Compact Soft X-Ray Microscopy: Image Processing and Instrumentation

Sammanfattning: Soft x-ray microscopy is a powerful technique for natural-contrast, high-resolution imaging of organic materials. This Thesis describes new instrumentational and new image-processing methods to improve the image quality of the compact x-ray microscope at the Biomedical & X-Ray Physics division at KTH. The microscope is based on a laser-plasma source combined with different condenser optics, either multilayer mirrors or zone plates. Imaging is performed by micro zone plates. The microscope works in the water window (\lambda = 2.3-4.4 nm), where the attenuation lengths of oxygen and carbon differ strongly, providing high natural contrast for carbon-containing specimens in an aqueous environment.By optimizing the properties of the laser-plasma source and fabricating multilayer mirrors with high, uniform reflectivity, the performance of the microscope's illumination system could be improved and exposure times decreased significantly to about 2 min for imaging dry samples and 5 min for imaging wet samples. For imaging of wet samples, a wet-specimen chamber was developed, which is vacuum-compatible. Since it is horizontally mounted in the microscope, it offers advantages for investigations in polymer and soil science.To improve the quality of images taken by the compact x-ray microscope an image-restoration algorithm was developed. Denoising is performed by a filtering algorithm based on the discrete wavelet transform. This algorithm shows advantages compared to Fourier-based algorithms, since the filtering of spatial frequencies is done locally. An improvement in exposure time by a factor of about 2 could be realized without loss of image information.To stimulate experiments on functional imaging in x-ray microscopy an image-analysis algorithm for identifying colloidal-gold particles was developed. This algorithm is based on a combination of a threshold with respect to the local absorption and a shape discrimination, realized by fitting a Gaussian profile to the potential particles. The algorithm was evaluated and optimized on images taken by the transmission x-ray microscope at BESSY II. The size-selective identification and localization of single gold particles down to a diameter of 50 nm was demonstrated.