The oxidation state of iron in Mid-Ocean Ridge Basaltic (MORB) glasses: Implications for their petrogenesis and oxygen fugacities

Abstract

The oxidation state of a basaltic liquid may be obtained from the average valence state of Fe in its quenched glass. Suitable glasses are widely available from the basalts erupted at Mid-Ocean Ridges (MORB). Measurements of Fe3+/∑Fe, where ∑Fe=Fe2++Fe3+, by XANES spectroscopy of a globally representative sample of these MORB glasses are in good agreement with the most recent literature values determined by wet chemistry, but have improved precision, with an expected standard deviation of ±0.01 in Fe3+/∑Fe for each independent measurement. These precise data allow the geochemical controls on MORB redox systematics at both global and local scales to be evaluated. At the global scale, the relationship between log[Fe2O3] and [MgO] shows that Fe3+ behaves like a lightly incompatible element (LICE) during the crustal evolution of MORB, with its incompatibility between those of the redox-insensitive elements Li and In. The variability of Fe3+ about this global trend is also consistent with that of redox-insensitive elements of similar incompatibility, implying no external buffering of oxygen fugacity (fO2). The variability of Fe3+ is anti-correlated with the variabilities of Sr and Na, which are compatible in plagioclase, but is positively correlated with the variabilities of elements that partition preferentially into clinopyroxene and/or olivine rather than plagioclase, showing that its variability is controlled at least partly by varying ratios of plagioclase to clinopyroxene and olivine during the crustal evolution of MORB. Deviations of Fe3+ concentrations from the global trend are, like those of other LICE, anticorrelated with the deviations from their global trends of the very incompatible elements (VICE). Extrapolation of the global trend to an assumed parental melt at 10.4 wt% [MgO] gives [Fe2O3]o, the concentration of Fe2O3 in the global average MORB parental melt, of 0.6 wt%, which is consistent with 15–20% partial melting of a spinel lherzolite source with 0.2 to 0.3 wt% Fe2O3. This estimate of source Fe2O3 agrees with the estimate for fertile upper mantle lherzolite deduced from measurements of Fe3+/∑Fe and total Fe in the minerals of spinel peridotite xenoliths from the lithosphere.The fO2 of the MORB glasses at 1 bar may be calculated from their measured Fe3+/∑Fe using a new parameterization of 478 experimental data from the literature with <60 wt% SiO2. The need for the new parameterization arises from recent experimental studies, which, among other aspects, are consistent with the ideal stoichiometry governing the thermodynamic relationship between Fe3+/Fe2+ and fO2, namely Fe3+/Fe2+ ∝ (fO2)0.25. The parameterization gives:log10⁡(Fe3+/Fe2+)=0.25ΔQFM−1.35+0.034[Na2O]+0.044[K2O]+0.023[CaO]−0.18[P2O5] where ΔQFM is the difference between the fO2 of the silicate melt and the quartz–fayalite–magnetite buffer in log10 units, logfO2(QFM) =8.58−25050/T, relative to the conventional standard state of pure O2 at 105 Pa, T is in K, and [Na2O] etc., are the concentrations of the oxide components in weight percent. The global average ΔQFM recorded by MORB glasses in the compositional range of 5 to 10 wt% MgO is +0.2 ±0.3, with a small but resolvable increase with decreasing MgO. There is no evidence that Fe3+/∑Fe systematics in MORB are influenced by interactions with other polyvalent elements like S or Cr.