As a major volatile in volcanic systems, water has a significant influence on the rheological properties of silicic magmas. This is especially so at minor water contents (< 0.5 wt%) relevant to the emplacement of rhyolitic lavas. Rheological measurements on water-bearing glasses are rare owing to the kilobar pressures needed to dissolve significant amounts of water in silicic melts and the likelihood of vesiculation occurring at high temperatures changing the physical properties of the melt. To investigate the influence of water on the viscosity of natural rhyolitic magmas, a novel strategy has been adopted employing both parallel-plate and micropenetration viscometry techniques.
Viscosity measurements have been performed on three types of materials: raw water-bearing obsidians; annealed anhydrous obsidian glasses and hydrothermally-hydrated obsidians. Ten natural rhyolitic glasses (of peraluminous, peralkaline and calc-alkaline compositions) were used in experiments: seven originated from lava flows and contained < 0.2 wt% water, two samples were F-rich originating from pyroclastic successions, and one was an obsidian cobble with 1.5 wt% water also associated with pyroclastic units. Anhydrous glasses were prepared from pre-foamed fragments of the original sample at 1 atmosphere pressure and 1650°C. One peralkaline and one calc-alkaline rhyolite were hydrothermally-hydrated with 0.5 and 1.0 wt% added water at high pressures and temperatures (1300°C at 3 kbar for 60 - 90 hours) in an internally-heated pressure vessel (CRSCM-CNRS Orleans, France). Melt compositions and water contents (FTIR) were stable during viscometry.
Results show that the addition of small amounts of water cause a decrease in activation energies of viscous flow, viscosities and glass transition temperatures (Tg). Furthermore, a loss of 1.5 wt% water during subvolcanic degassing of a rhyolitic magma causes Tg to increase by 315°C and viscosities to increase by 5 - 6 orders of magnitude. A marked non-linear decrease in Tg, and viscosity (at a constant temperature) occur with increasing water contents, from 0 to 1.5 wt% water. The widely-used Shaw calculation scheme deviates significantly from experimental data, whereas the recent Hess and Dingwell scheme, based on synthetic leucogranitic melts, closely agrees with experimental data for calc-alkaline rhyolites and andesites. This scheme, however, should not be used for water-bearing peralkaline rhyolitic melts.
These experimentally determined viscosities span conditions corresponding to the near-surface pre-eruptive degassing of magmas to post-eruptive cooling of lava flows. Glasses from magmatic intrusions contain 1.0 - 1.5 wt% water and lava flow obsidians never contain more than 0.5 wt% water. These results could be used to improve existing calculations of the viscosity of water-bearing magmas used for numerical modelling of volcanic eruptions. Furthermore, these data permit the testing of calculational models for viscosity largely based on synthetic systems.