Continuing at the third session, we showed how the spreading at 660 km can be explained by the presence of a crystalline phase transition in the main upper mantle mineral, olivine, at this depth. This phase transition, with its negative Clapeyron slope, slows plate penetration into the deeper mantle. However, to date, there is no clear explanation for the spreading of these plates to depths of 1, ,000 km.
Over the past fifteen years, advances in image resolution have made it possible to quantitatively combine tomographic images of tectonic plate pieces identified as such in the deep mantle (by their compressional and shear moduli indicating below-average temperature) with geological reconstructions of plate positions. The aim was to extend the history of subduction into the past, showing in particular that the plates currently arriving at the bottom of the mantle date from around 300 Ma, which provides information on the average speed of material flow in the deep mantle.
The infinite-frequency approximation underlying grape theory is a good approximation for wave travel in the cold parts of the mantle (where local propagation velocities are faster than in neighbouring regions). However, this approximation, which relies on measuring the arrival time at the very beginning of the wave, prevents us from " see " warmer regions. In this third session, we thus introduced theoretical advances in the interpretation of teleseismic wave travel times, with the introduction of sensitivity kernels " of finite frequency ", which have made it possible to obtain the first reliable images of hot mantle plumes, which rise from the depths of the mantle and whose surface expression is a " hot spot " volcano, like that of Hawaii, located in the middle of a tectonic plate. These images, based only on a small number of seismic wave types (mainly " P-waves "), do not, however, provide good resolution of these plumes on their own, due to the lack of illumination of the deep structure beneath the oceans.