The deepest phase transition in the Earth's mantle

The D"-layer is a 100-300 km thick layer lying just above the core-mantle boundary at 2900 km below the surface of the earth, under a pressure of ~100 to 135 GPa. In this region, seismic observations provide evidence of heterogeneities in wave propagation and sound velocities. Since the perovskite (Pv) to post-perovskite (PPv) phase transition was discovered for MgSiO₃ [1], various authors attempted to explain all peculiar seismic features on the basis of this transformation. Indeed, the local structure is quite different in these two phases; the PPv is layered (Figure 1), which enhances significantly its elastic anisotropy, and could help to explain the seismic heterogeneities [2,3]. However, these heterogeneities are observed within the D"-layer and at some restricted geographic positions only. This contrasts with the sharp seismic discontinuities observed all over the globe at different mantle depths, such as the transition-zone extending from depths of 410 to 660 km. Thus, based on MgSiO₃ studies, the post-perovskite transformation is unlikely to be the only source of the seismic anomalies in the D" region and other hypotheses remain plausible, such as the presence of a primitive and chemically different mantle reservoir, the presence of a cemetery of oceanic plates subducted by mantle convection, or the occurrence of partial melting at the core-mantle boundary (CMB).

Figure 1. The post-perovskite structure, analagous to CaIrO3, shows an anisotropic compression behaviour with b axis more compressible than the others [1, 2].

Most of the previous experimental investigations of the Pv to PPv transition mechanism suffer from the limitation that the MgSiO3 composition is too simple. When more relevant mantle compositions were investigated, the chemical analyses were only performed ex situ on samples recovered at ambient conditions. In this work, we investigated in situ at high P and T the mineralogical content and the Fe distribution in various geological samples, using tandem analyses of X-ray absorption spectroscopy at the Fe K-edge (ID24) and X-ray diffraction (ID27). Thanks to these techniques, phase relations between the minerals could be determined for bulk samples, in opposition to the thin films that were analysed in previous studies. Pressures up to 180 GPa were obtained using diamond-anvil cells, exceeding the pressure of 135 GPa found at the CMB. YAG laser heating was used to reach temperatures of more than 3000 K.

Synthesis of the Pv and PPv grains was carefully monitored using X-ray diffraction, for variable heating duration and temperatures. The sharpness of the Bragg lines and occurrence of a multitude of small spots on the 2-dimensional X-ray pattern indicate efficient atomic diffusion at the scale of the grain size and quasi-equilibrium crystallisation of the silicate phases. The diffraction patterns reveal the presence of two crystalline phases, the Pv and PPv, up to the maximum pressure investigated. Relative proportions of the different minerals were quantified based on Rietveld refinements. We then performed Fe K-edge XANES on the same samples in order to retrieve the Fe coordination site and thus the preferential Fe insertion in Pv or PPv phases (Figure 2). Combined with PPv fractions determined using X-ray diffraction, we calculated the Fe-partition coefficient between the two phases.

Figure 2. X-ray absorption spectra recorded at the Fe K-edge as a function of the synthesis pressure (black lines). Experimental data are compared with calculations (colour lines). XANES spectrum of the enstatite starting material (lowest) and quenched amorphous (highest) are shown for comparison. The features B and C shift significantly with increasing pressure while A and D remain unchanged. This is associated to changes in the Fe local structure when the Fe environment changes progressively from Pv to PPv with increasing pressure from 106 to 184 GPa.

Our results show that the main PPv phase is largely depleted in Fe, compared to the Pv phase. Also, we report the proportion and composition of each mineral in the P-T conditions of the D"-region (Figure 3). These results have severe geophysical consequences. First, we show that Fe and Al produce a progressive phase transition from Pv to PPv, in contrast with pure MgSiO₃. This transition mechanism is incompatible with the presence of any sharp seismic reflectors in the D"-layer. More progressive seismic anomalies should be expected, which is compatible with the seismic observations in this mantle region. Then, the occurrence of a PPv-phase with a minimum Fe-content has two major implications: (i) it should enhance the radiative thermal conductivity in the D"-layer, since Fe-free minerals are generally transparent to visible light. This can induce a much higher temperature in the lowermost mantle that is in contact with the outer core; (ii) The FeO activity in the mantle is much enhanced at the CMB, as evidenced by a FeO content up to ~0.30 mole percent in ferropericlase (See Figure 3). Thus, FeO should diffuse to the outer core, with a concomitant loss of FeO in the lowermost mantle. This can affect the density of the material located in the D"-layer and can induce unstable dynamical behaviour in this region.

Figure 3. a) Fractions of lower mantle minerals and b) Fe concentration in minerals plotted as a function of pressure for the four phases expected in the lowermost mantle region. Pv, PPv, Fp and Ca-Pv stand for Al-bearing (Mg,Fe)SiO3 perovskite and post-perovskite phases, (Mg,Fe)O ferropericlase and CaSiO3 perovskite, respectively. The Earth's core-mantle boundary is located at 135 GPa.

References
[1] M. Murakami, K. Hirose, K. Kawamura, N. Sata, Y. Ohishi, Science 304, 855 (2004).
[2] N. Guignot, D. Andrault, G. Morard, M. Mezouar, Earth and Planetary Science Letters 256, 162 (2007).
[3] K. Hirose, Y. Fei, P. Ma, H.K. Mao, Nature 397, 53 (1999).
[4] Derived from L.H. Kellogg, B.H. Hager, R.D. van der Hilst, Science 283, 1881-1884 (1999).

Principal publication and authors
D. Andrault (a), M. Muñoz (b), N. Bolfan-Casanova (a), N. Guignot (c), J.P. Perrillat (d), G. Aquilanti (d), S. Pascarelli (d), Earth and Planetary Science Letters 293, 90-96 (2010).
(a) LMV, Université Blaise Pascal, Clermont-Ferrand (France)
(b) LGCA, Université Joseph Fourier, Grenoble (France)
(c) SOLEIL, Gif-sur-Yvette (France)
(d) ESRF

Top image: Experiments at extreme conditions provide insight into the composition of the D''-layer just above the core-mantle boundary at 2900 km below the surface of the Earth [4].