0
Article ? AI-assigned paper type based on the abstract. Classification may not be perfect — flag errors using the feedback button. Tier 2 ? Original research — experimental, observational, or case-control study. Direct primary evidence. Sign in to save

Rapid identification of microplastic particles in water using geometric polarization characterization and quantum neural networks

Optics Letters 2026
Xueqiang Fan, Liming Zhu, Yan Wu, Daoyou Guo, Zhongyi Guo, Rui-Pin Chen

Summary

Scientists have developed a fast, AI-powered method to detect microplastics in water by analyzing how light bends and scatters off tiny plastic particles, achieving about 95% accuracy in under 12 milliseconds per sample. This matters because microplastics are increasingly found in drinking water and their health effects are still being studied — faster, more reliable detection tools could help researchers and water treatment facilities monitor contamination more effectively, which is a key step toward better understanding and reducing our exposure to these particles.

Polymers
Body Systems

The widespread pollution of water environments by microplastics (MPs) is a critical global issue. The identification of MPs remains challenging, primarily due to their limited texture and color information. To address this issue, this Letter proposes a learning-based approach that incorporates what we believe to be a novel geometric polarization characterization of MPs with quantum neural networks (QNNs). Our method comprises three stages. First, we establish a novel Polarization Angle-Weighted Geometric (PAWG) feature derived from the geometric relationship of polarization information. Second, a novel polarization-driven quantum convolutional network (PQCNet) is designed based on pure variational quantum circuits. Finally, the PAWG features are fed into the PQCNet to train an effective variational quantum classifier for intelligent identification of MPs. Experimental results show that the proposed method yields an average accuracy of approximately 95% across three MP types (i.e., PP, LDPE, and HDPE) in water. Further, the model with only 0.0023 M parameters enables rapid inference with an average prediction time of 11.3 ms per sample. Such a success rate and computational efficiency indicate that our method can be used to rapidly identify MPs.

Share this paper