From geometry to dynamics of microstructure: using boundary lengths to quantify boundary misorientations and anisotropy

The microstructure of a quartzite experimentally deformed and partially recrystallised at 900 °C, 1.2 GPa confining pressure and strain rate 10~ /s was investigated using orientation contrast and electron backscatter diffraction (EBSD). Boundaries between misoriented domains (grains or subgrains) were determined by image analysis of orientation contrast images. In each domain, EBSD measurements gave the complete quartz lattice orientation and enabled calculation of misorientation angles across every domain boundary. Results are analysed in terms of the boundary density, which for any range of misorientations is the boundary length for that range divided by image area. This allows a more direct comparison of misorientation statistics between different parts of a sample than does a treatment in terms of boundary number.

The strain in the quartzite sample is heterogeneous. A 100 x 150 Am low-strain partially recrystallised subarea C was compared with a high-strain completely recrystallised subarea E. The density of high-angle (>10°) boundaries in E is roughly double that in C, reflecting the greater degree of recrystallisation. Low-angle boundaries in C and E are produced by subgrain rotation. In the low-angle range 0–10° boundary densities in both C and E show an exponential decrease with increasing misorientation. The densities scale with exp( — 9/X) where 1 is approximately 2 ° in C and 1 ° in E; in other words, E has a comparative dearth of boundaries in the 8-10° range. We explain this dearth in terms of mobile high-angle boundaries sweeping through and consuming low-angle boundaries as the latter increase misorientation through time. In E, the density of high-angle boundaries is larger than in C, so this sweeping would have been more efficient and could explain the relative paucity of 8-10° boundaries.

The boundary density can be generalised to a directional property that gives the degree of anisotropy of the boundary network and its preferred orientation. Despite the imposed strain, the analysed samples show that boundaries are not, on average, strongly aligned. This is a function of the strong sinuosity of high-angle boundaries, caused by grain boundary migration. Low-angle boundaries might be expected, on average, to be aligned in relation to imposed strain but this is not found.