Inter-instrument definition of valid criteria for the automatic identification of microplastics by micro-Raman spectroscopy

Abstract: Assessing the environmental and human health risks of microplastic contamination requires a reliable identification of polymer types. µ-Raman spectroscopy is widely used for this purpose by comparing sample spectra with reference databases. However, automatic identification demands well-defined match algorithms and thresholds for this parameter to ensure high true positive rates (TP) with minimal false positives (FP), regardless of spectrometer or acquisition conditions. This study proposes a robust methodology to optimise microplastic identification by µ-Raman spectroscopy using data from different instruments and laboratories. A bootstrap approach was applied to determine the optimal match threshold without assumptions about match distribution. For polyethylene terephthalate (PET) microparticles, the best performance was achieved using Pearson’s correlation coefficient (P5»P = 0.6244), yielding a TP of 95 % and an FP of only 4.9×10⁻⁷ %. Spectra with signal-to-noise ratios below 10 were flagged for manual verification. The proposed method proved effective, providing reliable, reproducible microplastic identification across varying experimental conditions.