Phase separation, degassing, and resorption of H2O in phonolitic melt investigated by decompression experiments

Abstract

Explosive volcanic eruptions are driven by dynamic processes in the Earth’s interior. Within volcanic systems, magmas contain volatiles of which H2O is the most abundant. At magmatic pressures and temperatures, up to several wt% H2O can be dissolved in silicate melt. The strong pressure dependence of H2O solubility causes melt degassing by formation and growth of vesicles when a melt is decompressed by, for instance, magma ascent. The vesicle number density and the porosity of a melt are important parameters that strongly influence the dynamics of volcanic eruptions. Vesiculation decreases the melt density. This accelerates magma ascent by increased buoyancy, eventually causing explosive volcanic eruptions. Decompression experiments at high temperature and pressure are used to simulate magma ascent in the laboratory. This enables the investigation of melt degassing processes that are not directly accessible for observation in nature. Within the frame of this work, the homogeneous vesicle formation in hydrous phonolitic melt with white pumice composition of the AD79 Vesuvius eruption was investigated by means of decompression experiments. A constant high vesicle number density (~105 mm-3) independent from decompression rate was found (Study I). This contrasts nucleation theory that is commonly used to describe vesicle formation in silicate melts. Based on these new experimental data, spinodal decomposition was proposed as an alternative phase separation mechanism in phonolitic melts. This spontaneous phase separation may cause sudden melt degassing, triggering explosive eruptions, even at low decompression rates. While the decompression rate does not influence the vesicle number density in phonolitic melt, it was shown that the vesicle number density varies by one order of magnitude with initially dissolved H2O concentration (3.3–6.3 wt%) (Study II). Furthermore, a change in degassing evolution at 5.6 wt% initial H2O concentration was found. At lower concentration, the textures observed in natural volcanic products can represent the initially formed vesicles. At higher concentration or if heterogeneous nucleation on crystals is involved, vesicle textures are likely obscured by coalescence or secondary vesicle formation. Besides the investigation of melt degassing, this work focusses on the experimental problem of vesicles shrinkage during cooling (Study III), which may occur when samples are rapidly quenched to glass in order to facilitate analysis at ambient conditions. The study shows that vesicles can significantly shrink during cooling, resulting in a glass porosity less than half the melt porosity. Vesicle shrinkage is governed by the decrease in molar volume of H2O and the resorption of H2O from fluid vesicles back into the melt during cooling. Consequently, the glass porosity and H2O concentration in the glass do not necessarily represent the melt porosity and the H2O concentration in the melt before cooling. This poses a problem in the interpretation of degassing processes derived from quenched decompression experiments. Especially the comparison of experimental samples with natural volcanic products can lead to misinterpretations when melt porosity is directly derived from glass porosity. Therefore, new methods for the analysis and interpretation of vesicle shrinkage and H2O resorption are developed, which are useful for future studies to correctly apply experimental results to natural volcanic processes.