Machine Learning-Assisted Microwave-Microfluidic Platform for Microplastic Detection.

Microplastics (MPs) have emerged as a critical environmental concern due to their ubiquitous presence in ecosystems and potential impacts on human health. However, conventional MP detection and identification methods are often constrained by procedural complexity, limited particle size coverage, and the lack of continuous monitoring capability. In this proof-of-concept study, we present a novel microwave-microfluidic platform for flow-through detection and identification of MPs in aqueous media. In this system, engineered, size-categorized particles are passed through disposable microfluidic chips positioned over split-ring resonator (SRR) sensing hotspots, generating particle-specific perturbations in resonance frequency and transmission amplitude. We first evaluated single-particle detection performance across four particle-size ranges spanning 20-300 μm. Next, we focused on distinguishing between particle types in the 165-300 μm size range using a lightweight k-nearest neighbors (k-NN) model to recognize material-specific resonance-derived features for polyethylene, polystyrene, soda lime glass, and brine shrimp eggs. We further examined robustness across multiple aqueous carrier liquids through carrier-specific baseline referencing. Overall, our work demonstrates the feasibility of microwave-microfluidic sensing for continuous, in-flow, single-particle monitoring and materials classification under controlled conditions, thus, establishing the foundation for future analysis of MPs in complex environmental matrices.