Effect of variable Na/K ratio on CO2 solubility in slab-derived rhyolitic melts – An experimental study

Monday, 15 December 2014
Michelle Muth, Megan S Duncan and Rajdeep Dasgupta, Rice University, Houston, TX, United States
Subduction zones are an important part of global carbon cycling and understanding the agent through which carbon is transported from the slab to the mantle wedge is key. Previous work has constrained the role of P, T, and melt-H2O content on CO2 solubility in a model, slab-derived rhyolitic partial melt [1, 2]. However, the effect of composition, i.e., the variation in cation ratios on CO2 solubility remains unknown. In particular, variation in Na/K, which can be highly variable in rhyolitic melts, can have a large effect on CO2 solubility [3, 4]. Here we investigate the effect of variable alkali ratio on the CO2 solubility in rhyolitic melt at sub-arc depths.

CO2-saturated experiments were conducted at 3 GPa, 1300 °C on 3 rhyolite compositions, similar to low-degree partial melt of pelitic sediments, with fixed total alkalis (Na2O + K2O ~11.5 wt.%, volatile-free), but molar Na# = Na/(Na+K) varying from 0.15 to 0.88. Experimental run products show glasses with void spaces, suggesting the presence of CO2-rich fluid during experiments. The glasses were analyzed using EPMA and FTIR spectroscopy. All glasses show peaks for H2O at ~3540 cm-1. CO2 is dissolved as molecular CO2 (CO2mol.) at ~2350 cm-1 and carbonate (CO32-). Peaks for CO32- shift in position and shape: high Na# glass peaks are asymmetrical and closely spaced at ~1504 cm-1 and 1439 cm-1, while low Na# glass peaks are symmetrical and separated at ~1425 cm-1 and ~1530 cm-1.

Our data show a strong positive correlation between Na# and CO2tot. contents from 1.0 wt.% for Na# = 0.15 to 3.0 wt.% for Na# = 0.88, at a fixed melt NBO/T ≈ 0.07. CO2mol./CO2tot. decreases with increasing Na#, from 0.74 to 0.09. These data suggest that Na has a larger effect than K on carbonate dissolution in rhyolitic melts. This is in agreement with the low P experiments on phonotephrite compositions [3], but in contrast with the observation for synthetic nephelinitic compositions [4], possibly indicating the effect of Na/K on CO2 solubility is influenced by other major element concentrations. Our results suggest that compositional variability in silicic melts must be considered to place constraints on the limit of CO2 transfer in subduction zones.

[1] Duncan & Dasgupta (2014) GCA 124, 328-347. [2] Duncan & Dasgupta (sub.). [3] Vetere et al. (2014) CMP 167, 1014. [4] Morizet et al. (2014) GCA 141, 45-61.