Self-Absorption Corrected X-Ray Microtomography with Structural and Elemental Contrast

X-ray fluorescence tomograms of large or highly absorbing samples often suffer from self-absorption effects. For qualitative and quantitative results, it is necessary to correct these effects.In this thesis, a fully 3D self-absorption corrected reconstruction algorithm for X-ray fluorescence tomography datasets is presented. As an extension of the tomographic maximum-likelihood expectation-maximization algorithm, it is an iterative algorithm that reconstructs the density volumes of all measured elements in parallel. The algorithm includes multiple optional parts to aid its convergence, including density and background constraints. Noise may be reduced by using a priori information of the sample to introduce a weighted penalty term. All calibration parameters are derived from experimental measurements and small inconsistencies in those values may be corrected by a rudimentary adaptive calibration. Parallel calculations on CPU and GPU as well as downscaling approximations are used to assure high performance of the algorithm and to allow for online reconstruction during experiments.Numerical simulations of multi-element phantoms without and with added noise confirmed the qualitative as well as quantitative properties of the self-absorption correction. Additionally, tomographic STXM and XRF measurements of a barred olivine micrometeorite were performed at beamline P06 of the PETRA III synchrotron storage ring at DESY in Hamburg, Germany. Although the algorithm was able to correct parts of the self-absorption effects in the micrometeorite data, the qualitative and quantitative properties of the final reconstructions were severely limited by inconsistencies in the experimental calibration, mainly the deadtime and pile-up corrections