High-pressure experiments cast light on deep-Earth mineralogy

A technically challenging analysis has revealed the physical properties of a mineral at pressures and temperatures as high as those in Earth’s mantle. The findings have implications for our understanding of Earth’s deep interior.
Johannes Buchen is in the Seismological Laboratory, California Institute of Technology, Pasadena, California 91125, USA.

Search for this author in:

The movement of tectonic plates carries parts of Earth’s crust back into the deep interior of the planet. Earlier in Earth’s history, processes that preceded the present regime of tectonic motion might also have forced crustal rocks to sink into Earth’s mantle1. The fate of crustal material recycled in the mantle, however, is unknown. Writing in Nature, Gréaux et al.2 present measurements of the speed of sound in one of the minerals that are thought to form in recycled crustal rocks, at pressures and temperatures that resemble those in Earth’s mantle. The findings promise to facilitate the use of seismic waves to track crustal rocks in the mantle.

Beneath the oceans, Earth’s crust consists mainly of a rock called basalt. When exposed to high pressures as a result of tectonic processes, basalt transforms into an assemblage of minerals that is denser than the rocks of the underlying mantle. Basaltic rocks can therefore sink into the mantle, where they are exposed to rising pressures and temperatures that drive further mineral transformations3,4. One of these reactions produces the mineral calcium silicate perovskite (CaPv).

The crystal structure of CaPv has cubic symmetry at temperatures greater than about 600 kelvin, and this cubic form is probably adopted at the high temperatures present in the mantle. However, the crystal structure of CaPv spontaneously distorts to a tetragonal form at lower temperatures. CaPv must therefore be held at high temperatures and pressures to determine the physical properties of the cubic phase.

In their experiments, Gréaux et al. synthesized CaPv, and analysed both the cubic and the tetragonal forms at high pressures and temperatures in a high-pressure apparatus. They measured the time taken for ultrasonic waves to travel through the CaPv at different pressure–temperature combinations, and irradiated the samples with intense X-rays to generate images and diffraction patterns. By combining these measurements, the authors derived sound-wave velocities and elastic moduli (which quantify the resistance of a solid material to small, non-permanent deformations) for CaPv.

Gréaux and colleagues found that the shear modulus of cubic CaPv, which specifically measures the resistance of the mineral to reversible deformation caused by distortions (shear deformation), is substantially lower than estimates5,6 calculated using first-principle computations. The difference reflects the fact that the computed sound-wave velocities were significantly higher than those measured in the experiments. The experimental findings highlight the importance of assessing the elastic and acoustic properties of mantle minerals at relevant pressures and temperatures that, in the case of CaPv, would stabilize the cubic form. The authors went on to use the new data for cubic CaPv to improve estimates of the velocities of seismic waves travelling through rocks in Earth’s mantle.

The boundary between Earth’s upper and lower mantle is marked by steep gradients in density and seismic-wave velocities at depths of around 660 kilometres. The pressures and temperatures thought to prevail at this depth coincide with major changes in the mineral assemblage of pyrolite — the hypothetical rock that is often used as a model for the rocks that constitute the bulk of the mantle. These mineral transformations increase the density of pyrolite, so that basaltic materials become buoyant at depths of between 660 km and 750 km4,7,8. Geodynamic simulations9 show that the sequence of density changes in this region can effectively trap recycled oceanic crust.

Gréaux and colleagues’ results suggest that seismic waves travel much more slowly through recycled oceanic crust than through pyrolite at depths of between 660 km and 770 km. Consequently, the authors propose that the presence of trapped basaltic rocks could explain the reduction of seismic-wave velocities that has been observed locally at these depths (Fig. 1). The reduction was previously attributed10,11 to deep dehydration melting — the process in which water is released from the crystal structures of hydrous minerals, causing melting. The report last year of the discovery12 of a fragment (an inclusion) of CaPv in diamonds that formed in the deep mantle supports Gréaux and colleagues’ hypothesis. However, inclusions of a hydrous mineral13 and of pressurized ice14 in two other diamonds point to the presence of water-containing fluids at similar depths, in support of the alternative hypothesis.

Figure 1 | Current theories for the recycling of oceanic crust in Earth’s mantle. Slabs of oceanic crust (which is formed mainly of a rock called basalt) and underlying mantle rocks (harzburgite) sink into Earth’s mantle, which is often modelled as being formed from a hypothetical rock called pyrolite. Basalt accumulates within and beneath the transition zone between the upper and lower mantle. In downwelling regions, hydrated rocks in the transition zone are pushed into the lower mantle, where they release water bound in their minerals. The resulting aqueous fluids can trigger melting. This dehydration melting could explain10,11 why seismic waves have low velocities in some regions at depths greater than 660 kilometres. Gréaux et al.2 report that basaltic rocks in the mantle will also have slow seismic waves. The presence of basaltic rocks at depths greater than 660 km could therefore be an alternative explanation for the low seismic-wave velocities in these regions.

Global-scale geodynamic simulations9 indicate that oceanic crust descending into the mantle accumulates to form a layer that is enriched in basaltic rocks, centred at a depth of around 600 km — that is, at shallower depths than would be inferred from experimentally derived rock densities alone4,7,8. Gréaux and colleagues’ results show that, in basalt, the velocities of the two types of seismic wave (known as shear (S) waves and compressional (P) waves) remain lower than global average seismic velocities at that depth, although the reduction in velocity is less than the reduction that occurs at depths greater than 660 km. It is known that P waves are converted to S waves at depths of around 600 km by globally distributed zones that have below-average S-wave velocities15. The idea that a layer of basaltic rocks scatters seismic waves at around 600 km depth could reconcile the seismic observations with the geodynamic predictions and with Gréaux and colleagues’ models derived from the measured physical properties of minerals.

Further seismological studies are necessary to map zones that have low seismic velocities through a range of depths, and to better constrain their characteristics — for example, to measure differences in the velocities of P and S waves relative to the surrounding mantle. Measurements of sound-wave velocities in single crystals of CaPv (rather than in polycrystalline samples, as studied by Gréaux et al.) would also reveal how such velocities depend on the direction of passage through the crystal lattice. The effect of the crystal lattice might give rise to an observable direction dependence of seismic-wave velocities in the mantle. Devising models that combine seismological data with constraints derived from geodynamic simulations and data for the physical properties of minerals will aid the search for recycled oceanic crust in Earth’s mantle.

Nature 565, 168-170 (2019)

doi: 10.1038/d41586-018-07864-2
Nature Briefing

Sign up for the daily Nature Briefing email newsletter

Stay up to date with what matters in science and why, handpicked from Nature and other publications worldwide.

Sign Up


  1. 1.

    Johnson, T. E., Brown, M., Kaus, B. J. P. & VanTongeren, J. A. Nature Geosci. 7, 47–52 (2014).

  2. 2.

    Gréaux, S. et al. Nature 565, 218–221 (2019).

  3. 3.

    Irifune, T., Sekine, T., Ringwood, A. E. & Hibberson, W. O. Earth Planet. Sci. Lett. 77, 245–256 (1986).

  4. 4.

    Irifune, T. & Ringwood, A. E. Earth Planet. Sci. Lett. 117, 101–110 (1993).

  5. 5.

    Stixrude, L., Lithgow-Bertelloni, C., Kiefer, B. & Fumagalli, P. Phys. Rev. B 75, 024108 (2007).

  6. 6.

    Kawai, K. & Tsuchiya, T. Geophys. Res. Lett. 42, 2718–2726 (2015).

  7. 7.

    Kesson, S. E., Fitz Gerald, J. D. & Shelley, J. M. G. Nature 372, 767–769 (1994).

  8. 8.

    Hirose, K., Fei, Y., Ma, Y. & Mao, H.-K. Nature 397, 53–56 (1999).

  9. 9.

    Ballmer, M. D., Schmerr, N. C., Nakagawa, T. & Ritsema, J. Sci. Adv. 1, e1500815 (2015).

  10. 10.

    Schmandt, B., Jacobsen, S. D., Becker, T. W., Liu, Z. & Dueker, K. G. Science 344, 1265–1268 (2014).

  11. 11.

    Liu, Z., Park, J. & Karato, S. Geophys. Res. Lett. 43, 2480–2487 (2016).

  12. 12.

    Nestola, F. et al. Nature 555, 237–241 (2018).

  13. 13.

    Pearson, D. G. et al. Nature 507, 221–224 (2014).

  14. 14.

    Tschauner, O. et al. Science 359, 1136–1139 (2018).

  15. 15.

    Shen, X., Yuan, X. & Li, X. Geophys. Res. Lett. 41, 836–842 (2014).

Download references