| dc.description.abstract | Carbonates may be incorporated in the mantle at collisional plate boundaries by many processes, including subduction of weathered oceanic lithosphere and fault-bounded tectonic slivers of seamounts. Once magnesite is formed in subducting slabs, it is likely to remain as an important carbon-bearing phase, as its stability extends over a wide range of mantle conditions. High-magnesium carbonates, including magnesite and dolomite, have been observed in ultra-deep metamorphic collisional terrains and in mantle xenoliths. Recent deformation experiments of magnesite indicate that it is weaker than mantle phases such as olivine and that it is likely to affect the geodynamics of subduction. Microstructures and lattice orientations of experimentally deformed samples have been investigated in this study, using optical and electron microscopes and electron backscatter diffraction. Microstructures of samples deformed at lower temperatures (400-600 ˚C), at strain rates of 10^-4-10^-6 s^-1 and an effective pressure of 900 MPa are dominated by undulose extinction, kink bands, and mechanical twins. At higher temperatures (>750 ˚C), microstructures include undulatory extinction and extensive dynamically recrystallized grains, subgrains, and core-and-mantle structures. Mechanical twins are not observed at these higher temperatures. External rotations of magnesite c- and a-axes show that dislocations glide on the c, r, and f planes by slip systems that are similar to those reported for calcite and dolomite. Additionally, magnesite mechanically twins on the e- and f-planes. The CRSS values for magnesite twins exceed 300 MPa, significantly greater than twinning CRSS reported for either calcite or dolomite. I propose that twinning CRSS in the carbonate system is controlled, not by crystal class or cationic ordering, but by cationic size, which affects CO3^2- anion-anion repulsion at twin boundaries. This hypothesis is consistent with lowest energy calcite twin modeling of Bruno et al. (2010). In a similar fashion to twinning, I predict that dislocation movement is also controlled by cation size. Lastly, I observe that this effect is not limited to carbonates. Cationic size may also affect CRSS values for twinning in oxides and the strength of olivine with different Fe and Mg contents deforming in the dislocation creep regime. | en |
