Proof of principle of a high-spatial-resolution,
resonant-response γ γ γ γ-ray detector for Gamma
Resonance Absorption in 14N
M. Brandisa,b, M.B. Goldberga, D. Vartskya, E. Friedmanb, I. Kresloc, I. Mardora, V.
Dangendorfd, S. Levia, I. Mora and D. Bara
a 1Nuclear Physics Division, Soreq NRC,
Yavne 81800, Israel
b 2Racah Institute of Physics, The Hebrew University of Jerusalem,
Jerusalem 91904, Israel
c 3Laboratory for High Energy Physics (LHEP), University of Bern,
Bern 3012 Switzerland
d 4Physikalisch-Technisch- Bundesanstalt (PTB),
Braunschweig 38116, Germany
ABSTRACT: The development of a mm-spatial-resolution, resonant-response detector based on a
micrometric glass capillary array filled with liquid scintillator is described. This detector was
developed for Gamma Resonance Absorption (GRA) in 14N. GRA is an automatic-decision
radiographic screening technique that combines high radiation penetration (the probe is a 9.17
MeV gamma ray) with very good sensitivity and specificity to nitrogenous explosives.
Detailed simulation of the detector response to electrons and protons generated by the 9.17 MeV
γ-rays was followed by a proof-of principle experiment, using a mixed γ-ray and neutron source.
Towards this, a prototype capillary detector was assembled, including the associated filling and
readout systems. Simulations and experimental results indeed show that proton tracks are
distinguishable from electron tracks at relevant energies, on the basis of a criterion that
combines track length and light intensity per unit length.
KEYWORDS: Detection of explosives; Inspection with gamma rays.
– 1 –
1. Introduction 1
2. Detector principle of operation
2.1 Track formation by 9.17 MeV γ-rays
2.2 Scintillator light yield and light propagation in capillaries
2.3 Spatial resolution of the track
3. Simulation of the detector response 6
3.1 Simulation detector geometry
3.2 Electron and proton tracks
3.3 Track-based particle discrimination
3.4 Determination of the point of interaction 10
4. Experimental 10
4.1 Capillary array unit
4.2 Liquid scintillator
4.3 Description of the detector
4.4 Filling procedure
4.5 Light readout system
4.6 Visualization of particle tracks
5. Conclusions 15
GRA is an element-specific, automatic-decision, radiographic-transmission imaging
technique, capable of revealing the spatial density distribution of the element in question within
inspected objects first proposed by Soreq NRC (SNRC) as a means of detecting explosives in
airline baggage via nitrogen-specific radiographic imaging [1,2]. High nitrogen density is
characteristic of most explosives but not of commonly transported benign materials; the GRA
method relies on this property to distinguish between them. The GRA method is uniquely well-
suited to inspecting large, massive objects such as aviation/marine containers, heavy vehicles
and rolling stock, the resonant probe being a high energy γ-ray of 9.17 MeV.
The method was successfully taken under FAA sponsorship through several experimental
feasibility rounds, including a proof-of-principle laboratory test launched in collaboration with
Los Alamos National Laboratory (LANL) (1989) , a blind test on aviation baggage
aggregates (1993)  and a preliminary run on an aviation container (1998) .
For detection of nitrogenous explosives via GRA, the inspected object is scanned by a
beam of γ-rays, a fraction of which represents on-resonance flux at 9.172 MeV that is capable of
being absorbed resonantly by 14N nuclei. Thus, in addition to the familiar non-resonant atomic
– 2 –
processes that attenuate the transmitted flux, such as Compton scattering and pair production,
these γ-rays will also undergo a nuclear resonance component of attenuation that is proportional
to the density of nitrogen in the line-of-sight between radiation source and detector. By
measuring the transmitted γ-rays flux at energies on and off resonance and normalizing
appropriately, both the non-resonant and net resonant (total-nitrogen) components of gamma
attenuation can be extracted.
The optimal γ-ray source for the nitrogen GRA method is the de-excitation spectrum of the
excited 14N* 9.17 MeV level following 1.746 MeV proton capture via the reaction 13C(p,γ)14N.
Since the lifetime of the 9.17 MeV level (5.1x10-18 s) is very short compared to ion stopping
times (typically ~1x10-12 s) the emission of the γ-ray occurs in-flight following the recoil of the
excited 14N nucleus, resulting in Doppler-shifting of the γ-ray. At the resonant angle θR = 80.66o
with respect to the proton beam, the nuclear recoil energy losses that occur during emission and
absorption of the γ-ray by the 14N nucleus are precisely compensated by the Doppler-shifted
energy component. Since the 9.17 MeV γ-ray yield into 4π is ~6x10-9 γ-rays per proton, it is
estimated that an operational explosives detection system with realistic throughput based on this
reaction will require a 3-5 mA proton accelerator.
For the kinematic reasons mentioned above the resonant flux is confined to a narrow cone
around polar angle θR = 80.66o relative to the proton beam forming a slightly curved fan beam.
A linear detector array is positioned on the perimeter of this cone as shown in Fig. 1. The
inspected object passes through the fan beam and a radiographic image is formed slice by slice..
Fig. 1: Schematic of GRA imaging system showing the 9.17 MeV resonant gamma-ray beam cone.
The detector array is positioned on the perimeter of the resonant cone. The inspected object is
being moved in a scanning motion between the source and the detector.
The width ∆θ of the resonant cone has been determined by several investigators [8-10]. All
experiments consistently show that, the angular aperture size required to include most of the
resonant photons is ∆θ ~0.75o. This aperture corresponds to an energy spread of 520 eV, which
is a factor of ~4 larger than the total intrinsic width of the level ΓT=122 eV . The
consequence of the emission-line broadening is that only a small intensity fraction within the
cone (of order ~0.23) is on-resonance radiation. Thus, measuring the total intensity of the
emission line with a conventional-response detector such as NaI, BGO or Ge, will result in
reduced nitrogen contrast sensitivity by this fraction .
Soreq NRC has hitherto circumvented this problem by developing nitrogen-rich resonant-
response detectors, that are selectively sensitive to photons in the 9.17 MeV region [2,11]. With
– 3 –
such detectors the resonant flux is sampled, on an event-by-event basis, by counting 1.5 MeV
photo-protons internally-produced via the resonant absorption reaction 14N(γ,p), the inverse
reaction to 13C p-capture:
γ (hν 9.17 MeV) + 14Ng.s. → 13Cg.s (0.12 MeV) + p (1.5 MeV)
The separation of photo-protons from non-resonant events (electrons and positrons) is
realized in currently existing detectors  through Pulse Shape Discrimination (PSD) [11,12], a
method that utilizes the difference in scintillation light decay time to distinguish electron and
proton events. The latter give rise to a somewhat longer signal, spread over a wider time range.
The large angular width of the flux cone required to include most of the resonant photons
limits the achievable GRA spatial resolution. Thus the current Soreq resonant detectors shown
in Fig. 2 are composed of an Aluminum vessel of 20×20×240 mm3 in dimensions with a
rectangular glass window, filled with a special-purpose, nitrogen-rich liquid scintillator.
Fig. 2:. Left: schematic representation of the detector setup. Right: Components of the
nitrogen-rich, liquid scintillator resonant detector
The vessel is coupled via a parabolic Perspex light-guide to a 2” diameter photomultiplier
tube. The 20x20 mm2 detector face is presented normal to the γ-ray beam, forming a pixel of
these dimensions. The spatial resolution of the GRA system with the present detectors is 5-15
mm, depending on the location of the voxel of interest within the source-detector space.
Although Soreq has demonstrated that it is possible to detect thin sheet explosives in cargo
using such detectors , it would be clearly beneficial to improve the spatial resolution.
In November 2005, within the framework of a DHS/TSA Risk Reduction program Soreq
NRC launched an R&D effort aimed at developing novel resonant response detectors with sub-
mm resolution, based on imaging the photo-proton and electron-induced tracks in a glass
capillary array filled with liquid scintillator .
In parallel, an independent approach to high-spatial-resolution resonant detector is
developed by the LHEP group, University of Bern, using a Time Projection Chamber filled with
a mixture of liquefied Argon and Nitrogen 
This paper summarizes the work performed at Soreq NRC so far on the capillary tracking
detector and is organized as follows: Sect. 2 describes the concept and the physical processes
taking place within the detector. In Sect. 3 a simulation of detector response to 9.17 MeV γ-rays
and 1.5 MeV protons is performed. Sect. 4 describes the apparatus, experimental procedures and
the preliminary experimental results obtained with the first prototype of such a detector. Sect. 5
provides a short summary.
– 4 –
2. Detector principle of operation
The concept of capillary detectors and most of the hardware used in this work are based
on the R&D effort performed in CERN by CHORUS and RD46 collaborations [18-20]. Fig. 3
describes the detector concept schematically.
Fig. 3 Schematic representation of the detector
The GRA methodology requires that any resonant-response detector, and in particular such
a capillary array, should permit separation of resonant interactions from non-resonant
interactions on an event-by-event basis [1-4,11].
To this end, the detector consists of a coherent bundle capillary array filled with nitrogen
rich liquid scintillator. γ-rays incident on the detector can create photo-protons if they are at the
resonant energy. At off-resonance energies, they can create only electrons or electron/positron
pairs. The light created in the detector by the particles is transported into an intensified CCD
camera which images the projection of particle tracks . Since equal amounts of scintillation
light travel along the capillaries in each direction, a good quality reflecting mirror is coupled to
one end of the capillary bundle, in order to reflect the light back towards the CCD, thus nearly
doubling the amount of light.
Discrimination between particle types produced in the capillary array detector is based on
the difference in track projection length and light intensity per unit track length. Light charged
particles, such as electrons and positrons, create tracks that are centimeters long while 1.5 MeV
protons give rise to much shorter tracks, only ~50 µm long. However, protons leave much more
light per unit length along their tracks than electrons. Thus, proton events will give rise to strong
bursts of light over a short trajectory of typically a single capillary, which are very distinct from
the long, fainter tracks of electrons. Analysis of track projection length and the intensity of light
per capillary should permit reliable proton event discrimination with sub-mm spatial resolution.
2.1 Track formation by 9.17 MeV γ γ γ γ-rays
The interactions via which the energy is transferred to the scintillator depend on the
incident γ-ray energy. Since the special-purpose scintillator inside the capillaries is itself
nitrogen rich, resonant γ-rays can undergo the nuclear resonant reaction, producing 1.5 MeV
photoprotons and 13C atoms. The proton path in the scintillator is essentially straight (apart from
the extremely rare event of a nuclear collision, when large angle scattering occurs) as the
momentum transfer in these ion-ion collisions is relatively small.
Non-resonant γ-ray interactions at ~10 MeV, in organic scintillators are predominantly
Compton scattering and, to a lesser extent, pair production. In both interactions the secondary
charged particles created are free leptons at various energies. An electron passing through
matter dissipates its energy in excitation and ionization, and is scattered in collisions with
electrons and nuclei. In its interactions with the matter, the electron may undergo appreciable
– 5 –
deflections, so that its path through the scintillator (at energies up to ~10 MeV) is not straight
The primary proton or electron can transfer energy directly to molecule excitation or in a
more complicated process, by ionizing a molecule and creating a secondary electron of lower
energy that will transfer its energy to molecular excitation. Several generations of such electrons
(δ-electrons) can be produced and contribute to the fluorescent excitation process, by which
light is emitted in the scintillator.
These secondary and δ-electrons cause track branching and broadening. The branching is
more common in primary electron tracks than in proton tracks, whereas the widening is
common to both. However, in terms of the measurable quantities, the broadening has an effect
on track shape only if, on average, it is larger than a capillary radius.
2.2 Scintillator light yield and light propagation in capillaries
The light emission in the scintillator is followed by propagation of this light through the
capillaries. The most important figure-of-merit in the capillary detector scintillator is the light
output at the end of each capillary. It is dependent on the following parameters: scintillation
light yield, the refractive indices of the scintillator and the capillary material, as well as the
optical transparency of the scintillator liquid.
The refractive index of the liquid scintillator should be as high as possible relative to that
of the capillary material in order for the critical angle of total internal reflection to be as large as
possible, allowing a greater portion of the isotropically emitted scintillation light to reach the
end of the capillary. This portion, called the trapping fraction, is only a small percentage of the
total light yield. For an ideal circular fibre it is equal to 1/2[1-(nC/nL)2] , where nC and nL
are the refractive indices of the capillary wall and liquid, respectively. In our experiment we
used glass capillaries and a liquid scintillator with refractive indices of 1.49 and 1.62
respectively. In such a configuration the trapping fraction is expected to be up to 7.7% in each
direction along the capillary. In practice, capillary imperfections and partial filling may reduce
the trapping fraction to about 4-5% . If a good quality reflecting mirror is attached to one
end of the capillary array we expect about 10% of the overall created light to be available for the
2.3 Spatial resolution of the track
Instrumentally, the resolution with which the track can be observed is determined by the
diameter of the constituent capillaries. The narrowest capillaries available are 10 µm in
diameter; that is therefore the upper limit on the resolution. The glass partition between
capillaries needs to be taken into account when considering the spatial resolution. in the
experiment conducted at the Laboratory for High-Energy Physics (LHEP), University of Bern,
it had a thickness of ~11 µm for the 20 µm capillary diameter array.
In addition there are several physical factors that can cause degradation in spatial
resolution: 1) the track properties of the secondary ionizing particles. For electrons, under
certain conditions, insufficient energy may be deposited in some of the capillaries to be
detected, thus creating an intermittent track. 2) the amount of light traversing through the
capillary walls into adjacent capillaries will cause pixel cross-talk. In order to minimize this
effect, Extra-Mural-Absorbers (EMU) should be placed between the capillaries and preferably
integrated into the capillary matrix. 3) in practice, the spatial resolution is also determined by
the optical readout system.
– 6 –
3. Simulation of the detector response
The simulations were performed using Geant4 simulation Toolkit, version 4.9.2,
distributed by CERN [22,23].
At 9.17 MeV γ-ray energy, the dominant non-resonant interaction process is the Compton
Effect, pair production being weak (as the scintillator is low-Z) and photoelectric contributions
essentially negligible. All these interactions are included in the simulation application written
here. Secondary and tertiary electron interactions in this simulation include: ionization (the
principal process inside the scintillator), multiple scattering and Bremsstrahlung radiation.
Processes undergone by positrons include annihilation as well.
3.1 Simulation detector geometry
Sssc In order to simulate a capillary array, equi-distant cylinder shaped "incisions" of
scintillator-like material were incorporated in a square block of glass-like material, 25×25×350
mm3 in dimensions. Two capillary arrays of different basic cell dimensions were used in the
simulation runs (see Table I).
Table I: Details of different simulated array geometries
# Diameter (µm)
The first variant simulates the capillary array that was available from the CHORUS project
(see Sect. 5). Smaller capillaries offer superior tracking resolution, but the amount of light
created by the particle per capillary scales with capillary diameter and may thus be insufficient
for reliable track identification. This is due to the fact that, below a certain diameter, electrons
may deposit insufficient energy in some of the capillaries, thus giving rise to a discontinuous
track. The second variant employs capillaries with larger diameters and thinner walls. It
represents a much better tradeoff between full tracking capabilities and high spatial resolution.
The simulation was performed in two parts: 1) a pencil beam of 9.17 MeV γ-rays
impinged on the central capillary and the non-resonant reaction products (electrons and
positrons) were tracked; 2) 1.5 MeV proton source which was disc-shaped and located halfway
along the central capillary was simulated. The simulated protons velocity direction was
distributed isotropically. This simulates quite well the photo-protons created by the 9.17 MeV
γ-rays that interact resonantly with the nitrogen in the scintillator via the photo-nuclear reaction.
Based on the cross-sections for creation of electrons and protons by a monoenergetic 9.17 MeV
γ-ray beam  the total number of produced electrons is calculated to be 11 times larger than
that of protons, for a 35 cm long detector containing 15% nitrogen by weight. This factor has
been taken into account in our simulations when evaluating the particle discrimination
For each event the length of track projection (in number of capillaries) is determined and
the total energy deposited in the track is calculated. The energy is then converted into number of
light photons, taking into account the difference in light yield between electrons and protons,
the non linear behaviour of the light yield with proton energy  and the fact that the light
output of our 15%-nitrogen rich scintillator is reduced by a factor of ~1.6 relatively to a nitrogen
Basic Cell Wall width
806 x 806
454 x 454
– 7 –
3.2 Electron and proton tracks
Example projections of electron tracks in a 20 µm capillary array (Variant I) are presented
in Fig. 4. Simulated 9.17 MeV γ-rays impinge on the center of a 25×25 mm2 detector. The
tracks are color coded as following: red - electrons, blue - positrons, green - γ-rays or X-rays.
Fig. 4a shows a case of a Compton scattering followed by production of Bremsstrahlung
radiation. Fig. 4b shows a case of pair production followed by an annihilation of the positron.
Fig. 4c shows a more complicated case of multiple Compton scattering, Bremsstrahlung
radiation and the photoelectric effect.
Fig. 4: Three simulated track projections of electrons and positrons created by 9.17 MeV γ γ γ γ-rays in a
20 µ µ µ µm capillary (Variant I) array, 25×25 mm2 in dimensions. The red tracks represent electrons,
the blue tracks represent positrons
Fig. 5 shows a magnified simulated track projection created by a 9.17 MeV -ray
secondary electron (Fig. 5(a)) and a 1.5 MeV proton (Fig. 5(b)) in the capillary array of Variant
#1. The length of the proton projection is two capillaries. One can observe the effect of the
Bragg distribution, in that the proton deposits most of its energy at the end of the track.
The length of track projection is determined by summing the capillaries along the particle
track. Measured track lengths could thus exceed real ones by as much as 1 capillary.
Fig. 5: Magnified view of simulated tracks in 20 µ µ µ µm (Variant I) capillary array. The color code on
the right shows the number of photons/capillary. (a) Part of electron track. (b) 1.5 MeV proton
track. Fig 5a and 5b do not have the same magnification.
Fig. 6 shows the distribution of track length projections (in number of capillaries) of
electrons/positrons for the 20 µm diameter capillary. Electron tracks show a wide distribution of
lengths with a peak at around 325 capillaries
– 8 –
The protons deposit their energy in up to two capillaries. About 70% of them do so in a
single capillary and the remainder - in 2 capillaries.
Fig. 6: Distribution of track length projections (in number of capillaries) for electrons (red) and
protons (blue) in Variant I capillary array.
Fig. 7: Distribution of track length projections (in number of capillaries) for electrons (red) and
protons (blue) in Variant II capillary array.
The Variant II capillary array (50 µm in diameter) shows a similar electron and proton
track length (in number of capillaries) distributions (as can be seen in Fig. 7).
Due to the fact that the capillaries are larger, the fraction of a single capillary proton track
is slightly higher here.
3.3 Track-based particle discrimination
The discrimination between electrons and protons is based on the projection of track length
(number of capillaries in track) and the average amount of light per capillary generated by each
species. The average amount of light per capillary is derived by calculating the total amount of
light in the track and dividing by the number of capillaries in the track projection. In order to
calculate the amount of light/capillary that reaches the photocathode of the electro-optical
readout we must take into account the light yield of the nitrogen-rich (15% N by weight)
scintillator which is ~6100 photons/MeV electron, the difference in light yield between
electrons and protons, the non linear behaviour of the light yield with proton energy and the
trapping fraction of our capillaries (~10%). Fig. 8 shows the distribution of the average
– 9 –
light/capillary on the optical readout's input in Variant I capillaries for electrons (red) and
protons (blue). As expected the electrons light/capillary is much lower than for protons.
Fig. 8: Distribution of average light/capillary for 20 µ µ µ µm diameter capillary array. Red-electrons.
Blue-protons, Green- electrons with track length of up to 2 capillaries.
The electron distribution peaks at 3 photons/capillary and drops to a negligible value at
around 13 photons/capillary. Only 0.4% of the total number of created electrons are above this
value, while 90% of protons are accounted for. One can observe a peak at 160 photons/capillary
in the proton distribution. These are events where the proton deposited all its energy in a single
capillary. However, there is an additional piece of information that can be invoked for
discrimination, namely, the fact that the proton can produce a track length of only up to two
capillaries. The green curve shows the fraction of electrons with a track length of up to 2-
capillaries, which could be interpreted as protons. These events can be considered as
background. When applying this condition together with the 13 photons/capillary threshold, the
proton to electron counts ratio is 53. This is a very good signal to background ratio, about a
factor of 2 better than that obtained with our present resonant detectors, based on PSD.
Fig. 9: Distribution of average light/capillary for 50 µ µ µ µm diameter capillary array. Red - electrons.
Blue - protons, Green - electrons with track length of up to 2 capillaries.
Fig. 9 shows the distribution of the average light/capillary in Variant II capillary array. In
this variant, the distribution is shifted towards larger values of light/capillary for both electrons
and protons. Here, the electron distribution peaks at about 7 photons/capillary. With a threshold
– 10 –
of about 31 photons/capillary the fraction of detected events is 90% and 0.4% for protons and
electrons respectively. When the 2-capillaries condition is applied in addition to this threshold
the ratio of proton to electron number of events is 56.
In a practical light detector with a quantum efficiency of about 15%, some electrons will
form weak and intermittent tracks and will not be visible. Using a binomial distribution to
calculate the probability of generating one or more photoelectrons at the photocathode for the
light emitted from a capillary we estimate that 38% and 95% of the electron and proton tracks
respectively for Variant I capillary array will be observed. Due to the larger amount of light
emitted from the capillaries in Variant II array the corresponding fractions are 70% and 98%.
3.4 Determination of the point of interaction
In order to determine the pure resonant attenuation due to nitrogen in the inspected object
in each capillary both on-resonance (protons) and off-resonance (electrons) events must be
determined with the same position resolution. Thus, it is important to locate the point of
interaction of the 9.17 MeV γ-ray in the detector for both types of events with similar precision.
The proton track is short and its point of creation is determined with a precision of a single
capillary. The Bragg peak, visible on the proton tracks can help locate the end of the track.
Thus, the on-resonance point of interaction can be located with a precision of about 50 µm.
On the other hand, electrons have long and complex tracks without a visible Bragg
distribution. Moreover, the start and end of a track is not so obvious. Several criteria for making
a decision regarding the point of non-resonant interaction of the γ-ray have been devised:
• Complex tracks or events that produce several non-adjacent tracks are rejected.
• For electrons that escape the detector it is possible to locate the start position.
• The amount of energy deposited at the point of interaction is usually greater than at
other positions along the track, due to the fact that the primary electron is strongly forward
peaked and is thus moving along the length of the capillary. This phenomenon becomes more
pronounced as the diameter of capillary increases.
By calculating the center-of-gravity for the simulated track deposited energy we obtain an
estimate of point-of interaction.
We tested these criteria on 100 simulated electron events. Of these, 11% were rejected.
Using the center-of-gravity for the remaining events, we have been successful in determining
the interaction location of 85% to within a distance of 3 mm from the true point of interaction.
In conclusion, with the capillary array detector the GRA nitrogen image resolution has the
potential of being 3×3 mm2. As the number of electrons is much higher than that of the protons,
the 11% rejection will not affect significantly the statistical precision of the nitrogen
The experimental setup used in this work is based on the expertise and equipment
derived from the CHORUS experiment and RD46 program at CERN [18-20]. The expertise
accumulated at that time was recently recovered and reactivated at LHEP, University of Bern. In
the following section we shall describe the various parts of the detector used for the proof-of
– 11 –
4.1 Capillary array unit
The capillary array unit is composed of ~20 µm diameter capillaries with wall thickness of
~11 µm, manufactured by Schott Fiber-Optics. The basic capillary array unit used here has a
rectangular cross-sectional dimensions of 5×5 mm2 as shown on Fig. 10 and was assembled into
a larger area array. It was manufactured in lengths of ~1 meter and cut for our application to ~11
Fig. 10: A capillary array unit used in the experiment
The capillary matrix is made of Schott Technical Glass Type 8250, and incorporates extra
mural absorbers (EMA) made to absorb light in order to prevent light cross-talk. The matrix
glass has relatively low refractive index of 1.49. The capillaries are arranged in hexagonally
shaped aggregates, also named “multies”. These multies are then brought together in a beehive
structure to create larger arrays of millimetric size. Fig. 11 shows a Scanning Electron
Microscop (SEM) picture of the capillary array.
Fig. 11: SEM images of capillary array used in the experiment. Scale: left-10 µm, right-
4.2 Liquid scintillator
The principal characteristics required of the new scintillator, in addition to high light yield,
are: high refractive index, optical transparency, quench resistance and a strong solvent, capable
of dissolving large concentrations of nitrogen-rich additives. These are very different properties
from those of the scintillator currently used in the GRA application detectors.
The development and optimization of the special-purpose, nitrogen rich, high-refractive-
index scintillator involves a tradeoff between the different requirements in order to achieve the
composition that is most suitable for the application. This work will be described in detail in a
For the purpose of experimental proof-of-principle a non-nitrogenous scintillator was
prepared. It consists of 1-Methylnaphtalene (1MN), as a solvent, doped with 5g/l and 3/l of 2,5-
– 12 –
Di-phenyl oxazole (PPO) and 1,4-bis-(2-methylstyryl)-benzene (bis-MSB) as primary and
secondary phosphors respectively. Its properties are: refractive index-1.62, density-1.001 g/cc,
light yield-77% of anthracene, peak emission wavelength-420 nm.
As mentioned in section 2.2 if a good quality mirror is used on the edge of the capillary
array opposite to the optical readout, up to ~10% of the created light is expected to exit the
capillary at the latter edge. In our experiment a reflecting mirror was not used.
4.3 Description of the detector
In order to construct a GRA detector, 4 square capillary array units depicted in Fig. 10
were glued together to form a larger 10 x 10 mm2 assembly. This capillary assembly was placed
inside an aluminum tube, and the space between them filled with silicon RTV elastosil E41 to
prevent its filling with liquid scintillator. This material was chosen for its durability when
exposed to the liquid scintillator.
Fig. 12 (left) illustrates a schematic diagram of the detector structure. The array is pressed
against a Fiber Optic Plate (FOP) on the bottom side and a viewing glass window from the top,
using aluminum flanges. The contact between the flanges and the tube is sealed using silicon
gaskets. The filling inlet and outlet tubing are incorporated into the upper and lower flanges.
During the filling procedure the upper and lower windows are shifted a short distance away
from the array, to allow entry of the liquid into the capillaries. After the filling this gap is closed
by tightening the plastic rod screws, such that both windows are in close contact with the
capillary array. Fig. 12 (right) displays the assembled detector.
Fig. 12: Detector schematic (left) and photograph (right). (1) liquid inlet/outlet with copper feed-
through and plastic tubes; (2) glass window; (3) Aluminium tube housing for the capillary array,
spaces filled with silicon RTV; (4) FOP; (5) aluminium flanges; (6) tightening plastic screw rods; (7)
The connection to the filling system is via a copper feed-through and the plastic tubes seen
in Fig. 12 - right which can connect to the filling system and to each other, creating a stand-
alone detector cell.
4.4 Filling procedure
The filling of the capillaries is similar to the one used by the RD46 collaboration . The
filling system is described schematically on Fig. 13 (left). It consists of 2 glass reservoirs, a
vacuum pump, pressurized Ar bottle and a series of valves. The system is first flushed with
argon gas that eliminates the oxygen in the system and in the liquid scintillator stored in one of
the vessels. The liquid scintillator is then introduced into the detector by applying argon
– 13 –
pressure and the capillary array is repeatedly flushed with the liquid scintillator, which is shunt
circularly between the two blue reservoirs. It takes several hours until all capillaries are filled.
Fig 13: Schematic diagram (left) and photo of the filling system
The filling procedure is monitored by illuminating one end of the detector and observing
the light emerging from the other end. For empty capillaries no light is internally reflected and
they appear darker. Therefore, looking at a light source through the glass window of the
capillary array detector reveals whether the filling process was successfully completed. Fig. 14
shows a photograph taken through the looking window of the filled detector.
Fig. 14: A successfully filled array
Although this crude method cannot detect individual empty capillaries, it does give a good
indication if the filling process was adequate for the requirements of this preliminary
4.5 Light readout system
The light readout was refurbished from the CHORUS system . Fig. 15 is a schematic
diagram of the readout system.
Fig. 15: Schematic diagram of the readout system.
– 14 –
The detector FOP was coupled to the entrance fiber-optics window of an electrostatically
focused Image Intensifier (I-I) having a demagnification of m1=0.62 and a green K-67
phosphor. The phosphor of the first I-I is viewed by an electron-bombarded-CCD (EBCCD)
with an electrode gating, which combines the functions of an electrostatically-focused I-I and a
conventional CCD. It is essentially an I-I with the phosphor screen replaced by a back-thinned
CCD. The demagnification of the EBCCD is m2=0.83 and it consists of 1024×1024 pixels, each
with dimensions of 13×13 µm2. The EBCCD output is displayed on a computer screen using a
ROOT software interface. Both the I-I and the EBCCD were produced by Geosphera. The
readout system was originally designed to be used with green-emitting (490-500 nm) scintillator
so, for the blue-emitting scintillator employed here, its efficiency is lower by ~20%.
The detector was positioned between two plastic scintillator plates that served as cosmic-
ray rejection system. Simultaneous signals in both plates provided an anticoincidence event veto
which blocked the readout system.
4.6 Visualization of particle tracks
The experimental observation of proton and electron tracks was performed using a 239Pu-
Be neutron source. The neutrons emitted from 239Pu-Be source have energy range of 0.2-10
MeV . The average neutron energy is about 4.2 MeV . The predominant γ-ray energy is
4.43 MeV, produced during the decay of the first excited state of 12C, populated in the
During the experiment the readout system recorded the image on the capillary array face.
The images were taken automatically with pre-set exposure gate times and intervals between
exposures. The gates were in the range of 300 µs -100 ms. The lower limit on gate time was
determined by the first I-I, the decay time of its phosphor being ~200 µs. The threshold for
taking an event image was set so that the maximum numbers of tracks are visible while most of
the noise (predominately single-pixel) is eliminated. Eventually, the optimal conditions were
found, and the images were recorded with 1 ms gate and 500 mV threshold.
Fig. 16 shows the tracks obtained using the 239Pu-Be source. As the FOP was not in good
physical contact with the back end of the array some of the liquid scintillator migrated into the
gap. Thus, the light created in a single capillary expands as it traverses this scintillator, causing
a "blooming" effect". This has the effect of delocalizing the pixel information and tends to blur
the track images.
Fig. 16: (a),(b),(c),(d) - tracks of recoil-protons produced by neutrons; (d) simultaneous events: high
energy Compton electron and a recoil-proton.
As seen in Fig. 16 (a)-(d), recoil-protons produced by neutrons give rise to bright, short
tracks. In all the proton events displayed, there are indications of the Bragg peak, which
manifests itself as a bright agglomeration of light near the end of the particle range.
– 15 –
This paper introduces a novel resonant detector for the GRA method with mm spatial
resolution based on a micrometric glass capillary array filled with liquid scintillator. A detailed
simulation of the response of the detector to electrons and protons generated by the 9.17 MeV γ-
rays was followed by a proof-of principle experiment at LHEP, University of Bern, using a
mixed γ-ray and neutron source. Towards this, a prototype capillary detector was assembled,
including the associated filling and readout systems. Simulations and experimental results show
that proton tracks are distinguishable from electron tracks at relevant energies, on the basis of a
criterion that combines track length and light intensity per unit length.
The agreement between simulation and experiment is the crucial necessary (and, indeed,
almost sufficient) condition for demonstrating the proof-of-principle in question. The sole
remaining knowledge gap pertains to the optimization of the dedicated nitrogen-rich cocktail, an
issue that all the commercial companies active in the field (as well as this group) have always
solved satisfactorily and without undue effort. The development of a suitable high-refractive
index, nitrogen-rich scintillator will be described in our next publication.
For the purpose of particle discrimination it is preferable to use larger capillaries which
give rise to brighter electron tracks that could be analyzed with higher accuracy. It appears that
a 50 µm diameter capillary with 5 µm walls represents a good compromise between resolution
and light output.
As the proton track is short the on-resonance point of interaction can be located with a
precision of about 50 µm, however the point of interaction of the non-resonant events can only
be located with a resolution of 3×3 mm2. This dictates the resolution of the nitrogen image.
Thus, with the capillary array detector the nitrogen image resolution will be 3×3 mm2. This
compares very favorably with the 20×20 mm2 resolution obtained with our present resonant
In an operational GRA system operating with a 5 mA proton beam, the total event rate
(electrons and protons) in a 20×20×240 mm3 resonant detector positioned at about 250 cm from
the 13C target, is expected to be about 500 events/s. Thus in order to obtain an image with not
more than 2-3 particle tracks the exposure time of the readout system should be not more than
about 6 ms/frame. It is then necessary for the image to be analysed or transferred, such that an
automated track recognition procedure can be performed without introducing a significant dead-
time or delay.
In summary, the advantages of the proposed detector are: Excellent separation between
electrons and protons, improved spatial resolution (3×3 mm2), possibility of constructing large
area array detectors.
This work was funded by DHS Grant 071905-001. The authors would like to thank Drs Ronald
Krauss and Curtis Bell from the Transportation Security Laboratory, Science and Technology
Directorate, U.S. Department of Homeland Security, for their continuing interest in the GRA
Thanks are due to Drs. Guy Van Beek, Pierre Vilain and Gaston Wilquet of IIHE, the Inter-
University Institute for High Energy Physics, ULB-VUB, Brussels, for many discussions and
recommendations that proved critical for this research.
– 16 –
Our gratitude is also due to the X-ray microscopy group in Soreq, and especially to Dr. Irina
Gouzman for taking the SEM images.
 D. Vartsky and M.B. Goldberg, An Explosives Detection system for Passenger Baggage Based on
Nuclear Resonance Absorption in 14N, A proposal to the FAA, (March 1986).
 D. Vartsky, M.B. Goldberg, G. Engler, et al. Method and System for Detection of Nitrogenous
Explosives by Using Nuclear Resonance Absorption, Israeli Patent No. 86826, June 1988, U.S.
Patent No. 4,941,162, June 1989.
 D. Vartsky et al., A Detection System for Explosives in Passenger Baggage based on Nuclear
Resonance Absorption in 14N, Progress Report IV to FAA (June 1991).
 D. Vartsky et al., The Development of a Prototype High-Explosives Detection System based on
Nuclear Resonance Absorption, Summary Report to the FAA (December 1993).
 D. Vartsky et al., Advanced Feasibility Study on Explosives Detection in Aviation Cargo via
Nuclear Resonance Absorption (NRA), Progress Report to FAA (June 1998 – March 1999)
 A.S. Kuznetsov, Yu.I. Belchenko, A.V. Burdakov et al. Nucl. Instr. and Meth. in Phys. Res. A 606
 M.B. Goldberg, V. Dangendorf, D. Vartsky et al. A dual purpose Ion-Accelerator for Nuclear-
Reaction-Based Explosives and SNM-Detection in Massive Cargo, International Topical Meeting on
Nuclear Research Applications and Utilization of Accelerators, AccApp ’09 Vienna, 4-8 May 2009,
 S.S. Hanna and L. Meyer-Schützmeister, Phys. Rev. 115, pp. 986-989, 1959
 W. Biesiot and P.B. Smith, Phys. Rev., C24, pp. 2443-2457, 1981
 D. Vartsky, M.B. Goldberg, G. Engler et al. , Nucl. Phys., A505, pp. 328-336, 1989
 D. Vartsky, M.B. Goldberg et al., Detectors for the gamma-ray resonant absorption method of
explosives detection in cargo: A comparative study Proc. SPIE Int. Soc. Opt. Eng 5198 (2004) 243,
 J.B. Birks, The theory and Practice of scintillation counting, Pergamon Press 1964
 R.E. Morgado, G.J. Arnone, C.C. Capiello, et al. Prototype explosives detection system based on
nuclear resonace absorption in nitrogen, SPIE Conference on Substance Detection Systems, Proc
SPIE Vol. 2092, pp. 492-502, Innsbruck, Austria, October 1993.
 Schmor P.W., Buchmann L.R. and Rogers J.G., Nitrogen Detector and Method of Detecting
Explosives, US Patent No. 5,282,235, January 1994
 J.G. Rogers, C. Moisan, A. Altman, et al. IEEE Trans. on Nucl. Science, Vol. 44, (1997) 994-1000
 I. Kreslo et al, Study of ionization signals in a liquid Argon TPC doped with Nitrogen JINST
3:P10002,2008. Aug 2008. 18pp, arXiv:0808.3939
 M. Brandis, Particle Identification via Track Imaging in Liquid-Scintillatior-Core Capillary Arrays,
M.Sc thesis at the Hebrew university of Jerusalem, (May 2008) Published at JINST thesis archive
– 17 – Download full-text
 P. Annis et al., Nucl. Instr. & Meth. A386 (1997) 72
 P. Annis et al., Nucl. Phys. B 61B (1998) 390
 P. Annis et al., Nucl. Instr. & Meth. A449 (2000) 60
 C.P. Achenbach and J.H. Cobb, J. Opt. A: Pure Appl. Opt. 5 (2003) 239
 Geant4 simulation toolkit user's guide: For application developers,
 Geant4 Physics reference guide, http://geant4.web.cern.ch/geant4/
 G.V. O’Reilly, N.R. Kolb and R.E. Pywell,. Nucl. Instr. & Meth. A 368 (1996) 745
 H.R.Vega-Carrillo, E.Manzanares-Acuna, A.M. Becerra-Ferreiro, A. Carrilo-Nunez, Appl.
Rad.&Isotopes 57 (2002) 167.
 L. Stewart, Phys.Rev. 98,(1955), 740