are competing processes and that both diffusion creep and dislocation creep can operate simultaneously in the same sample.
In the lower mantle, seismic anisotropy is observed at the top of the lower mantle near subducting slabs (Ferreira et al., 2019; Lynner & Long, 2015; Mohiuddin et al., 2015; Wookey et al., 2002) and also at the base of the mantle in the D” region (for recent summaries of D” anisotropy, see Creasy et al., 2019 and Romanowicz & Wenk, 2017). In contrast to these regions, the bulk of the mantle appears to be seismically isotropic. This has led to the suggestion that the bulk of the mantle deforms by diffusion creep (Karato et al., 1995), whereas regions of high strain near subducting slabs and at boundary layers such as near the CMB undergo deformation by dislocation creep (McNamara et al., 2002). However, others have suggested that isotropy might be due to patterns of anisotropy for Brg and Fp that cancel each other leading to an isotropic aggregate (Wenk, Speziale, et al., 2006). In any case, it appears that dislocation creep is active in at least some regions of the lower mantle, but it might also be likely that the mantle is close to the transition between diffusion creep and dislocation creep. As such, strain rate gradients in the mantle may result in transitions between diffusion‐dominated and dislocation‐dominated regimes.
2.3 METHODS: EXPERIMENTAL DEFORMATION OF LOWER MANTLE MINERALS
Deformation studies of lower mantle phases are complicated by the fact that many phases are unstable (Brg) or unquenchable (Ca‐Pv and pPv) to ambient conditions. Both Ca‐Pv and pPv become amorphous during pressure release and cannot be preserved to ambient pressures. Thus study of recovered samples (from high pressure and/or temperature experiments) is difficult or impossible, necessitating in‐situ analysis. A technique that has led to significant advances in deformation studies of these phases is the use of synchrotron x‐ray diffraction in conjunction with high pressure and temperature deformation devices.
2.3.1 The Diamond Anvil Cell as a Deformation Device
The diamond anvil cell (DAC) uses two gem‐quality diamonds with flat tips (culets) in opposed anvil geometry to generate high confining pressure (e.g., Jayaraman, 1983). In DAC experiments, a tiny sample (< 0.1 mm) is confined within a gasket and the anvils impose pressure and stress on the sample. The DAC is capable of generating pressures in excess of those experienced at the center of the Earth. Most commonly, the DAC is used in axial diffraction geometry where the x‐ray beam is parallel to the compression axis. This is advantageous as the diamonds are nearly x‐ray transparent. In these types of experiments the sample is typically surrounded by a soft pressure medium to reduce differential stress to near hydrostatic conditions. This type of experiment has been highly successful for studying phase equilibria and equations of state at high pressure and temperature. The DAC has also been used in axial geometry without a pressure medium to infer yield strength from radial pressure gradients (e.g., Meade & Jeanloz, 1988).
The DAC may also be used in radial geometry to study deformation properties at high pressures. In radial diffraction, the x‐ray beam is directed through the gasket, orthogonal to the compression direction. For these experiments, no pressure medium is used so that the diamonds impose pressure and deviatoric stress to deform the material elastically and plastically. This geometry requires x‐ray transparent gaskets such as beryllium or amorphous boron (Hemley et al., 1997; Merkel & Yagi, 2005) to allow passage of the x‐ray beam across the sample. Radial diffraction technique and the DAC has now been used for around two decades to make in‐situ measurement of lattice strains and texture development at high pressure (Hemley et al., 1997; Wenk et al., 2000).
Although the DAC reaches the highest pressure range of all the deformation devices, it has many limitations. Generating high temperatures during DAC deformation is technically challenging; most DAC deformation experiments are room temperature. High temperatures are generated in the DAC through the use of electrical resistive heating or by focusing an infrared laser beam through the diamonds and onto the sample. Resistive heating produces relatively uniform temperatures within the sample, but temperatures are significantly lower than those that can be achieved through laser heating (Figure 2.1). For radial diffraction resistive heating is also complicated by the fact that any heater surrounding the sample must allow passage of x‐rays in the radial direction. A graphite resistive heater has been developed for radial geometry (Liermann et al., 2009; Petitgirard et al., 2009) and has been used statically (without active compression) to generate temperatures as high as 1785 K at 40 GPa (Miyagi et al., 2013). More recently, Immoor et al. (2018) used this technique to deform a sample of Fp along a 1400 K isotherm to ~ 70 GPa (Figure 2.1).
Laser heating can reach temperatures as high as ~ 6000 K (Boehler, 2000) but generally has large axial and radial temperature gradients (Boehler, 2000; Manga & Jeanloz, 1996). Axial gradients can result due to the high thermal conductivity of the diamond anvils, creating lower temperatures near the diamonds. Radial temperature gradients are more problematic for radial diffraction and arise because the laser hotspot is typically smaller than the sample dimensions. Gasket material may also conduct heat away from the edges of the sample. Additionally, temperatures may not be uniform throughout the laser hotspot. Advances such as double‐sided heating (Shen et al., 2001) and beam shaping (Prakapenka et al., 2008) have significantly reduced temperature gradients in the DAC. Surrounding the sample in an insulating material serves to both increase heating efficiency and reduce thermal gradients by reducing heat conduction away from the hotspot. However, for deformation experiments in radial geometry, this can be problematic. In addition to extra diffraction peaks from the insulating material, which complicates lattice strain and texture analysis, the insulation material must also be strong enough to impose deformation on the sample. If the insulation material is weak, strain will partition into the insulation material and the harder phase will remain relatively undeformed. As most lower mantle phases have high strength, this significantly limits the choice of insulation material, and in practice most radial diffraction measurements with laser heating do not use an insulating material. Additionally, laser heating in radial geometry is challenging, as most synchrotron beamlines are set up for laser heating in axial geometry, and reconfiguring the laser system for heating in radial geometry is difficult and time consuming (Kunz et al., 2007). Thus, only a few studies have performed radial diffraction experiments coupled with laser heating (Kunz et al., 2007; Merkel et al., 2004; Miyagi et al., 2013; Miyagi, Kunz, et al., 2008; Wu et al., 2017). The highest pressures and temperatures reached for radial diffraction and laser heating in the DAC are ~2500K and 150 GPa (Figure 2.1; Wu et al., 2017).
Figure 2.1 Approximate pressure and temperature capabilities of various deformation devices including both large volume techniques and the radial diffraction diamond anvil cell (R‐DAC). Also plotted for reference is a “cold” mantle geotherm (Brown and Shankland, 1981).
In compression, experiments, diamonds induce both pressure and deviatoric stress on the sample, and these cannot be easily decoupled. Thus, studying deformation properties at a fixed confining pressure in the DAC is currently not possible. This means that flow stresses estimated from DAC experiments convolute strengthening due to pressure and strain hardening. Controlling strain rates is difficult in the DAC. Often, deformation is controlled through manual increase of pressure/stress through turning load screws. Not only does this result in a discontinuous increase in stress and strain, but it is incompatible with simultaneous heating and deformation. Furthermore, the small sample size makes it difficult to measure sample dimensions during experiments, and commonly strain (and thus strain rate) cannot be measured. The use of a gas membrane has somewhat improved on this, as compression is