This doctoral thesis was written by Johan de Hoog and describes the different steps that were taken for developing a new method for the semi-quantitative analysis and characterization of (environmental) particles by TW-EPMA.

First, we evaluated the need for such a new method by showing the different technological evolutions in electron microscopy. We saw that with the arrival of a new generation of energy-dispersive X-ray detectors, it became possible to determine light elements. Their ultra-thin polymer windows allow for a better transmission of low-energy X-rays, which made sure that at least qualitative analysis of elements with an atomic number Z > 5 was possible. Since these light-element compounds are very abundant in environmental particles, the breakthrough in detector technology was also a big step forward for enhancing our knowledge about their composition. The use of a so-called cold stage for cooling down samples with liquid nitrogen to –193°C also allowed us to analyze volatile species, for example ammonium and nitrate compounds, since we could considerably reduce the beam damage effects.

Soon the idea rose to investigate the possibility to develop a method for the quantification of both light and heavy elements contained in environmental particles. A recently developed Monte Carlo model for the simulation of electron-solid interactions at low energies was adapted for particulate samples by implementing the necessary features for taking into account particle geometry effects. The Monte Carlo model was integrated into an algorithm for the iterative quantification of particle compositions based on reverse simulations. While testing the quantification method using particle standards made of inorganic and organic salts, several directly or indirectly critical aspects were investigated in order to adapt the involved analytical procedures like the sample preparation (the quality of substrates and the use of coatings), the beam damage effects, the spectrum evaluation (peak identification and corrections for peak overlap) and the physical parameters used in the model. A comparison with other quantification methods showed that our method produced quite satisfying results, also considering the uncertainties and assumptions we had to deal with.

Then, we tried to go one step further by exploring several possibilities to get the most out of the obtained concentrations for light and heavier elements. For the characterization of environmental samples, huge datasets have to be dealt with in a fast, efficient and intelligent way to get a correct profile of the analyzed particles. Experiments with artificial particles showed that the calculation of light element concentrations by our iterative Monte Carlo quantification procedure indeed produced an added value for differentiating between different particle types when applying data reduction tools. These mathematical methods are able to reduce large datasets to smaller sets of specific characteristic parameters. The purely mathematical data treatment was then extended with an expert system for the elucidation of particle composition. The mental algorithms used to identify particle types were translated into an automatic procedure that could do the same job of the analyst in an unsupervised and computerized way. This method starts on the individual particle level and ends with the classification of similar particles into typical groups.

Two methods were developed and tested for the additional characterization of the internal particle composition. Grazing-Exit EPMA (GE-EPMA) could offer specific information about core-shell structured particles, but some disadvantages due to beam damage and the instrumental setup strongly limited the possibilities and made data interpretation quite difficult.  Multiple-Voltage EPMA (MV-EPMA) appeared to have more potential, since particles are analyzed at different voltages to vary the information depth, so that layered particles could be qualitatively differentiated from homogeneous particles. The technique was also used to determine the surface layer thickness of artificial, heterogeneous particles, but beam damage again played a disadvantageous role. It is clear that the application of GE-EPMA or MV-EPMA to real atmospheric particles will probably be limited to qualitative analysis due to the complexity involved in the analysis, since both their complex chemical composition and their structural heterogeneity cannot be known a priori. It was however shown that both techniques, in combination with light-element analysis, are able to extract information that could not be obtained before.

The developed and optimized method was also put into use by applying it in projects that required the characterization of different environmental samples. The semi-quantitative determination of light elements by TW-EPMA was applied to the analysis of North Sea aerosol particles, and it showed to be an efficient tool for the investigation of a huge number of individual microparticles. The method offered good statistics to study their chemical properties and their behavior in the atmosphere. The results of the collected aerosol samples showed that the light-element analysis of single particles gave additional useful chemical information to marine aerosol studies, compared to the conventional EPMA methods. The detection of characteristic X-ray radiation of low-Z elements, which was not possible before, offered the possibility to identify more compounds (e.g. NH₄NO₃, organic particles), or it enabled the differentiation between particle types, like e.g. (NH₄)₂SO₄ and sulphur-containing organic particles which previously only showed a sulphur peak in their X-ray spectra. The extended, semi-quantitative knowledge of low-Z element concentrations led to a more relevant particle classification; therefore, the possible sources and interactions occurring in the troposphere can be traced in much more detail by the analysis of large numbers of individual particles. Through the combination of chemical data with meteorological data, the environmental link was found between the chemical composition of the different samples and the environmental conditions during sampling.

TW-EPMA has already been applied in many projects for the characterization of different sample types from around the world, like atmospheric aerosols (Amazon region, Brazil; Lake Balaton, Hungary; North Sea; sugar cane plantation, Brazil; indoor and outdoor environments in residential areas, Flanders; indoor and outdoor environments near historical monuments all over Europe) and sediment particles (North Sea; Tisza river, Hungary). For this thesis, the method was compared to and combined with different complementary bulk techniques. As well in the abovementioned North Sea project, as in another project for the characterization of PM2.5 in Flanders, it was shown that TW-EPMA could offer new insights in the composition and (trans)formation of particulate matter in the atmosphere. The combination with ion chromatography showed that the matching of bulk and microanalytical results could lead to mutual benefits in aerosol characterization. TW-EPMA proved to be a valuable element in the multi-analytical approach of the PM2.5 study in Flanders.

Should the method from now on be used in every project about the characterization of environmental particles?  No, because for some problems a qualitative approach is already enough or the knowledge about light elements is not necessary. For example, studies about the distribution of heavy metals in river sediments will probably not always require an exhaustive and time-consuming quantitative characterization of the light-element composition. The analyst will have to evaluate what is the scope of the projects and what are the capabilities of the available methods in order to choose the right level of characterization. TW-EPMA combined with our quantification method has proven to be a valuable tool for single particle analysis, with still some considerable potential for improvements in the future … when better technology will again lead to new insights.