The development of x-ray backscatter imaging systema through simulation

X-ray backscatter has applications in defence and security, medical imaging, astrophysics and industry. The development and testing of X-ray backscatter imaging systems can be achieved not only by experiment, but also by using Monte-Carlo modelling. The PENELOPE simulation package was chosen for its versatility and transparency. However, PENELOPE is a radiation transport package that is not user-friendly, is not inherently compatible with parallel processing, and is not equipped with the facility to process output data in a way that replicates the output from imaging plates or energy dispersive detectors. Tools called PENMAT and PAXI were written in MATLAB to extend the capability of PENELOPE and so enable the efficient exploration of X-ray backscatter imaging which is the focus of this study. The enhanced PENELOPE suite was used to model a real thermionic source to validate the process by comparison with experiment, and model virtual sources suitable for exploring fundamental principles of backscatter. Virtual sources were conceived and designed to efficiently characterise various imaging system features. These include mono-directional and mono-energetic sources (to isolate energy dependant scattering cross sections), flat spectrum sources (to objectively characterise transmission through mask materials) and thin ‘wire form’ sources (to simultaneously characterise the spatial resolution and field of view of X-ray optics). A process of using virtual detectors to feed the input of virtual sources was used to shortcut the repeated computationally expensive modelling of a thermionic tube. With this efficient process and parallel computing, various combinations of pinhole and Coded Aperture optics could be efficiently tested and compared. To enable systematic comparisons the image quality metrics of signal, noise, contrast, resolution, field of view etc. are identified and procedures developed to extract them from images. ii For the experimental energy range of likely practical use, it was found that pure tungsten masks were superior to other alloys studied and that a 2mm pinhole gave the most generally suitable resolution/signal compromise. The results were consistent with physical experiment. A range of Coded Apertures were also modelled and compared favourably to experiment. The pinhole work on field of view informs the envelope within which coded apertures could avoid partial coding. The HEXITEC energy dispersive image plate was used to collect experimental images from a multi material quadrant. The image was simulated accurately using PAXI. Further, modelling with PAXI allowed the distinct interaction processes giving rise to image characteristics to be isolated. This concept was extended with a unique and innovative 2π hemispherical detector, which efficiently captured backscatter X-rays from carbon, copper, manganese dioxide, and lead when shielded and unshielded. This process allowed the brightness of materials to be studied, as governed by the complex combination of attenuation and cross section with angle. Further, the relative contributions from Compton, elastic and fluorescent processes to image brightness and spectral features could be isolated and compared with angle. This was conducted with/without shielding. This cannot be achieved by experiment, and pilots how modelling can inform the best beam energies and detector angles where the backscatter X-rays contain the right information to characterise materials and structures. This work includes significant use of simulation and also a strong supporting element of physical experimentation. The development of modelling techniques and their exploitation can give information that physical experiment cannot, whilst experimentation has been shown to validate the use of simulation and identify some limitations