Partitioning of carbon between Fe-rich alloy melt and silicate melt in a magma ocean - Implications for the abundance and origin of volatiles in Earth, Mars, and the Moon

Our results show that carbon in the metallic melt varies between 4.4 wt.% and 7.4 wt.% and increases with increasing temperature and modestly with increasing pressure but decreases with increasing Ni content of the alloy melt. Carbon concentration in the silicate melts, on the other hand, varies from 11 ± 1 ppm to 111 ± 7 ppm and is negatively correlated with pressure but positively correlated with temperature, the NBO/T, the oxygen fugacity and the water content of the silicate melts. Raman and FTIR results show that at our experimental conditions, carbon in silicate melt is dissolved both as hydrogenated species and CO32-. The calculated carbon partition coefficient DCmetal/silicate varies from 510 ± 53 to 5369 ± 217 and varies systematically as a function of P, T, fO2, water content, the composition of the silicate melt (expressed using NBO/T), and Ni content of alloy melt (XNi). The range of DCmetal/silicate measured in our study with carbonated and hydrogenated carbon species in silicate melt is similar to that reported in the literature for experiments where carbonyl complexes are the chief carbon species in silicate melts. A parameterization was derived using the data from this and existing studies such as lnDCmetal/silicate=a/T+b·P/T+c·ln(fO2)+d·(NBO/T)+e·ln(1-XNi)+fwhere a = −33,510, b = 1357, c = −0.596, d = −1.182, e = 4.15, f = 13.38, the temperature is in Kelvin, and the pressure is in gigapascal. Using this parameterization and the estimated conditions for the base of the MOs, the average DCmetal/silicate value for Earth, Mars, and the Moon can be predicted. The deep MO of Earth is predicted to cause the strongest depletion of its silicate carbon budget, closely followed by Mars with intermediate depth MO, and then the Moon with a shallow MO. We predict that the lunar mantle carbon budget, similar to that of the Earth's present-day upper mantle, might have been set by equilibrium core-mantle fractionation in MO; whereas for Earth, later processes such as ingassing from a proto-atmosphere and late-stage accretion of volatile-rich material was necessary for delivery of carbon and other volatiles. Finally, the comparison of our measured and predicted value of DCmetal/silicate for terrestrial MO with similar constraints on DNmetal/silicate from the literature suggests that the apparent depletion of nitrogen relative to carbon for the bulk silicate Earth and the Earth's upper mantle is unlikely to be caused by preferential partitioning of nitrogen to alloy melt during core formation.