V14A-01 INVITED
Solubility of H2O-CO2 in Basalts and the Effect of Melt Composition: Assessing Current Solubility Models with Experimental Data
Our ability to use measured volatile contents (H2O and CO2) in basaltic melt inclusions to estimate useful parameters such as saturation pressures, fluid compositions, and degassing paths is dependent on understanding the volatile solubility relations between fluid and melt as a function of temperature, pressure, and importantly, melt composition. Recent mixed-volatile solubility models such as Volatile Calc (Newman and Lowenstern, 2002) and Papale et al (2006) can be used to estimate these inter-dependencies, but both are limited by the extent and quality of the experimental data used to calibrate them. Therefore, the correct application of the solubility models to melt inclusion data requires an understanding of the behavior of these models (e.g. compositional dependence), their reliability over varying conditions (e.g. pressure, fluid composition) and especially the potential accuracy of their parameter estimates. For example, experimental results involving the solubility of H2O and CO2 in calcic and calc-alkaline basalts indicate that the effect of variable melt composition is significant. This result, coupled with the observation of the large compositional variation in natural melt inclusions, raises concern about the application of the models to compositions outside of the range of the experimental data. To test this, we assess the ability of the Volatile Calc and Papale solubility models to reproduce experimental H2O-CO2 solubility data on various basalt compositions, both recent and from the literature. The results highlight not only the significant differences between the two models, but also identify important compositional variables and pressures that are critical to melt inclusion studies, but that are poorly constrained by available experimental data.
V14A-02
Basaltic volcanoes as large-scale aerators: the example of Mt. Etna
The recognition and simulation of the patterns of gas release from active volcanoes in relation to those of magma supply and transfer are a major geochemical goal. At basaltic volcanoes such as Mt. Etna (Sicily, Italy) this knowledge would greatly assist our comprehension of the mechanisms of magma rise and injection at different storage levels, from depth up to the shallow systems feeding lava fountains and flows. In this contribution we investigate the H2O-CO2-SO2-H2S-silicate melt system by integrating theoretical models on volcanic degassing with data from plume chemistry, fumarole sampling, chemistry and volatile contents of melt inclusions (MIs). Given an initial bulk composition, we show that the degassing processes behind this ensemble of data can be quantitatively assessed by carefully evaluating the interplay of 1) crystallization, hence phase proportions, 2) redox variables, 3) gas addition (flushing) occurring at different steps along the magmatic column. Because of pervasive CO2-flushing through the magma, we picture Mt. Etna as a big aerator, by analogy with gas absorption techniques in chemical process engineering. CO2-flushing is particularly efficient at P > 140 MPa, where the volatile influx generates a family of degassing paths that embrace the range of variability displayed by H2O, CO2 and S contents dissolved in MIs. The flushing mechanism can work under two extreme scenarios: in one case the rising volatile phase is completely blocked by the shallower magmas, whereas in the other one at each addition of deep gas the pre-existing gas phase is completely separated from the flushed magma. This study shows that equilibrium thermodynamics provides reasonable physico-chemical constraints to interpret the ensemble of data observed, without invoking diffusive regimes acting far from the equilibrium. This is a strong argument for the joint adoption of MI-based volatile contents and volatile-melt saturation algorithms. The results of this study are expected to shed light on the close association between basaltic volcanism and the release of large amounts of carbon dioxide.
V14A-03 INVITED
Experimental Constraints on Non-Equilibrium Degassing of Bubbly Magmas
The extreme ranges in sizes and numbers of vesicles in volcanic pumice has been cited as evidence that magma does not degas in equilibrium as it ascends during eruption. Such non-equilibrium allows volatiles to supersaturate in the melt, which triggers gas bubbles to nucleate. If no bubbles are present in the magma when it starts to rise, then non-equilibrium degassing is likely. But, copious evidence points to most magmas being gas saturated before they erupt, and thus were already bubbly liquids at depth. Those bubbles play a critical role in eruption dynamics, because they can allow magma to degas, which causes it to accelerate as it ascends towards the surface by increasing the viscosity of the melt. There must be a limit to that acceleration, however, because gas bubbles cannot grow infinitely fast. Hence, a key to understanding the conditions needed for non-equilibrium degassing during ascent is knowing how fast must magma rise to exceed that limit. To explore that limit, we are carrying out series of hydrothermal decompression experiments to determine the maximum rate at which bubbly silicate melts can degas in equilibrium. Preliminary results using bubbly rhyolitic melt show that such melts containing 7 to 18 vol.% bubbles can degas in equilibrium at 875° C when decompressed at rates up to 1.2 MPa s-1 from 150 to 78 MPa, and up to 1.8 MPa s-1 when decompressed further to 42 MPa. In contrast, that same rhyolite cannot degas in equilibrium at 750° C if decompressed faster than 0.015-0.025 MPa s-1. When compared to predictions from conduit flow models that assume equilibrium degassing, it is found that such models over-estimate the rate at which relatively cold rhyolite can decompress, and so such magmas may be able to ascend fast enough to supersaturate and nucleate new bubbles. In contrast, relatively hot bubbly rhyolite can decompress almost as fast as flow models allow, suggesting that if bubbles are present at depth then non-equilibrium degassing may only occur near the level of fragmentation for such melts. Importantly, most other magmas are less viscous than even hot rhyolite, and thus those may degas essentially in equilibrium under most conditions.
V14A-04 INVITED
Water diffusion, Viscosity and Bubble Growth in Silicate Melts
For quantitative modeling of bubble growth and volcanic eruption dynamics, it is necessary to know H2O diffusivity in the melt. Over the years, we have been experimentally and systematically investigating H2O diffusion in rhyolite, dacite, andeside, basalt, and a per-alkaline rhyolite (1-7; as well as work in progress). We have also investigated viscosity of hydrous melts and developed a viscosity model for all natural silicate melts (8-10). In this report, we discuss the compositional dependence of H2O diffusivity and the relation between H2O diffusivity and viscosity. Furthermore, we explore how these parameters affect bubble growth rate in various melts. Experimental data show that in contrast to the large differences in viscosity of various melts, the variation of H2O diffusivity with melt composition is in general small, especially at super-liquidus temperatures. For example, when per-alkaline rhyolite is compared with calc-alkaline rhyolite, the viscosity difference is large but the diffusivity difference is small. Comparison between rhyolite and dacite is more complicated. At 1423 K (super-liquidus) and 1.0 wt percent total H2O, the viscosity decreases by a factor of 80 from rhyolite to dacite, but the diffusivity increases by less than a factor of 2. However, at 873 K (sub- liquidus) and 1.0 wt percent total H2O, the difference in the calculated viscosities of rhyolite and dacite is negligible, but the diffusivity decreases by a factor of 6 from rhyolite to dacite. Hence, there does not seem to be a consistent relation between viscosity and H2O diffusivity. When modeling bubble growth rate in different melts, the effect of viscosity variation can change bubble growth rate significantly, but the effect due to variation in diffusivity is small at super-liquidus temperatures. References: (1) Behrens et al. (2004) Geochim. Cosmochim. Acta, 68, 5139-5150. (2) Behrens et al. (2007) Earth Planet. Sci. Lett., 254, 69-76. (3) Liu et al. (2004) Chem. Geol., 209, 327-340. (4) Ni and Zhang (2008) Chem. Geol., 250, 68-78. (5) Zhang et al. (1991) Geochim. Cosmochim. Acta, 55, 441-456. (6) Zhang and Stolper (1991) Nature, 351, 306-309. (7) Zhang and Behrens (2000) Chem. Geol., 169, 243-262. (8) Zhang et al. (2003) Am. Mineral., 88, 1741-1752. (9) Zhang and Xu (2007) Geochim. Cosmochim. Acta, 71, 5226-5232. (10) Hui and Zhang (2007) Geochim. Cosmochim. Acta, 71, 403-416.
V14A-05
Growth and Resorption of Bubbles by Chemical Exchange between Disequilibrium Fluids
Volatile components in magmas show compositional diversity due to pressure dependence of solubility and variety of distribution along depth: the shallower and silicic magmas are generally rich in H2O, while the deeper and mafic ones are relatively CO2-rich. Previous studies on melt inclusions and pyroclastic obsidians suggested that CO2-rich fluids, probably originated from deeper basaltic magmas, could be transported to the shallower H2O-rich magma chambers or volcanic conduits. In such a process, CO2-rich fluid would interact with H2O-rich melt to attain re-equilibrium. Exchange flux of the volatile components depends on their diffusivities in the melt, thus difference in the diffusivity may cause bubble growth or resorption. To investigate the kinetics of interaction between disequilibrium fluids, we have performed volatile exchange experiments between H2O-rich rhyolitic melt and CO2-rich fluid in both bubble-present and -absent cases using a cold-seal pressure vessel. In the bubble-absent experiments, we pre-synthesized H2O-saturated bubble-free glass at 100 MPa and 800 degC, and then placed the glass in the CO2-rich fluid at the same PT conditions. We found that H2O diffused out rapidly to the surrounding fluid while CO2 diffused slowly into the melt, resulting in temporal undersaturation with the COH-fluid. In bubble-present experiments, we first synthesized COH-fluid-saturated melt at 100 MPa and 800 degC, and then decompressed to 50 MPa to promote vesiculation. The bubbles formed in the melt consisted almost of H2O, while the excess fluid around the melt was enriched in CO2. The bubbles have once grown, then dissolved into the melt, because the melt temporarily became undersaturated due to the diffusivity difference between H2O and CO2. These results carry two implications for natural COH-fluid bearing systems: First, according to the solubility law of COH- fluid, large amount of H2O is extracted from the hydrous melt to the CO2-rich bubbles during this interaction, while only small amount of CO2 dissolves into the melt, leading to the increase in total vesicularity. Thus, this process might trigger the volcanic eruption upon injection of mafic magma. Second, flow of CO2-rich fluid in the permeable conduit margin near the wall would cause dissolution of H2O-rich bubbles and thus enhance the formation of dense, CO2-rich magma that often erupted as obsidian pyroclasts.
V14A-06
Gas Loss and Resorption beneath Mount St. Helens, 1980-1986
We report the first direct determinations of CO2 and H2O in melt inclusions from the 1980-86 eruption of Mount St. Helens. 12C and 1H were measured by ion-microprobe on 70 melt inclusions in plagioclase and orthopyroxene and converted to CO2 and H2O using working curves based on well-characterised rhyolite glasses. Minimum detection limits are 50 ppm CO2 and 120 ppm H2O. This dataset is supplemented by broad-beam FTIR analyses of the groundmass of selected samples, with a novel correction procedure to account for the presence of microlites. The highest combined H2O and CO2 contents (6.1 wt% and 350 ppm, respectively) were found in a melt inclusion from the May 18th 1980 plinian eruption. At 900 ° C these correspond to a saturation pressure of 270 MPa and a coexisting vapour composition of 81 mol% H2O, in reasonable agreement with experimental phase equilibria. Other plinian sample melt inclusions extend to lower H2O and CO2, broadly consistent with closed system degassing. Melt inclusions from subsequent eruptive phases do not, however, conform to simple degassing. Most strikingly, melt inclusions from the May 25th eruption extend to high CO2 at lower H2O than the plinian samples. These trends can only be ascribed to repressurisation of residual magma left behind after the plinian eruption. During repressurisation residual vapour bubbles become resorbed leading to a pronounced enrichment in CO2 relative to H2O. The mixing line for this repressurisation process extends to the composition of the fumarole gases. We propose that following depressurisation of the chamber during the plinian eruption the magma chamber degassed. Upon re-sealing of the conduit system pressure returned to lithostatic pre-eruptive values leading to resorption of residual vapour bubbles. The magnitude and timing of the inferred pressure change is consistent with numerical calculations. A similar story is revealed by all subsequent dome-forming eruptions, although the magnitude of the pressure changes is significantly smaller. Our data suggest that syn-eruptive pressure loss followed by repressurisation and bubble resorption drives a chemical pump in which initially exsolved gases passing through the system become resorbed, enriching the magma in volatile species. This process accounts for anomalous enrichment of Li in post-plinian melt inclusions and may also enrich the melt in volatile species such as Cu. This model provides a driving force for the formation of hydrothermal ores, suggesting that an embryonic ore body is forming beneath Mount St. Helens.
V14A-07
When Does Magma Become Permeable?
It is commonly assumed that vesiculating magma will become permeable when its porosity reaches ~ 30%, the percolation threshold for spheres. However, porosity-permeability measurements of volcanic rocks suggests that the critical porosity for permeability development varies widely, from close to 0% in mafic lava flows and domes to >95% in basaltic reticulite. These differences can be explained by consideration of magma vesiculation and degassing histories. Rapid bubble nucleation and growth within volcanic conduits during pumice- and scoria-producing eruptions generates vesicularities of 60-70% before the bubble-bubble connectivity is sufficient for permeable gas flow. When syn-eruptive expansion is constrained by conduit walls, elongation of bubbles to form tube pumice increases permeability along the elongation direction, thus permitting efficient degassing. When syn-eruptive expansion is unconstrained (as in the case of basaltic fire fountains), magma vesicularities may exceed 95% before bubble-bubble connectivity is achieved, as seen in basaltic reticulite. When syn-eruptive expansion is slow, as in the case of breadcrust bombs, the porosity threshold for permeability approaches the theoretical value of ~30% as bubble coalescence rates approach rates of bubble growth. Maximum values of permeability in these samples vary as a function of the tortuosity of permeable pathways, as determined from measurements of electrical conductivity. Deformation and collapse of bubbles during emplacement of lava flows and domes further modifies the vesicle structure from connected spheres to disk-like cracks, as indicated by low measured electrical tortuosities. The effect on permeability is a hysteresis such that the magma maintains a high permeability to much lower vesicularities than under conditions of expansion. In low crystallinity melts, this hysteresis decreases the apparent percolation threshold from >60% to ~ 30%; in high crystallinity melts it appears that the crystals may aid the development of a crack-like bubble network such that percolation thresholds drop to < 10%. We conclude that both the percolation threshold and the maximum permeability, and thus the efficiency of degassing, depend on the decompression and deformation conditions. For this reason, consideration of controls on permeability development is critical for accurately modeling magma ascent and eruption. Furthermore, our data show that both the textures and permeabilities of clasts provide important insights into eruption dynamics.
V14A-08
Eruption dynamics of Hawaiian-style lava fountains
Hawaiian-style lava fountains are episodic, not continuous. Existing models view a typical Hawaiian fountain as driven by a sustained, steady-state process. However, eyewitness observations show that, within a single fountain, the height pulsates implicating fluctuations in the fluxes of gas and magma The short-lived but powerful 1959 eruption at Kīlauea Iki contained 17 fountaining episodes and produced a cone and tephra blanket as well as a lava lake which interacted with the vent during all but the first episode of the eruption. We have conducted a detailed investigation of tephra from the episode 1 deposit in order to examine ascent and eruption conditions of Hawaiian lava fountaining. Density measurements (300 to 1600 kg/m3) and vesicularity calculations (42% - 88%) of closely spaced lapilli-sized samples from the episode 1 tephra indicate a wide textural diversity reflecting variability in vesiculation processes. Micro-textures in the least vesicular clasts have polymodal size distributions of bubbles with thick walls, whereas clasts with higher vesicularity offer bimodal distributions of bubbles with thin walls. Modal vesicularity clasts have bi- to poly-modal vesicle size distributions which lie between these two end members as well as exterior macro-textures of frequent occurrences of centimeter-sized bubbles. Early quenched rims of the highest vesicularity clasts have vesicle-number densities approaching 107 cm- 3 which offers a valid approximation to magma conditions immediately at fragmentation. However, modal vesicularity clasts with vesicle-number densities of 106 cm-3 strongly indicate that this number density, represented by the majority of magma driving the eruption, has dropped away from the maximum. Partial bubble loss occurred through the process of coalescence and outgassing during two-phase flow in the conduit. Combining the macro- and micro-textural observations we therefore propose a more complex story for Hawaiian lava fountaining from conduit ascent through to eruption than currently exists within the literature.