Abstract
The main aim of this study was to show that scattered radiation in the backward direction can be used as an additional source of material-specific information about a sample measured simultaneously to a primary method in the forward direction, e.g. X-ray diffraction imaging. Many applications for material and structural discrimination employing X-ray scattering and absorption collimate the emitted radiation in a defined direction. Scattered radiation into other directions often remains disregarded or is shielded. The major physical effects characterising the backward-scattered spectra are, besides attenuation, Compton shift and material-specific Doppler broadening of the photon spectrum in the sample. The shape of a Compton-broadened peak can reveal information about the electron momentum of an element or, even to some extent, of a chemical compound. Since most applications utilise polychromatic X-ray sources, the focus of this work was the analysis of Compton scattered characteristic anode peaks from sample materials. The multiplet structure of these characteristic peaks superimposed to a broad Bremsstrahlung spectrum is challenging for peak-shape analysis. The initial part of this work was the design and characterisation of an experimental set-up for measuring energy-resolved Compton scattered spectra of sample materials. Subsequently different calculation models were developed for describing the broadening of the characteristic lines of the anode material. These models were adapted to the specific experimental conditions. Essentially, two types of methods were considered. The first calculation procedure evaluated scattered spectra based on tabulated material-specific Compton profiles. The second model was a phenomenological approach. It fitted a function consisting of Gaussian curves above a linearly approximated background to the curvature of the 2nd derivative of the Compton spectrum. These models were experimentally validated on Compton profiles of a variety of sample materials containing period 2 and period 3 elements. It was proven that in principle comparing measured spectra with calculated spectra provides high material differentiation capabilities, but for most molecules tabulated Compton profiles are not available and the independent atom approximation causes deviations. The phenomenological method employed extracted Gaussian curve fit-parameters to distinguish measured materials quantitatively. Most of the samples could be distinguished from each other based on their profile structure. A principal component analysis of corrected spectra provided an alternative approach and showed capabilities to separate scatter-spectra of different materials. In principle, more sample-specific information valuable for material separation could be extracted as compared to a simple integration method of backscattered radiation.