Simultaneous Determination of Small Microplastics' Size, Type, Charge, Number and Mass Concentration by Machine-Learning Driven Single-Particle Sensing

Microplastic (MP) pollution is a growing global concern, with increasing evidence confirming that its complex environmental behaviors and ecotoxicity are influenced by multiple properties such as size, type, charge, number or mass concentration. However, none of the existing technique is able to measure all of them concurrently. Herein, using representative small microplastics with various sizes (3/4/5/6 μm in diameter), types (polystyrene, polyvinyl chloride, polyethylene terephthalate, and polymethyl methacrylate), and charges (pristine, carboxyl-modified, and amine-modified), we demonstrate for the first time that micropore-based single-particle electrochemical sensing enables high-fidelity recording of the above parameters in distinct current signal features, which thus leads to nearly 100% accurate identification for each individual MP through multi-feature classification. More importantly, we develop a vibration-induced capture strategy that can considerably enhance the mass transfer of particles towards micropore, forming a universal calibration curve for all MPs. Hence, by counting the event frequency of a particular MP, the value of its number concentration is readily convertible to mass concentration. As a proof-of-concept, mixtures of four MPs sharing the same mass concentrations but of different sizes, charges, or types, are examined by both single-particle sensor and pyrolysis-gas chromatography-mass spectrometry. The results of total mass are almost identical for two methods, but only the former is capable of differentiating size and charge in the meantime. The unique ability of multi-parameter analysis by machine learning driven single-particle sensing offers a transformative approach towards precision characterization of source, fate and effect of environmental microplastics, and is transferrable to nanoplastics as well.