In 1996 and 1997, when seismictomography began producing much improved images of Earth's mantle, many researchers thought they were witnessing the resolution of the debate over whether mantleconvection takes place across the entire mantle, or rather within—but not across—chemically distinct layers. The images revealed convincing evidence of slabs of subducted oceanic lithosphere penetrating through the boundary between the upper and lower mantle at a depth of 670 km. If slabs of oceanic crust, which formed, in part, from the upper mantle, could penetrate so easily into the lower mantle, it was difficult to see how the two regions could differ dramatically in composition. (See PHYSICS Today, August 1997, page 17.) High‐pressure mineral physicists had already provided a suitable explanation for the discontinuity in seismic‐wave speeds at the boundary: The boundary corresponded well to the pressure where the dominant phase in the mantle changes from spinel to perovskite—an isochemical, pressure‐induced phase transformation.
As geochemists, modelers, and seismologists try to make sense of data from Earth's mantle, a new model poses challenges to each group and suggests that progress in understanding the deepest regions of the mantle can occur only on a broad front.