Grain growth kinetics of bridgmanite under topmost lower-mantle

The absence of seismic anisotropy in most regions of the lower mantle suggests that diffusion creep may be the dominant mechanism in the lower mantle. Because the diffusion-creep rate is inversely proportional to the 2~3 power of grain size, knowledge of the grain-growth kinetics is crucial for studying lower-mantle dynamics. For these reasons, this study determined the grain-growth kinetics of bridgmanite at a pressure of 27 GPa using advanced multi-anvil techniques.

We first measured the grain sizes of bridgmanite in an olivine bulk composition with various annealing durations at 2200 K. The results were fitted to an equation dⁿ – d₀ⁿ = kt, where d and d₀ are the final and initial grain sizes, respectively, n is the grain-size exponent, t is the annealing duration, and k is the growth-rate constant. This fitting yielded n = 5.2 ± 0.3, which is much smaller than given by a previous study [Yamazaki et al., 1996], n = 10.6 ± 1.1. This discrepancy may be because Yamazaki et al.’s [1996] olivine starting material may have contained adhesive water, which enhanced grain growth at the beginning of annealing. We then conducted runs at various temperatures, yielding the activation energy of 260 ± 20 kJ/mol. These results suggest that the bridgmanite grain sizes over 0.1 – 1 Gyr should have grain sizes of 150-230 μm, which is one order of magnitude larger than Yamazaki et al.’s [2006] estimation. Consequently, the lower mantle should be much harder than previously considered.

Furthermore, we measured the grain-growth kinetics as a function of the fraction of coexisting ferropericlase. Although the grain-growth kinetics is almost independent of the ferropericlase fraction down to 20 vol.%, it rapidly increases with decreasing ferropericlase fraction at lower fractions. Over 0.1~4.5 Gyr, the bridgmanite grain sizes in pure-bridgmanite rock should be 2 ~ 3 orders of magnitude larger than those coexisting with 20 vol.% of ferropericlase. These results suggest that pure-bridgmanite rock has 4 ~ 9 orders of magnitude lower flow rates than pyrolite if the diffusion creep is dominant. Since the diffusion creep rate in pure-bridgmanite rock is so low, the dislocation creep should dominate in pure-bridgmanite rock. We estimated that the pure-bridgmanite rock should have 1 ~ 2.5 orders of magnitude more viscous than pyrolite if the stress condition is 0.1~0.5 MPa in the lower mantle. This variation may interpret the viscosity variation in the lower mantle inferred from the geoid analysis [Rudolph et al., 2015], subduction speed [van der Meer et al., 2018], and plume morphology [French & Romaniwicz, 2016].