UNIVERSIDAD NACIONAL AUTÓNOMA DE MÉXICO
POSGRADO EN CIENCIAS FÍSICAS
INSTITUTO DE CIENCIAS NUCLEARES
PHYSICS WITH REACTOR NEUTRINOS USING
CHARGE-COUPLED DEVICES AND ITS SYNERGY WITH
DARK SECTOR SEARCHES
TESIS
QUE PARA OPTAR POR EL GRADO DE:
DOCTORA EN CIENCIAS (FÍSICA)
PRESENTA:
BRENDA AUREA CERVANTES VERGARA
TUTOR PRINCIPAL:
DR. JUAN CARLOS D’OLIVO SÁEZ, ICN, UNAM
MIEMBROS DEL COMITÉ TUTOR:
DR. ALEXIS ARMANDO AGUILAR ARÉVALO, ICN, UNAM
DR. ERIC VÁZQUEZ JÁUREGUI, IF, UNAM
CIUDAD UNIVERSITARIA, CD. MX., SEPTIEMBRE 2023
 
UNAM – Dirección General de Bibliotecas 
Tesis Digitales 
Restricciones de uso 
  
DERECHOS RESERVADOS © 
PROHIBIDA SU REPRODUCCIÓN TOTAL O PARCIAL 
  
Todo el material contenido en esta tesis esta protegido por la Ley Federal 
del Derecho de Autor (LFDA) de los Estados Unidos Mexicanos (México). 
El uso de imágenes, fragmentos de videos, y demás material que sea 
objeto de protección de los derechos de autor, será exclusivamente para 
fines educativos e informativos y deberá citar la fuente donde la obtuvo 
mencionando el autor o autores. Cualquier uso distinto como el lucro, 
reproducción, edición o modificación, será perseguido y sancionado por el 
respectivo titular de los Derechos de Autor. 
 
  
 
ii
Acknowledgements 1
Mis logros personales y profesionales, incluida esta tesis doctoral, son producto de un
esfuerzo personal alentado, en varias ocasiones, por diferentes personas en mi vida. Es
a ellas a quienes quiero agradecer en estos párrafos.
A mi FAV, por tus constantes porras, por esperarme para comer-cenar después de las 10
p.m. en los días de redacción intensa, por escucharme hablar de los temas de esta tesis,
aunque pareciera "hablar en chino"; gracias por compartir la vida y, particularmente,
esta aventura conmigo, Blacky y Holly; tenerlos cerca me alimenta el corazón.
A mi pequeño ved, por sentarte junto a mí, físicamente o a la distancia, a escribir nuestras
respectivas tesis; escucharte hablar de tu trabajo apasionadamente es inspiración para
mí. A mis padres, Maripaz y Aurelio, por dar lo mejor de ustedes, transformarse y
acompañarme; gracias por promover nuestro crecimiento en un ambiente enriquecedor.
A ustedes tres, mi ohana, les agradezco su escucha, aliento y apoyo incondicional.
A mi tutor principal, el Dr. Juan Carlos D’Olivo, por su total apoyo durante mis estudios
de posgrado, la realización de esta tesis y por alentar mi desarrollo profesional. A mi
comité tutor, por su asesoramiento durante mi doctorado. A los miembros de mi sínodo,
por el tiempo dedicado a la revisión de esta tesis.
A Juan Estrada, por su guía, asesoramiento y constante apoyo en mis actividades profe-
sionales; gracias por fomentar mi crecimiento laboral. A los miembros de las colabora-
ciones de los experimentos Oscura, CONNIE y DAMIC, por el conocimiento intercam-
biado diariamente en fructíferas discusiones.
A la UNAM como institución, por ser mi casa de formación educativa desde los 11 años
y por las oportunidades y vivencias que me enriquecieron personal y profesionalmente.
A mis amigos en México, con quienes en algún momento compartí aula en la UNAM y
que han seguido acompañándome día a día, gracias por estar y permanecer.
A Fermilab como institución, por las oportunidades profesionales que me ha ofrecido. A
mis amigos latinos en Fermilab, con quienes he compartido a diario en los últimos dos
años, por crear un ambiente motivador, de cariño y apoyo, aún estando lejos de casa.
Al CONAHCYT, a través del PNPC, por la beca recibida durante mis estudios de
doctorado y por el proyecto CF-2023-I-1169. Los temas de investigación de esta tesis
han sido apoyados por los proyectos DGAPA-UNAM PAPIIT IN106322 y IN104723.
1This section is intentionally written in spanish.
iii
iv
Contents
Introduction 1
1 New physics in the low-energy neutrino sector 3
1.1 The discovery of the neutrino . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Natural and artificial sources of neutrinos . . . . . . . . . . . . . . . . . 4
1.3 Low-energy neutrino interactions . . . . . . . . . . . . . . . . . . . . . . 5
1.3.1 Elastic neutrino-electron scattering (EνES) . . . . . . . . . . . . 6
1.3.2 Coherent elastic neutrino-nucleus scattering (CEνNS) . . . . . . 7
1.3.3 Quasielastic neutrino-nucleon scattering . . . . . . . . . . . . . . 9
1.4 Beyond the Standard Model (BSM) physics . . . . . . . . . . . . . . . . 10
1.4.1 Neutrino non-standard interactions (NSI) . . . . . . . . . . . . . 11
1.4.2 New light mediators . . . . . . . . . . . . . . . . . . . . . . . . . 12
Scalar mediator . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Vector mediator . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4.3 Neutrino electromagnetic properties . . . . . . . . . . . . . . . . 15
Magnetic moment . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Millicharge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2 Reactor neutrino experiments 17
2.1 Reactor ν̄e flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.1 Commercial reactor flux from a summation approach . . . . . . . 19
2.2 Ongoing/completed low-threshold experiments . . . . . . . . . . . . . . . 21
2.2.1 CONUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
v
vi CONTENTS
2.2.2 NCC-1701 (Dresden-II) . . . . . . . . . . . . . . . . . . . . . . . 22
2.2.3 νGEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.4 RED-100 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.5 NEON . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3 CONNIE with standard CCDs 25
3.1 CCDs structure, functioning and packaging . . . . . . . . . . . . . . . . 25
3.2 Experimental setup, operation and data calibration . . . . . . . . . . . . 28
3.3 Expected event rate in CONNIE . . . . . . . . . . . . . . . . . . . . . . 30
3.3.1 Nuclear recoil quenching factor in Si . . . . . . . . . . . . . . . . 31
3.4 Results from the 2016-2018 run . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 Results from the 2019 run . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.6 Search for light mediators with CEνNS channel . . . . . . . . . . . . . . 37
4 Skipper CCDs 41
4.1 Standard vs. Skipper CCD output stage and readout . . . . . . . . . . . 42
4.2 Instrumental sources of Single-Electron Events (SEE) . . . . . . . . . . . 44
4.2.1 Thermal dark current . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2.2 State traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2.3 Spurious charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.4 Amplifier light . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3 Few-electron events from instrumental SEE . . . . . . . . . . . . . . . . 48
4.3.1 Accidental coincidences . . . . . . . . . . . . . . . . . . . . . . . 48
4.3.2 Misidentified events due to readout noise . . . . . . . . . . . . . . 49
5 CONNIE with skipper CCDs 51
5.1 Data taking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.2 Data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
CONTENTS vii
5.2.1 Masking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3 Event extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.4 Data analysis and selection cuts . . . . . . . . . . . . . . . . . . . . . . . 55
5.4.1 Noise and SER stability . . . . . . . . . . . . . . . . . . . . . . . 56
5.4.2 Detection efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.4.3 Impact of cuts on ROFF background spectrum . . . . . . . . . . 57
5.5 CEνNS forecasted sensitivity . . . . . . . . . . . . . . . . . . . . . . . . 58
6 Synergy with dark sector searches 61
6.1 Direct dark matter searches . . . . . . . . . . . . . . . . . . . . . . . . . 61
6.1.1 Skipper CCD experiments at underground laboratories . . . . . . 63
SENSEI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
DAMIC-M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
DAMIC at SNOLAB . . . . . . . . . . . . . . . . . . . . . . . . . 65
Oscura . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
6.2 Search for millicharged particles with skipper CCDs . . . . . . . . . . . 68
7 Future large skipper CCD experiment 71
7.1 Characterization of skipper CCDs from new foundries . . . . . . . . . . 71
7.1.1 Readout noise and speed . . . . . . . . . . . . . . . . . . . . . . . 72
7.1.2 Exposure-dependent single-electron rate . . . . . . . . . . . . . . 73
7.1.3 Spurious charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.1.4 Traps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.1.5 Charge transfer inefficiency . . . . . . . . . . . . . . . . . . . . . 78
7.1.6 Buried-channel potential . . . . . . . . . . . . . . . . . . . . . . . 78
7.1.7 Transistor curves . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
7.1.8 Amplifier light . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
viii CONTENTS
7.2 Massive skipper CCD instrument . . . . . . . . . . . . . . . . . . . . . . 83
Final remarks 87
Bibliography 89
Introduction
The study of neutrinos, fundamental particles with intriguing properties, has been a
subject of great interest in the field of particle physics in the last ∼90 years, since
Pauli proposed their existence [1]. The discovery of neutrino oscillations [2,3], implying
a non-zero neutrino mass, and the fact that their mass seems to be very tiny, have
challenged the completeness of the most well-established theory up-to-date that describes
the elementary particles and their interactions, the Standard Model (SM).
Reactor neutrinos, decay products of processes occurring at nuclear reactors’ cores, have
been historically used to study neutrino properties and interactions, as reactors are one
of the most intense, pure, and well-understood neutrino sources. Reactor neutrinos have
low energies, below 4 MeV, and their interactions with matter, well-described by the SM
theory, can only be studied with low-threshold detectors. Any experimental observation
of a deviation from the SM prediction would be significant evidence of a new discovery.
Charge-coupled devices (CCDs), pixelated semiconductor sensors, have found extensive
use in astronomy since the ∼1980s as they present appealing characteristics such as low
readout noise and excellent spatial resolution. The development of thick CCDs and the
introduction of the skipper CCD technology, which offers electron-counting capability,
have made possible to study weakly interacting particles using these sensors [4,5]. This
includes the study of a possible “dark sector”, consistent of new particles and forces,
which has emerged as a strong research area due to its potential to provide insights into
some experimental findings, including dark matter and neutrino masses.
In this work, we present the capability of CCD and skipper CCD detectors to contribute
to the understanding of the Universe by studying reactor neutrino interactions, within
and beyond the SM, focusing on the results of the Coherent Neutrino-Nucleus Interaction
Experiment (CONNIE) [6–9]. Additionally, we provide an overview of the use of skipper
CCDs to explore the dark sector at other facilities [10–14] and present the current efforts
towards building massive skipper CCD detectors with reduced backgrounds [15–17].
This thesis is structured into seven chapters. Chapter 1 provides a brief historical review
of the neutrino’s discovery and its flavors. It discusses the energy range of various neu-
trino sources. The effective cross sections of the main low-energy neutrino interactions
with matter are presented within the framework of the SM. Furthermore, the modifica-
tion of these cross sections is discussed when considering physics beyond the SM, such
as new interactions, new mediators, or electromagnetic properties of neutrinos.
Chapter 2 introduces the experiments that study neutrinos from nuclear reactors. It
1
2 INTRODUCTION
presents the calculation of the neutrino flux from a commercial reactor and summarizes
the results of low-threshold experiments, highlighting their capability to explore new
physics.
Chapter 3 provides a detailed presentation of the CONNIE experiment. This exper-
iment, in which the Institute of Nuclear Sciences (ICN) at UNAM has been actively
involved since 2014, focuses on the search for reactor neutrino interactions using CCDs.
The chapter provides a description of the structure and functioning of charge-coupled
devices developed for particle physics applications. It outlines the general procedure for
calculating the expected event rate in CONNIE given an interaction. Furthermore, it
presents the results of the CONNIE data analysis corresponding to two acquisition peri-
ods using standard CCDs, 2016-2018 [7] and 2019 [9]. Finally, it presents the CONNIE
exclusion limits in the parameter space of light mediators [8], scalar and vector, which,
at the time of publication, were the most stringest constraints among experiments study-
ing the coherent elastic neutrino-nucleus scattering (CEνNS) in the low-mass mediator
region.
Chapter 4 introduces the electron-counting skipper CCD technology. It highlights the
differences between the standard and the skipper output stages in a CCD and explores
the main instrumental sources of single-electron events in these sensors. The chapter
also discusses how these sources can lead to events with few electrons due to accidental
coincidences or the effect of readout noise.
Chapter 5 focuses on the CONNIE detector upgrade with two skipper CCDs, carried
out in July 2021. It provides an overview of the methods used for data acquisition, pro-
cessing, and analysis. Preliminary results obtained from the analysis of data collected
between November 2021 and May 2023 are presented. Finally, based on the detection
efficiency of these sensors, the confidence level at which CONNIE with skipper CCDs
could measure the CEνNS as a function of exposure is calculated. The CONNIE collab-
oration expects to publish the results with skipper CCDs soon.
Chapter 6 focuses on the synergy between skipper CCDs and the exploration of the dark
sector. It summarizes the current status of direct dark matter searches, with emphasis
on underground laboratory experiments using skipper CCDs to explore dark matter
interactions with electrons. It also discusses the capability of using skipper CCDs to
explore millicharged particles [18].
Lastly, chapter 7 presents the recent advancements in building massive (multi-kilogram)
skipper CCD experiments. The characterization of the first mass-produced sensors from
new foundries is described in detail, and their performance is discussed [16, 19]. Ad-
ditionally, this chapter presents the performance of the largest skipper CCD detector
(∼80 g and 160 channels) [17], recently commissioned at the Fermi National Accelerator
Laboratory (FNAL).
Chapter 1
New physics in the low-energy neutrino sector
Neutrinos are light neutral leptons that interact via the weak and gravitational forces.
They are the most abundant massive particles in the Universe, until now known. How-
ever, as the weak force is short-range and neutrinos have very small masses, less than
1.1 eV [20], they rarely interact with other particles. This is why their detection is
challenging. The study of these elusive particles is of great interest because they are a
window to new physics.
1.1. The discovery of the neutrino
In 1930, the Austrian physicist Wolfgang Pauli proposed the existence of a light neu-
tral particle to compensate the apparent loss of energy and momentum in the β decay
of neutrons [21]. Enrico Fermi called Pauli’s particle the “neutrino” and, in 1933, he
presented his theory of β decay [1] which assumed the existence of this particle. In the
current terminology, this process is written as n → p+ + e− + ν̄e. Therefore, Pauli’s
particle corresponds to the electron antineutrino, ν̄e.
The first experimental confirmation of its existence occurred more than 20 years later, in
1956, when Clyde Cowan and Frederick Reines [22]1 observed the signature of the inverse
beta decay, i.e. ν̄e + p+ → n+ e+, produced by the interaction of electron antineutrinos
coming from the Savannah River nuclear reactor with protons in two water tanks.
In 1962, L. M. Lederman, M. Schwartz and J. Steinberger experimentally showed the
existence of more than one type of neutrino when detecting muons, associated to muon
neutrinos, in an aluminum spark chamber at the Brookhaven National Laboratory2.
The muon neutrinos were the decay products of pions and kaons that were produced by
15 GeV protons hitting a Be target [24].
Almost 40 years later, in 2000, the DONUT collaboration announced the existence of the
tau neutrino. Using a nuclear emulsion they detected taus, associated to tau neutrinos
coming from the decays of charmed mesons that were produced by 800 GeV protons
11995 Nobel Prize in Physics: “for the detection of the neutrino” [23]
21988 Nobel Prize in Physics: “for the neutrino beam method and the demonstration of the doublet
structure of the leptons through the discovery of the muon neutrino” [23]
3
4 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
from the Tevatron at the FNAL hitting a tungsten beam dump [25]. This announcement
confirmed the prediction of the existence of three active neutrinos from the results of
the Z boson decay studies at CERN’s Large Electron-Positron Collider [26–29].
1.2. Natural and artificial sources of neutrinos
At Earth, the neutrino flux that we constantly receive has its origin in a wide variety of
sources, both natural and human-made, and covers a broad energy range [30]. This is
illustrated in the Grand Unified Neutrino Spectrum. shown in Fig. 1.1.
Figure 1.1. Taken from [30]: Grand Unified Neutrino Spectrum.
There are many neutrinos coming directly from nature. The Big Bang model predicts the
existence of neutrinos with energies below 100 meV. These are relic neutrinos, decoupled
approximately 1 second after the Big Bang, which conform the Cosmic Neutrino Back-
ground (CNB), and neutrinos coming from the decays of neutrons and tritons created
during the Big Bang Nucleosynthesis (BBN). Despite having the highest flux at Earth,
these neutrinos have never been directly measured because their low energy makes their
detection extremely difficult.
The Sun and the Earth are huge natural sources of neutrinos. The Sun converts hydrogen
into helium mainly via the p-p chain, producing keV to MeV electron neutrinos from
either proton β+ decays or electron captures. Also, eV to keV neutrino pairs of all
flavours are produced by thermal processes, e.g. plasmon decay, Compton scattering and
bremsstrahlung. Geoneutrinos, with energies between keV to MeV, are decay products
of natural long-lived radioactive elements, typically 238U, 232Th and 40K.
Supernovae release neutrinos and antineutrinos of all flavors within a few seconds. The
1.3. LOW-ENERGY NEUTRINO INTERACTIONS 5
Diffuse Supernova Neutrino Background (DSNB), yet to be measured, is a prediction
of the MeV neutrinos that have been produced by all supernovae in the universe. At-
mospheric neutrinos, with energies between MeV to PeV, are produced when pions and
muons, from cosmic ray cascades, decay. Finally, neutrinos with energies above TeV are
naturally produced by astrophysical objects such as active galactic nuclei, star-forming
galaxies and gamma-ray bursts.
Artificial sources of neutrinos include particle accelerators and nuclear reactors. Neutri-
nos from accelerators are usually the decay products of pions produced after a collision
of a proton beam with a target. Their energies range from MeV to GeV. Nuclear re-
actors can produce big fluxes of keV to MeV electron antineutrinos. These come from
the β decays of fission fragments originated during the fission of some isotopes, typically
235U.
1.3. Low-energy neutrino interactions
When talking about low-energy neutrinos we are referring to neutrinos with energies
of tens-of-MeV and below. Sources of these neutrinos include supernovae, the Sun, the
Earth, accelerators and reactors. Neutrinos undergo weak interactions which can be
classified as charged current or neutral current depending on the nature of its mediator,
W or Z, respectively. At low energies, there are multiple processes that can occur
enabling neutrino detection. However, the dominant ones are coherent elastic scatterings
in which the nature of the particles in the initial and final states is the same. Figure 1.2
shows the Feynman diagrams associated to these processes in the context of the Standard
Model of particle physics, considering typical targets.
Z
e−, p,N
νl, ν̄l
e−, p,N
νl, ν̄l
1
W
e
−
νe
νe
e
−
1
W
e
−
ν̄e
ν̄e
e
−
1
Figure 1.2. Feynman diagrams of the NC (left) and CC (center and right) coherent elastic
scattering interactions in which the nature of the particles in the initial and final states is the
same, considering usual targets. Here, l = e, µ, τ .
In the laboratory frame and assuming that the targets are at rest, the SM differential
6 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
cross sections of low-energy neutrino weak interactions can be written as [31,32]
dσSM
dER
=
2G2
Fm
π
[
(GV +GA)
2 + (GV −GA)
2
(
1− ER
Eν,ν̄
)2
−
(
G2
V −G2
A
) mER
E2
ν,ν̄
]
,
(1.1)
where GV (A) is the vector (axial-vector) coupling, ER is the target recoil energy, Eν,ν̄ is
the incident (anti)neutrino energy, GF is the Fermi coupling constant and m is the target
rest mass. Details on the cross sections for specific processes are given next. Figure 1.3
shows the SM cross sections for different low-energy neutrino interactions.
Figure 1.3. Taken from [33]: SM differential cross sections of different low-energy neutrino
interactions: νe − e− elastic scattering per e− (dashed red), ν̄e − p quasielastic scattering (solid
red), CC reaction 127I (νe, e−)
127 Xe (green) and CEνNS for Cs, I, Ge, Ar and Na (blue).
1.3.1. Elastic neutrino-electron scattering (EνES)
In this process a(n) (anti)neutrino elastically scatters off an electron. The weak inter-
actions associated to it can be [34]:
• Pure neutral current (NC)
It occurs for muon and tau (anti)neutrinos. Its cross section is described by equa-
tion 1.1 with GV ≡ gν,ν̄V geV , GA ≡ gν,ν̄A geA and m ≡ me the electron mass. Neglecting
radiative corrections, the vector (axial-vector) NC couplings are given by
gνV = 1/2 gν̄V = 1/2 geV = −1/2 + 2 sin2 θW (1.2)
gνA = 1/2 gν̄A = −1/2 geA = −1/2 (1.3)
1.3. LOW-ENERGY NEUTRINO INTERACTIONS 7
where θW is the weak mixing angle. At low energies, sin2 θW = 0.23857 [35].
Replacing the given couplings, the cross section of pure NC (anti)neutrino-electron
scattering is given by
dσ
νµ,τe
SM
dER
=
2G2
Fme
π
[
(
1
2
− sin2 θW
)2
+ sin4 θW
(
1− ER
Eν
)2
+ sin2 θW
(
1
2
− sin2 θW
)
meER
E2
ν
]
(1.4)
dσ
ν̄µ,τe
SM
dER
=
2G2
Fme
π
[
sin4 θW +
(
1
2
− sin2 θW
)2(
1− ER
Eν̄
)2
+ sin2 θW
(
1
2
− sin2 θW
)
meER
E2
ν̄
]
(1.5)
• Combined neutral and charged current (NC+CC)
The electron (anti)neutrino elastic scattering off an electron has the contribution of
both, NC and CC interactions. In this case, the SM cross section is given by equa-
tion 1.1 with GV ≡ gν,ν̄V (geV + 1), GA ≡ gν,ν̄A (geA + 1) and m ≡ me the electron mass.
The extra factor in GV (A) derives from the interference term.
Replacing the given couplings, the cross section of the electron (anti)neutrino-electron
scattering is given by
dσνee
SM
dER
=
2G2
Fme
π
[
(
1
2
+ sin2 θW
)2
+ sin4 θW
(
1− ER
Eν
)2
− sin2 θW
(
1
2
+ sin2 θW
)
meER
E2
ν
]
(1.6)
dσν̄ee
SM
dER
=
2G2
Fme
π
[
sin4 θW +
(
1
2
+ sin2 θW
)2(
1− ER
Eν̄
)2
− sin2 θW
(
1
2
+ sin2 θW
)
meER
E2
ν̄
]
(1.7)
Note that the cross sections in Eqs. (1.4)–(1.7) are derived under the free electron ap-
proximation (FEA) hypothesis. When considering EνES in a medium where electrons
are bounded to atoms, atomic many-body effects should be accounted for.
Although the EνES cross section is usually small compared to the cross sections of other
low-energy neutrino interactions (see Fig. 1.3), this process is highly directional, i.e. the
electron is scattered at very small angles with respect to the direction of the incident
(anti)neutrino. This feature has been widely used by solar and supernova neutrino
experiments to identify the sources and to differentiate signals from backgrounds.
1.3.2. Coherent elastic neutrino-nucleus scattering (CEνNS)
The coherent elastic neutrino-nucleus scattering, predicted by Daniel Z. Freedman in
1974 [36], is a weak flavor-independent neutral current interaction. In this process the
8 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
incident (anti)neutrino elastically scatters off a nucleus N , causing all the constituent
nucleons of the nucleus to recoil as a whole. For it to occur, the three-momentum transfer
q = |q| should be small enough so that q2R2 < 1, with R being the nuclear radius. In
the laboratory frame and for a Si nucleus, this corresponds to an energy of the incident
(anti)neutrino below ∼60 MeV.
Due to coherence, the total scattering amplitude is the sum of the amplitudes associated
to the scattering of the (anti)neutrino with each nucleon. This results in an enhancement
of the effective cross section which can be considered proportional to the square of the
number of neutrons in the nucleus. However, despite CEνNS cross section is large
compared to the cross sections of other low-energy neutrino interactions (see Fig. 1.3),
this process took a long time to be observed due to the difficulty of measuring the
low-energy nuclear recoils produced. In fact, the maximum nuclear recoil energy is
Emax
R = 2E2
ν,ν̄/(M + 2Eν,ν̄) where M is the nuclear mass. For a Si nucleus and Eν,ν̄ =
2 MeV, Emax
R ∼ 300 eV.
In 2017, the COHERENT collaboration reported the first experimental observation [37]
of the CEνNS using a CsI[Na] scintillator crystal with a ∼5 keV energy threshold ex-
posed to neutrinos (νµ, ν̄µ and νe), products of the Spallation Neutron Source (SNS)
at the Oak Ridge National Laboratory. In 2021, the same collaboration reported the
CEνNS observation in a LAr detector with a ∼20 keV energy threshold [38]. So far,
only COHERENT has experimentally seen this process but a huge effort to cross-check
these results with multiple current and future experiments is ongoing.
In the laboratory frame, the SM differential cross section of the coherent elastic scattering
of (anti)neutrinos off a nucleus at rest, with Z protons, N neutrons and mass M is given
by equation 1.1 with m =M and
GV ≡ gν,ν̄V gNV = gν,ν̄V
(
gnVNF
n
V (q
2) + gpV ZF
p
V (q
2)
)
(1.8)
GA ≡ gν,ν̄A gNA = gν,ν̄A
[
gnA(N+ −N−)F
n
A(q
2) + gpA(Z+ − Z−)F
p
A(q
2)
]
(1.9)
where N± and Z± refer to the number of nucleons with spin up (down) +(−) and
Fn,p
V (A)(q
2) is the nuclear vector (axial-vector) form factor for nucleons. Neglecting ra-
diative corrections, the vector (axial-vector) NC nucleon couplings are given by
gnV = (guV + 2gdV ) = −1/2 gpV = (2guV + gdV ) = 1/2− 2 sin2 θW , (1.10)
gnA = (guA + 2gdA) = −1/2 gpA = (2guA + gdA) = 1/2 , (1.11)
as
guV = 1/2− 4 sin2 θW /3 gdV = −1/2 + 2 sin2 θW /3 guA = 1/2 gdA = −1/2 .
(1.12)
1.3. LOW-ENERGY NEUTRINO INTERACTIONS 9
The vector contribution is dominant for most nuclei and the only contribution for spin-
zero nuclei. Neglecting GA, the CEνNS SM differential cross section can be written as
dσνNSM
dER
(Eν,ν̄) =
G2
FM
π
Q2
W
(
1− MER
2E2
ν,ν̄
− ER
Eν,ν̄
+
E2
R
2E2
ν,ν̄
)
, (1.13)
where
QW ≡ gn
VNF
n
V (q
2) + gp
V ZF
p
V (q
2) =
[(
1
2
− 2 sin2 θW
)
Z − 1
2
N
]
F (q2) (1.14)
is the weak nuclear charge. For the last step in Eq. (1.14) we assume that the nuclear
vector form factor for protons and neutrons is the same, i.e. F (q2) ≡ Fn,p
V (q2). This
factor can be expressed as in Ref. [39]
F (q2) =
4πρ0
Aq3
(sin qR− qR cos qR)
1
1 + a2q2
, (1.15)
where A is the atomic mass of the nucleus, a = 0.7× 10−13 cm is the range of the Yukawa
potential considered, R = r0A
1/3 is the nuclear radius and ρ0 = 3/4πr30 is the nuclear
density, with r0 = 1.3× 10−13 cm being the average radius of a proton in a nucleus.
For low neutrino energies F (q2) ≃ 1; hence, the cross section uncertainty associated to
the nuclear structure becomes negligible.
1.3.3. Quasielastic neutrino-nucleon scattering
This process is a weak charged-current interaction, mediated by a W boson, in which
a(n) (anti)neutrino scatters off an entire (proton) neutron; see Fig. 1.4. In this process
the nature of the particles in the initial and final states is not the same but, it is worth
discussing it as this detection channel dominates the expected neutrino signal in water
Cherenkov and liquid scintillator neutrino detectors.
W
n
νl
p
l−
1
W
p
ν̄l
n
l+
1
Figure 1.4. Feynman diagrams of the QE (anti)neutrino-nucleon scattering.
For it to occur, in the laboratory frame, Eν,ν̄ > ((ml +mo
n)
2 −mi
n
2)/2mi
n whereml is the
rest mass of the outgoing lepton and mi
n (mo
n) is the rest mass of the incoming (outgoing)
nucleon. This process can be considered a low-energy neutrino interaction only for
10 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
electron (anti)neutrinos. In fact, the quasielastic scattering of the electron antineutrino
on proton, better known as inverse beta decay (IBD), occurs when Eν̄e
>1.806 MeV, in
the laboratory frame.
At tree level and for low neutrino energies, the SM differential cross section of the IBD
can be written as [40]
dσν̄ep→en
SM
dER
(Eν̄e) =
G2
F cos2 θC
16πmpE2
ν̄e
[
(
g2V + g2A
)
[
(
2mp(Eν̄e − ER)−m2
e
)2 −
(
m2
p −m2
n
)2
]
−
(
g2V − g2A
)
(mp +mn)
2 (m2
p −m2
n + 2mp(Eν̄e − ER) +m2
e
)
]
, (1.16)
where ER is the positron recoil energy, Eν̄e is the incident antineutrino energy, θC is the
Cabibbo angle (cos θC = 0.97373 [20]), gV (A) are the vector (axial-vector) couplings and
mp, mn and me are the masses at rest of the proton, neutron and positron, respectively.
When considering the quasielastic electron antineutrino-nucleus scattering, i.e. ν̄e +
A
ZX → A
Z+1X + e+, nuclear effects must be taken into account.
1.4. Beyond the Standard Model (BSM) physics
The study of neutrinos is directly linked to new physics. In the SM of electroweak
interactions, developed in the 60s by Glashow [41], Weinberg [42] and Salam [43]3,
neutrino masses are zero by construction, as their fields are considered to be purely left-
handed, motivated by the absence of a experimental hint of a non-zero neutrino mass
at that time. The discovery of neutrino oscillations [2, 3]4 implies that neutrinos have
mass and mix, evidencing the incompleteness of the SM. Neutrino oscillations have been
vastly studied since then and the robust experimental program in this area is providing
tighter experimental constraints on the neutrino mass-squared differences and mixing
parameters. However, the origin of the very small neutrino mass is still unknown and
its study could provide insight into new mass generation processes beyond the standard
Higgs mechanism.
Moreover, extensions of the SM accounting for massive neutrinos can give rise to neu-
trinos having electromagnetic properties. The study of these properties could help us
understand fundamental unknowns such as the nature of the neutrinos, which is related
to the matter-antimatter asymmetry observed in the universe. Also, studying neutrino
interactions is a way to explore the weakly interacting dark sector, a proposed collec-
31979 Nobel Prize in Physics: “for their contributions to the theory of the unified weak and elec-
tromagnetic interaction between elementary particles, including, inter alia, the prediction of the weak
neutral current” [23]
42015 Nobel Prize in Physics: “for the discovery of neutrino oscillations, which shows that neutrinos
have mass” [23]
1.4. BEYOND THE STANDARD MODEL (BSM) PHYSICS 11
tion of yet-unobserved quantum fields and mediators linked to new interactions and
strengths. Here, we discuss how non-standard interactions (NSI), new mediators and
neutrino electromagnetic properties can be probed using low-energy neutrino detection
channels, e.g. EνES, CEνNS.
1.4.1. Neutrino non-standard interactions (NSI)
Neutrino NSI affect their production, propagation and detection. For example, when
propagating, neutrino NSI with matter can modify their oscillation probabilities. A
model-independent approach to search for NSI is to extend the SM interactions with
further four-fermion interactions, where new mediators are generally assumed to be much
heavier than the characteristic momentum transfer [44]. The new NSI couplings ϵqXll′ ,
quantified with respect to GF , can be either flavor-preserving (l = l′) or flavor-changing
(l ̸= l′), with l, l′ = {e, µ, τ} and X = {L,R}, associated to the chirality projection
operators. The vector and axial-vector couplings are defined as ϵqVll′ ≡ ϵqLll′ + ϵqRll′ and
ϵqAll′ ≡ ϵqLll′ − ϵqRll′ . This framework has been useful to connect constraints on new physics
from scattering and oscillation experiments.
• EνES channel
When NSI are introduced, the EνES differential cross section can be obtained with
Eq. (1.1) by replacing [45,46]
GV (A) → GNSI
V (A) = gν,ν̄V (A)

geV (A) + δle + ϵ
eV (A)
ll +
∑
l ̸=l′
ϵ
eV (A)
ll′

 . (1.17)
• CEνNS channel
When NSI are introduced, the CEνNS differential cross section can be obtained with
Eq. (1.1) by replacing [47]
GV → GNSI
V = gν,ν̄V



gpV + 2ϵuVll + ϵdVll +
∑
l ̸=l′
(
2ϵuVll′ + ϵdVll′
)

ZF p
V (q
2)
+

gnV + ϵuVll + 2ϵdVll +
∑
l ̸=l′
(
ϵuVll′ + 2ϵdVll′
)

NFn
V (q
2)

 , (1.18)
12 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
GA → GNSI
A = gν,ν̄A



gpA + 2ϵuAll + ϵdAll +
∑
l ̸=l′
(
2ϵuAll′ + ϵdAll′
)

 (Z+ − Z−)F
p
A(q
2)
+

gnA + ϵuAll + 2ϵdAll +
∑
l ̸=l′
(
ϵuAll′ + 2ϵdAll′
)

 (N+ −N−)F
n
A(q
2)

 . (1.19)
1.4.2. New light mediators
Light mediators are often explored in the framework of simplified models. These models,
unlike full BSM models, only account for representative few new hypothetical BSM
particles and their interactions, allowing to characterize new physics with a small number
of parameters [48]. Some of these models can be understood as a limit of a more general
BSM theory. Theories including light mediators have been vastly proposed [49–54],
motivated by the experimental status in various fields including dark matter, cosmology
and neutrinos. Here, we consider the presence of new mediators in the framework of a
simplified model that assumes couplings to the first-generation SM fermions.
Scalar mediator
Let us consider pure scalar-type interactions associated to a mediator ϕ, with mass
Mφ. Since the scalar-neutrino interaction flips chirality, it does not interfere with the
chirality-conserving SM weak interactions [55]. Therefore, the differential cross section
of the EνES (CEνNS) with the presence of a new light scalar mediator ϕ can be written
as the sum of the SM contribution and the one associated to the new mediator:
dσSM+φ
dER
(Eν,ν̄) =
dσSM
dER
(Eν,ν̄) +
dσφ
dER
(Eν,ν̄)
=
dσSM
dER
(Eν,ν̄) +
G2
Fm
4π
G2
φ
(
2mER
E2
ν,ν̄
)
,
(1.20)
where dσSM/dER is given by the expressions in Section 1.3.1 (1.3.2).
• EνES channel
In this case, m ≡ me and
Gνe
φ =
gν,ν̄φ geφ√
2GF
(
2meER +M2
φ
) . (1.21)
• CEνNS channel
1.4. BEYOND THE STANDARD MODEL (BSM) PHYSICS 13
Here, m ≡M and
GνN
φ =
gν,ν̄φ QN
φ√
2GF
(
2MER +M2
φ
) . (1.22)
The nuclear scalar charge QN
φ is written as [56]
QN
φ =
(
∑
q
gqφ ⟨n|q̄q|n⟩
)
NFn
S (q
2) +
(
∑
q
gqφ ⟨p|q̄q|p⟩
)
ZF p
S(q
2)
=
(
mn
∑
q
gqφ
fnq
mq
)
NFn
S (q
2) +
(
mp
∑
q
gqφ
fpq
mq
)
ZF p
S(q
2)
≃ gqφ (14N + 15.1Z)F (q2) ,
(1.23)
where fn,pq express the quark contributions to the nucleon masses. For the last step
in Eq. (1.23), we assume a universal coupling of the scalar to the quarks, the same
nuclear scalar form factor for protons and neutrons, i.e. F (q2) ≡ Fn,p
S (q2), and the
approximation in Ref. [57].
Vector mediator
Let us consider pure flavor-conserving vector interactions associated to a mediator Z ′,
with mass MZ′ , with no kinetic or mass mixing with the SM weak bosons. Since both
the SM weak interactions and the Z ′ interactions are of vector type, they contribute
coherently to the cross section.
• EνES channel
The EνES differential cross section with the presence of a new light vector mediator
can be written as
dσνeSM+Z′
dER
(Eν,ν̄) =
dσνeSM
dER
+
G2
Fme
π
[
(2GV + Gνe
Z′ )
2 − 4G2
V
]
, (1.24)
where dσνeSM/dER and its corresponding GV are given by the expressions in Sec-
tion 1.3.1, and
Gνe
Z′ =
gν,ν̄Z′ geZ′√
2GF
(
2meER +M2
Z′
) . (1.25)
• CEνNS channel
At tree level, the net effect on the CEνNS cross section due to the new mediator is a
modification of the global factor Q2
W in Eq. (1.13) [58]. Thus, the CEνNS differential
14 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
cross section becomes
dσνNSM+Z′
dER
(Eν,ν̄) =
(
1 +
GνN
Z′
QW
)2
dσνNSM
dER
(Eν̄e) , (1.26)
where dσνNSM/dER is given in Eq. (1.13) and
GνN
Z′ =
gν,ν̄Z′ QN
Z′√
2GF
(
2MER +M2
Z′
) . (1.27)
The nuclear vector charge QN
Z′ is written as
QN
Z′ = gnZ′NFn
V (q
2) + gpZ′ZF
p
V (q
2)
=
(
guZ′ + 2gdZ′
)
NFn
V (q
2) +
(
2guZ′ + gdZ′
)
ZF p
V (q
2)
≃ gqZ′3(N + Z)F (q2) ,
(1.28)
where, for the last step, we assume a universal coupling to the quarks, as in Ref. [57],
and the same nuclear vector form factor for protons and neutrons, i.e. F (q2) ≡
Fn,p
V (q2). Note that gp,nZ′ are given by the sums of the couplings of their constituent
quarks since the vector current is conserved.
The quadratic dependency on the prefactor (1 + GνN
Z′ /QW ) in Eq. (1.26) generates
a degeneracy in the vector parameter space, since values of ±1 result in a SM-like
interaction. The case GνN
Z′ = 0 is the pure SM interaction as g2Z′ ≡ gν,ν̄Z′ g
q
Z′ = 0,
whereas the case GνN
Z′ = −2QW corresponds to a non-exclusion region where the light
vector mediator contribution can not be distinguished from the SM CEνNS interaction.
In models with an additional massive Z ′ vector boson associated with a new U(1)′
gauge group, the coupling constants are proportional to the charges of leptons and
quarks under the new gauge symmetry [59], i.e. gνl,ν̄lZ′ = gZ′Ql
Z′ , geZ′ = gZ′Qe
Z′ and
gqZ′ = gZ′Qq
Z′ , where gZ′ is the coupling constant of the symmetry group. A review
of such models can be found in Ref. [60]. Within the most constrained ones are the
universal model in which Ql
Z′ = Qq
Z′ = 1 and the B − L model in which Ql
Z′ = −1 and
Qq
Z′ = 1/3. The universal model is not anomaly-free, but it is often considered with the
understanding that the contributions of other non-standard particles in the full theory
cancel the anomalies.
It is worth to mention that, when the mediator is significantly heavier than the momen-
tum transfer in the scattering process, it can be integrated out, allowing for a mapping
between the mass and couplings of the Z ′ and the ϵ-parameters of vector NSI [61],
1.4. BEYOND THE STANDARD MODEL (BSM) PHYSICS 15
discussed in Section 1.4.1, as
ϵ
q(e)V
ll′ ∝ (gν,ν̄Z′ )ll′g
q(e)
Z′
GFM2
Z′
. (1.29)
1.4.3. Neutrino electromagnetic properties
In the general parametrization of the neutrino electromagnetic interactions, the form
factors related to the electric charge, magnetic and electric moments are represented
as matrices in the mass eigenstate space [62]. The study of these properties can be
used to unveil the neutrinos fermionic nature: Dirac neutrinos can have both diagonal
(“intrinsic”) and off-diagonal (“transition”) elements whereas, for Majorana neutrinos,
the diagonal ones vanish [63].
Moreover, if neutrinos had electromagnetic properties, there could be observable effects
from their interaction with electromagnetic fields or charged particles. This is particu-
larly interesting in astrophysical contexts, where neutrinos travel over great distances in
such environments.
Magnetic moment
In the simplest SM extension that accounts for right-handed neutrinos and lepton num-
ber conservation, the Dirac massive neutrino acquires a very small intrinsic magnetic
moment through radiative corrections given by [63]
µν =
3eGF
8
√
2π2
mν ≃ 3.2× 10−19
(mν
eV
)
µB , (1.30)
with µB =
√
4πα/2me the Bohr magneton, α the fine structure constant, e the electric
charge and mν the neutrino mass. Taken into account the current upper limit on the
neutrino mass, the value of µν in Eq. (1.30) is still orders of magnitude below the latest
experimental bounds (∼ 10−12µB) [20]. However, larger values of µν are consistent with
other theoretical beyond the SM ideas; see Ref. [63] for a discussion.
The presence of a non-zero magnetic moment contributes to the detection channel cross
section. Since the neutrino magnetic moment interaction changes the neutrino helic-
ity, it does not interfere with the helicity-conserving SM weak interaction; hence, both
16 NEW PHYSICS IN THE LOW-ENERGY NEUTRINO SECTOR
contributions add incoherently in the cross section, which can be written as [64]
dσSM+µν
dER
(Eν,ν̄) =
dσSM
dER
(Eν,ν̄) +
dσµν
dER
(Eν,ν̄)
=
dσSM
dER
(Eν,ν̄) + καµ2ν
(
1
ER
− 1
Eν
)
,
(1.31)
where dσSM/dER is given by the expressions in Section 1.3.1 (1.3.2) for EνES (CEνNS).
The κ factor is 1 for EνES and Z2F 2(q2) for CEνNS, with F (q2) the same form factor
as in the SM cross section. As these cross sections scale inversely with ER, low-energy
threshold detectors can provide strong bounds on µν .
Millicharge
In the SM, neutrinos are neutral particles, consequence of the quantization of electric
charge [63]. However, some SM extensions, such as those that account for right-handed
neutrinos and lepton number conservation, can lead to the dequantization of electric
charge, thereby enabling neutrinos to be electrically charged. In fact, if we consider one
of the strongest upper bounds on the neutrino electric charge qν,ν̄ ≤ 3×10−21 e, coming
from the neutrality of matter [63, 65], neutrinos may be millicharged particles. Here, e
is the electron charge.
The EνES (CEνNS) differential cross section accounting for the contribution from a
millicharged neutrino can be obtained by modifying gν,ν̄V (gpV ) in the equations in Sec-
tion 1.3.1 (1.3.2) through
gν,ν̄V → gν,ν̄V +
2
√
2πα
GF (−2meER)
2 q
ν,ν̄ gpV → gpV − 2
√
2πα
GF (−2MER)
2 q
ν,ν̄ (1.32)
where q2 ≡ −2mER is the momentum transfer. Here, we have just considered interac-
tions with flavor-diagonal electric charge. Note that, as the neutrino electric charge can
interfere with the SM coupling, its sign is important.
Chapter 2
Reactor neutrino experiments
Nuclear reactors are powerful pure sources of MeV electron antineutrinos whose emis-
sions have been studied with many different reactor neutrino experiments distributed
worldwide; see Fig. 2.1. These experiments use mainly two types of nuclear reactors as
ν̄e source: 1) commercial reactors, with gigawatt thermal capacity (∼GWth) that use
low-enriched 235U fuel (LEU), and 2) research reactors, with lower capacity (∼MWth)
that use highly enriched 235U fuel (HEU). Due to the difference in fission isotopes and
their yields, the number and energy spectrum of the ν̄e produced per fission differ slightly
between commercial and research reactors.
Reactor experiments use various ν̄e detection channels including inverse beta decay
(IBD), NC inelastic nuclear scattering, EνES and CEνNS. IBD has been the most
observed interaction so far, due to its advantage of excellent background reduction.
Experiments looking at other detection channels usually require low-energy detection
thresholds and low radioactive contamination.
Figure 2.1. Taken from Ref. [66]. Map showing the planned (yellow), current (green) and
completed (red) reactor antineutrino experiments taking data after 2010. The arrows’ and
markers’ styles indicate their main detection channel and their type of nuclear reactor as ν̄e
source, respectively.
17
18 REACTOR NEUTRINO EXPERIMENTS
Reactor antineutrino experiments have played a crucial role in advancing our under-
standing of these particles. In fact, the first experimental confirmation of the neutrino
existence in 1956 came from the results of an IBD-based reactor experiment [22]. In
the first decade of the 2000s, the KamLAND experiment demonstrated for the first time
that reactor neutrinos oscillate [67–69]. In 2012, more recent (now completed) reactor
experiments, including DayaBay [70], RENO [71] and Double Chooz [72], did a pre-
cise measurement of the neutrino mixing angle θ13 confirming its non-zero value, which
turned out to be smaller compared to the values of θ12 and θ23, according to current up-
per limits [20]. This result allows for the possibility of observing CP violation in neutrino
oscillations, i.e. a difference in the oscillation probabilities between neutrinos and their
antineutrinos, which could help explain the observed matter-antimatter asymmetry in
the Universe.
Results from reactor experiments have validated the SM theory of weak interactions [73]
and have set stringent constraints on important parameters such as the electroweak
mixing angle [74, 75]. Moreover, reactor experiments have served as powerful tools
for exploring physics beyond the SM. They have been used to probe the existence of
sterile neutrinos (DayaBay [76], NEOS [77], DANNS [78], STEREO [79], RENO [71],
PROSPECT [80], Neutrino-4 [81]), hypothetical particles that do not interact via the
weak force but could mix with the active neutrinos, and to put competitive limits on po-
tential hidden sector couplings to neutrinos (CONNIE [8], NCC-1701 [82], CONUS [83]).
2.1. Reactor ν̄e flux
Nuclear reactors generate power through fission. The β decays of the fission products,
following the fission of four principal fissile isotopes (235U, 238U, 239Pu and 241Pu),
produce most of the ν̄e in the reactor flux. The β decays of 239U nuclei (239U →239Np
→239Pu), generated from 238U neutron capture, also produce ν̄e. The contribution of
each process to the total reactor ν̄e emission spectrum depends on the fuel composition
of the reactor core (LEU/HEU) and on time, since the composition changes with time.
The ν̄e flux as a function of Eν̄e at a detector located at a distance d from the reactor
core is given by
dΦ
dEν̄e
=
nf
4πd2
(
dNν̄e
dEν̄e
)
=
Pth (6.24× 1021MeV/s)
4πd2 Efis
(
dNν̄e
dEν̄e
)
, (2.1)
where nf is the number of fissions per second, Pth is the reactor’s thermal power in GWth,
Efis ≃ 205.24 MeV/fission is the average energy released per fission and dNν̄e/dEν̄e is
the total reactor antineutrino energy spectrum per fission.
Uncertainties on the reactor ν̄e spectrum depend on the approach for its calculation [66].
In the summation or ab initio approach, the spectrum is computed by summing the
2.1. REACTOR ν̄E FLUX 19
contributions from all fission products. This method requires extensive information on
the thousands of beta branches involved and the weighting factors of the fission products,
known as the fission yields. The conversion approach, generally considered to be more
precise, converts the measured electron spectra from β decays of a limited number of
individual beta branches into antineutrino spectra. Uncertainties of this method arise
from the experimental measurements themselves.
As experiments become more sensitive to these uncertainties, a need to improve the
neutrino flux and spectra calculations arises. In fact, depending on the method used
to compute the reactor antineutrino flux, the so-called reactor neutrino anomaly, i.e.
a “missing” measured flux to match the theoretically expected flux, could increase or
disappear [84].
2.1.1. Commercial reactor flux from a summation approach
Here, we describe the summation method used in Refs. [6–9] to compute the ν̄e energy
spectrum per fission from a commercial reactor with LEU fuel.
The ν̄e spectrum coming from the β decays of the fission products of each fissile isotope
was taken from Ref. [64]. For energies below 2 MeV, these spectra are given as tabulated
values in Table 2.1,
Eν̄e
(MeV) 235U 239Pu 238U 241Pu
7.813 ×10−3 0.024 0.14 0.089 0.20
1.563 ×10−2 0.092 0.56 0.35 0.79
3.12 ×10−2 0.35 2.13 1.32 3.00
6.25 ×10−2 0.61 0.64 0.65 0.59
0.125 1.98 1.99 2.02 1.85
0.25 2.16 2.08 2.18 2.14
0.50 2.66 2.63 2.91 2.82
0.75 2.66 2.58 2.96 2.90
1.0 2.41 2.32 2.75 2.63
1.5 1.69 1.48 1.97 1.75
2.0 1.26 1.08 1.50 1.32
Table 2.1. Tabulated values of the antineutrino spectrum of each fissile isotope in units of
ν̄e/MeV/fission taken from Ref. [64].
while for energies above 2 MeV these spectra are described by the parametric expression
dNν̄e
dEν̄e
= aea0+a1Eν̄e+a2E2
ν̄e , (2.2)
where the fitted parameters are listed in Table 2.2.
The antineutrino spectrum coming from the β decays of 239U, generated from the neutron
20 REACTOR NEUTRINO EXPERIMENTS
Parameter 235U 239Pu 238U 241Pu
a 1.0461 1.0527 1.0719 1.0818
a0 0.870 0.896 0.976 0.793
a1 −0.160 −0.239 −0.162 −0.080
a2 −0.0910 −0.0981 −0.0790 −0.1085
Table 2.2. Fitted parameters of the antineutrino spectrum of each fissile isotope taken from
Ref. [64]. The a constant was fitted so that the values in Table 2.1 matched Eq. (2.2) at 2 MeV.
capture of 238U nuclei, was extracted from Ref. [85], as well as the fission yields of each
of the processes considered to compute the ν̄e energy spectrum per fission, which are
shown in Table 2.3.
Process Relative rate per fission Nν̄e
per process Nν̄e
per fission
235U fission 0.55 6.14 3.4
239Pu fission 0.32 5.58 1.8
238U fission 0.07 7.08 0.5
241Pu fission 0.06 6.42 0.4
238U(n, γ)239U 0.60 2.00 1.2
Table 2.3. Time-averaged relative rates and ν̄e yields per fission for each process considered in
the dNν̄e
/dEν̄e
computation. Values taken from Ref. [85].
To obtain the total reactor ν̄e energy spectrum per fission, dNν̄e/dEν̄e shown in Fig. 2.2,
the individual spectra were summed after being normalized and multiplied by their
corresponding ν̄e yield per fission.
All
235U
239Pu
238U
241Pu
238U(n,γ)239U
0 1 2 3 4
0.1
0.5
1
5
10
Eυe
(MeV)
d
N
υ
e
/d
E
υ
e
(N
υ
e
/f
is
s
io
n
)
Figure 2.2. Total ν̄e energy spectrum per fission (solid) of a commercial reactor showing each
process contribution (dashed).
2.2. ONGOING/COMPLETED LOW-THRESHOLD EXPERIMENTS 21
2.2. Ongoing/completed low-threshold experiments
Recent progress in low-threshold technologies, initially developed for direct DM searches,
has enabled the experimental exploration of low-energy neutrino interactions. While
some of these interactions have already been observed, they have not been proved across
the whole experimentally accessible energy range. A specific example is CEνNS, ob-
served using 10-50 MeV neutrinos [37, 38], energies at which incident neutrinos become
sensitive to nuclear structure, provoking the loss of coherence. Detecting CEνNS at
lower energies, such as those of reactor neutrinos (≤2 MeV) where the process is fully
coherent, requires O(eV) threshold detectors. To minimize backgrounds coming from the
continuous reactor operation, these detectors usually need a thick shielding and over-
burden. Here, the state-of-the-art ongoing low-threshold reactor neutrino experiments,
besides CONNIE, is presented as well as their more recent results.
2.2.1. CONUS
The COherent Neutrino nUcleus Scattering (CONUS) detector operated in the Brokdorf
nuclear power plant, in Germany, from 2018 to 2022, when the plant stopped working.
It was located inside the reactor dome, beneath the fuel cooling pool (∼24 m.w.e. over-
burden), ∼17 m away from the 3.9 GWth reactor core. CONUS sensors are four p-type
point contact high-purity germanium (HPGe) sensors, leading to a total fiducial mass
of ∼3.7 kg. CONUS suppresses backgrounds via passive and active shields, see Fig. 2.3
(left), reaching a background rate of ∼10 DRU1. The collaboration did a detailed back-
ground study [86], showing that their main contribution (∼40%) at low energies comes
from electromagnetic particles and neutrons that are generated from cosmic muon in-
teractions in high-density materials, such as lead, even with the active muon veto; see
Fig. 2.3 (right).
Figure 2.3. Taken from Refs. [86,87]. Left) Overall design of CONUS detector highlighting the
active and passive shields. Right) CONUS background model showing the main contributions.
11 Differential Rate Unit (DRU) corresponds to 1 event/kg/day/keV.
22 REACTOR NEUTRINO EXPERIMENTS
In their first published results [87], CONUS collected an exposure of 248.7 (58.8) kg-day
with RON (ROFF). Looking at energies between ∼0.3 and 1 keV, they found no hint
for a CEνNS signal and disfavored any Lindhard-like nuclear recoil quenching factor
model for germanium with a parameter k > 0.27. Assuming k = 0.16, value coming
from measurements using CONUS sensors as target [88], the 90% C.L. limit established
by their results is ∼17 times above the SM prediction of the CEνNS rate. From these
results, CONUS put limits on NSIs, light mediators [83] and neutrino electromagnetic
properties: µν < 7.5× 10−11 µB and |qν | < 3.3× 10−12 e [89]. Fig. 2.4 shows examples
of CONUS sensitivity to new physics.
Figure 2.4. Taken from Ref. [83]. CONUS limits on vector-type NSIs (left) and light vector
mediator parameters (right). For details please refer to Ref. [83].
Although CONUS has ended its operations, the successor CONUS+, with improvements
in the HPGe sensors, electronics and shields, is going to be installed at the Leibstadt
nuclear power plant, in Switzerland, during summer 2023.
2.2.2. NCC-1701 (Dresden-II)
NCC-1701 hosts a ∼2.9 kg inverted coaxial p-type point contact germanium detector
with an energy threshold of ∼200 eV, installed 8 m away from the core of the 2.96 GWth
Dresden-II power reactor inside the dome [90]. To suppress backgrounds, this detector
has passive and active shields. In the most recent publication with results from this
system, corresponding to 96.4 days of exposure, authors claim a “very strong preference
for an interpretation that includes the SM CEνNS signal, present during periods of
reactor operation only” [90].
From these results, new physics has been explored [91–94], resulting in constraints on
NSIs, light mediators and neutrino electromagnetic properties: µν < 2.13 × 10−10 µB
and |qν | < 8.6 × 10−12 e [59]. Fig. 2.5 shows examples of NCC-1701 sensitivity to new
2.2. ONGOING/COMPLETED LOW-THRESHOLD EXPERIMENTS 23
physics.
Figure 2.5. Taken from Refs. [93, 94]. NCC-1701 limits on vector-type NSIs (left) and light
vector mediator parameters (right). For details please refer to Refs. [93,94].
2.2.3. νGEN
The νGEN detector is located inside the dome (∼50 m.w.e overburden) of the 3.1 GWth
reactor unit #3 at the Kalinin nuclear power plant in Udomlya, Russia. It hosts a
1.4 kg HPGe detector surrounded by active and passive shields, as shown in Fig. 2.6
(left). During the first runs [95], the detector was ∼12.2 m away from the reactor’s core,
reaching a background level of 30 DRU at around 1 keV that increased at low energies,
see Fig. 2.6 (right).
Figure 2.6. Taken from Ref. [95]. Left) Diagram of the νGEN detector. Right) RON and ROFF
spectra using νGEN data from the first data taking period.
The νGEN detector collected 94.50 (47.09) days of data with the reactor ON (OFF),
24 REACTOR NEUTRINO EXPERIMENTS
observing no significant difference between the spectra of the two data sets. Assuming a
SM CEνNS signal and looking between 320 and 360 eV, νGEN disfavored any Lindhard-
like nuclear recoil quenching factor model for germanium with a parameter k > 0.26 at
90% C.L..
For the new data taking period, the νGEN detector has been placed on a lifting structure
which allows it to be positioned within 12.5 and 11 m away from the reactor’s core. This
feature allows to suppress systematic errors coming from the reactor flux. Also, several
new detectors with masses between 1 and 1.4 kg will be added to the current setup and a
new inner veto and new electronics could be placed to further reduce the background [95].
2.2.4. RED-100
The RED-100 detector is a dual-phase liquid xenon time projection chamber (TPC),
with ∼100 kg of fiducial volume, surrounded by a ∼5 cm-thick copper and ∼60 cm-thick
water shields. It is placed under the 3 GWth reactor unit #4 at the Kalinin nuclear
power plant in Udomlya, Russia, ∼19 m away from the core. The detector has a vertical
overburden of ∼65 m.w.e. and an energy threshold of 4 ionization electrons2. RED-
100 completed its first period of data taking, corresponding to ∼8.9 (6) days in the
RON (ROFF) mode, finding no significant correlation in external background rate with
reactor operation [97]. Currently, data is being analyzed and an upgrade of the detector
is ongoing.
2.2.5. NEON
The Neutrino Elastic scattering Observation with NaI (NEON) detector is located 23.7 m
away from the core of the 2.8 GWth reactor unit 6 at the Hanbit nuclear power complex
in Yeonggwang, Korea. The experimental site, which is a gallery 10 m underground,
has a ∼20 m.w.e. overburden. NEON employs a 13.3 kg array of NaI(Tl) scintillation
crystals, with an energy threshold of ∼200 eV, and has passive and active shields. In
the initial results [98], a background level of 6 DRU in the 2 to 6 keV energy region was
reported. Data taking is ongoing.
21 ionization electron in liquid xenon corresponds to ∼15 (250) eV for electronic (nuclear) recoils [96].
Chapter 3
CONNIE with standard CCDs
The Coherent Neutrino-Nucleus Interaction Experiment (CONNIE) is located ∼30 m
away from the core of the 3.95 GWth Angra 2 nuclear reactor at the Almirante Álvaro
Alberto nuclear power plant in Rio de Janeiro, Brazil, inside a shipping container (see
Fig. 3.1). In steady-state operation, the total neutrino flux at CONNIE produced by
the reactor is ∼7.8×1012 ν̄e/cm2/s. CONNIE employs low-noise Charge-Coupled De-
vices (CCDs)1 to search for low-energy nuclear recoils produced by the coherent elastic
scattering of reactor antineutrinos with silicon nuclei, and to probe new physics. CON-
NIE is the first reactor neutrino detector to deploy a multi-CCD array. The engineering
detector prototype was installed in the container in late 2014 and the results of that run
are discussed in Ref. [6]. A complete first upgrade of the sensors was done in mid 2016,
installing fourteen 16 Mpix standard CCDs, reaching a total mass of ∼80 g [7].
Figure 3.1. Left) Almirante Álvaro Alberto nuclear power plant in Angra dos Reis, Rio de
Janeiro, Brazil. Right) Dome of the Angra 2 nuclear reactor and the shipping container where
CONNIE is located.
3.1. CCDs structure, functioning and packaging
CONNIE standard CCDs are thick (∼675 µm), fully-depleted silicon sensors designed at
the Lawrence Berkeley National Laboratory (LBNL) in collaboration with the FNAL,
12009 Nobel Prize in Physics: “for the invention of an imaging semiconductor circuit – the CCD
sensor” [23]
25
26 CONNIE WITH STANDARD CCDS
starting from an existing design [99,100] for the Dark Energy Survey camera [101]. These
sensors were fabricated at Teledyne DALSA Semiconductor and have the same design
as the standard CCDs used in the DArk Matter In Ccds (DAMIC) experiment, briefly
discussed in Chapter 6. Each CCD has 4116×4128 pixels of 15×15 µm2 and a p-type
buried channel implant on a high-resistivity n-type substrate (∼14 kΩ-cm). The high
resistivity allows to fully deplete the substrate of majority charge carriers when applying
a substrate bias voltage Vsub of ∼70 V.
Each pixel has in its front surface three polysilicon gate electrodes over an insulator
layer consisting of SiO2 and Si3N4. Along each parallel register, i.e. a column of pixels,
a p-type Si layer (buried channel) separates the insulator layer from the substrate. The
channel stops are n+ Si regions between the substrate and the insulator layer that de-
limit the CCD columns, separating the buried channels from each other. Early in the
fabrication process, a ∼1 µm thick layer of in-situ polysilicon doped with phosphorous
capped with Si3N4 is deposited on the back side of the substrate for extrinsic getter-
ing [102], i.e. removing most harmful impurities in the substrate. This layer serves also
as the backside ohmic contact to apply the substrate bias voltage. Additional layers of
polysilicon and SiO2 are added to the backside.
The substrate bias voltage is applied from the front side of the device through an im-
planted n+ substrate contact ring that surrounds the array of pixels creating an equipo-
tential surface that extends from the front n+ ring to the backside ohmic contact. A
set of floating p+ guard rings, enclosed by the n+ contact ring, gradually decrease the
potential from Vsub to the inner p+ ring connected to ground. Drawings of the structure
of a CONNIE-like CCD are shown in Fig. 3.2.
Figure 3.2. Images adapted from Ref. [103]. Left) Cross-sectional drawing (yz plane) of a
CCD pixel showing the three-phase gate electrodes (14), the p-type buried channel (15) over
the Si substrate (18) and the backside n+ ohmic contact (12). Center) Cross-sectional drawing
(xz plane) of a CCD showing the frontside SiO2 layer (32), the n+ channel stops (34) and
the unimplanted gaps (36) between the buried channels (15). Right) Cross-sectional drawing
showing the n+ ring (22) in which Vsub is applied, the floating p+ guard rings (27) and the
grounded inner ring (26). The depleted (23) and undepleted (24) regions are shown separated
by the depletion edge (25).
Ionizing radiation interacting in the substrate creates electron-hole pairs. The charge
carriers (holes for a n-type substrate) are drifted along the substrate towards the CCD
3.1. CCDS STRUCTURE, FUNCTIONING AND PACKAGING 27
surface due to the electric field generated when Vsub is applied. While being drifted,
charge carriers diffuse transversely until being collected in the potential minima gener-
ated under the gate electrodes when biased, near the buried channel-substrate junction.
The lateral charge spreading follows a two-dimensional Gaussian distribution with a
spatial variance proportional to the transit time. The variance is related to the depth
at which the interaction takes place, z, as [99]
σ2x = σ2y = α ln (1− βz) , (3.1)
where α and β are related to the sensor’s physical properties and operating parameters.
In three-phase CCDs, every third electrode in the same register is at the same potential,
whose value is controlled over time by the same clock. During readout, charge carriers
are transferred from pixel to pixel along each parallel register to the serial register, i.e.
the last row of pixels, by sequentially clocking the three-phase electrodes. Pixels in the
serial register are wider than pixels in the parallel registers to accommodate binning,
i.e. join the charge of adjacent pixels. The same transfer process happens in the serial
register, where charge carriers are moved from pixel to pixel to an output stage, where
they are read. CONNIE CCDs have two serial registers, each of them connecting to
output stages at both ends. This allows to read out the pixel array in quadrants, halves
or in a whole when using four, two or one output stages, respectively (see Fig. 3.3).
Figure 3.3. Image adapted from Ref. [100]. Diagram of a CONNIE-like CCD. Arrows denote
the flow of charge during readout when using the four output stages.
28 CONNIE WITH STANDARD CCDS
In order to be placed in the experimental setup, CONNIE CCDs are packaged. The
packaging was done at the Silicon Detector Facility (SiDet) at the FNAL. The back of
each 6 cm × 6 cm CCD is glued to a 7 cm × 7 cm Si frame coming from the same ingot
used for the fabrication of the sensors. A Kapton flex cable, which carries the signals to
drive and read the CCD, is also glued to the Si frame and connected to the sensor with
micro-wire bonds. The sensor, frame and flex cable are mounted on a two-piece copper
tray to provide mechanical support, thermal connection and infrared shield. A picture
of a CONNIE packaged standard CCD is shown in Fig. 3.4 (left).
3.2. Experimental setup, operation and data calibration
CONNIE packaged CCDs are mounted horizontally inside a copper cold box with capac-
ity to hold 20 packages that shields the sensors from environmental infrared radiation,
shown in Fig. 3.4 (center). This box is connected with a cold finger to a closed-cycle
helium cryocooler to cool down the sensors to typical operating temperatures, around
120 K. The cold box is suspended below a 15 cm height lead cylinder, which shields the
sensors from radiation coming from the electronics. Both elements are placed inside a
∼20 cm diameter, ∼80 cm long cylindrical copper vacuum vessel that is continuously
evacuated using a turbo-molecular pump, reaching typical pressures of ∼10−7 torr. The
Kapton flex cables connect to second-stage Kapton flex extensions which connect to the
vacuum-side of a vacuum interface board (VIB) to provide the electronics feedthrough.
The air-side of the VIB connects to the Monsoon acquisition system [104], developed for
the DECam imager [105].
Figure 3.4. CONNIE: Packaged standard CCD (left), Cu cold box with 14 CCDs installed
(center) and experimental setup showing the passive radiation shield (right).
CONNIE has a passive radiation shield, shown in Fig. 3.4 (right). It consists of, from
the outermost part, 30 cm of polyethylene, 15 cm of lead and, finally, 30 cm more of
polyethylene. Lead is used to shield γ rays while polyethylene is efficient for shielding
3.2. EXPERIMENTAL SETUP, OPERATION AND DATA CALIBRATION 29
neutrons. The inner polyethylene layer is placed to shield the neutrons produced when
cosmic muons pass through the lead layer.
CONNIE is operated remotely and its operating parameters, e.g. pressure, temperature,
readout noise, are continuously monitored. To reduce external noise sources, during the
sensors readout, all circuits are disconnected from the AC power network and connected
to an uninterrupted power supply (UPS); also, the cryocooler is turned off. To minimize
the total time spent reading out the CCDs and increase the signal-to-noise ratio, it is
desirable to have long exposures. However, since the experiment is located at surface,
the background events, mainly cosmic muons, quickly populate the pixels. The exposure
time is chosen such that the occupancy, i.e. the fraction of pixels associated to events,
remains less than ∼10%.
CONNIE CCDs are read out using only one serial register and one output amplifier,
the left (L). Few columns of pixels are readout before the active area (prescan region),
and more pixel values are registered after the active area by overclocking the registers
beyond the physical extent of the CCD (overscan region). The data are stored as a FITS
image. Data taking periods are divided into runs, i.e. sets of images sharing a common
detector configuration. The acquired images are processed to remove unwanted offsets
and noise. The image processing chain is discussed in detail in Ref. [7].
From the processed images, a catalog file containing the information of reconstructed
events, i.e. pixel clusters associated with energy depositions in the CCD, is obtained.
From this file, a background spectrum is computed. Energy calibration is performed
by fitting the position of fluorescence peaks in the data: Cu Kα, Cu Kβ and Si, corre-
sponding to 8.047 keV, 8.905 keV and 1.740 keV, respectively. Depth-size calibration is
obtained by fitting the width of cosmic muon tracks as a function of depth with Eq. (3.1).
Depth is calculated from the track length on the xy plane using trigonometry, as muons
are assumed to straightly cross the entire detector thickness leaving a nearly constant
deposition per unit length. Stability of the background radiation is monitored by looking
at the intensity of the Cu Kα fluorescence peak and the muon rate, from ∼200 keV to
∼400 keV, over time.
CONNIE acquires data during reactor on (RON) and reactor off (ROFF) periods to
perform a model-independent search of events associated to possible interactions of the
reactor electron antineutrinos in the sensors. This is done by looking for an excess of
events in the subtraction of the ROFF spectrum from the RON one. ROFF periods
are scheduled shutdowns to provide maintenance to the reactor and occur for ∼1 month
every ∼13 months.
30 CONNIE WITH STANDARD CCDS
3.3. Expected event rate in CONNIE
The differential event rate as a function of the target recoil energy ER in the CONNIE
experiment is
dR
dER
= NT
∫ ∞
Emin
ν̄e
dΦ
dEν̄e
dσ
dER
dEν̄e , (3.2)
where dσ/dER is the differential cross section of the interaction (see Chapter 1), dΦ/dEν̄e
is the reactor antineutrino flux as a function of the antineutrino energy, NT is the number
of targets in the detector and Emin
ν̄e is the minimum antineutrino energy that can produce
a target recoil with energy ER.
Depending on the nature of the target, part or all of its recoil energy is used to ionize.
The ionization yield, also called the quenching factor Q, is defined as the fraction of the
total recoil energy that goes into ionization: Q = EI/ER. If the target is an electron,
Q = 1; however, if it is a nucleus, Q can be described as a function of either EI or ER,
see Section 3.3.1.
Taking Q into account, the differential event rate as a function of EI is
dR
dEI
=
dR
dER
dER
dEI
=



dR
dER
1
Q
(
1− EI
Q
dQ
dEI
)
if Q ≡ Q(EI)
dR
dER
1
Q
(
1 + ER
Q
dQ
dER
)−1
if Q ≡ Q(ER)
. (3.3)
The expected event rate R in CONNIE between E1 and E2 is given by
R =
∫ E2
E1
ε(EM )
dR
dEM
dEM , (3.4)
where EM is the measured energy and ε(EM ) is the detection efficiency, accounting for
the selection cuts and reconstruction efficiency. Assuming a Gaussian detector response,
the differential event rate as a function of EM is
dR
dEM
=
∫∞
0 G
(
EM , EI ;σ
2
I
) dR
dEI
dEI
∫∞
0 G
(
EM , EI ;σ2I
)
dEI
, (3.5)
where
G
(
EM , EI ;σ
2
I
)
=
1
√
2πσ2I
exp
{
−(EM − EI)
2
2σ2I
}
, (3.6)
with σ2I = σ20+FEehEI characterizing the energy resolution of the CCD. At low energies,
the resolution is dominated by the sensor readout noise, characterized by σ0, while, at
high energies, it is proportional to EI through the product of the Fano factor, F ≃
3.3. EXPECTED EVENT RATE IN CONNIE 31
0.133 [106], and the mean ionization energy required for photons to produce an electron-
hole pair in silicon, Eeh ≃ 0.003745 keV [107].
3.3.1. Nuclear recoil quenching factor in Si
When a nuclear recoil is produced inside the detector, a part of its energy generates
charge carriers and the rest contributes to the increase of the system’s thermal energy.
The nuclear recoil quenching factorQ accounts for this and it is believed to be an intrinsic
property of each material. In the case of Si, experimental measurements of Q for ER ≳ 4
keV [108–111] have been consistent with a model developed by Lindhard et al. [112].
However, recent measurements at lower energies [113–116] show a discrepancy with this
model, evidencing the need to remodel from basic principles the energy deposition of a
recoiling nucleus traveling in a crystal lattice. Some efforts have been made towards this
end [117–119]. Recent measurements and models of the nuclear recoil quenching factor
in Si are shown in Fig. 3.5.
Two measurements of the quenching factor for ER below 4 keV were performed using
silicon CCDs in different experiments [113, 114]. An analytical fit to the measurements
in Ref. [113] is considered in this work, parametrized as
Q(EI) =
p3EI + p4E
2
I + E3
I
p0 + p1EI + p2E2
I
, (3.7)
with p0 = 56 keV3, p1 = 1096 keV2, p2 = 382 keV, p3 = 168 keV2 and p4 = 155 keV.
●
●
●
●
●
●
●
●●
●●
●
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼
▼ ▼
■
■
■
■
■
■
Lindhard
Chavarria fit
Sarkis2020
Sarkis2022
● Chavarria
▼ Izraelevitch
■ Albakry
0.05 0.10 0.50 1 5 10
0.001
0.010
0.100
1
10
ER (keV)
E
I
(k
e
V
)
Figure 3.5. Recent measurements (data points) [113, 114, 116] and models (orange and purple
solid lines) [118,119] of the nuclear recoil quenching factor in Si. The Lindhard model (dashed)
is shown for comparison. Also, the analytical expression that fits the measurements in Ref. [113],
given in Eq. (3.7), is shown in red.
32 CONNIE WITH STANDARD CCDS
3.4. Results from the 2016-2018 run
We took 3-h exposure images in the standard CCD readout mode, i.e. the charge
collected by each pixel is read out individually. The readout of the whole CCD array,
optimized to achieve a low pixel noise, took ∼16 min. CONNIE took data continuously
from August 2016 to August 2018, with short interruptions due to planned on-site
interventions or power cuts. Data was divided in three seasons: 1) August 2016 to
March 2017 - Includes one of the reactor shutdowns and ended with a planned period
for detector maintenance; 2) March to December 2017 - An infraestructure upgrade of
the laboratory was made; and 3) January to August 2018 - Includes the second reactor
shutdown.
For the analysis, we used data from seasons 1 and 3, when the selected CCDs (8 out of
14) showed a good performance, i.e. a readout noise below 2.2 e− and dark counts less
than 0.3 e−/pix/h (quality cuts). We also applied geometrical cuts to the data removing
the volume corresponding to the edge of the sensors and hot columns, i.e. columns with
an excess of pixels with high dark current compared to the rest of the sensor. The data
passing the cuts correspond to a total exposure of 3.7 kg-days: 2.1 kg-days with RON
and 1.6 kg-days with ROFF.
We selected events with energy above 10 e−, corresponding to ∼4 times the readout noise.
To identify neutrino-like events, a statistical test, based on the likelihood of the pixel
values of an event to not follow the distribution function of on-chip noise sources events
simulated over blank images, was performed. Events passing this test were selected.
The RON and ROFF energy spectra of selected events are shown in Fig. 3.6.
Figure 3.6. Taken from Ref. [7]. Energy spectra of selected events from the periods of RON and
ROFF in the CONNIE 2016-2018 run. Cu Kα, Cu Kβ and Si fluorescence peaks are seen.
Neutrino-like events with a uniform random probability in the active volume and a uni-
3.5. RESULTS FROM THE 2019 RUN 33
form distribution in energy up to 2.5 keV were simulated over images from the ROFF
period. The size of the simulated events was computed according to the depth-size cal-
ibration. The images with simulated neutrino events were processed using the standard
chain and the selection cuts were applied. From these images the overall detection ef-
ficiency ε(EM ) was computed, see Ref. [7]. This efficiency was used to compute the
CEνNS expected event rates considering Lindhard [112] and Chavarria [113] nuclear
recoil quenching factors, shown in Fig. 3.7.
The search for a signal was done by looking at the energy spectrum difference of RON
minus ROFF selected data. We found no significant excess allowing us to put a 95% one-
sided confidence level (C.L.) upper limit on the rate, shown in Fig. 3.7 (right). Compared
to the CEνNS expected event rate between 0.075 keV and 0.275 keV considering the
Chavarria [113] quenching factor, the limit established by the analysis of the CONNIE
2016-2018 run data is 41 times greater.
Figure 3.7. Taken from Ref. [7]. 95% C.L. upper limit from RON-ROFF measurements (solid
line) and CEνNS expected event rates using the Chavarria [113] (dashed line) and Lindhard [112]
(dotted line) nuclear recoil quenching factors.
3.5. Results from the 2019 run
In the 2019 run, a readout mode performing 1×5 binning, i.e. 5 pixel rows are transferred
into the serial register before the charge is read out, was implemented. This readout
mode increases the signal-to-noise ratio, as the effective readout noise per pixel is 5
times smaller, at the cost of losing spatial resolution, see Fig. 3.8. To have a low pixel
occupancy, we took 1-h exposure images. The readout time of the full CCD array with
1×5 binning took ∼3.5 min, 5 times smaller compared to the one in standard readout.
We performed a hidden-data analysis looking only at images corresponding to the ROFF
34 CONNIE WITH STANDARD CCDS
Figure 3.8. Taken from Ref. [9]. Diagram illustrating 1×5 binning. Five pixel rows in a standard
read out image correspond to one row in a 1×5 binned image (left). Parts of a standard
image (middle) and a binned image (right) are shown for comparison. The binned image seems
compressed in the vertical direction; the full 1×5 binned image has 5 times fewer pixels than
the full standard image. Pixels in the binned image have an effective size of 15 µm×75 µm.
period. We updated the techniques and tools to calibrate the sensors and monitor their
performance. The temporal and geometrical selection cuts are similar to the analysis
of the 2016-2018 run. Any image with a readout noise or dark counts 5 standard de-
viations above the measured mean values was excluded. Also, we removed the volume
corresponding to 140 columns and 10 rows from the edge of the sensors, as well as the
hot columns. The data passing these cuts correspond to a total exposure of 2.7 kg-days:
1.4 kg-days with RON and 1.3 kg-days with ROFF.
We studied the low-energy background events in the ROFF data. We compared the
size distribution of data events with energy between 0.1 and 0.2 keV to the expected
distributions from simulated events uniformly distributed in the front, bulk and back of
the sensor, with data showing a clear excess of backside events, see Fig. 3.9 (left). From
size studies of low-energy data events, we identified two main background sources: large
low-energy (LLE) events, with sizes greater than 1.2 pixels and energies below 0.4 keV,
and events from a partial charge collection (PCC) layer, with sizes between 0.9 and 1.2
pixels, see Fig. 3.9 (right).
Regarding LLE events, two production mechanisms were identified based on the study
of the low-energy background: 1) charge from a large ionization packet that is released
up to a hundred pixels after the main charge packet, and 2) charge that was diffused
to the serial register from charge depositions in the inactive volume of the sensor, also
referred to as serial register events (SRE). Due to their characteristic one-dimensional
shape, LLE events can be rejected by requiring a maximum event size.
The PCC layer in analogous CCDs has been deeply studied by the DAMIC collabora-
tion [120, 121]. When creating the gettering layer during the CCD fabrication process,
some phosphorous diffuses into the substrate leading to a ∼5 µm thick region with a
transient in the phosphorous concentration, going from ∼1020 atoms/cm3 in the back-
3.5. RESULTS FROM THE 2019 RUN 35
side contact, where all free charge recombines, to ∼1011 atoms/cm3 in the fully-depleted
active region, where charge recombination is negligible [121]. A fraction of the charge
carriers produced by ionizing radiation in this region may recombine before reaching the
fully-depleted region, leading to a partial charge collection (PCC). This effect distorts
the observed energy spectrum as the ionization signal of an event would be smaller if
it deposited energy in the PCC layer than in the fully-depleted region. This layer can
be removed from the CCDs in an intermediate step during the fabrication process by
polishing the backside surface of the sensors [99]. In this case, the CCD backside ohmic
contact is formed by depositing a ∼20 nm thick layer of in-situ polysilicon doped with
phosphorous, significantly reducing the partial charge collection effect.
CONNIE CCDs were fabricated without the backside treatment; therefore, PCC events
are observed. Since the PCC layer is on the backside, the associated events have a size
close to 1 pixel. As in the case of LLE events, most of the PCC events can be rejected
by requiring a maximum event size.
Figure 3.9. Taken from Ref. [9]. Left) Size distribution of events with energy between 0.1
and 0.2 keV in ROFF data (blue), compared with distributions from simulated events from the
front (black), bulk (cyan) and back (pink) regions. The red line is the sum of the simulated
distributions that fits the data. Right) Size versus energy histogram of low-energy events in
ROFF data. Distributions coming from LLE and PCC events are shown.
We selected events with energy above 45 eV, corresponding to 4-5 times the readout noise
and a size below 0.95 pixels. Spatial and temporal uniformity of the selected events in
ROFF data was statistically tested, showing no trend. With the cuts fixed, we analyzed
the RON data and computed the energy spectra of selected events during the ROFF
and RON periods, shown in Fig. 3.10. A substantial improvement in the background
event rate below 1 keV can be seen in the spectra compared to that of the 2016-2018
run.
Following the same procedure as in the analysis of the CONNIE 2016-2018 run data,
the overall detection efficiency ε(EM ) was computed, see Ref. [9]. We used this effi-
ciency to compute the CEνNS expected event rates considering Chavarria [113] and
36 CONNIE WITH STANDARD CCDS
Figure 3.10. Taken from Ref. [9]. Energy spectra of selected events from the periods of RON
and ROFF in the CONNIE 2019 run. Cu Kα, Cu Kβ and Si fluorescence peaks are seen.
Sarkis2020 [118] quenching factors, shown in Fig. 3.11. From the energy spectrum dif-
ference of RON minus ROFF selected data, we found no significant excess allowing us
to put a 95% one-sided confidence level (C.L.) upper limit on the rate (observed limit),
shown in Fig. 3.11. For comparison, we computed the expected 95% C.L. limit assum-
ing zero as the mean. The observed upper limit between 0.05 and 0.18 keV is larger
than the CEνNS expected rate by a factor of 66 (75) considering the Sarkis2020 [118]
(Chavarria [113]) quenching factor.
Figure 3.11. Taken from Ref. [9]. The 95% C.L. observed (expected) upper limit from RON-
ROFF measurements is shown in blue (orange). The CEνNS expected event rate considering
the Sarkis2020 [118] (Chavarria [113]) quenching factor is shown in green (red).
3.6. SEARCH FOR LIGHT MEDIATORS WITH CEνNS CHANNEL 37
3.6. Search for light mediators with CEνNS channel
We use CONNIE results to search for new physics using the CEνNS detection channel.
The expected event rates in CONNIE accounting for the interaction of the neutrino
with a light scalar and a light vector mediator were computed, considering the cross
sections in Eqs. (1.20) and (1.26). The 95% C.L. exclusion limits for the light mediators
parameters, in the framework of the universal simplified model described in Section 1.4.2,
are calculated as the curve in the 2D space for which the rate is
RSM+X(MX , gX) =
∫ E2
E1
ε(EM )
dRSM+X
dEM
dEM ≥ R 95%C.L. , (3.8)
with X = {ϕ, Z ′}. Here, g2X ≡ gν,ν̄X gqX and R 95%C.L. is the CONNIE 95% C.L. upper
limit on the event rate.
We consider the results from the analysis of the 2016-2018 data (1 × 1) and the 2019
data (1× 5). As the expected rate rapidly decreases with energy, the integration limits
we consider are given by the lowest energy bins in Figs. 3.7 and 3.11, corresponding to
E1 = 0.075 (0.050) keV and E2 = 0.275 (0.180) keV for the 1 × 1 (1 × 5) data. For
the overall detection efficiency ε(EM ), we consider an analytical expression to fit the
CONNIE published efficiency of the 1× 1 (1× 5) data [7,9], valid for EM > 64 (35) eV,
which is given by
ε(EM ) = b0 −
[
1 + eb1(EM−b2)
]−1
, (3.9)
where b0 = 0.74 (0.64), b1 = 17.47 (34.55) keV−1 and b2 = 0.12 (0.05) keV. Fig. 3.12
shows the CONNIE detection efficiencies for the 1× 1 and 1× 5 data and their fits with
Eq. (3.9).




















 
  

  









 
 

       

   
1x1
1x5
0.00 0.05 0.10 0.15 0.20 0.25 0.30
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
EM (keV)
ϵ
(E
M
)
Figure 3.12. CONNIE detection efficiencies corresponding to the 1×1 (blue) and 1×5 (orange)
data. The published efficiencies (•) and the fits with Eq. (3.9) (solid line) are shown.
38 CONNIE WITH STANDARD CCDS
The resulting exclusion regions are shown in Figs. 3.13 and 3.14. The latter shows a
comparison of these regions corresponding to the 1 × 1 and 1 × 5 results. The most
significant systematic uncertainty in these limits comes from the nuclear recoil quench-
ing factor. To quantify it, we include in the plots in Fig. 3.13 the exclusion regions
corresponding to the Lindhard [112] and Chavarria [113] quenching factors. Additional
systematic uncertainties of the CONNIE 95% C.L. upper limits in Refs. [7, 9] are sub-
dominant. These are related to the reactor neutrino flux, the detection efficiency and
the energy calibration stability, contributing less than 5%, 10% and 2%, respectively, to
the event rate in the lowest-energy bin. These lower level effects will become relevant
once the uncertainty in the quenching factor is significantly reduced in future analyses.
COHERENT
CONNIE Chavarria
CONNIE Lindhard
1 10 100 1000 104
10-6
10-5
10-4
0.001
0.010
Mϕ (MeV)
g
ϕ
COHERENT
CONNIE Chavarria
CONNIE Lindhard
Δaμ
1 10 100 1000
1.×10-5
5.×10-5
1.×10-4
5.×10-4
0.001
0.005
0.010
Mz' (MeV)
g
z
'
Figure 3.13. Taken from Ref. [8]. Exclusion regions in the (MX , gX) plane from CONNIE 1× 1
results using as quenching the fit to the measurements in Ref. [113] (Chavarria) in orange and
the one in Ref. [112] (Lindhard) in red. The top (bottom) plot corresponds to the scalar (vector)
mediator. The COHERENT exclusion regions for the scalar [122] and vector [58] parameters,
in blue, and the 2σ allowed region in the Z ′ parameter space that could explain the anomalous
magnetic moment of the muon (see reviews [123,124]), in green, are shown for reference.
3.6. SEARCH FOR LIGHT MEDIATORS WITH CEνNS CHANNEL 39
COHERENT
CONNIE 1x1
CONNIE 1x5
1 10 100 1000 104
10-6
10-5
10-4
0.001
0.010
Mϕ (MeV)
g
ϕ
COHERENT
CONNIE 1x1
CONNIE 1x5
1 10 100 1000 104
1.×10-5
5.×10-5
1.×10-4
5.×10-4
0.001
0.005
0.010
Mz' (MeV)
g
z
'
Figure 3.14. Exclusion regions in the (MX , gX) plane from CONNIE 1×1 (1×5) results, shown
in orange (red). The top (bottom) plot corresponds to the scalar (vector) mediator. These curves
were computed using as quenching factor the fit to the measurements in Ref. [113] (Chavarria)
for the scalar mediator, and the one in Ref. [118] (Sarkis2020) for the vector mediator.
For both mediators, scalar and vector, the contribution to the event rate is proportional
to g2X/(2MER +M2
X). For light mediators, i.e. MX ≪ √
2MER, the rate contributions
depend only on gX and the limits become independent of MX . For heavy mediators, i.e.
MX ≫ √
2MER, the rate contributions are proportional to the ratio gX/MX . These two
cases are readily seen in Figs. 3.13 and 3.14. The CONNIE exclusion region in the Z ′
parameter space confirms the statement in Ref. [58], disfavoring a light vector mediator
to explain the discrepancy in the anomalous magnetic moment of the muon.
The exclusion regions derived from the CONNIE 1×5 results do not significantly improve
the bounds imposed with the 1 × 1 results because, despite having higher detection
efficiency at low energies, the observed 95% C.L. limit on the rate was significantly
higher. At the time they were published, the exclusion regions from the 1× 1 results [8]
were the best limits among the experiments searching for CEνNS in the low-mass regime,
40 CONNIE WITH STANDARD CCDS
Mφ (Z′) < 30 (10) MeV for a scalar (vector) mediator, extending beyond the region
excluded by the COHERENT results in Refs. [58,122]. Since reactor antineutrinos have
lower energies than neutrinos from a pion decay-at-rest (π-DAR) source, our bounds
on light scalar or vector mediators are stronger (weaker) at smaller (larger) mediator
masses, making evident the complementarity of two different techniques to explore new
physics. These results were the first search for non-standard interactions with reactor
neutrinos and CCDs, and demonstrated that reactor experiments provide a powerful
probe for new physics at low energies.
Note that the presented search for new physics was based on a counting experiment,
comparing the rate above threshold in CONNIE with the expectations from the univer-
sal simplified model with light mediators. Stronger limits are expected when spectral
information of the CONNIE data is included in the analysis. It should also be high-
lighted that there are limits from different experiments on the universal simplified model
with light mediators which can be recasted for comparison [59,125]. In Fig. 3.15 we show
a plot from Ref. [125] in which authors have put together the existing limits from CEνNS
experiments (COHERENT-CsI+LAr, CONNIE, CONUS and Dresden-II), from multi-
ton DM experiments (XENONnT and LZ), from solar neutrino data (Borexino), from
collider experiments, from beam-dump experiments and from the Big Bang Nucleosyn-
thesis (BBN). For further details, reader should refer to the original source.
Figure 3.15. Taken from Ref. [125]. Existing limits on the universal simplified model with a
light scalar (left) and light vector (right) mediators.
Chapter 4
Skipper CCDs
The readout noise in standard CCDs can not be reduced below ∼2 e− RMS/pix because
of the presence of low-frequency noise, also called 1/f noise. In 1990, Janesick et al.
first proposed the idea of the skipper CCD to circumvent this problem [126,127]. They
presented a sensor in which multiple (Nskp) non-destructive measurements can be made
for each pixel. By averaging the Nskp independent samples of the same pixel off-chip, the
impact of the low-frequency readout noise can be reduced, as the noise would decrease
as σ = σ1/
√
Nskp where σ1 is the readout noise of one sample. By increasing Nskp,
sub-electron levels of noise can be achieved, allowing to precisely count the number of
electrons in each pixel. However, measuring multiple times each pixel increases the
readout time by ∼Nskp, limiting the sensor’s time resolution. As many applications
do not require a single-electron resolution in every pixel, multiple sampling could be
performed only in the areas of interest, whereas the rest of the pixels could be single-
sampled, or “skipped”. This is the origin of the term “skipper”.
Janesick’s very first experimental attempt of a skipper CCD was reported in 1990 [126,
127] and patented in 1993 [128]. It was designed “for ultra low-signal level imaging and
spectroscopy applications that require sub-electron read noise floors” [127]. To perform
multiple non-destructive measurements of the same charge packet, a floating-gate output
circuit [129, 130] was surrounded with CCD gates so that the charge can be moved to
and from the sense node. In this first attempt, the lowest noise achieved was 0.5 e−
RMS performing 256 samples per pixel. However, the noise of this device could not be
further reduced impeding the detection of the single photo-electron.
Since 2009 scientists from the LBNL and the FNAL have collaborated developing and
testing skipper CCDs [131]. In 2017, the sub-electron readout noise of the skipper CCD
was demonstrated using a large-area detector of 4126×866 pix, designed by Stephen
Holland [5]. A readout noise of 0.068 e− RMS/pix was measured with 4000 samples/pix
demonstrating the electron-counting capability of the sensor in a wide energy range;
see Fig. 4.1. Because of this feature, the skipper CCD technology is very attractive for
experiments where a low-energy threshold is required.
41
42 SKIPPER CCDS
(a) (b)
Figure 4.1. Figures taken from [5]. (a) Measurements of a skipper CCD readout noise as a
function of the number of samples per pixel. The red line represents the theoretical expectation.
(b) Charge distribution with single-electron resolution in a wide energy range.
4.1. Standard vs. Skipper CCD output stage and readout
The fundamental difference between a standard and a skipper CCD lies in the output
stage. The first element in this stage is a MOS capacitor with an independent clock.
This capacitor allows to store charge coming from the serial register; that is why its
electrode is usually known as summing gate (SG). The second element is another MOS
capacitor whose main function is to transfer the stored charge to the sense node. Its
electrode is often referred to as output gate (OG). These first two elements are common
to the standard and skipper output stages.
In a standard CCD, the sense node is actually the source terminal of the reset MOSFET
(RT) and the gate terminal of the output MOSFET (OT). It is commonly referred to
as a floating diffusion region because its potential is allowed to float when the RT is off
and because of its structure, usually a n-type implant allowed to diffuse into a p-type
substrate.
To start the readout process, at t0 the RT is switched on, draining any charge in the
sense node and setting its potential to a reference voltage (Vref). Then, at t1 charge is
transferred to the sense node by raising VSG. The OT, configured as a common-drain
amplifier, proportionally converts the charge into a voltage, performing a non-destructive
readout. In the standard output stage, moving the charge backwards from the sense node
under the OG is not possible. This is because, if the potential beneath OG ever became
equal or lower than Vref, charge would be injected through the RT. This cycle is repeated
destroying each charge packet after reading it.
4.1. STANDARD VS. SKIPPER CCD OUTPUT STAGE AND READOUT 43
In a skipper CCD, the sense node is a MOS capacitor with a floating gate embedded in
the SiO2 layer. This gate is capacitively connected to the source terminal of the reset
MOSFET and to the gate terminal of the output MOSFET. Next to the sense node,
a last MOS capacitor allows to drain the charge towards an ohmic contact at Vdrain.
Its electrode is usually known as drain gate (DG). Any packet of charge in the sense
node, being isolated from its floating gate, can be moved back and forth by changing
the potentials in the adjacent electrodes.
To start the readout process, at t0, any charge in the sense node is drained by applying
a pulse to the DG. At the same time, the potential in the floating gate is fixed to Vref by
switching on the RT. By raising VSG, at t1, charge is transferred to the sense node and
the OT performs a first non-destructive readout. To perform a second measurement of
the same charge packet, at t2, VSG and VOG are lowered to move the charge from the
sense node under the SG. Then, VOG is raised and RT is switched on to fix again the
potential in the floating gate. This cycle can be repeated to sample the same charge
packet multiple times (Nskp).
Figure 4.2. Simplified diagrams of a standard CCD output stage (left) and a skipper CCD
output stage (right).
Regardless of the CCD output stage, a correlated double sampling method is usually
used to obtain the value of each pixel sample, as it eliminates the high-frequency noise.
The output signals of the OT are integrated for a period of time, usually a few µs, after
t0 to register the voltage associated to the pedestal, and after t1 to register the voltage
associated to the charge. The voltage associated to the pixel sample is, simply, the
difference between these two signals. Then, an analog-to-digital converter (ADC) links
this voltage to a binary number. These numbers can be used to construct a bidimensional
image in which each Nskp consecutive pixels in the same row correspond to the Nskp
performed measurements of the same charge packet (“raw” images).
44 SKIPPER CCDS
4.2. Instrumental sources of Single-Electron Events (SEE)
4.2.1. Thermal dark current
“Dark current,” i.e., charge carriers going from the valence to the conduction band
because of thermal fluctuations, presents an irreducible source of 1e− events in skipper
CCDs, constraining the lowest 1e− rate that can be achieved. To minimize it, sensors
are cooled to temperatures around 120 K. The 1e− rate coming from dark current (DC)
in a CCD at temperature T , in units of e−/pix/day, is [132]
RDC,1e− = q−1CT0PS
(
T
T0
)
e
Eg
2kB
(
1
T0
− 1
T
)
× 86400 s/day , (4.1)
where q = 1.602×10−19 C is the electric charge, CT0 is the “dark current figure of merit”
at T0 = 300 K in nA/cm2, PS is the pixel size in cm2, kB = 8.617 × 10−5 eV/K is the
Boltzmann constant and Eg is the silicon band gap energy in eV, which varies with T
following the empirical formula:
Eg = 1.1557− 7.021× 10−4T 2
1108 + T
. (4.2)
The generation of DC carriers can be enhanced by the presence of intermediate energy
levels within the Si bandgap, i.e. state traps, caused by defects in the silicon lattice.
These defects are more prone to be in the CCD layer interfaces, e.g. the Si-SiO2 interface,
near the sensor surfaces. This is why “surface DC” is usually higher than “bulk DC”.
The former can be temporarily reduced by filling the traps with majority charge carriers
when inverting the sensors surface [132]. As they are gradually emptied, surface traps,
and therefore surface DC, return to equilibrium.
4.2.2. State traps
State traps are usually modeled using the Shockley-Read-Hall theory for carrier gen-
eration and recombination [133]. The ones that lie within the charge transfer channel
usually capture one charge carrier from charge packets as they are transferred through
the device, process characterized by the capture time constant τc, and release it at a
later characteristic time τe. These time constants are given by
τc = 1/σcvthnt and τe =
1
σcvthnc
e
Et
kBT , (4.3)
where T is temperature (K), Et is the trap energy level (eV), σc is the charge carriers
cross section (cm2), vth is the charge carriers thermal velocity (cm/s), nt is the trap
4.2. INSTRUMENTAL SOURCES OF SINGLE-ELECTRON EVENTS (SEE) 45
density (1/cm3) and nc is the effective density of states in the conduction band (1/cm3);
vth and nc depend on the rest mass of the charge carriers me and T as
vth =
√
3kBT/0.26me and nc = 2
[
2π(1.08me)
kBT
h2
]3/2
. (4.4)
The probability of a trap to capture (c) or emit (e) one charge carrier within the time
interval [t1, t2] is given by
Pc,e = e−t1/τc,e − e−t2/τc,e . (4.5)
By measuring τe as a function of T , the trap energy and the carrier cross section can
be extracted using Eq. (4.3). These parameters can be compared with known silicon
trap characteristics in the literature. Fig. 4.3 shows an illustration of typical point-like
defects in silicon lattices and Table 4.1 shows the parameters of some of the state traps
found in buried-channel CCDs.
Figure 4.3. Taken from Ref. [134]. Illustration of point defects in silicon: a) vacancy, b)
divacancy, c) self-interstitial, d) interstitialcy, e) interstitial impurity, f) substitutional impurity,
g) impurity-vacancy pair, h) impurity-self-interstitial pair.
The number of 1e− events coming from state traps per exposure can be assumed to
be Nhits × Ntraps, where Nhits is the number of hits, i.e., pixels with more than 2e−,
in one exposure and Ntraps is the mean number of state traps that a hit traverses
during readout, which equals the total number of carriers trapped per hit assuming that
each trap captures one charge carrier. Only state traps with a τe larger than the pixel
readout time are considered because faster traps will release the trapped carrier in the
pixel containing the hit.
Nhits can be estimated from the background rate Rbkgd (DRU). Considering that each
46 SKIPPER CCDS
Defect Et (eV) σc (cm2)
Si-A 0.17 1× 10−14
Si-E 0.46 5× 10−15
(V-V)− 0.39 2× 10−15
(V-V)−− 0.21 5× 10−16
CiPs (III) 0.23 3× 10−15
CiPs (IIB) 0.32 1.5× 10−14
BiOi 0.27 5× 10−16
Table 4.1. Parameters of state traps commonly found in buried-channel CCDs: variations of
the Si self-interstitial defect (Si-A and Si-E), the single− and double−− donor configurations of
the divacancy (V-V), two of the possible configurations of the carbon-interstitial-phosphorus-
substitution defect (CiPs) and the boron-oxygen-interstitial pair (BiOi).
event with energy in the interval [E1, E2] keV activates a mean of n pixels,
Nhits = n×Rbkgd
mpixNpix (E2 − E1)
Nexp
, (4.6)
where Nexp is the number of exposures per day, Npix is the number of pixels corre-
sponding to one exposure and mpix = ρSiPsh is the mass of one pixel in kg, with
ρSi = 2.3× 10−3 kg/cm3 the silicon density and h the pixel thickness in cm.
The rate of 1e− events produced from state traps in e−/pix/day is
RT,1e− =
Nhits ×Ntraps
Npix
Nexp = n×Rbkgd mpix (E2 − E1)×Ntraps . (4.7)
4.2.3. Spurious charge
The high electric field generated in the CCD when the gate voltages change to move the
charge from one pixel to the other can lead to the production of spurious charge (SC),
also known as clock-induced charge [132]. This can happen either in the active area or
in the serial register and it strongly depends on the clock rise time and the clock swings.
The primary source of SC is the clocking of the serial register, which tends to dominate
over the slower vertical clocks due to the higher capacitance of the line across the CCD
active region.
Assuming that the probability of generating one charge carrier on a single pixel transfer
PSC is the same in both registers and considering that each pixel is shifted Nshifts times
during readout, the number of 1e− events from spurious charge in e−/pix/day is
RSC,1e− = Nexp Nshifts PSC . (4.8)
4.2. INSTRUMENTAL SOURCES OF SINGLE-ELECTRON EVENTS (SEE) 47
As RSC,1e− is proportional to Nexp, its contribution to the total 1e− rate can be re-
duced by taking longer exposures (small Nexp). Also, RSC,1e− is proportional to Nshifts;
therefore, reducing Nshifts, by performing binning in the parallel registers, decreases its
contribution. Let us assume that when reading in 1 × 1 mode the number of transfers
in the serial (parallel) register is Nser (Npar); hence N1×1
shifts = Nser + Npar. By doing
1× 10 binning, N1×10
shifts = (Nser/10) +Npar reducing the overall SC generation.
4.2.4. Amplifier light
The ouput MOSFET in a CCD output stage can emit low-energy photons by different
mechanisms [135]. One of them, and probably the primary light production mechanism
in CCDs, is the light emission in saturation [136] as, when reading out the charge,
the output CCD MOSFET is operated as a source-follower amplifier, i.e. in saturation
mode. When the MOSFET works in this mode, a pinched-off region appears in the
conducting channel giving rise to a high electric field region where charge carriers are
drifted promoting electron-hole pair creation, see Fig. 4.4 (left). When free carriers
recombine, a broad spectrum of low-energy photons, with wavelengths ranging from
the visible to the near-infrared, is emitted. It has been observed that the intensity of
the emitted photons from p-channel transistors is lower than the one from n-channel
transistors [137]; this has been attributed to differences in the ionization potential of
electrons and holes.
Figure 4.4. Left) Taken from Ref. [138]. Diagram of the light emission from a saturated n-type
MOSFET. Here, Vg is the gate voltage, Vt is the threshold voltage, Vds is the drain-to-source
voltage and Vdsat ≈ Vds−Vt. Right) Taken from Ref. [139]. Photon absorption length in silicon
as a function of wavelength from measurements (solid) and phenomenological fits (dashed).
The low-energy photons emitted in saturation travel a finite range through the active
area of the CCD, defined by the light absorption length in silicon, see Fig. 4.4 (right).
These photons deposit their energy generating single-electron events which are spatially
localized in the region near the CCD readout stage. The number of 1e− events from
48 SKIPPER CCDS
amplifier light is expected to increase linearly with the time the output transistor re-
mains in saturation mode. Therefore, to minimize it, the transistor can be switched
off during exposure and moved to saturation mode only when reading out the charge.
Also, as the amplifier light intensity depends on the transistor’s polarization voltages,
an optimization of their values can be made to minimize this background source.
4.3. Few-electron events from instrumental SEE
4.3.1. Accidental coincidences
Thermal dark current and spurious charge generation are random processes that can
produce pixels with 2e− or more by accidental coincidences. The count of ne− single
pixel events from these processes per year in a skipper CCD detector with Npix pixels
can be calculated assuming a Poisson distribution,
Kn =
λne−λ
n!
Npix Nexp × 365 days/year , (4.9)
with λ = λDC+λSC , where λDC(SC) = RDC(SC),1e−/Nexp is the mean 1e− rate from DC
(SC) per exposure. Assuming a 10 kg detector with 725 µm-thick sensors composed of
15×15 µm2 pixels, i.e. Npix = 26.65 Gpix, we computed Kn for different run conditions
defined by the parameters on which λ depends (see Table 4.2). As λSC = Nshifts PSC ,
for a given PSC we can compute the maximum Nshifts allowed to meet the different
run conditions. The last column in Table 4.2 shows these values for PSC = 1 × 10−8,
assuming λ ≃ λSC .
Run conditions 2e− 3e− 4e− PSC = 1× 10−8
R1e− = 1× 10−2
Nexp = 1 481M 1.6M 4k Nshifts = 1M
Nexp = 12 40M 11k 2.3 Nshifts = 83k
R1e− = 1× 10−4
Nexp = 1 48.6k 1.6 0 Nshifts = 10k
Nexp = 12 4.1k 0 0 Nshifts = 833
R1e− = 1× 10−6
Nexp = 1 4.9 0 0 Nshifts = 100
Nexp = 12 0.4 0 0 Nshifts = 8
Table 4.2. Counts of 2e−, 3e−, and 4e− single pixel events generated by accidental coincidences
depending on R1e− and Nexp. Nexp = 1 (12) exposure(s)/day means that the full readout of the
detector takes 24 (2) hours. The last column shows the maximum Nshifts allowed to comply
with the run conditions for PSC = 1× 10−8 assuming λ ≃ λSC .
Table 4.2 shows that the number of accidental coincidences decreases with shorter ex-
4.3. FEW-ELECTRON EVENTS FROM INSTRUMENTAL SEE 49
posures (large Nexp). This is the best scenario to minimize accidental coincidences from
DC events. However, as can also be seen from Table 4.2, the maximum Nshifts allowed
for the SC contribution to meet the run conditions, which constrains the number of
effective pixels that can be read out per amplifier, decreases with shorter exposures.
Then, depending on the sensor’s size, performance and run conditions, a balance should
be made between DC and SC generation when choosing Nexp and the readout mode.
Also, Nexp is constrained by the readout rate, which totally depends on the electronics.
The time that takes to readout a whole skipper CCD detector where each amplifier
is read simultaneously is proportional to the number of effective pixels read out per
amplifier and the number of skipper samples taken per pixel. Assuming 1.35 Mpix
sensors and 1× 1 readout mode, a readout rate higher than 188 (32) pix/s is needed to
readout the whole detector array in less than 2 (24) hours, which corresponds to a pixel
readout time of 5.3 (31.2) ms if sensors are read out using a single amplifier.
The lowest R1e− achieved in skipper-CCD detectors, reported by SENSEI [12], is 1.6×
10−4 e−/pix/day. Given this rate and the assumptions we made in Table 4.2, less
than one accidental 3e− event is expected if we operate with 2-hour exposures, i.e.
Nexp = 12 exposures/day. However, for RSC,1e− to be consistent with the run conditions
and considering a 1× 1 readout mode using a single amplifier, PSC should be less than
5.7× 10−9 e−/pix/transfer, value that has not been achieved with skipper CCDs yet.
4.3.2. Misidentified events due to readout noise
The readout noise and the threshold used to determine if a pixel has ne− define the
number of (n − 1)e− single pixel events that fall above the threshold that are counted
as ne− events (misidentified events). In principle, a skipper CCD readout noise can be
made extremely small when multiple Nskp are collected, as it drops as 1/
√
Nskp [5].
However, adding skipper samples makes the readout slower which can be inconsistent
with the desired run conditions. An optimization between the readout noise and speed
should then be considered when choosing Nskp.
The total count of (n − 1)e− single pixel events counted as ne− events comes from
integrating the tail of the (n − 1)e− single pixel event normal distribution from the
threshold for counting ne− events. It is given by
Ln =
1
2
[
1− erf
(
eth/
√
2σnoise
)]
K(n−1), (4.10)
where K(n−1) is the total number of (n−1)e− single pixel events, see Eq. (4.9), erf is the
error function, σnoise is the electronic readout noise in units of electrons, and (n−1)+eth
is the threshold used to determine if a pixel has ne−. For example, for n = 2, eth = 0.5,
if the threshold is set to 1.5e−.
50 SKIPPER CCDS
As the noise increases, we need to increase eth to keep Ln < 1 for a given K(n−1), but
higher values of eth produce inefficiency for counting ne− single pixel events. In fact,
the efficiency (eff) is the integral of the ne− single pixel event normal distribution from
the given threshold, and can be calculated as
eff =
1
2
[
1 + erf
(
(1− eth)/
√
2σnoise
)]
. (4.11)
From Eq. (4.11) we see that eth = 1 corresponds to 50% efficiency, independent of the
value of σnoise. Fig. 4.5 shows the efficiency for counting ne− events as a function of
the readout noise after imposing the conditions to keep Ln < 1 for a given K(n−1). For
this efficiency to be higher than 80% we need σnoise below 0.18, 0.22 and 0.32 e− when
having 100, 10k and 1M K(n−1) events, respectively.
K(n-1) = 100
K(n-1) = 10k
K(n-1) = 1M
0.10 0.15 0.20 0.25 0.30 0.35
0.0
0.2
0.4
0.6
0.8
1.0
σnoise (e-)
E
ff
ic
ie
n
c
y
Figure 4.5. Efficiency for counting ne− single pixel events as a function of the electronic readout
noise when the threshold is set such that Ln < 1 for 100 (cyan), 10k (blue) and 1M (black)
K(n−1) events.
Chapter 5
CONNIE with skipper CCDs
In July 2021, CONNIE underwent its second upgrade. In an effort to reduce the de-
tection energy threshold and increase sensitivity, we installed two thick (∼675 µm),
fully-depleted silicon skipper CCDs, designed at the LBNL in collaboration with the
FNAL during the R&D phase of the SENSEI experiment [10]. Both sensors, fabricated
at Teledyne DALSA Semiconductor, consist of 1022×682 pixels of 15×15 µm2 and have
a p-type buried channel implant on a high-resistivity n-type substrate (∼18 kΩ-cm).
Each sensor and a new dedicated Kapton flex-cable, connected to it with micro-wire
bonds, were glued to a 7 cm × 7 cm Si substrate and packaged in the two-piece copper
trays used for the CONNIE standard CCDs, see Fig. 5.1 (left). The packages were in-
stalled in the two bottom slots of the detector’s inner copper box. The 14 standard CCDs
from the first CONNIE upgrade remained in their original locations, see Fig. 5.1 (center).
The skipper CCDs flex cables connect through new dedicated second-stage flex exten-
sions to the vacuum side of a new fabricated vacuum-interface board, see Fig. 5.1 (right).
This VIB enabled us to connect two new skipper CCDs and two old standard CCDs.
The air-side of the VIB connects four Low-Threshold-Acquisition (LTA) boards [140] via
long cables to the four CCDs. These LTA boards replace the old Monsoon acquisition
system. We kept the entire CONNIE passive radiation shield unchanged.
Figure 5.1. CONNIE: Packaged skipper CCD (left), Cu cold box with 14 standard CCDs and 2
skipper CCDs installed (center) and new VIB board on top of the Cu vacuum vessel (right).
51
52 CONNIE WITH SKIPPER CCDS
5.1. Data taking
After installing the skipper CCDs, we took data for debugging and commissioning from
mid-July 2021 to early November 2021, when the full passive shield was assembled.
Throughout this period, we operated for some time with either no shield or a partial
shield, the latter consisting of 30 cm of inner polyethylene and 5 cm of lead. Since
November 2021, we have been continuously taking science data with short interruptions
due to technical issues, on-site interventions and power cuts. During the entire science
data taking, we had three periods when the reactor was off: 1) from June 12 to July 25,
2022, 2) from November 10 to 13, 2022, and 3) from November 15 to 29, 2022.
We will refer to the newly installed skipper CCDs as ACDS-10 and ACDS-11 from now
on. ACDS-11 is located on top of ACDS-10 within the Cu box. Although we have two
usable standard CCDs within the system, we decided to disconnect them from their
assigned LTA boards due to difficulties synchronizing the readout sequences between
the standard and skipper CCDs. Then, we are using only the skipper CCDs and two
out of four LTA boards to collect data. In the final setup, these boards operate in a
leader-follower configuration, where ACDS-11(10) is connected to the leader (follower)
board.
Each skipper CCD has four output amplifiers (chIDs), one in each corner. We use only
two amplifiers to readout each sensor: ACDS-10(11) is read through chIDs 0 and 3 (1
and 2), as illustrated in Fig. 5.2 (left). Particle tracks are only seen in chID 0 (2) for
ACDS-10(11) because of issues in the sensors; see Fig. 5.2 (right).
Figure 5.2. Left) Diagram illustrating the readout mode of the newly installed skipper CCDs:
ACDS-10(11) is read out using amplifiers 0 and 3 (1 and 2). Right) Particle tracks in processed
images from the second ROFF period corresponding to chID 0 (2) for ACDS-10(11). We use
the SAOImage DS9 software [141] to visualize the images.
5.2. DATA PROCESSING 53
The data-taking cycle consists of two phases: cleaning and readout. During cleaning,
the procedure to reduce “surface DC” is performed, see Section 4.2.1, in which Vsub is set
to 0 V whereas the vertical clock voltages are set to 9 V inverting the sensor’s surface.
Then, after setting the voltages to their usual operation values, the charge in the active
area of the CCD is clocked and discarded so that the readout starts with a “clean” CCD.
We read out both sensors simultaneously in 1 × 1 mode with Nskp = 400 samples/pix,
taking ∼2.6 hours per raw image. The data-taking periods are divided into “runs”, which
are usually sets of ∼500 images sharing a common detector configuration.
5.2. Data processing
The raw images are bidimensional FITS images in which each Nskp consecutive pixels
in the same row correspond to the Nskp performed measurements of the same pixel.
We first process these images using a script developed by members of the SENSEI
collaboration [10] which constructs a bidimensional image assigning to each pixel the
average of the Nskp measurements (proc*.fits). These processed images are Nskp times
shorter in the x-axis, compared to the raw images, and the pixel values are in analog-
to-digital units (ADUs).
Over the processed images we compute the horizontal baseline in each row as the median
of the pixels in the overscan (OS) region. We also compute the vertical baseline in
each column as the median of the first 100 rows. The horizontal (vertical) baseline is
subtracted row by row (column by column). We use the pixel charge distribution of the
active area of the image, excluding a 5 pixels border, to compute a local gain per image
in ADU/e−. We fit the peaks corresponding to 0 and 1e− pixels with two gaussians.
The local gain corresponds to the distance between the means. Then, we compute the
readout noise per image, σ, by fitting, with a gaussian function, the 0e− pixels peak of
the OS region pixel charge distribution, calibrated with the local gain.
Using all the images in one run, we compute the pixel charge distribution and fit, with
gaussian functions, the first 100 consecutive peaks. We plot the fitted means, in analog-
to-digital units (ADUs), versus their expected values, in e−. We perform a linear fit
to this plot. The slope corresponds to the run global gain g in ADU/e−, which we use
to generate the calibrated images (cal*.fits). We evaluate the differential nonlinearity
(DNL) value of this calibration method for each ke− peak as
DNLk =
(
µk+1 − µk
g
− 1
)
× 100 , (5.1)
where µk is the fitted mean corresponding to the ke− peak. The histogram of the DNLk
values follows a gaussian distribution. The mean value corresponds to the DNL value of
the method, which we found to be less than 0.1%.
54 CONNIE WITH SKIPPER CCDS
5.2.1. Masking
The first mask per image generated in our processing chain is the serial register event
(SRE) mask. To build it, we use the baseline corrected images. We developed an
identification algorithm for SREs based on their geometry. We expect the charge of an
isolated SRE to be distributed along one single row, within a certain range of pixels,
with empty pixels in the rows above and below. The algorithm searches within a row
and a window of 10 consecutive pixels (main train) any pixel with charge above a given
threshold, set to 1.5 e−, calibrated with the local gain. When found, the pixels are
identified as belonging to a SRE candidate when there is no charge in the two auxiliary
trains running above and below the main train. After identifying pixels potentially
belonging to a SRE in the entire image, we flag the rows that contain, at least, two
identified pixels separated at most by two pixels.
The second mask per image generated in our processing chain is the hot row/column
(HRC) mask. To build it, we use the calibrated images. First, we mask in each image
all pixels with more than 3.52 e− and their nearest neighbours. Then, we flag all rows
(columns) in the images with more than 12 pixels with charge above 0.6 e−, as well as
their adjacent rows (columns).
Then we create two master masks per run: the master hot row/column (MHRC) mask
and the master hot pixel (MHP) mask. In the MHRC mask, we flag all the rows/columns
that were flagged in more than 5% of the individual HRC masks. In the MHP mask,
we flag pixels with more than 2.56 e− that appear in more than 10% of the calibrated
images in the run. If these pixels are adjacent in a column, we flag the entire column
and its adjacent ones. However, if they are isolated, we flag the pixel and a halo of 2
pixels around it.
Figure 5.3. Generated masks for one image from ACDS-11 during a ROFF period.
5.3. EVENT EXTRACTION 55
Finally, we compute the global (GB) mask per image as the sum of the SRE, HRC,
MHRC and MHP masks, see Fig. 5.3. For each image, we generate a pixel exposure
map where the time assigned to the nth read pixel is tro(n − 1)/(Npix − 1), where tro
is the image readout time and Npix is the total number of pixels in the image. We
compute the effective exposure time per image as the mean value of the exposure times
of all unmasked pixels after applying the GB mask. Note that the masking algorithms
were developed using only ROFF data.
5.3. Event extraction
Once we generate the calibrated images, the next step is to extract events, i.e. pixel
clusters associated to energy depositions, and to create the event catalogs per image.
The event extraction algorithm first searches for a “seed” pixel with charge above 3.52 e−,
corresponding to 13.12 eV or 22σ. When found, a new event ID is created in the catalog.
From this “seed” pixel, the algorithm searches, within the nearest neighbours, for pixels
with more than 0.64 e−, corresponding to 2.4 eV or 4σ, and adds them to the event.
The algorithm continues repeatedly searching, from the newly added pixels, for pixels
satisfying these conditions and adding them to the event until the conditions are no
longer met, finishing the first level of reconstruction. To avoid losing any information
about the charge deposited by a particle in the sensors, we add, to the first-level cluster,
its adjacent pixels, finishing the second level of reconstruction. Pixels that are part of
an event are not considered anymore when searching for a new “seed” pixel.
The extraction algorithm also computes some parameters regarding each event in the
image. One of these is the event size, which is computed for events with energy below
5000 e−. As we fit each event with a 2D gaussian function, the event size information is
embedded in the σx and σy parameters. Another important variable is the barycenter,
whose position (xbary, ybary) is computed as
xbary =
∑
i xiEi
∑
iEi
and ybary =
∑
i yiEi
∑
iEi
, (5.2)
where (xi, yi) is the position of the ith pixel in the event and Ei its energy. We use the
GB masks to flag the events, in each event catalog, whose barycenter corresponds to
a masked pixel. The flagged events are removed from the background spectrum when
applying the GB mask cut.
5.4. Data analysis and selection cuts
We perform a hidden-data analysis, without looking at the RON data, to establish data
selection cuts. To remove edge effects, we exclude events whose barycenter lies within a
56 CONNIE WITH SKIPPER CCDS
10 pixels border around the active area. In images from ACDS-11, due to issues in the
sensor, we also exclude the region x<135 and y>900. These are our geometrical cuts.
5.4.1. Noise and SER stability
We looked at the noise and SER stability in each image. The noise per image is com-
puted during the data processing using the baseline corrected images, see Section 5.2.
The single-electron rate (SER) per image is computed using the same images and their
corresponding SRE masks. For its computation and, in addition to applying the SRE
mask, we exclude, from the active area, a 5 pixels border, pixels with more than 5 e−,
calibrated with the local gain, and a halo of 10 pixels around each of them. We fit the
charge distribution of the remaining pixels with a set of gaussian functions convoluted
with a Poisson distribution, as we expect the sources contributing to the SER to follow
this distribution, i.e.
f(x) = A
∞
∑
k=0
e−λλk
k!
1
σ
√
2π
e−
(x−kg)2
2σ2 , (5.3)
where k corresponds to the number of electrons, A is a scaling constant, σ is the readout
noise in ADU, g is the global gain and λ is the mean charge per pixel in e−/pix. The
local SER corresponds to the fitted λ divided by the exposure time; note that we use the
term “single-electron rate” because, in the pixel distribution after masking, we mostly
have pixels with 0 and 1e−. The exposure time is computed as the mean value of the
exposure times of all unmasked pixels from the pixel exposure map, after applying the
masks used in the SER computation.
Figure 5.4. Noise and SER distributions of the images from the ROFF period. We show the
mean values per run (dashed) and the established performance cut (solid blue).
From the noise and SER distributions of the images from the ROFF period, shown in
Fig. 5.4, we established the performance cuts: exclude images with noise above 0.164 e−
5.4. DATA ANALYSIS AND SELECTION CUTS 57
and SER above 0.1 e−/pix/day. Note that the performance, geometrical and GB mask
cuts only affect the effective data exposure.
5.4.2. Detection efficiency
We evaluate the overall detection efficiency using the same methodology as in previous
analyses [7,9]. We simulate 100 neutrino-like events per image with a uniform probability
in the active volume, using the depth-size calibration based on muon tracks, see Eq. (3.1),
and a uniform energy distribution, between 5 and 2005 eV, over images from the ROFF
period. By comparing the energy and 3D position of the simulated events before and
after the image processing and event extraction, we found out that almost all of them
were extracted with ∼100% acceptance. By looking at the σx and σy distributions of
the simulated events, we established the lower limit for the size cut, i.e. 0.2 < σx,y. We
kept the upper limit as in the analysis of the CONNIE 2019 data [9], i.e. σx,y < 0.95,
to avoid events from the PCC layer. The overall detection efficiency, shown in Fig. 5.5,
accounts for the event extraction acceptance and the size cut.
Figure 5.5. CONNIE with skipper CCDs overall detection efficiency (red), accounting for the
event extraction acceptance and the size cut. The CONNIE with standard CCDs 1×1 efficiency
is shown for reference.
From Fig. 5.5 we can see that skipper CCDs allow us to reduce the energy threshold
to ∼15 eV and increase the detection efficiency at low energies, increasing CONNIE’s
sensitivity to low-energy neutrino interactions.
5.4.3. Impact of cuts on ROFF background spectrum
We computed the ROFF background spectrum, whose evolution when different cuts are
added is shown in Fig. 5.6. We see that, after applying the GB mask cut, the spectrum
gets flat towards low energies. Also, this is the cut that has more impact in the effective
data exposure. At ∼1.8 keV we can identify the silicon X-ray fluorescence peak.
58 CONNIE WITH SKIPPER CCDS
Figure 5.6. CONNIE with skipper CCDs ROFF spectrum including all events (light gray) plus
the performance cuts (light purple) plus the geometrical cuts (light green) plus the GB mask
cut (red) plus the size cut (blue).
We fix all the processing and analysis parameters before processing the RON data.
The ROFF data passing all cuts correspond to an effective exposure of 3.21 g-days.
Accounting for this exposure, the ROFF background spectrum, shown in Fig. 5.6, shows
a mean level of ∼3.2 kDRU, consistent with the analyses in Refs. [6,7,9]. The CONNIE
collaboration will report the results after the analysis of the RON data in a future
publication.
5.5. CEνNS forecasted sensitivity
Using the CONNIE with skipper CCDs overall detection efficiency, shown in Fig. 5.5, we
computed the expected CEνNS event rates considering different nuclear recoil quenching
factors; see Fig. 5.7. Between 20 and 120 eV, we expect 8.6, 8.1, 6.2 and 3.4 events/kg-day
from CEνNS considering Lindhard [112], Albakry [116], Sarkis2022 [119] and Chavar-
ria [113] quenching factors, respectively.
The confidence level of a CEνNS signal measurement as a function of the effective data
exposure ϵ is the integral of a gaussian distribution with σ ≃ √
Nb +Ns in the range
[−Ns, Ns], where Nb is the number of background events and Ns is the number of
expected CEνNS events for a given ϵ, i.e.
C.L.(ϵ) = erf
(
Ns
√
2(Nb +Ns)
)
= erf


ϵRs|E2
E1
√
2ϵ(Rb +Rs)|E2
E1

 . (5.4)
5.5. CEνNS FORECASTED SENSITIVITY 59
●
●
●
●
●
● ● ● ● ●
●
●
●
● ● ●
●
●
●
●
●
● Lindhard
● Chavarria
● Sarkis2022
● Albakry
0.2 0.4 0.6 0.8 1.0
0
20
40
60
80
E (keVee)
d
R
/d
E
(e
v
e
n
ts
/k
g
/d
a
y
/k
e
V
)
Figure 5.7. Expected CEνNS event rates from 20 to 1020 eV, assuming CONNIE with skipper
CCDs overall detection efficiency and different quenching factors: the Lindhard model [112]
(blue), the expression in Eq. (3.7) fitting the Chavarria measurements [113] (orange), the
Sarkis2022 model [119] (green) and an extrapolation to the Albakry measurements [116] (red).
Assuming ∼4 kDRU of background, E1 = 20 eV, E2 = 120 eV and the expected CEνNS
event rates in Fig. 5.7, we computed the C.L. of a CEνNS signal measurement in CON-
NIE as a function of the effective data exposure ϵ, shown in Fig. 5.8. From this figure we
see that the required exposures to observe CEνNS at a 90% C.L. are 7.5, 8.3, 14.3 and
46.3 kg-day considering Lindhard, Albakry, Sarkis2022 and Chavarria quenching factors,
respectively. If we could increase the mass in CONNIE to ∼1 kg, we could overcome
the background rate we observe and search for a CEνNS signal with a high level of
confidence (>90%) running for ∼2 months, for the less favorable quenching factor.
Lindhard
Chavarria
Sarkis2022
Albakry
0 5 10 15 20 25 30
0.0
0.2
0.4
0.6
0.8
1.0
Exposure (kg-day)
C
.L
.
Figure 5.8. Confidence level of a CEνNS signal measurement in CONNIE as a function of the
effective data exposure ϵ, assuming the same quenching factors as in Fig. 5.7.
60 CONNIE WITH SKIPPER CCDS
Chapter 6
Synergy with dark sector searches
6.1. Direct dark matter searches
The existence of Dark Matter (DM) is strongly motivated by numerous astronomical and
cosmological observations that cannot be explained with the known laws of gravity taking
into account only the visible mass. These observations cover a wide range of astrophysical
and cosmological scales, from galactic rotation curves [142] to the anisotropy of the
Cosmic Microwave Background (CMB) [143,144]. Despite efforts to describe the effects
attributed to DM with modified theories of gravity, these theories do not work at all
scales [145], which discourages this approach. Unveiling the nature of DM is critical
to understanding the formation and evolution of our Universe and is currently a top
priority within the physics community.
The mass of DM candidates which could explain the astrophysical observations lay
within a broad range, from ∼10−22 eV, which is the lightest mass consistent with galactic
structure, to several solar masses. However, DM is best described as a particle when its
mass lies within ∼eV and the Planck scale (∼1019 GeV) [146]. The Standard Model of
particle physics does not contain a DM candidate, pushing the need to explore beyond
the SM scenarios.
Figure 6.1. Adapted from Ref. [147]. Representative mass ranges for dark matter and mediator
particle candidates.
61
62 SYNERGY WITH DARK SECTOR SEARCHES
During the last decades, a strong experimental effort was dedicated to the search for
“heavy” DM, with masses between ∼GeV to ∼TeV. Weakly Interacting Massive Particles
(WIMPs) have been one of the most theoretically motivated DM candidates to search
for in this mass range, as they naturally arise in SM extensions that aim to solve the
electroweak hierarchy problem [148]. Also, the relic density for WIMPs in thermal equi-
librium in the early Universe undergoing freeze-out matches the observed cosmological
dark matter abundance, according to the standard cosmological model (ΛCDM).
However, different experiments have excluded a vast region of the WIMP parameter
space, motivating the search for other DM candidates in a different mass range. Of
particular interest is sub-GeV DM, consistent with the hypothesis that DM could be
one or more particles from a “dark sector”, neutral under the SM forces, that feebly
interact with SM particles through “portals” associated to specific mediators. In these
scenarios, DM could undergo different production mechanisms to match the observed
cosmological DM abundance [147].
Direct detection experiments are crucial to identify the nature of DM as they test sce-
narios in which DM interacts with SM particles. Traditional searches focus on looking at
the elastic scattering of a DM particle off a target nucleus. Ton-scale detectors lead this
search at low cross-sections and DM masses above ∼10 GeV (LZ [149], XENONnT [150],
PandaX-4T [151]). Future experiments aim to explore lower cross-sections, where the
CEνNS of solar and atmospheric neutrinos becomes a significant background [152]. In
the low-mass regime, below ∼10 GeV, small-scale low-threshold detectors are more sen-
sitive (CRESST-III [153], EDELWEISS [154], SuperCDMS [155], DAMIC [120], NEWS-
G [156]). However, the best limits below ∼1 GeV correspond to high cross-sections and
are nonexistent below ∼90 MeV.
Sensitivity to sub-GeV DM could be achieved in direct searches by exploring inelas-
tic processes such as DM absorption or DM scattering off bound electrons, individual
nuclei accompanied by a bremsstrahlung photon or a Migdal electron, or a condensed-
matter system. Signals from these processes consist of one to a few electrons, photons
or collective excitations [157]. Different technologies with single-electron and single-
photon resolution have been demonstrated, e.g. dual-phase time projection chambers
(TPCs) [158], skipper CCDs [5], transition edge sensors (TESs) [159] with cryogenic
targets and superconducting nanowire single-photon detectors (SNSPDs) [160]. The in-
trinsic characteristics of these technologies make them sensitive to different mass ranges.
For example, the minimum energy needed to free the least-bound electron from a nu-
cleus in noble gases, e.g. Xe, is ∼10 eV, in semiconductors, e.g. Si and Ge, is ∼1 eV
and in low-gap materials, e.g. superconductors, is ∼meV, allowing experiments using
these technologies to be sensitive to DM masses of ∼10 MeV, ∼500 keV and ∼1 keV,
respectively, if considering DM-electron scattering.
6.1. DIRECT DARK MATTER SEARCHES 63
6.1.1. Skipper CCD experiments at underground laboratories
The skipper CCD technology has demonstrated to be highly competitive in searching
for sub-GeV DM, leading to the current state-of-the-art limits in several DM-electron
interactions for masses below ∼5 MeV [12, 14]. Several active and planned experiments
form part of the ongoing effort to search for sub-GeV DM with skipper CCDs, aiming to
increase their sensitivities through understanding and reducing their low-energy back-
grounds and having larger detector masses. Here, we give an overview of the ongoing
and planned skipper CCD experiments for DM detection.
SENSEI
The Sub-Electron-Noise Skipper ccd Experimental Instrument (SENSEI) is the pioneer
experiment using skipper CCDs to search for sub-GeV DM interactions with electrons.
SENSEI has a shallow-underground (∼104 m) setup located at the MINOS cavern at
the FNAL. It consists of a copper vacuum vessel that hosts one copper tray housing the
sensors. This tray connects through a cold finger to a cryocooler to bring the sensors
to a temperature of ∼135 K. A LTA board [140] is used to control and readout the
skipper CCDs. Inside the vessel, the sensors are surrounded by a 1-to-3-inch lead shield.
Fig. 6.2 (left) shows pictures of the setup. Results from an exposure of 0.069 g-days using
a prototype skipper CCD led to world-leading constraints on DM-electron interactions,
see Ref. [11] for details.
Figure 6.2. Taken from Ref. [12]. Left) Photographs of the SENSEI at MINOS setup from
outside and inside. Right) Single-electron rate dependence on high-energy background rate.
The measurement without external shield and amplifier off (on) during exposure is shown in
green (black), and the one with external shield is shown in red.
A new science-grade skipper CCD, with ∼2 g of mass, was installed in the same setup
in 2019. This 675 µm-thick sensor, fabricated at Teledyne DALSA Semiconductor, has
6144×886 pixels and four amplifiers, one in each corner. SENSEI collected 24 days of
64 SYNERGY WITH DARK SECTOR SEARCHES
science data and reported the lowest rates in silicon detectors of events containing 1 to 4
electrons, allowing them to place the best limits on DM-electron scattering via a heavy
(light) mediator, DM-nucleus scattering through a light mediator and DM absorption on
electrons for DM masses between 500 keV to 10 MeV (above 500 keV), 600 keV to 5 MeV
and 1.2 eV to 12.8 eV, respectively [12]. In their setup at MINOS and with an external
2-inch lead shield, SENSEI reported a background rate of ∼3400 DRU between 500 eV to
10 keV, and a exposure-dependent single-electron rate of ∼1.6×10−4 e−/pix/day. They
show that external background radiation generates single-electron events, see Fig. 6.2
(right). SENSEI also did a detailed characterization of single-electron events using the
science-grade skipper CCD at MINOS, identifying 3 independent contributions: dark
current, amplifier light and spurious charge [161]. Recently, using the science data
collected at MINOS, SENSEI put world-leading exclusion limits on the parameter space
of millicharged particles with masses between 30 to 380 MeV [13].
The SENSEI collaboration has deployed a low-radiation setup capable of holding ∼50
skipper CCDs (∼100 g) at SNOLAB in Canada, a ∼2 km underground laboratory, where
a lower single-electron rate is expected. On 2022, SENSEI at SNOLAB underwent a
partial commissioning and started taking data.
DAMIC-M
The DArk Matter In Ccds at Modane (DAMIC-M) experiment will use thick silicon
skipper CCDs to search for sub-GeV DM particles under the French Alps (∼1.7 km
underground) at the Laboratoire Souterrain de Modane (LSM) in France. The complete
experiment will feature ∼700 g of target mass and an expected external background rate
of ∼0.1 DRU.
Figure 6.3. Taken from Ref. [14]. Photographs of the Low Background Chamber installed at
the Laboratoire Souterrain de Modane (left) and of the copper box housing two DAMIC-M
prototype skipper CCDs.
6.1. DIRECT DARK MATTER SEARCHES 65
DAMIC-M is currently in the development phase and has installed at LSM a prototype
detector, the Low Background Chamber (LBC). In this detector, two thick (670 µm)
skipper CCDs with 6144×4128 pixels, fabricated at Teledyne DALSA Semiconductor,
were installed reaching a total target mass of ∼18 g. The sensors are mounted in a high-
purity, oxygen-free copper box and placed inside a copper vacuum cryostat. The LBC
has a 7.5 cm inner lead shield surrounding the copper box and an external shield of 15 cm
of lead and 20 cm of high-density polyethylene surrounding the cryostat. A commercial
CCD controller is used to operate the skipper CCDs. DAMIC-M reported a background
level of ∼10 DRU and a dark current of ∼4.5×10−3 e−/pix/day. With a total exposure
of 85.23 g-days, DAMIC-M put exclusion limits on DM-electron scattering via a heavy
(light) mediator improving the SENSEI limits for masses above ∼1.5 MeV.
DAMIC at SNOLAB
Installed in 2012 at SNOLAB, Canada, DAMIC was the first experiment to use thick
CCDs to study particle physics, specifically to directly search for DM. DAMIC standard
CCDs are analogous to CONNIE CCDs, described in Section 3.1, and its detector design
is also very similar to CONNIE detector, see Fig. 6.4 (left). DAMIC published exten-
sive studies on measurements [162, 163] and characterization [121] of its background.
Also, it put limits on DM-nucleus [106,120] and DM-electron interactions [164,165]. In
2020, DAMIC reported a statistically significant (∼3.7σ) excess of bulk events above the
background model between 50 and 200 eV [120].
Figure 6.4. Taken from Ref. [166]. Left) Photographs of the DAMIC at SNOLAB detector with
two DAMIC-M skipper CCDs installed. Right) Comparison of the parameters of the observed
excess with standard and skipper CCDs.
In 2021, DAMIC decommissioned its standard CCDs and installed two DAMIC-M skip-
per CCDs, as the ones installed by DAMIC-M in the LBC, to explore the excess of
events previously reported; see Fig. 6.4 (left). A LTA board [140] was used to control
and readout the sensors. The dark count rate achieved was ∼2.5 × 10−3 e−/pix/day.
66 SYNERGY WITH DARK SECTOR SEARCHES
The collected data, corresponding to a 3.1 kg-days exposure, also show an excess of bulk
events above the background model at energies below 200 eV [167], statistically compati-
ble with the previous result. Instrumental events are discarded as a possible explanation
as the excess was also observed when the standard DAMIC CCDs were operating. The
excess can be modeled by a decaying exponential with decay length ϵ = 89 eV; see
Fig. 6.4 (right).
Oscura
Oscura is a planned multi-kg skipper CCD experiment for direct DM search [15]. It aims
to collect a 30 kg-year exposure with less than one background event in each electron
bin in the 2–10 electron ionization-signal region using a low-background 10 kg detector.
To achieve this background, a radiation background below 0.025 DRU is needed, as well
as an instrumental single-electron event rate below 1× 10−6 e−/pix/day.
1 10 102 103
10-43
10-42
10-41
10-40
10-39
10-38
10-37
10-36
10-35
10-34
10-33
10-32
10-31
10-30
10-29
10-28
10-27
mχ [MeV]
σ e[c
m
2
]
XENON10
X
E
N
O
N
1
0
0
D
ark
S
id
e5
0
C
D
M
S
-
H
V
eV
D
A
M
IC
@SN
O
LA
B
protoSE
NSEI@S
urface
protoSE
NSEI@M
INOS
X
E
N
O
N
1t
KeyMilestone
FDM=1
SENSE
I
DAMI
C-M
Oscura
SENSE
I@MIN
OS
DAMI
C-M
100 101 102 103 104
10-43
10-42
10-41
10-40
10-39
10-38
10-37
10-36
10-35
10-34
10-33
10-32
10-31
10-30
10-29
10-28
mχ [MeV]
σ e[c
m
2
] XENON10
XEN
ON1
00
XEN
ON
1tDark
Side
50
DAMIC-SNOLAB
proto
SEN
SEI@
Surfa
ce
CDM
S-H
VeV
proto
SEN
SEI@
MIN
OS
FDM=(αme/q)
2
Key Milestone
SEN
SEI@
MIN
OS
DAM
IC-M
SEN
SEI
DAM
IC-M
Oscu
ra
PA
N
D
A
X
-
II
Figure 6.5. Taken from Ref. [16]. Approximate projected sensitivity for Oscura to DM-electron
scattering at 90% C.L. assuming a 30 kg-year exposure, zero background events with 2e− or
more, a 1e− threshold and a fixed 1e− event rate of 10−6e−/pix/day (blue). The left (right)
plot assumes a heavy (light) mediator in the DM-electron interaction. Approximate projected
sensitivities for SENSEI with 100 g (DAMIC-M with 1 kg) are shown in cyan (red). Existing
constraints from skipper CCD experiments [10–12,14] are shaded in pink. Shaded gray regions
represent existing limits [165,168–173]. Orange regions labeled “Key Milestone” represent well-
motivated sub-GeV DM models [174].
Oscura will reach unprecedented sensitivity to sub-GeV DM-electron interactions. As
an example, we show in Fig. 6.5 the approximate projected sensitivity for Oscura to
DM-electron scattering through a “heavy” or “ultralight” mediator [175–178], particu-
6.1. DIRECT DARK MATTER SEARCHES 67
larly probing DM masses in the range of 500 keV to 1 GeV. For these projections, we
assume the QeDark cross section calculation for DM-electron scattering [178] and the
astrophysical parameters considered in Ref. [12].
As the first multi-kg experiment, Oscura has conducted a major R&D to address the
requirements for scaling the mass of skipper CCD experiments. This includes ensuring a
mass production of science-grade skipper CCDs [16,19], devising effective packaging and
cooling methods for a large number of sensors [15], developing multiplexing and scalable
electronics [179, 180], among others. Oscura is currently in its design stage. We plan
to complete an engineering test, the “Oscura Integration Test”, with a 1 kg detector by
2026 and to start operations with the whole 10 kg array by 2028.
The Oscura detector design is based on 1.35 Mpix sensors packaged on a Multi-Chip-
Module (MCM), see Fig. 6.6 (left). Each MCM consists of 16 sensors epoxied to a
150 mm diameter silicon wafer, with traces connecting the sensors to a low-radiation
background flex cable [181, 182]. MCMs will be integrated into Super Modules, where
each of them will hold 16 MCMs using a support and shielding structure of custom
ultrapure electro-deposited copper [183], see Fig. 6.6 (center). The Oscura experiment
needs ∼80 Super Modules to reach 10 kg of active mass. The full detector payload
consists of 96 of them, assuming a yield above 80%, surrounded by an internal copper
and lead shield, arranged in six columnar slices forming a cylinder, see Fig. 6.6 (right).
Figure 6.6. Taken from Refs. [15,16]. Left) A fully assembled Si-MCM in a copper tray. Center)
Oscura Super Module design with 16 MCMs supported and shielded with electroformed copper.
Right) Model showing one of the columnar segments with 16 Super Modules each and the full
assembly of all six segments to form the full cylindrical Oscura detector payload.
The current strategy for the cooling system is to submerge the full detector array in a
Liquid Nitrogen (LN2) bath operated with a vapor pressure of 450 psi to reach skipper
CCDs operating temperatures (between 120 to 140 K). Closed-cycle cryocoolers will
provide the full system cooling capacity. A schematic of the pressure vessel and its
radiation shield is shown in Fig. 6.7.
68 SYNERGY WITH DARK SECTOR SEARCHES
Figure 6.7. Taken from Ref. [16]. Left) Design of the Oscura pressure vessel for the operation
of the 26 gigapixel skipper CCD detector array. Right) Cross section of the Oscura vacuum
vessel showing the internal lead and copper shield (dark/light pink), the external high-density
polyethylene shield (dark/light blue), and the region filled with LN2 (green).
6.2. Search for millicharged particles with skipper CCDs
Experimentally, the electric charge of all known elementary particles in the SM is an
integer multiple of 1/3 e, where e is the electron electric charge. Hence, charge seems
to be quantized. We call millicharged particles (mCPs) to those with a small electric
charge ε, on the order of a fraction of e. These particles can arise naturally in many
SM extensions [184]; see for example Section 1.4.3 about the discussion of the neutrino
millicharge. Also, in the context of dark sector particles, a new fermion, charged under
a new gauge boson Z ′ that kinetically mixes with the SM photon γ, can become mCP
if Z ′ is massless [185], i.e. if Z ′ is the “dark photon” γ∗.
Figure 6.8. Taken from Ref. [13]. Diagram of mCPs production in a proton beam via a photon-
mediated meson decay.
It has been recently explored the capability of a skipper CCD experiment to search
for mCPs produced in photon-mediated decays of mesons (see Fig. 6.8) and Drell-Yan
processes occurring in high-energy proton collisions with a target. In fact, the SENSEI
experiment at MINOS, where the Neutrinos at the Main Injector (NuMI) beam passes
through, has recently published world-leading exclusion limits on the parameter space
6.2. SEARCH FOR MILLICHARGED PARTICLES WITH SKIPPER CCDS 69
of millicharged particles with masses between 30 to 380 MeV [13]. These results have
motivated the development of future skipper CCD experiments specifically designed to
enhance the search for mCPs.
Derived from the ideas to produce early science with the Oscura experiment, we explored
the capability of the Oscura Integration Test (OIT) detector to search for millicharged
particles from proton beams. First, we consider as a source the NuMI beam at MINOS
and as a possible detector site the MINOS near-detector hall, located ∼1 km away from
the target [18]. In this beam, around 1018 120 GeV protons strike a graphite target per
day. If mCPs are produced, we expect them to be highly boosted (∼GeV), collinear
with the proton beam and to have a uniform flux on the detector. The OIT detector
payload is expected to be one of the six columnar slices of the whole Oscura detector,
containing 16 Super Modules. One slice corresponds to a ∼2 kg detector but we will
assume a 50% yield. If this column is vertically placed, each mCP will traverse two
Super Modules, see Fig. 6.9 (left). Hence, the OIT detector would work as a 32-layer
silicon tracker for mCPs, each layer being ∆z = 725 µm thick. The mCPs are expected
to interact electromagnetically, generating ionization signals.
In the work in Ref. [18], we considered three different approaches to search for mCPs:
looking for single, double and triple hits. The probability of having at least k hits in a
N -layer tracker is calculated as,
ξ(k|µ) = 1−
k−1
∑
m=0
µme−µ/m! , (6.1)
where µ = NP is the expected number of hits in the tracker and P is the probability
of having a charge deposition in a single pixel. For a mCP, P = ∆z/λ, where λ is the
mean free path of the particle, which depends on the transferred energy to the electron
and the millicharge ε via the interaction cross section. With the expression in Eq. (6.1),
we can also compute the probability of having k hits from backgrounds taking P as the
background rate of ne− single pixel events, i.e. P ≡ Rne− . We can use this probability
to compute the number of background tracks with k hits with ne− in the tracker as
Ntracks = Npix ξ(k|µ), with Npix the number of pixels in the detector. For example, for
Rne− = 1 × 10−4 e−/pix/day, we expect 441 (0.47) background tracks with 2 (3) ne−
hits in the 32-layer tracker of the OIT detector. Therefore, the approach of looking at
multiple hits could help us reduce backgrounds.
We computed the projected sensitivities at a 90% C.L. for a 1 kg-year exposure collected
with the OIT detector assuming 2 × 1018 protons-on-target per day, see Fig. 6.9 (left).
For mCP masses below 220 MeV, the 1-hit strategy provides the strongest constraints.
Because of the dependence of λ on ε, higher values of ε result in lower values of λ and,
consequently, the probability of charge deposition increases. This, together with the
background reduction, makes the multiple-hit strategy more advantageous for higher
masses. A combination of these approaches should be made in order to get the most out
70 SYNERGY WITH DARK SECTOR SEARCHES
of the 32-layer tracker. These limits are significantly better than existing limits and, for
masses below ∼300 MeV, surpass the projected limits for planned experiments searching
for beam-produced mCPs, such as FORMOSA [186] and FerMINI [187].
Figure 6.9. Taken from Ref. [18]. Left) mCP track crossing the OIT detector. Right) Projected
sensitivities at a 90% C.L. when searching for single, double and triple hits, assuming a back-
ground similar to that of SENSEI at MINOS (solid) and no background (dashed). Previous
limits are enclosed in the gray line, which does not account for the recent SENSEI’s limit.
We can think of implementing a skipper CCD experimental program to explore a broader
region in the parameter space of mCPs by using different sources. For example, to
explore the high-mass region, we could consider installing a skipper CCD detector in the
proposed Forward Physics Facility (FPF) at the High-Luminosity Large Hadron Collider
(HL-LHC) [188], where mCPs with masses below ∼100 GeV could be produced from the
∼TeV proton-proton collisions. Also, to explore lower mCP masses, we could use as mCP
sources the PIP-II linear accelerator at the FNAL [189], in which accelerated 800 MeV
protons colliding with a target could produce mCPs with masses below ∼100 MeV, or
nuclear reactors, where mCPs with keV masses could be produced through Compton-like
processes [190] (see Fig. 6.10).
Figure 6.10. Taken from Ref. [190]. Diagram of mCPs production in a nuclear reactor through
Compton-like processes.
Chapter 7
Future large skipper CCD experiment
Nowadays, g-size skipper CCD detectors in which tens of sensors and readout channels
are involved can be built with the current technology. Increasing the detector mass
imposes new requirements on sensors, design, cryogenics and electronics. The Oscura
experiment has conducted a major R&D in these areas to make kg-size skipper CCD
detectors a reality, ensuring a sensor mass production in new foundries. Here, we discuss
the characterization of the first batch of skipper CCDs fabricated in new foundries and
the implementation of the largest skipper CCD instrument with scalable electronics.
7.1. Characterization of skipper CCDs from new foundries
Before Oscura, skipper CCDs were fabricated at a 150 mm diameter wafer foundry (Tele-
dyne DALSA Semiconductor) that is in the process of discontinuing the CCD processing
line. The development of large-scale CCD fabrication techniques in partnership with new
foundries is essential to build kg-size skipper CCD detectors. Oscura has addressed this
technical risk developing a skipper CCD fabrication process on 200 mm diameter wafers
with a new industrial partner (Microchip Technology Inc.) and also with a government
laboratory (MIT-LL).
The overall design of the Oscura prototype sensors resembles that of the skipper CCDs
used in the SENSEI and DAMIC-M experiments. Oscura skipper CCDs are small for-
mat thick sensors (∼725 µm), with 1278×1058 pixels, and four skipper CCD amplifiers,
one in each corner. The new three-phase skipper CCDs were made using high-resistivity
silicon wafers as a starting material, which is similar to the material used for the sen-
sors’ fabrication in previous DM skipper CCD experiments. In the first iteration, the
foundries produced sensors with two different p-type buried-channel ion implantation
doses, namely the standard and the low dose. The standard dose corresponds to an in-
tegrated doping density of 1.3× 1012 atoms/cm2 in the fabrication from both foundries,
while the low dose corresponds to 1 (0.9)× 1012 atoms/cm2 in the Microchip (MIT-LL)
fabrication.
In this section, we present the performance of the first fabricated skipper CCDs for
Oscura. The characterization tests discussed here were done using individual Oscura
prototype skipper CCDs packaged in copper trays and installed in dedicated testing
71
72 FUTURE LARGE SKIPPER CCD EXPERIMENT
setups at the Silicon Detector Facility, at the FNAL. Using a cryocooler, the sensors were
operated at around 150 K. A LTA board [140] was used for reading out the collected
charge. Fig. 7.1 shows pictures of an Oscura prototype skipper CCD (left) and a 200 mm
diameter wafer with ∼50 Oscura sensors (right), both from the Microchip fabrication.
Figure 7.1. Left) Skipper CCD fabricated for Oscura at Microchip in 2021 (design from S.
Holland - LBNL). The upper structures in the picture are for performing tests. Right) 200 mm
diameter wafer with ∼50 skipper CCDs fabricated for Oscura at Microchip.
After a first optimization of the potentials needed to effectively move the charge towards
the sense node we performed multiple characterization tests on the first Oscura prototype
skipper CCDs fabricated in the new foundries. Here, we discuss them and present the
results.
7.1.1. Readout noise and speed
We measured the readout noise as a function of Nskp using an optimized set of voltages
that allows to have single-electron resolution and reduces the instrumental low-energy
events. We took images with zero exposure time and ∼50 rows, changing Nskp. First,
we fitted the 0 and 1e− peaks of the charge distribution of the active area of an image
with large Nskp with two gaussians. We computed the gain, in ADU/e−, as the dis-
tance between the means. Then, for each image we computed the readout noise. We
fitted, with a gaussian function, the charge distribution of the active area of the image
surrounding the expected 0e− pixels peak. The readout noise, in e−RMS, corresponds
to the standard deviation divided by the gain.
The results, shown in Fig. 7.2, demonstrate that the readout noise of the sensors from
7.1. CHARACTERIZATION OF SKIPPER CCDS FROM NEW FOUNDRIES 73
both foundries follow the expected 1/
√
Nskp dependence. Moreover, Nskp = 400 (225)
are enough to reach a noise of 0.15 (0.19) e−. In these measurements, the pixel readout
time for Nskp = 400 (225) was 15.3 (9) ms. This corresponds to a pixel readout rate of
65 (111) pix/s, allowing to read out a whole sensor using one amplifier in 5.8 (3.4) hours.
skp
N1 10 210
3
10
 R
M
S
)
-
N
o
is
e 
(e
1−10
1
Energy (e-)
0 1 2 3 4
N
u
m
b
er
 o
f 
p
ix
el
s
0
0.5
1
1.5
2
2.5
3
3
10×
Figure 7.2. Left) Oscura prototype skipper CCD readout noise as a function of Nskp. The
expected 1/
√
Nskp dependence is shown in red. Right) Charge pixel distribution from image
with Nskp = 1225 samples/pix; this result demonstrates the electron-counting capability of the
first Oscura prototype skipper CCDs fabricated in new foundries, see Ref. [19].
7.1.2. Exposure-dependent single-electron rate
To compute an upper-limit on the Oscura prototype skipper CCDs dark current (DC),
we measured the exposure-dependent 1e− rate as a function of temperature in a ded-
icated setup with 2 in of lead shield at surface. At a given temperature, we acquired
images with different exposure times, from 0 to 30 min, with Nskp = 200. To increase
statistics, multiple images with the same exposure time were acquired. To decrease the
image readout time and the resulting probability of having high-energy depositions and
exposure non-uniformity, we did a 5× 1 binning, i.e., the charge of 5 consecutive pixels
in the same row was summed before readout.
For the analysis, we select the first rows as they are mostly free of high-energy events.
After masking some unwanted features in the images such as SREs, high-energy pixels
and borders, we compute the charge distribution of the unmasked pixels in the active
area per set of images with the same exposure time. We fit each of these distributions
with a set of gaussian functions convoluted with a Poisson function. The Poisson’s
lambda parameters correspond to the SERs (R1e−). We plot the SERs, in e−/pix,
74 FUTURE LARGE SKIPPER CCD EXPERIMENT
versus exposure time per given temperature. We perform linear fits to these plots,
where the slopes correspond to the exposure-dependent 1e− event rate. Fig. 7.3 (left)
shows one of these plots corresponding to images taken at T = 150 K. We performed
these measurements at different temperatures; the results are shown in Fig. 7.3 (right).
The lowest value achieved was 0.03 e−/pix/day, at 140 K.
Exposure (days)
0 2 4 6 8 10 12 14 16
3−
10×
 (
e-
/p
ix
)
-
1
e
R
0.8
1
1.2
1.4
1.6
1.8
2
3−
10×
 / ndf 2χ  0.3287 / 2
p0       05− 5.725e± 0.0008304 
p1        0.01474± 0.0545 
Expected
Model fit at 160K
Data
140 145 150 155 160
0.001
0.010
0.100
1
T [K]
E
x
p
o
s
u
re
-
d
e
p
e
n
d
e
n
t
S
E
R
[e
-
/p
ix
/d
a
y
]
Figure 7.3. Left) 1e− event rate as a function of exposure time for images taken at T =
150 K. The linear fit is shown in red. Right) Exposure-dependent SER measurements as a
function of temperature using a Oscura prototype skipper CCD operating at surface (Data).
For comparison, we show the expected DC in dashed blue, computed using Eqs. (4.1) and (4.2),
with C300 K = 4 nA/cm2. For this model to fit the measurement at 160 K, we would need
C300 K = 0.11 nA/cm2 (dashed red).
We see, from Fig. 7.3, that the measurements of the exposure-dependent SER do not
follow the dark current model from Eqs. (4.1) and (4.2). First, for this model to be
consistent with the exposure-dependent SER measurement at 160 K, the dark current
figure of merit should be 0.11 nA/cm2 instead of 4 nA/cm2. This last value was measured
at LBNL in the tests structures on the wafers. From the model we expect the DC to
decrease approximately one order of magnitude per ∼10 K. This is not the trend that
the data show. Actually, the exposure-dependent SER from the data starts getting
flat below ∼160 K. However, it is well known that, at surface, the main contribution
to the exposure-dependent 1e− rate for T < 160 K does not come from dark current
but from low-energy radiation that is created when high-energy events interact with the
detector components [12,191,192]. All reported measurements at surface of the exposure-
dependent SER are above ∼1×10−2 e−/pix/day, consistent with ours. Lower exposure-
dependent 1e− rates are expected when measurements can be made underground, in
lower-background environments.
7.1. CHARACTERIZATION OF SKIPPER CCDS FROM NEW FOUNDRIES 75
7.1.3. Spurious charge
From the measurements described in Section 7.1.2, we extract an upper limit for the
generated spurious charge from the y-intercept (exposure-independent SER) of the linear
fits. The average value is 8.4× 10−4 e−/pix. Assuming that the source of this exposure-
independent SER is only spurious charge and considering that in these measurements
each read pixel underwent Nshifts = 1168 transfers/exposure, the measurement is con-
sistent with PSC = 7.2× 10−7 e−/pix/transfer.
Following an analogous procedure as the one described in [161], we measured the exposure-
independent SER in the output stage. We read out 20 pixels with Nskp = 5M, clocking
only the output stage. We obtained an average rate of ∼1× 10−7 e−/pix/sample, which
is an upper limit for PSC ; we can consider 1 sample ≃ 1 transfer.
Note that, in the exposure-independent SER measurements, we should also have a contri-
bution from the amplifier light generated during readout, which we are not subtracting.
However, if we consider PSC = 1× 10−7 e−/pix/transfer, the maximum Nshifts allowed
to have R1e− = 1×10−4 is 1000 (83) if doing 24 (2) hour exposures, evidencing the need
to reduce it to not be constrained by such low maximum allowed Nshifts.
Nowadays we are realizing that the exposure-independent SER could be a big source
of instrumental background and its understanding is essential to achieve experiments’
goals. Regarding spurious charge, we are considering several approaches to reduce its
generation, including the use of filtering techniques to decrease the slew rates of hori-
zontal clock signals, as well as implementing shaped clock signals [193].
7.1.4. Traps
We use the charge pumping technique [132, 194–196] to localize and characterize traps
in the new fabricated Oscura prototype skipper CCDs. This popular method consists
of filling the traps and allowing them to emit the trapped charge in their neighbor pixel
multiple times. This is done by repeatedly moving, back and forth in one pixel, a uniform
illuminated field creating “dipole” signals relative to the flat background. The method
is illustrated in Fig. 7.4.
Using a violet LED externally controlled by an Arduino Nano, we uniformly illuminated
the skipper CCDs and performed a charge pumping sequence that probes traps below
pixel phases 1 and 3, such as the one illustrated in Fig. 7.4. We collected images varying
the time that charge stayed below the pixel phases, dtph. We identified and tracked the
position of the dipoles in the set of images. Each dipole, composed of a bright and a
dark pixel with Sb and Sd e−, respectively, has an intensity given by
Idip =
1
2
|Sb − Sd| = NpumpsPcPeDtrap , (7.1)
76 FUTURE LARGE SKIPPER CCD EXPERIMENT
Figure 7.4. Taken from Ref. [196]. Sub-sequence (1-2-3-2) of a three-phase pumping sequence
to identify traps under phases 1 and 3. The diagram shows a trap under phase 1 that is being
filled multiple times. Here, tph is the time spent under each pixel phase.
where Npumps is the number of times we repeated the pumping sequence, Pc(e) is the
probability of the trap to capture (emit) a charge packet, given by Eq. (4.5), and Dtrap
is the trap depth which we assume to be 1e−. For each trap, we computed Idip as a
function of dtph and fitted it with the function of the second step in Eq. 7.1. From
the fits, we extracted the trap characteristic release time, τe. We did this at different
temperatures.
Fig. 7.5 shows images revealing traps with τe > 3.34 ms, corresponding to dtph =
50000 clocks, for two Oscura prototype sensors fabricated at different foundries. The
four images in the top row, corresponding to each of the four amplifiers in prototype-A,
show a much more significant density of dipoles compared to the images in the bottom
row, corresponding to each of the amplifiers in prototype-B.
Figure 7.5. Section of images corresponding to each of the 4 amplifiers after performing pocket
pumping with dtph = 50000 clocks (3.34 ms) in Oscura prototype-A (top) and prototype-B
(bottom), at 170K. The dipoles seen in the images correspond to charge traps under pixel
phases 1 and 3.
7.1. CHARACTERIZATION OF SKIPPER CCDS FROM NEW FOUNDRIES 77
The histograms in Fig. 7.6 show the number of traps per pixel as a function of τe for
these Oscura prototypes and a SENSEI skipper CCD (for comparison), at two different
temperatures. These histograms correspond to traps below pixel phases 1 and 3 and no
detection efficiency was taken into account. Considering a uniform density of traps below
the three phases in each pixel and assuming a conservative 10% detection efficiency with
a flat profile, the y-axis in Fig. 7.6 should be multiplied by a factor of 15 to obtain a
more realistic trap density.
Figure 7.6. Number of traps per pixel as a function of the release characteristic time τe for:
Oscura prototype-A at 150 K (blue) and at 170 K (green); Oscura prototype-B at 170 K (red);
and a SENSEI skipper CCD at 150 K (orange).
From Fig. 7.6 we can see that Oscura prototype-A has almost two orders of magnitude
more traps than the other sensors. This was unexpected and, in a cooperative effort
with the foundry that fabricated prototype-A, we are trying to implement a different
gettering method during the fabrication process to reduce possible impurities in the
silicon. Also, from Fig. 7.6 we can see that with lower T , the trap distributions move
towards higher τe, which is expected because of the dependence of τe with temperature,
see Eq. (4.3).
Traps with a release time greater than the pixel readout time will generate 1e− events in
the images. Eq. (4.7) determines the allowed mean number of traps that a hit traverses
during readout to be consistent with RT,1e− . Assuming RT,1e− = 1× 10−4 e−/pix/day,
Rbkgd = 100 dru and events below 100 keV that activate 10 pix in mean, the allowed
Ntraps is 2.7. Considering that, in these sensors, each hit traverses a maximum of
2336 pix, the allowed density of traps is ρtraps = 2.7/2336 ≃ 1.2 × 10−3 traps/pix.
In this case, all the characterized sensors could reach RT,1e− = 1 × 10−4 e−/pix/day,
78 FUTURE LARGE SKIPPER CCD EXPERIMENT
with prototype-A being close to the limit. However, as ρtraps is directly (inversely)
proportional to RT,1e− (Rbkgd), if we aim for a lower RT,1e− , the background rate should
also decrease for these sensors to reach the desired run conditions.
7.1.5. Charge transfer inefficiency
We exposed a Oscura prototype skipper CCD to a Fe55 X-ray source and took images
with Nskp = 4 samples/pix with zero exposure to avoid an overpopulation of X-ray
events. We computed the parallel and serial registers CTI at different operational tem-
peratures by linearly fitting the pixel population associated with X-ray depositions from
X-ray transfer plots [132]. In Fig. 7.7, we show the scatter plots of the pixel values versus
its column (left) and row (right) numbers from a set of images acquired at 170 K. In all
cases, we computed a CTI below ∼5× 10−5 per pixel transfer. As these measurements
were done using a sensor with high-density of traps and, as traps contribute to CTI, we
expect this number to be smaller when addressing the traps issue and optimizing the
clocking sequence.
Figure 7.7. Scatter plots of the pixel values in analog-to-digital units (ADU) versus its column
(left) and row (right) numbers for CTI measurements from an Oscura prototype skipper CCD
exposed to a Fe55 X-ray source. The linear fits to the X-ray pixel population are shown in red.
7.1.6. Buried-channel potential
Assuming the CCD surface as the plane xy, the potential/electric field at any depth (z)
is known from the solutions to the corresponding electrostatic Poisson’s equations, see
7.1. CHARACTERIZATION OF SKIPPER CCDS FROM NEW FOUNDRIES 79
Ref. [99]. Considering only 1D equations, the potential at the buried channel substrate
junction is roughly equivalent to the potential minimum Vmin, and can be expressed as
V (zJ) ≈ Vmin ≈ VG − VFB − qNA
2ϵSi
z2J
(
1 +
2ϵSid
ϵSiO2zJ
)
, (7.2)
where VG is the applied gate voltage, VFB is the flat-band voltage [197], q is the electron
charge, NA is the p-type buried channel doping density, zJ is the buried channel depth,
d is the gate insulator thickness and ϵSi = 103.6 × 10−14 C/V cm (ϵSiO2 = 34.5 ×
10−14 C/V cm) is the silicon (silicon dioxide) permittivity. Note that Vmin is independent
of Vsub. In Fig. 7.8 we show a CCD cross section along the z axis for reference.
Figure 7.8. Illustration adapted from Ref. [99]. P-type buried channel CCD cross section along
the z axis. Drawing is not to scale.
We developed a method to measure the potential minimum in the newly fabricated
Oscura prototype skipper CCDs for a given VG. As the electrode in which Vdrain is
applied is the only ohmic contact in the skipper CCD output stage, we use it as a
voltage reference. In a normal operation, Vdrain is usually kept at negative values, below
-22 V, to ensure efficient discarding of charge after each pixel measurement. When only
the output stage is clocked, we do not expect any measured charge unless it is being
injected from the Vdrain contact to the sense node. The point in which charge injection
starts is when Vdrain equals the potential minimum under the floating gate.
We took images with Nskp = 225 samples/pix, clocking only the output stage, increasing
Vdrain in steps of ∼0.2 V. We identified the voltage at which charge injection started,
i.e. Vdrain = Vmin, and plotted it against the applied Vref ≃ VG. We did this for
different VG using sensors with two different buried channel integrated doping doses. The
results are shown in Fig. 7.9. Data points are compared against the expected potential,
computed using Eq. (7.2), assuming VFB = 2 V,NAzJ = 1.3×1012 (1×1012) atoms/cm2,
NA = 2× 1016 atoms/cm3 and d = 75 nm.
As shown in Fig. 7.9, measurements of the potential minimum are consistent with the
expectations. According to Eq. (7.2), Vmin as a function of Vref is a linear function with
slope 1. However, data points seem to prefer a lower slope. We expect the potential
minimum under the sense node to be affected by the voltages applied to the adjacent
80 FUTURE LARGE SKIPPER CCD EXPERIMENT
Data 1.3 e-12
Data 1.0 e-12
Expected 1.3 e-12
Expected 1.0 e-12
-10 -9 -8 -7 -6
-22
-20
-18
-16
-14
Vref (V)
V
m
in
(V
)
Figure 7.9. Buried channel potential minimum under the sense node against Vref : measurements
using Oscura prototype skipper CCDs (data points) and expectations (solid line). The error
bars in the data points account for the non-uniformity between sensors, and amplifiers, with the
same doping dose.
gates. In fact, we have verified this with the Oscura prototype sensors obtaining that
1 V change in any adjacent electrode results in a ∼0.1 V change in the sense node
potential minimum. One last thing to notice is that higher buried channel doping doses
correspond to more negative values of Vmin, which is consistent with having the potential
minimum and, consequently, the collected charge, farther from the surface.
7.1.7. Transistor curves
Each output MOSFET in the CCDs exhibits characteristic curves (Id vs Vgs and Id
vs Vds) that account for its performance. From these curves we can extract the best
operating voltages for the transistor to work as an amplifier.
We measured the characteristic curves of the output transistors in the newly fabricated
Oscura prototype skipper CCDs. For these measurements, we set the reset MOSFET
to work in conduction mode so that Vref = Vg in the output MOSFET. We use an
external power supply for Vref to not be limited by the allowed voltage range from the
electronics. For each pair of Vdd = Vd and Vref , we measured Vs = −Id (20 kΩ) and
computed Vds = Vdd−Vs and Vgs = Vref −Vs. Results are shown in Figs. 7.10 and 7.11.
From Fig. 7.10 we see that the output transistors of sensors with the same buried channel
doping dose but fabricated at different foundries have slightly different characteristic
7.1. CHARACTERIZATION OF SKIPPER CCDS FROM NEW FOUNDRIES 81
Figure 7.10. Output transistor characteristic curves of Oscura prototype skipper CCDs fabri-
cated at Microchip (solid) and MITLL (dashed): Id vs Vgs for Vds = −7 V (left) and Id vs Vds
for Vgs = 7 V (right). The transistor load line for Vdd = −20 V is shown for reference (dotted
gray). The corresponding integrated buried channel doping densities are specified in the legends.
Figure 7.11. Output transistor characteristic curves of Oscura prototype skipper CCDs
fabricated at MITLL with two different integrated buried channel doping densities: 0.9 ×
1012 atoms/cm2 (solid) and 1.3 × 1012 atoms/cm2 (dashed). The curves in the left (right)
correspond to Id vs Vgs (Id vs Vds) for three different values of Vds (Vgs). The transistor load
line for Vdd = −20 V is shown for reference (dotted gray).
curves. From Fig. 7.10 (left) we can extract the transistors threshold voltage, i.e. the
minimum Vgs that is needed to create a conducting path between the source and drain
transistor terminals. For the Microchip sensors with an integrated buried channel doping
density of 1× 1012 (1.3× 1012) atoms/cm2, Vth ≃ 8.8 (11.3) V. For the MIT-LL sensors
with an integrated buried channel doping density of 0.9× 1012 (1.3× 1012) atoms/cm2,
Vth ≃ 8.8 (12) V. In Fig. 7.11 we show that the transistor characteristic curves exhibit
a similar behavior for different Vgs (Vds) as expected.
In Figs. 7.10 and 7.11 (right) we show for reference the transistor load line corresponding
82 FUTURE LARGE SKIPPER CCD EXPERIMENT
to Vdd = −20 V, a typical bias voltage. The transistor operation point lies within this
line. We found the characteristic curves of different output MOSFETs within one sensor
to be uniform; this is also true for output MOSFETs of different sensors, except for the
ones that were identified as not working properly.
7.1.8. Amplifier light
The intensity of the light coming from the output amplifier depends on the transistor
bias voltages. Depending on whether photons deposit their energy during exposure or
during readout, we observe different spatial distributions of charge deposition in the
images, see Fig. 7.12. Nevertheless, in both cases, the light intensity exhibits a spatial
exponential decay.
Figure 7.12. Left bottom part of images acquired with Oscura prototype skipper CCDs demon-
strating the spatial distribution of charge deposition from amplifier light during exposure (left)
and readout (right). The physical location of the output amplifier in the sensors is next to the
left bottom corner.
To characterize the light coming from the output amplifier, we took images varying the
bias voltages of the output transistor, i.e. Vref and Vdd. To study the effect during
exposure, we took full images with Nskp = 25 samples/pix and 10 min exposure, time
in which the reset transistor was set to conduction mode and the output transistor was
biased with the chosen Vref and Vdd. To readout the charge, we moved Vref and Vdd
to an operational point that minimizes amplifier light maintaining high gain. For each
pair of Vref and Vdd we took 5 images and then computed a median to remove any
effect from particle tracks. We did the analysis on the amplifier light during exposure
over these median images. To study amplifier light during readout, we took images
with Nskp = 324 samples/pix, 50 rows and zero exposure varying Vref and Vdd. We
did an absolute calibration of the images with the region corresponding to a “reverse”
overscan, i.e. extra columns in the image where the charge in the serial register was
clocked towards the center. We did the analysis on the amplifier light during readout
over these images.
First, we computed the total intensity of the images by summing the charge of all the
pixels. We found that more negative values of Vdd and/or more positive values of Vref
lead to a higher intensity, consistent with what is expected. To get some information of
7.2. MASSIVE SKIPPER CCD INSTRUMENT 83
the photons wavelength, assuming a monochromatic signal, we performed an exponential
2D (1D) fit to the light intensity in the images accounting for amplifier light during
exposure (readout),
I(r) = I0e
−αr , (7.3)
where I0 is the incident light intensity and α is the absorption coefficient, the reciprocal
of the absorption length. We related this coefficient to the emitted photons wavelength,
using Fig. 4.4 (right), resulting in infrared photons with a wavelenght between 1000
and 1050 nm. A more detailed study on the amplifier light emission should consider a
multichromatic signal an effects due to temperature.
7.2. Massive skipper CCD instrument
As part of the effort to increase mass in skipper CCD experiments, we put together the
largest instrument, in terms of active mass (∼80 g) and number of channels (160), at
SiDet at the FNAL; details can be found in Ref. [17]. The vacuum vessel, shown in
Fig. 7.13 (a), is similar to the SENSEI’s vessel at SNOLAB. It has two main volumes: a
cylinder at the top, which holds a copper box with room for 16 modules with 16 sensors
each, shown in Fig. 7.13 (b), and a box at the bottom, housing the front-end electronic
boards that connect to a VIB, shown in Fig. 7.13 (c). The air-side of the VIB connects to
a LTA board [140] with an expansion board, developed for this instrument, see Fig. 7.13
(a). A cryocooler is used to cool down the sensors to operating temperatures, around
150 K, and a vacuum pump is employed to maintain vacuum inside the vessel.
The array of sensors of this instrument consists of 160 Oscura prototype skipper CCDs
arranged in 10 ceramic MCMs. Each MCM consists of a single-layer printed circuit
board, made using a 635 µm thick ceramic substrate (96% Al2O3), over which the
sensors are glued and wire-bonded. This ceramic offers electrical isolation and a good
thermal conductivity. A Kapton flex cable is also glued to the ceramic board and wire-
bonded. As each sensor is read using only one of the four available amplifiers, each of
the MCMs corresponds to 16 readout channels.
The MCMs flex cables, routed to the bottom of the vessel, are connected to the front-
end electronic boards, one for each MCM. Each board has 16 analog channels (one per
sensor), to compute the pixel values, and an analog multiplexer, to read one channel at a
time. The analog processing chain of each channel implements a pile-up technique which
reduces the data rate, as only the final averaged pixel value is readout [180]. Details of
the operation, capabilities and advantages of using this front-end circuit can be found
in Ref. [179]. These electronic boards connect to a VIB which has a second stage of
analog multiplexing. An expansion board was designed to control the additional clock
signals required for the front-end boards and the two multiplexing stages. It also allows
the use of external power supplies for reference voltages exceeding the output current
capabilities of the LTA board. With these implementations, only a single channel and
84 FUTURE LARGE SKIPPER CCD EXPERIMENT
Figure 7.13. Taken from Ref. [17]. Photographs of the largest skipper CCD instrument at
the FNAL: a) vacuum vessel and outer electronics b) copper box and copper trays holding the
ceramic MCMs and c) front-end electronic boards connected to the VIB.
ADC of a LTA board, originally designed for reading a single four-channel CCD, is
needed to control the full instrument.
We took images with the full instrument to evaluate its performance. As seen in Fig. 7.14,
above 90% of the sensors worked, even though they were not preselected or pretested
before being assembled into the MCMs. A few malfunctioning sensors were disconnected,
by removing the wire-bonds. If images were arranged in the sensors physical positions,
we could, for example, track muons, as their straight charge depositions appear in more
than one sensor.
We took images varying Nskp to characterize the readout noise. For Nskp > 20, the noise
in all working channels follows the expected 1/
√
Nskp reduction rate, see Fig. 7.15. As
discussed in Ref. [180], for low Nskp, the noise is dominated by the analog readout
electronics rather than the sensors performance. For Nskp = 400 samples/pix, the
average channels noise is ∼0.175 e− and the pixel readout time is tpix = 14 ms, plus
an additional tmux = 0.64 ms for multiplexing the 16 MCMs. This corresponds to a
pixel readout rate of 68 pix/s, allowing to read out the whole array in 5.5 hours. These
results demonstrate the sensors single-electron resolution and the instrument’s capability
to achieve a similar performance as detectors with a smaller number of sensors.
7.2. MASSIVE SKIPPER CCD INSTRUMENT 85
Figure 7.14. Taken from Ref. [17]. 160 images showing particle tracks corresponding to each of
the sensors in the largest skipper CCD instrument ever built.
Figure 7.15. Taken from Ref. [17]. Noise as a function of Nskp for all 10 MCMs and channels.
The expected 1/
√
Nskp dependence is shown in pink. The blue histograms correspond to the
gain and noise distributions computed for Nskp = 400 samples/pix.
86 FUTURE LARGE SKIPPER CCD EXPERIMENT
Final remarks
The study of reactor neutrinos with CCDs has demonstrated to be a powerful tool to
explore physics within and beyond the SM, as shown by the results of the CONNIE
experiment [7–9]. Particularly, CONNIE’s limits in the parameter space of light medi-
ators [8] were the most stringest constraints among experiments looking at the CEνNS
detection channel in the low-mass mediator region at the time of their publication. These
results marked a milestone as the first search for new physics with reactor neutrinos and
CCDs, demonstrating that reactor neutrino experiments are a powerful tool to explore
new physics at low energies and highlighting their complementarity with neutrino beam
experiments. Furthermore, these limits showed that the low-energy threshold of the
CCD technology is advantageous when searching for signals whose cross section is in-
versely proportional to the recoil energy. CONNIE could put more powerful limits in
the parameter space of light mediators if spectral information is included in the analysis,
as some non-standard interactions produce a change in the spectral shape.
The development of the skipper CCD technology and its electron-counting capability has
revolutionized the field, providing a better understanding of the low-energy backgrounds
present in CCD detectors [16, 161] and high-precision measurements. In particular, the
upgrade of the CONNIE detector with skipper CCDs was a success, allowing us to
develop more precise analysis tools to get rid of unwanted backgrounds. Moreover, the
preliminary results show that higher efficiencies at lower energies can be reached with
this technology, compared to what we had with standard CCDs. This is important
because the expected number of events from reactor neutrino interactions increases at
low energies, hence, the sensitivity for detection increases.
CONNIE’s main drawback is the external background. As CCDs are slow-readout sen-
sors, taking approximately 40 µs to readout one pixel sample, we can not discard external
backgrounds with timing. This is particularly significant for CCD experiments at sur-
face or shallow-depths, where the main backgrounds are secondary particles produced by
cosmic ray interactions. Performing smart readout [198] has been proposed to decrease
skipper CCDs readout time. Moreover, emerging fast-readout semiconductor technolo-
gies [199] such as CCDs with Single-electron Sensitive ReadOut (SiSeRO) stages [200],
Multi-Amplifier Sensing (MAS) CCDs [201] and Complementary Metal-Oxide Semicon-
ductor (CMOS) sensors with skipper output stages [202], all of them with single-electron
counting capability, are being developed and could be used in future experiments as
fast-readout sensors or complementary active vetoes. Since CONNIE can not overcome
backgrounds with timing resolution, we can increase sensitivity to reactor neutrino in-
87
88 FINAL REMARKS
teractions by increasing its mass. Particularly, in chapter 5 we concluded that CONNIE
would need a 1 kg skipper CCD detector and data collected during two months to search
for a CEνNS signal with more than 90% C.L. for the less favorable nuclear quenching
factor. Now that a strong experimental program aiming to build massive skipper CCD
detectors is ongoing, motivated by the remarkable results achieved by using skipper
CCDs to explore low-energy interactions, we could think of installing at CONNIE a
large-mass detector. What seems to be plausible, according to CONNIE’s experimental
setup, is to install a ∼100 g detector. With this mass, a 90% C.L. CEνNS measurement
could be achieved in a 1.5-year run. The probability with which reactor neutrino inter-
actions happen in the CONNIE detector could be incremented by increasing the flux
it receives. This could be done by moving the detector inside the dome, closer to the
reactor core. However, this will imply dealing with reactor-induced backgrounds, which
will need to be understood.
Regarding the use of skipper CCDs in dark sector searches, dark matter experiments at
underground laboratories have published world-leading results on DM-electron interac-
tions [10–12, 14], positioning this technology as a leading frontrunner on a global scale.
Also, recent results searching for millicharged particles using skipper CCDs [13] and
the explored projected sensitivities to mCPs of a massive skipper CCD detector at an
accelerator facility have proved the skipper CCD technology capability in this area. The
latter study has also demonstrated that multi-layer skipper CCD experiments, which are
expected to have low instrumental backgrounds, could use tracking as a tool to identify
possible signals.
In summary, this work provides an overview of the physics within and beyond the SM
that can be probed with reactor neutrinos and highlights the potential of skipper CCD
technology in both reactor neutrino studies and dark sector searches.
The main contents of this thesis have been published in Refs. [8, 16], of which I am
the principal author. Also, several other publications are referenced throughout the
chapters, e.g. Refs. [6, 7, 9, 15, 18, 19, 106, 120, 121, 163–165, 167, 198, 201], of which I am
a co-author.
Bibliography
[1] E. Fermi. Tentativo di una Teoria Dei Raggi β. Il Nuovo Cimento 11 (1934) 1.
[2] Y. Fukuda, T. Hayakawa, E. Ichihara, K. Inoue, K. Ishihara, H. Ishino et al..
Super-Kamiokande collaboration. Evidence for Oscillation of Atmospheric
Neutrinos. Phys. Rev. Lett. 81 (1998) 1562.
[3] Q.R. Ahmad, R.C. Allen, T.C. Andersen, J.D. Anglin, G. Bühler, J.C. Barton
et al.. SNO collaboration. Measurement of the Rate of νe + d → p + p + e−
Interactions Produced by 8B Solar Neutrinos at the Sudbury Neutrino
Observatory. Phys. Rev. Lett. 87 (2001) 071301.
[4] J. Estrada, H. Cease, H.T. Diehl, B. Flaugher, J. Jones, D. Kubik et al..
Prospects for a Direct Dark Matter Search Using High Resistivity CCD Detectors.
arXiv:0802.2872.
[5] J. Tiffenberg, M. Sofo-Haro, A. Drlica-Wagner, R. Essig, Y. Guardincerri,
S. Holland et al.. Single-Electron and Single-Photon Sensitivity with a Silicon
Skipper CCD. Phys. Rev. Lett. 119 (2017) 131802.
[6] A. Aguilar-Arevalo, X. Bertou, C. Bonifazi, M. Butner, G. Cancelo,
A.C. Vázquez et al.. Results of the engineering run of the Coherent Neutrino
Nucleus Interaction Experiment (CONNIE). JINST 11 (2016) P07024.
[7] A. Aguilar-Arevalo, X. Bertou, C. Bonifazi, G. Cancelo, A. Castañeda,
B. Cervantes Vergara et al.. CONNIE collaboration. Exploring low-energy
neutrino physics with the Coherent Neutrino Nucleus Interaction Experiment.
Phys. Rev. D 100 (2019) 092005.
[8] A. Aguilar-Arevalo, X. Bertou, C. Bonifazi, G. Cancelo, B.A. Cervantes-Vergara,
C. Chavez et al.. Search for light mediators in the low-energy data of the
CONNIE reactor neutrino experiment. JHEP 2020 (2020) 54.
[9] A. Aguilar-Arevalo, J. Bernal, X. Bertou, C. Bonifazi, G. Cancelo, V. Carvalho
et al.. Search for coherent elastic neutrino-nucleus scattering at a nuclear reactor
with CONNIE 2019 data. JHEP 2022 (2022) 17.
[10] M. Crisler, R. Essig, J. Estrada, G. Fernandez, J. Tiffenberg, M.S. Haro et al..
SENSEI collaboration. SENSEI: First Direct-Detection Constraints on Sub-GeV
Dark Matter from a Surface Run. Phys. Rev. Lett. 121 (2018) 061803.
89
90 BIBLIOGRAPHY
[11] O. Abramoff, L. Barak, I.M. Bloch, L. Chaplinsky, M. Crisler, Dawa et al..
SENSEI collaboration. SENSEI: Direct-Detection Constraints on Sub-GeV Dark
Matter from a Shallow Underground Run Using a Prototype Skipper CCD. Phys.
Rev. Lett. 122 (2019) 161801.
[12] L. Barak, I.M. Bloch, M. Cababie, G. Cancelo, L. Chaplinsky, F. Chierchie et al..
SENSEI collaboration. SENSEI: Direct-Detection Results on sub-GeV Dark
Matter from a New Skipper CCD. Phys. Rev. Lett. 125 (2020) 171802.
[13] L. Barak, I. Bloch, A. Botti, M. Cababie, G. Cancelo, L. Chaplinsky et al..
SENSEI: Search for Millicharged Particles produced in the NuMI Beam.
arXiv:2305.04964.
[14] I. Arnquist, N. Avalos, D. Baxter, X. Bertou, N. Castelló-Mor, A.E. Chavarria
et al.. DAMIC-M collaboration. First Constraints from DAMIC-M on Sub-GeV
Dark-Matter Particles Interacting with Electrons. Phys. Rev. Lett. 130 (2023)
171003.
[15] A. Aguilar-Arevalo, F.A. Bessia, N. Avalos, D. Baxter, X. Bertou, C. Bonifazi
et al.. The Oscura Experiment. arXiv:2202.10518.
[16] B.A. Cervantes-Vergara, S. Perez, J. Estrada, A. Botti, C.R. Chavez,
F. Chierchie et al.. Skipper-CCD sensors for the Oscura experiment:
requirements and preliminary tests. JINST 18 (2023) P08016.
[17] F. Chierchie, C. Chavez, M.S. Haro, G.F. Moroni, B. Cervantes-Vergara, S. Perez
et al.. First results from a multiplexed and massive instrument with sub-electron
noise Skipper-CCDs. JINST 18 (2023) P01040.
[18] S. Perez, D. Rodrigues, J. Estrada, R. Harnik, Z. Liu, B.A. Cervantes-Vergara
et al.. Early Science with the Oscura Integration Test. arXiv:2304.08625.
[19] B. Cervantes-Vergara, S. Perez, J. D’Olivo, J. Estrada, D. Grimm, S. Holland
et al.. Skipper-CCDs: Current applications and future. Nucl. Instrum. Meth. A
1046 (2023) 167681.
[20] R.L. Workman, V.D. Burkert, V. Crede, E. Klempt, U. Thoma, L. Tiator et al..
Review of Particle Physics. Progress of Theoretical and Experimental Physics
2022 (2022) 083C01.
[21] L.M. Brown. The idea of the neutrino. Phys. Today 31 (1978) 23.
[22] C. Cowan, F. Reines, F. Harrison, H. Kruse and A. McGuire. Detection of the
free neutrino: A Confirmation. Science 124 (1956) 103.
[23] All nobel prizes in physics. Nobel Prize Outreach AB 2023. Accessed on May 30,
2023, URL: https://www.nobelprize.org/prizes/lists/all-nobel-prizes-in-physics/.
BIBLIOGRAPHY 91
[24] G. Danby, J.-M. Gaillard, K. Goulianos, L.M. Lederman, N. Mistry, M. Schwartz
et al.. Observation of High-Energy Neutrino Reactions and the Existence of Two
Kinds of Neutrinos. Phys. Rev. Lett. 9 (1962) 36.
[25] K. Kodama, N. Ushida, C. Andreopoulos, N. Saoulidou, G. Tzanakos, P. Yager
et al.. Observation of tau neutrino interactions. Phys. Lett. B 504 (2001) 218.
[26] P. Aarnio, P. Abreu, W. Adam, P. Adrianos, T. Adye, G. Akopdzhanov et al..
Measurement of the mass and width of the Z0-particle from multihadronic final
states produced in e+e- annihilations. Phys. Lett. B 231 (1989) 539.
[27] D. Decamp, B. Deschizeaux, J.-P. Lees, M.-N. Minard, J. Crespo, M. Delfino
et al.. A precise determination of the number of families with light neutrinos and
of the Z boson partial widths. Phys. Lett. B 235 (1990) 399.
[28] B. Adeva, O. Adriani, M. Aguilar-Benitez, H. Akbari, J. Alcaraz, A. Aloisio
et al.. A determination of the properties of the neutral intermediate vector boson
Z0. Phys. Lett. B 231 (1989) 509.
[29] M. Akrawy, G. Alexander, J. Allison, P. Allport, K. Anderson, J. Armitage
et al.. Measurement of the Z0 mass and width with the opal detector at LEP.
Phys. Lett. B 231 (1989) 530.
[30] E. Vitagliano, I. Tamborra and G. Raffelt. Grand unified neutrino spectrum at
Earth: Sources and spectral components. Rev. Mod. Phys. 92 (2020) 045006.
[31] C. Giunti and C.W. Kim. Fundamentals of Neutrino Physics and Astrophysics.
Oxford University Press (2007),
DOI: 10.1093/acprof:oso/9780198508717.001.0001.
[32] J.A. Formaggio and G.P. Zeller. From eV to EeV: Neutrino cross sections across
energy scales. Rev. Mod. Phys. 84 (2012) 1307.
[33] D. Akimov, J.B. Albert, P. An, C. Awe, P.S. Barbeau, B. Becker et al..
COHERENT collaboration. COHERENT 2018 at the Spallation Neutron
Source. arXiv:1803.09183.
[34] W.J. Marciano and Z. Parsa. Neutrino–electron scattering theory*. J. of Phys. G
29 (2003) 2629.
[35] M. Tanabashi, K. Hagiwara, K. Hikasa, K. Nakamura, Y. Sumino, F. Takahashi
et al.. Particle Data Group collaboration. Review of Particle Physics. Phys.
Rev. D 98 (2018) 030001.
[36] D.Z. Freedman. Coherent effects of a weak neutral current. Phys. Rev. D 9
(1974) 1389.
92 BIBLIOGRAPHY
[37] D. Akimov, J.B. Albert, P. An, C. Awe, P.S. Barbeau, B. Becker et al..
Observation of coherent elastic neutrino-nucleus scattering. Science 357 (2017)
1123.
[38] D. Akimov, J.B. Albert, P. An, C. Awe, P.S. Barbeau, B. Becker et al..
COHERENT collaboration. First Measurement of Coherent Elastic
Neutrino-Nucleus Scattering on Argon. Phys. Rev. Lett. 126 (2021) 012002.
[39] S.R. Klein and J. Nystrand. Interference in exclusive vector meson production in
heavy ion collisions. Phys. Rev. Lett. 84 (2000) 2330.
[40] A. Strumia and F. Vissani. Precise quasielastic neutrino/nucleon cross-section.
Phys. Lett. B 564 (2003) 42.
[41] S.L. Glashow. Partial-symmetries of weak interactions. Nuc. Phys. 22 (1961) 579.
[42] S. Weinberg. A Model of Leptons. Phys. Rev. Lett. 19 (1967) 1264.
[43] A. Salam. Weak and Electromagnetic Interactions. Conf. Proc. C 680519 (1968)
367.
[44] T. Rink. Investigating neutrino physics within and beyond the standard model
using CONUS experimental data. Ph.D. thesis. Heidelberg University. Germany.
(2022), DOI: 10.11588/heidok.00031274.
[45] B. Dutta, S. Liao, L.E. Strigari and J.W. Walker. Non-standard interactions of
solar neutrinos in dark matter experiments. Phys. Lett. B 773 (2017) 242.
[46] D.W.P. Amaral, D. Cerdeno, A. Cheek and P. Foldenauer. A direct detection
view of the neutrino NSI landscape. arXiv:2302.12846.
[47] J. Barranco, O.G. Miranda and T.I. Rashba. Probing new physics with coherent
neutrino scattering off nuclei. JHEP 2005 (2005) 021.
[48] E. Morgante. Simplified Dark Matter Models. Advances in High Energy Physics
2018 (2018) 5012043.
[49] P. Fayet. U boson interpolating between a generalized dark photon or dark Z, an
axial boson, and an axionlike particle. Phys. Rev. D 103 (2021) 035034.
[50] D.E. Kaplan, M.A. Luty and K.M. Zurek. Asymmetric dark matter. Phys. Rev.
D 79 (2009) 115016.
[51] M. Pospelov, A. Ritz and M. Voloshin. Secluded WIMP dark matter. Phys. Lett.
B 662 (2008) 53.
[52] D. Hooper and K.M. Zurek. Natural supersymmetric model with MeV dark
matter. Phys. Rev. D 77 (2008) 087302.
BIBLIOGRAPHY 93
[53] P. Fayet. U -boson production in e+e− annihilations, ψ and Υ decays, and light
dark matter. Phys. Rev. D 75 (2007) 115017.
[54] P. Fayet. Light spin-12 or spin-0 dark matter particles. Phys. Rev. D 70 (2004)
023514.
[55] Y. Farzan, M. Lindner, W. Rodejohann and X.-J. Xu. Probing neutrino coupling
to a light scalar with coherent neutrino scattering. JHEP 2018 (2018) .
[56] M. Cirelli, E.D. Nobile and P. Panci. Tools for model-independent bounds in
direct dark matter searches. JCAP 2013 (2013) 019.
[57] Cerdeño, David G. and Fairbairn, Malcolm and Jubb, Thomas and Machado,
Pedro A. N. and Vincent, Aaron C. and Bœhm, Céline. Physics from solar
neutrinos in dark matter direct detection experiments. JHEP 2016 (2016) 118.
Erratum: JHEP 09 (2016) 048.
[58] J. Liao and D. Marfatia. COHERENT constraints on nonstandard neutrino
interactions. Phys. Lett. B 775 (2017) 54.
[59] M. Atzori Corona, M. Cadeddu, N. Cargioli, F. Dordei, C. Giunti, Y.F. Li et al..
Impact of the Dresden-II and COHERENT neutrino scattering data on neutrino
electromagnetic properties and electroweak physics. JHEP 2022 (2022) 164.
[60] P. Langacker. The physics of heavy Z
′
gauge bosons. Rev. Mod. Phys. 81 (2009)
1199.
[61] P.B. Denton, Y. Farzan and I.M. Shoemaker. Testing large non-standard
neutrino interactions with arbitrary mediator mass after COHERENT data.
JHEP 2018 (2018) 37.
[62] M.S. Dvornikov and A.I. Studenikin. Electromagnetic form factors of a massive
neutrino. J. Exp. Theor. Phys 99 (2004) 254.
[63] C. Giunti and A. Studenikin. Neutrino electromagnetic interactions: A window
to new physics. Rev. Mod. Phys. 87 (2015) 531.
[64] P. Vogel and J. Engel. Neutrino electromagnetic form factors. Phys. Rev. D 39
(1989) 3378.
[65] G. Bressi, G. Carugno, F. Della Valle, G. Galeazzi, G. Ruoso and G. Sartori.
Testing the neutrality of matter by acoustic means in a spherical resonator. Phys.
Rev. A 83 (2011) 052101.
[66] C. Awe et al.. CHANDLER, CONNIE, CONUS, Daya Bay, JUNO, MTAS,
NEOS, NuLat, PROSPECT, RENO, Ricochet, ROADSTR Near-Field
Working Group, SoLid, Stereo, Valencia-Nantes TAGS, vIOLETA,
WATCHMAN collaboration. High Energy Physics Opportunities Using Reactor
Antineutrinos. arXiv:2203.07214.
94 BIBLIOGRAPHY
[67] K. Eguchi, S. Enomoto, K. Furuno, J. Goldman, H. Hanada, H. Ikeda et al..
KamLAND collaboration. First Results from KamLAND: Evidence for Reactor
Antineutrino Disappearance. Phys. Rev. Lett. 90 (2003) 021802.
[68] T. Araki, K. Eguchi, S. Enomoto, K. Furuno, K. Ichimura, H. Ikeda et al..
KamLAND collaboration. Measurement of Neutrino Oscillation with
KamLAND: Evidence of Spectral Distortion. Phys. Rev. Lett. 94 (2005) 081801.
[69] S. Abe, T. Ebihara, S. Enomoto, K. Furuno, Y. Gando, K. Ichimura et al..
KamLAND collaboration. Precision Measurement of Neutrino Oscillation
Parameters with KamLAND. Phys. Rev. Lett. 100 (2008) 221803.
[70] F.P. An, J.Z. Bai, A.B. Balantekin, H.R. Band, D. Beavis, W. Beriguete et al..
Observation of Electron-Antineutrino Disappearance at Daya Bay. Phys. Rev.
Lett. 108 (2012) 171803.
[71] J.K. Ahn, S. Chebotaryov, J.H. Choi, S. Choi, W. Choi, Y. Choi et al.. RENO
collaboration. Observation of Reactor Electron Antineutrinos Disappearance in
the RENO Experiment. Phys. Rev. Lett. 108 (2012) 191802.
[72] Y. Abe, C. Aberle, T. Akiri, J.C. dos Anjos, F. Ardellier, A.F. Barbosa et al..
Double Chooz collaboration. Indication of Reactor νe Disappearance in the
Double Chooz Experiment. Phys. Rev. Lett. 108 (2012) 131801.
[73] E. Pasierb, H.S. Gurr, J. Lathrop, F. Reines and H.W. Sobel. Detection of Weak
Neutral Current Using Fission νe on Deuterons. Phys. Rev. Lett. 43 (1979) 96.
[74] F. Reines, H.S. Gurr and H.W. Sobel. Detection of νe − e Scattering. Phys. Rev.
Lett. 37 (1976) 315.
[75] M. Deniz, S.T. Lin, V. Singh, J. Li, H.T. Wong, S. Bilmis et al.. TEXONO
collaboration. Measurement of νe-electron scattering cross section with a CsI(Tl)
scintillating crystal array at the Kuo-Sheng nuclear power reactor. Phys. Rev. D
81 (2010) 072001.
[76] F.P. An, A.B. Balantekin, H.R. Band, M. Bishai, S. Blyth, D. Cao et al.. Daya
Bay collaboration. Improved Search for a Light Sterile Neutrino with the Full
Configuration of the Daya Bay Experiment. Phys. Rev. Lett. 117 (2016) 151802.
[77] Y.J. Ko, B.R. Kim, J.Y. Kim, B.Y. Han, C.H. Jang, E.J. Jeon et al.. NEOS
collaboration. Sterile Neutrino Search at the NEOS Experiment. Phys. Rev. Lett.
118 (2017) 121802.
[78] I. Alekseev, V. Belov, V. Brudanin, M. Danilov, V. Egorov, D. Filosofov et al..
Search for sterile neutrinos at the DANSS experiment. Phys. Lett. B 787 (2018)
56.
BIBLIOGRAPHY 95
[79] H. Almazán, L. Bernard, A. Blanchet, A. Bonhomme, C. Buck, P. del
Amo Sanchez et al.. STEREO collaboration. Improved sterile neutrino
constraints from the STEREO experiment with 179 days of reactor-on data.
Phys. Rev. D 102 (2020) 052002.
[80] M. Andriamirado, A.B. Balantekin, H.R. Band, C.D. Bass, D.E. Bergeron,
D. Berish et al.. PROSPECT collaboration. Improved short-baseline neutrino
oscillation search and energy spectrum measurement with the PROSPECT
experiment at HFIR. Phys. Rev. D 103 (2021) 032001.
[81] A.P. Serebrov, R.M. Samoilov, V.G. Ivochkin, A.K. Fomin, V.G. Zinoviev,
P.V. Neustroev et al.. Search for sterile neutrinos with the Neutrino-4 experiment
and measurement results. Phys. Rev. D 104 (2021) 032003.
[82] J. Colaresi, J.I. Collar, T.W. Hossbach, A.R.L. Kavner, C.M. Lewis,
A.E. Robinson et al.. First results from a search for coherent elastic
neutrino-nucleus scattering at a reactor site. Phys. Rev. D 104 (2021) 072003.
[83] H. Bonet, A. Bonhomme, C. Buck, K. Fülber, J. Hakenmüller, G. Heusser et al..
Novel constraints on neutrino physics beyond the standard model from the
CONUS experiment. JHEP 2022 (2022) 85.
[84] C. Giunti, Y. Li, C. Ternes and Z. Xin. Reactor antineutrino anomaly in light of
recent flux model refinements. Phys. Lett. B 829 (2022) 137054.
[85] H.T. Wong, H.B. Li, S.T. Lin, F.S. Lee, V. Singh, S.C. Wu et al.. TEXONO
collaboration. Search of neutrino magnetic moments with a high-purity
germanium detector at the Kuo-Sheng nuclear power station. Phys. Rev. D 75
(2007) 012001.
[86] H. Bonet, A. Bonhomme, C. Buck, K. Fülber, J. Hakenmüller, J. Hempfling
et al.. Full background decomposition of the CONUS experiment. Eur. Phys. J. C
83 (2023) 195.
[87] H. Bonet, A. Bonhomme, C. Buck, K. Fülber, J. Hakenmüller, G. Heusser et al..
CONUS collaboration. Constraints on Elastic Neutrino Nucleus Scattering in
the Fully Coherent Regime from the CONUS Experiment. Phys. Rev. Lett. 126
(2021) 041804.
[88] A. Bonhomme, H. Bonet, C. Buck, J. Hakenmüller, G. Heusser, T. Hugle et al..
Direct measurement of the ionization quenching factor of nuclear recoils in
germanium in the keV energy range. Eur. Phys. J. C 82 (2022) 815.
[89] H. Bonet, A. Bonhomme, C. Buck, K. Fülber, J. Hakenmüller, J. Hempfling
et al.. CONUS collaboration. First upper limits on neutrino electromagnetic
properties from the CONUS experiment. Eur. Phys. J. C 82 (2022) 813.
96 BIBLIOGRAPHY
[90] J. Colaresi, J.I. Collar, T.W. Hossbach, C.M. Lewis and K.M. Yocum.
Measurement of Coherent Elastic Neutrino-Nucleus Scattering from Reactor
Antineutrinos. Phys. Rev. Lett. 129 (2022) 211802.
[91] P. Coloma, I. Esteban, M.C. Gonzalez-Garcia, L. Larizgoitia, F. Monrabal and
S. Palomares-Ruiz. Bounds on new physics with data of the Dresden-II reactor
experiment and COHERENT. JHEP 2022 (2022) 37.
[92] J. Liao, H. Liu and D. Marfatia. Implications of the first evidence for coherent
elastic scattering of reactor neutrinos. Phys. Rev. D 106 (2022) L031702.
[93] D. Aristizabal Sierra, V. De Romeri and D.K. Papoulias. Consequences of the
Dresden-II reactor data for the weak mixing angle and new physics. JHEP 2022
(2022) 76.
[94] A. Majumdar, D.K. Papoulias, R. Srivastava and J.W.F. Valle. Physics
implications of recent Dresden-II reactor data. Phys. Rev. D 106 (2022) 093010.
[95] I. Alekseev, K. Balej, V. Belov, S. Evseev, D. Filosofov, M. Fomina et al.. νGeN
collaboration. First results of the νGeN experiment on coherent elastic
neutrino-nucleus scattering. Phys. Rev. D 106 (2022) L051101.
[96] A. Kopec, A. Baxter, M. Clark, R. Lang, S. Li, J. Qin et al.. Correlated single-
and few-electron backgrounds milliseconds after interactions in dual-phase liquid
xenon time projection chambers. JINST 16 (2021) P07014.
[97] D. Akimov, I. Alexandrov, R. Alyev, V. Belov, A. Bolozdynya, A. Etenko et al..
The RED-100 experiment. JINST 17 (2022) T11011.
[98] J.J. Choi, E.J. Jeon, J.Y. Kim, K.W. Kim, S.H. Kim, S.K. Kim et al.. Exploring
coherent elastic neutrino-nucleus scattering using reactor electron antineutrinos
in the NEON experiment. Eur. Phys. J. C 83 (2023) 226.
[99] S.E. Holland, D.E. Groom, N.P. Palaio, R.J. Stover and M. Wei. Fully depleted,
back-illuminated charge-coupled devices fabricated on high-resistivity silicon.
IEEE Transactions on Electron Devices 50 (2003) 225.
[100] S.E. Holland, W.F. Kolbe and C.J. Bebek. Device Design for a 12.3-Megapixel,
Fully Depleted, Back-Illuminated, High-Voltage Compatible Charge-Coupled
Device. IEEE Transactions on Electron Devices 56 (2009) 2612.
[101] J. Estrada, T. Abbott, B. Angstadt, L. Buckley-Geer, M. Brown, J. Campa
et al.. CCD testing and characterization for dark energy survey in Ground-based
and Airborne Instrumentation for Astronomy. I.S. McLean and M. Iye (Eds.).
Vol. 6269. p. 62693K. International Society for Optics and Photonics. SPIE.
(2006), DOI: 10.1117/12.672596.
BIBLIOGRAPHY 97
[102] S. Holland. Fabrication of detectors and transistors on high-resistivity silicon.
Nucl. Instrum. Meth. A 275 (1989) 537.
[103] S. Holland. Fully depleted back illuminated CCD. U.S. Patents No. 6,259,085.
(2000).
[104] P. Moore, N. Buchholz, M. Hunten and D. Sawyer. MONSOON image
acquisition system: control techniques for application to the orthogonal transfer
array detectors in Ground-based and Airborne Instrumentation for Astronomy II.
I.S. McLean and M.M. Casali (Eds.). Vol. 7014. p. 70147O. International Society
for Optics and Photonics. SPIE. (2008), DOI: 10.1117/12.802254.
[105] T. Shaw, O. Ballester, L. Cardiel-Sas, J. Castilla, S. Chappa, J. de Vicente et al..
The Dark Energy Camera readout system in High Energy, Optical, and Infrared
Detectors for Astronomy V. A.D. Holland and J.W. Beletic (Eds.). Vol. 8453.
p. 84532Q. International Society for Optics and Photonics. SPIE. (2012),
DOI: 10.1117/12.926284.
[106] A. Aguilar-Arevalo, D. Amidei, X. Bertou, M. Butner, G. Cancelo, A. Castañeda
Vázquez et al.. DAMIC collaboration. Search for low-mass WIMPs in a 0.6 kg
day exposure of the DAMIC experiment at SNOLAB. Phys. Rev. D 94 (2016)
082006.
[107] R. D. Ryan. Precision Measurements of the Ionization Energy and Its
Temperature Variation in High Purity Silicon Radiation Detectors. IEEE
Transactions on Nuclear Science 20 (1973) 473.
[108] A.R. Sattler. Ionization Produced by Energetic Silicon Atoms within a Silicon
Lattice. Phys. Rev. 138 (1965) A1815.
[109] G. Gerbier, E. Lesquoy, J. Rich, M. Spiro, C. Tao, D. Yvon et al.. Measurement
of the ionization of slow silicon nuclei in silicon for the calibration of a silicon
dark-matter detector. Phys. Rev. D 42 (1990) 3211.
[110] P. Zecher, D. Wang, J. Rapaport, C.J. Martoff and B.A. Young. Energy
deposition of energetic silicon atoms within a silicon lattice. Phys. Rev. A 41
(1990) 4058.
[111] B.L. Dougherty. Measurements of ionization produced in silicon crystals by
low-energy silicon atoms. Phys. Rev. A 45 (1992) 2104.
[112] J. Lindhard, M. Scharff and H. Schioett. Range concepts and heavy ion ranges
(Notes on atomic collisions, II). Kgl. Danske Videnskab. Selskab. Mat. Fys.
Medd. 33 (1963) .
[113] A.E. Chavarria, J.I. Collar, J.R. Peña, P. Privitera, A.E. Robinson, B. Scholz
et al.. Measurement of the ionization produced by sub-keV silicon nuclear recoils
in a CCD dark matter detector. Phys. Rev. D 94 (2016) .
98 BIBLIOGRAPHY
[114] F. Izraelevitch, D. Amidei, A. Aprahamian, R. Arcos-Olalla, G. Cancelo,
C. Casarella et al.. A measurement of the ionization efficiency of nuclear recoils
in silicon. JINST 12 (2017) P06014.
[115] R. Agnese, A. Anderson, T. Aramaki, W. Baker, D. Balakishiyeva, S. Banik
et al.. Nuclear-recoil energy scale in CDMS II silicon dark-matter detectors. Nucl.
Instrum. Meth. A 905 (2018) 71.
[116] M.F. Albakry, I. Alkhatib, D. Alonso, D.W.P. Amaral, T. Aralis, T. Aramaki
et al.. SuperCDMS collaboration. First measurement of the nuclear-recoil
ionization yield in silicon at 100 ev. Phys. Rev. Lett. 131 (2023) 091801.
[117] P. Sorensen. Atomic limits in the search for galactic dark matter. Phys. Rev. D
91 (2015) 083509.
[118] Y. Sarkis, A. Aguilar-Arevalo and J.C. D’Olivo. Study of the ionization efficiency
for nuclear recoils in pure crystals. Phys. Rev. D 101 (2020) 102001.
[119] Y. Sarkis, A. Aguilar-Arevalo and J.C. D’Olivo. Ionization efficiency for nuclear
recoils in silicon from about 50 eV to 3 MeV. Phys. Rev. A 107 (2023) 062811.
[120] A. Aguilar-Arevalo, D. Amidei, D. Baxter, G. Cancelo, B.A.C. Vergara,
A.E. Chavarria et al.. DAMIC collaboration. Results on Low-Mass Weakly
Interacting Massive Particles from an 11 kg d Target Exposure of DAMIC at
SNOLAB. Phys. Rev. Lett. 125 (2020) 241803.
[121] A. Aguilar-Arevalo, D. Amidei, I. Arnquist, D. Baxter, G. Cancelo,
B.A.C. Vergara et al.. DAMIC collaboration. Characterization of the background
spectrum in DAMIC at SNOLAB. Phys. Rev. D 105 (2022) 062003.
[122] A.N. Khan and W. Rodejohann. New physics from COHERENT data with an
improved quenching factor. Phys. Rev. D 100 (2019) 113003.
[123] F. Jegerlehner and A. Nyffeler. The Muon g-2. Phys. Rept. (2009) 1.
[124] A. Keshavarzi, K.S. Khaw and T. Yoshioka. Muon g-2: A review. Nuc. Phys. B
975 (2022) 115675.
[125] V. De Romeri, O.G. Miranda, D.K. Papoulias, G. Sanchez Garcia, M. Tórtola
and J.W.F. Valle. Physics implications of a combined analysis of COHERENT
CsI and LAr data. JHEP 2023 (2023) 35.
[126] C.E. Chandler, R.A. Bredthauer, J.R. Janesick and J.A. Westphal. Sub-electron
noise charge-coupled devices in Charge-Coupled Devices and Solid State Optical
Sensors. M.M. Blouke (Ed.). Vol. 1242. pp. 238 – 251. International Society for
Optics and Photonics. SPIE. (1990), DOI: 10.1117/12.19457.
BIBLIOGRAPHY 99
[127] J.R. Janesick, T.S. Elliott, A. Dingiziam, R.A. Bredthauer, C.E. Chandler,
J.A. Westphal et al.. New advancements in charge-coupled device technology:
subelectron noise and 4096 x 4096 pixel CCDs in Charge-Coupled Devices and
Solid State Optical Sensors. M.M. Blouke (Ed.). Vol. 1242. pp. 223 – 237.
International Society for Optics and Photonics. SPIE. (1990),
DOI: 10.1117/12.19452.
[128] J.R. Janesick. Ultra Low-Noise Charge Coupled Device. U.S. Patents
No. 5,250,824. (1993).
[129] D. Wen and P. Salsbury. Analysis and design of a single-state floating gate
amplifier in 1973 IEEE International Solid-State Circuits Conference. Digest of
Technical Papers. Vol. XVI. pp. 154–155. (1973),
DOI: 10.1109/ISSCC.1973.1155181.
[130] D. Wen. Design and operation of a floating gate amplifier. IEEE Journal of
Solid-State Circuits 9 (1974) 410.
[131] G. Fernandez Moroni, J. Estrada, G. Cancelo, S.E. Holland, E.E. Paolini and
H.T. Diehl. Sub-electron readout noise in a Skipper CCD fabricated on high
resistivity silicon. Experimental Astronomy 34 (2012) 43.
[132] J. Janesick. Scientific Charge-Coupled Devices. Vol. 83 of Press Monograph. SPIE
(2001), DOI: 10.1117/3.374903.
[133] W. Shockley and W.T. Read. Statistics of the Recombinations of Holes and
Electrons. Phys. Rev. 87 (1952) 835.
[134] P. Pichler. Intrinsic Point Defects, Impurities, and Their Diffusion in Silicon.
Computational Microelectronics. Springer Vienna. 1 ed. (2004),
DOI: 10.1007/978-3-7091-0597-9.
[135] S. Polonsky, M. Bhushan, A. Gattiker, A. Weger and P. Song. Photon emission
microscopy of inter/intra chip device performance variations. Microelectronics
Reliability 45 (2005) 1471.
[136] S. Tam and C. Hu. Hot-electron-induced photon and photocarrier generation in
Silicon MOSFET’s. IEEE Transactions on Electron Devices 31 (1984) 1264.
[137] D. Barton, P. Tangyunyong, J. Soden, A. Liang, F. Low, A. Zaplatin et al..
Infrared Light Emission From Semiconductor Devices in ISTFA 1996:
Conference Proceedings from the 22nd International Symposium for Testing and
Failure Analysis. International Symposium for Testing and Failure Analysis.
pp. 9–17. (1996), DOI: 10.31399/asm.cp.istfa1996p0009.
[138] S. Polonsky. Time-Resolved Emission Microscopy of Silicon Integrated Circuits.
APS News 14 (2005) .
100 BIBLIOGRAPHY
[139] H. Oluseyi, A. Karcher, W. Kolbe, B. Turko, G. Aldering, C. Bebek et al..
Characterization and deployment of large-format, fully depleted back-illuminated
p-channel CCDs for precision astronomy in Sensors, Systems, and
Next-Generation Satellites VIII. R. Meynart, S.P. Neeck and H. Shimoda (Eds.).
Vol. 5570. pp. 515 – 524. International Society for Optics and Photonics. SPIE.
(2004), DOI: 10.1117/12.566976.
[140] G.I. Cancelo, C. Chavez, F. Chierchie, J. Estrada, G. Fernandez-Moroni,
E.E. Paolini et al.. Low threshold acquisition controller for Skipper charge-coupled
devices. JATIS 7 (2021) 015001.
[141] W.A. Joye and E. Mandel. New Features of SAOImage DS9 in Astronomical
Data Analysis Software and Systems XII. H.E. Payne, R.I. Jedrzejewski and
R.N. Hook (Eds.). Vol. 295 of Astronomical Society of the Pacific Conference
Series. p. 489. (2003).
[142] V.C. Rubin, J. Ford, W. K. and N. Thonnard. Rotational properties of 21 SC
galaxies with a large range of luminosities and radii, from NGC 4605 (R=4kpc)
to UGC 2885 (R=122kpc).. The Astrophysical Journal 238 (1980) 471.
[143] G. Hinshaw, D. Larson, E. Komatsu, D.N. Spergel, C.L. Bennett, J. Dunkley
et al.. Nine-year Wilkinson Microwave Anisotropy Probe (WMAP) observations:
Cosmological parameter results. The Astrophysical Journal Supplement Series
208 (2013) 19.
[144] Aghanim, N., Akrami, Y., Ashdown, M., Aumont, J., Baccigalupi, C., Ballardini,
M. et al.. Planck 2018 results - VI. Cosmological parameters. A&A 641 (2020)
A6. Erratum: A&A 652 (2021) C4.
[145] S. Dodelson. The real problem with MOND. International Journal of Modern
Physics D 20 (2011) 2749.
[146] J. Cooley, T. Lin, W.H. Lippincott, T.R. Slatyer, T.-T. Yu, D.S. Akerib et al..
Report of the Topical Group on Particle Dark Matter for Snowmass 2021.
arXiv:2209.07426.
[147] M. Battaglieri, A. Belloni, A. Chou, P. Cushman, B. Echenard, R. Essig et al..
US Cosmic Visions: New Ideas in Dark Matter 2017: Community Report.
arXiv:1707.04591.
[148] D.G. Cerdeno. WIMPs: A brief bestiary in 4th Patras Workshop on Axions,
WIMPs and WISPs. pp. 9–12. (2009),
DOI: 10.3204/DESY-PROC-2008-02/cerdeno_david.
[149] J. Aalbers, D.S. Akerib, C.W. Akerlof, A.K. Al Musalhi, F. Alder, A. Alqahtani
et al.. LUX-ZEPLIN collaboration. First Dark Matter Search Results from the
LUX-ZEPLIN (LZ) Experiment. Phys. Rev. Lett. 131 (2023) 041002.
BIBLIOGRAPHY 101
[150] E. Aprile, K. Abe, F. Agostini, S. Ahmed Maouloud, L. Althueser, B. Andrieu
et al.. XENON collaboration. First Dark Matter Search with Nuclear Recoils
from the XENONnT Experiment. Phys. Rev. Lett. 131 (2023) 041003.
[151] Y. Meng, Z. Wang, Y. Tao, A. Abdukerim, Z. Bo, W. Chen et al.. PandaX-4T
collaboration. Dark Matter Search Results from the PandaX-4T Commissioning
Run. Phys. Rev. Lett. 127 (2021) 261802.
[152] D.S. Akerib, P.B. Cushman, C.E. Dahl, R. Ebadi, A. Fan, R.J. Gaitskell et al..
Snowmass2021 Cosmic Frontier Dark Matter Direct Detection to the Neutrino
Fog. arXiv:2203.08084.
[153] A.H. Abdelhameed, G. Angloher, P. Bauer, A. Bento, E. Bertoldo, C. Bucci
et al.. CRESST collaboration. First results from the CRESST-III low-mass dark
matter program. Phys. Rev. D 100 (2019) 102002.
[154] E. Armengaud, C. Augier, A. Benoît, A. Benoit, L. Bergé, J. Billard et al..
EDELWEISS collaboration. Searching for low-mass dark matter particles with a
massive Ge bolometer operated above ground. Phys. Rev. D 99 (2019) 082003.
[155] R. Agnese, T. Aralis, T. Aramaki, I.J. Arnquist, E. Azadbakht, W. Baker et al..
SuperCDMS collaboration. Search for low-mass dark matter with CDMSlite
using a profile likelihood fit. Phys. Rev. D 99 (2019) 062001.
[156] Q. Arnaud, D. Asner, J.-P. Bard, A. Brossard, B. Cai, M. Chapellier et al.. First
results from the NEWS-G direct dark matter search experiment at the LSM.
Astroparticle Physics 97 (2018) 54.
[157] R. Essig, G.K. Giovanetti, N. Kurinsky, D. McKinsey, K. Ramanathan, K. Stifter
et al.. Snowmass2021 Cosmic Frontier: The landscape of low-threshold dark
matter direct detection in the next decade. arXiv:2203.08297.
[158] E. Aprile, M. Alfonsi, K. Arisaka, F. Arneodo, C. Balan, L. Baudis et al..
Observation and applications of single-electron charge signals in the XENON100
experiment. J. of Phys. G 41 (2014) 035201.
[159] K. Irwin and G. Hilton.Transition-Edge Sensors in Cryogenic Particle Detection.
C. Enss (Ed.). pp. 63–150. Springer Berlin Heidelberg (2005),
DOI: 10.1007/10933596_3.
[160] Y. Hochberg, I. Charaev, S.-W. Nam, V. Verma, M. Colangelo and
K.K. Berggren. Detecting Sub-GeV Dark Matter with Superconducting
Nanowires. Phys. Rev. Lett. 123 (2019) 151802.
[161] L. Barak, I.M. Bloch, A. Botti, M. Cababie, G. Cancelo, L. Chaplinsky et al..
SENSEI collaboration. SENSEI: Characterization of Single-Electron Events
Using a Skipper Charge-Coupled Device. Phys. Rev. Appl. 17 (2022) 014022.
102 BIBLIOGRAPHY
[162] A. Aguilar-Arevalo, D. Amidei, X. Bertou, D. Bole, M. Butner, G. Cancelo et al..
Measurement of radioactive contamination in the high-resistivity silicon CCDs of
the DAMIC experiment. JINST 10 (2015) P08014.
[163] A. Aguilar-Arevalo, D. Amidei, D. Baxter, G. Cancelo, B.C. Vergara,
A. Chavarria et al.. Measurement of the bulk radioactive contamination of
detector-grade silicon with DAMIC at SNOLAB. JINST 16 (2021) P06019.
[164] A. Aguilar-Arevalo, D. Amidei, X. Bertou, M. Butner, G. Cancelo, A. Castañeda
Vázquez et al.. DAMIC collaboration. First Direct-Detection Constraints on
eV-Scale Hidden-Photon Dark Matter with DAMIC at SNOLAB. Phys. Rev.
Lett. 118 (2017) 141803.
[165] A. Aguilar-Arevalo, D. Amidei, D. Baxter, G. Cancelo, B.A. Cervantes Vergara,
A.E. Chavarria et al.. DAMIC collaboration. Constraints on Light Dark Matter
Particles Interacting with Electrons from DAMIC at SNOLAB. Phys. Rev. Lett.
123 (2019) 181802.
[166] A. Piers. Results from a 3.1 kg day Target Exposure with Skipper CCDs from
DAMIC at SNOLAB and other Beyond the Standard Model Searches with
Semiconductor Detectors. Ph.D. thesis. University of Washington. (2023),
URL: https://digital.lib.washington.edu/researchworks/handle/1773/49954.
[167] A. Aguilar-Arevalo, I. Arnquist, N. Avalos, L. Barak, D. Baxter, X. Bertou et al..
DAMIC, DAMIC-M, SENSEI collaboration. Confirmation of the spectral
excess in DAMIC at SNOLAB with skipper CCDs. arXiv:2306.01717.
[168] J. Angle, E. Aprile, F. Arneodo, L. Baudis, A. Bernstein, A.I. Bolozdynya et al..
XENON10 collaboration. Search for Light Dark Matter in XENON10 Data.
Phys. Rev. Lett. 107 (2011) 051301. Erratum: Phys. Rev. Lett. 110 (2013)
249901.
[169] E. Aprile, M. Alfonsi, K. Arisaka, F. Arneodo, C. Balan, L. Baudis et al..
XENON100 collaboration. Dark Matter Results from 225 Live Days of
XENON100 Data. Phys. Rev. Lett. 109 (2012) 181301.
[170] E. Aprile, J. Aalbers, F. Agostini, M. Alfonsi, L. Althueser, F.D. Amaro et al..
XENON collaboration. Light Dark Matter Search with Ionization Signals in
XENON1T. Phys. Rev. Lett. 123 (2019) 251801.
[171] P. Agnes, I.F.M. Albuquerque, T. Alexander, A.K. Alton, G.R. Araujo,
D.M. Asner et al.. DarkSide collaboration. Constraints on Sub-GeV
Dark-Matter–Electron Scattering from the DarkSide-50 Experiment. Phys. Rev.
Lett. 121 (2018) 111303.
BIBLIOGRAPHY 103
[172] D.W. Amaral, T. Aralis, T. Aramaki, I.J. Arnquist, E. Azadbakht, S. Banik
et al.. Constraints on low-mass, relic dark matter candidates from a
surface-operated SuperCDMS single-charge sensitive detector. Phys. Rev. D 102
(2020) 091101.
[173] C. Cheng, P. Xie, A. Abdukerim, W. Chen, X. Chen, Y. Chen et al.. PandaX-II
collaboration. Search for Light Dark Matter–Electron Scattering in the
PandaX-II Experiment. Phys. Rev. Lett. 126 (2021) 211803.
[174] R. Kolb, H. Weerts, N. Toro, R. Van de Water, R. Essig, D. McKinsey et al..
Basic Research Needs for Dark-Matter Small Projects New Initiatives: Report of
the Department of Energy’s High Energy Physics Workshop on Dark Matter.
United States (2018), DOI: 10.2172/1659757.
[175] R. Essig, J. Mardon and T. Volansky. Direct detection of sub-GeV dark matter.
Phys. Rev. D 85 (2012) 076007.
[176] P.W. Graham, D.E. Kaplan, S. Rajendran and M.T. Walters. Semiconductor
probes of light dark matter. Physics of the Dark Universe 1 (2012) 32.
[177] S.K. Lee, M. Lisanti, S. Mishra-Sharma and B.R. Safdi. Modulation effects in
dark matter-electron scattering experiments. Phys. Rev. D 92 (2015) 083517.
[178] R. Essig, M. Fernández-Serra, J. Mardon, A. Soto, T. Volansky and T.-T. Yu.
Direct detection of sub-GeV dark matter with semiconductor targets. JHEP 2016
(2016) 46.
[179] C.R. Chavez, F. Chierchie, M. Sofo-Haro, J. Lipovetzky, G. Fernandez-Moroni
and J. Estrada. Multiplexed Readout for an Experiment with a Large Number of
Channels Using Single-Electron Sensitivity Skipper-CCDs. Sensors 22 (2022) .
[180] M.S. Haro, C. Chavez, J. Lipovetzky, F.A. Bessia, G. Cancelo, F. Chierchie
et al.. Analog pile-up circuit technique using a single capacitor for the readout of
Skipper-CCD detectors. JINST 16 (2021) P11012.
[181] I.J. Arnquist, C. Beck, M.L. di Vacri, K. Harouaka and R. Saldanha. Ultra-low
radioactivity Kapton and copper-Kapton laminates. Nucl. Instrum. Meth. A 959
(2020) 163573.
[182] I.J. Arnquist, M.L. di Vacri, N. Rocco, R. Saldanha, T. Schlieder, R. Patel et al..
Ultra-low radioactivity flexible printed cables. arXiv:2303.10862.
[183] A.M. Suriano, S.M. Howard, C.D. Christofferson, I.J. Arnquist and E.W. Hoppe.
Developing radiopure copper alloys for high strength low background applications
in Low Radioactivity Techniques 2017. Vol. 1921 of American Institute of Physics
Conference Series. p. 080001. (2018), DOI: 10.1063/1.5019009.
104 BIBLIOGRAPHY
[184] R. Foot, H. Lew and R.R. Volkas. Electric-charge quantization. J. of Phys. G 19
(1993) 361.
[185] B. Holdom. Two U(1)’s and Epsilon Charge Shifts. Phys. Lett. B 166 (1986) 196.
[186] S. Foroughi-Abari, F. Kling and Y.-D. Tsai. Looking forward to millicharged dark
sectors at the LHC. Phys. Rev. D 104 (2021) 035014.
[187] K.J. Kelly and Y.-D. Tsai. Proton fixed-target scintillation experiment to search
for millicharged dark matter. Phys. Rev. D 100 (2019) 015043.
[188] J.L. Feng, F. Kling, M.H. Reno, J. Rojo, D. Soldin, L.A. Anchordoqui et al.. The
Forward Physics Facility at the High-Luminosity LHC. J. of Phys. G 50 (2023)
030501.
[189] P. Derwent, S. Holmes and V. Lebedev. An 800-MeV superconducting LINAC to
support megawatt proton operations at Fermilab. arXiv:1502.01728.
[190] L. Singh, J.W. Chen, H.C. Chi, C.-P. Liu, M.K. Pandey, H.T. Wong et al..
TEXONO collaboration. Constraints on millicharged particles with low-threshold
germanium detectors at Kuo-Sheng Reactor Neutrino Laboratory. Phys. Rev. D
99 (2019) 032009.
[191] P. Du, D. Egana-Ugrinovic, R. Essig and M. Sholapurkar. Sources of Low-Energy
Events in Low-Threshold Dark-Matter and Neutrino Detectors. Phys. Rev. X 12
(2022) 011009.
[192] G.F. Moroni, F. Chierchie, J. Tiffenberg, A. Botti, M. Cababie, G. Cancelo
et al.. Skipper Charge-Coupled Device for Low-Energy-Threshold Particle
Experiments above Ground. Phys. Rev. Appl. 17 (2022) 044050.
[193] R. Smith and S. Kaye. CCD speed-noise optimization at 1 MHz in High Energy,
Optical, and Infrared Detectors for Astronomy VIII. A.D. Holland and J. Beletic
(Eds.). Vol. 10709. p. 1070910. International Society for Optics and Photonics.
SPIE. (2018), DOI: 10.1117/12.2314261.
[194] M.M. Blouke, F.H. Yang, D.L. Heidtmann and J.R. Janesick. Traps and Deferred
Charge in CCDs in Instrumentation for Ground-Based Optical Astronomy.
L.B. Robinson (Ed.). pp. 462–485. Springer. (1988). ,
URL: https://link.springer.com/chapter/10.1007/978-1-4612-3880-5_45.
[195] D.J. Hall, N.J. Murray, A.D. Holland, J. Gow, A. Clarke and D. Burt.
Determination of In Situ Trap Properties in CCDs Using a “Single-Trap
Pumping” Technique. IEEE Transactions on Nuclear Science 61 (2014) 1826.
[196] P. Bilgi. Optimization of CCD charge transfer for ground and space-based
astronomy. Ph.D. thesis. California Institute of Technology. (2019),
URL: https://thesis.library.caltech.edu/11574/9/ThesisV4-6Final.pdf.
BIBLIOGRAPHY 105
[197] C. Hu. Modern Semiconductor Devices for Integrated Circuits. Prentice Hall
(2010).
[198] F. Chierchie, G. Fernandez Moroni, L. Stefanazzi, C. Chavez, E. Paolini,
G. Cancelo et al.. Smart-readout of the Skipper-CCD: Achieving Sub-electron
Noise Levels in Regions of Interest. arXiv:2012.10414.
[199] M. Sofo Haro. Development of new CCDs with non-destructive readout mode.
Fermi National Accelerator Laboratory. (2023),
URL: https://indico.fnal.gov/event/58707/.
[200] M. Sofo-Haro, K. Donlon, B. Burke, J. Estrada, F. Fahim and C. Leitz. Design
and Simulation of a Highly Sensitive Charge Detector With Nondestructive
Readout Mode for Fully Depleted Thick CCDs. IEEE Transactions on Electron
Devices 70 (2023) 563.
[201] A.M. Botti, B.A. Cervantes-Vergara, C.R. Chavez, F. Chierchie,
A. Drlica-Wagner, J. Estrada et al.. Fast Single-Quantum Measurement with a
Multi-Amplifier Sensing Charge-Coupled Device. arXiv:2308.09822.
[202] B. Parpillon and L. Rota. Design of a skipper CCD-in-CMOS active pixel sensor.
CPAD 2022. Stony Brook University. (2022),
URL: https://indico.bnl.gov/event/17072/contributions/70259/.