·Table of Contents
·Workshop - Neutron workshop
Neutron Tomography on a Low Power research reactor
Atominstitut der österreichischen Universitäten, Stadionallee 2; A - 1020 Wien,
Paul Scherrer Institute, CH - 5232 Villigen PSI
TU München, Fakultät für Physik E21, D - 85747 Garching
Paul Scherrer Institute, CH - 5232 Villigen PSI
Atominstitut der österreichischen Universitäten, Stadionallee 2; A - 1020 Wien
At the 250 kW TRIGA Mark II reactor of the Atominstitut in Austria a neutron tomography facility has been installed in front of the thermal column. First neutron tomography experiments showed promising results. The neutron flux at this beam position is 1,3 · 105 neutrons / (cm2s), the L / D - ratio is 125, the Cd - ratio is 20 and the beam diameter is 8 cm. For a three-dimensional tomographic reconstruction of the sample interior, transmission images of the object taken from different view angles are required. Therefore a rotary table driven by a step motor connected to a computerized motion control system has been installed at the sample position. In parallel a suitable electronic imaging device has been designed. It can be controlled by a computer in order to synchonize the software of the detector and of the rotary table with the aim of an automation of measurements. Further requirements for a detector suitable for neutron tomography are: exact and reproducible positioning, easy handling, high spatial resolution and dynamic range, high efficiency and a good linearity. According to these requirements a new CCD-camera NR detector has been designed, and optimized. The principle of this detector is the following: The neutron beam reaches the neutron sensitive scintillator screen (Levy Hill, ZnS(Ag)-6LiF) where each detected neutron triggers a photon cascade. The photons are reflected to a CCD-camera by a mirror to allow placing the camera out of the direct neutron beam. All components are placed in a light tight box. The camera is an Astrocam nitrogen cooled slowscan CCD-camera with a thinned SITe SI502A/T chip of the format (512 x 512) pixels and a pixel size of (24 x 24) mm2. We have the possibility to chose between two lenses, a Nikon NOKT 58 mm F/1.2 or a Nikon 180mm F/2,8 lens, depending on the sensitivity and resolution needed for each application. The high precision CCD driver electronics provides 16-bit digitization (65535 gray levels). The detector has an excellent linearity and sensitivity. Reasonable exposure times can get as low as 20 s per image. This means that a complete tomography of a sample can easily be performed within one working day.
Any method that reconstructs internal structural information within an object by mathematically reconstructing it from series of projections is generally referred to as tomography. In principle it may be used to get true three-dimensional (3D) arrays of voxels, or to obtain a two-dimensional (2D) image from a series of one dimensional line projections . In the case of neutron computed tomography (N-CT), we usually speak about the reconstruction of a 3D voxel array representing the inner structure of the sample on the basis of 2D neutron transmission images of an object taken from different view angles. To obtain these images it is principally possible either to rotate source and detector around the object, as it is usually done with X - ray tomography or to rotate the object between taking the images, which is usually the only practicable possibility with N-CT, as (despite mobile sources) most neutron sources, as reactors or spallation sources are at a fixed position, whereas the object can easily be placed on a rotary table.
Two neutron radiography facilities are located at the 250 kW TRIGA Mark II reactor (training, research and isotope production; General Atomic) at the Atominstitut in Vienna, Austria. Main data of these facilities is shown in tab. 1 .
|FLUX DENSITY (cm-2s-1)
|| 3 ×105
|BEAM DIAMETER (cm)
|Cd - RATIO
|tab. 1 : main characteristics of the neutron radiography facilities at the Atominstitut|
Even though station 2 has a lower neutron flux and a smaller beam diameter, we decided to build the first neutron tomography set-up at the Atominstitut at station 2. The main reason for this decision was the demand for easy access to the sample position during reactor operation to allow adjustment and operation-control of the new experimental setup. This is only possible at station 2, whereas at station 1 the sample position is located in an irradiation room, that can not be entered during reactor operation or even for several hours after shutdown, due to the high radiation level. (Samples are transported to the beam position using a vertical elevator system). Another advantage of station 2 is the high collimation ratio (L/D = 125). As shown in fig. 2, station 2 is located at the thermal column of the TRIGA reactor. This is a cube of nuclear grade graphite with dimensions of about 1,2 m at each side and inside this cube a collimator has been installed.
Fig 1: neutron radiography station 1
Fig 2: neutron radiography station 2|
Neutron tomography setup
The first step was the design and optimization of a suitable imaging device, fulfilling the following requirements:
- High Efficiency:
As we installed the tomography setup at a low power reactor with very low neutron flux at the sample position, a high detector efficiency is one of the most important demands in order to be able to perform at least one tomography within one working day.
- Exact and Reproducible Positioning:
It is evident, that for all exposures of one sample (from different view angles), the detector needs to be in exactly the same position. Corrections of different detector positions during the reconstruction calculations would be very time consuming, not precise and for many cases even impossible. (I. e. for a detector consisting of film and a converter in a cassette, this requirement can hardly be realized, as after each exposure the cassette has to be removed and the film changed and it is quite difficult to place the cassette as well as the films inside it always at exactly the same position.)
- Easy Handling:
The enormous number of digitized images needed for each tomography requires a detector that can be synchronized with the rotary table by a computer to automate the whole set of measurements needed for the tomography of each sample. Also the detector should produce digitized image data to avoid external digitization (as i. e. scanning of films).
- High Spatial Resolution, Large Dynamic Range, Good Linearity:
These points are always very important for the quality of images gained by NR. But the influence on the quality of neutron tomography is even higher.
At the Atominstitute two neutron radiography stations are available. Therefor we wanted to design an imaging device that can easily be transferred to station 1 or to the Atominstitut's permanent beamline S18 at the ILL, Grenoble in France .
Based on these requirements a CCD-camera detector has been chosen . Furthermore, a rotary table, driven by a stepping motor has been installed at the sample position. A computer synchronizes the rotary table and the electronic imaging device, so that a sequence of images from certain different view angles is taken automatically . The principle of this N-CT facility is shown in fig. 3.
Fig 3: neutron tomography facility|
A neutron beam penetrating the sample is attenuated according to the sample material and - geometry and reaches the neutron sensitive scintillator screen, where each detected neutron triggers a photon cascade. The light emitted by the scintillator is reflected to the CCD - camera by a mirror. This detector design allows to place the camera out of the direct neutron beam to protect the chip from radiation damage. The chip of the camera is read out by the computer. After exposure the computer sends a trigger signal to the rotary table, the sample is rotated a defined value and the next exposure starts automatically.
Details of the CCD neutron radiography detector
Due to the low neutron flux the most important requirement for the detector is its sensitivity. Therefor we decided to chose a thinned (backilluminated) chip with a pixel array format of 512 x 512 pixels with a pixel size of (24 x24) mm2 (SITe SI502A). The Quantum Efficiency of the CCD camera is in the range of 80 - 90 % for wavelengths from (350 - 800) nm, as shown in fig. 5. The camera is nitrogen cooled to reduce dark current, which is extremely important for long exposure times caused by the low neutron flux. The high precision CCD driver electronics provides 16-bit digitization (65535 gray levels).
Fig 4: spectral sensitivity function of SITe chips
Fig 5: MTF of the CCD neutron radiography detector with the use of the 58mm F/1,2 and the 180mm F/2,8 lens|
Depending on the desired image size, resolution and detector sensitivity, we can choose between a NIKON 58mm F 1,2 and a NIKON 180mm F 2,8 lens. After extensive test measurements performed at Paul Scherrer Institute (PSI), Switzerland , concerning sensitivity, linearity, resolution, homogeneity and after glow, a neutron sensitive Levy Hill ZnS(Ag)-6LiF scintillator screen has been chosen. For the mirror the main requirements are a high reflectivity (95 %) of the light emitted by the scintillator, generation of as few g-rays as possible and no lasting activation of the materials of the mirror. According to that a 2 mm thick glass plate coated with Al and with TiO2 as protective layer was manufactured at PSI. A light shielding tube which also serves as a positioning device for the detector components has been designed according to the boundary conditions given by the size and shape of the individual detector components, the desired image size and the space available at the facility. The material for the whole box is Al to avoid lasting activation. It has also been manufactured at PSI. Light tightness of the box has been successfully tested. The image size is (21 x 21) cm2 for using the 58mm F/1,2 lens (1) and (5 x 5) cm2 for the 180mm F/2,8 lens (2). This means that an area of (428 x 428) mm2 (1) or (80 x 80) mm2 (2) of the object is projected on one camera pixel. We will only be able to use the larger image size when the detector will be transferred to station 1, but nevertheless, we also employed lens (1) at station 2 (with a beam diameter of 8 cm) for investigations with lower demands for the resolution, because with that lens the detector is much more sensitive and exposure times can get as low as 20 s (3 min. for lens (2)) per image. This means that for both cases one set of measurements necessary for a high quality tomography (200 images) can be performed easily within one working day, despite the low neutron flux.
Realization of the experimental work
An extremely important point before the measurements can be performed is to adjust the rotary table so that the rotation axis is parallel to the vertical lines between the camera pixels. This has been done by placing a test object (steel cube) on the rotary table, taking an image, rotating the object by 180 degrees and taking another image . The second image can then be flipped and subtracted from the first image. If the rotary table is adjusted well, the resulting image is all black, if not, the resulting image shows wedge shaped black or white areas and the rotary table has to be adjusted accordingly.
For the N-CT of each sample we took the following images:
- 5 open beam images
- 5 dark images (with closed camera shutter)
- 200 images at different view angles of the object (object has been rotated 0,9 degrees between each exposure)
Preparation of the data for the reconstruction
- White spot correction :
All images taken by the CCD - camera show white spots caused by g - radiation hitting the CCD chip. For the dark and the open beam images this correction is done by applying a median filter over 5 images. For each pixel the median pixel value (within the 5 values of the 5 images) is taken to obtain the resulting open beam- and dark image, respectively. For the correction of the images of the object, a threshold value filter is applied, which replaces values above a certain threshold by a constant, chosen value.
- Dark image correction:
The dark image is subtracted from the open beam image and the images of the object
- Flat field correction
- Search of the rotation axis :
The first (1) and the last image (200) of the set of measurements of the object are identical but flipped (rotation of 180 degrees). Therefor image 200 can be flipped and subtracted from image 1. If the rotation axis is already located in the center, the resulting image is all black, if not, a white or black stripe appears in the resulting image. From this stripe the number of rows that have to be cut can be derived and the images are cut accordingly.
- Selection of the area of interest (AOI):
To reduce the size of the matrices of the 2D images, an AOI around the object is selected, so that unnecessary open beam regions are removed and that the rotation axes stays within the center.
- Calculation of the attenuation coefficient m:
m= -ln (I / I0), where I0 is the Open Beam signal and I the signal after the attenuation of the beam in the object.
- Creation of Sinograms
- Calculation of the 3D voxel array using the filtered backprojection algorithm
- Visualization of the N-CT
Fig 6: neutron tomography of a diode|
First reconstructions have been made at the TU Munich and in parallel at Paul Scherrer Institute. Fig. 6 shows the N-CT of a diode, as an example of a N-CT made at the low power TRIGA reactor at the Atominstitute.
A neutron tomography facility has successfully been installed at the low power TRIGA reactor at the Atominstitute of the Austrian Universities in Vienna. Due to the high sensitivity of the new CCD camera neutron radiography detector, reasonable exposure times are as low as 20 s per image. This means that one tomography consisting of a set of 200 images can easily be made within one working day. First reconstructions have been made in parallel at the TU Munich and at PSI. For the close future it is planned to transfer the tomography equipment to station 1 at the Atominstitute, where a larger beam diameter and a higher neutron flux is available.
Special thanks are devoted to Dr. E. Lehmann for his support with the measurements at PSI, his ongoing advice and for the box and the mirror for the CCD detector.
Part of this project was financed by EURATOM, UT 4.
- J. C. Russ; The Image Processing Handbook; IEEE press, 1999, p. 575 - 616
- S. Körner, H. Pleinert, H. Böck, Review of Neutronradiography Activities at the Atominstitut of the Austrian Universities, Proceedings of the 5th World Conference on Neutron Radiography, Berlin 1996
- M. Hainbuchner, M. Villa, G. Kroupa, G. Bruckner, M. Baron, H. Amenitsch, E. Seidl, H. Rauch; The new high resolution ultra small angle neutron scattering instrument at the High Flux Reactor in Grenoble; Journal of Applied Crystallography (2000). 33, p. 851 - 854
- S. Körner, H. Böck, H. Rauch, E. Lehmann, A New CCD - Camera Neutron Radiography Detector at the Atominstiute of the Austrian Universities, Proceedings of the conference on Nuclear Energy in Central Europe 99, Portoroz, Slovenia, 1999
- B. Schillinger, 3D Computer tomography with thermal neutrons at FRM Garching, Journal of Neutron Research Vol. 4, 1996, p. 57 - 63
- S. Koerner, E. Lehmann, P. Vontobel, Design and Optimisation of a CCD - Neutron Radiography Detector, accepted for publication in Nuclear Instruments and Methods A.
- B. Schillinger, Neue Entwicklungen zu Radiographie und Tomographie mit thermischen Neutronen und zu deren routinemäßigen Einsatz, Dissertation TU München (Theses in German), Mensch & Buch Verlag Berlin, 1999;www.menschundbuch.de