Fluid escape from subduction zones controlled by channel-forming reactive porosity

Abstract

Earth would be uninhabitable if water was not returned to exogenous reservoirsat subduction zones, preventing global ocean drainage. Yet the bottleneck mechanism that couples initial fluid release from subducting, zero-porosity rocks with chemically bound water to rocks with high-permeability fluid escape channels is unknown. Using multiscale rock analysis combined with thermodynamic modelling we show that fluid flow initiation in dehydrating serpentinites is controlled by intrinsic chemical heterogeneities, localizing dehydration reactions at specific microsites. Porosity generation is directly linked to the dehydration reactions and resultant fluid-pressure variations force the reactive fluid release to organize into vein networks across a wide range of spatial scales (µm to m). This fluid channelization results in large-scale fluid escape with sufficient fluxes to drain subducting plates. Moreover, our findings suggest that antigorite dehydration reactions do not cause instantaneous rock embrittlement, often presumed as the trigger of intermediatedepth subduction zone seismicity.

Watching Dehydration: Seismic Indication for Transient Fluid Pathways in the Oceanic Mantle of the Subducting Nazca Slab

“…It forms at low temperatures when the mineral assemblage of olivine and pyroxene (i.e., a dry peridotite) reacts with water. Even though the discussion regarding the nucleation mechanism of LSZ earthquakes remains controversial, the observations of the dehydration structures of serpentinite are consistent throughout literature: In serpentinized peridotites from paleosubduction zones they have the form of vein-like porosity structures that are filled with nanocrystalline reaction products and exhibit small amounts of porosity (Healy et al, 2009;Plümper et al, 2017). The details of the relative contribution of emerging void pore space, buildup of pore fluid pressure, reduction of grain size, and stressing of load-bearing relict minerals have been investigated using data of outcrop studies, laboratory experiments, or numerical simulations.…”

“…The details of the relative contribution of emerging void pore space, buildup of pore fluid pressure, reduction of grain size, and stressing of load-bearing relict minerals have been investigated using data of outcrop studies, laboratory experiments, or numerical simulations. Combined thermodynamic and hydraulic modeling indicates that reactive fluid release during deserpentinization indeed forms interconnected pore networks in a self-organized manner (Plümper et al, 2017), and fluid flow experiments on pressurized serpentinites suggest that they maintain a certain permeability even under the confining pressure present within the LSZ (Katayama et al, 2012). Similar structures have been observed in laboratory experiments that investigated the breakdown of antigorite (e.g., Chollet et al, 2011;Jung et al, 2004;Proctor & Hirth, 2015).…”

“…Under only partially drained conditions ( < 0.9) they close. These temporal permeability fluctuations would more likely allow for the development of short-living transient dehydration events (Connolly, 2010;John et al, 2012;Plümper et al, 2017;Taetz et al, 2018). This relation has implicitly been used to directly map the V P ∕V S ratio to , without the assertion pores being compliant (e.g., Audet et al, 2009;Moreno et al, 2014).…”

Devolatilization of Subducting Slabs, Part II: Volatile Fluxes and Storage

“…In this paper, we provide a two‐dimensional model of reactive fluid flow in subducting slabs. Since the dynamics of within‐slab flow remains highly uncertain (Faccenda et al, ; Morishige and van Keken, ; Plümper et al, ; Wilson et al, ), we prescribe the flow direction in our model and investigate model behavior as a function of this parameter, rather than solving equations for momentum conservation. The model incorporates open‐system equilibrium thermodynamics and enables us to assess the factors controlling slab dehydration and decarbonation.…”

Water Migration in the Subduction Mantle Wedge: A Two‐Phase Flow Approach

“…Moreover, variations in fluid content (e.g., water or melts) that may affect viscous flow in the mantle (Hirth & Kohlstedt, ; Karato, ), and cause phase changes such as serpentinization and eclogitization (Hacker et al, ; Poli & Schmidt, ) were not explicitly included. We furthermore assumed no chemical transport of solutes with the fluid phase, thus ignoring the effects of reactive transport—a process that is likely important for channelized reactive flows in nature (Arkwright et al, ; Plümper et al, ) . The effect of fluid pressure on frictional yield, which may weaken for instance the subduction interface and lead to dehydration embrittlement (Jung et al, ; Podladchikov & Miller, ; Skarbek & Rempel, ), is also missing from our models.…”