Source Characterization Of African Soils Using Computer Controlled Sem Analysis

Abstract

An investigation is underway to determine whether African dust can be chemically distinguished by regions and whether it is developed through Pedogenic or Aeolian processes. A total of 29 dust samples were taken from the Sahel and Saharan region of Africa and are to be analyzed using computer controlled scanning electron microscopy (CCSEM). CCSEM collects data on size shape and chemical composition on individual microscopic dust particles present in a media sample analyzed in a scanning electron microscope (SEM). Back-scattered electron imaging (BEI) in the SEM is accomplished using a four-quadrants backscatter electron detector. This method of imaging the particles in the SEM provides particle morphology data. Compositional information on individual particles will be collected with a 10 mm2 OmegaMax Silicon Drift Detector (SDD) with ultra thin window (UTW) (capable of registering light element X-rays) Peltier-cooled energy dispersive x-ray spectrometer. A variable pressure SEM will be used for the analysis of insulating materials, which eliminated the need for special specimen coating to dissipate charge and remove artifacts. Data from these samples are being used to address the primary question: (1) Can CCSEM technology accurately describe elemental compounds derived from soil samples collected from Africa. The principal goal of the research is to attempt to identify (by cluster analysis) homogenous groups of particles in each soil sample. By homogenous we mean a group of particles that separates from other groups based on the information gained on the individual particles. The homogenous group will occupy a unique position in information space where the information that defines the group is primarily the composition data from the particles (data on up to 25 elements in each particle), but can also include particle size, shape (e.g., aspect ratio), and backscatter signal strength (a measure of the particle's average atomic number composition). Obviously, the greater the amount of individual particle data in the information space, the greater the likelihood of separating homogenous groups of particles. The term particle "class" is used here to represent homogenous elemental particle groups. Pilot data analysis made use of a 19-class element classification scheme to separate particles in a typical analysis of some 4000 particles in a single dust sample. Initial findings of six samples showed large amounts of Fe, Si, and Al-rich minerals. The Al-Si-rich minerals show a close correlation in relative elemental amounts. This is to be expected from clay minerals of the pyroxene group. The Fe, Si-rich minerals trend towards an inverse relationship, which is also consistent with iron oxides of the spinel group that generally consist of magnetite. Other elemental constituents within the samples include varying amounts of Ti, Ca, and K. An initial run of samples, 2 Burkina Soils and 2 Burkina Laterites, show a similarity in chemical composition, leading to the hypothesis that the Burkina Soils may have originated from the Burkina Laterites. As the experiment progresses we expect to see similar Aeolian processes contributing to the mineral content of other surface dusts. Further research on the effects of these wind driven dusts is needed to assess the potential health and nutrient impacts on the Americas as it is blown across the Atlantic Ocean and deposited in the ocean, Caribbean islands, Florida, the Amazon and Texas.