The mineral olivine dominates the composition of the Earth's upper mantle and hence controls its mechanical behaviour and seismic anisotropy. Experiments at high temperature and moderate pressure, and extensive data on naturally deformed mantle rocks, have led to the conclusion that olivine at upper-mantle conditions deforms essentially by dislocation creep with dominant [100] slip. The resulting crystal preferred orientation has been used extensively to explain the strong seismic anisotropy observed down to 250 km depth. The rapid decrease of anisotropy below this depth has been interpreted as marking the transition from dislocation to diffusion creep in the upper mantle. But new high-pressure experiments suggest that dislocation creep also dominates in the lower part of the upper mantle, but with a different slip direction. Here we show that this high-pressure dislocation creep produces crystal preferred orientations resulting in extremely low seismic anisotropy, consistent with seismological observations below 250 km depth. These results raise new questions about the mechanical state of the lower part of the upper mantle and its coupling with layers both above and below.

No abstract

[1] We use forward models based on recent high-pressure experimental data on mantle minerals to predict the seismic anisotropy produced by plastic strain of orthorhombic wadsleyite, the dominant mineral in the upper transition zone. These models predict a weak seismic anisotropy for a polycrystal of pyrolitic composition (60% wadsleyite, 40% garnet) at transition zone conditions: $2% for P and $1% for S waves for a shear strain of 1. Both P and S wave anisotropy patterns show an orthorhombic symmetry. P waves propagate faster at low angle to the shear direction and slower at high angle to the shear plane. S wave anisotropy is characterized by faster propagation of waves polarized at low angle to the shear direction. Horizontal shearing results therefore in higher velocities for horizontally propagating P waves (PH ) and horizontally polarized S waves (SH ), as well as in weak azimuthal variation of SV and SH velocities. On the other hand, vertical flow leads to higher velocities for vertically propagating P waves (PV ) and vertically polarized S waves (SV) and to a weak azimuthal variation of SV velocity but to a roughly constant SH velocity. Analysis of global observations of seismic anisotropy in the transition zone in the light of these models supports dominant horizontal flow in the uppermost transition zone, in agreement with predictions of geodynamical models that explicitly introduce phase transitions.