Re–Os fractionation by sulfide melt–silicate melt partitioning: A new spin

Experiments have been done to assess the role of residual sulfide (melt and crystalline monosulfide solid solution, or MSS) in controlling the behavior of rhenium and osmium during basalt petrogenesis. In order to facilitate efficient separation of the sulfide and silicate phases, sulfide melt–silicate melt partitioning experiments were done at 1200 °C, 105 Pa and high gravitational acceleration using a furnace mounted in a centrifuge. Additional (static) high pressure experiments (1.5 GPa; 1200, 1250 °C) were performed to measure both sulfide–silicate and MSS–sulfide partitioning. Results from high pressure experiments show that MSS–sulfide melt partitioning does not significantly fractionate Re from Os, so the behavior of these elements during mantle melting will not be sensitive to the identity of the residual sulfide phase. In contrast, most experiments produced minimum values of Dsulfide/silicate, Os/Dsulfide/silicate, Re > 1, with some values > 150, which is the requisite minimum to produce the observed Re/Os fractionation in mantle-derived magmas. Dsulfide/silicate for Re varies over a wide range, from > 20,000 to ∼ 20, depending on the fO2 − fS2 conditions imposed on an experiment, and defines two coherent groupings, based on fS2, described by the expressions:log Dsulfide/silicate = − 6.95(± 0.20) + 2.39(± 0.04) {1/2 log fS2 − 1/2 log fO2} (ΔFMQP > 1.5)log Dsulfide/silicate = − 7.79(± 0.44) + 2.90(± 0.11) {1/2 log fS2 − 1/2 log fO2} (ΔFMQP < 1)in which ΔFMQP is the difference between the log fS2 of the sample and that of the fayalite–magnetite–quartz–pyrrhotite buffer at the same temperature. of Dsulfide/silicate for Re which apply to a specific igneous system can be predicted using the results of this study, combined with estimates of the prevailing fO2 and fS2. By this approach, Dsulfide/silicate determined from the rhenium content of coexisting sulfide globules and silicate glass for MORB (FAMOUS locality) is similar to predicted values, whereas Dsulfide/silicate measured for Loihi is somewhat lower than expected. Generally speaking, values of Dsulfide/silicate for Re of ∼ 400–800 are expected for melting of oceanic basalt sources, with deviations in bulk partitioning of Re corresponding to the effect of fO2 on mineral–melt partition coefficients, or by changes in modal sulfide content. Using the experimental partitioning data, along with reasonable estimates of source sulfur content and degree of melting, the Re–Yb systematics for MORB can be reproduced from a depleted mantle composition provided magmas undergo extensive fractional crystallization and/or magma mixing.