U NIVERSITY OF C OLORADO B OULDER

H ONORS T HESIS

Gas Cherenkov Muon Monitor for

the DUNE Neutrino Oscillation
Experiment
by
Max Weiner

Reviewed by:

Professor Alysia Marino, Advisor, Physics

Professor Paul Beale, Honors Council Representative, Physics

Professor Mark Rast, Astrophysical and Planetary Sciences

April 6, 2018
 iii

UNIVERSITY OF COLORADO BOULDER

Abstract

Gas Cherenkov Muon Monitor for the DUNE Neutrino Oscillation

Experiment

by Max W EINER

The upcoming Deep Underground Neutrino Experiment will begin to

take data in the 2020s and will be capable of making the most precise mea-
surements of neutrino oscillation parameters which will help answer some
of the most fundamental questions in physics today. In order to accomplish
this task, it is essential to know the neutrino flux at both the near and far
detectors. To aid in this difficult measurement will be a system of muon
monitors to characterize the associated muon beam. This paper concentrates
on measuring the muon distribution at the beginning of the beam with a
gas Cherenkov detector with the goal of constraining the neutrino energy
and momentum distribution. Currently there is a prototype detector at the
Fermi National Accelerator Laboratory in the NuMI beamline that makes
these measurements. We attempt to extrapolate a muon distribution by ob-
serving and analyzing these signals over various gas pressures and detector
orientations. I will discuss how such a model is created and how to go about
comparing the simulations with real data as well as what future work is nec-
essary.
 v

Acknowledgements
I would like to acknowledge my deepest gratitude to those who have of-
fered me their support and guidance throughout this research project. First,
and foremost, I want to thank Alysia Marino for being such an outstanding
adviser and leader. She has put aside countless hours educating and navigat-
ing me since I started here over two years ago. Her enthusiasm and passion
for what is experimental neutrino physics is truly contagious. I cannot thank
her enough for turning me on to such a fascinating field and for helping me
along with this project, and more. I would also like to thank Jeremy Lopez,
without whom this thesis would not have been possible. Jeremy helped me
develop the complicated computer code used for this research. The entire
Neutrino Group has been nothing but a pleasure to work with. Finally, I
want to thank my committee members, Paul Beale and Mark Rast, for taking
the time to participate in my honors research project.
 vii

Contents

Abstract iii

Acknowledgements v

1 Introduction 1
1.1 Neutrinos and The Standard Model of Particle Physics . . . . 1
1.1.1 Standard Model . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.2 Neutrinos . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.3 Neutrino Oscillations . . . . . . . . . . . . . . . . . . . . 5
1.2 The Deep Underground Neutrino Experiment . . . . . . . . . 9
1.2.1 Long-Baseline Neutrino Facility . . . . . . . . . . . . . 10
1.2.2 Far Detector . . . . . . . . . . . . . . . . . . . . . . . . . 12

2 Gas Cherenkov Muon Monitor 15

2.1 Cherenkov Radiation . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Gas Cherenkov Detector . . . . . . . . . . . . . . . . . . . . . . 18

3 Predicting a Muon Distribution 25

3.1 Pressure Scans at NuMI . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Monte Carlo Simulations . . . . . . . . . . . . . . . . . . . . . . 27

4 Conclusion and Future Work 39

 ix

List of Figures

1.1 Standard Model of Particle Physics . . . . . . . . . . . . . . . . 2

1.2 Mass Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Neutrino Beam . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Illustration of Electromagnetic Horn . . . . . . . . . . . . . . . 12

2.1 Cherenkov Radiation Illustration . . . . . . . . . . . . . . . . . 16

2.2 Super-Kamiokande Detector . . . . . . . . . . . . . . . . . . . . 18
2.3 Cartoon Depicting How a Gas Cherenkov Detector Operates . 19
2.4 Cherenkov Detector at NuMI . . . . . . . . . . . . . . . . . . . 20
2.5 Muon Alcove . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.6 Stable Beam Profile . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.7 Erratic Beam Profile . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.1 Plots From Yaw Scans . . . . . . . . . . . . . . . . . . . . . . . . 26

3.2 Flow Chart Describing Yaw Monte Carlo Scan Simulation . . . 28
3.3 Simulated Yaw Scans Compared to Real Data . . . . . . . . . . 29
3.4 Yield Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5 Muon Distribution . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.6 Muon Distribution in Lower Momentum Regime . . . . . . . . 33
3.7 Muon Distribution in Upper Momentum Regime . . . . . . . . 34
3.8 Best-fit Parameters for Muon Distribution Function . . . . . . 35
3.9 Scans Comparing Monte Carlo Simulation with Yield Matrix
Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.10 Fitting for a Muon Distribution . . . . . . . . . . . . . . . . . . 37
 xi

List of Tables

1.1 Neutrino Oscillation Parameters . . . . . . . . . . . . . . . . . 9

 1

Chapter 1

Introduction

In this paper I will summarize my research in the past year-and-a-half

which involves a muon monitoring system for the future Deep Underground
Neutrino Experiment. This work involved recording and analyzing data as
well as helping develop a fitter so as to compare real data with simulations.
Chapter One introduces the Standard Model of particle physics as well
as the history of neutrinos and their importance to physics today. Next, the
Deep Underground Neutrino Experiment is introduced along with its sci-
ence goals and how it will accomplish these goals. Chapter Two discusses
Cherenkov radiation and how it is utilized in particle physics. The relevance
of a gas Cherenkov detector for neutrino physics is revealed in this chapter.
Chapter Three details how to build a model which predicts a muon flux with
a gas Cherenkov detector. Chapter Four is the conclusion and future work
needed to further this project.

1.1 Neutrinos and The Standard Model of Particle

Physics

The Standard Model (SM) of particle physics is the theoretical frame-

work which describes elementary particles and their interactions. Within this
2 Chapter 1. Introduction

framework is a tiny neutral particle called the neutrino which has been baf-
fling scientists since it was first postulated to exist in 1930 [1]. This conflict
between the SM and the neutrino eluding a complete understanding sug-
gests neutrinos may light the way to a more fundamental theory of nature
beyond the SM.

1.1.1 Standard Model

The Standard Model of particle physics describes matter at its most fun-
damental level. It postulates that the universe is composed of a handful of
matter and antimatter particles as well as force-carrying particles (see Figure
1.1). The matter particles are spin- 21 fermions and are divided into leptons
and quarks while the force-carriers are spin-1 and spin-0 gauge bosons. The
leptons are further divided into three generations according to their flavor:
e, µ, and τ. Each flavor forms a pair containing a charged lepton (l) and an
associated neutral particle, a neutrino (νl ).

F IGURE 1.1: List of particles in the Standard Model. Parti-

cles are grouped into force particles (purple), leptons (green),
quarks (red), and the Higgs boson (grey). Not shown are each
particle’s antimatter counterpart.

The SM was developed in the 1970s and has successfully explained al-
most all experimental results and predicted a wide variety of phenomena [2].
It describes all of the four fundamental forces: electromagnetism, the weak
1.1. Neutrinos and The Standard Model of Particle Physics 3

force, the strong force, and gravity1 . Each force has a particle which is in
charge of mediating that interaction: the strong force is carried out by glu-
ons, the weak force is mediated by the W and Z bosons, gravity interacts via
the graviton, and the electromagnetic force is governed by the photon. Forces
manifest via the exchange of force particles (e.g. a photon is exchanged be-
tween two electrons when they repel one another). The quarks experience all
four forces while the leptons are invisible to the strong force altogether. All
leptons feel the weak force, and because neutrinos have zero charge they do
not interact electromagnetically. It has been established that neutrinos have
a tiny mass, about one millionth the mass of the electron [3]. Neutrinos are
special in that their most significant interaction is comes from the weak force,
and here their interaction range is minuscule (∼ 10−18 m) due to the W and
Z bosons being so heavy (∼90 GeV/c2 ). This combination of being neutral,
having a tiny mass, and a minuscule interaction range makes neutrino detec-
tion formidable. In the next section I will explain what a neutrino is and the
problems it presents to the SM.

1.1.2 Neutrinos

What is a neutrino and how does it conflict with the SM? Neutrinos were
first hypothesized in 1930 by Wolfgang Pauli to save the laws of conserva-
tion of energy and momentum in β decay [1]. Radioactive nuclei will have
a neutron decay into a proton and emit an electron (historically known as
a β particle). The expected outcome was that in each decay an electron of
constant energy would be emitted; the electron’s energy being the difference
between the initial and final energies of the nucleus. Surprisingly, a contin-
uous spectrum of electron energies was measured. Where was the missing
energy? Some (including Niels Bohr [3]) claimed this was proof that the law
1 Although gravity is included in the SM, along with its hypothetical force particle the
graviton (not included in Figure 1.1), an adequate quantum theory of gravity compatible
with general relativity has not yet been established.
4 Chapter 1. Introduction

of conservation of energy is wrong. Pauli suggested that an undetectable,

tiny, neutral particle must be carrying off the missing energy and momenta
(later named the neutrino). Soon after, Enrico Fermi put together a relativis-
tic quantum field theory describing β decay which incorporated neutrinos.
But it wasn’t until decades later, in 1956, when Frederick Reines and Clyde
Cowan confirmed their existence by detecting antineutrinos via inverse β
decay (νe + p → n + e+ ) at the Savannah River nuclear plant in South Car-
olina [1].
The theoretical models assumed neutrinos came in three flavors (νe , νµ ,
and ντ ) and were massless because they travel at a speed indistinguishable
from the speed of light [4]. In the 1960s, Ray Davis studied solar neutrinos
with a giant underground detector at the Homestake Mine in Lead, South
Dakota. The Sun undergoes nuclear fusion and thus emits an extraordinary
number of neutrinos (∼ 1012 pass through your body every second, day and
night!) yet Davis detected only about one third of the predicted neutrinos [5].
What was wrong? Were the theorists incorrect or was their something awry
in Davis’ measurements? This is what is now known as the "Solar Neutrino
Problem." It turns out Davis’ detector (a tank containing 100,000 gallons of
dry cleaning fluid) was only sensitive to measure electron neutrinos. The
physicists Bruno Pontecorvo, Ziro Maki, Masami Nakagawa, and Shoichi
Sakata [6, 7] proposed that neutrinos undergo oscillations (i.e. an electron
neutrino can suddenly transform into a muon neutrino) but this can only
happen if neutrinos have mass. Neutrino oscillations weren’t confirmed un-
til within the last couple of decades by the Super-Kamiokande (1998) and
Subdury Neutrino Observatories (2002) whose discovery was awarded the
Nobel Prize in 2015.
1.1. Neutrinos and The Standard Model of Particle Physics 5

1.1.3 Neutrino Oscillations

What are neutrino oscillations and why are they important? Neutrino
flavor states are not equivalent to neutrino mass eigenstates. Instead, they
are superpositions of the mass eigenstates ν1 , ν2 , and ν3 :

3
|νl i = ∑ Uli∗ |νi i , l = e, µ, τ (1.1)
i =1

and so the masses of the flavor states are not well defined. The matrix U is
the Pontecorvo-Maki-Nakagawa-Sakata (PMNS) Matrix:

 
 Ue1 Ue2 Ue3 
 
U= U
 µ1 U µ2 U 
µ3 
 
Uτ1 Uτ2 Uτ3
   
0   c13 0 s13 e − iδ
1 0   c12 s12 0
CP

   
=
 0 c 23 s 
23   0 1 0  −s c
  12 12  0 
   
0 −s23 c23 −s13 eiδCP 0 c13 0 0 1

and cij and sij are shorthand for sin θij and cos θij . The mixing angles θij
relate the amount of mass eigenstates in each flavor state and the Dirac CP
violation phase δCP suggests neutrinos and antineutrinos oscillate differently
for a nonzero value (neutrinos mix by U, antineutrinos mix by the complex
conjugate U ∗ ).
To illustrate the oscillatory behavior of neutrinos, let us assume that there
are only two flavor states and two mass states. Then we can write our flavor
states as [8],
6 Chapter 1. Introduction

   
θ θ
|νe i = cos |ν1 i + sin |ν2 i
2 2
   
θ θ
νµ = sin |ν1 i − cos |ν2 i
2 2

where θ is the parameter describing the mixing between flavor and mass
states. So if we have an electron neutrino as our initial state, what is the prob-
ability of measuring it to be a muon neutrino at some later time t? Neutrino
flavor states are eigenstates of the weak interaction, whereas the mass states
are the eigenstates of the free-particle Hamiltonian [9]. Thus, the electron
neutrino evolves in time (in vacuum) according to the Schrödinger equation,

   
θ −iE1 t/h̄ θ −iE2 t/h̄
|ψ(t)i = cos e |ν1 i + sin e |ν2 i
2 2

and the probability of measuring a muon neutrino is given by,

Pνe →νµ = | νµ ψ(t) |2

sin2 (θ )
= |1 − e−i(E2 −E1 )t/h̄ |2
4
2 ( E2 − E1 ) t
 
2
= sin (θ ) sin .
2h̄

In the relativistic limit where the neutrinos are traveling at near the speed of
light,

( m i c2 )2
Ei ≈ pc + ,
2pc

p ≈ E/c,

t ≈ L/c.
1.1. Neutrinos and The Standard Model of Particle Physics 7

Thus,

 (m2 − m2 ) Lc3 
Pνe →νµ = sin2 (θ ) sin2 2 1
. (1.2)
4Eh̄

Equation 1.2 applies to neutrinos in vacuum, corrections must be applied

when they travel through matter.
The full three neutrino probability function is more complicated,

 ∆m2 L 
ij
P(νl → νl 0 ) =δll 0 − 4 ∑ Re(Uli∗ Ul 0 i Ulj Ul∗0 j ) sin2
i> j
4E
 ∆m2 L 
ij
+ 2 ∑ Im(Uli∗ Ul 0 i Ulj Ul∗0 j ) sin2 ,
i> j
2E

where ∆m2ij = m2i − m2j . The main idea is the same for both cases: the os-
cillatory nature depends on the parameter L/E, where all other variables
are constant. As you can see from the oscillation equations, only ∆m2ij can
be measured, not the absolute value of the neutrino masses. These masses
remain unknown, in fact it is a question of whether the masses fall into a
normal hierarchy (m1 < m2 < m3 ) or an inverted hierarchy (m3 < m1 < m2 )
as shown in Figure 1.2. Long-baseline neutrino experiments measure these
oscillation parameters by shooting a neutrino beam to a large detector hun-
dreds of kilometers away. This, along with solar neutrino and nuclear re-
actor experiments, has resulted in the best-fit values as shown in Table 1.1.
The Deep Underground Neutrino Experiment is one such project that will,
among other things, make the most precise measurement of the CP viola-
tion parameter δCP . This is an important measurement as it may explain the
matter/antimatter asymmetry of the observable universe.
CP violation is the notion that under both charge conjugation (changing
8 Chapter 1. Introduction

F IGURE 1.2: Illustration of normal and inverted hierarchies.

Only difference of mass squares can be measured via neutrino
oscillations, not absolute masses. Colors represent amount of
flavor in each mass state.

a particle to its antiparticle) and parity (inverting spatial coordinates), par-

ticles do not behave symmetrically. So far, only left-handed neutrinos and
right-handed antineutrinos have been observed [3]. This means that neutri-
nos maximally violate charge conjugation and parity separately, yet it is un-
known if neutrinos violate the combination, CP. If we act CP on a left-handed
neutrino:

CP |νL i = C |νR i

= |νR i .

So if the state |νL i oscillates differently than |νR i, then neutrinos violate CP
symmetry. Since it is a complex phase, the sign of δCP is different for neutri-
nos and antineutrinos and a nonzero (as well as non-π) value manifests itself
in different oscillatory behavior. CP violation has been observed for quarks
but never for leptons. It is in this respect that neutrinos may explain the mat-
ter/antimatter asymmetry of the observable universe (the mixing of quarks
1.2. The Deep Underground Neutrino Experiment 9

TABLE 1.1: Current best-fit values on neutrino oscillation

parameters[12]. Values are given as: normal hierarchy (in-
verted hierarchy) and ∆m2 = m23 − (m22 + m21 )/2.

Parameter Best-Fit 3σ
∆m221 [10−5 eV 2 ] 7.37 6.93 - 7.97

|∆m2 |[10−3 eV 2 ] 2.50 (2.46) 2.37 - 2.63 (2.33 - 2.60)

sin2 (θ12 ) 0.297 0.250 - 0.354

sin2 (θ23 ) 0.437 (0.569) 0.379 - 0.616 (0.383 - 0.637)

sin2 (θ13 ) 0.0214 (0.0218) 0.0185 - 0.0246 (0.0186 - 0.0248)

δCP Unknown

is too small to explain this asymmetry).

1.2 The Deep Underground Neutrino Experiment

The Deep Underground Neutrino Experiment (DUNE), utilizing the Long-

Baseline Neutrino Facility (LBNF), is an upcoming long-baseline neutrino
project. DUNE will make the most precise measurements of the oscillation
parameters as well as address the following big questions [10]

1. Matter/Antimatter Asymmetry. Equal parts matter and antimatter were

created in the Big Bang but we live in a universe of matter. What ex-
plains this asymmetry? A possible explanation is a value of the CP
violating phase δCP 6= 0 or π. This would be the first evidence of CP vi-
olation in the lepton sector. It is observed in the quark sector but is not
enough to explain the matter/antimatter asymmetry of our universe.

2. Nature’s Fundamental Underlying Symmetries. Insight into new sym-

metries beyond the Standard Model can be gained by measuring the
10 Chapter 1. Introduction

mixing parameters, determining the ordering of the neutrino masses,

and understanding how this relates to the analogous quark mixings.

3. Grand Unified Theories (GUTs). Theories that encapsulate all of the

forces predict protons should decay, which DUNE will be sensitive
enough to measure.

4. Supernovae. When stars explode they release an intense burst of neu-

trinos. Measuring this phenomena will help further understand how
stars collapse.

The world’s most intense neutrino beam will be fired from the Fermi Na-
tional Accelerator Laboratory (FNAL) near Batavia, Illinois 1300 km to the
Sanford Underground Research Facility (SURF) in Lead, South Dakota (Fig-
ure 1.3) [11]. The far site neutrino detector will be a multi-kiloton liquid ar-
gon time projection chamber (LArTPC) housed at SURF approximately one
mile underground (to shield from cosmic rays in our atmosphere). The ad-
vantage of having a giant detector and high-intensity beam is a high event
rate, which is essential for neutrino experiments since they hardly interact.
DUNE will be the longest baseline neutrino experiment which will aid in
measurements of δCP and the determination of the neutrino mass ordering [10].

1.2.1 Long-Baseline Neutrino Facility

LBNF will provide the infrastructure for DUNE including the neutrino
beam, a beam monitoring system, as well as the near and far sites which will
house the near and far detectors. These facilities will be located at Fermilab
(near site) or SURF (far site). The far site will host the LArTPC far detector
while the near site will be home to the beam, beam monitoring system, and
near detector.
1.2. The Deep Underground Neutrino Experiment 11

F IGURE 1.3: A neutrino beam will be fired from Fermilab in

illinois 1300 km to the Sanford Underground Research Facility
in Lead, South Dakota.

Near Site Facilities

The neutrino beam is created by accelerating protons to high energies
(60 to 120 GeV) by a series of accelerators and colliding them into a target
creating charged hadrons (mostly pions with some kaons). The pions and
kaons are focused with a series of electromagnetic horns into a 200 meter
pipe where a large fraction decay into muons and muon neutrinos. The mag-
netic horn focuses for either positive or negative charges depending on the
direction of the current and thus can select for a neutrino or antineutrino
beam:

π + → µ+ + νµ

π − → µ− + ν̄µ .

The resulting neutrinos and their associated muons continue in the same di-
rection where they encounter a hadron absorber designed to capture any re-
maining baryons or mesons. Neutrinos pass straight through to the near and
far detectors. Muons do make it through the absorber, but not much farther.
12 Chapter 1. Introduction

A system of monitors are set up in alcoves just beyond the absorber to mea-
sure these muons. An illustration of this process is shown in Figure 1.4.

F IGURE 1.4: High-energy protons collide with a fixed target

producing (mostly) charged pions. The direction of the horn’s
current selects for either positive or negative pions which cor-
respond to a neutrino or antineutrino beam. The selected pions
are focused into a 200 meter pipe where they each decay into
one muon and one neutrino. Any remaining hadrons are sub-
tracted by the absorber, the muons travel as far as the alcoves,
and the neutrinos make the journey to the far detector.

The near detector will be stationed near the origin of the neutrino beam
(beyond the alcoves) and will measure the beam’s initial flux and energy
spectrum. There are, however, uncertainties in the interaction rates and neu-
trino event rates are small. Muons don’t suffer this problem, the changes in
the beam can be seen much more quickly with muon monitors. This is also
a nice independent cross check of the beam flux from the near detector. To
measure the oscillation parameters it is crucial to understand the initial en-
ergy distribution and flavor composition of the neutrino beam. In addition,
data from the near detector will be used to measure neutrino interaction cross
sections.

1.2.2 Far Detector

The far detector is located 1300 km from the beam at SURF and will con-
sist of four 17 kiloton liquid argon time projection chambers (LArTPC) 1.5
km underground [10]. The detectors are underground to shield against any
1.2. The Deep Underground Neutrino Experiment 13

background noise from cosmic rays in our atmosphere. By the time the neu-
trinos arrive at the far detector from the near site (∼0.004 seconds) they will
have travelled mostly uninterrupted through Earth’s crust. They will enter
the LArTPC and undergo weak interactions ultimately producing a signal.
Time projection chambers are volumes with a constant electric field main-
tained by a series of anode and cathode wire planes. The electric field is in the
transverse direction relative to the beamline. Each volume will be filled with
liquid argon, which must be kept below ∼ −186◦ C in order to stay a liquid.
Neutrinos will undergo a weak interaction with an argon nucleus producing
a charged lepton dictated by the neutrino’s flavor as well as transmutating
the argon atom. The charged byproducts will create ionized electrons which
will drift to the detection planes. Combining these detections with timing in-
formation enables a three-dimensional reconstruction of the particle tracks.
Although most neutrinos undergo zero interactions on their way to the
far detector, their oscillations are affected as they travel through Earth. In
fact, neutrinos and antineutrinos are affected differently. This matter effect
typically inhibits measurements of δCP ; DUNE, due to its large baseline (1300
km), will be able to resolve this issue and measure δCP .
 15

Chapter 2

Gas Cherenkov Muon Monitor

In this chapter I will overview the physics of Cherenkov radiation and

how it can be applied to experimental particle physics. I will then describe
the specific detector that has been studied for this thesis.

2.1 Cherenkov Radiation

Pavel Cherenkov won the 1958 Nobel Prize in physics for the discovery
that charged particles moving in a medium faster than light emit electro-
magnetic radiation [13]. This phenomenon is called Cherenkov radiation. A
charged particle moving through a dielectric medium with index of refrac-
tion n polarizes the molecules in the medium. Once the particle has passed,
the molecules return to their unpolarized state by emitting photons. If the
particle’s velocity is greater than the speed of light in that medium (v > c/n)
the result is constructive interference of Cherenkov radiation as a coherent
wavefront at an angle θc relative to the particle’s trajectory (Figure 2.1) [9].
The light is emitted in a cone, much like shockwaves are sent out in a cone
when a jet travels faster than the speed of sound, and is characterized by the
Cherenkov angle θc . From this geometrical relationship we see that cos(θc ) =
1/nβ. We can derive the particle’s momentum threshold p T , the minimum
momentum a particle must have to undergo Cherenkov radiation. Note that
cos(θc ) is bounded above by one, which means at threshold v T = c/n and,
16 Chapter 2. Gas Cherenkov Muon Monitor

F IGURE 2.1: A charged particle traveling in a medium with in-

dex of refraction n emits light in a cone at the Cherenkov angle,
θc . The particle and photon travel a distance βct and ct/n, re-
spectively, where β = vc .

p T = γmv T
mc
pT = p
n 1 − β2
mc
pT = √ .
n2 − 1

Therefore, only a particle of momentum p > p T will produce Cherenkov ra-

diation. For some experiments, this fact is exploited to help identify particles
of momentum p: only particles with mass

mc < (n2 − 1)1/2 p

will emit Cherenkov raditation. Further, the number of photons produced

per unit length (via this process) is given by the Frank-Tamm formula [12]
2.1. Cherenkov Radiation 17

dN 2παz2  1
Z 
= 1 − dλ
dx λ2 β2 n2 ( λ )

dλ
Z λ2
= 2παz2 sin2 (θc )
λ1 λ2

where z is the charge of the incoming particle, α is the fine structure constant,
n is the index of refraction as a function of photon energy, and the photon
wavelength is integrated over values for which β > 1/n(λ). The intensity
of the emitted light grows with frequency, this is why nuclear reactors make
water glow blue (blue and violet being the highest frequency the eye can
detect). The number of radiated photons also grows with the Cherenkov
angle; particles near the momentum threshold will emit very little light.
Cherenkov detectors can be utilized in different ways for neutrino physics.
The Super-Kamiokande experiment in Japan is a 50 kiloton water Cherenkov
detector. The water is in a large cylindrical volume with thousands of photo-
mulitplier tubes (PMTs) lining the inside walls. A neutrino will undergo an
elastic scattering interaction producing a super-luminal electron. This elec-
tron emits Cherenkov radiation which is detected as rings by the PMTs. This
is called ring-imaging and the number of detected photons provides a mea-
sure of the neutrino energy while the direction of the electron can be deter-
mined from the orientation of the Cherenkov ring (Figure 2.2) [9].
The detector discussed in this paper consists of a pipe filled with argon
gas. Its ability to pivot relative to the beamline and adjust the gas pressure
(i.e. change n) offers the potential to constrain the muon (and ultimately the
neutrino) beam profile as I will discuss next.
18 Chapter 2. Gas Cherenkov Muon Monitor

F IGURE 2.2: (a) Schematic of the Super-Kamiokande water tank

and (b) illustration of ring-imaging from an electron neutrino
event. Photo from Modern Particle Physics by Mark Thomson. [9]

2.2 Gas Cherenkov Detector

The detector used for this project consists of an L-shaped chamber filled
with argon gas (Figure 2.3). It is located in the NuMI beamline at Fermilab.
NuMI (Neutrinos at the Main Injector) is an existing facility which provides
a neutrino beam pointed to detectors in northern Minnesota. Attached at the
elbow is a mirror and at the end of the leg perpendicular to the beamline is a
photo-multiplier tube (PMT) which counts the number of photons. The PMT
is sensitive to the visible spectrum (300 to 700 nm). Charged particles (in
our case, muons) enter the one-meter-long leg along the beamline and emit
photons at some angle θc . Photons are then reflected towards the PMT. The
perpendicular leg is long (∼5 m) to filter out any noise from the high radia-
tion of the beam (it is important that our signal comes only from Cherenkov
radiation due to the beam). The fact that the muons emit photons at some
angle θc which must strike the mirror and be perfectly reflected to hit the
PMT make this design sensitive to muon momentum and incident angle, de-
tector orientation, as well as the pressure of the gas. The detector is allowed
2.2. Gas Cherenkov Detector 19

to move in pitch (0.828◦ to 6.011◦ ) and yaw (-6.064◦ to 4.695◦ ) via actuators.
The gas inside the chamber is allowed to vary in pressure (0 to 200 psi) which
in turn changes the index of refraction n.

HIT!

𝜃
PMT

Muon

Photon

Mirror

F IGURE 2.3: A detector filled with argon gas designed to mea-

sure Cherenkov radiation from a muon beam (red). Muons
enter the argon medium and undergo Cherenkov radiation
(green) which reflect off the mirror (blue) to be recorded by a
PMT.

We have been taking data from the gas Cherenkov detector in the NuMI
beamline (Figure 2.4). It is located in Alcove 2, approximately 14 meters
downstream from the hadron absorber (Figure 2.5). It is roughly centered
with the beamline. One application of this device is to monitor beam sta-
bility. Since muons and their neutrinos are produced in a one-to-one ratio
and (for the most part) share the same direction, a measurement of the muon
beamline is a good approximation of the neutrino beamline; their energies
20 Chapter 2. Gas Cherenkov Muon Monitor

are correlated in the two body decays (π → µν). Using data from the detec-
tor, profiles of the beam were created (shown in Figure 2.6). Note the signal is
negative; a large signal corresponds to a more negative value. This is because
the PMTs read out a negative voltage due to the photoelectric effect. This is
a nice check to see that the beam is running properly. We can also point to
where the beam deviates and, from being aware of what was going on at the
time, what the cause was (Figure 2.7).

F IGURE 2.4: This is the gas Cherenkov detector currently in

the NuMI beamline at Fermilab located in Alcove 2. It is opera-
tional and the data is used to monitor beam stability. It will also
be used to extrapolate the flux of the muon beam by performing
pressure scans.

In the next chapter I will discuss further utility of a gas Cherenkov detec-
tor; specifically, how it can be used to extrapolate a muon distribution of the
beam. This information will help constrain the flux of the neutrino beam.
2.2. Gas Cherenkov Detector 21

F IGURE 2.5: Top view of particle beam from end of decay pipe
through muon alcoves. Undecayed hadrons are absorbed after
decay pipe, leaving a beam of muons and neutrinos. Neutri-
nos travel through unhindered to near and far detectors while
muons are measured before being absorbed by Earth.
22 Chapter 2. Gas Cherenkov Muon Monitor

F IGURE 2.6: Here is a profile of a relatively stable muon beam.

This is used as a nice, and quick, check to ensure the neutrino
beam is operating as expected.
2.2. Gas Cherenkov Detector 23

F IGURE 2.7: This is a profile of an "erratic" muon beam labeled

with events to explain such behavior.
 25

Chapter 3

Predicting a Muon Distribution

In this chapter I will discuss how to develop a model to predict a muon

distribution (in momentum and direction) utilizing a gas Cherenkov detector
and what future work will be needed to accomplish this task. The general
idea is that particular muon distributions will generate unique signals in our
detector over various pressures and yaws (the signals are like a "finger print"
of some muon distribution). Our goal then is to extrapolate, or constrain,
the muon beam distribution by observing the subsequent signal from a gas
Cherenkov detector over various orientations and pressures (look at output
signal and reconstruct muon distribution in momentum and direction). This
in turn will help constrain the associated neutrino distribution. This model
is computationally intensive and I will show how time-consuming Monte
Carlo simulations can be significantly shortened.

3.1 Pressure Scans at NuMI

The first step was to take data with the gas Cherenkov detector in the
NuMI beamline over several yaws and pressures. The detector was operated
remotely from Boulder, Colorado and the pitch was fixed to be (roughly)
centered with the beamline. We took data over seven pressures (8, 16, 32,
60, 100, 150, and 200 psi) and let the yaw vary from approximately -5 to
3 degrees relative to the beamline center. Data was taken from a neutrino
26 Chapter 3. Predicting a Muon Distribution

( A ) Yaw scan taken at 60 psi.

( B ) Yaw scan taken at 200 psi.

F IGURE 3.1: Yaw scans taken during neutrino mode (i.e. an-
timuon beam) at 60 and 200 psi. Note the signal reads out
from a negative voltage; a more negative value corresponds to
a larger signal.

(and later from an antineutrino) beam shown in Figure 3.1. The plots show
the integrated signal per protons on target (POT) versus the detector’s yaw
orientation which is measured in degrees relative to the beamline center (note
that a more negative value means a larger signal).
The signal grows and makes a more pronounced "W" shape at higher
pressures. This makes sense: Cherenkov radiation, as well as the number
of photons radiated per muon, depends on the particle’s momentum and the
index of refraction n of the medium (which is related to pressure). We know
3.2. Monte Carlo Simulations 27

that only high momentum muons will contribute to the signal at low pres-
sures [14],

mc  P 
pT = √ , n( P) = (n Ar − 1) × +1
n2 − 1 14.7 atm

where n Ar is the index of refraction of argon at λ ∼400 nm, and P is the

pressure of the argon inside the detector in pounds per square inch (psi).
Note that n grows with pressure.
The "W" shape makes sense because the detector is highly sensitive to the
Cherenkov angle; the largest signals will occur where the Cherenkov angle is
"just right" relative to the muon’s incident angle and detector orientation. The
signal is symmetric relative to the beamline center because light is radiated
in a cone. At larger pressures the momentum threshold decreases and more
muons will exhibit Cherenkov radiation. A greater Cherenkov angle (which
increases with pressure for a given muon momentum) also means the num-
ber of emitted photons per muon will increase according to the Frank-Tamm
formula; these factors explain the larger signal and more pronounced "W" at
higher pressures.

3.2 Monte Carlo Simulations

These plots can be recreated with Monte Carlo simulations using the com-
puter program Geant4, a platform designed to simulate particles through
matter an ubiquitously used throughout the particle physics community. Us-
ing the same pressures that were used with the detector at Fermilab, pres-
sure scans were performed over similar yaw angles in half-degree increments
(Figure 3.2). The muon flux in the decay pipe used for this recreation came
from a beam simulation. The muons were then simulated through a gas
28 Chapter 3. Predicting a Muon Distribution

Cherenkov detector in Alcove 2 and a "signal" was produced by counting

the number of photons hitting the PMT. As is shown in Figure 3.3, we were
able to produce the same shapes generated from the real detector. More sim-
ulations (statistics) would probably result in a better shape.

F IGURE 3.2: An initial distribution of neutrinos in the decay

pipe was obtained from a beam simulation. This was converted
into a muon distribution via conservation of energy and mo-
mentum. Yaw scans were then performed over seven different
pressures and eleven yaw orientations each. Each blue square
corresponds to a Monte Carlo simulation involving over one
million muons and takes approximately two hours to run.

The simulations accurately describe our setup at NuMI. But how do we

extrapolate a muon distribution from a signal? Well, we will ultimately have
to fit a simulated signal to the real signal by adjusting the muon distribu-
tion. This is in essence extremely time consuming: simulate a large number
of muons through the detector about 70 times (seven pressures and ten yaw
orientations), see if the signal agrees with the real output, if not adjust muon
distribution and repeat until they match, then you have found your distri-
bution. Simulating one million muons for a single configuration takes about
3.2. Monte Carlo Simulations 29

( A ) Yaw Scan at 150 psi.

( B ) Yaw Scan at 8 psi.

F IGURE 3.3: These plots show that yaw scans from Monte Carlo
simulations (dark blue squares) recreate the same shape taken
with real data (blue crosses).

two hours, so time adds up quickly (could run in parallel, still long).
To reduce the computing time, we decided to build a four-dimensional
yield matrix. For a given muon momentum and direction (p, θ) as well as
detector configuration in yaw and pressure (Y, P), our yield matrix assigns
each muon a "hit" or signal. To build this matrix we ran simulations over
discrete values in (p, θ, P, Y):
30 Chapter 3. Predicting a Muon Distribution

p = (1, 2, 4, 8, 12, 18) GeV

θ = (0.1, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5) degrees from beam center

P = (8, 16, 32, 60, 100, 150, 200) psi

Y = (0.5, 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5) degrees from beam center.

This comes out to 4,620 files with 100,000 muons in each file. This allowed us
to build a matrix in (p, θ)-space such that for some detector configuration in
yaw and pressure, we can assign each muon a "signal" based on the muon’s
momentum and direction as shown in Figure 3.4. The signal we create with
the yield matrix sums over all muon momenta and direction for a given pres-
sure and yaw,

Signal( P, Y ) = ∑ Nµ ( p, θ ) × YM( p, θ, P, Y )
p,θ

where Nµ is the number of muons as a function of p and θ and is the im-

portant parameter we are attempting to extract (we want the Nµ that pro-
duces the signal which best matches our data). The function YM is our four-
dimensional yield matrix in charge of assigning a signal. Although the ma-
trix is built from discrete values, we can assign any muon a number of hits
(signal) via interpolation. If we find that we are low on statistics, or resolu-
tion in (p, θ), we can always run more muons per file (or add more values
of p and θ). Comparing the time it takes to output a signal via Monte Carlo
simulations (which takes on the order of hours for one million muons) with
this yield matrix technique (assigns a signal in a matter of minutes) it is easy
to appreciate the advantages of the latter method.
3.2. Monte Carlo Simulations 31

F IGURE 3.4: Yield matrix in (p, θ)-space at 200 psi and yaw of -
4◦ . The z-axis is the number of photon detections (or hits) which
correspond to an output signal. Although this matrix was built
with discrete values of p and θ, muons with intermediate values
are assigned a signal via interpolation.

Using the yield matrix is a quick way of creating a signal for a large num-
ber of muons, but we still have yet to predict a muon distribution. To do
this, I created a function fitted to the "realistic" distribution file from NuMI
(Figure 3.5). The end result is a distribution that says the number of muons
as a function of momentum and direction is,

 
Nµ ( p, θ ) = a( p)θ + b( p)θ 2 + c( p)θ 3 + d( p)θ 4 × 2.2−θ/σ( p) (3.1)

where the coefficients of the polynomial of θ are functions of the muon’s mo-
mentum (and so is the exponent). The function was then fitted over several
momentum regimes (divided in such a way that each regime had approx-
imately the same number of muons) shown in Figures 3.6 and 3.7. Each
regime resulted in different parameters, and we fitted for the parameters
themselves (Figure 3.8).
The end result is the following eight parameter function,
32 Chapter 3. Predicting a Muon Distribution

F IGURE 3.5: Muon distribution used to run Monte Carlo sim-

ulations. The colors indicate the number of muons in each bin
according to the particle’s momentum and direction, θ. Most
muons have an angle less than 5◦ off from beamline center.


Nµ ( p, θ ) = ([0] p + [1] p3 )θ

+ ([2] p2 + [3] p3 )θ 2
(3.2)
2 3 3
+ ([4] p + [5] p )θ

+ ([6] p2 + [7] p3 )θ 4 ∗ 2.2−θ ∗0.558p

Shown below in Figure 3.9 are plots of simulated yaw scans using the
same distribution of 100,000 muons distributed according to Equation 3.2.
There are two output signals, one generated the long way with Geant4, and
the short way using the yield matrix. The disagreement is the largest where
the signal is the biggest (most negative). The error can probably be reduced
with a higher resolution in (p, θ) for the yield matrix.
To demonstrate that a muon distribution is capable of being extrapolated
from a yaw scan, Figure 3.10 shows how the signal changes when the param-
eters in Equation 3.2 are altered. If there was more time, a fitter would have
3.2. Monte Carlo Simulations 33

F IGURE 3.6: Number of muons vs θ in different momentum

ranges.

been constructed to match an output signal with a best-fit muon distribution.

The ultimate goal is to fit for the parameters in 3.2 to data (the output signal).
34 Chapter 3. Predicting a Muon Distribution

F IGURE 3.7: Number of muons vs θ in different momentum

ranges.
3.2. Monte Carlo Simulations 35

F IGURE 3.8: Plots showing best-fit values for parameters.

36 Chapter 3. Predicting a Muon Distribution

Yaw Scan at 60 psi

0

-1000

-2000

-3000
Nγ

-4000

-5000

-6000

-7000

-5 -4 -3 -2 -1
Yaw (Degrees)

Yaw Scan at 150 psi

0

-2000

-4000

-6000
Nγ

-8000

-10000

-12000
-5 -4 -3 -2 -1
Yaw (Degrees)

F IGURE 3.9: Yaw scans comparing Monte Carlo simulations

(blue) with yield matrix technique (red). The largest disagree-
ment ( 20%) occurs where the signal is the greatest. This can
probably be corrected for better (p, θ) resolution when building
the yield matrix.
3.2. Monte Carlo Simulations 37

Yaw Scan at 100 psi

0

-1000

-2000

-3000
Nγ

-4000

-5000

-6000

-7000

-5 -4 -3 -2 -1
Yaw (Degrees)

F IGURE 3.10: This plot illustrates how altering the muon dis-
tribution (in this case the exponential in Equation 3.2) changes
the output signal from a yaw scan. The red line corresponds
to a nominal value for the exponent. The green and pink lines
correspond to larger and smaller values, respectively, relative
to the nominal value. Future work will build upon this model
to develop a fitter which matches an output signal with a muon
flux.
 39

Chapter 4

Conclusion and Future Work

This thesis has presented the utility of a gas Cherenkov detector for the
upcoming DUNE neutrino beam. It has been shown that the detector in the
NuMI beamline is able to monitor the beam’s stability. Also presented is
a method for extracting a muon flux, which would ultimately be used to
help constrain the beam’s neutrino flux. Although this ultimate goal was not
accomplished, the method was proven to be promising. There is no doubt
that future work to build upon this model will succeed in characterizing the
muon distribution from the output signal.
Although the near detector can be used as a beam monitor, muon moni-
tors accomplish this task much quicker and at a fraction of the cost. Neutrino
detection is difficult because they hardly interact and so the near detector re-
lies on a low event rate to monitor the beam. This is not so for muons, they
are easy to observe and can measure the beam (as a byproduct of the neutrino
beam) much quicker. This will help ensure the beam is working as expected.
Knowing the initial neutrino beam energy spectrum and flavor composi-
tion is crucial for DUNE to make accurate measurements of neutrino oscilla-
tion parameters. This is also done at the near detector but can be aided with
a gas Cherenkov muon detector. Specific muon distributions correspond to
unique signals in the detector over several pressures and orientations. A
computer model has been developed which will ultimately result in fitting a
muon distribution to any given signal. This can be done, in principle, with
40 Chapter 4. Conclusion and Future Work

Monte Carlo simulations but it would take forever. This paper has shown
that generating a muon distribution from some function and turning it into a
signal (the fitter would actually do the reverse) saves a significant amount of
time.
Future work will be needed to finish this project. The next step would
be to develop a fitter which takes a signal (over various detector orientations
and pressures) and matches it with some muon distribution according to the
functional parameters. This would be done with computer simulations first
to make sure the predicted muon flux is accurate. After this, the model will
be applied to real data.
 41

Bibliography

[1] Los Alamos Science No. 25. Celebrating the Neutrino: The Reines-Cowan
Experiments (1997).
http://permalink.lanl.gov/object/tr?what=info:lanl-
repo/lareport/LA-UR-97-2534-02.

[2] European Organization for Nuclear Research (CERN).

https://home.cern/about/physics/standard-model.

[3] D. Griffiths, Introduction to Elementary Particles. Wiley-VCH, 2 ed., 2010.

[4] R. Garisto, Focus: Neutrinos Have Mass. Phys. Rev. Focus 2(10), Septem-
ber 1998. https://physics.aps.org/story/v2/st10.

[5] R. Davis, Physical Review 12(11). 303 (1964).

[6] B. Pontecorvo, Inverse beta processes and nonconservation of lepton charge.

Soviet Physics JETP (1958).

[7] Z. Maki;M. Nakagawa;S. Sakata, Remarks on the Unified Model of Elemen-

tary Particles. Progress of Theoretical Physics, Vol. 28, No. 5, November
1962.

[8] D. H. McIntyre Quantum Mechanics. Pearson Education Inc.

[9] M. Thomson, Modern Particle Physics. Cambridge University Press, 2013.

[10] Deep Underground Neutrino Experiment. http://www.dunescience.org/.

[11] Long-Baseline Neutrino Facility. http://lbnf.fnal.gov/beam.html.

42 BIBLIOGRAPHY

[12] Particle Data Group, Physical Review D: Particles, Fields, Gravitation, and
Cosmology. American Physical Society, July 2012.

[13] P. A. Cherenkov, Physical Review 52(4), 378 (1937).

[14] Bideau-Mehu et al., Measurement of refractive indices of neon, argon, kryp-

ton and xenon. Journal of Quantitative Spectroscopy and Radiative Trans-
fer, 25(5), 1981.

[15] J. Boissevain et al., NuMI Alcove 1 Prototype Cherenkov Muon Monitor.

November 2015.

[16] E. Kemp, The Deep Underground Neutrino Experiment

DUNE: the precision era of neutrino physics. September 2017.
https://arxiv.org/pdf/1709.09385.pdf.

[17] C. Giganti, Neutrino oscillations: the rise of the PMNS paradigm. November
2017. https://arxiv.org/pdf/1710.00715.pdf.

[18] P. S. Madigan, Measuring the muon flux of neutrino beams with a novel gas
Cherenkov detector. November 2015. Undergraduate Honors Theses. Pa-
per 986.

[19] K. Dochen, Diamond Muon Monitors for the Deep Underground Neutrino
Experiment. April 2017. Undergraduate Honors Theses. Paper 1326.

[20] B. T. Cleveland et al., Measurement of the Solar Electron Neutrino Flux

with the Homestake Chlorine Detector. American Astronomical Society 496
(March 1998). http://iopscience.iop.org/article/10.1086/305343/pdf.