Enhanced Catalytic Depolymerization of Mixed Polyethylene Terephthalate (PET) Waste using Dynamic Solvent-Assisted Supercritical CO2 Extraction and Machine Learning Predictive Control

**Abstract:** This paper presents a novel approach to the complete depolymerization of mixed PET waste streams, incorporating dynamic solvent-assisted supercritical carbon dioxide (scCO2) extraction and machine learning (ML) predictive control for optimized catalyst performance and monomer recovery. Conventional PET depolymerization struggles with mixed feedstocks and incomplete reactions. We address this limitations through a feedback-controlled process that monitors reaction progress using Raman Spectroscopy and dynamically adjusts extraction solvent ratios and scCO2 conditions to maximize monomer yield and minimize byproduct formation. The result represents a pathway toward a cost-effective and scalable process for circular PET recycling, converting waste into valuable raw materials.