Effects of H2O on the solution behavior of fluorine and chlorine in peralkaline sodium aluminosilicate glasses quenched from melts at high temperature (1400 °C) and pressure (1.5 GPa) were studied by combining solubility measurements and Raman spectroscopy. With increasing H2O content from 0 to ~10 wt%, the fluorine solubility increases from 3.3 to 4.4 mol% in Al-free glasses and from 6.3 to 9.3 mol% in Al-rich glasses (10 mol% Al2O3). In contrast, in the same H2O concentration range the chlorine solubility decreases from 5.7 to 3.4 mol% in Al-free glasses and from 3.6 to 1.7 mol% in Al-rich glasses. In Al-free glasses, interaction between H2O and the silicate to depolymerize the network is Q4 + H2O " Q2(H) and Q3 + H2O " Q2(H). The effect of water on silicate melt structure is different in halogen-bearing melts because in hydrous melt systems both F and Cl can act to depolymerize the melt further. For fluorine, this is accomplished via formation of Si-F, Al-F, and Na-F bonding in addition to Si-OH, whereas in chlorine-bearing hydrous melts, there is no interaction between Si4+ and Cl-. The halogen solubility in the magmatic liquid influences mineral/melt partition coefficients of chlorine and fluorine and implies partition coefficients different from unity. Moreover because of the contrasting effects of H2O on fluorine and chlorine solubility, the Cl/F ratio in magmas formed in water-rich environments such as subduction zones can be a sensitive indicator of H2O content during arc magmas genesis. Transport properties of melts, such as diffusion and viscosity, also vary differently in halogen-bearing hydrous melts compared with halogen-free systems. Moreover, those effects on melt properties are the strongest in F-bearing systems.
Bibliographical noteFunding Information:
This research was conducted with partial support from NSF grant EAR-1212754 (to B.O.M. and C.D.), and NAI grant to the Geophysical Laboratory. Reviews by Hélène Balcone-Boissard, two anonymous reviewers, and A.E. Daniel Neuville were very helpful. We also thank J.T. Armstrong and K. Crispin for their help with the optimization of the SEM/EDS analytical protocol, and R. Bowden for her help in analyzing samples H2O content on the mass spectrometer.
- aluminosilicate melt
- solution mechanism