Melt inclusion constraints on petrogenesis of the 2014–2015 Holuhraun eruption, Iceland

The 2014–2015 Holuhraun eruption in Iceland was the largest volume eruption on the island since the 1783–1784 Laki erution (e.g., Neave et al., 2013; 2017), and was one of the most closely monitored eruptions ever to have taken place (Gudmundsson et al., 2016). In this paper, lead by Margaret Hartley and Enikö Bali (who also recently published an associated paper), we present melt inclusion data from a suite of samples collected throughout the eruption.

Variability in melt inclusion compositions indicates that the erupted magma evolved from diverse primary melts by concurrent mixing and crystallisation. Using a refined method of olivine–plagioclase–augite–melt (OPAM) barometry, we place this evolution at mid-crustal depths, in agreement with geophysical indicators of magma storage. Re-equilibration of melt inclusion H2O contents indicates that crystals spent at least 1–12 days in their carrier liquid before eruption, consistent with lateral transport in a mid-crustal dyke from the Bárðarbunga central volcano to the eruption site.

A conceptual model for the entrapment of melt inclusions erupted during the 2014–2015 Holuhraun eruption. Figure from Hartley et al. (2018).


Hartley, M.E., Bali, E., Maclennan, J., Neave, D.A. & Halldórsson, S.A. 2018. Melt inclusion constraints on petrogenesis of the 2014–2015 Holuhraun eruption, Iceland. Contributions to Mineralogy and Petrology, 173: 10. <Open Access>

Mantle-derived trace element variability in olivines and their melt inclusions

Olivine is almost ubiquitous in primitive basalts, making it an excellent tool for investigating early phases of magmatic evolution and mantle melting conditions. For example, studies of olivine-hosted melt inclusions have provided crucial insights into primitive melt variability, deep magma mixing (e.g., Neave et al., 2013) and the behaviour of volatiles during magma transport from the mantle to the surface (e.g., Neave et al., 2012; 2014). It has also been proposed that the compatible trace-element (CTE) content of olivines themselves provides information about lithological heterogeneity in the mantle (e.g., Sobolev et al., 2007), though the causes of such CTE variability remain highly debated (e.g., Matzen et al., 2017).

In this paper with Oliver Shorttle and Martin Oeser, I present both CTE and incompatible trace-element (ITE) data from primitive Icelandic olivines that we use check the validity of melt inclusion records and investigate causes of geochemical variability in olivine macrocrysts themselves. We demonstrate that olivine macrocrysts are capable of preserving similar patterns of compositional variability to melt inclusions on intra- and inter-eruption lengthscales, and may allow degrees of magma enrichment to be reconstructed in samples where matrix glasses are degraded or absent.

An X-ray map of P in an olivine from the Stapafell eruption. Almost no P zoning can be observed in the olivine; boundary layer crystallisation seems unimportant. This image is approximately 1 mm across.

Although olivines from our enriched case study eruption, Stapafell, are slightly richer in Ni than those from our depleted case study eruption, Háleyjabunga, the CTE content of both eruptions are wholly consistent with melt supply from a peridotitic source. However, independent constraints from the combined major and trace element systematics of Icelandic basalts indicate that enriched melts come from a modally enriched source (Shorttle & Maclennan, 2011); enriched Icelandic basalts are too rich in iron to be derived by melting of depleted mantle. We therefore conclude that enriched domains in the Icelandic mantle are composed of modally enriched peridotite not pyroxenite, and that olivine CTE contents provide an incomplete picture of lithological heterogeneity in the mantle.


Neave, D.A., Shorttle, O., Oeser, M., Weyer, S. & Katsura, K. 2018. Mantle-derived trace element variability in olivines and their melt inclusions. Earth and Planetary Science Letters 483, 90–104.

Continuous mush disaggregation during the long-lasting Laki fissure eruption, Iceland

Igneous rock textures encode important information about magma reservoir dynamics. Specifically, the size, shape and abundance of crystals can record multiple phases of crystallisation and magma mixing. However, characterising rock textures using traditional manual methods is extremely time consuming. However, the potential for quantifying textures with automated mineralogical methods, which have seen widespread use in the ore petrology community for some time, has yet to be evaluated.

We investigated samples from across the long-lasting Laki fissure eruption, Iceland, in order determine whether crystal mush occurred at the start of the eruption, or throughout its eight-month duration – an important consideration for understanding magma reservoir dynamics and geometry. We did this by using traditional approaches to determine phase proportions and plagioclase size distribtuions, as well as novel QEMSCAN-based approaches. Although we found significant differences between the manaul and automated datasets, largely because of the inability to easily segment glomerocrysts in the latter, being able to easily combine textural and compositional data was a powerful advantage of the automated approach.

Combined composition-size distributions of plagioclase in samples from the Laki eruption. A0.5 is the square root of crystal area. Figure from Neave et al. (2017).

By fitting high-quality, manually derived plagioclase size distributions, we estimated that mush disaggregation occurred around ten days before the eruption of each sample. These observations, which align well with findings from other stidies (Hartley et al. 2015; 2016), suggest that mush disaggregation was progressive and occurred throughout the eruption: the total volume of eruptable magma active at any given time was much less than the final erupoted volume of 15.1 km3.


Neave, D. A., Buisman, I. & Maclennan, J. 2017. Continuous mush disaggregation during the long-lasting Laki fissure eruption, Iceland. American Mineralogist 102, 2007–2021. <Open Access>

The effect of anorthite content and water on quartz–feldspar cotectic compositions in the rhyolitic system and implications for geobarometry

Since moving to Hannover, I have become involved in a number of exciting new projects. One project, lead by François Holtz and carried out by Sören Wilke, involved carrying out a large number of experiments to determine how variations in the anorthite (i.e. calcium) and water contents of rhyolites affects the position of quartz–feldspar cotectics. This is important because the position of quartz–feldspar cotectics can be used as a geobarometer, especially is systems lacking pressure sensitive minerals such as amphibole, but only if the effects of anorthite and water contents are appropriately accounted for.

Our experiments allowed us to define thermal minima and quartz–sanidine–plagioclase triple points on quartz–feldspar cotectics at various pressures, water contents and anorthite contents. This information was then used calibrate an empircal barometer (DEtermination of Rhyolite Pressures; DERP) to esimtate the storage pressure of rhyolitic glasses in equilibrium with quartz and at least one feldspar. DERP is calibrated in the range 50–500 MPa and for any H2O content. Importantly, our findings suggest that rhyolite-MELTS may underestimate the storage pressures of rhyolitic magmas. Bringing emprical and thermodynamic geobarometers into alignment thus represents a key next step in the investigation of rhyolitic magmas.


Wilke, S., Holtz, F., Neave, D. A. & Almeev, R. R. 2017. The effect of anorthite content and water on quartz–feldspar cotectic compositions in the rhyolitic system and implications for geobarometry. Journal of Petrology 58, 789–818.

How to fragment peralkaline rhyolites: Observations on pumice using combined multi-scale 2D and 3D imaging

Thanks to their alkali-rich compositions, pantellerites have much lower viscosities than other rhyolites. As a consequence, these peralkaline magmas erupt in myriad ways to create volcanic landforms that range from lava domes to ignimbrites and fountain-fed lava shields. However, the mechanisms by which such low viscosity melts are able to fragment in explosive eruptions are poorly understood despite the hazards presented by pantellerite volcanoes, above all in the East African Rift.

Building on her MSci project, Ery Hughes investigated this problem of magma fragmentation under the supervision of Marie Edmonds, Kate Dobson and myself by combining 2D (electron microscopy) and 3D (X-ray microtomography) measurements of pumice samples during my MSci.

Slices through an X-ray tomography volume of a pumice from Pantelleria (Hughes et al. 2017).

We found that pantelleritic pumices from Pantelleria are texturally indistinguishable from calc-alkaline pumices from a range of rhyolitic systems, implying that our peralkaline pumices fragmented in a brittle fashion and that their unusual chemistry had little effect on their syn-eruptive textural evolution. We therefore propose that the observed pumice textures developed in response to high decompression rates and that peralkaline rhyolite magmas can fragment when strain localisation and high bubble overpressures develop during rapid ascent.


Hughes, E.C., Neave, D.A., Dobson, K.J., Withers, P.J. & Edmonds, A. 2017. How to fragment peralkaline rhyolites: Observations on pumice using combined multi-scale 2D and 3D imaging. Journal of Volcanology and Geothermal Research 336, 179–191. <Open Access>

Volatile and light lithophile elements in high-anorthite plagioclase-hosted melt inclusions from Iceland

Geochemcial records of mantle processes are progressively degraded as magmas differentiate and ascend towards the Earth’s surface. This degradation is particularly severe in the case of volatiles (H2O, CO2, F, S and Cl) that decouple from melts upon reaching vapour saturation. Melt inclusions – pools of silicate liquid that are partially insulated from changes in the external magmatic enronment by their host crystals – are thus appealing targets for investigating the behvaiour of magmatic volatiles. Although numerous recent studies have critically evaluated the effects of syn- and post-entrapment modification on olivine-hosted melt inclusion compositions, little comparable information is available for plagioclase-hosted systems, depite plagioclase’s abundance in mafic magmas.

In order to address this imbalance in undertanding between olivine-hosted and plagioclase-hosted systems, we present volatile and light lithophile element analyses from a large number of mainly plagioclase-hosted melt inclusions from the 10 ka Grímsvötn tephra series from Iceland. Major and trace element data have already  been presented in study into the pre-eruptive evolution and storage of the tephra series (Neave et al., 2015).

Volatile-trace element systematics in matrix glasses and melt inclusions used to distinguish between pre-, syn- and post-entrapment signals of variability. Figure from Neave et al. (2017).

The uniformly low CO2 content of melt inclusions cannot be explained by either shallow entrapment or shrinkage bubble formation, suggesting that inclusion CO2 contents were controlled by decrepitation instead. High H2O/Ce values in primitive plagioclase-hosted inclusions (182–823) are most easily accounted for by diffusive H2O gain following the entrainment of primitive macrocrysts into H2O-rich melts a few days before eruption (e.g., Hartley et al., 2015). Extreme F enrichments in primitive plagioclase-hosted inclusions (F/Nd = 51–216 versus 15 in matrix glasses) possibly reflect the entrapment of inclusions from high-Al/(Al+Si) melt pools formed by dissolution-crystallisation processes (as indicated by HFSE depletions in some inclusions), and into which F was concentrated by uphill di ffusion: F is highly soluble in Al-rich melts. The high S/Dy of inclusions (300) indicates that primary melts were rich in S in comparison with most oceanic basalts. Although primitive plagioclase-hosted melt inclusions from the 10 ka Grímsvötn tephra series record few primary signals in their volatile element contents they nevertheless record information about crustal magma processing that is absent from olivine-hosted melt inclusions suites.


Neave, D.A., Hartley, M.E., Maclennan, J., Edmonds, M. & Thordarson, T. 2017.  Volatile and light lithophile elements in high-anorthite plagioclase-hosted melt inclusions from Iceland. Geochimica et Cosmochimica Acta 205, 110–118. <Open access>

A new clinopyroxene-liquid barometer, and implications for magma storage pressures under Icelandic rift zones.

Pressure is one of the key intensive variables that controls magmatic phase equilibria, which thus raises the possibily of using mineral and melt compositions to estimate magma storage pressures from erupted products. Such estimations are crucial for addressing geological problems ranging from understanding crustal accretion through to interpreting signals of unrest at active volcanoes.

In this paper with Keith Putirka, I assess the performance of some commonly used barometers that exploit the pressure-sensitive incorporation of jadeite (Jd) into clinopyroxene. We find that many current barometers overestimate the pressure of phase equilibria experiments carried out on H2O-poor basalts at 1–7 kbar by up to 3 bar. Many published magma storage pressure estimates may thus need to be re-evaluated, and revised towards lower pressures.

In order to resolve the of barometer inaccuracy at low pressures, we thus present a newly calibrated Jd-in-clinopyroxene. Our new barometer is suitable for use on hydrous and anhydrous samples that are ultramafic to intermediate in composition. However, we do not recommend using the barometer at temperatures below 1100 °C and at oxygen fugacities above QFM+1 because of reduced accuracy under these conditions. The barometer reproduces its calibration data with a standard error of estimate (SEE) of 1.4 kbar, and tests performed using experiments on bastilc compositons confirm that it is significantly more accurate than previous models.

Calibration and test datasets used to develop and assess a new Jd-in-clinopyroxene barometer for use on ultramafic to intermediate compositions.

We apply our new barometer to a range previously studied eruptions from Iceland’s neovolcanic rift zones. Most eruptions preserve records of mid-crustal crystallisation at 2.6–3.6 kbar; only the highly primtive Borgarhraun eruption recrds crystallisation in the lower crust at 5.7 kbar. While some magma processing takes place immediately beneath Iceland’s central volcanoes, magma evolution under the island’s neovolcanic rift zones is thus strongly mediated by mid-crustal processes.

Spreadsheet for estimating P-T conditions from clinopyroxene-liquid equilibria

Clinopyroxene P-T Dec16


Neave, D.A. & Putirka, K.D. 2017. A new clinopyroxene-liquid barometer, and implications for magma storage pressures under Icelandic rift zones. American Mineralogist 102.

Note that there is an erratum concerning a typographical error in the barometric equation published in this paper

On the feasibility of imaging carbonatite-hosted rare earth element (REE) deposits using remote sensing

Rare earth elements (REEs), which are essential for many modern technologies such as batteries and wind turbines, have characteristic absorption features in visible to shortwave infra-red (VNIR-SWIR) reflectance spectra. Neodymium (Nd) has amongst the most prominent absorption features of the REEs and thus represents a key pathfinder element for the element group as a whole. Given that the world’s largest REE deposits are associated with carbonatites, we collected spectral, petrographic and geochemical data from a predominantly carbonatitic suite of rocks to assess the feasibility of imaging REE deposits using remote sensing. Rock samples were drawn from a number of sources including the Harker Collection in Cambridge, Cambourne School of Mines and the Natural History Museum in London. REE ores from the Bayan Obo (China) and Mountain Pass (USA) mines, as well as REE-rich alkaline rocks from the Motzfeldt and Ilímaussaq intrusions in Greenland were included in the sample suite.

Hand specimens of REE ores from Mountain Pass, USA.
Hand specimens of REE ores from Mountain Pass, USA. Samples from the Harker Collection.

By simulating the response of a number of remote sensing instruments, we demonstrated that hyperspectral instruments with capabilities equivalent to the operational Airborne Visible-Infrared Imaging Spectrometer (AVIRIS) and planned Environmental Mapping and Analysis Program (EnMAP) instruments have the spectral resolution necessary to detect Nd absorption features, especially in high grade samples with economically relevant REE accumulations.

REE ore from Bayan Obo, China. The fibrous and equant bright phases are bastnäsite-(Ce) and monazite-(Ce) respectively. Sample from the Natural History Museum.

Unfortunatly, most REE-rich outcrops are too small to be detected by satellite-based platforms. However, Nd absorption features should be identifiable in high-quality, airborne, hyperspectral datasets collected at meter-scale spatial resolutions. Future deployment of hyperspectral instruments on Unmanned Aerial Vehicles (UAVs) could enable REE grade to be mapped at the cm-scale across whole deposits.

This research also featured in this article.


Neave, D.A., Black, M., Riley, T.R., Gibson, S.A., Ferrier, G., Wall, F. & Broom-Fendley, S. 2016. On the feasibility of imaging carbonatite-hosted rare earth element (REE) deposits using remote sensing. Economic Geology 111, 641–665. <Open Access>

The evolution and storage of primitive melts in the Eastern Volcanic Zone of Iceland: the 10 ka Grímsvötn tephra series (i.e. the Saksunarvatn ash)

The environmentally impacting AD 1783–84 Laki eruption was the largest Icelandic eruption to have been directly obseved by humans (Thordarson et al., 1996). However, it is by no means unique in Iceland’s volcanic history: Thordarson & Höskuldsson (2008) note that over 50 eruptions >1 km3 in volume have taken place in Iceland since the end of the last glaciation. The 10 ka Grímsvötn tephra series, or Saksunarvatn Ash, which is distributed across the North Atlantic from Greenland to Germany, is thought to have been generated in a series of large, phreatomagmatic eruptions within the Grímsvötn volcanic zone at the end of the last glacial period (Grönvold et al., 1995; Thordarson, 2014). In this first petrological study of the tephra, we (a team from the universities of Cambridge, Manchester and Iceland) exploited the abundance of primitive crystals and melt inclusions in samples from Lake Hvítárvatn in central Iceland in order to investigate magma evolution and storage processes.
Crystal textures in the 10 ka Grímsvötn tephra series from lake Hvítárvatn in central Iceland. a) zoning in plagioclase, b) melt inclusions in plagioclase, c) sector zoning in clinopyroxene and d) inclusions in olivine. Figure from Neave et al. (2015).

Following the approaches laid out by our recent work on Laki and Skuggafjöll, we defined evolved and primtive macrocryst assemblages in tephra samples, the latter of which was out of equilibrium with the matrix glass and probably derived from disaggregated crystal mushes (e.g., Halldorsson et al., 2008). High-anorthite plagioclase-hosted melt inclusions provided the first direct evidence for the supply of high-Mg#, incompatible trace element-depleted mantle melts to the base of the lithosphere in Iceland’s Eastern Volcanic Zone. Through the critical application of clinopyroxene-melt and melt barometers (Putirka, 2008; Yang et al., 1996) , we suggested that the primtive macrocryst assemblage formed within the mid-crust (4±1.5 kbar) and that the evolved assemblage formed in the shallow crust (<2 kbar) shortly before eruption. We showed, however, that clinopyroxene-melt equilibria are not well calibrated at conditions relevant for the tephra’s pre-eruptive storage. We therefore made the case for further exploration of basalt phase equilibria in the critical 1–7 kbar interval, which is a primary aim of my Humboldt Research Fellowship in Hannover.


Neave, D.A., Maclennan, J., Thordarson, T. & Hartley, M.E. 2015. The evolution and storage of primitive melts in the Eastern Volcanic Zone of Iceland: the 10 ka Grímsvötn tephra series (i.e. the Saksunarvatn ash). Contributions to Mineralogy and Petrology 171, 21. <Open Access>

Diffusive over-hydration of olivine-hosted melt inclusions

Olivine-hosted melt inclusions are ofen used to estimate the pre-eruptive H2O content of magmas (Métrich & Wallace, 2008). However, it has been noted for a number of years that H2O appears to ‘leak’ out of melt inclusions during ascent and eruption (Massare et al., 2002; Chen et al., 2013). Rare cases of H2O gain have also been noted (Kolezsar et al., 2009). Recent experiments and modelling has clarified the mechanisms of H2O loss– by diffusive re-equilibration through the host crystal – and has opened up the possibily of extracting timescales from the extent of H2O exchange (Gaetani et al., 2012; Bucholz et al., 2013).

In this study, led by Margaret Hartley at the University of Manchester, we showed that different populations of melt inclusions from the Laki and Skuggafjöll eruptions in the Eastern Volcanic Zone of Iceland experienced diffusive loss or diffusive gain of H2O. Some rapidly quenched melt inclusions from the Laki tephra and subglacially-quenched pillow glasses from Skuggafjöll had coherent H2O/Ce values of ~180 that we interpreted as the primary mantle value (e.g., Michael, 1995). However, many inclusions from the Laki lava flow had very low H2O/Ce values consistent with H2O loss during transport in an extensive lava tube network at the surface. Conversely, most inclusions from Skuggafjöll, as well as most low-Ce, primitive inclusions from Laki, had elevated H2O/Ce values of up to ~1000 that are indicative of H2O gain during storage in the crust.

A figure from Hartley et al. (2015) summarising the processes by which diffusive loss and gain of H2O has been observed in the Laki and Skuggafjöll eruptions.

Using the diffusive re-equilibration model of Bucholz et al. (2013), we placed minimum constraints on the residence times of dehydrated inclusions in the Laki lava flow and over-hydrated inclusions in evolved melts immediately prior to the eruptions. The timescales were on the order of days to tens of days in both cases. Finally, we demonstrated that diffusive gain, as well as diffusive loss, can be observed in a number of global datasets where primitive, H2O-poor inclusions are mixed into more enriched and/or evolved melts before eruption. Thus, rather than viewing the open system nature of olivine-hosted melt inclusions as weakness, it can be exploited to gain further insights into pre-eruptive magma processes.


Hartley, M.E., Neave, D.A., Maclennan, J., Edmonds, M. & Thordarson, T. 2015. Diffusive over-hydration of olivine-hosted melt inclusions. Earth and Planetary Science Letters 425, 168–178. <Open Access>