The Fate of Sulfur During Fluid-Present Melting of Subducting Basaltic Crust at Variable Oxygen Fugacity

JOURNAL OF PETROLOGY VOLUME 55 NUMBER 6 PAGES 1019^1050 2014 doi:10.1093/petrology/egu016 The Fate of Sulfur During Fluid-Present Melting of Subducting Basaltic Crust at Variable Oxygen Fugacity 1 DEPARTMENT OF EARTH SCIENCE, RICE UNIVERSITY, 6100 MAIN STREET, MS-126, HOUSTON, TX 77005, USA 2 INSTITUT DES SCIENCES DE LA TERRE D’ORLEANS (ISTO), UNIVERSITE D’ORLEANS, CNRS: UMR7327, INSU, BUREAU DE RECHERCHES GEOLOGIQUES ET MINIERES (BRGM), ORLEANS, FRANCE RECEIVED AUGUST 13, 2013; ACCEPTED MARCH 14, 2014 To constrain the effect of redox state on sulfur transport from subducting crust to mantle wedge during fluid-present melting and the stability of sulfur-bearing phases in the downgoing ocean crust, here we report high-pressure phase equilibria experiments on a H2Osaturated mid-ocean ridge basalt with 1wt % S at variable oxygen fugacity (f O2 ). Double-capsule experiments were conducted at 2·0 and 3·0 GPa and 950^10508C, using Co^CoO, Ni^NiO, NixPd1^x^NiO, and Fe2O3^Fe3O4 external f O2 buffers. Sulfur content at sulfide saturation (SCSS) or sulfur content at sulfate saturation (SCAS) of experimental hydrous partial melts was measured by electron microprobe. All experiments were fluid-saturated and produced either pyrrhotite- or anhydrite-saturated assemblages of silicate glass, clinopyroxene, garnet, and rutile or titanomagnetite,  amphibole  quartz  orthopyroxene. The silicate partial melt composition evolves from rhyolitic at 9508C to trachydacitic and trachyandesitic at 10508C with increasing f O2. At pyrrhotite saturation, melt S contents range from 30 ppm S at f O2 5FMQ ^ 1 to 500 ppm S at FMQ5f O2 FMQ þ1·1, whereas at anhydrite saturation (f O2  FMQ þ 2·5) melt S concentrations range from 700 ppm S to 0·3 wt % S. Mass-balance calculations suggest that the aqueous fluid phase at equilibrium may contain as much as 15 wt % S at 10508C at pyrrhotite saturation (f O2  FMQ þ1·1), in agreement with previous estimates, and up to 8 wt % S at anhydrite saturation. Our data also show fluid=melt decreases markedly with increasing f O2 at pyrrhotite that DS saturation, from several thousand at f O2 5FMQ ^ 1 to  200^400 at FMQ5f O2 FMQ þ1·1, owing to the increase of melt S confluid=melt is very low (5100) but intent. At anhydrite saturation, DS creases with decreasing temperature, in an opposite way to previous *Corresponding author. Present address: Institut des Sciences de la Terre d’Orle¤ans (ISTO), Universite¤ d’Orle¤ans, CNRS: UMR7327, INSU, Bureau de Recherches Ge¤ologiques et Minie'res (BRGM), Orle¤ans, France. Telephone: þ33-2-38-25-53-99. Fax: þ33-2-38-44-49-76. E-mail: observations at pyrrhotite saturation. As a consequence, at fluid=melt might be in the range 200 100, irrespective T  9008C, DS of f O2. The present study confirms that slab partial melts saturated with pyrrhotite are unable to efficiently transport S from slab to mantle wedge, and suggests that slab partial melts in equilibrium with anhydrite also have very limited power to enrich the mantle wedge in S. Importantly, slab-derived aqueous fluids appear to be efficient vectors for the transport of sulfur from slab to mantle wedge at all f O2 .Therefore, S transfer from ocean crust to wedge mantle is not f O2 dependent and could take place over a range of f O2 conditions, and oxidized slab conditions are not necessarily required to enrich the mantle wedge in S. Finally, depending on the initial amount of sulfur in the slab, the proportion of residual anhydrite and pyrrhotite in the dehydrated slab below the region of formation of arc magmas is likely to be significant and may efficiently be recycled into the deep mantle. KEY WORDS: oxygen fugacity; slab-derived fluid; slab partial melt; subduction zone; sulfur cycle; anhydrite; pyrrhotite I N T RO D U C T I O N Sulfur (S) is one of the major volatiles that control fundamental magmatic processes, including (chalcophile) elemental partitioning, redox evolution of magma and melt^mantle systems, mantle metasomatism, magma degassing, and the dynamics of volcanic eruptions. One of the main tectonic settings where surficial reservoirs and ß The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@ oup.com SE¤BASTIEN JE¤GO1,2* AND RAJDEEP DASGUPTA1 JOURNAL OF PETROLOGY VOLUME 55 JUNE 2014 (Cervantes & Wallace, 2003). Therefore, S is likely to be scavenged in significant amounts by hydrous slab-derived fluids, from either sulfide or sulfate mineral phases contained in the subducting lithologies. However, it remains unclear whether fluid-mediated sulfur transfer happens at relatively low temperatures and shallow depths or at relatively high temperatures at sub-arc depths. If the former, it is difficult to envision how such low-temperature, shallow fluids, perhaps released in forearcs, contribute to arc volcanism. If the latter, then the presence of hydrous fluid is also expected to trigger partial melting and the relative mobility of sulfur in fluid versus melt is unconstrained. Thus the relative contribution of slab fluid versus slab partial melt in sulfur transfer needs to be constrained at subarc depth conditions. Recently, two studies have provided experimental data that constrain the transport of S at sub-arc depths during hydrous partial melting of the subducting slab (Je¤go & Dasgupta, 2013; Prouteau & Scaillet, 2013). However, even though their experimental results are partially mutually consistent, their conclusions about the capacity of slab partial melts to efficiently transport S at high pressure are divergent and thus provide contrasting views on the transfer of S from the slab to the mantle wedge. Starting material compositions and ranges of pressure^temperature conditions are very similar in both studies, although Prouteau & Scaillet (2013) performed experiments at 700^9508C with addition of 1^2 wt % elemental S, whereas Je¤go & Dasgupta (2013) conducted experiments at 800^10508C with 1wt % bulk S added as pyrite (FeS2). Also, in the Je¤go & Dasgupta (2013) experiments, the fO2 was imposed by using Ni^NiO (NNO) and Co^CoO (CCO) external oxygen buffers, so that all experimental charges were pyrrhotite-saturated, but variably reduced. Prouteau & Scaillet (2013) did not control the fO2 (except for one experiment, buffered with a Pt^graphite capsule at NNO ^ 2), which introduces significant uncertainty about the actual redox state of their experiments and thus the interpretation of their data. In most of their runs, the experimental products show the coexistence of pyrrhotite and anhydrite, implying that the fO2 domain may be close toçor right onçthe sulfide^sulfate transition (NNO þ1); that is, more oxidized than the Je¤go & Dasgupta (2013) experiments. In both studies, the composition of the partial melt, quenched to glass, evolves from rhyolitic to dacitic with increasing temperature and melting degree. Je¤go & Dasgupta (2013) reported ve (...truncated)