Bubble-crystal interactions during magma degassing

One of the critical processes that controls volcanic eruption style is exsolution of volatiles (e.g., H2O, CO2) from a melt, or magma degassing. The exsolved, pressurized volatiles provide much of the driving force for explosive volcanic eruptions. In addition to melt and bubbles, many magmas also contain crystals, and bubbles will interact with crystals during degassing. One of the important steps in the degassing process is the formation of an interconnected, permeable bubble network as that allows pressurized volatiles to escape from the system. I have primarily focused on the influence of crystals on this particular step of magma degassing, though certain crystal phases also promote nucleation of bubbles very early on in the degassing process.

I have conducted decompression experiments at the University of Alaska Fairbanks, in collaboration with Jess Larsen and Nate Graham, to examine these bubble-crystal interactions during magma degassing. My experiments focused on the influence of equant crystals (see panel b to the right), while experiments led by Nate Graham focused on elongate crystals (panel c to the right) and mixtures of elongate and equant crystals (panel d to the right). In all cases we find that the presence of crystals promotes bubble coalescence earlier in the degassing process and the formation of permeability, as compared to the crystal free case (panel a to the right). However, the shapes of the crystals influences the structure of the permeable network. Take a look at the stark differences in the bubble textures (bubbles are in black) in the far right column!

If interested in learning more, take a look at deGraffenried et al. (2019) and Graham et al. (2023)!
​Figure to the right is from Graham et al. (2023), using some data from deGraffenried et al. (2019).

---

​Using melt embayments to calculate magma decompression rates

Melt embayments are pockets of magma that are trapped in crystals but remain open to the host melt (see photo A on the left for an example). For reasons that are still unclear, bubbles do not always nucleate within the embayments. In these cases, during magma decompression, volatiles will diffuse towards a bubble that typically forms at the mouth of the embayment to equilibrate dissolved volatile concentration with the decreasing pressure. This leaves behind a concentration gradient of volatiles that can be measured once the melt is quenched upon eruption. Using diffusion modeling, researchers can calculate the time needed to form the concentration gradient and infer a decompression rate.

However, embayments come in many shapes, and some of those shapes are not ideal for the 1D diffusion models that are preferred in the community. I have conducted 3D diffusion models with embayment geometries that are increasingly constricted at the mouth and shown that 1D diffusion models retrieve increasingly faster decompression rates the more constricted the mouth is (see plot in figure B on left). In collaboration with Tom Shea, I am currently working on constructing correction factors for these more constricted embayments so that the community can continue to use 1D diffusion models

Plot on left from deGraffenried and Shea (2021).

Petrologic trackers of magma storage processes

Crystals are excellent recorders of the conditions/processes that they experience, and targeted studies of crystal compositions and textures can be used to untangle the history that a crystal has experienced. The processes that are recorded include changes to magma storage conditions (e.g., temperature, pressure, oxygen fugacity, composition of surrounding melt, etc.), changes to those storage conditions from system perturbations (e.g., injection of a new magma, evolution of a magma through equilibrium processes, etc.), and timescales between changes in the system. I have two ongoing projects in this area of research:

1) Applications of machine learning to magma plumbing system structure and magma migration timescales

Petrologic monitoring of eruptions often involves tracking the different batches of magma that feed an eruption to make a forecast about future eruption behavior. This is important both for short-term fluctuations in eruption behavior within a single eruptive sequence and long-term behavior of a volcano (see work from Dr. Maren Kahl). Machine learning techniques can produce large amounts of data from which to infer behavior. Applying machine learning techniques to acquisition of chemical data from thin sections using electron microscopy (see Leichter et al., 2022; Frontiers in Earth Science), we are able to map Fe-Mg concentrations in olivine in 2D in thin sections very rapidly (within ~2 days) to gain information about magma migration in the subsurface prior to eruption - this informs the magma bodies involved in the eruption. Diffusion modeling can be applied to these Fe-Mg concentration gradients to also gain information about the timescales between perturbation of a magmatic system and eruption.  We find that machine learning-derived data sets provide broad-scale constraints on eruption timing and magma migration that is similar to what is derived from electron microprobe techniques (see abstract deGraffenried et al., 2022, presented at Goldschmidt 2022).

2) How does eruption initiation mechanism influence volcanic eruption style at subduction zone volcanoes?

**This question is the subject of my current NSF EAR Postdoctoral Fellowship at Arizona State University.

Volcanic eruptions can be be initiated by a range of processes, from internal buildup of volatiles to instabilities developed by the injection of new magmas, and this is recorded by crystals in the host magma. The mechanism by which an eruption is initiated has a direct impact on the eruption style (see Kent et al., 2020; Earth ArXiv; https://eartharxiv.org/repository/view/1772/), and I hypothesize that this is relevant to the ascent rate produced by the different eruption initiation mechanisms. I will be investigating  this relationship for both Mt. St. Helens (Loowit/Lawetlat’la/Louwala-Clough) and Mt. Hood (Wy’east). I will also be using multiple decompression rate meters to investigate how magma decompression rate evolves through time. Stay tuned to this page to see how this project evolves!

Rheology of lavas and magmas

The material properties of lavas and magmas (e.g., viscosity, yield strength) influences how lavas/magmas behave. This underpins much of my research described above, as well as other projects that I am/was involved in, including lava flow propagation (deGraffenried et al., 2021), magma ocean behavior (Chao et al., 2021), and magma-water interactions (deGraffenried et al., 2020; Goldschmidt 2020 abstract, manuscript in progress).

Figure on the left is from deGraffenried et al. (2021) and shows the evolution of the lava flow from the Early Fissure 8 flow of the 2018 Kilauea eruption.

---

Developing analog models for geoscience education

Many processes in the geosciences occur at either very large scales or very small scales, and it can be difficult for newcomers to those fields to visualize the processes in play. One method to overcome this challenge is analog models. These models scale relevant geological processes to something that is much easier to visualize or interact with, and thus work well for educational purposes at all educational levels. One phenomenon that I frequently leverage in my own research is diffusion of elements or molecules to extract timescales of igneous process. However, fine scale (micron to nanometer scale) processes such as diffusion can be difficult to visualize or understand, much less the computations associated with the extraction of timescales. I am currently working with a team of researchers from both the University of Hawaii at Manoa and Ruhr-Universitaet Bochum to design simple experiments using just gelatin and food coloring (see photo below) to visualize the diffusion of food coloring in gelatin. This exercise will teach not only diffusion but also diffusion modeling to extract timescales of diffusion, and will be tailorable to students of different educational levels.

If you're interested to hear more, come see my poster at the IAVCEI 2023 General Assembly Meeting (explicit time TBD right now)!