The temperature of the Icelandic mantle from olivine-spinel aluminum exchange thermometry

A comparison of previous estimates of mantle potential temperature at Iceland (red horizontal bars) to our new estimate based on combining crystallisation temperatures with crustal thickness and geochemical constraints (red/yellow histograms).  Estimates of mantle potential temperature for MORB are shown for reference in blue.

Variations in mantle temperature are a primary control on the melting behaviour of the mantle. Despite its importance for understanding present day volcanism and the thermal evolution of the Earth, mantle temperature has remained difficult to quantify. Proxies, such as crustal thickness, seismic velocity, and melt chemistry must be used; however, each suffers from its own uncertainties and trade-offs with other equally uncertain parameters. Melting anomalies, such as Iceland, have been variously linked to raised mantle temperature, unusually fusible mantle, or enhanced mantle flow.
Several studies have recently used olivine crystallisation temperatures, derived from olivine-spinel aluminium-exchange thermometry, as a proxy for mantle temperature. When offsets in olivine crystallisation temperatures are used to infer mantle temperature variation directly, it is implicitly assumed the method does not suffer from trade-offs arising from greater mantle fusibility or enhanced mantle flow.

Summary of the new crystallisation temperature estimates we made in this study.  Crystallisation temperatures were calculated from the composition of olivine-spinel pairs using the Wan et al. (2008) and Coogan et al. (2014) Al-exchange thermometer.

Using a new set of crystallisation temperatures determined for four eruptions from the Northern Volcanic Zone of Iceland, we demonstrate crustal processes, rather than mantle processes, are responsible for the crystallisation temperature variation within our dataset. However, the difference between Icelandic crystallisation temperatures and those from MORB, are most easily accounted for by substantial mantle temperature variations between the two locations.

The thermal structure of the mantle melting region will determine the chemical and thermal properties of the melts entering the crust. As lithological heterogeneity can exert a large effect on the thermal structure of the melting region, we assess its effect on crystallisation temperature using a forward thermal model of multi-lithology melting. Using crystallisation temperature estimates from Iceland and MORB as examples, we demonstrate that in the absence of further constraints on the thermal structure of the melting region (e.g. crustal thickness), crystallisation temperature provides only a weak constraint on mantle temperature.

By inversion of our thermal model, fitting for crystallisation temperature, crustal thickness, and fraction of bulk crust derived from pyroxenite melting, we demonstrate that a mantle temperature excess over ambient mantle is required for Iceland. We estimate a mantle temperature of $\mathsf{1480^{+37}_{-30}}$°C for Iceland, and $\mathsf{1318^{+44}_{-32}}$°C for MORB.

Online [publisher]: https://dx.doi.org/10.1002/2016GC006497

Reference: Simon Matthews, Oliver Shorttle, John Maclennan. The temperature of the Icelandic mantle from olivine-spinel aluminum exchange thermometry. Geochemistry, Geophysics, Geosystems (2016)

A significantly hotter mantle beneath Iceland

A false color image of an olivine crystal (centre) found in the Borgarhraun eruption of north Iceland. The color picks out variations in the crystal’s composition. We estimated the temperature at which the crystal grew by comparing the composition of the olivine to that of the spinel crystal which has been trapped inside it (small red circle inside crystal) . Horizontal scale = 1.5mm.

We have shown that the Icelandic mantle is unusually hot, a result which has been featured in AGU’s Eos magazine.  We measured the chemistry of olivine and spinel crystals that grew from magmas sourced directly from the Icelandic mantle. These crystals recorded crystallisation at almost 1400°C, indicating that the underlying mantle must be at least this hot. By developing a model to account for how the temperature of the mantle changes during melting, we were able to show that this crystallisation temperature is consistent with a mantle temperature prior to melting of closer to 1500°C.  This is more than 160°C hotter than the mantle underlying most regions on Earth.

And head here for a more detailed summary of the science.

Thesis: Characteristics of a heterogeneous mantle

A copy of my PhD thesis, carried out at the University of Cambridge between Oct. 2009 and Oct. 2012 can be found here. I was supervised by John Maclennan and Alex Piotrowski.

A more digestible read of Chapters 4, 5 and 6 can be found in the associated papers, which are respectively:

Plume-ridge interaction

Map of melt region traversal distances for mantle flow paths extending away from the Iceland plume centre. The flow of plume material beneath spreading centres, with the concomitant decompression and partial melting progressively extracts the enriched components giving the plume mantle its distinctive trace element and isotopic character. Melt regions are marked out as regions in black, 120 km wide. Figure modified from Shorttle et al. (2013), based on the model presented in Shorttle et al. (2010).

The dispersal of mantle plumes in the shallow mantle causes excess volcanism and surface uplift over thousands of kilometres. These phenomena result primarily from a mantle plume’s excess temperature with respect to ambient mantle, which will be the main reason a plume is buoyantly ascending through the mantle in the first place. However, mantle plumes also tend to produce basalts that have distinct trace element and isotopic compositions, that are unlikely to be due solely to changing mantle potential temperature. Like surface uplift, these geochemical characteristics of a mantle plume can be used to trace its dispersal in the shallow mantle if it intersects passive melting features like mid-ocean ridges.

In the simplest model of plume dispersal, the geophysical and geochemical observables recording the presence of a mantle plume will track each other, so that when a mid-ocean ridge is shallow because of underlying hot mantle, it will also erupt trace element enriched basalts. However, in two classic cases of plume-ridge interaction, the Galapagos and Iceland, geophysical and geochemical tracers of plume dispersal in the shallow mantle are decoupled and apparently asymmetric about the plume axis. These observations have led to models of asymmetric plume flow in response to prevailing mantle convection, fracture zone blocking of plume outflow, or tilted mantle plumes.

In this paper we show that by allowing a realistically located centre of plume symmetry to be found for Iceland and the Galapagos, geophysical indicators of plume dispersal can be shown to be radially symmetric. However, geochemical enrichment along the ridges either side of the Iceland and Galapagos plumes remain highly asymmetric. These observations can be reconciled by considering that the enriched plume component, carrying much of the trace element load that gives the plume its distinctive geochemical character, is more fusible (i.e. is a pyroxenitic heterogeneity) than ambient mantle. The implication of this is that partial melting during outflow of the plume material will preferentially deplete it in the enriched component, leaving any basalts that the source goes on to produce relatively depleted. It is then the asymmetry in the distribution of spreading centres about the Iceland and Galapagos plumes that ingrows asymmetry in the chemistry of the plume material (see figure above for Iceland), as spreading centres cause the laterally flowing plume material to decompress and undergo small degrees of melting.

Our model therefore explains decoupling between geophysical and geochemical indicators of plume dispersal without requiring complex dynamics, just the observation that enriched mantle domains will be more readily extracted from a source than depleted mantle.

Online [Department repository, full text]: http://eprints.esc.cam.ac.uk/1400/2/G3_Shorttle_Maclennan.pdf

Online [publisher]: http://dx.doi.org/10.1029/2009GC002986

ReferenceShorttle, Oliver, J. Maclennan, and S. M. Jones. Control of the symmetry of plume‐ridge interaction by spreading ridge geometry. Geochemistry, Geophysics, Geosystems 11, no. 7 (2010).