Experimental Constraints on CO2 Solubility in Rhyolitic Slab Melts – Implications for Carbon Flux in Subduction Zone

Understanding the fate of carbon in subduction zones is critical to understand carbon cycle on a global scale. The amount of carbonate and reduced (organic) carbon that is subducted and the amount of CO₂ that is released from arc volcanoes vary for subduction zones around the globe. If the agent of carbon transfer from slab to sub-arc mantle is a partial melt of either ocean-floor sediments [1] or hydrous basalt [2], we need to know the solubility of CO₂ in rhyolitic slab melt to constrain the flux of carbon in subduction zones.

Our previous experiments have constrained CO₂ content in silicic slab melts as a function of P (1.5-3.0 GPa) and melt H₂O content (0.5-3.0 wt.%) [3]. Here we extend our experiments to constrain the effect of temperature (1100-1400 °C) and fO₂ (CO₂ vapor-saturated [3] and graphite-saturated) on CO₂ solubility and speciation in natural rhyolitic melts. From our data, we constructed empirical and thermodynamic models to calculate CO₂ content in slab melts at P and T appropriate for the sub-arc region of the subducting slab at variable fO₂ [4]. These experiments and models show that CO₂ solubility increases with increasing P, fO₂, and melt H₂O contents to ~3.5 wt.%, while there is a only slight increase in CO₂ solubility with increasing T though the effect is much smaller.

Our study constrains the extent of C-cycling to the deep interior and to the arc source for graphite-saturated domains of the downgoing crust. Further, there is a general correspondence between CO₂ solubility in slab-derived, rhyolitic melts at sub-arc depth with measured CO₂ outflux at arcs [5]. For hotter slabs (T>800 °C) the calculated CO₂ contents using our thermodynamic model, for example, are 1.5-3.4 wt.% for a low-H₂O melt generated near the FMQ buffer and correspond to arc fluxes of 50-500 × 10⁹ mol/yr. For colder slabs (T<800 °C) the calculated CO₂ contents are 0.9-1.6 wt.% for a low-H₂O melt generated near the FMQ buffer and correspond to arc fluxes of 0.1-15 ×1 0⁹ mol/yr. This correspondence suggests that slab-derived silicic melt may be the chief agent of C-transport from slab to mantle wedge.

[1] Plank and Langmuir (1993) Nature, 362, 739-743; [2] Prouteau et al. (2001) Nature, 410, 197-200; [3] Duncan and Dasgupta. (2014) GCA, 124, 328-347; [4] Syracuse et al. (2010) PEPI, 183, 73-90; [5] Fischer et al. (2008) Geochem. J., 42, 21-38.