The partitioning of sulfur between multicomponent aqueous fluids and felsic melts

Abstract

Sulfur partitioning between melt and fluid phase largely controls the environmental impact of volcanic eruptions. Fluid/melt partitioning data also provide the physical basis for interpreting changes in volcanic gas compositions that are used in eruption forecasts. To better constrain some variables that control the behavior of sulfur in felsic systems, in particular the interaction between different volatiles, we studied the partitioning of sulfur between aqueous fluids and haplogranitic melts at 200 MPa and 750–850 °C as a function of oxygen fugacity (Ni–NiO or Re–ReO2 buffer), melt composition (Al/(Na + K) ratio), and fluid composition (NaCl and CO2 content). The data confirm a first-order influence of oxygen fugacity on the partitioning of sulfur. Under “reducing conditions” (Ni–NiO buffer), Dfluid/melt is nearly one order of magnitude larger (323 ± 14 for a metaluminous melt) than under “oxidizing conditions” (Re–ReO2 buffer; 74 ± 5 for a metaluminous melt). This effect is likely related to a major change in sulfur speciation in both melt and fluid. Raman spectra of the quenched fluids show the presence of H2S and HS− under reducing conditions and of SO42− and HSO4− under oxidizing conditions, while SO2 is undetectable. The latter observation suggests that already at the Re–ReO2 buffer, sulfur in the fluid is almost completely in the S6+ state and, therefore, more oxidized than expected according to current models. CO2 in the fluid (up to xCO2 = 0.3) has no effect on the fluid/melt partitioning of sulfur, neither under oxidizing nor under reducing conditions. However, the effect of NaCl depends on redox state. While at oxidizing conditions, Dfluid/melt is independent of xNaCl, the fluid/melt partition coefficient strongly decreases with NaCl content under reducing conditions, probably due to a change from H2S to NaSH as dominant sulfur species in the fluid. A decrease of Dfluid/melt with alkali content in the melt is observed over the entire compositional range under reducing conditions, while it is prominent only between the peraluminous and metaluminous composition in oxidizing experiments. Overall, the experimental results suggest that for typical oxidized, silicic to intermediate subduction zone magmas, the degassing of sulfur is not influenced by the presence of other volatiles, while under reducing conditions, strong interactions with chlorine are observed. If the sulfur oxidation state is preserved during an explosive eruption, a large fraction of the sulfur released from oxidized magmas may be in the S6+ state and may remain undetected by conventional methods that only measure SO2. Accordingly, the sulfur yield and the possible climatic impact of some eruptions may be severely underestimated.