High sulfur solubility in subducted sediment melt under both reduced and oxidized conditions: With implications for S recycling in subduction zone settings

The relative enrichment of sulfur (S) observed in arc magmas when compared to MORB, reflects the addition of slab-derived S to the mantle wedge source region. However, the mechanisms and efficiency of such S recycling remain poorly constrained. In this study, sediment melting experiments have been conducted using a synthetic pelite starting composition containing similar to 7 wt% H2O and similar to 1.9 wt% S, at 3 GPa, 1050 degrees C and variable oxygen fugacity (fO(2)), to investigate the effect of fO(2) on S solubility in sediment melts. To assess temperature and concentration effects, selected experiments were repeated either at lower temperatures of 950 degrees C and 1000 degrees C, or with a higher bulk S content of similar to 4 wt%. All experiments produced hydrous rhyolitic melts, saturated with either pyrrhotite under reduced conditions or anhydrite under oxidized conditions. For 3 GPa, 1050 degrees C experiments, the sulfur content at sulfide saturation (SCSS) in melt is found to increase with decreasing fO(2), from similar to 200 ppm at FMQ-0.5 to similar to 1900 ppm at FMQ-7.5. The highest S solubility is achieved at FMQ + 1.6 where melt is saturated with both anhydrite and pyrrhotite. The sulfur content at sulfate saturation (SCAS) decreases from similar to 2600 ppm at FMQ + 1.6 to similar to 2000 ppm at FMQ + 7. Increasing either bulk S content or temperature produces a positive effect on SCSS and SCAS. Raman spectra of our experimental melts show that S exists as H2S/HS- under reduced conditions and as SO42- under oxidized conditions. The solubility minimum, i.e., the onset of transition from S2- to S6+ is estimated to occur at similar to FMQ, with full transition to S6+ by similar to FMQ + 2. While the SCAS values are in good agreement with previous reports, the distinct increase of SCSS with decreasing fO(2) (when fO(2) < FMQ) has not been observed in previous slab melting experiments. Furthermore, we report for the first time that dissolution of H2S and SO2 in hydrous rhyolitic melt follows the Fincham-Richardson relationship; and propose the definition of a hydro-sulfide capacity as C-HS = [HS]*(fO(2)/fS(2))(1/2); where [HS] is the concentration of S in melt (in ppm) dissolved as HS- and H2S. SCSS for H2S dissolution can therefore be modeled using C-HS in an analogous fashion to modeling SCSS for anhydrous melts using the sulfide capacity (C-s2), with A, the relation ln[HS](SCSS) = -Delta G degrees(Fe)(S - FeO)/RT - lnC(Hs) - lna(FeO)(melt) + lna(FeS)(sulfide). As predicted by such a model framework, we indeed observe a linear correlation between 1ogSCSS and logXFeo (the mole fraction of FeO in melt) with a slope close to -1, i.e., SCSS experiences a sharp increase when FeO in melt falls below 1 wt%. Therefore, both our experimental results and model predictions suggest hydrous low-Fe rhyolitic melt produced by sediment melting under reduced conditions has the required S solubility to account for the relative enrichment of S observed in arc magmas. (C) 2021 Elsevier Ltd. All rights reserved.