Earth science

Deeper understanding

The boundary between the core and mantle is one of the most enigmatic regions of Earth's interior. Analyses of a newly discovered crystalline phase should yield a fuller understanding of this region.

The lowermost 250 km or so of Earth's mantle, known for historical reasons as D″, is comparatively small in volume but potentially holds the key to understanding a host of geophysical phenomena — among them the formation of plumes in the mantle, interactions between core and mantle, and the ultimate fate of subducting slabs of crust that are driven into the interior by tectonic forces. Investigations of this region largely depend on interpreting the behaviour of seismic waves, which have shown that it is highly complex. Until recently, however, studies of the region's mineral properties at high pressures and temperatures had been unable to provide satisfying explanations for much of this complexity. Part of the problem is that the extreme conditions in D″ — pressures up to 135 gigapascals and temperatures probably ranging between 2,000 K and 4,000 K — are difficult to reach in the laboratory. However, laboratory experiments and theory are finally coming together to bring this region into sharper focus.

Beginning on page 442 of this issue, papers by Iitaka et al.1 and Oganov and Ono2 provide insights that link the calculated physical properties of a newly discovered3 high-pressure crystal structure with seismic observations of the deep lower mantle. Earth's mantle is composed mostly of dense silicate minerals containing magnesium, iron, calcium and aluminium. Experiments have shown that the lower mantle, extending from 660 km depth to the base of the mantle at about 2,900 km (Fig. 1), is mainly composed of (Mg,Fe)SiO3 in a crystal structure known as perovskite. Although the properties of this material are compatible with most observations for the lower mantle, the abrupt change in properties near the mantle's base defied explanation in terms of perovskite behaviour.

Figure 1: The principal regions of Earth's interior.

The latest interpretations of the lowermost mantle, the D″ layer, are shown in Fig. 2.

Thus, Murakami and colleagues' experimental discovery3 of a ‘post-perovskite phase’ in MgSiO3, at conditions comparable to the D″ region, has stimulated considerable interest in the physical properties of the new phase. Given the difficulty of performing direct experiments under these conditions, first-principles quantum mechanical calculations of the type carried out by Iitaka et al.1 and Oganov and Ono2 are especially useful for studying the deep Earth. In contrast to the perovskite structure that is widely adopted by many compounds, the post-perovskite phase seems to be rather uncommon. In this structure, each silicon cation remains surrounded by six oxygen anions — producing the octahedral coordination that is characteristic of the lower mantle. But rather than forming a corner-linked, three-dimensional network as in perovskite, in the post-perovskite phase the silicon octahedra share edges and corners to form a sheet-like structure with alternating magnesium and silicon layers (see Fig. 1 of Iitaka and colleagues' paper on page 442).

This much could be demonstrated by laboratory experiment3. But theoretical calculations now shed light on several other properties of the post-perovskite phase. First, by taking account of thermodynamic considerations, Iitaka et al.1 and Oganov and Ono2 show that it is expected to be stable at pressures above about 100 GPa (at 0 K). Oganov and Ono2 also include experimental observations of the new phase after heating at near 118 GPa. For experiments under these extreme conditions, reports of structural changes are all too frequently not verified. So multiple experimental observations of the post-perovskite phase2,3,4, together with the theoretical predictions of its stability, mean that the implications of the new phase need to be seriously considered. Indeed, the two theoretical reports here1,2 are also generally compatible with the findings from another study5,6.

In conjunction with experimental findings, the theoretical results at 0 K indicate that the transition has a positive pressure– temperature slope. At mantle temperatures, the phase transition is then expected to occur about 200–300 km above the base of the mantle (Fig. 2), consistent with evidence for a sudden change, or discontinuity, in the velocity of seismic waves there7,8, possibly global in extent9,10. The positive slope of the phase boundary is even compatible with seismic-wave evidence10 that the D″ discontinuity is elevated in seismically fast (and presumably cold) regions and depressed in seismically slow (hot) areas. The new phase is also found to be about 1–2% denser than perovskite at D″ conditions.

Figure 2: New model for the mantle's base.

The D″ discontinuity is now thought to be due to a transition from the perovskite to a post-perovskite structure in (Mg,Fe)SiO3 about 200–300 km above the base of the mantle. The phase boundary is elevated in locally cooler regions (blue) and depressed in locally hotter regions (red). A tendency for the layers of the post-perovskite phase to align parallel to Earth's core can help to explain the faster propagation of horizontally polarized (vSH) than vertically polarized (vSV) shear waves. Ultralow-velocity zones are thin (5–40-km thick) regions, located directly above the core, where shear-wave velocities are strongly depressed.

Calculations of the elastic properties confirm that the post-perovskite phase is anisotropic, being more compressible normal to layering than parallel to it. It also exhibits considerable anisotropy in seismic-wave velocities, especially for the type of waves known as shear waves. Given the nature of the structure, it is plausible that post-perovskite crystals will develop a lattice-preferred orientation under compression such that the direction normal to layering will tend to lie along the vertical. If the layering is imperfect, and taking into account the presence of other phases, the calculations show that a 2–3% seismic discontinuity consistent with deep-mantle observations7,8,9,10 would result from this phase transition. The transition is expected to produce a larger discontinuity for shear waves than for compressional waves, the other main type of seismic wave.

This form of texturing will also result in horizontally polarized shear waves (of velocity vSH) propagating faster than vertically polarized shear waves (vSV). Seismic anisotropy in D″ is complex, but this sense of anisotropy has been well documented in certain regions11. Previously, this behaviour was difficult to reconcile with the known elastic properties and deformation behaviour of perovskite and other lower-mantle minerals. Instead, it was proposed that the anisotropy resulted from aligned inclusions or layering of minerals with dissimilar seismic velocities. The discovery of the post-perovskite phase may provide a simpler explanation.

The proposed transition between perovskite and post-perovskite will not resolve all questions about the D″ region. But it clearly provides a new framework for studying the region and is sure to stimulate further geophysical observations, laboratory experiments and computer calculations. From a mineral-physics viewpoint, studies of texture development in the new phase, as well as constraints on the behaviour of more chemically complex systems, are clearly needed. Also, the elastic anisotropy has only been calculated at 0 K, yet in some cases temperature can drastically change the magnitude and even orientation of anisotropy. The theoretical studies1,2,5,6 are in remarkably good agreement. But they all used similar techniques involving some degree of approximation, which will also necessitate further examination.

Nevertheless, a new era in the study of Earth's deepest mantle has begun. An explanation for both the D″ discontinuity and the onset of seismic anisotropy in the region may finally be within our grasp.


  1. 1

    Iitaka, T., Hirose, K., Kawamura, K. & Murakami, M. Nature 430, 442–445 (2004).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Oganov, A. R. & Ono, S. Nature 430, 445–448 (2004).

    ADS  CAS  Article  Google Scholar 

  3. 3

    Murakami, M., Hirose, K., Kawamura, K., Sata, N. & Ohishi, Y. Science 304, 855–858 (2004).

    ADS  CAS  Article  Google Scholar 

  4. 4

    Shim, S. -H., Duffy, T. S., Jeanloz, R. & Shen, G. Geophys. Res. Lett. 31, L10603 (2004).

    ADS  Article  Google Scholar 

  5. 5

    Tsuchiya, T., Tsuchiya, J., Umemoto, K. & Wentzcovitch, R. M. Earth Planet. Sci. Lett. 224, 241–248 (2004).

    ADS  CAS  Article  Google Scholar 

  6. 6

    Tsuchiya, T., Tsuchiya, J., Umemoto, K. & Wentzcovitch, R. M. Geophys. Res. Lett. 31, doi:10.1029/2004GL020278 (2004).

  7. 7

    Lay, T. & Helmberger, D. V. Geophys. J. R. Astron. Soc. 75, 799–837 (1983).

    ADS  Article  Google Scholar 

  8. 8

    Wysession, M. E. et al. in The Core-Mantle Boundary Region (eds Gurnis, M. et al.) 273–297 (Am. Geophys. Un., Washington DC, 1998).

    Google Scholar 

  9. 9

    Nataf, H. -C. & Houard, S. Geophys. Res. Lett. 20, 2371–2374 (1993).

    ADS  Article  Google Scholar 

  10. 10

    Sidorin, I., Gurnis, M. & Helmberger, D. V. Science 286, 1326–1331 (1999).

    CAS  Article  Google Scholar 

  11. 11

    Kendall, J. -M. & Silver, P. G. Nature 381, 409–412 (1996).

    ADS  CAS  Article  Google Scholar 

Download references

Author information



Rights and permissions

Reprints and Permissions

About this article

Cite this article

Duffy, T. Deeper understanding. Nature 430, 409–410 (2004).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing