Kenneth S. Befus

Experimental constraints on rhyolite-MELTS and the Late Bishop Tuff magma body

Thermodynamic models are vital tools to evaluate magma crystallization and storage conditions. Before their results can be used independently, however, they must be verified with controlled experimental data. Here, we use a set of hydrothermal experiments on the Late-erupted Bishop Tuff (LBT) magma to evaluate the rhyolite-MELTS thermodynamic model, a modified calibration of the original MELTS model optimized for crystallization of silicic magmas. Experimental results that are well captured by rhyolite-MELTS include a relatively narrow temperature range separating the crystallization of the first felsic mineral and the onset of the ternary minimum (quartz plus two feldspars), and extensive crystallization over a narrow temperature range once the ternary minimum is reached. The model overestimates temperatures by ~40xa0°C, a known limitation of rhyolite-MELTS. At pressures below 110xa0MPa, model and experiments differ in the first felsic phase, suggesting that caution should be exercised when applying the model to very low pressures. Our results indicate that for quartz, sanidine, plagioclase, magnetite, and ilmenite to crystallize in equilibrium from LBT magma, magma must have been stored at ≤740xa0°C, even when a substantial amount of CO2 occurs in the coexisting fluid. Such temperatures are in conflict with the hotter temperatures retrieved from magnetite–ilmenite compositions (~785xa0°C for the sample used in the experiments). Consistent with other recent studies, we suggest that the Fe–Ti oxide phases in the Late Bishop Tuff magma body are not in equilibrium with the other minerals and thus the retrieved temperature and oxygen fugacity do not reflect pre-eruptive storage conditions.

Compositional gradients surrounding spherulites in obsidian and their relationship to spherulite growth and lava cooling

Spherical masses of crystal fibers (spherulites) crystalize from rhyolitic melt/glass mainly in response to significant undercooling while lava cools. Spherulite growth should induce compositional gradients in the surrounding glass from expulsion of incompatible constituents and diffusion of those constituents away from the spherulite. Finite-difference numerical modeling of one-dimensional diffusion, in which diffusivities are allowed to vary with temperature, is used to investigate how compositional gradients reflect spherulite growth and lava cooling. Overall, three forms of gradients are identified. Elements that diffuse quickly are expelled from the spherulite but then migrate away too quickly to become enriched at the boundary of the spherulite. Elements that diffuse slowly are trapped within the growing spherulite. Between those endmembers are elements that are not trapped, yet diffuse slow enough that they become enriched at the contact. Their slow diffusion away then elevates their concentrations in the surrounding glass. How enriched those elements are at the spherulite-matrix interface and how far their enrichments extend outwards into the glass reflect how spherulites grow and thermal conditions during growth. Concentrations of H2O, Rb, F, Li, Cl, Na, K, Sr, Cs, Ba, and Be were measured in and around spherulites in obsidian from a 4.7u2009±u20091xa0km3 rhyolite lava dome erupted from Tequila volcano, Mexico. Measurable concentration gradients are found for H2O, Rb, and F. Attributes of those gradients and the behaviors of the other elements are in accord with their experimentally constrained diffusivities. Spherulites appear to have grown following radial, rather than volumetric, growth. The observed gradients (and lack of others) are more consistent with growth mainly below the glass transition, which would necessitate the dome cooling at ca. 10−5 to 10−7xa0°Cxa0s−1. Such slow cooling is consistent with the relatively large volume of the dome.

Stable chlorine isotope behavior during volcanic degassing of H2O and CO2 at Mono Craters, CA

Trends in CO2 and H2O concentrations and δD values of obsidian clasts from Mono Craters volcanic field, California demonstrate clear chemical and isotopic evidence for eruptive degassing. However, neither Cl concentrations nor stable isotopes (35Cl and 37Cl) track the degassing process, which is likely because of disequilibrium due to slow diffusion of Cl in the cooling melt. Obsidian pyroclasts (nu2009=u200929) were collected from tuff layers representing a single eruptive sequence that occurred circa 1340 A.D., as well as, rhyolitic obsidian samples (nu2009=u200912) were collected from three high-silica (>74xa0% SiO2) flows forming the domes and coulees in the region. The Cl, H2O, and CO2 concentrations recorded by the eruptive pyroclastic obsidians track the chemical evolution of the magmatic system during eruption, whereas the concentrations of the dome samples represent the final degassed product. The H2O and CO2 concentrations of the pyroclastic samples range from 0.49 to 2.13xa0wt% and 2 to 35xa0ppm, respectively; whereas concentrations in the dome glasses range from 0.17 to 0.33xa0wt% and 1 to 3xa0ppm, respectively. H2O and CO2 concentrations in the pyroclastic fall and dome samples are strongly correlated and reflect the degassing trend of the eruptive sequence. Chlorine concentrations of the pyroclastic fall samples and the domes range from 609 to 833xa0ppm and 681 to 872xa0ppm, respectively. Cl concentrations do not display a strong correlation with either H2O or CO2 concentrations. δD values of the pyroclastic fall obsidians vary between −84u2009‰ and −55u2009‰, whereas the δD values of the dome obsidians vary between −117u2009‰ and −91u2009‰. D/H ratios decrease with total water content following a distillation trend controlled by both closed and open system degassing. δ37Cl values of pyroclastic fall obsidians (−1.9u2009‰ to −0.1u2009‰) overlap with those of dome samples (−1.2u2009‰ to 0.0u2009‰). The similar Cl concentrations between the pyroclastic fall and dome obsidians argue for lack of Cl degassing, despite H2O and CO2 loss. These observations can be explained by disequilibrium effects due to the slow diffusion rate of Cl compared to H2O and CO2 in silicate melt, buffering by a separate brine phase, or by fluxing of Cl from a deeper magma source, with the slow diffusion rate of Cl being the preferred explanation. The wide range in δ37Cl values may be indicative of isotopic compositional heterogeneities in the magma source due to assimilation of sedimentary material or fluxing of mantle-derived Cl to a crustal melt.

Pre-eruptive storage conditions and eruption dynamics of a small rhyolite dome: Douglas Knob, Yellowstone volcanic field, USA

The properties and processes that control the size, duration, and style of eruption of rhyolite magma are poorly constrained because of a paucity of direct observations. Here, we investigate the small-volume, nonexplosive end-member. In particular, we determine the pre-eruptive storage conditions and eruption dynamics of Douglas Knob, a 0.011-km3 obsidian dome that erupted from a 500-m-long fissure in the Yellowstone volcanic system. To determine pre-eruptive storage conditions, we analyzed compositions of phenocrysts, matrix glass, and quartz-hosted glass inclusions by electron microprobe and Fourier-transform infrared analyses. The pre-eruptive melt is a high-silica rhyolite (∼75xa0wt.% SiO2) and was stored at 760u2009±u200930xa0°C and 50u2009±u200925xa0MPa prior to eruption, assuming vapor saturation at depth. To investigate emplacement dynamics and kinematics, we measured number densities and orientations of microlites at various locations across the lava dome. Microlites in samples closest to the inferred fissure vent are the most aligned. Alignment does not increase with distance traveled away from the vent, suggesting microlites record conduit processes. Strains of <5 accumulated in the conduit during ascent after microlite formation, imparted by a combination of pure and simple shear. Average microlite number density in samples varies from 104.9 to 105.7xa0mm−3. Using the magma ascent model of Toramaru et al. (J Volcanol Geotherm Res 175:156–157, 2008), microlite number densities imply decompression rates ranging from 0.03 to 0.11xa0MPaxa0h−1 (∼0.4–1.3xa0mmxa0s−1 ascent rates). Such slow ascent would allow time for passive degassing at depth in the conduit, thus resulting in an effusive eruption. Using calculated melt viscosity, we infer that the dike that fed the eruption was 4–8xa0m in width. Magma flux through this dike, assuming fissure dimensions at the surface represent its geometry at depth, implies an eruption duration of 17–210xa0days. That duration is also consistent with the shape of the dome if produced by gravitational spreading, as well as the ascent time of magma from its storage depth.

Magma storage and evolution of the most recent effusive and explosive eruptions from Yellowstone Caldera

Between 70 and 175xa0ka, over 350xa0km3 of high-silica rhyolite magma erupted both effusively and explosively from within the Yellowstone Caldera. Phenocrysts in all studied lavas and tuffs are remarkably homogenous at the crystal, eruption, and caldera-scale, and yield QUILF temperatures of 750xa0±xa025xa0°C. Phase equilibrium experiments replicate the observed phenocryst assemblage at those temperatures and suggest that the magmas were all stored in the upper crust. Quartz-hosted glass inclusions contain 1.0–2.5xa0% H2O and 50–600xa0ppm CO2, but some units are relatively rich in CO2 (300–600xa0ppm) and some are CO2-poor (50–200xa0ppm). The CO2-rich magmas were stored at 90–150xa0MPa and contained a fluid that was 60–75xa0mol% CO2. CO2-poor magmas were stored at 50–70xa0MPa, with a more H2O-rich fluid (

Analyzing water contents in unexposed glass inclusions in quartz crystals

X_{{{text{CO}}_{2} }}

Nonequilibrium degassing, regassing, and vapor fluxing in magmatic feeder systems

XCO2xa0=xa040–60xa0%). Storage pressures and volatiles do not correlate with eruption age, volume, or style. Trace-element contents in glass inclusions and host matrix glass preserve a systematic evolution produced by crystal fractionation, estimated to range from 36xa0±xa012 to 52xa0±xa012xa0wt%. Because the erupted products contain <10xa0wt% crystals, crystal-poor melts likely separated from evolving crystal-rich mushes prior to eruption. In the Tuffs of Bluff Point and Cold Mountain Creek, matrix glass is less evolved than most inclusions, which may indicate that more primitive rhyolite was injected into the reservoir just before those eruptions. The presence and dissolution of granophyre in one flow may record evidence for heating prior to eruption and also demonstrate that the Yellowstone magmatic system may undergo rapid changes. The variations in depth suggest the magmas were sourced from multiple chambers that follow similar evolutionary paths in the upper crust.

Crystallization kinetics of rhyolitic melts using oxygen isotope ratios

To assess the eruption and emplacement of volumetrically diverse rhyolite lavas, we measured microlite number densities and orientations from samples collected from nine lavas in Yellowstone Caldera and two from Mono Craters, USA. Microlite populations are composed of Fe-Ti oxidesu2009±u2009alkali feldsparu2009±u2009clinopyroxene. Number densities range from 108.11u2009±u20090.03 to 109.45u2009±u20090.15xa0cm−3 and do not correlate with distance from the vent across individual flows and are remarkably similar between large- and small-volume lavas. Together, those observations suggest that number densities are unmodified during emplacement and that ascent rates in the conduit are similar between small domes and large lava flows. Microtextures produced by continuous decompression experiments best replicate natural textures at decompression rates of 1–2xa0MPaxa0hr−1. Acicular microlites have a preferred orientation in all natural samples. Because the standard deviation of microlite orientation does not become better aligned with distance travelled, we conclude that microlites exit the conduit aligned and that strain during subaerial flow was insufficient to further align microlites. The orientations of microlite trend and plunge in near-vent samples indicate that pure shear was the dominant style of deformation in the conduit. We speculate that collapsing permeable foam(s) provides a mechanism to concurrently allow microlite formation and alignment in response to the combination of degassing and flattening by pure shear.