Fe3+ partitioning between pyroxene and melt at shallow upper mantle conditions

Oceanic basalts offer direct clues to the oxidation state of the Earths shallow upper mantle. Although their fO2 is known with high precision, the processes of acquiring their Fe3+ budget is not well understood1,2,3. Observation of peridotite xenoliths and thermodynamic modeling of DMM fO2 point towards different concentrations of the primitive mantle Fe3+, which vary by a factor of 34,5. Since cpx and opx together host ~75% of the Fe3+ reserve in peridotite xenoliths6, they play crucial roles in controlling Fe3+ concentration in the equilibrium partial melt. To this end, we have performed high P, T, fO2 monitored piston cylinder experiments to understand Fe3+ partitioning behavior between cpx, opx and mafic melt. Experiments were dynamically cooled to grow large, homogeneous pyroxene crystals and the fO2 was monitored by using Fe-Pt alloy capsules with variable initial enrichment of Fe7. Fe3+/FeT of the pyroxenes and glass were measured by XANES spectroscopy using the newly developed Mossbauer-based calibrations8,9. Our experimental results show that DFe3+ cpx/melt is not constant and increases with increasing P and fO2. DFe3+ cpx/melt correlates strongly with the Al3+ and moderately with the Fe3+ content of cpx suggesting the non-ideal nature of Fe3+ substitution in cpx. Compared to the widely used thermodynamic models in pMELTS and perple_X10,11, the experimentally derived DFe3+ pyroxene/melt are systematically lower at all fO2 and P. We model the variation in Fe3+ in liquid by equilibrium batch melting of the DMM mantle with Fe3+ = 0.3 wt.% between 1-2 GPa by varying DFe3+ cpx/melt and mineral modes. The decrease in bulk DFe3+ with increasing F reproduces the negative correlation between Fe3+ and Na2O observed in N-MORB and can explain nearly 60% of the variability observed in the MORB data set. The Fe3+ content of the OIBs is too high to be generated from a source with Fe3+ similar to DMM. However, more specific estimates of OIB-source Fe3+ will require improved estimates of DFe3+ peridotite/melt at pressures above 2 GPa. 1. Cottrell et al., (2011) 2. Shorttle et al., (2015) 3. Brounce et al., (2017) 4. Canil et al., (1994) 5. Gaetani (2016) 6. Woodland (2006) 7. Davis and Cottrell, (2021) 8. Rudra et al., (in review) 9. Zhang et al., (2018) 10. Ghiorso et al., (2002) 11. Jennings and Holland (2015)