USING MINERAL EQUILIBRIA TO CONSTRAIN THE NATURE OF MANTLE FLUIDS

H₂O plays an important role in many mantle processes (e.g., melting and deformation) and, therefore, it is important to determine values of the activity of H₂O (aH₂O) in samples from the earth’s mantle. The H₂O content of the upper mantle has been estimated from the H₂O contents of nominally anhydrous mantle minerals (NAMMs). However, for many common NAMMs, the relation between H₂O activity and H₂O content is not well known, and these mantle minerals may be prone to H₂O loss during emplacement on Earth’s surface. Amphibole equilibria can also be applied to estimate values of aH₂O, and this may be a more robust method to determine values of aH₂O. We have, therefore, used mineral equilibria to estimate values of aH₂O, hydrogen fugacity (ƒH₂), and oxygen fugacity (ƒO₂) in mantle samples.

The chemical compositions of olivine + orthopyroxene + clinopyroxene + amphibole + spinel were characterized in eleven amphibole-bearing mantle xenoliths from the southwestern U.S.A. These mineral chemistries were used to estimate values of temperature (T), pressure (P), aH₂O, ƒO₂ and ƒH₂. Estimated P-T conditions for the 11 samples range from 1.1 to 1.6 GPa, and 900 to 1020˚C, respectively. Application of pargasite dehydration equilibria yields values of aH₂O ranging from 0.03 to 0.18. The compositions of coexisting spinel + olivine + orthopyroxene yield DlogƒO₂(FMQ) of -1 to +0.7. For nine of the samples, values of ƒH₂ were estimated using amphibole dehydrogenation equilibria, which required determining the ratios of Fe²⁺ to Fe³⁺ in the amphiboles. Values of ƒH₂ ranged from 6 to 117 bars, and these values were combined with values of ƒO₂ to estimate aH₂O using the relation 2H₂O = 2H₂ + O₂. Values of aH₂O using this method range from 0.02 to 0.12 for these nine samples.

A potential problem with estimating aH₂O using amphibole equilibria is possible retrograde H-loss from amphibole. However, values of aH₂O estimated from dehydration equilibria are less sensitive to this H-loss as compared to values of aH₂O estimated from the combination of ƒH₂ and ƒO₂. Therefore, equilibria based on amphiboles that suffered H-loss should yield different values of aH₂O when these two methods are applied to the same sample. This difference in values of aH₂O is < 0.05 for all samples, suggesting that the amphiboles have experienced little or no H-loss.