Se–Te fractionation by sulfide–silicate melt partitioning: Implications for the composition of mantle-derived magmas and their melting residues

Abstract

Partitioning of Se and Te has been measured between coexisting sulfide liquid, monosulfide solid solution (MSS) and silicate melt at 0.9–1.5 GPa, 1200–1300 °C, fO2 controlled near the fayalite–magnetite–quartz buffer (FMQ-1.2 to -1.6) and 3–22 wt% FeO in the silicate melt. Both elements are highly compatible in the sulfide phase relative to silicate liquid (Dsulfide phase/silicate liquid > 600), with the identity of the sulfide dictating the sense of Se–Te fractionation. Whereas the measured DTe/DSe is ∼5–9 for sulfide liquid/silicate liquid partitioning, MSS/silicate melt partitioning fractionates Te from Se in the opposite sense, with DTe/DSe of ∼0.5–0.8. At fixed fO2, DSulLiq/SilLiq values for both Se and Te decrease ∼8-fold over the range in silicate melt FeO content investigated. The relative values of DSulLiq/SilLiq for Cu to Se increase with increasing FeO in the silicate melt, such that DCu exceeds DSe only for melts with >11 wt% FeO. Hence the standard belief that DCu>DSe as indicative of sulfide removal should be carefully assessed in the context of the FeO content of the magmas involved.Assuming a chondritic mantle Se/Te, predicted MSS and sulfide liquid compositions are generally in accord with natural mantle sulfides, in terms of their designation as MSS or sulfide liquid, based on independent criteria. However, additional variability is likely due to Te redistribution in accessory platinum group minerals (PGM), or that some sulfides are metasomatic. Calculations show that the Se/Te ratio of silicate melt derived from a sulfide liquid-saturated mantle is significantly higher, and more variable, than for silicate melt in equilibrium with residual MSS; modest sulfide liquid removal at low pressure, however, likely obscures the Se/Te fractionation imposed by the source sulfide phase. Models indicate that the composition of MORB is consistent with melts produced from sulfide-bearing sources with chondritic Se/Te, and source sulfur contents higher than estimates for depleted mantle. The calculated composition of sulfide-saturated melting residues show relatively little deviation in Se/Te if MSS is residual, but a sharp drop in this ratio for sulfide liquid control. Although the data are scattered, a portion of the peridotite array near the primitive mantle composition is consistent with model trends, and suggests control by residual MSS.