Reconciling bubble nucleation in explosive eruptions with geospeedometers

Abstract

Magma from Plinian volcanic eruptions contains an extraordinarily large numbers of bubbles. Nucleation of those bubbles occurs because pressure decreases as magma rises to the surface. As a consequence, dissolved magmatic volatiles, such as water, become supersaturated and cause bubbles to nucleate. At the same time, diffusion of volatiles into existing bubbles reduces supersaturation, resulting in a dynamical feedback between rates of nucleation due to magma decompression and volatile diffusion. Because nucleation rate increases with supersaturation, bubble number density (BND) provides a proxy record of decompression rate, and hence the intensity of eruption dynamics. Using numerical modeling of bubble nucleation, we reconcile a long-standing discrepancy in decompression rate estimated from BND and independent geospeedometers. We demonstrate that BND provides a record of the time-averaged decompression rate that is consistent with independent geospeedometers, if bubble nucleation is heterogeneous and facilitated by magnetite crystals.

Fig. 1. Plinian silicic eruptions, their observed bubble number densities, and their inferred decompression rates.

a Spatial distribution of Plinian (Volcanic Explosivity Index ≥ 4) silicic eruptions over the past 100 kyr, based on Crosweller et al.. Red symbols are eruptions for which bubble number density and H₂O saturation pressure are documented. They are: 1875 Askja^,; 2008 Chaiten^,; 7.7 ka Mount Mazama^,; 1980 Mount St. Helens (MSH)^–; 1912 Novarupta^,; 1991 Pinatubo^,; 1.8 ka Taupo eruptions. b Bubble number density versus the maximum potential H₂O supersaturation pressure for eruptions (red symbols) and for homogeneous nucleation experiments (blue symbols)^,,,,–. Only experiments with supersaturation pressure of ≥150 MPa overlap with eruptions. c Decompression rate values estimated from observed bubble number density and homogeneous nucleation. There is a large gap between these estimates and those calculated by independent geospeedometers for the same eruptions^,–.

Fig. 2. Heterogeneous nucleation on magnetite reconciles H₂O saturation pressure with bubble number densities.

Heterogeneous nucleation factor, α, required for each eruption to match bubble number densities. Pinatubo can be reconciled with homogeneous nucleation (α = 1), whereas all other eruptions require heterogeneous nucleation (α < 1). α shown are for magnetite^,, hematite^,, pyroxene, and feldspar. Magnetite is the only mineral phase that allows heterogeneous nucleation to simultaneously match observed bubble number densities in all eruptions.

Fig. 3. Estimated decompression rates for silicic Plinian eruptions.

The blue symbols show the time-averaged estimates from bubble number density with heterogeneous nucleation on magnetite. The error bars represent uncertainties in magnetite contact angle. Heterogeneous nucleation substantially reduces the gap in decompression rate between homogeneous nucleation and independent geospeedometers that include: diffusion in melt inclusions and melt embayments for May 18th, 1980 Mt. St. Helens, 0.77 Ma Bishop tuff^,, 27 ka Oruauni^–, and 2 Ma Yellowstone; conduit models for May 18th, 1980 Mt. St. Helens^,; crystal rims for May 18th, 1980 Mt. St. Helens and 2008 Chaiten; and groundmass crystallization for May 18th, 1980 Mt. St. Helens.

Fig. 4. Illustrative model results of the feedback between water exsolution, decompression rate, and magma fragmentation during heterogeneous nucleation.

The contact angle is θ = 160° (α = 0.003). Nucleation first occurs at low supersaturation. Subsequently H₂O concentration remains close to equilibrium because of H₂O diffusion into nucleated bubbles. This results in a progressive increase in viscosity and hence decompression rate. Supersaturation pressure increases gradually, leading to a second nucleation peak, followed by magma fragmentation.