Grain‐Boundary Diffusion Creep of Olivine: 1. Experiments at 1 atm

Abstract

We conducted one‐atmosphere uniaxial compression experiments on fine‐grained (~1 μm) Fe‐bearing olivine (Mg1.8Fe0.2SiO4) aggregates that were variably doped with CaO ± Al2O3. We identified power‐law interface‐controlled creep at low stresses and grain‐boundary diffusion creep at high stresses, which operate as mutually coupled, that is, sequential processes. We established constitutive equations for interface‐controlled creep and diffusion creep of undoped olivine and used the combined rate equation as a reference to examine the effect of doping on creep rates. Ca and Al were found to enhance rates of both interface‐controlled creep and diffusion creep above certain temperatures, and this effect becomes significant with increasing temperature. We attribute the rate enhancements to grain‐boundary disordering promoted by grain‐boundary segregation of the dopants at near‐solidus conditions. The enhancements are well described in relation to the sample solidus temperature and an additional activation energy relative to that of the reference creep state.

Grain‐Boundary Diffusion Creep of Olivine: 2. Solidus Effects and Consequences for the Viscosity of the Oceanic Upper Mantle

“…Thus, we consider the shear viscosity ( ) of the upper mantle based on Equation 6, in which an activation volume ( V ) is added and the temperature is replaced with the geotherm ( T geo ): where We used 4 cm 3 /mol for V , which was determined by Si grain‐boundary self‐diffusion experiments on our synthesized Fe‐free olivine aggregates at pressures ranging from 0.1 MPa to 13 GPa (Fei et al, 2016). In Part 1 of this study (Yabe et al, 2020), we showed good reproductions of our obtained creep rates for Fe‐free and Fe‐bearing olivine with their reported diffusivities (Figure 14 in Yabe et al, 2020). As seen in Equations 20a–20c, grain size, temperature ( T geo ), and normalized temperature ( T geo / T s ) are the primary factors controlling viscosity.…”

“…where A diff ¼ 1.12 × 10 10 μm 3 /MPa/s and Q diff ¼ 470 kJ/mol. We showed that the _ ε ref diff values from Equation 3 are comparable to diffusion creep rates of Fe-free olivine and are also consistent with the creep rates predicted from Si grain-boundary self-diffusivity (Fei et al, 2016) (see Figure 14 in Yabe et al, 2020).…”

“…We determined the following constitutive equations for each mechanism: where A int = 2.12 × 10 12 μm/MPa 3 /s and Q int = 610 kJ/mol, and where A diff = 1.12 × 10 10 μm 3 /MPa/s and Q diff = 470 kJ/mol. We showed that the values from Equation 3 are comparable to diffusion creep rates of Fe‐free olivine (Nakakoji et al, 2018) and are also consistent with the creep rates predicted from Si grain‐boundary self‐diffusivity (Fei et al, 2016) (see Figure 14 in Yabe et al, 2020).…”

“…We limited the plot to data collected from stepped‐temperature tests that were aimed at deforming the samples well within the diffusion creep regime (see Yabe et al, 2020). We used a T s of 1540°C (Bowen & Schairer, 1935) and the experimentally determined T s for our undoped and doped olivine, respectively, all of which are listed in Table 2 of Yabe et al (2020). The T / T s of RG‐olivine and our undoped olivine are all <0.9, while T / T s is >1 for SC‐olivine except for one datum of BK.…”

“…Plot of χ as a function of experimental temperature normalized by the sample solidus ( T / T s ). Data from our undoped and doped olivine given in Yabe et al (2020) are included. The predicted χ diff ‐ T / T s relationships (Equation 6) for T s of 1000°C, 1160°C, and 1350°C are shown.…”