Until recently, most of the lower mantle was generally considered to be well-mixed with strong heterogeneity restricted to the lowermost several hundred kilometres above the core–mantle boundary, known as the $D$″ layer. However several recent studies have started to hint at a potential change in Earth's structure at mid-mantle depths beneath the transition zone.

Here we present a continental-wide search of Europe and the North Atlantic for mid-mantle P-to-s wave converted phases. Our data set consists of close to 50,000 high quality receiver functions. These are combined in slowness and depth stacks to identify seismic discontinuities in the range of 800–1400 km depth to determine at which depths and in which tectonic settings these features exist. Receiver functions are computed in different frequency bands to resolve the sharpness of the observed discontinuities. We find most seismic velocity jumps are observed between 975–1050 km depth, localised beneath western Europe and Iceland. The shear wave velocity jumps are roughly 1–2.5% velocity increase with depth occurring over less than 8 km in width. The most robust observations are coincident with areas of active upwelling (under Iceland) and an elongate lateral low velocity anomaly imaged in recent tomographic models which has been interpreted as diverted plume material at depth.

The lack of any suggested phase change in a normal pyrolitic mantle composition at around 1000 km depth indicates the presence of regional chemical heterogeneity within the mid-mantle, potentially caused by diverted plume material. We hypothesise that our observations represent either a phase change within chemically distinct plume material itself, or are caused by small scale chemical heterogeneities entrained within the upwelling plume, either in the form of recycled basaltic material or deep sourced chemically distinct material from LLSVPs.

Our observations, which cannot be directly linked to an area of either active or ancient subduction, along with observations in other hotspot regions, suggest that such mid-mantle seismic features are not unique to subduction zones despite the large number of observations that have previously been made in such settings.