Investigating the Role of Fluid–Solid Coupling on Subduction Dynamics and Fluid Pathways

Constraining the complex nonlinear feedbacks between patterns of fluid transport and solid deformation in subduction systems remains a key area of research towards understanding subduction zone seismicity, magmatism, and volatile cycling. In this study, we use 2D geodynamic simulations to constrain how distinct physical approximations for reactive volatile transport and fluid-solid coupling affect both long-term subduction dynamics and fluid transport pathways. 

The simulations use the open-source geodynamic software package ASPECT, which provides a framework for modeling coupled nonlinear viscoplastic deformation and reactive fluid transport in combination with a free surface, adaptive mesh refinement, advanced nonlinear solvers, and massive parallel scaling. Fluid–rock interaction follows a previously published parameterization of volatile–rock interaction within subduction systems (Tian et al., 2019), which provides an analytical solution for water partitioning between bound and free water phases across pressure–temperature space for sediment, mid-ocean ridge basalt, gabbro, and peridotite lithologies. We simulate fluid transport as either partially coupled Darcy flow (ignoring compaction terms) or fully coupled two-phase flow following the McKenzie equations (including compaction terms) (McKenzie 1984). In both cases, fluid–solid coupling also occurs through exponential reduction of the solid viscosity as a function of the volume of free-water. Furthermore, we examine the additional fluid-solid coupling through a reduction in the brittle strength of the solid in the presence of free-water and of the solid viscosity as a function of the bound H2O content.

Consistent with previous work, our model results demonstrate that the choice of partially or fully coupled two-phase flow significantly impacts fluid pathways, and that increased fluid–solid coupling leads to increased convergence rates between the subducting and overriding plates. When ignoring compaction terms, the partially coupled Darcy models promote vertical fluid pathways as the slab dehydrates, while including compaction prevents immediate release of the fluid from the subducting plate, promoting updip fluid pathways within the slab before fluids are released into the mantle wedge. Significantly, fluid release into the mantle wedge in the deeper and mechanically strong portions of the slab does not occur until a sufficiently high porosity is reached to locally reduce the solid viscosity and thereby enable the compaction pressure to overcome compaction viscosities. 

Extensive serpentinization of the subducting mantle lithosphere enables the transport of large fluid volumes to beyond the arc. When including the full degree of fluid–solid coupling (including additional brittle and ductile weakening), this large volume of fluid carried to the back-arc promotes sufficient weakening of the overriding plate to drive the dynamic initiation of back-arc spreading. In contrast, reduced degrees of serpentinization inhibit back-arc rifting. We propose that variations in mantle lithosphere hydration provide a fundamental control on the occurrence of back-arc spreading, with less hydrated subducting plates corresponding to subduction zones lacking back-arc extension.