A Metamodel for Crustal Magmatism: Phase Equilibria of Giant Ignimbrites

Diverse explanations exist for the large-volume catastrophic eruptions that formed the Bishop Tuff of Long Valley in eastern California, the Bandelier Tuff in New Mexico, and the tuffs of Yellowstone, Montana, USA. These eruptions are among the largest on Earth within the last 2 Myr. A common factor in recently proposed petrogenetic scenarios for each system is multistage processing, in which a crystal mush forms by crystal fractionation and is then remobilized to liberate high-silica liquids. Magma evolves in the lower crust in earlier phases. We have tested these scenarios quantitatively by performing phase equilibria calculations (MELTS) and comparing the results with observed liquid (glass) and phenocryst compositions. Although comparison of tuff samples from each ignimbrite reveals distinct phenocryst compositions and proportions, the computed results exhibit a remarkable degree of congruity among the systems, pointing to some underlying uniform behavior relevant to large-volume silicic ignimbrites. Computed liquid compositions derived from more than ∼25% fractional crystallization of the parental melt in the deep crust are marked by SiO2 concentrations several weight per cent too low compared with the observed compositions, suggesting a limit on the extent of magma evolution by crystal fractionation in the deep crust. In all cases, the phase equilibria results and related considerations point to evolution dominated by crystal fractionation of a water-saturated mafic parental melt at shallow depths (∼5 km). Parental melt compositions are consistent with those of observed regional primitive basalts erupted prior to ignimbrite eruption for each system in each region. Fractional crystallization of water-rich mafic melt at shallow levels leads inherently to destabilization near thermodynamic pseudoinvariant points at around 800°C within the melting interval close to, but above, the solidus. For each system, the magmas evolve to states of high exsolved H2O volume fraction even at 5 km depth, eventually exceeding the criterion for magma fragmentation of ∼60 vol. % near the pseudoinvariant point temperature. Copious exsolution and possible expulsion of fluid occurs at this temperature, where the solid fraction in the magma changes almost discontinuously (isothermally) to significantly higher values. This instability mechanism acts as an eruption trigger by generating a gravitationally unstable arrangement of low-density, water-saturated magma beneath a thin (several kilometres) crustal lid. The trigger mechanism is common to fractional crystallization scenarios based on a variety of conditions, when crystallized solids and/or exsolved fluids are fractionated from residual melt isobarically (constant pressure) or isochorically (constant volume). In a single system, differences in liquid compositions resulting from constant volume versus constant pressure crystallization and expulsion versus retention of exsolved H2O are small compared with those arising from variations in initial water concentration, lithostatic pressure, and oxygen fugacity. It is these latter quantities that lie at the crux of the commonality in large-volume ignimbrite-forming eruptions, with a reasonable range of metamodel parameters. Scale analysis provides thermal timescales for fractional crystallization, including age ranges for discrete crystal populations. For the Bishop Tuff, the overall timescale for the Bishop magma body is >1 Myr. For the Yellowstone Tuffs, calculated thermal timescales are consistent with recurrence intervals of ∼600 kyr between successive caldera collapses. Although it is recognized that petrogenetic processes other than perfect fractional crystallization play a role in ignimbrite petrogenesis, by emphasizing common features the uniqueness of each system can be brought into better focus by sound and quantitative analysis.