Treffer: A Novel Framework for Electron Microscopy‐Based Atmospheric Particulate Matter Analysis: Ensuring Representativeness and Quantifying Uncertainty.

Scanning electron microscopy (SEM) is a critical tool for characterizing the morphology, elemental composition, and size characteristics of atmospheric single particles. Although advancements in computer‐controlled technologies have significantly improved analytical throughput (>1,000 particles/hour), the analysis of particulate matter (PM) samples still faces two fundamental challenges: first, determining how many particles need to be analyzed (i.e., the analysis threshold) to ensure statistical representativeness and second, quantifying the data uncertainty caused by the limited number of particles analyzed. Herein, we established an innovative framework addressing both challenges: a multicriteria analysis threshold evaluation system was developed to determine analysis thresholds, and a cyclic overlapping block bootstrap (COBB) method was proposed to quantify data uncertainty arising from finite particle counts. Analysis of 38 PM samples (479,200 particles) encompassing diverse emission sources, urban environments, and seasonal variations revealed that sample complexity dictated analysis thresholds. Environmental samples required higher thresholds (approximately 4,300 particles for active sampling and 5,000 for passive sampling) than source samples (approximately 3,600 particles) primarily due to their more complex composition. COBB analysis demonstrated an inverse correlation between component abundance and relative uncertainty. Notably, trace components (abundance <1.0%) exhibited persistently high uncertainty even with 2,000‐particle analyses. This framework establishes systematic methodologies spanning standardized SEM data acquisition to uncertainty quantification, substantially enhancing the scientific rigor, and cross‐study comparability of SEM‐based atmospheric PM research. Plain Language Summary: Scanning electron microscopy (SEM) has become a critical technology for particulate matter (PM) analysis, enabling the reconstruction of atmospheric PM characteristics through single‐particle observation. This technique not only reveals the physical and chemical properties of individual atmospheric particles but also advances research in PM source apportionment and aerosol‐climate interactions. However, a major limitation lies in its capacity to analyze only a tiny fraction of atmospheric particle populations. This constraint underscores the need to determine the minimum number of particles required (i.e., the analysis threshold) to ensure data representativeness and to quantify the uncertainties arising from limited particle counts. Herein, we innovatively established a multicriteria analysis threshold evaluation system based on morphological, elemental, and size‐related multidimensional features. Using 38 samples, we established a reference threshold library for five common types of PM samples, offering universal benchmarks for SEM‐based analyses. Furthermore, we proposed a novel method for quantifying uncertainties in SEM data. These breakthroughs collectively form a systematic SEM analytical framework, significantly enhancing the reliability of electron microscopy data in critical applications such as PM source apportionment and aerosol‐climate modeling. Key Points: Environmental samples are more complex than source samples, and more particles should be analyzed to ensure data representativenessDeveloped an analysis threshold evaluation method and established a threshold library for five common particulate matter sample typesProposed a method to quantify uncertainty in scanning electron microscopy data arising from limited analyzed particle numbers [ABSTRACT FROM AUTHOR]

Copyright of Journal of Geophysical Research. Atmospheres is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)