Water has a strong influence on the chemical and physical properties of silicate minerals, fluids, and melts in the mantle ¹. The presence of water affects how the mantle melts by lowering the temperature of the peridotite solidus. The bulk concentration of water in the Earth's mantle may be as high as 1000 ppm ². The origin of mantle water is still being debated. Part of it is likely primordial in that it may be the remnants of incomplete degassing of the Earth, or the contribution of an extraterrestrial body colliding with the Earth early in its history ³. However, water released from the mantle via melt transport to the crust, or degassing, may be recycled back into the mantle in subduction zones. Subduction processes likely add tremendous quantities of water back into the mantle (about 6 times what is released by arc volcanoes ⁴). The reactions occurring during the dehydration of a down-going slab, the quantity of water being kept by the slab and that expelled to the overlying mantle wedge, how water is stored in these various components, and in what form it can be released, are therefore important factors in understanding the water budget of the mantle. This project proposes to estimate the water content of the down-going slab and that of the overlying mantle wedge, in order to shed light on the water transport budget at subduction zones.

Much of the understanding of the water budget in subduction zones comes from measurements of water contents of arc volcanic rocks, and studies of metamorphic reactions in subducting slabs. Here, it is proposed to directly measure the water contents of the rocks at the very source where these processes occur; mantle peridotite xenoliths coming from subduction zone environments, and eclogites considered to be from slabs. The measurements on minerals of the rocks chosen for this investigation will be made using Fourier Transform Infrared Spectrometry (FTIR) at the Smithsonian Institution.

Water is present in the mantle in hydrous minerals, as water-bearing fluids, in silicate or carbonate melts, and also as hydroxyl groups in the nominally anhydrous minerals ^{1, 2}. Nominally anhydrous minerals are of special interest because they may represent an important way of storing water in subducted slabs at depths where hydrous minerals may not be longer stable, and in the mantle above slabs. Pyroxenes (diopside) from peridotitic mantle xenoliths in particular, appear to be able to contain 500 ppm H ₂ O in weight ¹, and up to 1000 ppm has been measured in clinopyroxenes (omphacite) in eclogite xenoliths from kimberlite pipes ⁵. Olivine and garnet can retain up to 50 ppm H ₂O ¹.

Here, it is proposed to attempt to correlate the water content of continental mantle samples with their tectonic environment, i.e. subduction and off-subduction. Melt inclusions trapped in olivine are richer in water in subduction zone lavas than those that occur in MORB ⁶. Although the higher water content of garnets from Colorado Plateau peridotite xenoliths, as well as the strong occurrence of hydrous minerals in peridotite xenoliths from this region have been interpreted as the signature of subduction during the Cretaceous in this region ¹, an unambiguous relation between the water content of nominally anhydrous minerals and the subduction environment has not yet been demonstrated in peridotites. Here, peridotite xenoliths from areas of present day subduction will be analyzed for their water contents: lherzolites from Ichinomegata, a xenolith site in the Japanese arc, and harzburgites from Simcoe, a xenolith location in Washington State (USA) ^7-9. The peridotites from both of these locales have high f _O2, characteristic feature of subduction zone mantle rocks and their isotopic systematics (Sr, Os, Pb) are also consistent with such an interpretation. Other spinel mantle xenoliths are also available for comparison with off-subduction peridotites: Canadian Cordillera mantle xenoliths (my PhD thesis), which oxygen fugacities are low in the Southernmost part ¹⁰, and thus appear not to record subduction influence (further oxygen fugacity investigation would be needed for the northern sites), and Mexican xenoliths which oxygen fugacity increases toward the west, possibly related to increasing subduction signature ¹¹. Finally, the analysis of water contents of minerals (omphacite and garnet) of the well documented eclogites from the Mönchberg Gneiss Massif (Germany), Cabo Ortedgal (Spain) and Lago di Cignana (Italy) ¹², will permit investigation of the potentiality of a slab to transport water into the deep mantle. Estimating the water content of the slab (eclogites from Europe), of the mantle wedge above the slab (Ichinomegata, Simcoe), and in the mantle far from any slab influence (selected samples of the Canadian Cordillera and Mexico) will allow to constrain the water budget of a subduction zone.

Samples for this investigation are made available from Alan Brandon, Harry Becker, and James Luhr, and one of the world's largest collection of ultramafic xenoliths is present at the Smithsonian Institution if more samples are needed. All laboratory facilities necessary to this project are present at the Smithsonian Institution (new Bio-Rad Excalibur FTIR for the water content analysis, and a new JEOL 8900 electron microprobe for thorough mineral analysis). Minerals are hand-picked and chosen for being free of any defect, impurities or inclusions through various optical methods ¹³. The Beer-Lambert Law relates the IR spectra with the water content, but is only valid directly for measurements isotropic materials. Most crystals being anisotropic, various methodological constraints are necessary on preparing the sample: (1) the sample needs to be properly oriented, with the principal optical directions determined. Extremely thin, oriented single-crystal samples are prepared as double-sided slabs of down to a few mm ¹⁴. Proper orientation of the crystal can be achieved with an X-ray precession camera ⁵ or using a petrographic microscope ¹³. (2) A polarizer is necessary for water measurements in all non-isotropic minerals, otherwise sample thickness and water concentration cannot be scaled with the absorbance ¹⁵. (3) Integrated absorbance values, measured in all three orthogonal polarization directions of the crystal and summed, are used because they correlate with water content ¹⁶. Concerning mantle minerals, water absorption coefficients are relatively well determined for clinopyroxenes, orthopyroxene and garnet ¹³, but olivine has not be so far specifically calibrated ¹, although estimation of its water content can still be done using other calibrations ^{16, 17}. Further investigation of water storage in the mantle can also include the relation of water content in nominally anhydrous minerals with cation sites ⁵ and structure defects in minerals ¹⁸