Quantitative X-ray mapping using an electron probe enables quantitative evaluation of inhomogeneities within rocks. Recent studies have proposed methods to construct quantitative chemical maps by combining X-ray maps with referential spot analyses within a mapped area. These approaches address matrix effects by assuming each pixel in the mapped area represents a single phase. In such cases, the spatial resolution of the X-ray maps must be sufficiently high to separate mineral phases. This study proposes a new procedure to reliably quantify centimeter-scale X-ray maps even if the maps contain an ineligible number of pixels analyzing multiple phases because of a large mapping probe diameter. Such multi-phase pixels are statistically classified into their constituent phases by introducing a distribution-based clustering analysis. Furthermore, based on referential spot analyses, we implemented corrections for matrix effects and the backgrounds of single- and multi-phase pixels. Our technique, termed QntMap, was developed as an open source R package and distributed on a social coding platform, GitHub (https://github.com/atusy/qntmap). We applied QntMap to calculate local bulk compositions within an ultrahigh-pressure eclogite from Nové Dvory, Czech Republic. The studied sample is a garnet-rich bimineralic eclogite that includes a 3 mm thick pyroxene-rich layer. A mapped area is approximately 3 × 1 cm in size and oriented normal to the layer. A profile normal to the layer shows increases in Cr2O3 (0.0 to 0.3 wt%) and XMg [Mg/(Fe+Mg) = 0.5 to 0.8] from the garnet-rich matrix toward the pyroxene-rich layer. A large variation in XMg and high-Cr2O3 contents in the pyroxene-rich layer are inconsistent with a cumulate origin. We suggest that the pyroxene-rich layer was derived from a pyroxenitic melt that intruded the eclogite.
ASJC Scopus subject areas