**5.1 Prompt gamma neutron activation analysis**

Admittedly, Prompt Gamma Neutron Activation Analysis (PGNAA) is usually thought of as a continuous (or DC) source technique. In this technique, fast neutrons from a radioisotope, a neutron generator or a reactor impinge upon a sample. The sample then emits gamma rays through (1) prompt gamma ray emission from gamma decay, (2) gamma ray emission through short-lived beta decay, or (3) prompt gamma ray emission due to exoergic nuclear reactions. By the latter, we mean gamma ray emission to satisfy conservation of energy (or mass) in a nuclear reaction. The most well-known gamma rays from this type of reaction are the 2.22 MeV gamma ray from the 1H(n,) reaction and the 10.8 MeV gamma ray from the 14N(n,) reaction.

Pulsed neutron sources can be used in PGNAA systems. However, using pulsing neutron generators without taking advantage of the pulsing mechanism is not efficient.

### **5.2 Pulse fast neutron analysis**

Tsahi Gozani, Peter Sawa, and Peter Ryge conceived of Pulsed Fast Neutron Analysis (PFNA) in 1987 while they were working at SAIC (Gozani, 1995). In PFNA, a large accelerator creates a deuteron beam which is directed at a deuterium gas target. A chopper, which consists of strong electric field that periodically sweeps the deuteron beam away from the target, creates neutron pulses of a few nanoseconds duration. In described system, user has control of the chopper, so moment when neutrons are created and duration of the neutron pulse are known precisely. The PFNA uses time-of-flight (TOF) methods to obtain a favourable signal-to-noise ratio (SNR) to detect various chemicals. The accelerator used in PFNA is typically an 8 MV Van de Graff accelerator. The D(d,n) reaction has a low Q-value (approximately 100 keV). Thus any energy above the Q-value is mostly transferred to the kinetic energy of the neutron, producing a neutron with a maximum kinetic energy of 8 MeV. The velocity of 8 MeV neutrons is about 6 cm/ns. The user, then, knows exactly when the neutron pulse is made and now can estimate its position at any time after its creation.

Furthermore, due to the high kinetic energy of the deuteron beam, the neutron's momentum is parallel to the momentum of the deuteron. The developers took advantage of this fact and used a sophisticated system to "raster" the neutron beam across the object of interest. The

This section covers a number of systems that perform bulk material analysis using neutron induced gamma spectroscopy. Although this is by no means a complete list, it represents systems and techniques that have been utilized in the past twenty years. Some of these systems are still in the market and some never made it to market. Some of the systems have

Admittedly, Prompt Gamma Neutron Activation Analysis (PGNAA) is usually thought of as a continuous (or DC) source technique. In this technique, fast neutrons from a radioisotope, a neutron generator or a reactor impinge upon a sample. The sample then emits gamma rays through (1) prompt gamma ray emission from gamma decay, (2) gamma ray emission through short-lived beta decay, or (3) prompt gamma ray emission due to exoergic nuclear reactions. By the latter, we mean gamma ray emission to satisfy conservation of energy (or mass) in a nuclear reaction. The most well-known gamma rays from this type of reaction are the 2.22 MeV gamma ray from the 1H(n,) reaction and the 10.8

Pulsed neutron sources can be used in PGNAA systems. However, using pulsing neutron

Tsahi Gozani, Peter Sawa, and Peter Ryge conceived of Pulsed Fast Neutron Analysis (PFNA) in 1987 while they were working at SAIC (Gozani, 1995). In PFNA, a large accelerator creates a deuteron beam which is directed at a deuterium gas target. A chopper, which consists of strong electric field that periodically sweeps the deuteron beam away from the target, creates neutron pulses of a few nanoseconds duration. In described system, user has control of the chopper, so moment when neutrons are created and duration of the neutron pulse are known precisely. The PFNA uses time-of-flight (TOF) methods to obtain a favourable signal-to-noise ratio (SNR) to detect various chemicals. The accelerator used in PFNA is typically an 8 MV Van de Graff accelerator. The D(d,n) reaction has a low Q-value (approximately 100 keV). Thus any energy above the Q-value is mostly transferred to the kinetic energy of the neutron, producing a neutron with a maximum kinetic energy of 8 MeV. The velocity of 8 MeV neutrons is about 6 cm/ns. The user, then, knows exactly when the neutron pulse is made and now can estimate its position at any time after its creation.

Furthermore, due to the high kinetic energy of the deuteron beam, the neutron's momentum is parallel to the momentum of the deuteron. The developers took advantage of this fact and used a sophisticated system to "raster" the neutron beam across the object of interest. The

generators without taking advantage of the pulsing mechanism is not efficient.

different acronyms or different trade names but rely on the same physical principles:

PGNAA - Prompt Gamma Neutron Activation Analysis

PFNTS - Pulsed Fast Neutron Transmission Spectroscopy

PFTNA - Pulsed Fast/Thermal Neutron Analysis

**5.1 Prompt gamma neutron activation analysis** 

MeV gamma ray from the 14N(n,) reaction.

**5.2 Pulse fast neutron analysis** 

**5. Methods and applications** 

PFNA - Pulsed Fast Neutron Analysis

API – Associated Particle Imaging

neutron beam also has a small angular divergence and estimates (Strellis, 2009) are that the beam is 9 cm × 12 cm in the center of the object of interest.

PFNA systems can be used to screen very large cargo shipments such as tractor-trailer shipping containers and airport shipping containers. A large, 2-dimensional array of NaI gamma ray detectors covers the cross-section area of the object under scrutiny.

The energies of the gamma rays emitted from the object are plotted against the TOF of the neutron. This creates a two-dimensional array of data that looks similar to a spectrogram in that the intensity of the gamma ray at a particular TOF is represented by a color using the RGB color scheme. In this array of data, color bands parallel to the TOF axis indicate constant gamma ray background such as from normally occurring radioactive materials (NORM). Color bands parallel to the energy axis represent the gamma ray spectra of volume elements ("voxels") within the object of interest. The volume element size is based on the time resolution of the system so the voxels are approximately 5-cm thick. For example, in the center of the container the voxel is 9 cm × 12 cm × 5 cm. The small voxel size increases the SNR of the system. Another benefit is that the lifetime of certain activation products can be measured and this gives more data upon which to identify the material.

At early development stage, price and size were the drawbacks of using PFNA. However, since the 9/11 attacks the main challenge is the system cost. The cost includes installation and maintenance of this complex system. In 2009, there was a single system working at the George Bush Intercontinental Airport (Strellis, 2009). As of this writing, we are aware of no other installations.

#### **5.3 Pulsed fast / thermal neutron analysis**

The Pulsed Fast/Thermal Neutron Analysis (PFTNA) is a technique used in conjunction with small, portable electronic neutron generators. It was originally developed by George Vourvopoulos, Phillip Womble, and Frederick Schultz and presented in (Womble et al., 1995). Unlike PFNA, which has pulse duration of approximately 2 ns, PFTNA employs pulses with a minimum duration of 5 s. Longer pulse duration significantly reduces cost of PFTNA systems. The PFNA system can be used in a "macro-pulse" mode, in which the neutron beam is turned off for a period of 100 s. This "macro-pulse" mode mimics the PFTNA system's mode.

The advantage of the PFTNA systems is an ability to separate the gamma ray spectrum of inelastic scattering reactions (n,n') from thermal neutron capture (n,) and activation reactions (e.g. (n,p)) gamma-ray spectra. The data acquisition system collects data during the neutron pulse at one memory address and then switches to another memory address to acquire data between pulses. The data collected during the pulse is primarily from (n,n') reactions and the data collected between pulses is primarily from (n, reactions. Often systems are designed to be shut off for a few minutes to collect short-lived activation products such as 16O(n,p) (t1/2 ≈ 16 s). It is a common misconception that the frequency and duration of the neutron pulses is chosen to maximize the data from the (n,n') reactions. In fact these parameters are chosen to maximize the (n,) reactions or more precisely the thermal neutron flux. The neutron pulse frequency determines whether the thermal neutron flux is kept near constant or if it is allowed to diffuse. Applications such as differential die-

Material Analysis Using Characteristic Gamma Rays Induced by Neutrons 31

Another reason to use fast data acquisition electronics in PFTNA systems is high counting rate in the detector when the neutron generator is producing neutrons. Data acquisition rates in the system during this period can exceed 100 kcps. High rates such as these can overwhelm HPGe detectors and analog amplifies. Modern digital electronics can cope with these rates but as rates approach 1 Mcps, scintillation detectors such as sodium iodide or bismuth germanate can be overwhelmed. Count rate limitations force PFTNA system designers to place shielding material between the detector and the neutron generator that

The d-d and d-t fusion reactions take place at low momentum which means that the neutrons are emitted isotropically. These systems typically consume about 100 W during operation. The average neutron outputs for PFTNA systems are 108 n/s for d-t based systems and 106 n/s for d-d based systems. For d-t systems, this leads to radiological concerns to personnel which can be mitigated by distance (approximately 8 meters stand-off

The benefits of PFTNA systems are their smaller size and relatively low cost. However, these features lead to a lower SNR compared to PFNA systems. Some research teams have suggested combining PFTNA method with the associated particle imaging technique to

In associated particle imaging (API), the recoiling residual nucleus, e.g. the alpha particle for d-t reaction, is used to perform time-of-flight and direction selectivity. SNR could be greatly improved for (n,n') gamma ray spectra by measuring gamma ray signals that are emitted only from the selected volume. However, application of this technique would have no effect

The scheme of API technique is shown in Fig.6. The d-t fusion reaction produces alpha particle and 14.1-MeV fast neutron that are emitted in opposite directions due to linear momentum conservation. The segmented alpha detector installed inside the sealed neutron generator tube is used for detection of the -particle event's position and time to "tag" the direction of the 14.1-MeV neutron (Koltick et al., 2009). The geometry of segments of the alpha detector and the neutron's times-of-flight define the geometry of "voxels" for the 3D analysis. ZnO(Ga) detector was used as an alpha detector. It was found that detector's efficiency is about 90% for 3.49-MeV alpha particles. The phosphor coating emits ~15 photoelectrons / alpha; its scintillation emission peaks at 390 nm with ~3.3-ns decay time, allowing up to ~21010 n/s output for 2% tagged solid angle without significant pile-up

The alpha particle detection event and gamma ray detection event are both stamped with the timing signals. The DAQ system is set up to produce the logic signal when both events (the alpha particle and the photon detection) are recorded within a short time interval – the "coincidence window". This logic signal is used to select those gamma ray signals in the energy spectrum that arrive from the tagged voxel. The 14.1-MeV neutron travels in air with the velocity ~5 cm/ns. The 3.49-MeV -particle has the velocity ~1.3 cm/ns. Thus the coincidence window should be in the order of nanoseconds. The quality of the timing

for unshielded operation) or shielding (approximately 30-50 cm shielding).

on the SNR of the (n,) or the time delayed activation gamma ray spectra.

adds to system weight.

improve the SNR.

(Cooper et al., 2003).

**5.4 Associated particle imaging** 

away analysis (DDA), a method of measuring fissile content, allow the thermal neutron flux to completely diffuse and use pulse frequencies less than 1 kHz. PFTNA systems use pulse frequencies greater than 5 kHz to ensure that the thermal neutron flux is nearly constant for the entire period of measurement (Vourvopoulos & Womble, 2001). Our personal experience in this area has shown that frequencies higher than 10 KHz may be desirable as well.

For a d-t or d-d neutron generator, the typical pulsing method is to clamp the so-called source voltage using "clamping circuits". The source voltage causes ionization of the deuterium gas before the ions are accelerated. A consequence of the higher pulse frequency is the shorter pulse duration. This is due to the fact that these clamping circuits operate at a constant duty cycle. The source voltage duration must be a few microseconds (approximately 4 s) for the deuterium gas to reach a pressure where ionization occurs (the "fill time"). With this condition along with the constant duty cycle, the maximum neutron pulse frequency is about 20 kHz since higher frequencies (>25 kHz) will not have a sufficiently long fill time. Thus PFTNA pulses are typically 10 s in duration with a pulsing frequency of 10 kHz. The PFTNA scheme with the neutron pulse's time structure is shown in Fig.5.

Fig. 5. Pulse fast thermal neutron analysis scheme

As discussed earlier, PFTNA method uses two different memory addresses depending whether the neutron generator is on or off. The use of two memory addresses is sometimes described as "ping-ponging" since the data "bounces" between two addresses. A gate signal is sent from the neutron generator to the data acquisition to indicate whether the generator is on or off. The gate signal is usually delayed from the rise of the source voltage by the filltime. Furthermore the gate signal will extend past the fall-time of the source voltage by a few microseconds. This lag is due to the processing time of the data acquisition system. One of the reasons that PFTNA systems use fast data acquisition electronics is to minimize this lag.

Another reason to use fast data acquisition electronics in PFTNA systems is high counting rate in the detector when the neutron generator is producing neutrons. Data acquisition rates in the system during this period can exceed 100 kcps. High rates such as these can overwhelm HPGe detectors and analog amplifies. Modern digital electronics can cope with these rates but as rates approach 1 Mcps, scintillation detectors such as sodium iodide or bismuth germanate can be overwhelmed. Count rate limitations force PFTNA system designers to place shielding material between the detector and the neutron generator that adds to system weight.

The d-d and d-t fusion reactions take place at low momentum which means that the neutrons are emitted isotropically. These systems typically consume about 100 W during operation. The average neutron outputs for PFTNA systems are 108 n/s for d-t based systems and 106 n/s for d-d based systems. For d-t systems, this leads to radiological concerns to personnel which can be mitigated by distance (approximately 8 meters stand-off for unshielded operation) or shielding (approximately 30-50 cm shielding).

The benefits of PFTNA systems are their smaller size and relatively low cost. However, these features lead to a lower SNR compared to PFNA systems. Some research teams have suggested combining PFTNA method with the associated particle imaging technique to improve the SNR.

#### **5.4 Associated particle imaging**

30 Gamma Radiation

away analysis (DDA), a method of measuring fissile content, allow the thermal neutron flux to completely diffuse and use pulse frequencies less than 1 kHz. PFTNA systems use pulse frequencies greater than 5 kHz to ensure that the thermal neutron flux is nearly constant for the entire period of measurement (Vourvopoulos & Womble, 2001). Our personal experience in this area has shown that frequencies higher than 10 KHz may be desirable as

For a d-t or d-d neutron generator, the typical pulsing method is to clamp the so-called source voltage using "clamping circuits". The source voltage causes ionization of the deuterium gas before the ions are accelerated. A consequence of the higher pulse frequency is the shorter pulse duration. This is due to the fact that these clamping circuits operate at a constant duty cycle. The source voltage duration must be a few microseconds (approximately 4 s) for the deuterium gas to reach a pressure where ionization occurs (the "fill time"). With this condition along with the constant duty cycle, the maximum neutron pulse frequency is about 20 kHz since higher frequencies (>25 kHz) will not have a sufficiently long fill time. Thus PFTNA pulses are typically 10 s in duration with a pulsing frequency of 10 kHz. The PFTNA scheme with the neutron pulse's time structure is shown

As discussed earlier, PFTNA method uses two different memory addresses depending whether the neutron generator is on or off. The use of two memory addresses is sometimes described as "ping-ponging" since the data "bounces" between two addresses. A gate signal is sent from the neutron generator to the data acquisition to indicate whether the generator is on or off. The gate signal is usually delayed from the rise of the source voltage by the filltime. Furthermore the gate signal will extend past the fall-time of the source voltage by a few microseconds. This lag is due to the processing time of the data acquisition system. One of the reasons that PFTNA systems use fast data acquisition electronics is to minimize this

well.

in Fig.5.

lag.

Fig. 5. Pulse fast thermal neutron analysis scheme

In associated particle imaging (API), the recoiling residual nucleus, e.g. the alpha particle for d-t reaction, is used to perform time-of-flight and direction selectivity. SNR could be greatly improved for (n,n') gamma ray spectra by measuring gamma ray signals that are emitted only from the selected volume. However, application of this technique would have no effect on the SNR of the (n,) or the time delayed activation gamma ray spectra.

The scheme of API technique is shown in Fig.6. The d-t fusion reaction produces alpha particle and 14.1-MeV fast neutron that are emitted in opposite directions due to linear momentum conservation. The segmented alpha detector installed inside the sealed neutron generator tube is used for detection of the -particle event's position and time to "tag" the direction of the 14.1-MeV neutron (Koltick et al., 2009). The geometry of segments of the alpha detector and the neutron's times-of-flight define the geometry of "voxels" for the 3D analysis. ZnO(Ga) detector was used as an alpha detector. It was found that detector's efficiency is about 90% for 3.49-MeV alpha particles. The phosphor coating emits ~15 photoelectrons / alpha; its scintillation emission peaks at 390 nm with ~3.3-ns decay time, allowing up to ~21010 n/s output for 2% tagged solid angle without significant pile-up (Cooper et al., 2003).

The alpha particle detection event and gamma ray detection event are both stamped with the timing signals. The DAQ system is set up to produce the logic signal when both events (the alpha particle and the photon detection) are recorded within a short time interval – the "coincidence window". This logic signal is used to select those gamma ray signals in the energy spectrum that arrive from the tagged voxel. The 14.1-MeV neutron travels in air with the velocity ~5 cm/ns. The 3.49-MeV -particle has the velocity ~1.3 cm/ns. Thus the coincidence window should be in the order of nanoseconds. The quality of the timing

Material Analysis Using Characteristic Gamma Rays Induced by Neutrons 33

PFTNS was proposed as a primary or secondary screening system for airline security. Designs were proposed which would handle a large number of bags per minute. This would be achieved by having the bags ride a carousel around accelerator. The neutron detectors would be placed in a "wall configuration" and the neutron beam would raster through a number of bags. The National Academy report (NNMAB-482-6, 1999), written in 1999, was extremely critical of the utilization of PFTNS for airport security. However, Overley suggests that detection rates of 93% and false alarm rates of 4% are possible with this

Neutron based material analysis methods generally require a skilled analyst to interpret the gamma ray spectral data collected, and to classify the interrogated object using the elemental parameters extracted from the spectral data. Automatic spectral analysis algorithms and the object's classification algorithms are required for real world applications where access to nuclear spectroscopy expertise is limited, or the autonomous and/or the

The first step in the data analysis process is to extract the sample's elemental information from the measured gamma ray spectra. The spectrum analysis algorithms that are used for that purpose should simultaneously provide quick, accurate, and objective analysis of gamma ray spectra by evaluating the intensities of the characteristic photon peaks. For spectra measured with high resolution detectors such as HPGe, the approach can be based on the peak finding algorithm using the regions of interest (ROIs). Usually, the "blank" spectrum (measured with no sample present) is subtracted from the "sample" spectrum (measured with the sample) before the spectral analysis. It takes into account the signatures of the same elements that are present in surrounding materials, and in the sample. The "nuclear" ROI parameters such as the net peak area in counts /second units are proportional to the number of isotopes in the sample that emitted the fingerprint gamma rays. The "nuclear" parameters may be converted into other appropriate units, if needed, using the elemental calibration library (for example, "chemical" parameters accepted in the coal or the cement analysis industry, etc.). These libraries are created for the system using

The simple ROI-based method may be appropriate for non-complicated spectra with the peaks that are well resolved. For spectra with many closely positioned peaks, or low resolution spectra with overlapping peaks, the peak-shape fitting algorithms are required. The mathematical method of measured spectrum fitting as the linear combination of single element's detector responses, that are measured experimentally, was developed by George Vourvopoulos and Phillip Womble (Vourvopoulos & Womble, 2001). To use this method, one must first measure the response of the low resolution detector to -rays from pure elements. For example, a block of pure graphite is used to determine the detector's elemental response to the carbon -rays. To determine the detector's elemental response to hydrogen, a response is measured from a water sample, and so on. The counts in i-th

channel of the spectrum of a sample S can be represented by the equation:

technique (Overley et al., 2006).

robotic operation is necessary.

**6. Gamma radiation spectral analysis** 

**6.1 Analysis of neutron induced gamma ray spectra** 

calibration measurements using known samples.

signals for both detectors should be very high, without jitter. The width of the coincidence window and the neutron flux are interconnected: the random coincidence rate increases with the higher neutron flux thus limiting neutron yield of the generator.

The API technique was used in such systems as SENNA (Vakhtin et al., 2006), EURITRACK (Perret et al., 2006), and UNCOSS (Eleon et al., 2010).

Fig. 6. Associated particle imaging technique
