May 18, 2026 15:38

In Germany, the SpectralAIge research project is developing a new method based on spectroscopy and artificial intelligence (AI) algorithms to more reliably assess the aging and degradation of plastics subjected to multiple recycling cycles, which can significantly impair the performance of finished products.

At present, the researchers say, there is no industry-ready method for reliably assessing the degradation status of plastic waste and recycled materials. It is known that quality declines over multiple recycling cycles, but so far this has not been reliably predictable.

The SpectralAIge project combines hyperspectral imaging (HSI) systems, integrated with LED light sources for targeted fluorescence excitation, to record the fluorescence spectra of aged samples of common plastics such as PE, PP and PET.

Based on the measurement data, both chemometric models for polymer analysis and artificial intelligence (AI) approaches will be developed and then systematically compared. The most effective modelling approach will then be transferred into a software application enabling the automated determination of process-induced aging.

Since HSI technology offers high performance but is also costly, the German researchers say, low-cost inverse spectrometers will also be developed and used. These are designed for specific use cases with a limited scope, for example to assess PET in bottle loops managed through deposit-return systems.

The SpectralAIge project involves the SKZ Plastics Center in Würzburg, the Fraunhofer Institute for Factory Operation and Automation IFF in Magdeburg, and the German companies Silicann Systems and HAIP Solutions.

“The findings from the project make a valuable contribution to guaranteeing high product quality and safety in a circular economy,” comments Cosima Güttler, scientist at the SKZ Plastics Center. “At the same time, in SpectralAIge we also focus on the practical, industrial implementation of the results.”

© Polimerica -  Reproduction prohibited, all rights reserved