Solubility, diffusivity, and O isotope systematics of H2O in rhyolitic glass in hydrothermal temperature experiments

Abstract

In many volcanic settings, eruptive deposits experience prolonged cooling in the presence of water, such as in subglacial or submarine eruptions. Under these conditions, volcanic glass will rehydrate and record the isotopic composition of the water. This isotope exchange is moderated by H2O solubility and diffusivity in the glass. In this study, we report results from glass hydration experiments conducted at 175–375 °C to constrain H2O solubility and diffusivity under these hydrothermal conditions over timescales lasting hours to months. We use anhydrous high and low silica rhyolites as well as hydrous high silica rhyolite (perlites) with isotopically labeled water as starting materials. Measurements of bulk H2O by TC/EA of experimental glasses provide minimum H2O solubility estimates. High-Si rhyolitic glass has an H2O solubility between 2.75 wt.% (175 °C, 0.89 MPa) and 4.1 wt.% (375 °C, 21 MPa) while low-Si rhyolite H2O solubility is uniformly ∼0.5 wt.% higher at each temperature. We find a roughly linear relationship of solubility vs 1/T that is ∼1–2 wt.% greater than extrapolations from magmatic temperature solubility relationships. Furthermore, three independent methods of diffusion modeling – one in situ and two mass balance approaches – all produce H2O diffusivity (DH2O) values that up to 5.5 times greater than predicted by extrapolation of the 1/T – DH2O relationships above 400 °C to the experimental P-T-XH2O conditions. In situ H2O profiles in rhyolite particles measured by NanoSIMS have the characteristic “snowplow ” functional form that arises from the H2O concentration dependence of DH2O. We cannot detect diffusively driven kinetic fractionation of D relative to H with the NanoSIMS data. Diffusion and mass balance calculations that fit TC/EA time series of bulk H2O in particles of a single size distribution, and calculations that reconcile two sets of different sized particles at a single experimental duration, return similar DH2O constraints. We also present time series δ18O of bulk glass (δ18Obulk) and the δ18O of water-in-glass (δ18Owig) measurements, which indicate that molecular water (H2Om) dissolved in the glass is the primary driver of subsequent oxygen isotope exchange between glass and an external fluid. Local equilibrium between the δ18Owig and the δ18Obulk is rapidly established and ranges from approximately −14‰ at 175 °C to −10‰ at 375 °C. Both the δ18Obulk and δ18Owig then increase with time moving slowly towards estimated bulk glass δ18O equilibrium with the external experimental water. Oxygen isotope exchange between glass and a fluid is therefore strongly linked to – and is limited by – H2O diffusivity, which is slower at lower P-T conditions and lower H2O solubilities as H2Om diffusion is the main exchange mechanism.