Temperature Control on the Source of Slab Fluids: Constraints from Nd and Hf

The composition of fluids evolved from dehydration reactions during subduction depends critically on the temperature structure and evolution of the slab lithosphere. For example, fluids generated below the H2O-saturated solidus are solute-poor aqueous fluids (H2O > 90 wt%), while those generated above are hydrous melts (H2O generally < 30 wt%). Continuous variation between these extremes may exist at pressures > ~ 4 GPa in super-critical fluids, but compositions are still a strong function of temperature. Some element tracers are also predicted to vary strongly as a function of fluid temperature, such as the REE elements, with concentrations strongly controlled by the temperature-dependent solubility of accessory minerals allanite and monazite. Here we use this prediction, along with the isotopes of Nd, to constrain slab temperatures beneath different volcanic arcs. Slabs with a greater thermal parameter (the product of the age and descent rate) will be colder at any given depth than slabs with a low value for the thermal parameter. Hotter slabs will not only have a higher slab surface temperature (in the subducting sedimentary layer), but also a higher interior temperature (in the subducting oceanic crust). Thus, cold slabs may generate fluids that carry Nd only from the sedimentary layer, whereas hotter slab will generate fluids that mobilize Nd from the oceanic crust as well. Because these layers are isotopically distinct, we can use Nd isotopes to test this prediction against estimates of slab fluid temperatures. We take advantage of new data sets on the Nd and Hf isotopic composition of marine sediments [1-4] to evaluate isotopic mixing and elemental fractionation together by examining the systematics of arc volcanics and subducting sediments on Hf/Nd vs. eNd diagrams. Mixing on this diagram is linear and therefore unambiguous to interpret. Many arcs form linear arrays that are consistent with mixing between depleted mantle (high Hf/Nd and eNd) and a low Hf/Nd and low eNd end-member, which surprisingly, in every case fails to intersect bulk sediment. Highly fractionated sediment compositions (i.e., even taking Hf/Nd to zero) also fail as end-members. Instead, varying amounts of MORB Nd and Hf (with low Hf/Nd) are required from the subducting plate. The proportion of Nd that is slab MORB vs. sediment increases from 0-20% MORB Nd (i.e., 100-80% sediment Nd) for central Tongan volcanoes and Seguam (Aleutians) to 79-90% for Shishaldin (Aleutians) and Adams (Cascades). These quantities correlate with recent estimates of slab surface temperatures derived from the H2O/Ce thermometer (from ~750°C at 4 GPa for Tonga to ~900°C for the Cascades [5]), and consistent with numerical models of slab thermal structure [6]. Thus, Nd and Hf isotopes and ratios may be useful proxies for slab fluid compositions, with a greater proportion of MORB being more consistent with high temperature melts (< 30% H2O) as the main agent of mantle wedge hydration beneath arcs. [1] Chauvel et al. (2008) G3. [2] Carpentier et al. (2009) EPSL. [3] Vervoort et al. (2011) GCA. [4] Plank (2012) Treastise on Geochem. [5] Cooper et al. (2012) G3. [6] Syracuse et al. (201) PEPI.