Startseite der Alexander von Humboldt-Stiftung Humboldt Kosmos Startseite


Hunting for Earth's primary melts

The challenge of understanding the deep earth interior - a direct approach

By Alexander V. Sobolev

Alexander V. Sobolev (Fotos: Sobolev)
Alexander V. Sobolev
More than 90 percent of the Earth's mantle (the region outside its liquid core) consists of solid but hot and plastic silicate rock, which is able to move in response to temperature and gravitational gradients. Solid-state convection (moving in a viscous state) in the mantle is the major engine of global geological processes, such as the spreading of ocean floors, the drift of continents and mountain building. Global geophysical observations on the velocities of seismic waves in the Earth interior, as well as gravity and heat flow distribution on the Earth surface all suggest impressive movements of hundreds of kilometers in scale in the Earth's mantle.

To better understand how this works, one should know the mantle composition and it's physical parameters. However, the convecting mantle is not available for direct observation in its original form because normally it lies more than 10 km beneath the ocean ridges and at least 100 km below the continental surface. The only available samples of mantle rocks carried as xenoliths by magmas or exposed in outcrops are (with rare exceptions) related to the viscous and brittle part of the mantle (called lithosphere), which has been protractedly isolated from convection and has been thus significantly modified by secondary processes. This situation leaves unresolved principal questions of Earth dynamics, including the scale and origin of mantle heterogeneities, and the scale and the rate of material fluxes within the mantle.

Fig. 1. Schematic figure, showing a recent view of the structure and dynamics of the Earth' s mantle. Convecting mantle is shown in yellowish colors with arrows indicating major directions of material fluxes. Oceanic lithosphere originates by decompression melting of the upper convecting mantle. After some tens of millions of years of horizontal moving and cooling oceanic lithosphere sinks under continental or oceanic lothosphere into the hot convecting mantle (subduction) and then is partly recycled back to the surface by mantle plumes. The latter originate mainly in the outer core-mantle boundary and generate on the surface oceanic islands like Hawaii. The figure was designed in the geochemical department of Max Planck Institute for Chemistry. Click here for a larger image.
Figure 1
Nevertheless, samples from the mantle are carried to the surface by melting and volcanism. In fact, melting is a universal feature of the convecting mantle. Primary melts (in equilibrium with mantle sources) are formed by partial (1 - 30 percent) melting of mantle rocks. Thus, primary melts are direct witnesses of processes in the mantle, and they potentially represent powerful probes into the composition of the parental mantle and the physical parameters of melting (e.g temperature, pressure and degree of melting), as well as dynamic features of magmatic processes (e.g permeability of source, separation efficiency of melt, timescales, etc.). Recent developments in experimental petrology easily quantify these entire mantle parameters from the known primary melt composition. But do we know the composition of primary melt? The answer to this question even 10 years ago would have been quite easy. Just analyze basaltic lavas and take into account rather small effects of their crystallization on major and trace elements content and you will get an estimate of primary melt composition.

This approach was based on the assumption that the main processes, which could affect the composition of primary melt, are limited by its crystallization and the separation of crystals. In this line ratios of incompatible elements (which strongly prefer melt over crystalline phases) and ratios of radiogenic to stable isotopes of the same element in basalts would be representative for their mantle source (see review of Hofmann A.W., 1997, Nature 385, 219-229).

However in the early 90's it was clearly shown that after extraction from the mantle source, primary melts are commonly modified either by mixing with other melts or by reaction en route (e.g. Johnson K. T. M. et al, 1990 J. Geophys. Res. 95, 2661-2678 and Sobolev A. V. and Shimizu N., 1993, Nature 363, 151-154). Even until recently the scale of this modification was severely underestimated.

Fig. 2. Inclusions of melt (brown) in olivine crystal (bright and transparent) from mid-oceanic ridge basalt. Inclusions possess negative shape of host mineral and consist of homogeneous melt, which was quenched to glass by submarine eruption. Click here for a larger image.
Figure 2
Consequently, the only available option for studying primary melts directly is to find a way to sample them before they have been modified. This is possible by exploiting a natural sampling mechanism: growing crystals frequently trap traces of the surrounding melt (or fluid) deep in the magmatic system. The basic idea of this approach is to use the unique ability of host minerals to protect melt inclusions from subsequent, complicated processes which take place in the bulk magma en route to the surface (see Roedder E., Fluid Inclusions. Reviews in Mineralogy 12, 644, 1984).

Thus, melt inclusions can record and retain information on the composition of primary melts, which is blurred or lost altogether in the erupted lava. The problem is, however, how to analyze those inclusions, which are normally very small in size (e.g. 0.03 - 0.1 millimeters). The present revolutionary development of in-situ micro analytical tools, namely laser ablation multi-collector mass spectrometers and new generation ion microprobes, are now enabling us to exploit the potential scientific benefits of melt inclusion studies to an unprecedented extent.

Systematic studies of melt inclusions in the earliest olivine crystals from mid-oceanic ridges (such as the Mid-Atlantic ridge) and mantle plumes (such as Hawaii) were carried out by state of the art in-situ micro analytical tools and modeling at Vernadsky Institute of Geochemistry, Russian Academy of Sciences (Russia), Woods Hole Oceanographic Institution (USA) and Max Planck Institut for Chemistry (Germany). These reveal the following major results:

1. Each volcanic plumbing system or particular lava represents a dynamic mixing of numerous primary melts, which often generated in different isolated mantle sources. Unmixed primary melts are rare and show extreme compositional ranges far exceeding those of bulk surface lavas. They are usually found as inclusions in the earliest crystals formed in the deepest parts of the plumbing system.

2. The compositional and isotopic ranges of the recovered primary melts suggest both highly efficient open system melting, fast melt transport and small-scale compositional heterogeneity of mantle sources in all volcanic environments studied so far (e.g. Sobolev, 1996, Petrology 4, 209-220; Saal A. E. et al., 1998, Science 282, 1481-1484; Sobolev A. V. et al, 2000, Nature 404, 986-990).

Fig. 3. Cumulative number of papers in journals covered by the Science Citations Index Expanded from the Institute for Scientific Information (ISI) citing melt or glass inclusions in their titles, abstracts or key words. Data extracted from Web of Science, ISI. Click here for a larger image.
Figure 3
3. A detailed case study of one mantle plume (Hawaii) suggests that compositional heterogeneities in the mantle source beneath Hawaiian volcanoes are produced by mixing and reaction of mantle with recycled fragments of old oceanic lithosphere, which has been descended (subducted) into convecting mantle (Sobolev A. V. et al, 2000, Nature 404, 986- 990). The latter has directly supported the idea proposed by A.W. Hofmann and B.W. White, 1982 (Earth Planet. Sci. Lett. 57, 421-436).

The melt inclusions approach of studying processes in the Earth interior is becoming increasingly popular. This is clearly seen in the nearly exponential growth of a number of publications on the subject (Fig. 3). Our approach in the Max Planck Institut for Chemistry in Mainz was recently supported by the Wolfgang Paul Award from the Alexander von Humboldt Foundation. In this project, named "Melt inclusion studies to assess the nature of primary melts in the mantle", we plan to concentrate on considerably increasing the database of primary melts by studying new samples from our current type localities such as the Hawaii, Kamchatka and mid-ocean ridges, and from studying volcanic environments such as Iceland that are new to us. In fact these investigations are in progress already.

 zum Seitenanfang

Prof. Dr. Alexander V. Sobolev
Fachgebiet: Geochemie
Förderprogramm: Wolfgang Paul-Preis
Gastinstitution: MPI für Chemie/Mainz
Heimatuniversität: Russian Academy of Sciences
Vernadsky Institute of Geochemistry and Analytical Chemistry, Moscow/Russische Föderation

Calcite in granite: why does it matter? Predicting the effect of acidity from the atmosphere on the chemistry of lakes. weiter >>

Kann der Computer den Weg weisen? Neuronale Netzwerke reagieren ähnlich wie das menschliche Gehirn. weiter >>

From cells to cities: The evolution of composite individuality - early life leaves its record in the rocks. weiter >>

"Wir brauchen den Methodentransfer": Interview mit Professor Dr. Ralf Emmermann, Präsident der Alfred Wegener-Stiftung. weiter >>

A million people would be threatened by an eruption: Prehispanic disasters at Popocatépetl. weiter >>

Playing with fire: Mitigating the effects of volcanic disasters in the south pacific. weiter >>

Climate Change Detection from Space: Satellite-based monitoring of the earth's polar regions. weiter >>

Immer in Bewegung: Die Erde als dynamisches System. weiter >>

Titelthema 14:2008:
Wissen schafft Entwicklung
weiter >>
Titelthema 13:2007:
Kontinent der Geisteswissenschaften
weiter >>
Titelthema 12:2007:
Das Rennen um die Qualität
weiter >>
Titelthema 11:2006:
Wissenschaft und Politik
weiter >>
Titelthema 10:2006:
Die Macht der Bilder
weiter >>
Titelthema 9:2005:
weiter >>
Titelthema 8:2005:
Der Griff nach den Sternen
weiter >>
Titelthema 7:2004:
weiter >>
Titelthema 6:2004:
weiter >>
Titelthema 5:2003:
50 Jahre Humboldt-Stiftung
weiter >>
Titelthema 4:2003:
weiter >>
Titelthema 2:2001:
weiter >>
Titelthema 1:2001:
Bio- und Gentechnologie in Deutschland
weiter >>