No effect of H2O degassing on the oxidation state of hydrous rhyolite magmas: a comparison of pre- and post-eruptive Fe2+ concentrations in six obsidian samples from the Mexican and Cascade arcs

The extent to which degassing affects the oxidation state of arc magmas is widely debated. Several researchers have examined how degassing of mixed H-C-O-S-Cl fluids may change the Fe³⁺/Fe^T ratio of magmas, and it has been proposed that degassing may induce either oxidation or reduction depending on the initial oxidation state. A commonly proposed oxidation reaction is related to H₂O degassing: H₂O (melt) + 2FeO (melt) = H₂ (fluid) + Fe₂O₃ (melt). Another mechanism by which H₂O degassing can affect the iron redox state is if dissolved water affects the activity of ferrous and/or ferric iron in the melt. Although Moore et al. (1995) presented experiments showing no evidence of an affect of dissolved water on the activity of the ferric-ferrous ratio in silicate melts, other experimental results (e.g., Baker and Rutherford, 1996; Gaillard et al., 2001; 2003) indicate that there may be such an effect in rhyolite liquids. It has long been understood that rhyolites, owing to their low total iron concentrations, are more sensitive than other magma types to degassing-induced change in redox state. Therefore, a rigorous test of whether H₂O degassing affects the redox state of arc magmas is best evaluated on rhyolites. In this study, a comparison is made between the pre-eruptive (pre-degassing) Fe²⁺ concentrations in six, phenocryst-poor (<5%), fresh, glassy obsidian samples with their post-eruptive (post-degassing) Fe²⁺ concentrations. Near-liquidus co-precipitation of two Fe-Ti oxides allows the pre-eruptive oxygen fugacity and temperature to be calculated in each rhyolite using the oxygen barometer and thermometer of Ghiorso and Evans (2008). Temperatures range from 793 (± 19) to 939 (± 15) °C, and ΔNNO values (log₁₀fO₂ of sample - log₁₀fO₂ of Ni-NiO buffer) range from -0.4 to +1.4. These ΔNNO values allow the ferric-ferrous ratio in the liquid to be calculated, using the experimental calibration of Kress and Carmichael (1991), which relates melt composition (not including dissolved water), oxygen fugacity and temperature to melt ferric-ferrous ratios. With temperature known, the plagioclase-liquid hygrometer of Lange et al. (2009) was applied and maximum melt water concentrations range from 4.2 to 7.5 wt%. Both the oxidation state and water concentration are known prior to eruption, at the time of phenocryst growth. After eruption, the rhyolites lost nearly all of their volatiles, as indicated by the low loss on ignition values (LOI ≤ 0.7 wt%). In order to test how much oxidation of ferrous iron occurred as a consequence of that degassing, we measured the ferrous iron concentration in the bulk samples by titration, using the Wilson (1960) method, which was successfully tested again three USGS and one Canadian Geological Survey standards. Our results indicate no detectable change within analytical error between pre- and post-eruptive FeO concentrations, with an average deviation of 0.09 wt% and a maximum deviation of 0.15 wt%. Our results show that H₂O degassing has no effect on the redox state of rhyolite magmas, which requires that dissolved water has no resolvable affect on the activity ratio of the iron oxide components in melt.