2:15 PM - 2:30 PM
[SVC28-03] Nature of slab-fluid sulfur indicated from sulfur isotope and trace element systematics of olivine-hosted melt inclusions from Kyushu Island volcanoes, Japan
★Invited Papers
Keywords:subduction zone, seawater, slab-derived fluid, trace element, volatile element
The subduction zone magmatism is a system of sulfur cycle providing its flux to the Earth atmosphere. Sulfur behaves in the most complex manner among magmatic volatile elements, with its multiple valencies and its partitioning into melt, gas, fluids and S-bearing minerals. The stable isotopes of sulfur (e.g., 34S/32S, expressed as δ34S) are useful indicator to assess the source of sulfur and chemical exchange among sulfur species. Yet, the nature of sulfur in slab-fluid metasomatizing sub-arc mantle and its isotopic composition are highly debated topics. For example, recent studies of metamorphic rocks and thermodynamic modelling proposed two opposite hypotheses: A) Slab-derived sulfur is released as reduced sulfur species and slab-fluid has negative δ34S value (e.g., Li et al., 2020), B) Slab-fluid is sulfate dominant and has various sulfur isotope compositions from negative to positive δ34S values (e.g., Walters et al., 2019). The one reason of this disparity is the complexity of the fluid generation and transport processes and the diversity of the source rocks. Therefore, those hypotheses must be evaluated with a novel approach to determine the isotopic composition of slab-derived component and magma source materials from the magmatic products.
Determination of sulfur isotopic composition of primitive magma involves at least two problems. First, magmatic volatiles are degassed after the eruption because of their strong pressure-dependence of solubility in silicate melt (e.g., Dixon et al. 1995). This degassing process causes significant sulfur isotope fractionation of magma (e.g., Marini et al., 2011). Second, possible crustal magma process, such as fractional crystallization, magma mixing, and crustal assimilation, obscure the primitive magma composition (e.g. Sakuyama, 1979; Hildreth and Moorbath, 1988). At present, a wide range of δ34S values have been reported from arc volcanic rocks, from values close to the MORB mantle (δ34S = -0.91 ± 0.50‰) to values as high as seawater (δ34S = ~+21), but most are from whole rocks that have experienced degassing and are not representative of the primitive values (e.g., Ueda and Sakai, 1984; Ohmoto, 2020). Instead, the systematic study of sulfur isotope composition in olivine-hosted magmatic inclusions is much more revealing, as the olivine can prevent diffusive sulfur-loss through the host crystal (e.g. Bucholz et al., 2013).
Kyushu Island in Japan hosts 11 active volcanoes and is perhaps the most suitable site for the study of the sulfur cycle for its availability of recent volcanic deposits and its many highly monitored volcanoes. We investigated sulfur and sulfur isotope compositions of the magma source of Kyushu Island arc using olivine-hosted melt inclusions in mafic Holocene tephras and lavas, from 8 volcanoes going from Northern Kyushu with Oninomi, Yufu, Kuju, and Aso, to Southern Kyushu volcanoes such as Kirishima-Ohachidake, Kirishima-Shinmoedake, Sumiyoshi-ike, and Kaimondake, and one back arc volcano, Fukue-Onidake. We measured major, trace and volatile elements and S isotopes (δ34S) in 130 melt inclusions. Melt compositions were basalt to andesite with SiO2 ranging from 40.3 to 60.7 wt. %. The obtained δ34S values in melt inclusions range between -0.32±0.79 ‰ and +9.43±0.47 ‰ (2σ) with high S concentrations up to ~3800 ppm. For each edifice, we identified the least degassed and least differentiated compositions and we selected the melt inclusions the closest to the primitive melt composition, based on volatile and trace elements systematics.
Positive correlations between δ34S and indicators of slab flux (e.g., Th/Nb, Pb/Ce) suggest that slab fluids play an important role in the high δ34S signature of Kyushu arc magmas. In our dataset, the δ34S does not correlate with other indicators, such as redox state of mantle (V/Sc), melting of slab (Cl/F), residual hydroxyl minerals or breakdown (F/Nd), and degree of mantle melting (La/Sm). Thus, the δ34S variation must be related to the compositional variation of slab component. The positive correlation between δ34S and Rb/Ba requires at least two mixing endmembers in addition to the deplete mantle: low-Rb/Ba material, such as sediment, and high-Rb/Ba material, which could potentially be seawater. The lack of a negative δ34S component suggests that the dominant sulfur mobility from the slab is carried out by sulfate rather than sulfide, and the negative δ34S of sulfide is perhaps not incorporated into the slab flux.
Determination of sulfur isotopic composition of primitive magma involves at least two problems. First, magmatic volatiles are degassed after the eruption because of their strong pressure-dependence of solubility in silicate melt (e.g., Dixon et al. 1995). This degassing process causes significant sulfur isotope fractionation of magma (e.g., Marini et al., 2011). Second, possible crustal magma process, such as fractional crystallization, magma mixing, and crustal assimilation, obscure the primitive magma composition (e.g. Sakuyama, 1979; Hildreth and Moorbath, 1988). At present, a wide range of δ34S values have been reported from arc volcanic rocks, from values close to the MORB mantle (δ34S = -0.91 ± 0.50‰) to values as high as seawater (δ34S = ~+21), but most are from whole rocks that have experienced degassing and are not representative of the primitive values (e.g., Ueda and Sakai, 1984; Ohmoto, 2020). Instead, the systematic study of sulfur isotope composition in olivine-hosted magmatic inclusions is much more revealing, as the olivine can prevent diffusive sulfur-loss through the host crystal (e.g. Bucholz et al., 2013).
Kyushu Island in Japan hosts 11 active volcanoes and is perhaps the most suitable site for the study of the sulfur cycle for its availability of recent volcanic deposits and its many highly monitored volcanoes. We investigated sulfur and sulfur isotope compositions of the magma source of Kyushu Island arc using olivine-hosted melt inclusions in mafic Holocene tephras and lavas, from 8 volcanoes going from Northern Kyushu with Oninomi, Yufu, Kuju, and Aso, to Southern Kyushu volcanoes such as Kirishima-Ohachidake, Kirishima-Shinmoedake, Sumiyoshi-ike, and Kaimondake, and one back arc volcano, Fukue-Onidake. We measured major, trace and volatile elements and S isotopes (δ34S) in 130 melt inclusions. Melt compositions were basalt to andesite with SiO2 ranging from 40.3 to 60.7 wt. %. The obtained δ34S values in melt inclusions range between -0.32±0.79 ‰ and +9.43±0.47 ‰ (2σ) with high S concentrations up to ~3800 ppm. For each edifice, we identified the least degassed and least differentiated compositions and we selected the melt inclusions the closest to the primitive melt composition, based on volatile and trace elements systematics.
Positive correlations between δ34S and indicators of slab flux (e.g., Th/Nb, Pb/Ce) suggest that slab fluids play an important role in the high δ34S signature of Kyushu arc magmas. In our dataset, the δ34S does not correlate with other indicators, such as redox state of mantle (V/Sc), melting of slab (Cl/F), residual hydroxyl minerals or breakdown (F/Nd), and degree of mantle melting (La/Sm). Thus, the δ34S variation must be related to the compositional variation of slab component. The positive correlation between δ34S and Rb/Ba requires at least two mixing endmembers in addition to the deplete mantle: low-Rb/Ba material, such as sediment, and high-Rb/Ba material, which could potentially be seawater. The lack of a negative δ34S component suggests that the dominant sulfur mobility from the slab is carried out by sulfate rather than sulfide, and the negative δ34S of sulfide is perhaps not incorporated into the slab flux.