The macroscopic strength of metals is determined by the dislocation arrangements that are formed when dislocations slip in the crystal lattice in response to the applied stress. Despite the extensive research carried out, the transition from uniform to non-uniform dislocation structures is not yet fully understood. This information is however essential to support the development of computational models that aim to predict dislocation patterning. Lattice rotation caused by the presence of dislocations is, for instance, a parameter that is often not taken into account in computational models although it plays a role in the development of dislocation ensembles. Experimentally following in-situ dislocation patterning including involved lattice rotations at lengthscales comparable to those in simulations is not straightforward. In fact, the current available techniques that have the necessary spatial and angular resolution are either destructive (3D-EBSD) or very time consuming (3D X-ray Laue diffraction). In this dissertation we present a new experimental approach that allows following lattice rotation and dislocation ensembles time-resolved during deformation. The technique is based on X-ray Laue diffraction scanning in transmission mode, which provides a sub-micron spatial resolution in 2D and statistical information on orientation spread in the third dimension. The in-situ mechanical tests are performed at the microXAS beamline of the Swiss Light Source Synchrotron. The method has been applied to understand dislocation pattering of copper single crystals occurring in the earlier phases of low cycle fatigue. Samples with different crystal orientations (single crystal, coplanar and collinear) have been cyclically deformed up to a maximum of 120 cycles. A dedicated miniaturized shear-fatigue device compatible with Laue diffraction has been constructed for that purpose and a sample preparation based on picosecond laser ablation has been developed. The developed procedure analyzes the evolving dislocation microstructures in terms of lattice rotation, lattice curvature and geometrically necessary dislocation densities at lengthscales similar to those addressable in computational models. It sacrifices on the fully 3D aspect but it brings the time resolution required to validate ongoing simulations. It has been shown that the error made by integration depends on the dislocation network expected. For instance, the technique can be used to validate 3DDD models or density based dislocation dynamics models when applied in deformation geometries where the formation of 2D dislocation patterns are expected (e.g. vein-channel structure in single slip oriented fatigued samples). The greatest advantage of this technique is that the evolving regions are easy to follow due to the high rotational sensitivity and time resolution of the method. What is more, the influence of the initial microstructure can be evaluated and quantitative values can be provided. On the other hand, the principal drawbacks are related to the gauge thickness of the samples. Even if the technique can estimate the orientation spread along the third direction, the approach cannot physically resolve the integrated signal along the thickness. This is source of uncertainty when providing actual values of lattice curvatures and geometrically necessary dislocation densities. Another shortcoming is the sensitivity of the method to the energy distribution of the X-ray beam provided in the beamline, which determines the collected signal - the principal basis of the subsequent analysis and interpretation.
EPFL2016
