Monte-Carlo Simulations of Nuclear Reactions at

Relativistic Energies
Bachelor Thesis in Engineering Physics

STEFAN BULLER
JOEL MAGNUSSON
TORBJÖRN RATHSMAN

Department of Fundamental Physics

Chalmers University of Technology
Gothenburg, Sweden 2013
 Monte-Carlo Simulations of Nuclear
Reactions at Relativistic Energies

Bachelor Thesis

by

Stefan Buller, Joel Magnusson, and Torbjörn Rathsman

Supervisors:

Andreas Heinz
Håkan Johansson
Ronja Thies

Examiners:

Christian Forssén
Daniel Persson

Department of Fundamental Physics

Chalmers University of Technology
Gothenburg, Sweden 2013
Monte-Carlo Simulations of Nuclear Reactions at Relativistic Energies
Stefan Buller (bstefan@[Link])
Joel Magnusson (joelm@[Link])
Torbjörn Rathsman (torbjorr@[Link])

c Stefan Buller, Joel Magnusson, Torbjörn Rathsman, 2013.

FUFX02 - Bachelor thesis at Fundamental Physics

Bachelor Thesis No. FUFX02-13-04

Supervisor: Andreas Heinz, Håkan Johansson, Ronja Thies

Examiners: Christian Forssén, Daniel Persson

Department of Fundamental Physics

Chalmers University of Technology
SE-412 96 Göteborg
Sweden
+46 (31) 772 1000

Printed by Chalmers Reproservice

Göteborg, Sweden 2013

Cover: An event created by using GGLAND. A 14 B nucleus is fired into a partial setup
of the S393 experiment. The colour coding is as follows: traces of electrons (e− ) are
red, positrons (e+ ) blue, photons (γ) green, protons (p) cyan, neutrons (n) yellow and
pions (π) magenta. Traces of all other particles are gray.
 Abstract
This thesis describes Monte-Carlo simulations of the S393 light exotic-ion beam experiment that
was carried out at GSI Helmholtz Centre for Heavy Ion Research, using the simulation software
GEANT4 via the wrapper program GGLAND. Specifically, it focuses on the fragment arm of the
detector setup in Cave C and the ability of the RALF’S TRACKER software to track and identify
simulated ions.
Two separate cases are considered. Firstly, the trackers ability to identify particles ejected
forward along the beam-axis with incorrect information of the magnetic field in ALADiN (A
LArge DIpole magNet). Secondly, the trackers ability to identify particles ejected with an angle
α from the beam-axis is studied. The simulations were carried out by firing single, neutron rich
4 ≤ Z ≤ 9 nuclei with a magnetic rigidity of 8.8 Tm from the target location, corresponding to
particles bending off along the fragment arm at a 15◦ deflection angle.
It is found that the tracker, calibrated in accordance with a pre-described routine1 , is able
to correct for the incorrect magnetic field by effectively rotating the fragment arm for the cases
simulated. The resulting errors are small enough as not to affect identification of the simulated
ions. The tracker is also able to identify most particles fired at an angle α as long as all detectors
along the trajectory are hit. The size of the GFIs (Großer FIber detektor) is the first limiting
factor for detecting particles with α & 2.5◦ . However, some non-geometric effects cause some
particles to miss the TFW (Time of Flight Wall) at roughly the same angle. These effects remain
unidentified.
In addition, the geometry of detectors defined in GGLAND together with algebraic approxi-
mations for choosing an appropriate magnetic rigidity in the simulations for a given ALADiN
current and for determining the charge of particles, given the energy deposits in the TFW, are
presented in appendices.

Sammandrag
Denna tes beskriver Monte-Carlo simuleringar av S393 -experimentet utfört på GSI Helmholtz
Centre for Heavy Ion Research med en stråle av lätta exotiska nuklider, utnyttjande simulerings-
mjukvaran GEANT4 genom wrapper-programmet GGLAND. Fokus ligger på fragmentarmen hos
detektoruppställningen i Cave C och förmågan hos mjukvaran RALF’S TRACKER att spåra och
identifiera simulerade joner.
Två separata fall har undersökts. Det första av dessa är trackerns förmåga att identifiera
partiklar utskjutna framåt längs strålaxeln då den matas med felaktig information om ALADiNs
(A LArge DIpole magNet) magnetfält. Det andra fallet är trackerns förmåga att identifiera
partiklar som skjutits ut med en vinkel α från strålaxeln. Simuleringarna utfördes med enskilda,
neutronrika 4 ≤ Z ≤ 9 nuklider med en magnetisk rigiditet på 8.8 Tm, avfyrade från strålmålets
position. Med denna magnetiska rigiditet böjdes jonerna av längs med fragmentarmen, med en
avböjningsvinkel på 15◦ .
Resultaten är att trackern, kalibrerad för fragment enligt en föreskriven rutin1 , kan korrigera
för ett felaktigt magnetfält genom att effektivt rotera fragmentarmen för de fallen som simulerats.
De resulterande felen är små nog att inte påverka identifieringen av den avfyrade jonen. Trackern
identifierar även stora delar av partiklar avfyrade med en vinkel α, såvida alla detektorer träffas.
Storleken på GFI:erna (Großer FIber detektor) fanns vara den första begränsande faktorn för att
detektera partiklar med α & 2.5◦ , men oidentifierade icke-geometriska effekter får vissa partiklar
att missa TFW:n (Time of Flight Wall) vid ungefär samma vinkel.
Utöver detta presenteras, i appendix, geometri för detektorer definierade i GGLAND samt med
algebraiska approximationer för att välja ett lämpligt bρ i simuleringar för en given ALADiN
ström och för att avgöra laddningen hos partiklar, givet energidepositioner i TFW:n.
1
[Link] (19-05-2013)
Acknowledgements
This project was realised thanks to the help and support from a few, but important,
people.
We would first like to thank our supervisors: Andreas Heinz, whose enthusiasm and
guidance kept us going to the very end. Håkan Johansson, who acted as tech support
and advisor, at any time of the day. And last but not least, Ronja Thies, for going to
lengths to find out what no one else seemed to know.
We would also like to thank Sven Eriksson, for proof-reading our report and for pro-
viding the point of view of a fellow student.
Finally, we would like to thank all of those who have contributed to the various soft-
ware products employed during this project.

The Authors, Gothenburg, May 2013

Contents

1 Introduction 1
1.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Theoretical Background 4
2.1 Particle Interaction Processes . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 Photons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.2 Charged particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Detector Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Scintillators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 The S393 Experiment 10

3.1 Source of Unstable Nuclei . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Cave C Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

4 Simulation and Analysis Tools 14

4.1 GEANT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.2 GGLAND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.3 RALF’S TRACKER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.3.1 Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.3.2 Outline of a Possible Tracking Algorithm . . . . . . . . . . . . . . 17
4.4 ROOT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5 Simulation and Analysis Procedure 18

5.1 Determining an Appropriate bρ . . . . . . . . . . . . . . . . . . . . . . . . 18
5.2 Configuring and Running GGLAND . . . . . . . . . . . . . . . . . . . . . . 19
5.3 Preparing and Running RALF’S TRACKER . . . . . . . . . . . . . . . . . . 19
5.3.1 Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
5.3.2 Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.3.3 Running the tracker . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.4 Angular Dependency in Detected Events . . . . . . . . . . . . . . . . . . . 22
5.5 Mass Deviations for Tracker Calibrated With Incorrect Current . . . . . . 24

6 Results 27
6.1 Angular Dependence of Detected Events . . . . . . . . . . . . . . . . . . . 27
6.1.1 Reasons for the Decrease in Events with Spread . . . . . . . . . . . 28
6.2 Effects of Incorrect Currents on Tracking . . . . . . . . . . . . . . . . . . 32

i
7 Discussion 38
7.1 Angular Spread, Detected Events . . . . . . . . . . . . . . . . . . . . . . . 38
7.1.1 Energy Cuts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
7.1.2 Dip in Detected Events . . . . . . . . . . . . . . . . . . . . . . . . 39
7.2 Compensating for Incorrect Current . . . . . . . . . . . . . . . . . . . . . 40
7.3 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Glossary 43

A Approximating the Magnetic Rigidity 48

A.1 Analytical Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
A.2 Angular Approximations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
A.3 Linear Correlation Between the Current and the Magnetic Field Strength 54
A.4 The Magnetic Field of ALADiN . . . . . . . . . . . . . . . . . . . . . . . . 56

B An Example Detector: POS 59

C Obstacles Defined in GGLAND 62

C.1 ALADiN — A LArge DIpole magNet . . . . . . . . . . . . . . . . . . . . . 62
C.2 DTF — Dicke ToFWand . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
C.3 GFI — Großer FIberdetektor . . . . . . . . . . . . . . . . . . . . . . . . . 65
C.4 NTF — New Time-of-Flight wall . . . . . . . . . . . . . . . . . . . . . . . 66
C.5 PDC — Proton Drift Chamber . . . . . . . . . . . . . . . . . . . . . . . . 67
C.6 POS — POSition detector . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
C.7 ROLU — Rechts, Oben, Links, Unten . . . . . . . . . . . . . . . . . . . . 69
C.8 SST — Silicon STrip detectors . . . . . . . . . . . . . . . . . . . . . . . . 70
C.9 TFW — Time of Flight Wall . . . . . . . . . . . . . . . . . . . . . . . . . 71

D Function Used to Calibrate RALF’S TRACKER 72

D.1 Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
D.2 Step 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
D.3 Step 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
D.4 Step 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

E RALF’S TRACKER — Used experiment_specific.hh 76

F Energy Depositions for Z 79

F.1 Energy Ranges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
F.2 MATLAB Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

G Software Resources 84
G.1 GGLAND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
G.2 RALF’S TRACKER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
G.3 A Program Integrating GGLAND and RALF’S TRACKER . . . . . . . . . . . 84

ii
H Svensk Sammanfattning 85
H.1 Inledning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
H.2 Kort om detektorprinciper . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
H.3 S393 -Experimentet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
H.4 Mjukvara . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
H.4.1 GEANT4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
H.4.2 GGLAND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
H.4.3 RALF’S TRACKER . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
H.4.4 ROOT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
H.5 Simulering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
H.5.1 Kalibrering av RALF’S TRACKER . . . . . . . . . . . . . . . . . . . 90
H.6 Analys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
H.6.1 Vinkelberoende hos detekterade händelser . . . . . . . . . . . . . . 91
H.6.2 Utvärdering av trackerkalibrering . . . . . . . . . . . . . . . . . . . 91
H.7 Resultat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
H.7.1 Vinkelberoende hos detekterade händelser . . . . . . . . . . . . . . 92
H.7.2 Inverkan av felaktig ström hos partikelspårning . . . . . . . . . . . 93
H.8 Diskussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
H.8.1 Spridning och antalet träffar . . . . . . . . . . . . . . . . . . . . . 93
H.8.2 Kompensering för inkorrekt kalibreringsström . . . . . . . . . . . . 98

iii
Notations Throughout the Report
Term Type Definition
u,x,b geometric object Three-vectors
U ,X,P geometric object Four-vectors
P ,Q,N physical quantity Four-momentum
u,v,v 0 , etc. physical quantity Speed
c physical constant The speed of light
E physical quantity Energy
p physical quantity Linear momentum
m,M physical quantity Mass
m0 ,M0 physical quantity Rest mass, also known as invariant mass
me ,mX physical quantity Rest mass of the electron/particle ”X”
Z physical quantity Atomic number, the number of protons in
a nucleus
A physical quantity Mass number, the number of nucleons in
a nucleus.
β physical quantity Beta, the quotient vc
γ particle Gamma, a photon, a quantum of energy
γ physical quantity Lorentz gamma factor.
γ = (1 − v 2 /c2 )−1/2
α physical quantity Opening angle, the angle between the par-
ticles initial direction and the beam axis.
b physical quantity Magnetic field
T physical quantity Kinetic energy
θ, φ, ϕ physical quantity Angles; θ often being taken as the angle
from the z-axis (the beam axis); ϕ often
from the x-axis
ρ physical quantity Bending radius of a particle due to a mag-
netic field
bρ physical quantity Magnetic rigidity of a particle. bρ = p/q
I physical quantity Current
z direction The direction of the beam axis
y direction The direction towards the ceiling
x direction Direction chosen to yield a right-hand co-
ordinate system together with y and z
q physical quantity Charge

iv
1. Introduction

All of the matter around us are made out of atoms, consisting of electrons and different
nuclei. These nuclei are in turn made out of nucleons, which have been bound to each
other through fusion of lighter elements in astrophysical processes. While the stability
of elements may be understood in terms of minimising energy, their relative abundance
can only be understood by looking at how they were formed in these processes.
A large part of understanding how elements are synthesised is to determine how likely
certain reactions are to take place. These likelihoods are often given in terms of cross
sections, σ, defined by
R
σ= , (1.1)
NΦ
where Φ is the intensity of an incoming beam of particles ([number of particles]/[area ×
time]), R is the number of reactions r per unit of time and N is the amount of particles
being illuminated by the beam. In order to measure σ, the numbers of reactions r have
to be measured. If the reaction r results in a set of reaction products, this is often
done in practice by detecting and counting the product particles. To determine σ with
absolute accuracy, each and every one of these particle must be detected or otherwise
accounted for.
In order to detect the results of various interactions, experimentalists have often come
to rely on increasingly complex detector setups. This has the benefit of allowing more
particles to be detected with increased resolution, but the additional data require more
elaborate analysis techniques. Moreover, the physical quantities of interest may not be
directly retrievable from the data of a single detector, and heuristic methods must be
used to figure out which event in different detectors are caused by the same particle.
Nevertheless, it is not feasible to detect every single produced particle. In order to
still be able to determine cross sections, the experimental setup needs to be evaluated
to get a grasp on how often a particle is not detected.
To evaluate the experimental setup, the setup may be tested with known reactions.
This makes the experiment take longer time and may not always reveal what actually
happened. Another way to evaluate the setup is to simulate it by using computer
programs. In this case, all particles can be traced, since a particle track is just represented
by bits in readable memory. This difference is illustrated by Figure 1.1.
Once the experimental setup is evaluated, the ratio of the detected and undetected
events - the acceptance of the setup - may be used to compensate for the particles that

1
1 Introduction

Experiment Simulation

Figure 1.1: By simulating the experiment, the experimental data can be related to the
inputs of the simulation, giving a better understanding of the behaviour of the
experimental setup. This in turn makes it easier to disentangle the underlying
physics.

avoid detection. The acceptance may be formally defined as

E[Ndetected events ]
Acceptance = , (1.2)
E[Nevents ]

where E[X] denotes the expected value of X [1]. Generally, the acceptance will depend
on several variables, such as the momentum of the detected particle. Another thing to
note is that since the relevant physical parameters for identifying certain particles only
may be determined by extensive analysis of the experimental data, the acceptance must
take the analysis methods into account.

1.1 Scope
This bachelor thesis describes Monte-Carlo simulations of the fragment arm (described
in Section 3.2) of the S393 experiment performed at GSI Helmholtz Centre for Heavy
Ion Research, Darmstadt, Germany. The main focus has been on how well the analysis
tools behave under different circumstances with single ions fired forward, from the target
location, and with a magnetic rigidity (defined in Section 2.1.2) of 8.8 Tm when trying
to identify what kind of particle that was fired. This information may then be used as
a step towards determining the acceptance of the setup.
One part of the analysis chain is a tool used to evaluate particle paths and find their
speed, mass, and charge. It has been investigated how many detector events it can
identify correctly and if there are some situations when it will fail, by delivering an
incorrect value. An important part of the experimental setup for identifying particles
is a magnetic field, which causes charged particles to deviate from their original path
depending on their mass-to-charge ratio.
One of the presented results is how the setup with analysis tools is able to detect
and identify forward travelling ions depending on differences between the real magnetic
field and the field used in the analysis. Another result is how the initial angle between
particles trajectories and the beam axis affects the amount of detected particles.
Apart from the aforementioned results, there are also results in the form of detector
definitions, which was created in the scope of this project. These definitions have been
thoroughly used in, and were crucial for, the simulations that generate the data that
were eventually analysed.

2
 1.2 Outline

1.2 Outline
In order to understand the experiment, a brief background in detector physics is neces-
sary. A short introduction on that subject is found in Section 2. After that, in Section 3,
the S393 experiment is described detector by detector. Then follows a survey of the tools
used for our simulations and analyses in Section 4 and, in Section 5, it is described how
these tools were used. The results are presented in Section 6 and discussed in Section 7.
Possible future investigations and challenges are discussed in the concluding section,
Section 7.3.

3
2. Theoretical Background

In the following sections, a few mechanisms useful for understanding the principles of ra-
diation detection are presented. Section 2.1 provides a theoretical basis for understanding
the principles behind detection of particles, the latter being the focus of Section 2.2.

2.1 Particle Interaction Processes

This section describes how particles interact with other particles and external fields. This
is mostly done within the framework of special relativity and the underlying mechanics
for the interactions are not considered.

2.1.1 Photons
The most common interaction processes of photons — photoelectric absorption, Comp-
ton scattering and pair production — are presented in this section. Different interaction
processes dominate within different energy ranges, as seen in Figure 2.1. Pair production
and Compton scattering both yield photons of lower energy, and therefore energetic pho-
tons are able to trigger a cascade of events, until the produced photons are at sufficiently
low energy to be likely to be absorbed.

Photoelectric Absorption
A photon in the vicinity of an atom may be absorbed by transferring all of its energy to
a bound electron. The probability for this to occur is highly dependent on the charge
of the atomic nucleus and the shell of the electron, with more tightly bound electrons
being more likely to receive the energy [3], and the interaction not occurring at all with
free electrons. At sufficiently high energies, the electron is emitted from the atom.

Compton Scattering
Compton scattering is due to a photon scattering on a charged particle (often an elec-
tron). The energy of the photon after the collision, assuming the particle was initially
at rest, can be expressed as
E
E0 = E
,
1+ m 0 c2
(1 − cos θ)

4
 2.1 Particle Interaction Processes

Dominant Photon Interaction Processes

100

80

Atomic Number, Z
Pair-
production
60

Photoelectric absorption

40

20

Compton scattering

0
0.001 0.01 0.1 1 10 100
Photon Energy (MeV)

Figure 2.1: Representation of the relative predominance of the main photon interac-
tion processes with an atom. Data were calculated using the NIST XCOM
database [2].

where E is the initial energy of the photon, m0 is the rest mass of the charged particle
and θ is the scattering angle in the initial rest frame of the particle, as illustrated in
Figure 2.2.

Pair Production
A photon with sufficient energy can in the presence of a nucleus spontaneously create
an electron-positron pair, as illustrated in Figure 2.3. This may occur if

E > 2me c2 ,

where me is the rest mass of the electron (and positron), c the speed of light. However,
it is only at still higher energies, a few MeV, that pair production makes a notable
contribution to the energy depositions of photons in matter, as can be seen in Figure 2.1.

2.1.2 Charged particles

Bremsstrahlung
Bremsstrahlung is the result of acceleration or deceleration of a charged particle. As
such, it occurs in the simulation when a particle is decelerated through interaction with
other particles. For this case, a collision between a pair of particles is considered, in which
an incoming particle is decelerated and a photon is emitted. This can be understood
from 4-momentum conservation. The energy of the emitted photon is given by

mo Mo c2 γ(u) − γ(u0 )γ(v 0 )(1 − u0 v 0 cos θ/c2 )

E= ,
M0 + m0 γ(u)(1 − u cos φ/c)

5
2 Theoretical Background

Before After γ
γ θ
m0 m0

Figure 2.2: A photon is scattered by a massive particle initially at rest.

Before After
M0 M0

γ +
me -
me

Figure 2.3: A photon creates an electron-positron pair near a stationary massive particle.

with notations as in Figure 2.4 and with u0 being the speed of the incoming particle and
v 0 being the speed of the initially stationary particle, both after the collision.

Response to a Magnetic Field

A charged particle with a constant velocity v travelling in a magnetic field b will experi-
ence a force f = qv ×b. Due to this force a particle moving with a velocity perpendicular
to a homogeneous magnetic field, as illustrated in Figure 2.5, will be deflected in a cir-
cular arc with the radius ρ given by
s
T 2m0 c2
ρ= 1+ , (2.1)
qbc T

where T is the kinetic energy of the particle, q its charge, b the strength of the magnetic
field, and m0 its rest mass. Equation 2.1 can alternatively be written as

p
bρ = , (2.2)
q

where p = γm0 v is the momentum of the particle. From this it can be concluded that
particles with the same pq will bend off with the same radius in a given homogenous
magnetic field. For this reason, pq is known as the magnetic rigidity of a particle.

6
 2.1 Particle Interaction Processes

Another way of formulating Equation 2.1 is

v
(Zebρ)2
u
u
β=t . (2.3)
(Zebρ)2 + (cm)2

Before After
Q0 , v 0
P,u Q θ
m0 M0 φ P 0 , u0
N0

Figure 2.4: Deceleration and deflection of a charged particle with rest mass m0 , initial
4-momentum P and speed u. The angle θ is the angle between the directions
of the two massive particles after the collision. The angle φ is the angle
between the direction of the emitted photon and the original velocity of the
incoming particle. Everything takes place in the original rest frame of the
M0 particle.

ẑ
bẑ
rr̂
uθ̂
m0 q

Figure 2.5: A particle with constant velocity moving through a magnetic field perpendic-
ular to the plane of motion.

Energy Loss of a Charged Particle Passing Through Matter

When a charged particle passes through matter, it excites or knocks nearby electrons
into motion, whereas the several orders of magnitude heavier nuclei are mostly left
unaffected. By conservation of energy, the charged particle loses energy equal to the
energy given to the electrons. Since the interaction is mainly due to the Coulomb force,
the particle continuously interacts with the surrounding matter. If the charged particle
is much heavier than the electron, then the energy loss of the particle due to Coulomb
interactions as a function of travelled distance through matter is given by the Bethe

7
2 Theoretical Background

formula !2 " ! #
dE 4π nz 2 e2 2me c2 β 2
− = N Z ln − β2 , (2.4)
dx me c2 β 2 4π0 I(1 − β 2 )
where β = v/c is the dimensionless speed relative to the speed of light, and ze is the
charge of the particle. The parameters N Z is the charge density of the medium and I
is the ionisation energy of the atoms in the medium. The Bethe formula is mainly valid
for high energies and low z, which is the case in the S393 experiment.
A consequence of charged particles passing through matter, and ionising nearby atoms,
is that the particles travelling through gives rise to scores of freed electrons. These
electrons will trace paths originating and branching off from the path of the original
particle, but only rarely straying far from that path due to their low kinetic energy.
However, among these are some with a sufficiently high energy to in turn ionise additional
atoms. These more energetic electrons are referred to as δ-electrons, or δ-rays.

2.2 Detector Principles

The information about radiation detection found in this section is based on Radiation
detection and measurement by G.F. Knoll [3].
It is often desirable to convert measured quantities into electronic signals, since those
can easily be manipulated and later digitised. This can be achieved in various ways,
depending on the properties of the particles to be detected.
Charged particles, like protons and ions, can be detected directly through their in-
teractions with the active volume1 of a detector [4]. On the other hand, uncharged
particles, like neutrons, are detected indirectly when they interact with individual nuclei
through the resulting motion of other (charged) particles [4].

2.2.1 Scintillators
In an inorganic crystalline scintillator, the valence electrons in the detector medium
closest to the passing particle will occasionally be excited to a higher energy band [5].
The excited electrons will eventually return to the conduction band by spontaneous re-
laxation, with a photon being emitted in accordance to ∆E = ~ω. Due to additional
energy levels between the bands, introduced by defects in the scintillator-material, elec-
trons may deexcite in several steps, emitting photons with energies that are not able to
excite electrons in the valence band and are thus unlikely to be absorbed [5]. These pho-
tons can then be detected by photomultiplier-tubes (PM-tubes), producing an electric
signal.
By conservation of energy, the charged particles passing the detector medium must
deposit energy equal to the energy used to excite the electrons, whereby the energy depo-
sition of the particle can be measured. These are the basic principles behind scintillation
detectors.
1
The active volume of a detector is the volume which is able to transform energy deposits within itself
into read-out signals.

8
 2.2 Detector Principles

The same principles apply to other types of scintillators as well, such as organic
scintillators, where the energy-level structure of the molecules allows an excited molecule
to emit photons not absorbable by molecules in the ground state [5].

9
3. The S393 Experiment

The S393 experiment is a nuclear physics experiment which was conducted at GSI
Helmholtz Centre for Heavy Ion Research in 2010. Its purpose is to increase the knowl-
edge about astrophysical reaction rates for the synthesis of neutron-rich nuclei and about
the structure of the exotic nuclei in that region. In the experiment, unstable, neutron
rich 4 ≤ Z ≤ 10 nuclei were shot at light target nuclei. [6]

3.1 Source of Unstable Nuclei

The accelerator facility at GSI is illustrated in Figure 3.1 [7]. In order to produce a
beam of unstable nuclei, stable ions are accelerated in a linear accelerator (UNILAC)
before injection into a synchrotron accelerator SIS (SchwerIonenSynchrotron, heavy ion
synchrotron). In SIS, the stable ions are accelerated to relativistic energies. After the
ions have gained sufficient energy in the SIS, they are directed at a light, but thick,
production target in the FRS (FRagment Separator). In S393, the stable ion of the
beam was 40 Ar and the production target used was made of 9 Be with a density of
4.011 g/cm2 [8, 9].
The collisions in the FRS can result in many fragments. Among these fragments are
the interesting unstable 4 ≤ Z ≤ 10 nuclei. In order to filter out all but a specific kind
of nuclei from the different fragments, a combination of magnetic fields and degraders
are used [10]. The magnetic fields will filter out ions depending on their velocity and
charge-to-mass ratio [10]. The degraders consist, in principle, of a layer of matter with
adjustable thickness, allowing ions to be filtered depending on the energy lost in the
material, in accordance with Equation 2.4 [10]. With this setup, ions with specific
charge and mass can be selected from the fragments [10].

3.2 Cave C Setup

The experimental setup, as illustrated in Figure 3.2, is located in Cave C. Additional
details about the geometry of the detectors can be found in Appendix C and on the
land02 unofficial site [11].
Ions from the FRS are sent into Cave C, typically at energies on the order of a few
hundred MeV/nucleon. As the ions enter Cave C, they hit the POS (POSition) detector,
which starts a timer used to measure the velocity of the particles. It also stops a timer

10
 3.2 Cave C Setup

SIS
Heavy
Sources
FRS

UNILAC
Experimental hall
Sources ESR

Cave C

Target hall

Figure 3.1: A sketch of the accelerator facility at GSI [7]. In the S393 experiment, stable
ions were accelerated in the UNILAC and SIS and collided with a produc-
tion target in the FRS to produce the beam of unstable nuclei used in the
experimental setup in Cave C.

LAND
ROLU XB
POS ALADiN
Nuclei from
FRS Neutron arm

x
z
GFI
y Target
SST TFW

PDC Fragment arm

DTF

Proton arm

Figure 3.2: The experimental setup for S393 in Cave C at GSI [7]. The filled obstacles
were already defined in GGLAND (see Section 4.2), while the others were
implemented as part of this project.

11
3 The S393 Experiment

started in the FRS, which gives a measurement of the speed of the incoming particle.
POS is a thin sheet of scintillator material with light guides and PM-tubes at all four
sides perpendicular to the ion-beam.
After the ion-beam has passed through POS, it passes through ROLU (Rechts, Oben,
Links, Unten), which measures the spread1 of the beam and can be used as a veto-trigger
that tells the data acquisition to reject coming events, if the beam particles are too far off
the beam axis. ROLU consists of four plastic scintillators defining a rectangular opening.
By moving the scintillators, the position and size of the opening can be adjusted, and
only events related to ions passing through it are recorded and used for analysis.
After ROLU and POS, the ions arrive at the target located inside the XB (Crystal
Ball) detector. The beam reacts with the target and new particles are ejected. To detect
these, the XB has a 4π-coverage with blind spots along the beam axis for the incoming
beam and outgoing forward reaction products. The detector consists of 159 removable
pentagonal and hexagonal NaI crystals placed in a spherical shell (there is room for
162 crystals, although 3 are removed). The XB is mainly used to detect photons and
protons, since these often are emitted at large angles relative to the beam axis.
Inside XB are the SSTs (Silicon STrip detectors), consisting of planar silicon detectors
divided into many sections by the segmentation of the read-out electrodes. There are
eight of these detectors in total: two are positioned before the target to determine the
angle of the incoming beam. Likewise, two are positioned behind the target determine
the angle of the outgoing beam, whereas the remaining four surround the target in
directions orthogonal to the beam axis.
Although the particles may be emitted in all directions, most products continue for-
ward in the beam direction. These particles exits the XB and may reach ALADiN (A
LArge DIpole magNet). Inside this magnet, charged particles are bent in almost circu-
lar arcs due to a nearly-constant magnetic field in the y-direction, defined as pointing
upwards, orthogonal to the floor. Although the field is almost constant within the mag-
net, there is a notable fringe-field, which simulations need to take into account. The
actual magnetic field experienced by particles passing through ALADiN is presented in
Section A.3. The field in ALADiN is used to separate different particles depending on
their mass-to-charge ratio. Most particles entering ALADiN travel at approximately the
same speed as the ions of the beam, which together with the mass-to-charge ratio and
the magnetic field determines the path the particles take. The setup branches of into
different arms for particles with different mass-to-charge ratio:

• Proton arm: Protons, which have a low mass-to-charge ratio, are heavily affected
by the magnetic field and their paths are bent into the PDCs (Proton Drift Cham-
ber). The PDCs are basically a collection of vertical and horizontal charged wires,
generating an electric field, allowing charged particles to be detected by collecting
freed electrons in their tracks. After passing through the PDCs, the protons hit
the DTF (Dicke ToF-wall), a detector made out of scintillator-bars.
• Fragment arm: Heavier ions are bent to somewhere between LAND and the
1
The maximum angle between the beam axis and a particles trajectory.

12
 3.2 Cave C Setup

PDC, and hit the two GFIs (Großer FIberdetektor). A GFI is a collection of
vertical scintillating fibres packed closely together into one layer. Around and
between the fibres is a thin layer of wrapping material to keep the light from
leaking into adjacent fibres.
After passing the GFIs, the ions hit the final detector, TFW (Time of Flight Wall),
which consists of two planes. Each plane in turn consists of multiple vertical or
horizontal paddles of scintillating material.

• Neutron arm: Neutrons, which are uncharged particles, are not affected by the
magnetic field of ALADiN and thus continue forward into LAND (Large Area
Neutron Detector) where they may be detected.

The final steps for all three paths measure the time since the incoming ion passed the
POS-detector, which gives the speed of the particle. In real life, things are not always as
simple as described above, as particles behave stochastically and may stray from their
paths or interact with matter in an unintended way. This is one of the reasons why
ROLU and other veto-triggers are in place. This is also why simulations need to be
performed, to better understand the influence of these effects.

13
4. Simulation and Analysis Tools

When simulating a nuclear physics experiment like S393, there are primarily two software
solutions available: FLUKA and GEANT. The software chosen for this project was GEANT.
Reasons for this choice include the fact that the Subatomic Physics Group at Chalmers
University of Technology has experience with GEANT, ease of acquiring the source-code
and the more permissive license. GEANT4 was chosen over GEANT3 since GEANT4 is
under active development, and as a consequence has more up-to-date physics models.
A wrapper library and program called GGLAND was used to access the functionality of
GEANT4.
After generating data using GGLAND, the data has to be analysed. This was done by
using various tools including ROOT, RALF’S TRACKER and MATLAB.

4.1 GEANT
GEANT is a toolkit for Monte-Carlo simulations of particles [12]. Monte-Carlo simula-
tions are stochastic methods for simulating physical processes using random numbers.
Depending on the physical process, different probability distributions are used. Using
a large set of data points, the method should converge to the expectation value in the
same way as the real process.
While GGLAND may be used both together with GEANT3 and GEANT4, this section
focuses on details in GEANT4. Particle tracking (navigation) in GEANT4 is based on a
tree structure of volumes. Such a structure is illustrated in Figure 4.1. The outermost
volume is called the world volume. Within the world volume, a set of daughter volumes
are declared and placed, which in turn may contain other daughter volumes [12]. A
“shadow world” can also be specified where a set of daughter volumes can be declared
correspondingly. GEANT4 will then, in addition to the physics and geometries of the
world volume together with its sub volumes, also consider the boundaries of the shadow
world. This is useful, since volumes in the same world may not overlap. The tree
structure makes it less complex to track particles, making the simulations faster. This is
because GEANT4 does not need to check for collisions in more complex subvolumes if the
mother-volume has not been hit. A special case of such world subdivision approaches
are also used in real time 3d engines such as id tech 1-3, which suggests that such trees
provide an efficient solution for location look-up [13].
GEANT4 uses different models for various physics events and effects. The models to

14
 4.2 GGLAND

be used are specified in a physics list. From the physics list models, GEANT4 generates
random time steps at which discrete physical events take place for each particle, given
certain conditions, and then it determines what will actually happen and moves the
particle along its track until the first event is scheduled to occur or the particle reaches
a geometric boundary [14]. In addition, GEANT4 takes into account specified continuous
interactions during the steps, such as Bremsstrahlung or interactions with external fields;
and interactions that take place when a particle is approximately at rest, such as a
positron annihilation [14].
Some of the processes included in GEANT4, for example Bremsstrahlung, give rise to an
infrared divergence, in which an infinite number of massless particles with infinitesimal
energies are created. These particles, although infinitely many, sum up to finite energies,
and are thus not a problem physically. However, since it is impossible to simulate an
infinite number of particles, GEANT4 has a lowest energy for which to generate particles.
This lowest energy is determined by a range-cut. The range-cut is specified as a length,
and particles with energy too low to travel longer than this length are not generated.
The physics in GEANT4 is not fully validated [15], and since the underlying physics
itself is not completely understood, there are limitations on what GEANT4 is capable of
simulating. For example, the experiments running at the LHC are not fully evaluated
and hence, one cannot expect correct results in such cases, as of now. This is also the
case for the reactions studied in the S393 experiment, which is why we have to start
simulating reaction products after the target. As an added benefit, since the reaction
products are known, we know what to look for in the simulation output. This is in
contrast to having GEANT4 randomly generate them according to a model.

world shadow
world

1 2 3 1 2

1-1 1-2 1-1 1-2

Figure 4.1: An illustration on how GEANT4 divides the world volume into several daugh-
ter volumes. A “shadow world” can be created which can be divided corre-
spondingly.

4.2 GGLAND
GGLAND is designed to allow for command line simulations of subatomic particle ex-
periments [16]. In order to simulate an experiment with GEANT4 through GGLAND, the

15
4 Simulation and Analysis Tools

obstacles1 must first be specified in GGLAND. This is done by adding modules to the GG-
LAND source tree and recompiling the program. For a short example of a C++ source
file defining a detector, see Appendix B. When all obstacles necessary for the planned
simulation are defined, they can be placed in the world volume by providing appropriate
command line options to GGLAND, specifying the position and rotation of each detector,
among other things such as detector material.
As mentioned in the paragraph about delta-electrons in Section 2.1.2, a particle travel-
ling through matter is often accompanied by several short-range electrons. Due to these
electrons, a particle hitting a detector is usually accompanied by several electrons, also
hitting the detector, giving rise to multiple energy deposits. To deal with this, GGLAND
will cluster nearby energy deposits within the same segment and store information on
which particle made the biggest contribution to each of the clustered hits.
For a more detailed description of GGLAND, including a quick reference on how to
define detectors, see Håkans Johansson’s GGLAND write-up [16].

4.3 RALF’S TRACKER

As part of the event reconstruction, the trajectory of a particle, its track, needs to be
calculated. RALF’S TRACKER, hereafter called “the tracker”, is a tool under development
used to calculate particle tracks from experimental data. As input, besides a ROOT file
(see Section 4.4) containing measurement data, the tracker takes a special configuration
file describing detector positions and the ALADiN current.
The tracker tries to find the path for individual particles through the experimental
setup that best matches detector hits in input data. From this track, the mass number
can be determined for a particle with a given charge, as indicated by Equation 2.1. The
charge can be identified independently of the track, by, for example, energy deposits
in the TFW. Thereby allowing the tracker to identify both the mass and charge of the
particle.
By using the tracker with simulated data, the methods used to analyse actual ex-
perimental data can be evaluated. Since the analysis plays a large part in identifying
events, this is essential in order to determine the acceptance of a setup. A version of
RALF’S TRACKER, modified by Håkan Johansson to accept simulated data, was used for
tracking.

4.3.1 Configuration
To tailor the tracker for the needs of a specific experiment, the user is to make changes
to a specific source-file, known as experiment_specific.hh, by default. These changes
are made to tell the tracker to skip events which fulfil certain conditions. For example,
events with energy depositions outside a set of intervals may not correspond to the
expected energy deposition of particles of any Z, as predicted by Equation 5.3, and thus
1
An obstacle is a part of the experimental setup that the particle has to pass through.

16
 4.4 ROOT

the charge cannot be identified. The tracker can be told to discard such events rather
than trying to identify them.
After compiling the tracker, configuration files are used to describe different exper-
imental setups. These files also contain calibration data (see Section 5.3.2) such as
detector position offsets in the detector coordinate system, as well as custom defined
variables.

4.3.2 Outline of a Possible Tracking Algorithm

The algorithm outlined in this section is intended to explain the needed inputs and inner
workings of a fully realised tracking algorithm, as implemented in, for example, RALF’S
TRACKER. It is an interpretation of the algorithm outlined in a presentation given by
Ralf Plag at the S393/land02 Workshop 2011 [17].
This algorithm assumes that the particle travels straight when not affected by any
external field, that is, the magnetic field from ALADiN. This is not the case in reality
or in our GEANT4 simulations, since the particles interact with the surrounding medium
at random. Nevertheless, the track given by this algorithm may be thought of as an
approximate mean trajectory.
In order to solve the problem of a particle travelling through the magnetic field of
ALADiN, β, m0 and Z of the particle need to be determined. The β can be measured
from the time it takes for a particle to travel to the TFW relative to a reference time.
With an approximate β, Z is determined from the energy depositions in the TFW
using Equation 2.4. Given the β and Z approximations, the tracker numerically solves
the equation describing a particle travelling through the magnetic field of ALADiN, for
different m0 [17]. The m0 corresponding to the track yielding detector hits best matching
to the hits of the input data is then used.
The tracker can calculate the tracks both forward (defining the direction via the
intersections at the SSTs), or backwards (defining the direction via the intersections at
the GFIs), or a combination of both (using all four intersections). Those are the different
tracking modes used by RALF’S TRACKER [18].

4.4 ROOT
ROOT is a CERN developed C++ framework well suited for data analysis involving
large amounts of list-mode data from nuclear and particle physics experiments [19]. The
native ROOT file format is used by both GGLAND and the tracker, which makes ROOT
suitable for extracting information from these programs. Files in ROOTs native format
are hereafter called ROOT files. ROOT can be used interactively through an interpreter,
CINT, which allows the user to make quick histogram and scatter plots, as well as to
apply various filters (known as “cuts”) to the data. It is also possible to use the ROOT
libraries from a C++ program or script to extract information.

17
5. Simulation and Analysis
Procedure

The only part of the S393 experiment that was simulated were the obstacles used to
identify outgoing fragments. Namely: the fragment arm (GFIs and TFW), ALADiN
and the two forward SSTs. The obstacle positions used in the simulation are listed in
Table 5.1.
When analysing the outputs of the simulations, we exclusively looked at the clustered
hit with the highest energy contribution, the primary hit, for a given event and detector.
This will, in most cases, correspond to energy depositions by the original particle. It is
also the standard procedure in the tracker at the moment. In rare cases, however, the
original particle will just barely miss a detector, while one of the created δ-electrons will
hit the detector, thus giving the major energy contribution. Such events are rare and
easily filtered with appropriate energy cuts, since they give a low energy deposition.

Table 5.1: Obstacle positions used in simulations. The positions are given as (x, y, z)
coordinates relative to the beam target. The rotations are made around the
internal negative y axes of the obstacles.
Obstacle Position/cm Rotation/deg
ALADiN (0, 0, 255) 7
FTF (−232.94, 0, 1124.33) 15
GFI1 (−56.94, 0, 467.504) 15
GFI2 (−72.469, 0, 525.450) 15
SST1 (0, 0, 11) 0
SST2 (0, 0, 14) 0

5.1 Determining an Appropriate bρ

According to Equation 2.1, charged particles with a constant magnetic rigidity, bρ =
p/q, are equally affected by a homogenous magnetic field. While the magnetic field
generated by ALADiN is not perfectly homogenous, it may nevertheless be used as an

18
 5.2 Configuring and Running GGLAND

approximation to predict the outgoing angle of charged particles after ALADiN.

To ensure that the particles fired in the simulation hit the fragment arm, an appropri-
ate bρ needs to be determined. In Appendix A, an expression for determining a bρ for a
desired outgoing angle and a given magnetic field is derived. The resulting expression is
bdz
bρ = , (5.1)
sin(θA ) + sin(ν − θA )

where ν is the angle of the outgoing fragment relative to the initial beam direction,
θA = 7◦ the orientation of ALADiN’s gas tube relative to the beam axis and dz = 2.42 m
the length of ALADiN’s gas tube.
The magnetic field b is controlled by the ALADiN current I. Therefore, different I
result in a different fields. Given I is in amperes, I can be approximated by
T
b = 2.187 · 10−4 I + 0.402T (5.2)
A
This relation is also derived in Appendix A.

5.2 Configuring and Running GGLAND

Before any simulation could be done, it was first necessary to define the geometry of all
obstacles used in the simulation. This was done by following the procedure described
in Section 4.2. The obstacles added to GGLAND as part of this project, the non-filled
obstacles in Figure 3.2, are documented in Appendix C.
We used the GEANT4 physics list QGSP_BERT (Quark, Gluon, String Precompound
with Bertini cascade) in our simulations (the default in GGLAND). This is expected to
give results which are in better agreement with experiments than with just the QGSP list
for lower energies, but does not employ the data driven high precision neutron package
of QGSP_BERT_HP [20]. The choice to use QGSP_BERT instead of QGSP_BERT_HP reflects
the fact that the part of the setup that was simulated is unable to detect neutrons.
The range-cut for particles used in the simulations were the GGLAND defaults: 1 cm
for the world-volume and 0.1 cm for other volumes.

5.3 Preparing and Running RALF’S TRACKER

As mentioned in Section 4.3, some tasks has to be performed before the tracker can be
used. This section describes the steps taken to get the tracker work as intended.

5.3.1 Configuration
The changes made to the file experiment_specific.hh was the addition of a method
for extracting the particle charge from the energy deposited in the TFW, and support
for some extra variables used in communication with external sources. For the modified
source file, see Appendix E.

19
5 Simulation and Analysis Procedure

Charge determination
To be able to extract the correct Z, we used Equation 2.4 with the assumption that β is
approximately constant, which is a valid assumption for sufficiently thin detectors. By
using this approximation Equation 2.4 gives
!2 " ! #
4π z 2 e2 2me c2 β 2
−∆E ≈ ∆x N Z ln − β2 ,
me c2 β 2 4π0 I(1 − β 2 )

where the derivative has been replaced by ∆E

∆x . In order to determine the detector specific
constants, N and I, this equation was rewritten as
" ! #
2me c2
− ∆E ≈ C ln /β 2 − 1 z 2 . (5.3)
I(1/β 2 − 1)

The values of C and I were found by minimising the 2-norm of the difference between
vectors of the most likely energy deposits in the detector of interest (in our case the
TFW) for given pairs (Z, β), and the values predicted by Equation 5.3. For the TFW,
this resulted in C = 0.060319339946223 MeV and I = 9.352906073747982 × 10−7 MeV.
For more details on how these values were found, see Appendix F.
In the tracker, −∆E is known. To calculate Z, we rewrote Equation 5.3 to yield the
formula v
u −∆E
Z=t h  i . (5.4)
u

2me c2
C ln I(1/β 2 −1) /β 2 − 1

In order to reject particles depositing energies far from the predicted peaks, the ad-
ditional condition that |Z − Zint | ≤ 14 , where Zint is the “nearest” integer to Z 1 , was
imposed on the calculated Z. If the calculated Z passed this condition, it was identified
as Zint .
In addition, if −∆E ≤ 10 MeV, the event is rejected. This mostly filters δ-electrons
and perhaps protons and helium, thus it should not have a significant impact on the
results. This cut is a remnant from an earlier experiment_specific.hh that was not
noticed until after some time.

Extra communication variables

The extra communication variables were used to control which β to use as tracker input.
These variables are listed together with other variables in Table 5.2.

5.3.2 Calibration
In order to work optimally with experimental data, the tracker needs to be calibrated
with input data. This is due to the fact that the tracker needs its own internal repre-
sentation of the experimental setup, which can not be made to match the actual setup
1
“nearest” is here defined as floor (Z + 0.5)

20
 5.3 Preparing and Running RALF’S TRACKER

since it cannot be measured with sufficient precision. Calibration is essentially to let the
tracker manipulate its internal experimental setup so that it can identify known input
data. The calibration procedure, whose output depends on the given ALADiN current
and detector positions, aims to minimise the difference between tracked data and input
data.
Below follows a description of the calibration procedure. The calibration steps essen-
tially follow the steps for calibrating the tracker with real experimental data, currently
presented on the tracker website [18]. Some alterations had to be made in order to ac-
count for the differences between our simulations and the situation in a real experiment.
For one, we simulated only one kind of ion in a single simulation and thus no angular
calibration was carried out by comparing the angles of fragments, protons and neutrons.
For the time calibration of the TFW, the actual β of the simulated particle was used as
a first approximation, while a real first approximation would need to contain errors.
In the following list, text written in teletype font refers to variables listed in Ta-
ble 5.2, or command line arguments to the tracker. The algorithm is represented graph-
ically in Figure 5.1. The C++ implementation used in this project is found in Ap-
pendix D.

1. To calibrate the time offset, the tracker is run with --ignore-tof and --track-forward.
--ignore-tof tells the tracker to override the β-identification from the input data
and instead use a specified value. This value is set with Inbeta, and was set to the
β used in the simulation. Forward mode is used in accordance with the outlined
calibration procedure [18]. The value of Truebeta, used for Z identification, is also
set to the simulated β. Truebeta_valid is set to true in order to tell the tracker
that Truebeta can be used for charge identification.
The resulting mean difference between the time of impact of the simulation and the
calculations of the tracker (that is, the mean value of offset_ftft) is then added
to the OFFSET_FTFT variable in the configuration file of the tracker, to compensate
for the difference. This procedure is repeated until the resulting mean difference is
less than 10−5 ns; or until a maximum of four iterations is reached, at which point
the calibration procedure is aborted and the calibration is considered to have failed.

2. After the time calibration, the x coordinate of the TFW is calibrated. The goal
in this step is to bring the mean value of fres_ftfx close to 0. To achieve this,
the tracker is run with the --track-backward option and adding the mean of
fres_ftfx to the x offset of the TFW until the absolute value of the mean of
fres_ftfx is less than 10−5 cm. If this criterion is not fulfilled in 4 iterations, the
procedure is aborted and the calibration is considered to have failed.

3. The final step is to calibrate the mass identification of the tracker. Using the output
file from the previous step, the mass-histogram given by the tracker is shifted to
yield a peak-value matching the known mass of the simulated particle. This is done
by having the tracker calculate the track of a particle with the mass identified by
the tracker during the previous step, and a particle with the correct mass. The

21
5 Simulation and Analysis Procedure

Table 5.2: Configuration variables used by the tracker. The symbol → indicates that the
variable is used as an input to the tracker and is defined in the experiment
configuration file. The symbol ← indicates that the variable is an output of
the tracker. Bold-face variables are variables added for the simulation task.
Name Description
→ Inbeta Used to determine the time of flight when using --ignore-tof
→ Truebeta The actual β. This value is used to determine Z from the
energy deposited in the TFW.
→ Truebeta_valid A Boolean that is true if and only if the value of Truebeta is
deemed correct. When Truebeta_valid is true, the tracker
uses Truebeta to identify the Z of the particles using Equa-
tion 5.3.
→ OFFSET_FTFT The time offset added to the time of flight to make the time
of flight approximated with Inbeta match with the time of
flight in the input data.
← fres_ftft The difference between tracked time of flight and time of flight
in the input data.
← fres_ftfx The difference between tracked x position in TFW and x po-
sition in TFW in the input data

tracker then identifies x-offsets to adjust the positions of the GFIs and the TFW
within the tracker calibration in order to make a particle with the correct mass
hit the detectors where the particle with the identified mass hit. This procedure
is referred to as playing the have/want game on the tracker website [18]. These
offsets were then added in the calibration file to the corresponding offset-vector for
the involved detectors. The tracker is now considered to be calibrated.

After calibration, a new experiment configuration file was created for the tracker,
containing the adjusted detector positions together with calibration data.

5.3.3 Running the tracker

When analysing data, the tracker was run twice in mixed-mode. The first time with
Truebeta_valid set to false to get an approximation of the β of the simulated particle,
and a second time with Truebeta_valid set to true and Truebeta set to the mean β
from the first time, to identify Z of the particles as described in Section 5.3.

5.4 Angular Dependency in Detected Events

In order to find how the spread of the beam affects the ability of the setup to identify
fired particles, a simulation with detector positions according to Table 5.1, firing 220
14 B particles with bρ = 8.8 Tm with a spread of 8◦ , was carried out.

22
 5.4 Angular Dependency in Detected Events

Start

1
3
*
|fres_ftft| < 10−5
true Run tracker
--have=peak(fra_A),Z,beta
false

Run tracker
--ignore-tof
--track-forward Extract offsets
from stdout

Add
mean(fres_ftft) to
OFFSET_FTFT

2
|fres_ftfx| < 10−5
true
false

Run tracker
--track-backward

Add
mean(fres_ftfx) to
x offset of FTF

*
Figure 5.1: Flowchart of the calibration procedure for the tracker. Numbers in shadowed
boxes refers to items in the list in Section 5.3.2

23
5 Simulation and Analysis Procedure

z (beam axis)

Figure 5.2: The opening angle, α, is used as a measurement of the spread of the beam
and the initial direction of the fired particle.

The tracker was not calibrated since the detector positions in the simulation are ex-
actly known, negating the need to adjust detector positions to get good results. Time
calibration was also omitted since it was not needed, although it could in principle be
used to compensate for the additional physics processes considered in the simulation.
Tracking was still done twice in mixed-mode, first in order to get a more realistic value
for β, as identified by the tracker, to use for Z identification during the second run.
Information on what kind of particle caused the major energy depositions in the TFW
for each event could be retrieved from the simulation output ROOT file. This data could
then be used together with the ROOT file generated by the tracker to determine whether
a particle was correctly identified or not.
To obtain the dependence of the number of identified particles on the angle between
the initial trajectory of the particles and the beam axis, α, illustrated in Figure 5.2,
a histogram of cos α was plotted with appropriate cuts. More specifically, cos α was
plotted for particles having been identified by the tracker as having mass m satisfying
|m−14 u| < 0.5 and Z = 5, corresponding to particles having been identified as 14 B. This
was done with additional cuts on the type of particles causing the major energy deposit
in the TFW to see how often the tracker made a correct or incorrect identification.
Due to the way the α of the fired particle is generated, if the spread had no bearing
on the acceptance, the events would have an even distribution over cos α. According to
Equation 1.2, the histogram then immediately gives an unnormalised acceptance as a
function of cos (α).

5.5 Mass Deviations for Tracker Calibrated With

Incorrect Current
One of the goals of the experiment is to identify the particle that left the target according
to mass and proton number. The actual magnetic field is only roughly known due to,
among other things, magnetic hysteresis. Therefore, it was investigated how well the
tracker calibrated with an incorrect ALADiN current works. The process is described
below and illustrated in Figure 5.3. All simulations were done√using 15
 2 events. The
number of bins used for mass peak extraction were floor 100 215 = 18101 on the

24
 5.5 Mass Deviations for Tracker Calibrated With Incorrect Current

interval [0, 30]. The peak was detected by looking for the largest peak in an associated
TSpectrum object with at most 1000 peaks and a resolution of 1.
To have simulated data for analysis as well as to calibrate the tracker, known input
data were generated by GGLAND. The isotopes

SB = {10 B, 11 B, 12 B, 13 B, 14 B, 15 B, 17 B, 19 B}

were simulated with a current of I0 = 2500 A and bρ = 8.8 Tm. The reason for excluding
16 B and 18 B is that these isotopes do not survive the flight time to the TFW because
they are unbound.
After the simulations mentioned above were done, the tracker was calibrated using the
procedure outlined in Section 5.3.2, with the data from the 14 B simulation. The tracker
was calibrated 8 times, every time given a different current Ic , the currents being the
following: 2500 A, 2462.5 A, 2425 A, 2387.5 A, 2350 A, 2312.5 A, 2275 A, and 2237.5 A,
with I0 = 2500 A being the actual ALADiN current used in the simulation and the others
being increasingly incorrect. A calibration profile was created for each of these currents.
Given the created calibration profiles, the simulation output for each of the nuclei in
SB were tracked. This was done for each calibration profile. The calibration profiles
created for SB were also used for different elements with a fixed mass number; the nuclei
22 C , 22 N , 22 O , and 22 F were simulated and tracked.

Deviations might also occur when the tracker has been calibrated as above, but is
given data acquired with a different current than the current used in the calibration
experiment. Therefore, simulations for SB were done with current Isim = 2300 A. Then,
the output data were tracked by using each calibration profile, but giving the tracker a
current
Isim
Itrack = Ic , (5.5)
I0
where Ic is the current given to the tracker during calibration and I0 is the current
used when creating the calibration data. The quantity Itrack was chosen so that the
relative change in current in the tracker is equal to the relative change in current in the
simulation.

25
5 Simulation and Analysis Procedure

14
Simulation B Tracker Calibration

SSTs SSTs

ALADiN ALADiN

m0
Calibration GFI GFI

Gun
(target)
TFW ∆I Gun
(target)
TFW

m0 m0

I0 = 2500 A Ic = I0 − ∆I

Simulation Tracking

∆I 0
SSTs SSTs

ALADiN ALADiN

GFI

Tracking GFI
TFW

Gun TFW Gun

(target) (target)
∆x

Isim Itrack

Figure 5.3: A procedure to investigate the effects of an incorrect tracker current during
calibration. The simulation is run with a current I0 . The tracker, which
has its own experimental setup with an incorrect current Ic , is then given
the output from the simulation and is calibrated by manipulating detector
positions of its experimental setup. The calibrated tracker is then used for
tracking outputs from various simulations, often but not necessarily with the
same currents as was used during calibration.

26
6. Results

In the following section, results from the investigations in Section 5.4 and Section 5.5 are
presented, starting with Section 6.1 on how the number of detected events in a simulation
with 220 14 B , fired with bρ = 8.8 Tm, vary with the opening angle α, as defined in
Figure 5.2. In this section, it is also presented how the tracker is able to identify the
fired 14 B when using mixed-mode tracking.
In Section 6.2, the effects of giving an incorrect current to the tracker during cali-
bration are presented. The identified mass of the particle is taken as the peak in the
mass spectrum returned by the tracker given input data from simulations with 215 fired
particles when using different calibration profiles, identified by the currents Ic given to
the tracker during calibration.

6.1 Angular Dependence of Detected Events

As can be seen from the “over 0 MeV in all” histogram in Figure 6.1a, the number
of events hitting all the relevant detectors is constant for low angles, only to drop at
cos α ≈ 0.9988, corresponding to α ≈ 2.5◦ . This behaviour is slightly modified when
more strict energy cuts are applied to the deposits in the detectors, as can be seen
from the other histograms in Figure 6.1a. Most notably, the cut at 3 MeV for the SSTs
greatly reduces the amount of particles detected, while the number of events is here
dependent on the spread for low α. The 3 MeV cuts in the SSTs roughly corresponds
to the beginning of a high peak in the energy deposition spectrum of the primary hits,
as seen in Figure 6.2. However, the cuts exclude numerous events below the peak value
that would be included in the less strict 2 MeV cut. This gives a plausible explanation
for the drops in the number of accepted events.
In Figure 6.1b, the number of particles hitting the different detectors for various cos α
is presented. It can be seen that particles start to miss the GFI2 at an angle that
corresponds to the drop in the amount of detected events in Figure 6.1a, suggesting
that missing the GFI2 is a cause for the decrease in the amount of events hitting all the
detectors.
Since particles start the miss the GFI2 first, more particles should be eligible for
forward tracking than mixed or backward. Figure 6.1c shows that this is indeed the
case.
Figure 6.1d shows the particles that contributed mostly to the major energy deposits

27
6 Results

in the TFW. Since there are ions other than 14 B giving major energy contributions,
we can conclude that not all of the simulated 14 B nuclei reach the TFW as 14 B . This
indicates that some particles undergo reactions along their flight path resulting in new
fragments. From Figure 6.1d, it is possible to determine the maximum tracking efficiency
of the experimental setup with analysis methods rejecting fragments that decay or react
during the flight.
Also revealed by Figure 6.1d is that at sufficiently large α, almost no ions give a
primary contribution to the energy deposits in the TFW. On the other hand, the fired
14 B ion dominates for low spread angles.

The histogram of the successfully mixed-mode tracked particles, presented in Fig-

ure 6.3, largely follows the same trends as the histograms of simulated events in Fig-
ure 6.1. This suggests that the tracker is efficient in handling trackable events.
Tracking with some of the stricter energy cuts seemingly offers no advantages for the
given simulation, since the amount of incorrectly identified 14 B hardly decreases, while
the numbers of unidentified 14 B notably increases for both cuts. The same can be said
about a stricter A identification, shown in Figure 6.3d, where particles not fired exactly
straight along the z-axis seemingly are easier to identify.

6.1.1 Reasons for the Decrease in Events with Spread

Figure 6.4a together with Figure 6.5 gives a plausible explanation for the drops in events
in the GFI2 for cos α < 0.9988, as seen in, for example, Figure 6.1a. Figure 6.4a depicts
the number of events as a 2D-histogram with respect to the spread cos α and the y
coordinate of the GFI2. For a given spread, the particles are expected to be distributed
approximately homogeneously on a circle, as can be seen in Figure 6.5a. This explains
the fact that the events for a given spread in Figure 6.4a are found within intervals
around y = 0, with most events at the end-points. The points where the bright curve,
defined by the edge of the event-interval for different cos α, intersects with the edge
of the plot corresponds to particles on the expected circle missing the GFI2 in the y
direction, which can be seen in Figure 6.5b. Plots corresponding to Figure 6.4a and
Figure 6.5a display the same characteristics for the other detectors, with one notable
exception: the TFW. As can be seen in Figure 6.4b and Figure 6.5, particles with a
cos α < 0.9988 in the negative y-direction, as well as particles with cos α < 0.9982 in the
positive y-direction seem to have trouble reaching the TFW.
From this it can be concluded that at cos α ≈ 0.9988, particles start to miss the GFI2
due to passing beyond the limiting dimensions of the detector. In the same way, it can
be seen that at cos α ≈ 0.9987 and cos α ≈ 0.9985, GFI1 is frequently missed in the y
and/or x direction, followed by SST2 at cos α ≈ 0.9979, SST1 at cos α ≈ 0.9969, both
only in the y direction, and finally the particles that miss the TFW at cos α ≈ 0.9966 in
the x direction.

28
 6.1 Angular Dependence of Detected Events
Counts

Counts
7000 Hit the SST1 u+v
7000
No cut Hit the SST2 u+v
6000 over 0MeV in all 6000 Hit the TFW
over 2MeV in all
TFW 10MeV, GFIs 2MeV, SSTs u+v 2MeV Hit the GFI1
5000 TFW 10MeV, GFIs 6MeV, SSTs u+v 2MeV 5000
Hit the GFI2
TFW 20MeV, GFIs 6MeV, SSTs u+v 2MeV
4000 TFW 20MeV, GFIs 6MeV, SSTs u+v 3MeV 4000

3000 3000

2000 2000

1000 1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(a) ...caused energy deposits above certain values in (b) ...hit the different detectors.
the detectors. “No cut” shows every particle
fired.
Counts

Counts
7000 Hit all (mixed) All particles
7000
B14
6000 Hit all except perhaps GFI1 or GFI2 (forward) 6000 not B14
Hit all except perhaps SST1 u+v or SST2 u+v (backward) Ions
5000 5000
Electrons
4000 4000

3000 3000

2000 2000

1000 1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(c) ...were eligible for the different tracking modes in (d) ...caused primary deposits in the TFW at-
RALF’S TRACKER. tributed to different particles.

Figure 6.1: Histograms with 150 bins depicting the number of particles in a simulation
with 220 fired 14 B with bρ = 8.8 Tm and with a maximum α of 8◦ for different
cos α that...

×10
3
120

100

80

60

40

20

0
0 1 2 3 4 5 6 7 8 9 10
Energy [MeV]

Figure 6.2: Number of primary energy deposits with certain energies in the SST1u from
a simulation with 220 fired 14 B nuclei with bρ = 8.8 Tm and with a maximum
α of 8◦ . The peak around 0 can most likely be attributed to δ-electrons, while
the other peak is due to 14 B passing through the detector.
29
6 Results

Counts

Counts
6000
No cuts No particle cuts
6000
Correctly identified as B14 5000 Correctly identified as B14
5000 Incorrectly identified as B14 multiplied by 10 Incorrectly identified as B14 multiplied by 10
4000 B14 not identified
B14 not identified
4000

3000
3000

2000
2000

1000 1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(a) Over 0 MeV in all detectors (b) Over 2 MeV in all detectors
Counts

Counts

5000
No particle cuts No particle cuts
6000
Correctly identified as B14 Correctly identified as B14
4000 Incorrectly identified as B14 multiplied by 10 5000 Incorrectly identified as B14 multiplied by 10
B14 not identified B14 not identified
4000
3000

3000
2000
2000

1000
1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(c) Over 20 MeV in TFW, 6 MeV in the GFIs 3 MeV (d) Over 0 MeV in all and |A − 14| < 0.1
in the SSTs (u+v)

Figure 6.3: Histograms with 150 bins depicting the number of particles identified as var-
ious particles by RALF’S TRACKER from a simulation with 220 14 B nuclei
simulated for bρ = 8.8 Tm with a maximum spread of 8◦ for different cos α.
Different parts of the figure corresponds to additional cuts on the energy de-
posits in various detectors. The histogram labelled “No particle cuts” corre-
sponds to every particle the tracker managed to track for the given additional
energy cuts. The “B14 not identified” histogram includes 14 B nuclei, which
were not included in the energy cut. The energy cuts of the different parts of
the figure correspond to different the histograms in Figure 6.1a. Figure (d) is
with a stricter condition for A to be considered correctly identified, otherwise
the condition |A − 14| < 0.5 was used.

30
 25 250

TFW y (cm)
GFI2 y (cm)

400
20 60
350
15 200
40
300
10
20
5 150 250

0 0 200

-5 100
-20 150
-10
-40 100
-15 50
50
-20 -60

-25 0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(a) The y coordinate of the GFI2 against cos α (b) The y coordinate of the TFW against cos α

Figure 6.4: Histograms with 150 × 150 bins depicting the number of primary energy de-
posits at different y-coordinates at different cos α ranges in different detectors
in a simulation with 220 fired 14 B nuclei with bρ = 8.8 Tm and with a maxi-
mum α of 8◦ . Brighter bins indicate more events.

25 25 12
GFI2 y (cm)

GFI2 y (cm)

12
20 20
10
15 10 15

10 10 8
8
5 5
6
0 6 0

-5 -5 4
4
-10 -10

2 2
-15 -15

-20 -20
0 0
-25 -25
-20 -15 -10 -5 0 5 10 15 20 -20 -15 -10 -5 0 5 10 15 20
GFI2 x (cm) GFI2 x (cm)

(a) GFI2: 0.9988 ≤ cos α < 0.9989 (b) GFI2: 0.9987 ≤ cos α < 0.9988

80 80
TFW y (cm)

TFW y (cm)

25

60 60 20

20
40 40

15
20 20
15

0 0
10
10
-20 -20

-40 -40 5
5

-60 -60

0 0
-80 -80
-100 -80 -60 -40 -20 0 20 40 60 80 100 -100 -80 -60 -40 -20 0 20 40 60 80 100
TFW x (cm) TFW x (cm)

(c) TFW: 0.9988 ≤ cos α < 0.9989 (d) TFW: 0.9986 ≤ cos α < 0.9987

Figure 6.5: Histogram wih 150×150 bins depicting the number of primary energy deposits
for x and y-coordinates at different cos α ranges in different detectors in a
simulation with 220 fired 14 B nuclei with bρ = 8.8 Tm and with a maximum
α of 8◦ . Brighter bins indicate more events.
6 Results

6.2 Effects of Incorrect Currents on Tracking

We begin this section by defining ∆m = mdet − msim , where mdet is the mass peak
detected by the tracker and msim is the actual mass of the ion used in the simulations.
The notations used for currents are also the same as those in Section 5.5, with ∆I
referring to the difference in the current of the simulation and tracker during calibration,
but not necessarily during tracking.
The graphs in Figure 6.6 show the results of using different calibration profiles on
simulation output files from simulations with Isim = 2500 A for the isotopes in SB defined
in Section 5.5. The figure shows the mass deviation plotted against the current deviation,
where the different graphs correspond to different isotopes of boron. It is clear that all
isotopes have been identified correctly since the absolute value of the mass deviation is
in the range [0, 0.05 u], which is significantly less than the mass of the isotopes. This
conclusion holds regardless of the current used when calibrating the tracker. If instead,
as in Figure 6.7, A is shown on the x-axis, the spread in mass deviation seems to increase
as A increases.
By trying to track the output from simulations for A = 22 on Z ∈ {6, 7, 8, 9} with
the same I, the results presented in Figure 6.8 were obtained, where, again, the mass
deviation is plotted against the current deviation. It is in this case shown that also all
isobars are correctly identified, as the mass deviations are in the range [−0.02, 0.08 u].
In Figure 6.9 the same output is plotted as a function of Z. In this figure, the spread in
mass deviation seems to decrease as Z increases.
When using a simulation with a different current, Isim = 2300 A, and running the
tracker with that simulation output, using the old calibration profiles together with a
tracker-current Itrack according to Equation 5.5, the results presented in Figure 6.10
were obtained, where the mass deviation is plotted against the current deviation for
eight different isotopes of boron. Still, all isotopes are identified correctly, but the mass
deviation seems to be dependent on how much deviation in current there was during
calibration.
As can be seen in Figure 6.11, where the rotation is plotted against the deviation in
current, the tracker moves the detectors during the calibration procedure, so that the
line from the centre of ALADiN through the centre of the detectors becomes rotated by
an angle that compensates for the deviation in current. Since the tracker only modifies
the x coordinate of the detectors, this makes the arm slightly longer or shorter, and the
detectors themselves are not rotated. This effect of an incorrect current is in agreement
with what Ralf Plag presented at the S393/land02 Workshop 2011 [17]. A visualisation
of this is presented in Figure 6.12, where zdist is the distance from the detector to the
centre of ALADiN, xoff is the x-offset of the detector, and
xoff
 
θoff = arctan (6.1)
zdist
is the resulting effective rotation.
Additional parameters of interest, such as the initial direction of the outgoing frag-
ment, presented in Figure 6.13, are also identified by the tracker. As seen in Figure 6.13b

32
 6.2 Effects of Incorrect Currents on Tracking

0.05
A = 10 Z = 5
0.045
A = 11 Z = 5
0.04 A = 12 Z = 5
0.035 A = 13 Z = 5
(u)

0.03 A = 14 Z = 5
A = 15 Z = 5
∆m

0.025
A = 17 Z = 5
0.02
A = 19 Z = 5
0.015
0.01
0.005
0 50 100 150 200 250 300
∆I (A)

Figure 6.6: Mass deviation, ∆m = mdet − msim , between the simulated and tracked mass
for Z = 5 and different A, plotted against the difference, ∆I = I0 − Ic ,
between the current used to generate calibration data and the current used
when calibrating the tracker. The data was acquired by running the tracker,
using different calibration profiles, on simulation data with the same current
as the input used for calibration. The quantities I0 and Ic are defined in
Section 5.5.

0.05
Ic = 2237.5 A
0.045
Ic = 2275 A
0.04 Ic = 2312.5 A
0.035 Ic = 2350 A
(u)

0.03 Ic = 2387.5 A
Ic = 2425 A
∆m

0.025
Ic = 2462.5 A
0.02
Ic = 2500 A
0.015
0.01
0.005
10 11 12 13 14 15 16 17 18 19
A

Figure 6.7: The same data as in Figure 6.6, plotted against A.

33
6 Results

0.08
A = 22 Z = 6
0.07
A = 22 Z = 7
0.06
A = 22 Z = 8
0.05
A = 22 Z = 9
(u)

0.04
0.03
∆m

0.02
0.01
0
-0.01
-0.02
0 50 100 150 200 250 300
∆I (A)

Figure 6.8: Mass deviation, ∆m = mdet − msim , between the simulated and tracked mass
for a A = 22 and different Z, plotted against the difference, ∆I = I0 − Ic ,
between the current used to generate calibration data and the current used
when calibrating the tracker. The data was acquired by running the tracker,
using different calibration profiles, on simulation data with the same current
as the input used for calibration. The quantities I0 and Ic are defined in
Section 5.5.

0.08
Ic = 2237.5 A
0.07
Ic = 2275 A
0.06
Ic = 2312.5 A
0.05
Ic = 2350 A
(u)

0.04 Ic = 2387.5 A
0.03 Ic = 2425 A
∆m

0.02 Ic = 2462.5 A
0.01 Ic = 2500 A
0
-0.01
-0.02
6 7 8 9
Z

Figure 6.9: The same data as in Figure 6.8, plotted against Z.

34
 6.2 Effects of Incorrect Currents on Tracking

0.04
A = 10 Z = 5
0.03
A = 11 Z = 5
0.02
A = 12 Z = 5
0.01
A = 13 Z = 5
(u)

0 A = 14 Z = 5
-0.01 A = 15 Z = 5
∆m

-0.02 A = 17 Z = 5
-0.03 A = 19 Z = 5
-0.04
-0.05
-0.06
0 50 100 150 200 250 300
∆I (A)

Figure 6.10: Mass deviation, ∆m = mdet −msim , between the simulated and tracked mass
for different A and Z = 5, plotted against the difference, ∆I = I0 − Ic , be-
tween the current used to generate calibration data and the current used
when calibrating the tracker. The simulations were done with a current
of 2300 A and the current given to the tracker was calculated from Equa-
tion 5.5. The quantities I0 and Ic are defined in Section 5.5.

1
TFW
0.9
GFI1
0.8
GFI2
0.7
Predicted
(deg)

0.6
0.5
θoff

0.4
0.3
0.2
0.1
0
0 50 100 150 200 250 300
∆I (A)

Figure 6.11: Fragment arm rotation after tracker calibration, as a function of the differ-
ence, ∆I = I0 − Ic , between the current used to generate calibration data
and the current used when calibrating the tracker. The angle was calculated
using Equation 6.1. The “expected” line is the predicted difference between
the angle of the outgoing fragment and the fragment arm for the trackers
incorrect current, as given by Equation 5.1. The quantities I0 and Ic are
defined in Section 5.5.
35
6 Results

x
z
ALADiN
y
xoff
θoff
zdist
I2 < I1

θoff
I1

Gun
(target) SSTs GFI1 xoff
GFI2
TFW

Figure 6.12: The effect of different currents on a particles track. Since the tracker uses
the incorrect current I2 , it must move the detectors of the fragment arm
during calibration step 3 in order to identify the desired mass.

and Figure 6.13d, the x and y-components of the initial direction corresponds well with
the initial direction on the simulation, with the exception of a few outliers. This be-
haviour does not seem to depend on the current given to the tracker, as can be seen by
for example comparing Figure 6.13b and Figure 6.13d.

36
 6.2 Effects of Incorrect Currents on Tracking

0.025 0.025
Simulated initial normalised fragment direction, x-component

Simulated initial normalised fragment direction, y-component

0.02 0.02

0.015 0.015

0.01 0.01

0.005 0.005

0 0

-0.005 -0.005

-0.01 -0.01

-0.015 -0.015

-0.02 -0.02

-0.025 -0.025
-0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025 -0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025
Reconstructed initial normalised fragment direction, x-component Reconstructed initial normalised fragment direction, y-component

(a) x direction, Itrack = 2500 A (b) y direction, Itrack = 2500 A

0.025 0.025
Simulated initial normalised fragment direction, x-component

Simulated initial normalised fragment direction, y-component

0.02 0.02

0.015 0.015

0.01 0.01

0.005 0.005

0 0

-0.005 -0.005

-0.01 -0.01

-0.015 -0.015

-0.02 -0.02

-0.025 -0.025
-0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025 -0.025 -0.02 -0.015 -0.01 -0.005 0 0.005 0.01 0.015 0.02 0.025
Reconstructed initial normalised fragment direction, x-component Reconstructed initial normalised fragment direction, y-component

(c) x direction, Itrack = 2237.5 A (d) y direction, Itrack = 2237.5 A

Figure 6.13: Reconstructed x and y-components of the direction of the outgoing fragments
at the target plotted against the initial normalised x and y-components mo-
mentum of the outgoing fragment in the simulation. Itrack refers to the cur-
rent given to the tracker. The data presented is from a simulation with 215
14 B fired with bρ = 8.8 Tm with a maximum α of 16 mrad and an ALADiN

current of I = 2500 A.

37
7. Discussion

In this section, the results mentioned in the previous section are discussed. In Section 7.1
it is discussed how the number of detected events in a simulation, vary with the opening
angle. It is also presented how the tracker is able to identify the fired 14 B when using
mixed-mode tracking. After that, in Section 7.2, the effects of giving an incorrect current
to the tracker during calibration are discussed.

7.1 Angular Spread, Detected Events

From comparing Figure 6.3 with Figure 6.1, it is clear that the tracker is efficient at
identifying fired 14 B hitting the detectors for the |A − 14| < 0.5 identification condition.
However, Figure 6.3d indicates that the tracker has problems identifying the mass of the
particle to high precision, but this surprisingly gets better for moderate α. This might be
due to differences in how the tracker and GEANT4 calculate tracks. Increased precision
will probably not be necessary for simulations where single particles are fired, as done in
this project, but it may be needed for cutting down the number of incorrectly identified
particles in simulations where more than one particle is fired simultaneously.

7.1.1 Energy Cuts

Although it was concluded in Section 6.1 that energy cuts do not improve the efficiency
of the tracker, some energy cut above zero should be used to simulate the fact that the
detectors in the experimental setup have a threshold energy. The 2 MeV cut is possibly
too harsh, but it indicates that such cuts need not distort the dependence of α on the
number of detected hits. There is still a danger though, since particles with lower Z
and higher β would deposit lower energy, thus bringing the peak closer to the threshold
energy, which may give rise to behaviour similar to the one seen in the strictest cut
in Figure 6.1a. To determine when and if this could become a problem, values for the
constants in Equation 5.3 for the SSTs would be required, as well as a value for the
threshold energy.
There are also other, “hidden”, energy cuts in the simulation. For one, the Z-
identification described in Section 5.3 will only identify particles depositing energies
in the TFW around the values predicted by Equation 5.3. These cuts make the Z-
identification more well-behaved, but systematically reject outliers. It may thus be worth

38
 7.1 Angular Spread, Detected Events

trying out other ways to identify Z. The energy deposits of charged particles for medium-
thick detectors and certain energy ranges supposedly follow a Landau-distribution [21].
This may be useful to define a more sophisticated method to determine the most likely
charge of a particle given its energy distribution, without rejecting any particles.
Another “hidden” energy cut is due to the range-cut. As mentioned in Section 4.1,
GEANT4 will not generate particles with too low energy to travel a certain distance.
This will bring the effective energy cut to somewhere above zero. A comparison between
different range-cuts can be found in the GGLAND write-up [16]. The primary effects
of different energy cuts seems to be that a lower range-cut increases the amount of
secondary hits in the detectors [16]. Since the secondary hits were not considered in the
analysis, as described in the introduction in Section 5, this should not affect the results
considerably.
In any case, the tracker is often able to track most of the particles after the cuts and
the output histograms closely reflect the input simulated data, as can be seen from the
similarity of the histograms in Figure 6.3 with corresponding histograms in Figure 6.1a.

7.1.2 Dip in Detected Events1

As was seen in Section 6.1.1, the dip in the amount of events detected when α & 2.5◦
(cos α . 0.9988) can probably be attributed to particles starting to miss the GFI2 and
then other detectors. However, the fact that particles stop hitting the TFW in an ellipse
for a given α at cos α ≈ 0.9988 cannot be explained by the geometry. Some plausible
explanations could be

1. Particles with a certain θ and φ are bent off in ALADiN in a way that gives rise
to the effects observed in Figure 6.4b.

2. GGLAND treats particles missing the GFIs differently, since they do not cross
boundaries as often as the ones which hit the GFIs, making them miss the TFW
as well as the GFIs.

The first explanation could be investigated by looking at the particles fired with cos α ≈
0.9988. By plotting the number of hits in the TFW against arctan (py0 /px0 ) of the
fired particles, problematic φ-angles could be identified. For these angles, a GGLAND-
simulation could be run with ALADiN compiled to dump the field-strength at the points
the particle travel. These would reveal if there are anomalies in the ALADiN field
map causing a loss of particles. Even without anomalies, the possibility of the field of
ALADiN causing particles to miss the TFW cannot be ruled out, as even comparatively
small initial differences in particle trajectories in ALADiN could give a notable spread
in the distant TFW.
The second explanation has been ruled out by a simulation2 with the GFIs removed,
in which the hit-pattern on the TFW was reproduced.
1
This section contains conclusions based on simulations performed after the deadline for the first version
of the report.
2
This simulation was carried out after the first version of this report was handed in.

39
7 Discussion

It has also been concluded that the geometry of the iron-core of ALADiN is not the
cause for the reduced amounts of hits in the TFW. This was investigated by running
a simulation2 in which the geometry of ALADiN was removed, and only the field was
present.
Although interesting, the particles missing the TFW are not the primary constraints
to the ability of the simulated setup to detect particles with a α & 2.5◦ , as this is due
to particles missing the GFI2.

7.2 Compensating for Incorrect Current

As seen in Figure 6.7 and Figure 6.9, the spread in mass deviation increases with mass
and decreases with charge, respectively. This can possibly be explained by the fact that
a smaller charge to mass ratio gives rise to a larger uncertainty in mass. This can be
seen by noting that bρ = mγ
q and combining this with Equation 5.1. Then

qbdz
m=
γ (sin(θA ) + sin(ν − θA ))
From this it follows that when ν is small, the mass deviations becomes large. This is
what happens for heavy particles with low charge.
As can be seen from Figure 6.11, the tracker is able to compensate for an incorrect
current by moving the detectors of the fragment arm in a way that roughly corresponds
to rotating the entire fragment arm around the centre of ALADiN. Since the tracker
does not really rotate the fragment arm, there are a at least two errors introduced,
disregarding the incorrect detector positions:

1. The detectors of the fragment arm will be further away from the centre of ALADiN.

2. The detectors of the fragment arm will not face the centre of ALADiN.

Since the particles move at a considerable fraction of the speed of light, the resulting
time-offset from moving the TFW will be negligible compared to the time resolution of
the TFW and can be re-calibrated. Orders of magnitude larger time-offsets would cause
problems in determining β, making the tracker unable to correctly identify particles, but
such errors in the ALADiN current would be unrealistic.
Likewise, the fact that the detectors are not directly facing ALADiN anymore could
cause problems, but even for an error of 10% in current, the detectors would be merely
a degree off, just barely distorting the beam area of intersection with the detectors.
Thus, neither of these errors should considerably impact ability of the trackers to
identify particles. The fact that the tracker is using incorrect detector positions after
ALADiN is still worrying, and it may introduce other errors that perhaps could be
revealed by other, less idealistic, simulations. Less idealistic as in with spreads in all
parameters, and with more than one particle fired at once.
A possibly better solution for calibration would perhaps be to adjust the current
instead of detector positions in cases where this would lead to a good fit. A simple way

40
 7.2 Compensating for Incorrect Current

to implement this for our simulation and tracking scheme would be to look at how many
degrees the tracker effectively has rotated the fragment arm with, and use that angle
together with the by-the-tracker given bρ = p/q to find a current corresponding to the
offsets of the detectors.
A typical indication that it is time to adjust the current would be if each detector of the
fragment arm have been approximately rotated with respect to the centre of ALADiN
by the same angle, that is, if

x offset
 
θoff = arctan
distance from the centre of ALADiN

for the different detectors are roughly the same. If this is the case, we can get an
approximation for the difference in I by using Equation 5.1 for both the real and tracked
current and outgoing angle of the particles. This yields:

dz
(breal − btrack ) = sin (νreal − θA ) − sin (νtrack − θA ) ≈ νreal − νtrack ≡ θoff . (7.1)
bρ

Inserting Equation 5.2 together with numerical values for dz = 2.42 m gives

Ioff = (Ireal − Itrack ) = Cbρθoff , (7.2)

A
where C = 1888.80350899349 Tm , bρ is given in Tm and θoff in radians. This gives
an approximation of the real current from the x-offsets of the detectors given by the
tracker and the current specified in the trackers config-file. As a quick verification,
θoff = 1◦ for bρ = 8.8 Tm gives Ioff ≈ 230 A, which is reasonably close to the actual Ioff
at approximately 250 A, as seen in Figure 6.11. This first-order approximation does not
rely on any measurements from the real experimental setup. It should be possible to
track again after adding the current offset and iterate until the real current is known.
Equation 7.2 can perhaps be used to understand the drift in mass deviation, as seen
in Figure 6.10, when the tracker and simulation was run with different currents than
during calibration. Since the difference in tracker current, Itrack − Ic , is not equal to
the difference in simulation current, Isim − I0 , as stated by Equation 5.5, the rotation
angle of the fragment arm no longer corresponds to the current used in simulation. The
current used for tracking, Itrack , was chosen since the relative difference in field strength
might be easier to relate to the change in current between two runs during an actual
experiment than the absolute difference.
Equation 5.1 should be used with care, since it relies on approximating the effective
magnetic field felt by the particle with a homogeneous field within ALADiN. Since the
real field felt by the particle will vary depending on the particles trajectory, the formula
may not be valid for particles with a large α or a bρ far away from the bρ predicted
by Equation 5.1 (8.8 Tm for I = 2500 A). It has nevertheless successfully predicted bρ
for different currents, and particles with a α up to about 2.5◦ do on average end up
hitting ellipses aligned around the centre of the detectors of the fragment arm. This
suggests that the field felt by the particle does not change too drastically for particles

41
7 Discussion

well behaved enough to hit the detectors of the fragment arm, which is a condition for
Equation 5.1 to remain valid.
Another parameters of interest, the angle of the outgoing fragment at the target
is also reconstructed by the tracker. This can be seen from the line in Figure 6.13,
where components of the reconstructed initial direction match the simulated, with the
exceptions of a few outliers.
This identification does not seem to be related to the current given to the tracker,
as can also be seen in Figure 6.13. This is reasonable, since the tracker does not move
the SSTs during any of the steps in the calibration scheme. Not moving the SSTs is
reasonable, since uncertainties in the magnetic field mostly would affect the track after
ALADiN, those uncertainties should be compensated for by either moving the detectors
after ALADiN or adjusting the current. Trying to compensate for these errors by moving
the SSTs would cause these uncertainties to affect the entire setup and instead of just
the parts that would be affected by these uncertainties in the real experiment.

7.3 Outlook3
There is still work to be done before GGLAND and RALF’S TRACKER will be applicable
to the entire experimental setup used in S393. For one, the PDC, XB and LAND
detectors as defined in GGLAND currently do not give output data compatible with
RALF’S TRACKER, preventing tracking of off the full setup.
There is also work to do relating to detector geometries in GGLAND. The most ap-
parently problematic detector is the PDC, for which the planes used to identify the x
and y coordinates of a passing charged particle currently have to be placed as sepa-
rate detector and rotated relative to one another. Apart from that, more details of the
detectors, such as read-out electronics, could always be added, although this has the
downside of making the detectors very specific, making it difficult to reuse the code for
other detectors. A possible solution to this and the PDC problem would be to allow
users to define detectors in terms of other, already defined, detectors. This would make
it easy to add additional details to a basic frame, or to merge the two planes of the PDC
into one detector.
Related to the analysis of the S393 experiment, the acceptance of fragment arm to-
gether with the other arms and the XB needs to be investigated in order to determine
the acceptance of the setup. With the acceptance, cross sections of the studied reactions
could be determined.
In the far future, as part of the expansion of GSI into FAIR (Facility for Antiproton and
Ion Research), new detectors such as NeuLand, will eventually need to be implemented
if GGLAND is to remain a usable tool for simulating future experiments.

3
A part of this section regarding an incorrect rotation of SST2 in the tracker and its effect on the results
have been removed in this version of the report, as it no longer applies to the data presented in this
version of the report.

42
Glossary
CINT A relaxed C++ interpreter used in ROOT, page 17

GEANT A toolkit for Monte-Carlo simulation of particles, page 14

GGLAND A wrapper library and program used to communicate with GEANT3 or GEANT4,
page 14

RALF’S TRACKER A program used to find particle tracks best matching input data
from detectors., page 16

ROOT A C++ framework for data analysis, page 17

experiment_specific.hh A file used by RALF’S TRACKER where to user can add extra
conditions for rejecting events and identifying Z, page 16

Acceptance The ratio between the expectation value of observed events and the ex-
pectation value of all events, page 1

Active volume The volume of a detector able to detect energy deposits of particles.,
page 8

Clustered hit Nearby energy depositions within a detector segment are grouped in a
cluster, a clustered hit, page 16

Command line option Value given to a program at start-up, page 16

Cut A filter on a data set that yields a data subset satisfying the filter conditions.,
page 17

Daughter volume A volume located inside another volume in GEANT4, page 14

Delta-electron Freed electron due to a passing charged particle with enough energy
to ionise nearby atoms, page 8

GSI GSI Helmholtz Centre for Heavy Ion Research, a research centre in Darmstadt,
Germany, page 2

Heuristic method A method for finding good enough solutions to problems that may
lack exact solutions., page 1

Magnetic rigidity The magnetic rigidity of a particle, bρ = p/q, determines how it

will travel through a magnetic field, page 6

43
Glossary

Monte-Carlo simulations Stochastic methods for simulating processes using random

numbers, page 14

obstacle A part of the experimental setup that the particle has to pass through., page 16

Primary hit The clustered-hit with the highest energy contribution, page 18

Range-cut A threshold energy for generation of particles in GEANT4, given by how far
the generated particle is likely to travel, page 15

ROOT file A file used by ROOT, page 17

S393 An experiment with neutron rich 4 ≤ Z ≤ 10 nuclei conducted at GSI, page 2

Spread The maximum angle between the beam axis and a particles trajectory, page 12

Wrapper A function or library that simplifies calls to other more elementary functions,
page 14

44
Bibliography
[1] R. K. Bock. Acceptance — The Particle Detector Brief Book, 1998.
[Link] (accessed 26 Feb. 2013).

[2] M. J. Berger, J. H. Hubbell, S. M. Seltzer, et al. XCOM: Photon cross sections

database. National Institute of Standards and Technology, 2010.
[Link] (accessed 17 Apr. 2013).

[3] G. Knoll. Radiation detection and measurement. Wiley, 2000.

[4] G. Knoll. Radiation detection and measurement, chapter 2. Wiley, 2000.

[5] G. Knoll. Radiation detection and measurement, chapter 8. Wiley, 2000.

[6] PIANO. S393 Details, oct 2010.

[Link] (ac-
cessed 5 Mars 2013).

[7] R. Thies. Prototype tests and pilot experiments for the R3 B scintillator-based
detection systems. Master’s thesis, Chalmers University of Technology, 2011.

[8] Expert team. Status report after exp. S393/S306 (25 Aug-17 Sep 2010), sep 2010.
[Link]
pdf. (accessed 24 Apr. 2013).

[9] FRS website: S393/S306.

[Link]
(accessed 24 Apr. 2013).

[10] H. Geissel, P. Armbruster, K. Behr, et al. The GSI projectile fragment separator
(FRS): a versatile magnetic system for relativistic heavy ions. Nuclear Instruments
and Methods in Physics Research Section B: Beam Interactions with Materials and
Atoms, 70(1–4):286 – 297, 1992.
[Link]

[11] The land group. land02 — the unofficial guide, 2013.

[Link] (accessed 16 May 2013).

[12] J. Allison, K. Amako, J. Apostolakis, et al. Geant4 developments and applications.

IEEE Transactions on Nuclear Science, 53(1):270 –278, Feb. 2006.

45
Bibliography

[13] Wikipedia. Binary space partitioning — Wikipedia, The Free Encyclopedia, 2013.
[Link]
partitioning&oldid=542008108. (accessed 5 March 2013).

[14] H. Kurashige. How to Specify Physics Processes.

[Link]
ForApplicationDeveloper/html/GettingStarted/[Link].
(accessed 6 May 2013).

[15] A. Ribon. Geant4 physics: issues, progress, development. Presented at LPCC

Detector Simulation Workshop, Oct 2011.
[Link]
2&resId=0&materialId=slides&confId=144956. (accessed 4 Mars 2013).

[16] H. Johansson. ggland - command-line simulation wrapper, Apr 2013.

[Link]

[17] R. Plag. Current status of: Tracking at the LAND/R3B setup. Given as talk at
the S393/land02 November 2011 workshop, nov 2011.
[Link]
[Link].
(accessed 18 May 2013).

[18] R. Plag. Some documentation on Ralf’s tracker, apr 2013.

[Link]
(accessed 6 May 2013).

[19] The ROOT Team. ROOT | A Data Analysis Framework, 2013.

[Link] (accessed 8 May 2013).

[20] Reference Physics Lists, mar 2013.

[Link]
[Link].
(accessed 6 May 2013).

[21] J. Beringer, J. F. Arguin, R. M. Barnett, et al. Review of Particle Physics. Phys.

Rev. D, 86:010001, Jul 2012. [Link]

[22] ALADiN Collaboration, 2000.

[Link] (accessed 19
May 2013).

[23] NUSTAR Collaboration, 2013.

[Link] (accessed 19
May 2013).

46
 Bibliography

[24] The land group. land02 — the unofficial guide, 2010.

[Link]
dtf. (accessed 19 May 2013).

[25] J. Cuba, G. Stengel, A. Gtinschlofiß, et al. A large-area scintillating fibre detector

for relativistic heavy ions. Nuclear Instruments and Methods in Physics Research
A, 402:010001, Jan 1998.
[Link]

[26] The land group. land02 — the unofficial guide, 2011.

[Link]
ntf. (accessed 19 May 2013).

[27] Christine. PDC_LAND02_workshop-2, 2009.

[Link]
[Link]. (accessed 19 May 2013).

[28] The land group. land02 — the unofficial guide, 2008.

[Link]
pos. (accessed 19 May 2013).

[29] The land group. land02 — the unofficial guide, 2012.

[Link]
rolu. (accessed 19 May 2013).

[30] The land group. land02 — the unofficial guide, 2009.

[Link]
ssd. (accessed 19 May 2013).

47
A. Approximating the Magnetic
Rigidity

This section is dedicated to the calculation of the constant bρ, the magnetic rigidity, used
in the simulations to get the fragments into the fragment arm. The correlation between
the current put into ALADiN and bρ is also calculated here, for currents ranging from
2300 A to 2500 A.

A.1 Analytical Solution

To make analytical calculations feasible, the magnetic field generated by ALADiN is
taken to be enclosed in a cuboid. The magnetic field is also taken to be homogenous and
only acting in the (positive) y-direction, for simplicity. For the full geometry considered,
see Figure A.1. A good starting point is to get a grip on what it is we are looking for.
We want to be able to express the angle of deflection of a positively charged particle in
terms of the known parameters. For a full list of these, see Table A.1.
As previously stated, the magnetic field we consider is

b = bŷ (A.1)

with b being the magnetic field strength. The magnetic field is centred along the z-axis,
at z0 and the enclosing cuboid is rotated around the negative y direction by the angle
θA . To avoid having to take heed to a large number of restrictions, and the tedious work
of calculating these, we assume that the cuboid is only limited in the z direction (before

Table A.1: A list of the known parameters. They are described in the text and some of
them can also be found in Figure A.1.
Parameters
z0 φ A θA
θ φ v
m q b

48
 A.1 Analytical Solution

rotation) with a length of dz . The consideration of other fields such as gravitation is

omitted.
The considered particle emerges from the target located in the origin with a (known)
initial velocity
v = v(cos φ sin θx̂ + sin φ sin θŷ + cos θẑ) (A.2)
where θ and φ are the standard spherical angles (θ being the polar angle and φ the
azimuthal angle) whereas v is the velocity of the particle. The particle is hereafter
considered to be massive and carry a charge, as a neutral particle would be unaffected
by the magnetic field. Relativistic effects are taken into account as the speed of the
particles considered is close to the speed of light (that is v & 0.1c )
Given the geometry of the problem, it is easiest to consider it in the xz plane and only
afterward calculate the y components, as the velocity of the particles along the y-axis
will remain unchanged throughout the entire trace. We therefore make use of the angle
of emergence of the particle projected on the xz plane, ϕ, defined by

tan ϕ = cos φ tan θ . (A.3)

The first step is now to acquire the point of entry into the magnetic field. We do
this by first calculating a few geometric distances marked in Figure A.1. Namely z1 , the
distance from target to the magnetic field along the z-axis, and x0 , the distance along
the field border between the point of entry and the z-axis.
It is easily shown that
dz
z1 = z0 − (A.4)
2 cos θA
and with some trigonometry
sin ϕ
x0 = z1 . (A.5)
cos (ϕ + θA )
The point of entry, x0 , can thus be written as

x0 = x0 cos θA x̂ + y0 ŷ + (z1 + x0 sin θA )ẑ . (A.6)

We can quickly determine the y0 component by multiplying the particles travelled dis-
tance in the xz plane with a scaling factor, which has a θ and φ dependence. This factor
can be extracted from the particles angle of emergence (eq. A.2) and we yield

sin φ sin θ q
y0 = p x20 + z12 + 2x0 z1 sin θA , (A.7)
1 − sin2 θ sin2 φ

where the fraction is the scaling factor after simplifications and the square root comes
from the travelled distance.
The next step is to acquire the particles trajectory within the magnetic field. Given
that the speed remains constant, the trajectory is fully determined by a radius, ρ, and
the centre of the circular motion. As the magnitude of the force acting on the particle by
the magnetic field equals the centripetal force, we acquire both the radius of the motion

49
A Approximating the Magnetic Rigidity

x̂
ŷ

ẑ b = bŷ

z0

x0

x0
ϕ
z1 φA θA
x = (0,0,0) 0 x
x

ρ dz
r
ρ
ν

xr
xρ
Figure A.1: A schematic drawing of a charged particle travelling from the target through
a magnetic field. The magnetic field is enclosed in a cuboid, rotated about
its centre by an angle θA , the centre being located a distance z0 away from
the target along the z-axis. The particle bends off in the magnetic field by an
angle ν and with a radius of curvature ρ. The fragment arm is located along
the line with long dashes, ending with the TFW at xr . The arm is rotated
around the centre of the cuboid, making an angle, φA , with the z-axis.

as well as the direction towards the centre (from the point of entry). Keep in mind,
however, that the speed we have to consider is the speed projected on the xz plane. We
do as follows

|F | = qb|v × ŷ| = qbv| − cos θx̂ + cos φ sin θẑ| =

v2
q
1 − sin2 θ sin2 φ = m 1 − sin2 θ sin2 φ ⇒


qbv
ρ
mv
q
ρ= 1 − sin2 θ sin2 φ , (A.8)
qb
where q is the charge of the particle and m its relativistic mass. Hereafter, we require q
to be positive. We now also get the centre of motion, xρ , in the xz plane by the following
translation.

50
 A.1 Analytical Solution

F
xρ = x0 + ρ
|F |
mv mv
= (x0 cos θA − cos θ)x̂ + (z1 + x0 sin θA + cos φ sin θ)ẑ (A.9)
qb qb
= xρ x̂ + zρ ẑ ,

where we have introduced the parameters xρ and zρ to simplify further calculations.

To calculate the point of exit, x0 , we need an expression for the back-most boundary
of the magnetic field, still in the xz plane. By remembering that the magnetic field is
rotated about z0 ẑ by the angle θA , this can be achieved by applying a proper rotational
matrix to the straight line xx̂ + dz/2ẑ and then translating, where x is a parameter
along the boundary which needs to be determined. The rotational matrix, A, is written
as !
cos θA sin θA
A= . (A.10)
− sin θA cos θA
h iT
This will rotate a vector of the form x , z counter clockwise. Applying it to
the already mentioned line will rotate it about the centre of the magnetic field. The
necessary translation is thus only z0 ẑ and we have
" # " #
0x 0
x =A +
dz/2 z0
dz dz
= (x cos θA − sin θA )x̂ + (z0 + cos θA + x sin θA )ẑ (A.11)
2 2
= x0 x̂ + z 0 ẑ ,

where we again have introduced two new parameters, x0 and z 0 , for later convenience.
The next step is to decide what value x has, which then will give us the point of exit
through Equation A.11. We do this by requiring that the distance between the point of
exit and the centre of motion is the radius ρ. That is

|x0 − xρ | = ρ (A.12)

⇒ ρ2 = (x0 − xρ )2 + (z 0 − zρ )2 . (A.13)
By defining
dz dz
x0ρ = xρ + sin θA , zρ0 = zρ − z0 − cos θA (A.14)
2 2
we get

(A.13) : 0 = (x cos θA − x0ρ )2 + (x sin θA − zρ0 )2 − ρ2

= x2 − 2(x0ρ cos θA + zρ0 sin θA )x + x02 02
ρ + zρ − ρ
2
(A.15)

51
A Approximating the Magnetic Rigidity

with the solutions

q
x = x0ρ cos θA + zρ0 sin θA (+
−) (x0ρ cos θA + zρ0 sin θA )2 + ρ2 − (x02 02
ρ + zρ ) . (A.16)

The negative solution, for our case of a positive particle, corresponds to the re-entry into
the magnetic field if the particle continues on the circular path even after exiting the
field, and is therefore neglected.
The total angle of deflection, ν, from the particles initial trajectory is now calculated
by utilising the scalar product between the two vectors (x0 − xρ ) and (x0 − xρ ). That
is, the vectors stretching from xρ to the point of entry and the point of exit respectively.
We have
(x0 − xρ ) · (x0 − xρ )
cos ν =
ρ2
1 mv
(cos θx̂ − cos φ sin θẑ) · (x0 − xρ )x̂ + (z 0 − zρ )ẑ .


= 2 (A.17)
ρ qb
Using Equation A.9 and A.11, and simplifying, we acquire
1 mv  mv dz

cos ν = 2
cos θ (x − x0 ) cos θA + cos θ − sin θA
ρ qb qb 2
mv dz 1

− cos φ sin θ (x − x0 ) sin θA − cos φ sin θ + (cos θA + ) (A.18)
qb 2 cos θA
where x is taken as the positive root from Equation A.16. One quickly sees that writing
out the full expression will not bring us any joy and since everything in the expression
is known, we wont need to.
Only a few short steps remain. We can, similarly to what we did before, get the y
coordinate of x0 as a scaling factor (same as before) and the length of the curved path,
which is simply the angle of deflection, ν, multiplied by the radius of curvature. This
will then have to be added to the y coordinate of the entry point, x0 ,

sin φ sin θ mv
y 0 = y0 + p 2 2
νρ = y0 + ν sin φ sin θ , (A.19)
1 − sin θ sin φ qb
where Equation A.8 was used in simplifying the expression.
To wrap things up, we also write down an expression for the particles velocity after
it has exited the magnetic field. This is done similarly to what we did in getting Equa-
tion A.11. However, we need to take the angle to be the angle of deflection, ν, and we
end up with

v 0 = v (cos ν cos φ sin θ − sin ν cos θ)x̂ + sin φ sin θŷ

+ (cos ν cos θ + sin ν cos φ sin θ)ẑ , (A.20)

where ν can be found from Equations A.18 and A.16. Given the length of our final
expression, some approximations are in order to acquire something useful. We proceed
with this in the continuing sections.

52
 A.2 Angular Approximations

A.2 Angular Approximations

In the S393 experiment, the spread of the beam is around 0.016 rad. With this, we can
make a large number of simplifications to the expressions in the previous section by the
following approximation

θ ≈ 0 ⇒ sin θ ≈ 0, cos θ ≈ 1, tan θ ≈ 0 . (A.21)

If we apply this, rather crude, approximation to Equation A.3 through A.6 we get

tan ϕ = 0, ϕ = 0, x0 = 0 (A.22)

and
dz
x0 = z1 ẑ = z0 − ẑ . (A.23)
2 cos θA
The radius, from Equation A.8, is now simply written as
mv
ρ= (A.24)
qb

along with the centre of motion, Equation A.9,

mv dz
xρ = − x̂ + (z0 − )ẑ = xρ x̂ + zρ ẑ (A.25)
qb 2 cos θA

and the point of exit, Equation A.11,

dz dz
x0 = (x cos θA − sin θA )x̂ + (z0 + cos θA + x sin θA )ẑ = x0 x̂ + z 0 ẑ . (A.26)
2 2
We get the equivalent of Equation A.14

mv dz dz 1
x0ρ = − + sin θA , zρ0 = − ( + cos θA ) (A.27)
qb 2 2 cos θA

and can now calculate the value of x from Equation A.16 by utilising the following
simplified expression.

dz mv
x0ρ cos θA + zρ0 sin θA = −( tan θA + cos θA ) . (A.28)
2 qb

With this we get

dz mv mv mv
r
x=−( tan θA + cos θA ) (+
−) −d2z + ( cos θA )2 + 2dz sin θA , (A.29)
2 qb qb qb

where Equation A.27 and A.24 were used in getting the final expression. As before, we
only consider the positive root for our case of a particle, carrying a positive charge.

53
A Approximating the Magnetic Rigidity

It can now also be shown that

(x0 − xρ ) · (x0 − xρ )
cos ν =
ρ2
qb qb mv mv
r
2
= sin θA − dz sin θA + cos θA −d2z + ( cos θA )2 + 2dz sin θA
mv mv qb qb
s
dz dz 2 dz
= sin2 θA − sin θA + cos θA −( ) + cos2 θA + 2 sin θA , (A.30)
ρ ρ ρ

by using Equation A.18 together with the positive root from Equation A.29.
We can now acquire a final expression for the required radius ρ to get the particle to
deflect from its initial trajectory by an angle ν, in the approximation that the angular
spread from target is very small.

dz
ρ= . (A.31)
sin θA + sin ((+
−) ν − θA )
The calculations in getting this expression from Equation A.30 are elementary but te-
dious, which is why the intermediate steps have been omitted. The two solutions come
from the fact that the angle ν can be both positive and negative. If we require the angle
to be positive and that the radius be positive as well, we end up with

dz
ρ= . (A.32)
sin θA + sin (ν − θA )

A.3 Linear Correlation Between the Current and the

Magnetic Field Strength
In the previous sections we acquired an expression for the ρ part of the magnetic rigidity,
bρ. In this Section we will finish up the calculations by establishing a linear relation
between the current fed to ALADiN, I, and the magnetic field strength, b, from the
previous sections.
The magnetic field felt by a particle, fired from target along the z-axis, as it passes
through ALADiN can be seen in Figure A.2 for two different currents, 2500 A and 2300 A.
It is in this interval that we will assume a linear relationship. The reason for this comes
from the fact that field maps exist for these two currents and both the tracker and
GGLAND use the very same assumption.
The particle used in the case presented in the figure was a 15 B nucleus, with the mass
m0 = 15.00307 u and charge Ze = 5e, e being the elementary charge. The speed of
the particle, v0 , was in both cases calibrated as to give it an exit angle equal to φA ,
set to 15 ◦ . This was easily done by simply having it centre on the TFW, located at
xr in Figure A.1. The radial distance between the centre of ALADiN and the centre of
the TFW, r, was taken to be large enough as to make the angular discrepancy ν − φA
negligible.

54
 A.3 Linear Correlation Between the Current and the Magnetic Field Strength

The first step will be to correlate the current to some representative values of the
magnetic field in ALADiN. We take these to be the maximum values of the magnetic
field in the y direction experienced by the particle, shown in Figure A.2 as b2500 and
b2300 for the respective currents. That is, we assume the entire magnetic field to scale
the same way as these maximum values for varying currents. Thus, we get the relation
b2500 − b2300
bALADiN (I) − b2500 = ( ) · (I − 2500) ⇒
2500 − 2300
b2500 − b2300
bALADiN (I) = ( ) · (I − 2500) + b2500 . (A.33)
200
where bALADiN (I) signifies the discussed maximum value at a given current. The param-
eters b2500 and b2300 are found to be
b2500 = 1.6593 T and b2300 = 1.5828 T (A.34)
from the calibration runs shown in Figure A.2.
We now want to relate the magnetic field strength from ALADiN to the homogenous
field strength, b from the previous sections. We do this by assuming that they are
proportional with respect to each other. In doing this, we only need to calculate b from
the known values for one of the calibration currents, using Equations A.32 and A.24.
We end up with
m0 v0 sin θA + sin (ν − θA )
bhom = (A.35)
Ze dz
where, apart from the aforementioned values, we use values from the 2500 A calibration.
These are
v0 = 0.6855c , ν ≈ φA = 15 ◦ , θA = 7 ◦ , and dz = 2.42 m , (A.36)
where c is the speed of light and dz has been taken to be the full length of ALADiN.
Using these values we acquire
bhom = 0.9487 T. (A.37)
Given that we used the values from the 2500 A calibration, we now get the relation
between the homogenous magnetic field and the current
bhom bhom  b2500 − b2300 
b(I) = bALADiN (I) = ( ) · (I − 2500) + b2500
b2500 b2500 200
 b2300 I − 2500 
= (1 − ) + 1 bhom . (A.38)
b2500 200
We finally arrive at our final expression for the magnetic rigidity bρ
dz  b2300 I − 2500 
bρ = · (1 − ) + 1 bhom . (A.39)
sin θA + sin (ν − θA ) b2500 200
By putting in values for the various constants in the equation we yield
5.2925 · 10−4 AT
I + 0.9728 T
bρ = . (A.40)
sin θA + sin (ν − θA )

55
A Approximating the Magnetic Rigidity

A.4 The Magnetic Field of ALADiN

Given that the magnetic field is a vector field, there is no obvious way to present the
magnetic field of ALADiN in the three spatial dimensions. However, what is of primary
importance is the magnetic field felt by a particle travelling through ALADiN, which is
what is presented in this section.
The magnetic field in the x-, y- and z-directions are presented in Figure A.2, as felt
by a 15 B nucleus that travels through ALADiN, for the two currents, 2500 A and 2300 A.
The nucleus initially traces the z-axis, where the axes are equivalent to those defined
in Figure 3.2, and ALADiN is rotated 7 ◦ . The initial speed of the nucleus is such that
it hits the centre of the TFW located at the end of the fragment arm, which makes an
angle of 15 ◦ with the z-axis.
It is, from the figures, possible to extract that the experienced field is not homogenous
and that ALADiN actually generates quite notable fringe fields. Another important
result is that the field in the x-direction is not negligible, whereas the field in the z-
direction is several orders of magnitude smaller than that of the other two directions.
However, as the integral over the magnetic field in the x-direction gets rather small,
due to its change of sign and especially compared to the integral over the field in the
y direction, it will not affect the particle’s direction by much (again compared to that
caused by the field in the y-direction).
Two equivalent simulations were made, but with 14 B nuclei with a magnetic rigidity
of 6.56 Tm and 13.24 Tm respectively. These values were chosen so as to make the nuclei
barely miss one detector while travelling through the others in the setup and it turns
out that the detector it misses first is GFI2. This is, in a way, seen as the two extremes
of when the particle traces a much desirable path, one that includes all detectors in the
fragment arm.
The, by the particles, experienced magnetic fields in these two simulations are similar
to the field shown in Figure A.2. In fact, they are so similar that they can not be told
apart when drawn in the same figure. In light of this, the difference between the mag-
netic fields over travelled distance for the two simulations are presented in Figure A.3.
Comparing these with Figure A.2 it is discovered that the experienced magnetic field
hardly changes at all, as the difference is several orders of magnitude smaller than the
corresponding fields.

56
 A.4 The Magnetic Field of ALADiN

Figure A.2: Magnetic field strength experienced by a 15 B nucleus travelling through AL-
ADiN, for two different currents. The nucleus initially traces the z-axis.
ALADiN is rotated 7 ◦ and the initial speed of the nucleus is such that it hits
the centre of the TFW, located at a radial distance of 9 m from ALADiN’s
centre. The fragment arm makes an angle of 15 ◦ with the z-axis. ...

b2500 bx (2500 A)
1.6 b2300
by (2500 A)
1.4 bx (2300 A)
by (2300 A)
1.2
Magnetic field strength (T)

0.8

0.6

0.4

0.2

0 50 100 150 200 250 300 350 400 450 500 550
Travelled distance (cm)
(a) ... The axes are equivalent to those defined in Figure 3.2.
0.5
0.1 bx (2500 A) bz (2500 A)
bz (2500 A) 0.4 bz (2300 A)
0.08 bx (2300 A)
bz (2300 A) 0.3
0.06
Magnetic field strength (mT)
Magnetic field strength (T)

0.04 0.2

0.02 0.1

0 0

-0.02 -0.1

-0.04
-0.2
-0.06
-0.3
-0.08
-0.4
0 50 100 150 200 250 300 350 400 450 500 550 0 50 100 150 200 250 300 350 400 450 500 550
Travelled distance (cm) Travelled distance (cm)

(b) ... The axes are equivalent to those defined in (c) ... The axes are equivalent to those defined
Figure 3.2 and it is a zoomed version of Fig- in Figure 3.2 and it is a zoomed version of Fig-
ure A.3a but now looking at the magnetic field ure A.3b but only looking at the magnetic field
in the x and z directions. in the z direction.

57
A Approximating the Magnetic Rigidity

Figure A.3: Difference between the magnetic field strengths experienced by two 14 B nuclei
travelling through ALADiN, with a current of 2500 A. The geometry is the
same as that specified in Figure 3.2. The magnetic field, b, felt by the particle
with a bρ of 6.56 Tm and 13.24 Tm is given the index 1 and 2 respectively.
The figure presents ...

400

300

200

100
by,1 − by,2 (µT)

-100

-200

-300

-400
0 50 100 150 200 250 300 350 400 450 500 550
Travelled distance (cm)
(a) ...the difference in the y component of the magnetic field.
80 12

60 10

40
8

20
bx,1 − bx,2 (µT)

bz,1 − bz,2 (µT)

6
0
4
-20

2
-40

-60 0

-80 -2
0 50 100 150 200 250 300 350 400 450 500 550 0 50 100 150 200 250 300 350 400 450 500 550
Travelled distance (cm) Travelled distance (cm)

(b) ...the difference in the x component of the mag- (c) ...the difference in the z component of the mag-
netic field. netic field.

58
B. An Example Detector: POS

This appendix shows how to implement a simple but not trivial detector in GGLAND.
The example given is the POS detector, the first detector we implemented. The source
file for the detector should be saved as geom_pos.cc. Also, the geom_pos.o dependency
needs to be added to the GGLAND makefile.
#define __STDC_FORMAT_MACROS
#include <i n t t y p e s . h>

#include " gg_geom_help . hh "

#include " d e t e c t o r _ h i t . hh "

#include " gg_ tree . hh " b2500 bx ( 2 5 0 0 A)

1 . 6 b2300
by ( 2 5 0 0 A)

#include " gg_step . hh "

The detector parameters are declared inside a struct.

struct spec_POS_t
{
SPEC_FLOAT( _dx , 2 . 5 , "cm" , " f u l l ␣ width ␣x␣ o f ␣ a c t i v e ␣ volume " )
;
SPEC_FLOAT( _dy , 2 . 5 , "cm" , " f u l l ␣ width ␣y␣ o f ␣ a c t i v e ␣ volume " )
;
SPEC_FLOAT( _dz , 0 . 0 3 , "cm" , " f u l l ␣ width ␣ z ␣ o f ␣ a c t i v e ␣ volume " ) ;
SPEC_FLOAT( _ l g h e i g h t , 2 , "cm" , " l i g h t ␣ g u i d e ␣ h e i g h t ␣ o v e r ␣ a c t i v e ␣
volume " ) ;
SPEC_FLOAT( _lgheadd , 0 . 7 0 , "cm" , " s i z e ␣ o f ␣ s q u a r e −shaped ␣ l i g h t ␣ guide
−heads " ) ;
SPEC_MEDIA( _type , " plastic " ," " , "POS␣ a c t i v e ␣ volume ␣ m a t e r i a l . " ) ;
SPEC_MEDIA( _lgtype , " p l a s t i c " , " " , " Light ␣ guide ␣ material " ) ;
};

#include " auto_gen / s p ec _ in f o _p o s . hh "

#define UNUSED_PARAM( x )

This function is called when GGLAND creates the detector.

gg_geom_obj ∗make_POS( void ∗ vspec , u in t 3 2 _ t UNUSED_PARAM( mask_set )
, const t r a n s f o r m _ m a t r i x ∗ l o c _ r o t

59
B An Example Detector: POS

, det_name_no_info ∗name_no )
{

The vspec points to a sepc_POS_t object allocated by GGLAND.

spec_POS_t ∗ s p e c = ( spec_POS_t ∗ ) v s p e c ;

In order to place our volumes, we need to create a volume “mother volume” to contain
daughter volumes. The d_full class is used to tell GGLAND that we are dealing with
“diameters” rather than “radii”.
gg_geom_obj ∗goPOS=makebox ( "POS" , _world_media_type
, d _ f u l l ( spec −>_dx+2∗ spec −>_ l g h e i g h t )
, d _ f u l l ( spec −>_dy+2∗ spec −>_ l g h e i g h t )
, d _ f u l l (__max( spec −>_lgheadd , spec −>_dz ) ) ) ;

The POS detector has a cuboid in the middle that acts as active material. Therefore we
create a box containing the active material.
gg_geom_obj ∗ g o A c t i v e=makebox ( "POS" , spec −>_media_type , d _ f u l l ( spec −>
_dx)
, d _ f u l l ( spec −>_dy) , d _ f u l l ( spec −>_dz ) ) ;

Around the cuboid, there are four trapezoidal light guides that transmit light emitted
from the active volume to photo multipliers. Since volumes can be rotated, it is enough
to create two volumes.
gg_geom_obj ∗ goLgHorz = maketrd2 ( "TRD1" , spec −>_media_lgtype , d _ f u l l (
spec −>_dz )
, d _ f u l l ( spec −>_lgheadd ) , d _ f u l l ( spec
−>_dy)
, d _ f u l l ( spec −>_lgheadd )
, d _ f u l l ( spec −>_ l g h e i g h t ) ) ;

gg_geom_obj ∗ goLgVert = maketrd2 ( "TRD2" , spec −>_media_lgtype , d _ f u l l (

spec −>_dz )
, d _ f u l l ( spec −>_lgheadd ) , d _ f u l l ( spec −>
_dx)
, d _ f u l l ( spec −>_lgheadd )
, d _ f u l l ( spec −>_ l g h e i g h t ) ) ;

Now place and rotate all volumes into the mother volume. Note that the rotation
matrices are owned by GGLAND.
p o s v o l ( goActive , 4 , goPOS , 0 . , 0 . , 0 . ) ;

gg_rot_obj ∗ l g 0 r o t=makezyzrotm (M_PI/ 2 , 0 . , 0 . ) ;

p o s v o l ( goLgHorz , 0 , goPOS , 1 ∗ spec −>_dx/2 +1∗ spec −>_ l g h e i g h t / 2 , 0 . , 0 . ,
lg0rot ) ;

gg_rot_obj ∗ l g 1 r o t=makezyzrotm(−M_PI/ 2 , 0 . , 0 . ) ;
p o s v o l ( goLgHorz , 1 , goPOS, −1∗ spec −>_dx/2 −1∗ spec −>_ l g h e i g h t / 2 , 0 . , 0 . ,
lg1rot ) ;

gg_rot_obj ∗ l g 2 r o t=makezyzrotm (M_PI/2 , −M_PI/ 2 , 0 . ) ;

60
 p o s v o l ( goLgVert , 2 , goPOS , 0 . , − 1 ∗ spec −>_dy/2 −1∗ spec −>_ l g h e i g h t / 2 , 0 . ,
lg2rot ) ;

gg_rot_obj ∗ l g 3 r o t=makezyzrotm(−M_PI/2 , −M_PI/ 2 , 0 . ) ;

p o s v o l ( goLgVert , 3 , goPOS , 0 . , 1 ∗ spec −>_dy/2 +1∗ spec −>_ l g h e i g h t / 2 , 0 . ,
lg3rot ) ;

Place the mother volume into the world. The argument name_no contains a unique
identifier identifying this detector.
place_in_world ( goPOS , l o c _ r o t , name_no ) ;

Finally, make GGLAND collect events detected in this detector, and return its mother
volume.
c o l l e c t _ t r e e _ v o l _ i n f o ∗ c o l l e c t _ i n f o=r e g i s t e r _ c o l l e c t _ t r e e _ i d ( name_no
,1) ;
collect_tree_reduced_vars ( c o l l e c t _ i n f o
,COL_TREE_VAR_T | COL_TREE_VAR_E | COL_TREE_VAR_X |
COL_TREE_VAR_Y) ;

vect_step_copyno_path path ;

r e g i s t e r _ s t e p ( goActive , NULL, c o l l e c t _ i n f o ,& path ) ;

return goPOS ;
}

61
C. Obstacles Defined in GGLAND

In this Appendix, obstacles are presented as they are defined in GGLAND along with
a list of variables with default values. The obstacles presented are those implemented
as a part of this project, hence the exclusion of LAND and XB. The illustrations are
constructed to show all features of the geometry of an obstacle, rather than being scale
accurate.
The obstacles in GGLAND are only approximations of the actual obstacles using simple
geometric shapes. Most of them lack geometry for read-out electronics and other non-
active parts. Obstacles that have such features are ALADiN, which only consists of
non-active material, and POS, whose light guides are implemented.
Given that not all specifications are well documented, some data has been taken from
personal communications with Ronja Thies. Information given by her has been marked
with a † . Other information — marked with a ‡ —was acquired through a note from
Håkan Johansson, who got the information from Michael Heil via Ralf Plag. Some
measurements have been improvised, these have been marked with a *.

C.1 ALADiN — A LArge DIpole magNet

ALADiN is a large dipole magnet with a core made of iron, surrounding a tube for
particles to pass through. The tube can be filled with gas other than air to make
interaction less likely. In reality, the tube is a trapezoid. In the model, the trapezoid is
approximated by a cuboid with a width taken as the average width of the trapezoid.
Measurements for ALADiN were obtained by combining pixel counts from two pic-
tures. For the top view, a technical drawing with a resolution of 59 px/m was used [22].
For the front view, a photograph with ALADiN in the background was used [23]. To
get something meaningful from the picture, it was deformed to compensate for the per-
spective view. The width and depth of ALADiN were then measured and the height
was calculated to preserve the aspect ratio for the perspective corrected image. From
this procedure, it follows that the values of the width and depth of ALADiN are more
accurate than those of the height.

62
 C.1 ALADiN — A LArge DIpole magNet

z g wch g

tp tp
x

tw

dco dch dco

tw

wco wco
tp tp

y
hcob

111111
000000 x
000000
111111
000000
111111
hch 000000
111111 hco
000000
111111

hcob

Key Parameter Default value

wco core_width 93 cm a
dco core_depth 176 cm a
hco core_height 266 cm b
hcob core_height_bar 50 cm b
tp core_thickness_plate 3 cm a
 core_type Fe *
wch chamber_width 138 cm a
dch chamber_depth 242 cm a
hch chamber_height 81 cm b
g chamber_gap 46 cm a
 chamber_type He *
tw chamber_window_thickness 0.0005 cm *
 chamber_window_type plastic *

a
This measurement was from reference [22].
b
This measurement was from reference [22] and [23].

63
C Obstacles Defined in GGLAND

C.2 DTF — Dicke ToFWand

DTF is a row of vertical scintillator paddles inside a thin wrapping. Data without note
are taken from land02 — the unofficial guide [24].
0
1
1
0
0
1 w
dw 0
1
0
1
0
1 0
1 N0
d0w 0
1
0
1 0
1
0
1 000000000000
111111111111
0
1
000000000000
111111111111
0
1 0
1
000000000000
111111111111
N 111 1111
000 0000
000
111 0
1
11111
00000
0000
1111 0
1
1111
0000 0000
1111 000
111
1111
0000
000000000000
111111111111
0
1
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
0000
1111 000
111
000000000000
111111111111
000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
s
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
0000
1111 000
111
000
111
000
111 0000
1111 000000000000
111111111111
0
1 0000
1111 000
111
000000000000
111111111111
000
111 0000
1111 0
1 0000
1111 000
111
000000000000
111111111111
l 0
000
111 0000
1111 0
1
0
1 0000
1111 000
111
000000000000
111111111111
000
111 0000
1111 0000
1111 000
111
000000000000
111111111111
0
1
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
0000
1111 000
111
000
111
000
111 0000
1111 1
0 0000
1111 000
111
000
111 0000
1111 0 1111
0000 000
111
000
111 0000
1111 s 1111
0000 000
111
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
0000
1111 000
111
000
111
000
111 0000
1111 0
1 0000
1111 000
111
000000000000
111111111111
000
111
1111
0000 1111
0000
1111
00000
1 0000
1111 000
111
1111
0000
000000000000
111111111111
w 000
111 0000
1111 0
1 0000
1111 000
111
000000000000
111111111111
0
1
000000000000
111111111111
d0
x x
d S l

z y

Key Parameter Default value

w0 paddle_width1 20 cm
d0 paddle_depth1 1.5 cm
l0 paddle_length1 120 cm
s0 paddle_spacing1 0.3 cm a
d0w paddle_wrapping1 0.3 cm *
N0 paddles1 6
 type1 plastic *
 wrapping_type1 plastic *
w paddle_width2 10.4 cm
d paddle_depth2 0.5 cm
l paddle_length2 140 cm
s paddle_spacing2 43 cm a
dw paddle_wrapping2 0.2 cm *
N paddles2 3
 type2 plastic *
 wrapping_type2 plastic *
S plane_spacing 1 cm *

a
This measurement approximated from pictures at land02 — the unofficial guide [22] using similar
techniques as mentioned in Section C.1.

64
 C.3 GFI — Großer FIberdetektor

C.3 GFI — Großer FIberdetektor

GFI is a row of vertical scintillator fibres inside a thin wrapping. Data without note are
taken from Cuba [Link]. [25].
ds /2
z

d
x
ds

111
000 111
000 1111
0000 1111
0000
000
111 000
111 0000
1111 0000
1111 y
000
111
000
111 000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111
000
111
000
111 000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111
000
111 000
111 0000
1111 0000
1111
000
111 000
111 0000
1111 0000
1111
000
111 000
111 0000
1111 0000
1111 x
000
111 000
111 0000
1111 0000
1111
000
111 000
111 0000
1111 0000
1111 h
000
111 000
111 0000
1111 0000
1111
000
111 000
111 0000
1111 0000
1111
000
111
000
111 000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111
000
111 000
111 0000
1111 0000
1111
000
111 000
111 0000
1111 0000
1111
000
111
000
111 000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111
000
111
000
111 000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111
000
111 000
111 0000
1111 0000
1111

w ds N

Key Parameter Default value

w fiber_width 0.1 cm
d fiber_depth 0.1 cm
h fiber_height 50 cm
ds fiber_spacing 0.005 cm †
N fibers 475
 type plastic †
 wrapping_type plastic †

65
C Obstacles Defined in GGLAND

C.4 NTF — New Time-of-Flight wall

NTF is a row of vertical scintillator paddles inside a thin wrapping. Data without note
are taken from land02 — the unofficial guide [26].

1
0
0
1
0
1 w
0
1
dw 0
1
0
1 0
1 N0
d0w 0
1
0
1 0
1
0
1 000000000000
111111111111
0
1
000000000000
111111111111
0
1 0
1
000000000000
111111111111
N 111 1111
000 0000
000
111 01111
1
11111
00000
0000 0
1
1111
0000 0000
1111
000000000000
111111111111
0
1 000
111
1111
0000
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
000000000000
111111111111
0000
1111 000
111
000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 1
0 0000
1111 000
111
000
111 0000
1111 0000
1111 000
111
s 000
111 0000
1111 0000
1111 000
111
000
111 0000
1111 0000
1111 000
111
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
000
111 0000
1111 0
1 0000
1111
000000000000
111111111111000
111
000
111 0000
1111 0
1 0000
1111
000000000000
111111111111
0
1 000
111
000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
l 0
000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 1
0 0000
1111 000
111
000
111 0000
1111 0000
1111 000
111
000
111 0000
1111 0 1111
0000 000
111
000
111 0000
1111 s 0000
1111 000
111
000
111 0000
1111 0000
1111 000
111
000
111 0000
1111 0000
1111 000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
0000
1111 000
111
000
111
000
111
1111
0000 1111
0000
1111
0000000000000000
111111111111
0
1 0000
1111 000
111
1111
0000
000
111 0000
1111 000000000000
111111111111
0
1 0000
1111 000
111
w 000000000000
111111111111
0
1
000000000000
111111111111
d0
x x
d S l

z y

Key Parameter Default value

w0 paddle_width1 6 cm
d0 paddle_depth1 0.5 cm
l0 paddle_length1 48 cm
s0 paddle_spacing1 0.2 cm *
d0w paddle_wrapping1 0.05 cm *
N0 paddles1 8
 type1 plastic †
 wrapping_type1 plastic *
w paddle_width2 6 cm
d paddle_depth2 0.5 cm
l paddle_length2 48 cm
s paddle_spacing2 0.2 cm *
dw paddle_wrapping2 0.05 cm *
N paddles2 8 
 type2 plastic †
 wrapping_type2 plastic *
S plane_spacing 1 cm *

66
 C.5 PDC — Proton Drift Chamber

C.5 PDC — Proton Drift Chamber

A PDC is a collection of charged long wires in a hexagonal pattern inside a gas chamber.
The thicker wires are positively charged to attract released electrons from ionised atoms
in the gas. The field near the wires is strong enough to accelerate the electrons sufficiently
to ionise other atoms, causing detectable current pulses in the thicker, positively charged
wires.
In reality, the active gas volume is surrounded by dead material to contain the gas
and regulate the pressure. There are also two sets of wires, turned about 90◦ around the
beam axis relative to each other in order to give the x and y coordinates of a particle
passing through.
The square grid in the illustration represents the repeated building blocks of the
detector. Data without note are taken from PDC_LAND02_workshop-2 [27].

h
d2 r

R
N
d1

n z z

x y

Key Parameter Default value

d1 d1 1.6 cm
d2 d2 1.6 cm
h cell_height 80 cm
n n_cells_plane 144
N n_planes 2
r r_w1 0.01 cm
 type_w1 Cu *
R r_w2 0.02 cm
 type_w2 Cu *
 type_cell Ar

67
C Obstacles Defined in GGLAND

C.6 POS — POSition detector

The POS detector is a thin sheet of scintillator material with a light guide at each end
orthogonal to the beam axis. Data without note is taken from land02 — the unofficial
guide [28].

11
00
000000
111111 00
110
1
00
11
000000
111111 00
110
1
00
11
000000
111111 00
110
1
00
11
000000
111111
00
11 00
11
001
0
000000
111111
00
11
000000
111111
110
1
0000001
00
110
111111 0
1
000000
111111
000000
111111 0
1
0000001 0
h dlg
111111 0
1
000000
111111
00
11
000000
111111 00
110
1
0
1
00
11
000000
111111
00
11 00
11
00
110
1
000000
111111
00
11 00
110
1
000000
111111
00
11 00
110
1
000000
111111
00
11 00
110
1
000000
111111
00
11 d lg
001
110
wlg

w hlg d
y y

x z

Key Parameter Default value

w dx 5.5 cm ‡
h dy 5.5 cm ‡
d dz 0.03 cm †
dlg lgheight 4 cm *
hlg lgheadd 1.4 cm *
 type plastic
 lgtype plastic *

68
 C.7 ROLU — Rechts, Oben, Links, Unten

C.7 ROLU — Rechts, Oben, Links, Unten

ROLU are two pairs of scintillator plates that can be moved independently to define a
gap which the beam has to pass through. The measurements are taken from land02 —
the unofficial guide [29].

wv

yo hv

ho
s
hh

wh dh

xo
wo dv
y y

x z

Key Parameter Default value

wv dx_v 10 cm
hv dy_v 9.5 cm
dv dz_v 0.5 cm
wh dx_h 9.5 cm
hh dy_h 10 cm
dh dz_h 0.5 cm
xo offset_x 0 cm
yo offset_y 0 cm
s space 1 cm
ho opening_v 2 cm
wo opening_h 2 cm
 type plastic

69
C Obstacles Defined in GGLAND

C.8 SST — Silicon STrip detectors

SST is a thin slab of silicone divided into many sections by the segmentation of the read-
out electrodes. This segmentation is represented by the square grid in the illustration.
The material surrounding the grid on the right hand side of the illustration is a dead-
layer. The measurements are taken from land02 — the unofficial guide [30].

w d

wc
Ny

hc Nx

ddl

y y

x z

Key Parameter Default value

w width 7.2 cm
h height 4.0 cm
d thickness 0.03 cm
ddl deadlayer 0.001 cm *
wc pitchx 0.0104 cm
hc pitchy 0.00275 cm
Ny stripsx 384
Nx stripsy 640
 type Si

70
 C.9 TFW — Time of Flight Wall

C.9 TFW — Time of Flight Wall

The TFW consists of 2 planes of scintillator paddles. One plane has horizontal paddles
while the other has vertical, giving the x and y coordinates of the detected particles.

1
0
0
1
0
1 w
0
1
dw 0
1
0
1 0
1 N0
d0w 0
1
0
1 0
1
0
1 000000000000
111111111111
0
1
000000000000
111111111111
0
1 0
1
000000000000
111111111111
N 111 1111
000 0000
000
111 01111
1
11111
00000
0000 0
1
1111
0000 0000
1111
000000000000
111111111111
0
1 000
111
1111
0000
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
000000000000
111111111111
0000
1111 000
111
000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 1
0 0000
1111 000
111
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
s
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
000
111
000
111 0000
1111
0000
1111 0
1 0000
1111
0000
1111 000
111
000
111
000
111 0000
1111 000000000000
111111111111
0
1 0000
1111
000000000000
111111111111000
111
000
111 0000
1111 0
1 0000
1111
000000000000
111111111111
0
1 000
111
l 0
000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 1
0 0000
1111 000
111
000
111 0000
1111 0000
1111 000
111
000
111 0000
1111 0 0000
1111 000
111
000
111 0000
1111 s 0000
1111 000
111
000
111
000
111 0000
1111
0000
1111 0000
1111
0000
1111 000
111
000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 0
1 0000
1111 000
111
000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
000
111
1111
0000 0000
1111
1111
0000 0000
1111
000000000000
111111111111
0
1 000
111
1111
0000
w 000
111 0000
1111 0000
1111
000000000000
111111111111
0
1 000
111
000000000000
111111111111
d0
x x
d S l

z y

Key Parameter Default value

w0 paddle_width1 10 cm †
d0 paddle_depth1 0.5 cm †
l0 paddle_length1 147 cm †
s0 paddle_spacing1 0.2 cm †
d0w paddle_wrapping1 0.001 cm *
N0 paddles1 18 †
 type1 plastic *
 wrapping_type1 plastic *
w paddle_width2 10 cm †
d paddle_depth2 0.5 cm †
l paddle_length2 189 cm †
s paddle_spacing2 0.2 cm †
dw paddle_wrapping2 0.001 cm *
N paddles2 14 †
 type2 plastic *
 wrapping_type2 plastic *
S plane_spacing 1 cm *

71
D. Function Used to Calibrate
RALF’S TRACKER

Below is a C++ function that can be used to run our calibration scheme found in
Section 5.3.2, where all steps listed below are described in common language. output is a
handle to a log file. The ProgramData object contains information used to communicate
with the tracker as well as, nuclei masses, and methods for calling the tracker. The
string_format function is a sprintf variant for C++ strings. ResourceManager is a
class that helps to clean up resources.
bool c a l i b r a t e ( FILE∗ output , ProgramData& data , unsigned i n t Z , unsigned
i n t A)
{

D.1 Preparation

s t d : : s t r i n g f i l e n a m e _ i n = . . . ; // your f i l e t h a t c o n t a i n s d a t a t o
calibrate for
s t d : : s t r i n g f i l e n a m e _ o u t = . . . ; // temporary o u t p u t f i l e

double mass=data . masses−>massGet ( I s o t o p e ( Z ,A) ) / C o n s t a n t s : : mev ;

double brho = 8 . 7 ; //Tm
double b e t a=betaGet ( Z , mass ∗ C o n s t a n t s : : u , brho ) ;
data . experiment −>variableCustomGet ( " I n b e t a " ) . x=b e t a ;
data . experiment −>variableCustomGet ( " Truebeta " ) . x=b e t a ;
data . experiment −>variableCustomGet ( " T r u e b e t a _ v a l i d " ) . x=1;
// Save c u r r e n t c o n f i g u r a t i o n b e f o r e t o u t c h i n g i t .
{
std : : s t r i n g experiment_file_in = . . . ;
, C o n s t a n t s : : e l e m e n t s [ Z ] , data . experiment −>v a r i a b l e G e t ( " I " ) . v a l u e )
;
data . experiment −>s a v e ( e x p e r i m e n t _ f i l e _ i n . c _ s t r ( ) ) ;
}

D.2 Step 1

72
 D.3 Step 2

s t d : : s t r i n g o p t i o n s=s t r i n g _ f o r m a t ( "−−t r a c k −f o r w a r d ␣−−output=%s ␣−−

i g n o r e −t o f ␣%s ␣>␣ / dev / n u l l ␣2>&1"
, filename_out . c_str ( ) , filename_in . c_str ( ) ) ;

double m=1;
f o r ( unsigned i n t k=0;k<4 && m>f a b s ( 1 e −5) ; ++k )
{
data . t r a c k e r −>commandBuild ( ∗ data . experiment , o p t i o n s . c _ s t r ( ) ) ;
f p r i n t f ( output , " S t a r t i n g ␣ t r a c k e r \n{\n%s \n}\n " , data . t r a c k e r −>
commandGet ( ) ) ;
i f ( system ( data . t r a c k e r −>commandGet ( ) ) )
{
return 0 ;
}
data . t r a c k e r −>c l e a r ( ) ;
m=getMean ( f i l e n a m e _ o u t . c _ s t r ( ) , " h509 " , " f r e s _ f t f t " ) ;
f p r i n t f ( output , "<m␣ %.15g>\n " ,m) ;
double o f f s e t=a t o f ( data . experiment −>v a r i a b l e G e t ( "OFFSET_FTFT" ) .
value ) ;
o f f s e t+=m;
s p r i n t f ( data . experiment −>v a r i a b l e G e t ( "OFFSET_FTFT" ) . v a l u e , " %.15 g "
, offset ) ;
remove ( f i l e n a m e _ o u t . c _ s t r ( ) ) ;
}

i f (m>f a b s ( 1 e −5) )
{ return 0 ; }
}

D.3 Step 2
{
s t d : : s t r i n g o p t i o n s=s t r i n g _ f o r m a t ( "−−output=%s ␣−−t r a c k −backward ␣%s ␣
>␣ / dev / n u l l ␣2>&1"
, filename_out . c_str ( ) , filename_in . c_str ( ) ) ;

double m=1;
f o r ( unsigned i n t k=0;k<4 && m>f a b s ( 1 e −5) ; ++k )
{
if (k)
{ remove ( f i l e n a m e _ o u t . c _ s t r ( ) ) ; }
data . t r a c k e r −>commandBuild ( ∗ data . experiment , o p t i o n s . c _ s t r ( ) ) ;
f p r i n t f ( output , " S t a r t i n g ␣ t r a c k e r \n{\n%s \n}\n " , data . t r a c k e r −>
commandGet ( ) ) ;
i f ( system ( data . t r a c k e r −>commandGet ( ) ) )
{
return 0 ;
}
data . t r a c k e r −>c l e a r ( ) ;

//By d e f i n i t i o n f r e s _ g f i 1 x and f r e s _ g f i 2 x a r e 0 when r u n n i n g t h e

tracker
// w i t h −−t r a c k −backward . E x t r a c t them f o r t h e s a k e o f c o n s i s t e n s y

73
D Function Used to Calibrate RALF’S TRACKER

double m_1=getMean ( f i l e n a m e _ o u t . c _ s t r ( ) , " h509 " , " f r e s _ g f i 1 x " ) ;

double m_2=getMean ( f i l e n a m e _ o u t . c _ s t r ( ) , " h509 " , " f r e s _ g f i 2 x " ) ;
double m_3=getMean ( f i l e n a m e _ o u t . c _ s t r ( ) , " h509 " , " f r e s _ f t f x " ) ;
data . experiment −>o b j e c t G e t ( " GFI1 " ) . ofs_x+=m_1;
data . experiment −>o b j e c t G e t ( " GFI2 " ) . ofs_x+=m_2;
data . experiment −>o b j e c t G e t ( "FTF" ) . ofs_x+=m_3;
m=s t d : : max( s t d : : max( f a b s (m_1) , f a b s (m_2) ) , f a b s (m_3) ) ;
f p r i n t f ( output , "<m␣ %.15g>\n " ,m) ;
}

i f (m>f a b s ( 1 e −5) )
{ return 0 ; }
}

D.4 Step 3

{
double m_have=f i n d P e a k ( f i l e n a m e _ o u t . c _ s t r ( ) , " h509 " , " fra_A " ) ;
f p r i n t f ( output , "<m_have␣ %.15g>\n " , m_have ) ;
double beta_have=getMean ( f i l e n a m e _ o u t . c _ s t r ( ) , " h509 " , " f r a _ b e t a " ) ;
f p r i n t f ( output , "<b e t a ␣ %.15g>\n " , b e t a ) ;
remove ( f i l e n a m e _ o u t . c _ s t r ( ) ) ;

s t d : : s t r i n g o p t i o n s=s t r i n g _ f o r m a t ( "−−have=%.15g,%u , % . 1 5 g ␣ "

"−−want=%.15g,%u , % . 1 5 g ␣ "
"−−output=%s ␣ "
">/tmp/ t r a c k e r −c a l i b −r e s u l t −%p . t x t ␣2>/dev / n u l l "
, m_have , Z , b e t a
, mass , Z , b e t a
, filename_out . c_str ( )
, data . e x p e r i m e n t ) ;

data . t r a c k e r −>commandBuild ( ∗ data . experiment , o p t i o n s . c _ s t r ( ) ) ;

f p r i n t f ( output , " S t a r t i n g ␣ t r a c k e r \n{\n%s \n}\n " , data . t r a c k e r −>
commandGet ( ) ) ;
system ( data . t r a c k e r −>commandGet ( ) ) ;

s t d : : s t r i n g cmdbuff=s t r i n g _ f o r m a t ( " / b i n / bash ␣−c ␣ \ " g r e p ␣ b e t a \\ ␣ a t ␣−

A3␣ "
" /tmp/ t r a c k e r −c a l i b −r e s u l t −%p . t x t \ " ␣ | ␣ g r e p ␣OFFSET" , data .
experiment ) ;

FILE∗ n e w _ o f f s e t s=popen ( cmdbuff . c _ s t r ( ) , " r " ) ;

i f ( n e w _ o f f s e t s==NULL)
{
return 0 ;
}
ResourceManager<FILE ∗ , i n t ( ∗ ) ( FILE ∗ )> rm( n e w _ o f f s e t s , p c l o s e ) ;
E x p e r i m e n t A b s t r a c t ∗ e_ c lo ne=data . experiment −>c l o n e ( ) ;
e_clone−>l o a d ( n e w _ o f f s e t s ) ;
data . experiment −>o b j e c t G e t ( " GFI1 " ) . ofs_x=e_clone−>o b j e c t G e t ( " GFI1 " )
. ofs_x ;

74
 D.4 Step 3

data . experiment −>o b j e c t G e t ( " GFI2 " ) . ofs_x=e_clone−>o b j e c t G e t ( " GFI2 " )
. ofs_x ;
data . experiment −>o b j e c t G e t ( "FTF" ) . ofs_x=e_clone−>o b j e c t G e t ( "FTF" ) .
ofs_x ;
e_clone−>c l o n e D e s t r o y ( ) ;
}
s t d : : s t r i n g e x p e r i m e n t _ f i l e _ o u t = . . . ; // Your f i l e f o r t h e c a l i b r a t e d
tracker
, C o n s t a n t s : : e l e m e n t s [ Z ] , data . experiment −>v a r i a b l e G e t ( " I " ) . v a l u e ) ;
data . experiment −>s a v e ( e x p e r i m e n t _ f i l e _ o u t . c _ s t r ( ) ) ;

return 1 ;
}

75
E. RALF’S TRACKER — Used
experiment_specific.hh

Before the tracker can be used, changes have to be made to a file called experiment_specific.hh,
by default. These changes are made to tell the tracker to skip events which do not fulfil
certain conditions. In this Appendix, the modified source file is listed.
In experiment_specific.hh:
#i f n d e f EXPERIMENT_SPECIFIC_HH
#define EXPERIMENT_SPECIFIC_HH

#define TRACK_EXP −1 // <−−− make s u r e h e r e i s your e x p e r i m e n t

number ! ! ! !
// e . g . 393 f o r s393

#i f d e f EXTEND_EVENT_DATA

#include <cmath>

bool Event_data : : i n i t _ c u t s ( )
{
// l o a d your g r a p h i c a l c u t s h e r e i f needed . Return f a l s e i f l o a d i n g
// f a i l e d .
return true ;
}

void Event_data : : t r a c k _ e v e n t ( )
{
// a p p l y your own c u t s h e r e
// Tpat , incoming , c h a r g e ( e n e r g y ) on t o f −w a l l , SSTs , . . .
// l e a v e t h i s f u n c t i o n w i t h r e t u r n ; i f t h e e v e n t i s n o t good . This
section
// c o u l d be s i m i l a r t o t h e f o l l o w i n g ( a d j u s t t o your n e e d s ) :

// o u t g o i n g c h a r g e
i f ( ( Tfmul <1) | | ( Tfe [ 0 ] < 1 0 ) ) return ;

// ∗∗∗ you might want t o k e e p t h e s t u f f b e l o w : ∗∗∗

// c h e c k GFI p o s i t i o n s
i f ( ( Gf1n <1) | | ( Gf2n <1) ) return ;

76
 i f ( i s n a n ( Gf1x [ 0 ] ) | | i s n a n ( Gf2x [ 0 ] ) ) return ;
i f ( Gf1x [0] < −25.0 | | Gf1x [ 0 ] > 2 5 . 0 | | Gf2x [0] < −25.0 | | Gf2x [ 0 ] > 2 5 . 0 )
return ;

// c h e c k SSTs
i f (SS01SMUL==0 | | SS02SMUL==0) return ;
i f (SS01KMUL==0 | | SS02KMUL==0) return ;

//OK, s i n c e c a l i b r a t i n g w i t h B14 . Used as f i r s t g u e s s ?

I n z =5;
Inaoverz =14./5.;

// I n b e t a i s used as s t a r t a p p r o x i m a t i o n o f b e t a when c a l i b r a t i n g t h e
FTF (−−i g n o r e −t o f ) .
// Use t h e v a l u e from c o n f i g f i l e .
I n b e t a=MY_X( " I n b e t a " ) ;

// o u t z =5;

// c a l c u l a t e Z from e n e r g y r an g e i n TFW.
// u s e s known b e t a t o c h e a t some more .
i f (MY_X( " T r u e b e t a _ v a l i d " ) )
{
double b e t a=MY_X( " Truebeta " ) ;
double I i o n E =9.352906073747982 e −07; //mean i o n i z a t i o n e n e r g y (MeV)
double C_tfw = 0 . 0 6 0 3 1 9 3 3 9 9 4 6 2 2 3 ; //MeV
double BConst=C_tfw ∗ ( l o g ( 1 . 0 2 2 / ( I i o n E ∗ ( pow ( beta , −2) − 1 ) ) ) ∗pow (
beta , −2) − 1 ) ;
double Z=s q r t ( Tfe [ 0 ] / BConst ) ;
unsigned i n t Z_int=(unsigned i n t ) ( Z + 0 . 5 ) ;
i f ( f a b s ( Z−Z_int ) >0.25)
{ return ; }
o u t z=Z_int ;
}

// h e r e we a s s i g n t h e c h o s e n h i t s t o t h e c o r r e s p o n d i n g d e t e c t o r
planes .
// d e t e c t o r s b e f o r e a l a d i n ( and a f t e r t a r g e t ) :
t r a c k −>f t r a c k _ b e f o r e . set_measured ( SS01X ( 0 ) , SS01Y ( 0 ) ,& s s t 1 ,
SS02X ( 0 ) , SS02Y ( 0 ) ,& s s t 2 ) ;

// d e t e c t o r s a f t e r a l a d i n :
t r a c k −>f t r a c k _ a f t e r . set_measured ( Gf1x [ 0 ] , Tfy [ 0 ] ,& g f i 1 ,
Gf2x [ 0 ] , Tfy [ 0 ] ,& g f i 2 ,
Tfx [ 0 ] , Tfy [ 0 ] ,& f t f ) ;

T0 = 0 ;

// t r a c k i t !
track_fragment ( 0 , 0 , 0 , 0 , 0 ) ;

77
E RALF’S TRACKER — Used experiment_specific.hh

#e l s e
#d e f i n e EVENT_DATA_OMITTED
#endif

#endif //EXPERIMENT_SPECIFIC_HH

78
F. Energy Depositions for Z

By Equation 5.3, there is a relation between the Z and β of a charged particle passing
through a thin detector and its energy loss, but in order to use this relation, the material
specific parameters of Equation 5.3, C and I, need to be determined. These parameters
were determined for the TFW by observing the amount of energy deposited in the
TFW by most particles with Z ∈ {7, 6, 5, 4, 3, 2} and β ≈ 0.761; Z ∈ {5,4,3,2} and
β ≈ 0.745; and Z ∈ {8, 7, 6, 5, 4, 3, 2} and β ≈ 0.859. The different Z for a given β
come from particles undergoing charge-changing reactions. Regardless of whether the
particle changed charge or not, the β at the TFW was assumed to be the same β
as the particle of the fired particle. The β ≈ 0.761 simulation was made with 20 N ,
the β ≈ 0.745 simulation with 15 B and the β ≈ 0.859 simulation with 16 O. The
most likely energy deposits for particles of different Z in the mentioned simulations
are presented in Table F.1, together with the energies predicted by Equation 5.3 with
CTFW = 0.060319339946223 MeV and ITFW = 9.352906073747982 × 10−7 MeV.
These values of CTFW and ITFW were found by minimising the 2-norm of the difference
between vectors of the most likely energy deposits in the TFW for a given Z and the
values predicted by Equation 5.3. This was done using MATLABs fminsearch-function.
The MATLAB code used can be found in Section F.2.
With these values for CTFW and ITFW , Equation 5.3 was used to predict the energies
that particles of a given Z and β would deposit in the TFW, which in turn was used
to determine a best-match Z given the energy the particle deposited and the β of the
particle.
In order to verify our method for determining the energy deposition ranges of ions
with a given β and Z, an additional simulation with 17 C at β = 0.764 was used. The
resulting simulated energy depositions are presented in Figure F.1, and the positions of
the peaks corresponding to different Z are presented in Table F.2 alongside the predicted
values by Equation 5.3 with the constants delivered from MATLABs fminsearch.

F.1 Energy Ranges

From for example Figure F.1, depicting the energy depositions in the TFW, it is clear
that particles of a given charge and β mostly deposit their energy the TFW in a narrow
range. However, the energy depositions are not entirely restrained to a certain range,
and there are several particles depositing energies far away from the predicted peak. This

79
F Energy Depositions for Z

Table F.1: The simulated most likely energy peaks for different Z and β compared to
the values given by Equation 5.3 with CTFW = 0.060319339946223 MeV and
ITFW = 9.352906073747982×10−7 MeV. These values where obtained by min-
imising the 2-norm between the simulated and calculated values with MATLABs
fminsearch.
β Z Simulated energy (MeV) Bethe energy (MeV)
0.761 7 70 69.7
0.761 6 52 51.2
0.761 5 36 35.5
0.761 4 22 22.7
0.761 3 12 12.8
0.761 2 6 5.69
0.745 5 36 36.9
0.745 4 23 23.6
0.745 3 13 13.3
0.745 2 6 5.90
0.859 8 74.1 74.3
0.859 7 56.9 56.9
0.859 6 42.0 41.8
0.859 5 29.1 29.0
0.859 4 18.7 18.6
0.859 3 10.1 10.5
0.859 2 4.85 4.65

Table F.2: The positions of the peaks for different Z in Figure F.1 presented alongside
the calculated values using Equation 5.3 with the values of the constant given
by MATLABs fminsearch. Each value is in agreement within 1 MeV, which
is considered good enough given the width and spacing of the peaks to predict
the energy deposition of fired particles in the TFW.
Actual peak Calculated value
50.6 50.9
35.6 35.3
23.0 22.6
13.0 12.7
5.95 5.65

80
 F.1 Energy Ranges

Energy deposition in TFW

Counts

105 Z=2
Z=3
Z=4
Z=5
Z=6
104

103

102

10
0 10 20 30 40 50 60 70 80 90 100
Energy [MeV]

Figure F.1: A stacked histogram of the major energy depositions in TFW for ions of
a given charge when 17 C were fired with β = 0.764. The peaks for each Z
agree with the energy deposit predicted by Equation 2.4.

81
F Energy Depositions for Z

makes it somewhat more difficult to identify charges of individual particles from their
energy deposition alone; since the peaks will overlap, mistakes are unavoidable. Because
A is determined from the charge-to-mass ratio, this also means that the mass will be
incorrectly identified by the tracker in these cases. The amount of incorrectly identified
particles can however be minimised by only accepting particles of a given β with energy
deposits closely matching integer values of Z. Thus, Z calculated from Equation 5.3
were only accepted if they deviated from an integer value Zi by less than 0.25, that is,
|Z − Zi | ≤ 0.25. If so, then Z from Equation 5.3 was identified as Zi , otherwise the
particle responsible for the energy deposition was rejected by the tracker since it could
not be identified with sufficient certainty. The value 0.25 was chosen arbitrarily, but
was found to yield accurate results without rejecting overly many events. More accurate
energy ranges could possibly account for the fact that the peaks are not symmetric, and
that particles tend to deposit energy above the peak value.
In all the energy deposition spectra in the TFW when a single particle was fired,
the charge of the fired particle can be determined with certainty since particles of that
charge give a dominating contribution to the total number of events within it is energy
range, as shown in Figure F.1. The charge-changed particles can also be identified, but
doing so requires more statistics, as those events are orders of magnitude rarer.

F.2 MATLAB Code

%preamble
format long
u=1.6605389e−27;
c=299792458;
e=1.6021765e−19;
brho=10.4;

%function giving beta^2 for a previously defined !!! fixed brho !!!
beta2=@(A,Z)[1./(1+(brho.∗Z.∗e./(u.∗A.∗c)).^(−2))];
%Z, A and simulated E−peaks of sim1
Z=[7 6 5 4 3 2];
A=[20 17 18 10 8 6];
E=[70 52 36 22 12 6];
%beta^2 of sim1
bet21=beta2(20,7);
%beta^2 of sim2
bet22=beta2(15,5);

%beta^2 of both sim1 and sim2

bet2all=vertcat(bet21∗ones(6,1),bet22∗ones(4,1));
%Z of sim2
Z2=[5 4 3 2];
%Z of both sim
Zall=horzcat(Z,Z2)’;
%E of sim2
E2=[36 23 13 6];
%E of both sim

82
 F.2 MATLAB Code

Eall=vertcat(E’,E2’);

%E of sim3 (more decimals then before since I used to be lazy)

E3=[74.1 56.9 42.0 29.1 18.7 10.1 4.85];
%beta^2 of sim3
bet23=beta2(16,8);
%Z of sim3
Z3=[8 7 6 5 4 3 2];
%all Z, E and beta
Zall=vertcat(Zall,Z3’);
Eall=vertcat(Eall,E3’);
bet2all=vertcat(bet2all,bet23∗ones(7,1));

%function giving predicted energy deposit for these values of Z and beta^2
%with material parameters as argument
h=@(x)[x(1).∗(log((1.022./x(2))∗1./((1./bet2all )−1)).∗(Zall .^2./ bet2all ) ...
− Zall.^2 ) ];
% initial guess, based on previous experience
x0=[0.8;0.013e−3];
%find optimal material parameters with fminsearch by minimising 2norm of
%calculated and simulated energies
X=fminsearch(@(x)[norm(h(x)−Eall)],x0)
%calculated energies with these parameters
EallCalc=h(X)

%to test the parameters with additional data:

%most likeley energies and Z and beta^2 of sim4
E4sim=[50.6 35.6 23.0 13.0 5.95]’;
Z4=[6 5 4 3 2]’;
bet24=ones(5,1)∗beta2(17,6);
%calculated energies and difference check
E4calc=[X(1).∗(log((1.022./X(2))∗1./((1./bet24)−1)).∗(Z4.^2./bet24) − Z4.^2 )]
E4calc−E4sim
norm(E4calc−E4sim)

83
G. Software Resources

This appendix contains links to the not-commonly available software products that were
used for simulation and analysis during this project.

G.1 GGLAND
GGLAND can be downloaded by running:
export CVS_RSH=ssh
cvs -d :ext:land@[Link]:/u/johansso/CVS co land02
(2013-05-17, password protected)
A write-up and some simple examples may be found at:
[Link]

G.2 RALF’S TRACKER

The main version of RALF’S TRACKER can be obtained via:
git clone land@[Link]:/u/rplag/tracker
(2013-05-17, password protected)
The version used in this project is a slightly modified version, available at the computer
network of the subatomic physics group from:
git clone /net/home/htj/[Link] (2013-05-17)
Documentation on how to use the tracker may be found at:
[Link] (2013-05-17, password protected).

G.3 A Program Integrating GGLAND and RALF’S TRACKER

GLUE is a program for gluing together GGLAND and RALF’S TRACKER. It has plug-ins
for automating simulation, tracker calibration and tracking.
A download link for GLUE as well as other related resources can be found at:
[Link]

84
H. Svensk Sammanfattning

H.1 Inledning
Tvärsnittet för en reaktion σ, definieras enligt

R
σ= , (1)
NΦ

där Φ är intensiteten hos den inkommande partikelstrålen ([antal partiklar]/[area × tid]),

R är antal reaktioner r per tidsenhet och N är antalet partiklar som bestrålas. Tvärsnit-
tet kan ses som en proportionalitetskontant mellan intensiteten hos strålen och antalet
reaktioner normerat med antalet bestrålade partiklar. Ett exempel på en reaktion skulle
kunna vara att en atomkärna absorberar en extra neutron då den bestrålas av neutro-
ner, eller att kolliderande kärnor fusioneras. Tvärsnitten för sådana reaktioner är av stort
intresse, då de kan förklara hur de olika grundämnena uppstår och varför de finns i de
proportioner som de gör.
För att räkna ut absoluta tvärsnitt för kärnreaktioner krävs det att antalet reaktio-
ner kan räknas. Detta görs genom att detektera produkterna av dessa reaktioner. För
att ta hänsyn till att inte alla produkter kan detekteras, definieras acceptansen hos en
uppställning som
E[Ndetekterade händelser ]
Acceptans = , (2)
E[Nhändelser ]
med vilken det går att kompensera för att inte allting detekteras. Då antal händelser
inte är känt i praktiken krävs det i regel simuleringar för att bestämma acceptansen.
Denna kandidatarbetsrapport beskriver Monte-Carlo-simuleringar av delar av S393 -
experimentet som har utförts vid GSI Helmholtz Centre for Heavy Ion Research i tyska
staden Darmstadt. Huvudfokus har legat på hur bra analysverktygen beter sig under
olika omständigheter. Testfallen har varit enstaka joner, som från strålmålets position
avfyrats framåt med en magnetisk rigiditet på 8,8 Tm.
En del av analysen består i att finna partiklars väg, massa och laddning. Det har
undersökts hur många detektorhändelser som identifierats korrekt och om det finns några
situationer där identifikationen inte fungerar. En viktig del i experimentuppställningen
för identifieringen är ett magnetfält, som ser till så att laddade partiklar avviker från sin
ursprungliga väg.

85
H Svensk Sammanfattning

Ett resultat som presenteras är hur väl experimentuppställningen tillsammans med

analysverktyg kan identifiera framåtgående joner om skillnaden mellan det riktiga mag-
netfältet och det magnetfält som används i analysen varierar. Ett annat resultat beskriver
hur utgångsvinkeln mellan partikelbanan och strålaxeln påverkar mängden detekterade
partiklar.
Vid sidan om ovannämnda resultat, presenteras också resultat i form av detektordefini-
tioner, som skapades inom ramen för detta projekt. Dessa definitioner har genomgående
använts i, och var nödvändiga för, simuleringarna som genererade den data som analy-
serades.

H.2 Kort om detektorprinciper

Det finns olika typer av detektorer. I projektet har främst detektorer utnyttjande scin-
tillatorer simulerats. I en scintillator blir elektronerna närmast den passerande partikeln
exciterade till en högre energinivå [5]. De exciterade elektronerna kommer tillslut att
återgå till grundtillståndet via spontan relaxation varpå en foton emitteras. På grund av
ytterligare energinivåer mellan dessa nivåer, kan elektroner återgå i flera steg och avge
fotoner med energier som inte kan excitera elektroner i grundtillstånd. Dessa fotoner har
därför låg sannolikhet att absorberas. Istället kan fotonerna detekteras med hjälp av en
fotomultiplikator, vilket skapar en elektrisk signal [5].
Från energiprincipen följer att laddade partiklar som passerar detektormaterialet mås-
te deponera energi lika stor som den energi som används till att excitera elektronerna,
varför energidepositionen från partikeln kan mätas [5].

H.3 S393 -Experimentet

I figur H.1 presenteras en schematisk bild av experimentuppställningen i Cave C med
ingående detektorer.
Instabila joner kommer in i Cave C från FRS1 , där de skapats i kollisioner och filtrerats
beroende på deras laddning och massa via magnetfält och degraders. De inkommande
jonernas hastighet bestäms genom att en klocka som startats i FRS stoppas då jonerna
passerar POS2 . En ny klocka startas även då jonerna åker igenom POS.
Efter att jonerna passerat POS kommer de till ROLU3 , som består av fyra scintillator-
plattor som definierar en öppning för strålen att åka igenom. Om strålen blir för ofoku-
serad detekterar ROLU detta och skickar en signal, som säger åt systemet att inte lagra
data kring tillhörande händelser.
Efter ROLU går strålen in i XB4 -detektorn. Denna detektor består av pentagonala och
hexagonala NaI-kristaller placerade i ett sfärisk skal, med öppningar för inkommande
stråle och framåtgående produkter.
1
FRagment Separator
2
POSition
3
Rechts, Oben, Links, Unten
4
Crystal Ball

86
 H.3 S393-Experimentet

LAND
ROLU XB
POS ALADiN
Nuclei from
FRS Neutron arm

x
z
GFI
y Target
SST TFW

PDC Fragment arm

DTF

Proton arm

Figur H.1: Experimentuppställningen för S393 i Cave C vid GSI [7]. Strålen med jo-
ner kommer från vänster (från FRS) och träffar strålmålet (Target) innanför
detektorn XB. Resulterande partiklar detekteras sedan antingen av XB, eller
separeras beroende på laddning av ett magnetfält inuti ALADiN och detek-
teras i de olika armarna. Alternativt detekteras de inte alls. De icke-ifyllda
hindrena definierades i GGLAND som en del av detta arbete.

Inuti XB omringas strålmålet av 8 SST5 :er. Dessa kan användas tillsammans med
varandra eller XB beroende på vilken riktning SST:n är i, för att bestämma infallsvinkeln
hos strålen och utgående vinklar hos produkterna.
De produkter som lämnar XB i framåtriktningen når ALADiN6 . ALADiNs magnetfält
får produkterna att böja av beroende på kvoten mellan deras massa och laddning. Detta
får produkterna att åka in i en av tre olika armar:

• Protoner som har ett lågt förhållande mellan massa och laddning påverkas starkt
av magnetfältet. På grund av detta hamnar de i PCD7 :erna. Dessa består av en
uppsättning laddade vertikala och horisontella trådar vilka genererar ett elekt-
riskt fält som gör det möjligt att detektera laddade partiklar genom att samla in
frigjorda elektroner i deras väg.
Efter dessa når protonerna DTF8 , vilken mäter x och y positionen hos protonerna
samt stoppar den klocka som startats av POS, vilket ger medelhastigheten genom
uppställningen hos de enskilda protonerna.

• De tyngre laddade produkterna hamnar i fragment-armen, där de träffar de två

GFI9 :erna. Dessa består av vertikala parallella scintillator-fibrer, vilka kan ge en
x-position för fragmenten.
Efter detta träffar de TFW10 , vilken består av två plan av parallella scintillator
5
Silicon STrip detector
6
A LArge DIpole magNet
7
Proton Drift Chamber
8
Dicke ToF-wall
9
Großer FIberdetektor
10
Time of Flight Wall

87
H Svensk Sammanfattning

plattor, vridna 90◦ grader relativt varandra. Denna detektor ger x, y samt ener-
gidepositionen, och stoppar klockan startad av POS, vilket ger medelhastigheten
för enskilda fragment genom uppställningen.

• Neutroner, som är oladdade, böjs inte av av magnetfältet och fortsätter rakt fram
in i LAND11 , där de detekteras och en av POS startad klocka stoppas för att ge
deras medelhastighet.

Detta beskriver hur experimentuppställningen idealt ska fungera. I verkligheten beter

sig laddade partiklar stokastiskt, och på grund av växelverkan med till exempel luften
kommer de inte att åka helt rakt. Detta är anledningen till att experimentet behöver
simuleras.

H.4 Mjukvara
För att utföra simuleringarna användes GEANT4 tillsammans med en wrapper, GGLAND.
Detaljer om dessa återfinns i avsnitt H.4.1 och avsnitt H.4.2. Datan analyserades med
RALF’S TRACKER och ROOT, vilka presenteras i avsnitt H.4.3 avsnitt H.4.4.

H.4.1 GEANT4
GEANT4 är ett toolkit för att simulera växelverkan av partiklar med materia. De fysika-
liska processer och tillhörande modeller som en simulering ska ta hänsyn till definieras
i GEANT4 av en physics list, vilket ger villkor och sannolikheter för att vissa fysikaliska
processer ska inträffa. Vid en simulering slumpar GEANT4 fram tider då olika händel-
ser inträffar utgående från dessa modeller, och om partikeln i simuleringen är kvar då
denna tid inträffar så inträffar händelsen. I våra simuleringar användes physics list:an
QGSP_BERT, vilken är standardalternativet i GGLAND.

H.4.2 GGLAND
Då GEANT4 är väldigt allmän krävs det mycket arbete för att få GEANT4 att simulera
ett specifikt experiment, varför det är önskvärt att använda en wrapper. GGLAND är
ett kommandoradsverktyg, som använder GEANT4, för simulering av subatomära parti-
kelexperiment [16]. GGLAND låter användaren snabbt ställa upp vissa typer av vanliga
simuleringar genom att låta användaren placera ut fördefinierade hinder12 och simulera
av användaren specificerade spektrum av partiklar avfyrade mot dessa. Vidare går det
att modifiera mått på definierade hinder via kommandoraden. När detta inte räcker till
kan helt egna hinder specificeras i en modul som placeras i GGLAND:s källkodsträd. Som
utdata genererar GGLAND filer som kan analyseras med samma metoder som verkliga
experimentdata.
11
Large Area Neutron Detector
12
Ett hinder är en del av experimentuppställningen som en partikel måste passera.

88
 H.5 Simulering

H.4.3 RALF’S TRACKER

RALF’S TRACKER är ett program som försöker återskapa partikelbanor (tracks) genom
att kombinera data ifrån enskilda detektorer. Dessa partikelbanor används sedan för att
bestämma andra egenskaper hos partiklarna. Detta görs i princip genom att lösa rörel-
seekvationen för laddade partiklar genom ALADiNs magnetfält och försöka hitta den
bana som ger minst avvikelse från de experimentellt uppmäta positionerna i detekto-
rerna. Detta ger kvoten mellan en partikels massa och laddning. För att få massan och
laddningen räcker det således med att utöver detta bestämma partikelns laddning, vilket
kan göras med Bethes formel och energidepositioner i en detektor. För att göra allt detta
behöver RALF’S TRACKER även partikelns hastighet, vilket den kan få genom tiden det
tar för partikeln att nå TFW.
För att spåra partiklar ges RALF’S TRACKER information om experimentuppställning-
en i en konfigurationsfil. För att få indatan att stämma bra med det trackern beräknar
kan trackern tillåtas manipulera sin experimentuppställning, vilket görs då den kalibre-
ras. Detta kan motiveras med att det inte går att få perfekt information om den verkliga
experimentuppställningen, varpå trackern försöker lista ut denna från sin indata under
kalibrering.

H.4.4 ROOT
För att analysera den stora mängden data som simuleringarna och RALF’S TRACKER ge-
nererar används det för ändamålet framtagna C++-ramverket ROOT. ROOT är framtaget
för att hantera de stora datamängder som är typiska för experiment inom subatomär
fysik. Vidare kan både GGLAND och RALF’S TRACKER ge utdata i form av ROOT-filer,
varför det är lämpligt att använda ROOT.

H.5 Simulering
Den delen av figur H.1 som simulerats är de två SST:erna efter strålmålet, ALADiN samt
fragmentarmen. Detta möjliggör undersökandet av hur olika framåtgående fragment kan
detekteras givet förutsättningarna i simuleringen. I Table H.1 presenteras placeringen av
de olika detektorerna i simuleringen.
Vid simuleringarna användes ALADiN med en ström på I = 2500 A, vilket är den
högsta ström för vilken magnetfältet uppmätts. Detta svarar ungefär mot den ström som
användes vid det faktiska experimenten. För att få partiklarna att träffa fragmentarmen
avfyrades de iväg med en magnetisk rigiditet på bρ = 8,8 Tm, vilket ungefär svarar mot
att partiklarna träffade GFI1, GFI2 och TFW i mitten.
Nedan presenteras vad som avfyrades i olika simuleringar och vad som undersöktes i
dessa simuleringar:
• Olika partiklar avfyrades med en slumpmässig spridningsvinkel α ≤ 8◦ från z-
axeln. Detta för att undersöka hur olika spridningsvinklar påverkar uppställningens
förmåga att detektera partiklar samt RALF’S TRACKERs förmåga att analysera
resultaten. Endast en sorts partikel användes per simulering.

89
H Svensk Sammanfattning

• Olika 5 ≤ Z ≤ 9 joner avfyrades med en slumpmässig vinkel α ≤ 0,016 rad från

z-axeln. Detta användes för att undersöka hur ofta RALF’S TRACKER korrekt kan
identifiera olika partiklar beroende på med vilken ALADiN ström den kalibrerats.
Endast en sorts partikel användes per simulering.

Den analys som gjordes på de olika simuleringarna presenteras i avsnitt H.6. Resulta-
ten för de olika simuleringarna efter analys återfinns i avsnitt H.7.

H.5.1 Kalibrering av RALF’S TRACKER

Trackern kalibrerades för fragment i den grad det var möjligt enligt den procedur som
finns beskriven på trackerns hemsida [18]. Då proton- och neutronarmen inte simulerats
kunde inte vinkel kalibreras, varför detta steg hoppades över. Kalibreringen gjordes med
indata från en simulering då 14 B med bρ = 8,8 Tm avfyrades från strålmålets position
i z-axelns riktning med en spridning på 0,016 rad och ALADiN hade en ström på I =
2500 A. Utöver denna ström gavs trackern andra ALADiN strömmar än den som använts
i simuleringen, detta för att med facit i hand undersöka hur osäkerheter i ALADiNs
magnetfält påverkar trackerns förmåga att efter kalibrering identifiera olika partiklar.
Vidare gavs trackern Z hos partikeln som användes för kalibrering.
Kalibreringen kan kort beskrivas enligt följande:

1. Tracka i forward-läge med avfyrade partikelns β givet till trackern för att se hur
tiden för att nå TFW enligt trackern skiljer sig från tiderna i indatan. Addera
medelvärdet av denna skillnaden som kompensering och iterera tills felet blivit
litet.

2. Tracka i backward-läge och låt trackern denna gång avgöra hastigheten från tiden
det tar att nå TFW. Addera medelvärdet av avvikelsen mellan den trackade x-
koordinaten för energidepositioner i TFW och den i indatan och iterera tills felet
är litet.

Tabell H.1: Hinderplacering använd vid simulering. Placeringen är given som (x, y, z)-
koordinater relativt strålmålet. Rotationen ges med avseende på hindrens ne-
gativa y-axel.
Hinder Placering/cm Rotation/grader
ALADiN (0, 0, 255) 7
FTF (−232,94, 0, 1124.33) 15
GFI1 (−56,94, 0, 467.504) 15
GFI2 (−72,469, 0, 525.450) 15
SST1 (0, 0, 11) 0
SST2 (0, 0, 14) 0

90
 H.6 Analys

3. Notera toppen i masspektrumet som trackern gav från den tidigare gjorda simule-
ringen. Detta är massan trackern identifierat. Låt trackern köra en simulering med
denna massa och med den faktiska massan hos den avfyrade partikeln. Trackern
kan sedan ges instruktioner för att flytta detektorerna i sin uppställning för att
få en partikel med den korrekta massan att träffa detektorerna på samma ställen
som partikeln med den identifierade massan har träffat.
När trackern är kalibrerad kommer den således ha en annan experimentuppställningen än
den som användes i simuleringen. Med denna experimentuppställning kommer partikeln
som trackern kalibrerats för, trots felaktig ALADiN ström, att få den track som bäst
matchar indatan.

H.6 Analys
En träff är i föreliggande arbete definierat som det kluster av energidepositioner gruppe-
rade av GGLAND som tillsammans gav den största energidepositionen. Denna definition
görs då laddade partiklar som färdas genom ett medium joniserar närliggande atomer
och därför ger upphov till sekundära partiklar i form av frigjorda elektroner kring parti-
kelns spår. Då dessa elektroner oftast deponerar mindre energi än den primära partikeln
kan deras bidrag filtreras genom att bara betrakta den största gruppen av depositioner.
Den primära partikelns typ kan sedan fås från GGLANDs utdata genom att se vilken
sorts partikel som gav största bidraget till den primära energidepositionen.

H.6.1 Vinkelberoende hos detekterade händelser

För att få beroendet mellan detekterade händelser och vinkeln α, definierat som vinkeln
mellan utgående fragmentets ursprungliga riktning och strålaxeln, görs ett histogram
med cos α mot antal träffar i detektorerna. Detta kan göras med villkor för vilka typer
av partiklar som ska ha detekterats samt vilken mängd energi som ska ha deponerats.
För att se hur bra trackern är på att identifiera olika partiklar utnyttjas GGLANDs
förmåga att rapportera vilken partikel som ger största bidraget till den primära deposi-
tionen.

H.6.2 Utvärdering av trackerkalibrering

Hur bra en kalibrering är utvärderas genom hur bra den är på att identifiera partiklar
korrekt. För att kunna testa detta kördes simulering på isotoperna i mängden

SB = {10 B, 11 B, 12 B, 13 B, 14 B, 15 B, 17 B, 19 B}

med en ström I0 = 2500 A och ett bρ = 8.8 Tm. Orsaken till att isotoperna 16 B och
18 B exkluderades är att dessa är obundna.
Efter simuleringarna kalibrerades trackern med data från simuleringen av 14 B . Trac-
kern kalibrerades åtta gånger, med åtta olika strömmar till trackern. De strömmar som
användes var 2500 A, 2462,5 A, 2425 A,2387,5 A,2350 A, 2312,5 A, 2275 A, and 2237,5 A,

91
H Svensk Sammanfattning

där strömmen I0 = 2500 A är den ström som användes vid simulering och de övriga
strömmar listats med ett stigande fel. För varje ström skapades en kalibreringsprofil.
Givet de skapade kalibreringsprofilerna spårades utdata från simuleringarna för alla
isotoper i SB . Detta gjordes för varje kalibreringsprofil. Kalibreringsprofilerna som ska-
pats för SB användes också för olika atomnummer med ett fixt masstal. Isotoperna 22 C
, 22 N , 22 O och 22 F simulerades och spårades.
Avvikelser kan också inträffa när trackern har kalibrerats men ges data insamlad vid en
annan ström än den som användes vid kalibreringsförsöket. Därför gjordes simuleringar
för alla isotoper i SB med strömmen Isim = 2300 A. Sedan spårades utdata med samtliga
kalibreringsprofiler fast med en ström

I0 − Isim
 
It = Ic 1− , (3)
I0

där Ic är den ström som skickades till trackern vid kalibrering och I0 den ström som
användes för att generera data till kalibreringen.

H.7 Resultat
H.7.1 Vinkelberoende hos detekterade händelser
I figur H.2 presenteras resultatet från en simulering där 220 stycken 14 B avfyrats med
bρ = 8.8 Tm och en maximal spridningsvinkel α på 8◦ . Figuren illusterar hur antalet
avfyrade partiklar som uppfyller olika vilkor varierar med avseende på vinkeln.
I figur H.2a visas antalat partiklar som gav upphov till energidepositioner överskri-
dande vissa tröskelvärden i olika detektorer. Låga tröskelvärden i förhållande till de mest
troliga energidepositionerna i de olika detektorerna visar sig inte förvränga vinkelbero-
endet i antalet detekterade händelser.
Figur H.2b visar antalet partiklar som ger upphov till träffar i de olika detektorerna
för olika cos α. Ifrån detta verkar det som att GFI2 är den detektorn som partiklarna
börjar missa först.
Då partiklar kan missa olika detektorer före andra kan de olika spårningsmetoder-
na användas för olika mängder av partiklar. Partiklar som kan spåras med de olika
metoderna visas i figur H.2c, där det syns att flera partiklar kan spåras i forward-läget.
Slutligen visar figur H.2d den andel primära energidepositioner i TFW som tillskrivs
olika partiklar för olika cos α. Från denna figur är det tydligt att en andel av de 14 B
som avfyras når TFW som andra fragment, samt att främst elektroner når TFW för
stora α.
Med simuleringsresultaten som bakgrund kan trackern utvärderas genom att jämföra
dess resultat med simuleringens. Trackerns förmåga att identifiera partiklar avfyrade vid
olika vinklar presenteras i figur H.3. De olika figurerna visar bland annat antal korrekt
identifierade partiklar för olika vinklar givet extra filter med avseende på energideposi-
tioner i olika detektorer, svarande mot olika histogram i figur H.2a. I figur H.2d visas
antal identifierade partiklar med det striktare villkoret att |A − 14| < 0,1 för att en

92
 H.8 Diskussion

partikel ska anses vara korrekt identifierad. Med detta villkor identifierar trackern färre
partiklar.

H.7.2 Inverkan av felaktig ström hos partikelspårning

I figur H.4 visas massavikelser ∆m = midentifierad −msimulerad när olika kalibreringsprofiler
har använts på utdata från simuleringar med strömmen I0 = 2500 A för isotoperna
i SB , som definieras i avsnitt H.6.2. Det är tydligt att alla isotoper har identifierats
korrekt och att absolutbeloppet av massavvikelsen är mindre än 0,05 u. Man ser även
att massavvikelsen tycks öka när A ökar.
Genom att försöka spåra utdata från simulering med A = 22 och Z ∈ {6, 7, 8, 9} vid
samma ström, erhölls resultatet i figur H.5. Även i detta fall har samtliga isobarer iden-
tifierats korrekt med en maximal massavvikelse mindre än 0,08 u. I båda fallen tenderar
trackarna att identifiera massor som är lite tyngre än de egentliga massorna.
När annan ström användes vid simulering och trackern körs på genererade utdata med
de tidigare kalibreringsprofilerna och en ström som beräknats enligt ekvation 3 erhölls
massavikelser enligt figur H.6. Även i detta fall har samtliga isotoper identifierats korrekt,
absolutbeloppet av massavvikelsen är här under 0,06 u. Vi noterar även att den tycks
vara mer centrerad kring ∆m = 0.
Som illustreras i figur H.7 flyttar trackern detektorerna under kalibrering på ett sätt
som gör att linjen från ALADiNs mitt genom detektorernas, roteras med en vinkel som
kompenserar för avvikelsen i ström. Vinkeln har beräknats enligt

xoff
 
θoff = arctan (4)
zdist

Eftersom trackern enbart ändrar detektorernas x-koordinat (i detektorsystemet) blir

armen något längre eller kortare, och själva detektorerna roteras inte. Denna inverkan hos
en inkorrekt ström stämmer överens med vad Ralf Plag har presenterat på S393/land02
Workshop 2011 [17].

H.8 Diskussion
H.8.1 Spridning och antalet träffar
Som synes genom att jämföra figur H.3 med figur H.2a lyckas trackern korrekt identifiera
stora delar av de avfyrade 14 B som träffar de detektorer som används för fragment-
identifiering. En del av de partiklar den inte lyckas identifiera kan också tillskrivas till
det faktum att alla 14 B som avfyras inte når TFW som 14 B , vilket syns i figur H.2d
då andra joner än 14 B ger upphov till den primära energidepositionen.
I figur H.2b syns det att GFI2 är den detektor som missas för minst vinkel, vid
α ≈ 2,5◦ . Detta svarar mot vinkeln då antal partiklar trackern lyckas rekonstruera
minskar, varför det verkar som att partiklar som missar GFI2 är orsaken till att trackern
ofta inte kan identifiera partiklar med α > 2,5◦ .

93
H Svensk Sammanfattning

Counts

Counts
7000 Hit the SST1 u+v
7000
No cut Hit the SST2 u+v
6000 over 0MeV in all 6000 Hit the TFW
over 2MeV in all
TFW 10MeV, GFIs 2MeV, SSTs u+v 2MeV Hit the GFI1
5000 TFW 10MeV, GFIs 6MeV, SSTs u+v 2MeV 5000
Hit the GFI2
TFW 20MeV, GFIs 6MeV, SSTs u+v 2MeV
4000 TFW 20MeV, GFIs 6MeV, SSTs u+v 3MeV 4000

3000 3000

2000 2000

1000 1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(a) ...orsakade energidepositioner över visa värden (b) ...träffade de olika detektorerna.
i alla detektorer. “No cut” visar alla avfyrade
partiklar.
Counts

Counts

7000 Hit all (mixed) All particles

7000
B14
6000 Hit all except perhaps GFI1 or GFI2 (forward) 6000 not B14
Hit all except perhaps SST1 u+v or SST2 u+v (backward) Ions
5000 5000
Electrons
4000 4000

3000 3000

2000 2000

1000 1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(c) ...vore möjliga att tracka med de olika track- (d) ...orsakad primära energidepositionerna till-
metoderna i RALF’S TRACKER. skrivna olika partiklar.

Figur H.2: Antal partiklar i en simulering där 220 stycken 14 B avfyrats med bρ = 8,8 Tm
och en spridningsvinkel α på 8◦ för olika cos α som...

94
 H.8 Diskussion
Counts

Counts
6000
No cuts No particle cuts
6000
Correctly identified as B14 5000 Correctly identified as B14
5000 Incorrectly identified as B14 multiplied by 10 Incorrectly identified as B14 multiplied by 10
4000 B14 not identified
B14 not identified
4000

3000
3000

2000
2000

1000 1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(a) Över 0 MeV i alla detektorerna (b) Över 2 MeV i alla detektorerna
Counts

Counts

5000
No particle cuts No particle cuts
6000
Correctly identified as B14 Correctly identified as B14
4000 Incorrectly identified as B14 multiplied by 10 5000 Incorrectly identified as B14 multiplied by 10
B14 not identified B14 not identified
4000
3000

3000
2000
2000

1000
1000

0 0
0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1 0.99 0.991 0.992 0.993 0.994 0.995 0.996 0.997 0.998 0.999 1
cos(α ) cos(α )

(c) Över 20 MeV i TFW, 6 MeV i GFI:erna 3 MeV i (d) Över 0 MeV i alla samt |A − 14| < 0.1
SST:erna (u+v)

Figur H.3: Antal av RALF’S TRACKER identifierade partiklar från en simulering där 220
stycken 14 B avfyrats med bρ = 8,8 Tm och en spridningsvinkel på 8◦ för olika
cos α. Olika underfigurer visar hur trackern presterade då händelser ytterli-
gare filtrerats. “B14 not identified” histogramet inkluderar 14 B som uteslutits
av dessa extra filter. Energi filtreringarna i de olika delfigurerna svarar mot
de olika histogrammen i figur H.2a. Underfigur (d) är med ett striktare vill-
kor för att A ska anses vara korrekt identifierat, i övrigt användes vilkoret
|A − 14| < 0,5.

95
H Svensk Sammanfattning

0.05
Ic = 2237.5 A
0.045
Ic = 2275 A
0.04 Ic = 2312.5 A
0.035 Ic = 2350 A
(u)

0.03 Ic = 2387.5 A
Ic = 2425 A
∆m

0.025
Ic = 2462.5 A
0.02
Ic = 2500 A
0.015
0.01
0.005
10 11 12 13 14 15 16 17 18 19
A

Figur H.4: Massavvikelse mellan simulerad och spårad massa för Z = 5 som funktion
av A för olika ∆I = I0 − Ic . Resultatet erhölls genom att köra trackern med
samtliga kalibreringsprofiler på utdata från simulering vid samma ström som
användes vid kalibrering. Storheterna I0 och Ic definieras i avsnitt H.6.2.

0.08
A = 22 Z = 6
0.07
A = 22 Z = 7
0.06
A = 22 Z = 8
0.05
A = 22 Z = 9
(u)

0.04
0.03
∆m

0.02
0.01
0
-0.01
-0.02
0 50 100 150 200 250 300
∆I (A)

Figur H.5: Massavvikelse mellan simulerad och spårad massa för A = 22 och olika Z
funktion av ∆I = I0 − Ic . Resultatet erhölls genom att köra trackern med
samtliga kalibreringsprofiler på utdata från simulering vid samma ström som
användes vid kalibrering. Storheterna I0 och Ic definieras i avsnitt H.6.2

96
 H.8 Diskussion

0.04
A = 10 Z = 5
0.03
A = 11 Z = 5
0.02
A = 12 Z = 5
0.01
A = 13 Z = 5
(u)

0 A = 14 Z = 5
-0.01 A = 15 Z = 5
∆m

-0.02 A = 17 Z = 5
-0.03 A = 19 Z = 5
-0.04
-0.05
-0.06
0 50 100 150 200 250 300
∆I (A)

Figur H.6: Massavvikelse mesllan simulerad och spårad massa för olika A och Z = 5 som
funktion av ∆I = I0 −Ic . Simuleringarna gjordes med en ström på 2300 A och
strömmen som skickades till trackern beräknades ur ekvation 3. Storheterna
I0 and Ic definieras i avsnitt H.6.2.

1
TFW
0.9
GFI1
0.8
GFI2
0.7
Predicted
(deg)

0.6
0.5
θoff

0.4
0.3
0.2
0.1
0
0 50 100 150 200 250 300
∆I (A)

Figur H.7: Fragmentarmens rotation efter kalibrering som funktion av ∆I = ∆I =

I0 − Ic . Rotationsvinkeln har beräknats ur ekvation 4. Linjen “Predicted”
är den beräknade skillnaden mellan vinkeln för utgående fragment och frag-
mentarmens vinkel vid efter kalibrering vid felaktig ström. Storheterna I0 och
Ic definieras i avsnitt H.6.2.

97
H Svensk Sammanfattning

Något som inte lyckats förklaras är varför antal partiklar som träffar TFW:n inte
uppvisar samma beteende som de som träffar de andra detektorerna. Övergången mellan
att partiklar missar och träffar TFW:n är mycket mjukare än för GFI:erna. Förklaringar
skulle kunna vara att ALADiNs magnetfält får partiklar att inte nå den avlägsna TFW:n.

H.8.2 Kompensering för inkorrekt kalibreringsström

Trackern lyckas kompensera för felaktiga strömmar under kalibreringen genom att flytta
på detektorerna i fragmentarmen på ett sätt som ungefär svarar mot att rotera armen
kring ALADiNs centrum, vilket ses i figur H.7. Detta medför visa fel, förutom i detek-
torpositioner. För det första kommer inte detektorerna vara riktade mot ALADiNs mitt,
vilket distorderar strålens skärning med detektorerna något. För det andra blir armen
lite längre. Dessa effekter är dock små för de ∆I som undersökts, och ger inte upphov
till fel stora nog för att orsaka att partiklar felidentifieras.
Det faktum att trackern flyttar detektorerna till felaktiga positioner kanske utgör ett
problem, även om det inte påvisats i de simuleringar vi har utfört. Det vore då kanske
bättre att korrigera för strömmen istället. En indikation på att så vore fallet är om
trackern flyttar detektorerna så att arctan (x förflyttning/avstånd från ALADiN) är
lika för de olika detektorerna. Om så är fallet bör förflyttningen inte utföras, utan istället
bör ett nytt I för trackern beräknas.

98