Abstract:

The use of Gamma-ray Compton backscattering as the main tool to construct an imaging system able of identifying landmines buried in soil has proven very promising in its first prototype developments. These prototypes take advantage of the back-to-back positronic 511 keV γ-decay. One of the gammas is used to tag the backscattering signal from the other gamma-ray. By sampling coincidence events, an image is built reflecting mainly the distribution of the electron density of the matter placed in front of the prototype device. Laboratory and field tests have shown that although an image of enough quality to guess the presence of objects buried into the soil is achieved, closer characterization of the potential mine is hampered by difficulties related to our rather poor knowledge of two main processes: (i) the γ-soil interaction, and (ii) the effects that the different hardware pieces and software procedures have on the final image.

Regarding the first point, the problem is that although the Klein-Nishina cross-section predicts the result of a one-step γ-electron interaction, the result of such a complex process as a multi-step collision of a γ-ray with a target as complex as the multielement, macroscopic soil is very difficult to predict. We have performed both experiments and Geant4 simulations, focused on understanding two processes: (a) the effect of soil composition on the tail of the 511 keV quasi-gaussian peaks, due to low-angle Compton dispersion. (b) Characterization of the backscattering spectrum with the aim of obtaining a prediction of its shape and a quantitative evaluation of the absolute backscattering probability.

For the point (ii) above, we simulate images whose quality can be tuned by hand. These images are then processed by especially designed software with the purpose of extracting physical dimensions: approximate shape, dimensions and eventually an estimation of the depth of deployment. By tuning the input image’s quality we can distinguish the effect of the different software steps on the final image.