The orientation-optimized electron channeling contrast imaging (ooECCI) was employed for the investigation of individual dislocations and forming subgrain boundaries. A conventional field-emission scanning electron microscope (FE-SEM),Zeiss LEO 1530 FE-SEM with the GEMINI column, equipped with an electron-backscatter-diffraction (EBSD) system, was used for ECCI with a backscattered electron (BSE) detector. The acceleration voltage and the beam current are 20–30 kV and 11 nA, respectively. The working distance is 7–10 mm. The detectors were a 120-mm² annular scintillator-type detector (KE Development CENTAURUS) placed directly under the lens’ polepiece for ECCI (backscattered geometry) and a low-light camera with a phosphorus screen (Nordlys II of Oxford Instruments) for EBSD. Before ECCI, the crystal orientations of single-crystal garnets were determined by using off-line indexing of Kikuchi lines of (hkl) in each Kikuchi pattern acquired with the EBSD camera system. The SFC software (Kogure, 2003) was used for indexing Kikuchi bands to identify crystal planes nearly parallel to the incident beam at the 0-degree-microscope stage (or sample-tilt stage) in the FE-SEM. The detailed procedure for the ooECCI of a large single crystal was described in case 2 for a natural olivine (Miyajima et al., 2018).

The specimens for ooECCI are experimentally deformed single-crystal pyrope-rich garnets. Single crystals of a natural pyrope-rich garnet (Py₇₃Al₁₅Gr₁₂, averaged Z-numbers, AZs 10.4) were deformed at 3.5 GPa and 1273 K in a pure-shear geometry on a multi-anvil apparatus-in house (MAVO 6-ram press; Manthilake et al., 2012). The total strain and strain rate of the M643 and the M633 runs are 0.52 and 8.3·10⁻⁵ s⁻¹ and 0.30 and 5.0·10⁻⁵ s⁻¹, respectively. The crystal orientations were not controlled in the deformations but, in the M633 run, the single crystal was incidentally forced to be sheared nearly in the <100> direction on the {011} plane.

As a test of the applicability limit of ECCI into low-atomic-number minerals (averaged Z-numbers, AZs ∼ 10) and iron-free minerals, e.g., quartz, a polycrystalline forsterite (Bollinger et al., 2015) was also observed in ECCI. The specimen was deformed in a deformation DIA high pressure apparatus at 5.5 GPa and 1373 K with strain rate = 4.4·10⁻⁵ s⁻¹ and stress = 263 MPa. The detailed procedure in the ooECCI of a polycrystalline mineral was described in case 1 for torsion-deformed ferropericlase (Miyajima et al., 2018). All the experimental conditions for each specimen are described in Table 1.

After the selection of targeted microstructures in the ooECCI, a site-specific TEM thin foil (that we call “TEM lamella” in the FIB session) of pyrope-rich garnet (M643) was prepared with a dual-beam (Ga and field emission electron guns) focused ion beam (FIB) milling machine, FEI Scios, equipped with a sample-lift-out manipulator system, FEI Easylift. The Ga-beam conditions of the trench milling process in the M643 sample are 30 kV for the acceleration voltage and 5–30 nA for the beam current. After cleaning the TEM lamella surface by a 1-nA Ga beam, the lamella was lifted out and glued on a TEM grid by the Easylift system. The final thinning conditions are 30 kV and 300 pA to obtain an electron transparency against a 200-kV accelerating electron beam in TEM. The TEM lamella was milled normal to the main compression axis in the pure shear geometry.

The TEM characterization was performed by using a 200-kV scanning transmission electron microscope, FEI Titan G2 80-200 S/TEM. Crystal defect microstructures in the pyrope-rich garnet were observed in a conventional TEM mode with bright-field (BF), dark-field (DF) and weak-beam dark-field (WBDF) images to characterize dissociated dislocations, stacking faults between short spreading partial dislocations and long stacking fault ribbons. To determine the crystal orientations and diffraction vectors in TEM imaging, selected-area electron diffraction (SAED) was used. The scanning transmission electron microscopy (STEM) mode with BF, DF and high-angle annular dark-field (HAADF) detectors was also employed to obtain a whole view of the TEM lamella on the distributions of defects and chemical homogeneities.

Fig. A.1 (multimedia component No. 1) shows the BSE image of a whole view of the deformed single-crystal pyrope-rich garnet (M643). Striped contrasts are visible along the direction from upper left to lower right, which are likely to be a deformation-related feature; we call it “deformation band” because it has no remarkable chemical difference in energy-dispersive X-ray analysis. The contrasts are originated from a small orientation misfit between the band and the matrix (Fig. 1a and Fig. A.1c). The acquired Kikuchi pattern in the EBSD indicates a Kikuchi line (KL) of the (046) plane close to the sample's normal position (Fig. A.2, multimedia component No. 2), which means a two-beam condition for KL(046) (or g = 046 in the TEM technical term) available by an orientation optimization within a few degrees in the sample-tilt system of the FE-SEM. We followed the procedure to find an “inversion of contrast” in the BSE intensity (figs. 9 & 10 in Miyajima et al., 2018). Note that the intensity of the KL(046) band is stronger than that of the other Kikuchi bands close to the sample's normal position. After orientation optimization, dislocation-like contrasts were visible in the deformed single-crystal pyrope-rich garnet by using the ECCI excited with KL(046), which have about 33% of the intensity of the maximum structure factor of KL(400) (Fig. 1b). The kinematical diffraction intensity was calculated based on the electron wave length at 20 kV and a pyrope-end member structure. In comparison, dislocations in the deformed single-crystal pyrope-rich garnet (M633) were more visible in the ECCI excited with the strongest KL(400) in the structure (Fig. 2). A Kikuchi band with a high structure factor of more than 30% of the strongest one was mandatory for performing ooECCI with high quality.

In the TEM on the M643 sample (Fig. 3), dislocation microstructures with dissociated dislocations and stacking fault ribbons were confirmed, which correspond to the dislocation-like contrasts in the ooECCI image (Fig. 1b). The microstructure in TEM is very complicated, and thus the stacking fault fringes (ribbons) are profuse and curved (Fig. 4). The dark field TEM image with excitation error, s = 0 (Fig. 4c) was better to visualize the stacking fault ribbons than weak-beam dark-field TEM image (Fig. 4a and b). The defects might be a super-partial dislocation and its movement can drag antiphase domain boundaries (APBs) between partial dislocations. Current TEM characterizations using diffraction vectors g = 400 and 444 indicated that all the dislocations are visible and curved and elongated along the 01¯1 plane. The stacking fault ribbons can be compared with the traces of (100) and {111} planes on the 01¯1 plane (Fig. 4c), but the curved nature of this stacking faults would represent a non-crystallographically controlled motion of the dislocations (Smith, 1982).

The dislocation microstructures might be also related to high water activities in the deformation. Some dislocations interact with strain-free bubbles like deformed wet-quartz (McLaren et al., 1989) or precipitates (Voegelé et al., 1999) (Fig. 4b). Although we did not investigate water activity on the natural garnet by infra-red spectroscopy, high water activities in pyrope-rich garnet under high-pressure and high-temperature conditions were reported by Mookherjee and Karato (2010). The dissociated <110> dislocations into 1/n <110> partial dislocations with widely spreading stacking fault ribbons were also reported in low-temperature eclogitic garnets with some potential water activity (Voegelé et al., 1998). The unusual stacking fault ribbons might be related to low-stacking-fault energy in hydrous and high-temperature-deformation conditions of a natural almandine garnet (Smith, 1982; Smith, 1985). Further characterization of the complex dislocation microstructure sheds light on the deformation mechanisms of pyrope-rich garnets in the Earth's mantle.

For further application, we tested the ooECCI method with an iron-free forsterite from the deformation experiments at high pressure and high temperatures (Bollinger et al., 2015). Some of the forsterite grains with internal sub-grain boundaries are clearly observed in a BSE image at low magnification (Fig. 5a). An ooECCI image of one of the grains displayed individual dislocations (Fig. 5b). The visibility of individual dislocations in forsterite strongly supports potential applications of ECCI into iron-free high-pressure polymorphs. Note that ECCI of dislocations in forsterite denies the necessity of iron decollation and enrichment on dislocations due to pipe diffusion and Cottrell atmosphere in natural processes, when we try to observe individual dislocations in the mantle olivine by using ECCI. The ooECCI of individual dislocations in iron-bearing minerals (Miyajima et al., 2018) is not a composition-dependent BSE imaging, but an orientation-contrast microscopy.