Professional Documents
Culture Documents
Afhandling
Afhandling
PhD Dissertation
by
Jesper Halkjær Jensen
October 2018
v
Clever guy,
“Clever guys book”, unpublished [Citation].
Contents
Contents i
Abstract v
Resumé vii
Acknowledgments ix
List of Publications xi
1 Preface 1
2 Reaction theory 3
2.1 Kinematics . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Kinematic compression . . . . . . . . . . . . . . . . . . . . . 5
2.3 Scattering theory . . . . . . . . . . . . . . . . . . . . . . . . 10
3 Experimental technique 13
3.1 ISOLDE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.2 Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 ∆E − E telescope . . . . . . . . . . . . . . . . . . . . . . . 18
4 Beam diagnostics 21
i
ii CONTENTS
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3 Setup and simulation . . . . . . . . . . . . . . . . . . . . . . 24
4.4 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5 AUSAlib 43
5.1 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5.2 Energy loss . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.3 Analysis pipeline . . . . . . . . . . . . . . . . . . . . . . . . 47
6 Data acqusition 53
6.1 Triggers and acqusition control . . . . . . . . . . . . . . . . 54
6.2 Analogue chain . . . . . . . . . . . . . . . . . . . . . . . . . 55
6.3 Readout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.4 Multicrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
6.5 Dead time ratio . . . . . . . . . . . . . . . . . . . . . . . . . 58
6.6 Pulsed beam . . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.7 Back end . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
6.8 DaqC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.9 RunDB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
6.10 Real time monitoring . . . . . . . . . . . . . . . . . . . . . . 66
7 Acceptance 69
7.1 MC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.2 Rutherford estimation . . . . . . . . . . . . . . . . . . . . . 73
8 Experimental setups 77
9 IS367 79
9.1 Data analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 80
9.2 Intensity estimation . . . . . . . . . . . . . . . . . . . . . . 85
10 IS561A 91
10.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
CONTENTS iii
11 IS561C 101
11.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
11.2 Main target . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
11.3 Secondary target . . . . . . . . . . . . . . . . . . . . . . . . 110
12 Results 115
12.1 Elastic scattering . . . . . . . . . . . . . . . . . . . . . . . . 116
12.2 10 Li transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
12.3 8 Li transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Abbreviations 143
Abstract
v
Resumé
vii
Acknowledgments
ix
List of Publications
The following manuscripts form part of this thesis
xi
xii CONTENTS
Preface
1
2
Reaction theory
2.1 Kinematics
The reactions we study are usually two body reactions. When we populate
unbounds states such a description will not suffice due to break up into
multiple constituents, but depending on the break up mechanics, we may still
be able to learn something from a simple two body picture. In this chapter
we will develop the equations we need to describe the reactions as well as
discuss the kinematic considerations required to design and undertake an
experiment in inverse kinematics, that is when the projectile is significantly
heavier than the target.
Since our experiments are carried out a relatively low energies, we will
treat the kinematics in a classical fashion.
Consider a beam particle, B, impinging on a target particle t, with some
kinetic energy TB in the laboratory frame. We assue the target is at rest
and thus no kinetic energy Tt = 0.
They react in some way and produce a heavy projectile-like recoil R and
a light target-like ejectile e (if we were in normal kinematics, it is customary
to denote the light particle as the projectile-like instead)
B + t → R + e. (2.1)
We assume that both the projectile and the target are initially in the
ground state. Conservation of energy gives
3
4 CHAPTER 2. REACTION THEORY
∗
TB = TR + Te + ER + Ee∗ − Q, (2.2)
where Ee∗ is the excitation energy of the ejectile, ER
∗ is the excitation energy
solution. In this case the a given angle corresponds to two energies. The
maximal angle this can occur is when the two terms inside the square root
just cancels and the two "branches" of solutions meets. Since the low-energy
solutions corresponds to large-angle ejectiles, we say that the kinematic
curve bends back.
An example of both single valued and double valued kinematic curves
can bee seen in Figure 2.1 (A) for elastic scattering of 9 Li (d,d)9 Li.
In the case of two body reactions, we can instead limit our measurement
to the ejectile and use energy and momentum conservation to infer the
kinematics of the recoil, which is usually the particle of interest. If we
assume that the ejectile stays in the ground state (which is resonable in the
case of deuterons and tritons since they would be unbound otherwise) we
can calculate the excitation energy of the recoiling nucleus with 2.2 and 2.5
if we know the masses of all four constituents
∗
ER = TB − TR − Te + Q (2.7)
(~
pB − p~e )2
= TB − Te − + Q, (2.8)
2mR
Let us consider elastic scattering 9 Li (d,d)9 Li. In this case the projectile
and the heavy recoil have the same mass mB = mR and Q = 0. The
kinematic curve reduces to
√ √
1/2 mB me TB cos θ ± mB me TB cos2 θ
Tb = , (2.9)
mR + me
which vanishes when θ ≥ 90◦ . We can see the resulting kinematic curves for
the ejectile and the recoil in Figure 2.1 (A).
All the deuterons are scattered in the first 90◦ in LAB and we have
effectively compressed the full angular range in CM to only 90◦ in LAB.
For the heavy recoil we see the back bend in the kinematic curve. The
heavy recoil is focused a small LAB angles, since it eventually will bend
back toward 0◦ . This has two important implications if we want to measure
both the ejectile and the recoil. Firstly, we need to put detectors close
to the beam axis cover a resonable area. Secondly, our angular resolution
suffers close to the turning points, since we have a lot of CM-angles that
are compressed into a small angular range of the LAB frame.
When we transfer a neutron we change the Q-value for the reaction.
Depending on whether we change transfer it to the target or the projectile,
we get two different situations.
In Figure 2.1 (B) we see the kinematic curves for 9 Li (d,p)10 Li. The
ejectile is single valued all the way to 180◦ . The energy at backward angles
are, however, much lower. We can increase the proton energy by increasing
the beam energy. An experimental setup will have some threshold that
determines how low energy protons we can measure. It is thus necessary to
increase the beam energy enough to be able to actually measure the protons
in the backward directions if desired. This asymmetric energy distribution
is important to consider when designing the setup.
The recoil is scattered even closer to the beam axis, and coincidence
measurements requires coverage even closer to the beam axis than in the
case of elastic scattering. Moreover will the angular resolution be even
worse, especially at the region around the maximally scattered recoils.
In Figure 2.1 (B) we see the kinematic curves for 9 Li (d,t)8 Li. In this
case we even see the back bend for the light ejectile. We notice that the
2.2. KINEMATIC COMPRESSION 7
25 30
Ejectile Ejectile Ejectile
25 Recoil Recoil Recoil
25
20
20
20
15
15
15
10
10 10
5 5
5
0 0 0
0 25 50 75 0 50 100 150 0 50 100
Figure 2.1: Kinematic curves for ejectile and recoil of the reactions 9 Li
(d,t)8 Li, 9 Li (d,d)9 Li and 9 Li (d,p)10 Li at 2.68 MeV/A.
recoil scatters further from the beam axis, making this situation better for
coincidens measurements.
Detector coverage
When designing an experiment it is important to consider what region of
4π that should be covered by the detectors. The physics of the reaction will
guide what angles are the most interesting. Usually, however, theoretical
calculations are performed in CM and thus we need to convert this to
LAB to design the detector setup, and this conversion is governed by the
kinematic of the reaction.
In the previous section we discussed how the angular resolution is
compressed in the forward LAB angles when we work in inverse kinematics.
This means that the angular resolution will be different across the angular
range if similar detectors are used. Take for instance the case of 9 Li (d,p)10 Li.
We can calculate the angular resolution as dθCM /dθLAB . The result at
2.68 MeV/A is shown in Figure 2.2. We can see that we need a much better
angular resolution to at forward angles compared to backward angles if we
want to have compareable CM resolution in both regions. The required
8 CHAPTER 2. REACTION THEORY
dσ dσ
J(θLAB ) = J(θCM ) J(θCM ), (2.10)
dΩ LAB dΩ CM
where
(1 + γ 2 + 2γ cos θCM )3/2
J= , (2.11)
1 + γ cos θCM
and
s
mB me Ech
γ= . (2.12)
mt mR Ech + Q
A B
2.0
150
1.5
dθLAB
100
dθCM
θCM
1.0
50 0.5
0 0.0
0 50 100 150 0 50 100 150
θLAB θLAB
C D
4 2.0
θLAB
θCM
3 1.5
J sin θ
2 1.0
J
1 0.5
0 0.0
0 50 100 150 0 50 100 150
θ θLAB
Figure 2.2: Kinematic effect on angular resolution and solid angle coverage.
All the plots corresponds to a proton from 9 Li (d,p)10 Li at 2.68 MeV/A.
A: The relation between the scattering angle in LAB and CM. B: The
differential, which gives a measure of the angular resolution in LAB needed
to recover a given angular resolution in CM. C: The Jacobian that transforms
from CM to LAB plottet. Since there is a 1-1 correspondance between
θCM and θLAB it is plotted as a function of both. D: The Jacobian times
sin θ plottet as a function of θLAB .
10 CHAPTER 2. REACTION THEORY
p+t→R+e (2.13)
The scattering is described by the total wavefunction of the system and
the potential that describes the interaction between the projectile and the
target via the time-independent Schrödinger equation
where T̂R~ is the kinetic operator, Hp (xp ) and Ht (xt ) are the internal Hamil-
tonians of the projectile and target, Vα (R, ~ xp , xt ) is the interaction and
Ψ(R,~ xp , xt ) is the total wavefunction. The internal coordinates of the pro-
jectile and the target are denoted xp and xt . The total energy of the system
is the sum of kinetic and internal energies.
To calculate the cross sections of a given reaction, we need to solve ??
to obtain Ψ(R, ~ xp , xt ).
Since we measure the scattered particles far away from the reaction
site, we consider the asymptotically behavior of the solution. The incident
projectile can be represented as a plane wave approaching the target. During
the collision the wave will be distorted, due to the interaction described by
Vα (R,
~ xp , xt ). The outgoing wavefunction is assumed to be a radial wave
and we can write the total asymptotic wave function as
~ ~
~ R~ eiK·R
Ψ (+)
→e iK·
+ f (θ) , (2.15)
R
2.3. SCATTERING THEORY 11
where (+) refers to the fact that we have outgoing boundary conditions.
f (θ) is the socalled scattering amplitude and we can calculate the scattering
cross sections from this
dσ
= |f (θ)|2 . (2.16)
dΩ
Optical potential
The simplest approach to determine elastic cross sections is the Optical
Model (OM). In this model only the elasic channel is calculated, and all
other channels are treated as loss of flux from the the elastic channel.
In general the scattering problem is well understood.
3
Experimental technique
In this chapter we will discuss some more technical topics that were common
for all three experiments. There has been a varity of upgrades to the various
techniques and equipment between the experiments, but the main concepts
and objectives has remained the same.
Performing experiments with highly unstable isotopes are no trivial task,
due to their very short lifetime. In the case of 9 Li we have τ = 172ms to
perform the experiment before half the nuclei has decayed. This can not be
done in ordinary kinematics since we can not create a target with such short
lived isotopes. Instead we can inverse the kinematics and use a radioactive
ion beam (RIB) of 9 Li impinging on hydrogen isotopes. Deuterons can easily
be implanted in a plastic foil, taking the place of protons. A small percentage
of the protons will remain, giving rise to a contamination that must be
assessed in the analysis. Tritons can, on the other hand, be implemented
on a titanium backing [REF]. Several RIB-facilities across the world exists.
The experiments of this thesis were all performed at the ISOLDE facility
at CERN, in Switzerland. Between IS367 and IS561A there were major
upgrade of the facility, the HIE-ISOLDE upgrade, which meant we could
reach higher beam energies.
To detect the particles we used standard silicon detectors. The signals
were amplified with a series of analogue modules and finally digitized with
VME-compatible ADC’s and TDC’s. The readout of these methods were
optimized between the experiments. As I did not take part in the IS367
13
14 CHAPTER 3. EXPERIMENTAL TECHNIQUE
experiment I will not discuss the data acqusition system from this experiment,
however, it was by large the same modules and logic that was used in the
two latter experiments.
3.1 ISOLDE
The radioactive beams at ISOLDE are produced using the isotope seperation
on-line (ISOL) method[REF]1 . A pulse of ≈ 1.4 GeV protons from the PS
Booster impinges on a primary target of a heavy element, for instance UCx
or Tantulum. This produces a wide array of different isotopes through
fragmentation, spallation and fission. The target is heated which makes the
products diffuse into the ion source. Inside the ion source they are laser
ionized and subsequently accelerated to ≈ 60 keV through a high voltage.
They are sent to one of the two mass seperators GPS or HRS. The GPS
(General purpose seperator) have a single bending magnet and the HRS
(High resolution seperator) have two bending magnets. The mass seperators
seperate the particles according to their masses and the isotope of interest
can be sent to one of the experiments.
In order to do reactions we need a higher beam energy than 60 keV.
The first installation to perform post-acceleration of the beam was REX-
ISOLDE and it was intalled between 2001 and 2005. Firstly the beams reach
a trap where the beam particles are trapped, cooled down and bunched in a
Penning trap. A buffer gas is used in the cooling process. These bunches
are continued into an electron beam ion source (EBIS). Inside the EBIS the
beam is highy ionized before it is sent through another seperator, which
seperates according to A/q to get rid of the buffer gas residues. After the
seperator the beam is sent into a linear accelerator which can accelerate
them up to ≈ 3 MeV/A.
The REX-ISOLDE was used for post acceleration in IS367. In the decade
following this experiment a series of superconducting cavity moduels was
installed to increase the postacceleration capabilities. This was part of the
HIE-ISOLDE upgrade, which aims to provide High Intensity and Energy.
1
FiXme Note: Find reference!!!
3.1. ISOLDE 15
HRS
1.4 GeV
GPS protons
60keV
The superconducting cavities were parts of the energy upgrade and was
completed with the seventh module in (WHEN?)3 . In 2016 only three of
the modules were installed, and 6.72 MeV/A was the highest energy that
was reached until then.
chamber. We call this time the TPROTON. Due to the large timescale of the
supercycle, we clearly see the decay characteristics of the beam. This is a
good way to access whether the beam is indeed 9 Li. In Figure 3.2 we see the
TPROTON-spectrum for 9 Li induced reactions. The clear steps in increments
of 1.2 s corresponds to the super cycle. The fast rise in the beginning is
the production, extraction and acceleration time, before they end up in the
chamber. This is followed by a exponential slope, which corresponds to
the decay of 9 Li. This is key indication the 9 Li composition of the beam.
This can also be used to limit background contributions. Backgrounds
from the buffer gases would be uncorrelated with TPROTON and give a flat
contribution. By restricting the TPROTON to the first 3-4 life times, we can
omit a considerable amount of background, however, we will also discard
some false negatives. This will, however, be a small fraction since to most
of the 9 Li arrives in the first second or so.
A better constraint can be achieved from the EBIS release. The trap
releases a bunch with a well defined repetition frequency (typically around
30 Hz) and each of these "spills" have a well defined length. We also record
a timestamp when the trap is opened, so we can calculate the time from
the begin of the spill and a given event. We refer to this time as TEBIS,
and a spectrum can be seen in Figure 3.2. The events are nicely constained
to a region of ≈ 700 µs, corresponding to the release time. We can thus
also use this to exclude background components which are recorded outside
this window. Background sources such as noise is in general uniformely
distributed and can by suppressed significantly in this way.
Since the EBIS release is relatively short and the release curve is concen-
trated in the first 100 µs, we effectively get very high instanous rates. With
a repitition rate of 30 Hz and a release of ≈ 700 µs we recieve all the beam
in only 700 µs · 30 Hz = 21 ms of each second. This mean that the average
intensity during a release will be two orders of magnitude larger than the
reported intensity. It will be even larger in the begining of the release. This
put extra requirements on the data acqusition system in order to keep up.
3.2. DETECTORS 17
105 103
104
102
Counts/20 ms
103
Counts/1 µs
1.2s
102 101
2.4s
101
100
100
10−1 10−1
0 1 2 3 4 5 6 0 100 200 300 400 500 600 700 800 900 1000
TPROTON [s] TEBIS [µs]
3.2 Detectors
Our main workhorse for charged particle spectroscopy is silicon detectors.
The silicon detectors works as a pn-juntion in reverse bias. When a charged
particle enteres the detector an electron-hole pair is created. By applying a
bias across the detector, these carges generate a signal which we can pick
up an digitize. We use generally two types of silicon detectors.
To get position information we use double sided silicon strip detectors
(DSSD). The electrodes that read out the charges are segmented into a
number of strips. On one side of the detector these strips are oriented
horizontally and on the other side they are oriented vertically. When a
particle hit the detector, ideally the charge is split evenly between a strip
on each side (electrons goes to one side, holes to the other side) and we get
a signal in two orthogonal strips. This lets us determine which pixel, i.e.
which small area that overlaps with both strips that was hit.
The second kind of detectors are the simpler unsegmented detectors.
They consist of is a single silicon wafer, and we do not have any positional
information from these. We will refer to these as a pad.
DSSD’s comes in many different sizes and geometries, and I will shortly
describe the kinds we use. I will refer to their unique product names.
18 CHAPTER 3. EXPERIMENTAL TECHNIQUE
W1 Square DSSD with 16 strips on each side. This gives 256 pixels. Each
strip has a width of 3 mm and an interstrip region of 100 µm. They come in
different thicknesses, but during these experiments we only used 60 µm.
BB7 Another square DSSD but with 32 strips on each side. This gives us
a greater angular resolution compared to the W1. The strip width is 2 mm
and the interstrip region is 100 µm.
S3 A round DSSD with a hole in the middle. The front side consists of 32
radially oriented strips (spokes) and the back side consists of 24 rings. Each
ring is 886 µm wide and each radial spoke covers 11.25◦ . The radius of the
inner hole is 11 mm. This detector was used in two of the experiments, and
were 1 mm thick.
3.3 ∆E − E telescope
We can use a combination of two silcon detectors to distinguish between
different particle types. If a particle has enough energy it will penetrate
the detector and only deposit a part of its energy in the detector. If we
place a thick pad detector behind which stops the particle, we can use the
correlation between the energy deposited in the front detector and the back
detector to uniquely identify which particle it was while still measuring the
full energy. This relationship is described by the Bethe Bloch equation. We
call such a setup for a ∆E − E telescope.
If we plot the energy deposited in the DSSD vs the energy deposited in
the pad we get Figure 3.4 (A). The upper feature corresponds to α and the
lower feature corresponds to hydrogen isotopes.
The energy deposited in the front detector, however, dependes on the
distance traveled and thus on the incoming angle,
t
teff = , (3.1)
cos θ
where teff is the distance traveled, t is the thickness of the detector and θ is
the angle shown in Figure 3.3. Each particle, depending on the angle, will
3.3. ∆E − E TELESCOPE 19
teff
DSSD t
θ Particle
Figure 3.3: The angle, θ, between the normal of the detector and the particle
determines the actual length teff the particles travel in the detector.
(A) (B)
12.5 12.5
10 10
∆E · cos θ [MeV]
∆E [MeV]
7.5 7.5
5 5
2.5 2.5
0 0
0 5 10 15 20 25 0 5 10 15 20 25
E [MeV] E + ∆E · (1-cos θ) [MeV]
Figure 3.4: Energy correlation plots for a silicon telescope. In (A) we see
the energy deposited in the thin detector versus the energy in the back
detector. In (B) we have made the geometrical correction described in 3.4
to obtain a better resolution.
thus deposit a different amount of energy even if they are identical. This is
why we can not distinguish between the hydrogen isotopes in 3.4 (A).
Let us assume that the energy loss is constant through the front detector.
The energy deposited in the front detector will then be proportional to the
thickness,
dE
∆E = · t. (3.2)
dx
20 CHAPTER 3. EXPERIMENTAL TECHNIQUE
dE dE t ∆E 0
∆E = · teff = · = , (3.3)
dx dx cos θ cos θ
where ∆E 0 is the corresponding energy loss if the particle had entered
the front detector orthogonally. With this correction, we get an improved
∆E − E spectrum.
The remaining energy is deposited in the back detector, and the total
correction is
dE 0 = dE · cos θ
E 0 = E + dE · (1 − cos θ), (3.4)
where E is the measured energy in the back detector and dE is the energy
measured in the front detector and θ is the angle between the particle and
the normal of the detector. In Figure 3.4(B) we see an example of an with
this correction. Notice how the three particles in the lower regions are much
better resolved.
4
Beam diagnostics
4.1 Introduction
In experimental work with radioactive ion beams (RIB) the beam is often
assumed to be ideal point-like and incident in the target at a well known
position. This is, however, often not the situation and the beam’s emittance
and displacement are nonzero. Minimizing the emittance of the beam is a
major challenge in the design and operations of accelerator facilities.
In the resulting analysis of the experimental data, and in particular
single event measurement, the energy is deduced based on the assumption
that we know the geometrical profile and energy distribution of the beam.
When this is not the case we introduce a bias in the analysis leading to an
unnecessary decrease in resolution of our excitation spectra.
To correct this, we propose a method to deduce the beam’s prop-
erties from coincidence measurements. The method is partly based on
[johansen2013]. It has the advantage that it is applicable directly on the
primary experiment of scientific interest. The method applies to reactions
of the type
A+a→B+b
where both particles are detected. This of course includes elastic scattering.
We start with a theoretical description of the method. This is followed
by a demonstration of the method on simulations of the reaction 9 Li(d, t)8 Li
using a 5 MeV beam incident on a deteronated carbon foil.
21
22 CHAPTER 4. BEAM DIAGNOSTICS
4.2 Theory
Firstly, we present a short review of the method described in [johansen2013].
Thereafter we describe the new method. Both methods are derived for a
general reaction of the form
A + a → B + b,
Old method
The azimuthal angle of the two outgoing particles are given as
y
tan(φ) = . (4.1)
x
In the center-of-mass frame, the two particles will move back to back,
due to momentum conservation, i.e.
|φb − φB | = π. (4.2)
This is, however, unaffected by the transformation along the z-axis, and
hence, combining (4.1) and (4.2) we get the relation
yb yB
= ,
xb xB
which holds as long as the beam was incident in (0,0). If this is not the
case, but it is incident in (xA , yA ) we get the following relation
yb − yA yB − yA
= .
xb − xA xB − xA
We now assume that the offset is as small as possible, that is we need
to minimize x2A + yA 2 . This results in the following two relations (note that
x b yB − x B yb
xA = (yB − yy )
(yb − yB )2 + (xb − xB )2
x b yB − x B yb
yA = (xb − xB )
(yb − yB )2 + (xb − xB )2
New method
In the new method we can calculate the momentum of the beam particles
from momentum conservation in the LAB frame (the target, b, is at rest)
(xi − xA , yi − yA , zi − zA )
v̂i = p
(xi − xA )2 + (zi − zA )2 + (zi − zA )2
The energy of the incoming beam is given as
p2A
EA = (4.3)
2mA
1
(2EB mB +2Eb mb
= 2mA (4.4)
+2 4EB mB Eb mb v̂B · v̂b ).
p
EA = EB + Eb − Q. (4.5)
Subtracting (4.5) from (4.4) should thus equal zero. We can now do a
numerical minimization to find the position of the beam that best full fills
this criterion.
Angle determination
One we have estimated the displacement of the beam, we can estimate the
angle, since they are given as
q
p2A,x + p2A,y
tan(θA ) = (4.6)
pA,z
p2A,y
tan(φA ) = (4.7)
pA,x
In the analysis we will not use these angles, but the angles projected
onto the x an y-axes
θx = θA cos(φA ) (4.8)
θy = θA sin(φA ) (4.9)
4.4 Analysis
To apply the method we iterate over all events, check for coincidence and
find the numerical solution to
p2A
= EB + Eb − Q. (4.10)
2mA
26 CHAPTER 4. BEAM DIAGNOSTICS
This gives the position of the incident beam and thus the incoming
momentum. From this we calculate the corresponding angles, using (4.8)
and (4.9). We will firstly look at an ideal beam and restrict us to the
position estimate in order to learn a important feature of the method.
With this knowlegde we can continue to look at an angled beam. In real
application we will experience both displaced and angled beams, so we will
afterwards look at an combination of the both displacement and angling.
To assess the geometrical dependencies we will then look some of the factors
that determine the accuracy. Finally we will discuss energy estimation
distribution.
4.4. ANALYSIS 27
20 6
20 12 15
5
15 10
10
4
10 5
y[mm]
8
5 0 3
y[mm]
0 6 −5
2
−5 − 10
4
1
− 10 − 15
2
− 15 − 20 0
−20 − 15 −10 −5 0 5 10 15 20
x[mm]
− 20 0
−20 − 15 −10 −5 0 5 10 15 20
x[mm]
Figure 4.3: The total po-
Figure 4.2: The total po- sition distribution for all
sition distribution for all concidence measurements
concidence measurements. with a solution to (4.10).
500 X projection
400
Y projection
300
N
200
100
0
−20 − 15 −10 −5 0 5 10 15 20
[mm]
Beam position
We apply the method to a simulation of an ideal beam and the previous
described setup. This position estimates are shown in Figure 4.2. Firstly,
we notice that the setup structure is really pronounced. Secondly we notice
that the distribution is broad. It extends to approximately 15 mm in both
directions, even though our target only is 10 mm in both directions. This
can, however, be enhanced by noticing that (4.10) is quadratic in pA . There
may thus not be a solution or we can get two solutions. If there is no solution
we ignore the measurement, since it can not reproduce our conditions. If
we have two solutions we take the one closest to the z-axis, assuming we
are as close to the z-axis as possible.
If we disregard the coincidences that yields no solution to (4.10) we get
the plot in Figure 4.3. We removed many of the outer estimates, confining
the estimates within approximately 10 mm. In Figure 4.4 are the projections
on the x-axis and y-axis shown respectively. The distributions are centered
around (0 mm, 0 mm) but the distributions are so broad that it is hard to
tell whether the centers of the distributions are shifted slightly.
Let us consider what is actually happening in the reaction. Each reaction
spans a reaction plane, defined by three points: the reaction position (i.e.
the beam position) and the two detected particle positions. The method
estimates one of these points, namely the reaction position. Ideally we
should take the intersection of all these planes in the target plane. This is,
however, not feasible since we do not have an infinite resolution. Instead we
average over ensembles of measurements. We do this by defining specific
planes of interest, namely the primary axes of the detector pairs, and make
an averaged estimate from each of these. The intersection in the target
plane will then be our estimate of the beam’s position.
More specifically, we plot the position estimates from the method for
each detector pair separately. This is shown in Figure 4.5. We see that we
get the best estimate along the secondary axis of the two detectors. The
reason for this is twofold. It is partly because of the detectors geometry.
We have "more detector" along the primary axis than perpendicular to it.
The second reason is the method itself. We have much more freedom to
4.4. ANALYSIS 29
match momenta in the reaction plane than perpendicular to it. This is due
to the minimization criterion. A projection onto the primary axis and the
secondary axis clearly show this effect, see Figure 4.6.
To make an averaged estimate over the ensemble, we fit a straight line
with a fixed slope (the primary axis) for each of the pairs. This means we
only estimate the displacement along the secondary axes. This line is our
first estimate; the reaction took place somewhere on that line in the targets
xy-plane. If we plot each of these lines we see that their intersection is very
close to origin, see Figure 4.7. We can thus estimate the position of the
beam as the intersection point by minimizing the sum of the square of the
distances to each line, yielding (0.02 mm,0.01 mm).
If we displace the beam to be incident in (1 mm, 1 mm) we get the result-
ing lines in Figure 4.8. The displacement is estimated to be (0.99 mm,0.98 mm).
The method thus give a good estimate of the displacement.
Angled beam
We now turn to an angled beam. As described we can calculate the angle of
the beam from the estimated beam position and conservation of momentum.
For this analysis we also isolate each detector pair. Figure 4.10 shows
the angle spectrum from the detector pairs of a beam angled θ = 50 mrad
and φ = 45◦ , i.e. θx = 50 mrad · cos(45◦ ) = 35.3 mrad and θy = 50 mrad ·
sin(45◦ ) = 35.3 mrad. We notice that many counts are in (0,0). The method
are thus insensitive to coincidences angled in the inferred reaction plane, as
we also saw in the position estimate. This means we must look at the angle
relative to the reaction plane, i.e. the angle with the local x-axis, θx .
In Figure 4.11 are the projections onto the primary and secondary axes
shown. We see the y-projections are mainly peaked in zero, whereas the
x-projections show the same behaviour as with the displacement analysis.
That is, they are shifted and we can once again fit a line. This time it is,
however, the axis orthogonal to the primary axis. We also note that the
distributions are very broad. Gaussian fits to the x-axis give a σ in the order
of ≈ 45 mrad. This is the order of error we expect due to the resolution of
our detectors. This will be discussed in more detail later. We can, however,
30 CHAPTER 4. BEAM DIAGNOSTICS
Det1 Det2
15 7 15 6
6
10 10 5
5
5 5 4
4
y [mm]
y [mm]
0 0 3
3
−5 −5 2
2
−10 −10 1
1
−15 0 −15 0
−15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15
x [mm] x [mm]
Det3 Det4
15 6 15 6
10 5 10 5
5 4 5 4
y [mm]
y [mm]
0 3 0 3
−5 2 −5 2
−10 1 −10 1
−15 0 −15 0
−15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15
x [mm] x [mm]
Det1 Det2
200
200
X projection X projection
180
180
160
160
Y projection 140
Y projection
140
120
120
N
N
100 100
80 80
60 60
40 40
20 20
0 0
−15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15
x [mm] x [mm]
Det3 Det4
200
X projection 180 X projection
180
160
160
Y projection 140 Y projection
140
120
120
100
N
100
80
80
60 60
40 40
20 20
0 0
−15 −10 −5 0 5 10 15 −15 −10 −5 0 5 10 15
x [mm] x [mm]
3 3
2 2
1 1
y [mm]
y [mm]
0 0
−1 −1
−2 −2
−3 −3
−4 −4
−4 −3 −2 −1 0 1 2 3 4 −4 −3 −2 −1 0 1 2 3 4
x [mm] x [mm]
60
40
θ y [mrad]
20
− 20
− 40
− 60
− 80
− 100
−100 −80 −60 −40 −20 0 20 40 60 80 100
θ x
[mrad]
Det1 Det2
100 100 35
35
80 80
30
60 30 60
40 40 25
25
20 20
θ y [mrad]
θ y [mrad]
20
20
0 0
15
−20 15 −20
−40 −40 10
10
−60 −60
5 5
−80 −80
−100 0 −100 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x [mrad] θ x [mrad]
Det3 Det4
100 100 40
40
80 80
35
35
60 60
30 30
40 40
25 25
20 20
θ y [mrad]
θ y [mrad]
0 20 0 20
−20 −20
15 15
−40 −40
10 10
−60 −60
5 5
−80 −80
−100 0 −100 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x [mrad] θ x [mrad]
Figure 4.10: The angle distributions from the new method. There are many
counts in (0,0), meaning the method is rather insensitive to coincidences
angled in the reaction plane. This means we should use the angle relative to
the inferred reaction plane, i.e. θx .
34 CHAPTER 4. BEAM DIAGNOSTICS
Det1 Det2
250 400
X projection X projection
350
200
300
Y projection Y projection
250
150
N
N
200
100
150
100
50
50
0 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x [mrad] θ x [mrad]
Det3 Det4
300 350
X projection X projection
250 300
Y projection Y projection
250
200
200
N
150
150
100
100
50
50
0 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x [mrad] θ x [mrad]
Figure 4.11: Projection of the angle distributions from the new method. The
x-projection shifts, and can thus be used to estimate the total angle of the
beam.
4.4. ANALYSIS 35
6000
µ=44.9MeV 1800 µ=44.9MeV
σ =0.0615MeV σ =0.255MeV
1600
5000
1400
1200
4000
1000
N
N
3000
800
2000 600
400
1000
200
0 0
36 38 40 42 44 46 48 50 52 54 36 38 40 42 44 46 48 50 52 54
E[MeV] E[MeV]
µ=44.9MeV
500 σ =0.979MeV
400
300
N
200
100
0
36 38 40 42 44 46 48 50 52 54
E[MeV]
(c) σ = 1 MeV
Figure 4.12: The energy estimates for a) a point beam b) a gaussian energy
distribution with σ = 0.25 MeV and c) a gaussian energy distribution with
σ = 1 MeV
36 CHAPTER 4. BEAM DIAGNOSTICS
improve these distribution by using the old method. If we use this instead
we get the projections show in Figure 4.13.
We see that our x-projections are more well defined. If we use these
spectra instead, fit straight lines and estimate the intersection we get an
estimate of (32.8 mrad, 33.0 mrad), see Figure 4.9. We underestimate slightly,
but we are close to the expected angling.
Broad beam
Typically a beam is also spatially distributed. The exact shape is due to
various accelerator effects and therefore very complex. We will, however,
simulate a simple spatial distribution to see how our method responds.
In Figure 4.16 we see the distribution of one detector pair using a beam
4.4. ANALYSIS 37
Det1 Det2
800
120 X projection X projection
700
100
Y projection 600 Y projection
80 500
N
N
400
60
300
40
200
20
100
0 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x [mrad] θ x [mrad]
Det3 Det4
140 800
X projection X projection
700
120
400
60
300
40
200
20 100
0 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x [mrad] θ x [mrad]
Figure 4.13: The projection of the angle distributions using the old method.
We see that the x-projection shifts and we can thus use that for our angle
estimate.
38 CHAPTER 4. BEAM DIAGNOSTICS
Figure 4.14: We can look a narrower slices by only considering the center
half (the red and green area) or the center fourth (the green area).
with a radius of 2.5 mm. The position distributions get slightly broader on
the secondary axis, and the angle distribution is slightly elongated on the
primary axis. We can thus not quantify the beams spatial distribution. It
does, however, not affect the results of the position and angle estimates. We
can thus still estimate the primary component of the beam.
Position estimate
100
Angle estimate
4
80
3
60
2
40
1
θ y [mrad]
20
y [mm]
0 0
− 20
−1
− 40
−2
− 60
−3
− 80
−4 − 100
−4 −3 −2 −1 0 1 2 3 4 −100 −80 −60 −40 −20 0 20 40 60 80 100
x [mm] θ x
[mrad]
Figure 4.15: Position and angle estimate of a beam which was both displaced
and angled. We do not recreate the correct values. The method are thus
biased when both effect are present.
most sensitive secondary axis when estimating position and angles. It thus
seems logical to make the analysed pairs "thinner". This can be done by
looking at slices of the segmented detectors. That is, we might only look
at the center strips of the surrounding W1’s in the setup. In this way we
should narrow the spread in the estimates from the coincidences.
Another limiting factor is the resolution of the detectors. We can only
detector particles within one pixel. This gives a positional resolution in the
order of the pitch and an angular resolution in the order of 2 · tan−1 ( p/2
d ),
where p is the pitch and d is the distance from the target. For the W1’s in
our setup the positional resolution is ≈ 3.1 mm and the angular resolution
is ≈ 100 mrad. By using detectors with a better resolution, we thus also
expect to lower the spread in the estimates.
To investigate the effect of looking at narrower slices, we perform the
analysis only using the center half and the center quarter of the back strips
of the surrounding detectors as sketched in Figure 4.14. To investigate the
effect or better resolution, this is done for both W1 and an identical setup
40 CHAPTER 4. BEAM DIAGNOSTICS
Det1 Det1
15 7 15 6
6
10 10 5
5
5 5 4
y [mm]
y [mm]
0 0 3
3
−5 −5 2
2
− 10 − 10 1
1
− 15 0 − 15 0
− 15 −10 −5 0 5 10 15 − 15 −10 −5 0 5 10 15
x [mm] x [mm]
Det1 Det1
100 100 100
80 120 80
60 60 80
100
40 40
80
θ y [mrad]
θ y [mrad]
20 20 60
0 0
60
− 20 − 20 40
− 40 40 − 40
− 60 − 60 20
20
− 80 − 80
− 100 0 − 100 0
−100 −80 −60 −40 −20 0 20 40 60 80 100 −100 −80 −60 −40 −20 0 20 40 60 80 100
θ x
[mrad] θ x
[mrad]
Figure 4.16: Position and angle distributions for a point beam and a spatial
uniformly distributed beam with a radius of 2.5 mm.
with BB7 1 detectors instead. The results are given in Table 4.1.
We see that an increase in resolution is to be preferred since it decreased
the spread in both position and angle. Looking a narrower slices decreases
the position, but not as much. We do, however, see that the spread in angle
increases. This means we need to do slices in the other direction if we want
it to narrow the angle interval.
Energy distribution
We can estimate the energy distribution of the beam by calculating back-
wards from our position and angle estimate. For each coincidence we
calculate the energy using (4.4). In Figure 4.12 we see the results for a
1
A square DSSSD detector with 2 mm pitch and 32x32 strips.
4.5. CONCLUSION 41
µ σ
Point 44.9 MeV 0.06 MeV
σ = 250 keV 44.9 MeV 0.26 MeV
σ = 1 MeV 44.9 MeV 0.98 MeV
Table 4.2: The deduced energy distributions of three beams with different
energy distributions.
point beam, a gaussian shaped energy distribution with σ = 250 keV and a
gaussian shaped energy distribution σ = 1 MeV, respectively. Applying a
gaussian fit get the results in Table 4.2. There are clear very close to the
simulated values. We can thus accurately determine the beam energy from
coincidence measurements.
4.5 Conclusion
The proposed method can be used to estimate a beam’s properties and in
particular estimate a displacement and angling of a beam using coincidence
measurements in segmented detectors. The method can not, however,
distinguish between the effects. One can, however, use simulations of the
42 CHAPTER 4. BEAM DIAGNOSTICS
setup to find the beam that yields same results as the analysis of the actual
experiment. In order to get the best intersection estimates we need multiple
different primary axes. The setup should thus cover multiple directions in
the azimuthal angle. Four detectors are for instance sufficient.
5
AUSAlib
5.1 Geometry
A core feature of AUSAlib is the generalization of geometrical properties
of the detectors. When a particle hits a double sided detector, the front
43
44 CHAPTER 5. AUSALIB
3 2 1 2 3 ... Orientation
1
4 1
2
n
tio
ta
ri en 3
O
..
...
.
Normal Normal
0 Center Center
1
2
3
4
...
strip and the back strip is recorded in the data stream. This translate to a
position in space, but the transformation depends on the specific geometry
of the DSSD. In Figure 5.1 we see to very different geometries that can
still be described orthogonal front and back strips. The interface in the
analysis is thus the same. Given a front strip and a back strip, calculate a
position. We can thus have a single superclass that exposes the interface
for a DSSD. Concrete implementations, such as the round DSSD, can then
specify the exact transformation. For the experiments in this thesis we can
use the square DSSD to describe both the W1 and the BB7, whereas we
can describe the S3 with the round DSSD.
Even more general is the position and orintation in space. This is
required for all types of detectors, segmented or unsegmented, in order to
place them in a global coordinate system and rotate them correctly.
This leads to a class hierachy as shown in Figure 5.2.
This approach has to main advantages. Firstly the data is decoupled
from the geometry in the sense that a hit only specify the segments that were
hit, i.e. the strips. Then it is up to the concrete detector implementation to
5.1. GEOMETRY 45
Detector
SingleSidedDetector DoubleSidedDetector
calculate what position this corresponds to. Secondly is it easy to add new
detectors. It just a matter of implementing the correct transformation, and
then it can be used in an existing analysis through the common interface.
This favors code reuse and modifications are done by addition, keeping the
codebase backward compatible.
Setup file
For a given experiment the position and type of the different detectors
must be specified to the analysis. To ease this process, AUSAlib uses a
socalled setup file, that encompass all the geometry specific details for each
detector, as well as which detectors that are meant to be a telescope. An
example of the description of a detector is shown in Figure 5.3. Each
detector has a name. The "file" contains details such as detector type
(from the detector hierachy), number of strip, deadlayer thickness, etc. The
"calibration" file contains calibrations for each strip. The "position",
"normal" and "orientation" specify the position and orientation in the
global coordinate system. The data mapping to the unpacked files are
described in "frontmapping" and "backmapping".
Once such a file is written, typically during the setup of an experiment,
it can be used throughout all analysis and even the online analysis. AUSAlib
simply loads this file making it easy to makes changes without having to
46 CHAPTER 5. AUSALIB
1 {
2 "name ": "Det1",
3 "file ": "DSSD1.json",
4 " calibration ": "Det1.cal",
5 " position ": {"x": "-4mm", "y": " -42.0 mm", "z": "39.0 mm"},
6 " normal ": {"x": "0.0 mm", "y": "1.0 mm", "z": "0.0 mm"},
7 " orientation ": {"x": "0.0 mm", "y": "0.0 mm", "z": "1.0 mm"},
8 " frontMapping ": {
9 " multiplicity ": "DET1F",
10 " segment ": " DET1FI ",
11 "adc ": " DET1F_E ",
12 "tdc ": " DET1F_T "
13 },
14 " backMapping ": {
15 " multiplicity ": "DET1B",
16 " segment ": " DET1BI ",
17 "adc ": " DET1B_E ",
18 "tdc ": " DET1B_T "
19 }
20 }
Custom analysis
Calibrator .cal
Unpacking
The DAQ saves the files in a GSI list mode format (.lmd). The first step
is to convert these files to ROOT files that can be read into the rest of the
AUSAlib pipeline. This is done with ucesb. The user is required to make a
precise specification of which modules are used, how the data is layed out in
the data stream as well as the mapping from the individual ADC channels
to the detector segments. Typically this is done during the setup of the
DAQ, such that it can be used during the experiment and in the subsequent
analysis. One major strength of ucesb is that it checks that every single
48 CHAPTER 5. AUSALIB
bit in the data is consistent with the data layout of the modules. If the file
unpacks, the data is thus not corrupt.
Energy calibration
The first task after the unpacking is to energy calibrate the detectors. We use
an α-source with well defined peaks. In the ISOLDE experiments we used
a source with 244 Cm, 241 Am, 239 Pu and 148 Gd which all give well defined
nearly monoenergetic peaks. The source is placed in a well known position.
The energy response in a silicon detector is highly linear [REF??] and we
can now fit a linear relationship between the ADC channel corresponding
to the centroid of the peak and the well known tabulated energy. There are,
however, a few small corrections we need to make to get a better calibration.
Deadlayer
There is small deadlayer on both sides of a silicon detector. This is a small
layer of inactive material. The α-particle loses energy, but the created
electron-hole pairs are not collected. This means the measured energy is
slightly lower than the actual energy
dE ∆t
E 0 = Eα − , (5.1)
dx cos θ
where ∆t is the thickness of the deadlayer and θ is the angle shown in
Figure 3.3 and Eα is the tabulated α energy. The dead layer is assumed
to be so small that the energy loss is constant. It is thus important that
the geometry of the detector as well as the source position is known, so θ is
calculated properly.
Shift calculation
Each segment is calibrated individually, but extends over many angles. The
measured energies should therefore be averaged over the full surface of the
segment. Since our positional resolution are limited to the pixels the average
5.3. ANALYSIS PIPELINE 49
can be approximated with a weighted average with the solid angle of the
pixels a weight
X dE ∆t Ωpixel
E 0 = Eα − , (5.2)
pixel
dx cos θpixel Ωstrip
Sorting
The sorting step has two main jobs, and is performed by a general purpose
tool, the Sorter. Firstly it applies the calibrations. These can come from
the Calibrator as described above, or other means of calibration. Only
linear calibrations are, however, implemented.
Secondly it performs matching to determine which pixels was hit by a
particle. When a particle hits a DSSD a signal is detected in both sides.
If there is only one hit in the back side and one hit in the front side they
must match. If, however, there are more than one hit in one or both sides
we need to determine which pairs that match. There might also be noise
that only shows up in one side.
The Sorter tool implement a greedy algorithm that computes the dif-
ference between all pairs. The pair with the smallest difference is selected
as the first pair and those hits are excluded from the pool of candidates.
The next pair is now the pair with the smallest energy difference from the
remaining hits. This continues until there are no more pairs with an energy
difference between a certain thresholds.
50 CHAPTER 5. AUSALIB
The majority of hits will hit a strip in both the front and the back, but
a minor part will hit in an interstrip region. This results in a sharing of the
energy between two strips. It is also possible that multiple particles hit the
same front strip but different backstrip and we see a summing of the energies.
Neither of these situations can be handled by the current algorithm, but they
can mostly be recovered by a more sophisticated algorithm that considers
all pairs and tries to reconstruct both normal, sharing and summing.
Identification
After the sorting, the energy and the position of the hits are determined.
This enables the next step, which is trying to identify what particle a
given hits corresponds to. There are different techniques to perform this
identification. When the detectors are organized as telescopes we can use
the ∆E − E method to seperate the different isotopes. If the energy is
too low to penetrate the ∆E detector, it may be possible to exclude some
particle types simply based on how much energy they lost in the detector.
This is, however, not unambigious, and a hit can have several candidate
identifications. Based on each of these identifications, the energy is corrected
for energyloss in deadlayers and the target.
The identification and energyloss is performed by another tool, the
Identifier. It requires details on the target and the particles too look for,
which are supplied either as a command line argument or throguh simple
text files.
Eventbuilding
From the files with identified particles, the eventbuilding stage attempts to
reconstruct acutal physical events from the detected hits. It requires one or
more final states to look for or reconstruct. If an event contains an identified
deuteron as well as hit that is consistent with a 9 Li, the four-momentum
for both are computed based on the position of the hit and the energy in
the detector. If only a deuteron is detected it can try to reconstruct the
5.3. ANALYSIS PIPELINE 51
Data acqusition
53
54 CHAPTER 6. DATA ACQUSITION
TDC trigger
TDCs
Busy
Trigger requests
Acquisition Gate
Discriminators ADCs
control Busy
Readout
VULOM 4b running the TRLO II firmware. The AcqC get trigger inputs
from all our input sources, i.e. the detectors. If the DAQ is ready, the
AcqC generates a MS that is sent to the digitizing modules. While they
are digitizing they assert a logic BUSY signal. As long as any module has
asserted BUSY, the AcqC can not accept a new trigger. The readout software
also requires time periods to check that all modules agree on the number of
events they have digitized. If they disagree, there is a fatal flaw somewhere
that needs to be addressed immidiately. During these times, the DAQ can
also not accept new triggers. Lastly, a module may need to be emptied
before it can store a new event. In this case the DAQ also needs to wait in
order to stay synchronized. We will refer to these times as dead time (DT).
The time the DAQ can accept a new trigger is the life time.
a fast rise time and a long exponential tail and is sent into a secondary
combined amplifier and signal discriminator. The signal is further amplified
and bandwith limited, producing a signal of 1 µs − 2 µs1 . The discriminator
produces logic signal if the shaped signal passes a certain condition. We use
both constant fraction and simple threshold modules. For each channel a
shaped energy signal, a logical discriminator signal is produces. These are
sent to the digitization modules. Furthermore is a logical OR from all the
channels produced, and this is used by the trigger logic.
ADC
To measure the energy we use peak sensing Analog-to-Digital coverters.
They charge a capacator internally to capture the full peak, but then during
of this integration period is up to the experiment. This was done by the
AcqC that sent a logical GATE signal. After this periode the module converts
the value and stores in a buffer. During this time it asserts its BUSY to tell
the AcqC to wait. If the buffer gets full it can not and also asserts BUSY
and waits to be emptied. The GATE should be wide enough to capture the
peak of the signal. For charged particles in a silicon detector the width of
the shaped signals is a few µs, and the GATE was adjusted accordingly.
Both IS561A and IS561B was implemented with CAEN 7xx. They have
a 7.06 µs conversion time and 32 event buffer. In the IS561C they were
exchanged with MESYTEC ADC instead, since they had a faster conversion
time of 1.2 µs and significantly deeper buffers. Two different revisions were
used and the newer version could store 4096 events whereas the older version
could only store 1024 events in the buffer.
TDC
The leading edge from the discriminators is sent to Time-to-Digital converter
which timestamp the arrival time. For each channel it keeps the timestamp
from all leading edges it recieved within some time periode. When it
1
FiXme Fatal: Check det her
6.3. READOUT 57
recieves a trigger from the AcqC it saves all channels that fired within some
predefined time window.
6.3 Readout
The readout is responsible for reading the data from the modules, move
them over the network and save to it a physical disk. The DAQ was VME-
based in all cases meaning that all the modules that acquired data was
housed in a VME crate[ref] and the communication with the modules was
via the VME bus. To empty the modules in an orderly fashion, the readout
must communicate with the AcqC. The transport of the data as well as the
communication with AcqC is handled by available DAQ frameworks and
we used MBS[ref] for IS561A and drasi[ref] for IS561B and IS561C. They
require a piece of code from the user that have the following responsibilities
These tasks were handled by the Nustar Readout Library [ref] and the
readout code ran on Single Board Computer (SBC), located in the VME-
crate. More specifically it was a CES RIO4-8072 [ref].
6.4 Multicrate
As the number of detectors increases, so does the required number of ADC
channels. At one point a single VME-crate can not house enough modules
and we have two use one more. This introduces the complexity of having
58 CHAPTER 6. DATA ACQUSITION
Slave
Gate
Readout trigger ADC
Slave
Readout Busy
AcqC
Deadtime TDC
Trigger
Event Data Readout trigger Deadtime
builder Master start Slave busy
Gate
Readout trigger ADC
Master
Readout Busy
AcqC
Deadtime TDC
Trigger
Master
Discriminators
Trigger requests
two crates, two SBC and two AcqC to commincate. The single crate design,
however, scales nicely and it is rather easy to extend to two crates running
in a master/slave configuration.
Both the master and the slave crate is more or less identical to the
single crate setup. Figure 6.3 shows an illustration of such a system. The
triggers logic is now only handled by the master. The master propagates
the generated MASTER START to the slave crate, much like if was just a
module. Moreover is the readout trigger propagated. The slave adheres to
this, however, the master must respect the busy and DT from the slave. If
it generated MASTER START in while the slave crate was busy, the modules
would get out of sync.
The readout in each crate does not know about the other, and an event
builder therefore collects the events and merge them. The events are tagged
with a crate number.
This multicrate design was used in IS561C.
and the dead time ratio is the ratio of rejected triggers to the total incoming
triggers
Nacc
L= , (6.1)
Ntot
where Nacc is the accepted number of triggers and Ntot is the total incoming
number of triggers. These numbers were recorded by the AcqC for each
trigger input. In all three experiments this corresponded to one detector.
Throughout the analysis this is used to correct for dead time, by dividing
the actual count number in a given detector by L.
If a stream of Poisson distributed trigger the life time ratio can also be
evaluated analytically [ref]
1
L= , (6.2)
1 + fr ∆t
where fr is the request frequency and ∆t is the dead time per event. De-
pending on the operational mode of the readout, ∆t can be estimated.
Single cycle mode When the DAQ recieves a trigger it issues a MS. After
the GATE, the ADC’s converts the data and the readout empties the buffer
and check synchronicity. The main contributions to the total dead times is
the sum of the gate, the conversion time and the readout time. On top of
this, there is some readout overhead that adds up to a significant amount.
Multi event mode The readout overhead can be reduced by using the
internal buffers in the modules. If we wait until the buffers are full we can
effectively split the readout overhead across many events. The deeper the
buffers, the bigger the advantage. The DAQ ran in this mode in IS561A
and IS561B.
Multi event
80
60
40
20
0
0 10 20 30 40 50 60 70
Trigger accept frequency, f a [kHz]
Figure 6.4: MMh
the buffers as empty as possible. If the events come in at a slower pace than
the readout can empty it from the buffer, a continous readout effectively
removes the dead time associated with readout, including the overhead.
This was implemented by M. Munch after the idea of H. T. Johansson and
was used in IS561C.
In particular the last scheme gives large improvements of L, and in
particular for large trigger frequencies. A more detailed investigation of this
was carried out and published[ref]. A simplified figure from the paper is
showed in Figure 6.4. The three different schemes are calculated with 6.2
for a crate with six CAEN V785s and a RIO4 SCB. The reader is referred
to the paper for the technical details. The data points are generated with a
poisson distributed pulser. The shadow reaout mode performs significantly
better, especially for large request frequencies. Close to 55 kHz the curve
drops - this is data bandwith limit. The DAQ can simply not move data
faster. For a typical ISOLDE reaction experiment the rate is typically much
lower. At 20 kHz the accepted trigger rate is increased by roughly 30 %,
which is not an unrealistic rate in a down stream detector where Rutherford
scattering dominates.
6.6. PULSED BEAM 61
105
104
Counts/1µs
103
102 12.1 µs
101
100
0
0 50 100 150 200 250 300
TEBIS [µs]
Figure 6.5: TEBIS spectrum from a down stream detector in IS561B. The
time period between the peaks corresponds to the effective DT of the DAQ.
The main contribution is the GATE and the conversion time.
to reduce the conversion time from 7 µs to 1.6 µs. The deeper buffers also
made sure the DAQ could accept a full EBIS release without having to be
emptied. To be sure that the modules were mostly empty at the begining
of an EBIS release, a readout trigger was produced at a fixed time interval
before the EBIS pulse. With a 30 Hz release cycle there were approximately
30 ms between each release which is plenty of time to empty the buffers.
With the lower intensity 9 Li beam there were no major issues with the
dead time. In the detectors of interest we had a dead time ratio in the order
of 5 %-10 % in IS561A and it was reduced to 1 %-2 % in IS561C.
Relay
The relay is an ucesb instance that connects to the readout data stream
from and fan the data out to two streams. One data stream is in "weak-hold"
mode, which means that if a client can not keep up with the data it blocks
the acqusition. This is used by the file taking.
The second stream is in "no-hold"-mode and does not require the clients
to consume all data. This is used for the online analysis.
The relay have the additional feature that it validates the data on the
fly. This means we do not have to save data and unpack it to detect corrupt
data.
6.8. DAQC 63
File taking
The file taking is another ucesb instance that connects to the relay and
writes the data to a file. It saves the data in the GSI list mode (.lmd)
format [ref]. It is furthermore compressed with gzip.
To ease the subsequent analysis the files are written in chunks of typically
100 MB. In case of a failure it also makes recovery easier, since all the data
from a measurement are not in the possibly corrupted file.
Online analysis
The online analysis presents the incoming data to the operators. This is
handled by the go4cesb project, which combines ucesb, GO4 and AUSAlib.
GO4 provides an advanced user interface that lets the operator interact with
different spectra produced by the analysis routines. The data is piped from
the relay to a ucesb process which performs the unpacking of the data.
This process is spawned by go4cesb and the unpacked data is subsequently
piped back into GO4 as an event stream.
This design has two main advantages. The unpacker only has to be
written once for both offline analysis and online analysis. Furthermore does
the unpacked data have the same AUSAlib-friendly format, so it is trivial
to apply AUSAlib analysis routines in the online analysis.
The simplest analysis includes mapping the ADC channels to detector
strips, applying calibrations and matching. Since this is built in to AUSAlib,
this can automatically be applied to the unpacked data. More sophisticated
analysis such as ∆E − E spectra and kinematic curves can now be imple-
mented by the user, but since everything is in the AUSAlib format, it is
easy to reuse such analysis components acrosse different experiments. Such
analysis components can be turned on and off in a simple configuration file.
6.8 DaqC
To have a fully functional DAQ all the described components must be running
and continously monitored. Since a typical experiment spans several days,
64 CHAPTER 6. DATA ACQUSITION
C
A
D
B
Figure 6.6: DAQC
non-experts should also be able to start, operate and even recover the DAQ.
To ease this, we have developed DaqC.
Internally it consists of interfaces start, stop and monitor all the dif-
ferent services. It implements the correct startup order and keeps track
of dependencies between the different services. When started the startup
sequence is executed. When all the services is up and running, the internal
watchdog monitors each service to check that is up and healthy. If a service
crashes it tries to revive it.
When running, an operator can interact via a simple web interface,
as shown in Figure 6.6. It provides controls to all services managed by
DAQC as well as simple status report. Panel (A) shows the status shown
6.9. RUNDB 65
for the different componenets. In panel (B) is a control panel for the trigger
logic. The operator can quickly include or exclude detectors in the global
trigger between measurements. In (C) is a simple overview of previous
measurements shown.
Panel (D) provide a button to start measurements, i.e. start the file
taking. When a new measurement is started, the operator is prompted with
a form to fill in a run sheet with information about target, beam, operators
on shift, etc. This is saved in log file, and effectively compiles a detailed
measurement log. Furthermore, when a measurement is started and stopped,
DaqC dumps the configuration of all the ADCs, TDCs, amplifiers as well as
other available meta data. This is saved together with the data files. Both
the data and the log files are continuously backed up to a remote storage
facility in case of hardware failure.
I started the development of DaqC together with M. Munch after IS561A.
It has been used for IS561B, IS561C and multiple other ISOLDE experiments
and it is the control program for the DAQ at the 5 MeV accelerator at Aarhus
University.
6.9 RunDB
The information from the run sheet is essential for the subsequent analysis of
the data, both during the experiment and after the experiment. Traditionally
this has been written by hand in a paper log by the operators, when a
measurement has been started and stopped. With the implementation of
DaqC, it was natural to compile this data in a database. This led to the
development of RunDB, which stores the meta data and file location of the
data in a SQL database.
RunDB thus provides an easy way of browsing through the different
measurement configuration and in particular the time spent. The automated
upload of the run sheet has moreover reduced the number of errors in the
log such as wrong time stamps.
A web interface makes it easy to search for specific targets, time periods
etc. It is shown in Figure 6.7.
66 CHAPTER 6. DATA ACQUSITION
Acceptance
NR = NI nt σ, (7.1)
69
70 CHAPTER 7. ACCEPTANCE
d2 Nr (θ, φ, E) d2 σ(θ, φ, E)
= NI nt , (7.2)
dΩdE dΩdE
and the total number of detected particles can be calculated by integrating
over the solid angle covered by our detectors, ΩS , and the energy range that
is allowed by the kinematics of our experiment Er
d2 σ(θ, φ, E) 0 0
Z Z
NR = NI nt dΩ dE . (7.3)
Er ΩS dΩdE
Since we want to calculate the cross section we rearrange 7.2 to get
d2 σ(θ, φ, E) 1 d2 Nr (θ, φ, E)
= . (7.4)
dΩdE NI nt dΩdE
In an actual experiment we can not measure infitesimal range in angle
and energy, and we therefore bin our data into a suitable bin size, usually
depending on our amount of statistics. Choosing too small bins, and we get
large uncertainties. Choosing too big, we average out possible interesting
features. In either way, we can approximate the cross section for each point
in a given bin by the average
d2 σ(θ, φ, E) 1 Nr (θ, φ, E)
≈ , (7.5)
dΩdE NI nt ∆Ω∆E
where Nre is the detected numbers in a given bin, ∆Ω is the solid angle
corresponding to the bin and ∆E is the energy range covered by the bin.
In reality we do not have full coverage of all the bins. Some bins will be
outside of detectors and some will suppressed due to kinematics. This will
alter ∆Ω and ∆E corresponding to a particular Nr , but proportionally. We
can thus write
7.1 MC
RAMBLE ABOUT WHY IT IS DIFFICULT
To get a reliable estimate of the cross section, each of the values on the
right side must be evaluated. NI and nt does not depend of the geometric
setup of our detectors or the reference frame we choose to investigate. Nre ,
∆Ω and ∆E are, however, more difficult. First of all these depends on the
geometry of the setup. The particular geometry of the detectors dictates
how large a solid angle it covers. This may not be the same in all angles,
and we thus have to take an angular dependence of the solid angle into
account. Since the detected particles scales with the solid angle, so does
Nre . If we try to do our analysis in the CM-frame the solid angles need to
be transformed correspondingly.
Moreover, the effects of a broad divergent beam, finite angular resolution
in the detectors and the like, will change the effective solid angle. These
effects are not easy to incorporate in an analytic way, however, we can solve
it with a Monte Carlo simulation.
Consider a simulation with a uniform distribution. We should see a
constant fraction of the the simulated particles in each bin, proportional to
the size of the bin
Ns (θ, E) ∆Ω∆E
= , (7.7)
Ns 4πEr
where Ns (θ, E) is the number of simulated particle in a specific bin, Ns
is the total number of simulated particles and Er is the energy range we
sample uniformly from.
If we also include our setup in the simulation and keep track of how
many particles we detect, we will instead get a measure of the effective bin
size
Nsd (θ, E) ¯ ∆E
∆Ω ¯
= , (7.8)
Ns 4πEr
where Nsd (θ, E) is now the detected amount of a the simulated reactions
and ∆Ω¯ ∆E
¯ is the effective bin size. It is, however, the same ∆Ω
¯ ∆E
¯ as in
¯ ¯
7.6. This means we can isolate ∆Ω∆E in 7.8 and substitute into 7.6 and
get
72 CHAPTER 7. ACCEPTANCE
where the sum is over all energy bins and we have assumed that each bin
have the same width ∆E.
Similarily for the single differential energy cross section we get
1 Ns X Nrd (θ, φ, E) 1
!
dσ(E)
≈ ∆Ω , (7.11)
dE nt NI i Ns (θ, φ, E)
d 4πEr
Bound states
We can do a similar analysis to get the single differential angular cross section
of bound states. In this case we simulate a single energy, corresponding to
the resonance energy, and the energy dependence ∆E/Er in 7.8 vanishes.
This gives us
!2
dσR e2 zZ 1
(θ, φ) =
dΩ 4π0 E sin (θ/2)
2
α~czZ 1 2
=
4E sin2 (θ/2)
1.44MeV · fm
2
zZ
= (7.13)
4E sin2 (θ/2)
where z is the charge of the beam, Z is the charge of the target and E is
the available energy in the CM system
mtarget
E = E LAB (7.14)
mtarget + mbeam
dσR
Z
NR = NI nt (θ, φ)dΩ0 (7.15)
Ωsetup dΩ
We can now use two different strategies to determine Ωsetup . The first
method is a MC methods like the one we used for the acceptance. This
method makes it easy to incorporate imperfect beams, etc. The second is a
simpler geometrical method, which is good to consistency check, but lacks
the sophistication we can achieve with a simulation.
Monte Carlo
Imagine we simulate Ns reactions. We distribute them accordingly to the
Rutherford formula in CM with the MC method and count how many we
detect, Nsd . Notice that it diverges when θ → 0, hence we only simulate in
the range [θmin , 180].
The ratio of detected scattered beam particles (analogous to 7.8) to
simulated particles is
We can rearrange to
Nd
Z θmin
dσR dσR
Z
(θ, φ)dΩ0 = s 2π (θ, φ)sinθdθ0 (7.18)
Ωsetup dΩ Ns 180 dΩ
Nr Ns 1
NI = (7.19)
2πnt Nsd 180
R θmin
dΩ (θ, φ)sinθdθ
dσR 0
Geometric
We can also calculate the solid angle of the individual pixels with a simple
geometrical approximation where we project the pixel onto the tangent
plane of the unit sphere and calculate the area on this plane. If the pixels
are small enough, this is a sufficiently good approximation.
7.2. RUTHERFORD ESTIMATION 75
2 2
r !
LAB 2 cos θ + 1− m1
m2 sin θ
dσR zZα~c
(θ, φ) = ·
2E
r
dΩ 2
sin4 θ 1− m1
m2 sin θ
We thus get
Z
dσR X dσ LAB
(θ, φ)dΩ0 ≈ R
(θpixel )Ωpixel (7.20)
Ωsetup dΩ pixels
dΩ
Experimental setups
77
9
IS367
79
80 CHAPTER 9. IS367
12000
10000
8000
∆E · cos θ [MeV]
6000
4000
2000
0
0 5000 10000 15000 20000 25000 30000 35000
E + ∆E · (1-cos θ) [MeV]
Energy calibration
The calibration of the detecters was done with a standard α source, as
described in Section ??. Each segment was calibrated individually and the
pads were calibrated with a source from the back side.
Since the energy of the particles were that stops in the pad, are up to
4 times higher than the highest calibration point, the calibrations must
be consistency checked at higher energies. Luckily, as we will see in the
following analysis, the kinematics and the ∆E − E spectra are consistent,
so we accept the energy calibrations.
Particle identification
Using the methods described in ??, we can identify the different particles
detected. The scattered 9 Li form both the heavy target and the primary
target does not have enough energy to penetrate the ∆E-detector. We can
82 CHAPTER 9. IS367
thus only identify these based on kinematics, which we will discuss as part
of the kinematic analysis.
In the right panel in Figure 9.1 we see the ∆E − E plot for the physics
runs. We clearly see protons, deuterons, tritons and α. There is also a small
contribution of 6 He. With proper gates, we can pick out the individual
particles.
Kinematics
The kinematic curves for protons, deuterons and tritons are show in the
upper panel in Figure 9.2. For the protons we see a clear component
from elastic scattering of protons. This is expected, since the target have
≈ 1 % protons. Since we do not know the exact amount, we instead did
a measurement of CH2 which we can subtract. Also notice that is well
seperated from the continuum expected from 9 Li (d,p)10 Li reactions.
The deuteron spectra have two well defined components, correspond to
the ground state and 1st excited state respectively.
The tritons have components from the three first excited states. At
larger excitation energies we exceed the neturon seperation energy and we
see a continuum. There a various channels that may contribute to this
continuum, but this has not been investigated in detail. The corresponding
excitation spectra, calculated with 2.8 are shown in the lower panel in Figure
9.2.
The width of the 9 Li ground state peak can be used to estimate the
resolution of our setup. A simple gaussian fit gives a resolution of FWHM=
551(4) keV.
15
10
0
0 10 20 30 40 50 60 70 80
0 10 20 30 40 50 60 70 80
0 10 20 30 40 50 60 70 80
θ LAB [deg]
800
900
700
800
600 500
500
400
400
100 300
300
200
200
100 100
0 0 0
-4 -2 0 2 -4 -2 0 2 4 -4 -2 0 2 4 6
Ex [MeV]
25
25000
d
9 Li
20
20000
15
E [MeV]
15000
E [MeV]
10
10000
5
5000
0 0
0 10 20 30 40 50 60 70 80 90 0 20 40 60 80
θCM [deg] θCM
the feature below the main curve. This is an artifact of a faulty strip in one
of the detectors.
We also measure some amount of coincident recoils from the light par-
ticles. An example is shown in the left panel in Figure 9.3. The spectra
here fits the curves nicely, and they are very clean when the coincidence
requirements have been made. Once again, we can use the kinematic region
as a gate.
Background subtraction
In the 10 Li spectrum we see a peak just below 0 followed by large tail at
negative excitation energies. These events are unphysical in the sense, that
they can not origin from the 9 Li (d,p)10 Li reaction. They must instead come
from various background contributions. The peak below 0 corresponds to 9 Li
(p,p)9 Li as we saw in the kinematic curve. Since we did a measurement with
both CH2 and C we can pipe these measurements through the same analysis
9.2. INTENSITY ESTIMATION 85
300 100
50
200 100
0
100
−50
0 0
-4 -2 0 2 -4 -2 0 2 -4 -2 0 2
Ex [MeV]
Figure 9.4: A: Excitations spectra for protons for the main target (CD2 )
and the two background targets (CH2 and C). B: The combined background
on top of the signal. C: The resulting background subtracted spectrum.
Nd Rd
Nd = = . (9.2)
N9 Li I
In the above ratio, we have assumed that the intensity did not change
during two adjecent measurements, which is generally not correct. In
particular will the super cycle at CERN determine how many proton bunches
that reaches ISOLDE. This changes on a timescale that is comparable to
the length of the runs. However, by looking at the distribution across all
intensity measurements, we may be able to treat this systematic error as a
statistical error.
In Figure 9.6 N is plotted as a function of measurement number. The
upper panel is the deuterons and the lower panel is for α. The median as
well as the 1σ interquartile range is shown. We use the median, since it is
more robust to outliers. In practice, however, the median turned out to be
very close to the weighted average.
9.2. INTENSITY ESTIMATION 87
The distribution does not seem to have a clear systematic trend. We can
make a simple χ2 test, to test is they are consistent with being from the same
mean with a gaussian distribution. For deuterons we get a pval = 0.048.
This is on edge of being consistent with coming from the same mean, but
due to the expected systematic errors we accept it, with a correspondingly
larger uncertainty on the integrated intensity.
The medians for the two particle types are respectively
Using the ratio we can calculate the integrated beam across all the
measurements, by simply counting scattered particles.
N9 Li = Np Nip , (9.3)
X
40
Beam intensity [kHz]
35
30
25
20
0.000010
0.000008
N
0.000006
0.000035
N
0.000030
IS561A
The second experiment in this series was carried out in 2016 and this was
the first experiment I took part in. This was the first experiment with the
HIE-ISOLDE upgrade. The full upgrade was not completed at this time, but
we were still able to get a significantly higher beam energy of 6.72 MeV/A.
This was requested to be able to measure the protons in the backward
directions, which has quite low energy due to the kinematic compression.
With 6.72 MeV/A the protons should reach 0.5 MeV-1.0 MeV, which should
be possible to measure. Therefore the setup was redesigned and in particular
included a telescope to cover the backward directions.
During the run we encountered problems both with beam alignment
and low yields from the ISOLDE part. This resulted in significantly lower
statistics than expected. The beam alignment problems also led us to steer
the beam around during the experiment, which in turn requires a more
careful analysis.
The experimental program for this experiment was very similar to the
older IS367. We had a deuterated polyethylene target as the main target.
For background measurements we used a pure carbon target and a normal
polyethylene target. For intensity measurements we used a gold target. A
table of the targets is shown in Table 10.1.
The operators used 12 C as a pilot beam to steer the beam into the
chamber. We were allowed to take some data with this beam, which turned
out to be very useful since the intensity is rather large, which means we can
91
92 CHAPTER 10. IS561A
10.1 Setup
The experimental setup for IS561A was updated compared to IS367. A
picture is shown in ??. The setup had detector coverage in both forward
and backward directions. In the forward direction (1) there was 5 telescopes
placed symmetrically around the beam axis in a pentagon structure. Each
of these were 60 µm thick and had 16x16 strips. Perpendicular on the beam
axis were a round DSSD with a thickness of 1 mm. The idea with this design
was inspired by the method discussed in Section ??. The thick DSSD should
capture the recoils and the 5 pentagon detectors would capture the light
ejectiles. With the high degree of symmetry we should be able to capture
the beam spot and beam direction. These detectors were mounted in a
3D-printed plastic structure, which provided good geometrical constraints
on the detector positions.
In the backward direction was a single telescope place (3). The DSSD
was 60 µm. The purpose of this detector was to measure the ejectiles in
the backward lab direction, which corresponds to the forward CM angles
due to the inverse kinematic. We expect there to be larger cross sections
in this area, and we should be able to get a better resolution due to the
higher angular resolution. Due to the kinematic compression the energy of
10.2. CALIBRATION AND BEAM PROPERTIES 93
10000
8000
∆E · cos θ [MeV]
6000
4000
2000
0
0 5000 10000 15000 20000 25000 30000 35000 40000 45000
E + ∆E · (1-cos θ) [MeV]
the ejectiles from the primary channels will be 500 keV-1000 keV, and will
thus not penetrate the DSSD in the backward telescope.
The targets was places on a moterized wheel (2) which could be controlled
from outside of the vacuum chamber. This means we could change the
target easily without having to open up the vacuum chamber.
X
theta_h pd
d
theta
Z pbs
to determine the beam spot and the beam direction. There is five parameters
we need to determine
p~bs = (x, y, z)
dˆ = (θ, φ),
where p~ is the position where the beam hits the target in a global coordinate
system and dˆ is unit vector in the direction of motion of the beam. See Figure
4.14. When we measure a particle in a detector and want to determine
the kinematics of the reaction it originated from, we need to use the angle
relative to the beam axis
pd − p~bs ) · dˆ
(~
cos θ = , (10.1)
||~
pd − p~bs ||
where p~ is the vector to the detection position.
An initial guess of p~bs = (0, 0, 0) and dˆ = ẑ is good enough for us
to identify the elastic scattered deuterons from the 12 C(d,d)12 C. We use
the 12 C beam since the intensity was sevel orders of magnitude larger. In
Figure ?? (A) we can see the initial guess and a kinematic curve. The fit
10.2. CALIBRATION AND BEAM PROPERTIES 95
of the curve is not particulary good, but we can clearly identify the elastic
component. We can now calculate the a χ2 value for each point, based on a
given set of beam parameters
where Ei is the total measured energy in the telescope (corrected for dead
layers), θi is the angle calculated with 10.1 and Td (θi ) is the kinematic curve
from 2.9. We can now minimize this function to get an estimate of the
beam parameters. Notice that this methods does not account for divergence,
spatial distribution or energy distribution. It turned out, however, that
we are limited by the detector resolution, unless the beam is unphysically
large. This resolution also means that the uncertainties on the estimated
parameters have a similar magnitude.
Another crucial part is the symmetrical coverage from the pentagon.
With just one telescope, the orthogonal direction were too unconstraint,
but having five telescopes covering most of φ gave nice constraints on the
parameters.
The result is shown in Figure ?? (B), with one important extension. The
pad calibrations were also adjusted, since they did not extrapolate nicely to
large energies. This imposes an issue, since Ei depends on the pad energy.
We can, however, also adjust the pad calibrations based on reactions.
Consider again the elastic scattering of deuterons and 12 C. The scattered
deuterons penetrate the DSSD, and the energy deposited depends on the
distance traveled in the DSSD, which in turn dependes on the geometry of
the detector and the angle of the deuteron, θ̂.
We can, however, determine what the initial energy of a deuteron would
have been, given the energy deposited in the DSSD based on energy loss
tabulations.
We can calculate the effective distance it travels in the detector in the
same way as in Section ??, teff (θ̂) which depends on the angle with the
detector θ̂. We can then numerically find Einit , the energy that corresponds
to the energy we measure in the DSSD
96 CHAPTER 10. IS561A
Z teff (θ̂)
dE
∆E(θ̂) = (x) (10.3)
dx.
0 dx Einit
60
60 30
50
40 20
40
30
20 20 10
10
0 0 0
-6 -4 -2 0 2 4 6 8 -6 -4 -2 0 2 4 6 8 -6 -4 -2 0 2 4 6 8
Ex [MeV]
Forward telescopes
The background subtraction in this experiment turned out to be more
difficult, due to the low statistics. In all three plots are the background
from the carbon target is shown. The statistics from CH2 was unfortunately
essentially zero.
For the deuterons and the tritons, we expect the background from
protons to be vanishing, and the C background should be sufficient. For
the protons, however, they have a significant contribution. Moreover, due
to the larger beam energy, the background from elastic protons is now in
the middle of our excitations spectrum for 10 Li. To perform a subtraction
of this, we have instead simulated the elastic contribution and estimated
the contribution. The result of the simulation is shown in Figure 10.2 (A).
In the protons spectrum we see a rather large feature without a lot of
structure. The elastic proton peak is, however, visible around 3 MeV as a
peak on top of a broader structure beneath.
In the deuteron spectrum we see a the ground state around 0 MeV and
a broader structure at higher energies. The 1st excited state at 2.7 MeV is
easily embedded in this region. The neutron seperation energy for 9 Li is
4.06 MeV[TILLEY], so start to get into the continuum, and thus expect little
structure, even if there are 9 Li resonances up to 6.43 MeV. The resolution
98 CHAPTER 10. IS561A
8 8
15
∆E · cos θ [MeV]
E [MeV]
E [MeV]
6 6
10
4 4
5
2 2
0 0 0
5 10 15 20 120 125 130 135 140 145 150 120 125 130 135 140 145 150
E + ∆E · (1-cos θ) [MeV] θLAB [deg] θLAB [deg]
is clearly worse than in IS367, but this is expected from the higher beam
energy.
In the triton spectrum we see much less structure, than we did in IS367.
!!!MMMH; WHY!?
Backward telescope
In the backward telescope we only expect to see the protons from the 9 Li
(d,p)10 Li reaction. In Figure 10.3 we see the ∆E − E plot for the telsecope
to the left and the kinematic plot in the middle with the condition of no
particle in the pad. The protons we are interested in are so low in energy,
that they should stop in the DSSD. The kinematic curve for 9 Li (d,p)10 Li
is drawn on top, to show in which region we should see the protons. On the
right we see corresponding kinematic plot for 12 C(d,x).
There is not much structure in either of the two. The features down in
the right corner is unfortunately a noise peak. This effectively renders half
of the detector incapable of detecting the protons from 10 Li. In the other
half we see very few events and in particular not enough to extract any
structure in terms of angular distributions. This could be a DAQ related
10.3. DATA ANALYSIS 99
Intensity
In this experiment we only performed a single measurement on a heavy
target to estimate the intensity. The scattered 9 Li are very forward focused,
so we only see them in the downstream DSSD. The spectrum is shown in
Figure 10.4. Notice that the beam energy have been reduced to account
for energy loss in the target of the kinematic curve. We perform the same
analysis as we did in Section 7.2, but only for a single run. During the
measurement we get an intensity of
70
65
60
55
50
45
E [MeV]
40
35
30
25
20
15
10
5
0
0 5 10 15 20 25 30
θCM [deg]
Figure 10.4: Kinematic spectrum for elastic scattered 9 Li off of a gold target.
The kinematic curve has been corrected for energy loss in the target.
11
IS561C
The last experiment in this series was carried out in the fall of 2018. The
main goal of this experiment was the two neutron transfer with a tritium
target to study 11 Li. The tritium target was backed by titanium so, we also
used a pure titanium target to estimate the background. For this experiment
the energy upgrade of HIE-ISOLDE was complete and we managed to get
a beam energy of 8.0 MeV/A. This energy was required to be able to
measure the protons from the reaction in the backward direction. Due to
the kinematic compression, would their energy be relatively low.
In case of problems with the primary target, we also had a deuterated
plastic target in the chamber. In case this was required, a normal plastic
target was also present for background.
This, unfortunately, turned out to be necessary. The main target turned
out to be damaged from an earlier experiment, and the repair made it mor
or less unuseable for our purpose. I will discuss this in more detail in Section
11.2.
For the stable pilot beam a heavy target was also used, in this case a
lead target. It was, however, not in the chamber during the radioactive
beam.
A table of all the targets can be seen in 11.1.
With the high intensity of the stable beam, we could use the same method
for calibrating pads and determine the beam spot and beam direction as we
did in 10.2. The beam spot and beam direction was estimated to be
101
102 CHAPTER 11. IS561C
11.1 Setup
The setup for this experiment is very similar to the one used in IS561A.
The only major difference is another backward telescope, that was placed
symmetrically to the previous one, effectively doubling our coverage in the
backward directions. A sketch of the setup is shown in Figure ??.
The target wheel was replaced by a target stick, that could still be
operated from outside the chamber.
At the beam dump, we installed a small silicon telescope. The motivation
was to differentiate between beam components. The large instantanous
intensities meant that it could not be used for intensity measurements, the
electronics could simply not keep up. Moreover were there very severe pileup
effects during the experiment and the detectors deteriorated significantly
during the experiment. It was thus useful as a dianostics tool during the
experiment, but not for quantitative analysis.
An alternative beam dump was a Faraday Cup, which we used as a
diagnostics tool during the experiment. It was, however, not sensitive
enough to give a reliable estimate of the beam intensities.
11.2. MAIN TARGET 103
Figure 11.1: The target ladder slot with the 3 H foil. The three regions we
investigated is indicated.
Pentagon
If the target only had the expected 3 H and 48 Ti components, all the protons
we would see would come from a varity of reactions with 9 Li. If there is
a contaminant with a lot of proton (glue has a major organic part) we
would see a lot of elastically scattered protons. We can thus get an idea
of the background by looking at the kinematics of the protons we see in
104 CHAPTER 11. IS561C
Steel H3 Glue
20
18
16
14
E [MeV]
12
10
2
40 50 60 70 40 50 60 70 40 50 60 70
θLAB [deg]
Figure 11.2: Kinematic curves for 9 Li beam on the main target. The red
curves corresponds to elastic protons.
the forward direction. Kinematic curves for the three regions are shown in
Figure 11.2 together with the expected kinematic curve for elastic scattered
protons.
There is a big contaminant in both the region we expect to be the "good"
region and where we expect to see the glue. Notice the energy offset to the
kinematic curves. This indicated that the glue component is rather thick,
but we will return to this.
If we assume the two neutron transfer reaction we can plot the corre-
sponding excitation energy for 11 Li based on the measured protons, see
Figure 11.3. Unfortunately they are too similar in the region of interest and
we can not do a simple background subtaction.
S3
To get a more clear picture of the situation, we can also look in the round
DSSD placed downstream. We can not identify protons in this detector,
but we can see the elastic scattered 9 Li. To simplify the analysis we restrict
us to a single ring. The uncalibrated energy spectra from the three regions
as well as an energy spectrum from a measurement on the pure 48 Ti is
shown in Figure 11.4. Notice that the pure 48 Ti target was not modified in
11.2. MAIN TARGET 105
H3
103
Glue
Steel
Counts/81 keV
102
101
100
0
-8 -6 -4 -2 0 2 4 6
Ex [MeV]
Figure 11.3
in the Glue spectrum. This is consistent with smeared out elastic peak due
to a large energy loss, for instance a thick layer of flue.
At the low channel numbers, up until channel 1000, we see a feature
that resembles the same region in the Steel spectrum. This feature falls off
to quickly to also be present in the Glue spectrum.
Neither of these two features are present in the pure 48 Ti spectrum. The
three spectra from the 3 H target thus all seem to be largely dominated by
reactions with the glue and steel.
Backward detectors
The elastic protons will not scatter beyond 90◦ , and they will therefore not
contaminate the backwards detectors. Reactions from 9 Li with steel, the
106 CHAPTER 11. IS561C
103
H3 Glue
103
102
Counts/4ch
102
101
101
100 100
0 0
0 1000 2000 3000 4000 0 1000 2000 3000 4000
Steel 102 Ti
103
Counts/4ch
102
101
101
100
100
0 0
0 1000 2000 3000 4000 0 1000 2000 3000 4000
Channel Channel
Figure 11.4
60 BB7L
H3
50 Glue
Steel
40
30
20
Counts / 100 keV
10
0
BB7R
30 H3
Glue
Steel
20
10
0
-2 -1 0 1
E [MeV]
Figure 11.5
Trigger thresholds
We experienced another issue with the backward detectors, limiting our
effective coverage region. The trigger thresholds was, due to noise issues
during the setup phase, simply set to high for the backward detectors. Our
trigger effiency, that is the fraction of hits that we trigger on at a certain
energy, can be calculated from the TDC and ADC values. The TDC have a
value whenever a channel had a signal to trigger on. The ADC can, however,
digitize non-zero values if the DAQ was trigger from another source. In a
physics run this could be a coincident particle or random noise.
For a more precise measurement of this we can use a pulser, and make
sure we trigger the DAQ every time we send in a pulse. By varing the pulse
height we can simulated different energies and we can see if a particular
channel produces a trigger as well, i.e. a TDC value. We can calculate the
108 CHAPTER 11. IS561C
0.8
Counts/2keV
0.6
0.4
Pulser Pulser
0.2
Physics Physics
0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 8
E [MeV]
Figure 11.6
Thickness of contaminant
We can try to estimate the thickness of the glue component based on the
energy loss. Since we do not know exactly what the glue is made of we model
it as plastic. We thus assume that the target consists of a think titanium
foil and a thick layer of plastic. As usually we identify all the protons, and
do the energy correction, assuming that that the reaction happend in the
middle of the thick plastic part.
11.2. MAIN TARGET 109
20
10
5
E [MeV]
200
10
0
40 50 60 70 80 40 50 60 70 80
θLAB [deg]
Figure 11.7
We can now try to get the kinematic curve for the elastic proton for
both 12 C(Glue) and 9 Li(Glue) to fit, by adding a layer of plastic. We should
be able to agree on a thickness which fits both beams. With a thickness of
90µm we get Figure ?? (C) and (D). This looks consistent, considering the
material is not the correct one.
110 CHAPTER 11. IS561C
p d t
CD2 CD2
120 CD2
200
CH2 CH2 CH2
250
175 100
150 200
80
125
150
100 60
75 100 40
50
50 20
25
0 0 0
-4 -2 0 2 4 6 8 -4 -2 0 2 4 6 8 -4 -2 0 2 4 6 8
Ex [MeV]
Figure 11.8: Excitation spectra from the identified protons, deuterons and
tritons.
Intensity
In this experiment we did not have a heavy target due to target ladder
limitations. This also means that we do not have a good measurement
of the intensity. The titanium target is too light for the the 9 Li beam to
scatter according to the Rutherford distribution in our coverage region. The
grazing angle for 9 Li is only 9.1◦ in LAB and we do not have coverage here.
We can make a crude estimate by extrapolating the scattering on ti-
tanium into the Rutherford region. Beyond the grazing angle, we see
Fresnel and Fraunhofer types of diffraction patterns. We do not have
the angular resolution to resolve these, but they do follow an exponential
trend[HODGESON, pp533]. At angles up to the grazing angle the the
ratio is just one. At angles above the the grazing angle the ratio follow an
exponential decay
dσ/dΩ
∝ e−λ(θ−θgz ) ,
dσR /dΩ
I 0 = I · e−λ(θ−θgz ) . (11.2)
We can now measure in a number of different angles and fit to an
exponential to determine λ and in particular the real intensity I, which
corresponds to the value at θ = θgz .
We had a varity of problems with the down stream detector, and in par-
ticular the calibrations. This meant that the matching algorithm discarded
112 CHAPTER 11. IS561C
most of the elastic events. I therefore performed this analysis using just
the rings which are mostly symmetric in θ. If we assume that each ring
corresponds to a single angle we can plot the deduced intensity as a function
of the central angle in each ring. See Figure 11.9 for the plotted data and
the subsequent fit to 11.2. The fitted value is
where the error comes from the fit. There are, however, several systematic
errors, which must also be accounted for. First of all are the rings not fully
symmetrical, and the central angles does not reflect this. Secondly are we
averaging over the ring, which becomes more problematics as we approach
0◦ . Most importantly, however, is our model crude, and there are many
details about the diffraction pattern that we have left out. We do also not
fit over many angles. Due to these systematics we take slightly conservative
approach and report an uncertainty on 25 %.
102
Fit
Intensity [kHz]
1σ
θgz = 9.1◦
101
7 8 9 10 11 12 13 14
θLAB
Results
The first simplest reaction is the elastic scattering of deuterons. Due to the
inverse kinematics the deuterons only scatter at θLAB ≤ 90◦ in LAB, and
we detect a large range of the CM angles. We only observed the ground
state at 2.68 MeV/A, but at 6.72 MeV/A and 8.0 MeV/A we also observed
inelastic scattering to the excited states in 9 Li and even break-up channels.
The width of the ground state can give us an estimate of the resolution of
the experiment. The FWHM yields a resolution of INSERT RESOLUTION.
The elastic channel can be described by the optical model, and we
can calculate the expected angular distribution as described in Section ??.
Furthermore, we can try to describe the scattering using a CDCC approach,
where we treat the scattering of a neutron and a proton respectively and
introduce a secondary deuteron interaction. Such a description would be
similar to our model of the neutron transfer to 10 Li, and may provide some
good consistency checks.
With the elastic scattering results, we can continue to the more difficult
task of describing 10 Li, which will be done in the framework described in
Section ??. We can, however, use the optical potentials for 9 Li-p scattering,
which we measured with the plastic target.
Another open channel in our experiments is one neutron transfer from
9 Li to the deuteron, and thus the measurement of 8 Li. For this reaction we
have several bound states, and will use a DWBA approach, as described in
Section ??. This has, however, not been a focus point.
115
116 CHAPTER 12. RESULTS
ell, Watson and KD can not reproduce the correct form, but the magnitude
is not too far of. Making a fit to the data, we get a set of parameters that
reproduces the data better. This would in principle be the correct input
for the CDCC calculation. Since we do not have a corresponding fit for the
neutrons, and we need to use some parametrization to describe the 9 Li-n
interaction.
CDCC
For the CDCC calculation we need to include both the 9 Li-n and the 9 Li-p
interaction. Further more we need to specify the bins we use for the p-n
continuum. In the present calculation we make sure to include energies
high enough to let the solution converge. In practice we used 6 MeV for
2.68 MeV/A and 10 MeV for 6.72 MeV/A and 8.0 MeV/A. Furthermore we
need to specify the spin and parity states for the individual bins. Once again,
we simply need to include enough partial waves to make the calculation
converge. The deuteron ground state is J π = 1+ , and with the coupling
scheme discussed in Section ?? we include states up to l = 3.
For the 9 Li groundstate we use the Reid-Soft-Core as implemented in
FRESCO by AM. We include the coupling to each of the bins in the p-n
continuum.
The result of the CDCC calculations with different optical potentials is
shown in Figure 12.3 for the three different energies. In the 2.68 MeV/A case,
the Watson parametrization is far off. From the elastic protons calculation,
Watson was the parametrization which was furthest from reproducing the
shape. If the 9 Li-n interaction have a similair descrepency, it is not surprising
that the CDCC calculation using these optical models are worse. It is,
however, a noteworthy that it overestimates the magnitude by a large factor.
The KD parametrization, on the other hand, does a good job reproducing
the shape and magnitude of the data. The first minimum is particularly
underestimated.
If we use the fitted potential for the 9 Li-p interaction we get closer in
this region but at the expense of agreement in the larger angles. Since
118 CHAPTER 12. RESULTS
20
(a) 2.68 MeV/A
15 P1
DA1P
P2
10
Fit
250
(b) 6.72 MeV/A
20 P2
dΩ [mb/sr]
P1
15
P3
10
dσ
300
(c) 8.0 MeV/A
25
P2
20 P1
15
10
0
20 40 60 80 100 120 140 160 180
θCM
Figure 12.1: Angular cross section for elastic deuterons at three different
energies. Data for 2.68 MeV/A is shown in (a) together one parametrizations
of the optical potentials as well as two fits from [Jeppesen]. The last curve
corresponds to the a fit to the data. In (b) we see the data for 6.72 MeV/A
with three fits to the optical potentials. In (c) we see the data for 8.0 MeV/A
with two fits. P1 us fitted to the full dataset whereas P2 is fittede to the
first feature below 60◦ . The parameters used in the calculations are listed
in 12.1.
12.1. ELASTIC SCATTERING 119
100
(a) 2.68 MeV/A
80 Powell
Watson
60
KD
40 Fit
20
2000
(b) 6.72 MeV/A
150 Watson
dΩ [mb/sr]
KD
100
dσ
50
2000
(c) 8.0 MeV/A
150 Watson
KD
100
50
0
20 40 60 80 100 120 140 160 180
θCM
Figure 12.2: Angular cross section for elastic protons at three different
energies. Data for 2.68 MeV/A is shown in (a) together with different
parametrizations of the optical potentials. The last curve corresponds to
the a fit to the data. In (b) and (c) we see the data for 6.72 MeV/A and
8.0 MeV/A respectively together with two parametrizations of the optical
potentials. The parameters used in the calculations are listed in 12.1.
120 CHAPTER 12. RESULTS
10
12.2 Li transfer
The analysis of the one-neutron transfer is more difficult, since the resulting
10 Li is unbound. This means that the calculation is more involved due to
fact that we need to describe both the reaction mechanism as well as the
12.2. 10 LI TRANSFER 121
20
(a) 2.68 MeV/A
15 Watson
KD
KD+Fit
10
250
(b) 6.72 MeV/A
20 Watson
dΩ [mb/sr]
KD
15
10
dσ
300
(c) 8.0 MeV/A
25
Watson
20 KD
15
10
0
20 40 60 80 100 120 140 160 180
θCM
Figure 12.3: Angular cross section for elastic deutrons calculated with CDCC.
Data for 2.68 MeV/A is shown in (a) together with two parametrizations of
the optical potentials for 9 Li-n and 9 Li-p. The KD+Fit uses the 9 Li-p that
was fitted to elastic protons. In (b) and (c) we see two parametrizations of
the 9 Li-n and 9 Li-p optical potentials.
122 CHAPTER 12. RESULTS
Table 12.1: Parameters for optical potentials for elastic scattering. Parame-
ters in bold were fitted otherwise they were fixed.
Acceptance
The correction of the acceptance is also slightly more complicated since we are
not looking at bound states and thus not on single excitation energies. This
requires a more explicit treatment of the the double differential acceptance
as described in 7.9.
One challenge with the double differential acceptance is the amount of
statistics. Nrd (θ, φ, E) and in particular Nsd (θ, φ, E) can get quite small,
which can lead to large uncertainties in those bins.
We have large kinematic and geometric areas that our setup does not
cover. If we follow the procedure outlined in Section ?? we can investigate
our acceptance region by simply plotting the number of detected particles
given a uniform distribution. In the case of 10 Li we thus perform the
simulation uniformly in 4π and uniformely on Er = 2.01 MeV which is the
available energy in CM. We can now analyze the simulation and plot the
number of detected particles as a function of CM angle and excitation energy
and the result for 2.68 MeV/A and 6.72 MeV/A is show in Figure 12.4.
We see that it is a non-triviel region that is covered. Traditionally
theorist are aware and will report the integrated cross section in some
region, for instance angular distributions between 0 MeV-2 MeV. There are,
however, a complicated correlation between and angle and energy in terms
of the acceptance.
We can approach this in two ways. The simples way is to gate on a
"simple" region. In this case we would have coverage and we can safely
integrate the theoretical calculations to get single differential cross sections.
An alternative approach is to use the acceptance to calculate the expected
number of counts and then compare this the experimental result. This has
the advantage that we can use the full dataset, however, it makes it difficult
to compare to other experimental datasets and models that have not been
folded with our acceptance.
124 CHAPTER 12. RESULTS
7 8
1.5 6
6
5
E [MeV]
1.0 4
4
3
0.5 2
2
1
0.0 0 0
Figure 12.4: Acceptance for 9 Li (d,p)10 Li at 2.68 MeV/A 6.72 MeV/A and
8.0 MeV/A respectively. Notice lines in the two first panels. They corre-
sponds to dead strips of the detectors.
Let us consider the case of 2.68 MeV/A, the left panel in Figure 12.4.
Suppose we want to get the differential cross section the angular axis. The
upper left area in the acceptance plot makes it impossible to define a "simple"
region that includes all the data. We can choose θCM = 90◦ − 150◦ and
E = 0 MeV − 1.5 MeV. In this case we have full coverage, and we can
compare to theoretical predictions integrated over either θCM = 90◦ − 150◦
or E = 0 MeV − 1.5 MeV. This would, however, be on the expense of some
data. If we instead include angles down to θCM = 70◦ and project onto the
angle axis, we are missing a lot of cross section at the low angles but high
energy due to the vanishing acceptance.
Given a range we now have to choose our bin size. To the lowest
approximation, we may simply choose one energy bin to increase the statistics
in each of the angular bins. This effectively changes (7.10) to
where ∆E is now the energy range we chose from the acceptance plot. Nrd (θ)
12.2. 10 LI TRANSFER 125
S 90
0.010 0.05 S 70
C 90
C 70
0.008 0.04
Arb. units
Arb. units
0.006 0.03
0.004 0.02
S 90
0.002 S 70 0.01
C 90
C 70
0.000 0.00
80 100 120 140 0.0 0.5 1.0 1.5 2.0
θ CM E [MeV]
Figure 12.5: Acceptance corrections for 2.68 MeV/A for both angular and
energy differential cross sections.
Singles
Based on the acceptance we limit our analysis to θCM = 90◦ − 150◦ and
E = 0 MeV − 1 MeV. The final result is shown in Figure 12.6 (A). There
are two main observations in the excitation spectrum.
The first is the significant strength at the threshold in the excitation
plot (right panel). This is on the debated points, but it is clear that these
data show a significant contribution from a s1/2 state in order to reproduce
the data. The overall shape of the spectrum seems to fit the calculation
nicely, however, the magnitude if off by a factor of ≈ 2. One possible
explanation is a sizeable contribution from CN reactions. There is multiple
ways of estimating the expected contribution of compound nucleus channels,
however, we will not pursue such calculations in this works.
The second main observation is the shape of the peak. Even though the
data give a clear peak just below 0.5 MeV, we can not directly conclude that
we need two resonant p1/2 contributions to reproduce it. The resolution of
the setup is not good enough to distinguish two close resonant states. This
will instead require a more detailed invesitagtion of the model.
For the d-wave contributions, we still have to little energy available
12.2. 10 LI TRANSFER 127
dσ/dEx [mb/MeV]
p1/2 2+
dσ/dΩ [mb/sr]
101 20 Sum
15
A)
10
100
0
0 25 50 75 100 125 150 175 0.0 0.5 1.0 1.5 2.0
θCM [deg] Ex [MeV]
p1/2 2+
dσ/dΩ [mb/sr]
d5/2 1-
15 d5/2 2-
101
d5/2 3-
B) 10
d5/2 4-
Sum
100
5
10−1
0
0 25 50 75 100 125 150 175 0 1 2 3 4 5 6 7
θCM [deg] Ex [MeV]
p1/2 2+
dσ/dΩ [mb/sr]
10 d5/2 1-
d5/2 2-
101 8 d5/2 3-
C) 6
d5/2 4-
Sum
100 4
2
10−1 0
0 25 50 75 100 125 150 175 0 2 4 6 8
θCM [deg] Ex [MeV]
Figure 12.6: Differential cross for 2.68 MeV/A (A), 6.72 MeV/A (B) and
8.0 MeV/A (C). Right panel shows the angular cross section and the left
panel shows the energy cross section. The title of each panel indicates the
integration range in the other variable.
128 CHAPTER 12. RESULTS
in CM at 2.68 MeV/A. We can instead look at the results from the two
experiments at higher energies. In these two experiments we have a few more
limitations. For 6.72 MeV/A the statistics are limited and the the intensity
are more uncertain, giving a larger systematic error. For 8.0 MeV/A we
have decent statistics, but the intensity estimate is very uncertain. For
both cases, the energy resolution is not as good as the low energy, which
is mainly an effect of the higher beam energy, giving a larger kinematic
compression. This means that we can not say much about the low energy
region (0 MeV-2 MeV), but we can use it to investigate the indications of a
d-wave which may lie in the 2 MeV-10 MeV range.
We need to choose a safe range like we did before. For 6.72 MeV/A
AND 8.0 MeV/A we chose θCM = 65◦ − 100◦ and E = 0 MeV − 6 MeV. The
results are shown in Figure 12.6 (B) and (C), respectively. We see a broad
feature above 2 MeV in both cases. For 6.72 MeV/A the shape of the angular
distribution seems to match well, but the magnutide is too low. The missing
cross section in the model comes mainly from the high energy region, and
in particular in the area where we could expect to see the d-wave resonance.
This could indicate the exsistence of such a resonance, and this is backed
by a similar feature at 8.0 MeV/A. The normalization of this data set is,
however, rather uncertain, so it is difficult to conclude much from magnitude.
Without the measurement of coincidences, it is difficult to determine how
much of this belongs to other reactions mechanisms.
Slices
We can learn more about the data if we only look at the angle integrated or
energy integrated differential cross sections in certain bins. This is basically
an attempt to look at the double differential cross section. We can divide
our acceptance region into bins of 10◦ and calculate the angle integrated
cross section. Likewise can we divide the energy range into bins of 200 keV
for 2.68 MeV/A and 1 MeV for 6.72 MeV/A and 8.0 MeV/A and calculate
the energy integrated cross sections. The results are shown in Figure 12.7,
12.8 and 12.9 respectively.
12.2. 10 LI TRANSFER 129
6.72 MeV/A The angle integrated cross sections (left column) shows an
interesting dip in the 60◦ -70◦ bin, clearly indicating some structure. This
could be due to the suppresion of the s-waves even though this does not fit
with the position from the current calculation. The energy integrated cross
sections shows a similar dip in the 3 MeV-4 MeV bin. In general, however,
the angular distribution looks more consistent with the model, since all the
bins increases as we approach lower angles.
8.0 MeV/A The energy integrated bins (right column) looks more struc-
tureless. At the higher energy bins, the angular distributions again increases
at lower angles, but it does look more flat at low energies. The angle
integrated (left column) cross sections looks structurally very similar.
Coincidences
In IS367 we did measure a small amount of 9 Li in coincidens with protons.
We now consider all protons that was detected together with one other hit.
If we assume that the other hit is a 9 Li, we can calculate the missing energy
and momentum and deduce the neutrons energy as well as direction. More
importantly, however, can we look at the correlation between the neutron
and the 9 Li or the neutron and the proton.
To explain the descrepency we see in the cross section we may explore
some other possible channels that may contribute to our reaction, and to
which extend we can understand the contributions from an experimental
point of view.
130 CHAPTER 12. RESULTS
5 1
0.0-0.2MeV
θCM =80-90 0
0 1
5
0.2-0.4MeV
0
0
θCM =90-100 1
5
0.4-0.6MeV
0
1
θCM =100-110
0
dσ/dΩ [mb/sr]
5
0.6-0.8MeV
0
dσ/dE [mb/MeV]
1
θCM =110-120
0
5 0.8-1.0MeV
0
1
θCM =120-130
0
1.0-1.2MeV
5 0
1
θCM =130-140
0 1.2-1.4MeV
0
5 1
Figure 12.7: Differential cross section for angular (left panel) and energy
(right panel) bins for 2.68 MeV/A. The distributions seems very similar
across both axes, indicating that we are, in fact, not measuring a direct
reaction.
12.2. 10 LI TRANSFER 131
5 5
θCM =50-60
0 0 0.0-1.0MeV
5 5
θCM =60-70
0 0 1.0-2.0MeV
5 5
dσ/dE [mb/MeV]
dσ/dΩ [mb/sr]
θCM =70-80
0 0 2.0-3.0MeV
5 5
θCM =80-90
0 0 3.0-4.0MeV
5 5
θCM =90-100
0 0 4.0-5.0MeV
5 5
=100-110
0 5.0-6.0MeV
θ
0 CM
0 2 4 6 50 60 70 80 90 100 110
E [MeV] θCM [deg]
Figure 12.8: Differential cross section for angular (left panel) and energy
(right panel) bins for 6.72 MeV/A.
132 CHAPTER 12. RESULTS
5.0 5
2.5
0 0.0-1.0MeV
θCM =50-60
0.0 5
5.0
0 1.0-2.0MeV
2.5
5
θCM =60-70
0.0
5.0 0 2.0-3.0MeV
5
2.5
dσ/dΩ [mb/sr]
θCM =70-80
0.0 0 3.0-4.0MeV
5.0 5
2.5
0 4.0-5.0MeV
dσ/dE [mb/MeV]
θCM =80-90
0.0 5
5.0
0 5.0-6.0MeV
2.5
5
θCM =90-100
0.0
5.0 0 6.0-7.0MeV
5
2.5
=100-110
0 7.0-8.0MeV
θ
0.0 CM
0 2 4 6 8 50 60 70 80 90 100 110
E [MeV] θCM [deg]
Figure 12.9: Differential cross section for angular (left panel) and energy
(right panel) bins for 8.0 MeV/A.
12.2. 10 LI TRANSFER 133
50 Singles
Coincidences (Phase space)
Coincidences (Breakup)
40
dσ/dEx [mb/MeV]
30
20
10
0
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75
Ex [MeV]
Figure 12.10: Differential cross section for 10 Li for singles and coincidens
measurements. The singles spectrum is only based on protons. The second
dataset (squares) are coincidences with the acceptance of C. The last dataset
(triangles) are coincidences but with the acceptance of 10 Li breakup A.
A) 9
Li(d, p)10 Li
n + 9 Li
B) 9
Li(d)11 Be 9
Li + n + p
C) 9
Li(d, Li)d 9
n+p
D) 9
Li(d, n)10 Be 9
Li + p
In Figure 12.10, we see the singles spectrum together with the coincidens
spectrum. We have corrected for acceptance with two assumed reaction
mechanisms, A and C. Our first observation is that both magnitude and
shape is consistent with each other and the singles spectrum. The chan-
nels that contribute must be within our acceptance for both singles and
coincidences. A simulation of D showed that we have no acceptance for
coincidences in this reaction, and we can thus rule out D (At least at
the precision of our uncertainties). Otherwise we would see a descrepency
between singles and coincidences. It should be noted that our acceptance
region is the same as for the singles, and we can use the same region and
compare the integrated cross sections. 1
To investigate the reaction mechanism, we can exploit the kinematic
differencens and therefore the difference in acceptance. We can simulate the
three first different situations, uniformly in CM, and analyse the results as
9
Li + d → 9 Li + p + n, (12.3)
where we measure 9 Li and p and deduce n.
The result of the simulation is shown in Figure 12.11, where we have
plotted the correlation between the deduced neutron as a function of either
the proton or 9 Li. The first row, corresponding to situation A, have a
1
FiXme Fatal: En figur?
12.3. 8 LI TRANSFER 135
8
12.3 Li transfer
136 CHAPTER 12. RESULTS
100
50
0
150 (B) (B)
100
50
0
150 (C) (C)
θn
100
50
0
150 Data Data
100
50
0
0 50 100 150 0 50 100 150
θp θ9Li
Figure 12.11: Acceptance for angle correlations between the neutron and
either the ejected proton or the ejected 9 Li based on different simulated
kinematic situations. The letters correspond to the situations outlined in
the text. The left column is the θn versus θp and the right columns is θn
versus θ9 Li . All angles are in center of mass.
List of Figures
2
FiXme Fatal: Tegn ind hvor eksperimental setup er!
137
138 List of Figures
3.4 Energy correlation plots for a silicon telescope. In (A) we see the
energy deposited in the thin detector versus the energy in the
back detector. In (B) we have made the geometrical correction
described in 3.4 to obtain a better resolution. . . . . . . . . . . 19
4.13 The projection of the angle distributions using the old method.
We see that the x-projection shifts and we can thus use that for
our angle estimate. . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.14 We can look a narrower slices by only considering the center half
(the red and green area) or the center fourth (the green area). . 38
4.15 Position and angle estimate of a beam which was both displaced
and angled. We do not recreate the correct values. The method
are thus biased when both effect are present. . . . . . . . . . . . 39
(a) Position . . . . . . . . . . . . . . . . . . . . . . . . . . 39
(b) Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.16 Position and angle distributions for a point beam and a spatial
uniformly distributed beam with a radius of 2.5 mm. . . . . . . . 40
(a) Point beam: Position . . . . . . . . . . . . . . . . . . 40
(b) Broad beam: Position . . . . . . . . . . . . . . . . . . 40
(c) Point beam: Angle . . . . . . . . . . . . . . . . . . . . 40
(d) Broad beam: Angle . . . . . . . . . . . . . . . . . . . 40
11.1 The target ladder slot with the 3 H foil. The three regions we
investigated is indicated. . . . . . . . . . . . . . . . . . . . . . . 103
11.2 Kinematic curves for 9 Li beam on the main target. The red
curves corresponds to elastic protons. . . . . . . . . . . . . . . 104
List of Figures 141
11.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
11.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
11.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
11.6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
11.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
11.8 Excitation spectra from the identified protons, deuterons and
tritons. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
11.9 Calculated intensity based on 9 Li scattered off of 48 Ti. The data
has been fitted to an exponential fit and extrapolated to the
grazing angle at 9.1◦ . . . . . . . . . . . . . . . . . . . . . . . . . 113
143
145
Rettelser
Note: Find reference!!! . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Fatal: Tegn ind hvor eksperimental setup er! . . . . . . . . . . . . . 15
Fatal: MMH!! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Fatal: Check det her . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Fatal: En figur? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Fatal: Tegn ind hvor eksperimental setup er! . . . . . . . . . . . . . 137