- Home
- Users & Science
- Scientific Documentation
- ESRF Highlights
- ESRF Highlights 2003
- High Resolution and Resonance Scattering
- Iron Partitioning in the Earth's Mantle: Layering and Deep Lower-Mantle Heterogeneities
Iron Partitioning in the Earth's Mantle: Layering and Deep Lower-Mantle Heterogeneities
Recent seismic observations [1] suggest that compositionally distinct domains exist in the Earth's lower mantle, with a boundary located between 1700 and 2300-km depths. In order to interpret these observations, the chemical and physical properties of the dominant phases in the lower mantle, namely (Mg,Fe)SiO3 magnesium silicate perovskite and (Mg,Fe)O ferropericlase have to be determined at the pressure and temperature conditions of the deep mantle. Due to the thermodynamic stability of these phases at the pressure and temperature conditions of the lower mantle, one cannot explain such geophysical observations on the basis of a chemical breakdown or a structural phase transition, as in the case of other seismic discontinuities. On the other hand, more subtle effects, driven by the chemistry of iron in the lower mantle, can affect the iron content in perovskite and ferropericlase, i.e. the partition coefficient of iron between the two compounds.
We measured the spin state of iron in ferropericlase (Mg0.83 Fe0.17)O at high pressure and found a high-spin to low-spin transition occurring in the 60 to 70 gigapascal pressure range (Figure 1), corresponding to depths of 2000 kilometers in Earth's lower mantle. The measurements were performed on beamline ID16 at the ESRF. It has been predicted [2] that such a transition would increase the partition coefficient of iron between ferropericlase and perovskite by several orders of magnitude, almost entirely depleting perovskite from its iron. Since the transition pressure increases with temperature and with iron content, the transition will be gradual and spread over a large range of depths rather than a sharp discontinuity, consistent with seismic observation of heterogeneities [1]. The transition could also be accompanied by phase separation between iron-rich high-spin and magnesium-rich low-spin ferropericlases.
Fig. 1: X-ray emission spectra collected on ferropericlase (Mg0.83 Fe0.17)O at different pressures. The presence of a satellite structure (Kß' line) on the low energy side of the iron main emission line (Kß1,3 line) is characteristic of a high spin 3d magnetic moment. This structure collapses at high pressure upon compression, and then reforms upon pressure decrease (spectra annotated with letter "R" in the legend). |
Perovskite is the major lower-mantle phase, and iron-free perovskite is much more viscous that iron-rich perovskite, implying the transition could have a fairly strong rheological signature, and could affect the geodynamics in the lowermost mantle. Investigation of that matter could contribute to constrain geodynamical interpretations of the seismic observations, and quantify the effect of such a viscous layer on the dynamics of plumes. It should also be noted that such a layering model requires no isolated convection cells, as the chemistry of the two layers is reversible as a function of depth (the transition is reversible upon decompression); uplifted materials will recover the partitioning properties of the top layer.
In conclusion, we provide a mineral-physics basis for lower-mantle layering and chemical heterogeneities. The upper layer would consist of a phase mixture with about equal partitioning of iron between magnesium silicate perovskite and ferropericlase, whereas the lower layer would consist of almost iron-free perovskite and iron-rich ferropericlase. This stratification is likely to have profound implications for the transport properties of the Earth's lowermost mantle.
References
[1] R. van der Hilst and S. Kárason, Science 283, 1885 (1999).
[2] E. Gaffney and D.L. Anderson, J.Geophys. Res. 78, 7005 (1973).
Principal Publication and Authors
J. Badro (a), G. Fiquet (a), F. Guyot (a), J.-P. Rueff (b), V.V. Struzhkin (c), G. Vankó (d), and G. Monaco (d), Science 300, 789 (2003).
(a) LMCP, Université Paris VI, Inst. de Phys. du Globe de Paris (France)
(b) LMR, Université Paris VI (France)
(c) Geophysical Labo., Carnegie Institution of Washington (USA)
(d) ESRF