Sample-based subsampling strategies to identify microplastics in the presence of a high number of particles using quantum-cascade laser-based infrared imaging

Microplastics (MPs) are ubiquitous in all ecosystems, affecting wildlife and, ultimately, human health. The complexity of natural samples plus the unspecificity of their treatments to isolate polymers renders the characterization of thousands of particles impractical for environmental monitoring using conventional spectroscopic techniques. Two primary solutions are to analyze a small fraction of the sample or to measure only a subset of particles present over a holder, known as subsampling. A strategy to subsample reflective Kevley slides and gold-coated filters using quantum-cascade laser-based infrared imaging is proposed here, as this technology is a promising tool for MPs monitoring. In contrast to most previous approaches that struggle to propose general subsampling schemes, we introduce the concept of sample-based subsampling. This can be applied ex-ante always and it highlights the best subsampling areas for a sample after a preliminary assay to count the total number of particles on a holder. The error at this stage acts as a proxy to minimize errors when evaluating the number of particles and MPs, significantly enhancing the feasibility of large-scale MPs monitoring. The predictive ability of the approach was tested for fibres and fragments, for total amounts of particles and MPs. Further, the evaluations were disaggregated by size and polymer type. In most situations the reference values were contained in the confidence intervals of the predicted values (often within the 68 % ones) and relative errors were lower than 25 %. Exceptions occurred when very scarce (one or two) items of a given size or polymer were present on the overall holder. The approach was compared to other systematic ad-hoc strategies.