X-ray backscatter is currently used in a wide range of imaging applications for noninvasive inspection, such as industrial inspection and portal security, where it is used to scan shipping containers, vehicles, and people.1, 2 In single-sided imaging the x-ray source and detector are on the same side as the object being imaged. This allows imaging to be conducted in situations where two-sided transmission x-ray imaging is not feasible (e.g., when detector access behind or inside an object is impossible without intervention). There are several issues, however, that limit the efficacy of this single-sided methodology.
Conventional x-ray backscatter systems have limited stand-off distances because of the x-ray intensity loss that occurs with increasing range (which follows an inverse square law). Unless x-ray backscatter systems are operated in combination with computed tomography techniques, they can only be used to provide 2D images. This requires two-sided transmission radiography, which presents further practical limitations. Furthermore, traditional x-ray backscatter techniques use 10–100s of x-rays (at kiloelectronvolt energy levels) to increase the fraction of backscattered x-rays.3 At these energies, however, the x-rays are highly attenuated by the sample, which causes a reduction in the maximum imaging depth that can be achieved.4
We have recently developed a new x-ray backscatter approach in which we use an ultra-short pulse of highly directional electrons. These electrons are able to penetrate to significant depths within a sample.5 The transmitted electrons emit bremsstrahlung (‘braking radiation’) x-rays in the sample. This radiation is subsequently backscattered and detected to provide ranging information (see Figure 1). We can vary the electron energy that we use so that the x-ray backscatter intensity at depth is optimized. In this work, we have conducted a series of simulations for this approach and have compared the results with our experimental data.6
Laser wake field acceleration experimental setup for the generation of backscattered x-rays. To produce a highly directional relativistic electron beam, a 55fs laser pulse (1019
) is focused into a helium (He) and 5% nitrogen (N) supersonic gas jet. The electron beam propagates out of the vacuum chamber through a 3mm-thick steel window into air and then to samples that are positioned 0.4–1.5m from the end of the chamber. The electrons produce bremsstrahlung x-rays in the imaging target, some of which are backscattered, and subsequently detected and temporally resolved by a multi-channel plate photomultiplier tube (MCP-PMT) 2.2m away from the sample. The MCP-PMT is placed within a lead shielding to restrict the field of view and suppress unwanted background x-ray emission.5
f: Focal length.
In March 2014 the Astra Gemini laser facility at the Rutherford Appleton Laboratory was used to demonstrate the feasibility of our stand-off 3D x-ray imaging approach. For these tests the laser wake field acceleration (LWFA) technique8 was used to generate a 140MeV electron beam. The resulting ‘x-ray radar’ image that we obtained during the experiment is shown in Figure 2(b).5 To produce this x-ray image, 40 laser pulses—each producing 4×109 source electrons—were used to acquire range information for each of the 11 horizontal scan positions. At each scan position, a specific configuration of objects in the scene was interrogated: see Figure 2(a). The backscattered x-rays were time stamped and plotted against position to produce the image.
(a) Experimental layout of imaging objects during stand-off 3D x-ray imaging tests. (b) Experimentally obtained x-ray backscatter image (consisting of 11 horizontal scan positions). The objects with different mass densities and atomic numbers are represented in this image, i.e., 38mm-thick aluminum (Al), 0.14m-thick insulation foam, and a low-density organic compound.5
Arb: Arbitrary units.
To further understand the results of our experiment, we used the GEANT 4 high-energy physics modeling package to simulate the experimental setup shown in Figure 1. In our model, we simulated the effect of firing a short (less than 100fs) burst of electrons into the experimental geometry. Our model also included the vacuum chamber, sample objects, lead collimator around the detector, as well as the surrounding concrete walls, floors, and ceiling. With GEANT 4 it is not possible to directly model the non-linear dynamics associated with laser/plasma interactions. Our simulation, therefore, begins at the point of electron emission from the plasma. For simplicity, we assumed a mono-energetic electron beam (based on measurements made during the experiment). We randomly assigned the directions of the electrons within a predefined cone beam, which had a divergence and beam spot size that were representative of our experimental measurements. An idealized detector volume was also placed in the simulation, at the same position and angle as the detector that was used in the experiment. The detection area for our simulation was 100cm2, whereas a 1cm2 detector area was used in the experiment. This increased area enabled the simulation of fewer source electrons compared with the number in the experiment, and thus reduced the simulation time. To recreate the image shown in Figure 2(b), we simulated 107 source electrons and the subsequent backscattered x-ray photons were time stamped for each of the 11 horizontal positions.
The resulting x-ray radar image from our simulation is shown in Figure 3. This image is broadly similar to the experimental image (Figure 2), but with some additional artifacts. A low-intensity response from the foam object is observed, as well as a response from the aluminum at about 17ns and with the low-density organic compound at about 15ns. Both the aluminum target and the low-density organic compound appear to have trailing artifacts that follow their main responses. The large signal at about 23ns is from the simulated lead wall, followed by a delayed secondary response. We have investigated the low-density organic compound target in more detail so that we can identify the source of the observed delayed responses. The effect of different object materials, within air or a vacuum, on the observed imaging artifacts is illustrated in Figure 4. We find that the main contributions of the secondary scattering in the simulation—see Figure 4(a)—originate from the concrete walls, floors, and ceiling. The use of a lead collimator in the model—see Figure 4(b)—suppresses the secondary artifacts. Our results—see Figure 4(c) and (d)—also indicate that the weak delayed signals are associated with scattering effects from surrounding air molecules. To explain these results, we suggest that following the interaction of the electron beam with the sample, some of the bremsstrahlung-generated photons undergo secondary scattering by air molecules at random ranges.
Simulated x-ray radar image of the entire test object scene (see Figure 2
), modeled with a lead collimator around the detector. The image shows a low-intensity response from the foam wall, which is associated with the low-density organic compound response at about 15ns and with the aluminum response at about 17ns. It is then followed by the response from the rear lead wall. The image is shown on a log scale to make low counts more visible. Arrows mark a single image slice that is examined further in Figure 4
Figure 4. Series of independently simulated x-ray radar images for various scenes with low-atomic-number targets. (a) Simulation includes the concrete floors, walls, and ceiling, as well as the rear lead wall, but no collimator around the detector. (b) Simulation includes the concrete floors, walls, and ceiling, as well as a lead castle collimator around the detector, but no rear lead wall. (c) Simulation does not include the concrete wall, floor, or ceiling, the rear lead wall, nor the collimator around the detector. (d) Simulation is the same as for (c), but for a vacuum rather than atmospheric pressure.
In this work we have developed an x-ray radar particle physics model. We have used this model to simulate our new x-ray backscatter approach and have compared the simulation results with experimental data. We find that the computational and experimental results are in broad agreement. We are thus able to readily identify various sources of secondary scattering in the experiments. We are now working to further improve the accuracy of our model. This will involve simulating more representative beam parameters (e.g., energy distribution) and incorporating a scintillator response. Our ultimate aim is to achieve a sufficiently accurate model that can be used to predict the performance of alternative scenarios and thus inform future x-ray backscatter experiments.
This content includes material that is subject to Crown Copyright (2015), DSTL. This material is licensed under the terms of the Open Government Licence,9 except where otherwise stated.
David Lockley, Robert Deas, Robert Moss
Counter Terrorism and Security Division
Defence Science and Technology Laboratory
Sevenoaks, United Kingdom
Dean Rusby, Lucy A. Wilson, David Neely
Central Laser Facility
STFC Rutherford Appleton Laboratory
Didcot, United Kingdom
1. A. Chalmers, Rapid inspection of cargos at portals using drive-through transmission and backscatter x-ray imaging, Proc. SPIE
5403, p. 644, 2004. doi:10.1117/12.548958
2. A. Chalmers, Three applications of backscatter x-ray imaging technology to homeland defense, Proc. SPIE
5778, p. 989, 2005. doi:10.1117/12.607184
3. O. Klein, Y. Nishina, Über die Streuung von Strahlung durch freie Elektronen nach der neuen relativistischen Quantendynamik von Dirac, Zeitschrift Phys. 52, p. 853-868, 1929.
4. E. T. Dugan, A. M. Jacobs, Z. Su, L. Houssay, D. Ekdahl, S. Brygoo, Development and field testing of a mobile backscatter x-ray lateral migration radiography land mine detection system, Proc. SPIE
4742, p. 120, 2002. doi:10.1117/12.479082
5. R. M. Deas, L. A. Wilson, D. Rusby, A. Alejo, P. P. Black, S. E. Black, M. Borghesi, et al., A laser driven pulsed x-ray backscatter technique for enhanced penetrative imaging, J. X-ray Sci. Technol.
23(6), p. 791, 2015. doi:10.3233/XST-150520
6. D. Lockley, R. M. Deas, R. Moss, D. Neely, D. Rusby, L. A. Wilson, Laser induced x-ray radar particle physics model. Presented at SPIE Security + Defence 2015, in Toulouse, France.
7. C. J. Hooker, J. L. Collier, O. Chekhlov, R. Clarke, E. Divall, K. Ertel, B. Fell, et al., The Astra Gemini project---a dual-beam petawatt Ti:sapphire laser system, J. Phys. IV France 133, p. 673-677, 2006.
8. S. Kneip, S. R. Nagel, S. F. Martins, S. P. D. Mangles, C. Bellei, O. Chekhlov, R. J. Clarke, et al., Near-GeV acceleration of electrons by a nonlinear plasma wave driven by a self-guided laser pulse, Phys. Rev. Lett.
103, p. 035002, 2009. doi:10.1103/PhysRevLett.103.035002