·Table of Contents
·Late Received Papers
NON-DESTRUCTIVE EXAMINATION OF SiC NUCLEAR FUEL SHELL USING X-RAY FLUORESCENCE MICROTOMOGRAPHY TECHNIQUEK.NECIB ; M.A. Belouchrani ; A. BRITAH
Fort Valley State University
Computer Technology Department
Fort Valley, GA 31030
Oak Ridge National Laboratory
Metals and Ceramics Division
Oak Ridge, TN 37831-6118
Carl J. McHargue
University of Tennessee
Materials Science and Engineering Department
Knoxville, TN 37996
The advent of third generation synchrotrons and the ongoing revolution in x-ray focusing optics have contributed to the development of x-ray probes with micron and submicron resolution. As a result, it is now possible to perform micro-structural analysis of materials with techniques previously not possible. In this study, x-ray fluorescence microtomography, a comparatively new technique, coupled with a synchrotron x-ray microprobe of micron resolution were used to study the trace element distributions in a SiC shell of an advanced nuclear fuel particle, known as TRISO. The TRISO fuel particles contain a small kernel of nuclear fuel encapsulated by alternating layers of C and a barrier layer of SiC. The SiC shell provides the primary barrier for radioactive elements in the kernel and metallic fission products. The performance of this barrier under adverse conditions is key to containment.
The tomographic measurements were made with an ~1x3 mm2 x-ray probe on beamline 2-ID at the Advanced Photon Source (APS). The distribution of trace elements in the SiC shell was reconstructed after correcting the data for artifacts arising from absorption and scattering off the Kapton tape used to encapsulate the sample. The tomographic reconstruction was carried out, both with and without absorption correction. The images obtained in both reconstruction cases show that the SiC shell is ~38 mm thick, and the observed trace elements are distributed in small <1µm regions through the SiC shell. The results also indicate that most trace elements have localized distributions within the SiC shell. The trace elements can be attributed to radiation enhanced diffusion of elements in the kernel or to trace elements introduced during fabrication. The investigation of the reconstructed images with absorption correction indicates that for low energy fluorescence lines the effect of SiC absorption is rather significant in estimating the concentration of trace elements.
The overall result of this research study verifies that X-ray fluorescence microtomography is an ideal tool for detecting high Z trace elements in a low Z matrix because it is a sensitive and penetrating nondestructive probe and because it provides a picture of the elemental distribution with micron spatial resolution. The advantages of x-ray fluorescence tomography for the non-destructive 3-D characterization of materials will continue to grow as x-ray source brilliance increases.
|Fig 1: Trends in brilliance of x-ray sources over the last 100 years.|
In recent years, there has been a growing need in various areas of science and technology to non-destructively examine the internal structure of advanced materials with high spatial resolution. This need is especially great in the nuclear industry where ultra-reliable and high performance materials are needed for operation in hostile environments. Since many advanced materials depend on structures that are as small as 1 micron or even less, inspection probes on the order of micron or submicron sizes are required to characterize the relevant microstructural features. Conventional inspection techniques such as scanning tunneling microscopy or electron and optical microscopy can provide 3-D micron to submicron resolution only in a destructive manner. These methods require serial sectioning, etching and staining of the sample, which can alter (or even destroy) the original structural features. In some cases, even destructive methods may be impractical because of physical and /or environmental restrictions (e.g.; radioactive samples). Conventional x-ray based non-destructive methods such as transmissive radiography and laminographic imaging can achieve spatial resolutions on the order of hundreds of microns but new synchrotron based methods are required to approach the micron or submicron regime.
An extension of conventional tomography for nondestructive and noninvasive imaging, known as microtomography now exists which can provide high spatial resolution and detailed imaging information about the internal structures of objects. Microtomography utilizes synchrotron x-rays, advanced focusing optics, and precision x-ray detectors to achieve micron size spatial resolution. Synchrotron radiation possesses unique characteristics which allow the measurement of internal structures with micron resolution. Synchrotron radiation is extremely intense and highly collimated. It covers a wide range of energy spectra from the infrared to 200 keV. In addition, it can provide a tunable monochromatic x-ray beam with minimum beam hardening effects. For example, synchrotron x-ray sources can provide x-ray brilliance up to 14 orders of magnitude higher than from conventional laboratory or medical sources. Figure 1 presents the trends in the x-ray brilliance over the last century. As can be seen in the figure, there has been significant progress in improving the x-ray brilliance since the advent of synchrotron technology. Presently, the Advanced Photon Source (APS) provides the greatest x-ray brilliance at ~5´1011 (photons/s/3eV/mrad2/mm2). Since the figure-of-merit for microfluorescence is brilliance , 12 to 14 orders of magnitude increase in the brilliance will allow for x-ray fluorescence microtomography with micron resolution. However, with laboratory and medical x-ray sources, insufficient x-ray brilliance limits the achievable spatial resolution to approximately one millimeter or slightly better.
Microtomography technology is currently gaining popularity as a result of its unique and attractive features. It has become more widely recognized as an alternative to destructive testing for microstructural analysis of advanced materials since it allows repeated measurements on the same samples. Microtomography has found a wide range of applications in biological, medical, chemical, and materials sciences. From internal imaging of a bee’s head to complex and heterogeneous microelectronic chips, this imaging technology has proven to have a major impact for both science and industry. With current and future advances in synchrotron technology and x-ray focusing optics, microtomography can help revolutionize the design and development of advanced materials.
This paper takes advantage of an ongoing revolution in x-ray science to nondestructively measure trace chemical inhomogenities with near micron resolution. This work illustrates the unique advantages of x-ray microtomography for nondestructive trace element analysis. The ability to nondestructively measure trace element inhomogenities in 3-D will have a major impact on our understanding of advanced materials.
A study of TRISO fuel particles provides a good opportunity to demonstrate the unique advantages of x-ray fluorescence microtomography on a complex material system. TRISO fuel particles, used in High-Temperature Gas Cooled Reactors(HTGR) are composite structures with a nuclear fuel kernel surrounded by alternating layers designed to contain fission products and compensate for radiation damage. As shown in Fig 2, a typical fuel particle contains an inner kernel of nuclear fuel, a low density buffer layer of pyrocarbon, dense layer of pyrocarbon coating, an interlayer of SiC and an outer dense layer of pyrocarbon. Depending on the type of reactor core design, the fuel kernel is chosen from UCO, UC2, ThO2, or UO2. In addition, fuel kernel size, the thickness of the various layers, and the overall size of the TRISO fuel particle can vary with the type of fuel kernel. TRISO coated fuel particles are compacted into fuel rods designed for passive containment of the radioactive isotopes. The SiC layer provides the primary barrier for both radioactive elements in the kernel and gaseous and metallic fission products. The effectiveness of this barrier layer under adverse conditions is critical to containment and has been the subject of previous studies.[1,2]
We report on measurements of the elemental distribution in a SiC shell after exposure to a fluence of ~1025 (neutrons/m2). X-ray fluorescence microtomography is an ideal tool for this work because it is nondestructive, it is sensitive to heavy elements in a low Z matrix, and because it can provide a 3-D picture of the elemental distribution; the observed elemental distribution can be correlated with flaws and defects in the SiC shell.
|Fig 2: Schematic of TRISO fuel element.|
Previous Studies and Sample Preparation
The shell examined came from a previous study of diffusion through SiC shells. In the previous study, a total of 19 shells were exposed to varying fluences (Table 1). The C buffer layers and nuclear kernels of the TRISO particles were removed by laser drilling through the SiC and then leaching the particle in acid. The shells were repeatedly leached until the activity remained constant. At this point it was assumed that any remaining activity was due to daughter products which had migrated into the SiC shell. The shells were then analyzed to determine the total number of the various daughter products in each shell (Table 1). This method provides an accurate absolute measurement of the total loading of radioactive elements but gives no information about the distribution in the shell.
|Ball ID||Kernel Type||Fluence|
|Atoms in particles
||9.91 x 1015
||4.95 x 1015
||1.04 x 1014
||1.01 x 1016
||5.58 x 1015
||1.14 x 1014
||7.69 x 1015
4.37 x 1015
1.02 x 1014
||7.82 x 1015
||4.39 x 1015)
||9.11 x 1013
||9.05 x 1015
||4.36 x 1015
||1.19 x 1014
||9.14 x 1015
||4.38 x 1015
||1.05 x 1014
||9.52 x 1015
||4.60 x 1015
||1.10 x 1014
||9.55 x 1015
||4.89 x 1015
||1.20 x 1014
[used in this experiment]
||6.81 x 1014
||1.37 x 1014
||2.96 x 1012
||6.56 x 1014
||1.36 x 1014
||3.56 x 1012
||2.79 x 1015
||5.71 x 1014
||2.20 x 1013
||3.22 x 1015
||6.30 x 1014
||2.16 x 1013
||Table 1: Selected isotopes and irradiation conditions for irradiated TRISO particles.|
X-ray fluorescence microtomography
The advent of intense synchrotron radiation sources has improved the resolution of tomography into the mm regime. Although most tomographic measurements use transmission tomography methods [3-5] experiments as early as 1985 demonstrated the elemental sensitivity of fluorescence tomography to high Z trace elements in a low Z matrix. These pioneering measurements of the Fe distribution in a honey bee (Apis mellifera) found Fe concentrations at the surface and in the abdomen of the bee with a spatial resolution of ~150mm. With new x-ray focusing optics and sources, mm3 fluorescence microtomography is now practical. Fluorescence tomography however is very slow compared to transmission tomography because the number of volume elements (voxels) which can be resolved scales roughly with the number of data points collected. In contrast, transmission tomography measures the attenuation through all translation positions simultaneously and therefore has an ~ 106 faster collection time. Although transmission measurements are much faster, fluorescence measurements have better signal-to-noise for trace element detection.We estimate below the relative sensitivity of transmission microtomography to fluorescence microtomography in order to justify the enormous increase in time for fluorescence measurements.
Comparison of transmission contrast microtomography to fluorescence microtomography
The most sensitive way to measure elemental distributions with transmission tomography is by comparing the transmitted beam intensity between x-ray energies above and below the absorption edge of a trace element. In this experiment, the transmitted beam intensity is attenuated along the volume elements in the sample matrix (Fig. 3) and is detected with a CCD area detector with ~14-16 bits resolution.
|Fig 3: Schematic of a transmission tomography experiment. The beam intensity observed by an element of the CCD detector is attenuated by all elements along the direct beam path to the detector.|
We can estimate the minimum detectable limit in the most favorable (but impractical) transmission case where the CCD detector is near the saturation limit. For example, at 10 keV, the linear absorption coefficient for Z=6 is ~3 cm-1, the linear absorption coefficient scales roughly like Z4 and there is an ~ factor of 8 increase in absorption at the K edge. For a small pixel with length DX (see Fig. 2) and with trace-element concentration CT, the transmitted beam intensity above the K absorption edge is therefore attenuated by an additional amount compared to the transmission below the absorption edge. Here mT is the trace-element linear absorption coefficient which we estimate to be ~2.3x10-3Z4 (cm-1). At the limit of detectability, the additional absorption is small (e-e~1-e) and is at the limit of the resolution of the detector, 1.6x10-2 (cm-1) Z4 DX(cm)CT~2-16. Changing length units to microns, we find,
For example, with Z~26 and for feature size of ~1mm the estimated detectable limit cannot be less than 20 PPM. In a real measurement the detectable limit must be much higher, since the CCD must operate away from the detector saturation limit to avoid non-linear performance.
By contrast, fluorescence tomography is much more sensitive to trace elements. The estimated minimum detectable limit (MDL) of a modern x-ray fluorescence microbeam is ~ 5-80 PPB/s/mm2 for a uniform sample. Since the MDL increases as the square-root of the background, and decreases linearly with signal, the MDL of a small particle inside a larger sample will scale inversely with the ratio of particle size to sample size. The estimated MDL for fluorescence x-ray microbeam is therefore ~500-8000 PPB/s/mm2, or ~2 orders of magnitude better than for transmission microtomography. In actual tomographic reconstruction, the elemental sensitivity is degraded slightly due to noise introduced by the reconstruction algorithm. Nevertheless, x-ray fluorescence tomography is at least 100-1000 times more sensitive to elemental distributions than transmission tomography.
The experimental setup for an x-ray fluorescence tomography measurement is conceptually very simple (Fig. 4). The sample is placed on a stage which rotates and translates the sample. A detector is placed in the plane of the storage ring and at 90° to the incident beam. This detector geometry allows efficient measurement of the characteristic fluorescence from the trace elements while minimizing the x-ray elastic and Compton scattering. The elemental distribution in a slice through the sample can be reconstructed after translating the sample through the x-ray beam and rotating the sample at least 180° for every x position. Finer resolution is achieved by decreasing both the translation and rotational step size.
|Fig 4: Key elements of an x-ray fluorescence tomography experiment. The x-ray fluorescence is monitored while the sample is rotated and translated by a precision stage. Both the step size of the stage motions and the focal spot size of the probe beam determine the spatial resolution in the reconstructed image.|
The experiment was performed on beamline 2-ID of the Advanced Photon Source. Beamline 2-ID uses a low bandpass x-ray mirror to define a beam axis, followed by a Si 111 perfect crystal monochromator and a hard x-ray zone plate. The x-ray energy was set at 10.5 keV. For this experiment a 40 cm focal length zone plate was used which produced a spot size of approximately 1x3 mm2. The sample was epoxied to a glass fiber and sandwiched between 2 mil kapton tape to simplify handling. The fiber was mounted on a small goniometer head which allowed the ball to be positioned at the center-of-rotation of the rotation stage. The ball was then centered on the x-ray beam so that the translation range of the measurement passed completely through the center of the ball.
Because a tomographic reconstruction requires consistent measurement conditions, the incident beam intensity was measured with an AMPTEC model XR-100T PIN Diode detector with 250 eV energy resolution. The detector was placed at ~90° 2q to the beam (see Fig. 3) but out of the plane of the x-ray ring (to optimize scattering efficiency). Scatter from the air between the sample and the zone plate was monitored. This crude incident beam monitor worked well after a backscatter shield was installed between the sample and the air volume viewed by the detector. In a previous fluorescence tomography attempt with the same sample, the measurements were rendered useless for reconstruction by large and unmonitored variations in the incident beam intensity.
The reconstruction was compromised by both limitations of the equipment and by limited counting time. For example, the stepping rate of the sample stage and the detector readout imposed an overhead of about 2 seconds for each measurement step. Because of the limited beam time available, a compromised data collection scheme with 8 mm translation steps (101 translation steps) and 3° rotation steps (101 rotational steps) was used[To improve the quality of reconstructed images, it is recommended that the number of projection data (corresponding to the rotational steps) be greater than the sampling points (corresponding to the translational steps). Reference 14 presents the optimal relationship between the sampling points M (grid size) and the projection number, N, in the following form. The grid size, M, is equal to 2R/Dx where R is the maximum length of the sample from the center of rotation and Dx is the sampling interval (mesh size)]. This step size was much larger than the probe beam size which complicated the data analysis. In addition, we note that 60° of rotation was inaccessible due to the design of the rotation stage.
Before the tomographic measurements were begun, the unfocused beam was centered on the fuel ball shell to determine the detectable trace elements. The measureed fluorescence spectrum is shown in Fig. 5. Regions-of-interest, ROI’s, were set around the dominant fluorescence lines. Because of software limitations only 10 (ROI’s) could be stored. The 10 ROI’s stored are listed in Table 2. Unfortunately, Cs, Ce and Eu L lines lie in the region from ~4.2-7 keV. This region is substantially masked by intense K fluorescence from Cr and Fe (Fig. 5). Therefore the distribution of radioactive elements, which was most interesting, was not measureable due to the Cr and Fe concentrations in the shells.
A single line scan with 2 mm step size was then made to test the data collection software and hardware and to estimate typical feature sizes. For example, the measured linescan intensity of Zn is shown in Fig. 6. As can be seen, there are numerous small Zn features through the shell, some of which are smaller than the 2 mm step resolution. The origin of this Zn is not known.
|Fig 5: Fluorescence spectrum from an unfocused beam on the fuel ball shell.|
|ROI||Element||MCA Channel Number||Energy (KeV)|
|Table 2: Measured regions of interests (ROIs) in the fuel ball spectrum|
|Fig 6: Measured fluorescence intensity of Zn in a linescan experiment on TRISO fuel particle.|
The absolute elemental concentrations were estimated by comparing the observed fluorescence signal to the x-ray elastic and Compton scattering intensities. The total scattering cross section of SiC at 90° was estimated from Ref. 10. The beam polarization was estimated at ~5% and multiple scattering and absorption were assumed to be small. With these approximations, the factor I0W was determined, where I0 is the incident beam flux in photons/sec/mm2 and W is the detector solid angle. The trace element concentrations were then estimated again assuming negligible absorption from the following equation.
|I Fluorecence »I0W(Cs)||(2)|
Here C is the elemental concentration of the trace element and s is the fluorescence cross-section at the incident beam energy. Self absorption in the sample was later corrected during the tomographic reconstruction.
Plots of the ROI intensities as a function of angle and position are shown in Fig. 7. These raw images reveal a surprising amount of information about the sample. Attention to the correlated patterns in the various ROI’s is also required to obtain a meaningful reconstruction of the elemental distributions. For example, the signwave like patterns observed in the images of Fig. 6 are typical of point inhomogeneities. As seen in Fig. 7, the number of inhomogeneities varies depending on the ROI. In addition as seen in Fig. 7, the sign patterns are not always continuous. This behavior results because the step size used was larger than the convolution of the size of the x-ray probe and the inhomogeneities.
Another feature seen in Fig 7, is a marked intensity change in all the images at q~190°. This artifact arises due to scatter from the kapton tape used to encapsulate the sample. Over a small angular range, the kapton tape was at near glancing angle which deflected and absorbed the incident beam. These conditions produced large artifacts in the data which were removed by excluding the data points from a small angular range and averaging neighboring data points to estimate the fluorescence intensities in the missing region. Of the 104 data points for each ROI, ~25 were corrected for the effect of the kapton film.
In addition to the kapton artifact, during the measurements, the detector was occasionally saturated by an intense fluorescence signal from an unusually high concentration of one or more trace elements. Under these conditions, the ROI’s from all the elements were effected. The signature for detector saturation was either a decrease in the intensity of many ROI’s (parallized detector) or an increase in the intensity of many ROI’s correlated with background from an intense fluorescence line. The elements responsible were identified by observing the angle/translation pattern of each element and comparing them to the intense signal coordinates. Approximately 20 points for each ROI were corrected for saturation effects using average neighboring data points.
Tomographic reconstructions were carried out, both with and without absorption correction, using the library suite RECLBL. As described above, prior to the image reconstruction phase, the scanned projection data were corrected for kapton tape and saturation artifacts. After correcting the artifacts, the Fourier filtered backprojection routine of the library was incorporated into a computer program which reconstructed the 2-D images of the trace element distributions from the projection data on a Pentium PC. The mode of reconstruction was based on 360° rotations and the choice of filter was Hamming with a cutoff frequency of 0.25. As mentioned previously, the stage design limited rotations to ~300°. Because the reconstruction with 360° rotations appeared more robust, the missing 60° data was estimated from the 180° symmetric data.
The reconstructed spatial distributions of the monitored elements are presented in Fig. 8 and Fig. 9. Figure 8 shows the reconstructed distribution without absorption correction and Fig. 9 shows the distribution with absorption correction. Although the fuel ball shell is only ~38 mm thick, absorption corrections can be large, particularly for the low energy fluorescence lines. For example, ~97% of the Ba fluorescence (4.48 keV) can be absorbed by the SiC shell. As a result, the trace elements distributions are most easily detected near the surface of the SiC shell.
Fig 7: Raw intensity data for the various regions of interest. The artifact due to kapton tape scatter is clearly seen at ~190°. Saturation artifacts are also evident in the Ar and elastic scattering patterns. Saturation artifacts can appear as decreases in intensity when the detector is paralyzed, or as increases in intensity due to energy resolution degradation at high count-rate
The analysis of the images indicates that the spatial distributions of the trace elements are mostly localized at the outer edge of the SiC shell. The reconstructed images with the absorption correction implies that a higher counting rate is required to improve the signal-to-noise ratio for low energy fluorescence lines. The quality of the images in reconstruction with absorption correction depends also on the selection of a proper number for intensity levels of object/background in the reconstruction routines. An elaborated trial and error procedure is required to obtain the proper number for a desired image quality. Though, at the present time, it is not possible to confidently identify the origin of the trace elements, they may have been originated from the impurities associated with the fuel, SiC, and/or fabrication processes. Measurements planned for non-irradiated fuel balls and on fuel balls exposed to different fluences will help settle this issue.
Fig 8: Reconstructed images of the spatial distribution of trace elements with no absorption correction. The reconstructed elemental distributions are superimposed on the reconstructed shell from the elastic/Compton scattering.
Fig 9: Reconstructed images of the trace elements with absorption correction.
X-ray microtomography is an emerging technique made practical by high brilliance x-ray sources, advanced x-ray focusing optics and high-performance x-ray detectors. We have demonstrated that fluorescence microtomography technique can be a powerful tool for investigating elemental distributions in materials. The technique is nondestructive and noninvasive with high spatial resolution. It is sensitive to high Z trace elements in a low Z matrix with PPM accuracy, and has a good signal-to-noise ratio. Fluorescence microtomography can be used to simultaneously identify the elemental distributions of many trace elements. The main drawback of fluorescence microtomography is the slow data collection rate. The results of this experiment point to several ways to greatly accelerate data collection. For example, with a broad-bandpass monochromator, the fluorescence signal can be increased by 2-3 orders of magnitude with no loss in spatial resolution. With such an intense probe, it will be necessary to use wavelength dispersive spectrometers with integrating rather than single photon detectors. Such an arrangement will simultaneously increase spatial resolution and will greatly increase the sensitivity for minor trace elements of interest.
It should also be noted that in our experiment the beam energy was much lower than the K absorption edges of heavy elements such as Cs, Ce, Eu, and Ru. We therefore could not excite the K lines of these elements. The L lines are masked by overlapping K lines of less interesting trace elements. To measure the elemental distribution of Cs for example, either a crystal spectrometer or a high energy (>36 keV) x-ray probe is required. We intend to try bothmethods in future measurements.
Research sponsored in part by the Laboratory Directed Research and Development Program of the Oak Ridge National Laboratory and the Division of Materials Sciences, U.S. Dept. of Energy under contract DE-AC05-96OR22464 with Lockheed Martin Energy Research Corporation. Experimental measurements were made on beamline 2-ID-CD at the Advanced Photon Source ANL which is supported by the U.S. Dept. of Energy Basic Energy Science. The authors express their appreciation to Dr. E. D. Specht for his technical assistance and comments.
|© NDT.net - |||Top||