Abstract

We review our current understanding of the rheological properties of the lower mantle based both on materials science and geophysics points of view. We assume a simple model of the lower mantle that is made of only two minerals: bridgmanite (Br) (Mg,Fe)SiO3 and ferropericlase (Fp) (Mg,Fe)O, and address a question of (i) which mineral is weaker (lower viscosity), (ii) how does lower mantle viscosity change with depth and location, and (iii) discuss implications for shear localization. We first review plausible mechanisms of deformation based on the deformation mechanism map on the normalized stress and temperature space. We conclude that likely mechanism of deformation in the lower mantle is either diffusion creep or power-law dislocation creep. Based on this review, we discuss recently proposed models by Cordier and his group (Cordier in Nature 481:177–181, 2012; Cordier in Nature 613:303–306 , 2023) where either asthermal creep (i.e., low-temperature plasticity) or pure climb creep (not power-law dislocation creep) would play an important role. We conclude that these models are not acceptable because (1) many aspects of their models are incompatible with experimental observations and theoretical models of deformation of most materials including oxides and metals and (2) these models are not consistent with the distribution of seismic anisotropy. Hence, we focus on power-law dislocation creep and diffusion creep. We review previously published results on deformation (by dislocation creep) and diffusion, we conclude that Fp is weaker than Br. The radial (depth) depth and lateral variation of viscosity is discussed based on the estimated activation volume and estimated variation of grain-size. Geophysical studies suggest only modest depth variation of viscosity that demands relatively small activation volume (V* (< 3 × 10–6 m3/mol)). Plausible models to explain small activation volume are discussed including the role of extrinsic diffusion. Grain-size also controls viscosity if deformation is by diffusion creep. Okamoto and Hiraga (J Geophys Res, 2024. 10.1029/2023JB027803), Solomatov et al. (Phys Earth Planet Inter 129:265–282, 2002) estimated the grain-size evolution in the lower mantle based on the kinetics of grain-growth and the role of a phase transformation. In contrast, there are other papers (e.g., Paul et al. in Prog Earth Planet Sci 11:64, 2024; Rozel in Geochem Geophys Geosyst, 2012. 10.1029/2012GC004282) where grain-size distribution is estimated assuming that grain-size is controlled by dynamic recrystallization. The validity of assumption is questionable because dynamic recrystallization occurs due to deformation by dislocation creep but not by diffusion creep and the absence of seismic anisotropy indicates that diffusion creep dominates in most of the lower mantle. Finally, we review the published models of shear localization that would explain the long-term preservation of geochemical reservoirs in the lower mantle. Accepting that two minerals (Fp and Br) in the lower mantle have largely different viscosity, Ballmer et al. (Nat Geosci 10:236–240, 2017) proposed that the presence of regions of compositional difference (difference in Fp/Br ratio) leads to localized deformation (deformation mainly in the weaker regions). However, in addition to the ad hoc nature of this model, there is no strong evidence for the presence of large variation in Fp/Br in the lower mantle that makes the validity of this model questionable. There are some papers where processes of shear localization are explored without invoking the presence of regions of large rheological contrast. Thielmann et al. (Geochem Geophys Geosyst, 2020. 10.1029/2019GC008688) presented the results of theoretical study of deformation of initially homogeneous two-phase mixture (Fp and Br) and showed that deformation causes the elongation of a weak Fp that promotes shear localization. In this model, the rheological contrast between Fp and Br was assumed to be independent of strain. However, Cho and Karato (J Geophys Res 2022. 10.1029/2021JB022673 ; Phys Earth Planet Inter, 2024. 10.1016/j.pepi.2024 ) showed that when deformation is by diffusion creep, the rheological contrast increases with strain due to the evolution of stress concentration caused by grain elongation. They showed that this will promote strain weakening particularly in simple shear that would lead to shear localization. Consequently, the tendency for shear localization is stronger in their model than a model where rheological contrast is assumed to be independent of strain.