UPTEC ES07 002
Examensarbete 20 p Januari 2007
The intensity profile of a neutron beam of 96 MeV at TSL
Andreas Eriksson
Teknisk- naturvetenskaplig fakultet UTH-enheten
Besöksadress:
Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0
Postadress:
Box 536 751 21 Uppsala
Telefon:
018 – 471 30 03
Telefax:
018 – 471 30 00
Hemsida:
http://www.teknat.uu.se/student
Abstract
The intensity profile of a neutron beam of 96 MeV at TSL
Andreas Eriksson
The interest in high-energy neutrons is rapidly growing due to applications that primarily fall into three sectors: nuclear energy and waste management, medicine and electronics. To meet these demands, a neutron-beam facility has recently been developed at The Svedberg Laboratory (TSL). When the neutron beam is used at the TSL facility it is important to know how the intensity varies over the collimator opening.
In this thesis the neutron flux over two different collimator openings (diameters of 158 mm and 300 mm) has been reconstructed. For both openings, the results are compatible with uniform intensity.
Handledare: Stephan Pomp Ämnesgranskare: Jan Blomgren Examinator: Ulla Tengblad
ISSN: 1650-8300, UPTEC ES07 002
Sammanfattning
Intresset för högenergirika neutroner över 20 MeV växer snabbt idag, då olika tillämpningar innefattande dessa neutroner är under utveckling eller har identifierats. Dessa tillämpningar finns vanligast inom tre olika områden, kärnavfallshantering, sjukvård samt elektronik.
Inom området avfallshantering finns förhoppningar att med snabba neutroner göra det möjligt att transmutera kärnavfall till ämnen med kortare halveringstider. Skulle detta kunna göras ekonomiskt försvarbart, så skulle den erfordrade lagringstiden för kärnavfall kunna bli mycket kortare. Den långa lagringstiden är annars idag ett av huvudargumenten mot kärnkraft.
Att behandla vissa cancertumörer med snabba neutroner har visat sig vara mer effektivt än vanliga kommersiella metoder. Detta har medfört att neutronterapi idag är den näst största icke-kommersiella behandlingsformen för hjärntumörer.
Den tekniska utvecklingen, har medfört teknik som är mer känslig för neutronstrålning vilket främst har märkts hos flygplan eftersom de utsätts för hög kosmisk strålning. Den mest kända effekten idag, är den så kallade ”single-event” effekten. Den uppkommer genom att en
neutron skapar en kärnreaktion i en kiseldel tillhörande en minneskomponent. Den skapade laddningen från reaktionen ändrar minnesinnehållet vilket medför till felaktigt resulterad data.
Denna effekt kallas för ”single-event” och kan t.ex. orsaka fel på flygplans färddatorer, med ökad risk för krascher. Det har även visat sig att elektronik på marken är sårbar för den kosmiska strålningen, som då tycks bidra till hårdvarufel.
I The Svedberglaboratoriet (TSL) i Uppsala finns en anläggning som kan bilda neutronstrålar för bland annat kommersiell testning av kraftelektronik. För att erhålla de önskade
neutronstrålarna i denna anläggning måste flera steg genomgås. Det första steget är att med en cyklotron skapar protonstrålar, i energiintervallet 20 till 180 MeV. Genom att fokusera dessa strålar på en litiumbit, så kan vissa av protonerna frigöra neutroner. Dock, så reagerar inte majoriteten av protonerna med litiumkärnorna och måste därför länkas av från huvudstrålen efter litiumbiten för att man ska erhålla en ren neutronstråle. Detta görs med hjälp av en magnet efter litiumbiten, som böjer av de laddade protonerna ner i en tio meter lång dumpningstunnel.
Den rena neutronstrålen passerar sedan en kollimator som har till uppgift att forma strålen geometriskt. Kollimatorn fungerar som en tunnel som ska avskärma alla neutroner som hamnar utanför själva öppningen.
Efter den meterlånga kollimatorn, når de kvarvarande neutronerna laboratoriet där den används vid olika experiment.
Det är önskvärt att kunna använda hela strålens tvärsnittsyta vid dess tillämpningar. Detta medför att det är viktigt att veta hur intensiteten ser ut över denna tvärsnittsyta. Om den är högre i mitten jämfört med kanterna, så skulle det medföra en ojämn bestrålning vilket inte är önskvärt. T.ex., om flera tekniska komponenter är placerade i strålens väg, så skulle de bli bestrålade med olika mängd neutroner om en ojämn profil användes vilket skulle medföra att de olika komponenterna inte kan jämföras med varandra. Därför är det viktigt att säkerställa hur profilen ser ut i TSL och om den varierar då olika kollimatoröppningar används.
För att undersöka hur strålen ser ut, så har två olika kollimatorer används, det första med en öppning på ∅158 mm och det andra med en öppning på ∅300 mm.
Neutroner är oladdade partiklar vilket gör att de är svåra att detektera. Därför sattes plast för öppningarna i de två experimenten, vilket medför att en del av neutronerna sprider protoner elastiskt. Dessa protoner kan sedan detekteras med en detektor eftersom de är laddade. Har detektorn även egenskapen att kunna mäta protonernas riktning, så blir det möjligt att avgöra varifrån de kommer. Dessa egenskaper har detektorn SCANDAL och användes därför i detta arbete för att rekonstruera neutronprofilerna över de olika kollimatoröppningarna.
Det visade sig att för de båda kollimatorena att intensiteten är konstant över öppningen.
Mätfelen blev dock något större med den större öppningen vilket gjorde att den blev svårare att analysera. Detta beror förmodligen på att den använda detektionsuppställningen inte var optimalt utformad samt att det fanns järnringar placerade i början av kollimatorn som skulle kunna påverka profilens utseende.
Skulle intensiteten behöva undersökas noggrannare, så rekommenderas det att utföra Monte Carlo simuleringar. Dessa simuleringar kan bestämma skillnad i neutronflödet vid
förändringar i kollimatorn, t.ex. om järnringarna plockas bort. Dessutom kan nya experiment med bättre detektionsförhållanden utföras.
Contents
1 The new Uppsala neutron beam facility ... 2
1.1 Introduction... 2
1.2 High-energy neutron production... 3
1.3 The TSL neutron beam ... 4
1.3.1 The lithium target ... 5
1.3.2 The collimator... 5
2 Detection of neutrons... 6
2.1 Elastic np scattering ... 6
3 The SCANDAL setup ... 9
3.1 CsI hodoscope... 10
3.2 Drift chambers ... 11
3.3 The trigger scintillators ... 11
4 The experimental campaign... 12
5 Data analysis... 14
5.1 Data cuts... 14
5.2 Background elimination... 14
5.3 The Time Of Flight (TOF) cut - part 1 ... 15
5.4 CsI detection ... 16
5.5 Particle identification... 18
5.6 The TOF cut – part 2... 19
6 Profile reconstruction ... 20
6.1 Result after cuts... 21
7 Resolution folding ... 22
7.1 Gauss simulation... 22
7.2 Fitting simulated profiles with the experimental ones... 22
7.3 Intensity variation over the collimator opening ... 24
7.4 The 40 mm cut ... 26
8 Setup 2 - the ∅300 mm experiment... 28
8.1 Data cuts for setup 2 ... 28
8.2 Intensity variation over the collimator opening ... 29
9 Geometrical effects... 32
10 Discussion... 34
11 Summary and outlook ... 37
12 Acknowledgements ... 37
References... 38
1 The new Uppsala neutron beam facility
1.1 Introduction
The interest in neutrons above 20 MeV is rapidly growing, since a number of potential large-scale applications involving fast neutrons are under development, or at least have been identified. These applications primarily fall into three sectors: nuclear energy and waste management (1), medicine (2) and electronics (3) [1].
1. The recent development of high-intensity proton accelerators has resulted in ideas to use subcritical reactors, fed by external spallation-produced neutrons, for transmutation of waste from nuclear power reactors or nuclear weapons material.
This might result in less problematic waste material and/or energy production [2].
2. Neutron therapy for treatment of cancer tumours is today the second largest used non-conventional therapy worldwide. Very good treatment results from this kind of therapy have been achieved for some tumors that cannot be treated successfully with conventional radiation treatment, i.e., by photons and electrons [3, 4].
Neutrons at aircraft altitudes give a significant radiation dose to aeroplane personnel. This poses a relatively new dosimetry problem, which is currently under intensive investigation [5].
3. During the last few years, it has become evident that electronics in aeroplanes suffer effects from cosmic-ray neutrons. The presently most well-known effect is that a neutron can induce a nuclear reaction in the silicon substrate of a memory device, releasing a free charge, which in turn flips the memory content. This effect is called a single-event upset, which causes a not wanted random re- programming. Similar effects causing hardware damage have recently been identified also on ground level [6, 7].
Obviously, fast neutrons may have a major influence in certain areas in the future. In the energy region up to 20 MeV, there exist extensive evaluated data libraries, which have been established for the development of nuclear fission applications. Very little high- quality neutron-induced data exist above this energy today. So, to achieve the best possibly results in these different application areas, it is important to get more reaction data of neutrons with energies above 20 MeV. Such data can be obtained at, e.g., the neutron beam facility at The Svedberg Laboratory (TSL) in Uppsala.
The purpose and goal of this thesis is to determine the beam profile for the high-energy neutrons in the neutron beam when the facility produces neutrons of 96 MeV. The profile is the intensity of the neutrons across the collimator opening. The collimator is the part of the TSL facility that shapes the beam and will be discussed closer in chapter 1.3.2.
1.2 High-energy neutron production
There are traditionally two different types of neutron beam facilities for energies above 50 MeV, white sources and monoenergetic ones. White spallation sources have been used at, e.g., Los Alamos [8] and also a new source has recently been installed at CERN [9].
Spallation neutrons from a white source arise when high-energy protons bombard heavy nuclei. Some neutrons are knocked out, in a nuclear reaction process called spallation.
Other neutrons are "boiled off" as the bombarded nucleus heats up. For a 1 GeV proton striking into lead, 20 to 30 neutrons are emitted and these neutrons are distributed all over the energy spectrum. However, with highest intensity at lower neutron energies, it is similar to the cosmic-ray induced neutron spectrum [10].
So-called quasi-monoenergetic neutron beams use specific nuclear reactions such as the
7Li(p,nx) reaction. This type of beam is used at TSL where protons from the cyclotron bombard 7Li targets, generating neutrons. The 7Li(p,nx) reaction produces a neutron spectrum consisting of a full-energy peak and a continuum of neutrons at lower energies, the so-called reaction tail. About 40 % of the neutrons are obtained in the high-energy peak and the rest of the neutrons are distributed in the reaction tail [11]. A 7Li(p,nx) reaction spectrum for incoming protons of 97.9 MeV is shown in figure 1.
The high-energy peak is at a slightly lower energy than the initial proton energy because of the Q-value for the 7Li(p,n)7Be reaction (Q=1.64 MeV). The energy peak also depends on the thickness of the lithium target. The typically energy width is about 1-4 MeV [1].
The resulting high-energy peak is the reason why quasi-monoenergetic sources are being used. Instead of using neutrons distributed over the whole spectrum, it is possible to use neutrons at specific energies. Events from neutrons in the reaction tail can be removed using time-of-flight or unfolding techniques. Therefore only neutrons from the high- energy peak will be used in this thesis for the characterization of the beam and the rest of the neutrons with lower energies will be discarded.
Figure 1: The energy spectrum for monoenergetic neutrons produced by protons of 97.9 MeV in an eight mm thick lithium target [11].
1.3 The TSL neutron beam
In early 2004, the first measurements for the beam characterization of the new neutron beam facility at TSL were performed. The new facility, which is shown in figure 2, utilizes an already existing cyclotron and a flexible lithium target in a rebuilt beam line.
Besides cross-section measurements, the new facility has been designed specially to provide optimal conditions for testing of single-event effects in electronics and for dosimetry development [12].
Figure 2: The neutron beam facility [13]. For the measurements discussed in this work, the MEDLEY setup has been moved out of the beam. The used special arrangement of
the SCANDAL setup is discussed later.
There are two major steps that create and shape the neutron beam at the TSL facility, the production of the neutrons by protons and the collimation that defines the beam
geometrically.
1.3.1 The lithium target
The proton beam is produced by the Gustav Werner cyclotron with the energy variable in the 20-180 MeV range. The beam is incident on a target of lithium that is enriched to 99.99% in 7Li. The available targets are 2, 4, 8, 16 and 24 mm thick and are rectangular in shape, 20×32 mm. The targets are mounted in a remotely controlled water-cooled copper frame [12].
Most of the protons do not react inside the Li-target. Therefore a magnet deflects the protons and bends them into a 10 meter long dumping line after the target. In the end of the tunnel the protons are stopped in a water-cooled graphite beam dump. With all protons removed, the beam consists of almost only neutrons resulting from the 7Li(p,nx) reaction when it exits the collimator[12].
1.3.2 The collimator
A cylindrically shaped iron collimator block with a diameter of 50 cm and a length of 100 cm forms the neutron beam geometrically. A modular construction of the collimator allows the user to adjust the diameter of the collimator to the needs of a specific
experiment. Openings with diameters of 2, 3, 5.4, 10, 15, 20 and 30 cm are available. In addition it is possible to provide a beam of quadratic shape with a 1 cm2 opening. In principle, any beam shape can be obtained upon request up to 30 cm maximum width. To achieve efficient shielding from the production target region, the collimator is surrounded by concrete to form the end wall of the production line towards the experimental area [12]. While this shielding is sufficient for most experiments, it has turned out that it is not adequate for all MEDLEY experiments. Therefore, the collimator block is presently being reconstructed to consist of iron only. First runs with the new collimator will be performed in February 2007.
2 Detection of neutrons
The goal of this work is to measure the neutron beam profile at the collimator exit, a task that requires detection of neutrons with sufficient spatial resolution.
The neutron is a neutral particle. This absence of charge makes neutrons harder to detect than particles with charges because neutrons cannot ionize atoms. High-energy particles with charge can dislodge electrons from the atoms it strikes, producing pairs of ions. This ionization can create light or an electrical signal, which can be read.
Neutrons are detected in two steps. First they have to undergo a nuclear reaction that produces energetic charged particles that can be detected. This can happen through several nuclear reactions. Examples of possible reactions
• 10B + n → 7Li + α
• 3He + n → 3H + p
• Fission, e.g., 235U + n → 140Xe + 94Sr + 2n
[14]
To obtain charged particles in this project, elastic scattering with neutrons have been used.
2.1 Elastic np scattering
Nuclear reactions can be characterized by two different categories, elastic and non-elastic ones. For elastic scattering, there is no difference in the total kinetic energy and
momentum before and after a reaction [15]. This is shown in figure 3, where an incoming particle collides with a particle without momentum. Kinetic energy from the first particle is transferred to the second one without any energy loss.
Figure 3: Elastic scattering of two particles, where the kinetic energy (Ek) and the momentum (p) for the system is conserved. In this project the incoming particle is a neutron from the beam and the particle with p2=0 is the target proton.
p1 + p2 = p´1 + p´2
Ek1 + Ek2 = E´k1 + E´k2
The total kinetic energy is conserved when a neutron elastically scatters a proton. This is shown in figure 4 for a neutron with the kinetic energy of 96 MeV.
To get this elastic scattering, a plastic (CH2) target is placed directly after the collimator.
When a neutron enters the plastic, three different things can happen. The first and most probable (>99 %), is that the neutron passes through the plastic undisrupted. The second alternative is that the neutron reacts with or scatters off a 12C nucleus. The final
alternative, which is the wanted one in this thesis, occurs when a neutron hits a proton in the plastic (hydrogen is a proton with an electron) and scatters off the proton.
The mass of a neutron (939.565 MeV) and a proton (938.272 MeV) is about the same.
This similarity in mass results in an almost 100% energy transfer from the neutron to the proton if the scattering angle is 0 degrees, i.e., the neutron “pushes” away the proton in the same direction as the neutron moved in before the reaction. With increasing scattering angle, to 90°, the energy transfer between the two particles decreases, as shown in figure 4.
Figure 4: The energy of elastically scattered protons by 96 MeV neutrons. Also the kinetic energy of the neutron for different angles and the total kinetic energy are plotted.
If the neutron hits a carbon nucleus in the plastic target, a reaction can be induced which may result in a proton (12C(n,p)). This reaction has a Q value of –12.6 MeV, which makes it easy to separate these protons from protons from elastic np scattering at small angles.
With increasing angles, the energies for the protons with different origin get more equal, which make them harder to separate from each other [16].
The SCANDAL setup described in next chapter has been used to reconstruct the origin of elastically scattered protons from the plastic target. This is done by measuring the
trajectory for each proton, which is shown in figure 5.
3 The SCANDAL setup
To detect the protons from neutron scattering, a detection setup called SCAttered Nucleon Detection AssembLy (SCANDAL) is placed at a distance from the CH2 target.
SCANDAL can also be used for (n,n) and (n,d) reaction experiments. Figure 2 shows a setup with two SCANDAL arms on each side of the beam. In this thesis the left detector is removed and only the right one is being used (see figure 5). The detector is not placed directly in the beam because the high intensity of the beam would otherwise contribute to pile-up effects, where events are overlapping each other and too high energies are
recorded in the setup.
The SCANDAL detector is supposed to detect charged particles and measure their energies and directions. By measuring a number of interaction points the trajectory for the particle can be calculated and this makes it possible to track down the geometrical origin of each neutron reaction. Finally SCANDAL must have the ability to separate protons from other charged particles like deuterons. SCANDAL must attain all these properties to measure the profile for the high-energy neutrons over the collimator opening. Therefore, the setup consists of several different detector elements.
CsI crystals are placed at the end of SCANDAL with the purpose to stop and measure the energy of each detected particle. The setup contains two drift chambers, so the trajectory for each particle can be determined. Two trigger scintillators are used for both detecting events and also measure the energy loss of the particles when they pass the triggers.
These components are presented in figure 5 and are described more detailed in the chapters 3.1-3.3.
Figure 5: Schematic layout of SCANDAL and its placement relative to the neutron beam. The figure is not to scale.
3.1 CsI hodoscope
SCANDAL has in total 12 CsI detectors at the end of the setup (see figure 5). These detectors are trapezoidal in shape, 30 cm high with a 7 × 7 cm2 cross-section area at one end, and a 5 × 5 cm2 area at the other end. By placing the detectors in alternating
directions, they create together a 72 cm wide CsI hodoscope. The trapezoidal shape is not optimal, but they are being reused from another experiment. The CsI crystals are doped with sodium, which makes them more resistant to radiation damage [1]. For the present experiment, this is however not an important issue.
Every crystal is equipped with a photomultiplier (PM) tube placed at the 7 × 7 cm2 cross- section end (see figure 6). When a proton hits the CsI crystal, the proton will loose energy when it ionizes the crystal. This ionization creates photons that can be detected by the PM tube. The PM tube converts the incoming photons to an electrical signal. The pulse height of the electrical signal is proportional to the number of incoming photons, which in turn is proportional to the energy deposition of the charged particles in the CsI [1].
It is very important that the protons are completely stopped in the crystals because otherwise it does not measure the complete energy and the event reconstruction is incomplete. If a proton hits close to an edge, there is a possibility that the proton leaves the crystal before it is stopped and the crystal does not detect the full energy of the proton. Therefore, it is desirable to eliminate the hits of protons that are too close to the edges of the crystals. The geometrical cut used for each crystal is shown in figure 6, where all events outside the grey area are rejected.
Figure 6: Geometrical cut of events in a CsI crystal.
3.2 Drift chambers
To be able to measure the proton trajectories, the SCANDAL setup is equipped with two drift chambers in front of the CsI detectors. Every drift chamber is filled with gas and contains an active detection area of 960 × 192 mm2. This detection area contains wires that create 40 horizontal and 8 vertical drift cells, where each cell is 24 mm wide. When the scattered protons pass through the drift chambers, the gas gets ionized and creates free electrons that can be detected by the wires. The time it takes for the electron to reach a certain wire is measured and the coordinate of the proton can be computed. By using the two coordinates a proton creates in the drift chambers, the trajectory can be
reconstructed, all the way from the drift chambers to both the CH2 target and the CsI hodoscope. For optimum conditions, the position resolution is 0.3 mm (FWHM) for each coordinate [17].
3.3 The trigger scintillators
There are two trigger scintillators in the SCANDAL setup, with the first placed before the drift chambers and the second one between the CsI hodoscope and the drift chambers.
These detectors serve the purpose of defining events. Both scintillators are 30 cm high and are 2 mm thick. The widths, however, are different and are 60 cm for the first and 75 cm for the second one [1].
The trigger scintillators have two PM tubes each mounted adjacent to each other on one of the longer horizontal sides. The signals from the PM tubes are lead to a remote counting room, where they are handled using mainly standard electronic units. When both triggers detect a particle, a computer starts saving events detected by SCANDAL to magnetic tapes. All these stored events create a raw data file [1].
During the time the computer is busy with a detected event it cannot handle other
detected events by SCANDAL. This time while the computer rejects new events is called the computer dead time.
4 The experimental campaign
Two experimental runs with different collimator openings have been evaluated in this thesis. Different openings are used to determine if the intensity profile over the opening differs for different openings. These experiments were performed at the TSL facility, during February 2004. The beam energy was 96 MeV and the lithium target was 4 mm thick. A summary of the experimental runs is shown in table 1.
Setup File name Collimator opening
(mm) Run time
(h:min) Computer busy
(h:min) CH2 target
run 25 ∅ 158 0:53 0:16 YES
1 run 26 ∅ 158 0:48 0:14 NO
run 28 ∅ 300 1:44 0:32 YES
2 run 29 ∅ 300 0:40 0:09 NO
Table 1: Information of the different experimental runs.
For each setup, two different runs were executed, one with aCH2 target and one without to measure the background.
The SCANDAL setup was placed differently for the 158 and 300 mm opening measurements, which is presented in the figures 7, 8 and 9. To be able to measure the intensity of a beam profile, several different evaluation steps must be implemented. When these are presented in this thesis, they will be illustrated with the data from setup 1. The evaluation steps are practically the same for the two setups but to avoid repetition, the evaluation of setup 2 is presented in chapter 8 less detailedly than for setup 1.
Figure 7: When the target runs (run 25 & 28) are seen from the side, the collimator openings are the only thing that differs.
Figure 8: The setup for run 25 as seen from above.
Figure 9: The setup for run 28 as seen from above.
5 Data analysis
The data acquisition system SVEDAQ [18] was used to collect and save the raw data from the experiment. The principle for this collection is described in chapter 3.3. To be able to analyze the resulting data from the experiment, the analysis program ROOT [19]
was used. Therefore, the raw data must be converted to ROOT files. This conversion to ROOT files is made with a program called s2r4s (SVEDAQ to ROOT for SCANDAL).
This program has sub-programs that describe the setup after the target, calculate the trajectory for the protons etc.
5.1 Data cuts
From each plastic target experiment, the major part of detected events (charged particles) by SCANDAL is unwanted particles. Protons scattered by low-energy neutrons and scattered particles from other present material than the plastic are examples of these unwanted events. Therefore, before the beam profile can be reconstructed all these unwanted events must be discarded. This is done with several different cuts and elimination methods.
5.2 Background elimination
Using data only from a run with a CH2 target is not enough to reconstruct the profile of the neutron beam. When the facility is running, SCANDAL does not just detect particles with origin from the CH2 target. After the deflection magnet, there is a probability for the neutrons to interact with, e.g., nuclei in the collimator. These reactions may produce charged particles that can be detected in the SCANDAL setup. The cross section and therefore the amount of the background radiation depends on the energy of the neutrons.
The amount of background should be constant, no matter whether the run is made with or without a CH2 target as long as all the other parameters are kept the same. To be able to determine the amount of background that affects the setup, an experiment without a CH2
target must be executed. This background run may also give an indication of the shielding efficiency of the material after the deflection magnet, besides to eliminate background from the target run.
To be able to eliminate the background radiation from setup 1, two things must be considered. For the first, the experimental run time was longer for the run without the target than for the one with the target, which is shown in table 1. Therefore, the target run has shorter radiation exposure time. To make these two runs comparable with each other, the number of detected events from the longer run must be scaled down with the ratio in exposure time between the runs. However, the exposure time is not the same as the running time of the experiment. When SCANDAL detects a particle, the computer system cannot store other events during the time it handles the first event, which is described in chapter 3.3. This computer dead time must be taken into account.
The background run must undergo the same data cuts as the target run before the background can be subtracted from the target run.
5.3 The Time Of Flight (TOF) cut - part 1
The neutron production in the lithium target results in neutrons with different energies.
By using a method called Time Of Flight (TOF), high-energy neutrons can be separated from low-energy neutrons. When the cyclotron at TSL produces the proton beam, this beam is not continuous. Instead, groups of protons arrive from the cyclotron in short pulses. Knowing the exact time when every neutron is produced makes it possible to measure the flight time to the SCANDAL setup. High-energy neutrons travel with higher velocity than neutrons with lower energies. Thus, if a cut on the transportation time is set, separation between neutrons with different energies is achieved [1].
For longer flight paths, the time difference between different energies gets larger and makes it easier to discard unwanted neutrons. This is true as long the energy difference does not become so large that the fast neutrons from one pulse reach the detector at the same time as the slower neutrons from the previous pulse, the so-called wrap around effect. This overlapping would make it impossible to separate high-energy neutrons from low-energy neutrons with the TOF method. Scattered deuterons in the CH2 target can also create wrap around, because they travel slower than protons with the same energy.
Therefore, deuterons must be discarded, which is described in section 5.5 [1].
In this thesis the TOF value is presented as the time between a trigger signal and the next RF signal from the cyclotron. This subtraction makes the faster neutrons corresponding to higher TOF values, which is backwards relative the ordinary TOF value. The TOF
distribution for events from setup 1 is shown in figure 10.
Figure 10: The TOF spectrum.
It seems that the high-energy neutrons corresponds to a TOF value around 340 ns. It is difficult to determine where to set the cut for high-energy neutrons from this plot alone.
Therefore, all events under 320 ns will be discarded for the moment. This cut eliminates many unwanted events, which makes other cut methods easier to evaluate. After the other cuts have been made, it will be possible to determine a better TOF cut.
5.4 CsI detection
There are 12 CsI detector crystals in the SCANDAL setup (see section 3.1) and they are numbered from left to the right starting with zero. Number zero detects protons with the smallest scattering angle and number eleven at the largest angles. From the earlier discussion in chapter 2.1, it is known that protons from np scattering with low angles have higher energies than protons with larger scattering angles. Also the cross section for np scattering is decreasing with larger angles, which makes the number of events from np scattering lower in the crystals with higher numbers.
The energies for the detected protons in every crystal vary significantly. The major part of these events is unwanted hits. The only wanted events are the elastically scattered protons by neutrons from the high-energy peak. If the spectrum of detected energy is plotted for every crystal, a peak should be found in each crystal that corresponds to the neutron high-energy peak.
In figure 11, the detected energy for CsI crystal 8 is presented. This crystal is selected because it got one of the most illustrative peaks. The darker line is the energy detected from the target run and the other is from the background run. The TOF cut chosen in the previous chapter is used to make it easier to find the wanted peak.
Figure 11: The energies of the detected particles in CsI crystal 8.
In figure 11 the high-energy peak is visible but can be even more pronounced if the background is eliminated, which is done in figure 12.
Figure 12: Background-subtracted data for CsI crystal 8.
To determine which energy range that will be selected for every specific crystal, a Gaussian is fitted for each peak. This fitting gives a mean value and a standard deviation for each crystal. In this thesis, the adopted energy range of the peak ranges from 1σ below the centroid to 1σ above.
Using one sigma, 68 % of the events in a Gaussian distribution are accepted [20].
Therefore, with this narrow energy range, 32 % of the wanted events are discarded, but this also makes the contribution from unwanted events less. Each energy range for every crystal is presented in table 2.
CsI crystal
Mean value
(MeV) σ (MeV) Energy interval (MeV)
0 - - -
1 - - -
2 97.0 5.0 92.0 - 102.0
3 101.7 2.5 99.2 - 104.2
4 85.9 2.6 83.3 - 88.5
5 82.8 3.2 79.6 - 86
6 82.0 3.0 79.0 - 85.0
7 93.7 2.6 91.1 - 96.3
8 86.1 2.0 84.1 - 88.1
9 94.2 4.0 90.2 - 98.2
10 88.5 3.5 85.0 - 92.0
11 96.2 3.4 92.8 - 99.6
Table 2: The energy ranges for setup 1 determined with Gaussian fitting, using one sigma.
It might seem strange and impossible to detect particles with energies above 96 MeV, which it also is. The energies for the CsI detectors have not been calibrated for these
experiments and are therefore presenting false energies. This absence of calibration does not matter for this thesis. The only necessary thing for the profile characterization is to be able to determine which events that are elastically scattered by high-energy neutrons and not to determinate the exact energy for each particle. Therefore, if the high-energy peak is found, the wanted events are found, because they are presented in that peak, no matter the purported energy.
SCANDAL is constructed to have a target right in front of it. With the setup in the experiments, the first trigger scintillator cannot detect protons that are traveling towards some crystals. The CsI hodoscope is wider than the first trigger scintillator, which makes it impossible for scattered protons to hit both the scintillators and the first crystals
according to its trajectory from the CH2 target (see figure 5). This is why the high-energy peak cannot be found for the first two crystals, which makes it necessary to discard all events detected in these CsI crystals.
5.5 Particle identification
Both protons and deuterons are charged particles and can be detected in SCANDAL.
Deuterons have larger mass than protons, which makes it possible to separate the different particles from each other. The difference in mass makes the energy loss for deuterons larger than for protons with the same energy when they interact with other nuclei. If the total energy loss in both scintillators for each event is plotted versus the detected energies in the CsI crystals two bands of events gets visible. These bands are presented in figure 13 and contain events that represent the energy loss for each proton and deuterons in both scintillators. So if all events outside the proton band for each crystal are discarded, there should only be protons left for the profile reconstruction [1].
Figure 13: Particle identification. A weak deuterium band is visible for CsI 5 (see the arrow). Below this band, a clear proton band is found.
Using the estimated TOF cut from section 5.3, most of the deuterons are rejected and the proton band gets narrower, which makes it easier to analyze, see figure 14.
Figure 14: Particle identification after TOF cut. Protons detected in CsI crystal 5, contributes to the dark event band.
If each crystal is analyzed in its energy range that was determined in chapter 5.4, the energy loss in the two scintillators for each proton is between 2 and 8 MeV.
5.6 The TOF cut – part 2
With the determined cuts from section 5.4 and 5.5, the TOF can be presented in a new plot, which is shown in figure 15.
Figure 15: The TOF plotted using the cuts determined in section 5.4 and 5.5. The plot is fitted with a gauss distribution.
The major part of the TOF peak comes from np-scattered protons in the high-energy peak, according to the earlier cuts. Discarding too many of these events would decrease the statistics and therefore are two standard deviations for the fitted Gaussian used to determine the cuts for TOF. This satisfies about 95 % [20] of the events under the curve.
This discussion together with the information box in figure 15, a TOF cut between 325 and 365 ns is chosen.
6 Profile reconstruction
With the trajectory from every specific event detected in SCANDAL, profiles of the beam intensity across the opening can be reconstructed, one for the x- and one for the y- axis. Negative numbers on the x-axis represent events from the left of the facility. The y- axis represents the vertical direction with positive direction upwards. These coordinates are also presented in figures 7, 8 and 9.
If the profiles for setup 1 are reconstructed with all detected events and none of the presented cuts in the previous chapter are used, the profiles look as shown in figures 16- 19.
Figure 16: The x-axis profile for the target run without cuts.
Figure 17: The y-axis profile for the target run without cuts.
Figure 18: The x-axis profile for the background without cuts.
Figure 19: The y-axis profile for the background without cuts.
According to figure 16 and 18, there seems to be more detected events from the left side of the opening compared to the right side. This phenomenon did only disappeared when the CsI cut was implemented.
Neutrons moving in the beam line can interact with the nucleus in the air that creates charged particles. This happens mostly to the left of the detector and is probably the reason for the asymmetrical appearance. Neither the TOF cut nor the particle- identification cut can eliminate all these events.
6.1 Result after cuts
Implementing the determined cuts and the ratio between the target- and the background run, comparable profiles can be reconstructed for setup 1. In figure 20, the x-axis profile is plotted for both runs and in figure 21 both y-axis profiles are presented. Figures 22 and 23 shows the profiles after the background subtraction.
Figure 20: The x-axis after the cuts. The profile with the darker line represents the target run and the other is due to the background run.
Figure 21: The y-axis after the cuts. The profile with the darker line represents the target run and the other is due to the background run.
Figure 22: The final x-axis profile. The center of the profile is situated at the coordinate –35 mm. The FWHM value is about 150 mm.
Figure 23: The final y-axis profile. The center of the profile is situated at the coordinate –8 mm. The FWHM value is about 150 mm.
The Full Width at Half Maximum (FWHM) is a simple measure of the width of a
distribution. The width across the profile when it drops to half of its peak is the full width of the image. It is a simple and well-defined number that can be used to compare the quality of images obtained under different observing conditions. In figures 22 and 23, a FWHM value around 150 is presented, which is very close compared with the opening.
Both the profiles are to different degrees misplaced from the center of the opening, which is discussed in chapter 10.
7 Resolution folding
The final intensity profiles detected by SCANDAL for setup 1 are presented at the end of the previous chapter. To evaluate these in detail, a profile for each axis is simulated. By comparing the agreement in width between an experimental profile and a simulated one, the assumed detection resolution for SCANDAL can be controlled. To determine the flux intensity for the experiment, the simulation assumes a flat intensity distribution over the opening.
7.1 Gauss simulation
To simulate a profile, the assumption has been made that the SCANDAL detection resolution for a certain point at the collimator end is Gaussian with a certain standard deviation. Thus, for each square centimeter over the opening, a Gaussian distribution is placed, which will contribute to the profile characteristic (see figure 24). A comparable profile with the experiment should be found if the correct resolution and the same number of events are used.
When a projection is simulated, the contribution from Gaussians are larger in the middle because of the larger area compared to the edges of the opening and therefore the opening must be folded.
Figure 24: Each square over the opening corresponds to one Gaussian distribution. The figure is not to scale.
Adding all Gaussians into one profile and setting the amount of events equal to the experimental profile, the two profiles can be compared with each other.
7.2 Fitting simulated profiles with the experimental ones
Before the simulated profiles can be compared with the measured ones, the center of the simulations must be placed at the same coordinates as the experimental profiles.
According to figures 22 and 23, these centers are situated at the coordinate -35 mm on the x-axis and at -8 mm for the y-axis.
Figure 25: The x-axis. The darker line represents the measured profile and the other is simulated with a 10 mm
standard deviation.
Figure 26: The y-axis. The darker line represents the measured profile and the other is simulated with a 10 mm standard deviation.
According to chapter 3.2, the resolution for the two drift chambers (ΔX1 and ΔX2) is for perfect conditions 0.3 mm (FWHM) each. The resolution for the drift chambers is probably worse than 0.3 mm for the experiments in this thesis, but it is used anyway as the lower limit.
When the total drift chamber resolution for reconstructing the profile is calculated, the coordinate in the second drift chamber is assumed to be well known and its uncertainty is added to the uncertainty in the first drift chamber (DC1).
3 . 0
* 2
* 2
1= ΔX12 +ΔX22 = ΔX =
DC
The distance between the two levels is 78 mm and the total distance from the second drift chamber to the CH2 target is about 3700 mm. This gives a total resolution (Tot) of
mm
Tot * 2*0.3 20
78
3700 ≈
= .
The resolution can be converted from FWHM to standard deviation with the formula
FWHM mm 35 9 .
2 ≈
σ = .
Therefore, if the uncertainty of the drift chamber hits were the only reason for the resolution when a profile is reconstructed, the total calculated resolution for setup 1 would be about σ = 9 mm. According to figures 25 and 26, a standard deviation of 10 mm seems to be too small to describe the total resolution for the setup. The calculated
resolution of 9 mm is obviously too small. So, other parameters must affect the resolution negatively, such as straggling in air and a resolution worse than 0.3 mm for the drift chambers.
After several tests, a standard deviation of 40 mm for the simulated profile results in the best fit, which is shown in the figures 27 and 28.
Figure 27: The x-axis profile. The darker line represents the measured profile. The other is simulated with σ = 40 mm.
Figure 28: The y-axis profile. The darker line represents the measured profile. The other is simulated with σ = 40 mm.
From the results in figure 27 and 28, it can be determined that the profile reconstructed in setup 1 with SCANDAL corresponds to a standard deviation of 40 mm in resolution.
7.3 Intensity variation over the collimator opening
The simulated profile illustrates a beam with constant intensity flux over the collimator opening. This can be used to measure how the intensity varies over the opening in reality.
If the measured profile is divided with this constant beam over the opening, an intensity ratio will be the result, which is shown in figures 29-32.
When a profile for a certain axis is reconstructed, ROOT places all events in so-called bins. In this thesis when an axis is plotted, 200 bins are used per plot, which results in one bin per four mm over the axis when it starts at –400 and ends at +400 mm (see for
example figure 27). The number of events in each bin over the axis represents the intensity over the opening detected by SCANDAL. The detected events in each bin are Poisson distributed, which results in statistical errors that are equal to the square root of all detected events in its bin. Therefore, the higher number of detected events in a bin, the lower is its relative error.
When the reconstructed profile is divided with the simulated profile, the errors still remain and can be plotted as error bars. With these error bars it is possible to fit a function over the opening that presents the intensity profile. The goodness of the fitting function can be decided with a chi-square test. To implement the chi-square test, the values for chi-square and υ (the number of degrees of freedom) must be known. The chi- square is the weighted value of the distance between the calculated values and the adjusted function. The υ is the number of measure points subtracted with the number of the adjusted functions parameters [21]. The equation for chi-square and υ is
2
1
2 ∑( )
=
= k −
i i
i
Xi
σ
χ μ and υ =k−a.
If the chi-square value is divided with υ, a quotient around one represents a good adjustment. A quotient much larger than one implies that the error bars are too small or that the model poorly describes the data for the adjusted function. A much smaller quotient than one represents too large errors or a description using too many parameters.
Figure 29: A function with 1 parameter fitted over the x-axis resulting in χ2/υ
≈0.89.
Figure 30: A function with 1 parameter fitted over the y-axis resulting in χ2/υ ≈1.18.
Figure 31: A function with 2 parameters fitted over the x-axis resulting in χ2/υ
≈0.91. The intensity difference between the left and the right edge is about 2%.
Figure 32: A function with 2 parameters fitted over the y-axis resulting in χ2/υ ≈1.21.
The intensity difference between the lower and upper edge is about 3%.
Using one parameter for the fitting over both axes seems to give a good description according to the chi square test, which is shown in figures 29 and 30. Adding one parameter to the fitting, which is done in figure 31 and 32, does not improve the
description significantly. Therefore, the one-parameter fitting represents a good model of describing the intensity over the opening.
7.4 The 40 mm cut
The method, comparing the measured profile with a constant one, can be improved.
Instead of using the whole collimator opening to reconstruct a profile, a cut at the middle of the opening can be implemented (the principle for this cut is shown in figure 33). For example, when the x-axis is plotted, only events from the middle of the y-axis are used.
Besides of knowing which y-axis value that contributes to the x-axis profile, this cut will make the contribution from the Gaussian distributions evenly distributed over the x-axis.
The x-axis profile will get a more flat appearance over the opening, which may make it easier to analyze.
Figure 33: The principle with the 40 mm cuts over the opening. The figures are not to scale.
Figure 34: The x-axis with a 40 mm cut on the middle of the y-axis. The darker line represents the measured profile and the other is simulated.
Figure 35: The y-axis with a 40 mm cut on the middle of the x-axis. The darker line represents the measured profile and the other is simulated.
The profile over the opening had been more flat for the two profiles in figures 34 and 35 if the resolution had been narrower than 40 mm or if the opening had been bigger.
In figures 38-41, the measured profiles are divided with the simulated profiles using the 40 mm cut. This presents the intensity over the middle of the opening for the two axes.
Figure 36: The x-axis with the 40 mm cut. The function with 1 parameter gives a χ2/υ ≈1.13.
Figure 37: The y-axis with the 40 mm cut.
The function with 1 parameter gives a χ2/υ
≈1.38.
Figure 38: The x-axis with the 40 mm cut. The function with 2 parameters gives a χ2/υ ≈1.16. The intensity
difference between the left and right side is about 5%.
Figure 39: The y-axis with the 40 mm cut.
The function with 2 parameters gives a χ2/υ
≈1.27. The intensity difference between the lower and upper edge is about 12%.
The differences between using one or two parameters in the fit is still very small (see figures 36-39), so the intensity over the opening is assumed to be constant.
8 Setup 2 - the ∅300 mm experiment
The wideness of the beam increases after the collimator. To ensure that SCANDAL is not placed directly in the beam when the collimator opening is increased to ∅300 mm, SCANDAL was moved 110 mm sideways. This new setup, also called setup 2, is shown in figure 9.
It is better to use the 40 mm cut compared with using the whole collimator opening when the flux distribution is examined. Therefore, for setup 2, the 40 mm cut will be used from the start.
8.1 Data cuts for setup 2
Comparing with setup 1, both the TOF cut and proton energy loss in the triggers are the same for setup 2 (see chapter 5.4 and 5.5). For the crystals on the other hand, differences between the setup are notable. For the first, the high-energy peaks for each crystal are found in different energy intervals between the setups (compare table 2 with table 3).
Also, a peak can be found in crystal 1 for setup 2, so the geometrical conditions seem to be better for this setup compared with setup1.
CsI crystal
Mean value
(MeV) σ (MeV) Energy interval (MeV)
0 - - -
1 100.9 4.4 96.5 - 105.3
2 87.3 4.2 83.1 - 91.5
3 91.8 3.6 88.2 - 95.4
4 83.8 3.6 80.2 - 87.4
5 75.6 3.2 72.4 - 78.8
6 78.6 3.4 75.2 - 82
7 82.7 3.5 79.2 - 86.2
8 81.1 4.1 77 - 85.2
9 81.7 5.3 76.4 - 87
10 80.7 3.5 77.2 - 84.2
11 77.9 4.7 73.2 - 82.6
Table 3: The energy ranges for setup 2 determined with Gaussian fitting, using one sigma.
8.2 Intensity variation over the collimator opening
When the 40-mm cut for the experimental and simulated profiles is used, the profiles look like in figures 40 and 41.
Figure 40: The x-axis with a 40 mm cut on the middle of the y-axis. The centre of the profile is situated at the coordinate 250 mm.
Figure 41: The y-axis with a 40 mm cut on the middle of the x-axis. The centre of the profile is situated at the coordinate -5 mm.
It is obvious in figures 40 and 41 that the profiles are mismatched. Dividing the
experimental profiles with the simulated ones results in a non-constant intensity over the collimator opening (see figures 42-47).
Figure 42: The x-axis with the 40 mm cut. The function with 1 parameter gives a χ2/υ ≈4.14.
Figure 43: The y-axis with the 40 mm cut.
The function with 1 parameter gives a χ2/υ
≈2.64.
Figure 44: The x-axis with the 40 mm cut. The function with 2 parameters gives a χ2/υ ≈3.02. The intensity
difference between the left and right side is about 32%.
Figure 45: The y-axis with the 40 mm cut.
The function with 2 parameters gives a χ2/υ
≈2.57. The intensity difference between the lower and upper edge is about 11%.
Figure 46: The x-axis with the 40 mm cut. The function with 3 parameters gives a χ2/υ ≈1.50. The intensity difference between the centeroid and 150 mm away is 40 %.
Figure 47: The y-axis with the 40 mm cut.
The function with 3 parameters gives a χ2/υ
≈0.88. The intensity difference between the centeroid and 150 mm away is 35 %.
The intensity difference between the center and edges of the opening is very high for both of the axes. This can be seen in figures 46 and 47, which show the best-adjusted
functions. Therefore, new profiles were simulated with other resolutions. This just resulted in worse adjusted functions and the intensity difference between the center and the edges increased.
In figures 40 and 41 the simulated profiles look wider than the experimental profiles. So, instead of changing the resolution, the widths for the simulated profiles were decreased to 280 mm. This resulted in a much better agreement between the different profiles (see figures 48 – 53).
