Microplastics identification by Infrared spectroscopy – Evaluation of identification criteria and uncertainty by the Bootstrap method

Abstract The assessment of microplastic contamination in an environmental compartment involves identifying and counting microplastics in a representative fraction of the compartment. Microplastics can be identified by μFTIR spectroscopy where spectra are manually examined for characteristic polymer bands or by an automatic comparison of particle spectrum with reference spectra of polymers. The automatic spectra comparison can involve calculating a correlation coefficient, CC, between particle and reference spectra where a minimum correlation above which identification is adequately reliable should be defined. Correlation can be calculated from original or transformed signals, such as taking the first derivative, and by using unweighted or weighted CC. Weighted CC can highlight the spectral features more relevant to distinguish polymers. This work describes a methodology for setting the minimum CC, P5»P, associated with a true positive result rate, TP, of 95% and for checking if this threshold allows identifications with a false positive result rate, FP, not greater than 5%. This methodology was successfully applied to the use of various CC determined from original or transformed spectra for the identification of polyethylene, PE, and polypropylene, PP, microplastics in river sediments by μFTIR. The analytical portions of sediments were digested with H2O2 and microplastics separated from the remaining particles by density using a saturated NaCl solution. Pearson’s, Spearman’s and Alternative unweighted and weighted correlation coefficients were studied. The P5»P was estimated by the Bootstrap method that resamples spectra CC between a reference material and microparticle of the same polymer collected from the environment. This resampling allows simulating CC distribution required to estimate its 5th percentile (i.e. P5»P). The FP was estimated from the probability of a particle not from the same polymer type of the reference material producing a CC greater than P5»P. Some unweighted and weighted CC determined from original or transformed spectra were successfully used to identify PE or PP particles in river sediments. More particle spectra need to be collected to ensure performance is assessed from a representative diversity of aged polymers with different additives. The spreadsheets used for CC calculations and Bootstrap simulations are made available and can be used for the validation of the identification of other polymer types by μFTIR or ATR-FTIR spectroscopy.