PyMelt: An extensible Python engine for mantle melting calculations

Matthews, Wong, Gleeson. Preprint available on EarthArxiv, doi: 10.31223/X5JP7X.

PyMelt is our new open-source python library for calculating the melting behaviour of lithologically heterogeneous mantle. Built into pyMelt are a number of published models for the melting behaviour of individual lithologies, including the Katz et al. (2003) lherzolite melting model, and the lherzolite and pyroxenite melting models that we developed in Matthews et al. (2021), and others. PyMelt implements the equations of Phipps Morgan (2001) to calculate the melting behaviour when these lithologies are in complete thermal equilibrium with one another.

There are also numerous methods built on top of this for calculating other melting region parameters, for example the trace element abundances in lavas, the crustal thickness produced at spreading ridges, or the magmatic productivity at intra-plate settings. To get started check out our interactive cloud-based tutorials on myBinder!

VESIcal: 2. A Critical Approach to Volatile Solubility Modeling Using an Open-Source Python3 Engine

Wieser, Iacovino, Matthews, Moore Allison, Earth & Space Science 9(2), e2021EA001932. doi: 10.1029/2021EA001932

One of the new capabilities offered by our VESIcal magma solubility modelling software is the ease with which we can compare the predictions of different solubility models and perform sensitivity tests on parameters about which we make assumptions. In this article we review the most widely used solubility models and examine the origins of the differences in their predictions. We also use VESIcal to demonstrate the effect of neglecting the contribution of dissolved CO2 when calculating saturation pressures in H2O-rich arc systems. VESIcal also makes it very easy to rapidly calculate many isobar-curves, and we use this functionality to critically assess the utility of plotting melt inclusion suites on top of a single set of isobars.

To read more about the VESIcal software, check out the first part of the VESIcal publications. VESIcal can be used in the cloud by signing up to the ENKI server.

Global trends in novel stable isotopes in basalts: Theory and observations

Soderman, Shorttle, Matthews, Williams, Geochimica et Cosmochimica Acta, 318, 388-414 (2021). doi: 10.1016/j.gca.2021.12.008

In this paper we assessed the utility of novel stable isotopes (Mg, Ca, Fe, V, and Cr) in lavas for tracing mantle lithological heterogeneity and melting processes, and in particular the prospects for combining multiple stable isotope proxies to uniquely identify these processes. Major element isotope systems may better respond to lithological heterogeneity because, unlike trace elements, their concentrations do not vary by orders of magnitude between different mantle components.

This work (led by PhD student C. Soderman) significantly expanded the capabilities of the stable-isotope fractionation code I originally developed (and published a proof-of-concept).

VESIcal Part I: An Open-Source Thermodynamic Model Engine for Mixed Volatile (H2O-CO2) Solubility in Silicate Melts

Iacovino, Matthews, Wieser, Moore, Bégué, Earth and Space Science 8 (11), e2020EA001584 (2021) doi: 10.1029/2020EA001584

An interactive version of this manuscript is available on myBinder:

We introduce our open-source solubility modelling engine, VESIcal, in this manuscript. Many of the most commonly used volatile solubility models are built into VESIcal and calculations can be performed using a common interface. This is the first time multiple solubility models have been so easily accessible at the same time, allowing their full potential to be exploited. Furthermore, calculations on large datasets are automated and can be performed in minutes or less.

I was one of the principle developers of the code and oversaw the construction of the generic model engine, and the implementation of several of the empirical models. The code is hosted in a GitHub repository (, and the documentation is available at

Geochemical Constraints on the Structure of the Earth’s Deep Mantle and the Origin of the LLSVPs

Gleeson, Soderman, Matthews, Cottaar, Gibson, G3, 22, e2021GC009932 (2021). doi: 10.1029/2021GC009932

The Galápagos islands preserve significant geochemical variability in their lavas, which varies consistently with spatial position. It is thought that much of this heterogeneity derives from melting of recycled and primordial mantle components. At the base of the mantle from which the Galápagos plume rises is a large low shear velocity province (LLSVP), however its origin remains enigmatic. LLSVPS have variously been interpreted as primordial mantle heterogeneities or piles of recycled oceanic crust.

If recycled oceanic crust is contributing to the magmatism, we should expect to see evidence for pyroxenite melting, This study demonstrated that the strongest pyroxenite signatures are found in a narrow band offset from the centre of the mantle plume. The lack of evidence for pyroxenite in the centre of the plume, the part most likely to be sampling the LLSVP, could indicate that the LLSVP material is not recycled oceanic crust, but this recycled material might be present on the margins of the LLSVP.

This study made use of the pyMelt mantle melting package ( and the THERMOCALC results we published in Soderman et al. (2021).

Conditions and dynamics of magma storage in the snæfellsnes volcanic zone, Western Iceland: insights from the Búðahraun and Berserkjahraun eruptions

Kahl, Bali, Guðfinnsson, Neave, Ubide, Van Der Meer, Matthews, Journal of Petrology 62(9) (2021). doi: 10.1093/petrology/egab054

The Snæfellsness volcanic zone in Iceland is an example of off-rift magmatism. Melt eruption rates are much lower than in Iceland’s neo-volcanic zones, and the eruptive products are more alkalic and show greater geochemical enrichment. It might, therefore, be expected that magma storage and transport processes work differently beneath Snæfellsness. In this manuscript the crystal cargoes of two eruptions Búðahraun and Berserkjahraun are analysed.

The chemically diverse crystals record complex petrogenetic histories most likely occurring during magma storage in both the lower- and mid-crust. The Berserkjahraun crystals I analysed for melt inclusions (published in Matthews et al., 2021) contributed to this study.

The photo shows the panoramic view across the mountains of Snæfellsness, with the Buðahraun lava flow in the foreground.

Iron isotopes trace primordial magma ocean cumulates melting in Earth’s upper mantle

Williams, Matthews, Rizo, Shorttle, Science Advances 7(11), eabc7394 (2021). doi: 10.1126/sciadv.abc7394

Evidence for an ancient magma ocean on Earth is preserved in the geochemistry of 3.7 billion year old metabasalts from Isua, Greenland. Previously work suggested these rocks are derived from melting a mantle source formed by Bridgmanite crystallisation and accumulation in the lower mantle. Bridgmanite crystallisation has previously been proposed to result in oxidation of the mantle, as it incorporates Fe3+ into its crystal structure, even when the magma it crystallises from contains only Fe2+. To balance the reaction Fe-metal is produced, which could be extracted efficiently to the core.

This reaction is thought to be associated with a fractionation in Fe-isotopes, such that the remaining bridgmanite enriched mantle should have an excess of 57Fe and 56Fe over 54Fe. In this study we demonstrated that such a fractionation is present within the Isua rocks, providing further confirmation for these magma ocean processes having taken place on the early Earth. However, the concentration of trace elements in the lavas suggest a more complex process, involving remelting and recrystallisation in the lower mantle.

To explore the consequences of these processes further, we performed THERMOCALC phase-equilibria modelling to determine how these magma-ocean derived mantle lithologies will melt in the upper mantle. We found that our hypothesised source for the Isua metabasalts melts at an anomalously low temperature. This would likely mean that such heterogeneities were rapidly melted out of the mantle, their evidence being largely erased.

The image is Figure 5 in the manuscript. It shows one such iteration of the phase-equilibria calculations. See the manuscript for further explanation.

Do olivine crystallization temperatures faithfully record mantle temperature variability?

Matthews, Wong, Shorttle, Edmonds, Maclennan, G3 (2021). doi: 10.1029/2020GC009157

Mantle temperatures are thought to vary substantially in the present day throughout the Earth, as a consequence of the vigorous convective cycling within our planet’s interior. Additionally, mantle temperatures are thought to vary through time. Estimating mantle temperatures in the ancient Earth can be more complex than for the present day Earth. In many cases the constraints we would like to use have to be indirectly inferred for the Earth’s past.

However, petrological techniques can be applied equally to both modern and ancient volcanic rocks. The temperature at which a magma starts crystallising is determined in large part by the temperature of the mantle whence it derived. Mantle composition also exerts a control (which we explored for Iceland in Matthews et al., 2016). In this new contribution we assess to what extent crystallisation temperatures can be relied on as a proxy for mantle temperature in the absence of information about mantle composition.

Central to our methods is an open source python module that we have developed for calculating mantle melting behaviour- pyMelt. To use the model without installing the python module, see our web-based app:

The image is from the Supplementary Information for the manuscript. It shows how we can use an electron probe to map the aluminium contents of olivine crystals. By measuring the aluminium content of coexisting olivine and spinel we can estimate the temperature they crystallised from magma.

Heavy 𝛿57Fe in ocean island basalts: A non-unique signature of processes and source lithologies in the mantle

Soderman, Matthews, Shorttle, Jackson, Ruttor, Nebel, Turner, Beier, Millet, Widom, Humayun, Williams, GCA 292, 309-332 (2021). doi: 10.1016/j.gca.2020.09.033

Despite being many tens of kilometres beneath our feet, Earth’s mantle plays an important role in the development of our planet. It acts as a vast chemical reservoir, exchanging with the surface through volcanism and subduction of tectonic plates. The mantle may also act as an archive of the chemical and tectonic changes that have occurred during our planet’s life.

An important tracer of past tectonics is the presence of ancient recycled crust, returned to the mantle by subduction. The presence of recycled crust has been inferred beneath volcanic islands such as Hawaii and Iceland, thought to have been transported in hot upwellings from the base of the mantle. However, most of the geochemical tools available to us only imply the presence of recycled crust indirectly. A more direct observation must relate to the mineralogical makeup of the mantle component, also known as its lithology.

Fe makes up a significant proportion of mantle rocks, and (in-part) determines their lithology. Subtle fractionations in its isotopes exist between different minerals due to variations in the way Fe atoms are bonded in their crystal structures. ‘Heavy’ Fe-isotope signatures have been linked to melting of recycled crust, but lithology is not a unique control on Fe-isotope fractionations. Here we combined new data with some novel models to assess the role of recycled crust in generating the Fe-isotope variability we observe in erupted lavas.

The image is Figure 6 from the manuscript. This shows the results of THERMOCALC phase-equilibra calculations for two mantle lithologies: KLB-1 (more ‘normal’ mantle) and G2 (‘recycled’ mantle material). Software we developed allowed us to calculate the degree of Fe-isotope fractionation that would be generated as these mantle components produce magma.

Reconstructing Magma Storage Depths for the 2018 Kīlauean Eruption from Melt Inclusion CO2 Contents: The Importance of Vapor Bubbles

Wieser, Lamadrid, Maclennan, Edmonds, Matthews, Iacovino, Jenner, Gansecki, Trusdell, Lee, Ilyinskaya, G3 22(2), e2020GC009364 (2020). doi: 10.1029/2020GC009364

The CO2 content of magmas is controlled primarily by pressure. Since pressure increases with depth in the crust, the CO2 contents of magmas can be useful tool for estimating how deep magma chambers sit beneath volcanoes. However, by the time magma reaches the surface it has lost most of its CO2. A widely used to tool to get around this is to measure the CO2 content in tiny droplets of magma trapped within crystals (melt inclusions) as they grew within the magma chamber.

However, crystals never behave as perfect pressure vessels. In many cases they have been shown to leak, removing the signal of magma chamber depth they once preserved. In other cases, they don’t leak, but the CO2 from the magma exsolves to form a bubble within the melt inclusion. This study analyses this behaviour in detail, and derives estimates of magma chamber depths beneath Kīlauea.

This study provided an ideal testing ground for the new VESIcal software I am involved in developing. In particular it allowed comparison of different CO2 solubility models and how they take into account the secondary control of magma composition on CO2 solubility.

The image is Figure 10 from the manuscript. It demonstrates the disparity in pressure estimates derived from different CO2 solubility models. Whilst none of the models are incorrect, they all work best in different situations. Something important to consider when using melt inclusions as barometers.