Abstract
The water treatment for microplastics (MP) at a Danish groundwater-based waterworks was assessed by Fourier-Transform IR micro-spectroscopy (μFTIR) (nominal size limit 6.6 μm) and compared to results from Raman micro-spectroscopy (μRaman) (nominal size limit 1.0 μm) on the same sample set. The MP abundance at the waterworks' inlet and outlet was quantified as MP counts per cubic metre (N/m(3)) and estimated MP mass per cubic metre (μg/m(3)). The waterworks' MP removal efficiency was found to be higher when analysing by μFTIR (counts: 78.14 ± 49.70%, mass: 98.73 ± 11.10%) and less fluctuating than when using μRaman (counts: 43.2%, mass: 75.1%). However, both techniques pointed to a value of ∼80% for the counts' removal efficiency of MPs >6.6 μm. Contrarily to what was shown by μRaman, no systematic leaking of MPs from the plastic elements of the facility could be identified for the μFTIR dataset, either from the counts (inlet 31.86 ± 17.17 N/m(3), outlet 4.98 ± 2.09 N/m(3)) or mass estimate (inlet 76.30 ± 106.30 μg/m(3), outlet 2.81 ± 2.78 μg/m(3)). The estimation of human MP intake from drinking water calculated from the μFTIR data (5 N/(year·capita)) proved to be approximately 332 times lower than that calculated from the μRaman dataset, although in line with previous studies employing μFTIR. By merging the MP length datasets from the two techniques, it could be shown that false negatives became prevalent in the μFTIR dataset already below 50 μm. Further, by fitting the overall frequency of the MP length ranges with a power function, it could be shown that μFTIR missed approximately 95.7% of the extrapolated MP population (1-1865.9 μm). Consequently, relying on only μFTIR may have led to underestimating the MP content of the investigated drinking water, as most of the 1-50 μm MP would have been missed.