In-situ determination of the oxidation state of iron in Fe-bearing silicate melts

Terrestrial lavas commonly contain up to 10 wt% of iron. Furthermore, rocks returned from the Moon indicate lunar lava containing up to 25 wt% of iron and planetary scientists estimated that the martian mantle has about 18 wt% of iron. An experimental challenge in dealing with Fe-bearing silicate melts is that the oxidation state, controlling the proportions of ferric and ferrous iron, is a function of composition, oxygen fugacity and temperature and may vary significantly. Further complications concerning iron originate from its potential to be either four-, six- or even five-fold coordinated in both valence states. Therefore, the oxidation state of iron was determined in air for various Fe-bearing silicate melts. Investigated samples were Na-disilicate (NS), one atmosphere anorthite-diopside eutectic (AD) and haplogranitic (HPG8) melts containing up to 20, 20 and 10 wt% of iron, respectively. XANES spectra at the Fe K-edge were collected for all the melts at beamline A1, HASYLAB, Hamburg, using a Si(111) 4-crystal monochromator. Spectra were collected for temperatures up to 1573 K using a Pt-Rh loop as heating device. The Fe oxidation state was determined from the centroid position of the pre-edge feature using the calibration of Wilke et al. (2004). XANES results suggest that oxidation state of iron does not change within error for NS melts with addition of Fe, while AD and HPG8 melts become more oxidised with increasing iron content. Furthermore, NS melts are well more oxidised than AD and HPG8 melts that exhibit relatively similar oxidation states for identical iron contents. The oxidation state of iron for NS melts appears to be slightly temperature-dependent within the temperature range investigated (1073-1573 K). However, this trend is stronger for AD and HPG8 melts. Assuming that glass reflects a picture of the homogeneous equilibria of the melt, the present in-situ Fe-oxidation states determined for these melts were compared to those obtained on quenched glasses from different temperatures, when possible, using wet-chemical analysis (i.e., K-dichromate potentiometry). Both datasets agree reasonably well (within 10 %). References: Wilke et al. (2004) Chemical Geology, 213, 71-87.