Full size image available through contact
The roots of most volcanoes lie in the Earth's mantle, often at depths of about 100 km where melts are produced from solid, hot mantle rock rising from even greater depths, hundreds or even thousands of kilometers. The melt, called magma, rises through fissures in the uppermost mantle into magma chambers located somewhere near the base of the volcano itself (5 to10 km deep).
Magma chambers are mixing bowls where the incoming batches of fresh mantle magma start to crystallize and to be blended into the material that finally erupts as surface lava. When the first crystals form, they trap small parcels of melt (measuring only one tenth to one hundredth of a millimeter in diameter). These microscopic melt inclusions are quenched as glasses, which preserve the primary composition of the magma at the time of trapping. They have been analyzed using modern micro-analytical tools, such as the "ion microprobe," by Alexander Sobolev, (Russian Academy of Sciences, Moscow), visiting scientist at the Max Planck Institute for Chemistry in Mainz and recent winner of the Alexander von Humboldt Research Award.
Fig.1: Melt inclusions in an olivine crystal from Mauna Loa Volcano, Hawaii (in a hundred-fold magnification).
Analyses of hundreds of melt inclusions have yielded normal, basalt-like compositions. However, six "exotic" inclusions were found which have highly unusual trace-element contents. The chemical "fingerprints" of these trace elements match those of only one possible rock type, a so-called "gabbro", which could have produced these melts. But this rock type is not found in the Earth's mantle where the Hawaiian magmas are produced, and it cannot even exist there, because the most abundant mineral making up the gabbro, feldspar, breaks down under (mantle) conditions of high pressure (and temperature) and is transformed to new minerals, for example garnet.
The solution to this paradox was found by matching Sobolev's new observations with an old hypothesis of Albrecht Hofmann, director at the Max Planck Institute. Hofmann speculated many years ago that the oceanic crust, which is known (from plate tectonic theory) to be newly formed at mid-ocean ridges and returned to the mantle through "subduction" beneath deep-sea trenches, may be segregated and sink to the base of the mantle, perhaps 2900 km deep. After a long "rest" lasting perhaps a billion years or more, this ancient crust is swept up by rising "mantle plumes" and starts to melt again at a depth of about 100 km to produce ocean island volcanoes such as those forming the island of Hawaii.
This hypothesis, although it has received increasing support by indirect geochemical evidence such as isotope ratios of elements produced by natural radioactive decay, had never been confirmed by direct chemical data. Such confirmation is now provided by the unique trace-element fingerprint of the exotic melt inclusions.
The gabbro (which makes up the lower half of the oceanic crust world wide) is transformed during subduction into a new, feldspar-free rock type called eclogite, and this eclogite inherits the characteristic trace-element fingerprint of its parental gabbro. When it is recycled to the surface in a mantle plume, it melts together with normal mantle rocks. These different melts are normally thoroughly mixed in magma chambers before they erupt. Therefore, the geochemical fingerprints of the different melting rocks are not easily recognized because they are diluted and mixed. Only by studying melt inclusions trapped in crystals that were formed before this mixing process took place can the scientists catch the magma compositions in a sufficiently early stage, so that they can "see" the undiluted and obvious fingerprint of that ancient, recycled oceanic crust.