Skip to main content

The fate and behaviour of volatiles during subduction of oceanic crustal material towards greater mantle depths

Final Report Summary - FATEMANTLE (The fate and behaviour of volatiles during subduction of oceanic crustal material towards greater mantle depths)

Knowledge of the speciation, abundance and mobility of volatiles (such as H2O and CO2) in the Earth’s deep interior remains an area of great uncertainty, but is crucially important because the silicate mantle, which extends from approximately 30-2891 km, comprises the largest reservoir on Earth for most volatile elements. The exchange of volatiles between the surface and the interior has influenced the availability of volatile elements in the atmosphere, oceans and biosphere over geological time. It is possible, for examples, that the Earth’s sea level has been influenced over geological time by changes in the distribution of H2O between the surface and the interior. An important and open question is the extent to which the oceans are currently being either returned or expelled from the mantle. Volatile species inside the mantle also affect transport properties such as rheology and, therefore, affect mantle convection. They also promote the formation of melts in the deep interior and can, therefore, affect the geochemical cycling of incompatible elements. Volatiles escape the interior through volcanic eruptions. The chief means by which volatiles enter the interior is through subduction, i.e. sinking of altered oceanic lithosphere back into the Earth’s mantle. Whether volatiles can pass into the deep mantle will depend on the extent to which they can be incorporated into solid phases as the material sinks and heats up. If volatiles fail to remain in the solid state during subduction they will be expelled from the mantle on a relatively short timescale within volcanic magmas. Deeply subducted natural samples are inaccessible, however, so the only way to study this volatile transport process is to simulate it in high-pressure and temperature experiments.
In the first part of this study high pressure and temperature experiments have been performed to simulate the behavior of H2O within subducting oceanic crust as it sinks into the mantle. The ocean floor is hydrated as a result of alteration by cycling hydrothermal systems. As the entire oceanic plate sinks back into the mantle it heats up and releases H2O as the hydrous minerals become unstable. This H2O rises into the overlying mantle, which consequently melts producing magmas, which rise to the surface and form island arc volcanos. In this way much of the subducted H2O is returned to the Earth’s surface on a relatively short timescale. At the conditions where hydrous minerals break down in the oceanic crust, however, pressures are very high and some of the H2O produced can be dissolved as hydroxyl defects in the surrounding nominally anhydrous minerals i.e. the typical minerals of the mantle that normally contain no H2O. Determining the amount of H2O that can remain in these solid minerals of the subducting lithosphere is very important because this amount of H2O will continue to be carried back into the deep interior of the Earth. By examining the breakdown of hydrous minerals and determining the amount of water that partitions into nominally anhydrous minerals at these conditions, the amount of water entering the deep mantle has been constrained.
Samples with the composition of hydrated oceanic crust, but with variable H2O contents, were welded into metal capsules and compressed using a 1000 tonne multianvil press to pressures that correspond to depths in the mantle of 100-200 km. This is the depth interval over which most dehydration reactions take place within subduction zones. The samples were then heated to temperatures between 850-1500°C and equilibrated for several days in order for the mineral phases to achieve chemical equilibrium with each other. The experiments were then quenched by dropping the temperature rapidly and thus preserving the high pressure and temperature mineral assemblage. By recovering and analyzing the sample the stability fields of hydrous minerals could be determined and the trace quantities of H2O that dissolve in the normal mineral phases upon the breakdown of these hydrous minerals was determined. The recovered samples were very small, such that state-of the-art methods had to be employed to determine these H2O contents. By using synchrotron Fourier-transform infrared (FTIR) spectroscopy at the European Synchrotron Radiation Facility (ESRF) in Grenoble, France, small amounts of H2O could be measured in the mineral phases. Phengite mica is one of the few hydrous minerals stable to depths > 100 km. The amount of water that passes into the solid minerals at the conditions where phengite breaks down has been determined. The results show that the concentrations stored in the minerals are relatively low. While concentrations of over 1000 ppm had previously been reported, measurements in this study show values that are a factor of 2 smaller. This implies that the crustal portions of slabs may become relatively dry, which has important implications for the cycling of H2O in the mantle as it means that relatively little H2O is being subductred into the deep mantle. This then implies that sea levels have likely remained relatively constant over recent geological time.

The project has also examined homogenization processes between subducted oceanic crust and mantle rocks. The rocks of the basaltic oceanic crust are formed at mid ocean ridges by mantle melting and are richer in Si, Al, Ca and Fe compared to the residual mantle, which is composed of a rock called peridotite. Subduction, therefore, introduces basaltic chemical heterogeneities into the peridotite mantle, the fate of which is intriguing as they also carry volatile components and are more fertile, i.e. melt at lower temperatures to produce more magma, than normal mantle rocks. Once subducted the chemistry of crustal rocks will remain distinct from the surrounding mantle because in the solid state, diffusion processes are too slow to remove these heterogeneities. However, once these rocks melt they can interact with the surrounding mantle leading to a wider distribution of these chemical anomalies. Subducted basaltic oceanic crust forms a high pressure rock called an eclogite at depths >30 km, composed of garnet and clinopyroxene. Experiments were performed to examine the process of eclogite melting and reactive interaction of the resulting melt with surrounding mantle peridotite. One of the most interesting and unexpected results was that at conditions equivalent to depths >100 km eclogitic partial melts do not percolate through overlying peridotite layers. Instead the SiO2 rich partial melt reacts with olivine, the main mantle mineral, to form relatively narrow zones comprised dominantly of the new mineral orthopyroxene. No melt is observed to percolate through these zones, which effectively seal off the remaining peridotite mantle rocks from further melt reaction. Eclogitic melt fractions of over 40 % were totally isolated from the overlying peridotite by orthopyroxene reaction layers that were less than 10 microns in thickness. This isolation was maintained in experiments that were performed for over 5 days! Extensive analysis using electron backscattered diffraction failed to find any evidence that melt can percolate through these layers, which were shown to have randomly oriented mineral grains. This is a fascinating and unexpected result because it is normally assumed that such melts would react immediately with the surrounding mantle, rather than accumulating. During decompressive melting of eclogitic heterogeneities, this process of orthopyroxene formation would allow melts to accumulate. Numerous geophysical studies have identified the potential presence of melts in the deep mantle, which in the past have been attributed to the presence of volatile-induced melts. The accumulation of partial melts from eclogitic heterogeneities, however, could explain this observation and may have further consequences for the geochemistry and dynamics of mantle melting.
A final goal of the project was to experimentally investigate how partial melting of eclogite and peridotite rocks affects the abundance of oxygen in the mantle. Elements like iron can be bonded to different amounts of oxygen as a result of losing different numbers of electrons during bonding. Most iron in the mantle is termed ferrous iron and is bonded to one oxygen atom i.e. FeO. Ferric iron, Fe2O3, also exists, however, and the extra oxygen that is then bonded to iron can be exchanged with other elements particularly volatile elements such as carbon. The distribution of ferric iron between experimentally produced mineral and melts was measured for the first time using micro-analysis techniques at a synchrotron X-ray source. The results show that ferric iron is an incompatible component in the mantle and partitions more preferentially into melts than minerals. The extraction of partial melt is, therefore, one way in which oxygen is removed from regions of the mantle. Conversely small degree melts will cause localised oxidation. Removal of oxygen leads to the formation of so called reduced areas where carbon for example would be stable as graphite or diamond, rather than in oxidised form as CO2. Melting therefore can ultimately influence the stability of volatile element bearing compounds. While these ideas had been proposed in the past, based on some field observations, the actual proof that this occurs had been lacking. Eclogitic rocks subduced into the mantle are expected to be relatively oxidised. However, eclogitic rocks from the mantle are actually quite reduced and as a result contain large proportions of the mantle’s diamonds. These results explain how eclogitic rocks can become reduced and can, therefore, act as host rocks for the deep subduction of carbon as diamond.
More information at