Abstract
The presence of polyethylene (PE) in recycled polypropylene (rPP) is a critical quality concern, as the two polymers are inherently incompatible and PE contamination can degrade rPP mechanical performance. Building on our previous development of an AI-assisted partial least squares (PLS) predictive model using laboratory-prepared mixtures, the present study evaluates the model’s performance on real-world materials. Commercial rPP feedstock and finished product samples were obtained from a local plastic recycler, and their Raman spectra were collected and preprocessed through despiking, smoothing, baseline correction, and normalization. When applied to these spectra, the PLS model successfully predicted PE contamination levels in the commercial samples. The finished rPP products were estimated to contain approximately 7-8% PE; meanwhile, one PE piece might be detected in the tested feedstock (Sample# 12). These results demonstrate the model’s practical utility for rapid, non-destructive monitoring of PE contamination in industrial recycling streams.
Recommended Citation
Chatfield, Caleb; Mancos, Nathan; Le Bras, Julie; Higdon, Chloe; and Liu, Shaoyang
(2026)
"Determination of Polyethylene Contamination Level in Commercially recycled polypropylene with Raman Spectrometry Using AI-Assisted Modeling,"
Journal of the Alabama Academy of Sciences: Vol. 97:
No.
1, Article 88.
Available at:
https://jagworks.southalabama.edu/jaas/vol97/iss1/88
