The XAMS level meter and pressure safety system

The XAMS level meter and pressure safety

system

Verslag van Bachelorproject Natuur- en Sterrenkunde, omvang 15 EC, uitgevoerd tussen 07-04-2014 en 27-07-2014

Nikhef

Abstract

This Bachelor thesis is a report of a three month project at the Nikhef/UvA Dark Matter group on the XAMS setup.The main aim of the group consists of building and commissioning a liquid xenon par-ticle detector. The main part of this detector is a volume partly filled with liquid xenon and partly filled with gaseous xenon in a configu-ration called a dual-phase time-projection-chamber. For the detector to work properly, the level of the liquid needs to be monitored with a precision better than 1 mm. The main subject of this report is about how this liquid level was measured using a capacitance based level meter. Another important subject was the implementation of a safety system for the detector in order to prevent the pressure rising to high inside the detector.

Donkere materie en de XAMS level meter

De zoektocht naar donkere materie (dark matter) houdt veel ex-perimentele en theoretische fysici bezig. Uit waarnemingen van het heelal weten we dat zichtbare massa, zoals sterren, planeten en gas, maar 15% van alle massa in het universum uitmaakt. De overge-bleven 85% bestaat uit donkere materie. Het is nog niet bekend waar deze donkere materie precies uit bestaat en dat is een vraag die veel fysici proberen te beantwoorden. Een populaire hypothese is dat deze donkere materie uit een nieuw soort deeltjes bestaat en alom aanwezig is in de hele Melkweg en andere sterrenstelsels. In dat geval zouden er elke seconde velen duizenden dwars door de aarde heen vliegen. Normaal gesproken hebben deze deeltjes nauwelijks interactie met de deeltjes op aarde, maar het is mogelijk dat een donker deeltje precies op een atoomkern botst.

Er zijn verschillende projecten gaande die op deze manier donkere deeltjes proberen te ontdekken. De basis van veel van deze opstellin-gen wordt gevormd door een vat gevuld met vloeibaar xenon. Als in het vat een botsing tussen een donker deeltje en een xenon atoom plaatsvindt, wordt er licht geproduceerd. Dit licht kan gemeten wor-den en aan de hand van de intensiteit kan de energie van het donkere deeltje bepaald worden. De XAMS opstelling op het Nikhef is werkt precies op dezelfde manier, maar heeft alleen een veel kleinere schaal. Daardoor kan het geen eventuele donkere deeltjes meten, maar is het wel nuttig om meer over de werking van dit soort opstellingen in het algemeen te weten te komen.

Niet het hele vat, dat ongeveer de afmetingen van een koffiebekertje heeft, is gevuld met vloeistof. De bovenste laag bestaat uit xenon in gas vorm. Voor een optimale werking van de detector moet het niveau van het vloeistof constant zijn en zich in een vooraf bepaalde zone van 5 mm breed bevinden. Het vaststellen van het vloeistofniveau wordt gedaan met een zogenaamde level meter. De level meter die gebruikt is in het XAMS project, bestaat uit een metalen buis met een veel smallere metalen staaf erin. Deze opstelling wordt langs de lengte van de detector geplaatst en werkt op capacitieve basis. De capaciteit van een voorwerp is een maat voor hoeveel elektrische lading erin kan worden opgeslagen. Voor een voorwerp als de level meter hangt dit ook af van de dilektrische constante van het materiaal tussen buis en de staaf. De di¨elektrische constante is een grootheid die de sterkte van het elektrische veld in het betreffende materiaal omschrijft vergeleken met in het vacu¨um. Vloeibaar xenon heeft een hogere di¨elektrische

Figure 1: Een schematisch overzicht van de level meter met totale lengte l. Als het vloeistofniveau (h) in de detector toeneemt, wordt een groter gedeelte van de level meter gevuld met vloeibaar xenon en neemt, vanwege het verschil tussen de di¨elektrische constantes van vloeibaar gasachtig xenon, de capaciteit van de level meter toe.

constante dan xenon gas. Wanneer het vloeistofniveau stijgt, wordt een groter gedeelte van de level meter gevuld met vloeistof en dit heeft als gevolg dat de gemeten capaciteit toeneemt. In figuur 1 is het algemene principe van de level meter schematisch weergeven. Het doel van dit onderzoek is om een continue en nauwkeurige meting van de capaciteit van de level meter te maken, opdat het vloeistofniveau constant gehouden kan worden.

Een ander onderwerp waar in dit verslag aandacht aan zal worden besteed, is de implementatie van een veiligheidssysteem om de juiste druk binnen het detectorvat te waarborgen. In het geval dat door bi-jvoorbeeld een elektriciteitsstoring de koeling van de detector uitvalt, zou de druk kunnen stijgen. Om gevoelige apparatuur binnen de de-tector te beschermen tegen te hoge druk, zorgt het veiligheidssysteem er in dit scenario voor dat het xenon gekoeld wordt met vloeibaar

Contents

1 Introduction 5

2 Dark matter theory 6

2.1 Evidence for dark matter . . . 6

2.1.1 Rotation curves . . . 6

2.1.2 Gravitational lensing . . . 8

2.2 WIMPs . . . 11

3 The XAMS experiment 12 3.1 Xenon-based detectors . . . 12

3.2 XAMS setup . . . 15

3.3 XAMS results . . . 18

4 The level meter 19 4.1 XAMS level meter setup . . . 19

4.2 Measuring principle . . . 23

4.3 Stability analysis . . . 27

4.3.1 Recommendations . . . 29

4.4 Provisional level meter tests . . . 33

5 The safety system 36 5.1 Theory . . . 36

5.2 Practical considerations . . . 38

5.3 Emergency protocols . . . 39

1

Introduction

The search for dark matter is an important subject of astrophysics. From astronomical observations it is known that dark matter makes up 85% of all the matter in the universe. Still, it is not yet clear what particles dark matter is made of and what its exact properties are. In the last few years a number of experiments which strive to put a limit on the mass of dark matter by the use of liquid xenon have been designed. Examples are the LUX project or the XENON100 experiment.The basic idea is that a dark matter particle has an elastic collision with a xenon atom and as a result of this a measurable recoil is produced. The XENON100 and the future XENON1T experiment make use of a so-called dual-phase time-projection-chamber (TPC). The main part of such a construction consists of a volume partly filled with liquid xenon and partly filled with gaseous xenon. The XAMS experiment at Nikhef consists of a TPC like XENON100, but much smaller in size. Its aim is a better study of the working of such an installation and to provide a better understanding of the properties of liquid xenon. For a proper operation of the TPC the liquid level needs to be monitored with a precision better than 1 mm. The liquid level is measured using a capacitive level meter, which makes use of the difference in dielectric constants between liquid and gaseous xenon. This means that there is a direct relation between the capacitance of the level meter and the liquid level.

In addition to the work done on the level meter, a pressure safety system for the detector was implemented. This safety system starts to work when the pressure inside the detector reaches an inappropriate high value due to an unwanted or unforeseen increase in the temperature. When this happens, the safety system cools the detector with liquid nitrogen, in order to lower the temperature and thus the pressure.

This thesis is build up as follows. First there will be an introduction in the theory about dark matter and the means by which liquid xenon detectors try to detect it. Then the XAMS setup will shortly be introduced. The last two sections consist of a report about all the building, programming and testing involved in the realization of the measuring system for respectively

2

Dark matter theory

2.1

Evidence for dark matter

There are many pieces of evidence for the existence of dark matter. Still no dark matter particle has been identified. In this paragraph some of the more famous evidence material is given. This is by no means a complete overview of all the evidence of dark matter. For a good overview see [3] by Bertone et al..

2.1.1 Rotation curves

Probably the most classical and well known argument for the existence of dark matter lies in the observation of spiral galaxy rotation curves. A spiral galaxy generally consists of a flat rotating disk full of stars, gas and dust, with a bulge in the center. Most of the gas is concentrated in the disk, while the concentration of stars is the most dense inside the bulge. The radial velocity v of a star with orbit R is given by:

v = r

GM (R)

R . (1)

Here G is Newton’s gravitational constant and M (R) is the total mass inside a sphere centered at the galactic center with radius R. Most of the stars are concentrated in the center of the galaxy, so away from this center, the enclosed mass within this sphere increases in a negligible way. Thus, if the stars, gas and dust would contribute most, or all, of the mass, the velocity would fall as,

v ∝ √1

R. (2)

A rotation described by equation 2.2 is known as a Keplerian rotation. But when actual rotation velocities are measured, this relation is not observed. The rotational velocity of stars remains constant far away from the center of the galaxy. An example of such behaviour is given in figure 2. Therefore additional mass that does increase when R increases is needed. This gives rise to the idea of a dark matter halo, in which the visible parts of the spiral galaxy are embedded.

Figure 2: Rotation curve of the spiral galaxy NGC6503. The data points with the error bars are the observed circular rotation velocities as a func-tion of distance from the center of the galaxy. The dotted, dashed and solid lines are the contributions of gas, disc and the dark matter halo, respec-tively. It can be seen that the contributions of just the gas and the disc are not enough to construct the observed rotational velocity;an additional dark matter component is needed. Image taken from [4]

Figure 3: Gravitational lensing. In this example a galaxy bends the light coming from a distant light-emitting object. As a result, observers on earth see this object in two or more different locations on the sky. Image taken from http://casswww.ucsd.edu/archive/public/tutorial/GLens.html

2.1.2 Gravitational lensing

Another piece of evidence for dark matter comes from gravitational lensing. Gravitational lensing occurs when light is bent by the presence of mass. According to general relativity, when a photon moves closely to a strong gravitational potential, its trajectory is bent because of the curved space-time. As is shown in figure 3 heavy objects act as a lens when they are on front of light-emitting source. When the heavy object is somewhere directly between the source and the observer, the effect is clearly visible as in figure 4 and it is called strong gravitational lensing. Such circumstances are quite rare, but it gives physicists the occasion to calculate how heavy this object is.

An effect that is much harder to observe, but much less rare, is weak gravitational lensing. In this case the light is only deflected by a very small angle. A way to observe this effect is to look at the alignment of galaxies. When one looks at a certain patch of the sky, it is expected that the galaxies in that patch are randomly orientated. But when they are all lensed by a tiny bit, they will all be aligned more in the same direction. The lensing stretches them in the direction perpendicular to the gravitational field. In this way, it is still possible to examine the mass of objects using lensing, although no nice circles like in figure 4 are observable.

Recently, a very compelling piece of evidence for dark matter was found using gravitational lensing . In the merging cluster 1E0657558, also known as the bullet cluster, two clusters of galaxies moved trough each other. In such a collision, the galaxies themselves are largely unaffected and keep their own trajectories. Both clusters also contain a lot of gas, which gets ionized and produces an X-ray emitting plasma. In fact, this gas is the dominant baryonic

Figure 4: This is an example of strong gravitational lensing. The galaxy cluster Abell 1689 bends the light coming from more distant galaxies and projects them spread out in a circle. The image was taken with the Hubble Space Telescope. Image taken from http://hubblesite.org/newscenter/ archive/releases/2013/36/image/b/

component in the system. Baryonic mass consists of all normal matter, like atoms, build out of protons and neutrons. These plasmas do have interaction and this results in a large quantity of this plasma being left behind between the two galaxy clusters. In the absence of dark matter, the gravitational potential will coincide with the intra cluster X-ray emitting plasma, because this is the dominant mass in the whole system. When dark matter is present, the gravitational potential will coincide with the location of this dark matter, which is around the centers of the galaxies. When a gravitational potential map is made, this last situation seems to be the reality, as can be seen in figure 5. The main mass seems to be concentrated around the centers of the two original clusters, at the same place as the non-interacting galaxies. This indicates the presence of non-interacting dark matter.

Figure 5: The green contours in both panels are the gravitational lensing potentials. In the left panel a gravitational map with the light emitting sources is displayed. In the right panel the plasma is indicated, where the hottest regions are shown in white and the coolest regions in blue. The position of the plasma does not coincide with the contours of the gravitational potential, although this would be expected in the absence of dark matter. This discrepancy suggests the existence of non-interacting dark matter. The white bar corresponds to 200 kpc or 6.5 · 105 light year. Image taken from http://arxiv.org/pdf/astro-ph/0608407v1.pdf.

2.2

WIMPs

Using astronomical observations it is possible to calculate the contents of the universe. The WMAP is a satellite which observes the cosmic microwave background. By analyzing the nine year WMAP data of the cosmic mi-crowave background, the following values of the mass-energy densities have been calculated. These values are (4.6±0.2)% baryonic matter, (23.3±2.3)% dark matter and (72.1±2.5)% dark energy. Dark energy is a form of energy that accounts for the expansion of the universe, which will not be treated further in this thesis. From these values it can be calculated that dark matter particles makes up about 84% of all the matter in the universe, compared to 16% baryonic matter. Theorists have come up with a variety of candidates for dark matter particles. The properties that a candidate should at least satisfy will shortly be discussed. [2]

A dark matter particle has to have a certain mass, otherwise it could never account for 84% of all the mass in the universe. Interactions with other particles should be small. In combination with that, it is also needed that the particle is of neutral charge, otherwise it would be very easy to measure. One further condition is that it should be stable and not decay into other particles. The kind of particle that one tries to detect with liquid xenon detectors is the WIMP. WIMP stands for Weakly Interacting Massive Particle, which means that is a quite heavy particle (about ten to a few thousand proton masses) and that interacts via the weak force. The compelling issue about the existence of WIMPs is that extension of the Standard Model, for example by introducing Supersymmetry, also predicts such subatomic particles.

3

The XAMS experiment

3.1

Xenon-based detectors

Globally speaking, there are three ways to try to detect dark mater parti-cles. These are direct detection, indirect detection and by use of accelerator experiments. The XAMS experiment and other liquid xenon experiments are based on direct detection methods. Therefore, only this method will be discussed here. In a direct detection experiment, the idea is that a dark matter particle interacts with a known particle and afterwards the effects of this interaction are measured. Therefore, as few as possible other particles should interact with the detection medium. Examples of this background are cosmic rays produced in the atmosphere and radiation due to radioac-tive isotopes around the detector. To reduce the contribution due to cosmic rays, experiments are done deeply underground, in mine shafts. Shields to stop other particles are also installed. Materials that are used to build the detector and the shielding are selected for a low radioactive activity.

Because of the following properties, liquid xenon (LXe) is an excellent material to use as detection medium. In the first place it has a high density (3 g/cm3_{) and a high atomic number (54), which makes it relatively good}

at stopping particles that move trough the detection volume. Secondly the energy required to produce one electron-ion pair (also known as the W-value), is relatively small. For LXe this value is 15.6 eV, compared to liquid argon (23.6 eV) and liquid krypton (18.4 eV). [1] The most important property of LXe lies in the fact that it produces two signals when it is exposed to radiation. When LXe is excited due to a nuclear recoil, a scintillation as well as an ionization signal is produced. Physically this means that both photons and electrons are made free when a particle interacts with the xenon. The scintillation light is always produced in the following process. An excited xenon atom and a non excited xenon atom combine and then de-excite by releasing a photon,

Xe∗+ Xe → Xe∗₂

Xe∗₂ → Xe2+ hν. (3)

Excited xenon atoms can be produced by direct excitation, in this case only reaction 3.1 happens. They can also be produced from ionized xenon

atoms, according to reaction 3.2. Note that ionized xenon atoms are firstly a source of electrons and secondly a source of scintillation:

Xe++ Xe → Xe+₂ Xe+₂ + e− → Xe∗∗+ Xe

Xe∗∗ → Xe∗ + heat Xe∗+ Xe → Xe∗₂

Xe∗₂ → Xe2 + hν. (4)

These two processes are used in a dual phase time projection chamber (TPC). See figure 6 for a schematic view of the working of a dual phase TPC.When a particle interacts with xenon inside the detector, the scintil-lation produced is measured by both the bottom PMT’s (photo multiplier tubes) and the top PMT’s. This signal is referred to as the S1 signal. Any

electrons produced in the interaction are accelerated upwards by the drift field. This way they cannot recombine with any xenon ions and they arrive at liquid surface. The drift field is the electric field between the cathode and the gate mesh, as can be seen figure 6. When the electrons reach the liquid surface, they are extracted out of the liquid into the gas by the extraction field, a much higher field that is sustained between the anode and the ground. When the electrons are extracted into the gas again emission of scintillation takes place, in a process called electro-luminescence. This signal, that is also recorded by both PMT arrays, is called the S2 signal. The liquid level needs

to be somewhere between the gate and the anode for the extraction field method to work. The fact that two signals are produced, makes it possible to determine the z-position of the event. The S2 signal is observed a certain

amount of time later than the S1 signal by both PMT’s. This time is the

time that the electrons need to drift towards the extraction field. Out of this time discrepancy, which is called the drift time, the z-position of the event can be reconstructed.

Figure 6: A schematic illustration of a two-phase time projection chamber. An S1 signal is directly produced when a particle interacts with the liquid

xenon. The electrons produced at this interaction drift upwards and some time later produce an S2 signal when they are extracted into the gaseous

xenon. Image taken from http://www.physics.purdue.edu/darkmatters/ xenon1t/?p=44

Figure 7: A picture of the XAMS experiment. From left to right are shown: the xenon circulation system, the detector itself, housed in a stainless-steel cryostat, electronic equipment for data-taking and supplying the needed high voltages and the computer that monitors the temperature and pressure inside the detector.

3.2

XAMS setup

This section deals with the technical features of the XAMS detector. In figure 7 an overview of the experimental setup can be seen. The XAMS TPC is housed in a Teflon cylinder about 21 cm high. It was constructed by stacking Teflon rings upon each other. Space for any electronic equipment like meshes, resistors or temperature sensors was milled into the rings. A picture of this Teflon structure is shown in figure 8, a cross sectional view of the TPC is

Figure 8: A picture of the XAMS TPC while under construction. Note the Teflon ring structure which encloses the two PMT’s and the active volume.

ground) of 600 V/cm. The anode voltage is at + 3 kV. This constitutes an extraction field of 6000 V/cm. Any higher voltages on either the cathode or the anode, resulted in a spontaneous discharge between the meshes, also called a ‘trip’. The PMTs are operated at a voltage of 900 V. To minimize the electric fields of the cathode and anode around the PMTs, so called screening meshes were installed. The bottom one is kept at a potential of -900 V and the top one is kept at ground. This way the electric fields around the PMTs are reduced to a minimum.

Figure 9: A drawing of the cross section of the XAMS TPC. The bottom PMT is indicated by a 1 and the top PMT is indicated by a 2. The five meshes are shown and indicated. a: bottom screen; b: cathode; c: gate;d: anode; e: top screen. The bottom and top screen meshes shield the PMT’s

3.3

XAMS results

The XAMS detector has already produced results. To make sure to have a measurable signal, a gamma-ray source, Cs-137, was placed next to the detector. With the earlier described settings figure 10 was produced.

Figure 10: A plot of an S1 and S2 signal. The time between the two signals

is the drift time. The S2 signal is broader because not all electrons produce

a signal at the same time when they are extracted into the gas.Both the S1 and S2 signal are clearly visible, showing the successful operation of the

4

The level meter

4.1

XAMS level meter setup

As was seen in the previous chapter, for a clear measurement of the S2 signal,

it is necessary that the liquid level is somewhere between the gate mesh and the anode mesh. Since it is not possible to have a look inside the detector volume during filling because it has to be well- insulated, a more ingenious method is needed to measure the level. The level meter is exactly this; a device to measure the level of the liquid xenon inside the TPC. It indicates how high the liquid level is inside the detector. The device works in the following way. The level meter is a cylindrical capacitor that is placed in the Teflon hull of the TPC. A picture of the level meter (when it was not already placed inside the detector) can be seen in figure 11. It is aligned alongside the active volume. The cylindrical capacitor consists of a metal rod inside a metal tube, with space in between. The capacitance of a capacitor depends among other things on the dielectric constant of the medium between its two sides. The capacitance of a cylindrical capacitor of height l filled with a linear dielectric material with a dielectric constant ris given by the following

formula: C = 2π0rl 1 ln (Rb Ra) , (5)

where Ra and Rb are the radius of the rod and tube, respectively. The

XAMS level meter is filled with two dielectric materials as can be seen in figure 12. In order to calculate the capacitance of this configuration it is possible to view the level meter as two separate capacitors placed in parallel. The capacitor filled with liquid xenon has height h and dielectric constant l.

The capacitor filled with gaseous xenon has height l−h and dielectric constant g. when multiple capacitors are placed parallel, the total capacitance is just

the sum of all the independent capacitances:

C = 2π0lh

1

+ 2π0g(l − h)

Figure 11: A picture of the level meter used in XAMS. The mounting block on the end is not for measuring, but functions as a means to connect cables to the level meter. The lenght of the level meter is 275 mm.

Figure 12: A schematic view of the principle used by the XAMS level meter. Because of the difference between g and l, the

capac-ity increases when h, the liquid level, increases. Image taken from http://www.chegg.com/homework-help/questions-and-answers/

The last term of this equation is just the capacitance of a cylindrical capacitor fully filled with gas (compare with equation 4.1). So when the level meter is filled to height h with liquid, the change in capacitance is given by:

∆C = 2πh0∆lg

1 ln (Rb

Ra)

(7)

Where ∆lg is defined as l − g. For the XAMS level meter, Ra = 3 mm

and Rb = 4 mm and for xenon ∆lg = 0.879. [5] This gives for the change in

capacity per unit length: ∆C

h = 0.879 · 2π0 1

ln (4₃) = 0.170pF/mm. (8) Liquid xenon has a higher dielectric constant than gaseous xenon, which makes up the rest of the inside of the volume of the TPC. When the liquid level in the TPC increases, a larger volume of the space between the two charged cylinders of the level meter is filled with liquid xenon and thus the capacitance of the level meter will increase according to equation 4.4.

Unfortunately, the level meter stopped working the first time liquefied xenon was inside the detector. It stopped working in the sense that there was a short circuit between the ground and plus side of the level meter and thus it was no longer possible to measure the capacitance. When the xenon was recuperated, it started working again, so it is suspected that it’s failure has something to do with the decreased temperature inside the detector when filled with liquid xenon. There was no time to replace the level meter or to do additional testing. This will be done when the detector is opened again.

Fortunately, the gate mesh and the anode mesh could be used together as a provisional level meter. When the detector is taking data, the anode is at a voltage of around 3 kV in order to sustain the extraction field, which extracts electrons out of the liquid into the gas. But when this high voltage is removed, it is possible to use these two meshes as a parallel plate capacitor. A benefit of this construction is that it is precisely the liquid level between these two meshes that we are interested in, because in a normal operating

The electric fields follow from Gauss’ law for dielectric materials: Egas = σ 0g ~ z; Eliquid= σ 0l ~ z. (9)

The fields are aligned in the z direction, which is the direction perpendicular to the plates. Here σ is the surface charge density of the plates, which is defined as Q/A. Q is the total charge on a plate and A is its surface. Now the potential can be calculated using V (r) = R_OrE · dl. When the capacitor with height s is filled for percentage α with liquid this integral becomes:

V = Z αs 0 Eliquid· dl + Z s αs Egas· dl V = sσ ·  α 0l +(1 − α) 0g  . (10)

Now C = Q/V and σ = Q/A is filled in,

C = Q s(Q/A) ·_α 0l + (1−α) 0g  C = gl αg+ (1 − α)l · 0A s . (11)

This formula makes sense from a physical point of view. When the volume is completely filled with gas (i.e. α = 0) it just reads the capacitance of a capacitor filled with gaseous xenon. When the volume is completely filled with liquid (i.e., α = 1) it just reads the capacitance of a capacitor filled with liquid xenon. As seen in formula 4.7 and figure 13, there is no linear relation between the capacitance and the level of the liquid. This is no problem when you remember that when operating this level meter a change in the capacitance does not corresponds to a fixed change of the value of the liquid level. The capacitance is therefore still a useful way to determine the liquid level. A rough estimate that was made in this calculation is the fact that the meshes were taken to be infinite parallel plates. In fact the electrical field of a two meshes is quite complicated to calculate. A different electric field will also have an influence on the capacitance. It is expected that the capacitance of two meshes is lower than the capacitance of two comparable solid discs, because meshes just have less area and thus less space to allocate charge.

0.2 0.4 0.6 0.8 1.0 part filled with liquid 1.2 1.4 1.6 1.8 capacitanceHpFL

Figure 13: This graph shows how the capacitance of the provisional level meter depends on the fraction of the space between the two meshes that is filled with liquid.

4.2

Measuring principle

The next goal is to make a continuous measurement of the capacitance. In this way, any change in the capacitance will directly deliver information about any change of the liquid level. There are various ways to measure capacitance. One can for example measure a capacitance very precise with a LCR meter. This is a machine that uses an LCR circuit to calculate the capacitance of a given object. When this method was used at the XAMS experiment, it was found very difficult to continuously send information from the LCR device to the labview program. Labview is a program that is used to monitor all the different parameters of the experiment. It was decided to use an Arduino chip for reading out the capacitance. The Arduino is an open-source electronics prototyping platform. There are different versions available and in this project the Arduino UNO was used. It has various electronic pins that can be both configured as input or output. When configured as input the pins have a very high impedance. The Arduino works like a small processor and you can write your own program for it to execute in C language. The capacitance was measured with this device in the following way. When one applies a voltage over a capacitor, it will charge according to the following

1 2 3 4 5

time

HAUL

voltage

H%L

Figure 14: Charging curve for a capacitor.

capacitor is charged and C is the capacitance of the capacitor. An example of what such a graph looks like is given in figure 14. Formula 4.8 can be rewritten as a expression for C.

C = t/(R ln 1 − V (t) V0

) (13)

To calculate C one has to measure V and the corresponding t. V0 and R can

be measured beforehand. The RC-time is defined as the time were t = R ∗ C. At this time V = (1 − 1_e)V0.

Simply stated, the level meter will be charged up by the Arduino, which was measured to supply a basic voltage of 5.42 volts. The hardware config-uration of the circuit can seen in figures 15 and 16. The Arduino measures the voltage on the capacitor trough analog pin A0. This pin digitizes the voltage with 1024 bit accuracy. This means that the resolution for the volt-age measurement is 0.0053 volts per step. The Arduino can measure time with a resolution of 4 microseconds. It is expected that the most precise way to measure the capacitance would be to measure the voltage at different times and then fit these points to an exponential curve. Unfortunately, the software required to fit a curve to some points, was too large to upload to the Arduino. Therefore, it was decided to use a sampling method. There were taken some points along the curve, for each point the corresponding capacitance was calculated and then the average capacitance was calculated. In the final setup, 8 measurements from 3.2 V to 5.2 V were taken, with an interval of 0.25 V. As can be seen in figure 14, the capacitor charges faster at the start then at the end. With an 100 MΩ resistor, as was used in the

Figure 15: The Arduino circuit that was used to read out the capacity of the level meter. The level meter is charged using resistor R1, which is

typically 100 MΩ. While this happens, the voltage on the level meter is read by the readout pin A0.The level meter is then discharged using resis-tor R2, which is typically 220 Ω. Image partially taken from http://www.

logicaprogrammabile.it/come-creare-sistema-di-allarme-arduino/

final setup, the first measurement had to be after 3.0 V, because the readout of the Arduino was too slow. This means that when the Arduino measures a voltage it assigns a time to that voltage. This takes a few hundred mi-croseconds during which the voltage has already risen significantly higher.

Figure 16: The Arduino hardware setup. One can see the Arduino board, the two resistors and the Faraday box. A Faraday box is used to prevent any other electromagnetic radiation that might be produced near the experiment from influencing the Arduino circuit.

Figure 17: Measurement of the capacitance of a 220 pF capacitor using the Arduino. In the left hand panel, a plot of the capacitance against time is shown. A corresponding histogram of this measurement can be seen in right hand panel. The measured values adhere to a Gaussian distribution. For this measurement R1 = 100M Ω and R2 = 220Ω.

4.3

Stability analysis

One goal of the project was to find the optimal way for the Arduino to mea-sure the capacitance. Different setups were tested to find which set up would deliver the best precision. This was done by measuring constant capacitances for a certain amount of time with the Arduino. When doing these stability measurements the Arduino circuit was build on a bread board and not yet in its final form as can be seen in figure 16. This was done in order to easily change parts of the circuit. In these tests the values of R1 and R2 (see figure

15) where varied. There were also done various tests on the software part. When the capacitance of, for instance, a normal 220 pF capacitor is mea-sured, the values that are measured follow a Gaussian distribution. An ex-ample of such a measurement can be seen in figure 17.

Most measurements are close to 220 pF, but some vary quite a bit. It was investigated whether the small fluctuations in the capacitance were a

was found in either measurement. Rather, most measurements that were different, varied by a few bits. The precision was defined to be one standard deviation of the Gaussian. The main variable that was changed when testing different circuits, was the value of the resistors R1and R2. Resistor R1should

not have a very small resistance, because then the capacitor is charged too fast for the Arduino to do the necessary measurements. For R1the choice was

between a 100 MΩ and a 1000 MΩ resistor. Theoretically, if an infinite input impedance is assumed, the value for R2does not influence the measurement at

all, because no measurements are done when the capacitor discharges and the next measurement starts only when the capacitor is completely discharged. The capacitor is known to be discharged if the Arduino measures no more voltage on it. It was decided to use a 220 Ω resistor for R2. In figures 18 and

19 the results of two tests are shown. In the test in figure 18, the capacitance of a 220 pF capacitor was measured. The circuit with the 100 MΩ resistor reached a much smaller standard deviation. To confirm that this result also holds true for the level meter, a short test was done with the dry level meter as well. This means that the level meter was not filled with any xenon, be it gas or liquid. The results can be seen in figure 19. This test also suggests that the 100 MΩ resistor provides better results.

It is also possible to discharge the capacitor using the same resistor that was used to charge the capacitor. The result of a test using this setup is given in figure 20. It can be concluded that discharging the capacitor using a much lower resistance is preferred. This is probably due to the fact that it takes longer to discharge using a high resistance, and thus in the same time less measurements can be done, leading to a precise average. Still this setup is interesting, because a future idea might be to make measurements both when the capacitor charges and discharges.

The offset is a result of the so called stray capacitance of the Arduino circuit itself. The offset includes both the capacitance of the cables leading from the level meter to the Arduino and the stray capacitance of the Ar-duino itself. This offset is not a problem, because the ArAr-duino can easily be calibrated by unplugging any capacitance that is measured. In that case the Arduino just measures the capacitance offset. Because parallel capacitances add up, any measurement can be corrected by subtracting the offset. Actu-ally it is not necessary to do this, because for the level meter only differences in capacitance matter.

4.3.1 Recommendations

There are still some things that could be improved in the future or require additional testing. For the software part, doing measurements on both the charging and discharging of the capacitor could be an improvement. Another software improvement would be to send the voltages and corresponding times read by the Arduino to an program outside of the Arduino sketch. The benefit would be that this program could then fit these values to the curve seen in figure 14.

A hardware improvement could be to let the Arduino measure a known capacitor at the same time as it measures the level meter. This functions as a live calibration technique. When any irregularities occur at the measured value of the level meter, and such irregularities are at the same moment also seen in the measurement of the test capacitor, it can be concluded that this probably had nothing to do with anything physical concerning the level meter, like a rapid change of the liquid level. Rather, an electronic or en-vironmental phenomenon that influences the readout of the capacity can be considered. A test that also might be done is to look how the measured value of a constant capacitor changes when the temperature in the room of the Arduino changes.

Figure 18: Comparison between two configurations of the Arduino circuit. In the top graph a 100 MΩ resistor was used to charge the capacitor (in com-parison with figure 15, this is R1). In the bottom graph a 1000 MΩ resistor

was used to charge the capacitor. R2 = 220Ω for both tests. Both tests were

done with a static 220 pF capacitor and were done over the timespan of 8 hours. As can be seen, a lower σ was achieved using the first setup, which is would thus be preferred to use.

Figure 19: Comparison between two configurations of the Arduino circuit. In the top graph a 100 MΩ resistor was used to charge the capacitor (in comparison with figure 15, this is R ). In the bottom graph a 1000 MΩ

Figure 20: Comparison between two configurations of the Arduino circuit. In the top graph a 100 MΩ resistor was used to both charge and discharge the capacitor. In comparison with figure 15, this means that R1 = R2 = 100M Ω.

In the bottom graph a 220 Ω resistor was used to discharge the capacitor. Both tests were done with the XAMS level meter and were done over the timespan of 8 hours.The second setup achieved a better precision, which may indicate that the discharge resistor has some effect on the measurements and that in this case a small discharge resistor works better.

4.4

Provisional level meter tests

The provisional level meter, which was already discussed in section 4.1, was never intended as a level meter and thus needed to be tested. To do mea-surements with the provisional level meter, the anode was connected to the charging and discharging point of the Arduino circuit and the gate was con-nected to the ground of the Arduino circuit. A point measurement of the liq-uid level can be made using the Pt-100 temperature sensors inside the TPC. These sensors measure very locally the temperature of the xenon. Their main part is a resistor of which the resistance depends on the temperature. By measuring the resistance, you can make a measurement of the temperature. Because liquid xenon makes better thermal contact with the sensors, its tem-perature is measured a few degrees Celsius lower than when the temtem-perature of gaseous xenon is measured. When a Pt-100 is first located in the gas, but due an increase of the liquid level makes makes contact with the liquid, this effect can be observed in the form of a sudden drop in the measured temperature in the order of one or two degrees Celsius. A detailed overview of the exact locations of relevant Pt-100’s and the meshes is given in figure 22.

The test was done in the following way. It was first made sure that none of the Pt-100’s were submerged in the liquid by increasing the temperature in the volume. Because this had a quite slow effect, most of the time it was done by increasing the flow of xenon that was being pumped in and out of the detector for purification. It was found that the xenon came back warmer than it was pumped out, so cleanly cut there was an increase of the temperature inside the detector. Then the liquid level was increased by decreasing the temperature in the detector (or decreasing the flow). When all the Pt-100’s were clearly submerged in the liquid, the capacity read by the Arduino increased about 0.5 pF. This is less then was expected based on the rough estimate with the meshes assumed to be two massive parallel plates, but still a significant increase. Using formula 4.7 one can calculate that the expected increase is 12.23 pF. When the temperature was increased again, the temperature sensors got back into the gas and the capacity decreased.

Figure 21: Schematic view of the meshes and temperature sensors around the intended borderline of the liquid and gaseous xenon. The provisional level meter is made up of the gate mesh and the anode mesh and it is exactly between these two meshes were the liquid level should be. The locations of Pt-100’s numbered 17, 18, 19 and 20, respectively from bottom to top, are shown. The numbers refer to distances in mm.

With the finalized setup Arduino setup, the necessary precision of 1 mm was achieved and even improved. In tests with the provisional level meter, a precision of 0.030(7) pF was achieved. This corresponds to about 0.2 mm, which is more than sufficient for our goals.

Figure 22: Temperature versus time readout of the labview slow control program. As soon as sensors 18 and 19 are submerged in the liquid (this can be seen because the temperatures drop significantly), the capacity increases. In the graph this happens shortly after line A. Sensors 19 and 20 get back into the gas around line B. When the liquid level decreases and the PT-100’s one by one get back into the gas, the capacity decreases. When finally also sensor 18 is out of the liquid, shown shortly before line C, the capacity is

5

The safety system

5.1

Theory

In the XAMS detector liquid and gaseous xenon is kept under a pressure of about 2 bar. It is possible that the pressure inside the detector rises due to an increase of the temperature or an influx of xenon. Of course, normally these two parameters are monitored by the crew manning the experiment, but unforeseen situations can always happen and the crew can also make mistakes. During the night and in the weekends no one is in the laboratory. The most realistic emergency scenario is that during one of these moments the pulse tube refrigerator (PTR), which is responsible for maintaining the right temperature inside the detector, stops working. This can be either due to a power failure at the Nikhef building, or because of a failure of the PTR itself. In both cases the temperature inside the detector will rise and thus more LXe evaporates, which results in a higher pressure inside the detector. The steel construction of the detector has a security mechanism that will release all the xenon from the detector when a pressure of 3.5 bar is reached, in order to protect any vulnerable parts, like the PMTs, inside the detector. Any xenon released this way is lost. Xenon is quite expensive, so therefore it was decided to build a safety system to prevent any xenon spillage. The safety system is to make sure the pressure never reaches the value at which the security mechanism on the detector releases the xenon. The basic idea of the system is that, if the pressure inside the detector threatens to reach 3.5 bar, the safety system cools the xenon in order to lower the pressure instead. For this reason a steel coil was constructed in the upper part of the detector. This is called the cooling element. It was decided to use liquid nitrogen to cool the detector. To control the liquid nitrogen flow trough the pipe, the system shown schematically in figure 23 was build. The system works in the following way. A dewar with liquid nitrogen kept under pressure of about 2 bar is connected by means of a flexible steel pipe to the cooling element inside the detector. At the end a solenoid valve was placed. This is an electronically operated valve that can closed an opened by varying the voltage applied to the valve. The valve from the dewar to the cooling element is always open. This means that if the pressure is normal, there will be nitrogen inside the cooling element and the flexible pipe. This nitrogen will be in gaseous form, because these pipes are relatively warm when compared to the dewar. The no-return valve between the dewar and the pipe makes sure the dewar and

Figure 23: A schematic overview of the safety system. The solenoid valve controls the flow of liquid nitrogen. The rest of the valves are attached in principle for safety reasons.

the pipe are two separate volumes. When the pressure gets too high, the solenoid valve at the end is opened. As a result of the over pressure in the dewar, a flow of liquid nitrogen will now go trough the cooling element, cooling down the xenon. In figure 7 parts of the safety system implemented within the XAMS experiment can be seen. The liquid nitrogen dewar is the big dewar in the left-center.

The solenoid valve is closed as long as an voltage of 24 V is applied. The solenoid valve is connected to a relay on the pressure sensor. This relay works with an upper threshold and a lower threshold. Only when the pressure reaches the upper threshold, this relay is closed. There is no voltage over the valve and subsequently it opens. The relay opens only again when the pressure is lower than the lower threshold. When this happens, the current is restored and the solenoid valve is opened. The purpose of the upper and lower threshold is to prevent the relay from closing and opening every few seconds if the pressure varies a bit around the critical value. Typically, the

5.2

Practical considerations

This section deals with the practical considerations about the construction of the safety system described in subsection 5.1. In order to estimate how long it would take to cool down the pipes leading to the cooling element multiple tests were done.These tests are needed in order to determine how long it takes before liquid nitrogen is flowing trough the cooling element when the solenoid valve is opened. The tests used the components as the final version of the system, except for the cooling element, which was replaced by a comparable 1 meter long straight steel pipe. The first test was done by opening the valve on the dewar to the rest of the system and opening the solenoid valve. Some droplets of liquid at an irregular rate were observed to come out of the pipe after 15 minutes. The second test was done in exactly the same way, but the flexible pipe and the steel pipe were insulated with foam insulation material. This time it took about 7 minutes until a steady flow liquid was coming out. Before that, only nitrogen gas was released. This was found acceptable for our application. It was therefore decided to insulate the flexible pipe and the small bits of steel pipe that connect the cooling element to the flexible pipe and the solenoid valve. An additional benefit of this insulation is that it also stops most of the condensation forming on the pipes.

The flexible steel pipe is connected to the dewar with a connector piece that also includes a no-return valve and a safety valve. The no-return valve makes sure no nitrogen can flow back in the dewar. It has the consequence that the flexible pipe and the cooling element form a closed volume when the solenoid valve is closed. To prevent any build up of pressure inside this part due to heating of the nitrogen gas, there is the safety valve that will release gas when the pressure reaches 10 bar. The steel pipe and the cooling element are estimated to hold a pressure of about 50 bar. It is possible that if the solenoid valve was closed after being opened for some time, there is some liquid nitrogen left in the pipe. When liquid nitrogen expands to gas, its volume increases 694 times, so a safety valve is really needed. This number is called the expansion factor.

The dewar also includes a safety valve itself, in order to release any pres-sure build up caused by evaporating liquid nitrogen inside the dewar. The dewar comes attached with a pressure meter and a liquid level meter. To be absolutely sure of the amount of useful nitrogen left in the dewar, the dewar was mounted on a scale. The empty dewar was weighted beforehand. It is necessary to keep a close eye on the amount of liquid nitrogen left.

5.3

Emergency protocols

In addition to the safety system described above, there was more hardware and software installed to prevent any potential dangerous situations when there is no crew inside the laboratory. The main goal is to warn the crew when something has gone amiss, so that they can get to the laboratory as soon as possible.

When the measured pressure inside the detector gets too high, the lab-view program monitoring the pressure value immediately sends an email and a SMS text message to all crew members of the experiment. The SMS is sent using an Internet web service that converts an email into a SMS and sends it to the right mobile phone number. When the crew members arrive, they can recuperate the xenon from the detector. The computer, the solenoid valve and the pressure sensor are all connected to interruptible power supply (UPS). In case of a general power failure in the Nikhef building, the UPS maintains the same voltage as the power socket. The batteries of the UPS are sufficient to keep all components of the experiment working for 49 min-utes. When the UPS detects a power failure, email and SMS messages are immediately sent to the crew of the experiment, just like in the case of a overpressure. A dongle provides access to the Internet in the case the net-work also fails. A screen shot of the labview program monitoring the alarm system is shown in figure 24.

In addition to these warning systems used when no one is in the labo-ratory, there are also alarms that warn the experimentalists when they are working inside the room. For example when the percentage of oxygen in the room gets dangerously low. This could be due to an unforeseen release of other gases, such as nitrogen, that are stored and used in the room.

Figure 24: A screenshot of the labview program that controls all the different alamrs associated with the XAMS detector. Certain alarms, such as the email and sms warnng system, need not to be on when there are people in the lab and thus can be turned off using this program.

6

Acknowledgment

It was great to see how things worked in a research group and to be part of a real experiment. Sometimes it can be really frustrating when things suddenly stop working (like the level meter) but it can be really rewarding when they do work and show nice results. I want to thank especially Erik Hogenbirk, Rolf Schon, Matteo Alfonsi and Patrick Decowski for their readiness to help and clear explanations.

References

[1] E. Aprile and T. Doke. Liquid xenon detectors for particle physics and astrophysics. Reviews of Modern Physics, 82:2053–2097, July 2010. [2] C. L. Bennett, D. Larson, J. L. Weiland, N. Jarosik, G. Hinshaw, N.

Ode-gard, K. M. Smith, R. S. Hill, B. Gold, M. Halpern, E. Komatsu, M. R. Nolta, L. Page, D. N. Spergel, E. Wollack, J. Dunkley, A. Kogut, M. Limon, S. S. Meyer, G. S. Tucker, and E. L. Wright. Nine-year Wilkin-son Microwave Anisotropy Probe (WMAP) Observations: Final Maps and Results. The Astrophysical Journal, Supplement, 208:20, October 2013.

[3] Gianfranco Bertone, Dan Hooper, and Joseph Silk. Particle dark matter: Evidence, candidates and constraints. Phys.Rept., 405:279–390, 2005. [4] Douglas Clowe, Marusa Bradac, Anthony H. Gonzalez, Maxim

Marke-vitch, Scott W. Randall, et al. A direct empirical proof of the existence of dark matter. Astrophys.J., 648:L109–L113, 2006.

[5] Johannes Schulz. Design of a 2-Phase Xenon Time Projection Chamber for Electron Drift Length Measurements. Master’s thesis, Institut f¨ur Kernphysik Mathematisch-Naturwissenschaftliche Fakult¨at Westf¨alische Wilhelms-Universit¨at M¨unster, 2011.