Negligible sulfur isotope fractionation during partial melting: Evidence from Garrett transform fault basalts, implications for the late-veneer and the hadean matte

Abstract

We report the quadruple sulfur isotope compositions, sulfur contents and speciation major and trace elements (including copper and chlorine abundances) of eleven basalts collected in the Garrett transform fault. We combine these data to discuss the absence of S isotopic fractionation along both partial melting and low-pressure fractional crystallization.The variations of K2O/TiO2 and La/SmN-ratios (respectively between 0.017 and 0.067, and between 0.31 and 0.59) suggest a range of depletion in Garrett lavas that includes ultra depleted samples (K2O/TiO2<0.03). The remarkable level of incompatible element depletion is consistent with re-melting of a depleted source. Contrasting with incompatible element depletion, all samples display similar S and Cu abundance (at a given major-element composition) to mid-ocean ridge basalts (MORB). This indicates that Garrett Intra Transform Lavas (ITL) are sulfide saturated as MORB are. Copper content for Garrett parental melts (MgO >8%) are ∼80 ppm, indistinguishable from MORBs. This requires their mantle sources, variably depleted in incompatible element, to host residual sulfide buffering the Cu content of all erupted melts. We calculate a minimum S content for the source of ultra-depleted Garrett lavas of 100±40ppmS, i.e. roughly a factor of 2 below the MORB mantle source.After exclusion of a single sample with Cl/K ratio >0.1 that likely experienced hydrothermal sulfide assimilation, Garrett ITLs display homogeneous δ34S, Δ33S and Δ36S values with averages of −0.68±0.08‰, +0.010±0.005‰ and −0.04±0.04‰, respectively (all 1σ, n=10). The δ34S values display no relationship with either K2O/TiO2 variations or extent sulfide fractionation. From these observations, we derive a 34S/32S fractionation factor between exsolved sulfides and sulfide dissolved in silicate melts of 1.0000±0.0003. The S isotopic fractionation during partial melting can thus be considered as negligible, and both MORBs and ITLs record the 34S/32S ratio of their mantle source.The concept of sulfide melts segregating from the mantle, sinking and being added to the core during planetary differentiation was termed the ‘Hadean Matte’. The segregation of sulfides from the mantle to the core during planetary differentiation could account for various geochemical features of the Earth's mantle. Based on S isotopic mass balance, we derive a lower and upper limit for the hadean matte. While the lower bound corresponds to a virtually negligible hadean matte, the upper limit is 3.36×1024gS (i.e. ∼10% of the bulk terrestrial S), which remains 5 to 10 times lower than previous estimates. This upper bound nonetheless requires high mantle S content >1000 ppm S before the extraction of the hadean matte. This suggestion would have chronological requirements, requiring any sulfide melt to have formed after the core extraction but before late accretion of the highly siderophile elements.