摘要：Physical mechanisms of seismic anisotropy in the D” layer are examined based on seismological and mineral physics observations. The results of body-wave seismology on the fine structure of the D” layer and of mineral physics studies on the elastic constants and the lattice preferred orientation in lower mantle minerals as well as the shape preferred orientation of melt pockets are taken into account. Evidence of large but depth (pressure)-dependent elastic anisotropy of lower mantle minerals, particularly (Mg,Fe)O, and of tilted shape preferred orientation of sheared partial melts is summarized. It is shown that both shape preferred orientation of partial melts (or iron-rich secondary phases) and lattice preferred orientation of minerals with well-documented slip systems are difficult to reconcile with seismological observations. However, lattice preferred orientation of highly anisotropic mineral, (Mg,Fe)O, is consistent with most of the seismic observations if the dominant glide plane under the D” layer conditions is 100 rather than 110 as observed at lower pressures. Such a change in glide plane in MgO (or (Mg,Fe)O) is likely to occur as a result of pressure-induced change in elastic anisotropy and/or in the nature of chemical bonding (and possibly due to high temperatures). Both solid-state and partial melt mechanisms of anisotropy imply that the V SH > V SV ( V SV > V SH ) polarization anisotropy means horizontal (vertical) flow. In the solid-state mechanism, significant V SH > V SV in the D” layer beneath the circum-Pacific (Alaska and the Caribbean) implies horizontal shear at high stress caused presumably by the collision of subducting materials with the core-mantle boundary. Highly variable anisotropy beneath the central-Pacific can be attributed to solid-state fabrics caused by a complicated three-dimensional flow presumably related to the upwelling of plumes, but anisotropy in this region could also be attributed to the shape preferred orientation of melt pockets the presence of which is suggested by very low average velocities.
