Particle detection through proportional scintillation in liquid xenon: The development of novel anode techniques for XAMS

(1)

Particle detection through

proportional scintillation in liquid

xenon

The development of novel anode techniques for XAMS

Author

L.I. VERVELD, BSc

Supervisors

Prof. Dr. A.P. COLIJN

Dr. S. BR ¨

UENNER

Secondary examiner

Dr. H.L. SNOEK

(2)

(3)

Liquid Xenon dual-phase detectors have set the most stringent limits on WIPM dark matter. However, the dual-phase technology necessitates the handling of both liquid and gaseous xenon, which results in many engineering and operational challenges. This work explores a single-phase detector principle, where only liquid xenon is used. In this single-phase model, the two signals

characteristic for dual-phase liquid xenon detectors are preserved.

For this work, the XAMS (Xenon AMSterdam) detector was transitioned from a dual-phase to a single-phase detector. In order to implement a single-phase detector, new anodes were designed.

For this, a new anode technique based on photomasks is explored with both electric field simulations and significant experimental work.

Four anodes were designed and produced. Tests have been performed prior to installation in XAMS. These tests were promising, and XAMS should be operational as a single-phase detector

shortly. From electric field calculations, it was predicted that a photon gain of O(100) photons per electron can be achieved for an electron multiplication of ∼ 2. This would provide an S2 signal size similar to the S2 signal in dual-phase operation. The shadowing effect from the anode wires is predicted to be significant. When we optimize for minimal electron multiplication, which is necessary for reduction of statistical uncertainties, the shadowing fraction is ≈ 25 (fs=O(25)),

which means 75% of photons emitted towrds the top of the detector are stopped by the wire. When optimizing for a minimal shadowing effect, (fs=O(70)). A different photomask-based

geometry is suggested which is predicted to have a significantly lower shadowing effect (fs=O(64)) when optimising for electron multiplication.

(4)

(5)

1 Introduction

Dual-phase liquid xenon (LXe) detectors have set the most stringent limits on the cross-section of WIMP dark matter with ordinary matter in recent history. These detectors are time projection chambers (TPC’s) filled with liquid xenon and, near the anode, a smaller region of gaseous xenon. LXe TPC’s are extremely low background, have excellent 3D position-reconstruction and have a dual-signal principle which provides a measurement on whether an interaction is an electronic recoil (background) or a nuclear recoil. However, these detectors also have some drawbacks. One of these drawbacks is the liquid level management: maintaining two phases of xenon within the TPC provides operational difficulties. This thesis is a research and development study into a novel detector principle based on dual-phase liquid xenon detectors. This novel detector principle will be referred to as single-phase dual-signal, as it only employs liquid xenon in contrast with a combination of gaseous and liquid xenon, while preserving the dual-signal principle. Specifically, this thesis is aimed at transitioning the small XAMS (Xenon AMSterdam) detector from a dual-phase liquid xenon detector toward a single-phase liquid xenon detector. For single-phase dual-signal LXe detectors higher electric fields are required in the near-anode region than in the dual-phase model. In order to achieve these higher fields, four novel anodes were designed and produced.

Chapter 2 will offer a brief discussion of the context for this study. Different types of evidence for dark matter, the WIMP dark matter model and different types of dark matter detectors are discussed. Chapter 3 presents a discussion of liquid xenon time projection chambers. It will start with a general overview of LXe dual-phase TPC working principles. Then the XAMS detector TPC will be described. Finally, an extensive discussion of the single-phase dual-signal detector principle as well as an overview of the current status of single-phase dual-signal research will be offered. Chapter 4 discusses the upgrade of XAMS to a single-phase dual-signal detector. This includes a discussion of the physical properties of the monolithic anodes that were produced for XAMS and a discussion the experimental work done to test and install one of these anodes. Chapter 5 will discuss in detail the electric field study performed during the design-phase of the novel anodes. The main result from this study is the predicted gains of photon per electron, the expected electron multiplication and the shadowing-fraction. Finally, chapter 6 will summarise the findings of this study and chapter 7 will offer recommendations for further research with the novel anodes as well as a recommendation for an optimized anode design.

(8)

2 Dark Matter

What constitutes dark matter is one of the most significant open questions in physics. There is strong evidence on cosmic scales as well as on the scale of galaxies and galaxy clusters that there is non-luminous matter, which makes up 85% of the matter content of the universe.[1] However, a dark matter particle has never been observed directly and it is therefore still unclear whether one of the dark matter candidates (hypothetical particles which would explain the observations) currently theorised is indeed present in the universe, and whether it explains all the evidence for dark matter. In this chapter will first discus some of the strongest evidences for the existence of dark matter. Subsequently there will be a short discussion of the WIMP, one of the most popular classes of dark matter candidates and the class of dark matter candidate most relevant to this thesis. Finally, an overview of the different types of dark matter experiments currently operating.

2.1 Dark matter: evidence

Evidence for dark matter is found on scales ranging between the scale of galaxies and the largest observable structure in the universe. While there is a wide variety of evidences, here only two types of evidences for both galaxy cluster scales and cosmic scales are presented here, for the sake of brevity.

2.1.1 Evidence from galaxy (clusters) Rotation curves

Historically, the first evidence found for dark matter is the rotational speed of galaxies within clusters [3]. This was then expanded to the rotational speed of constituent parts of individual galaxies. The expected rotational speed of galaxies within galaxy clusters and the rotational speed of constituent parts of galaxies are given by Kepler’s law:

vr(R) =

r

G · M (R)

R , (2.1)

where vr(R) is the rotational speed for a object in stable orbit around a mass of M(R), at a

distance of R from the center of the mass. We can plot the rotational speed of constituent parts of galaxies as a function of their distance from the center of the galaxy. This is called a rotation curve, with the rotation curve of the M33 galaxy shown in figure 2.1. Here is plotted in blue the measured rotational speed. The luminous mass of the M33 galaxy is calculated independently and the expected rotational speeds calculated with equation 2.1.1. These are subsequently plotted in yellow for the luminous mass from starts and in green for the luminous mass from stars and gas. As is clear, there is significant disagreement between the rotational speeds from the data in blue and the expected rotational speeds from the mass of the stars and gas. This can be explained by dark matter: large amounts of non-luminous matter present in galaxies and galaxy clusters. Including

(9)

Figure 2.1: Rotation curve of the M33 galaxy in blue. The rotational speed is plotted as a function of the radius away from the center of the galaxy. In yellow, the expected rotation curve if only stars are considered, in green, the expected rotation curve of stars and gas and in red a combination of stars, gas and the dark matter halo. Here the gray lines are proportional to R−1/2_{. Figure taken}

from [2].

this so-called dark matter halo in the total mass gives us the rotational speed plotted in red in figure 2.1. The dark matter distribution in galaxies and clusters is called a dark-mater halo as it has a mostly (flattened) spherical shape and extends significantly beyond the luminous constituent parts of the galaxy or cluster.

Gravitational lensing

General relativity tells us that massless particles are also affected by gravity: the trajectory of light can by changed by massive objects. This means that a massive object in between a luminous object and the point of measurement will affect the perceived shape and location of the luminous object from the point of view of the measurement. A diagram of this process is shown in figure 2.2. The angle by which it is bent is given by the following formula:

α = 4 · G · M

b · c2 , (2.2)

where α is the angle of deflection, G the gravitational constant, M the mass of the object (which is approximated as a point-mass here), b the distance from the point-mass (called the impact parame-ter) and c the speed of light [5]. We can measure the angle of deflection α and the impact parameter

(10)

Figure 2.2: Diagram of process of gravitational lensing. The light originating in the luminous galaxy is bent around the massive galaxy cluster (orange light rays) before reaching the observatories, which records the image of the galaxy lensed by the galaxy cluster. Image taken from [4].

b. This provides us with an excellent way to measure the total mass M. Measuring masses with gravitational lensing can provide a measurement of the mass of whole objects, such as galaxies and galaxy clusters, as well as the mass distribution within an object. Evidence from gravitational lensing confirms that 95% of matter in galaxy (clusters) consists of non-luminous (dark) matter [1].

2.1.2 Cosmic evidence Structure formation

The epoch of structure formation in the universe is essential to the structure we currently observe in the universe. In this epoch, parts of the universe which are slightly more dense significantly increased in mass density through gravitational pull. Through this process, the seeds were planted for objects such as galaxies and galaxy clusters. However if only ordinary matter is present in the universe, radiation (photons) dominate the interactions between particles in the universe. This would result in an outward pressure, as opposed to gravity, which pulls the particles together. The outward pressure of radiation, due to its large outward pressure, is predicted to have prevented the formation of the structures we currently see in the universe. In order to make our predictions match the current structure of the universe, we need a different type of mass to be present in this epoch. This type of mass would not be affected by radiation and therefore provide a larger inward gravity without contributing to the outward pressure.

(11)

radiation. It is therefore able to increase the mass density locally, such that gravitational potential wells form which attract more dark matter as well as ordinary matter. An important distinction can be made between cold dark matter and hot dark matter: the former is non-relativistic and the latter is relativistic. From simulations that were performed in the 1980’s, it became clear that hot dark matter has too high an initial momentum to create the structure seen today, as high momen-tum can cause the dark matter to escape gravity wells. This means that for the amount and scale of structure we see in the universe today, cold dark matter is needed.

In this epoch, most of the universe was relativistic. Therefore cold dark matter cannot have been in thermal equilibrium with the relativistic parts of the universe; we say dark matter is ”frozen-out”. This requirement provides us with information about the properties of dark matter. The following section will expand on this.

The cosmic microwave background

The cosmic microwave background (the CMB) is our earliest image of the universe. The first photons able to escape the hot, dense early universe at 380 000 years after the Big Bang can still be seen as the CMB. In figure 2.3, the power spectrum of the CMB is shown. Here ∆2

T is plotted

Figure 2.3: Temperature fluctuations of the cosmic microwave background (CMB) plotted as a function of the multipole moment (l), where the angular scale (θ) is given by θ ≈ 180°/l. In blue the data taken by the Planck telescope and in red the best fit of the Lambda-CDM model, which describes the main constituents and evolution of the universe. The residuals are plotted in the lower panel. Taken from [6].

(12)

as a function of the multipole moment l which are related according to the following equation:

∆2T ≡

l(l + 1) 2π ClhT i

2_, _(2.3)

where Cl is the correlation function (a measure of δT/T, the temperature difference) for a given

angular scale [6]. What is measured in a power spectrum is the similarity temperatures of the CMB at different scales, with lower multiple moments corresponding to larger angles of the sky (θ ≈ 180°/l). The temperature of the CMB corresponds to the density of the specific regions the photons originate from, with hotter regions corresponding to higher density regions. As previously described, the early universe was dominated by gravity from baryons and dark matter, and pressure from radiation. The opposed forces of gravity (inward) and pressure (outward) create an oscillation: if the region is becoming denser due to gravity, the pressure will build up until the resultant force is reversed and the region will become less dense. These oscillations are shown in the peaks: those regions which reached maximum density once, are represented in the first peak. Those that From the power spectrum, many different measurements can be done. For example, from the angular scale of the first peak, it can be concluded that the universe is flat. The density of dark matter in the universe can also be determined from the CMB power spectrum. The power spectrum shown in figure 2.3 measured the energy content of the universe to be the following: ∼ 69% dark energy, ∼ 26% dark matter and ∼ 4% baryonic matter.

2.2 Dark matter candidates

A wide variety of dark matter candidates has been put forward, ranging from axions to dark stars and primordial black holes [1]. Here, we will focus on the Weakly Interacting Massive Particle, the WIPM, as it is one of the most popular candidates, and the main physics target of LXe dark matter detectors.

As mentioned in the previous section, evidence points towards dark matter freezing out before the structure-formation epoch of the universe. In order for this to happen, the dark matter would self-annihilate before the structure-formation epoch such that only a relic abundance is left. This relic abundance would be roughly equal to the abundance of dark matter today. If dark-matter particles self-annihilate at a cross-section roughly equal to the weak interaction (σv ≈ 10−26 cm3_/s)

[7], the relic abundance is indeed the same as the abundance observed today. This is called the ”WIMP-miracle”, as simply requiring a particle to freeze out with a self-annihilation roughly equal to the weak force gives us particles with an abundance which is predicted for dark matter from the aforementioned evidence. We call this type of particle the Weakly Interacting Massive Particle, the WIMP.

There are many different types of WIMP possible. For example, a sterile neutrino (which means it does not oscillate as do the electron, muon and tau neutrino) with a mass of a few GeV would satisfy the above conditions for a WIMP. Another popular WIMP model is the lightest particle predicted by a simple supersymmetric extension of the standard model: the LSP particle.

(13)

2.3 Dark matter experiments

Three main types of dark matter detection are currently in use: accelerators, indirect detection and direct detection.

Accelerators

Accelerators attempt to create dark matter in extremely high energy particle collisions. The Large Hadron Collider, for example, collides protons with a Center-of-Mass energy of 13 TeV. Particles are created in the interaction of two of these relativistic protons. The particles created may be dark matter particles, which could be identified as missing energy in the collision.

Indirect detection

Indirect detection is aimed at observing dark matter and anti-dark matter annihilation. This process is predicted to take place in regions of high density of dark matter and would provide astronomical evidence of dark matter. The annihilation can be measured by gamma rays, neutrino’s or radio waves created in the annihilation. This process is dependent on the density squared, and is therefore searched for in some of the densest regions of space. Searching in this region, however, introduces significant background as these dense regions have many possible sources of backgrounds.

From indirect detection, limits may be set on the self-annihilation cross-section of dark matter.

Direct detection

Direct detection dark matter experiments try to detect a collision of a dark matter particle with an ordinary matter particle. As WIMPs don’t carry electric charge, WIMPs would interact with atoms through elastic scatters of the nucleus. The WIMP scattering rate of a nucleus is given by:

dR dEnr = ρ0M mNmχ Z vesc vmin vf (v) dσ dEnr dv, (2.4)

where Enr is the nuclear recoil (NR) energy, ρ0 the local dark matter density, M the target mass

of the dark matter detector. The mass of the nucleon is given by mN, mχ is the WIMP mass, σ

the WIMP-nucleon scattering cross-section, f(v) the WIMP velocity distribution (normalized) and v the WIMP velocity, with vminthe minimum WIMP velocity necessary for a scatter and vesc the

WIMP velocity at which it is able to escape from the gravitational pull of the milky way [8]. From this equation, the expected number of events can be calculated using the detector properties and the predicted WIMP properties. A 100 GeV/c2 WIMP with a cross section of 10−44 cm2 has a rate of 18 events per 100 kg of xenon per year for a 5 keV recoil energy and 8 events per 100 kg of xenon per year for a 15 keV recoil energy [9]. As is clear from the above assumed cross section, WIMPS are predicted to have a very low cross-section to interact: (σ = 10−41− 10−51 _cm2_{) [8]. In}

order to observe such rare events, an experiment with very low background is required. Therefore the direct detection experiments are typically in underground laboratories, such as the Gran Sasso National Laboratory underneath the Gran Sasso mountain range in Italy or SNOLAB, which is

(14)

located in an active mine near Ontario. Typical backgrounds for direct detection experiments are radiogenic neutrons and electrons (neutrons and electrons created by radioactive atoms inherent to the detector materials), neutrino’s and muon-induced neutrons [10].

A wide range of different direct detection experiments are operated with different detection media. These are each optimized to measure either heat, charge or light yield from an interaction of a dark matter particle with the detector medium. Currently, the most sensitive detectors for WIMP-nucleon interactions are liquid xenon (LXe) dual-phase detectors, which read out both a charge and light signal from interactions in their LXe target. From direct-detection, limits may be set on the cross-section of dark matter with standard model particles. The limit published by XENON1T, a state-of-the-art LXe dual-phase detector, is shown in figure 2.4. Here the spin-independent cross-section is plotted as a function of the WIMP mass. The data was consistent with background models and therefore an upper limit was published. This upper limit is presented as a 90% confidence limit, shown in black in figure 2.4. The 90% confidence limits are also shown for the LUX and PandaX-II experiments. These results were both published in 2017. The confidence limit is calculated using the C.L. method, which differs from the median of the 1σ and2σ confidence intervals, shown respectively in green and yellow. These intervals are dependent on background fluctuations, while the 90% confidence limit is not. These differences are within a 2σ statistical uncertainty [11].

Figure 2.4: Limits set on spin-independent cross-section and mass of WIMP dark matter by the PandaX-II, LUX and XENON1T experiments. Here the spin-independent WIMP-nucleon cross-section is plotted as a function of the WIMP-mass. The 90% confidence limit set by XENON1T in black, with 1σ and 2σ sensitivity band in green and yellow. In red and blue the upper limits as measured by the LUX and PandaX-II collaborations. At the top, the same plot normalised to the median of XENON1T’s 1σ sensitivity band. Taken from [12].

(15)

3 Liquid xenon time projection chambers

Liquid xenon dual-phase detectors observe interactions in a target mass of liquid xenon. On top of the liquid xenon, there is a small region of gaseous xenon, which is used to convert electrons to photons, which can be measured. The largest LXe dual-phase detectors currently in use are employing a target mass of up to several tonnes of liquid xenon. LXe dual-phase detectors measure two distinct signals from each interaction, which provide the energy, position and information on the type of interaction. In this chapter a description of the general principles of dual-phase liquid xenon detectors is provided as well as a description of XAMS, the ”Xenon AMSterdam” LXe dual-phase research and development detector. The transition of XAMS from a dual-dual-phase TPC to a single-phase TPC is the main goal of this thesis and will be discussed in the final section of this chapter.

Figure 3.1: Diagram of LXe dual-phase TPC. In the middle is visible a cylindrical volume with LXe, with the blue disk at the top signifying the border of the GXe above and LXe below. The electrode is circular with grid-structure and depicted in gray. Photosensors gray blocks on top and bottom of TPC. On the left a small graph with signal strength of S1 and S2 interation plotted against the time difference on the y-axis. Taken from [13].

(16)

3.1 Detection principle of a dual-phase liquid xenon TPC

Dual-phase liquid xenon detectors are designed to observe interactions of particles within a liquid xenon target mass. A particle traveling through the detector can interact with a xenon atom, which provides two distinct signals, both of which are essential to the operation of dual-phase TPC’s. The first signal (S1), a light flash made up of 178 nm photons, comes directly from the interaction point and is caused by scintillation of xenon atoms. The S1 is indicated in figure 3.1 and can occur anywhere in the LXe volume. The proportional signal (S2) is caused by ionization of xenon atoms. In figure 3.1, these electrons are shown in red, above the S1 signal. The S2 is shown to occur between the gate electrode and anode. On the left of figure 3.1 a graph is shown of the relative size and timing of the S1 and S2 signals. The S2 is significantly larger and has a width of around ∼ 1 µs as opposed to a ∼ 29 ns width for the S1. The S2 arrives later than the S1, with delay times from a few µm up to ∼ 1 ms, depending on the TPC size and drift field [2]. These are freed from xenon atoms at the interaction point and drift upward through the liquid xenon by means of an electric drift field. The electrodes which create this electric field are shown in light grey on the top and bottom of the cylindrical TPC in figure 3.1. Near the top of the liquid xenon, the electrons pass through a grounded electrode, called the gate, which marks the end of the drift field. The electrons then enter the higher amplification-field between the gate electrode and the positively charged anode. There, the electrons leave the liquid xenon and are extracted into the gaseous xenon, where they will result in a proportional signal, also made up of 178 nm photons. The S1 and S2 signals are recorded with photomultiplier tubes (PMTs), state-of-the-art detectors optimized to be able to photons with minimal energy; O(keV). The working principle of a PMT is pictured in 3.2. An incident photon will pass the quartz class and hit the so-called photocathode. The photocathode can subsequently release an electron through the photoelectric effect. This electron is then accelerated towards the first dynode, through which it generates enough energy to create more electrons. After being accelerated towards multiple dynodes as pictured in figure 3.2, the electron avalanche is large enough be be read out as a charge measurement. The main quality evaluation of PMT’s is the quantum efficiency, the probability for a single photon to release an electron into the PMT:

QE = Ne/Nγ, (3.1)

where QE is currently ∼ 30% for state-of-the-art PMT’s in case of VUV photons. Concretely, the Hamamatsu R11410-21 PMT’s used for XENON1T have been measured to have a quantum efficiency of QE = (34 ± 3)% [14]. The PMT’s are shown as gray cubes on the top and bottom of the TPC in figure 3.1, which is similar to the arrangement of PMT’s in XENON1T. These are typically arranged close to one another at the top and bottom of the detector in order to maximise light detection. On the left hand side is a graph where the relative sizes and timings of the S1 and S2 signal are shown.

The following sections will offer a more detailed explanation of these processes and detector parts. Systems necessary for the operation of a LXe dual-phase detector, such as cryogenics, gas circulation

(17)

Figure 3.2: Diagram of the working principle of a photomultiplier tube (PMT). On the left the incident photon which has a probability of ∼ 30% to cause an electron avalanche, which is shown to happen on the dynodes. On the right the voltage supply and readout are pictured. Taken from [15].

and purification, liquid level management and the data acquisition system (DAQ) will not be discussed here.

3.1.1 Microphysics of the primary (S1) and proportional (S2) scintillation

A signal is created when a particle which is moving through the liquid xenon target has an inter-action with a xenon atom. This interinter-action can be either a nuclear recoil (NR) or an electronic recoil (ER). In case of a nuclear recoil, some of the energy of the initial particle is transferred to the xenon nucleus. In case of an electronic recoil, the xenon atom is instead ionized, which results in an electron moving through the liquid xenon. The xenon nucleus or the freed electron will deposit its energy, giving rise to a detectable signal, in three ways: it will excite xenon atoms along its path, it will ionize xenon atoms along its path and it will have elastic collisions with other xenon atoms. After a number of interactions, the xenon nucleus or electron is stopped as it has lost its initial momentum. A schematic of these three types of interaction is shown in figure 3.3.

The energy deposition through heat does not provide a signal that can be observed by the PMT’s. The second of these three possible interactions, excitation of xenon atoms, will exclusively con-tribute to the S1 signal. The excited xenon atom will form a bound state with another xenon atom in ground-state to form a so-called eximer (Xe∗+ Xe → Xe∗2). This can be done in either a singlet

(spin) state or a triplet (spin) state. Subsequently, this eximer will decay, the singlet state with a lifetime of ∼ 4 ns and the triplet state with a lifetime of ∼ 22 ns. When an eximer decays, it emits

(18)

Figure 3.3: Schematic of relevant particle interactions to S1 and S2 signal at initial interaction point. Figure taken from [2].

a 178 nanometer UV-photon. These photons give rise to the primary signal, the S1. The photons are not re-absorbed by other xenon atoms, as the excitation level of the eximer state is of lower energy than the excited state of a single xenon atom. This means that the LXe is transparent to the produced scintillation light.

The last energy deposition method is the ionization of a xenon atom, creating a positively charged ion and negatively charged free electron. The free electron is able to drift upward toward the pos-itively charged anode. Those electrons can contribute to the proportional scintillation (S2) signal. However, it is also possible for the free electrons to recombine with local ionized xenon atoms be-fore they are able to drift out of the region containing these ions. The ions have already formed charged eximer states Xe+_{+ Xe → Xe}+

2 with ground state xenon atoms. This charged eximer

state can recombine with a free electron (Xe+₂ + e−_{→ Xe}∗∗_{+ Xe), creating a doubly excited xenon}

atom and a ground state xenon atom. Subsequently the Xe∗∗ relaxes and forms an eximer state (Xe∗_{+ Xe → Xe}∗

2), which will contribute to the S1 signal just like the eximers created as described

in the primary signal section. If the electron escaped recombination it is drifted upward through the liquid xenon. During the drift time, there is a small probability for the electron to combine with an impurity in the LXe. This electron would therefore not contribute to the S2 signal. Within the active volume of the TPC, the electron will experience a homogeneous electric field, which is created by the cathode on the bottom and the gate electrode (held at ground potential) on top of the active volume. On the side of the TPC field shaping rings are placed in order to improve the homogeneity of the electric field inside of the active volume even further. The anode is typically O(10) millimeters above the gate, such that the electric field between the gate and anode (amplifi-cation field) is much higher than the drift field, between the cathode and gate. The amplifi(amplifi-cation

(19)

field is typically ∼ 10 kV/cm, while the drift field is typically ∼ 0.1 kV/cm. The liquid level is in between the anode and gate mesh. When an electron crosses the liquid level it is able to excite xenon atoms in the gas phase. These xenon atoms will subsequently emit scintillation light, ac-cording to the same principle as the S1 scintillation light; resulting in the S2 signal. This process is called proportional scintillation. The number of photons created through proportional scintillation is governed by the following equation:

Nγ = α · Ne·

E p − β

· p · d , (3.2)

where Nγ is the number of photons created, Ne is the number of electrons being accelerated

through the medium, α is the amplification factor in [(ph/e)/kV] and E is the electric field strength in [kV/cm. β is the electric field threshold for proportional scintillation (the minimum electric field strength for which photons are created), p the gas pressure and d the distance travelled by the electron [16]. From this equation, it is clear that the number of photons is linearly dependent on the number of electrons as well as the distance traveled. This is why this process is called proportional scintillation: from the number of photons a very precise measurement can be done of the number of initial electrons. Typically each electron gives rise to O(100) S2 photons in a dual-phase detector [2]. This means that the S2 signal is typically larger than the S1 signal. In addition to that, the S2 signal happens close to the top array of photon-sensors. Therefore the hit pattern in the top-array is very sensitive to the position of the S2 signal, and thus is used for position-reconstruction.

Property Value xenon Atomic number Z 54 [17] Mean atomic weight A 131.30 [17] Triple point temperature 161.3 K [17]

Triple point pressure 0.805 bar [17] Triple point density 2.96 g · cm−3 _[17]

Dielectric constant 1.95 [17] Path length 100 keV photon 0.18 cm [2]

Path length 1 MeV photon 5.9 cm [2] Path length 1 MeV neutron 10.6 cm [2] Path length 10 MeV neutron 15.9 cm [2]

Table 3.1: Selected properties of xenon: mass, triple point temperature, pressure and density and typical path length in liquid xenon for selected particles.

(20)

3.1.2 Liquid Xenon properties

Xenon is an excellent detector medium for dark matter detector for several reasons. Some essential properties of xenon can be found in table 3.1. The phase diagram of xenon is shown in figure 3.1.Aat atmospheric pressure, liquid xenon has a temperature in between 162 K and 165 K and a density of ∼ 3 g/cm3_{[17]. LXe detectors are typically oprated at around 2 bar.}

The high density of LXe in combination with its high atomic mass provides a very high

stopping-Figure 3.4: Phase diagram of xenon. Here the pressure is plotted on the y-axis and the temperature on the x-axis. The area’s in the plot signify whether xenon is in solid, liquid or gaseous phase for a given temperature and pressure combination. Figure taken from [17]

power. Typical path lengths for some particles are shown in table 3.1. These short path-lengths allow for ”self-shielding” of liquid xenon. The events which occur in the outer region of the de-tector volume are not used for the scientific analysis, improving the signal to background ratio significantly. This cutting of events from the outer regions of the detector is called fiducialisation. The fiducialisation makes liquid xenon detectors some of the lowest background detectors currently in operation.

Xenon is also produces more electrons and photons for a given energy deposition than other liquid noble gasses. This can be expressed with the so-called W-value:

W = E0 Ni

= Ei+ Ex(Nex/Ni) + , (3.3)

where E0is the energy deposition in [eV], Niis the average number of electron-ion pairs or photons

produced [17]. Ei is the average energy per electron pair, Nex the number of excited atoms, Ex

the average energy to excite an atom and the average kinetic energy of sub-excitation electrons. As can be seen in table 3.2, LXe has a lower W-value for each of the processes shown. This

(21)

Parameter LXe LAr Wph(α) 17.9 eV 27.1 eV

Wph(β) 21.6 eV 24.4 eV

We−ion 15.6 eV 23.6 eV

Table 3.2: W-values for liquid xenon and liquid argon. A lower W-value results in a higher produc-tion of photons for Wph and electron-ion pairs for We−ion. Values taken from [17].

means it creates more photons and more electrons per given energy deposition. In addition to the aforementioned advantages, xenon also has a high intrinsic radio-purity. This means that it has few unstable isotopes 124Xe and136Xe, which do not offer a significant background.

Lastly, there are some properties of LXe which make operation easier. Xenon is in liquid phase at around 160 K, which is approximately half the cooling of what would be required for liquid argon. Liquid xenon is also very dense, which means that the detector can have a moderate size with a high total target mass.

It should be noted that LXe is much more expensive than LAr and that, over all, LAr is also very well suited to dark matter experiments.

3.1.3 Analysis of S1 and S2 signals in LXe TPCs

The two signals S1 and S2, taken together provide a lot of information about the original interaction. Figure 3.5 shows the size and timings of a typical S1 and S2 signal in XENON1T, with the amplitude in P.E./10ns plotted as a function of the time-of-arrival. The round figures show the total number of hits for each PMT in the top and bottom array. The energy of the original interaction is extracted from both the S1 and the S2 signal size, which is found by integrating the amplitude over the total time of the signal. The total energy of an interaction (E) is given by :

E = W (nγ+ ne−), (3.4)

where W is as in equation 3.1.2, nγ the number of created photons and ne− the number of created

electrons [2]. In order to use equation 3.1.3 for an energy measurement, it can be re-written to include measurable quantities:

E = W cS1 g1 +cS2b g2 , (3.5)

where cS1 and cS2b are the S1 and S2 signal sizes, corrected for position dependent effects, and

g1 and g2 are the primary and secondary scintillation gain [2]. The primary and secondary

scin-tillation gain are conversion factors between the energy of the interaction and the amount of P.E. observed for the S1 and S2 signal. If one photon causes a P.E. in a PMT 10% of the time, we say g1= 0.1 P.E./γ. The S1 and S2 are corrected for position dependent effects on the scintillation

(22)

Figure 3.5: Diagram with typical data from a single event in XENON1T. The four circles show the the PMTs on the top and bottom array, with the number of p.e. signified in the color of each PMT, with the waveform of each PMT superimposed. On the x,y axis is the cumulative waveform from all PMTs. The S2 signal is clearly much larger (in nr of p.e.) than the S1. The S2 top array hit pattern shows a peak which is used for position reconstruction. Taken from [18]

(23)

gains and therefore denoted by cS1 and cS2b. The ”b” in cS2bdenotes the fact that only the

bot-tom array is used for the energy reconstruction from the S2, as the P.E. on the top array is very dependent on the position of the interaction and is therefore not a good measure of the energy. An example of this is that evens which occur lower in the detector have a relatively lower number of P.E. for the S2, as more electrons are lost to impurities during drift time, affecting the total size of the S2. The interaction position (in the lateral plane) is extracted from the S2 with an algorithm, which calculates the most likely position of the interaction from the hit pattern of the S2 on the top array. This algorithm typically uses a Light-Collection Efficiency (LCE) map of the detector to perform a likelihood fit or can also be a machine-learning algorithm trained on Monte Carlo simulations. The S2 pattern on the top array is visible on the top array in figure 3.5, where a localised peak in P.E. gives the information used to reconstruct the position of the original interaction. As the drift velocity is constant, the interaction position in the transverse plane (depth of interaction) can be determined from the drift time. As the electron drift velocity is constant in the active volume of the detector, the delay between between the S1 and S2 signal can be converted to the z-position of the interaction. Finally, an important measurable quantity is found in the ratio of S1 signal size to S2 signal size. As said before, one of the ways a particle can deposit energy is through heat (in addition to scintillation and ionization). Nuclear recoils lose a significantly larger fraction of their energy to heat than electronic recoils. This means that the S1 and S2 signal sizes are significantly reduced for NR interaction. In addition to that, an NR interaction is more localised. Therefore the ionizations occur closer to one another, making recombination of free electrons more likely. This means that in NR interactions, the size of the S2 with respect to the size of the S1 will be reduced. The ratio of S2 size and S1 size is used to distinguish between NR and ER interactions.

3.2 XAMS

XAMS is a small liquid xenon detector in Amsterdam (Xenon-AMSterdam) which is used for research and development studies for the XENON collaboration. It consists of a TPC, a cryogenic system and a gas system. This section will first present a detailed description of all the major parts of the XAMS TPC. Subsequently, a short discussion of the gas and cryogenic system will follow.

3.2.1 XAMS TPC

The XAMS Time Projection Chamber (TPC) can be seen in figure 3.6. XAMS is a cylindrical detector consisting of PTFE disks which hold together electrodes, photon-sensors and other small sensors (pt100s and a level meter) that make up the entirety of XAMS. The TPC is filled with liquid xenon, which results in a total active volume of around 434 grams of liquid xenon.

(24)

XAMS TPC: electric field components

The electric fields in XAMS are created by five electrodes: the cathode, gate and anode create two regions of electric field: the drift field between the cathode and gate and the amplification field between the gate and anode. The two other electrodes are the bottom screening electrode and the top screening electrode. These shield the photon-sensors at the top and bottom from the electric field created by the anode and cathode.1 All electrodes consist of stainless steel meshes, which have been chemically etched. Each mesh consists of wires of 150 µm by 150 µm, with a spacing between wires of 2.45 mm. The five electrodes are shown in figure 3.6 in blue-green. Around the active volume of the TPC copper rings are placed, called field shaping rings. These are held at the electric potential corresponding to the electric potential at that height in the drift field. This has the purpose of making the electric field inside the drift region more uniform. The rings are held at potential by connecting them in series, with a resistor of 1GΩ in between, with the bottom ring

1_{It should be noted here that the top PMT has recently been replaced by eight silicone photonmultipliers (SiPMs),}

which do not require shielding from the electric field.

Figure 3.6: Two-dimensional drawing of the cross-section of the XAMS TPC. Electrodes in blue-green, field shaping rings in dark red. Potentials for the electrodes are typical for dual-phase operation. Liquid xenon in green, with the liquid level between the gate electrode and the anode. In grey the PTFE disks. Surrounding the liquid xenon is the PTFE inner cylinder. In blue the photon sensors. Adapted from [2].

(25)

Figure 3.7: Diagrams of position of SiPMs in XAMS. LEFT: Cross-section of top half of XAMS TPC. LXe is shown in blue, with the liquid level between the gate electrode and anode. On top, the SiPM holder is shown in yellow, with the photo-sensitive areas of the SiPMs in green. RIGHT: The view of the SiPM holder from the top of the TPC downward. SiPM 20 is currently broken. Images taken from [19].

connected to the cathode and the top ring connected to the gate electrode. The field shaping rings are visualised in figure 3.6 in dark red.

XAMS TPC: PTFE structure

PTFE (Teflon) disks are used as the main structure of XAMS. They hold the electrodes, field shaping rings and sensors in place. PTFE is a very advantageous material, as it an excellent elec-trical insulator. In addition to that, its dielectric constant is similar to that of liquid xenon. This means that the PTFE disks will be able to withstand the high electric fields created in the TPC, without becoming conductive and without deforming significantly. In addition to that, PTFE has been chosen for the inner cylinder, which surrounds the active volume, because it also has excellent UV-reflection properties. The disks and inner cylinder are pictured in figure 3.6 in gray.

XAMS TPC: Photon-sensors

XAMS has two types of photon-sensors: a single photon multiplier tube (PMT) at the bottom and an array eight silicon photonmultipliers (SiPM’s) at the top of the TPC. PMT’s are generally

(26)

used in liquid xenon detectors as they have a significantly lower dark-count than currently available SiPMs. The dark-count rate is a measure of the number of P.E. registered without there being a signal photon to cause this [19].

The bottom PMT in XAMS, a type R6041-406 PMT from Hamamatsu, has a quantum efficiency of 30% for 178 nm photons [15].

The eight silicone photomultipliers (SiPM’s) are of type VUV4 S13370-3025CN. These detectors have a quantum efficiency of 24% for 178 nm wavelength which hits the sensitive area [19]. Position reconstruction can be performed in XAMS by analysing the hit patterns in the seven SiPM’s. These can be seen in figure 3.7.

XAMS TPC: pt100s and level meter

The XAMS TPC has two types of additional sensors to assist in the operation: the level meter and the pt100 temperature sensors. The level meter is a stainless steel cylindrical capacitor which measures the height of the liquid level. The level meter fills up with LXe as the level rises. It is able to measure the height because the dielectric constant of LXe is higher than GXe, which means the capacitance will increase linearly with the height of the liquid level.

There are also temperature sensors mounted between some of the PTFE disks. These sensors per-form a rough level measurement during the filling of XAMS with LXe, as the temperature will drop as soon as the temperature sensor is immersed in LXe. The pt100 temperature sensors measure temperature with a resistance which is directly proportional to the temperature and can therefore be calculated:

RT = 100Ω · (1 + 3.9 · 10−3T − 5.7 · 10−7T2− 4.2 · 10−12(T − 100)T3) , (3.6)

where RTis the resistance at temperature T for temperatures below 0°C. For temperatures above

0°C, a slightly different equation, which does not include the (RT∝ T3) is required.

3.2.2 Gas and cryogenic system

A LXe detector such as XAMS requires an extensive cooling (cryogenic) and gas system.

When not in use for XAMS, the xenon is stored at room temperature. The liquid xenon condenses on a so-called cold finger, from which it drips into the XAMS TPC in liquid form. This copper cold finger is cooled with a Iwatani PDC08 pulse tube refrigerator (PTR) down to −90°C. The gaseous xenon will condense onto this cold finger and drop into the TPC below. The temperature in XAMS is maintained with an isolation volume: XAMS is enclosed in two stainless steel cylinders. In between these cylinders, there is a volume which is pumped down to a pressure of 3 · 10−7 mbar [2]. This prevents conductive and convective heat exchange between the XAMS lab and the XAMS detector. A mylar foil coated with aluminium foil is also wrapped around the inner stainless steel cylinder in order to reduce radiative heat transfer. In case the cooling system would fail however, there is emergency liquid nitrogen cooling available during LXe operation of XAMS for approximately two

(27)

hours. Additionally, there is a so-called burst-disk, which will break at a pressure of 4 bar inside the TPC, letting the xenon gas escape and preventing a pressure build-up inside the detector. It is possible to recirculate the xenon through the SAES MonoTorr PS3-MT3-R-2 gas purifier during XAMS operation. The gas purifier removes impurities, such as residual gas or oxygen, from the liquid xenon. The recirculation is performed by the EMP MX-808ST-S diaphragm pump. The gas system also includes a radon monitor to measure the amount of radon (the main background for state-of-the-art LXe detectors) present in the xenon during XAMS operation. The radon detector is used for radon calibration measurements, for which a dedicated radon insertion system is present.

3.3 Single-phase TPC

The dual-phase detector design described in the previous sections is not the only way to construct a liquid xenon detector which measures both the S1 and S2 signal. Proportional scintillation, which gives rise to the S2 signal, has been verified in liquid xenon as well as gaseous xenon. Therefore it is in principle possible to transition from a dual-phase detector to a single-phase detector, where only liquid xenon is present. The S2 signal would then occur in the liquid phase.The single-phase model has several significant advantages; the liquid level has to be very precisely managed in a dual -phase detector, as a small change in the distance between the liquid level and the anode can result in a large change in the S2 signal size. It is therefore very valuable to investigate the possible applications of single-phase dual-signal LXe detectors. The single-phase model also has

Figure 3.8: LEFT: Picture of the XAMS set-up. On the right, in green, the TPC surrounded by the isolation volume. On top, in blue, the cryogenic system and on the left, in red, the gas system. Adapted from [19]. RIGHT: Picture of XAMS TPC with isolation volumes removed.

(28)

several significant disadvantages when compared to the dual-phase model. The most significant of these are the following: the reduced S2 signal size, which will be expanded on in the following sections. The following sections will present a discussion of the physics processes of proportional scintillation in liquid xenon and a discussion of the current status of research into single-phase dual-signal detectors. The last section discusses some of the benefits and challenges of a single-phase versus a dual-phase model.

3.3.1 Proportional scintillation in LXe

Proportional scintillation is the scintillation that occurs through a particle exciting the atoms of a medium it is being accelerated through. The word ”proportional” here refers to the fact that there is a number of photons created which is directly proportional to the number of particles which are being accelerated through the medium. The number of photons measured can therefore be con-verted to provide a measure of the original number of accelerated particles. In our case, this is an electron moving through xenon, which gives rise to the S2 signal. These xenon atoms subsequently fall back to the ground state, emitting the photons which make up the proportional scintillation signal. The scintillation process is the same as described in section 3.1.1. The electric field provides the force which moves the electron through the xenon. Hereby the electron gains momentum from the electric field and loses momentum through elastic scatters with xenon atoms. In a constant electric field, the electron will move at an average speed; here the force from the electric field and the average force from elastic scatters are in equilibrium. The higher the electric field, the higher this speed. There is a minimum electric field strength for proportional scintillation to occur, as the electron needs to have enough kinetic energy to excite a xenon atom. Therefore we can define a threshold electric field strength for proportional scintillation.

As this threshold is a function of the probability of an elastic scatter with a xenon atom, it is different for liquid xenon than for gaseous xenon, which is much less dense. Typically in dual-phase detectors an electric field of 10 kV/cm is applied in the gaseous xenon, which produces O(100) photons per electron for typical LXe dual-phase detectors, where the anode-gate gap is a few mil-limeters long. Proportional scintillation has been measured in liquid xenon as well, with the most precise measurement of the electric field threshold being 412+10₋₁₃₃ kV/cm, which is more than an order of magnitude higher than the electric field used for proportional scintillation in gaseous xenon [20].

This electric field strength threshold poses some challenges to application of single-phase dual-signal LXe detectors. Creating an electric field of O(500 kV/cm) is likely to lead to sparking with the currently used electrodes in LXe dual-phase detectors. An effective way to circumvent this problem in LXe single-phase detectors is to create the electric field of O(500 kV/cm) only in the region near the anode wires (near-wire region). In this region, the electric field has a 1/R dependence on the distance from the center of the wire (R). If the anode has wires of O(10 µm) diameter, an electric field sufficient for proportional scintillation can be achieved within a distance of approximately the

(29)

wire radius away from the surface of the wire under normal LXe detector operating conditions. A schematic of the main differences between a dual-phase detector and the described single-phase detector is shown in figure 3.9. Here the anode and gate electrode wires are signified by the black squares, with the electron paths in white (outside of proportional scintillation region) and orange (within proportional scintillation region). The thin-wires single-phase model has been used suc-cessfully implemented several times in R&D LXe detectors [20] [21] [22]. One important note is that it is also possible for an electron to have enough energy to ionize xenon atoms. This electric field threshold has been measured to be 725+48₋₁₃₉ kV/cm, significantly higher than the electric field threshold for proportional scintillation [20]. In this case, the electron freed through ionization will also be subject to the electric field, and accelerated towards the anode. It is then also able to ionize xenon atoms. This can therefore cause an electron avalanche: each electron freeing more electrons. Therefore this process is called electron multiplication. Electron multiplication amplifies the pro-portional scintillation signal, but also introduces a statistical uncertainty, as electron multiplication is an exponential process. It would therefore have a significant effect on the precision of the S2 signal size measurement in case of a significant amount of electron multiplication. A Monte-Carlo study would be a capable tool to quantify this effect.

The most challenging aspect of proportional scintillation is achieving a sufficient S2 signal size. Proportional scintillation only occurs in the near-wire field and as the path length of the electron is proportional to the S2 size, this already limits the total S2 signal size. The mechanism of electron multiplication limits the maximum electric field that can be achieved, as going too far into the electron multiplication regime will increase the statistical uncertainty to an undesirable amount.

Figure 3.9: Schematic of LEFT: dual-phase model and RIGHT: single-phase model. In black the anode and gate electrode wires. In white and orange the electron paths from the active volume to the anode, where orange signifies the proportional scintillation region. The main differences are the liquid level, proportional scintillation region (orange) and the wire size of the anode.

(30)

Lastly, there is the so-called shadowing effect. As the electrons are created very close, O(5 µm), to each wire, a significant amount of the scintillation photons will be stopped by the wire, creating a wire shadow on the top array of photon sensors. This would decrease the over-all size of the S2’s and worsen the position reconstruction. However, the anode geometry can minimize this effect, as will be discussed in the following section.

3.3.2 Current status proportional scintillation research

In 1979, the first study showing the possibility of a proportional scintillation LXe detector was published at the Waseda university [21]. Within this work, proportional scintillation signals were achieved employing a set-up equipped with anode wires of 4 µm thickness. Due to the engineering challenges, no further studies were published despite the promising results from the Waseda group. Recently, there has been renewed interest in the single-phase design. In 2014, two studies were published where measurements were performed with a single-phase LXe detector [20] [22]. Both reported significant S2 signals from their set-up.

The study performed by the XENON group at Columbia University used a single wire as the anode [20]. They measured the only published thresholds for proportional scintillation, at 412+10₋₁₃₃ kV/cm and electron multiplication, at 725+48₋₁₃₉ kV/cm. The maximum number of photons that was achieved per electron was 287+97₋₇₅, which is very similar to the number of photons per electron achieved in XENON100 [20]. However, this was only possible with a factor of ∼ 14 electron multiplication. In their study the model currently used for proportional scintillation in GXe (see equation 3.1.1) was successfully adapted to proportional scintillation in LXe. The adapted model (shown in equation 3.7 and equation 3.8) agreed well with data, as will be shown shortly. The following equation describes the proportional scintillation in LXe:

∆Nγ = Neθ3(E(~x) − θ4)∆~x, (3.7)

where the ∆Nγ is the change in the number of photons created over a path length ∆~x in an

electric field strength E. The number of electrons is given by Ne, θ3 [ph/e−/(kV/cm · µm)] is

the S2 gain factor for proportional scintillation in LXe and θ4 [kV/cm] gives the proportional

scintillation electric field threshold in LXe, which is the minimum electric field strength required for proportional scintillation in LXe. The above equation shows that proportional scintillation in LXe is still linearly dependent on the number of electrons, the electric field strength and the path length through the proportional scintillation region. The main differences with respect to gas phase proportional scintillation are the lack of dependence on the gas pressure and the changed electric field threshold and S2 gain factor. The process of electron multiplication is significant in LXe proportional scintillation (in contrast with GXe proportional scintillation). The XENON group at Columbia University included this in their model, providing the following equation for electron multiplication: ∆Ne= Neθ0 exp − θ1 E(~x) − θ2 ∆~x, (3.8)

(31)

where ∆Ne is the average change in a number of Ne electrons over a distance of ∆~x. Here

θ0[1/(µm · e−1)] is the charge gain factor in LXe, which determines how much electron

multi-plication will occur for a given electric field over a given distance. The slope in the charge gain is given by θ1 [kV/cm], which indicates how quickly the electrons freed through charge multiplication

are able to free electrons themselves. Lastly, θ2 [kV/cm] gives the electron multiplication threshold

in LXe, the minimum electric field necessary for electron multiplication. The main results from the study ,as published in [20], are shown in figure 3.10. On top, the charge multiplication (on the left) and the proportional scintillation (on the right) are plotted as a function of the anode voltages for two anode wires, of diameter 5 and 10 µm. The fit is performed using equations 3.8 and 3.7. The charge measurement was performed independently of the light measurement, with a charge-sensitive amplifier measuring the charge landing on the anode directly. From these measurements, the θ parameters were extracted for the 5 µm and 10 µm diameter wires. The parameters are show in table 3.3. As can be seen, the fit agrees well with the data. However, in the lower electric field regions, there is significant disagreement, as is shown in the lower two plots of figure 3.10. As is clear from the bottom left, an anode voltage of ∼ 1.5 kV, more S2’s are created (as indicated on the lower left plot) than expected from the model. The paper suggest that this may be due to the statistical nature of proportional scintillation: some electrons do gain enough momentum to cause proportional scintillation, while other electrons do not reach this momentum due to elastic scatters. This agrees with the data from the bottom-right plot. We see a wider spread of S2 widths at the same anode voltages where disagreement occurred in the bottom-left plot as compared to anode voltages above ∼ 1.5 kV. This is caused by the fact the amount of electrons which gain enough energy is a statistical process. At electric field strengths well above the threshold, all electrons will cause proportional scintillation. Therefore we see a minimum S2 width at 10% maximum of around ∼ 250 ns in this region. Below or near this threshold, a some S2’s will be larger than other S2’s because the electrons experiences fewer scatters and were therefore able to reach a higher momentum and cause more proportional scintillation than those electrons which experienced more

Quantity Fit parameter Measured value Charge gain factor θ0[1/(µm · e−1)] 0.80 ±0.10

Slope of charge gain θ1[kV/cm] 242 ±45

Electron multiplica-tion threshold

θ2[kV/cm] 725+48−139

S2 gain factor θ3 [ph/e−

/(kV/cm · µm)]

2.09+0.65_−0.47· 10−2

Proportional scintilla-tion threshold

θ4[kV/cm] 412+10−133

(32)

Figure 3.10: Charge and light measurement results from [20]. TOP LEFT:, the charge collected on the anode is plotted, which is a measurement of electron multiplication. TOP RIGHT: the total S2 signal on the bottom PMT is plotted, which is a measurement of the number of photons produced. Both are plotted as a function of the anode voltage applied for wires with a diameter of 5, in black and 10, in red, µm diameter wire. On the bottom, the disagreement between the fit and data at lower anode voltages is shown. BOTTOM LEFT: the total S2 size on the bottom PMT is plotted (the same quantify as shown on the top right) for the 10 µm wire, now on a log scale. This shows the disagreement at low anode voltages between the data in black and the fit in red. BOTTOM RIGHT: the S2 width at 10% max. is plotted as a function of the anode voltage on the 10 µm diameter wire. Here it is visible that at anode voltages where the data disagrees with the fit, the S2 signals have a much wider width range than in the region where the fit agrees with the data.

(33)

scatters. This feature of the data is not included in the model. This study found a significant shadowing effect (discussed in the previous section), with between 43% and 57% reduction of the light collection efficiency of the top PMT with respect to the bottom PMT. A measurement of the total reduction was not published. A second study, published in 2014 by K.L. Giboni et al., did not perform a threshold measurement over a large range of voltages due to some practical challenges they encountered [22]. However, they did achieve proportional scintillation in LXe, with S2 signals starting at ∼ 300 kV/cm. At a maximum electric field strength of ∼ 350 kV/cm, a gain of 42 γ/e is reported. Not enough information is given to be able to compare this quantitatively to the results from the Columbia group, however the gain factor of 42 is, from initial evaluations, in disagreement with the Columbia result, as it is a significantly higher.

Another, independent measurement of the thresholds with XAMS will be very valuable. One signif-icant contribution from the K.L. Giboni is the placement of the wires. As said before, the Columbia study reported a significant shadowing effect (of around 50%). The placement of the anode wire is the cause of this, as is visualised in figure 3.11. Here the electric field lines are shown for both the Columbia study and the K.L. Giboni et al. study. It should be noted here that only the central electron path was used for data analysis in the Columbia study. In the K.L. Giboni et al. study, no shadowing was reported due to the fact that the electron paths approach from the side instead of the bottom. This latter geometry is therefore distinctly preferred.

Figure 3.11: LEFT: Electric field lines, in red, from the localised (”needle”) alpha source toward the anode for the 2014 Columbia group study. Taken from [20]. RIGHT: Electric potential as calculated for the TPC used in the 2014 study by K.L. Giboni et al [22]. Electric field lines in black. The anode wire(s) are differently positioned, with the anode wire of the Columbia study off-centered and the anode wires from the K.L. Giboni et al. study centered with respect to the gate wires below.

(34)

3.3.3 Summary of benefits and drawbacks single-phase model

There are many notable advantages to the single-phase model., most of which are related to the liquid level management. The height of the liquid level is very important for the S2 signal size, as the path length of the electron through the gas phase is proportional to the number of photons created per electron. Therefore if the detector is not leveled, or the liquid level rises or falls, significant adjustments need to be made. In addition to that, the S2 signal size is also dependent on the pressure of the xenon gas, which has been known to fluctuate significantly and suddenly due to the recirculation flow rate [20] [22]. None of these effects will be present in a single-phase detector. Then there is also the significant advantage of increased freedom in the detector geometry. The gas phase necessitates a single, upward drift field. In a single-phase model, the electrons can be drifted downward, or horizontally. Multiple drift regions are in principle also possible, but this may not be desirable due to increased radioactive background.

Many of these advantages have become particularly relevant as dual-phase LXe detectors have been scaled-up. The potential difference and distance between the gate and anode should stay more or less the same as the detectors are becoming larger and larger. This makes the liquid level management more difficult, however the liquid level has so far been manageable in large detectors, such as XENON1T and XENONnT. In addition to that, the electrodes, which span the whole width of the TPC, also get larger with each new generation of detector. The weight of the mesh itself causes a sagging of the mesh, resulting in a higher electric field in the middle of the multiplication region. The single-phase detector would also be less sensitive to the sagging, as the electric field in the near-wire region is mostly dependent on the anode voltage and wire size.

There are also significant challenges to the implementation of single-phase dual-signal LXe detectors. First of all is the total S2 gain, as mentioned before. More research will need to be done in order to evaluate whether this is a limiting factor. In addition to that, the thin wires pose a significant practical challenge. The fragility could lead to wire breakage during operation. This thesis therefore studies a method of thin-wire proportional scintillation where the wires are stabilised onto a UV-transparent surface.

The most significant challenge to single-phase dual-signal TPC’s is the lack of research and expertise. Dual-phase TPC’s have been an area of research for decades and though they posses some significant drawbacks, these are well understood. There is no guarantee that single-phase dual-signal TPC’s will not also have similarly significant drawbacks when scaled-up to the size of the currently largest dual-phase TPC’s.

(35)

4 Monolithic thin-wire anodes

This chapter will discuss the specifications of the four novel anodes that were produced for XAMS in order to transition XAMS from a dual-phase detector to a single-phase detector. In addition to that, the results from dedicated setups used for the initial testing of these anodes and a discussion of the physical properties of photomasks, the base material from which these four anodes are made, will be presented. The final sections will discuss the details of the assembly of XAMS during the installation of the new anode and any changes made to XAMS.

The dedicated electric field calculations that were performed to design the exact geometry of the anodes (placement and size wires) can be found in 5. This exact geometry will not be relevant for the production and initial testing.

4.1 Anode requirements

The primary requirement of the new anodes is the ability to achieve an electric field strength above the threshold for proportional scintillation, which has been measured to be (412+10₋₁₃₃ kV/cm) [20].This can be achieved with a new anode with thin wires of diameter O(10 µm).

Wires of thickness of O(10 µm) are very vulnerable to breakages. If an anode wires should break during operation, this would most likely requires major interruption of the measurement. The wires of the electrodes currently in use in XAMS are 150 µm thick. The anode is fabricated out of a sheet of 150 µm thick stainless steel, which is chemically etched to achieve the wire structure visible in figure 4.1. This thin mesh is subsequently spot-welded to a stainless steel ring for structural in-tegrity. The full electrode is show on the left in figure 4.1.

Therefore a monolithic anode has be produced using a product called a photomask, which consists

Figure 4.1: Current XAMS electrode meshes. LEFT: Image of one of the electrode meshes used in XAMS. Middle region is stainless steel of 150 µm thick, spot-welded onto a stainless steel ring for stability. RIGHT: Detail of electrode mesh. Both images taken from [15].

(36)

of UV-transparent quartz glass, which is coated with ∼ 100 nm thick chromium. This chromium can be etched out into a desired pattern after it is applied to the quartz glass. This pattern will consist of a ring with the anode wires in the middle. It was chosen to use an industry-standard photomask, as this method is very novel and any further optimization may be done according to the outcome of this study. The physical properties the photomask which was produced for the XAMS set-up will be described in the next section.

4.2 A photomask as electrode

4.2.1 Physical properties of photomasks

Photomasks are essential components in the production process of integrated circuits (IC’s). IC’s are made in a repeated process, where each step consists of depositing a layer of material is to a wafer (the base of the integrated circuit) and subsequently selectively removing this layer such that only a desired pattern is left (pattering), which may be done through etching. Modern IC’s consist of many different material and therefore go through tens of application and patterning cycles as well as other steps aimed at modifying the (properties of) materials once they have been deposited. Photomasks are used for etching which is one of the main methods of patterning. Etching with photomasks consists of two main phases: the photolithography phase and the main etching phase. For the photolithography phase, a photo-sensitive layer (photo-resist) is applied to the wafer, which is selectively removed with light, leaving only the desired pattern of photoresist2_{. This}

projec-tion of a pattern onto a wafer is done with a photomask. The photomask is a plate consisting of

2_{For positive photo-resist, this is the case. However, for negative photo-resist only the photoresist which is not}

exposed to light is removed.

Figure 4.2: Photolitography step in integrated circuit production. The photomask (here: mask) filters light from a homogeneous source into a desired pattern, removing the photoresist according to this pattern. Etching the wafer is the subsequent step. Taken from [23].

Particle detection through proportional scintillation in liquid xenon: The development of novel anode techniques for XAMS