High Count Rate X-ray Detection for Liquid Threat Identification Applications

Abstract

This work presents the development of an x-ray detector system for the multispectral detection of x-rays used in a Bottle Scanner. This system is designed to identify liquid explosives concealed within bottles taken onto aircraft. The Bottle Scanner works by calculating the transmission spectrum for a given bottle and its contents. This is then used as a fingerprinting technique in order to identify the presence of liquid explosives.

The main focus of this work is the design of the detector signal chain and particularly the pulse height detection algorithms for the acquisition of 32 bin x-ray spectra. The pulse height detection algorithms are implemented on a field programmable gate array (FPGA). The performance of such algorithms at high count rate is a key requirement for this application. Four such algorithms are discussed in this work, each with varying complexity and different approaches to pile up handling.

The algorithms are simulated using a Monte Carlo approach where the random arrival rate of photons at the detector is modelled. Algorithms are then emulated using an Agilent DSO90254A oscilloscope before finally being programmed onto an FPGA and tested on a real system. The transmission curves produced in real Bottle Scanner systems show a deviation from those predicted by the underlying physics and exhibit a rollover at high energies. This is shown to be due to pulse pile up effects which are explored in detail.

Depth of interaction effects are also investigated experimentally and by simulation using the Geant4 software package. The results of this are used to design a biparametric type algorithm which is capable of simultaneous pile up rejection and depth of interaction correction.