QUANTITATIVE EPMA COMPOSITIONAL MAPPING OF NWA 2995: CHARACTERISATION AND PETROLOGIC INTERPRETATION OF MAFIC CLASTS

摘要

We present fully quantitative compositional maps of lunar meteorite NWA 2995 using electron microprobe stage mapping, and compare selected clast mineralogy and chemistry. NWA 2995 is a feldspathic fragmental breccia containing anorthosite, norite, olivine basalt, subophitic basalt, gabbro, KREEP-like basalt, granulitic and glassy impact melts, coarse-grained mineral fragments, Fe-Ni metal, and glassy matrix, and has previously been analysed by bulk methods instrumental neutron activation analysis (INAA) and fused-bead electron probe microanalysis (EPMA) [1,2]. The JEOL JXA-8200 electron microprobe at Washington University, with JEOL and Probe Software operating systems was used to acquire backscattered-electron (BSE) and wavelength-dispersive (WDS) X-ray maps, and point analyses as outlined in [3]. The CalcImage programme is used to perform a full φ(pz) matrix correction for each pixel according to C = k * ZAF, where C is the concentration, k is the background-corrected X-ray sample intensity relative to a calibration standard, and factors Z, A, and F correct for matrix effects. The mean atomic number (MAN) correction is based on the WDS standardisation and is part of the ZAF iteration. The resulting maps contain a complete elemental analysis at each pixel, and are used for further analysis using Golden Software Surfer, Matlab, and ENVI software packages. A "slab" of NWA 2995 approximately 19x11 mm was mapped at 15 kV and 100 nA probe current, using dwell times of 25 - 50 msec at 1024x1024 pixel resolution and stage step size of 2-7 µm per pixel, with run times of 16 - 24 hours. Data were collected in two passes of 5 WDS elements each, to obtain a 10 element data set of 10242 fully quantitative analyses per map. These analyses were classified by assignment to mineral phase categories and used to calculate the modal abundance and composition of olivine, low- and high-Ca pyroxene, plagioclase, ilmenite, and spinel. Mineral classification maps for three mafic clasts B, G, and H are shown in Fig. 1. A density-corrected bulk composition was calculated for each clast using the measured modal abundance, elemental composition, and an appropriate phase density. Depending on modal abundance, each phase represents from 10~3 -105 quantitative analyses. All compositional data are detailed in the upper part of Table 1, where the density-corrected bulk composition of each clast is compared with: (a) bulk data for NWA 2995 obtained from large area maps over most of the slab (maps 1&2); (b) the matrix data, which is representative of the slab with clasts subtracted; and (c) with the EPMA fused-bead analysis of [2]. The modal proportions of minerals in the clasts, and their mineral end-member representation are listed in the lower part of Table 1. Clasts B and G are sub-ophitic olivine basalts, and clast H is an ophitic olivine gabbro. Clast B contains subhedral olivine of two distinct compositions, low-and high-Ca pyroxene, ilmenite, and spinel, and has a basaltic composition with comparatively high-Fe content. Clast G contains olivine, low-Ca pyroxene, plagioclase, and spinel. Clast H contains low- and high-Ca pyroxene, plagioclase, minor olivine, and has a relatively aluminous composition. There is excellent agreement between the matrix data determined by compositional mapping and the fused-bead EPMA data acquired previously, indicating that both techniques can be used to accurately measure the bulk composition of meteorites as represented by subsampled areas. A main advantage of quantitative EPMA compositional mapping is in comparison of areas representing point analyses, phases, and lithic regions, in addition to the non-destructive nature of EPMA analysis. This advantage can be used to compare clast and matrix composition, characterise mineral zoning, and to determine the abundance of mineral fragments and glass, for example. In addition to treatment of map data using fully quantified concentrations, the procedure outlined here can be used to characterise samples based on the classification definitions developed on an initial set of maps, rather than requiring intensive inspection of each new map.