We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Physics > Applied Physics

Title: Characterizing soot in TEM images using a convolutional neural network

Authors: Timothy A. Sipkens (1 and 2), Max Frei (3 and 4), Alberto Baldelli (1 and 5), P. Kirchen (1), Frank E. Kruis (3 and 4), Steven N. Rogak (1) ((1) Department of Mechanical Engineering, University of British Columbia, (2) Department of Mechanical Engineering, University of Alberta, (3) Institute of Technology for Nanostructures, University of Duisburg-Essen, (4) Center for Nanointegration Duisburg-Essen, University of Duisburg-Essen, (5) Faculty of Land and Food Systems, University of British Columbia)
Abstract: Soot is an important material with impacts that depend on particle morphology. Transmission electron microscopy (TEM) represents one of the most direct routes to qualitatively assess particle characteristics. However, producing quantitative information requires robust image processing tools, which is complicated by the low image contrast and complex aggregated morphologies characteristic of soot. The current work presents a new convolutional neural network explicitly trained to characterize soot, using pre-classified images of particles from a natural gas engine; a laboratory gas flare; and a marine engine. The results are compared against other existing classifiers before considering the effect that the classifiers have on automated primary particle size methods. Estimates of the overall uncertainties between fully automated approaches of aggregate characterization range from 25% in d_{p,100} to 85% in D_{TEM}. A consistent correlation is observed between projected-area equivalent diameter and primary particle size across all of the techniques.
Comments: 23 pages, 8 figures, 5 tables, for associated code see this https URL and this https URL
Subjects: Applied Physics (physics.app-ph)
Journal reference: PowderTechnol. 387 (2021) 313-324
DOI: 10.1016/j.powtec.2021.04.026
Cite as: arXiv:2103.09272 [physics.app-ph]
  (or arXiv:2103.09272v1 [physics.app-ph] for this version)

Submission history

From: Timothy Sipkens [view email]
[v1] Tue, 16 Mar 2021 18:33:15 GMT (1359kb)

Link back to: arXiv, form interface, contact.