Post-spinel transition in Mg2SiO4

Purpose of this study

  • Studying detailed phase relations of post-spinel transition in Mg2SiO4
    • Checking the results of Ito & Takahashi (1989) and Irifune et al. (1998) by in situ X-ray diffraction in a multi-anvil apparatus
    • At what pressure is the phase boundary located at the realistic mantle temperature?
    • How is the Clapeyron slope of the post-spinel transition?

Experimental setup

  • The starting material:  mixture of Mg2SiO4 + Au.
  • X-rays are incidented in parallel to the axis of the cylindrical heater
  • P is calculated from T and unit cell volume of Au using EOS of Au by Anderson et al. [1989].


Main difficulty and its solution

  • Inertness of ringwoodite
    • Difficult to initiate dissociation of ringwoodite into perovskite + periclase
      • Especially when annealed at HT for a long time.
      • Crystalline perfection increases by annealing
  • Solution: rapid heating
    • 100-1000 K/min
    • thermal shock
      • May create dislocations in spinel grains
      • Once initiated, the reaction completes within a few minutes.
        • very fast

Examples of change in diffraction pattern

 Experimental results


  • P at 660-km discontinuity is 23 GPa, whereas P of the post-spinel transition observed in this study is 22 GPa.
    • 1 GPa higher than that of Irifune et al. (1998)
    • Still 1 GPa lower than 660-km discontinuity
      • Because of uncertainty of EOS of Gold?
      • 5% error is possible
      • Thermal pressure and/or pressure derivative of bulk modulus
  • Small temperature dependence of post-spinel transition
    • If steep negative slope (>|-2| MPa/K), the mantle convection can be prevented on the boundary
    • Post spinel transition cannot be a barrier of the mantle convection.
  • Completely against the previous understanding!

Successive studies

People doubted the results of this study, and the phase relation were repeatedly studied [cf. Fei et al., 2004; Litasov et al., 2005].

But the results are the same. Very small Clapeyron slope!

  • If dP/dT = -0.4 MPa/K → Ca. 1 km/100K
    • ±20 km topography → ±2000 K
      • Too large variation
      • The temperature at 660 km depth : ~2000K
  • If dP/dT = -1.0 MPa/K → Ca. 2.5 km/100K
    • ±20 km topography → ±800 K
    • Still very large
  • Temperature dependence of seismic wave velocity → 0.6 % / 100 K
  • Purturbation of seismic wave velocity : ±2 % → ± 300 K variation in temperature
  • Temperature variation inferred from the topography of 660-km discontinuity is too large (±800 K)
  • The temperature around 660 km depth is supposed to be 2000 K (1700℃)
  • If there is temperature variation of ±800 K, high temperature part of the mantle should melt.
  • However, mantle is solid, because shear wave propagates in the mantle
  • We cannot accept temperature variation of ±800 K.  We need some other factors to increase topology of the 660 km discontinuity



  • By combination of multi-anvil high-pressure experiment and synchrotron-based X-ray diffraction, it was found that the slope of breakdown of ringwoodite to perovskite + periclase is very shallow.
  • It is difficult to explain topography of the 660-km discontinuity and slab stagnation.
  • The negative slope is not a reason for the slab stagnation.  Other factors should be important.  However, there are no definite answer for them at present.