Nanoplastics analysis in real-world samples: From detection sensitivity to statistical rigor

초록

Nanoplastics (< 1 µm) have emerged as a new frontier in environmental pollution research, presenting analytical challenges distinct from microplastics due to their minute size and unique physicochemical behaviors. Recent innovations in vibrational spectroscopy, Surface-Enhanced Raman Scattering (SERS) and Pyrolysis-Gas Chromatography/Mass Spectrometry (Py-GC/MS) have established a technological foundation for identifying nanometer-scale particles. However, despite these dramatic improvements in detection sensitivity, the field continues to face severe bottlenecks regarding the reliability, reproducibility, and comparability of results in non-spiked environmental samples. This review argues that nanoplastics analysis must be redefined not merely as a deterministic task of detection but as a problem of probabilistic inference fraught with uncertainty. We clarify the conceptual distinction between Signal-level Limit of Detection (LOD) and Distribution-level LOD and critically examine the statistical limitations of currently prevalent analytical techniques. Specifically, we identify the quantification of nanoplastics as a parameter estimation problem for a multinomial distribution. By applying the theoretical framework of Thompson (1987), we mathematically demonstrate that a minimum sample size of approximately 510 particles is required to estimate the proportions of multiple polymer components with 95% confidence and a 5% margin of error. The fact that most current studies analyze only tens of particles suggests that the reported data contains non-negligible uncertainties. Consequently, we propose a new statistical reporting framework that includes distinguishing between exploratory and confirmatory research, explicitly quantifying uncertainty, and, most importantly, justifying sample sizes based on Thompson's theory. This is a prerequisite for nanoplastics research to mature beyond simple identification and produce robust quantitative data capable of informing policy decisions. © 2026 Elsevier B.V.

키워드