SOURCE ATTRIBUTION OF PB IN SIX SOILS FROM MITROVICA, REPUBLIC OF KOSOVO AT THE INDIVIDUAL PARTICLE LEVEL USING COMPUTER CONTROLLED SCANNING ELECTRON MICROSCOPY (CCSEM)

Samples of surface soils from Mitrovica, in northern Kosovo were examined by computer controlled scanning electron microscopy (CCSEM) to characterize the constituent types of Pb-bearing particles. Although the soils were in close proximity to each other, the soil Pb concentrations covered almost two orders of magnitude, and the relative bioaccesible (RBA) Pb ranged from 3 - 89%. To gain insights into why the Pb RBA varied so markedly CCSEM was used to identify differences in Pb-particle types in the soils.

CCSEM particle element composition data was obtained on >4,500 particles in each of 6 soil samples. Relative X-ray percentage data was collected for Na, Mg, Al, Si, P, S, Cl, K, Ca, Ti, Cr, Mn, Fe, Ni, Cu, Zn, As, Sb, and Pb. Homogenous groups of particles in each sample (based on similarities in element associations) were identified using supervised divisive hierarchical cluster analysis. In the samples most of the particles were crustal in origin and 35 homogenous groups accounted for the variety in these particles. A second cluster analysis was performed for the Pb-bearing particles and 49 groups of these were recorded. The element characteristics of each group were ordered in a linear sorting scheme (of classes). The data for each Pb-particle was sifted through the scheme and each was assigned to the closest matching class.

In the classification process 8 classes were found to have minimal discriminating power, so a 41 class scheme was used for particle differentiation. The data from the 41 classes were later concatenated into 18 groups of similar classes. It was expected that the group assignments for the Pb-particles from each soil would be different. With this detailed 41 class scheme minimal cross sample classification was accomplished. The major group assignments for the percentage of Pb particles in each soil (coded RM or BM) were as follows: RM77: 90.6% to groups 1 and 2; RM66: 60.1% to groups 3 and 4; RM72: 46.3% to groups 5, 6 and 7; BM5: 48.3% to groups 8, 9, 10, and 11; RM71: 40.4% to groups 12, 13, and 14; RM45: 64.7% to groups 15, 16, 17, and 18. The percentage of Pb-particles assigned to a group that was associated with a different soil was generally less than 4% and did not exceed 10%. The classes and Pb-particle types were quite specific to each soil. We posit that the differences in Pb-particle composition control the soil Pb RBA.