, and carbon emission.