Main content area

Comparison of operator- and computer-controlled scanning electron microscopy of particles from different atmospheric aerosol types

Eriksen Hammer, Stine, Ebert, Martin, Weinbruch, Stephan
Analytical and bioanalytical chemistry 2019 v.411 no.8 pp. 1633-1645
aerosols, altitude, chemical composition, coatings, computers, fly ash, forests, mixing, scanning electron microscopy, soot, traffic, troposphere, Alps region, Cyprus, Germany, Switzerland
Individual aerosol particles from an urban background site in Mainz (Germany), a traffic hotspot site in Essen (Germany), the free troposphere in the Swiss Alps (high altitude research station Jungfraujoch), a rural background/marine site on Cyprus (Cyprus Atmospheric Observatory) and a rural background site in the forested area of Odenwald (Germany) were characterised with two different scanning electron microscopy techniques, operator controlled (opSEM) and computer controlled (ccSEM). For all samples, about 500 particles were investigated by opSEM, and between 1103 and 6940 particles by ccSEM. Large systematic differences (in some cases a factor up to ~ 20) in the abundance of the various particle groups are observed in the results of the two techniques. These differences are dependent on particle type and size. With ccSEM, information on the mixing state of particles (e.g., presence of heterogeneous inclusions, surface coatings or gradients in chemical composition) cannot be obtained, and particle groups which are recognised by their complex morphology (e.g., soot and fly ash particles) are classified into other particle groups. In addition, highly volatile particles (i.e., particles which evaporate under electron bombardment within seconds) will be overlooked by ccSEM. If these limitations of ccSEM are not considered, normalising the particle group abundances to 100% (a popular practise in many publications) may lead to drastic misinterpretation of the real aerosol composition. OpSEM is indispensable when detailed information of particle composition is required, although it suffers from a much higher expenditure of time. In conclusion, both techniques might be used for single particle characterisation as long as drawbacks of each are considered.