skip to main content


Title: Melt Migration in Crystal Mushes by Viscous Fingering: Insights From High‐Temperature, High‐Pressure Experiments
Abstract

We conducted experiments to study melt migration in crystal‐rich mushes, with application to magma ascent within transcrustal magma reservoirs. Mushes with crystal volume fractions of 0.59–0.83 were prepared by hot‐pressing crushed borosilicate glass mixed with different proportions of quartz sand particles. Each experimental sample comprises stacked disks of mush and soda‐lime glass, a proxy for crystal‐free magma. Samples were subjected to confining pressures of 100–300 MPa and a temperature of 900°C (above the glass transition temperatures of the borosilicate and soda‐lime glasses) for up to 6 h. The bottom and circumference of the mush and soda lime disks experience the confining pressure, but the top of the mush disks is at room pressure, resulting in a pore‐pressure gradient across the mush layer. Following cooling and decompression, we determined the area fraction and morphology of soda‐lime melt that migrated into the mush layer during experiments. Melt fraction is more strongly correlated to crystal fraction than pore‐pressure gradient, increasing with crystal fraction before sharply decreasing as crystal fractions exceed 0.8. This change at 0.8 coincides with the transition from crystals in the mush moving during soda‐lime migration to crystals forming a continuous rigid network. In our experiments, melt migration occurred by viscous fingering, but near the mobile‐to‐rigid transition, melt migration is enhanced by additional capillary action. Our results indicate that magma migration may peak when rigid mushes “unlock” to become mobile. This transition may mark an increase in magma migration, a potential precursor to volcanic unrest and eruption.

 
more » « less
Award ID(s):
2052599
PAR ID:
10372042
Author(s) / Creator(s):
 ;  ;  ;  
Publisher / Repository:
DOI PREFIX: 10.1029
Date Published:
Journal Name:
Journal of Geophysical Research: Solid Earth
Volume:
127
Issue:
8
ISSN:
2169-9313
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. BSE mosaics of mushes and experimental products. Abstract: "We conducted experiments to study melt migration in crystal-rich mushes, with application to magma ascent within transcrustal magma reservoirs. Mushes with crystal volume fractions of 0.59 to 0.83 were prepared by hot-pressing crushed borosilicate glass mixed with different amounts quartz sand particles. Each experimental sample comprises stacked disks of mush and soda-lime glass, a proxy for crystal-free magma. Samples were subjected to confining pressures of 100 to 300 MPa and a temperature of 900°C (above the glass transition temperatures of the borosilicate and soda-lime glasses) for up to 6 h. The bottom and circumference of the mush and soda lime disks experience the confining pressure, but the top of the mush disks are at room pressure, resulting in a pore-pressure gradient across the mush layer. Following cooling and decompression, we determined the area fraction and morphology of soda-lime melt that migrated into the mush layer during experiments. Melt fraction is more strongly correlated to crystal fraction than pore-pressure gradient, increasing with crystal fraction before sharply decreasing as crystal fractions exceed 0.8. This change at 0.8 coincides with the transition from crystals in the mush moving during soda-lime migration to crystals forming a continuous rigid network. In our experiments, melt migration occurred by viscous fingering, but near the mobile-to-rigid transition, melt migration is enhanced by additional capillary action. Our results indicate that magma migration may peak when rigid mushes “unlock” to become mobile. This transition may mark an increase in magma migration, a potential precursor to volcanic unrest and eruption." Imaging: "Transverse sections cut from the top and/or bottom of the vacuum hot-pressed mushes were polished, carbon-coated, and imaged in BSE mode using the JEOL JXA-8530FPlus Electron Probe Microanalyzer (EPMA) at UMN (15 kV, 10 nA). About ten 50x magnification images were taken per sample and then compiled into BSE mosaics using Affinity Designer. The different compositions of the borosilicate glass and crystalline materials are distinguishable by greyscale in BSE images. [...] Following each experiment, sample assemblies were cut longitudinally along the cylindrical axis to produce sections for microstructural analysis. Scored samples (pHi-19s, Int-20s, Lo-21s) were cut again to produce sections tangential to the sample cylinder. Cut sections were vacuum impregnated with EpoFix resin and hand-polished on diamond lapping film from 30 to 0.5 μm grit. Polished and carbon-coated samples were imaged in BSE mode in the EPMA at UMN (15 kV, 10 nA). The different compositions of the soda-lime glass, borosilicate glass, and crystalline materials are distinguishable by greyscale in BSE images. Twenty to forty 50x magnification images were taken per sample and then compiled into sample-scale BSE mosaics using Affinity Designer." 
    more » « less
  2. Data accompanying manuscript of the same name. Here we present results of high-temperature, high-pressure experiments that test the conditions leading to melt migration in mushes. Our samples were made up of juxtaposed pierces of soda-lime glass and a densified mixture of borosilicate glass and quartz sand (X = 0.65 to 0.83). When these materials are subjected to high temperatures and confining pressures (900°C, 200 MPa) they are proxies natural silicate melts and mushes, respectively (Ryan et al., 2022). We deformed these samples in torsion and observed migration of melt into the mush as a result of shear. In samples with intermediate (X = 0.75) and high (X = 0.83) mush crystal fractions melt-filled dikes formed and propagated within the mush. To our knowledge these are the first instances of dike formation and propagation in high-temperature, high-pressure deformation experiments. The dikes formed as a result of shear-induced dilation, a process that was recognized in other granular media ∼150 years ago (Reynolds, 1885) but is rarely invoked as a potential deformation behavior for mushes (Petford et al., 2020). We use our experimental results to identify the conditions for shear-induced dilation and diking in mushes, apply this analysis to an active volcanic system (Soufrière Hills Volcano, Montserrat, W.I.) and, finally, consider the role of dike formation and propagation in mushes in the rapid generation and transport of crystal-poor magmas." Imaging: BSE mosaics of transverse sections of each experimental product were captured using a JEOL JXA-8530FPlus Electron Probe Microanalyzer (15 kV, 10 nA). Compositional differences between quartz, olivine, soda lime and borosilicate mean each phase is distinguishable based on its greyscale. Each sample was ground, polished and imaged four to ten times to produce serial sections. The area fraction of soda-lime glass that migrated into the mush (A) was quantified by thresholding and filtering BSE mosaics using ImageJ (Abramoff et al., 2004). Euclidean distance maps were thresholded to identify regions of soda-lime glass that have dimensions less than and greater than the estimated interparticle distance (40 μm; Supplement 2). Aintru is the area fraction of soda-lime glass in the mush with dimensions greater than the interparticle distance. The spatial distribution of soda-lime glass in the mush was quantified by overlaying rectangular grids on the BSE mosaics and measuring the area fractions greater and less than the interparticle distance (Supplement 2). 
    more » « less
  3. We present the results of high-temperature (900°C), high-pressure (200 MPa) deformation experiments that identify the processes and deformation conditions leading to melt migration in crystal-rich mushes. This study is relevant to transport of magmas in transcrustal magma reservoirs. Experimental samples comprise juxtaposed pieces of soda-lime glass and densified mixtures of borosilicate glass and quartz sand, which, at elevated temperatures and pressures, have melt and shear viscosities similar to natural silicate melts and crystal-rich mushes. The synthetic mushes have crystal fractions of 0.60 to 0.83. Samples were deformed in torsion at shear strain rates of 10-5 to 10-4 s-1 to shear strains up to 2.7. Image analysis of experimental samples shows melt migrates into the mush during shear. In mushes with crystal fractions ≥ 0.75 shearing causes melt-filled mm-scale dikes to form and propagate into the mush. To our knowledge, these features are the first dikes formed in high-temperature, high-pressure deformation experiments. Dike formation results from shear-induced dilation, which causes the mush to become underpressurized relative to the melt, at an estimated pressure differential of 10 MPa. Experimental conditions indicate shear-induced dilation and diking occur while the mush is still viscous (i.e., Weissenberg number < 10-2). We apply our results to Soufrière Hills Volcano (Montserrat, West Indies) and use our analysis to predict the deformation conditions that would lead to diking and rapid, voluminous melt migration in that active volcanic system. 
    more » « less
  4. SUMMARY

    The 2011–2012 eruption at Cordón Caulle in Chile produced crystal-poor rhyolitic magma with crystal-rich mafic enclaves whose interstitial glass is of identical composition to the host rhyolite. Eruptible rhyolites are thought to be genetically associated with crystal-rich magma mushes, and the enclaves within the Cordón Caulle rhyolite support the existence of a magma mush from which the erupted magma was derived. Moreover, towards the end of the 2011–2012 eruption, subsidence gave way to inflation that has on average been continuous through at least 2020. We hypothesize that magma segregation from a crystal mush could be the source of the observed inflation. Conceptually, magma withdrawal from a crystal-poor rhyolite reservoir caused its depressurization, which could have led to upward flow of interstitial melt within an underlying crystal mush, causing a new batch of magma to segregate and partially recharge the crystal-poor rhyolite body. Because the compressibility of the crystalline matrix of the mush is expected to be lower than that of the interstitial melt, which likely contains some fraction of volatile bubbles, this redistribution of melt would result in a net increase in volume of the system and in the observed inflation. We use numerical modelling of subsurface magma flow and storage to show under which conditions such a scenario is supported by geodetic and petrologic observations.

     
    more » « less
  5. Volcano inflation, for durations of months to years immediately following an eruption, has been observed at a number of volcanoes, including the 2011/12 eruption of Cordón Caulle, Chile. Such reinflation is often explained by replenishment of the magma reservoir from a deeper source. Whether and why that is the case remains uncertain in most instances, but the implications for renewed eruptive potential may be profound. Here, we posit redistribution of melt within a zoned magma reservoir consisting of a crystal rich mush overlain by an eruptible layer of crystal poor rhyolite as an alternate mechanism for reinflation. Such a zoned magma body is consistent with conceptual models for how crystal poor rhyolites form and with the presence of mafic enclaves within the Cordón Caulle rhyolite. The enclaves can be interpreted as pieces of mush entrained into the overlying rhyolite during its withdrawal from the reservoir. We test the hypothesis that melt from the inter-crystalline pores of the mush can redistribute by porous flow into the overlying crystal poor rhyolite, causing inflation after an eruption. We simulate the flow of melt within the zoned reservoir during and after eruption with a numerical model. As crystal poor rhyolite is erupted, magma pressure within the rhyolite layer above the mush decreases. Consequently, interstitial melt flows upward within the mush, toward the reduced pressure at the interface of mush and crystal poor rhyolite. The mush is treated as a poroelastic material, with interstitial melt flow governed by Darcy's law. Thus, the change in pressure caused by withdrawal from the overlying rhyolite diffuses downward into the mush as the interstitial melt flows upward. The change in pore pressure results in an elastic deformation of the mush matrix. Because pore pressure diffusivity is small, melt redistribution can persist for years after eruptive activity ends, leading to slow inflation compared to fast eruptive deflation. We predict a partial recovery of volume lost from the eruption. Reinflation occurs because the expansion of decompressing melt flowing from the mush into the crystal poor rhyolite exceeds compression of the poroelastic mush. For cases where the interstitial melt is moderately compressible due to exsolved volatiles, our model reproduces the deformation observed at Cordón Caulle. 
    more » « less