High resolution chemical mapping via scanning transmission x-ray microscopy
Abstract
Scanning transmission x-ray microscopy (STXM) combines imaging at sub-50 nanometer spatial resolution in the soft x-ray region with x-ray absorption spectroscopy at an energy resolution of 0.1 eV and better. The sensitivity to light elements and their chemical bonding states in the near-edge x-ray absorption region together with 2D microscopy allow to map out the chemical composition of microscopic samples. Over the past decades STXM has been used by scientists from various fields such as material sciences, biology, earth and environmental sciences. This thesis describes the upgrade of the STXM microscopes at the National Synchrotron Light Source in Brookhaven National Laboratory. The now 5 th generation was a retrofit of the existing microscopes with a three axes laser interferometer system for improved relative sample positioning. Included in the same project were a hardware change towards faster scan positioning and data acquisition electronics as well as a new software concept allowing for remote control of the microscopes. We also present the use of non-negative matrix factorization to analyze spectromicroscopic data sets. Since in many specimens of interest in biology or related fields a number of spectral signatures may be unknown, thickness maps of the chemical composition of a sample cannot be calculated directly. A previous method used principal component analysis to orthogonalize and noise filter spectromicroscopy data, followed by cluster analysis as a form of unsupervised pattern recognition to determine pixels with spectroscopic similarity. We tried a non-negative matrix factorization algorithm to find a parts-based representation of the same data. The non-negativity constraint allows for a physical interpretation of component spectra and thickness maps. The STXM technique was utilized to investigate the composition of human sperm samples. We describe why the analysis of human sperm via x-ray spectromicroscopy can provide valuable information and what the results of our efforts were.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- 2008
- Bibcode:
- 2008PhDT........80F