Analysis of nuclear quadrupole resonance signals from mixtures

Abstract

Nuclear quadrupole resonance (NQR) is a solid-state radio frequency (RF) technique, able to detect and identify compounds containing quadrupolar nuclei, including many high explosives, narcotics and pharmaceutical compounds. In addition to being able to identify entirely different compounds, the technique is also able to distinguish between different polymorphic forms of the same compound. Analysing the signals from mixtures, either of different compounds or of polymorphs, is important in several applications. Being able to exploit the signals from the different components of a mixture is important in the detection of explosives, whilst quantification of such components is important in several pharmaceutical applications. In this paper, we propose two hybrid detectors that can exploit the signals from multiple components, offering improved probability of detection, as compared to recently proposed detectors. The algorithms also provide estimates of the relative proportions of the components, as well as estimates of other important NQR signal parameters. The algorithms are evaluated on both real and simulated data. The former is measured from a sample of trinitrotoluene which contains at least two polymorphic forms with rather different NQR properties.