Previous page     Contents     Next page

A serious gap in our present knowledge of the properties of degassing subvolcanic silicic magmas is the description of melt viscosity at the P-T-X conditions immediately prior to and during volcanic eruptions. Modelling of the dynamics of such systems is flourishing at present and a further optimisation of models for the kinetics of melt degassing, vesiculation and fragmentation would be greatly aided by reliable viscosity data obtained under the appropriate P-T-X conditions.

The present measurements are focused on the volcanologically vital but previously under investigated region of water contents between 0.4 to 3.5 wt%. By investigating such melts at 1 bar pressure and very low temperatures of 500° to 750 °C we access a range of viscosities of 10⁹ to 10¹² Pa s which corresponds to those of calcalkaline rhyolites during the final stages of degassing, vesiculation and/or fragmentation. The results provided below are unfortunately not well-predicted by available calculation schemes based on data obtained at much higher water contents. We conclude that the discrepancies between experiment and calculation are large enough that the predictions of much modelling work regarding the degassing of rhyolitic melts, i.e. those melts involved in highly explosive, rhyolitic and dacitic volcanic systems, are in need of revision.

The viscosities of hydrous haplogranitic melts synthesized by hydrothermal fusion at 2 kbar pressure and 800° to 1100 °C have been measured at temperatures just above the glass transition and at a pressure of 1 bar using micropenetration techniques. For samples with up to 2.5 wt% H₂O, the water contents have been determined, using infrared spectroscopy obtained before and after each viscometry experiment, to be constant over the duration of the measurements. Above this water content a measurable loss of water occurs during the viscometry experiments.

The viscosity data illustrate an extremely non-linear decrease in viscosity with added water. The viscosity drops drastically with the addition of 0.5 wt% of water and then levels off at water contents of 2 wt%. An additional viscosity datapoint obtained from the analysis of fluid inclusions in a water-saturated haplogranite (HPG8) confirms a near invariance of the viscosity with the addition of water between 2 and 6 wt%.

There a number of aspects concerning the petrological significance of the present results which are worth noting here. Firstly, the severe disagreement between the model of Shaw with the present determinations means that we are not yet in a position to predict the viscosities of water-bearing granitic melts with the accuracy normally ascribed to that method. This disappointing conclusion leads to especially serious problems at water contents below 4 wt% and the discrepancy at even lower water contents is extreme (several orders of magnitude). The present observation of a much more extreme decrease of melt viscosity with water content than predicted by Shaw implies that the fluxing role of water in petrogenetic processes should be much more efficient at lower water contents than currently thought.

The present observations should have, in particular, consequences for the fate of subvolcanic calcalkaline rhyolitic magmas undergoing degassing. Many aspects of the physics of degassing, including vesicle nucleation and growth, are influenced by the non-linear response of melt properties, amongst them viscosity, to decreasing water content during the degassing process. The strong increase in viscosity during degassing of the melt can, through the rate-limiting control of viscosity on certain degassing processes, effectively act as a brake on the efficiency of degassing by diffusive means. It is for this reason that the functional form of the effect of water content on the viscosity is a critical similarity parameter in experiments involving analogue materials. The general effect of a correction of results generated from the Shaw method in the direction of the presently defined trend, combined with the likelihood of a strongly negative temperature-dependence of the water solubility, will mean that degassing processes linked kinetically to the viscosity should go much more effectively to completion than has been previously anticipated. The more effective degassing of rhyolites to be expected based on the viscosity curve of Fig. 3.7-2 may be a factor in the relatively efficient degassing inferred from the low water contents of many rhyolitic obsidians in comparison to the relatively high water contents inferred from other sources such as melt inclusions or phase equilibria. In particular, the possibility of a viscous resistance influencing the rate of bubble growth has been discussed extensively in the recent volcanological literature. It is of utmost importance to constraint the viscosity and water content at which viscous forces begin to impede the expansion of bubbles during magma ascent leading to eruption because such effects can potentially stop bubble growth leading to a fragmentation event. Such a scenario, termed a "viscosity quench" would generate brittle failure of rhyolitic foam at a certain critical viscosity. It has been suggested that this viscosity quench should occur at 10⁹ Pa s for decompression rates and timescales typical of explosive eruptions. The water contents corresponding to this viscosity value are drastically influenced by the discrepancy between the results of calculations using the Shaw method and the present data (Fig. 3.7-2). The Shaw calculation generates this critical viscosity value at water contents of 1.1 wt% for 800 °C whereas the present data indicate 0.55 wt%. Thus if we appeal to the scenario of the viscosity quench, (ignoring other possible causes of fragmentation) we observe that the magma is capable of degassing far more extensively than previously estimated.
 

Fig. 3.7-2: The influence of water on the viscosity of haplogranitic (HPG8) melt at 1 bar pressure expressed as isokoms (lines of constant viscosity). The estimated viscosity for this range of conditions using the method of Shaw at the 1010 Pa s isokom is included for comparison (line). The viscosity decrease is much more non-linear than the method predicts.