Characterisation of particulate matter from the Kara Sea using electron probe X-ray micro analysis

Automated electron probe X-ray micro-analysis (EPXMA) was used to characterise 17,600 individual particles collected in the Kara Sea and in the estuary of the Yenisey River. Samples were obtained by filtering and centrifugation of particles in suspension and from bottom sediments. Multivariate techn...

Full description

Bibliographic Details
Main Authors: Jambers, W., Smekens, A., Van Grieken, R., Shevchenko, V.P., Gordeev, V.
Format: Article in Journal/Newspaper
Language:English
Published: 1997
Subjects:
PNE
Online Access:http://www.vliz.be/nl/open-marien-archief?module=ref&refid=2954
Description
Summary:Automated electron probe X-ray micro-analysis (EPXMA) was used to characterise 17,600 individual particles collected in the Kara Sea and in the estuary of the Yenisey River. Samples were obtained by filtering and centrifugation of particles in suspension and from bottom sediments. Multivariate techniques were used to reduce the enormous data sets. By testing hierarchical, non-hierarchical and fuzzy clustering on the centrifuged suspension samples, it was shown that hierarchical clustering is most suitable. The two other clustering methods are very time consuming and in most cases do not add any additional information. This hierarchical clustering of the data matrix results in different particle types which can be apportioned to their possible sources. The aluminosilicate particle types dominate in all samples and suppress the relative abundances of other interesting groups like metal-rich particles. However, Fe-rich and Ti-rich particles are still frequently found. Since the nearest industry is 200 km from the sampling site, it is especially surprising to find high contributions for Ti-rich particles which, because of their small diameter (i.e. between 0.6 µm and 1.1 µm), are most likely of anthropogenic origin. To complete this study the homogeneity and morphology of 100 particles from water suspension samples were analysed using manual EPXMA.