From samples of ancient rocks on Kyushu, Japan, Tulley et al. (Sci. Advances, 2020) concluded that the subduction interface can flow viscously at low stress when temperatures exceed 300˚C. Furthermore, Tulley et al. (G-Cubed, 2022) found that transient, frictional slip occurred at temperatures where specific mineral dehydration reactions took place. Ujiie et al. (Geophys. Res. Lett., 2018) show that hat this transient slip can repeat at time intervals consistent with slow earthquakes. As such, this location is consistent with slow earthquakes being a consequence of local, fluid-driven weakening within a dominantly weak and viscous shear zone. Similarly, in the Kuiseb Schist, Namibia, we show that vein-filled fractures and viscous shear occurred at the same time, consistent with dehydration-triggered weakening by fluids at around 500˚C (Fagereng et al., Geology, 2018).
The active northern Hikurangi subduction margin, New Zealand, hosts some of the shallowest slow earthquakes recorded, and was target of scientific ocean drilling in International Ocean Discovery Program Expeditions 372 and 375. We found both brittle and ductile structures within a tens of metres thick fault zone, only ~ 300 m below the seafloor (Fagereng et al., Geology, 2019). We also studied the incoming seafloor sediments, and found that already before they are subducted, a mixture of brittle and ductile features have developed (Leah et al., Tectonics, 2020). The IODP expedition science team concluded that slow earthquakes in this location are an effect of geometrical and lithological heterogeneity (Barnes et al., Sci. Advances, 2020). We tested this idea using broadband seismic data, and concluded that roughness on the subducting plate is a key feature of this margin at least down to slow earthquake depths (Leah et al., Geology, 2022).
At Llanddwyn Island, Anglesey, UK, ancient oceanic crust records late Precambrian subduction. The shear zones here are dominated by basalt and volcanic sediments, and contain brittle and ductile structures developed simultaneously. In some carbonate rocks, crystal plastic flow was active despite relatively low temperatures (< 300˚C, Leah and Fagereng, Geophys. Res. Lett., 2022). In the other rocks, however, deformation was accommodated by competing frictional sliding and dissolution, transport, and precipitation of minerals - as in Japan, fluids promoted transient frictional sliding (Leah et al., J. Geophys. Res., 2022).
We show from the Kuckaus Mylonite, Namibia, that continental strike-slip faults can be very weak because of fine-grained mineral growth (Stenvall et al., Geophys. Res. Lett., 2019). This growth requires addition of small amounts of fluids from an external source (Stenvall et al., 2020; Geofluids). Oceanic transforms are thought to be weak based on their dominantly aseismic behaviour, and Cox et al. (J. Geophys. Res., 2021) show from the Troodos Ophiolite that this weakness may arise, at mantle depths, from alteration to serpentine. Serpentine may also host weak shear zones in the subduction zone mantle wedge (Tulley et al. Geophys. Res. Lett., 2022). At crustal depths, weakening occurs by alteration to frictionally weak and stably sliding chlorite, surrounding remnant, and potentially seismogenic, oceanic crust (Cox et al., Geophys. Res. Lett., 2021).
We generalise the slow earthquake source as a weak shear zone with relatively rigid inclusions. Interaction between these two components can generate force chains, where loss of continuous pathways of weak material leads to increases in stress. Beall et al. (Geophys. Res. Lett., 2019) show numerically that shear zones with more than 50 % strong material spontaneously generate force chains. Beall et al. (G-Cubed, 2019) suggest that if frictional yield is reached, then slip may accelerate within the rigid inclusions; depending on the length scale of the slipping zone, and the viscosity of the matrix, this could generate slow earthquakes. Fagereng and Beall (Phil. Trans. Royal Soc, 2021) make a further general case based on driving stress relative to frictional strength, and Lu et al. (in revision) use a micromechanical model to quantify when a shear zone scale, transient change from steady sliding to potentially fast, frictional slip can occur. Beall et al., (Geophys. Res. Lett., 2022) further apply a seismic cycle simulator to suggest that stress heterogeneity in a frictional-viscous shear zone may also control the size-distributions of earthquake magnitudes.
