Fast Neutron Scattering Analysis - University of Cape Town

Contraband detection,
using neutrons
Andy Buffler
Department of Physics
University of Cape Town

The non-intrusive characterization of bulk materials is required in
many industrial and commercial contexts including:
• mining
• food quality control
• containerized waste management
Very often an analysis of the elemental composition of the material
is needed, sometimes to form an image of the distribution of one or
more elements comprising the material.
In particular, there is a growing need for improved techniques to
detect explosives and narcotics hidden in airline luggage and other
containers, and landmines buried in the waters or soil of the earth.

Bombing of Pan Am Flight 103 (Boeing 747-121)
at Lockerbie, Scotland on 21 December 1988

Terror attacks on
World Trade Centre
and Pentagon
on 11 September 2001

Common approaches to detect hidden contraband include:
X-rays: Well established and highly penetrating but are not sensitive
to the light elements H , C, N and O, and object identification based
mainly on visual imaging.
Dogs: Extremely high success rate (esp. in South Africa) but their use
is disadvantaged by large workloads.
There is a need for alternative technologies which offer high:
• penetrability (to interrogate large containers)
• sensitivity (for low false alarm rates)
• specificity (to detect hidden contraband within a bulk
sample of primarily innocuous content)
Nuclear-based approaches offer all of these features.

Ammonium Nitrate
Composition B
Composition 4 (C-4)
Dynamite
EGDN
Nitrocellulose
Nitroglycerene
Octogen (HMX)
PETN
Picric Acid
RDX
TNT
Tetryl
Cocaine
Heroin
LSD
Mandrax
Morphine
PCP
Acetamide
Ammonium acetate
Barley
Cotton
Dacron
Ethanol
Lucite (Perspex)
Melamine
Methanol
Neoprene
Nylon
Orlon
Paper
Polyester
Polyethylene
Polyurethane
PVC
Rayon
Silk
Soybean
Sugar
Water
Wood
Wool
H C N O
0 20 40 60 80 100
Atom fraction (%)
Explosives
Illicit drugs
Miscellaneous
substances

Oxygen concentration (moles cm -3 )
0.08
0.07
0.06
0.05
0.04
0.03
0.02
Sand
Water
Sugar
Cellulose
Acryllic
Ethanol
PETN
Nitroglycerine
Nitrocellulose
Picric acid
EGDN
Datasheet
Dynamite
TNT
Black powder
Polyurethane
Tetryl
Ammonium nitrate
Composition B
Cocaine
Bulk nylon
0.01
Cotton
Polyester
Silk
Melamine
0.00 Wool
0.00 0.01 0.02 0.03 0.04 0.05 0.06
Nitrogen concentration (moles cm -3 )
C-3
C-4
HMX
Explosives
Fabrics
Polymers
Miscellaneous materials

Nuclear Physics approaches for elemental analysis
Charged particles p, d, α, … :
Good for surface analyses only
(e.g. Rutherford Backscattering)
Not suitable for bulk objects
α
α
sample
γ-rays:
Penetrating
Not easily applicable to
multi-elemental analyses
Neutrons:
Penetrating
Sensitive to both light and heavy elements
Undergo specific interactions with different nuclides
Easily produced and detected
γ
sample
γ’
γ
Transmitted
γ-rays
Compton
scattered γ-rays

Neutron – based interrogation methods
sample
thermal
neutrons
γ
Capture γ-rays
fast
neutrons
γ
Inelastic scattering
γ-rays
fast
neutrons
n
Transmitted
neutrons

slow neutrons in,
capture gamma rays out
(thermal neutron capture)
252
Cf source
HPGe
detector
neutron
Gamma
ray
CW agent

Thermal neutron capture
thermal
neutrons
γ
Capture γ-rays

fast neutrons in,
gamma rays out
(fast neutron inelastic scattering)
mono-energetic
fast neutrons
sample
γ
Inelastic
scattering γ-rays
Accelerator-based system using
mono-energetic 8 MeV neutrons
and arrays of NaI detectors to
detect de-excitation γ-rays from
neutron inelastic scattering
interactions.

Fast neutron
inelastic scattering
mono-energetic
fast neutrons
sample
γ
Inelastic
scattering γ-rays
Portable system based
around a 14 MeV sealed
tube neutron generator and
BGO detector to detect deexcitation
γ-rays from
neutron inelastic
scattering interactions.
Pulsed Elemental Analysis with Neutrons (PELAN)

fast neutrons in, fast neutrons out
(fast neutron transmission spectroscopy)
fast “white”
neutron beam
n
transmitted
neutrons
10
ENDF/B-VI
1
H
Scattering cross section (b)
8
6
4
12 C
14
N
16
O
Neutron beam transmitted
through the sample is
attenuated according to
total scattering cross
section for each element.
2
0
0 5 10 15
Incident neutron energy (MeV)

Neutrons in, neutrons out
fast
neutrons
n, n’
Elastically and
inelastically
scattered neutrons
The energy and intensity distributions of the scattered neutron
field are functions of the:
• incident neutron energy
• angle of scattering
• mass of the scattering nuclide

Fast Neutron Scattering Analysis (FNSA)
Andy Buffler , Frank Brooks, Saalih Allie
Department of Physics, University of Cape Town, South Africa
Krish Bharuth-Ram
Department of Physics, University of Durban-Westville, South Africa
Rudolph Nchodu
Department of Physics, University of the Western Cape, South Africa
See:
Buffler and Brooks et al, “Material classification by fast neutron scattering”,
Nuclear Instruments and Methods B 173 (2001) 483

National Accelerator Centre, near Cape Town, South Africa

Fast Neutron Scattering Analysis (FNSA)
Concept:
Basic approach is to irradiate the sample under interrogation
with a beam of fast mono-energetic neutrons and to detect the neutrons
scattered out of the beam.
Nuclide-specific information is present in the scattered neutron field.
By detecting the neutrons elastically and inelastically scattered at
different laboratory angles θ for different incident neutron energies E o
, the
amounts and positions of the scattering nuclides may be determined.
monoenergetic
neutron beam
shielding
Backward
detector (B)
scattering
sample
150°
45°
Forward
detector (F)

1500 mm
300 mm
600 mm
400 mm
310 mm
d
Dtarget:
2
H(d,n) He
reaction
2 3
Iron
Borated
wax
n
Sample
Rotating
Havar foil
(E = 4.7/4.0 MeV)
d
M
NE213
Detector B
θ
B = 150o
θ F
= 45 o
NE213
Detector F
Laboratory scattering experiments undertaken at the 6 MV
Van de Graaff accelerator of the National Accelerator Centre, South Africa.
• Neutron energy multiplexed between 7.5 MeV and 6.8 MeV by rotating
a 11µm Havar foil in and out of the 4.5 MeV deuteron beam.
• Pulse shape discrimination on all NE213 detectors.
• Pulse height and time-of-flight information measured for small monoelemental
samples, as well of compounds of known and “unknown”
chemical composition.

Neutron elastic scattering
(a) 12 C(n,n) 12 C
Differential
cross section
dσ/dΩ (b sr -1 )
4.0
3.0
2.0
1.0
0.0
1.0 0.5 0.0 -0.5 1.0
-1.0
2.0
3.0
6.0
5.0
4.0
E o
(MeV)
7.0
8.0
cos Θ
)
E scattered
/ E incident
for neutron
1.0
0.8
0.6
0.4
0.2
Kinematics
A=1
0.0
0 30 60 90 120 150 180
Neutron laboratory scattering angle (degrees)
A=207
A=56
A=27
A=16
A=14
A=12
A=10
A=7
A=4
A=2
(b) 14 N(n,n) 14 N
dσ/dΩ (b sr -1 )
4.0
3.0
2.0
1.0
0.0
1.0 0.5 0.0 -0.5 1.0
-1.0
(c) 16 O(n,n) 16 O
dσ/dΩ (b sr -1 )
4.0
3.0
2.0
1.0
cos Θ
0.0
1.0 0.5 0.0 -0.5 1.0
-1.0
2.0
2.0
3.0
3.0
5.0
4.0
E o
(MeV)
6.0
6.0
5.0
4.0
E o
(MeV)
7.0
7.0
8.0
8.0
cos Θ

Pulse height (L) channel
60 40 20 0
14E3
γ
7E3
0
0 10 20 30 40 50 60
Pulse shape (S) channel
Counts
80E3
40E3 γ n
0
γ n
50E3
25E3
Pulse height (L) channel
0 20 40 60
n
0

Pulse height and/or time-of flight measurements are made
at 2 scattering angles and 2 neutron beam energies.
Time-of-flight
Pulse height
Graphite scatterer
E n = 7.5 MeV
Detector at 150°

θ lab
: 150 o
45 o
Parameter: Pulse height Time-of-flight
R i
(n)
800
400
0
400
H
C
S D0
signature
(based on both pulse
height and time-of flight)
θ lab
: 150 o
800
400
0
400
H
C
45 o
0
400
N
R i
(n)
0
400
N
0
400
O
E o = 6.8 MeV
0
400
O
0
400
Al
E o = 7.5 MeV
0
400
Al
0
400
S
0
400
S
0
400
Fe
S L0
signature
0
400
Fe
0
400
Pb
0
0 150 300 450 600 750 900
Bin number n
(based on pulse height
only – does not require
a pulsed beam)
0
400
Pb
0
0 150 300 450 600
Bin number n

Scattering signatures measured
for unknown samples are
unfolded to determine the
elemental composition of the
sample:
N+
Sn ( ) f R( n)
= ∑ 4
i = 1
i
i
S(n)
200
150
100
50
Hist Data
Fit
Background
0
0 150 300 450 600 750 900
where:
S(n) : scattering signature measured for a sample of unknown composition
R i
(n) : scattering signatures measured for N individual elements
and 4 background components
(±∆f i
) : fitting coefficients
f i
n
The atom fraction a i
a
i
for each element in the unknown compound is simply
f
i
= N
∑
i = 1
f
i
The atom fractions a i
±∆a i
uniquely characterize the elemental composition
of the scattering sample.

Scattering signatures
measured for unknown
samples are unfolded to
determine the elemental
composition of the sample.
The measured atom
fractions uniquely
characterize the
elemental composition of
the scattering sample.
atom fraction
0.8
0.6
0.4
0.2
0.0
Methanol Ammonium Heroin
nitrate simulant
HCNOAlSFePb
HCNOAlSFePb
HCNOAlSFePb
The identification of specific materials from measured atom
fractions can be facilitated by introducing a χ 2 -based screening
procedure to compare the atom fractions measured for an
“unknown” sample with the known atom fractions of a large set of
candidate materials.

Results from analyses
using scattering
signatures based on:
atom
fraction
0.8
0.6
Methanol Ammonium Acetamide Ammonium
nitrate acetate
(a) both pulse height
and time-of-flight
0.4
0.2
a i
0.8
0.6
0.0
0.0
atom
fraction
Methanol Ammonium Acetamide Ammonium
nitrate acetate
(b) pulse height only
a i
0.4
0.2
Histogram: Expected
H C N O Al S Fe Pb

The sensitivity of the approach for detecting particular materials
is illustrated by introducing a function analogous to a weighted
chi-square coefficient:
2
N
⎛ ai
− bik
⎞
∑⎜
⎟
2
i = 1⎝
∆ai
χ =
⎠
x
( k)
N
−2
N ∑ai
i = 1
where:
a i
±∆a i
: are the measured atom fractions for N
candidate elements in sample x
b ik
: the corresponding atom fractions for a candidate
material k of known chemical composition.
More convenient to use a screening function P x
(x,k) defined by
normalizing each calculated χ 2 (k) value to a percentage value by
P
x
( x, k)
=
100
2
x
( k)
45
−1
(
2
∑ χ ( ))
x
k
k = 1
χ

Screening materials k = 1 to 45
The atom fractions measured for
each “unknown” sample x are
screened against candidate
materials, e.g. the following set
of 8 elements, 10 explosives,
5 illicit drugs and 22 common
substances.
Elements:
Common substances:
1 Hydrogen 24 Paraffin wax
2 Carbon 25 Polyethylene
3 Nitrogen 26 Ethanol
4 Oxygen 27 Methanol
5 Aluminium 28 Water
6 Sulphur 29 Ammonium acetate
7 Iron 30 Nylon
8 Lead 31 Lucite
32 Polyurethane
Explosives: 33 Acetamide
9 Ammonium nitrate 34 Benzene
10 C-4 35 Sugar
11 RDX/HMX 36 Iron sulphate
12 EGDN 37 Wood
13 PETN 38 Paper
14 Nitrocellulose 39 Cotton
15 Nitroglycerine 40 Silk
16 TNT 41 Orlon
17 Tetryl 42 Wool
18 Picric acid 43 Melamine
44 Polyester
Illicit drugs: 45 Aluminium oxide
19 Heroin
20 LSD
21 Cocaine
22 Morphine
23 Mandrax

P x
(x,k) represents the degree to which a sample x matches a
member k of the set of 45 candidate materials.
The distribution of P x
(x,k) versus k displays a profile of the
screening of x against the 45 candidate materials, with peaks at the
values of k corresponding to the most probable candidate
materials.
Acetamide
P(x,k)
100
80
Element Explosive Drug Common substance
Acetamide
60
40
20
0
0 5 10 15 20 25 30 35 40 45
Material k

P(x,k)
100
80
Ammonium nitrate
(Explosive)
Heroin simulant
(Illicit drug)
100
80
P(x,k)
60
60
40
40
20
20
P(x,k)
100
0
0 15 30 45
80
Methanol
(Common material)
Material k
0
0 15 30 45
Ammonium acetate
(Common material)
100
80
P(x,k)
60
40
20
60
40
20
0
0 15 30 45
Material k
0
0 15 30 45

Ammonium Nitrate
Composition B
Composition 4 (C-4)
Dynamite
EGDN
Nitrocellulose
Nitroglycerene
Octogen (HMX)
PETN
Picric Acid
RDX
TNT
Tetryl
Cocaine
Heroin
LSD
Mandrax
Morphine
PCP
Acetamide
Ammonium acetate
Barley
Cotton
Dacron
Ethanol
Lucite (Perspex)
Melamine
Methanol
Neoprene
Nylon
Orlon
Paper
Polyester
Polyethylene
Polyurethane
PVC
Rayon
Silk
Soybean
Sugar
Water
Wood
Wool
H C N O
0 20 40 60 80 100
Atom fraction (%)
Explosives
Illicit drugs
Miscellaneous
substances

In many practical situations it may not be
important to determine the actual type of
material, but rather the group (explosive,
illicit drug, common substance) to which
the material belongs.
This identification is assisted by
introducing a group profile function
G(x,g) for unknown sample x, by
summing P(x,k) over the members k of
each group g :
Gxg ( , ) = ∑ Pxk ( , )
over each
group
where the 4 groups are:
1. Elements
2. Explosives
3. Illicit drugs
4. Common materials
(a) Methanol
100
G(x,g)
50
0
(b) Acetamide
100
G(x,g)
50
0
(c) Ammonium nitrate
100
G(x,g)
50
0
(d) Drug simulant
100
G(x,g)
50
0
g = Element Drug
Explosive Common material

A group probability function
indicates the relative probability
that the atom fractions measured
by FNSA for a test sample match
those for compounds in each of
the four groups:
Pure elements
Explosives
Illicit drugs
Common substances
Probability that sample belongs to material group
100
Methanol
50
0
100
50
Ammonium nitrate
0
100
Heroin simulant
50
Methanol
Heroin simulant
Ammonium
nitrate
Results in a highly reliable
classification which is suitable to
the detection of explosives and
illicit drugs.
0
Pure
Element
Explosive
Illicit
drug
Common
substance

Panel on the assessment
of the practicality of
Pulse Fast Neutron
Transmission Spectroscopy
for Aviation Safety
Report: 1999
Washington D.C.
Recommendations:
1. Do not consider accelerator-based technologies to have promise for
deployment as a primary screening procedure for checked baggage
inspection. Any screening procedure relying on an accelerator cannot compete
with available technologies on either cost or practicality bases
2. Do not fund any large accelerator-based hardware development
projects. Combinations of experimental work with existing laboratory
equipment, mathematical modeling and simulation can better define the
potential of nuclear technologies without the expense or time required to design
and build new hardware.

Location of contraband in FNSA measurements
Rotate and translate package across collimated beam.
Backward
detector
Forward
detector
Package
Collimated
mono-energetic
neutron beam
Transmission
detector
Shielding
Rotating
turntable
Translating
trolley

Results from a Rotational-Translational scan
Counts measured at 45°
as a function of
x (translational coordinate) and
θ (rotational coordinate)
explosive
illicit drug
dense cotton clothing
packed in suitcase

Results from a
Rotational-Translational
scan
(a) transmission detector
(b) 45° detector
(c) 150° detector (H-suppressed)
(d) 150° detector (H-enhanced)
water
graphite
air

Results from a
Rotational-Translational
scan
(a) transmission detector
(b) 45° detector
(c) 150° detector (H-suppressed)
(d) 150° detector (H-enhanced)
explosive
illicit drug
dense cotton clothing

Detection of explosives in airline baggage by FNSA
Focus on screening of outgoing baggage – requirements include
high throughput and low false alarm rate.
Signal: High density region (1.6 – 2.0 g cm -3 )
Appropriate HCNO atom fraction
High concentration of HCNO
Effective explosive detection systems must offer:
(a) Rapid operation (< 20 s per bag)
(b) Reliable detection
Zero false negative rate (must detect all explosives > 200 g)
Small false positive rate (few false alarms)
(c) Safe use
(d) Ease of use
Not clear whether or not the new generation of X-ray systems will
fully meet all these requirements.

Multi-stage interrogation protocol for explosives in outgoing luggage
STAGE 1
First (rapid) screening using
X-ray method (10 seconds)
Bag “suspect” ?
No
Yes
(~3% of bags)
FNSA could be used as
the second-stage of a
second phase of a
two-stage screening
system, in order to
provide corroborative
evidence via an
independent and more
sensitive screening
Bag moved to STAGE 2 station
STAGE 2
Suspicious voxel(s) positioned in neutron beam
Thorough FNSA screening (1 – 2 minutes)
Bag still “suspect” ?
STAGE 3
Visual inspection
No
Yes
( < 0.1% of bags)

Detection of illicit drugs in airline baggage by FNSA
Focus on the screening of incoming baggage
Signal:
Appropriate HC atom fraction
High H concentration
Requirements less stringent and less precise than
for explosives detection.
(Detecting > 0.5 kg of illicit drugs with 90% efficiency is useful.)

Proposed system for the
detection of illicit drugs in
incoming airline luggage.
FNSA could be used as the
first stage of a two-stage
screening system.
The second stage is
manual inspection by
customs officials.
Neutrons scattered out
of the beam by a passing
suitcase are detected in
arrays of neutron detectors
placed at forward and
backward angles.
backward
neutron
detectors
fan
beam
collimator
fast
neutron
source
forward
neutron
detectors
neutron
beam
luggage
conveyor
backward
neutron
detectors
shielding

FNSA screening times per suspect item:
Neutron source type
STNG Accelerator Accelerator STNG
E (MeV) 2.5 4.2 7.2 14.1
φ (E ) (× 10 6 s -1 ) 1 0.145 1.13 10
t M (s) 2.8 18.8 3.5 0.94
E : neutron beam energy
φ (E) : number of source neutrons incident on sample per second
t M
: FNSA measuring time per suspect item
In addition, need to include time for:
Neutron fluence reduction due to attenuation:
Preliminary RT-scan:
Computing:
Mounting and positioning:
10 - 30 s
10 - 30 s
5 - 10 s
10 - 20 s
⇒
Total time for FNSA screening: 60 to 120 s per suspected item.

Conclusions
Laboratory tests have shown that FNSA is a highly reliable
method for characterizing materials in bulk.
FNSA measures the important elements for contraband detection
(H, C, N and O) with similar sensitivity.
FNSA is complementary to other techniques employing
mono-energetic neutron beams.
Present work on FNSA is being directed towards the
determination of the most suitable system parameters for practical
applications, before constructing a prototype to subject to further
testing and evaluation.