A transmission electron microscopy study of experimentally deformed quartzite with different degrees of doping
Résumé
TEM Study of Heavitree quartzite deformed at high temperatures and a pressure of 1.5 GPa with three different preparations: water-added, vacuum dried, and sodium doped show a wide range of dislocation microstructure. The vacuum dried sample has a very heterogeneous dislocation distribution varying from almost no dislocations to very high densities. Dislocation nucleation occurs by cross-slip mechanisms, such as Frank-Read sources and Orowan loops around hard inclusions. Shear bands develop in the basal plane composed of two closely spaced shear planes. Dislocations have straight segments and they are aligned along directions of dense packing, suggesting strong Peierls stress control due to the crystal structure. No voids, bubbles or dislocation walls where observed in the vacuum dried sample. The sodium-doped sample had a homogeneous high dislocation density. The water-added sample has a significantly lower, relatively uniform dislocation density. Bubbles associated with dislocations are seen in every grain. Dislocations were either present in loops around bubbles or as straight dislocations with small bubbles spaced along the dislocation line. Some of the long straight dislocations are aligned parallel to the c-axis. Sub-grains where very frequent, sometimes form cells of about 2 p,m in diameter with straight edges. Analysis of dislocations show 1/3 < a > and [c] dislocations in sub-grain walls and free 1/3 < a > dislocations on prism planes. There is clear evidence for dislocation climb with sub-grain walls, dislocation cells, dislocation junctions and dislocation debris of small loops. The major difference between vacuum dried and water-added samples are the homogeneity of the microstructure and evidence for climb in the water-added sample. Glide is clearly difficult, with a high Peierls stress in the vacuum dried sample as shown by areas of very low and very high dislocation density and the crystallographic control of dislocation line direction. The sodium-doped sample indicates that nucleation was easy by the high homogenous dislocation density, but no climb recovery has taken place. The density is high and the dislocations are strongly interacting causing tangles; no evidence for crystallographic control can be observed at these levels of strain. The activation energy for creep in Heavitree quartzite decreases with inferred water content of the specimens form 185 kJ mol(-1) for vacuum dried to 151 for 0.4 wt% water-added samples. Analysis of diffusion data for oxygen under hydrothermal conditions and inferred diffusion data for the hydrogarnet defect and dislocation velocity suggests activation energies for these processes are similar to the activation energy for the dislocation creep of Heavitree quartzite and other quartz aggregates. For data from the previously published experiments with the highest stress resolution there is a correlation between the A pre-factor in the power law creep equation and the activation energy for creep. It is speculated that the correlation may be due to variable hydrogarnet defect concentrations.