Increasing Recycling Quality by Sensor-based Sorting of Plastic Waste according to Property Profiles

To increase the use of secondary materials as substitutes for primary materials ¿ in the sense of a circular economy ¿ functionalities must be maintained during recycling. This dissertation investigated how this can be implemented for plastic packaging waste: First, it was determined whether sufficient quantities are available spatially and temporally for application-specific recycling (e.g. for food contact) as this is a basic economic prerequisite for such recycling routes. For this purpose, a sampling campaign was conducted in Austria, during which ~1,100 kg of light packaging waste was characterized at particle level (55,197 particles) regarding recycling-relevant properties. The resulting dataset was subsequently evaluated. The core finding was that the quantities collected in Austria for application-specific recycling were consistently available throughout the year. Subsequently, the investigation focused on polypropylene, which is often used for packaging due to its low-cost monomer and ease of modification. First, variability in rigid polypropylene packaging was investigated by analyzing technical data sheets of 113 virgin material grades and 20 selected post-consumer packages in the laboratory. The results revealed significant variations in technical properties, e.g. in post-consumer packaging in terms of impact strength (4¿10 kJ m-2), tensile modulus (800¿1,990 MPa), melt flow rate (2¿90 g 10 min-1), and the presence of 143 chemical substances and multilayer designs. Combining these results with the visual appearance of post-consumer packaging (e.g., processing method, application) allow for relating technical properties with visual appearance (e.g., chemical contamination of food packaging). Afterwards, sensor-based sorting strategies were investigated. By sorting 2,240 kg of polypropylene post-consumer packaging according to opacity/translucency, recyclates with different property profiles could be obtained. Since clear packaging contains a higher proportion of random copolymers, the clear recyclate had lower stiffness (1,154 MPa vs. 1,424 MPa). Additionally, the higher proportion of blow-molded and thermoformed articles in the clear recyclate reduced its melt flow rate (9.1 g 10 min-1 vs. 17.2 g 10 min-1). Finally, the relationship between the technical property of melt flow rate and the chemistry of post-consumer polypropylene packaging was investigated on a laboratory scale. Near-infrared spectra acquired through hyperspectral imaging were used to train regression models. The best results were achieved with white polypropylene, with an R² of 0.85 and an RMSE of 12.4 g 10 min-1. Subsequent binary classification at different thresholds (6, 12, 30, and 60 g 10 min-1) achieved balanced accuracies ranging from 0.82 to 0.92. These results demonstrate the effectiveness of property profiles in improving the quality of plastic recycling. Detailed material characterization at the waste and recyclate levels enables further development of sensor-based sorting and estimation of quantity potential. Achieving a circular economy of plastics requires interdisciplinary expertise in materials science and recycling technology, as well as communication between actors throughout the entire value chain.