Microplastics of the LABPLAS Project samples have been analyzed by different partners using different detection and identification techniques. Atmospheric, surface microlayer, and biota samples (except fish) were analyzed by UDC by QCL-LDIR using Agilent’s Clarity software, which is very fast. In addition, the sediment samples were analyzed by BfG and IOW applying the same Hit Quality Index (HQI) for identification ( >0.85 or >0.9 and for tyres >0.95) to make the analysis automatic and reduce overestimates as the HQI value can influence the number of particles, their size, and type of polymer identified. With these high correlations the identification usually matches well; particles with HQI 0.85-0.9 are checked manually for any inconsistencies, and lower values are discarded.
The manta net samples were analyzed at Geomar by HySpec and subsurface water and biota samples were identified and characterized (size, mass) by NOC using a freeware data analysis tool siMPle (https://simple-plastics.eu/; Primpke et al. 2020) and applying HQI of >0.70. The siMPle software is widely used in the microplastics research community as it is free of charge, allows sample analysis independently from the instrument manufacturer, and significantly reduces the analysis time (Primpke et al. 2020). The siMPLe reference spectra library is limited to 32 polymers (currently, 6 natural polymers and 26 artificial polymers/plastics). This software can thus be intrinsically prone to underestimating the number and the diversity of polymers compared to the alternative semi-automated approaches (here, Pabortsava and Lampitt 2020) utilizing larger libraries (e.g. 19,803 polymer types; spectra database from S.T. Japan-Europe GmbH, Germany/Japan) and a different identification framework which accounts for the alteration of chemical signature whilst in the environment/collected sample.
Such underestimation was evident from the inability of siMPle to detect the variations of acrylates in Mero Barces samples, e.g. Modacrylic polymer (Fig 1) or PTFE (Fig 2; green particle), which are both not in siMPle reference spectra library (the former however could come under a class of ‘Acrylates\Urethanes\Varnishes in simple).
Figure 1: Comparison of the methods for detection and identification of microplastics in the Mero Barćes samples. A. Visible image of the final particle sample deposited onto a silver filter; B) output from siMPle software; C) output from the PerkinElmer software following method in Pabortsava and Lampitt (2020). In all, the imaged area (depicted square) is 11 mm x 11 mm. In C) Green particles are the modacrylic microplastics not detected by siMPle.
Systematic underestimation of the microplastics counts was observed in all LABPLAS subsurface water samples (Figure 3) and included polymer types that were present in the siMPle reference spectra library, such as polyester fibers, polyethylene, and polypropylene (Figures 2 and 3). The best agreement between the two identification/quantification methods was achieved for polypropylene microplastics, where underestimation by siMPle was at most 4 times compared to the method adapted from Pabortsava and Lampitt 2020 (Figure 3) and HQI criteria of >0.80.
Figure 2: Comparison of the methods for detection and identification of microplastics in the Mero Barćes samples. A. Visible image of the final particle sample deposited onto a silver filter; B) output from siMPle software; C) output from the PerkinElmer software following method in Pabortsava and Lampitt (2020). In all, the imaged area (depicted square) is 11 mm x 11 mm. In A) the presence of the fibres is evident from the visible microscope image (see magnified region outlined in yellow), yet, they were not detected by siMPle (B).
Figure 3: Underestimation of microplastic counts in LABPLAS samples (Mero Barces (n=23), Thames (n=6), North Sea (n=7). Elbe (n=6)) when using siMPle software compared to the identification/quantification method in Pabortsava and Lampitt 2020.
If siMPle is to be used as a ‘go-to’ software for microplastic quantification and characterization, its spectral library should be significantly expanded and also include weathered/altered polymer spectra. Alternatively, but not mutually exclusive, the detection-identification algorithms in siMPle should be able to account for the weathered polymers in the samples, which is currently possible only through manual expert analysis. Before such capability becomes available, we recommend employing manual expert analysis of a subset of samples to QA/QC the performance of the siMPle software.
Cited literature:
1. Primpke S, Cross RK, Mintenig SM, et al. Toward the Systematic Identification of Microplastics in the Environment: Evaluation of a New Independent Software Tool (siMPle) for Spectroscopic Analysis. Applied Spectroscopy. 2020;74(9):1127-1138. doi:10.1177/0003702820917760
2. Pabortsava, K., Lampitt, R.S. High concentrations of plastic hidden beneath the surface of the Atlantic Ocean. Nat Commun 11, 4073 (2020). https://doi.org/10.1038/s41467-020-17932-9
