Seismic observations of an ultra-low velocity zone at the Core-Mantle Boundary (2900 km depth) suggest the presence of a dense silicate melt layer that could potentially be a remnant basal magma ocean (Labrosse et al, 2007). While this hypothesis is supported by geodynamical modelling and geochemical analyses, it remains problematic to explain the stability of silicate melts at such conditions. The preferred Fe partitioning in the liquid phase above ~60 GPa has been proposed as one explanation because it could induce a positive density contrast with crystalline mantle. However, data on the effect of Fe on silicate melt density are sparse due to the experimental difficulties.

To study the effect of Fe on the structure and density of silicates melts at lower mantle conditions, we conducted in-situ synchrotron XRD experiments on glasses doped with different Fe contents ((Mg1-x, Fex)SiO3, with x being 0, 0.2, 0.4) up to 135 GPa at ID27 of the ESRF. Data were acquired up to a very high Q range (170 nm-1) using an incoming X-ray beam energy of 55 keV. An optimized multichannel collimator system has been employed to further improve the signal to noise ratio (Mezouar et al, 2002). Measurements using either a monochromatic beam or a pink beam were conducted to identify the best measurement strategy for these samples.

Structural evolutions were followed qualitatively using the pair distribution functions, while density was extracted using a minimization of the figure of merit (Morard et al, 2013), through optimization of parameters as rmin and Qmax values. The present data allowed identifying the onset of the 4- to 6-fold coordination change of the silicate polyhedra at 35 GPa. The extracted density data suggest that the content of Fe does not have an effect on the atomic density, which is also consistent with recent molecular dynamic simulations.