Вы находитесь на странице: 1из 127

�������������������������������

��������������������������������������������

���������������������������������
���������������������������������������
�������������������������������������
����������������������

��������������

��

������������������������������������
����������� �������������������
����������������������
������������ ������������������
������������������
��������

�������������������������� ���������������� ��

�����������������������������������������

���������������������������������
���������������������������������������
�������������������������������������
����������������������

�������������������������

��������������

�����������������������

����������������������������

������������ ������������������
����������������������������
�����

������� ����������������������������������������������������������
����������������������������������������������������������
������ ����� � �������� � ��� � � � ������ � ���������� � �������� � ����
������ � ��� � ���������� � �� � ��� � ����� � ���� � ������ � �������
�����������������������������������������������������������
�����
����� ������� �������� ��������� ��� ���� � ����� �������� � �� � ��������
������������������������������������������

������ � �� � � � ������ � ����� � ������� � ���������� � �������� � ��������


����������� �������������������������������������������������������
����
��������������
��������������������������������
��������������������������������������
����������������������������
to Elisa and Antonio

i
ii
Contents

1 Introduction 1

2 The Prompt Gamma Neutron Activation Analisys 7


2.1 Characteristics of PGNAA . . . . . . . . . . . . . . . . . . . . . 8
2.1.1 The main neutron interactions occurring in a PGNAA
Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Analytical properties . . . . . . . . . . . . . . . . . . . . 10
2.1.3 Sensitivity and Detection limit . . . . . . . . . . . . . . . 11
2.1.4 Prompt gamma spectra . . . . . . . . . . . . . . . . . . . 12
2.2 Main components of a PGNAA facility . . . . . . . . . . . . . . 14
2.2.1 The Neutron beam . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 The shielding and the neutron absorber materials . . . . 16
2.2.3 The shutter, the sample holder and the beam stop . . . 19
2.3 Gamma-ray detector . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.1 The High Purity Germanium detector . . . . . . . . . . 20
2.3.2 Response function of a germanium detector . . . . . . . 23
2.3.3 Detector efficiency . . . . . . . . . . . . . . . . . . . . . 25
2.3.4 Compton-suppressed spectrometer . . . . . . . . . . . . . 27

3 Experimental boron concentration measurement 31


3.1 Boron measurement by α spectroscopy . . . . . . . . . . . . . . 32
3.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.1.2 Experimental set-up . . . . . . . . . . . . . . . . . . . . 33
3.1.3 Experimental spectra . . . . . . . . . . . . . . . . . . . . 35
3.1.4 Monte Carlo simulation . . . . . . . . . . . . . . . . . . 36
3.1.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2 Experience at NIST . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.1 PGNAA facilities at NIST . . . . . . . . . . . . . . . . . 41
3.2.2 Quantitative analisys . . . . . . . . . . . . . . . . . . . . 42
3.2.3 Experimental measurements performed at NIST . . . . . 48

iii
CONTENTS

4 The design of a PGNAA facility at Triga Mark II reactor 55


4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Materials and method . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 The Triga Mark II reactor . . . . . . . . . . . . . . . . . 55
4.2.2 MCNP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.3 The facility design . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.1 The Channel choice and its characterization . . . . . . . 65
4.3.2 Filtering of the neutron beam . . . . . . . . . . . . . . . 67
4.3.3 Collimator . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.4 The beam catcher design . . . . . . . . . . . . . . . . . . 81
4.4 The simulation of the germanium response function . . . . . . . 88

5 Conclusions and future perspectives 101

List of publications 116

Aknowledgments 117

iv
CONTENTS

vi
Chapter 1
Introduction

The aim of this project is the design of a Prompt Gamma Neutron Activa-
tion Analysis (PGNAA) facility at the Triga Mark II Research Reactor of the
university of Pavia, in Italy. This measurement technique will be used by the
BNCT research group in order to evaluate 10 B concentrations in biological
samples.
Boron Neutron Capture Therapy (BNCT) is a binary experimental radio-
therapy exploiting the combined action of the somministration of 10 B and a
low energy neutron irradiation to cause a lethal damage to the tumoral cells
[1].
Owing to the capture of a thermal neutron in 10 B, an α particle and a 7 Li
nucleus are produced. Since their range is comparable to the cellular diameter,
the release of this high LET radiation causes a localized damage in the tissue.
The efficacy of BNCT depends on 10 B concentration ratio between the tumor
and healthy tissue: if this ratio is high enough, the tumor cells are killed with
a substantial sparing of the normal ones. For this reason, the knowledge of the
boron concentration both in tumor and in normal tissue is a crucial issue for
preclinical studies and clinical BNCT application.

Currently, the BNCT research group working in Pavia is performing pre-


clinical studies for the in-situ treatment of the lung metastases within the
WIDEST1 (WidESpread Tumors BNCT) project funded by the National In-
stitute of Nuclear Physics (INFN). At the same time, feasibility studies of a
BNCT treatment for the limb osteosarcoma are being carried out.
Both the projects require measuments of 10 B in biological samples. For this
purpose two well established method are available in Pavia: α spectroscopy
and neutron autoradiography emplojed for quantitative and qualitative anal-
ysis respectively[2][3].
The neutron autoradiography is mainly used to evaluate the uptake of 10 B in
different kind of tissues, that may contain different structures such as tumour
cells, normal parenchyma, necrotic areas, blood vessels, connective tissue. The
information are further used to correct spectroscopic results, in order to pre-

1
1. Introduction

cisely evaluate the boron concentration values in the tumour and in the normal
tissues.
To perform α spectroscopy, a thin tissue slice (60 µm) cut from a cryogenic
sample is kept under vacuum and irradiated in the thermal column of the
TRIGA Mark II reactor. The samples are irradiated in front of a thin silicon
detector that collects the charged particles coming from neutron capture in
10
B and 14 N . A contiguous section of the sample is cut to perform standard
histological analysis. Then, the information resulting from the acquired charge
particles spectra are combined with spatial information from the histological
analysis to get the boron concentration in the different tissues types present in
the sample. With this method a sensitivity of 0.5 ppm of 10 B can be achieved.
The need to keep the sample under vacuum during the irradiation, makes the
analysis of biological liquid samples (such as blood and urine) impossible with
this technique.

Many other different techniques are also employed for boron measurement
in BNCT research, they are all described in [4]. These can be divided into
microscopic and macroscopic techniques, according to their spatial resolution.
The macroscopic analysis can provide an average boron concentration both in
tissue and in blood, whereas a microscopic analysis allows an evaluation of
boron distribution at cellular and subcellular level. The information obtain-
able by microscopic techniques are, of course, very important but, since they
are more expensive, they represent a further step to the macroscopic analysis.
Among the macroscopic techniques, Atomic Emission Spectroscopy (AES-
ICP), Inductive Couple Plasma-Mass Spectroscopy (ICP-MS), PGNAA and
α spectroscopy can be distinguished, while Secondary Ion Mass Spectroscopy
(SIMS), laser-Secondary Neutral Mass Spectroscopy (laser-SNMS) and High
Resolution Quantitative Autoradiograpy (HRQAR), can be mentioned for mi-
croscopic analysis.
The ICP-MS has an high sensitivity for 10 B and 11 B, it represents an efficent
method to evaluate the mean total B concentration and mean isotopic ratio in
biological matrix. On the other hand it is a destructive technique.
SIMS and laser-SNMS are both imaging boron techinques able to give the dis-
tribution map of the different element on the surface (2D image); the spatial
resolution achievable with SIMS is in the nanometer-scale. While the laser-
post ionization SNMS has an higher sensitivity allowing a subcellular distri-
bution image of 10 B and detection limits in upper-ppb range.
A very accurate radio-analytical techinque is represented by HRQAR. Boron
imaging at microscopic level and boron quantification can be performed at the
same time; this is possible by overlapping a neutron autoradiograph, i.e maps
of boron tracks, with a histological analysis, identifing the biological structures,
at the same tissue slice.
Other techinques, already well established in clinical practice and byologi-
cal studies such as Nuclear Magnetic Resonance (NMR) and Magnetice Res-
onance Imaging (MRI), Positron Emssion Tomography (PET), are also em-

2
ployed for BNCT studies. These are candidate to be applied for the real-time
10
B-enriched compound monitor, allowing the knowledge of the 10 B maps dur-
ing the treatment, leading to a more effective BNCT dose delivering.

PGNAA is based on the detection of gamma rays coming from the neu-
tron capture reaction taking place when a thermal neutron is captured in an
element. The high neutron capture cross section of 10 B (3837 b) makes boron
one of the most sensitive elements for this technique.
PGNAA has been used for boron concentration measurements for BNCT stud-
ies since many years. As an example, the PGNAA facility at the High Flux
Reactor in Petten allows accurate and non-destructive measurements of 10 B
concentrations down to 1 ppm in about 5 min for macroscopic samples (0.4 -
1.0 ml) of tissue, blood and urine [6]. In Petten prompt gamma analysis was
used to measure the 10 B concentration in blood after the administration of
sodium borocaptate, in order to to optimize the BNCT irradiation schedule
for the treatment of patients [5].
At the MIT Research Reactor a diffracted neutron beam PGNAA facility was
manly used for BNCT studies. It provided a thermal neutron flux at the sam-
ple position of 1.7 × 107 n/(cm2 s), with a sensitivity for natural boron in water
of 3750 cps/mg [18].
In Japan, maeasurements of 10 B concentration in blood, cerebrospinal fluid
(CSF) and brain tissue were performed from Mukay, Nakagawa [7]. At the
Musashi Institute of Technology evaluation of 10 B was performed in tumours,
tissue,blood and cultured cells [8]. Other institutions involved in BNCT projects
with PGNAA facilities are: the HANARO reasearch reactor in Korea Atomic
Energy Research Institute [9], the light water research reactor facility in Rez,
Czech Republic [10] and at the research reactor FRMII in Garching near Mu-
nich [11].
The implementation of a PGNAA facility at Univeristy of Pavia not only
allows improving the measurement of 10 B capability of the BNCT research
group, making possible the analysis of biological liquid samples, but it repre-
sents an important tool for a rapid evaluation of 10 B concentration during a
BNCT treatment.
A prompt gamma spectroscopy for BNCT purposes requires the evaluation of
a concentration of the order of µg/g of 10 B in tissue or blood sample with
measuring times of minutes. The mass of the sample could be of the order
of 0.05 g, such as healthy or tumoral biopsies with concentration of tenth of
ppms, or even smaller samples, of the order of 10 µl as in case of blood with-
drawal from small animal (100 ng of 10 B)). These BNCT requirements give
important constraint about some structural parameters of the facility, such as
the thermal neutron flux and the count rate background in the boron region,
being these directly related to the 10B detection capability of the measure-
ment method. In particular a minimum amount of detectable 10 B mass (DL )
of the order of hundreds of nanograms in times of minutes requires the use of

3
1. Introduction

a PGNAA facility highly sensitive to 10 B. Better characteristics of the facility


can be obtained by increasing the number of boron neutron capture reaction
by optimizing the thermal neutron component of the beam, maximizing the
detection efficency for 478 keV γ-rays and reducing the background count rate.

In the next chapter, a description of the measurement method based on


prompt gamma neutron activation analysis will be given. In particular, the
characteristics of the method, i.e. the basic physical processes and the ana-
lytical properties, the features of the main component of a PGNAA facility,
i.e the neutron beam, the shutter and the sample holder, the beam catcher
and the germanium detector, and some details about the spectrometry will be
described.
In the third chapter the experimental part of the PhD work is described. In
particular, the first section is devoted to the description of an experiment per-
formed to evaluate the α particle energy loss in lung tissue below 2 MeV. The
experimental set-up and the procedure are described and finally a curve of
the α particle energy loss as a function of the crossed lung tissue thickness is
presented. This curve allows the evaluation of a fundamental parameter for
the evaluation of 10 B concentration in lung sample by α spectroscopy. The
second section reports a series of experimental measurements performed at the
thermal neutron beam PGNAA facility at National Institute of Standards and
Technology (NIST, Gaithersburgh, USA), along with some general concepts
about the boron measurements by the prompt gamma method.
In the fourth chapter the design of the facility of the Pavia reactor will be
described, giving details about each component of the facility, pointing out the
requirements, the possible solutions proposed and the Monte Carlo techniques
used to achieve reliable results.
The last chapter is dedicated to the perfomances of the designed facility, draw-
ing some conclusions and anticipating some details about the future work to
be done in order to build the facility.

4
1. Introduction

6
Chapter 2
The Prompt Gamma Neutron
Activation Analisys

The Prompt Gamma Neutron Activation Analysis is a radioanalytical tech-


nique able to detect every element (except 4 He) in the periodic table. This
method is based on the detection of the radiation coming form the neutron
capture reaction (or (n,γ)) where a nucleus absorbs a neutron and successively
emits a characteristic γ radiation.
The first prompt gamma radiation evidence was in 1934 when a gamma ray
(2223.2487 keV) from the 1 H(n, γ)2 H was detected.
In 1966 the first PGNAA measurement with a reactor neutron beam was per-
formed by Isenhour and Morrison, where a NaI(Ti) scintillator was used to
detect the gamma ray. The potentiality of the PGNAA grew when the Ge(Li)
semiconductor detector with a high energy resolution became available (1967)
and as well as Ge(Li) Compton-suppressed system.
Moreover, in 1969 neutron beams with higher flux became available thanks
to the use of neutron guideS. At the same time, measurements of gamma ray
energies, intensities and plot of spectra helped the spreading of this measure-
ment technique. The “Lone Table“, published in a IAEA Technical Report
(1993) [13] had been for long time the only source of spectroscopic data. It
contained information of more than 10000 gamma rays of elements, obtained
mainly in the experiments performed at Massachusetts Insititute Technology
(MIT). Recently, a database of prompt gamma rays coming from the nuclear
reactions with all elements was published [14], the information about the re-
maining elements were obtained mainly by the effort of the group from the
Budapest Research Reactor[15]. The availability of a complete catalogue of
prompt gamma ray grew the interest of the scientists in PGNAA method.

7
2. The Prompt Gamma Neutron Activation Analisys

2.1 Characteristics of PGNAA


2.1.1 The main neutron interactions occurring in a PGNAA
Facility
The neutrons are commonly divided into three main groups according to their
energy: the slow neutrons produced during the thermalization process, their
energy is lower than 100 meV; these can be further classified in cold and
thermal neutrons; the epithermal neutrons with energy between 0.1 eV - 1
MeV and fast neutrons with energy higher than 1 MeV.

The main physical process occurring in a PGNAA is the radiative nuclear


capture reaction, or (n, γ): a slow neutron is absorbed by the interacting
nucleus giving rise to a compound nucleus whose exitation energy is the sum
of the binding energy plus the kinetic energy of the neutron. After 10−16 sec-
onds it decays and the nucleus reaches its groud state by the emission of 2
-4 gamma rays in cascade, (prompt gamma rays). If the nuclide coming from
the(n, γ) reaction is unstable, it further decays by a β particle emission and
the subsequently deexcitation of the daughter nuclide gives rise to the emission
of one or more gamma-ray photons, (delayed gamma rays). The detection of
this radiation is the basis of neutron activation analysis (NAA).The PGNAA
is based on the detection of both radiations but mainly the prompt one.
Another process taking place when a slow neutron interacts with a light ele-
ment is the capture reaction with the emission of a charged particle.
The most relevant cases are: 3 He(n, p)3 H with thermal cross section equal to
5333 b; 6 Li(n, t)4 He (940 b), 14 N (n, p)14 C (1.83 b) without gamma emission
and 10 B(n, α)7 Li (3837 b). With 94% of probability the 7 Li coming from the
interaction with 10 B is emitted in an excited state (7 Li∗ ) which decays with
the emission of a 478 keV gamma ray. For the listed nuclei the (n, γ) reaction
is less important.
The epithermal component of a beam may undergo radiative neutron cap-
ture reaction. The energy of the capture state is less defined than in the
previous (n, γ) case, it depends on the kinetic energy of the epithermal neu-
tron that can vary in a wide range. Thus, the energy of gamma emitted during
the transition of the excited state to the groud state is not always the same
except in case of resonaces. The analytical importance of these reactions has
not yet been studied systematically.
Moreover, a neutron can undergo an elastic scattering or inelastic process. The
epithermal and fast components are mainly elastically or inelastically scattered
by the nucleus. In case of elastic scattering, the result of the interaction is a
recoil nucleus with some kinetic energy transfered by the neutron while the
neutron itself changes its energy and direction. Whereas the inelastic scatter-
ing (n,n’) occurs when the neutron energy is above the first excited state of
the target nucleus, being thus relevant only for fast neutron beams. More de-
tailed descriptions of the neutron interactions occurring in a PGNAA system

8
2.1. Characteristics of PGNAA

are reported in [22].

Fundamental quantities The probability of each of these interactions is


characterized by the cross section that changes with the energy of the neutron.
In the slow neutron range the cross section follows the 1/v law, where v is
the neutron velocity. The capture cross section of the nuclides are usually
tabulated for 2200m/s monocromatic neutrons end these values are indicated
with σ0 . In case of regular nuclides, i.e. following the 1/v law, the σ for
different energies neutrons is:
v
σ(ν) = σ0 (2.1)
v0

Another important quantity in PGNAA is the partial cross section gamma ray
producion:

σγ = θσPγ (2.2)
where θ and Pγ are respectively the natural isotopic abundance of the given
nuclide and the emission probability of this gamma ray.

This quantity is used to determine (ργ ), that is the count rate (s−1 ) of a
gamma peak at a given energy of the spectrum. In the simplified case of a
monocromatic beam and an ideally thin and homogeneus sample, ργ takes the
form:
ργ = ε(Eγ )nσγ Φ (2.3)
ργ is the area of the peak divided by the measuring time (live time), while n is
the number of atoms, ε(Eγ ) is the counting efficency under the photoelectric
peak of the detector and Φ is the neutron flux.
This expression becomes more complicated when a real case is considered,
where, for instance, the neutron spectrum is not monoenergetic and the thick-
ness of the sample can introduce either gamma ray attenuation or neutron self
shielding. The formula including these effect is:
� � ∞
µ(r)
ργ = NA σγ (En )Φ� (En , r)ε(Eγ , r)dEn dr (2.4)
V 0 M

where µ(r) is the mass density as a function of the position in the sample, M
is the atomic mass of the element, NA is the Avogadro number, σγ (En ) is the
partial gamma ray production cross section as a function of the neutron energy.
The neutron flux Φ� and the efficency ε� are also expressed as a function of the
position inside the sample (r) and, respectively, of the energy of the neutron
beam (En ) and of the given gamma energy (Eγ ).
In practice, some tricks can be adopted to reduce these effects, for example, a
relatively large sample-to-detector distance avoids that the counting efficency
depends on differences in samples shape. Moreover, if the sample is relatively

9
2. The Prompt Gamma Neutron Activation Analisys

thin and small, the self absorption can be neglected; otherwise in case of bulk
samples a correction factor accounting for the self absorption can be applied.
Similarly a simple equation taking in to account the self shielding effect must
be used.
In case of thin sample approximation the expression 2.4 can be simpli-
fied: the efficency and the flux values do not depend on the position inside the
sample, thus the count rate can be expressed as:
� ∞
ργ = σγ (En )Φ(En )dEn dr = σγ0 Φ0 (2.5)
0

where σγ0 is the partial gamma ray production cross section for monocro-
matic neutrons with v = 2200ms−1 , and Φ0 is the thermal equivalent neutron
flux defined as:

π v0
Φ0 = ΦT (2.6)
2 vT
using Φ0 , the reaction rate and the count rate become respectively:
m
R0 = nσ0 Φ0 ργ0 =
NA σγ0 Φ0 ε(Eγ ) (2.7)
M
Thus, the count rate depends only on the mass of the sample and parame-
ters like shape or position do not appear in its formula.

2.1.2 Analytical properties


The main radiations involved in this measurement method are neutron and
gamma rays, both uncharged, therefore characterised by high penetration. As
a consequence, when the beam illuminates the sample, the interaction interests
the whole thickness, and the resulting gamma radiation carries the informa-
tion about the average composition of the sample. The energy of the emitted
gamma rays is wide, so the detected energy spectrum detected is not affected
by the self-absorption fenomena for gamma lines above 100 KeV. Nevertheless,
some attention must be paid in some particular cases: i.e. when the sample
under study is characterized by a high neutron absorption cross section, the
self-shielding effect can occur, and consequenty the result may be distorted.
In case of samples made of high Z elements, the gamma rays intensity is at-
tenuated travelling inside the sample.
The PGNAA method is non destructive because it does not require a partic-
ular sample preparation and the sample can be analized many times because
the irradiation does not change significantly the properties of the sample 1 .
Nevertheless a radiation damage can occur when the capture reaction gives
1
After a whole day irradiation with a neutron flux typical of prompt gamma (that is
lower than 109 n/(cm2 s)), the number of converted nuclide is 1014 if all the neutrons undergo
interactions, this number is 8 order of magnitude lowers than the number of the atoms in
the sample.

10
2.1. Characteristics of PGNAA

rise to charge particle emission such sample containing boron or lithium.


Another characteristc of the PGNAA method is the possibility to detect a low
amount of light elements in a matrix made of high Z elements. In particular,
traces of hydrogen can be detected with good sensitivity in almost every kind
of matrices.

2.1.3 Sensitivity and Detection limit


A PGNAA facility is characterized by two parameters the sensitivity and the
detection limit particular for an element.
The analytical sensitivity for a specific element is defined by:
NA
S = ργ /m = σγ,0 Φ0 ε(E) (2.8)
M

It depends on the partial gamma ray production cross section σγ,0 , the neutron
flux Φ0 and the counting efficency at the measured energy ε(E).
From the practical point of view, the sensitivity ((counts/s)/mg) for an element
is the net count rate under the measured peak, divided by the mass (m) of
that specific element in the sample:
NG − NB
S= (2.9)
tm
where NG and NB are respectively total and background counts under the peak
and t is the count time.

The detection limit (LD ) for an analyte represents the minimum detectable
amount of the element under study.
The espression for LD is derived from the theory of the ”hypotesis testing”
and from the definition of two fixed limit: α, indicating the probability of
false positive and β, that is the probability of a false negative. Their default
values are 0.05 [32]. The LD is associated to true value of the quantity of
interest. The assertion of a true value can be done comparing the outcome of
the measurement (L̂) with a critical value LC ; if L̂ > LC the decision ”detected”
can be assessed.
These quantities, LC and LD can be defined as multiple of a standard deviation
of the blank σ0 and of the sample σD respectively, according to the following:

LC = kα σ 0 (2.10)
LD = LC + kβ σD (2.11)
(2.12)

A simplified relation can be obtain under particular conditions, i.e assuming a


Gaussian distribution of the data, α = β, kC = kB = k, and assuming constant

11
2. The Prompt Gamma Neutron Activation Analisys

standard deviation σ0 ; the definition takes the following espression:

LC = 1.645σ0 (2.13)
LD = 2LC = 3.29σ0 (2.14)
(2.15)

In the case of prompt gamma, the probability distribution of the counts


is Poissonian, than σ is the squre root af the nuber of counts. Moreover, the
estimation of the mass of an element is obtained by the sensitivity reversing
the formula 2.9. Starting from these considerations, the σ0 can be expressed
as a function of the count time t (s) and the sensitivity S (cps/mg) according
to:
√ �
NB Rb /t
σ0 = = (2.16)
St S

where Rb is the background count rate under the photopeak.


Then, the detection limit DL (mg) for PGNAA can be determined by the
background rate as follows:

Rb /t
DL = 3.29 (2.17)
S

The highest analytical sensitivity and detection limit is shown in fig.2.1 for
different element in absence of matrix obtained with a cold neutron beam in
the JRR-3M facility at the Japan Atomic Energy Research Insititute (JAERI)
[16].
DL may vary of one order of magnitude accordingly to the matrices. Nev-
ertheless, it is possible to identify three levels of detection:

• ng/g, in case of B, Cd, Sm and Gd

• µg/g, for H, Na, S, Cl, K, Mn, Co and Hg

• some percents, for the remaining elements

2.1.4 Prompt gamma spectra


A tipical gamma ray spectrum in the PGNAA has an energetic range between
few keV and almost 12 MeV; it is made of several hundreds of peaks and
the Compton plateau is large. Moreover, the number of peaks is related to
the nuclear level structure of the emitting nuclide. In case of light elements,
the number of levels where the excited nucleus can transit before reaching
the ground state, can be zero ( 2 H, 3 H and 4 He nuclides), one or two. As

12
2.1. Characteristics of PGNAA

Figure 2.1: The highest analytical sensitivity and the lowest detection limit
of element by cold neutron beam PGNAA facility of JRR-3M at the Japan
Atomic Energy Research Insititute.

a consequence, the typical spectrum (fig.2.2) is constituted by high intensity


peaks with energies uniformly distributed around the capture state.

Figure 2.2: Sketches of the typical PGNAA spectra from light (left) end heavy
(right) elements. The number of full energy peaks is higher for heavy elements;
the peaks in the lower or higher energy intervals are selected for nuclide iden-
tification purpose. For light elements, the spectrum is more simple.

For heavy elements, the spectrum has a higher number of peaks overlaping
in the middle region of the energy range. Therefore, the only peaks that can
be used for the identification are located at low and high energies, and are
those originating form the direct transitions from an exited level to the ground
state. Anyway, since the high energy region has a lower number of peaks, for
a more reliable qualitative analysis it is better to use this zone.

13
2. The Prompt Gamma Neutron Activation Analisys

Figure 2.3: Configuration of PGAA facility at Dalat Research Reactor (DNRI).

2.2 Main components of a PGNAA facility


A PGNAA facility consists of a neutron beam, mainly composed by cold or
thermal neutrons, properly collimated, irradiating a sample. The sample is
located always at the same position, in a target assembly protected by a shutter
that turns on and off the beam allowing an easy change of the sample. A
gamma-ray detector provided of a shield for neutrons is located perpendicularly
to the beam axis, moreover a beamcatcher is required to stop the beam crossing
the sample. All the set-up must be surrounded by a biological shield to protect
the operators. In fig.2.3 a PGNAA facility is represented.

The main features of each component is described in detail.

2.2.1 The Neutron beam


The main characteristics of a PGNAA neutron beam are to be bright and
clean, that is, with a high cold or thermal neutron flux in order to get a high
neutron capture reaction probability, with low epithermal and fast components
and low reactor gamma radiation, in order to minimize the background. The
neutron flux must be stable over time and spatially uniform.
There are different neutron sources that can be used for a PGNAA facility,
but the mostly used are obtained in research reactors. The neutrons origi-
nated in the core are slowed-down by the moderator, generally water or heavy
water. The resulting energy spectrum is characterized by three components:

14
2.2. Main components of a PGNAA facility

a Maxwellian distribution in the eV region, a continuous slowing down in the


epithermal region (˜keV) and a typical fission distribution in the high energy
region (MeV). Even if the neutrons are moderated, a further optimization is
necessary in order to ensure a suitable peformance of the PGNAA facility. This
can be achieved introducing filters that partially absorb the unwanted compo-
nents, although this may imply a sacrifice in the desired intensity. Commonly,
a filter for the gamma radiation is used togeter with a thermal neutron filter
consiting of a single or poli-crystal block, that transmits a specific neutron
energy interval according to its properties. The theoretical and experimental
total cross sections for some of the most important single crystal filter, i.e sap-
phire, bismuth and silicon, are collected in [17].
Another approach used to obtain a neutron beam for PGNAA is to exploit the
Bragg diffraction: the primary beam inpinges on a crystal and a narrow range
of wavelenghts is removed from the beam and deviated by a precise angle. The
advantage is that the new beam is not contaminated by the gamma radia-
tion coming from the core. An example of PGNAA facility using a focused
diffracted neutron beam is at MITR II reactor in Boston.
When the beam tailoring is conceived, another essential element is the colli-
mator, that geometrically defines the diameter of the beam. It is normally
placed in a position inside the biological shield and it ends near the position
of the sample. It may affect the intesity of the beam depending on its shape
and lenght.
The flux of the nuclear reactor neutron beams for prompt gamma found in
literature range from 105 − 1010 n/(cm2 s) although the most used are among
107 − 108 n/(cm2 s). The intensity of a beam is an important parameter to get
a good analytical sensitivity, thus the distance from the core gives a constraint
in the design of a PGNAA facility. Indeed, the sample is usually as close as
possible to the reactor beam port. On the other hand, in that position the
background is higher making the shielding an important issue.
For many reasons, the beams obtained by the mirror guides are the best ones.
This method allows extracting a cold neutron beam with the advantage of a
higer neutron capture reaction cross section along with a higher intensity of the
beam. As an example, a thermal equivalent (see eq.2.6) flux of 8×108 n/(cm2 s)
at 40 m from the source was achieved at the cold-neutron PGNAA system at
National Institue of Standards and Technology (NIST) [19]. The basic princi-
ple of the neutron guide is similar to the one of the optical fiber light guide,
that is, the total reflection of neutrons from the surfaces. Only the neutrons
impinging on the guide wall at an angle below a characteristic critical value
are totally reflected. This critical angle is related to the material and it is
inversely proportional to the wavelenght; the result is a cold beam completely
free from the fast neutron component. Moreover, the natural decreasing of the
beam intensity with the square distance does not occur. By curved neutron
guides it is also possible to drive the neutrons far from the core also reducing
the gamma ray background. In table 2.1 some of the most important PGNAA

15
2. The Prompt Gamma Neutron Activation Analisys

system with the corresponding extraction methods are summarized.

2.2.2 The shielding and the neutron absorber materials


The problem of the gamma and neutron radiation shielding and absorption is
one of the most importants in the design of a PGNAA system. For example,
the shutter and the beam stopper must be designed in order to absorb the high-
flux neutron beam and the gamma radiation after the measurement process.
To absorb a non-monocromatic neutron beam is firstly necessary to slow
down the fast component of the beam and successively to absorb the thermal-
ized neutrons, exploiting the 1/v trend of the neutron capture reaction cross
section.

Neutron attenuation When a neutron interacts with an hydrogen nucleus,


it loses on average 50% of its energy in a single elastic collision (see §2.1.1).
For heavier nuclei, only lower transfer is possible, so the hydrogenous materials
are the most effective in slowing down neutrons.
The high ability of hydrogen in slowing down the neutrons, can be explained
as follows: when an interaction with a nucleus of hydrogen takes place, the
energy of the neutron is on average halved. Since the neutron cross section
for hydrogen increases with the lowering of the energy, the mean free path for
the next collision is highly reduced. Therefore, almost all the interactions of
the neutron are close to the position of the first one. When the energy of the
neutron is lower then 1 MeV, one single scattering event is enough to “remove“
the neutron from the fast energy region.
Among the moderating materials, the most used is the ordinary concrete, be-
cause it has an high percentage of hydrogen, it is cheap and easy to shape. The
use of massive blocks of concrete to protect the personnel from both neutron
and gamma radiation are commonly used.
For the attenuation of the fast neutrons it is more convenient to use material
with heavy elements (high A). In this case, the most significative processes
that contribute to lower the energy are the inelastic scattering and the (n,2n)
processes. The first process occurs at neutron energy higher than 0.6 - 0.8 MeV
in the iron and lead. The energy threshold for the (n,2n) reaction is higher.
The average energy E � of a neutron coming from an inelastic scattering with
a nucleus of atomic mass A is approximately:


E
E = 6.4 (2.18)
A
where E is the energy of the incoming neutron in MeV. In a neutron shielding
the heavy element can be inserted as a slab of iron or lead or alternatively
spread in concrete. Moreover, the presence of the high Z elements helps in
absorbing the gamma radiation. It was shown that the use of alternating
shields of heavy and hydrogenated materials (such as iron and concrete) is a
good strategy to attenuate neutron and gamma radiations [20].

16
2.2. Main components of a PGNAA facility

Facility Characteristics
SNU-KAERY Thermal beam extraction: diffraction (PG)
Beam flux: 8.2 · 107 n/(cm2 s) (thermal equivalent)
Beam size: 2 × 2 cm2
Cadmium ratio: 266 (for gold)
Effective temperature: 269 K
Titanium (1382 keV) sensitivity: 2020 counts·s−1 · g−1
Detection system: single HPGe with pulse processing system
Total background counting rate: 3000 counts/s

Dalat research reactor Thermal beam extraction: moderation (graphite) and filtering (Bi)
Beam flux: 2.1 · 107 n/(cm2 s)
Beam size: 2.5 cm
Cadmium ratio: 21 (for gold)
Detection system: single HPGe with pulse processing system

NIST (thermal) Thermal beam extraction: filtering (sapphire)


Beam flux: 3 · 108 n/(cm2 s)
Cadmium ratio: 160
Effective temperature: 300 K
Titanium (1382 keV) sensitivity: 890 counts ·s−1 · g−1
Detection system: HPGe and Compton suppression electronics

NIST (cold) Cold beam extraction: filtering (Be, Bi) and mirror giude
Beam flux: 9.5 × 108 n/(cm2 s)
Beam size: 2 cm or smaller
Effective temperature: 14 K
Titanium (1382 keV) sensitivity: 7700 counts·s−1 · g −1
Detection system: HPGe and Compton suppression electronics

Budapest research reactor Cold beam extraction: mirror guide


Beam flux: 5 · 107 n/(cm2 s) (thermal equivalent)
Beam size: 1 × 1 or 2 × 2 cm2
Effective temperature: ∼= 60K
Titanium (1382 keV) sensitivity: 750 counts ·s1 g 1
Detection system: HPGe and Compton suppression electronics

BARC (thermal 1) Thermal beam extraction: mirror guide


Beam flux: 1.4 · 107 n/(cm2 s) (total)
Beam size: 2.5 × 10 cm2
Cadmium ratio: 3.4 · 104 (for indium)
Detection system: single HPGe with pulse processing system

BARC (thermal 2) Thermal beam extraction: diffraction (graphite)


Beam flux: 1.6 · 106 n/(cm2 s) (thermal equivalent)
Beam size: 2.5 × 3.5 cm2
Detection system: single HPGe with pulse processing system

Table 2.1: Main characteristics of PGNAA facilities.


17
2. The Prompt Gamma Neutron Activation Analisys

The attenuation of neutrons passing through an hydrogenous material en-


riched with a heavy material, under specific conditions, can be described by
the removal cross section. The removal cross section is the probability per unit
of path for a neutron to be ”removed” from the primary neutron beam. Indeed,
it was experimentally evaluated that the fast neutrons attenuation factor for l
cm of a shield as previously described is eΣrem l where Σrem is the removal cross
section, that is characteristic of the heavy materials. It is connected to the
microscopic removal cross section by the following formula:

0.6202σrem ρ
Σrem = (2.19)
A

where ρ and A are the density and mass number of the absorber respectively.
When more than one heavy material layer is present, the macroscopic cross
section becomes: �
Σrem = Σi l i (2.20)
i

where Σi is the removal cross section of the generic i element, li thick.

Neutron absorption The most used materials for the neutron absorption
are those containing 6 Li, 10 B, 113 Cd because of their high neutron capture
cross section. Among these, only 6 Li absorbs the neutrons without producing
gamma rays,10 B absorbs the neutrons via the (n, γα) reaction. About Cd,
altough it is easy to shape and it is a good absorber, it is not indicated in
case of high flux because of its residual radioactivity in addition to the strong
prompt 559-keV capture gamma rays that are emitted. The characteristic of
the main neutron absorbers are illustrated in fig.2.4.
As consequence of the neutron capture in 10 B , 2.8 MeV are relased, shared

Figure 2.4: Characteristic of the nuclides manly employed as neutron absorber.

18
2.2. Main components of a PGNAA facility

between an α particle and a 7 Li recoil nucleus. With a probability of 94%,


7
Li is produced in a nuclear excited state, thus it emits a 478 keV prompt
gamma. Those gamma rays arising from the shield can be further source of
background in the measurement position, nevertheless 4mm of lead are enough
to reduce the radiation intensity by half. The collimating aperture near the
neutron source is commonly fabricated of Boral, a mixture of boron carbide
and aluminum powder rolled into sheets and cladded with aluminum.
As far as lithium is concerned, the high energy triton produced in the 6 Li(n, t)4 He
reaction gives rise to about 4 -10 energetic secondary neutrons by the inter-
action with the light elements. If the flux is high, the tritium produced may
exceed the limits. To increase the absorption power, materials isotopically en-
riched by 6 Li and 10 B are sometimes used. Where secondary radiation is of
no concern, sheets of cadmium or boron-loaded polymer are also used.

2.2.3 The shutter, the sample holder and the beam stop
The function of the shutter is to turn on and off the neutron beam when the
reactor is on. It is commonly located in front of the beam port and it is designed
to stop a neutron beam already filtered, therefore mainly composed of thermal
neutrons, but eventually containing also an epithermal and a fast components.
For this reason, the shutter must be made of a thermal neutrons absorber,
commonly a thick boron- or lithium-containing material, and supplemented
by thermalising materials and epithermal absorbers.
Some precautions must be taken at the beam tube, considering the scatter-
ing reactions occurring in the air contained in it: passing through 1 m of air,
about 5% of the neutrons are scattered into the surroundings. To avoid this,
the beam tube can be either evacuated or filled with helium, or alternatively
it can be lined with 6 Li enriched material.
The beam tube ends where the sample holder begins. To center the sam-
ple in front of the beam, it can be hold in a proper position by using ma-
terials that neither absorb or scattered the neutrons. Fluorocarbons such as
TeflonTM TFE or FEP (both with empirical formula CF2 ), or TeflonTM PFA
(approximately C4 F7 O), are commonly used for this purpose because of the
low capture-cross sections of fluorine and carbon. FEP is available in monofil-
ament fibers or in thin films reducing the amount of material used to hold the
sample.The target chamber should be lined with a neutron absorber, preferably
6
Li, to shield the neutrons scattered by the target. Other materials that have
been used as sample holder are: aluminum, magnesium, vanadium, graphite,
quartz, and silicon.

The final part of the facility is the beam catcher, this must be designed
in order to reduce the neutron and gamma radiation dose rate, outside the
PGNAA, to a minimum level, according to the radiation protection laws. This
structure is mainly constituted of blocks of materials properly arranged and
cutted in order to absorb those primary neutrons that crossed the sample

19
2. The Prompt Gamma Neutron Activation Analisys

without interact. Thus, the beam stop must absorb mainly thermal, but also
fast and epithermal neutrons. From the shielding point of view, the photon
sources must also be considered, that is: the gamma rays from both the core
and from the interaction of the neutrons with the surrounding materials.

2.3 Gamma-ray detector


One of the most important parts of the facility, is the detection system of the
gamma rays. There are two different classes of detectors used for photons: in-
organic scintillators, of which the NaI(Tl) is the most popular, and germanium
semiconductor detectors. Since the PGNAA spectrum produced by a sample
has a very large variety of peaks, a high energy resolution gamma ray detector
is required. This performance is fullfilled by the semiconductors detectors, the
modern ones are made of high purity germanium (HPGe) crystals.

The most important processes in the radiation measurement point of view


are: photoelectric effect, Compton scattering and pair production [21]. Each
of these interactions become significative in a specific range of energy for the
incoming photon: several hundreds of keV for photoelectric absorption, above
5-10 MeV for pair production and a middle range for Compton scattering.
The photon energy transfer to the medium occurs trough the production of
secondary electrons in the absorbing material whose atomic number also affects
the obtained spectra.

2.3.1 The High Purity Germanium detector


In a semiconductor detector the information carriers of the radiation are the
electron-hole pairs. The higher energy resolution of a semiconductor detector
compared to a scintillator is due to the larger number of carriers for a given
incident radiation event. In case of HPGe detector, the average energy of the
primary charged particle required to create an electron-hole pair (ionization
energy) is 3 eV. This value is also temperature-dependent: for the most signif-
icant materials it increases with the decreasing temperature.
The motion of both electrons and holes in an applied field generates the elec-
tric signal for the detection of the material: this current is further processed
by dedicated electronics.
The presence of a small amount of impurities in the semiconductor, that is:
metallic atoms occupying substitutional sites within the lattice structure, gives
rise to further recombination or trapping phenomena that contribute to de-
crease the number of the carriers in the detector.
Since 70s, particular tecniques have been developed in order to decrease the
level of impurities at a value of 10−10 , by locally heating the Ge and melting
a portion of its volume. This melted material is shifted from an end to the
other, dissolving the impurities. The molten Ge is then removed carring away

20
2.3. Gamma-ray detector

the metallic atoms. Single crystals of germanium are then slowly grown from
this purified block. If the low amount of remaning impurities are donor, i.e.
with an higher number of valence electrons, then the single crystal is called
n-type, if acceptors, i.e. atoms with less valence electrons, then it is p-type.

Configuration of the germanium detector In the case of n-type detec-


tors, the outer surface is doped with acceptor atoms, typically boron, with
a thickness of a few micrometers. They have been shown to have better re-
sistance against neutron damage than p-type detectors, doped with a thicker
layer of acceptor atoms. There are two possible type of HPGe configurations:
planar and coaxial (fig.2.5). In the planar configuration the electrical contacts
are placed on the two flat surface of a single crystal disk. In the coaxial con-
figuration a cylindrical germanium crystal is provided of an electrical contact
on its outer surface, then a second contact is fabricated at the inner surface
obtained removing the core of the crystal.

Figure 2.5: Planar and close ended coaxial germanium detector configurations.

Among the two configurations the coaxial one has a larger active volume,
that is the region of the crystal able to detect the signal, for this reason it
is the one used in PGNAA. The contacts of the crystal do not belong to the
active volume, but they represent dead layers on the surface of the detector.
Some techniques for the contact formation are able to reduce the thickness to
few tenth of micrometers in a way that it does not affect the resolution of low
energy gamma rays.

Operational condition The bandgap 2 of a crystal germanium is 0.7 eV,


this small value allows the electron transition to the valence gap already at
the thermal energy. The leakage current generated by these carriers makes im-
possible the operation of a germanium detector at room-temperature. For this
reason the crystal must be cooled at 77 K by an insulated dewar containing
liquid nitrogen in contact with the detector.
The energy resolution of a germanium detector is related to three factors:
2
The energetic gap between the valence and condution band of a crystallin material is
called bandgap

21
2. The Prompt Gamma Neutron Activation Analisys

the statistical fluctuation of the number of carriers produced by the radiation


(WD ); the efficency in the charge collection (WX ); the electronic noise (WE ).
The FWHM of a peak in the spectrum due to monoenergetic gamma rays is
expressed as:

WT2 = WD2 + WX2 + WE2 (2.21)

Each of these contributions depends on the energy of the gamma radiation and
on the dimension of the crystal. WX is the most significant for large volume
crystals; wherase WD and WE incerase with the gamma-ray energy. By the
use of the Poisson model, the fluctuation in the number of charge carriers is
E/�, where � is the ionization energy and E is the gamma-ray energy. This
evaluation requires that all the events along the track of the ionizing particle
are indipendent. Nevertheless the observed fluctuation is higher than the sta-
tistical prediction. The fraction between these two value is the Fano factor F:
observed statistical variance
F = (2.22)
E/�

A small Fano factor means a better energy resolution, being this value related
to WX2 according to the formula:

WD2 = (2.35)2 F �E (2.23)

Finally, WE2 represent the broadening effect due to all the electronic component
associated to the detector.

Neutron damage The performances degradation of a germanium detector,


both Ge(Li) and HPGe, are mainly related to the fast neutron fluence.
The damages due to the interaction of the neutrons with the crystal cause,
mainly, the hole trapping within the active volume of the detector.
The effect of the fast neutrons on a germanium detector are well known and
are described in detail in [35].
The first studies performed on a thick planar germanium detector showed
that a fast neutron fluence of about 109 n/cm2 may cause a detectable reduction
of the energy resolution while, with a fluence of about 1010 n/cm2 the detector
can become useless.
In case of coaxial detectors, the neutron damage depends on the electrode
configurations. The so called conventional Ge(Li) or HPGe detector consists of
a p-type germanium where the holes carrier type are collected to the p+ inner
contact electrode, while in n-type germanium the electrode are reversed and
the holes are collected to the outer surface (reverse electrode configuration).It
was shown (fig.2.6) that a conventional germanium is 28 times more sensitive
to the fast neutrons damage [36]. Werease, a germanium detector with the

22
2.3. Gamma-ray detector

Figure 2.6: Resoultion (FWHM) for the 1332 keV as a function of the neu-
tron fluence for a conventional and reverse electrode configuration coaxial Ge
detector.

outer p+ contact showed workable energy resolution (e.g. 3 keV) to neutron


fluences < 1010 n/cm2 .

An advantage related to the use of an HPGe instead of Ge(Li) is that a


moderate damage ( caused by a fluence < 4 × 109 n/cm2 ) can be repaired by
annealing the detector, that is by rising the temperature to 120o C for 72 h.
This process allows to restore the original energy resolution of the crystal.

2.3.2 Response function of a germanium detector


Even if the resolution of a germanium detector is high, the efficency of the
photopeak is tipically an order of magnitude lower if compared to a NaI scin-
tillator detector. This is related to the small size of the available crystals and
to the low atomic number of the germanium preventing the total absorption
of the gamma-ray energy. This disadvantage is still compensated by the high
energy resolution of a germanium, which allows detecting a weak source of
discrete energy when superimposed on a broad continuum.
The response function of a detector is the shape of the energy spectrum ob-
tained from a monoenergetic gamma source. It is strictly correlated to the
shape of the germanium, its electronics, the gamma energy and the surround-
ing materials.Fig.2.7 shows the different components that may be distinguished
in a tipical spectrum. It represents the response function in case of a 24 N a
source using a germanium detector in normal and Compton-suppressed modes
(see §2.3.4).

FE Full-energy peak (FEP), when the photon loses all its energy (E) within
the active volume of the detector.

23
2. The Prompt Gamma Neutron Activation Analisys

24
Figure 2.7: Normal and Compton-suppressed spectra of N a with 1369 keV
and 2754 keV gamma energies.

S Step below the full-energy peak, when the photons lose a small amount
of energy in a small-angle Compton scattering in the collimator or in the
dead region of the detector followed by full absorption of the scattered
photon.

MC Multiple Compton scattering region below the full energy peak and above
the Compton edge.

CE Compton edge about 220 - 250 keV below the full energy peak. It is a
result of a single Compton-event and corresponds to the highest energy
left in the detector in a single scattering. The Compton edge is below
the FEP with an energy of E/(1 + E/255.5keV ), as calculated for free
electrons.

CP Compton plateau is formed, when the Compton photon leaves the detec-
tor.

SE, DE Single and double escape peaks are 511 and 1022 keV below the full en-
ergy peak of high-energy photons that can produce an electron-positron
pair. These peaks are produced when one or both annihilation photons
leave the detector.

CF Between the single escape peak and the Compton edge there are events
from single escape processes. The escaping 511 keV photons may produce
a Compton scattering before leaving the sensitive volume.

AE Between the single escape and the double escape peaks there are events
from Compton scattering of both escaping 511 keV photons, which leave
the sensitive volume. Their Compton edge is 170 keV below the single
escape peak. Sometimes a back scattered peak of one of the escaping 511

24
2.3. Gamma-ray detector

keV photons can be identified at 34 keV below the single escape peak
(too weak to be seen in Figure 2.7).
A Annihilation peak at 511 keV, when an annihilation radiation produced
in the surrounding structural material or the sample is detected.
BS Back scattering peak at 220-250 keV is produced when back-scattered
Compton-photons from the sample chamber and other structural mate-
rials in front of the detector are observed. [The exact energy is E/(1 +
E/255.5keV ).] In the special case plotted in Fig2.7 there are two back
scattering peaks according to the two gamma energies, while in the typ-
ical prompt gamma spectra it covers a wider region.
B The strongly increasing background intensity towards low energies comes
from the bremsstrahlung from photo- and Compton-electrons and also
from the β particles emitted by radioactive nuclides. The baseline at
low energy can be further increased by the multiple Compont scattering
occurring in the shielding material placed close to the detector. he slope
of this region can be especially steep, when the spectrum contains intense
high-energy components.
X Characteristic X-ray peaks also show up below 100 keV from the shielding
and other structural materials.
In case of strong pile-up 3 or random coincidences, any combinations of
the above events can happen, which produces double or higher order Compton
edges or peaks at summed energies of full energy peaks. These random coin-
cidences and peak summing must be distinguished from the true coincidence
summing of gamma rays (see next paragraph) from cascades [22].

2.3.3 Detector efficiency


In order to perform a PGNAA measurement it is important to know the count-
ing efficency of the detector as a function of the energy. It represents the
number of recorded pulses under the photopeak divided by the number of the
gamma rays emitted by the source. The counting efficency depends on the
gamma-ray energy but it is also strictly connected to the germanium geome-
try, e.i distance detector-source, presence of shielding or collimator. It can be
evaluated experimentally by the use of a gamma source with known disinte-
gration rate, applying the following formula:
Cγ Rγ
ε(Eγ ) = = (2.24)
N Pγ K APγ K

where Cγ is the measured net peak, Rγ is the count rate, N is the number of
3
The pile-up of events occurs when a second signal arrives at the amplifier within its
resolution time, i.e. within the shaping width of a single pulse

25
2. The Prompt Gamma Neutron Activation Analisys

disintegrations; dividing by the time of counting N becames A: the source ac-


tivity. Pγ is the emission probability for that specific gamma. K is a correction
factor, it takes into account factors as the dead time and the coincidence sum-
ming. During the acquisition, the coincident detection of two or more gamma
ray peaks can take place, affecting the accuracy of the measurement. The rel-
ative number of coincidence events depends on different factors involving the
source but also the geometrical caracterisitc of the germanium. Two different
kinds of coincidence can be distinguished:
• True coincidence, that is the simultaneous detection of two gamma
rays emitted in coincidence. This effect occurs when a multi-gamma
source is used. If ε is the probability to detect a gamma ray of a given
energy, ε2 is the probability to detect two of these gamma rays at the
same time. Then reducing the geometrical efficency, the true coincidence
events decrease. For this purpose the use of a collimator can help.

• A random coincidence happens when a gamma ray is detected in


coincidence with a Compton background event of another gamma ray.
Its probability involves the total efficency, that is, the probability to
detect an event at any energy value, and it can be quantified by εεtot εtot .
Beacuse of the large number of peaks in prompt gamma spectra it is
advisable to keep the detector at large distance to reduce this effect.

Intrinsic and geometrical efficency It is common to distinguish between


intrisic and geometrical efficencies:

• The intrinsic efficency is the fraction between the number of events


recorded and the number of gamma rays incident on the detector sur-
face, it depends on the interaction between the gamma photons and the
material of the detector.

• The geometrical efficency is the ratio of the number of gamma rays reach-
ing the detector to the number of gamma rays started from the source.
It can be calculated by:
r2
�geom = 2 (2.25)
4d

where r is the radius of the detector surface facing the sample and d is
the source-detector distance.

The total efficency is given by the product of the two components:

� = �intr �geom (2.26)

In case of collimated detectors there is a strong dependence of the efficiency


on the position of the sample.

26
2.3. Gamma-ray detector

Figure 2.8: Layout of the collimated detector, and transmission geometry of the
collimator (0 = the geometric center, a = sample-to-shielding distance, b = thickness
of the collimator, r = radius of the collimator, e = end of the partial illumination
region, d = the distance of the inner aperture to the sample). The thick line shows
the geometry factor, it goes for 0 to 1 in the partially illuminated zone. The hatched
circle represents the front collimator seen by the the source point. The empty circle
is the source point view of the rear aperture. (Drawing is not to scale.)

In fig.2.8 a detector surrounded by a shieding and provided of a collimator


is represented. If a sample is located within the interval between -r and r (
where r is the radius of the collimator aperture) it illuminates the detector
with the highest possible intensity. Whereas going from r to e the collimator
partially shields the detector reducing the intensity of the source until the
extreme position e, where the detector does not see the sample anymore. A
parameter called geometry factor is used to quantify this reduction in the
counting efficency due to the collimator, its value is 1 in the central region and
gradually decreases to 0 in the partially illuminated zone.

2.3.4 Compton-suppressed spectrometer


In a PGNAA facility a Compton-suppresed detector system is commonly used.
It consist of a germanium detector surrounded by one or more scintillators, the
most used are NaI(Ti) or bismuth germanate (Bi4 Ge3 O12 ). The advantage of
this setup is the reduction of the compton continua associated with all peaks
(see § 2.3.2) and it also allows a reduction of the single and double peaks in-
tensities.
When a gamma ray, due to a Compton event, transfers only part of its energy
in the detector, it creates a Compton photon that interact with the scintillator
anullus. An anti-coincidence gate prohibits the registration of that event.
The anullus needs to be shielded since it can detect the gamma rays coming
from outside.
There are two possible compton-soppressed spectrometer arrangements: coax-
ial and perpendicular (fig.2.9).
In the coaxial geometry the curved surface of the detector is fully covered by

27
2. The Prompt Gamma Neutron Activation Analisys

Figure 2.9: Coaxial and perpendicular compton-suppressed spectrometer lay-


out.

the scintillators. In case of the perpendicular setup the germanium detector


is placed at right angle with respect to the axis of the scintillator anullus. in
this configuration the prompt gamma rays cross the germanium from its side.
The use of a HPGe with an anullus of BGO in coaxial configuration was shown
to be the most efficent Compton suppression system. The response function
of a normal and compton suppressed detector is illustrated in fig.2.10

Figure 2.10: Normal and Compton-suppressed spectra of a nitrogen-containing


compound (deuterated urea).

28
2. The Prompt Gamma Neutron Activation Analisys

30
Chapter 3
Experimental boron
concentration measurement

In this chapter the experimental part of the PhD work is treated.


The evaluation of 10 B concentration in biological tissue samples is a cru-
cial issue in BNCT, and until now these measurements have been performed
in Pavia by α spectroscopy technique. The first part of the PhD work was
devoted to the optimization of the spectroscopy method for the analysis of
lung tissue containing 10 B, required for the WIDEST research. This project,
funded by INFN since 2005, has the goal to study the application of BNCT
to lung disseminated tumours, starting from boron biodistribution studies in
animal models [23]. WIDEST foresees a BNCT irradiation by external ep-
ithermal neutron beams, exploiting the low density of the pulmonary tissue
with respect to the surrounding organs. This fact allows obtaining a uniform
thermal flux distribution inside the lung, when irradiating the thorax with op-
posite epithermal collimated neutron beams [24]. Comparing to the previous
BNCT application performed in Pavia, where an entire organ was treated in
an ex-situ irradiation [25], this strategy requires the knowledge of the boron
concentration in all the tissues irradiated, including blood and bones. These
tissues are not measurable using α spectrometry because liquids cannot be
irradiated under vacuum and bones are hardly sliced using the cryostat. This
is one of the reasons why the installation of a PGNAA facility represents an
important progress of the research activity in BNCT in Pavia.
Before starting the simulations for the design of the PGNAA, many studies
were performed on animal models to evaluate boron distribution obtained in
lung, in tumor and also in other solid tissues using α spectrometry. This tech-
nique is based on the analysis of the spectra obtained irradiating thin tissue
sections facing a thin silicon detector. A part of the spectra is selected where
only α particles contributed and the boron concentration is calculated using
the integral of the spectra in the selected interval, the capture cross section,
the efficiency of the set-up and the stopping power of α particles in tissue.
This last quantity is of high importance for the accuracy of the boron con-

31
3. Experimental boron concentration measurement

centration determination, in particular no tabulated values were avalilable for


true tissues: the tables found in SRIM [26] refer to uniform slabs of materials
and do not take into account the effect of the disomogeneities in the particles
transport. Thus, the energy loss of the α particles crossing lung tissue samples
of different thickness had to be experimentally evaluated. The results of this
experiment are descibed below in a dedicated paragraph.

The course of the PhD work was almost entirely dedicated to PGNAA, from
the beam tailoring to the measurement issues related to the analysis of boron
in biological samples. To gain experience about this last point, the author of
this thesis spent a research period at the Center for Neutron Research of the
National Institute of Standars and Technologies, where two prompt gamma fa-
cilities are available: CNPGNAA and TNPGNAA, using respectively a guided
cold neutron beam and a conventional thermal neutron beam. During the
stay at NIST the author participated to different boron measurement sessions,
mainly in metallic samples embedded in epoxy. This experimental activity
allowed facing the typical issues arisen in the analysis of hydrogenous samples,
such as the biological tissues. In this chapter the measurement performed
are described together with some theoretical aspects of the prompt gamma
spectrum analysis techniques.

3.1 Boron measurement by α spectroscopy


3.1.1 Introduction
As mentioned before, the α spectroscopy technique exploits the charged parti-
cle spectrum resulting from a neutron irradiation of a sample containing 10 B
to evaluate its boron concentration. Starting from a cryogenic tissue sample,
different slices, 60 µm thick, are obtained using a Leica criostat and are de-
posited on a mylar support. After the natural drying process, the samples are
irradiated in the thermal column of the Triga Mark II reactor at LENA, where
the thermal neutron flux is approximately 2 · 109 n/(cm2 s). In particular, ten
tissues samples are positioned in a special holder that, by a remote control, is
rotated in order to locate one sample a time in front of the Si detector. In this
way, the spectrum of each sample is collected on-line during the irradiation
time. The chamber containing the samples and the detector is kept under vac-
uum during the measurement process. More details about the measurement
process can be found in [2].
A typical spectrum for a lung-tissue sample is showed in fig.3.1; it is obtained
by the convolution of the spectra from α particles and 7 Li, arisen in the neutron
capture reaction in 10 B, 1 and the protons coming from the (n,p) reaction on
14
N ; their energy is equal to 580 keV. Their contribution to the total spectrum
1
The energy of α and 7 Li are respectively 1475 keV and 840 keV with 94% of probability
and 1777 keV and 1030 keV in the remaining 6%

32
3.1. Boron measurement by α spectroscopy

Figure 3.1: Example of a typical spectrum of α particles emitted from a lung-


tissue sample. The two vertical bars indicate the energy interval used for
calculating the boron concentration.

is not gaussian but it has the tipical shape of an absorbed spectrum; this is due
to the fact that the 10 B and the 14 N are uniformly distributed in depth into
the sample. The charge particles produced during the irradiation start from
different positions, therefore they reach the detector with different residual en-
ergy. The interval of the spectrum used to calculate the boron concentration,
indicated in fig.3.1 by the vertical bars, is located in a region whose contribu-
tion comes only from the α particle, with no counts from the 7 Li, (1030 keV
maximum energy) and the protons (580 keV maximum energy). The selected
energy interval does not contain the α coming from the overall sample, but
only the ones emitted in a particular depth range (Δ(ρx)). The evaluation of
the parameter Δ(ρx) is most important to obtain the concentration (C) of 10 B
in the sample that can be calculated by:
k PA ΔE
C(ppm) = (3.1)
η · σ · φ · S NA Δ(ρx)
where η is the geometrical efficiency of the apparatus, φ is the thermal neutron
flux, σ is the cross section of the (n, α) reaction, k is the number of events per
unit energy and time detected in the selected ΔE, NA is the Avogadro number,
PA is the atomic weight of boron.
Once the ΔE is fixed, the corresonding depth range (Δ(ρx)) can be eval-
uated by using a curve of the residual energy of the α particle as a function
of the crossed tissue thickness. This curve describing the α particles energy
loss in lung tissue was experimentally evaluated [27] and, the procedure and
results are the topics of the next sections.

3.1.2 Experimental set-up


Lung tissue from syngeneic BDIX rats, not treated with boronated substances,
was used for the evaluation of the α particles energy loss. The rat lungs were

33
3. Experimental boron concentration measurement

Figure 3.2: The experimental set-up used for measuring the residual energy of
α particles after crossing lung-tissue samples of different thicknesses.

explanted, frozen in liquid nitrogen and stored at −80o C. Sections with an


area of about 1 cm2 were obtained using a Leica cryostat at a temperature
of −20o C. A series of five groups of slices were prepared for each nominal
thickness (6, 16, 20, 25, 30, 35 and 40 µm) ranging from 6 to 40 µm. After the
cutting procedure, the samples were deposited on mylar disks 23 µm thick.
The samples so prepared where put in succession inside the experimentally
apparatus sketched in fig.3.2. It consisted of a vacuum chamber in which was
located an 241 Am radioactive source electrodeposited on a platinum support,
emitting α particles with different branching ratio: 5388.23 (1.6%), 5442.8
(13.1%), 5485.56 (84.5%), 5511.47 (0.22%) and 5544.5 (0.34%). The α particles
were collimated by a 100 µm mylar ring with an internal radius of 1.2 mm
placed at 2 cm from the source, providing a beam aperture of 16.7o . The
mylar support of the sample caused the lowering of the α most probable from
5485.56 keV to about 2000 keV that is similar to the α produced in the neutron
capture in 10 B. Finally a Si solid-state detector, with an area of 50 mm2 and
a thickness of 51.5 µm, operating at 25 V, was used to collect the α particles
spectrum. Two aluminum rings 1 mm thick, were used as spacers between
the collimator and the sample and between the sample and the Si detector
complete the assemblage. Outside the vacuum chamber, to analyse the signal
from the detector, a conventional electronic chain made up of a pre-amplifier
(PA), an amplifier (A), an analog-to-digital converter (ADC), a multichannel
analyser (MCA) and a system for data acquisition (MAESTRO-Ortec) was
used.
For each tissue sample the mass per unit area was calculated (mg/cm2 ), this
parameter was used to identify the “average thickness” of the sample instead
of the nominal one, that is the thickness set in the cryostat. The reason of
this choice is due to the fact that, after the samples are cut, a rapid drying
process occurs; as a consequence, the mass and the volume of the samples
change. Instead, the adesion of the slice to the mylar support keeps their area

34
3.1. Boron measurement by α spectroscopy

unchanged, as was verified by microscopic observation.


A disomogeneity in the thickness can be observed, different from point to point
and not regular between different samples (see fig.3.3). A further contribution
to the disomogenity comes from the biological differences in the lung tissue
samples; for these reasons a value of thickness expressed in µm cannot be used
to characterize the samples.

Figure 3.3: Enlargements of different tissue slice with nominal thickness of 6, 20


and 40µm; the true dimension of the images side is 350µm. The disomogenity
of the sample in the tissue increase with increasing thickness; whereas the
number of holes is higher for thinner samples.

Therefore, for each sample the mass and the area were evaluated by using,
respectively, a precision balance (Gibertini) and a Leyca stereomicroscope;
then the average thickness in mg/cm2 was calculated; the obtained values
range from 0 to about 0.7 mg/cm2 .

3.1.3 Experimental spectra


The first spectra were collected without interposing any sample between the
source and the detector (fig.3.4 (a)) and by placing a bare mylar support in the
sample position (fig.3.4(b)). The former spectrum was fitted with three gaus-

Figure 3.4: (a) Americium source spectrum. (b) Spectrum obtained with bare
mylar.

sian curves to take into account the 3 most probable branches of the α emitted

35
3. Experimental boron concentration measurement

by the 241 Am source; (the analysis of the experimental data, the calculations
and the plotting were performed using root [28]) . The most probable peak
was used to perform the energy calibration. The second spectrum was used to
evaluate the energy of the α particles crossing the tissue sample; a gaussian fit-
ting procedure was carried out and the resulting mean energy value was equal
to 2200 ± 80 keV.
In the next step, the spectra of the tissue samples were collected and the
residual energy of the α particles was evaluated by a fitting procedure. In
fig.3.5(a) and 3.5(b) the spectra respectively for a thin sample (0.13 mg/cm2 )
for a thick sample (0.60 mg/cm2 ) are shown. Two peaks can be observed in

Figure 3.5: (a) Spectrum of a thin sample. Two peaks are clearly visible; the
one at higher energy corresponds to a particles that just crossed the mylar; (b)
thick sample

the thinner sample; the one at higher energy value comes from the α particles
that crossed only the mylar support, this can be explained by the presence of
holes in the samples (as can be seen in fig.3.3). The broadening towards lower
energies of the most probable peak was due to the point to point variability of
the sample thickness, related to the heterogeneity of the tissue, which causes
a continuous change in the energy loss of the α particles. The increasing of
tissue thickness causes a higher broadening of the peak, as can be observed in
the fig.3.6 where spectra from samples with different thicknesses are compared.
The evaluation of the α particles energy loss crossing lung tissue samples is
affected by the broadening of the peak whose entity was supposed related to
the disomogeneus nature of the sample. Some Monte Carlo calculations were
performed in order to investigate this hypotesis.

3.1.4 Monte Carlo simulation


The simulated spectra for samples of different tissues were carried out by
MCNPX code (ver.2f) [29], reproducing in detail the geometry of the experi-
mental set-up. Some preliminary simulations were performed in order to repro-
duce the radioactive source electrodeposited on the platinum support. Firstly,
it was supposed that the source was infinitely thin and completely superficial,

36
3.1. Boron measurement by α spectroscopy

Figure 3.6: Experimental spectra of α particles that crossed a 23 µm thick


bare mylar support (the first on the right) and different tissue thicknesses (0,
0.1299, 0.1795, 0.4319, 0.4554, 0.5744, 0.6032 mg/cm2 from right to left). A
broadening of the spectra and a shift of the peak towards lower energies can
be noted for increasing sample thicknesses.

the resulting spectrum, obtained without any sample, was slightly different
from the experimental one expecially in the low energy region. Successively, a
depth distribution of the Americium source was introduced and its values were
gradually changed in order to obtain a good overlapping of the simulated and
experimental spectra, as can be seen in fig.3.7. Table 3.1 gives the percentages
and the corresponding depth intervals starting from the surface.
A tissue sample with thickness similar to an experimental one was simulated;

Figure 3.7: Comparison between experimental and simulated spectra. (a) α


particles from the 241 Am source; the agreement between the experimental and
simulated spectra was obtained assuming that the americium was distributed
at different depths in the platinum support. (b) α Particles from the 241 Am
source after their passage through a 23 µm mylar foil; in the simulation the
same source as for the previous step (a) was used.

the resulting spectrum reproduced well the energy of the most probable peak,

37
3. Experimental boron concentration measurement

Depth µm 0 0 ÷ .15 .15 ÷ .28 .28 ÷ .38 .38 ÷ .68 .68 ÷ 1 1÷2 2 ÷ 10
Source percentage 62.18 19.29 7.5 3.21 3.21 2.14 1.61 .86

Table 3.1: Spatial distribution of 241 Am source into the platinum support
assumed in the MCNPX simulation to reproduce the experimental spectrum.

but not its broadening. This difference was supposed to be related to the lack
of disomogenity in the simulated sample. To verify this hypotesis a sample,
made up of different zones with a proper density, was simulated. A single par-
allelepiped of tissue with a square base of 0.01 mm2 was simulated with a fixed
thickness according to the experimental sample that was under examination.
Circular zones of density ranging from 0 (no tissue) to almost 2 g/cm3 were in-
serted into the parallelepiped fig.3.8(a). Using the MCNPX repeated-structure
capability, the modular parallelepiped was reproduced to fill the whole area of
the tissue sample (fig.3.8(b)). The densities of the different circles were grad-

Figure 3.8: The geometry of the tissue sample as reproduced in the simulation:
(left) a square base of 0.01 mm2 filled with disks of different densities; (right)
the MCNP repeated-structure capability was used to fill the whole area of the
sample.

ually changed in order to reproduce the experimental spectrum, keeping the


average thickness (0.334 mg/cm2 ) of the simulated tissue equal to the mea-
sured one. The comparison between the simulated and experimental spectra
was done for a 0.334 mg/cm2 sample, it is showed in fig.3.9; in the same figure,
the spectrum obtained for an homogeneous sample, with the same thickness,
was also overlapped; the most probable peak was the same in all the spectra.
Thus it was possible to explain the observed broadening of the spectra as a
combined effect of the straggling, the beam collimation and the disomogeneity
of the sample itself.
Three more samples, with different thicknesses were simulated and compared
with the experimental ones; after the previous study of the peak broadening,

38
3.1. Boron measurement by α spectroscopy

Figure 3.9: Comparison between the experimental and the simulated spectra
of α particles crossing a lung-tissue sample 0.334 mg/cm2 thick: experimental
spectrum (dashed line); MCNPX simulation of a heterogeneous sample (dotted
line); MCNPX simulation of a homogeneous sample 0.334 mg/cm2 thick (full
line).

only uniform samples were simulted. The results of the simulations along with
the corresponding experimental results are collected in table 3.2.

Thickness Peak energy σ (%) Peak energy σ (%)


(mg/cm2 ) (keV) (MCNPX) (experimental) (experimental)
0.098 1943 3.7 1953.2 6.6
0.334 1481 5.4 1461.6 13.0
0.656 787.2 12 802.6 53.4

Table 3.2: Results (peak position and variance) of the simulation of alpha parti-
cles traversing a homogeneous tissue sample; three thicknesses are considered.
Experimental results for non homogeneous samples are reported in the last
two columns for comparison; a good agreement is visible for the energy peak
(columns 2 and 4); the contributions to the broadening of the spectrum due
to straggling and beam collimation from that due to the lack of homogeneity
of the sample can be separated by comparing columns 3 and 5.

3.1.5 Results
The experimental and simulated values of the residual energy of α particles as
a function of the crossed thickness of lung tissue sample were plotted (fig.3.10).
The residual energy is the most probable energy peak obtained by a gaussian
fitting procedure and the error bar is the associated standard deviation. As was

39
3. Experimental boron concentration measurement

Figure 3.10: Plot of the residual energy of the a particles as a function of the
distance crossed in dry rat-lung tissue. The MCNPX simulation results are
also shown. The continuous line is the linear best fit of the experimental data.

demonstrate by the Monte Carlo simulations, the peak broadening was related
to the heterogeneity of the tissue sample, that increases with the thickness
of the sample. The continuous line of the graph is the linear best fit of the
experimental points.

Eres = 2183(±50) − 1997(±178)Δ(ρx)[mg/cm2 ] (3.2)

The data reported in fig.3.10 were used to plot the thickness as a function of
the residual energy, which was linearly fitted:

Δ(ρx)[mg/cm2 ] = (4.7 ± 0.1) · 10−4 · Eres [keV ] + (1.03 ± 0.03) (3.3)

These results are used for the evaluation of the boron concentration present in
lung tissue treated with boron. The energy interval 1100 - 1350 keV (fig.3.1),
chosen as the portion of the spectrum to be used in the boron concentration
measurement, correspond to a tissue thickness Δ(ρx) = 0.12 ± 0.02mg/cm2 to
be used in Eq.3.1.
This experiment allowed the evaluation of boron concentration in lung tis-
sue treated with boronated substances by α spectroscopy technique.

3.2 Experience at NIST


In this section the measurement carried out in June 2010 by PGNAA system
at the National Institute of Standards and Technology (NIST, Gaithersburgh,
USA) are described. Firstly, a brief introduction about NIST and its facilities is
given, with a particular attention to the thermal neutron beam prompt gamma
facility. Successively, the methods used generally for the identification and

40
3.2. Experience at NIST

quantification of the elements from a PGNAA spectrum are described, pointing


out some aspects of the measurement process that are important in case of
analysis of boron in sample containing hydrogen. Finally, the experimental
measurements performed at NIST are desctribed.

3.2.1 PGNAA facilities at NIST


NIST is a non-regulatory federal agency within the U.S. Department of Com-
merce, was founded in 1901 as the nation’s first federal physical science re-
search laboratory. In the chemistry field NIST is the major supplies industry,
academia, government, and other users with over 1200 reference materials of
the highest quality and metrological value. The NIST Center for Neutron
Research (NCNR) is provided of different facilities using thermal and cold
neutrons. Thermal neutrons instruments are located in the confinement build-
ing, these are emploied for a wide variety of applications such as maintaining
radiation and dosimetry standards, developing of radiation detectors, and per-
forming neutron activation analysis for the Food and Drug Administration. In
the Guide Hall there are seven neutron guides that transport neutrons emerg-
ing from the cold source through distances of tens of meters suppling cold
neutron to sixteen different instruments.
As was mentioned before, two PGNAA facilities are available at NIST, one
using thermal neutrons, TNPGAA, and one using cold neutrons CNPGAA.
The CNPGAA is located 41 m from the cold source, at the end of the lower
half of neutron guide NG7 of the Guide Hall. [41]. The TNPGAA is located at
vertical tube 5 (VT5) on the top of the NIST Reactor. To perform boron mea-
surements, TPGNAA facility was emplyed. The neutron beam is extraxted by
using a sealed tube starting from the D2 O reflector approximately 0.5 m from
the nearest fuel elements. It contains collimating anully properly spaced and
in the final part, just below the floor, it is located a rotating beam shutter.
A sapphire filter 5.4 cm long and 4.3 cm diameter is installed in the shutter
assembly in order to lower the fast neutrons. In the external part of the reactor
four components can be distinguished (fig.3.11), an aluminum support, a beam
tube, a sample chamber containing the sample holder, a beam stopper and the
detection system [42]. The facility provides a thermal neutron flux at the sam-
ple position of 3 · 108 n/(cm2 s). A sample holder, consisting of an aluminum
frame equipped with two sets of double monofilament Teflon PFA lines, allows
a reproducible sample positioning. Aluminum rails welded in the internal part
of the chambers are used to place the frame, in a way that the sample is sus-
pended at 45o in the middle of the chamber. The detection system consists
of a Ge detector and a Compton suppression system. The Ge detector has a
relative efficency of 40% with a peak-to-Compton ratio of 70.6, and a resolu-
tion of 2.0 keV at 1333 keV. A BGO bismuth germanate 11cm×11cm×16cm
scintillator detector is used as Compton suppression system. The spectra can
be acquired in dual mode, that is with or without Compton suppressor. A fur-
ther optimization of the measurement procedure can be obtained keeping the

41
3. Experimental boron concentration measurement

Figure 3.11: Picture of the conventional thermal neutron beam PGNAA facility
installed at the top of the NIST reactor.

sample chamber under vacuum. The sensitivity of natural B for hydrogenous


material (6.2% H mass fraction) is 1159 ± 22 cps/mg; it becomes 981± cps/mg
for non-hydrogeous matrix.

3.2.2 Quantitative analisys


Elemental identification and quantification

The evaluation of the different nuclides contained in a sample can be performed


by the gamma-ray full energy peaks identification. The peak identification is
performed by comparing the obtained spectrum with known prompt gamma
ray data. A complete list of spectra and spectroscopic data can be found in
[22]. The single and double escape peaks located at 511 and 1022 keV below
the full energy peak can be also used for the elemental identification.
If the peak presents a good signal to noise ratio the quantitative measurement

42
3.2. Experience at NIST

can be preformed. The count rate of a particular gamma ray is related to the
mass of the nuclide according to the formula (already described in chapter 2):

ργ = ε(Eγ )nσγ Φ (3.4)

ργ is the area of the peak divided by the measuring time (live time), while n is
the number of atoms, ε(Eγ ) is the counting efficency under the photoelectric
peak of the detector and Φ is the neutron flux. This expression comes out to
be more complex in the real case, consequently the absolute method for the
mass nuclide determination is not so straigthforward. Nevertheless, in case of
thin sample the formula for the count rate is simplify (2.5) and the mass of
the nuclide is directly proportional to the net peak area A as follows :

M A
m= (3.5)
NA ε(Eγ )tσγ0 Φ0

where M is the atomic weight, ε(Eγ ) is the counting efficency for that specific
gamma ray energy of the nuclide, t is the measurement time (live time), Φ0 is
the thermal equivalent neutron flux and σγ0 is the partial gamma ray produc-
tion cross section.
This method can be used only if the following conditions are satified:

• the mass of the sample is well known;

• the sample is close to the ideally thin case;

• the self-shielding and the self-absorption do not occur;

• the neutron flux is stable in time;

• the live time is evaluated very accurately and the samples position does
not change for every measurement.

A more practical quantitative analysis can be performed in two different


ways: by using a reference standard for the nuclide under study, relative
standardization method, or through an universal parameter k0 properly
tabulated for each nuclide, k0 standardization.
By comparing the spectra obtained by the irradiation of the sample and of
the standard, it is possible to get the mass of the element in the sample. The
analytical sensitivity of an element x is determined by the net peak area (Nγ,x )
divided by the irradiation time (tm ) and the mass of the nuclide (mx ):

Nγ,x /tm
sx = (3.6)
mx

43
3. Experimental boron concentration measurement

If the neutron flux is constant and the geometry conditions are unchanged,
the sensitivity is constant, then it is possible to determine the mass of the
element in the sample by:

(Nγ,x)s /tm
(mx )s = (3.7)
sx
where sx is the sensitivity evaluated by the measurement of the standard.
Nevertheless, this method requires an accurate choice of the standard whose
chemical composition and geometry should match as much as possible with the
ones of the sample under study. Otherwise, parameters like, γ-rays detection
efficency, neutron scattering and self assorption of γ-rays and neutrons can
change, with the result of a different sensitivity for the element under study.
A more robust method consists in the internal standardization, by the use of
the parameter k0 . The relative elemental content (kr,c (x) factor) is the fraction
between the analytical sensitivity of the analyte x and a comparator element
c both contained in the standard. The elements used as internal standard
comparators are commonly H, Cl, Ti, Cr, Ni, Sm, and Hg. The k factor
is independent from the experimental conditions normalizing the sensitivities
with their detection efficencies, consequently the k0 factor is :
sx /ε(Eγ,x ) θx γx σ0,x /Mx σγ,x /Mx
k0,c (x) = = = (3.8)
sc /ε(Eγ,c ) θc γc σ0,c /Mc σγ,c /Mc
where θ is the abundance of the capturing isotope , γ is the number of γ-rays
emitted per capture, σγ is the partial gamma-ray production cross section (al-
ready defined in 2.2), M is the atomic weight.
Values of k0 and their uncertainties, relative to the hydrogen capture gamma
ray at 2223 keV, have been tabulated for more than 30000 capture lines and
are available at http://www-nds.iaea.org/pgaa.
A comparison of k0 factors obtained with four different neutron beams, the
thermal and cold neutron guide beam of JRR-3M, and cold neutron guided
beam and conventional thermal neutron beam can be found in [30]. The com-
parison shows that for elements with neutron cross section following the 1/v
law, the factors agree within 10%, whereas, in case of non-1/v elements, k0
depends on the characteristics of the neutron beam.
An effective g-factor is used in order to correct k0 for the neutron beam in
case of non-1/ν nuclides:
sx /ε(Eγ,x ) ĝc σγ,x /Mx ĝc
k0,c (x) = = (3.9)
sc /ε(Eγ,c ) ĝx σγ,c /Mc ĝx
The g-factor can be obtained by the following formula:
� ∞ �
1 2E
ĝ = σ(E) · · n(E)dE (3.10)
σ0 · ν 0 0 mN
where E is the neutron energy, σ(E) is the neutron capture cross-section as
a function of the neutron energy, mN is the mass of the neutron, and n(E) is

44
3.2. Experience at NIST

the actual neutron spectrum (neutron density) function.

The relative elemental contents of the element x with respect to element,


y in the same sample can be obtained by:

Ax /εx k0,c (x)


ρx:y = (3.11)
Ay /εy k0,c (y)

where Ax,y is the specific count rate of the element. The absolute contents
cannot be obtained directly by k0 , but require the determination of at least
one element in the sample, using other analytical methods.

Doppler broadening of B peak


The high neutron capture cross section (3837 b) for 10 B, makes the boron one
of the highest sensitive element in PGNAA. The signal due to the presence
of B in the sample is a 478 keV photopeak. It is characterized by a larger
broadening compared to the other peaks. 478 keV gamma rays are the result
of the diseccitation of the recoil 7 Li∗ produced with a 93% of probability by
the (n,α) reaction on 10 B. The Q value of this reaction is 2.78 MeV, divided
in 1.47 and 0.84 MeV for α particle and 7 Li∗ respectively. The high kinetic
energy of 7 Li∗ and the lifetime of 1.05·10−13 s (comparable to the slowing down
time of 7 Li in the target matrix) cause the 478 keV gamma ray to be emitted
in flight. Thus the Doppler effect causes the peak to be broadened; its line
shape is tipically rectangular and 8-10 times wider than a normal peak. The
width and the shape of the peak depend on the matrix containing boron as
represented in fig 3.12.

Spectral interference
The high number of peak in a PGNAA spectrum, causes the spectral interfer-
ence to be one of the highes sources of error.
The doppler broadening of B peak makes it more sensitive to the problem
of the spectral interference. The peaks that could overlap in the energy region
of boron come from 11 different elements, i.e Na, Si, P, Cl, Mn, Co, Ni, Sr, Cd,
Sm, and Hg. Among these, the only important in the analysis of biological
samples is the 472 keV Na peak.
The spectral interference can be corrected by a fitting procedure or by the
reference peak method that uses other interference-free gamma ray peaks.
An example of deconvolution procedure is shown in fig.3.13 [33] where the
B peak is fitted assuming that the velocity of 7 Li is constant (v0 ), than the
Doppler shifted energy of the gamma ray is given by
v0
ΔE = E0 ( cosθ) = kcosθ (3.12)
c
45
3. Experimental boron concentration measurement

Figure 3.12: Line shapes of the doppler broadened boron peak for different
samples containing boron; (a) gaseous boron trifluoride BF3 , (b) aqueous so-
lution of boric acid H3 B03 and (c) elemental boron. For comparison a normal
peak from a stationary 24 N a nucleus (N a2 CO3 target) is shown (d).

where E0 is the γ-ray energy without Doppler shift, c the velocity of light,
and θ the recoil angle between the fly-direction and the detector axis. The
probability of recoil with angle θ is proportional to sinθdθ. The line-shape
function for the broad boron peak can be simplified as:
� π
(E − E0 + Kcosθ)2
F (E) = N exp− sinθdθ. (3.13)
0 2σ 2

The other peaks were obtained by a normal Gaussian function [34].

While, in the reference peak method, the subtraction of an interfering peak


is performed using a correction factor, that is the ratio between the interfering
peak and a reference peak obtained by a previous measurement of the pure
element.

The case of hydrogenous sample


When the sample under study contains a relatively high amount of H, as oc-
curs in biological samples, the effect of the neutron scattering in the sample
must be taken into account. It was experimentally shown that the presence of
an element with an high incoherent elastic scattering cross section ( the value
for atomic H at thermal energies is 80 b, whereas it is equal to 20.5 b for free
H.) causes an enhancement of the analytical sensitivity and also of gamma
background. This effect also depends on the sample shape. It varies with the

46
3.2. Experience at NIST

Figure 3.13: Deconvolution of the broadened 478 keV peak by computer fitting.

neutron spectrum of the beam and were sistematically studied at the thermal
and cold neutron beam PGNAA facility.

The enhancement of the elemental sensitivity for 12 elements, including


boron, in liquid H2 O/D2 O mixture matrices was studied with the conventional
thermal neutron beam at NIST facility[37]. The results showed a linearly in-
crease of the sensitivity as a function of the hydrogen density; the enhancement
(expressed as (%) per increase of weight-volume of H) observed for B was about
1.8 (%/%H) that is comparable with the others 1/v element studied, wherease
a different enhancement was observed for the non-1/v elements. This variation
in the sensitivity occurs because the scattering events contribute to decrase the
energy of the epitermal and fast component increasing the fraction of neutrons
in the thermal energy range.
The change of the elemental sensitivity was also studied as a function of
the target shape of five hydrogenous material [38]. In particular the hydro-
gen sensitivity in disk-shaped and sphere-shape of THAM (tris-hydroxymethil-
aminomethane) was evaluated, the result is represented in fig.3.14.

Using 1,27 cm diameter disk-shaped samples of different thickness ranging


from 2 to 10 mm, placed at an angle of 45o to the neutron beam direction (flux
equal to 3.3 · 108 ), it was proved that the sensitivity increases with decreasing
thickness. The trend is reversed at thickness below about 3 mm. However,

47
3. Experimental boron concentration measurement

Figure 3.14: Hydrogen sensitivities for disk shaped samples (void square) and
for sphere (full circle) of THAM

the H sensitivity for spherical samples of different size remain constant within
the measurement uncertainty. Monte Carlo calculations [39] showed that the
scattering matrix causes a change in the neutron path lenght; depending on
the size and the shape of the target, this change in neutron path lenght may
increase or decrease the probability of neutron absorption, thus decreasing or
increasing the elemental sensitivity. This effect disappears in spherical targets
and it is relevant only for relatively small samples. Similar experiments were
also carried out with a cold neutron beam [40]. In order to minimize the errors
due to the neutron scattering, when a relative standardization is applied, it is
important to chose a standard with shape, size and H concentration as close
as possible to the sample.
The hydrogen content in the sample causes also an enhancement of the
background counting rate, due to the interactions of the neutron scattered
by the sample into the structure of the apparatus. For instance gamma rays
coming from the interactions with: H, B, C, Pb ( in the shielding materials);
N (in the liquid nitrogen); Al,Cl and Fe ( in the support structure); and N and
Ar (in the air) can be detected in the spectrum background.

3.2.3 Experimental measurements performed at NIST


Eight magnet samples (MAG1-#, MAG2-#) embedded in epoxy, of known
masses, were irradiated for an interval time between 45 and 75 minutes. The
samples had been previously packaged in Teflon-PFA bags and mounted in the
aluminum frame by the 2 couples of teflon filaments, some knots were used to

48
3.2. Experience at NIST

locate the sample always at the same positions. A blank sample and a stan-
dard boron sample containing 1.09 mg of boron were irradiate to evaluate the
boron content in the samples. Considering the magnet sample composition, i.e.
N d2 F e14 B, the mass of the magnet was determined for every sample, by the
estimation of the Nd mass from PGNAA analysis. The spectra were collected
in Compton suppression mode, without evaquating the sample chamber. The
amount of boron was obtained by dividing the net count rate of the 478 keV
peak by the analytical sensitivity obtained from the natural boron standard
applying the formula 3.7. In the evaluation of the boron net count rate, results
from blank were used for a more precise estimation of the background count
rate. The spectra obtained by the irradiation of the boron standard and the
blank are represented in fig.3.15.

Figure 3.15: The spectra from the irradiation of the boron standard (black)
and the blank (green).

For each sample the net area under a full-energy-peak was determined by the
Program SUM developed at NIST [43]. The program requires as input the
peak boudaries channels L and R, (left and right respectively) and the width
of the baseline on left and right sides (BL and BR) (see fig.3.16). Then the
program calculates the centroid, the background, the net area, the FWHM,
the energy, the count rate, the percent error of the net, and the Currie quality
indicator [32].

The net area under the peak is determined subtracting from the gross area a
trapezoidal baseline evaluated by averaging the BL and BR channels to the
left and right of L and R respectively, placing these averages at the positions

49
3. Experimental boron concentration measurement

Figure 3.16: Evaluation of the net area by SUM program. The input param-
eter, that is, peak boundaries (L and R) and widths of the baseline (BL and
BR), are shown.

L and R:
� L−1 R+BR

R−L+1 1 � 1 �
BKGN D = Y (C) + Y (C) (3.14)
2 BL C=L−BL BR C=R+1

where Y (C) are the counts for that channel.

Background Correction A correction was applied taking into account the


amount of hydrogen present in the sample.
The presence of H in the sample causes an increase of the background, mainly in
correspondence of the peaks originate by the elements present in the apparatus.
This enhancement is due to the interaction of the scattered neutrons by the
sample in the surrounding material. The background spectra obtained in the
absence of target and with the blank are shown in fig. 3.17. A blank, with
similar scattering characteristics in terms of shape and H content (consisting
of a paraffine sphere), was used to correct the net count rate under the boron
peak.
The boron background count rate of the sample due to its hydrogen con-
tent, is determined as a function of the count rate under the H and B peaks in
the blank spectrum. Indicating with the subscript S and W the sample and the
blank respectively, this contribution can be evaluated in cps by the following:

(NB,w − NB,bkg )
BkgH = NH,s × (3.15)
NH,w

50
3.2. Experience at NIST

Figure 3.17: Comparison between the spectrum background (black) and the
spectrum with the blank (green); the two spectra, plotted in cps, are almost
overlapped in the energy region higher than 2.2 MeV, instead, for lower energy
the Compton contribution from the 2.2 MeV peak appears in the blank sample.

where NB,bkg is the count rate of background under the boron peak without
any sample.
Thus the final net area of boron in cps is obtained by:

ℵB = NB,s − (NB,bkg + BkgH ) (3.16)

Results and discussion As mentioned before, the boron sensitivity allowed


the evaluation of natural boron mass present in each sample. Among the eight
samples, four showed an higher amount of natural boron compared to the
other elements under study. The spectrum resulting from the irradiation of
a sample containing boron in higher amount is represented in fig.3.18. The
aim of these measurements was the identification of the samples containing
boron with higher percentage. The results obtained are summarize in table
3.3. Even if this experiment did not measure biological samples, the issue of
boron concentration determination in hydrogenous matrix were similar. The
spectra showed also a Nd peak. The mean count rate fraction B/Nd, obtained
considering only the samples containing boron in higher amount, was used to
evaluate the sensitivity for Nd by the formula:
m B · SB
SN d = (3.17)
mN d · (cpsB)/(cpsN d)

where mB and mN d are the mass fraction respectively 26, 7% for Nd and 1%
for B, deduced by the chemical formula of the magnet, that is N d2 F e14 B. The
sensitivity previously determined was used to obtain the Nd mass, thus the

51
3. Experimental boron concentration measurement

Figure 3.18: Spectra from the irradiation of a higher mass boron contaning
sample compared to the other magnets under study. The live time of irradi-
ation is about 1 hour. The overlapped spectrum represents an enargement in
the boron energy peak region; the doppler broadened FEP of boron and the
618 keV Nd peak are shown.

mass of the magnet was consequently determined being on average equal to 1


mg. The sensitivity for H was used to determine the % in mass of hydrogen
contained in the epoxy of the different samples, the masses of the samples were
known. Its values range between 9 and 10% (see table 3.3).

FileName mg B mg Nd mg magnet %H
MAG1-3 0.35 · 10−3 ± 0.05 · 10−3 0.283±0.010 1.06 9.8%
MAG1-4 0.17 · 10−3 ± 0.04 · 10−3 0.299±0.010 1.12 9.5%
MAG1-5 0.25 · 10−3 ± 0.04 · 10−3 0.233±0.009 0.87 9.4%
MAG1-6 9.05 · 10−3 ± 0.04 · 10−3 0.365±0.007 1.37 9.1%
MAG2-3 10.16 · 10−3 ± 0.07 · 10−3 0.261±0.008 0.98 8.9%
MAG2-4 8.78 · 10−3 ± 0.06 · 10−3 0.229±0.007 0.86 8.7%
MAG2-5 15.59 · 10−3 ± 0.09 · 10−3 0.421±0.010 1.58 9.4%
MAG2-6 12.84 · 10−3 ± 0.09 · 10−3 0.360±0.010 1.35 9.2%

Table 3.3: Results of the PGNAA analisys of the magnet samples embedded
on epoxy; the masses of the boron, neodimium, and hydrogen contained in the
sample were derived, along with the mass of the N d2 F e14 B magnet.

The values of sensitivity for boron, hydrogen and neodimium obtained by the
PGNAA analisys and used for mass determination are collected in table:3.4.

52
Nuclide Sensitivity (cps)/mg
B 1101.6
H 2.23
Nd 3.53

Table 3.4: Sensitivity for the different element B, H, and Nd obtained from
the irradiation of respectively a boron standard, a blank and from the sample
itself.
3. Experimental boron concentration measurement

54
Chapter 4
The design of a PGNAA facility
at Triga Mark II reactor

4.1 Introduction
The implementation of a new measurement method for the evaluation of boron
concentration, based on prompt gamma spectroscopy, represents an important
step forward in BNCT studies. Indeed the new facility not only allows the
analysis of biological sample in liquid form, not feasible at the moment with
the available techniques, but also it provides a quick and accurate assessment
of the 10 B in blood useful in clinical phase for the correct estimation of the
delivered dose in the BNCT irradiation.
In this chapter the design of the PGNAA facility is dealt with. The first
part is dedicated to the Monte Carlo N Particle (MCNP) simulation code used
to develop the project, whereas in the following part the different components
of the facility are descried. In the final part, the performace capability of the
designed PGNAA are explained.

4.2 Materials and method


4.2.1 The Triga Mark II reactor
The Laboratory of Applied Nuclear Energy (LENA) of the University of Pavia
is equipped with a TRIGA MARK II research reactor. It operates at a max-
imum power of 250 kW. The Triga Mark II is made up of a light water tank
6.25m deep and with a diameter 2m where, next to the bottom, a graphite-
reflected core is located. 80 fuel elements, made up of a U (20% 235-U enriched)
and HZr Zirconium hydride mixture, are hold in the core. The tank is sur-
rounded by an ordinary concrete structure. It contains a thermal column and
4 horizontal channels; two of them, named A and B, are radial beam ports
that stop at the outer edge of the graphite reflector. Another channel named

55
4. The design of a PGNAA facility at Triga Mark II reactor

C is tangent to the outer edge of the core and a fourth, D, is a radial piercing
beam port that crosses the graphite reflector and stops in front of the core.

Among these channels, the one used to install the prompt gamma facility
is B; the graphite reflector thermalizes neutrons, making the epithermal and
fast components lower if compared to the others channels.

4.2.2 MCNP

The design of the PGNAA facility was achieved using a Monte Carlo particles
transport code MCNP, in the version MCNP-4c2 [44]. It is developed in the
Los Alamos National Laboratory and it can transport neutron, gamma rays
and electrons either in single or in coupled neutron/photon/electron mode.
MCNP simulates the path of each particles from the source to their death
(absorption, escape,...) and recording some important aspects (tally) specified
by the user. The average behaviour of the physical system can be deduced by
the average behaviour of the simulated particles.
The use of MCNP requires a deep knowledge of the source, the geometry of
the problem, the composition of the materials used, as well as the physical
data such as the cross sections of the different interactions radiation-material
involved in the problem.
In order to design the different components constituting a PGNAA system, an
input file describing the core and the geometry of the reactor, already built
and validated in Pavia [47], was used. In the different simulations each element
of the system was inserted in the geometry and proper tallies were chosen in
order to evaluate the suitability of the proposed solution for that particular
tasks (neutron filter, beam collimation, beam catcher design..). Moreover, the
use of non analogue tecniques1 was necessary in order to reach a good statistic
despite the long distance and the shielding material between the source and
the tally position. For the same reason, the shifting of the source from the core
to a surface close to the beam port was necessary both for the design of the
beam catcher and in the simulation of the response function of the germanium
detector.

A MCNP input file is divided in 3 main blocks so called: 1) Cell cards;


2)Surface cards ; 3) Data cards. The first two correspond to the geometry
definition, while the block 3 contains all the information related to the spec-
ification of the source, the definition of the materials, the cross sections, the
tallies and the variance reductions. Each of these specifications are introduced
by a proper “card“, those used for this thesis will be briefly described.

1
see the last paragraph of this section

56
4.2. Materials and method

Data card specification


Tallies
In a MCNP input file the tallies are the information that a user wants to obtain
by Monte Carlo calculation, such as current across a surface, flux averaged
over a surface. ecc. These are introduced by a specific card Fn:P ; where n
is a number identifing the type of information request and P is the type of
particle, e.i. N for neutron, P for photon and E for electron. There is a total
of 8 different tally types available, these can be combined with other option
cards to get more detailed information for that tally. The result of the tally is
by default normalized to the number of particles from the source, in this way
the result is indipendent from the source intensity. A more detaled description
of the tally type and options can be found in [44].
The tallies expolited for this work are: F1, current integrated over a surface;
F4, flux averaged over a cell, F6, energy deposition averaged over a cell and
F8, energy distribution of pulses created in a cell.
In particular, F1:N, that is applied for neutrons, was used in the shift
source operation. It gives the number of particles crossing a surface, it was
combined with the option cards En and Cn to get the energy and angle neutron
distribution. Cn (usable only for F1 tally) represents the angle defined by
the fly direction respect to the positive normal to the surface crossed by the
particle.
The tally F4 is, the most used in this work. It gives the average flux in
a cell, (particles/cm2 ) by calculating the average track lenght into the cell Tl ,
the track lenght density is related to the flux by the formula:
� � � � � �
dV dV
ϕ(�r, E, t)dEdt = N (�r, E, t)dEds (4.1)
V E t V V s E V
obtained by the flux definition ϕ(�r, E, t)dE = vN (�r, E, t) where N (�r, E, t) is
the particle density and v is their velocity, whereas the N (�r, E, t)dEds is the
track density. The estimation of the integral 4.1 is obtained by summing the
contributions W Tl /V related to the particles tracks lenght in to the cell; the
parameters W and V indicate respectively the particle weight and the volume.
By this definition of the flux, a rapid convergence of the result can be achieved
considering the high number of simulated tracks.
The evaluation of the energy deposition in a cell (F6) was highly used in
the design of the shielding in order to calculate the dose rate outside the walls
of the facility. The unit of measurement of this tally is MeV/g, that with a
proper FM (factor multipier) card can be converted in µGy/h.
The F6 tally can be used for photon and neutron only; all the energy trans-
ferred to electrons is assumed to be deposited locally. When a more detailed
knowledge of the photon and electron energy deposition is requred, the use of
a F8 tally is more appropriate. The outcome of a F6 tally can be obtained
also by appling the F4 tally (track length estimate of cell flux) with energy-

57
4. The design of a PGNAA facility at Triga Mark II reactor

dependent multipliers, FMn card. The FM card required the specification of


three different parameters: a normalization contant (C), in this case used to
used to convert the F4 result from particle/cm2 to MeV/g; the material num-
ber as already identified on the Mn card; and the number or, list of numbers,
identifing a particular reaction, that are used by the program to select the
proper cross section from the libray. The use of F4 combined with FMn card
to get the energy deposition in a cell, can be useful, for example, to distinguish
the dose contributions due to the different involved processes.

The pulse height tally (F8) was used to simulate the germanium detector
response, it provides the distribution of the pulse energy deposition occurring
in a cell. On the contrary of the other tallies, F8 records the energy deposited in
the detector by the single physical particles adding one score in the correspond-
ing energy bin. The pulse tally is an inherently analogue process, it does not
work well with neutrons and cannot be used with any variance reduction other
than source biasing. The sintax F8:P, is equivalent to F8:E and F8:P,E; that is,
there is not discrimation in the energy deposited by photons or electrons. F8
tally was used with a ”mode n,p“ that is with a coupled transport of neutrons
and photons omitting the electron transport. Moreover for the transport of
photons with energy lower than 100 MeV a ”detailed physics“ mode was cho-
sen, e.i. scattering Thomson and fluorescent photon production were included.
As concerning the simulation of the electron produced in photon interaction,
MCNP provides 3 different model, the ”thick target brehmmstralung” (TTB)
model was chosen. It works with mode e turned off, this model generates elec-
trons, but assumes that they travel in the direction of the incident photon and
that they are immediately annihilated. Any bremsstrahlung photons produced
by the non transported electrons are then banked for later transport. Thus
electron-induced photons are not neglected, but the expensive electron transport
step is omitted. This detail about the photon transport modality can be set
by the use of the PHYS:P data card.

Neutron and gamma source definitions


For the first part of the project, including: channel choice, filtering of the
neutron beam and implementation of a collimator, the source of neutrons and
gamma was the reactor core. The simulations were performed in single trans-
port mode, that is neutrons and photons separately. Neutrons are sampled
isotropically from the fuel elements with isotropic fly directions. The energy
was extracted according to the Watt spectrum with the probability distribu-
tion function:

p(E) = Ce−E/a · sinh(bE)1/2 (4.2)

where C is a normalization factor, a and b are two parameters related to the

58
4.2. Materials and method

material where the fission reaction take place.

Concering the simulations with the gamma transport from the core, some
preliminary consideration must be done. The reactor gamma rays are gener-
ated by three different processes occurring in the core:
• fission prompt γ, occurring 5 · 10−8 s after a nuclear fission takes place;

• γ rays emitted from the fission fragments;

• γ rays resulting from interactions of the neutrons with the surrounding


materials.
The intensity of the prompt γ component is higher compared to the others;
its value is 5.4 · 1016 s−1 , considering that 7-8 gamma rays are produced for
each fission reaction, the numbers of fissions occurring in the core at a power
of 250kW is equal to 0.76 · 1016 s−1 . The energy spectrum of the prompt γ is
represented by a negative exponential with energy values ranging between 0.2
MeV and 7.6 MeV [45]:

Φ = 6.67 × 1016 e−E M eV −1 s−1 (4.3)

with an average energy equal to 1.2 MeV.

The simulations with photons were performed implementing a γ source


that reproduce the prompt γ component generated by the fission production.
In particular, the γ rays were extracted from the fuel elements of the core
with the same spatial and fly direction distribution used for neutron. Whereas
the starting energy was sampled according to a decreasing exponential. This
source was used to evaluate the gamma attenuation obtained by the filters
added in the channel.

The design of the beam catcher and of the other PGNAA components lo-
cated outside the reactor biological shield, required the shift of the neutron
source from the core to a surface close to the end of the channel. This opera-
tion was mandatory because the neutron transport from the core through the
channel, almost 350 cm long, and through a shielding structure would not have
ensured sufficient statistc with the use of non analogue techinques. In practice
this operation consisted of a first run (with the source in the core) in which all
the information, e.i. current, energy and fly direction, of the neutron passing
through a surface are collected with high accuracy. Then, starting from these
results a new surface source can be implemented, shutting down the first one
and specifing the new probability distribution functions for energy, fly direc-
tions and position of the sampled neutrons. This new source was validated by
comparing the average neutron flux in a specific cell, using the original source
and the new one.
Then, with this procedure a surface source located 30 cm before the exit of

59
4. The design of a PGNAA facility at Triga Mark II reactor

the B channel, was used in order to design the beam catcher. The neutrons
are sampled uniformly from a circle with 10 cm of radius; the cosine of the
angle between the fly direction of the neutron and the normal to the surface
(forwards), is samples according to 3 intervals whose limits are 0, 0.98481,
0.99619, 1, each with its own probability and with its own energy spectrum.
The neutron energy ranges from 0 to 14 MeV divided by 7 different energy
bins, according to the following bounds: 0, 0.4 eV, 10 keV, 500 keV, 1 MeV,
2.5 MeV, 14 MeV.
The neutron spectra in corresponence of the source was evaluated in absence
of filters in the channel, with the aim to be conservative in the design of the
shielding structure. The validation of the source by comparing the average
neutron flux in cells located at the end of the channel, respectively at 346 cm
(cell 462) and and 352 cm (cell 463) from the center of the core are shown in
fig. 4.1. The surface source gives a neutron flux higher for each energy bin that
results, on average, increased of about 10% respect to the flux obtained with
the source in the core. This source, althought does not reproduce precisely
the results obtained with the source in the core, it was nevertheless considered
appropriate to design the beam cather since its slight overestimation of the
neutron flux can result in a further conservative approximation.
A further shift of the source from the core to a surface, whose normal was
on the beam axis, was necessary to evaluate the response function of the de-
tector. The new neutron source was located close to the measurement position
(about 21 cm distance source-sample) and to the germanium detector, it was
determined considering the effects of the filters and the collimator to the neu-
tron spectrum. To get reliable results in this geometry, the simulations with
the source in the core was performed applying a non-analogue technique, Dx-
tran sphere (see forward); a Dxtran sphere enclosing the tally was implemented
in order to bias the particles transport in a partially deterministic way that
will be described later. The new neutron source was implemented sampling
uniformly the neutron position in a circle with 2.12 cm radius corresponding
to the umbra section defined by the collimator and sampling the fly direction
with a uniformly angular distribution between 0 and 1.7 degrees corresponding
to the beam divergence in that position. The neutron energy ranges between 0
and 10 eV and it is sampled from a histogram with 25 energy bins. The source
was validated by the evaluation of the total number of (n,α) reactions in a pure
10
B sample (combining a F4 with a FMn card) located at the measurement
position, using the source starting from the core and the superficial one.

Thermal treatment
Some of the materials used for this work were defined in MCNP input file along
with their S(α,β) thermal treatment, by means of MT card, most important
when the thermal neutrons are transported.
The interactions of thermal neutrons with the materials are affected by
the thermal motion of the atom, that, if neglected, can be responsible of sig-

60
4.2. Materials and method

Figure 4.1: Comparison of the neutron flux (n/(cm2 s)) in cells 462 (up) and
463 (down), located respectively at 346 and 356 cm from the core, using the
original source and the new one.

nificative error. To take into account of this effect MCNP uses normally a
thermal treatment based on free gas approximation. Anyway this approach is
not always satisfactory, therefore specific S(α, β) cross section data for energies
neutrons lower than 4eV, are available in MCNP libraries, to take into account
the effects of the chemical bound of the crystal structure. These cross section
are available in MCNP only for a small number of material, but it is possible
to add the neeeded library as we did in this work for bimuth and sapphire.

Variance Reduction Techniques


Another important feature of MCNP is the availability of many variance re-
duction techniques, widely used in this work.
The results of a tally are reported along with the estimated
√ relative error
R, for a well-behaved tally, R will be proportional to 1/ N , where N is the

61
4. The design of a PGNAA facility at Triga Mark II reactor

number of histories. The statistically significance of the tally results can be


evaluated by this number; if it is < 0.10 the result can be consider generally
reliable (except for point detector) based on MCNP manual guideline. The R
is itself an estimation and it is indicated at 1 sigma level. √
The number of histories is related to computer time (T), thus R = C/ T
where C is a positive constant.
Two different strategies can be employed to reduce R: increasing the com-
puter time or reducing C. To achieve this last purpose the non-analog transport
must be performed, that is, the effectviness of the caluclations is improved by
applying variance reduction techniques. When no variance reduction technique
is used, the transport is analog, and it reproduces the physical situation: at
each event, the probability density function describing the interaction is sam-
pled and the random walk is thus simulated. Anyway, when a high amount
of absorbing material is interposed between the source and the tally, such as
in shielding calculations, or when the geometry is unfavorable (long distance
source-tally, small tally region etc.), the number of particles scoring the tally is
too low. In this situation (”high attenuation” or ”deep penetration” problems),
it is necessary to artificially increase the probability that the particles reach
the tally position, without affecting the tally result. The variance reduction
techniques act in a way that the most important particles are followed more
than the others, achieving a better statistic for the tally, and the result are
then normalized so that no bias is introduced. To ensure the reliability of the
tally, MCNP assigns to each particle a weight, that is equal to 1 in analog
calculations. In case of low scoring, it is possible to split one particle in more
histories, each one following a different random walk, but with a lower weight
(the weight of each new particle is obtaining by dividing the weight of the
parent particle by the number of splittings). All these new particles will score
the tally, allowing to lower the relative error because N is higher, but the re-
sult is re-normalized by the particle weight. This mechanism ensures that the
results of the analog and non analog transport are the same, but with a better
efficiency in the latter case.
Different Variance Reduction techniques are available in MCNP (”ranging
from the trivial to the esotheric”, as the manual states), that can be applied to
obtain reliable simulations in reasonable calculation times. Following, a brief
description of the techniques used in this work is reported.

• Geometry Splitting/Roussian Roulette: this technique is based on the


classification of the geometry according to the importance for the calcu-
lation that is performed. For example, all the regions between the source
and the tally should be privileged with respect to peripheral zones. A
parameter called Importance is thus assigned by the user to each cell of
the geometry indicating how the neutrons must be splitted to increase
the number of tracks scoring the tally. When a particle enters a cell
with a higher importance, it is splitted in a number of new particles
equal to the ratio between the importance of the cell that is entering

62
4.2. Materials and method

and the one of the cell that is leaving. Each particle will have a lower
weight, as explained before. On the contrary, when a particles enters a
cell with lower importance, it will undergo the Russian Roulette play:
it will survive with a probability equal to the ratio of the importances
with an increased weight, otherwise it will be killed. This play ensures
that for high statistis the total weight is conserved and that less particles
(with higher weight) will be transported in less important regions of the
problem. Obviously the user must pay attention in defying the impor-
tant cells, not excluding regions that could play a role for the tally score,
as for example regions which are far from the tally but could scatter
neutrons to that cell.

• Cell-based Weight Windows: this technique is a space-energy-dependent


splitting and Russian roulette play. For each cell, the user defines a lower
weight bound WL and an upper weight bound WU as a multiple of the
former. These limits identify a windows of acceptable particle weights.
If a particle reach a cell with a weight below WL , Roussian roulette is
played, and the particle survives with a weight within the window or it
is killed. If the particle has a weight above WU , the splitting is applied
so that more particles are generated with weights within the windows.
Nothing occurs if the particle weight is already within the window (fig
4.2). Besides the weights limits of the window, the user also defines the

Figure 4.2: Sketch of the weight particles assignment by the cell-based Weight
Windows variance reduction technique.

surviving weight in case of Russian roulette play. This technique helps


in performing more efficient calculations because it optimizes the weight

63
4. The design of a PGNAA facility at Triga Mark II reactor

changes along the transport, in comparison with simple splitting that


acts on the weight only according to the position in the geometry. This
fact avoids time expensive weight fluctuations. Moreover, the weight
windows can be also energy-dependent, allowing to act separately for
different energy bins defined by the user. This feature is useful when
only few particles score the tallies in an energy bin and the associated
relative error would be unaccetable for that portion of the spectrum.
It is often difficult to set manually the weight windows, this require a deep
knowledge of MCNP and of the transport in the problem under study.
MCNP provides a Weight Windows Generator, that, starting from the
importances assigned to each cell and optimizing the calculation for a
tally specified in the problem, builds a list of WL . This first result can
be further improved by a recursive procedure: the WL are used with the
generator turned on and a new list can be thus obtained.
The last releases of MCNP also offer the possibility to use a mesh-based
weight windows set, that is, a list of values for the weight windows defines
in a superimposed mesh grid. This is very useful because the geometry
can be defined in less cells regardless the variance reduction. Anyway,
this technique requires more expertise and in this work it was not used.
Weight windows were employed in the PGNAA simulations, especially
when the particles spectra were evaluated outside the facility shielding,
being the source of neutrons placed in the reactor core.

• DXTRAN: when a small region is being inadequately sampled because


particles have a very small probability of scattering toward it, the user
can specify a DXTRAN sphere that encloses the small region. When a
collision occurs outside the sphere, or when a particle that comes from
the source reaches the sphere, DXTRAN creates a “DXTRAN particle“.
This particle is deterministically scattered toward the sphere and trans-
ported without collision to its surface. The collision itself is otherwise
treated normally, producing a non-DXTRAN particle that is sampled in
the normal way, with no reduction in weight. The non-DXTRAN par-
ticle is killed if it tries to enter the DXTRAN sphere. DXTRAN uses a
combination of splitting, Russian roulette, and sampling from a nonana-
log probability density function. The details about the treatment of the
weights and the scattering angles can be found in [46].

4.3 The facility design


In this chapter the different steps followed in order to design the PGNAA
facility to be implemented in the Triga MarK II reactor in Pavia, are described.
The overall project was carried out starting from the PGNAA requirements for
a particular component of the facility, then, different possible solutions were

64
4.3. The facility design

proposed (as indicated in the PGNAA facility literature) and finally, the choice
of the most suitable solution was carried out by the Monte Carlo simulations
outcome and feasibility and costs considerations. Even if no experimental
validation was carried out until now, the accuracy of the physics result is
supported, not only by the use of a simulated neutron source already validated
[47], but also by the conservative criteria applied during the Monte Carlo
definition problem.

4.3.1 The Channel choice and its characterization

Figure 4.3: Top view of the Triga MarK II reactor in Pavia

In fig.4.3 the horizontal channels of the Triga Mark II reactor are illustrated.
The choice of the channel was based on the evaluation of the neutron spectra
at the beam port obtained for each channel in the original configuation. The
channel A faces directly the core, thus its neutron flux is higher compared to the
others but a moderator would be required to slow down the high fast neutron
component, moreover the γ rays flux coming from the core is also high. Similar
considerations can be done for the channel D, but with the difference that it is
possible to insert a block of material, or a sample, close to the outer edge of the
core; on the contrary, in the channel A an aluminum layer stops this access. At
the moment the channel D is employed for another experiment. The channel
C is the tangential one, it has the advantage to provide a beam with a low γ
rays contamination that is desirable for a PGNAA neutron beam, nevertheless
the thermal neutron flux is too low. Finally the channel B, characterized by
the precence of 30 cm of graphite was considered; the reflector provides a first

65
4. The design of a PGNAA facility at Triga Mark II reactor

attenuation of the reactor γ ray flux. By the use of F4 tally, the neutron flux,
for different energy bins, at the A, B and C beam port were simulated by
appling the geometry splitting/Russian roulette (GS/RR) to calculate a set of
cell-based weight windows. This variance reduction tecnique was mandatory
because of the long distance (almost 350 cm between the centre of the core and
the end of the channel) that the neutrons had to be cover to reach the tally
cell. The results of the Monte Carlo simulations, are summarized in table 4.1;
the corresponding calculated relative error is always < 10%.

Energy (MeV) A B C
< 2 · 10−7 3.79 · 108 1.07 · 108 2.61 · 108
2 · 10−7 ÷ 5 · 10−6 1.05 · 108 6.11 · 106 5.58 · 107
5 · 10−6 ÷ 5 · 10−1 4.69 · 108 1.47 · 107 2.38 · 108
5 · 10−1 ÷ 20 4.51 · 108 6.71 · 106 5.96 · 107
total 1.40 · 109 1.35 · 108 6.14 · 108
ϕth /ϕf ast 0.37 3.82 0.74

Table 4.1: Simulated neutron flux (n/(cm2 s)) at the beam port respectively
for channel A, B and C.

As is visible in table 4.1 the channel B provided the highest termal to


fast neutron ratio. The channels A and C are characterized by an higher
thermal flux, however, they were rejected because any solution used to stop the
fast component would inevitably cause a reduction of the thermal neutrons.
For these reasons, the channel B was chosen to install the facility, thus its
characterization and optimization were further carry out.
A more detailed F4 tally was requested at the the end of the channel B and
the spectrum shown in fig.4.4 was obtained. The typical features of a neutron
beam from a reactor can be recognized: a Maxwellian part in the thermal
region (meV), an epithermal slowing down region (� keV) and a fission fast
component in the high energy region (MeV) (the fits were perfomed by ROOT
[28]).
In particular, the energy region lower than 0.4 eV was fitted with the function
2πn � 2 �1/2 −E/kT
ϕM (E) = Ee (4.4)
(mkT )3/2 m
where m is the neutron mass, k is the Boltzmann constant, whereas n and T ,
respectively the neutron density and the temperature in K degree, were setted
as parameters of the fitting procedure. The interval from 0.4 eV to 0.5 MeV
was fitted with:
C1
ϕ(E) = p (4.5)
E
where C1 and p were setted as parameters. Finally the range between 0.5 MeV
and 15 MeV was fitted with the Watt spectrum:
ϕ(E) = C2 e−E/a · sinh(2E)1/2 (4.6)

66
4.3. The facility design

where the constant C2 and a were setted as fitting parameters.


The values assigned to the parameters by the fitting procedure are summarized
in Table 4.2.
Parameter Value
n 6.65 · 1017 ± 6 · 1015
T 335.8 ± 2.7
C1 1.00 · 106 ± 4 · 104
p 1.030 ± 7 · 10−3
C2 2.43 · 106 ± 1.8 · 105
a 2.64 ± 0.13

Table 4.2: Results of the fitting procedure of the simulated neutron spectrum
at the end of the channel B.

Figure 4.4: Neutron spectra at the beam port B with the corresponding fits.

4.3.2 Filtering of the neutron beam


The neutron beam provided by the B channel must be properly filtered in or-
der to get the characteristics required for a PGNAA facility ( § 2.2.1).

Filter for γ
In order to stop the reactor core γ rays, the addition of a proper shield in the
channel was studied. A poly-cristal bismuth block was chosen due to its high γ

67
4. The design of a PGNAA facility at Triga Mark II reactor

mass attenuation coefficent (Fig 4.3) and to its lower thermal absorption cross
section (0.034 b) compared, for example, to the one for lead (0.17 b) which
also stops effectively the γ rays. In Fig.4.6 the thermal neutron corss section
for Bi and Pb are comapred; the value for the total neutron cross sections at
the thermal energies are respectively 9.3 and 11.3 barns.

Figure 4.5: Values of bismuth mass attenuation coefficient, µ/ρ, and mass
energy-absorption coefficient, µen /ρ, as a function of photon energy.

Different simulations were performed in order to study the effect on neu-


trons and γ due to the bismuth shield. A bismuth block, 10 cm thick, was
simulated in the MCNP geometry, changing its position with respect to the
external graphite edge (fig.4.7, bottom); the distance was varied from 0 to 50
cm. Simulations in neutron mode, showed a lowering of the thermal neutron
flux at the beam port as the distance of the bismuth from the core increases.
The values of the thermal neutron flux range from 5.2 · 107 , when the bismuth
block was placed close to the graphite, to 1.2 · 107 n/(cm2 s) when the bismuth
block is 50 cm far from the graphite edge (see fig.4.7).
Inspite of the lower total neutron flux, the position of the Bi block was set to
50 cm from the outer reflector edge, close to the water tank , for safety reasons,
indeed in this case the bismuth load rests on the concrete biological shield.
In the fig.4.8 the thermal flux as a function of the distance from the core is
shown, for the empty channel and with the addition of a 10 cm thick Bi block.
Successively, the Bi thickness was varied and its combined effect on reactor
γ-rays attenuation and thermal neutron flux reduction, was studied. The sim-
ulations with photons were performed replacing the neutron source from the
core with a prompt-gamma source from fission reaction, as explained in § 4.2.2.
A F4 tally for photons was located in the middle lenght of the channel, almost

68
4.3. The facility design

Figure 4.6: Comparison of the absorption neutron cross section for bismuth
and lead.

1 meter far from the bismuth block, in order to study the attenuation of the
photons as a function of the bismuth thickness. The position of the tally was
chosen to avoid the contribution to the tally coming from the secondary ra-
diation scattered at high angles, occurring when the tally is too close to the
absorber; with this choise a more correct evaluation of the γ rays flux at the
measurement position, located almost 3 m from the bismuth block, can be
obtained. A non-analogue photon transport was required (GS/RR) to get the
gamma spectrum with reliable statistic. In fig.4.9 the γ rays spectra resulting
from the different bismuth thicknesses are compared.
The intensity of the gamma source undergoes a first attenuation by almost
30 cm of graphite that, considering an exponential attenuation e−µx , is about
10−2 for 1 MeV γ, according to the value reported in Table 4.3. With the
same calculation the attenuation of 10 cm Bi is about 10−3 . An average at-

Material ρ[g/cm3 ] µ [cm−1 ]


1 MeV 3 MeV 6 MeV
Bismuth 9.80 0.700 0.409 0.440
Graphite 2.25 0.143 0.080 0.055

Table 4.3: Linear absorbtion coefficent µ[cm−1 ] for bismuth and graphite

tenuation factor (F) for gamma with energy < 1 MeV and energy > 1 MeV
was obtained from the simulations by averaging the ratio of the flux with and
without bismuth filter; the results are reported in table 4.4.
The average attenuation of the γ rays intensity with energy lower than 1

69
4. The design of a PGNAA facility at Triga Mark II reactor

Figure 4.7: Top: Thermal neutron flux (n/(cm2 s)) at the beam port for dif-
ferent distance from the core of the bismuth block (10 cm thick); the x axis
indicate the distance between the graphite external edge and the center of
the Bi block. Bottom: the final position of the Bi block was setted at 50 cm
from the outer reflector edge; the bismuth load rests on the concrete biological
shield.

Bi Thickness [cm] F F
E < 1MeV E > 1 MeV
5 0.016 0.143
7 0.008 0.077
10 0.008 0.036

Table 4.4: Average attenuation factor (F) for γ due to 5, 7 and 10 cm of Bi


thickness.

MeV, that is the highest component, is the same with 7 cm or 10 cm of bismuth.

The effect of the bismuth on neutron beam was also studied; the optimal
thickness is the one ensuring a good gamma shielding without attenuating

70
4.3. The facility design

Figure 4.8: Thermal neutron flux as a function of the distance from the center
of the core for two different channel setup, that is with 10 cm thick bismuth
and without any filter.

Figure 4.9: Comparison of γ ray flux at the middle of the channel for 5, 7 and
10 cm of bismuth thicknesses and in absence of filter.

much the thermal neutron component. The thermal treatment card (MT card)
for bismuth was necessary to take into account the effects of the lattice in the
neutron interaction, that are important especially for neutron energy lower
than 4 eV. In table 4.5 the neutron flux at the end of the channel is listed for
the three different components set as follows: thermal energy lower than 0.2

71
4. The design of a PGNAA facility at Triga Mark II reactor

eV, epithermal energy between 0.2 eV and 0.5 MeV and the fast component
higher than 0.5 MeV.

Thermal F Epithermal F Fast F Total


no filter 1.07 · 108 - 2.26 · 10 7
- 5.80 · 106 - 1.35 · 108
5 cm Bi 3.53 · 107 0.33 6.72 · 10 6
0.29 3.07 · 106 0.53 4.51 · 107
7 cm Bi 2.34 · 107 0.22 4.27 · 10 6
0.19 1.98 · 106 0.34 2.82 · 107
10 cm Bi 1.33 · 107 0.12 2.2 · 10 6
0.09 1.28 · 106 0.22 1.68 · 107

Table 4.5: Neutron flux (n/(cm2 s)) at the end of the channel, without filters
and with different bismuth thicknesses. The corresponding attenuation factor
(F) for the three neutron energy components are also indicated.

The study of thicker filters is prevented by the higher attenuation that they
would provide to the thermal neutron beam. On the basis of the results of the
simulation, 7 cm thick Bi block was chosen because the thicker filter would not
produce any advantage in terms of attenuation of γ-rays (with energy lower
than 1 MeV) causing also a higher reduction of the thermal neutron flux.

Neutron filter
In order to absorb the unwanted fast and epithermal components of the neutron
beam a filter for neutron is necessary. The addition in the channel of a proper
single crystal material was considered. The use of a single crystal material as
filter for thermal neutrons is based on the fact that its neutron scattering cross
section is of the order of few barns for fast energies and it sharply decrases
at low energies. This behaviour is mainly due to the decreasing of the elastic
(or Bragg) scattering cross section components in a specific energetic range.
When a neutron beam interacts with a poly-cristal material, only the neutrons
with a wavelenght lower than a particular cutoff value λc 2 undergoes Bragg
scattering, while for higher wavelenghts the Bragg scattering does not take
place. For λ > λc the only contributions to the cross section come from
the neutron capture and inelastic scattering processes. The best materials,
for filtering purpose, are characterized by a low cross section for these last
processes; such as graphite, Be, BeO. Those in a polycristalline form have a
cutoff wavelenght of 0.67, 0.4, 0.45 nm (corresponding to the neutron energy
values of respectively 1.82, 5, 4 meV). In particular, when it is necessary to
transmit a specific energy range (such as thermal neutron energy λ = 0.1797
nm) a proper single crystal material should be selected. Parameters, such
as orientation, mosaic spread, temperature, cutting plane, also influence the
transmission properties of the crystal, thus they must be considered within
the beam tailoring procedure [17]. A theoretical formula for the total neutron
cross section in the energetic range of 10−4 eV < E < 10eV was determined
2
λc = 2d where d is the largest lattice spacing

72
4.3. The facility design

for a perfet single crystal by Freund [48], although it represents somehow an


approximation, it reproduces sufficently well the experimental results.
It has been shown that a single crystal sapphire filter (0001) cutting plane,
oriented along c-axis, can be effectively used to transmit the thermal compo-
nent of a reactor neutron beam with a significant attenuation of the epithermal
and fast components.

Figure 4.10: Comparison of the neutron attenuation as a function of the neu-


tron wavelenght for a single-crystal sapphire block at 300 and 80 K

Moreover, the transmission properties of the crystal do not change signifi-


cantly with the increase of temperature: a comparison of the total cross section
at 80K an 300K showed a farly higher transmission of thermal neutron at 80K,
but the difference does not justify the implementation of a cryogenic system.
The attenuation of neutrons for sapphire, at 80K and 300 K, as a function of
the neutron wavelenght is shown in fig.4.10
Commonly, the exposure of a crystal to a neutron beam causes damages
that deteriorate its transmission properties mainly related to the number of
displacements per atom (DPA). The neutron trasmission was measured for
sapphire blocks exposed to a thermal neutron fluence of about 3 · 1018 cm−2 ,
no significative difference appeared in the transmission properties compared to
an unirradiate block resulting in any degradation effect due to neutron beam
[49].

Given these properties, a single sapphire crystal was chosen as a filter for
fast neutrons. It was placed in the channel after the bismuth block, its diam-
eter was firstly set to 15 cm (equal to the diameter of the channel), its effect
on neutron beam was calculated for two different thicknesses: 5 and 10 cm

73
4. The design of a PGNAA facility at Triga Mark II reactor

(fig.4.11). Successively, the diameter of the sapphire block was reduced to 7


cm showing no significative difference in terms of neutron spectra. For the

Figure 4.11: A sapphire single crystal block was inserted in the channel, after
the Bi block, as filter for fast neutrons.

simulation of the neutrons transport through the crystal, an MT card, taking


into account the effects on thermal neutrons due to crystalline structure of the
sapphire blocks, was used. These data, not available in our MCNP libraries
, were calculate in Bariloche (Argentina) and experimentally verified as re-
ported in [50]. The S(α, β) cross section used for the simulation is represented
in fig.4.12.

In table 4.6 the combined effect on the neutron beam of the Bi shield (7
cm thick) and of the sapphire filter (5 and 10 cm thick) are summarized, along
with the attenuation factor for the different neutron components due to the
presence of the sapphire filter. A sapphire filter of 10 cm thick and 7 cm di-
ameter was chosen as neutron filter because it determines the decrease of the
unwanted fast neutron beam components of 1 order of magnitude, without af-
fecting too much the thermal neutron component. An HEM B5 sapphire filter
(0001) ± 30� with these dimensions was purchased to be inserted in the channel
B of the Triga reactor in Pavia. It is shown in fig.4.13. This crystal is made
of several single cristal disks, properly oriented, thus it is not a single crystal
block. Anyway it represents a less expensive option, showing not significative
differences in transmission properties.

This sapphire filter provides a further reduction of the γ rays intensity


increasing the overall quality of the beam (see fig:4.14); the attenuation factor
for γ-rays with energy lower than 1 MeV goes from 0.008, value obtained with

74
4.3. The facility design

Figure 4.12: Sapphire total neutron cross section, in the thermal energy region,
MCNP data.

Thermal F Epithermal F Fast F Total


7 cm Bi 2.34 · 107 - 4.27 · 106 - 1.98 · 106 - 2.96 · 107
7 cm Bi +
5 cm 1.93 · 107 0.83 9.65 · 105 0.22 4.91 · 105 0.25 2.07 · 107
Sapphire
7 cm Bi +
10 cm 1.62 · 107 0.71 3.53 · 105 0.19 2.11 · 105 0.11 1.67 · 107
Sapphire

Table 4.6: Neutron flux (n/(cm2 s)) at the end of the beam port with bismuth
and monocrystal sapphire blocks in the channel. The attenuation factor (F)
due to the implementation of the sapphire is also shown. It was calculated by
the fraction between the flux with both filters (Bi and sapphire) and with only
Bi filter inserted in the channel, for all the three components.

only the Bi block, to 0.0056, with both Bi and sapphire single crystal filters
inserted in the channel.

4.3.3 Collimator
The insertion of a collimator is required to reduce the diameter of the beam
at the sample position, delimiting a uniform beam intensity irradiating the
sample as well as providing a lower neutron beam intensity in the surrounding,

75
4. The design of a PGNAA facility at Triga Mark II reactor

Figure 4.13: Picture of the crystal sapphire block that will be inserted in the
channel, it is 10 cm in lenght, 7 cm diameter.

Figure 4.14: Comparison of γ ray flux at the middle lenght of the channel,
without filter, with 7 cm of poly-crystal bismuth and with the addition of 10
cm of single crystal sapphire block. It can be seen that the presence of the
sapphire further reduces the unwanted γ flux.

expecially at the detector position. A smaller beam diameter helps in shielding


the facility, in protecting the detector and in lowering the background during
the measurement.

Three different kinds of collimator were proposed, named A, B and C.


• A is a parallel collimator, concrete cylinder with iron, inner diameter 5

76
4.3. The facility design

cm, length L=90 cm, that end at the exit of the channel, fig.4.15;

• B is made up of convergent collimators, inner surface with cone shape


that gradually reduces the beam by using three separate rings 15 cm
thick, placed along the whole lenght of the channel. Material of the
collimators: concrete with iron and covered with a boral layer 1cm thick,
diameter at the exit is 5 cm, fig.4.16.

• C is a convergent collimator, length L=30 cm, 10 cm of iron and 20 cm


of lithium polyethylene, placed close to the end of the tube. Internal
maximum diameter: 6 cm, diameter at the exit: 5 cm, fig.4.17.

The insertion of a collimator, although necessary for the reason mentioned


above, it produces a further reduction of the flux according to the geometric
characteristics of the collimator, such as length (L) and diameter of the inlet
aperture (D). Thus the characteristics of the three collimator set-ups were
studied in terms of the uniformity of the beam obtained at the sample position,
the neutron background on the germanium and the thermal neutron beam
reduction.

Figure 4.15: Collimator A with the filter for γ, and fast neutrons; MCNP
geometry.

The first simulations were performed by using a F4 tally in a cell of radius


2.5 cm (equal to the outlet collimator aperture) and 0.5 mm thick placed at the
end of the channel, on the neutron beam axis. A DXTRAN sphere, coupled
with weight windows, was employed in order to overcome the disadvantageous
geometry. Firstly, the beam attenuation due to the collimator without any
filters in the channel was evaluated. The results of the thermal neutron flux
along with the attenuation factor for all the collimators are reported in table
4.7; the attenuation factor due the different collimators for the fast and ep-
ithermal neutron components is almost similar.

Successively the filters were inserted. While the position of the bismuth
remained fixed, the sapphire filter was moved along the channel according to
the collimators position. In the three configurations, no significant difference

77
4. The design of a PGNAA facility at Triga Mark II reactor

Figure 4.16: Collimator B; the shadowed area identify the two zones defined
by a collimator: umbra (in red), where the intensity of the beam is higher and
the penumbra where the intensity decrease towards the periphery.

Figure 4.17: Collimator C with the filters for γ and for fast neutrons; MCNP
geometry.

Type Thermal flux at the beam port Attenaution


No collimator 1.10 · 108 -
7
A 7.11 · 10 0.65
B 9.45 · 107 0.86
C 9.68 · 107 0.88

Table 4.7: Thermal neutron flux (n/(cm2 s)) at the end of the channel in
different geometry configuration, no filters are placed in the channel. The
neutrons attenuation due to the collimator is also shown.

occurred concerning the attenuation of the thermal neutron component caused


by the shift of the sapphire crystal. The neutron flux for different neutron en-
ergies and collimators are summarized in table 4.8

This set of simulation, performed with the tally close to the beam port,
confirmed that the collimator A provided the higher neutron attenuation, as
can be seen in table 4.7. Observing the filtered neutron spectra at the beam

78
4.3. The facility design

Neutron Energy (MeV) Void channel Collimator A Collimator B Collimator C


thermal 1.10 · 108 1.07 · 107 1.43 · 107 1.68 · 107
7 5 5
epithermal 2.37 · 10 1.54 · 10 3.37 · 10 2.28 · 105
6 5 5
fast 6.15 · 10 1.10 · 10 3.17 · 10 1.90 · 105

Table 4.8: Neutron flux (n/(cm2 s)) outside the channel for the different colli-
mator set-ups with the filters inserted in the channel.

port obtained with the three different collimator (table 4.8), the results ob-
tained with the collimator A are not satifactory; although it provided a lower
epithermal and fast neutron component, the thermal neutron flux at the beam
port is too low, considering that a value of about 107 n/(cm2 s) is advisable at
the measurement position. For this purpose the collimator A was discarded.
Before selecting between the remaining B and C collimator a further inves-
tigation was done observing the neutron flux at the measurement position.
In particular, the uniformity of the flux at the sample position and the fast
neutron flux directly irradiating the detector were evaluated. The collima-
tors define two zones along the beam path: the umbra, where the beam has
a constant intensity, and the penumbra, where the intensity lowers gradually
towards the periphery (see fig.4.17). To ensure the maximum uniformity, the
sample should be exposed to the umbra of the beam, while the penumbra
should be reduced as much as possible, to preserve the surrounding. Position-
ing the sample at a distance of about 50 cm from the beam port, geometrically
can be assest that both the collimator B and C fulfill this constraint. The next
step was to evaluate the neutron flux in the area outside the umbra at the
measurement position, this was done to understand which collimator provide
the best shielding of the germanium. For this purpose a new cilindrical tally
cell, 0.5 cm thick, was located at the sample position. It was divided into three
concentric circle section with radii corresponding to the umbra, penumbra and
a further halo sections defined by the collimator (Table 4.9). The so called
”halo” represents the solid angle determined by the intersection of the inlet
and outlet apertures of the collimator (see fig.4.18). Wherase, the penumbra
was evaluated by the intersection of the inlet aperture of the channel and the
outlet aperture of the collimator.

Figure 4.18: Scheme of the “halo” identified by the collimator; it is obtained


by the intersection of the inlet and outlet apertures of the collimator.

The tally cell was surrounded by a Dxtran sphere (see for example 4.19),

79
4. The design of a PGNAA facility at Triga Mark II reactor

the simulations were performed without any filter.


The tally used (as for both collimator B and C) are shown in table 4.9.
The results of the simulation showed that the collimator C provide a reduction

umbra penumbra halo


Radius (cm) 1.54 4.38 10.94

Table 4.9: Radius (cm) of the sections, at the sample position, of the optical
zones introduced by the collimator, that is umbra, penumbra and halo.

of the fast component in the halo region of a factor four; the most important
for the detector shielding purpose; moreover it assures a thermal neutron flux
at the measurement position (in the umbra) slightly higher than B. By these
consideration the collimator C was chosen, with the further advantage that
its implementation requres a lower amount of material to be inserted in the
channel.
To get a better shielding of the germanium detector the collimator C was
extended forward keeping the convercence of its inner surface. Then a new
pierced block 30 cm long, made up of 10 cm of iron and 20 cm of lithium
polyethylene was implemented outside the reactor along the beam axis fig.4.19
The radii of the three different zones due to the collimator were geometrically

Figure 4.19: External extention of the C collimator. The blue zone represents
the halo of the collimator.

defined and the new values are reported in table ??. The divergence of the
beam was considerably reduced and the radius of the halo was about 3.4 cm
in place of the previous 10.94 cm. The thermal neutron flux at the sample
position, with this new collimator configuration, was reduced by a very low
factor (less than 1%). The values for the umbra region, evaluated without the
filters in the channel, are summarized in table 4.11
Therefore the collimator in fig.4.19 was considered as the definitive solution.

80
4.3. The facility design

umbra penumbra halo


Radius (cm) 1.54 2.66 3.44

Table 4.10: Radius (cm) of the sections, at the sample position, of the opti-
cal zones introduced by the extended C collimator (fig.4.10) , that is umbra,
penumbra and halo.

Neutron energy Umbra


Thermal 7.27 · 107
Epithermal 1.44 · 107
Fast 3.88 · 106
Total 9.09 · 107

Table 4.11: Unfiltered neutron flux n/(cm2 s) for the three neutron components
at the measurement position obtained with the extended collimator.

4.3.4 The beam catcher design


The beam catcher is a structure designed to stop the primary beam and to en-
sure a level of radiation outside the shield lower than the radiation protection
prescriptions. Therefore, the beam catcher was projected in order to fix the
external radiation dose-rate not higher than 10 µSv/h; that is observing the
maximum annual dose allowed for the personnel, calculated considering 2000
working hours per year.

The simulations performed to design the shields and the beam catcher
required the shift of the neutron source from the core to a surface close to the
beam port (see § 4.2.2). The neutron source was determined in absence of
filters in the channel, to be conservative.
The different steps performed in the beam chatcher design included:

• The design of a structure able to absorbs the primary beam; for this
purpose the tally flux were expolited during the simulations;

• Optimization of the biological shield by monitoring the dose-rate outside


the structure. For this purpose, some soft tissue cells (ICRU 44) [51],
5 cm thick, were added all around the structure; the energy deposition
for neutrons and photons within these cells was calculated, this was dine
by using F6 tally in a coupled neutron/photon transport (mode n,p).
The outcome of the F6 tally, already converted in µG/h, was succes-
sively multiplied for the quality factor of the specific radiation to get
the equivalent dose-rate; the values used in this work are summarized in
table 4.12:

81
4. The design of a PGNAA facility at Triga Mark II reactor

Radiation Q factor
Photons 1
Neutrons
(En < 500 keV) 5
( 0.5 MeV< En < 1 MeV) 10
( 1 MeV< En < 2.5 MeV) 13
( 2.5 MeV < En < 20 MeV) 20

Table 4.12: Quality factor for photon and neutrons used to calculate the dose
equivalent in the design of the beam catcher of the PGNAA facility; the values
for neutrons depend on the energy range.

Beam Stop: first step


As was mentioned before the first step consisted in the determination of a beam
stop; for this purpose a cylindrical tally cell filled with air, 40 cm diameter and
0.5 cm thick was located 106 cm far from the outer reactor wall (fig.4.20).
A first shileding structure was simulated by evaluating the neutron energy

Figure 4.20: Position of the first tally cells used to evaluate the characteristics
of the unfiltered neutron beam that must be stopped.

spectrum in the position along the beam axis where the beam stop should
start. The caracteristics of the beam in this position determine the absorbing
materials to be used. A constraint in the beam stop development (shape and
dimention) was imposed by the available space, reduced by the thermal column
shutter when it is opened. Thus this geometrical information was implemented
in the MCNP file.
The unfiltered neutron flux was calculated at the tally cell. This gave the
information about the neutron beam that must be stopped; the results are
summarized in table 4.13
To stop the beam a cylindrical structure, 30 cm diameter and 120 cm long
was implemented (see fig.4.21), made up of a first layer of lithium carbonate
(Li2 CO3 ) and alternating layers of iron (10 cm thick) and borated concrete

82
4.3. The facility design

Neutron Energy (MeV) Flux (n/(cm2 s))


0 ÷ 2 · 10−7 1.04 · 107
−7 −6
2 · 10 ÷ 5 · 10 1.04 · 107
−6 −1
5 · 10 ÷ ·10 3.50 · 106
−1
5 · 10 ÷ 1 2.18 · 105
1÷2.5 4.02 · 105
2.5÷10 2.97 · 105
10÷20 1.61 · 105
totale 2.53 · 107

Table 4.13: Neutron flux evaluated at 106 cm far from the outer reactor wall,
in void channel setup.

(0.16% mass of 10 B) (20 cm thick). The first layer stops effectively the thermal
neutron component of the primary beam, avoiding also the activation of the
iron layer located after. Whereas, the following layers thermalize the fast
neutron component of the spectrum; the energy of the fast neutrons is firstly
reduced by the inelastic scattering with the iron and successively by the elastic
scattering with the hydrogen, finally, when the thermal range is reached, the
neutrons are captured in boron contained in the concrete.
The fast neutron flux into the cylindrical shielding structure is expected to

Figure 4.21: Cylindrical structure implemented to stop the primary beam. A


first lithium carbonate absorbs the thermal neutrons, successively an laternat-
ing layer of iron and borated concrete provide to slow down the fast comonent
of the beam until the thermal energy; the thermal neutrons are further cap-
tured in boron.

decrase exponentially according to the the removal cross section Σr theory (see
chapter 2); since the shield is made up of alternating layers of heavy and light
elements.
φ(r) = SG(r)eΣR t (4.7)
where G is:

AeΣR r
G(r) = (4.8)
4πr2
83
4. The design of a PGNAA facility at Triga Mark II reactor

To verify this a F4 tally cell was positioned for each iron/concrete layer; the
simulations showed the expected trend on the neutron flux (fig.4.22). In par-
ticular, the neutron fluxes at the beginning (tally cell 511) and at the end (tally
cell 515) of the thermalizing structure are reported in table 4.14; an overall at-
tenuation factor of 2.10 · 10−6 was obtained. This value is comparable with the
expected attenuation by applying the formula(4.7) that is equal to 9.75 · 10−7
3
.

Figure 4.22: Simulated fast neutron flux trend within the thermalizing struc-
ture of the beam catcher.

Energy (MeV) cell 511 cell 515


1 ÷ 2.5 4.97 · 105 1.2
2.5 ÷ 10 3.15 · 105 7.56 · 10−1
10 ÷ 20 1.30 · 105 2.63 · 10−2

Table 4.14: Neutron fluxes at the beginning (tally cell 511) and at the end
(tally cell 515) of the thermalizing structure.

As far as the thermal neutron attenuation concerns an attenuation factor


of the order of 1.6 · 10−5 was evaluated for the a 10 cm layer of lithium car-
bonate (Li2 CO3 ), natural composition 4 . This layer was located before the
thermalizing fast neutron cylinder.
3
using Σr(Iron) = 0.168cm−1 and Σr(Con) = 0.098cm−1 the removal cross section respec-
tively for iron and concrete
4
The natural isotopic composition is 92.5% for 7 Li and 7.5% for 6 Li

84
4.3. The facility design

Finally the neutron flux at the end of the cylindrical structure, for different
energy bins, (cell 515) assumed the value showed in table 4.15.

Neutron Energy (MeV) Flux (n/(cm2 s))


0 ÷ 2 · 10−7 1.45
2 · 10−7 ÷ 5 · 10−6 9.23 · 10−1
5 · 10−6 ÷ 5 · 10−1 3.45
5 · 10−1 ÷ 1 1.33 · 10−1
1÷2.5 4.40 · 10−2
2.5÷10 2.91 · 10−3
10÷20 9.85 · 10−3
totale 6.2

Table 4.15: Neutron flux evaluated at the end of the cylindrical structure.

Structure optimization
The next step dealt with the implementation of a biological shield to be lo-
cated all around the facility in order to ensure an external dose rate lower
than 10µSv/h. For this purpose the MCNP geometry was complemented by
the designed collimator and a massive borated concrete structure all around
the already described cylindrical shield. The position of the germanium de-
tector was also fixed at 15 cm (sample-detector distance) and a lead cilinder
4 cm thick was positioned along its axis. This set up was gradually modified
by monitoring the γ dose-rate at the external wall of the facility. For this pur-
pose, 24 tally cells, 5 cm thick, filled with soft tissue (ICRU44), were placed
all around the structure (see fig.4.23); 4 on the roof, 8 at the outside wall near
the detector dewar, 9 at the opposite wall and 2 at the wall ortogonal to the
beam axis. A coupled neutron/photon transport was simulated and a tally F6
for both neutrons and photons were request within the tissue cells.

The beam catecher design that gives the best results is shown in fig.4.23,
and its dimentions are summarized in table 4.16.

A Zone B Zone C Zone


Lenght 107.8 cm 61 cm 65 cm
Width 208.5 cm 106 cm 86 cm
Height 175.5 cm 175.5 cm 175.5 cm
Roof 20 cm
Right wall thickness 25 cm

Table 4.16: Geometrical details of the beam catcher shown in fig.4.23

85
4. The design of a PGNAA facility at Triga Mark II reactor

Figure 4.23: MCNP reconstruction of the beam catcher; the pink rectangulars
located around the outside wall are the soft tissue tally cells used to evaluate
the dose rate.

As can be seen in fig.4.23 the dewar of the germanium was moved back
from the crystal, and a concrete shield was placed all around the dewar. The
layers of iron and lithium polyethylene located in the surrounding of the de-
tector worked as a thermalizing structure for the fast neutron; A cavity in the
thermalizing structure was inserted to get a further reduction of the fast flu-
ence on the germanium, in table 4.17 the dimentions of the cavity are shown.

86
4.3. The facility design

Cavity
Lenght 46 cm
Width 17.5 cm
Height 30 cm

Table 4.17: Dimensions of the cavity holding the sample.

For the final set up, the simulated annual fast neutron fluence on the detector,
evaluated considering 750 working hours per year was equal to 2 · 1010 n/cm2 in
absence of filters in the channel; considering the attenuation factors introduced
by the filters the fast neutron fluence becomes 2.2·109 n/cm2 . This value is still
higher than the desirable value since a fast neutron fluence of 1010 n/cm2 was
estimated as the maximum limit above which the detector is damaged (see §
2.3.1). Nevertheless it must be considered that the studies about the detector
damages were carried out by using a unmoderated 252 Cf source; this source
provides a fast neutron spectrum with the most probable energy value of 0.77
MeV and the average energy about 1.98 MeV. When the simulated neutron
beam is considered, the average energy of the fast neutron component (rang-
ing between 500 keV and 20 MeV) was estimate being about 0.37 MeV, lower
compared to the one provided by a 252 Cf source, therefore can be assumed
that a less important damage is induced by fast neutron component of the
simulated PGNAA neutron beam. For this reason the structure designed can
be considered efficacy in shielding the crystal from the fast neutron.
A shield for the thermal neutrons was also implemented to protect the de-
tector, in particular a 2 mm thick layer of lithium carbonate (Li2 CO3 , 16%
6
Li mass enriched) was used to line the external holder of the germanium (see
4.24). Indeed, when the low energy neutrons interact with the crystal, they
are captured in Ge isotopes producing prompt and decay γ rays. The relative
intensities of the Ge detector peaks are different from those produced by a Ge
sample, due to the true-coincidence summing effect occurring with the gamma
ray cascades inside the detector. A 596 keV peak in the background is an in-
dication of the crystal activation [21]. The thermal neutron flux in the active
volume of the crystal is lowered by a factor of about 10−2 , bringing its mean
value 62 n/(cm2 s).

The neutron and γ dose rates obtained in the soft tissue cells located outside
the beam catcher, in its final configuration, range between 0.04 (in the frontal
wall) to 5.8 µSv/h remaining below the limit value of 10 µSv/h; in particular
the values for the tally cells showed in fig.4.23 are collected in table 4.18. The
highest dose value is in the zone where the primary beam is stopped within
the first layer of the cylindrical beam catcher (tally number 10 in fig.4.23).

87
4. The design of a PGNAA facility at Triga Mark II reactor

Cell number Equivalent dose (µSv/h)


1 0.63
2 1.22
3 0.23
Lateral walls 4 0.61
5 4.13
6 1.07
9 1.18
10 5.80
11 2.60
Frontal wall (7 8) 0.04
12 1.30
13 0.23
Roof 14 0.15
15 0.02

Table 4.18: Dose rate at the outside wall of the final beam catcher design; the
position of the numbered tally cells is shown in fig.4.23

4.4 The simulation of the germanium response


function
The simulation of the HPGe response function allowed both the study of the
background optimization and the evaluation of the detection capability of the
designed PGNAA facility. In particular, by simulating the response of the
detector, it was possible to improve the detector shielding reducing the back-
ground due to the neutron interaction with the surrounding materials.

Simulation conditions The simulation where performed by using a surface


source of 2.12 cm radius placed 21 cm before the sample position; it provides
a neutron beam, already filtered and collimated, with energy lower than 10
eV. The spectral distribution and the flight directions were chosen according
to previous simulations and in order to reproduce the beam divergence in that
point due to the collimator (see §4.2.2).
The detailed geometry of the coaxial HPGe N-type detector, model GR2519
(Canberra), was implemented according to the data provided by the construc-
tor; the MCNP reconstruction is visible in fig.4.24 and its characteristics are
summarized in table 4.19. It is provided of a carbon window 0.5 mm thick,
and its external electrode is very thin, being equal to 0.3 µm. Its performance
specification are listed in table 4.20

To simulate the response function, a pulse height energy tally (F8) was

88
4.4. The simulation of the germanium response function

Figure 4.24: MCNP scheme of the HPGe detector.

Parameter Description Value


Dc Cristal diameter 54 mm
Hc Cristal lenght 54 mm
Dh Core hole diameter 10 mm
Hh Core hole depth 42.5 mm
De endcap-crystal distance 5 mm
external dead layer 0.3 µm
internal dead layer 0.7 mm
endcap thickness 1.5 mm

Table 4.19: Geometric specification of the HPGe detector.

set within the active volume of the crystal. The cross section data for the
germanium isotopes used for Monte Carlo simulations 5 are shown in fig.4.25
The energy interval of the simulated spectra was set between 0 and 2.3
5
Data available from Evaluated Nuclear Data Files (ENDF) library http://www-
nds.iaea.org/exfor/endf.htm and kindly converted in a format readable by MCNP (ACE), by
Doct. Manuel L. Sztejnberg Gonçalves - Carralves, Comisión Nacional de Energı́a Atómica,
CNEA, Buenos Aires, Argentina, using NJOY program (http://t2.lanl.gov/codes/).

89
4. The design of a PGNAA facility at Triga Mark II reactor

Description Constructor values


Resolution (FWHM) at 1332 , 60 Co 1.80 keV
Resolution (FWHM) at 122 , 57 Co 0.850 keV
Peak/Compton ratio 61.4:1
Relative Efficency 28.2%
Amplifier time const 4 µs

Table 4.20: Performance specification of the HPGe detector.

Figure 4.25: Neutron cross section for the Ge isotopes; that, listed by higher
isotopic abundance, are: 74 (green), 72 (purple), 70 (light green), 76 (red), 73
(blue).

MeV, firstly a fine energy bin division was chosen, equal to 0.3 keV/ch, and
the gaussian energy broadening option (GEB) was used to simulate the normal
peak broadening occurring with a physical detector. This broadening is simu-
lated by MCNP sampling the tallied energies from a Gaussian distribution:
E−E0 2
f (E) = Ce−( A
)
(4.9)

where E and E0 are respectively the broadened and unbroadened energy of


the tally; C is a normalizzation constant and A is the Gaussian width. The
A of the distribution is proportional to the full-width-half-maximum, FWHM
provided by the user by defining a, b and c constant in the forumla:

F W HM = a + b E + cE 2 (4.10)

90
4.4. The simulation of the germanium response function

the parameters a, b and c can be specified by the GEB card.


The doppler broadening of the boron peak cannot be simulated by MCNP,
nevertheless the peak shape of the boron peak from boron in a water matrix
sample is almost rectangular and 15 keV wide centered in 478 keV. So a dif-
ferent energy binning was setted in order to distribute all the 478 keV event
within a bin 15 keV wide; therefore a calibration setted to 15 keV/ch was used
in the simulation.

22
Comparison between simulated and experimental spectra of a Na
radioactive source
A 22 N a source was used to evaluate the detector efficency at the 511 keV, that
could be approximately similar to the efficeny for the 478 keV boron peak. The
same source was successively simulated and the results between simulation and
experiment were compared.
The real detector was covered by a lead empty cilinder, 4 cm thick and the
22
N a radioactive source was located 15 cm far from the endcap of the detector
on its axis, in order to reproduce the design condition. The spectrum obtained
with 350 s of live time acquisition is shown in fig.4.26. In the spectrum the

Figure 4.26: Experimental spectrum of a 22 N a radioactive source located at


15 cm from the coaxial detector, along its axis. The 511 keV and 1275 keV
peaks are visible.

511 keV and the 1275 keV full energy peaks emitted by the radioactive source
are visible; by using these two peaks for the calibration procedure a value of
15.28 keV/ch was obtained.
The same measurement was simulated by MCNP; a similar efficiency was
obtained. In fig.4.27 the simulated and experimental spectra were compared.
To perform the comparison the room background spectrum was subtracted
from the experimental one, since the MCNP cannot consider that contribution.
A good agreement betweem the two the spectra was obtained.

91
4. The design of a PGNAA facility at Triga Mark II reactor

Figure 4.27: Comparison between the simulated (black) and experimental (red)
spectra.

Background Optimization The background spectrum represents the sig-


nal that is not related to the sample; it is desirable to reduce as much as
possible this contribution because it negatively affects the detection limit of
themeasurement system. Two background components can be distinguished:
the room-background, due to the γ naturally present in the room when the
neutron beam is off, and the beam-background that appears when the beam
is on and in absence of sample.
A room-background spectrum was experimentally measured by the HPGe de-
tector, it will be used in the evaluation of the designed PGNAA facility perfor-
mances; in particular, a count rate, in the region of the boron peak, of about
0.5 cps (count per second) was estimated. The spectrum corresponding to 2000
second of live time is represented in fig:4.28.

The beam background was simulated firstly without the sample, succes-
sively a D2 O sample was introduced in order to evaluate the enhancement
of the background due to the scattering of the sample. Indeed, the neutrons
undergoing scattering in the sample, interact with the surrounding material
increasing the background signal. The use of heavy water as scattering sample
prevents the 2.2 MeV capture gamma rays production, occurring in light water.

In fig. 4.29 the beam background simulated in absence of sample is shown.


The full energy peaks from boron and hydrogen, present in the beamcatcher
material are visible as well as the 511 keV annihilation peak. A less intense
peak is visible below the 2.2 MeV, it correspond to the single escape peak of
the hydrogen with energy of 1.712 MeV. Some changes in the structure were
required in order to lower the intensity of these background peaks; firstly the
reduction of the 478 keV peak was studied. The Cell-Flagging Card [CFn

92
4.4. The simulation of the germanium response function

Figure 4.28: Room-background spectrum obtained with 2000 seconds .

Figure 4.29: Beam-background spectrum evaluated without sample.

card] together with the F4 tally within the active volume of the detector al-
lowed defining the region of the beamcatcher where the 478 keV γ-rays came
from 6 . In particular, the first borated concrete wall encountered by the pri-
mary beam, located just after the polyethylene shield (region in green color in
fig.4.30), was identified as the boron peak source. Thus the boron in concrete
was replaced with Li (0.48% percentage mass of 6 Li) in the amount corre-
6
FCn is followed by cell identificators; when a particle track leave the designed cell it is
“flagged“ and its contribution to the tally is recorded separately.

93
4. The design of a PGNAA facility at Triga Mark II reactor

sponding to the macroscopic cross section of the borated concrete. After this
modification, the simulated response function of the germanium detector was
deprived of the boron peak, as shows fig.4.31, the intensity of the hydrogen
and the annihilation peaks remained unchanged.

Figure 4.30: Main structural improvements introduced to lower the background


count rate.

The origin of the 2.2 MeV γ rays is clearly due to the thermal neutron
capture in hydrogen present in the polyethylene; thus a lithium absorber, con-
sisting of a 2 mm Li2 CO3 (6 Li 0.666% mass percentage enriched), was used
to line the beam tube and the sample region in order to capture those neu-
trons scattered by the sample that may interact with the lithium polyethylene
(fig.4.30). The new beam-beackground is shown in fig.4.32; the intensity of
2.2 MeV full energy peak was decreased, with a consequent lowering of the
baseline; 511 KeV peak is the only one remaining.
A further investigation of the beam-background spectrum was carried out
by introducing a heavy water sample (1 ml), cilindrical shape, inclined by 45o
with respect to the beam line; the resulting spectrum is shown in fig.4.33, in
comparison with the beam background spectrum obtained in absence of sample
and without any structural improvements. The comparison of the two spectra
showed the effectiveness of the modifications implemented in the facility in
decrasing the background count rate, expecially in the region of boron and
hydrogen peaks.

94
4.4. The simulation of the germanium response function

Figure 4.31: Beam-background spectrum evaluated without sample after the


replacement of the B with Li in a region of the beamcatcher; the absence of
the boron peak can be noticed.

Figure 4.32: Beam-background spectrum after the introduction of a lithium


carbonate shield surrounding the measurement position; the 2.2 MeV hydrogen
peak was considerably lowered with a consequent reduction of the spectrum
baseline.

95
4. The design of a PGNAA facility at Triga Mark II reactor

Figure 4.33: Comparison of the original beam-background in absence if sample


(black) and after the structural improvement in presence of a D2 O sample
(1ml) (green). A significative reduction of the background count per second
was achieved by the structural optimization.

Detection and quantification capabilities After the reduction of the


background, a 0.5 ml sample containing 50 ppm of 10 B in a H2 O matrix was
added in the geometry and the germanium response function was simulated.
A 0.5 ml water sample was also simulated as a blanck for the measurement, in
order to correctly evaluate the background under the boron peak. To performe
these simulations, the MT card for water was specified correcting the thermal
neutron transport for the water chemical bounds.
The response function, bin normed, obtained with the boron containing sample
and with the blank, corresponding to almost 5 seconds of live time irradiation,
are represented in fig.4.34; the results are expressed in number of neutron per
particle source. About the boron containing sample (up), the simulated spec-
trum shown the 478 keV and 2.2 MeV peaks, respectively from the boron and
hydrogen, as well as the single and double escape peaks of hydrogen (respec-
tively at 1.712 and 1.201 MeV) and the 511 keV annihilation peak. As regard
the spectrum of the blanck, the same peaks are visible unless the boron one.
The counts per second under the boron peak for the different spectra, in-
cluding the background, are collected in table 4.21.
Starting from these simulations it was possible to calculate the sensitivity
for natural boron of the designed PGNAA by dividing the net count rate under
the boron peak with the mass of boron in the sample. The net area under the
boron peak was calculated by subtracting from the gross area the background

96
4.4. The simulation of the germanium response function

Figure 4.34: Germamium detector response function obtained with a 0.5 ml


sample; up 50ppm of 10 B in water matrix; down H2 O sample.
10
Room background Beam background H2 O blank 50 ppm B
cps 0.5 1.2 12.5 143.7

Table 4.21: Count per second under the boron peak for different spectra.

obtained by averaging the counts at right and left of the boron peak, only one
bin at each side was included. A sensitivity for natural boron of about 1033
cps/mg was obtained.
This value was compared to the one expected by applying the analytical
formula for the sensitivity (see chapter 2):

NA
S= σγ,0 Φ0 ε(E) (4.11)
M
the corresponding description and value of the parameters are collected in

97
4. The design of a PGNAA facility at Triga Mark II reactor

Parameter Description Value


σγ0 partial gamma ray production 716 b
Φ0 Thermal flux 1.27 · 107 n/(cm2 s)
ε(478keV ) efficency for boron peak 2.1 · 10−3 (simluated)

Table 4.22: Definition and value of the parameters used to calculate the sen-
sitivity.

table 4.22
By using the 4.11, the calculated reaction rate (number of reactions per sec-
ond) occurring for 1 mg of pure natural boron is 49.78 · 104 s−1 ; if a value of
2.1 · 10−3 is assigned to the detector efficency for the 478 keV region 7 , a sensi-
tivity for natural boron of about 1045 cps/mg can be estimated; this value is
comparable with the previous one.

The detection limit for boron (DL ), that is the smallest amount of boron
that gives rices to a true signal visible in the background, according to the Cur-
rie definition [32], was evaluated by the forumla already described in chapter
2: �
3.29 Rb
DL = (4.12)
S t
By using a sensitivity (S) of 1033 cps/mg and a background count rate, Rb =
13cps, given by the sum of the background count rate for the hydrogenous
blank sample and for the room, the detection limit for t=1000 seconds of live
time irradiation is 363 ng of natural boron, equivalent to a DL for 10 B equal
to 73 ng. With these value a detection of boron concentration of the order of
1 ppm, in sample of 0.5 ml and with irradiation times of few minutes can be
achieved.

7
The efficency for the 478 keV γ rays of the HPGe was estimated by the simulation,
that is extracting uniformly the 478 keV γ ray from the sample position and evaluating the
number of those counted by the detector.

98
4. The design of a PGNAA facility at Triga Mark II reactor

100
Chapter 5
Conclusions and future
perspectives

Boron Neutron Capture Therapy in Pavia University is a research activity per-


formed since the Eighties, and many progresses have been made in this field.
In particular, after the clinical application to two patients with a challenging
protocol of ex-situ irradiation, the research was focused on disseminated tu-
mours to be treated by external neutron beams. This new strategy presents
some advantage for the patients (avoids the invasive surgery) but it requires a
deeper knowledge of the boron biodistribution in all the irradiated tissues in
order to assess a safe and effective treatment plan. In this frame, a PGNAA
facility is the complementary instrumentation to analyze as deeply as possible
the variables involved in the BNCT dosimetry. The PhD work was almost
entirely dedicated to the project of a new PGNAA facility to be installed at
the TRIGA Mark II research reactor of Pavia University.
The Monte Carlo transport code MCNP4c2 was employed to design the
PGNAA facility expoliting the neutrons from the horizontal channel B. Its
final configuration is shown in fig.5.1.
The thermal neutron flux at the beam port, in the original configuration of
the channel, was equal to 1.07 · 108 n/(cm2 s) and the thermal-to-total neutron
flux equal to almost 0.8. The insertion of proper filters was studied in order
to obtain a neutron beam suitable for PGNAA purposes; with a high thermal
neutron component low epithermal, fast neutrons and γ ray contaminations.
In particular, a bismuth block, 7 cm thick, was inserted in the channel in order
to absorb the γ rays from the core, ensuring an attenuation factor of 0.008 for
γ with energy lower than 1 MeV. The thermal neutrons are attenuated by the
bismuth filter of a factor of about 0.22. A sapphire single crystal 10 cm thick
was tested as neutron filter, to reduce the epithermal and fast components of
a factor 0.11. The transmission of the thermal component is about 70%. A
convergent collimator, partially inserted in the channel, will be employed to
reduce the diameter of the neutron beam at the sample position to about 3
cm. In this configuration, the neutron flux at the irradiation position, setted

101
5. Conclusions and future perspectives

Figure 5.1: MCNP plot of the horizontal section of the Triga Mark II reactor
of the University of Pavia; the PGNAA facility was designed in the horizontal
B channel.

on 50 cm from the wall of the reactor, is about 107 n/(cm2 s) for the thermal
component, and the epithermal and fast neutron flux is about 105 n/(cm2 s).
With these values, the beam can be considered enough clean and potentially
able to provide a quick evaluation of 10 B-concentration of the order of ppm
required for BNCT purpose by the prompt gamma spectroscopy.

A high energy resolution semiconductor detector (HPGe) perpendicularly


positioned with respect to the beam direction is used to detect the prompt
γ rays from the sample. The sample-detector distance was setted to 15 cm.
The shield between the outside wall of the reactor and the germanium detector
was optimized in order to lower the fast neutron fluence per year irradiating
the germanium detector. With the designed shields configuration, the annual
fluence is equal to 2.2·109 n/cm2 . Moreover, a cylindrical lead shield surrounds
the detector in order to absorb the gamma rays coming from the interaction
of the neutrons with the structure. Finally, a 2 mm thick layer of enriched
lithium carbonate lines the holder of the detector, to absorb thermal neutrons
scattered by the sample. This set-up should prevent the detector damage and
lower the background signal.

102
Further precaution was used with the aim to decrase the background; a
shield of enriched lithium carbonate, 2 mm thick, was implemented along the
beam tube, in the surroundings of the irradiation position, in order to absorb
scattered thermal neutrons that may be captured in hydrogen of the shield
materials and give rice to a strong 2.2 MeV γ signal in the germanium detector.
The massive borated concrete used as biological shield was also a source of 478
keV γ rays; the replacement of the boron with the lithium, preserving the
same macroscopic absorption cross section, in the concrete wall that faces the
irradiation position, prevents most of the 478 keV signal background.
Next to the measurement position, an absorbing structure is required in
order to stop the beam, properly shielded to fulfill the radiation protection
indications. A cylindrical structure made of different materials in-axis with
the beam was implemented to stop the primary beam. It consists of a Li2 CO3
disk containing 1.22% (mass fraction) of 6 Li, to absorb the thermal neutrons
followed by an alternating layers of iron and borated concrete to thermalize
the fast neutrons (>1MeV) and then to absorb them. A massive structure
made of borated-concrete surrounds all the components of the facility, it was
designed to ensure an external dose rate lower than 10µSv/h.

Figure 5.2: Horizontal section of the PGNAA facility. Two filters was imple-
mented in the channel B: a bismuth block and a sapphire single crystal; all
the other components are: a convergent collimator, a HPGe detector, a beam
catcher.

For the projected facility, a sensitivity for natural boron in a light water
matrix of about 1033 cps/mg was estimated by the simulations. The detection
limit for natural boron, that is the minimum detectable mass, is about 363
ng, obtained considering a background count rate under the boron peak, of
about 13 cps (counts per second). This value of the background comes from
the counts per second under the boron peak obtained by the irradiation of a
light water sample, and the room background spectrum measured in the reac-

103
5. Conclusions and future perspectives

tor hall by the HPGe that will be used for the gamma spectroscopy. With this
facility, a detection of boron concentration of the order of 1 ppm, in sample of
0.5 ml and with irradiation times of few minutes will be possible.

The construction of the facility is scheduled for the near future; most of the
materials required to built the facility were purchased (sapphire crystal and
bismuth blocks). Some practical aspect of the facility construcion are still un-
der study, for example, the sample positioning system and the implementation
of a shutter. This device system, in particular concerning the turning on and
off the beam, will allow replacing the sample without the need to shutdown
the reactor.

104
5. Conclusions and future perspectives

106
Bibliography

[1] R. Barth, J. Coderre, M. Graca, H. Vicente and T. Blue, Boron Neutron


Capture Therapy of Cancer: Current Status and Future Prospects Clin.
Cancer Res. 11(11) (2005) 3987-4002.

[2] A. Wittig, J. Michel, R.L. Moss, F. Stecher-Rasmussen, H.F. Arlinghaus,


P. Bendel, P.L. Mauri, S. Altieri, R. Hilger, P.A. Salvadori, L. Menichetti,
R. Zamenhof and W.A.G. Sauerwein, Boron analysis and boron imaging
in biological materials for Boron Neutron Capture Therapy (BNCT) Crit.
Rev. Oncol. Hematol. 68, 66-90 (2008).

[3] S. Altieri, S. Bortolussi, P. Bruschi, P. Chiari, F. Fossati, S.Stella, U.Prati,


L. Roveda, A. Zonta, C. Zonta, C. Ferrari, A. Clerici, R. Nano, T.
Pinelli, Neutron autoradiography imaging of selective boron uptake in
human metastatic tumours Appl. Radiat. Isot. 66, 1850-1855 (2008) .

[4] T.U.Probst Methods for boron analysis in boron neutron capture therapy
(BNCT).A review Fresenius J Anal Chem 364 (1999), 391-403.

[5] Hideghéty K, Sauerwein W, Wittig A, Götz C, Paquis P, Grochulla F,


Haselsberger K, Wolbers J, Moss R, Huiskamp R, Fankhauser H, de Vries
M, Gabel D., Tissue uptake of BSH in patients with glioblastoma in the
EORTC 11961 phase I BNCT trial, J Neurooncol, 62, 145-56 (2003).

[6] Raaijmakers CP, Konijnenberg MW, Dewit L, Haritz D, Huiskamp R,


Philipp K, Siefert A, Stecher-Rasmussen F, Mijnheer BJ, Monitoring of
blood-10B concentration for boron neutron capture therapy using prompt
gamma-ray analysis, Acta Oncol.34(4), 517-23 (1995).

[7] Mukai K, Nakagawa Y, Matsumoto K, Prompt gamma ray spectrometry


for in vivo measurement of boron-10 concentration in rabbit brain tissue,
Neurol Med Chir (Tokyo) 35, 855-60 (1995).

[8] Matzumoto T., Aizawa O.,Prompt gamma-ray neutron activation analysis


of boron-10 in biological materials, Appl Rad. and Isotopes 41, 897-903
(1990).

107
BIBLIOGRAPHY

[9] Byun S.H., Choi H.D., Design of a prompt gamma neutron activation
analysis system at HANARO, J. of Radionanalytical Nuclear Chemistry,
Vol. 244, 413-416 (2000).

[10] Honzatko J., Tomandi I., Boron concentration measurement system for
the czeck BNCT Project, AIP Conference Proceeding (2000).

[11] P. Kudejova, G. Meierhofer, K. Zeitelhack, J. Jolie, R. Schulze, A. Türler,


T. Materna, The new PGAA and PGAI facility at the research reactor
FRM II in Garching near Munich , J of Radioanalytical and Nuclear
Chemistry, Vol. 278(3), 691â695 (2008).

[12] Imahori Y, Ueda S, Ohmori Y, et al., Positron emission tomography-


based boron neutron capture therapy using boronphnylalanine for high-
grade gliomas : PartII, Clin. Cancer Res. 4, 1833-41 (1994).

[13] IAEA Handbook on Nuclear Data for Borehole Logging and Mineral Anal-
ysis (TR- 357), International Atomic Energy Agency, Vienna, Austria
(1993).

[14] Database of Prompt Gamma Rays from Slow Neutron Capture for Ele-
mental Analysis, 2007 International Atomic Energy Agency (IAEA)

[15] G.Molnar et al The new prompt-gamma neutron activation analisys fa-


cility at Budapest, J of Radioanalytical and Nuclear Chemistry Vol.215,
111-115 (1997).

[16] Yonezawa, C.,A Magara, M.,A Hoshi, M.,A Tachikawa, E.,A Sawahata,
H.,A Ito, Y.D, Prompt gamma-ray analysis using cold and thermal guided
neutron beams J. of Radioanalytical and Nuclear Chemistry, 193, 171-178
(1995).

[17] Adib M.,Attenuation of Reactor Gamma Radiation and Fast Neutrons


through Large Single-Crystal Materials, Proceeding of the 3th Environ-
mental Physics Conference, 19-23 Feb.2008, Aswan, Egypt.

[18] Kent J.Riley, Otto.K.Harling, An inproved prompt gamma neutron acti-


vation analisys facility using a focused diffracted neutron beam, NIM B
143, 414-421 (1998).

[19] R.L.Paul, R.M.Lindstrom, C.Brocker and E.A.Mackey, Design of a new


instrument for cold neutron prompt gamma-ray activation analysis at
NIST , J of Radioanalytical Nuclear Chemistry Vol. 278 (3) 697-701
(2008).

[20] Lamarsh J.R., Introduction to Nuclear Reactor Theory, 1966 Addison-


Wesley Publishing Company Massachussets (USA)

108
BIBLIOGRAPHY

[21] Knoll G.F.; Radiation Detection and Measurement, 1989 Jhon Wiley and
Sons, (USA)

[22] G.L.Molnar; Handbook of Prompt Gamma Neutron Activation Analysis


with Neutron Beam, 2004 Kluwer Academic Publishers, Dordrecht.

[23] S.Bortolussi, J.G.Bakeine, F.Ballarini, P.Bruschi, M.A.Gadan, N.Protti,


S. Stella, A.Clerici, C.Ferrari, L.Cansolino, C.Zonta, A.Zonta, R.Nano and
S.Altieri, Boron uptake measurements in a rat model for Boron Neutron
Capture Therapy of lung tumours, Appl. Rad. Isot., 69, 394-398 (2011).

[24] S.Altieri, S.Bortolussi, P.Bruschi, P.Chiari, F.Fossati, A.Facoetti,


R.Nano, A.Clerici, C.Ferrari, A.Zonta, C.Zonta, A.Marchetti, E.Solcia
, J.J.Bakeine, O.Salvucci, Monte Carlo dose calculations for BNCT
treatment of diffuse human lung tumours, in Y.Nakagawa, T.Kobayashi,
H.Fukuda ed., Advances in Neutron Capture Therapy, Proceedings of
ICNCT-12, pp 500-503, Takamatsu, Japan, 9-14 October 2006.

[25] A. Zonta, T. Pinelli, U. Prati, L. Roveda, C. Ferrari, A.M. Clerici, C.


Zonta, G. Mazzini, P. Dionigi, S. Altieri, S. Bortolussi, P. Bruschi and
F. Fossati Extra-corporeal liver BNCT for the treatment of diffuse metas-
tases: What was learned and what is still to be learned Appl. Radiat. and
Isot. 67, 67-70 (2009).

[26] SRIM - The Stopping and Range of Ions in Solids, by J. F. Ziegler and
J. P. Biersack 2010

[27] S. Stella S. Bortolussi, P. Bruschi, C. Portella, S. Altieri. Measurement of


α particle energy loss in biological tissue below 2 MeV, NIM B, 267 (17),
2938-2943 (2009).

[28] Rene Brun and Fons Rademakers, ROOT - An Object Oriented Data
Analysis Framework, Proceedings AIHENP’96 Workshop, Lausanne, Sep.
1996, NIM A 389 (1997) 81-86. See also http://root.cern.ch/.

[29] L.S. Waters (Ed.), MCNPX USER’S MANUAL Version 2.1.5, Los
Alamos National Laboratory, TPO-E83-G-UG-X-00001, Revision 0, 14
November, 1999.

[30] H. Matsue, Ch. Yonezawa Neutron spectrum correction of k0-factors for


k0-based neutron-induced prompt gamma-ray analysis Journal of Radio-
analytical and Nuclear Chemistry, Vol. 255, 125-129 (2003).

[31] Z.Alfassi Non destructive Elemental Analysis, John Wiley & Sons, 2001.

[32] L.Currie ( Limits for qualitative detection and quantitative determination.


Application to radiochemistry) Anal. Chem. Vol.40, 586-593 (1968).

109
BIBLIOGRAPHY

[33] C. Yonezawa, P. P. Ruska, H. Matsue, M. Magara, T. Adachi Determi-


nation of boron in Japanese geochemical reference samples by neutron-
induced prompt gamma-ray analysis J. of Radioanalytical and Nuclear
Chemistry, Vol.239 (3), 571-575 (1999).
[34] M. Magara, C. Yonezawa Decomposition of prompt gamma-ray spectra in-
cluding the Doppler-broadened peak for boron determination Nuclear In-
struments and Methods in Physics Research A, Vol 411, 130-136 (1998).
[35] H.K.Kraner Fast Neutron Damage in Germanium Detectors, (1980) IEEE
Transactions on Nuclear Science, Vol NS-27, 218-234.
[36] R. H. Pehl, N.W. Madden, J. H. Elliott, T.W.Raudorf RADIATION
DAMAGE RESISTANCE OF REVERSE ELECTRODE GE COAXIAL
DETECTOR IEEE Transactions on Nuclear Science, Vol NS-26 (1979).
[37] E. A.Makey, G.E.Gordon, R.M.Lindstrom, D.L.Anderson Effect of Tar-
get Shape and Neutron scattering on Element Sensitivities for Neutron-
Capture Prompt γ-Ray Activation Analysis Analytical Chemistry, Vol.63,
288 (1991).
[38] E. A.Makey, G.E.Gordon, R.M.Lindstrom, D.L.Anderson Use of Spher-
ical Target To Minimize Effect of Neutron Scattering by Hydrogen in
Neutron-Capture Prompt γ-Ray Activation Analysis Analytical Chem-
istry, Vol.64, 2366 (1992).
[39] E. A.Makey, J.R.D.Coplay, Scattering and Absorption Effects in Neu-
tron Beam Activation Analysis Experiments J of Radioanalytical Nuclear
Chemistry, Vol.167, 127-132 (1993).
[40] E. A.Makey, J.R.D.Coplay Neutron Scattering by Hydrogen in Cold Neu-
tron Prompt Gamma Activation Analysis J of Radioanalytical Nuclear
Chemistry, 181, 321-333 (1994).
[41] R. L. Paul, R. M. Lindstrom, A. E. Heald Cold neutron prompt gamma-ray
activation analysis at NIST - Recent developments Journal of Radioana-
lytical and Nuclear Chemistry 215, 63-68 (1997).
[42] E.A. Mackey, D.L. Anderson ,P.J. Liposky, R.M. Lindstrom, H. Chen-
Mayer, G.P. Lamaze New thermal neutron prompt c-ray activation analy-
sis instrument at the National Institute of Standards and Technology Cen-
ter for Neutron Research Nuclear Instruments and Methods in Physics
Research B 226, 426-440 (2004).
[43] R. M. Lindstrom, Sum and mean. Standard programs for activation anal-
ysis, Biol Trace Elem. Res., Vol 43-45, 597-603 (1994).
[44] J. F. Briesmeister, Ed.,MCNP - A General Monte Carlo N-Particle Trans-
port Code, Version 4C, LA-13709-M (April 2000).

110
BIBLIOGRAPHY

[45] Reactor Shielding Design manual, 1965

[46] T.E. Booth, A Sample Problem for Variance Reduction in MCNP LA-
10363-MS (October 1985).

[47] S. Bortolussi and S. Altieri Thermal neutron irradiation field design for
boron neutron capture therapy of human explanted liver , Medical Physics,
34(12), 4700-4705 (2007).

[48] A.K. Freund, Cross-section of materials used as neutron monochromators


and filters, Nucl. Inst. and Method 213, 495-501 (1983).

[49] Mildner, D. F. R. and Arif, M. and Stone, C. A. and Crawford, R. K., The
neutron transmission of single-crystal sapphire filters, Journal of Applied
Crystallography, Vol.26, 438-447 (1993).

[50] F.Cantargi, F.Sanchez, J.R.Granada, R.E.Mayer, Seccione eficaces de


neutrones thermicos para un filtro de zafiro, AATN congress ”XXXVI
Reunión Anual” (2009).

[51] International Commission on Radiation Units and Measurements, Tissue


Substitutes in Radiation Dosimetry and Measurement, ICRU Report No.
44 (1989).

111
BIBLIOGRAPHY

112
List of publications

• S.Altieri, S.Bortolussi, P.Bruschi, P.Chiari, F.Fossati, S.Stella, U.Prati,


L.Roveda, A.Zonta, C.Zonta, C.Ferrari, A.Clerici, R.Nano, T.Pinelli,
Neutron autoradiography imaging of selective boron uptake in human
metastatic tumours, Appl. Radiat. and Isot. Vol.66; (2008), 1850-1855.

• S. Stella, S. Bortolussi, P. Bruschi, C. Portella, S. Altieri, Measurement


of a particle energy loss in biological tissue below 2 MeV, Nuclear Instru-
ments and Methods in Physics Research B 267, (2009), 2938-2943.

• N.Protti, S.Bortolussi, S.Stella, M.A.Gadan, A.De Bari, F.Ballarini, P.Bruschi,


C.Ferrari, A.M.Clerici, C.Zonta, J.G.Bakeine, P.Dionigi, A.Zonta and
S.Altieri, Calculations of dose distributions in the lungs of a rat model
irradiated in the thermal column of the TRIGA reactor in Pavia, Appl.
Radiat. Isot. Vol 67, (2009), S210-S213.

• L. Menichetti, L. Gaetano, A. Zampolli, S. Del Turco, C. Ferrari, S. Bor-


tolussi, S. Stella, S. Altieri, P.A. Salvadori, L. Cionini In vitro neutron
irradiation of glioma and endothelial cultured cells.Appl. Radiat. Isot.
67, Issue 7-8, (2009),S336-S340.

• C. Ferrari, C. Zonta, L. Cansolino, A.M. Clerici, A. Gaspari, S. Altieri,


S. Bortolussi, S. Stella, P. Bruschi, P. Dionigi, A. Zonta, Selective uptake
of p-boronophenylalanine by osteosarcoma cells for boron neutron capture
therapy; Appl. Radiat. Isot. 67, Issue 7-8, (2009),S341-S344.

• G.J. Bakeine, M. Di Salvo, S. Bortolussi, S. Stella, P. Bruschi, A. Bertolotti,


R. Nano, A. Clerici, C. Ferrari, C. Zonta, A. Marchetti, S. Altieri; Feasi-
bility study on the utilization of boron neutron capture therapy (BNCT)
in a rat model of diffuse lung metastases, Appl. Radiat. Isot. 67, Issue

113
List of publications

7-8, (2009),S332-S335.

• F.Ballarini, J.G.Bakeine, S.Bortolussi, P.Bruschi, A.M.Clerici, A.De Bari,


P.Dionigi, C.Ferrari, M.A.Gadan, N.Protti, S.Stella, C.Zonta, A.Zonta
and S.Altieri, Nuclear physics meets medicine and biology: Boron Neu-
tron Capture Therapy, Proceedings of the 12th International Confer-
ence on Nuclear Reaction Mechanisms, Varenna (Italy), 15-19/06/2009,
CERN-Proceedings-2010-001.

• M.Gadan, S.Bortolussi, P.Bruschi, N.Protti, S.Stella, D.Santoro, F.Ballarini,


J.G.Bakeine, C.Ferrari, A.M.Clerici, L.Cansolino, C.Zonta, A.Zonta and
S.Altieri, Neutron auto-radiography for boron concentration maesurements
in liquids and tissue biological samples, in S.Liberman, A.Kreiner, M.Casal,
eds. New challenges in neutron capture therapy, (p.554) Buenos Aires,
Argentina (2010).

• S.Stella, S.Bortolussi, N.Protti, A.Bazani, F.Ballarini, P.Bruschi and


S.Altieri, The design of a prompt gamma neutron activation analysis
facility at TRIGA Mark II reactor in Pavia, in S.Liberman, A.Kreiner,
M.Casal, eds. New challenges in neutron capture therapy, (p.569) Buenos
Aires, Argentina (2010)

• C.Ferrari, A.M.Clerici, C.Zonta, L.Cansolino, A.Boninella, J.Bakeine,


G.Mazzini, S.Altieri, F.Ballarini, S.Bortolussi, N.Protti, S.Stella, P.Bruschi,
P.Dionigi and A.Zonta, Recurrences following BNCT: a characterization
study of the refractory cell fraction, in S.Liberman, A.Kreiner, M.Casal,
eds. New challenges in neutron capture therapy, (pp. 55-58) Buenos
Aires, Argentina (2010).

• S.Bortolussi, S.Altieri, N.Protti, S.Stella, F.Ballarini, P.Bruschi, M.A.Gadan,


S.I.Thorp, M.Miller, E.C.C.Pozzi, V.A.Trivillin, M.A.Garabalino, A.J.Molinari,
A.Monti Hughes, E.M.Heber, M.E.Itoiz, R.F.Aromando, D.W.Nigg and
A.E.Schwint, 10B measurement by alpha spectroscopy and 10B imaging
by neutron autoradiography as a contribution to the understanding of
BNCT radiobiology in oral cancer and liver metastases animal models,
in S.Liberman, A.Kreiner, M.Casal, eds. New challenges in neutron cap-
ture therapy, 59-62 Buenos Aires, Argentina (2010).

• S.Geninatti-Crich, D.Alberti, I.Szabo, A.Deagostino, A.Toppino, A.Barge,


F.Ballarini, S.Bortolussi, P.Bruschi, N.Protti, S.Stella, S.Altieri, P.Venturello
and S.Aime, MR Imaging guided NCT by a dual Gd/B agent targeted to

114
List of publications

tumor cells via upregulated LDL transporters, Chem. Eur. J. 17, 8479-
8486 (2011).

• M.Corti, M.Bonora, F.Borsa, S.Bortolussi, N.Protti, D.Santoro, S.Stella,


S.Altieri, C.Zonta, A.M.Clerici, L.Cansolino, C.Ferrari, P.Dionigi, A.Marchetti,
G.Zanoni and G.Vidari, NMR Study of Gd-Based Nanoparticles to Tag
Boron Compounds in Boron Neutron Capture Therapy (BNCT), J. Appl.
Phys. 109, 07B302 (2011).

• S.Stella, S.Bortolussi, N.Protti, A.Bazani, F.Ballarini, P.Bruschi, and


S.Altieri: The design of a prompt gamma neutron activation analysis fa-
cility for BNCT purpose at the Triga Mark II reactor in Pavia; Proceed-
ing of RRFM (2011): European Research Reactor Conference - Roma.

• N.Protti, F.Ballarini, S.Bortolussi, P.Bruschi, S.Stella, S.Altieri, A.De


Bari, J.Bakeine, L.Cansolino, A.M.Clerici, C.Zonta, C.Ferrari, V.Conte,
J.Esposito, L.De Nardo, D.Moro, C.Fabris, M.Soncin, P.Colautti, G.Jori,
S.Geninatti Crich, D.Alberti and S.Aime, Biological tests for the Boron
Neutron Capture Therapy research at the TRIGA Mark II reactor in
Pavia, Proceeding of RRFM: European Research Reactor Conference,
Rome, 20-24/03/2011.

• N.Protti, F.Ballarini, S.Bortolussi, S.Stella, P.Bruschi, S.Geninatti-Crich,


D.Alberti, S.Aime and S.Altieri, Dose estimation in B16 tumour bearing
mice for future irradiation in the thermal column of the TRIGA reac-
tor after B/Gd/LDL adduct infusion, Appl. Radiat. Isot.69, Issue 12,
(2011), 1842-1845.

• F.Ballarini, J.Bakeine, S.Bortolussi, P.Bruschi, L.Cansolino, A.M.Clerici,


C.Ferrari, N.Protti, S.Stella, A.Zonta, C.Zonta and S.Altieri, Cell death
following BNCT: a theoretical approach based on Monte Carlo simula-
tions, Appl. Radiat. Isot.69, Issue 12, (2011), 1745-1747.

• M. Bonora, M. Corti, F. Borsa, S. Bortolussi, N. Protti, D. Santoro, S.


Stella, S. Altieri, C. Zonta, A.M. Clerici, L. Cansolino, C. Ferrari, P.
Dionigi, A. Porta, G. Zanoni, G. Vidari;1H and 10B NMR and MRI in-
vestigation of boron- and gadolinium-boron compounds in boron neutron
capture therapy.Appl. Radiat. Isot. 69, Issue 12, (2011),1702-1705.

115
List of publications

• S.Mandal, G.J.Bakeine, S.Krol, C.Ferrari, A.M.Clerici, C.Zonta, L.Cansolino,


F.Ballarini, S.Bortolussi, S.Stella, N.Protti, P.Bruschi and S.Altieri, De-
sign, development and characterization of a multi-functionalized gold
nanoparticles for biodection and targeted boron delivery in BNCT ap-
plications, Appl. Radiat. Isot.69, Issue 69, (2011), 1692-1697.

• N.Protti, F.Ballarini, S.Bortolussi, P.Bruschi, S.Stella, S.Geninatti, Crich,


D.Alberti, S.Aime and S.Altieri, In vivo efficacy test of the BNCT treat-
ment mediated by an innovative dual Gadolinium/Boron agent, Pro-
ceeding of 6th Young Researchers BNCT Meeting, 4-6 December 2011,
Hsinchu, Taiwan.

• S.Stella, F.Ballarini, S.Bortolussi, P.Bruschi, N.Protti, and S.Altieri; De-


sign of a facility for boron measurement based on prompt gamma neutron
activation analysis. Proceeding of 6th Young Researchers BNCT Meet-
ing, 4-6 December 2011, Hsinchu, Taiwan.

• M.A. Gadan, S. Bortolussi, I. Postuma, F. Ballarini, P. Bruschi, N.


Protti, D. Santoro, S. Stella, L. Cansolino, A. Clerici, C. Ferrari, A.
Zonta, C. Zonta, S.Altieri; Set-up and calibration of a Method to Mea-
sure 10B Concentration in Biological Samples by Neutron Autoradiogra-
phy.Nucl. Instr. and Meth. in Phys. Res. B, (2011).

116
Aknowledgments

My first thanks goes to my supervisor Saverio Alteri for supporting and en-
couraging me during these years, thank you for the possibility you gave me to
work with your group!
I am grateful to Kent Riley for the time he spent to read my thesis, for his
corrections and suggestions.

I want to thank all my group for the frendly and collaborative work atmos-
fere that I have always found. Thanks to Silva for her friendship and for her
constant enthusiasm in each occasion; thank you for helping me many times
both in professional and in personal issues, with you I found not only a great
collegue but also a sincere friend! Thank you Nicoletta for sharing with me the
”thesis writing” syndrome and for being a great neighbour; thank you for your
friendship and you helpfulness whenever I asked you! Thank you Francesca
for the time we spent together, good luck for your new adventure! Thank you
to Piero Bruschi, grazie per la tua disponibilitá e per le discussioni in labo-
ratorio sui temi piu’ svariati. Many thanks also to the medical and biology
staff; James, Cinzia, Anna and Celicia. Thankyou to all the guys with whom I
shared the office: thankyou to Mario and Ruben for their wonderful alfajores,
thankyou to Concetta and Alessia, it was nice and productive working with
you! And thanks to Ian, goodluck for your thesis!

Sincerely thanks to Dick Lindstrom, Greg Downing e Liz Makey for receiv-
ing me at NIST with such enthusiasm, it was an honour for me meeting you!
Thank you for the time you spent with me explaining many things and intro-
ducing me many persons, you made me feel immediately part of the group.
Thankyou to Jeremy for his precious hint about Monte Carlo simulation.
Many thanks to the guys with whom I shared the breck time at NIST, you
were precious! Thanks to Russel, Richard, Shruthi, Dmitry, special thanks to
Sophia for all the nice time we spent together chatting, roaming in Washington
and shopping, without you my period in USA would not be so fun! Finally
many many thanks to Cinthia for welcoming me with open harms at her house.

I want also to thank all the guys that work at LENA; un grazie particolare
a Michele per avermi telefonato tutte le volte che la mia macchina rischiava di
rimanere bloccata al parcheggio!

117
Aknowledgments

Many many thanks to all the friends I met in the department: Susanna,
Luca, Lorenzo, Davide, Francis, Alessandro, Jacopo, Andrea; grazie a Claudio
per aver sventato tutti i possibili cali di zucchero con i tuoi fantastici dolci da
tutto il mondo, continua cosi!

Grazie infinite alle irriducibili Laura, Giusi e Mary, grazie per esserci sem-
pre tutte le volte che torno nella mia Martano! Grazie mille “cumare“ per
le lunghissime telefonate spese a parlare dei nostri progetti sempre con tanto
entusiasmo, grazie per essere sempre un punto di riferimento tutte le volte che
ho bisogno di sfogarmi.

Infine devo ringraziare tutta la mia famiglia allargata per avermi sostenuto
ed aiutato tutte le volte che ne ho avuto bisogno, aiutandomi a conciliare il
mio corso di dottorato assieme all’importantissimo ruolo di mamma; senza di
voi, questa bella esperienza non sarebbe stata possibile! Grazie a Lucia e Luigi
per esservi precipitati a Pavia tutte le volte che sono dovuta partire, siete dei
nonni fantastici! A voi, papá e mamma l’ennesimo grazie per aver appoggiato
e sostenuto in tutte le mie scelte, per essermi sempre stati vicino e per avermi
infuso costantemente il coraggio e la fiducia che servono per percorrere il tor-
tuoso e affascinante viaggio che é la nostra vita!
Grazie ai nonni per il loro calore e per la loro saggezza.
Grazie a Maria Chiara e Francesco, per avermi accolto sin da subito come una
sorella. Grazie mille a tutti voi cari fratelli, per le risate, le discussioni, per i
pranzi a casa mia, per le chiaccherate al telefono del piu e del meno...grazie
Antonio, Marco, Simone e Fabrizio, non cambiate mai!

Un grazie infinite alla mia piccola Elisa che ha stravolto e arricchito la mia
vita; grazie per la tua pazienza e per tutte le volte che mi hai staccato la spina
del computer dicendo ”mamma basta sudiare!”. Grazie mille al mio Antonio
per tutto l’amore, il sostegno e la fiducia che hai sempre saputo darmi! A voi
due, Elisa ed Antonio, grazie per esserci!

118