High Resolution Three Dimensional X-ray Microscopy Simulations & Experiments

Deformation processing of metallic materials is one of the key routes for the production of industrial and consumer goods. The control of the microstructural evolution is essential for optimization of the materials’ service performance. Therefore, the characterization of the deformed microstructures has a key role to play in the understanding and subsequently the improvement of these materials.

Optical and electron microscopy techniques have been the dominant methods for mapping microstructures. However, these methods probe the (near) surface only. For these methods, 3D information can be reached by serial sectioning, but the destruction of the sample limits their use to static studies. Owing to their high penetration, Xray microscopy techniques offer an alternative non-destructive route for volumetric characterization of materials.

The development of three dimensional X-ray diffraction (3DXRD) microscopy created a new paradigm for 3D mapping of grain ensembles with spatial and angular resolutions of 2 − 10 μm and 0.01 − 1°, respectively. The reconstructed data provide a detailed 3D-EBSD type map of the microstructure, identifying the local orientation, the spatial position and shape of grains and their lattice parameters.

Deformation microstructures are hierarchical structures that spans from μ m-sized grains to atomsized crystal defects. Specifically, the deformation transforms an ideally strainfree grain structure into a banded microstructure of ≈1 μmsized crystallites called cells, or subgrains. Furthermore, the structure is locally heavily textured.

Therefore, a high angular resolution is required for differentiating the individual subgrains. Existing 3DXRD methods fail to meet both the requirement to spatial and angular resolution.  In order to enable in situ 3D mapping of deformed microstructures within representative volumes, in this thesis, I present work towards establishing a new high resolution modality for 3DXRD, HR3DXRD, with a 10 times superior spatial and angular resolution.

HR3DXRD is based on several concepts, that are novel in connection with 3DXRD. • Diffraction information acquired at an intermediate sampletodetector distance, L , • The use of a large (virtual) compound detector with a relation between         pixel size and field-of-view (FoV) that is much larger than what is commercially available,  • 3D mapping of submicrometer objects by means of a tessellation algorithm, requiring only the registration of the center-of-mass (CoM) and integrated intensity of diffraction spots.

Full scale numerical simulations of HR-3DXRD on phantoms representative of deformed microstructures reveal that diffraction spot densities are not prohibitive, and that indexing is possible. Moreover, while much outside of the range of applicability they were created for, existing 3DXRD indexing algorithms can be adapted to HR3DXRD. The resulting 3D maps are space-filling and exhibit a spatial and angular resolution of 0.1μm and 2×10−5 degrees.

The main limitation is shown to be the signal-to-noise ratio. A main challenge of HR3DXRD is the generation of the compound image, from a series of partial diffraction images, acquired at different positions of the 2D detector. It is shown that pattern recognition and stitching algorithms can provide the required subpixel accuracy.  Five experimental tests have been performed, all using ad hoc setups in four different grain-mapping beamlines.

The main sources for degradation of the quality of mapping are identified and remedies suggested. Preliminary results are presented for indexing and grain maps on one of the systems.