Repeatability and reproducibility of semi-automated measurements of soot primary particle size distributions from TEM images

摘要

The primary particle size distribution of fractal-like aerosols produced from combustion processes is an important property of their morphology. Manual analysis of transmission electron microscope (TEM) images remains a principal means for measuring this distribution, and the results are often used to quantify other aerosol features. However, a detailed uncertainty analysis of primary particle size distribution is needed, and this should account for both measurement repeatability and variation across different samples and operators. In this study, three soot samples are thermophoretically collected above a propylene/air co-flow diffusion flame and imaged using TEM. The primary particle size distribution of each sample is independently measured by four operators using a randomized, semi-automated procedure. The overall uncertainty considering sample, operator, and measurement system effects is quantified through a gage repeatability and reproducibility (gage R & R) study. The results from the semi-automated method are then compared with results from two automated methods employing Euclidian distance mapping and Hough Transform algorithms. We conclude that uncertainties uncovered by the gage R & R analysis can be substantial and must be taken into consideration when interpreting the results of TEM-derived properties.