The Use of Cosmic Ray Muon Tomography in the Detection of Concealed High-Z Materials

[pheloniusfriar]

It's been a while since I posted, and I'm going to go retro on this one. I had misplaced where I had put the source for this one and so can only post it now (I only had the PDF). I did this in the winter term of my first year at university. I had no idea how to use the library, I had no idea how to do citations, I had no idea how to write a research paper. But I ended up, I found out later, writing a 3rd year independent research paper (good enough for an honours project in Integrated Science at Carleton apparently, but I didn't know that at the time). I was asked to do it by a professor as a prelude to summer employment working on the CRIPT project as a research assistant. A gig that ran full time in the summers and a part time (mostly, sometimes more) through the school year until the end of November 2012, when the project started to wrap up and my contract was terminated. It was a good run, and I learned an amazing amount. I also feel privileged to have been able to have such an important role on such a huge project as an undergraduate student (mind you, it was my electronics and software experience they used, but still). I continue to volunteer on the project and am doing some additional volunteer work on another project studying the flux of horizontal cosmic ray muons leveraging the work I did on CRIPT (for developing a detector that can image the inside of nuclear reactors, for instance that have melted down, without having to actually get near them... naturally occurring muons are amazing particles!). Since the project proper has wrapped up and the new detector is still in the proposal stage, I decided to take a Teacher's Assistant position in the PHYS1004 class in the winter term (the money will help quite a bit even though it's not much). Anyway, this was submitted on April 30th, 2010. Oh, and I got an A- on it... :)


The Use of Cosmic Ray Muon Tomography in the
Detection of Concealed High-Z Materials
I. INTRODUCTION

A. The need for screening

It is becoming ever more important to monitor the flow of goods and people as a deterrent against state, criminal, or ideological organizations that may wish to wage war or cause serious disruption through the use of various asymmetric weapons systems within the territories we wish to consider secure. To that end, increasing surveillance and intrusive inspections have been implemented at points where the greatest risk exists, for instance at airports and border crossings. For an effective deterrent, all traffic through these key points of commerce and travel especially, as well as the appropriate measures for points between, require 100% screening to be maximally secure. For historic and economic reasons, this strategy of complete coverage presents an extreme challenge to even the most affluent and security conscious of societies. Furthermore, any onerous impediment to the efficient movement of goods and people elicits an economic cost of its own that can destroy the very prosperity that such security measures wish to protect.

While it can be argued that the smuggling of conventional weapons poses the greatest chance of occurring and resulting in harm being inflicted through their use, all but the largest of instances of such smuggling into otherwise stable countries are dwarfed by the already existent availability of these items within those countries. Where the national government of a country needs to protect its citizens against all forms of weapons smuggling, it has a special obligation to prevent the use of chemical, biological, radiological, and nuclear (CBRN) weapons against its population, infrastructure, services, and legitimate foreign interests: “Asymmetric CBR threats provide an adversary with significant political and force multiplier advantages, such as disruption of operational tempo, interruption/denial of access to critical infrastructure and the promulgation of fear and uncertainty in military and civilian populations. [...] Proliferation will continue to dramatically increase the threat from the use of CBR agents by states or terrorist organizations against unprotected civilian populations. Proliferation also poses an asymmetric threat against non-combatants outside the immediate theatre of conflict, including Canadians at home.”¹ As such, most functional nations have embarked on integrated strategies to minimize the chances of CBRN related incidents. In general, those efforts can be categorized in five ways: supporting or directing the improvement of foreign CBRN control, detection, and enforcement; border CBRN detection equipment and domestic law enforcement training; the securing of legitimate CBRN materials within the country’s borders; improved intelligence operations to detect potential smuggling operations before they occur; and various domestic and international research and development project to improve overall control and detection capabilities.²

Furthermore, of the CBRN threats, there are emergency measures and possible mitigation that can be taken to minimize the impact to the population and infrastructure of a successful attack with chemical, biological, or radiological weapons; however, the damage that would be inflicted should a nuclear device be detonated in a populated area would be devastating beyond measure to both the fabric and spirit of the country, its operation, and its people. Such results make special nuclear materials³ (as could be used in a nuclear bomb) particularly attractive targets for terrorists⁴ (“independent” or state sponsored): “Nuclear smuggling is an increasing concern for international security because creating a viable nuclear weapon only requires several kilos of plutonium or highly enriched uranium. The International Atomic Energy Agency has documented 18 cases of theft of nuclear [weapon grade] materials within the last decade, and probably more instances have occurred without report. This is especially prevalent within the former Soviet bloc, where large amounts of nuclear materials are insecurely guarded and inventories are often faultily kept.”⁵

Of particular concern is the realization that the view, held since World War II³, that the effort required to build a nuclear weapon was prohibitive, is no longer valid. This opinion had been based on the American experience of creating two small nuclear weapons, but it is now widely accepted that the expertise and technical capability to build a viable nuclear weapon is no longer the exclusive purview of large, economically advanced nation-states. In fact, the knowledge and infrastructure required is potentially within reach of any well-organized and funded group with sufficient long-term determination and resourcefulness: “The only real technological barrier to the clandestine construction of nuclear weapons is access to fissionable material itself. There is a growing black market for this material, and eventually demand will result in enough material reaching as-yet unidentified buyers to produce a nuclear weapon”³. In addition to the smuggling of processed special nuclear materials, given that uranium is roughly 40 times more prevalent in the Earth’s crust than is silver⁶, the smuggling of uranium ore or low quality extracted uranium from such ore is also a more likely possibility.

While it is widely acknowledged that “most known interdictions of weapons-useable nuclear materials have resulted from police investigations rather than by radiation detection equipment installed at border crossings”², the asymmetric nature of the threat calls for exceptional measures in the effective detection of smuggled special nuclear and radiological materials that might make it past the intelligence operations to a port of entry into the country. Per the U.S. Container Security Initiative Strategic Plan: 2006-2011, “the cost to the U.S. Economy resulting from port closures due to the discovery or detonation of a weapon of mass destruction or effect (WMD/E) would be enormous. In October 2002, Booz, Allen and Hamilton reported that a 12-day closure required to locate an undetonated terrorist weapon at one U.S. seaport would cost approximately $58 billion. In May 2002, the Brookings Institution estimated that costs associated with U.S. port closures resulting from a detonated WMD/E could amount to $1 trillion, assuming a prolonged economic slump due to an enduring change in our ability to trade.”⁷ While this is a U.S. figure, it can be scaled appropriately to reflect the impact of such an event on any trading nation, or the domino effect such an act would have on global commerce if it happened anywhere.

B. Screening technologies

Intrusive and time consuming inspections or screenings for people or goods create a negative impact on and perception from companies and individuals trying to conduct their legitimate business. For instance, the ever increasing security burden for air travellers has a seriously negative impact on travel and the infrastructure cost of the free movement of people and goods, costing upwards of 615 million person-hours per year in the U.S. alone, or roughly $230 billion since “9/11” for enhanced security related to the generic measures that have been implemented⁸. Any potential technologies or methodologies that can reduce risk to acceptable levels and increase the flow of people and goods will result in a commensurate boost in the efficiency of the overall economy, and thus research into those areas can yield large direct financial returns in addition to protecting against the potential losses brought about by a CBRN incident. In the area of CBRN protection, and special nuclear materials in specific, the reduction in risk needs to be substantial to allow free movement with confidence, and many of the existing technologies such as radiation sensors, while valuable as part of an overall security strategy, are only partially effective in the detection of those materials⁹.
The remainder of this report will focus exclusively on the detection of special nuclear (“nuclear”) and radiological materials, where the nuclear materials are those high-Z elements that could be used in or adapted to the construction of a functional nuclear weapon, and radiological materials are generally lighter, highly radioactive, materials that could be used as a source of contamination or in a “radiological dispersal” device. It should be noted that most radiological materials of interest, due to their high level of radioactivity and short half-life, are either easily detected with radiation sensors or have to be heavily shielded to prevent that detection, and it is this shielding that can then be the target for detection strategies.

Detection technology can be broken into two general types of systems: passive and active, with radiation sensors forming the bulk of deployed passive systems, and x-ray imagers and scanners forming the bulk of deployed active systems. Additionally, systems can be separated into three classes: sensing, imaging, and tomographic, where sensor systems usually only provide direction and strength of a signal (e.g. a Geiger-Müller counter), imaging systems provide a 2D composite density projection view through an object (with many enhanced systems providing additional computed information about details to assist with visualization), and tomographic systems provide a slice-by-slice view of the internals of an object which can be re-assembled into 3D model of the combined slices to allow for various visualizations. Finally, systems are either stationary or scanning (the detector and/or sample move relative to the other).

Each of these approaches have benefits and limitations and each has its place in the strategy to control materials of interest as determined by the cost vs. capabilities of the systems, the severity and complexity of the risk it needs to address, and environment it needs to be deployed to. Countries all over the world have implemented research projects and have deployed various technologies and methodologies to enhance the strengths and address the identified limitations of the prevalent detectors of the day. For instance, as part of the US’s Container Security Initiative⁷, new scanning systems are continually being evaluated and deployed. Of these, gamma ray systems are increasingly being used to scan cargo containers, and next-generation radiation sensors are being installed at critical traffic flow points, such as access roads to airports⁷. While improvements continue to be made, the issue of detecting high-Z elements that could be used in a nuclear device, as well as shielded radiological sources that could be used in a dispersal device, remain among the greatest challenges being faced in this field.

1. Radiation sensors

The first line of defence for any detection capability (of nuclear and radiological materials) needs to remain radiation sensors. Firstly, they will be able to detect any poorly shielded attempts to smuggle or even legitimately ship these materials. Secondly, they can provide an early indicator that a shipment is radiologically dangerous and that personnel have to invoke hazmat and/or security protocols to deal with the shipment. Effective radiation sensors were originally of the Geiger-Müller type, introduced in 1928¹⁰ and comprised of a high-voltage gas ionization tube operating near its breakdown region and an electronic readout system to provide an indication of the number of interactions detected (but not the energy of the detected events)¹¹. This device could be made portable and inexpensively, and was the workhorse of radiation detection until the nuclear age when more sophisticated methods were needed for many applications. The Geiger-Müller type detectors are still useful as a basic radiation monitoring device and for wide cost-effective deployment in the field. In the late 1940s, the development of the proportional chamber allowed the energy of radiation to be measured as well and modern variations with computer analysis can use the data to more selectively determine which events are background or expected sources of radiation (e.g. from kitty litter, which causes 34% of false detection alarms at the U.S. border)¹² versus from likely materials of interest (e.g. ¹³⁷Cesium, ⁹⁰Strontium, ¹⁹²Iridium, and ⁶⁰Cobalt)¹³. The proportional chamber is also relies on gas ionization and high voltages; however, it operates in the gas’ “proportional region” where the current of the signal generated is proportional to the energy of the ionizing radiation¹¹.

Two other commonly used technologies for radiation detection are scintillators, which emit light in response to radiation passing through them, and semiconductor sensors which directly produce an electronic signal. Scintillators must use technologies such as photomultipliers or semiconductor avalanche photodiodes in order to produce a usable electronic signal for further processing. Like proportional chambers, it is possible to extract information on the energy of a radiation event from the response of the integrated scintillator (material and electronics). Thallium-activated sodium iodide, NaI(Tl), based scintillators became the first commercially available method for gamma ray spectrometry in the 1950s¹⁰, and are now widely used for precision spectrographic analysis of radiation in advanced radiation detectors as part of efforts to reduce false alarms¹². Plastic scintillators are used where where large areas must be covered or the sensitivity of NaI(Tl) is not required, and liquid scintillators are available but not commonly used in security applications.

In the 1960s, practical methods of using solid semiconductors for the detection and analysis of radiation were developed¹⁰. Of these, silicon-based sensors form are at the heart of the bulk of deployed systems due to their ability to be used at room temperatures. Germanium sensors, on the other hand, require cryogenic cooling to operate, and thus tend only to be found in laboratory settings. In addition to the general utility of the materials in the detection of radiation, semiconductor fabrication advances have allowed numerous novel topologies to be developed that allow for spatial analysis of incident radiation. Silicon tends to be used for low energy x-ray and beta particle detection and analysis, whereas germanium is used for gamma rays.

Many radiation sensors in active use today use the same basic technology that has been in use for decades, but with more sophisticated readout electronics and computer interfaces. The coupling of this data acquisition and processing ability increases accuracy and can be used to cross-correlate data from multiple detectors, when available, to increase sensitivity and reduce false positives. For security applications, there is a trend towards the increasing use of sensor sophistication and powerful computer analysis of the resultant data¹². Two primary goals are being actively pursued in this area: the reduction of “nuisance alarms”, and sensitivity to the radiation signatures of specific materials of interest¹⁴. Particularly, sensors that have good directionality or even the ability to determine spacial information about the path of the event (e.g. some semiconductor detector architectures are able to extract spacial information¹¹), coupled with discrimination algorithms that can eliminate events that are likely from non-risk sources can drastically reduce “nuisance alarms” and thus greatly enhance the effectiveness of such measures in a repetitive security screening environment¹⁵. Research in this area is an ongoing effort; however, it is imperative to overcome the limitations of exiting systems. From a US General Accounting Office report: “current portal monitors deployed at U.S. borders can detect the presence of radiation but cannot distinguish between harmless radiological materials, such as ceramic tiles, fertilizer, and bananas, and dangerous nuclear materials, such as plutonium and uranium. DNDO is currently testing a new generation of portal monitors”². Once these new radiation sensors are deployed, and if they function as expected, their effectiveness in combating radiological and nuclear threats will be greatly enhanced.

2. 2D imaging systems

While 2D x-ray and gamma ray imaging systems are a critical part of border security, they play a limited role in the detection of nuclear and radiological materials as imaging systems per se. Where they provide intelligence of import is in identifying when a cargo might contain things that are not listed on its manifest, or by detecting that heavy shielding is present in the shipment, thus flagging the shipment as “suspicious” and in need of further inspection. However, because of ability of the 2D sensors to also detect radiation, these sensitive imaging systems can be used in much the same way as the simple radiation sensors described above, thus reducing the need for additional sensors, and in some cases with appropriate data processing may be able to provide some spacial information related to the strength of a detected radiation signature. Imaging systems can be stationary, where a “picture” is taken of the object of interest; or scanning, where the object and/or scanner are moved relative to each other. Scanning systems tend to be the only practical method of imaging large objects such as shipping containers.

All active imaging systems as described above suffer from two main drawbacks: the need for a radiation source, and the fact that a relatively high luminosity is required to “look through” most cargo (although the exposure times are kept as low as possible to minimize the overall dose if a person or animal is present). While radionuclides were once used as x-ray sources, most modern x-ray systems now use linear accelerators to generate the x-rays by colliding electrons into a solid target and then collimating the resultant beam. The primary source for gamma rays in security scanning applications, e.g. the VASIS system, are shielded radioisotopes, for instance ⁶⁰Cobalt or ¹³⁷Cesium radionuclides¹⁶. Obviously, the wide deployment of these systems actually introduces the issue of control of yet another radionuclide source, and must be carefully considered as part of the overall strategy of controlling the availability of dangerous materials for potential smuggling. In addition to the requirement of having to have a radiation source, concern about public and occupational exposure to radiation during travel (versus in a medical context), limits the acceptability of pervasive use of these active technologies. This consideration limits the potential applications of these technologies to applications where no human or animal exposure is likely, or in those situations were the risk is sufficient to justify exceptional procedures to evaluate the potential danger of a cargo.

3. Tomographic imaging systems

Tomography is a method of obtaining a clear image of a two-dimensional slice through a three-dimensional object by keeping the desired slice in the focal plane of the imaging medium or detector while moving the source and imaging medium around the object of interest. The focal plane remains clear while other planes are smeared and out of focus and, thus, amount to noise in the image, or tomograph. This technique, using x-rays or other penetrating radiation, to form images of the insides of objects was pioneered in the 1930s and has become a cornerstone of modern diagnostic medicine and non-destructive industrial material analysis. Modern tomography systems came of age in 1972 and use computer processing to process the tomographic information to enhance the signal to noise ratio of the image, identify critical features in enhanced visualizations, and to allow multiple slices of the object to be assembled and analyzed as a three-dimensional object¹⁷. Further data processing algorithms have allowed for the interpretation of scattering information as radiation passes through a material – versus attenuation information used in traditional x-ray tomographs – as a means of extracting additional information on the composition and structure of the internals of an object. These methodologies can now employed, with statistical analysis, to use the multi-path scattering of radiation to produce tomographic images provided sufficient information can be determined about the energy and direction of the radiation as it enters and leaves the object.

From a security perspective, active x-ray and gamma ray tomography could provide an excellent tool to non-intrusively verify the contents of traffic moving across borders and into and out of key transit hubs. The primary issue with these systems is that they require a relatively complex system to produce the tomographic imaging, with the radiation source and imaging units generally having to move around the scanning target to produce the image (although advances have allowed systems with fewer or even no moving parts to be needed, the principles remain the same). Regardless, the amount of time to produce the tomograph is relatively significant compared to 2D imaging techniques, and the amount of radiation exposure needed for a tomographic image is inherently higher than what is needed for a 2D image and thus presents health concerns. Furthermore, even gamma radiation, with an energy of in the low MeV range, can only penetrate roughly 1cm¹⁸ or 2cm¹⁹ of lead (or 15cm of steel²⁰) and provides only minimal improvements in the detection ability of nuclear materials or the shielding needed to smuggle radiological materials – while easier to find in a shipment of heavy equipment with the reduced clutter provided by tomographic imaging techniques, it would still be far too easy for a competent smuggler to use legitimate goods to mask the materials.

The issue remained that effective detection of properly shielded nuclear material is one of the critical challenges remaining to the international security establishment. From a 2008 report by the US General Accounting Office, “we found that a cargo container containing a radioactive source was not detected as it passed through radiation detection equipment that DOE had installed at a foreign seaport because the radiation emitted from the container was shielded by a large amount of scrap metal. Additionally, detecting actual cases of illicit trafficking in weapons-usable nuclear material is complicated: one of the materials of greatest concern in terms of proliferation—highly enriched uranium—is among the most difficult materials to detect because of its relatively low level of radioactivity”². It is in this environment of the need for a specific solution to this highly asymmetric threat that researchers have begun investigations into the use of passive cosmic ray muon tomography systems as a way of addressing the shortcomings of other technologies.

Muon detection and analysis is a field that evolved in high-energy physics laboratories in response to searches for new particles or decay mechanisms, and also became important in the characterization of cosmic rays. Through the 1970s to the mid-80s, the importance of muon analysis grew: “From the past practice: ‘Some outside chambers to also detect muons’, it changed to: ‘The need for accurate and complete muon detection by charge and momentum’”²¹. Work has continued in this field since and these advances, funded as part of pure research projects, are at a point of maturity where field deployable systems are within reach through modest research and development programmes. Passive Muon Tomography (PMT) systems are a relative newcomer into the arsenal of nuclear material detection in security applications; however, it is a technology that specifically addresses the limitations of other, existing, technologies by being able to use the pervasive cosmic ray background on Earth to “look through” vehicles and containers to detect the presence high-Z nuclear materials or shielding that might mask the radioactive signatures of other materials of concern. Although muons are unstable charged particles, those generated from cosmic ray showers have sufficient momentum to survive the trip to the surface of the Earth, even as they are slowed somewhat interacting with the atmosphere. Furthermore, since they are not subject to the strong force, they do not lose momentum through nuclear interactions but rather much more slowly through the Coulomb force due to their charge. Where gamma rays imaging systems can penetrate roughly 1cm to 2cm of lead shielding and 15cm of steel, muons that reach the Earth’s surface from cosmic ray interactions in the upper atmosphere can penetrate roughly 2 metres of lead¹⁸ and more steel than could be used inside a shipping container or truck.

The generic architectural requirement of PMTs is that they have to enclose the target on at least the top and bottom and require multiple layers of detectors to determine the trajectories of detected muons before they enter and after they leave the object being probed. Further layers of detectors are required to determine the momenta of the muons as required for use in the data analysis. Because each of these detectors must be several metres on a side, careful construction and calibration techniques must be employed. The more specific limitations are that, unlike a man-made tomography system which will rotate around the object of interest, the passive muon tomography system relies on cosmic ray muons which will naturally penetrate the object from multiple angles and allow a tomographic image to be constructed by employing statistical analysis of the multiple-scattering of the muons off the nuclei that comprise the object²². Besides the general complexity of a muon tomographic system, one final limitation needs to be addressed: the density of the muon flux at the Earth’s surface. Since there is roughly only one muon event per square centimetre per minute²³, the resolution requirements for detection have a profound impact on the exposure time required for a full “scan” (note however, that the size of the object does not affect the scan time as the number of muon events scales with the size of the detector surface area). Many of the challenges of using this new technology are compensated for by the relative sensitivity of muon tomography to the detection high-Z materials such as uranium, plutonium, and lead as those nuclei have relatively large effect on muon paths.

C. Muon Tomography Systems

Muon tomography systems must determine the trajectory of muons before they enter the target object and then again as they leave. In addition, the momentum of the particle needs to be determined in order to properly interpret the meaning of the measured angular deflection. As a muon passes through matter at relativistic speeds, it interacts with the atoms as it passes due to its charge and both loses momentum and has its trajectory altered (Coulomb scattering). At each interaction, a little more momentum is lost and the trajectory is altered again, and if the muon does not slow down sufficiently for it to decay (its relativistic velocity provides it with its “long” lifetime), it will emerge on the other side of the object for detection. Since the tomographic system determines the overall deflection of the particle from its original path and its final momentum, statistical analysis can be used, correlated with data from other measured muon paths, to produce a three dimensional image of the internal structure of the target object²².



Figure 1 - High-Z materials cause higher scattering angles than lower Z materials
Similar techniques of using multiple Coulomb scattering analysis – versus the more direct analysis of the attenuation characteristic of the radiation – is used in x-ray and gamma ray imaging analysis¹⁹. The momentum of the muons are generally determined by measuring the multiple scattering through several layers of known materials of known thickness interspersed with additional detectors that measure the 2D position of the particle as it passes through.

To determine the path angle of the muon, two 2D (x and y) detectors are needed before the muon interacts with the target object, and two similar detectors are required after the interaction to determine the scattering angle (x´ and y´). These pairs of detectors must be spaced sufficiently apart that the difference in (x, y) position registered in each detector, along with the precision of that measurement by the detector, provides an angular measurement of sufficient accuracy for the application²². Ideally, there would be no possibility of scattering between the individual detectors in the pairs (i.e. there would be a vacuum between them); however, the scattering produced by the air between the detectors is generally negligible.



Figure 2 - Determination of Muon Path and Scattering Angle
While a basic muon tomography system for security applications requires at least a set of detectors above the target object and another set below, with the momentum determination layers below the object to minimize the absorption of muons, it is desirable to surround the object on all four sides with detectors to maximize the muon flux that can be used. While incoming muons are concentrated towards the vertical direction (in a roughly cos²θ distribution), the intensity only drops to half maximum at angles of 45 degrees²⁴. Having vertical detectors in addition to horizontal detectors allows muons with high incident angles from the vertical to also be detected and incorporated into the tomographic calculations. The primary limitation is that if the momentum determination detectors are only horizontal under the overall detector, then the momentum of these high angle (from the vertical) muons cannot be determined as precisely. As such, their contribution to the model developed of the internals of the object is reduced, but also non-zero. Vertical detectors will also be able to catch muons that pass through the top detector, but either at too large an angle or too close to the edge to align with the bottom detector after passing through the object.

While there are many types of detectors appropriate for the detection of muons and their 2D spacial coordinates as they pass through, two primary detector types are being considered for this application that can determine 2D spacial position information about passing muons: drift chambers and scintillators. The details of each of these technologies are given in detail further below; however, a short summary is provided here for consideration at the architectural level.

Drift chambers use the ionization trail produced by a muon as it passes through a gas as the means of determining position. Specifically, an electric field inside the drift chamber accelerates the electrons and ions “knocked loose” by the passage of a high energy particle, like a muon, towards a central wire. By measuring the amount of time it takes for the electrons to travel to the central wire (“drift”) and the relative position of the electrons along the wire when they reach it, it is possible to determine the 2D position of the particle through the detector. Often, a simple scintillator is used to start the drift timing, although there are other methods to trigger the count. One of the primary advantages of drift chambers is the simplicity and cost effectiveness of the readout electronics, since there are only a small number of data channels required to extract all the needed information. The primary negatives of drift chambers are the high voltage power supply to create the necessary internal electric field, and the need for it to be filled with specialized gasses for proper operation²².

Scintillators of this size, and for this purpose, are usually comprised of plastic scintillating material extruded in a triangular shape a few centimetres on a side. Each of these scintillating strips has a special-purpose optical fibre running through its centre that shifts the frequency of the photons that impinge on it before it guides it to a photomultiplier or avalanche photodiode. Furthermore, each of these strips is optically isolated (coated in some opaque material) and then the triangles of scintillator are staggered in opposite vertical orientations from one another to form a horizontal sheet. By measuring the relative intensity of the photons given off between two adjacent scintillator strips, the relative distance travelled through each scintillator can be determined, and that ratio provides a position in one dimension. To determine the 2D spacial coordinate for the muon, a second scintillator is positioned orthogonally to the first, and in close proximity to determine the position in the other direction. The primary advantage of scintillators of this type are the cost effectiveness of the scintillating material and the fact that they are solid and easy to maintain (although they can degrade over time in high luminosity environments). The main disadvantage is the huge number of readout channels required (e.g. one channel every couple of centimetres over several metres in each of two directions times at least four for a tomographic system plus several more 2D pairs for momentum determination). While it is possible to bring several channels (from sufficiently spatially separated detectors in the array) to a single photomultiplier, that introduces some uncertainty in the measurement and, although it reduces the number of readout channels, there are still a large number required²².

Which detector technology is predominantly used for practical field-deployable muon tomography systems will have to be decided through careful cost and performance analysis and testing; however, both primary candidates ultimately provide similar data (x, y position data) that will feed into the same sort of back-end analysis software. All things considered, the elimination of the requirement for a flowing specialty gas in most drift chambers would result in a huge advantage for that technology; while the reduction of the number of readout channels required for large area scintillators would provide it with a huge advantage. Both of these areas should be subject of research projects for application in 2nd generation muon tomography systems.

D. Outline of Thesis

Because of the sensitivity of Passive Muon Tomography (PMT) systems to high-Z materials (versus lighter elements) they are a much more targeted solution than more indiscriminate imaging systems, and the lack of an active radiation source eliminates the potential health concerns associated with x-ray and gamma ray imaging systems. While PMT systems only address a particular class of risk, specifically the threat posed by the trafficking of special nuclear materials that could form the basis for a bomb or large well-shielded shipments of radionuclides that could be used in a “dispersal” device, the asymmetric nature of the threat justifies the commercialization of this technology to compensate for the serious limitations of existing technologies in this area of detection. Carleton University’s proposal to use large-area drift chambers for muon detection will result in a device that will provide excellent spacial and temporal resolution with very cost effective readout electronics and data processing requirements; however, the initial requirement for a flowing gas in the first generation solution presents a negative offset through higher infrastructure and ongoing maintenance costs that would need to be mitigated as part of a widespread deployment of this particular solution.

II. COSMIC RAYS

A. Overview

Primary cosmic rays are very high energy charged particles (into the range of many TeV²⁴) that originate mostly outside of the solar system, from astrophysical sources, and are comprised primarily of protons (~80%) and helium nuclei (~14%), with the remaining being heavier nuclei such as carbon, oxygen, and iron. These can also interact with interstellar gasses to create a much lower flux of secondary cosmic rays comprised mostly of anti-protons and lithium, beryllium, and boron nuclei²³. When cosmic rays interact with the Earth’s atmosphere at high altitudes, they produce showers of thousands of “secondary” particles, usually also called “secondary cosmic rays”. Most of the particles so generated decay or interact with atmospheric atoms before they can reach the surface of the Earth; however, a shower of gamma rays, electrons, neutrons, and muons²⁴ (due to relativistic time dilation) do reach the lower altitudes of the atmosphere and the surface itself. Of these, the cosmic ray muons are of primary interest in this application due to their high energy, penetrating power, and relative ease of precise determination of path and momentum.

B. Spectrum and properties

The energy spectrum of astrophysical cosmic rays obviously determines the energy spectrum of the resultant radiation that reaches the lower reaches of the atmosphere and the ground. Lower energy cosmic rays entering the solar system encounter the solar wind and are prevented from reaching the inner solar system and the Earth, so the cosmic rays that arrive tend to be of the more energetic type (a distribution between 10GeV to 100TeV). The intensity of primary nucleons over that energy spectrum is roughly given by:

(1)

where E is the energy per nucleon, and α (≡ γ + 1) = 2.7 is the differential spectral index of the flux and γ is the integral spectral index²³. The actual measured data is shown in Figure 3 below.



Figure 3 - Major components of the primary cosmic radiation²³
The primary and secondary cosmic ray particles cause a spray of other particles to be generated when they interact with nuclei in the upper atmosphere. The very short life particles (~26ns)²⁴ such as pions, despite their relativistic speeds, almost all decay before they reach the ground into gamma rays, electrons and positrons, and muons. The protons from the primary cosmic rays can also reach the ground, but at a much lower momentum. All of the particles and photons interact with the components of the atmospheric and lose momentum through electromagnetic interaction. If enough momentum is lost, they can be absorbed and will not make it to the ground. Muons, in particular, due to their longer lifetime (~2.2μs)²⁴ and higher momentum are able to pass through the atmosphere before they decay in sufficient numbers and with sufficient momentum for their use as tomographic probes at ground level.



Figure 4 - Particles and photons generated by cosmic rays¹⁸
From the initial cosmic ray interactions, three classes of radiation reach the ground: nucleons, the electromagnetic component, and muons. The nucleons are either remnants of the primary cosmic rays or are generated by atmospheric interactions. If measured in the vertical direction at sea level, about 1/3 are neutrons and 2/3 are protons. Nucleons over 1GeV/c have an integral intensity of roughly 1.35 (m² s sr)^-1 in the vertical direction. The electromagnetic component is comprised primarily of electrons, positrons, and high energy photons from the decay of mesons; however, it also has a component of low energy electrons from muon decay and of higher energy electrons knocked out of atmospheric constituents. The integral intensity in the vertical direction for electrons and positrons is roughly 30, 6, and 0.2 (m² s sr)^-1 above 10, 100, and 1000MeV respectively, where the photon to electron plus positron ratio is about 1.3 at 1GeV. Muons are generated in the upper atmosphere through the decay of mesons produced by the primary or secondary cosmic rays and lose about 2GeV through ionization before they reach the ground or decay themselves if they didn’t have a high enough energy to start. The integral intensity of muons in the vertical direction above 1GeV/c at sea level is about 70 (m² s sr)^-1 (≈1 (cm² min)^-1). The mean energy of muons that reach the ground is about 4GeV and have an angular distribution of roughly α cos²θ (approaching secθ distribution at higher energies and θ > 70º from the vertical). Figure 5 shows the distribution at the vertical and 75º from vertical²³.



Figure 5 - Muon spectrum at ground level (at θ = 0º and θ = 75º [◊])
The tomographic process is dependent on events occurring from multiple angles through each point so an analysis can be made of the intersecting paths of multiple events. As discussed above, there is a wide angular distribution of ground-level muons which is sufficient for tomography purposes. At 45º, the flux is roughly 0.5 what it is at the vertical, and even at 90º to the vertical, there is a non-zero muon flux:



Figure 6 - Horizontal Cosmic Ray Muon Energy Distribution²⁴
While the distribution all the way down to the horizontal is not going to be helpful in muon tomography systems using horizontal detectors that will form the basis of systems for security purposes, they can be used with a set of vertical detectors for content determination and auditing of fixed containers at nuclear storage facilities and in non-proliferation treaty verification and enforcement applications (see section “Use in sealed-container inventory determination and management” below).

The final important characteristic of cosmic rays at ground level is the phenomenon known as “air showers”, which are cascades of particles generated by particularly energetic cosmic rays that arrive at the ground in a tightly correlated “shower”. Air showers have a core of hadrons which acts as a collimated source for electromagnetic subshowers of photons, electrons, and positrons which form the most numerous components of the shower, with muons occurring at about one tenth the quantity of the electromagnetic components²³. Air showers can result in numerous near-simultaneous events being triggered in a detector and cause problems in determining the coordinates that they all passed through. In general, the data from the multiple triggerings in detectors by air showers have to be discarded because of the location of each of the triggers may not be discernible.

C. Multiple scattering and tomographic analysis

For x-ray and gamma ray based systems, the absorption signal is generally used for tomographic purposes; however, the Coulomb scattering signal can also be used to extract tomographic information. With muons, this provides the primary method being considered for tomographic systems (although some groups are proposing energy-loss systems for specific applications²⁴). Coulomb scattering of muons occurs with the electromagnetic interaction of charged muons with the atoms it goes past. Each interaction causes the muon to be deflected, or scattered, by a certain angle dependent inversely on the momentum of the muon and directly to the square root of the atomic number, Z, of the atoms it interacts with (through its inverse dependency on the radiation length, X, of the material)²². The radiation length for a given material is given by the approximation (good for all higher Z materials)²⁵:

(2)

where A is the mass number. For any given material, this multiple scattering produces a distinctive Gaussian distribution of scattering angles, θ¹⁹:

(3) and (4)

where p is momentum and β is v/c of the muon, and L is the path length through the object (logarithmic terms have been omitted, these contribute overall less than 10%¹⁹).
For any given material, if the scattering angle is measured, and its momentum can be determined, the path length through an object can be determined to a precision of where N is the number of transmitted muons. For instance, in 10cm of material, a 3GeV muon will scatter with a mean angle of 2.3 mrad in water (X = 36cm), 11 mrad in iron (X = 1.76 cm), and 20 mrad in tungsten (X = 0.56 cm)¹⁹. The problem is that in the real world, the material of interest will be non-homogenous, and from the perspective of each muon event, will contain multiple layers of different material:



Figure 7 - Multiple-path scattering through non-homogeneous materials²⁶
To address this complex analysis problem, the volume contained within the detectors is broken down into rectangular “voxels”, the 3 dimensional equivalent of a pixel. By performing iterative sets of “maximum likelihood” calculations based on estimates of which muon events went through which voxels, it is possible to assign each voxel with an estimated radiation length, and thus determine the Z of the material in that voxel.

As the number of muons through each voxel increases, and the size of each voxel can be decreased in the analysis (reducing the chances of non-homogeneity of material within a voxel), the Z of the material in the voxel can be estimated with increasing accuracy using relatively simple algorithms such as this Maximum Likelihood/Expectation Maximization (ML/EM) method based on algorithms used in medical emission tomography systems²⁶:



Figure 8 - Example tomographic analysis algorithm
The important thing to note is to recall that muons passing through a material are much more sensitive to deflection by high Z materials than by the lighter materials that comprise most cargoes and the containers that they are carried in. For instance, the mean square scattering angle at a nominal momentum through a unit depth of material in mrad² per centimetre is 3 for aluminum, 14 for iron, and 78 for uranium. By noting that large scattering angles likely indicate high-Z materials, it is possible to set the “stopping criteria” for the tomographic analysis algorithms such that a complete tomographic analysis of every voxel is performed down to a certain level of confidence – only down to a level of confidence that no voxel contains special nuclear materials or sufficient shielding to mask the radiation signature of a radiological substance of potential risk. Performing the analysis over a long period of time, with large numbers of muons, would allow a complete and accurate tomographic image to be built up of the contents of any arbitrary container; however, focusing on the detection high-Z materials allows a scan to be done in as little as 1 minute for voxels 10cm on a side.

To measure the momentum of each muon detected, multiple layers of detectors are placed with material of known composition and thickness between then (e.g. steel). By measuring the scattering angle through several layers of materials, an accurate value for the momentum can be determined. It should also be noted that in all of the above, it was assumed that the muons measured at the top set of detectors would penetrate through to the bottom of the momentum detector; however, muons that lose sufficient energy to decay – either because they had low momenta to start with or because they had to pass through a large amount of high-Z material – need to be taken into account in any practical algorithm. If nothing else, those events need to be discarded unless there are too many of them, in which case that could signal a large amount of high-Z material in the container being examined.

III. Detectors

A. Overview

For a muon detector to be usable in tomographic applications, it must be able to determine the position of the passage of an energetic muon in two dimensions while providing a minimum of interaction with the muon itself (in the sense of random scattering or absorption). For such a detector to be usable in a tomography system deployed to scan cargo containers and vehicles, such as transport trucks or trains, it must further be able to cover a very large area with relative economy and be manageable by the trained security staff where it is deployed. The two main competitors for large area muon detectors are drift chambers and scintillators. Each has its own set of advantages and disadvantages which need to be evaluated against the requirements of the application and the cost of widespread deployment.

Drift chambers function by measuring the time it takes for the ionization trail left by an energetic muon in a gas to be accelerated towards a high voltage anode wire suspended across and in the centre of the chamber (the “drift” time). As the electrons that were knocked loose by the muon are accelerated towards the anode wire, they knock loose new electrons from the intervening gas molecules and cause a large cascade of electrons towards the anode that is easily detected by relatively simple readout electronics. Usually a simple scintillation counter, above or below the drift chamber, is used to start the timer used to measure the length of the drift time. Knowing the speed with which the electrons propagate through the gas in the electric field of the chamber allows the distance of the event from the wire to be calculated, thus providing a measurement in one of the dimensions. The second dimensional measurement is achieved by determining the position along the anode wire of the arriving, accelerated, electrons associated with a cascade. This can be done in two ways: by using charge division along the anode wire, and/or by measuring the electric fields induced by the moving charge on a set of split cathode pads that run the length of the anode wire on the inner surface of the chamber²².

Large area scintillators are usually comprised of extruded lengths of plastic that will emit light when impinged upon by a high energy muon. These tiny flashes of light are guided to a photomultiplier or avalanche photodiode where they can then be detected electronically. To determine the one-dimensional location of the flash, a triangular cross section is used for the extrusions such that the point of one triangle is against the base of the next, and so on, separated by optically opaque material to form a flat sheet. When a muon passes through the sheet, it will generally pass through two of the extrusions. By comparing the intensity of the scintillation in the first extrusion to the intensity in the second, the relative distance travelled in each can be determined, and therefore the location of the event can be measured. To achieve a two-dimensional location, another scintillator assembly needs to be placed such that the lengths of the extrusions are orthogonal to those in the first. Because of the rapid attenuation of the scintillation light, the dimensions of the triangular cross section are limited. To that end, each extrusion that makes up the scintillator must have its own readout infrastructure, normally provided by special purpose optical fibres and some device capable of converting the small number of photons generated into a useable electronic signal²².

Both the drift chamber and scintillator in a two-layer orthogonal configuration provide roughly the same performance in terms of ability to determine the location of a muon event within its area of coverage, so the decision on which of the two technologies has to be taken based on other factors. The primary advantages of the drift chamber technology is the fact that one detector provides two dimensional information and that it has extremely simple and inexpensive readout electronics, with the requirement for only a few data channels that need to be analyzed. It’s chief disadvantages are the need to have it filled with a specific and pure gas and its need for a high voltage power supply. The primary advantages of the scintillator are the low cost of the materials and ease of construction of the sensor itself, and the fact that it’s a solid state device (although liquid scintillators do exist, they are not being considered for this application). The chief disadvantages are the huge number of data channels required to instrument a large area sensor, multiplied by two for the two planes of the overall detector (for 2D measurements)²², and the degradation that occurs over time in plastic scintillating material due to ionization²⁷. In addition to the implementation details, other factors such as the environmental suitability of the technology (in the sense of things such as temperature extremes, humidity, salt spray, vibration, etc.), and system-level issues such as fault coverage and error detection, need to be carefully evaluated for each solution before a choice can properly be made.

While both of these solutions need to be explored for performance, and ultimately for testing in real-world environments, Carleton University has been selected to build and test a drift chamber based solution while another institution was chosen to develop and test a scintillator based solution. As such, the remainder of this report will focus exclusively on drift chambers and the specific solution being developed at Carleton University.

B. Drift Chambers

1. Basic Operation

Early attempts for the precise detection of muons involved Multi-Wire Proportional Chambers (MWPCs)²¹. These devices were in turn built on the earlier work on single-wire cylindrical proportional gas ionization detectors that had a similar construction to the more well-known Geiger-Müller counters.



Figure 9 - Basic construction of a single-wire cylindrical gas ionization detector¹¹
Gas ionization detectors operate by accelerating electrons liberated by the ionizing passage of particles through a gaseous medium towards the anode wire. As these electrons accelerate through the gas, they develop sufficient energy to cause further ionization in the gas, and those electrons are also accelerated towards the anode wire, causing yet further ionization in an avalanche or cascade of electrons that is large enough that their motion induces an easily measurable electric potential on the anode wire¹¹.

Two key operating modes exist for these sorts of detectors: as proportional counters, and as Geiger- Müller counters. The primary difference in architecture is simply the operating voltage between the cathode and anode. When an event occurs in a device operated in the proportional counter region, the magnitude of the electron cascade is proportional to the number of electrons liberated by the primary ionization event, in what is essentially a linear multiplication of current. In the Geiger- Müller region, the electric field is sufficient that a chain reaction of avalanches occurs along the length of the anode wire that results in an electrical breakdown and a very strong signal is generated that is relatively independent of the magnitude of the initial event¹¹. While Geiger- Müller counters are ubiquitous in field detection of radiation, they have limited use in the laboratory or in any precision work due to their indiscriminate sensitivity, and so proportional chambers tend to be used in modern devices.



Figure 10 - The various operating voltage regions of a gas ionization detector¹¹
Gasses such as argon are often used in gas ionization devices because it is low cost and has a relatively high specific ionization potential; however, argon on its own is not enough for a practical device. The problem is that as atoms that become excited by the avalanche emit high energy photons that can ionize the cathode and cause secondary avalanches. By adding a second, polyatomic, gas that can absorb these photons and then dissipate the energy through dissociation or elastic collisions to act as a quencher, this issue can be controlled¹¹.

Multi-Wire Proportional Chambers can be through of as a large number of closely spaced single wire gas ionization detectors with their cathodes unrolled to form flat sheets above and below the anode wires. When an event occurs, it will induce a current on the anode wires, but the positive ions will also induce a current in the cathode strips used to generate the electric field. In advanced MWPCs, orthogonal cathodes can be used to get a two dimensional position of events in a single MWPC by using a “centre of gravity” localization of the cathode signals when a trigger pulse is received on the anodes.



Figure 11 - Advanced MWPC with orthogonal cathode signals used to localize event²⁸
While MWPCs revolutionized the particle detection and spacial localization, they suffered from the problem of having requiring complex readout electronics and many data channels to process the information they generated. In studying the temporal performance of MWPCs, it was realized quite early that the ionization from an event takes time to “drift towards the wire”, and that in turn led to the development of the drift chamber which has come to dominate the field²¹. The primary advantages of drift chambers are the huge reduction in the complexity of the required readout electronics and the much simpler and more forgiving construction requirements. The principle of a drift chamber is similar to all other gas ionization detectors; however, it operates by creating a shaped electric field that accelerates the electron cascade towards a single wire in the centre of the chamber where it generates a measurable current.



Figure 12 - Representative drift chamber architecture²⁸
Unlike MWPCs, the drift chamber requires that some mechanism is used to generate a timing signal coincident with the initial event in order that the time between the event and the arrival of the cascade at the anode can be measured (to determine the distance travelled from the initial ionization event):

(5)

where v(E(x)) is the drift velocity for a particular gas mixture, pressure, and electric field²¹. For example, for typical argon and isobutane mixtures, the drift velocity, v(E(x)), is given by:



Figure 13 - Drift velocities for argon/isobutane mixture at various electric potentials²¹
Like all gas ionization based detectors, drift chambers require some recovery time between events to properly record them. Firstly, the electrons must be re-absorbed into the gas to dissipate the charge on the anode wire. Secondly, if closely coincident events occur, the cascades could arrive at the same or nearly the same time, thus causing possible confusion as to the distance of the events from the anode wire – for instance, if the first event has a long drift time, but a second event will have a short drift time, the second event can arrive at the anode wire before the first, and there may be no way to associate the correct trigger with the arriving signal. This is especially an issue in very large area detectors such as those proposed for use in this security scanning application. In those instances, either additional detection is required, or that data must be discarded. Because of this limitation, a cosmic ray air shower over a large area muon detector will generally not generate any useful data because there will be multiple near-simultaneous events in the detector.

2. Specific Topology

The design being developed by Carleton University is for an eventual 4’ wide by 8’ long single-wire drift chamber based on the basic JADE flowing gas muon detector design²⁹ with enhancements pioneered in the Opal detector design²¹ to allow it to detect the precise position of the cascade along the length of the anode wire using both charge division and split cathode pads above and below the wire. Two of these chambers will be positioned side-by-side to provide the required 8’ by 8’ coverage area. Whereas the JADE detector’s timing was triggered by a master clock signal from the PETRA storage ring before an event was to occur, this design needs to measure naturally occurring random radiation, therefore the drift timing will be triggered by a scintillation counter placed above the entire system.



Figure 14 - JADE muon detector mechanical and high voltage details²⁹
The basic design will consist of a gas-tight chamber that will contain a flowing mix of argon (95%), isobutane (2%), and tetrafluoromethane (3%) introduced and exhausted through fittings on either end of the chamber. The top and bottom surfaces will consist of copper clad G10 fibreglass mounted to a frame that will separate them by 20mm. The external surface of the panels will be solid copper that will act as a grounded shield, and the internal surface will be patterned with strips running the width of the chamber (9mm of copper separated by 1mm of G10) to generate a uniform electric field, and then a special cathode pattern on the portion of the panels above and below the anode wire²². A 10kV power supply will provide the electric field between the cathode and anode, and the readout electronics will be isolated from the high voltages through the use of transformers, rather than direct coupling or coupling through a capacitor as was done with the JADE detector.



Figure 15 - Opal detector cathode pad patterns
The target performance for the Carleton University design is 1mm resolution in both directions. Current designs call for data channels with 100ns resolution (10MSamples/s). Given the drift velocity of the gas chosen for the detector is roughly 2-3mm/100ns, the analysis software will need to interpolate where the peak of the waveform is to get the desired accuracy. Given the well defined shape of the signal waveform, this should be possible; however, if it presents a problem, higher sampling rates can, of course, be specified and tested. In the perpendicular direction (along the direction of the anode wire), both charge division of the signal on the resistive wire and cathode pad pickups will be used to get the needed resolution.



Figure 16 - Split cathode pad design for spacial resolution (not to scale)²²
The proposed cathode pad design for the Carleton design will operate much like the one designed for the Opal detectors. As the electrons cascade travels to the anode wire, it will induce charge on the cathode pads as it passes. By measuring the difference in signal between the two pads, it will be possible to determine the position at which the cascade went past. The anode wire itself will operate like a resistor divider network with the voltage across the wire in each direction (there will be an amplifier at each end) a product of the current induced by the cascade times the resistivity of the wire times the length of the wire between the source of the charge induction and the end of the wire. Again, by comparing the difference between the voltage present at each end of the wire, the position of the cascade’s coupling to the wire can be determined²².

As shown in Figure 2, two detectors need to be placed above the target and two below to measure the incidence angles before and after interaction with the target. If the detectors achieve the desired resolution of 1mm in both directions, and the detector pairs are placed 1m apart from each other, then the system will have an angular resolution of 3 milliradians²². To complete the overall sensor, the scintillator will be placed on top, and at least three drift chambers separated by at least two layers of steel of known composition and thickness will be used to determine the momentum of the muons that were measured by the upper and lower pairs of detectors. Ideally, full systems will also contain pairs of detectors in the vertical direction as well to measure muons passing through the edges of the target or at high incident angles. Deployed systems will also need to contain various safety measures to protect personnel from the high voltages and the gasses employed in case of a leak (detectors and venting systems).

3. Readout Electronics and Data Processing

Each of the drift chambers has four data channels that will need to be analyzed to produce a spacial coordinate for a muon event. Further, at least one channel is required for the scintillator to be used as a data acquisition trigger. When the positions from all of the chambers have been determined, the incident and scattering angles are computed from those positions, along with the particle’s momentum. The information from the event is then sent to the tomographic analysis software which correlates all of the events to form a three dimensional model of the likely Z content of each computed voxel within the target. Further software analyzes this information to decide whether an alarm should be triggered to alert security personnel to a possible issue.

The analysis of the raw data from the drift chambers will have to be done with a system in proximity to the drift chambers due to the analogue nature of the signals at that point and the sheer volume of data generated at a full system installation. The position data from the detectors does not require much bandwidth to send for processing, so the computers that do the analysis of that data will likely be located in a convenient building nearby and with easy access by the personnel at the site.

C. Scintillation counters

When radiation passes through a scintillating material, such as a doped plastic, it causes excitation of the atoms in the material that is quickly released as high energy photons. Usually, a fluorescent material is included in the scintillating material or in special purpose optical fibres embedded in the material to absorb the photons emitted by the bulk material and re-radiate the energy as longer wavelength photons that are more easily amplified by a photomultiplier tube²⁷. As with any large area scintillator, an inherent problem is that the fluorescent material needed to re-emit photons is sensitive to the absorption of those very same photons, so the mean path length for photons is fairly short. To that end, optics that can guide the photons out of the material, such as special purpose optical fibres, are used at frequent intervals (every few centimetres in most materials). Unlike using large area scintillators for determining the spacial location of an event, using it as a counter greatly reduces the complexity of the readout optics and electronics because only a single channel of data is required, and all of the segments of the scintillator can be fed into a single photomultiplier to keep the cost at a minimum.

For the 8’ by 8’ scintillator needed for the proposed application, that is an area roughly 6000cm². For an average muon flux at sea level of roughly 1 (cm² min)^{-1 23}, that works out to about 1.7×10^-2 (cm² s)^-1. So for an area of that magnitude, the number of events per second is: 1.7×10^-2 (cm² s)^-1 × 6000cm² = 100 s^-1, or roughly one event every 10ms on the average. Since photomultipliers tend to have sub-μs response times, this should pose no issue for the technology being considered.

IV. Implementation

A. Description of prototype project

The three chambers that will be built for the proof-of-concept and to determine the performance of the manufacturing techniques and materials, the chosen gas mixtures, and the readout electronics are going to be 4 feet wide by 40cm long. A scintillator connected to a photomultiplier will be used as the timing trigger. In addition, an anode wire test jig was constructed to provide early integration testing of various electronics and to validate the charge division measurements by injecting charge at various points along the wire. Since the prototypes are also going to be 4’ wide, the anode wire tension and construction and assembly techniques will be validated. Additionally, the split cathode position sensor design (or multiple designs if necessary) will be validated as it will have the same dimensions as in the final design²².

B. Readout Electronics

Off-the-shelf amplifiers will be used on the initial prototypes to reduce the engineering work required to implement them. When the operation of the chambers has been characterized, it will be possible to design more specific amplifiers for the chambers and compare their performance with the integrated amplifier subsystems that were purchased for the laboratory. Each drift chamber, as described above, will have four data channels: two for the anode wire, one for each end, and two for the cathode pads, one for the top and one for the bottom. In addition to the readouts from the three chambers being built, each will need a channel from the scintillator that will be used to trigger the start of data acquisition.

Since the rise time of the anode wire waveform is expected to be roughly 100ns and to decay in roughly 10μs, a 10MS/s/channel data acquisition system was selected to provide 100ns sampling resolution on all data channels. Specifically, a National Instruments PXI system has been purchased to provide 16 data acquisition channels (4 × PXI-6115 cards) and an integrated Windows XP based controller (PXI-8110) in an 8-slot chassis (PXI-1042). The LabView software on the integrated controller will be configured to collect data from each of the chambers when an event is detected by the scintillator associated with the chamber. That data will be buffered for later processing on a separate system to perform a complete analysis using custom software and additional off-the-shelf software tools.

V. Further exploration

In addition to the use of the proposed muon tomography systems in border security and container/vehicle inspection, the basic technology can be useful in other applications as well. Furthermore, with appropriate research and development, enhancements to the basic technology are possible that will reduce the total cost of ownership and operation.

A. Use as a scientific instrument

With the possibility of large area muon detectors being deployed along borders and in key strategic locations, it should be noted that each one of these devices can be used as an element in a larger cosmic ray observatory. The information on incidence angle and momentum of incoming cosmic ray muons could provide a wealth of data to astrophysicists and particle physicists alike (who can analyze the data against various models developed for subatomic phenomena to support or discard various hypotheses). One major issue is that data on the contents of scanned targets cannot be shared with the general public due to security concerns. This can be addressed by sending data only when a scan is not in progress. Alternatively, if the initial momentum (before interaction with cargo) is reconstructed by projecting the final momentum backwards through the gathered tomographic data when cargo is present, there will be no way to determine anything about the contents of the scanned cargo from the data. In any case, the angular information from the top pair of detectors is gathered before any interaction with cargo and should not present any security risk as it is a purely astronomical data source at that point.

B. Developing a sealed chamber (no gas flow)

The major disadvantage of the drift chamber solution proposed by Carleton University is the need for a flowing gas mixture. If it were possible to seal the chamber and operate it for long periods without needing service, then it would be both cost effective from a readout electronics perspective and from the longer term operational cost and complexity perspective through the elimination of the need to manage gas supplies and disposal. Much work has been done over the years on sealed gas ionization based detectors, and research and development in this area could have a large impact on the cost of muon tomography systems in the field.

C. Use of active muon source system

One of the issues with using cosmic ray muons as a source of radiation for tomographic purposes is their relatively low flux (1 muon (cm² min)^-1). This low flux means that it takes at roughly a minute for a basic scan to determine whether there is any high-Z material of concern. By using an artificial source for a higher muon flux, it could be possible to do the scans faster or to build a more complete tomographic image of the contents of a shipping container or other target of interest. The issue is, of course, that this introduces a vary dangerous ionizing radiation source to the situation and the lack of any additional radiation is one of the attractive elements to using cosmic rays muons as the probe.

D. Use in sealed-container inventory determination and management

There are many installations, for instance Chalk River in Ontario, where there are sealed containers with unknown quantities of potentially dangerous materials in them. There are also situations where contents of containers are claimed to contain certain materials, but need to be verified as part of nuclear control treaties. In those cases, cosmic ray muon tomography could provide an excellent tool for cataloguing and monitoring the contents of these containers. Since this is more of an audit application, the lower flux and time to acquire the necessary level of data are not as much of an issue as for applications that impinge on commerce.

VI. Conclusion

Passive cosmic ray muon tomography systems present an excellent solution to the issue of deterring and detecting the trafficking in nuclear and radiological materials – in the first case through direct detection of high-Z materials, and in the second case, being able to detect high-Z shielding that might be hiding lower-Z radiological materials. The system further distinguishes itself by not introducing any new sources of radiation, thus sidestepping any potential health or safety concerns from the public or business. Carleton University’s proposed drift chamber muon detectors build upon decades of experience in implementing high resolution muon analysis systems, and can be used to determine to a high degree of accuracy both angular and momentum data on the muons passing through a detector system for analysis by the tomographic software. The low cost of readout electronics compensates for the higher cost due to the requirement for gas-filled chambers, and will result in a competitive solution for field-deployable systems.

VII. References

¹ DRDC, Technology Investment Strategy 2002 for the Evolving Global Security Environment. (Defence Research and Development Canada, 2002).

² D.C. Maurer, Nuclear Detection: Preliminary Observations on the Domestic Nuclear Detection Office's Efforts to Develop a Global Nuclear Detection Architecture (United States Government Accountability Office, 2008).

³ D.A. Robitaille and R. Purver, Commentary No. 57: Smuggling Special Nuclear Materials (http://www.csis-scrs.gc.ca/pblctns/cmmntr/cm57-eng.asp, 1995).

⁴ Nuclear Trafficking: A Radioactive Subject, The Economist 382, 60-61 (2007).

⁵ Transnational Threats Update January 2008 (Center for Strategic and International Studies, 2008).

⁶ Uranium, The Columbia Encyclopedia, Sixth Edition (http://www.encyclopedia.com/topic/uranium.aspx#1E1-uranium, 2008).

⁷ Container Security Initiative: 2006-2011 Strategic Plan (U.S. Customs and Border Protection, 2006).

⁸ S. Rein, Airport Security: Bin Laden's Victory, (Forbes.com, http://www.forbes.com/2010/03/03/airport-security-osama-leadership-managing-rein.html)

⁹ G. Aloise, Combating Nuclear Smuggling: Efforts to Deploy Radiation Detection Equipment in the United States and in Other Countries (United States Government Accountability Office, 2005).

¹⁰ F. Flakus, Detecting and measuring ionizing radiation - a short history, IAEA Bulletin 23, 31-36 (1981).

¹¹ W. Leo, Techniques for Nuclear and Particle Physics Experiments : A How-to Approach (Springer-Verlag, Berlin;New York, 1992).

¹² J. Ely and R.T. Kouzes, Spies, Lies, and Nuclear Threats (Pacific Northwest National Laboratory, http://hps.org/hsc/documents/Threats.pdf, 2005).

¹³ L. Zaitseva and K. Hand, Nuclear Smuggling Chains: Suppliers, Intermediaries, and End-Users, American Behavioral Scientist 46, 822-844 (2003).

¹⁴ G.A. Warren, L.E. Smith, C.E. Aalseth, E. Ellis, T.W. Hossbach, and A.B. Valsan, Determining activities of radionuclides from coincidence signatures, Nuclear Instruments and Methods in Physics Research Section A 560, 360-365 (2006).

¹⁵ N.A. Stanton, D.J. Harrison, K.L. Taylor-Burge, and L.J. Porter, Cognition, Sorting the Wheat from the Chaff: A Study of the Detection of Alarms, Technology & Work 2, 134-141 (2000).

¹⁶ Portal VACIS, FEMA Responder Knowledge Base Version 3.5.4 (https://www.rkb.us/contentdetail.cfm?content_id=97095, 2010).

¹⁷ S. Webb, From the Watching of Shadows : The Origins of Radiological Tomography (A. Hilger, Bristol;New York, 1990).

¹⁸ B. Fishbine, Muon Radiography, Detecting Nuclear Contraband, Los Alamos Research Quarterly 12-16 (Spring 2003).

¹⁹ C.L. Morris, Tomographic imaging with cosmic ray muons, Science Global Security 16, 1, 37 (2008).

²⁰ Portal VACIS gamma-ray imaging system, (SAIC Security and Transportation Technology, http://www.saic.com/products/security/portal-vacis/portalvacisbrochure.pdf, 2009).

²¹ U. Becker, Large Area and Muon Detectors in Instrumentation in High Energy Physics (World Scientific, Singapore, 1992), pp. 513-585.

²² J. Armitage, Personal communications, (2010).

²³ T. Gaisser and T. Stanev, Particle Data Group, Astrophysics and Cosmology, Physics Letters B 592, 228-234 (2002).

²⁴ P. Jenneson, W. Gilboy, S. Simons, S. Stanley, and D. Rhodes, Imaging large vessels using cosmic-ray muon energy-loss techniques, Chemical Engineering Journal 130, 75-78 (2007).

²⁵ K. Desler and D. Edwards, Particle Data Group, Accelerator Physics of Colliders, Physics Letters B 592, 235-274 (2003).

²⁶ L.J. Schultz, G.S. Blanpied, et al., Statistical Reconstruction for Cosmic Ray Muon Tomography, IEEE Transactions On Image Processing 16, 1985 -1993 (2007).

²⁷ C. Zorn, Plastic and Liquid Organic Scintillators in Instrumentation in High Energy Physics (World Scientific, Singapore, 1992), pp. 218-280.

²⁸ G. Charpak and F. Sauli, Multiwire proportional chambers and drift chambers, Nuclear Instruments and Methods 162, 405 - 428 (1979).

²⁹ J. Allison, J.C.M. Armitage, et al., The JADE muon detector, Nuclear Inst. and Methods in Physics Research A 238, August (1985).