Viscosity undulations in the lower mantle: The dynamical role of iron spin transition

https://doi.org/10.1016/j.epsl.2015.03.013Get rights and content

Highlights

  • Determination of the role of iron spin transition on ferropericlase viscosity.

  • Reduction of the free energies associated to mantle rocks rheology associated to iron spin transition.

  • Characterization of radial viscosity undulation in lower mantle based on diffusion creep and dilation factor.

  • Relationship between Large Igneous Provinces and lower mantle radial viscosity undulations.

Abstract

A proper determination of the lower-mantle viscosity profile is fundamental to understanding Earth geodynamics. Based on results coming from different sources, several models have been proposed to constrain the variations of viscosity as a function of pressure, stress and temperature. While some models have proposed a relatively modest viscosity variation across the lower mantle, others have proposed variations of several orders of magnitude. Here, we have determined the viscosity of ferropericlase, a major mantle mineral, and explored the role of the iron high-to-low spin transition. Viscosity was described within the elastic strain energy model, in which the activation parameters are obtained from the bulk and shear wave velocities. Those velocities were computed combining first principles total energy calculations and the quasi-harmonic approximation. As a result of a strong elasticity softening across the spin transition, there is a large reduction in the activation free energies of the materials creep properties, leading to viscosity undulations. These results suggest that the variations of the viscosity across the lower mantle, resulting from geoid inversion and postglacial rebound studies, may be caused by the iron spin transition in mantle minerals. Implications of the undulated lower mantle viscosity profile exist for both, down- and up-wellings in the mantle. We find that a viscosity profile characterized by an activation free energy of G(z0)300400kJ/mol based on diffusion creep and dilation factor δ=0.5 better fits the observed high velocity layer at mid mantle depths, which can be explained by the stagnation and mixing of mantle material. Our model also accounts for the growth of mantle plume heads up to the size necessary to explain the Large Igneous Provinces that characterize the start of most plume tracks.

Introduction

Lower mantle viscosity has been the subject of great debate over the last decades (Sammis et al., 1977, Ricard and Wuming, 1991, Forte and Mitrovica, 2001), and its determination would be fundamental to address a number of questions on the mantle, such as its composition, heterogeneity, and geodynamics. Interpretation of data, coming from geoid inversion and postglacial rebound studies, indicated undulations in the viscosity profile, with peaks around 1300 and 2000 km deep and a valley around 1600 km (Mitrovica and Forte, 2004). If viscosity is considered as controlled by thermally activated microscopic mechanisms (Sammis et al., 1977, Ellsworth et al., 1985), this viscosity profile could not be easily reconciled with a single diffusion creep mechanism taking place in the lower mantle. Those variations in the lower mantle viscosity suggest that several microscopic competing diffusion mechanisms could be taking place in the mantle. On the other hand, the recent discovery of the iron spin transition in major mantle minerals (Badro et al., 2003, Badro et al., 2004, Lin et al., 2005, Lin et al., 2007, Speziale et al., 2005, Tsuchiya et al., 2006, Fei et al., 2007), and the corresponding anomalies in their elasticity (Crowhurst et al., 2008, Wentzcovitch et al., 2009, Marquardt et al., 2009, Wu et al., 2009, Antonangeli et al., 2011, Wu and Wentzcovitch, 2014), could reconcile the description of viscosity with a single thermally activated mechanism by using the available information from simultaneous inversion of geoid and post-glacial rebound data (Mitrovica and Forte, 2004).

Van Keken et al. (1992) found that some radial viscosity profile would produce a pulsating diapiric rise. This work has simulated numerous investigations of the effects of non-monotonous viscosity profile in the lower mantle. Tomographic models of slabs that penetrated in the lower mantle show strong signals of large body lying between 1500 and 2000 km depth (Grand, 1994), such as the Farallon slab (Sigloch et al., 2008). This result has been confirmed by the analysis of a variety of global tomographic models, e.g. Tx2007 (Simmons et al., 2006), Rmsl-s06 (Li et al., 2007), Saw642an (Panning and Romanowicz, 2006), all finding a clear transition from fast to slow shear seismic velocities for degree up to ∼16 at a less than 1500 km depth (Boschi et al., 2008). Morra et al. (2010) showed that a sinking plate might penetrate, reorganize or even stall when crossing a 200 to 500 km high viscosity region in the middle of the lower mantle. Shahnas et al. (2011) used global mantle convection models to demonstrate that the only effects on the density of the iron spin transition enhances the vigor of rising plumes below 2000 km depth and slightly increases the temperature of the lowermost region of the mantle. Peltier and Drummond (2010) used glacial isostatic adjustment observations to infer a modest increase of the viscosity at mid mantle depths. Overall, those investigations have shown that a non-monotonic lower mantle viscosity profile would create substantial complications to the dynamics of sinking slabs and rising plumes. Here, we employ a standard scaling for plume head size evolution (e.g., Griffiths and Campbell, 1990, Ribe et al., 2007) integrating it along a one-dimensional vertical profile to calculate a broad range of solutions for the dynamics of a plume rise through a variety of physically based lower mantle viscosity profiles, obtained by first principles calculations of mineral elasticity.

We used the elastic properties of ferropericlase (Fp), Mg1  xFexO with x=0.1875, computed by a combination of first principles calculations and quasiharmonic approximation (Carrier et al., 2008, Wentzcovitch et al., 2009, Wu et al., 2013), to determine its viscosity under lower mantle conditions. Fp was treated as a solid solution in a mixed spin state, with the concentration of material with iron in high and low spin determined by the respective free energies. Although Fp is only the second most abundant lower mantle mineral, it is likely controlling deformation in the lower mantle (Zerr and Boehler, 1994, Yamazaki and Karato, 2001). This is justified by the fact that Fp is softer than the more abundant ferrosilicate perovskite under the same thermodynamic conditions. The viscosity of Fp was described within the elastic strain model (Sammis et al., 1977, Ellsworth et al., 1985), in which the activation energy parameters were computed along adiabatic (0.3 K/km) (Boehler, 2000) and superadiabatic (1.2 K/km) (da Silva et al., 2000) geotherms. The manuscript explores the role of dilatation and shear microscopic mechanisms (Ellsworth et al., 1985), variations in activation energies at the top of the lower mantle, and the Newtonian character of the mantle. The results show that the variations in Fp elasticity due to the iron spin transition can explain the undulations in the mantle viscosity, such as the viscosity hill about 800 km above the core–mantle boundary (Mitrovica and Forte, 2004).

Section snippets

Theoretical models

The viscosity (η) of Fp was described as a thermally activated process, as a result of diffusion of atomic species (Saha et al., 2013),η=f(σ)exp(GeRT) where Ge is the Gibbs free energy of activation and f(σ) is a function of stress. For a Newtonian fluid, f(σ) is a constant and G=Ge, where G is the activation energy of the appropriate dynamical mechanism (Ellsworth et al., 1985). On the other hand, the effective viscosity of a power law fluid of order n is equivalent to the viscosity of a

Viscosity of ferropericlase

Fig. 1 shows the adiabatic bulk and shear moduli and the respective sound wave velocities of Fp along two geotherms. The spin transition causes strong softening in the bulk modulus, but minor effect in the shear modulus. These results are consistent with recent experimental data on the elasticity of ferropericlase (Marquardt et al., 2009). The figure shows that the softening is more noticeable in the adiabatic geotherm than in the superadiabatic one. This is because the depth regions in which

Implications for mantle dynamics

Implications of the undulated lower mantle viscosity profile exist for both, down- and up-wellings in the mantle. Three dimensional models show that a viscosity “hill” of one order of magnitude through 500–1000 km of lower mantle is sufficient to slow down the sinking speed of a slab, and that a viscosity peak of two orders of magnitude can completely stop the sinking (Morra et al., 2010). Analytical calculations demonstrate that the sinking velocity of a slab is only weakly dependent on its

Summary

In summary, we have demonstrated that changes in the elastic parameters due to iron spin transition of Fp can produce dramatic variations in the rheological properties of the material, as to cause a non-monotonic behavior with depth. Our results indicate that, at lower mantle conditions, the effect of the spin transition in Fp is to create viscosity undulations respect to the radial profile and also horizontal variations. High viscosity regions would trap sinking slabs favoring chemical mixing,

Acknowledgments

This research was partially supported by grants NSF/EAR 0635990, NSF/ITR 0428774, and NSF/DMR 0325218 (ITAMIT) and the geochemistry and geophysics programs of the National Science Foundation. Computations were performed at the Minnesota Supercomputing Institute and on the Big Red Cluster at Indiana University. JFJ acknowledges partial support from Brazilian agencies FAPESP (grant number 2009/14082-3) and CNPq (grant number 473307/2013-8). GM acknowledges Louisiana Board of Regents – Research

References (50)

  • S. Saha et al.

    Effect of anomalous compressibility on Fe diffusion in ferropericlase throughout the spin crossover in the lower mantle

    Earth Planet. Sci. Lett.

    (2013)
  • N.A. Simmons et al.

    Constraining mantle flow with seismic and geodynamic data: a joint approach

    Earth Planet. Sci. Lett.

    (2006)
  • I. Stretton et al.

    Dislocation creep of magnesiowüstite (Mg0.8Fe0.2O)

    Earth Planet. Sci. Lett.

    (2001)
  • P. van Keken et al.

    Pulsating diapiric flows: consequences of vertical variations in mantle creep laws

    Earth Planet. Sci. Lett.

    (1992)
  • D. Antonangeli et al.

    Spin crossover in ferropericlase at high pressure: a seismologically transparent transition?

    Science

    (2011)
  • J. Badro et al.

    Iron partitioning in Earth's mantle: toward a deep lower mantle discontinuity

    Science

    (2003)
  • J. Badro et al.

    Electronic transitions in perovskite: possible nonconvecting layers in the lower mantle

    Science

    (2004)
  • R. Boehler

    High-pressure experiments and the phase diagram of lower mantle and core materials

    Rev. Geophys.

    (2000)
  • P. Carrier et al.

    Quasiharmonic elastic constants corrected for deviatoric thermal stresses

    Phys. Rev. B

    (2008)
  • M. Cococcioni et al.

    Linear response approach to the calculation of the effective interaction parameters in the LDA+U method

    Phys. Rev. B

    (2005)
  • J.C. Crowhurst et al.

    Elasticity of (Mg, Fe)O through the spin transition of iron in the lower mantle

    Science

    (2008)
  • K. Ellsworth et al.

    Viscosity profile of the lower mantle

    Geophys. J. R. Astron. Soc.

    (1985)
  • Y.W. Fei et al.

    Spin transitions and equations of state of (Mg, Fe)O solid solutions

    Geophys. Res. Lett.

    (2007)
  • A.M. Forte et al.

    Deep-mantle high-viscosity flow and thermochemical structure inferred from seismic and geodynamic data

    Nature

    (2001)
  • S.P. Grand

    Mantle shear structure beneath the Americas and the surrounding oceans

    J. Geophys. Res.

    (1994)
  • Cited by (12)

    • On the amplitude of dynamic topography at spherical harmonic degree two

      2019, Tectonophysics
      Citation Excerpt :

      An alternative explanation might be to alter the radial viscosity profile of the mantle in order to reduce degree-two dynamic topography. Viscosity variations within the mantle remain poorly constrained and a wide range of possibilities have been published, even within the last couple of years (e.g. Justo et al., 2015; Marquardt and Miyagi, 2015; Rudolph et al., 2015; King, 2016; Lau et al., 2016; Liu and Zhong, 2016). Importantly, altering the viscosity structure too severely is likely to erode the quality of fit to geoid anomalies.

    • Fine-scale structure of the mid-mantle characterised by global stacks of PP precursors

      2017, Earth and Planetary Science Letters
      Citation Excerpt :

      Regardless of where the jump in viscosity occurs, such an increase will also increase the lateral length scale of mantle flow compared to the upper mantle (Forte and Mitrovica, 2001) leading to a change of the radial distribution of heterogeneity between the mid-mantle (e.g. 1000 km) and the lower mantle (depths >2000 km). The radially varying small-scale mantle heterogeneity detected by PP precursors in this study is potentially a unique way to observe the effect of depth varying viscosity structure that has been proposed using a wide variety of different approaches (Forte and Mitrovica, 2001; Marquardt and Miyagi, 2015; Justo et al., 2015; Rudolph et al., 2015). Using the information contained in the short-period scattered seismic wavefield might deliver important complementary information to large-scale tomographic images allowing better insight into the dynamics and evolution of our planet.

    • Mid-mantle heterogeneities and iron spin transition in the lower mantle: Implications for mid-mantle slab stagnation

      2017, Earth and Planetary Science Letters
      Citation Excerpt :

      Our models assume a high-end value for the Clapeyron slope of deep mantle phase transition (e.g., Hirose, 2006), while some other studies suggest lower values (e.g., Tsuchiya et al., 2004). In the last sequence of models we employ a viscosity profile (V3) inferred from the spin transition in ferropericlase (Justo et al., 2015). The high spin to low spin transition in ferropericlase can cause viscosity undulations in the lower mantle as shown in Fig. 3-d that may enhance the flow stagnation at mid-mantle depths or slow the descending slabs.

    View all citing articles on Scopus
    View full text