Development of image processing systems for cosmic ray effect analysis

Abstract

Cosmic neutron radiation in the atmosphere is a harsh environment for modern aircraft electronics. High-energy neutron interactions with silicon produce spurious charge which can cause malfunctioning or even failure to advanced microelectronic devices. To study neutron interactions, a CCD based neutron monitor is developed which has found applications in neutron field characterisation and neutron event classification. The neutron monitor was irradiated in a
variety of neutron fields to be able to analyse high-energy neutron effects.
In order to reach the best performance of the neutron monitor three CCD sensors are evaluated for neutron detection. In particular CCD pixel size and anti-blooming structures are investigated. A CCD with 9 pm pixel size without hardware anti-blooming structure is chosen as a baseline device. Anti-blooming correction is performed by a post-processing algorithm which recovers the original shape of the event distorted by blooming. The algorithm shows that the blooming effects are amendable with the preservation of all deposited charge.
The main application of the neutron monitor is the characterisation of neutron facilities. The charge collecting capabilities of the neutron monitor allow the measurement of neutron fluence and the characterisation of neutron fields in terms of charge inducing capabilities and event spatial characteristics. The effects of dissimilar neutron spectra in two pseudo-atmospheric
neutron beams are demonstrated. The CCD data are used to measure neutron fluence at different locations along and across beam lines. Calibrating the detector against a suitable neutron beam allows it to be used as independent dosimetry in any neutron field. The neutron scattering measurements show that monitoring of the beam is of greatest importance when multiple experiments are irradiated simultaneously.
To exploit the imaging capability of the neutron monitor, it is applied to classify neutron events according to spatial and charge characteristics. Two feature spaces are used to do this, one that assigns events to clusters according to charge density and one that is sensitive to the spatial spread of charge within the event. The method provides useful statistical results which are directly applicable for example during radiation hard device design.