1 Risø National Laboratory for Sustainable Energy, Technical University of Denmark2 Department of Applied Mathematics and Computer Science, Technical University of Denmark
In the present thesis the development of a unique experimental method for volume characterisation of individual embedded crystallites down to a radius of 150 nm is presented. This method is applied to in-situ studies of recovery in aluminium. The method is an extension of 3DXRD microscopy, an X-ray diffraction technique for studies of the evolution of grains within polycrystalline materials. The much smaller volume of the crystallites of interest here in comparison to grains implies that the existing method is not applicable due to overlap of diffraction spots. In this work this obstacle is overcome by the combined use of X-ray micro focusing optics, new scanning algorithms and the use of foils. The ratio of foil thickness to crystallite size should be at least 10 such that the central ones are situated in a bulk environment. To avoid thermal drifts, gold reference markers are deposited onto the sample. The X-ray fluorescence from these markers defnes the position of the crystallites with respect to the beam to within 1 ¹m. Two types of data analysis approaches have been developed. The first one generates apparent size distributions of an ensemble of crystallites. These may be converted to true size distributions by stereological tools. Uniquely, this method enables in situ studies of the evolution in size distribution - at a specific sample location - with good statistics (5000-20000 per 20 minutes). The second approach generates growth curves (volume vs. time) of individual crystallites. This involves at all times 1) separating a given diffraction spot from neighbouring spots originating from other crystallites and 2) measuring the complete integrated intensity of the spot (as this is related to volume). This image analysis problem is formulated in a 5D observational space, where growth curves are represented as strings. To identify the strings a combination of a 5D connected component type algorithm and multi-peak fitting was found to be superior. The first use of the method was a study of recovery of a deformed aluminium alloy (AA1050). The aluminium alloy was deformed by cold rolling to a thickness reduction of 38%. The sample was annealed at 300±C for 3 hours. From the statistical analysis of the size distribution most of the recovery was found to occur during the first 3 minutes of annealing. Growth curves are presented for nine individual subgrains. A difference is observed between these experimental data and predictions from curvature-driven grain growth models. The observed individual subgrains showed no evidence of rotation. In outlook, several synchrotrons are presently developing nano-X-ray beams. Applying the methodology developed in this thesis to these beams will enable in-situ studies of the dynamics of bulk crystalline nano-structures down to the scale of »20 nm.