Subducted slabs are the driving force for mantle convection and plate tectonics. Lower mantle slab dynamics are critical to study how materials are transported and recycled between the surface and the CMB. However, in seismic tomography models, slabs in the Earth’s lower mantle are often under-resolved, hindering the understanding of how they deform and what they look like in the lower mantle. Seismic anisotropy is informative in the mantle flow patterns and the deformation in the region. The lowermost few hundreds of kilometers of the Earth’s mantle exhibit strong seismic anisotropy, which is hypothesized to be caused by lattice-preferred orientation, induced by the deformation of slabs impinging on the CMB. In this project, I investigated if slab trench length and the viscosity contrast between slabs and the background mantle (slab strength) can be determined by lowermost mantle flow patterns. For this purpose, I conducted 3D spherical convection calculations using CitcomCU and modified surface boundary conditions to induce subducting slabs. I found that weaker slabs and slabs with shorter trench lengths were deformed under pure shear, inducing azimuthal-divergent lowermost mantle flow patterns. On the contrary, stronger slabs and slabs with longer trench lengths underwent buckling or rigid bending, inducing bilateral lowermost mantle flow patterns. When combined with the lowermost mantle seismic anisotropy observations, this work is valuable in examining the dynamical behavior and morphology of slabs in the lower mantle. Moreover, it provides clues about the properties of slabs that reach the CMB.
Surrounded by subducted slabs, the two LLSVPs in the lowermost mantle beneath the Pacific and Africa have lower-than-average seismic wave velocities. One hypothesis for the origin of the LLSVPs is that they are caused by thermochemical piles of more primordial material from Earth’s early differentiation. Small patches of ULVZs are observed at the margin of and within the LLSVPs on the CMB. ULVZs are hypothesized to be caused by partial melting, ultra-dense materials distinct from the LLSVP compositions, or a combination of both. If LLSVPs and ULVZs are hypothesized to be compositionally distinct from the background mantle, they not only have a different composition but should also have a different mineralogical grain size. Both consequences affect the intrinsic viscosity contrast between LLSVPs and the background mantle, leading to different rheological formulations for piles and the ambient mantle. In particular, the LLSVPs may have an intrinsically higher viscosity than the ambient mantle. Because it is currently impossible to directly constrain the grain size/rheologies of the lower mantle from real Earth samples, it is important to study how the increased intrinsic viscosity of thermochemical piles would affect the dynamical behavior of piles (LLSVPs) and ultra-dense patches (ULVZs). However, composite-dependent viscosity is underestimated and usually not included in conventional thermochemical convection models.
One thing the increased pile intrinsic viscosity might affect is the shape of ULVZs. The shape of ULVZs is controlled by their properties (e.g., density, viscosity) and their interaction with the LLSVPs and the surrounding mantle. Therefore, ULVZ shapes provide information about the lowermost mantle flow and rheological properties of the LLSVPs. In this project, I explored if the symmetry of ULVZ cross-sectional shapes implies the viscosity of LLSVPs with respect to the ambient mantle. I found that if piles had the same intrinsic viscosity as the ambient mantle, ULVZ accumulations at pile edges were thicker on the outboard side of piles, leading to an asymmetric ULVZ shape. However, if compositional reservoirs had higher intrinsic viscosity than the surrounding mantle, ULVZs at pile edges typically became more symmetric or exhibited the opposite asymmetry, being thinner on the outboard side of piles. In addition, I found that ULVZs within the pile interior are consistently symmetrical. The finding is important, as it argues that even if ULVZs originate from ultra-dense materials, they can accumulate into symmetrical patches at LLSVP edges, just as if they are caused by partial melting. Furthermore, high-resolution seismic images of ULVZs may be available in the future so that the understanding of the viscosity, nature, and origin of LLSVPs can be advanced.
The increased pile intrinsic viscosity might also influence the stability of the LLSVPs. Such stability is usually discussed from two aspects: lateral mobility and morphological stability of the LLSVPs. Some paleomagnetic studies hypothesized that LLSVPs could be fixed at their current position for up to a few hundred million years. On the contrary, a recent study indicated that the paleomagnetic data could non-uniquely be explained by mobile LLSVPs. In geodynamic models, LLSVPs are envisioned as passive structures swept around by subducted slabs. However, there are other parameters such as compositional viscosity contrast between piles and the ambient mantle, that have not been sufficiently explored in these models. Therefore, it’s necessary and significant to explore if composition-dependent rheology could cause the stability of piles against changing mantle flow patterns. In this project, I investigated how dense thermochemical piles with increased pile intrinsic viscosity respond to changing convective flow patterns, in terms of lateral mobility and morphological stability. I found that the increased intrinsic viscosity of thermochemical piles doesn’t cause them to be more resistant to changing downwelling patterns. Additionally, piles with higher intrinsic viscosity are more resistant to changes in their cross-sectional morphology in response to changing upwelling flows in the background mantle. This study shows that if LLSVPs are indeed fixed, another mechanism must be found to explain the lateral immobility of piles.