New high pressure experiments on sulfide saturation of high-FeO∗ basalts with variable TiO2 contents - Implications for the sulfur inventory of the lunar interior

In order to constrain sulfur concentration in intermediate to high-Ti mare basalts at sulfide saturation (SCSS), we experimentally equilibrated FeS melt and basaltic melt using a piston cylinder at 1.0-2.5 GPa and 1400-1600 °C, with two silicate compositions similar to high-Ti (Apollo 11: A11, ∼11.1 wt.% TiO2, 19.1 wt.% FeO∗, and 39.6 wt.% SiO2) and intermediate-Ti (Luna 16, ∼5 wt.% TiO2, 18.7 wt.% FeO∗, and 43.8 wt.% SiO2) mare basalts. Our experimental results show that SCSS increases with increasing temperature, and decreases with increasing pressure, which are similar to the results from previous experimental studies. SCSS in the A11 melt is systematically higher than that in the Luna 16 melt, which is likely due to higher FeO∗, and lower SiO2 and Al2O3 concentration in the former. Compared to the previously constructed SCSS models, including those designed for high-FeO∗ basalts, the SCSS values determined in this study are generally lower than the predicted values, with overprediction increasing with increasing melt TiO2 content. We attribute this to the lower SiO2 and Al2O3 concentration of the lunar magmas, which is beyond the calibration range of previous SCSS models, and also more abundant FeTiO3 complexes in our experimental melts that have higher TiO2 contents than previous models' calibration range. The formation of FeTiO3 complexes lowers the activity of FeO∗, aFeO∗silicate melt, and therefore causes SCSS to decrease. To accommodate the unique lunar compositions, we have fitted a new SCSS model for basaltic melts of >5 wt.% FeO∗ and variable TiO2 contents. Using previous chalcophile element partitioning experiments that contained more complex Fe-Ni-S sulfide melts, we also derived an empirical correction that allows SCSS calculation for basalts where the equilibrium sulfides contain variable Ni contents of 10-50 wt.%. At the pressures and temperatures of multiple saturation points, SCSS of lunar magmas with compositions from picritic glasses, mare basalts, to young lunar meteorites vary from 2600 to 4800 ppm for basalt equilibration with a pure FeS melt and from 1400 to 2600 ppm for basalt equilibration with a Fe-rich sulfide melt containing 30 wt.% Ni. The measured S contents in these proposed near-primary lunar magmas are lower than the predicted SCSS at the conditions of their last equilibration with the lunar mantle, indicating no sulfide retention in the lunar mantle source during partial melting. Sulfide exhaustion during partial melting in the lunar mantle also supports the notion that the bulk silicate moon is depleted in highly siderophile elements. Based on the measured S contents and the estimated degree of melting, the estimated S contents for the mantle source of A15 green glass and A15 mare basalts is 10-23 ppm; for A17 orange glass is 25-62 ppm, for A12 mare basalts is 27-92 ppm, and for A11 basalt is 35-120 ppm. Consideration of SCSS decrease due to the presence of Ni in the sulfide melt does not change these mantle S abundance estimates for <30 wt.% Ni in the sulfide. The inferred S contents suggest that the lunar mantle is heterogeneous in terms of S. Although variable among different groups, the inferred S abundance of up to 120 ppm in the lunar mantle falls near the lower end of the S content of the depleted terrestrial mantle such as the MORB source.