Reduction of sampling and analytical errors for electron microscopic analysis of atmospheric aerosols

Electron microscopy-energy dispersive spectroscopy (EM/EDS) can be used to determine the elemental composition of individual particles. However, the accuracy with which atmospheric particle compositions can be quantitatively determined is not well understood. In this work we explore sources of sampling and analytical bias and methods of reducing bias. Sulfuric acid [H2SO4] and ammonium sulfate [(NH4)(2)SO4] particles were collected on beryllium, silicon, and carbon substrates with similar deposition densities. While (NH4)(2)SO4 particles were observed on all substrates, H2SO4 and ammonia-treated H2SO4 particles could not be found on beryllium substrates. Interactions between the substrate and sulfuric acid particles are implicated. When measured with EM/EDS, H2SO4 particles exposed to ammonia overnight were found having lower beam damage rates (0.000 +/- 0.002 fraction s(-1)) than those without any treatment (0.023 +/- 0.006 fraction s(-1)). For laboratory-generated [C10H6(SO3Na)(2)] particles, the composition determined using the experimental k-factors evaluated from independent particle standards of similar composition and size shows an error less than 20% for ail constituents, while greater than 78% errors were found when k-factors were calculated from the theory. This study suggests (1) that sulfate beam damage can be reduced by exposure of atmospheric particle samples to ammonia before analysis, (2) that beryllium is not a suitable substrate for atmospheric particle analysis, and (3) calibration (k-factor determination) using particle standards of similar size and composition to particles present in the atmosphere shows promise asa way of improving the accuracy of quantitative EM analysis. Copyright (C) 1996 Elsevier Science Ltd