Regenerable membrane sensors for ultrasensitive nanoplastic quantification enabled by a data-driven Raman spectral processing algorithm

The detection of nanoplastics (NPs) in complex natural water systems is hindered by matrix interferences and limitations in current analytical techniques. This study presents Pre_seg, a Raman spectral processing algorithm integrated with regenerable anodic aluminum oxide (AAO) membrane sensors, for ultrasensitive, rapid, and quantitative NP detection at the single-particle level. The AAO membranes function as both filtration substrates and Raman sensors, reducing sample loss and contamination. Pre_seg incorporates statistically determined thresholds for signal-to-noise ratios (SNRs) and full width at half maximums (fwhms) across segmented spectral ranges, effectively minimizing noise and enhancing accuracy and sensitivity of NP detection. Pre_seg achieved 93.5% prediction accuracy of NPs and ≥90.4% rejection accuracy for non-NP entries. Mixed NPs were quantified at the lowest concentration of 0.5 μg L^–1. The robustness of Pre_seg was validated in eutrophic and oligotrophic lake matrices following oxidation digestion pretreatment to mitigate organic interferences. Furthermore, the AAO membrane sensors demonstrated stability through multiple regeneration and reuse cycles. This innovative approach advances NP detection by enabling scalable, customizable, and environmentally relevant monitoring.