H₂O influences a number of processes that occur within the Earth's upper mantle. For example, H₂O strongly influences anatexis, such that the compositions and amounts of mantle derived melts are, in part, controlled by the availability of H₂O. H₂O also exerts a strong influence on deformation, as the addition of H₂O to mantle rocks produces a large weakening effect. For example, a significant difference exists between the viscosity of "wet" and "dry" olivine. This difference in viscosity may be of fundamental importance because the difference between the "strength" of the lithosphere and asthenosphere may be due to differences in H₂O content . As a first step toward understanding processes in the upper mantle it is important to know if the mantle is "wet" or "dry". However, a more complete understanding of upper mantle processes requires moving beyond a discussion of "wet" and "dry" end-members to a discussion of actual values of H₂O activity (aH₂O) or fugacity (fH₂O). If values of aH₂O are determined then it may be possible to determine how aH₂O affects mantle rheology and mineral or mineral-melt equilibria. Consequently, we plan to undertake research that will lead to quantification of aH₂O in the upper mantle.

We will use several different methods, where appropriate, to determine values of aH₂O on each suite of mantle rocks we examine. (1) Amphibole dehydration equilibria may be utilized to determine values of aH₂O. Use of any other H₂O -buffering equilibria, requires knowledge of pressure (P) and temperature (T) of mineral equilibration, as well as activities of end-member components in the mineral phases. Thus, chemical characterization of co-existing phases in mantle samples will be an important part of this study. (2) We will also use the concentration of H₂O in olivine, as determined via infrared spectroscopy, to determine aH₂O. This approach is possible because the relation between aH₂O in a fluid and the H₂O content of co-existing olivines has been determined experimentally . (3) Careful chemical characterization of natural amphiboles may be used to constrain values of fH₂ in the mantle via dehydrogenation/oxidation equilibria. If these values of fH₂ can be combined with values of oxygen fugacity (fO₂), values of aH₂O can be estimated. (4) The compositions of fluid in inclusions in mantle minerals may also provide some insight into values of aH₂O. Agreement between values of aH₂O, as determined using these methods on the same samples, would be strong evidence that our determinations of aH₂O are accurate. Disagreement may indicate re-setting has occurred in one or more of these systems.

Another goal of this study is to explore how variations in aH₂O may affect the rheology of mantle rocks. We have used currently available experimental data to derive a flow law for dynamic recrystallization of olivine that incorporates aH₂O. Using this flow law, we can explore how the values of aH₂O we determine for mantle samples may affect mantle rheology. In addition, many of the samples we intend to examine as part of this study exhibit evidence of deformation. Microstructures of these samples will be examined, petrographically and with electron-beam techniques, to determine the dominant deformation mechanisms and strengths that prevailed while these samples were undergoing deformation in the mantle. In this way we intend to examine the relation between deformation and values of aH₂O .

Below: Thin section photomicrograph from the Cassou peridotite body (French Pyrenees)

This photomicrograph (width corresponds to 870 microns) illustrates textures found in the Cassou peridotite body (French Pyrenees). finer grained recyrstallized zones are common (as shown here) and, in some cases, these zones apparently correspond to the development of hydrous phases (e.g., amphibole). Thus, fluids may have helped to promote recrystallization and the development of finer-grained material. However, amphibole equilibria indicate that the minerals in these rocks equilibrated at low values of aH₂O (<0.2). These results are from a recently completed (2001) M.S. thesis (by C. Petko).