Logo der Physikalisch-Technischen Bundesanstalt

Tomografie mit schnellen Neutronen an der PTB Beschleunigeranlage PIAF

16.03.2015

Bildgebende Durchleuchtungsverfahren mit Neutronen sind zu einem wichtigen Werkzeug in der zerstörungsfreien Werkstoffprüfung geworden. Schnelle Neutronen als Messsonde eröffnen gegenüber den üblichen thermischen Neutronen die Möglichkeit, auch größere Objekte zu durchdringen sowie kompaktere (und damit mobilere) Anlagen zu realisieren. In der PTB wurde eine Tomografieeinrichtung an der Beschleunigeranlage PIAF errichtet und experimentell demonstriert, dass Verteilungen von Objekten geringer Dichte hinter Abschirmungen aus schweren Metallen tomografisch dargestellt werden können.

Radiography and tomography with X-rays are well established and common tools in Non Destructive Evaluation (NDE. In industry they are used to assure quality during production, investigate defects and confirm reliability of critical components. In case of failure, they are instrumental to identify causes and scenarios. Based on their physical properties, X-rays have a limited penetration power in certain materials and weak contrast for materials made out of light elements (hydrogen, carbon, oxygen, nitrogen etc). Both limitations are most evident if relevant structures are composed of light elements while the surrounding structures are made of heavy materials (metals or compounds). Other applications require visualisation of functional or dynamic features such as moving parts, dynamic properties of layers or films (frequently of organic or hydrogenous liquids), or vapours inside heavy metallic compounds, which show hardly and contrast in X-ray transmission imaging. Neutron interaction, on the other hand, has a weak dependency on the atomic number if the material and can provide high contrast for low-Z elements encapsulated in massive shielding and because therefore a complementary tool to X-ray imaging where higher contrast of light elements is required.  However, the development and application of these neutron techniques in Europe (and worldwide) are concentrated at a few large-scale facilities. Due to the fact that powerful neutron sources are commonly based on nuclear reactors or large accelerator based spallation sources, neutron techniques did not find a widespread application in the field, in particular not into the industrial production process.

While most of the existing facilities focus on the use of slow (thermal) neutrons, fast-neutrons (1 MeV – 10 MeV) are rather scarce because of the rather weak sources available today and the difficult detection and imaging techniques for energetic neutrons. Nevertheless, fast-neutrons provide certain advantages which complement the use of thermal neutrons. Among these is the much higher penetration capability of large objects (container, barrels and drums, large engines), which is similar to high-energy (hundreds of keV-) X-rays and the possibility of element sensitive imaging, which was demonstrated by our group in the recent years. Furthermore, compared  to reactor or spallation source based neutron techniques fast neutron imaging can be performed at much smaller linear accelerators or even generators which, in principle, allow a certain amount of mobility and can make neutron imaging applicable outside large scale research facilities, e.g. in harbours, (nuclear) waste processing facilities, in industrial production or quality assurance.

In the previous years the focus of our work in fast-neutron imaging was the investigation of aircraft containers using Fast-Neutron Resonance Radiography (FNRR). In 2014 we have successfully implemented the instrumentation and demonstrated the capability of tomography with fast-neutrons at the PTB-accelerator facility PIAF. Though PIAF is not optimal suited for this purpose (too low neutron flux and therefore rather long measurement cycles) it is a perfect environment to develop the components and the technique. Tomography, in contrast to plain or few-view projection radiography,  requires multiple projections at different viewing angles on a precise tomographic sample manipulator (rotation  and translation) table, as well as appropriate data analysis and back-projection techniques. All these components were developed, procured and finally set up at PIAF and applied in two experiments to demonstrate the capability of the technique.

Figure 1 shows the beam target station (right) and the sample and detector setup (to the left) at beamline 8 in the experimental hall at PIAF. The fast-neutron beam was produced by a deuteron beam current of 20 mA and 11.5 MeV energy, hitting a thick, water cooled  Be target. The beam spot size was of about 3 – 5 mm in diameter. The resulting broad energy neutron beam of ca 6 MeV average energy was collimated either in cone-beam or in fan beam geometry and hit the sample in about 2 m distance from the target, directly in front of the imaging detector (right side of figure 1). For the tomographic exposure a sustained maximum flux of ca 2.5 x 106 cm-2s-1 can be achieved at PIAF. The detector system consisted of a plastic scintillator viewed by dedicated optics, image-intensifier and a CCD camera. Two scintillator screen types were tested during these experiments: a bulk screen (slab) of EJ2004 plastic scintillator, 20 mm thickness) and a fiber optic screen (FOS) of 50 mm thickness with diameter 0.7 mm fibers.  Both screens cover a field of view of 20 x 20 cm2. Due to the relatively large size of the neutron producing beam spot magnifying geometry was not applicable and the best position resolution is obtained by placing the detector close (ca 15 cm) behind the sample. Distance between source and detector were 200 cm.

Tomographic scans were taken in two collimation schemes, defining a cone and a fan beam neutron beam.  Neutron scattering by the object and inside the thick scintillator screens play a significant role in the quality of the tomographic reconstruction, which favor the fan beam configuration due to its lower scatter contribution in the images. Scattered neutrons and also the unavoidable gamma background in both irradiation scenarios demand also careful shielding of the CCD camera which captures the image of the scintillator screen, to avoid direct hits, which cause temporarily saturated pixels or even permanently destroy pixels. Direct hits and damaged pixels occur as black or white spots and cause reconstruction artifacts in the tomographic images. Since these effects cannot be avoided completely a Matlab based image acquisition, pre-processing and filtering scheme was developed which allows to remove almost entirely these effects. Image reconstruction was performed on usually 360 projections taken by rotating the sample in 1 degree steps, using a commercial software (Octopus), applying the Feldkamp algorithm. For visualísation and analysis the VG-Studio software was applied. Beam exposure time for a single tomography for 360 projections was of the order of 6 to 8 h.

The objects which were selected for the experiment were composed of typical materials where fast-neutron tomography can contribute and is expected to be superior to conventional X-ray and thermal neutron tomography. For this test and evaluation experiment the goal was to identify the best irradiation geometry (cone or fan beam), to select the optimum neutron converter in the detector (slab or fiber optical screen) and to determine the obtainable resolution in the reconstructed image. Figure 2 shows examples of a few results of test objects. The plastic (polyethylene or polyamide) phantoms with defined internal structures were inserted in a cylindrical lead (Pb-) shell of 10, alternately 25 mm wall thickness. Such a sample composition make it hard for X-rays (too low contrast) or thermal neutrons (too low penetration) to visualize the small voids and structures in the plastic matrix inside the Pb shielding. As expected, the comparison of the reconstructed images prove that cone beam is preferable to reproduce the fines structures, however exposure times are much longer as compared to cone geometry, since the object must be scanned in one dimension. Regarding the performance of the detector, the fiber optical screen (FOS) proves to be superior to the simple slab scintillator. The slab scintillator, in contrast to the FOS, requires to be viewed over its full depth (20 mm) which causes image blurring due to the low depth of field of our large aperture lens. The phantom contains different bore holes in tangential, radial and axial directions of the PE cylinder. In the tomographic reconstruction in fan beam geometry these structures down to 1 mm are visible.

As a first practical application we have also performed a tomography of a wooden artifact from an epitaph of the St Laurentius church in Tönning/Germany. This investigation was initialized as a comparative study of different radiation based techniques by Kurt Osterloh from BAM and the Rathgen Forschungslabor in Berlin. Due to an early restoration about 100 years ago the wooden epitaph was polluted by Carbolineum, a toxic chemical, which still evaporates from the object, distributing a stinging smell and destroys the surface coating. The restorers are interested to identify the distribution of Carbolineum inside the object. Three different radiographical techniques were tried on a sample of the epitaph, a massive wooden skull. These are X-ray tomography at 160 kV anode voltage (at BAM), tomography with neutrons from a reactor using unmoderated fission neutrons of ca 1.8 MeV energy (NECTAR facility at TU-Munich) and neutrons from PIAF, ca 6 MeV average energy. Figure 3 shows three different slices through the tomographically reconstructed skull, using X-rays and the 6 MeV neutron beam of PIAF. The fission neutrons could not penetrate the skull and did not yield useful images. The X-ray images, due to their high position resolution,  reveal nicely  the fine cracks and worm tunnels form earlier biological damages. Only the 6 MeV neutron images reveal a cloudy structure inside the wood which we attribute to different concentrations of Carbolineum. However, a quantitative correlation cannot be provided yet. For this suitable calibration measurements with defined samples must be performed which presently are not available.

Fig. 1: Set up of the tomographic system at PIAF. To the right is the target station, inside a massive beam collimator, here equipped with a fan beam collimator. The left side of the figure shows the target on a rotation and translation table in front of the detector box.

Fig. 2: Various slices of the tomographic images of plastic samples embedded in a lead cylinder of 10 and 25 mm wall thickness. The upper left shows a Siemens star made of polyamide, the others ones different projections  of the Polyethylene phantom, which contained various bore holes of different diameters and length, ranging from 1 mm in till 10 mm in diameter. The image in the middle row right is the PD phantom inside the 25 mm thick lead shielding.

Fig. 3: Three different slices of the tomographic reconstruction of the wooden skull, using X-rays from 160 kV anode voltage (upper row) and fast-neutrons of 6 MeV average energy (lower row). The photograph to the right indicates the position of the different slices.

Initiates file downloadDatei (pdf, 336 kB)