A microstructure reconstruction approach for three dimensional x-ray diffraction microscopy.

摘要

Three dimensional X-ray diffraction microscopy (3DXDM) is a set of techniques that can be used to explore the internal structure of polycrystalline materials in a non-destructive way. The microstructure mapping measurements of interest here are carried out at third generation synchrotron sources using high energy, focused x-rays, precision sample positioning hardware, and a high resolution area detector to image diffracted beams. The line focused beam illuminates a planar section of the sample and diffraction patterns are imaged at multiple sample-to-detector distances. The third dimension is probed by translating the sample and measuring successive layers. A variety of reconstruction algorithms for inverting observed data to deduce the microstructure that generated it have been developed but no quantitative maps of microstructure have yet been extracted. In this thesis, a software package that implements a new algorithm is presented and tested. The package uses image analysis for noise reduction and to identify and correlate diffraction spots observed at different sample-to-detector distances. Diffraction spots are back projected into the sample plane. Crystallographic orientation and grain shape are determined from multiple diffracted beams. An area filling growth algorithm determines the edges of individual grains. The program allows fully automatic analysis but also provides for user interactions during the process.; Several verification tests are presented. First, simulated data are generated for a simple but non-trivial microstructure and these data are used as input to the program. The software reproduces the correct microstructure within detector digitization errors. We use this test case to study the dependence of reconstruction errors on various mechanical translation stage imperfections, both systematic and statistical. Next, several data sets measuring successively more complex samples are tested. Data were collected by our group and collaborators at beamline XOR-1ID at the Advanced Photon Source. Samples include (i) a single crystal of silicon, (ii) a 20mum diameter gold wire, (iii) an aluminum alloy, Al1050, polycrystal, and (iv) a high purity aluminum polycrystal. While aspects of (i)-(iii) are well reconstructed, the soft high purity sample had broad mosaic structure that prevented the image analysis from identifying distinct diffraction spots.; The results described here show that direct inversion of 3DXDM data can be accomplished and can yield important microstructural information. The software presented is, as far as we know, the most successful to date at doing this type of reconstruction.