Research

The 2018 lower East Rift Zone (LERZ) eruption, and the accompanying collapse of the summit caldera, marked the most destructive episode of activity at Kīlauea Volcano in the last 200 years. I was fortunate enough to take part in an aerosol monitoring campaign during July-August 2018, allowing me to observe this spectacular phase of Kīlauea's history.

The eruption was extremely well-monitored, with continuous geodetic data capturing the caldera collapse and an extensive real-time lava sampling campaign. This multi-parameter dataset provides an exceptional opportunity to determine the reservoir geometry and magma transport paths supplying Kīlauea's LERZ. I use combined SIMS and Raman measurements to evaluate the total CO₂ of olivine-hosted melt inclusions from the dominant eruptive fissure (F8). This work shows that the highly primitive olivine crystals erupted at F8 have experienced large amounts of post-entrapment crystallization, driving 90% of the CO₂ into the vapour bubble. By quantifying the amount of CO₂ in the vapour bubble using Raman Spectroscopy, and critical evaluation of different volatile solubility models, I accurately reconstruct melt inclusion entrapment depths. This reveals that olivine crystals were supplied from two main reservoirs beneath Kīlauea's summit (at ~1-2km, and 3-5 km depth).

In addition to the more primitive samples erupted at Fissure 8 (~6-8 wt% MgO), variably evolved basaltic to dacitic lavas were erupted from a number of different fissures in the first month of the 2018 eruption. This project examines ~250 melt inclusions and ~150 matrix glasses from a suite of these variably evolved samples to investigate magma storage depths within the Lower East Rift Zone, and the evolution of sulfide-loving (chalcophile) elements during extensive differentiation of an ocean island basalt. This project will provide insights into:

1. the thermal and magmatic structure of Kīlauea's rift zone

2. The mechanisms producing the distinct chalcophile element signature of the continental crust

3. The potential for mineral chemistry to identify the onset of sulfide saturation in sample suites where melt inclusions are not available.

Samples were collected in collaboration with Hawai'i Volcano Observatory and the residents of Leilani Estates during 2 field campaigns in 2018 and 2019. SIMS analyses were conducted at the Edinburgh Ion Microprobe Facility, LA-ICP-MS analyses at the Open University, and EPMA and Raman analyses at the University of Cambridge.

This project examined a suite of melt inclusions and matrix glasses erupted between 1969-1974 at Kīlauea Volcano, Hawai'i, to understand how chalcophile elements (e.g., Cu, Ni, As, Se, Bi etc.) behave from the point of melt generation in the mantle source to the eruption of these melts at volcanic vents.

Simplified mantle melting models were run to compare the behaviour of highly incompatible elements such as Ba with chalcophile elements such as Cu and S during melting in the presence of mantle sulfide phases. Measured Cu and S concentrations in erupted lavas, and the co-variation of Cu and Ba, indicate that sulfide phases were present throughout the melting interval.

Models of sulfide solubility during fractional crystallization within shallow crustal storage reservoirs within the volcanic edifice show that melts become sulfide saturated at high MgO contents (~12 wt % MgO). As melts ascend to the site of an eruption, they begin to degas sulfur, driving the resorption of sulfides in contact with the degassing matrix glass. This obscures the textural and chemical record of sulfide saturation, explaining why previous studies suggested that sulfide saturation only occurred after extensive differentiation (~2 wt% MgO). Only careful examination of S and Cu contents of melt inclusions, and sulfides trapped within erupted crystals, reveal that magmas are sulfide saturated at depth prior to the onset of syn-eruptive degassing.

By utilising novel methodologies involving laser line scans and ICP-MS reaction cells, this work also provides the first petrological evidence that Se and S degas concurrently at low pressures.

Over the last decade, there has been a paradigm shift from the century-old model of magmatic systems consisting of large bodies of melt within the crust to a "mush-dominated" model, where plumbing systems predominantly consist of extensive piles of settled crystals with low enough melt fractions to form rigid frameworks. Crystals may be stored in mush piles for hundreds to thousands of years, before their eruption in unrelated carrier melts. The mush paradigm challenges the common assumption that careful petrological examinations of erupted crystals can be used to investigate specific eruption episodes.

This project examined crystal-melt relationships by assessing the degree of chemical equilibrium between melt inclusions, olivine crystals, and co-erupted carrier liquids at Kilauea Volcano. Despite significant chemical differences between erupted melts, melt inclusion records from 15 different eruptive episodes have indistinguishable Nb/Y ratios. Additionally, many of these inclusions have Nb/Y ratios that are significantly lower than the ratios in the melts that carried them to the surface. Host olivine crystals and carrier melts also exhibit prominent olivine-melt disequilibrium.

These observations suggest that olivines crystallise, trap melt inclusions, and settle into large mush piles at the base of the main storage reservoir at Kilauea. These crystals are then stored for long periods of time (centuries-millenia) within these mush piles, before being disturbed by the injection of chemically-unrelated melts, which carry these disturbed crystals to the surface. This mechanism explains the extremely high degree of major and trace element disequilibrium between erupted crystals and carrier melts. The random sampling of a portion of this trace element-diverse mush pile can account for the observation that many different eruptive episodes have melt inclusions with a similar range of Nb/Y ratios.

This study demonstrates that the fidelity of melt inclusion records must be carefully examined in mush-rich volcanic systems. At Kilauea, these inclusions record the evolution of the plumbing system as a whole, rather than eruption-specific processes (e.g., controls on eruption style, eruption triggering).

Olivine crystals with prominent distortions of the crystal lattice are near-ubiquitous in a number of systems worlwide (e.g., Picrites from West Greenland; Piton de la Fournaise, Reunion; Mull, UK; Cerro Azul, Galapagos; Kilauea Volcano Hawaii). This study examines these features at Kilauea Volcano using Electron Backscatter Diffraction, an SEM-based technique that allows the orientation of the crystal structure to be mapped.

Deformation features at Kilauea have been attributed to physical processes such as the plastic deformation of dunitic bodies located within Kīlauea's rift zones, collisions of crystals during transport, or crystal growth processes. This project quantifies the microstructures of deformed olivines from EBSD maps using techniques which are traditionally used to examine deformed olivine crystals within exhumed mantle peridotites. This inter-disciplinary approach reveals thatt Kīlauean olivines experienced differential stresses of ~3-12 MPa.

The absence of adcumulate textures, high angle grain boundaries, and the occurance of deformed grains at a large number of different eruption sites around Kīlauea's edifice, are inconsistent with the hypothesis that olivines are deformed within creeping, low melt fraction dunitic bodies in Kīlauea's deep rift zones. This model also cannot account for the occurrence of these features at a number of volcanic systems worldwide which do not possess prominent rift zones. Additionally, the major element chemistry of deformed and undeformed olivine crystals are statistically distinguishable, which is inconsistent with a model where olivines are deformed during crystal-crystal collisions during magma transport processes. Instead, I attribute deformation to gravitational loading within melt-rich cumulate piles with vertical extents of ~180-720 m located at the base of the deeper magma storage reservoir at Kilauea (~3-5 km).

Mass balance models of Kīlauea's plumbing system, accounting for magma input rates, the volume of olivine fractionated from each volume of melt, and the radii of the SC reservoir inferred from geophysical observations, demonstrates that these thickness's can accumulate within a few centuries

Like the deformation structures discussed above, clusters of aggregated crystals are a very common textural feature in basaltic lavas worldwide. Previously it has been suggested that aggregates form during growth processes such as heterogeneous nucleation (where a new crystal nucleates on the surface of an existing crystal), dendritic growth (which occurs when crystals grow extremely fast, forming finger-like dendrites - this comes from the greek for ‘tree’), or twinning (where two parts of a crystal grow together, with a specific orientation relative to one another). Alternatively, crystals may “swim” together as they settle through a liquid magma column (through a process known as synneusis), or land on top of one another after they settle to the solid base of the magma chamber. This study examines olivine aggregates from Kilauea Volcano and Chromite aggregates from the Bushveld Complex, using electron backscatter diffraction (EBSD) and surface energy calculations to differentiate between these hypotheses.

Neither olivine nor chromite aggregates show attachment geometries consistent with crystal growth processes such as twinning, epitaxial growth or heterogeneous nucleation. Additionally, I show that heterogeneous nucleation and epitaxial growth are associated with significant energy penalties compared with the continued growth of the pre-existing crystal (although these processes may well be dominant in multi-phase aggregates). Instead, the near-random attachment geometries of chromites indicate that aggregates formed as crystals landed on top of one another after settling to the base of the magma chamber. In contrast, olivine aggregates show a narrower range of geometries, with misorientation axes clustered at [100], and misorientation angles showing three dominant peaks. These attachment geometries, along with different core compositions of attached crystals, indicate that crystals aggregated during late-stage crystal settling or magma transport just prior to their eruption.

For more detail - see a popular science blog I wrote about this work! https://blogs.egu.eu/divisions/gmpv/2019/09/05/949/

Lavas erupted in continental arcs overlying thick crust show elevated incompatible element concentrations and enriched radiogenic isotope ratios relative to lavas erupted in thinner-crusted arcs, or ocean island arcs. However, it is uncertain whether this enrichment derives from the interaction of ascending magmas with the continental crust, or whether it reflects processes within the mantle, such as variable melt extents, mantle heterogeneity, or variable supply of materials from the subducted slab.

This project assesses the relative influence of these different processes using the Andean Southern Volcanic Zone (SVZ) as a case study. From South to North, crustal thickness increases from ~30 to 50 km. the depth to the slab increases from 70 to 120 km, and the amount of trench sediment declines. New analyses of trace elements and Sr-Nd isotopes were conducted on unusually primitive samples from Don Casimiro, a volcano at the northern terminus of the SVZ. These analyses were compared to previous literature data along the arc, with a specific focus on Villarrica volcano in the southern end of this arc segment (which is significantly depleted relative to Don Casimiro).

Mantle melting models with variable slab inputs and melt extents cannot recreate the extreme trace-element and Sr-Nd isotopic enrichment at Don Casimiro relative to Villarrica. Similarly, no known regional or global basement lithologies are sufficiently enriched in all the necessary incompatible elements and possesss suitable Nd and Sr isotope systematic to account for the northern enrichment by crustal assimilation. Crucially, we show that lavas from the rear-arc (which are highly primitive, and have experienced limited slab inputs) show similar geographic trends in geochemical signatures to arc-front lavas. Thus, we suggest that Northern arc-front and rear-arc lavas are sampling an enriched mantle component similar to the global EM1 component, possibly generated through recycling of metasomatized lithospheric mantle.

Until now, quantitative calculations of volatile solubility, saturation pressures, equilibrium fluid compositions, isobars and isopleths, and degassing paths has been a time-consuming endeavour. In many cases, no accompanying tool was provided alongside the publication of a new solubility model, requiring users to correctly combine and interpret the relevant equations. This is problematic from a perspective of reproducibility of the multitude of studies utilizing these models. Even when excel spreadsheets or web apps are provided, they require users to manually enter large amounts of data for each sample (e.g., the concentrations of 8-10 major oxides, volatile concentrations and temperature in Magmasat).

VESIcal is a Python3 code including 7 of the most popular solubility models. This project is lead by Kayla Iacovino (NASA). The provision of this code with easy-to-use Jupyter notebooks and a web portal allows uses with a wide range of coding experiences to be able to perform a wide range of calculations on large datasets provided as an excel spreadsheet, without the need for manual data entry. For the first time, this has allowed extensive comparisons to be drawn between the different models, and robust evaluation of their suitability in different settings.

In July-Aug, 2018, I took part in a field campaign to sample the Kilauean Plume (led by Evgenia Ilyinskaya from Leeds, in collaboration with Hawaii Volcano Observatory and the USGS). Our team of 5 female scientists from Leeds and Cambridge (#PlumeTeam) characterised the composition of the volcanic plume emitted at the Fissure 8 vent using a variety of aerosol sampling techniques (SIOTUS Impactors, filter packs), as well as MultiGas and PiCam measurements of volcanic gas fluxes.

In addition to characterising the near-source composition of the volcanic plume, we set up air quality monitoring stations at various sites around the island of Hawai'i, to track the chemical evolution of the plume during atmospheric dispersion.