Multiple sulfur isotope composition of oxidized Samoan melts and the implications of a sulfur isotope ‘mantle array’ in chemical geodynamics

•Sulfide segregation under a coexistence of dissolved sulfides and sulfates.•34S/32S for dissolved sulfides is correlated to the 87Sr/86Sr of the glasses relative abundance of sulfate is not buffered in the melt.•EM-1 and EM-2 mantle endmembers must share a common sedimentary nature.

To better address how subducted protoliths drive the Earth's mantle sulfur isotope heterogeneity, we report new data for sulfur (S) and copper (Cu) abundances, S speciation and multiple S isotopic compositions (32S, 33S, 34S, 36S) in 15 fresh submarine basaltic glasses from the Samoan archipelago, which defines the enriched-mantle-2 (EM2) endmember.Bulk S abundances vary between 835 and 2279 ppm. About 17±11%17±11% of sulfur is oxidized (S6+) but displays no consistent trend with bulk S abundance or any other geochemical tracer. The S isotope composition of both dissolved sulfide and sulfate yield homogeneous Δ33S and Δ36S values, within error of Canyon Diablo Troilite (CDT). In contrast, δ34S values are variable, ranging between +0.11 and +2.79‰ (±0.12‰ 1σ) for reduced sulfur, whereas oxidized sulfur values vary between +4.19 and +9.71‰ (±0.80‰, 1σ). Importantly, δ34S of the reduced S pool correlates with the 87Sr/86Sr ratios of the glasses, in a manner similar to that previously reported for South-Atlantic MORB, extending the trend to δ34S values up to +2.79±0.04‰+2.79±0.04‰, the highest value reported for undegassed oceanic basalts.As for EM-1 basalts from the South Atlantic ridge, the linear δ34S–87Sr/86Sr trend requires the EM-2 endmember to be relatively S-rich, and only sediments can account for these isotopic characteristics. While many authors argue that both the EM-1 and EM-2 mantle components record subduction of various protoliths (e.g. upper or lower continental crust, lithospheric mantle versus intra-metasomatized mantle, or others), it is proposed here that they primarily reflect sediment recycling. Their distinct Pb isotope variation can be accounted for by varying the proportion of S-poor recycled oceanic crust in the source of mantle plumes.