This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
Neutron detection is an essential part of the global struggle to avert the proliferation of special nuclear material. Applications dependent upon neutron detection technology vary from traditional nuclear non-proliferation purposes, such as safeguarding material and authenticating stockpile declines, to the authorative prohibition of SNM - an objective that has lately increased in importance to a level on equivalence with traditional missions. Large transnational programs focussing on interdiction and safeguards have deployed radiation detection assets across the world. In conjunction with these deployments of commercially available technology, significant research and development has been engaged toward the formation of next generation assets. Neutron detection technology plays a important role because of the proficiency of neutrons to penetrate materials that easily absorb gamma rays and the exclusive fission signatures neutrons possess. One particularly serious technology development task results from declining supplies of 3He, somewhat caused by extensive deployment of high efficiency systems for portal monitoring. Other evolving assignments, such as the wish to detect SNM at greater standoff distances, have also encouraged neutron detection technology development. In respect of these requirements, this thesis reviews the signatures of neutrons emitted by SNM, the principles of neutron detection, various strategies under investigation for detection in the context of nuclear non-proliferation and the experimental findings of the efficiency of a novel mixed-field organic liquid scintillator based detector.
The many different techniques for radiation detection have played a vital role in the international framework of deterring and detecting illicit uses of special nuclear material (SNM). The need for improvisation in radiation-detection technologies supporting interdiction, search and characterisation objectives is being placed by a growing emphasis on global nuclear threat reduction along with the expansion of nuclear power. The development of gamma-ray detection technologies has received a great deal attention; however less attention has been paid to improving the detection of SNM neutrons even though technology does not fully meet either existing or evolving mission needs.
Given the increased attention that nuclear security has received in the last decade and the resulting advances in the technologies to support them, it is the goal of this thesis to look at the efficiency of a novel mixed-field radiation detector to be used for the replacement of existing 3He-based neutron detectors due to the diminishing availability of 3He in the world.
All SNM emit neutrons, which includes both plutonium (Pu) and highly enriched uranium (HEU). The high rate of spontaneous fission of Pu (typically ~105 neutrons/s/kg) makes it the principal target of neutron-detection, as oppose to HEU (typically ~1 neutron/s/kg).  This does not mean to say that emissions from HEU cannot be exploited but they provide a usable signature in only a fraction of the application space.
While substantial SNM neutron emissions are not ubiquitous, neutrons have two intrinsic benefits: an exclusive signature and long attenuation lengths in materials that readily absorb gamma rays. Gamma-ray emissions are plenteous and span a wide range of energies but they are problematic when it comes to telling them apart from highly variable ambient backgrounds and can very easily be misread with nuisance sources ranging from medical isotopes to naturally arising radioactive material. In comparison, the sheer presence of neutrons above background levels is a brilliant indicator of SNM since ambient background emissions are meek, typically 0.005-0.01 neutrons / s cm2 from background at sea level.  Benign sources (from a non-proliferation point of view) that emit neutrons consist mainly of soil density gauges that use either a combination of americium and beryllium or 252Cf. These annoyances are sporadic in comparison to their gamma-ray counterparts. Moreover, SNM is distinct in that the nature of neutron emissions from the fission process provides unique signatures. Many applications do not manage the solid angle and measurement times required to exploit the correlations between neutron emissions, however when characterising SNM, these signatures possess striking fidelity.
The second advantage neutrons provide, long attenuation lengths, is beneficial in those applications where efforts are made to detect neutrons through attenuating materials - either incidental (such as a search application) or a product of engineered shielding (such as in the case of spent-fuel storage). Where gamma-ray emissions from SNM can penetrate 10s of centimetres of material under ideal conditions, fission neutrons with an average energy of approximately 1 MeV may penetrate through meters of the same material without being absorbed.
The very assets that make neutron detection eye-catching lead straight to two intrinsic challenges in detecting fission neutrons. First, unlike the circumstance of gamma-ray detection where it is possible to choose materials with advantageous properties (like high atomic number and high density), neutron scattering cross sections are diffident across the periodic table. Even when scattering does take place, such events are problematic to detect because of their limited energy depositions. Second, the detection of correlated neutron emissions needs large solid angles, hence restricting its applicability to situations that provide significant control over measurement conditions.
These properties levy the requirements of high intrinsic detection efficiency and the capability to scale to large areas (at a reasonable cost). Particularly in interdiction, exceptional gamma-ray rejection efficiency is also required. Since gamma-ray sources are omnipresent in many interdiction environments and they frequently possess large activities, false neutron alarms may result from exposure to such sources if gamma rays cannot be rejected in a proper manner. This is particularly important since the special nature of neutron sources often trips a series of instant responses. Scalability to greater area is a cross-cutting need that spans interdiction, search and characterisation. Intrinsic detection efficiency is vital in signal-starved environments, and it is especially relevant to technologies that may absorb neutrons with a high probability but do not have the capability to read out these events efficiency (such as traditional boron-lined proportional tubes).
Present day technology developments efforts must overcome three challenges. First, deployments of neutron-detection technology comprise largely (perhaps entirely) of 3He-based proportional counters that are sensitive primarily to thermal neutrons. But because of a recent realisation of an international shortage of 3He that is a result of cutbacks in tritium production in the post-cold-war era, deployment en masse of 3He is unsustainable, at least in the near term.  Second, improvisation in the capabilities of technologies that permit source characterisation in terms of mass, material matrix and geometry will be required to assist the renewed international focus on treaty verification and material safeguards. This is a timely issue that will likely receive a lot of consideration in the initiation of new agreements and a more prominent role of international bodies such as the International Atomic Energy Association (IAEA). Third, the development of directional detectors provides entirely new capabilities not accessible in the thermal-neutron-detection system, although the value of these systems has not yet been confirmed.
2 Physical characteristics of neutrons
2.1 Fission Neutrons
The fission process is complex and has been studied empirically with in-depth detail.  Summarised here is only the properties of neutron emission that affect the design and performance of neutron-detection systems, which includes the relevant parameters of neutron production rates, emitted energy spectra and multiplicities.
Neutrons are emitted by all SNM isotopes that undergo spontaneous fission but the rate of neutron production, rate of spontaneous fission, vary noticeably as a function of isotope. The even isotopes (238U, 238Pu, 240Pu, 242Pu) produce the vast bulk of spontaneous fission neutrons. The primary isotopes of interest have their neutron production rates listed in table I. The largest emitter is Pu and it emits ~60000 neutrons/s/kg at weapons grade, more than 90% of the isotope 239Pu and 6% of isotope the isotope 240Pu. In comparison, less than 1 neutron/s/kg is emitted by HEU, with the assumption of an isotopic composition of 93% 235U and 7% 238U.
Table I. Neutron emission properties of prominent SNM isotopes.
Average multiplicity b
Watt Spectrum c
Spontaneous a fission rate neutrons/ (s kg)
a Reference , b Reference , c Reference 
When considering the application space then the knowledge of two benchmarks (show by Kouzes et al. from the perspective of interdiction) proves useful.  The first is the emission of 106 neutrons/s that is given by a 4 kg weapons-grade sphere of Pu (90% 239Pu and 10% 240Pu). The second is the emission of 97 neutrons/s that is given by a 25 kg sphere of HEU (90% 235U and 10% 238U). These sources are comparable in mass to the significant magnitudes used by the IAEA to approximate the number of nuclear weapons that a country possesses.
2.1.1 Time correlation of fission neutrons
An important characteristic of SNM is the burst-like nature of the fission and multiplication processes that permits multiple neutrons to be detected from a solitary originating event. Multiplicity counting is often the nomenclature given to measuring the signature, which exploits the fact that the perceived neutron distribution does not obey the classic Poisson statistics. Poisson behaviour is observable in many other types of radioactive decay where either only a single particle is emitted or detector efficiencies are low enough that only a single particle will be observed in a given time interval. In the case of fission and multiplication processes are highly time correlated: emission of a single neutron has a distinct probability of being associated with one or more additional neutrons in a narrow time window
This non-Poisson behaviour makes itself apparent at several time scales, each of which is of interest. The fission timescale itself is extremely fast (â‰¤ picosecond), and the two to three neutrons emitted during the fission process known as prompt emissions. Since only a fraction of neutrons escaping a sample will be detected and most individual fission events are of low multiplicity, fission chains produced via (n,fission) reactions contribute significantly to observed multiplicity events. It is important to note that these fission chains need not be initiated by a fission event but may also be triggered by ambient background or (Î±,n) neutrons. The time between successive fissions is ~ 10ns. Chains of successive fissions become more prevalent and longer as the geometry dependent multiplication parameter of the system increases. Depending on the isotopic composition and geometry of the material, relatively rare but distinct ensembles of dozens of neutrons may occur in 10 to 100ns time scales. Note that thermalisation times (~10-100Âµs) are quite a lot slower than these time scales.
An important factor in the neutron production rate is the multiplicity of neutrons, whose means are listed in Table I. Fission events may emit significantly more neutrons than the average, with decreasing probability at higher multiplicity. Figure 1 shows the multiplicity distributions for several plutonium isotopes, which suggests that it not uncommon for 4 or more neutrons to be emitted per fission.
Figure 1: Comparison of multiplicity distributions for plutonium isotopes undergoing spontaneous fission (s.f), thermally induced fission (therm), and fission at 2 MeV (Ens98). The shift between the spontaneous fission and induced fission distributions plays an important role in multiplicity measurements.
In addition to mass, an important quantity in characterizing SNM is the sample multiplication, in other words the number of neutrons that exist in a sample per initial neutron. The convergence of three nuclear properties fundamentally defines the properties of fission chains, which macroscopically evident themselves as the sample multiplication, for a given geometry and isotopic composition. These are the number of neutrons emitted per fission, the emitted neutron spectrum, and the cross section for neutron-induced fission (as shown in Fig.2).
Fig 2: Plot of cross section for neutron-induced fission vs neutron energy. The multiplication of SNM is a function of these cross-sections and the energy spectrum as well as the multiplicity of emitted fission neutrons.
When detecting neutrons, a key observation is the leakage multiplication that is defined to be the number of neutrons emitted from a sample divided by the number of neutrons present at the beginning of a fission chain. The leakage multiplication is by definition greater than one and generally less than three. For example, Hollas et al. determined the leakage multiplication of subcritical HEU spheres to range from 1.16 for 0.6kg to 1.96 for 10.9kg samples. For the previously mentioned benchmarks presented by Kouzes et al., the leakage multiplications were 2.14 and 2.84 for the Pu and HEU, respectively.
2.1.2 Energy spectra of fission neutrons
A nucleus undergoing fission releases ~ 200MeV but neutrons carry away approximately 5 MeV.  How this energy is distributed among fission neutrons has an important effect on the nature of detection systems. The energy spectrum of fission neutrons has been described by various analytical and empirical models.  One suitable parameterisation is the Watt fission spectrum given by:
where E is the neutron energy, and a and b are fitted parameters, as shown in Table I. A lack of empirical data exists in the literature for the spontaneous emission spectra from isotopes prevalent in SNM,  but measurements of spontaneous emission from 252Cf  and induced emission from isotopes of SNM have been reported, most recently by Iwamoto.  Figure 3 compares the energy spectrum of neutrons, based on model parameters, for several isotopes. The flux falls below 10% of its maximum below 10 keV and above ~4.5 MeV. The majority of emitted neutrons initially possess energy of about 1MeV. The variations in the underlying emitted spectra amid different SNM isotopes are of little consequence in non-proliferation. The challenge of detecting SNM is significantly more complex than simply discriminating the fission spectra of Fig. 3 from the background distribution described below, however. Fission spectra are often significantly displaced to lower values by various processes, including down-scattering in the source itself or in material surrounding either the source or detector. For example, the low-Z explosives found in nuclear weapons, hydrogenous shielding used to shield material from discovery, or simply materials that happen to be near the object may all contribute to the down-scattering process. Figure 4 shows a spectrum of neutrons emitted by a sphere of 240Pu surrounded by a hydrogenous cover as modelled using the particle transport simulation code MCNP; the shift to lower energies is readily apparent. This softening of the energy spectrum has some important consequences for detector design and implementation in SNM detection, in particular the need for detection sensitivity in the epithermal energy region (1eV to 10KeV).
Fig 3: Plot of probability vs energy for spontaneous neutron emission of various isotopes. Curves are based on Watt parameters from Table I. Data points are measurements of 240Pu from [Vla01]. For the applications discussed in this thesis, the slight differences in these spectra are not important.
Fig 4: Comparison of fission neutron distributions stemming from a sphere of 240Pu without (un-moderated) and with (moderated) a hydrogenous cover.
2.1.3 Alpha-induced emissions
Fission neutrons are not the sole source of neutrons emitted by SNM. When SNM exists in an oxide form, fission neutrons are only a fraction of the total emitted flux. This is because the probability of decay via alpha emission is significantly greater than for fission in almost all SNM isotopes. After alpha emission, neutrons are created from (Î±,n) capture on low-Z materials, most commonly oxygen and fluorine. This is a particularly favourable process for 238Pu that may produce abundant amounts of neutrons even when present in fractions below 1%; 240Pu also has a large (Î±,n) contribution. These neutrons may create significant backgrounds that correlated fission events must be discriminated against. Since neutron detectors do not generally measure neutron energy and thus cannot distinguish between fission and (Î±,n) species, the presence of (Î±,n) neutrons levies the requirement that that detection systems possess, to the maximum extent possible, a flat response as a function of incident neutron energy. This requirement is mitigated to some degree by the unforeseen fact that the average energy of (Î±,n) neutrons emitted by oxygen is ~ 2MeV - very close to the average fission neutron energy - which allows one to assume equal detection probability for either species with minimal impact on measurement reliability. For a thorough discussion on the origins of (Î±,n) neutrons, consult Reilly et al. or Ensslin et al.
Most ambient neutrons are remnants of cosmic-ray interactions in the upper atmosphere but they are produced lose to the Earth's surface from (high-energy) secondary neutrons that multiply through secondary collisions; for details, see Ziegler,  and references therein. While extensive variability in the flux of ambient neutrons has been reported even at sea level measurements, Kouzes et al. adopt a representative value of ~100 neutrons neutrons/ (s m2) for the ambient sea-level flux that is distributed over a wide variety of energies. Attempts have been made to decipher the energy spectrum using data from various Bonner spheres supported by modelled detector response functions.  The results of this work imply energy dependence with four key features that include: (1) a large thermal component, (2) a plateau region encompassing from the thermal region up to approximately 0.1 MeV with a 1/E dependence, (3) an evaporation peak spanning from 0.1 to 10MeV that contains about 30% of the integrated flux, and (4) a high-energy peak centred at 100 MeV.
A large amount of background neutrons reside at thermal and epithermal energies but these neutrons can be removed from detector response functions by suitably designing moderators that absorb incident thermal neutrons, for example, by using materials such as cadmium. Most problematic to SNM detection is the evaporation peak in background, shown in Fig. 5, which resides at energies that overlap the fission neutron spectrum. In this energy regime, ambient background largely overlaps the fission energy distribution with a faintly lower (with respect to the resolution of neutron spectrometers) mean energy. The local maximum of the background occurs at ~ 0.5 MeV whereas the fission peak resides at ~ 0.8 MeV. Recent measurements of the energy dependence of background in this energy range, using a measurement technique based on multiple scattering, have confirmed the energy distributions reported by Gordon et al. but show excellent agreement with the normalisation of Ziegler.
The variability of the background flux has been measured as a function of altitude and latitude., ,  Mascarenhas et al. measured flux values at the Earth's surface, and they report results, integrated from 0.5 to 10 MeV, of approximately 0.005/ (cm2 s) at sea level to 0.035/ (cm2 s) at 2500m. The latitudinal variation in the flux between 19 Â°N and 56 Â°N was about a factor of 10 with a maximum at higher latitudes, as reported by Goldhagen et al. The flux increased by a factor of 1000 between sea level and altitudes of 10 to 20 km. While this variability has complications on the sensitivity of detection systems, local neutron backgrounds are mostly small and stable relative to gamma-ray backgrounds that give thousands of counts per second in detectors covering meter square areas and vary by factors of 10 over small geographic regions.
Benign sources of neutrons are somewhat rare (outside of nuclear laboratories) but are nonetheless present as part of moisture gauges used in construction, for example. These devices principally incorporate either 252Cf or americium and beryllium that together produce neutrons via the 9Be (Î±,n) reaction. Neutrons from 252Cf are emitted in a spectrum whose shape is quite similar to that of background in the region between 0.1 and 10 MeV but the 252Cf distribution has a longer tail at low energies. Neutrons from AmBe sources possess a distribution of energies ranging from 0.1 to 10 MeV but the huge majority of neutrons possess energies at the higher end of the fission spectrum, making them distinct from fission neutrons. Figure 5 compares both of these sources to background in the region of 0.1 to 20 MeV.
Fig 5: Contrast of ambient background energy distribution with those from AmBe and 252Cf fission sources
3 Principals of neutron detection
3.1 Interaction mechanisms
There are three principal mechanisms by which neutrons interact with nuclei of detection media: elastic scattering, inelastic scattering, and capture reactions. Elastic scattering is the primary energy-loss process for neutrons at fission energies, with archetypal cross sections of the order 1 barn, as shown for the case of 1H in Fig. 6.  Inelastic scattering in the form of (n,n') plays an important role at these energies in some nuclei, particularly heavy elements but more commonly inelastic cross sections are at least an order of magnitude lower than elastic cross sections.
Fig 6: Interaction cross sections as a function of energy for materials appropriate to neutron detection.
In designing a detector that exploits elastic and inelastic scattering, maximising the cross sections for these interactions is essential but not enough. Another key figure of merit is the amount of energy deposited in a scattering event. From Newtonian mechanics, the energy of the elastically scattered neutron is a function of the atomic number (A) of the target nucleus. If one inspects the case of maximum energy loss that occurs in backscattering, the scattered neutron's energy (Escattered) is:
Escattered = Einital
where Einitial is the energy prior to scattering. Maximising the amount of energy deposited (Einitial âˆ’ Escattered) requires minimizing 'A'. With the exception in the case of hydrogen, where a neutron can lose almost all its energy, the maximum possible energy deposition is only a small fraction of the initial energy, as shown in table II.
Table II: Maximum energy deposition in elastic scattering
Fraction of initial energy
In a detector that registers the deposited energy in the method of scintillation light or ionized charge, applied limits on the capability to discriminate gamma rays from neutrons, for example when using pulse-shape discrimination, are at greatest of the order of about 10 KeVee. These limits have important consequences on detector design. For example, assuming a conventional threshold of 500 keVee, no neutrons scattering off carbon with energies below approximately 2 MeV can be detected. This eliminates approximately half of the fission spectrum from consideration. The need for substantial energy deposition creates a reality where only hydrogenous materials can be formed into practical detectors of fast-neutron scattering events; at least until meaningfully increased gamma-ray rejection efficiency is achieved at lower thresholds.
The physical and practical challenges of fast neutron detection have compelled most detector development toward manipulating the greatly enhanced thermal neutron capture cross sections, commonly quoted at 0.025 eV, listed in Table III. The energy dependence in this area generally lacks structure and follows the reciprocal of the neutron velocity (see Fig. 6). In addition to increased cross sections, these capture reactions serve as intensifying reactions that, via their Q values, convert low-energy neutrons into energetic charged particles and photons. This is a critical property in forming a practical thermal neutron detector. The cost of exploiting the thermal neutron detection region is an intense reduction in the population of fission neutrons at these energies - only a tiny fraction of the Watt spectrum extends to this regime. This fact demands the use of moderation prior to detection.
Table III: Properties of common thermal neutron-capturre isotopes and 235U for comparison
Thermal capture cross section (b)
Major reaction products
1H (0.57 MeV)
3H (0.19 MeV)
3H (2.73 MeV)
4He (2.05 MeV)
4He (1.47 MeV)
7Li (0.84 MeV)
e- (29 & 72 keV, others)
Gamma rays (7.9 MeV cascade)
Fission fragments (168 MeV)
Neutrons (5 MeV)
The principle behind moderation is to use isotopes with fairly large scattering cross sections to slow down neutrons until they exist in the thermal regime where neutron capture agents, such as 3He, provide the main interaction channel. The efficiency of a moderator is given by the same figures of merit as detection media: the largest possible scattering cross sections and the maximum amount of energy deposited per interaction. It is also essential, however, to minimize the capture of neutrons in the moderator once thermalized. These constrictions quickly lead to hydrogenous materials since the capture cross section for 1H remains below 0.1 b down to 0.1 eV.
Organic compounds are the typical choice for moderators, for example, polyethylene (C2H2)n with a density of ~ 0.9 g/cm3. At least two studies of moderator geometry have been published. ,  Both wanted to determine what moderator geometry maximises the detection efficiency given a fixed amount of 3He. Reilly et al. modelled a 3He proportional tube inside a slab of polyethylene subject to a fission spectrum comparable with 252Cf. Simulated efficiencies as a function of moderator thickness in front and behind the tube showed a maximum efficiency with 6cm thick moderator in front and an indeterminately thick moderator in the back. Also it was observed that there were diminishing returns after a moderator thickness of approximately 4cm - a thickness where 90% of the maximum efficiency is achieved. In precisely modelling portal moderator designs, Kouzes et al. reached the same conclusion that moderators in front and behind the thermal-neutron absorber are essential for maximising efficiency.
3.3 Performance Metrics
There are numerous standard metrics for portraying detector performance, and there are a few critical metrics specific to detecting, locating, and characterizing SNM. Frequently, the first metric to cogitate is intrinsic detection efficiency that is defined as the number of counts recorded by a detector per incident neutron, in our case from a fission spectrum. (This is generally done assuming an incident flux normal to the detector as would largely be the case when the source-detector separation is knowingly larger than the detector's lateral extent.) This metric is especially relevant in thermal neutron detection where one can absorb an incident flux of thermal neutrons (such as those leaving a moderator) with relative ease but constructing an efficient system that can read out all of those captures has proven difficult (outside of the standard of proportional tubes). A good standard to keep in mind is the 77% intrinsic efficiency of 3He proportional counters (at 4 atm) for thermal neutrons.
When thinking of deployment to the field, collection efficiency is an important metric, especially in signal-starved interdiction circumstances. Collection efficiency, defined as the number of neutrons emitted from a source that enter a detector, is strictly a geometric property given by solid angle coverage. It plays a role, for example, when comparing gaseous detectors, which can simply be scaled at reasonable cost, to semiconductor-based detectors, which will likely have difficulty obtaining large collection efficiencies. The convolution of intrinsic detection efficiency and collection efficiency produce the absolute detection efficiency that is the number of counts recorded by a detector per neutron emitted from a fission source.
An equally important metric in certain applications is the capability to discriminate neutron and gamma-ray interactions. For example, plastic scintillation detectors, discussed below, have a high hydrogen content and consequently significant intrinsic neutron-detection efficiency. Problematic, though, is the point that they are also sensitive to gamma rays, especially low-energy gamma rays that are plentiful in background radiation. Neutron signals are consequently generally buried in a sea of gamma-ray interactions. Other detection systems discriminate neutrons from gamma rays by various methods, including the magnitude of pulse heights and the timing characteristics of pulses. Their ability to do this, the gamma-ray rejection efficiency, is the number of false neutron counts per incident gamma ray. This is generally computed using sources such as 137Cs or 60Co but measuring the discrimination of gamma rays from neutrons is not an easy process because numerous parameters affect its results. For example, 3He proportional tubes exhibit excellent gamma-ray rejection efficiencies in most applications but in high radiation fields ( ~1 R/h), they hold essentially no capability to discriminate events. Furthermore, there is a trade-off between the intrinsic detection efficiency and gamma-ray rejection efficiency when changing lower level discriminators. The same is true of many scintillation-based neutron detectors that fail to provide discrimination in much lower radiation fields. To be comprehensive, it is therefore necessary to characterise the gamma-ray rejection efficiency as a function of the incident gamma-ray flux at constant intrinsic neutron detection efficiency.
Another significant feature in the framework of measuring SNM is the temporal response of the detector. As discussed in section 2.1.1, fission neutrons are emitted in bursts, create chains of fissions on the scale of tens of nanoseconds, and thermalise and capture in time scales of 10 to 100 Âµs. Detectors with fast response times, such as liquid scintillators, are hence of particular interest for characterising events on the time scale of fission chains while thermal neutron detectors are sensitive only to the microsecond scales associated with thermalisation.
4 Fast Neutron Detection Method and Application
An ultimate fission neutron detector would be sensitive to the range of energies emitted in the fission process. The perplexing factors of meagre cross sections and modest energy deposition make realising fast neutron detection in a deployment system a prickly proposition. Deployments of neutron
detection technology have subsequently been composed of thermal neutron detectors enclosed by moderating material. There are numerous old and a few modern methods of detecting thermal neutrons (gaseous detectors, water-based Cerenkov detectors, conversion layer detectors, homogenous compound semiconductors, micro channel plates and scintillators), however in this thesis the discussion is of a fast neutron detector, which, after a significant amount of research and development, has matured to a prototype system available with decent to gamma-ray rejection efficiencies.
4.1 Organic Liquid Scintillation
Direct detection of fast neutrons is an vital technical challenge for two applications. Firstly, fast neutrons preserve directionality whereas the moderation process at the detector eradicates all "memory" of incident direction after a few scatters. Secondly, techniques manipulating time correlations among fission chains, such as the multiplicity counting discussed in section 5.3, would find it beneficial from systems with time resolutions significantly faster than the thermalisation process. For the purpose of this thesis, we define fast neutrons as those with energies â‰¥0.5 MeV, although this cut-off is rather arbitrary. Detecting neutrons in this energy range is fundamentally more difficult than at thermal energies because of relatively modest cross sections that are 2 to 3 orders of magnitude smaller, as shown in Fig. 7. Nonetheless, a collection of approaches has been developed to directly detect fast neutrons, each with its own quirks. One common thread exists: at the foundation of each approach is the detection of recoil nuclei, which are almost exclusively hydrogen.
Scintillators convert the energy in recoil nuclei into scintillation light, which is most frequently detected with photomultiplier tubes. In many scintillators, the time development of the scintillation pulse varies between electromagnetic interactions, such as those induced by gamma rays, and nuclear recoil interactions induced by neutrons. This provides a means for discriminating fast neutron recoils from gamma-ray or electron interactions and is a valuable feature in the framework of SNM detection.
Organic liquid scintillators are frequently used for fast-neutron detection; when doped with neutron
capture agents, they develop the ability to become sensitive to thermal neutrons as well.  They comprise of a hydrocarbon solvent base into which one or more organic phosphors are dissolved. The phosphors absorb energy from the solvent and emit ultraviolet radiation. Often, a second phosphor or wavelength shifter is added to shift the UV light to the blue wavelengths, which are best suited for photomultiplier readout. Many examples in long use, such as toluene and pseudocumene, are flammable, toxic, and can attack acrylic and other plastics over a period of time. More recently, effective replacements with reduced toxicity, reactivity and flammability have been discovered, such as phenylxylylethane and linear alkyl benzene scintillators.  Liquid organic scintillators are beneficial because they can be fabricated into large detectors and outfitted with pulse shape discrimination techniques.  These discrimination techniques are vigorous, with gamma-ray rejection efficiencies of the order of 10âˆ’4 but their response is energy dependent and operative only above ~ 500 keVee. Recent efforts have endeavoured to lower this threshold; see, for example, Prokuronov and Shabalin.
Organic liquid scintillators are a tool of increasing importance for measuring SNM because of their fast timing sensitivity. For example, arrays of liter-scale cells can be used to achieve large solid angle coverage and can trigger on the characteristic coincidence times of prompt fission events and
fission chains. This fast timing (0.1-1ns) allows for exploitation of the non-Poisson behaviour, described in section 2.1.1, using arrays.
Applications that aim at the detection of SNM include portal monitoring (such as of vehicles and cargo), wide-area search (such as in emergency response), and monitoring for the enrichment of 235U (via 234U (Î±,n) reactions in gas centrifuge plants). Following common practice in this field, the detection of SNM here does not imply unique identification thereof. As an alternative, detection is understood to indicate the presence of a raised neutron rate relative to backgrounds, consistent with SNM but also with harmless neutron sources.
Common measurement restraints may include relative source-detector motion that limits acquisition time to about 1-10s, for example, in portal monitoring and search. Limited access to the object of importance may also be a challenge, possibly because of standoff distance, intervening material, or other factors, including measurement protocols forced in verification regimes. These constraints often create signal-starved environments, for example, because of low collection efficiencies. Such low collection efficiencies in turn create the requirement for detection systems with high intrinsic detection efficiencies so as to make maximal use of neutrons entering the detector. Since high-activity gamma-ray sources are not scarce in the stream of commerce, and since ambient gamma-ray backgrounds are relatively high, gamma-ray rejection efficiency is a particularly important metric that has ill-fated otherwise promising alternatives to 3He.
Neutron sources in these applications are detected with moderated thermal neutron counters. The reliability and robust performance given by the simple design of 3He-based systems have allowed them to govern applications in need of large-area detection systems, such as those for border security. This makes the problem of a declining 3He supply particularly serious. The robustness and low power consumption of 3He-based systems is especially beneficial in unattended systems.
For conciseness, we concentrate here on interdiction since it is principally timely and illustrative, for an in-detail analysis of the performance of neutron detection systems in interdictions situations, see Kouzes et al.. From the decision making viewpoint, detecting the presence of a neutron source is a much simpler task than in the case of gamma rays since neutron backgrounds are lower in magnitude and less variable. Alarms may be issued under a simple analysis that compares the counts in a measurement period to an alarm threshold (Talarm) that is calculated from a premeasured background (B) and a multiplier (m):
As pointed out by Kouzes et al. since neutron count rates can be very low, constant offsets can be included in alarm thresholds that mitigate against alarms in cases where the pre-measured background is extremely small.
The assumption in this type of threshold formulation is that Poisson counting statistics govern variations in the background count rate. This lies in contrast to the use of gamma-ray detectors, whose response is significantly affected by the self-shielding related with the presence of massive conveyances.  But in the case of neutron sources, this assumption is well established since self-shielding effects are modest, and the above-mentioned benign fast-neutron sources of 252Cf and AmBe are relatively sporadic. Measuring neutron energy spectra is not practical in interdiction because of the limited number of measured events but threshold detectors or sensors that combine operational simplicity with rudimentary energy binning may be of some use if discriminating source types is a high priority.
Neutron source detection in the maritime environment presents a exclusive set of challenges. In particular, detection systems operating on-board large vessels experience what has been labelled the "ship effect" that brings a large degree of localised variability in background measurements. This makes the type of background estimate needed for equation 3 problematical. The origins of the ship effect arise from cosmic ray interactions in the upper atmosphere that manifest themselves at sea level in the form of secondary particles. These high-energy neutrons and charged hadrons induce spallation, particularly in materials such as iron, which in turn creates bursts of multiple neutrons that can mimic the burst-like nature of SNM signatures. For a discussion of this effect, a catalogue of references to cosmic-ray induced neutron fluxes, and their insinuations on interdiction, see Kouzes et al.
These events can intensely change background rates. One experiment compared the neutron flux at an air-steel interface to that of an air-water interface and showed a factor of 25 increase in the air-steel case  whereas another study reported equality in neutron emission between 227kg of lead and 1.87g of Pu.  In probing for SNM in maritime environments, these factors may require the use of higher threshold settings and/or a priori characterisation of the spatial dependence of backgrounds. Outside the problem of elevated rate, the time-correlated signature of local spallation events adds to the struggle of quashing this background.
Active interrogation technologies currently under development for detecting shielded SNM impose different necessities than passive detection systems. For detection (as opposed to characterisation of the sort described in Bertozzi et al.), the generic manner of processes in active interrogation is to stimulate fission in a target via either high-energy photons ,  or neutrons.  In certain cases, delayed neutrons can be counted following neutron-induced or photon-induced fission with modest modifications to conventional 3He technology  since detecting delayed fission products gives a measurement scheme akin to passive detection. One variation, for example, is that detectors must be robust enough to rapidly recover from interrogation pulses.
Preferably, detectors would be capable of operating in the signal-rich prompt regime during or directly (~ nanosecond) after interrogation pulses.  Fast-neutron detection systems might provide productive ground for operation in this environment.  Threshold detectors, for example, those founded on bubble-chamber technology,  in principal provide inherent insensitivity to photons and low-energy neutrons, such as those used in the "nuclear car wash." . A second importance of these detection schemes is their ability to be scaled to large areas at a practical cost, which is essential in the case of detecting the weak signals from shielded SNM. Other schemes could exploit nuclear reactions in detectors that quickly capture neutrons but have delayed emissions themselves, for example, via capture reactions with delayed emissions. Continuous investments in the development of active interrogation systems, along with the challenges this environment poses, may produce truly unique methods of neutron detection.
Directional and imaging detectors have the prospective to complete important fortes in the SNM application space but they have not yet been installed operationally. Two universal classes of applications somewhat divide the requirements imposed on these detection systems. In both classes, the detection function is implicit in this application. The first is standoff detection, defined here as detection in excess of ~ 3 m. This would be the case in a search situation where the source location is unidentified or in the external monitoring of a storage facility where the movement of sources may be of interest. At these distances, fast-neutron detection raises the potential for an improved standoff range compared with thermal neutron imagers since the fast neutrons have a significantly better mean free path in air (~10m). The second application class has the aim of generating a detailed map of the distribution of SNM within a nearby volume. Examples here include counting spent fuel assemblies in a safeguards regime or counting warheads as part of a dismantlement verification regime where standoff distances are commonly less than 10m. For the huge mainstream of situations in either type of application, static source-detector motion is necessary because of the need for extended acquisition times.
The first class of locating SNM at standoff is a research and development importance in the non-proliferation community where the operational objective is to identify an angular region with surplus neutron emissions so that other assets can be focussed toward determining the nature of the source. Therefore, moderate angular resolution (~10Â°) is necessary. Presently, the majority of search applications depend on on what has been termed "proximity imaging," where repeated measurements from a non-imaging detector inform source localisation, for example via a rate increase as a function of position in a simple neutron counter. Proximity imaging has the asset of simplicity but it suffers from long dwell times inherent when performing repeated measurements on a grid.
One important trade-off in system design for standoff detection occurs between angular resolution and the "imaging efficiency" - the efficiency with which systems can detect and rebuild events. For systems that depend on on the elastic recoil of neutrons, an average source direction is deduced from a statistical ensemble of scatters rather than from single scattering events. As a result, the collection efficiency must be large, consequently delegating a physically large system volume ~1m3 to attain sensible dwell times. This enables for rapid convergence on a mean angular direction for the above-mentioned distances and significant quantities of SNM. From a small number of interactions, these systems attain a modest angular resolution (~10Â°). Detector size is challenging, for example, in cases where access is limited to small entry apertures or where detectors must be human-carried.
The second application class of mapping the spatial distribution of neutron emissions exist in the opposite extreme in terms of standoff distance and spatial resolution. One particularly relevant example is the difficulty of counting warheads on a re-entry vehicle bus, which may be achieved as part of a confidence building measure in strategic arms reduction. In this case, distances and acquisition times are the topic of discussions but sensible starting parameters may be distances (1~m) and acquisition times (~1000s). This is also an active area of speculation within the non-proliferation community. The applicability of radiation detection to this problem was first proven using gamma-ray imaging technology pioneered by Ziock et al. but Sailor et al. considered the application of fast-neutron imaging technology toward this end. Renewed interest exists in addressing this challenge. Because of the limited angular resolution (10Â°) and the problem of excess intensity in the images from the projected cones that do not correspond to a source, the utility of fast-neutron imaging for warhead counting is a challenging proposition that will require systems with a significantly higher angular resolution (1Â°). This may be enabled by using thermal neutron imaging using apertures (pinhole or coded) that allow for higher resolution than fast neutron imaging. Next generation fast-neutron imagers could also meet this resolution, however, through the tracking of double-scattering events, the use of TPCs (time projection chambers) that incorporate 3He (assuming limited availability), or the development of fast-neutron pinhole apertures. The cost of these approaches is restricted efficiency and thus extended measurement times.
An additional current measurement objective is the certification of SNM in spent-fuel assemblies that are at risk for fuel-pin diversion. This environment challenges technology because fuel assemblies are stored collectively in large numbers (~20) and reside in environments with large ambient radiation
fields (spent-fuel cooling ponds or dry-cask storage areas). The task is further problematic since, for radiation safety, assemblies are enclosed by water or thick, often concrete, casks. Access is thus a problem, and even if access can be obtained, neutron scattering by the shield material makes recoil-based imaging hard or impossible. Wet storage in pools offers access to individual assemblies that can be characterised by merging neutron and gamma-ray measurements of sensors placed directly inside the assemblies, using empty control rod channels to permit proximity imaging. Ham et al., whose measurements in high radiation environments demanded the use of fission chambers, have carried out experiments recently.
The in-depth properties of fission neutrons are possibly most useful when the existence of a source and its location is known, and the aim is to determine the SNM mass and/or geometry. While source terms are a complex superposition of numerous neutron-emitting pathways, the resultant information is of great significance to a range of missions, including (1) safeguards where the objective is to determine mass for material bookkeeping, (2) dismantlement where the aim is to verify the disassembly of nuclear warheads, and (3) interdiction response where, after identifying the presence of SNM, it is desired to understand its form - especially if it exists in in a threatening conformation such as an assembled nuclear weapon.
Manipulating the rich nature of neutron signatures in the form of multiplicity counting permits for substantial insight into the source term at a significantly higher level of reliability than in the previously mentioned applications. This comes at the price of longer measurement times, typically ~1000s,  and large solid-angle coverage. These requirements occur for two reasons. Firstly, measuring multiple neutron events that are isotropically distributed is a function of the neutron efficiency raised to the power of the multiplicity moment. For example, in systems that have an absolute detection efficiency of 10%, the probability of measuring a triple-neutron emission is 0.001. The detection probability of three coincident neutron events rises significantly for higher multiplicity events but this comes at the cost of a lot less events to detect. These measurements thus integrally require large solid-angle coverage. Secondly, determining the source terms requires de-convolution of contributions to the neutron flux from spontaneous fission, induced fission (which is established
in the form of multiplication), and (Î± ,n) reactions. The correlated signal emerging from a source is consequently not a simple distribution consistent with an individual fission event, such as those multiplicity distributions of Table I. Additionally, the energy spectrum from spontaneous fission, induced fission, and especially (Î± ,n) events are different, thus requiring a detector system with a comparatively flat response to evade a complex or impossible de-convolution. Thirdly, higher backgrounds can significantly increase measurement times required for a given level of precision. This is a factor not normally present in the detect and locate applications but it is one that may occur if the detection system is operating in a facility with large amounts of SNM. For example, recent measurements made by Chichester and Seabury  were performed in a nuclear facility with a background approximately 20 times that of the surrounding environments. These background neutrons not only add to a baseline detector response but they also induce fissions in the SNM, which can be beneficial in the detection of fission chains. One specific challenge manifests itself in cases where a significant amount of high-Z material surrounds the measurement assembly. Cosmic-ray events, which induce spallation, can produce neutrons in large multiplicities that are correlated on the same time scale as fission events. 
Non-destructive assay techniques have long been applied in material bookkeeping, specifically with the aim of quantifying the mass of SNM. (For an in-depth overview of the methods and detection systems used in this application, refer to Ensslin et al.). Here, neutron detection offers a method that is not available in, yet complimentary to, gamma-ray spectroscopy, where the low-energy gamma rays exploited in isotopic characterisation of SNM (refer to Berlizov et al.) and Sampson and Kelley ) present to the detector only those emissions originating in a thin surface layer. The path length of fission neutrons frequently allows investigation of the flux from the complete source volume, assuming reasonable quantities. For example, in the case of 235U, the mean free path of a 1 MeV fission neutron is ~3 cm.
Passive neutron measurements infer SNM mass by assessing the spontaneous fission rate of a sample. These measurements have been limited to plutonium because of the low spontaneous fission rates of 235U. To analyse uranium samples, an uncorrelated neutron source such as AmBe must be added to the well so as to induce fission in the sample. In this approach, the instrument is sometimes called an Active Well Coincidence Counter and, subject on the analysis method, may be used as an active multiplicity counter. More recently, a 14 MeV neutron generator has been used to induce fission. Mass must be inferred with the assistance of several assumptions. The first is that the SNM isotopic fractions (f) are known, possibly based on the assumption of weapons grade material or from gamma-ray spectroscopy measurements. It is then possible to define an effective mass relation based on the relative fission rates. For the case of spontaneous fission of plutonium, where only the even isotopes contribute noticeably, the total mass of plutonium can be defined as:
TotalPu = 240Pueff / (2.52f238 + f240 + 1.68f242)
Where, 240Pueff is the mass of 240Pu that would create the same response as that acquired from all the even isotopes of the source in question. The measurement challenge is then to determine 240Pueff, and several methods that trade off accuracy of assumptions with measurement complexity have been developed. What makes this determination perplexing is the fact that the emitted neutron flux is not only a function of 240Pueff but also the source multiplication and the (Î±, n) reaction rate. Measurements must therefore either estimate these three parameters or eradicate them using assumptions or information provided a priori.
The simplest measurement approach is total neutron counting that connects the total neutron flux emitted by a sample to the material mass. This is an effective method for cases where the sample geometry (and thus multiplication) and embedded matrix are known. For example, consdering a uniform cylinder of plutonium oxide, it possible to determine the isotopics from gamma-ray spectroscopy, calculate the multiplication (via Monte Carlo techniques), and estimate the (Î±, n) reaction rate. In many cases, such as analysing plutonium metal where the (Î±, n) contribution can be overlooked, coincidence counting provides the necessary reliability. By measuring both the singles and coincidence (doubles) count rates, one can determine 240Pueff and source multiplication. Historically, coincidence counting has been used on reason of its capability to detect correlated neutron emissions with simple detector geometries that are cost effective. The use of multiplicity counting, a term earmarked for cases where triples (and higher moments) are measured, is most useful in challenging measurements either where accurate estimation of the (Î±, n) contribution is precluded, for example, because of impure or mixed oxides, or where increased accuracy in the determination of the multiplication is wanted.
Various models have been developed to offer analytical relationships between the observed rates of singles, doubles, and triples and the 240Pueff, the (leakage) multiplication, and the (Î±, n) reaction rate. These models rare based upon a host of assumptions with the most noticeable being that (1) all neutrons are emitted at the same time despite the length of the fission chain, (2) the probability of fission and the neutron detection efficiency are uniform over the source, (3) all emitted neutrons have the same energy spectrum, and (4) no correlation exists between neutron multiplicity and neutron energy. Given these and other assumptions, analytical expressions for the fission rate, the multiplication, and the (Î±, n) reaction rate have been derived as functions of the observed multiplicity moments and both the known spontaneous and induced neutron multiplicity distributions ( see fig 1). For a passive analysis of plutonium, reported accuracies using these methods are in the range of 2%.,  In the case of 235U using induced fission, Hollas et al. report accuracies within 5%. These results affirm to the robustness and richness of neutron fission signatures, despite the wealth of assumptions required in their analysis.
Depending on its source, SNM may exist in a assortment of forms, including metal castings, scraps, oxide powders, fluorides, oxalates, or waste that may be stored in containers variable in size from paint cans to oil drums. Many of these systems have been designed, optimized, and calibrated for a particular form of material and a typical range of masses, and therefore different calibration factors must be used depending on the enrichment, the alloy composition (for instance aluminium alloys), or the presence of oxide, fluoride, or neutron absorbers in the container matrix. Overall, independent knowledge of the chemical composition, impurities, inhomogeneity, and isotopic ratios are necessary to obtain reliable analyses. Despite these nuances, the well counter has been extensively used for inventories of SNM and is manufactured commercially for safeguards. 
The timing properties of these systems are restricted to the neutron moderation time (~ Âµs). With their nanosecond timing, systems based on fast neutron detectors, such as the organic liquid scintillators, portend to improve capabilities. Empirical data proposes that the capability to target events with event separation times from 10 ns to 100 Âµs will permit greater discrimination against cosmic-ray events with separation times (~ns) and random background events with typical separation times (~ms).  When operating in high-flux environments, fast timing has increasing benefits because the probability of random correlations from ambient background increases geometrically with the singles rate. Such systems are currently being evaluated for providing high reliability measurements. For example, a system has been tested with ~ 5% total efficiency and ~2Ï€ solid angle coverage. Upon testing with 242Cf sources, acquisition times have been condensed from a few hours to several minutes by reducing random background contributions.