Click for next page ( 15


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 14
2 Background on Radiation Detection Portal monitors contain detectors for both gamma rays and neutrons. There are thousands of known radionuclides, and most of them emit one or more gamma rays, so most radioactive materials emit some gamma rays. A single radionuclide can emit one or many distinct gamma rays, each having a characteristic energy and intensity, 14 resulting in a gamma ray spectrum 15 characteristic of that radionuclide. The intensity depends on the probability of emission in each decay event and the amount of the nuclide present. 16 The ability to reliably determine the presence of a radionuclide, especially in the actual or potential presence of other radionuclides, depends on having a detector with sufficient sensitivity and energy resolution. The neutron detectors employed in the radiation portal monitors (RPMs) do not resolve the energies of the neutrons. However, this is not a major drawback because relatively few radionuclides or combinations of radionuclides emit neutrons, and nearly all of those sources are of interest for security reasons. The detection of neutrons, then, is a strong indicator of the presence of threat material and the need to interdict the truck. Although many of these radionuclides and their daughters also emit alpha or beta particles, only gamma rays and neutrons are sufficiently penetrating to be detectable outside of a shipping container that holds the radiation source. When screening cargo, the Department of Homeland Security (DHS) tries to identify whether the cargo contains radionuclides useable in a radiological or a nuclear weapon. 17 The radionuclides of greatest concern for radiological attacks have been identified in several studies, including those by the U.S. Nuclear Regulatory Commission and the U.S. Department of Energy, the International Atomic Energy Agency, and others (see NRC-DOE 2003; IAEA 2003; NAS 2008), and include americium-241, cesium-137, cobalt-60, iridium-192, and strontium-90. Some of these radionuclides are easy to detect if they are present in significant quantities. For example, cobalt-60 emits two relatively high energy gamma rays with each disintegration, one at 1173 keV and one at 1333 keV. Strong cobalt-60, cesium-137, and iridium-192 (used in radiography) sources require heavy shielding to enable people to work near them. The materials of greatest concern for nuclear explosive devices are called direct-use nuclear materials—materials that are directly useable in a nuclear explosive device (this includes special nuclear material: uranium-233, uranium-235, and plutonium)—and do not necessarily require heavy shielding. For example, uranium-235 emits one intense gamma ray of energy 185.7 keV in 57.2% of its disintegrations. Plutonium-239 emits numerous weak (low-intensity or low-probability) gamma rays. The strongest, most readily detectable of these have energies of 14 For example, iridium-192 emits dozens of gamma rays as it decays, 4 of which are intense (iridium-192 emits them in approximately 30 percent or more decays).. Iodine-131, used in medicine, emits one intense gamma ray (emitted in over 80 percent of decays) and 17 other gamma rays (emitted in between 0.00009 and 7 percent of decays). 15 A gamma ray spectrum, the set of gamma rays of different energies emitted by a source, is represented as a plot of the number of gamma rays versus energy. 16 The relative intensities of multiple gamma rays from a single radionuclide can also be used to help identify it. 17 A radiological weapon uses radioactive material to cause harm based on the radiation the material emits. A nuclear weapon uses nuclear reactions to release large amounts of energy in a nuclear explosion, which also releases radioactive material. A radiological weapon is unlikely to kill many people, but can cause harm and economic damage. A nuclear weapon is the most devastating weapon in the U.S. arsenal. 14

OCR for page 14
CHAPTER 2: BACKGROUND ON RADIATION DETECTION 15 51.6, 98.4, 129.3, 375.1, 451.5, 650 (a “multiplet” containing about a dozen gamma rays of nearly the same energy), and 769.3 (doublet) keV. Plutonium also emits neutrons because of spontaneous fission. No real material is purely composed of one radionuclide. Other nuclides, including other radionuclides in many cases, are present because they are byproducts of the creation of the radionuclide or because they are decay products of the main radionuclide. Highly enriched uranium (HEU) contains, by definition, at least 20 percent uranium-235, with the rest typically being uranium-238 and trace quantities of uranium-234. Even weapons-grade uranium (generally considered to be at least 90 percent uranium-235, and what a weapons state would use in a nuclear weapon) may contain up to 10 percent uranium-238. The composition of plutonium typically has even more isotopes in measureable quantities: some mix of plutonium-238, -239, - 240, -241, and possibly -242. A notional gamma-ray spectrum would show up simply as a curve with peaks (spikes of counts) at the characteristic gamma-ray energies, but zero counts everywhere else. As discussed below, the width of the peaks, or “energy resolution” is different for different types of detectors. Real spectra are necessarily more complicated, due to the existence of alternative physical mechanisms for absorption and scattering of gamma rays in the detector and, to a lesser extent, to imperfections in the way different types of detectors and individual detectors of the same type operate. The most important difference between a real spectrum and the above-described “notional” spectrum is the presence of a broad continuum of gamma rays caused by Compton scattering. 18 For a gamma ray of given energy, the continuum lies below the peak and has a predictable shape, based on the gamma-ray energy and the composition and size of the detector. The continuum tends to fill in the regions between the peaks, and can make it difficult to identify peaks if the peaks are broad (i.e., in low-resolution detectors), and/or the peaks are weak compared to the continuum. Additionally, as noted, shielding can attenuate the peaks and add to the continuum, and additional radiation from natural background and masking materials can introduce additional gamma-ray peaks and add to the “Compton” continuum. The combination of these effects complicates the spectrum and creates a formidable challenge to the identification of radionuclides, especially with detection systems of relatively low resolution like thalium- activated sodium iodide [NaI(Tl), sodium iodide or NaI for short] detectors. Thus the challenge of testing the ability of a system to detect and identify a particular source under varying conditions is great. This chapter describes important technical aspects of passive detectors used to detect radiation from sources located in cargo containers. SHIELDING The observed gamma-ray spectrum from a source, (e.g., special nuclear material, SNM) is influenced by the presence and distribution of surrounding materials, which attenuate and scatter gamma rays by absorption and Compton scattering. These materials can include, containers, other materials being shipped with the source or placed around it in an attempt to shield it from detection (shielding), and an air gap. Attenuation even occurs in the radioactive material itself (self shielding). High-energy gamma rays are attenuated less than low-energy 18 Compton scattering is a fundamental physical process in which a gamma ray scatters off an electron, giving up some of its energy to the electron and retaining the rest in the scattered gamma ray. When this occurs in the detector (and the scattered gamma ray exits the detector without further interaction), the detector “sees” the energy of the scattering electron. When it occurs in material outside the detector (e.g., shielding), the scattered gamma-ray might be detected by the detector. Because the amount of gamma-ray energy given up to the electron varies continuously, the result in either case is a contribution to the continuum in the detector.

OCR for page 14
16 EVALUATING TESTING, COSTS, & BENEFITS OF ASPs: INTERIM REPORT gamma rays, and high-atomic number (high-Z) materials, such as tungsten and lead attenuate more than low-Z materials, such as aluminum and wood. High density materials also attenuate more than low-density materials, and there is a correlation between pure high-Z materials and high-density materials, and between pure low-Z materials and low-density materials. Shielding can reduce the intensity of gamma-rays observed by the detector (both peaks and continuum), and also shift the continuum to lower energies and increase its intensity relative to the peaks. Self Shielding An important consideration for detection of nuclear materials is self shielding. Because uranium and plutonium are very heavy elements with large number of electrons (high-Z), they strongly absorb gamma rays. Gamma rays produced in the interior of a thick piece of nuclear material are likely to be absorbed within the material. This effect makes it more difficult to detect these materials. HEU emits very few neutrons, virtually all of them from the small percentage of uranium-238 present in the material. 19 Consequently, one cannot reliably detect HEU with a neutron detector. Plutonium emits neutrons, most of which are emitted by the isotope plutonium- 240. Self-shielding has little effect in diminishing neutron emission, because sub-critical multiplication actually increases the neutron emission. Testing of RPMs carried out with SNM has demonstrated the detection of plutonium with some shielding and, in some tests, HEU. With sufficient shielding, passive detectors would fail to detect even large quantities of these materials. MASKING Masking is the phenomenon that occurs when benign radioactive materials obscure the signature of a radionuclide of interest. This occurs when the benign radionuclide either overwhelms the detector with a stronger signal or creates spectral signals that compromise the algorithm’s ability to analyze the spectra. Multiple radionuclide sources in the cargo, including masking materials, produce spectra that are linear sums of the spectra of individual radionuclides. The geometry of the source and masking materials can affect the spectrum, because different radioactive materials can be located in different positions relative to the detectors and any shielding materials. In addition, the interaction between the radioactive material and the shielding or masking material may result in secondary emissions that could confound identification. When testing for the effects of shielding, both high-Z and low-Z materials should be investigated. High-Z materials are more effective at attenuating gamma rays, especially low-energy gamma rays, whereas low-Z materials, notably materials containing hydrogen atoms, enhance the absorption of neutrons. 20 High-Z materials close to a source that emits beta particles, such as strontium-90, will enhance the production of Bremsstrahlung, a continuum spectrum of photons (gamma rays) resulting from the stopping of electrons. If shielding and masking are used in combination, it is important to consider scenarios where the masking material is closer to the source than the 19 HEU containing very small amounts of chemical impurities emits some neutrons, produced by reactions of alpha particles on light elements in these impurities. 20 Neutrons lose more of their kinetic energy in collisions with low-Z nuclei than with high-Z nuclei, and low-energy neutrons are much more likely to be absorbed than high-energy neutrons.

OCR for page 14
CHAPTER 2: BACKGROUND ON RADIATION DETECTION 17 shielding and vice versa, because the order may affect the observed gamma ray spectrum and the angular dispersion of scattered gamma rays. Finally, all radioactive decay data are subject to statistical variations, which are particularly significant for weak sources. 21 HOW DETECTORS WORK Gamma-ray detectors and neutron detectors are used in both the proposed Advanced Spectroscopic Portals (ASPs) and the currently-used PVT portals. Both systems use moderated helium-3 proportional counters for neutron detection. For gamma-rays, the ASP system uses thallium-activated sodium iodide [NaI(Tl), sodium iodide or NaI for short] detectors for gamma- ray detection, and the PVT system uses polyvinyl toluene based plastic scintillation detectors for gamma-ray detection. The plastic scintillator is made of a polyvinyl toluene solvent with a (typically) p-terphenyl solute. After mixing the two materials, the solvent is polymerized to make the plastic. Another system proposed for the ASP portals uses high purity germanium (HPGe) detectors for gamma-ray detection. How each detector works is discussed in the following paragraphs. Sodium Iodide Detectors Sodium iodide detectors consist of a NaI crystal containing approximately 0.1% thallium, coupled optically to a photomultiplier tube. 22 They are scintillation detectors: gamma rays interact with the detector to produce low-energy photons in the energy range of visible light, a phenomenon called scintillation. The NaI crystal is transparent to light. A scintillation photon is captured by the photomultiplier, which converts it into an electron, which is then accelerated and amplified in the photomultiplier to produce many electrons. The total electrical signal at the output of the photomultiplier is related to the sum of the photons from across the crystal and is roughly proportional to the energy deposited in the detector by the gamma ray, so the size of this signal is logged and tallied as the energy of one gamma ray. NaI detectors are expensive compared to plastic detectors, but inexpensive compared to HPGe detectors. (See below.) The range of energies detected for the full-energy gamma-ray peak in a NaI detector is typically around 8% of energy FWHM. 23 In other words, the peak in a NaI detector spectrum from a 1 MeV gamma ray might be 80 keV wide. This is relatively low (poor) energy resolution for gamma spectroscopy. When several different gamma rays are closer together in energy than the detector resolution, as can be the case with some sources observed by NaI detectors, it is difficult to identify them all. This is particularly true for a weak gamma ray (one with few counts observed) close to a stronger one. Another problem with a low-resolution detector, such as NaI, is that a weak gamma ray peak can be difficult to observe above the Compton continuum from 21 Radioactive decay is measured by detector counts of emitted particles and is modeled most naturally and rather faithfully by the Poisson distribution with standard deviation equal to the square root of its mean; hence a good estimate of the statistical variation in the number counts N is √N , and the variation relative to the count is √N/N = 1/√N. 22 Originally, single crystals were grown for the detectors. Currently many of the detectors are made of a polycrystalline material that has better resistance to cleavage from mechanical or thermal shock. 23 FWHM stands for full width at half maximum, the width (energy spread) of the peak in the spectrum at half the height of the peak above any underlying continuum. The low resolution of NaI detectors is the result of low efficiencies in the conversion of gamma-ray energy to energy of the light photons, and a low yield of electrons (around 0.15 per photon) at the photocathode.

OCR for page 14
18 EVALUATING TESTING, COSTS, & BENEFITS OF ASPs: INTERIM REPORT higher energy gamma rays, because the few counts in the peak are spread of a range of energies. Figure 2.1 shows a typical gamma-ray spectrum of naturally occurring radioactive material (NORM) measured with a NaI detector. Low detector resolution poses a challenge to analysis algorithms necessary to process the data and obtain meaningful conclusions, especially when there is a large statistical uncertainty in the data (i.e., few counts). NORM (Laboratory Room Background) 1000000 232 Th and daughters 238 U and daughters 100000 40 K 10000 Counts 1000 100 10 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 3200 3400 3600 Energy (keV) Figure 2.1 A gamma-ray spectrum gathered from the background radiation in a laboratory using a sodium-iodide detector. The x-axis is the energy in keV and the y-axis is the number of gamma-ray detections counted within a particular energy range. Gamma-ray peaks from common background radionuclides are labeled on the figure.

OCR for page 14
CHAPTER 2: BACKGROUND ON RADIATION DETECTION 19 PVT Detectors Like NaI detectors, PVT detectors are scintillation detectors. Unlike crystals, plastic scintillators can easily be fabricated into large detectors, and are relatively inexpensive. The larger size permits the detection of a larger number of events from the same gamma-ray source. However, the low density, low light yield, and especially the low atomic number 24 of the plastic scintillator combine to make the detector much less effective than NaI for spectroscopic measurements. They provide only crude information about the gamma-ray energy. Figure 2.2 shows a typical gamma-ray spectrum of radionuclides measured with a PVT detector, illustrating the absence of observable full-energy peaks. Figure 2.2. A calibration gamma-ray spectrum gathered by a PVT portal monitor. The background has been subtracted. SOURCE: Stromswold et al. (© 2004 IEEE). HPGe Semiconductor Detectors Gamma-ray spectrometers based on high-purity germanium (HPGe, or germanium) detectors are widely used as laboratory scientific instruments. Their energy resolution is typically around 0.1-0.2% FWHM of the gamma-ray energy, nearly two orders of magnitude better (narrower peaks) than a NaI detector. An HPGe detector is a semiconductor ionization-type detector, which operates on a different principle from NaI and PVT detectors. In an ionization detector, the gamma-ray energy is converted directly into electrons, which form the signal proportional to the energy deposited by the gamma ray. 25 Figure 2.3 shows the HPGe spectrum of the same NORM source whose measurement with an NaI detector was shown above (Figure 2.1). The advantage of the higher-resolution 24 Materials with low atomic numbers have almost no photoelectric interactions with gamma rays and therefore exhibit no full energy peak. 25 The high resolution of germanium detectors results from efficient conversion of the gamma-ray energy into electrons.

OCR for page 14
20 EVALUATING TESTING, COSTS, & BENEFITS OF ASPs: INTERIM REPORT NORM 1000 232 Th and daughters 238 U and daughters 235 U 40 K 100 Counts 10 1 Lead x-rays (from lead bricks near detector) 0.1 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 Energy (keV) Figure 2.3 A gamma-ray spectrum gathered with a germanium detector from the background radiation in the same laboratory mentioned in Figure 2.1. Again, the x-axis is the energy in keV and the y-axis is the number of gamma-ray detections counted within a particular energy range. Gamma-ray peaks from common background radionuclides are labeled on the figure. detector is evident. (Note that the presence of uranium-235, whose strongest gamma-ray is buried in the continuum with a NaI detector, is clearly revealed in the HPGe spectrum.) Although the higher energy resolution of HPGe detectors is essential in many laboratory measurements and would be desirable for detecting nuclear and radiological materials, especially under difficult conditions (e.g., masking), these detectors have other characteristics that make their widespread use in RPMs problematic. The main drawbacks are the difficulty of producing detectors in very large sizes needed to detect relatively small amounts of radiation in a short time, and the high cost per detector, which makes it expensive to use large numbers of them in an RPM. Also, the detectors must be cooled to low temperatures, requiring liquid nitrogen or special, mechanical or thermoelectric cooling devices. Neutron Detectors Neutron detection in the RPMs use helium-3 proportional counters, a type of gas-filled ionization detector that has built-in amplification caused by a complex process of charge multiplication. The detectors are embedded in polyethylene, which acts as a “moderator,” slowing (“thermalizing”) the neutrons emitted by sources of interest to low energies. 26 26 In the detector, the neutron reacts with a helium-3 nucleus to produce an energetic proton (p or hydrogen-1) and a triton (t or hydrogen-3). This reaction has a low cross-section (probability) for all but low-energy neutrons, so

OCR for page 14
CHAPTER 2: BACKGROUND ON RADIATION DETECTION 21 Because the neutrons must be slowed down to make them detectable, the counter does not measure the energy of the incident neutrons. This means that the detectors measure no useful spectroscopic information about the neutrons. Because very few radionuclides emit neutrons, and almost all of them are of security interest, this is not a serious drawback. Neutrons are simply counted. The mere detection of any neutrons above the low natural neutron background counting rate signals a likely cause for concern. 27 ANALYZING SPECTRA Once a gamma-ray spectrum has been collected, it must still be analyzed to identify the radionuclide(s) that generated the radiation. Two different strategies have been employed for this analysis in the ASPs: peak matching and template matching. Each has advantages in some configurations. It is also likely that the available algorithms could be improved by involving more of the science and engineering community to work on these problems. As described in the previous section, a detector pulse-height spectrum for a monoenergetic gamma-ray source has a peak centered on the full energy of the incident gamma rays and a continuous tail at lower energies caused by Compton scattering. A peak-matching algorithm identifies the full-energy peak and matches that energy to the signature energies in its library of radionuclides. Many radionuclides have multiple characteristic gamma rays, and more than one radionuclide may be present in cargo, so the algorithm must be able to identify and match multiple peaks in a single spectrum. The advantage of peak matching is that there is always a full-energy peak that is separate from the Compton tail. One disadvantage is that only a fraction of the detector counts are in the full energy peak. Peaks are also obscured by the Compton distribution of higher-energy gamma rays. Also, attenuating material between the source and the detector can drastically reduce the number of full-energy gamma rays that even reach the detector, making it difficult to differentiate full-energy peaks from the background. A template-matching algorithm has a library of energies of gamma rays emitted in radioactive decay and also a library of full detector spectra from radionuclides with intervening attenuating materials. Template matching compares not just the full-energy peak, but the whole spectrum to its libraries. An advantage of template matching is that all of the detector counts are used toward identification, and the effect of shielding can be accounted for, at least approximately. The challenge in template matching is the nearly limitless set of combinations of sources and attenuating materials and thicknesses, along with background radiation. Although software implementing algorithms for gamma-ray spectral analysis has been the subject of intense development in the national laboratories, and several vendors of spectrometers provide such software, there are in fact few commercial products available for radionuclide identification using gamma spectroscopy. 28 A particular problem is the dearth of neutrons must be slowed down for them to be detected well. The reaction energy is carried off by the proton and triton, which lose their energy by ionizing atoms in the detector gas. The ionized atoms make an electrical signal that is amplified by the proportional counter. 27 One likely source of neutrons is plutonium. Neutron sources such as the isotope californium-252 and mixtures of natural or man-made alpha-emitting isotopes with beryllium are used in some applications, including downhole measurements in oil wells. 28 Isotope identification software and algorithms are different from the ASP software for interaction with the operational hardware (occupancy sensors, gate arms, etc.). The former are exchangeable modules that analyze spectral data found in data files that follow standard formats. The latter are specific to each vendor’s ASP.

OCR for page 14
22 EVALUATING TESTING, COSTS, & BENEFITS OF ASPs: INTERIM REPORT commercial software for the complex problem of analyzing sources that can be shielded and masked with low-resolution (NaI) detectors—a problem of current interest mainly to detection systems for nuclear and radiological materials. Even the scientific literature on this topic is sparse. Engaging the broader science and engineering community in this challenge could lead to more sophisticated analytical methods from statistics and signal processing being applied to radionuclide identification, resulting in better algorithms. Radiation Detectors at Ports of Entry Today As noted in Chapter 1, the RPMs currently in use are PVT plastic scintillation detectors. Because these detectors are inexpensive and easily fabricated in large volumes, they can be made to be quite sensitive to radiation. But PVT detectors have very poor energy resolution; they cannot distinguish one gamma ray energy from another, except over broad energy ranges, so they have very limited ability to characterize the source of those gamma rays. The PVT detectors in the RPMs at most ports of entry have been equipped with crude energy resolution in the form of energy windowing: The gamma-ray events are binned into four large energy windows. Although these energy windows are too broad for isotope identification, the ratios of the counts in different windows and to background levels in the same window help to identify the presence of radiation sources that require further examination. 29 The RPMs are also equipped with moderated helium- 3 neutron detectors. The RPMs alarm if the container occupancy causes the RPM to exceed a gross gamma-ray counting threshold, exceed an allowed gamma-ray energy windowing ratio value, or exceed a gross counting threshold for neutrons. The spectrometer currently in use in the secondary inspection is a handheld radioisotope identification device (RIID) which contains a small NaI detector. At some ports of entry, the container may also be subjected to additional interrogation inspections such as imaging with an X-ray type machine (a radiography device with a gamma or X-ray source) to look for localized heavy metal objects (shielding or SNM), and direct examination of the cargo, including removal from the truck or shipping container. This or other suspicious results can trigger additional inspections. The gamma ray alarm threshold (the count rate above which the alarm is triggered) is established for each port based on the threat guidance and a number of other factors. 30 Performance of the RPMs relative to the threat guidance is tested by measurements using standard sources that are not special nuclear material, but have gamma-ray signatures that are similar to that of plutonium or uranium, and so can serve as surrogates. CBP has said that the threshold is selected to balance the needs for sensitivity for commerce to flow. Although most RPM gross-gamma-count thresholds are set to meet a particular guidance level, a fraction of them are set to a different level. Using the energy windowing mentioned above, PNNL reports that all of the RPMs can detect a plutonium surrogate source that is lower than the guidance activity (i.e., the RPM is more sensitive than the plutonium guidance). NEXT GENERATION RADIATION DETECTION TECHNOLOGIES: ASPs 29 All deployed SAIC RPM8 PVT systems have energy windowing algorithms that use four energy windows. CBP also has some Ludlum RPMs that only have two windows and hence 1 ratio on which to alarm. 30 The threat guidance, which is classified, was established by the Department of Energy in a 2003 letter to Parney Albright, assistant secretary of homeland security for science and technology.

OCR for page 14
CHAPTER 2: BACKGROUND ON RADIATION DETECTION 23 The goal of DHS in replacing the PVT RPM systems with the ASP technology is to address three perceived needs (Test and Evaluation Master Plan, August 2008):  “To improve the detection of nuclear weapons and radiological/nuclear threat sources  To reduce the burden associated with unnecessary inspection of conveyances with only naturally-occurring radioactive material (NORM)  To improve the consistency and accuracy of the identification of nuclear weapons and radiological/nuclear threat sources” Specifically, the ASP performance specifications called for systems that can detect and identify SNM, weapon-indicating radionuclides, NORM, medical radionuclides, and industrial radionuclides alone and in combination. According to DHS, the portal detection systems should respond consistently and predictably, and should assist CBP personnel in determining whether to release a conveyance or to detain it per the agency’s standard operating procedure or concept of operations (CONOPS, from the Performance Specifications July 2007). The ASP systems are expected to detect and identify these threat materials when surrounded by “engineered shielding or masking and/or significant amounts of cargo.” Improved RPMs, such as ASPs, should tolerate a wide range of conditions including variations in natural background radiation, environmental stress and weather conditions, and should be able to accommodate low- and high-volume traffic areas. Benign sources of radiation in normal commerce (such as medical radionuclides) would not need to be sent to secondary inspection if they could be identified in the primary inspection. The ASPs were developed to provide both detection and identification of radiation sources in cargo containers. The portals use NaI or HPGe detectors, which provide greater differentiation in the detector response to gamma rays of different energies than PVT. With suitable software to analyze the gamma-ray spectrum, the source of the gamma rays can, in principle, be identified. At the time that the committee prepared this report, the HPGe ASP had not met requirements to undergo full testing by CBP and DNDO, 31 so the committee’s report focuses on the NaI systems. There are reasons to believe that the ASP could perform the functions now being performed in both primary and secondary screening, in most cases. A confident identification of NORM in primary screening would significantly reduce the number of referrals to secondary screening. For cases in which primary screening determines that the cargo is suspect, an ASP could be used also in a secondary screening with the container moving at a lower speed to obtain greater statistical accuracy and hence more effective identification. CBP is also considering a hybrid deployment, with some ASPs deployed primarily in high traffic ports, and retaining PVTs in other, lower-traffic ports, and using ASPs for secondary screening at all ports. The ASP has advantages over the combination of a PVT portal and RIID detector in secondary inspections. The detection and identification feature of the ASP is enhanced by the slower speed in secondary as compared to the primary. The ASP is larger than the RIID and therefore can collect comparable or better statistical spectral data (e.g., higher counts and hence lower relative statistical variation), and the ASP has better identification software. The ASP 31 As noted above, gamma-ray energy resolution in a HPGe detector is far superior to that in a NaI detector, but the cost of HPGe crystals is much higher than sodium iodide crystals. The cost and the difficulty of growing large HPGe crystals resulted in the HPGe ASP having a much smaller detector volume than the others. A smaller detector requires more time being exposed to get a statistically useful number of counts (detection events). Consequently the HPGe ASP could not meet the requirement to screen cargo containers passing at the speeds required in the systems specifications. Because it could not meet these criteria, the contract with the vendor of the HPGe ASP was not extended. A change in CONOPs to allow for longer exposure times could enable the HPGe detectors to operate in secondary screening, but performance with different CONOPs has not been evaluated in the DNDO program.

OCR for page 14
24 EVALUATING TESTING, COSTS, & BENEFITS OF ASPs: INTERIM REPORT localization of the source is better than that of the RIID because the ASP uses data from a continuous screening and can analyze time-slices of data. The larger detector has much better coverage of the cargo container. Most containers will spend less time in secondary inspection with an ASP because the slow-speed scan will confirm the presence of radiation sources that are only NORM, so that the manual (hand-held detector) survey would not be necessary unless the container is unloaded and a RIID would be used to investigate specific packages in the container. The ASPs are required to identify as well as to detect the radioactive material. As a result, assessment of the performance of the instrument is not limited to the sensitivity of the detectors, but also includes determining the level of confidence in the threat identification algorithm for each system. Although there is evidence that the spectral analysis programs work remarkably well under challenging circumstances, the two vendors’ algorithms appear to yield somewhat different results, and it is not clear at this point that either is optimal.