Stable and Unstable Shear in Altered Downgoing Slabs: Predicted Strain Localization in Magnesian Carbonates and Wadati-Benioff Seismicity

Seismicity of subduction zones at upper-mantle depths is commonly explained by dehydration reactions of serpentine and hydrous silicates and reductions in effective pressure. However, the conditions of Wadati-Benioff zone seismicity do not strictly correspond to temperatures and depths of serpentine dehydration, and there is no independent evidence that seawater penetrates the lithosphere to form serpentine at depths >30km below the seafloor. Altered lithosphere may contain magnesian carbonates in addition to hydrous silicates, both at the top of plates, where CO₂ of seawater reacts with mantle rocks and at the base of plates where CO₂ is introduced by mantle plumes.

Adapting the thermal softening model of Kelemen and Hirth (2007), we model the strain localization and shear heating within magnesite horizons embedded within an olivine host using flow laws determined experimentally for dislocation creep and diffusion creep of the carbonate layer and olivine host (Hirth and Kohlstedt, 2003; Holyoke et al., 2014). Strain rates predicted within carbonate-rich layers of downgoing slabs are much higher than those of the surrounding olivine at all conditions. However, shearing may be either stable or unstable depending on the relative rates of shear heating and conductive heat loss from the shear zone. Localized strain rates reach a steady state when shear heating and heat flow are balanced, while unstable strain rates are calculated where shear heating exceeds heat flow. Modeled strain rates accelerate to 10⁺¹ s^-1, as temperatures reach melting conditions, and stresses drop, corresponding to a seismic event. Applications of this model to the double Benioff zones of the NE Japan trench predict unstable seismic shear for both upper and lower seismic zones to subduction depths of ~300 km. For cold downgoing slabs, such as the Tonga subduction system, unstable seismic shear is predicted for carbonate horizons of altered downgoing slabs to depths exceeding 400 km.