NIR Hyperspectral Imaging for Advanced Identification and Quantification of Materials in PV Recycling Fractions

Abstract

An advanced, layer-by-layer separation and recycling process for end-of-life (EoL) photovoltaic (PV) modules is developed, aiming for a material recovery ratio of over 95% by weight. The process includes input characterization, component separation, further processing, output characterization, and recycling of separated fractions. Three material groups have the potential to be recycled and reused in the sense of a circular economy: (i) front glass, (ii) metals and semi-conductors (Si) of the solar cell and connectors, and (iii) plastics from the backsheet. The major challenge is the separation of the insoluble, crosslinked encapsulant polymer (mostly EVA) from the other, reusable materials. To support non-destructive material identification and classification of the separated fractions, near-infrared (NIR) hyperspectral imaging (HSI) is incorporated into the recycling workflow. HSI enables the detection of residual encapsulant adhesions on glass, solar cell, or backsheet parts and supports the spatial mapping of materials. Machine learning algorithms process spectral data in real time, improving classification accuracy and recycling efficiency. Initial results show that HSI accurately identifies and quantifies materials in mixed streams. This approach increases material purity, supports scalable recycling workflows, and contributes to a circular PV economy by enabling the reuse of valuable components and minimizing waste.