Hyperspectral imaging-based strategy for detecting micro-bioplastics in anaerobic digestate matrices.

An efficient analytical strategy based on short-wave infrared hyperspectral imaging (SWIR-HSI: 1000-2500 nm) combined with chemometrics was developed to rapidly and non-destructively identify micro-bioplastics (MBPs) within the anaerobic digestate matrix. Four representative BP objects were selected as sources of MBPs: a coffee capsule (CA) made of poly(lactic acid)/poly(butylene succinate) (PLA/PBS) blend; a cup (CU), composed of a Mater-Bi® commercial formulation based on PLA and PBS; a plate (PL) made of PLA; and a shopper bag (SH), made of a Mater-Bi® blend containing PLA, thermoplastic starch (TPS), and poly(butylene adipate-co-terephthalate) (PBAT). Each object was ground to obtain powdered MBPs in two size fractions (< 0.5 mm and 0.5-1 mm). These MBPs were manually placed in digestate-coated cellulose filters and acquired at two spatial resolutions (150 and 30 μm/pixel), obtaining calibration and validation datasets. Principal Component Analysis (PCA) was performed with explorative purposes to highlight spectral variability of MBPs and the digestate matrix. A hierarchical classifier based on Partial Least Squares-Discriminant Analysis (Hi-PLS-DA) was then developed to identify 5 classes: digestate matrix, CA, CU, PL, and SH. The Hi-PLS-DA model achieved very promising identification results, with sensitivity and specificity values from 0.83 to 1.00 in prediction. Minor local effects, such as spatial resolution limits, MBP transparency, edge effects, and matrix variability, were observed but did not significantly affect the overall identification. These results highlight the potential of the developed HSI-based strategy as a rapid tool for effectively detecting MBPs in digestate matrix, supporting the evaluation of anaerobic digestion product quality.