Characterisation of a radon monitor and implementation into XAMS

(1)

Characterisation of a radon monitor

and implementation into XAMS

Bachelor Project Report

Physics and Astronomy

Isis P.M. Hobus

April 7, 2020

Student number 11328614

Daily supervision Alvaro Loya Villalpando MSc. Dr. Stefan Br¨unner

Supervisor Prof. dr. Auke-Pieter Colijn Second examiner Prof. dr. Marcel Vreeswijk Institute Nikhef

University Universiteit van Amsterdam, Faculty of Science Vrije Universiteit Amsterdam, Faculty of Science Credits 15 EC

(2)

Abstract

An important intrinsic background signal in direct dark matter detection is 222Rn. In this work a radon monitor is characterised and implemented into XAMS, the research and de-velopment dark matter detector at Nikhef. The characterisation is done by determining the detection efficiency of the radon monitor when it is filled with xenon at a pressure of 2.0 bar. This is shown to be Po-218 = 0.01994 ± 0.00018 and Po-214 = 0.02283 ± 0.00019. The

efficiency is seen to be dependent on the pressure inside the monitor, which can be described by a parabola. With the determined efficiencies and the detected polonium activities, the radon activity of the gas in the radon monitor can be calculated.

Populair wetenschappelijke samenvatting

Sinds vorige eeuw zijn natuurkundigen druk met het verklaren van de missende massa in ons heelal die ook wel bekend is als donkere materie. Een veelbelovende kandidaat is de ‘Weakly Interacting Massive Particle’, ook wel bekend als de WIMP. Het XENON-experiment hoopt dit donkere materie deeltje te observeren door zijn potentiele interactie met vloeibaar xenon te detecteren. In deze zoektocht naar donkere materie kampen de xenon-detectoren met een inwendige actergrondbron: het radioactieve radon, dat afkomstig is van de detectormaterialen. Om inzicht te krijgen in de bijkomstige verstoring is het handig om te weten hoe actief te radon in de detector is. Hiervoor kan een radonmonitor worden gebruikt. In deze scriptie is er onderzoek gedaan naar de efficientie van een radonmonitor. Na een karakterisatie is de monitor ge¨ınstalleerd in het gassysteem van de replica donkeremateriedetector XAMS, die door Nikhef gebouwd is voor het onderzoek en verbeteren van soortgelijke detectoren.

(3)

1 Radon in the XENON dark matter experiment

Over the past century evidence has been gathered to indicate the existence of dark matter. The amount of normal matter would not be enough to explain the current structure of our uni-verse, because the gravity of it is not enough to have formed galaxies. The weakly interacting massive particle (WIMP) is our best guess for a dark matter particle. As the name implies, the particle is relatively heavy and only interacts through the weak interaction or gravity. A possible way to study and prove the existence of WIMPs is through direct detection, which relies on the dark matter particle to interact with regular matter.

1.1 The XENON dark matter experiment

The XENON dark matter project is an experiment aiming to directly detect WIMPs. For this, a detector with a cylindrical shape is mostly filled with liquid xenon (LXe) and a layer of gaseous xenon (GXe) on top. A cryogenic system is used to liquefy the xenon, and the TPC is pressurized and insulated to maintain the liquid-gas phase. Xenon is a useful medium for direct detection because it is a heavy element with a large nucleus. This causes the cross section to be large and makes WIMP interactions more likely to happen. Liquid xenon has scintillation properties that can be used for dark matter detection purposes, and an added bonus is that the boiling point of xenon is relatively high with respect to other target materials, making it easy to liquefy.

In figure 1 the working principle of a time projection chamber as it is used for XENON is shown. Particles that enter the LXe can interact with xenon nuclei in two different ways: it can scatter off the nucleus, which is likely to happen with neutrons and is expected to happen for WIMPS. The particle can also interact with the electronic shell of the atom, which is more likely to happen for gamma rays and beta particles. These two interactions generate nuclear recoils (NR) and electronic recoils (ER) in the xenon atom respectively. Both NR and ER produce a light signal through prompt scintillation (S1), which will mostly be detected by an array of photosensors placed at the bottom of the TPC. Another array of photosensors, placed at the top of the TPC, observes the S1 signal at a lesser extent. This is due to the large refraction index of the LXe (1.69), that causes any light travelling towards the liquid-gas interface to reflect internally. Next to this, the particle interaction also ionizes the xenon atom, which will liberate electrons. Free electrons are drifted upwards towards the GXe by an electric field. This drift field is generated by a negatively charged cathode mesh that is

(5)

mesh and a positively charged anode mesh that is located just above the liquid-gas interface. This field, as its name implies, accelerates and extracts the electrons from the LXe into the GXe. The acceleration produces a second scintillation signal (S2), which is proportional to the amount of extracted electrons. The S2 signal is mostly detected at the top array of photosensors. The position of the particle interaction can be reconstructed in the horizontal plane by using the distribution of light on the array of sensors. The time difference between the first and second signal, the drift time, gives the depth of the particle interaction, since the drift velocity of the electrons is well known [1]. The positioning, based on a time delay of the second signal is the basic principle of a time projection chamber (TPC) as used in the XENON detectors. A detailed description of the instruments and detection principle of the XENON1T detector can be found in [2].

Figure 1: An illustration of particle interaction inside a time projection chamber. When a particle interacts with an atom in the liquid xenon volume it creates a light signal through scintillation (S1). Electrons which were liberated in the particle interaction are drifted to the liquid-gas interface by an electric field, where they will get extracted into the gaseous xenon by a second, stronger electric field. The extraction of the electrons creates another scintillation signal, S2, proportional to the amount of electrons involved. Arrays of photosensors at the bottom and top of the time projection chamber detect the S1 and S2 signals. Figure taken from [3].

The XENON experiment is operated in the underground laboratory of Gran Sasso in Italy, where 1400 meters of rock shield the detector from cosmic rays. XENON10 [4] was the first detector of its kind, which was installed in March 2006. It contained 15 kg of LXe it collected 59 live days of data. XENON100 [5] followed, which had a total xenon content of 165 kg. The next and most recently commissioned detector is XENON1T [2]. With a total xenon

(6)

volume of 3.2 tonnes it was the first WIMP dark matter detector operating at the tonne scale. XENONnT is the newest upgrade in the series of detectors, which will have a total xenon mass of 8 tonnes and as of writing the construction of the detector is in its final phase.

1.2 Radon as a background

One of the largest challenges experienced by sensitive, low-rate experiments like XENON is identifying and minimizing background events. The detector itself forms a background source due to trace amounts of uranium, among other things, that is inevitably present in its materials and constantly emanates222Rn (referred to as radon). The materials that were in close contact with the LXe caused a measured 222Rn activity concentration of 10 µBq/kg in XENON1T, where the concentration of other radon isotopes are negligible [6]. The 222_Rn

decay chain is shown in figure 2, with alpha emitters shown in purple, beta emitters shown in green and stable isotopes shown in brown. The half lives, decay energies and branching ratios are taken from [7] and are used throughout this thesis.

Radon has a lifetime of 3.82 days, which gives it time to travel longer distances through con-vection and therefore it can also be found homogeneously distributed inside the target LXe volume. Alpha decays in the radon decay chain have a large energy compared to the expected nuclear recoils originating from a WIMP interaction, which are in the 10 keV range [8]. Be-cause of this, alpha decays are easily distinguished and do not form a relevant background. Most of the beta interactions can also get distinguished and excluded because they happen through electronic recoil interactions. Electronic recoils produce significantly more free elec-trons than nuclear recoils, causing a characteristic large S2 signal, opposed to a similar S1, which makes them recognizable. Still, there is a possibility that only a fraction of the charge freed by the electronic interaction is accelerated and collected by the extraction field. This would be the case when impurities in the xenon capture the drifting electrons, or when the freed electrons get attracted to the TPC walls by plated, positively charged impurities. The smaller S2 signal causes the event to mimic a nuclear recoil.

The most critical background signal in the radon decay chain is 214Pb, which decays through beta decay. The subsequent214Bi beta decay can be distinguished, since it is quickly followed by the214Po decay (t1/2,P o−214= 164 µs). Due to insufficient time resolution, these two events

are mostly seen as one (BiPo event). The energy corresponding to a BiPo event is in the range of an alpha decay, which is why these isotopes do not form a threatening background. The radioactive isotopes following after the214Po decay are also not a relevant background source.

(7)

get eliminated with a fiducial volume cut, which excludes the edges of the detector from the target volume.

This intrinsic background is impossible to shield, and so detector materials need to be selected carefully based on their radio purity. To reduce the remaining radon in the detector cryogenic distillation is used, which has been proven to be an effective purification method for liquid xenon detectors [9].

Figure 2: The radon decay chain. Each square represents an isotope and has the corresponding mass number and half life displayed. Alpha decays are shown in purple, beta decays in green and stable isotopes in brown. The half lives, decay energies and branching ratios are taken from [7].

(8)

2 Radon monitoring in the XAMS detector

The dark matter research group at the Dutch national institute for subatomic physics (Nikhef) is part of the XENON collaboration. For research and development purposes a small dual-phase detector was build at Nikhef. The detector is called XAMS (Xenon AMSterdam) and holds roughly 6 kg of xenon. To maintain the liquid-gas phase the detector is operated at a pressure around 2.0 bar. In figure 3 a picture of the TPC and cryogenic system is shown. A detailed description of the system can be found in [10].

The goal of this thesis was to implement a radon monitor into the existing system for future radon measurements in XAMS. In this chapter the XAMS system is introduced, as well as the radon monitor itself.

(9)

2.1 XAMS

The TPC of XAMS is equipped with a photomultiplier tube functioning as the bottom pho-tosensor. Recently, the TPC was upgraded with an array of silicon photomultipliers placed at the top [3], which allows for three dimensional position reconstruction in XAMS.

While running XAMS, xenon is continuously purified by circulating it through the high tem-perature SAES MonoTorr 3000 getter [11]. The getter is a chemical filter that filters the xenon from electronegative impurities. Electronegative impurities are molecules or atoms that can absorb the drifting electrons in the TPC, which would cause a disturbed representation of the S2 signal. Next to this, the getter also filters out H2O, which is needed to keep the

xenon transparent for scintillation light. For circulation, the liquid xenon is pumped out of the TPC into a heat exchanger where it gets evaporated. A double walled membrane pump pushes the gas through the getter after which it will be directed back to the heat exchanger. Here, the gas will get liquefied again before re-entering the TPC. This cycle is also called the purification or circulation route.

In 2017, a radon insertion system (RIS) was installed into the gas system [12], which is illustrated in figure 4. The RIS was placed after the circulation pump and before the getter, and produces 222Rn using a 226Ra source with an activity of 22.2 kBq. The accumulated

222_{Rn can (partially) be injected into a volume called the sample cylinder by opening valve}

V23. If the sample cylinder was filled with xenon beforehand, the injected radon sample can distribute homogeneously in the xenon before getting introduced to the circulating gas by opening V22. An analog pressure gauge is available to monitor the pressure inside the sample cylinder. Valves V24 and V25 can be used to evacuate the RIS or to expose it to the ambient air.

Figure 4: A diagram of the radon insertion system, consisting of a sample cylinder, a pressure gauge and a226Ra source. The valves are marked V22-V25.

(10)

2.2 Radon monitor

The radon monitor is a gas-tight chamber with a volume of 2.13 liter in which a gaseous sample can be injected. It is illustrated in figure 5. Inside the chamber sits a smaller dome-shaped vessel on which a high voltage can be applied. The A300-17 ABM passivated implanted planar (PIPS) detector from Canberra [13] is placed in the middle of the dome and acts as a semiconductor alpha detector. Because the PIPS detector is mostly sensitive to alpha decays, we will focus on detecting the polonium isotopes. The radon itself cannot be detected, because its decay often happens too far away from the surface of the PIPS detector.

Figure 5: A render of the radon monitor. The gas-tight outer vessel has a smaller vessel inside in which an alpha PIPS detector is placed. A high voltage can be applied on the inner vessel. The gas gets injected into the middle of the monitor through a teflon pipe. The gas can spread throughout the monitor due to small openings in the inner vessel. The outlet is an opening in the outer vessel and is not included in this render as it is located on the other half of the monitor. In this figure the radon monitor is displayed upside down relative to how it is installed in the characterisation setup and XAMS gas system.

(11)

2.2.1 Detection principle

222_{Rn decays into}218_{Po through ejection of an alpha particle from its nucleus. Alpha particles}

are positively charged, so when they travel through the electron cloud of the daughter particle there is a high potential to drag some electrons along. This causes around 50% [14] to 87% [9] of the 218Po atoms to be positively charged. The ionization holds for some short amount of time before electrons in the surrounding material recombine with it to neutralize the ion. A positive ion that is produced inside the innermost vessel of the radon monitor can be mobilized by an electric field. This electric field is formed by the voltage that is applied on the inner vessel, together with the grounded surface of the detector. A COMSOL [15] simulation of the electric field distribution throughout the monitor is shown in figure 6. When the positive ion is created in a location where the field accelerates it in the direction of the PIPS, it will plate down on its surface where it will stay. When the atom then decays and emits an alpha particle in the direction of the PIPS detector it will move through the semi-conductor detector and transfer its energy to the depletion region of the pn-junction. This energy deposition will create electron-hole pairs. The produced electrons will get pulled towards the positively doped region and the ‘holes’ will get pulled toward the negatively doped region, creating signal. The detection of a 218Po decay is illustrated in figure 7.

Figure 6: A COMSOL simulation of the distribution of the electric field inside the radon monitor when a voltage of 2000 V is applied. The color scale represents the amplitude of the electric field at a specific location in the monitor. Image adapted from [16].

(12)

Figure 7: The detection of218_{Po decay illustrated. The radon monitor is filled with radon enriched}

xenon gas (1). When the radon decays, a fraction of the218_{Po atoms that are created are ionized (2).}

The ions get mobilized by the electric field in the inner vessel, and can plate down on the surface of the PIPS detector (3). When a plated isotope decays in the direction of the PIPS detector, it creates a signal that is proportional to the decay energy (4).

(13)

The signal that gets produced in the PIPS detector is amplified by the 2003BT Canberra silicon detector preamplifier, after which it is transmitted to a spectroscopy amplifier (model 2021 from Canberra). Then, the signal is directed to the EASY-MCA, a multichannel analyzer from ORTEC [17]. The energy spectrum of the detected events can be recorded and saved using MAESTRO software by ORTEC. The analysis of the data was done using python. Since the single measurements only give the accumulated number of events during the mea-surement period at a certain energy, the time evolution of the activity within a data set cannot be seen. In this work, the activity is averaged for each data set and is associated with the time

t = t0+

∆t

2 (2.1)

for investigating the time evolution between different data points. In equation 2.1 the time between the start of the measurement and the radon injection is given by t0 and the duration

of the measurement given by ∆t. In this estimation the decay is not taken into account, which holds for the measurements done in this work due to their relatively short length: in an hour time, a radioactive isotope with a decay constant of λRn = 2.098 · 10−6 s−1 loses 0.8% of its

activity. For longer measurements, the decreasing decay needs to be taken into account.

2.2.2 Implementation of the radon monitor in XAMS

Before integrating the radon monitor into XAMS, several characterisation measurements have been done. This was necessary to understand the response of the monitor to operational conditions as a part of the XAMS setup. The detailed description of all characterisation measurements is given in chapter 3. After the characterisation measurements the components of the radon monitor were cleaned in two ultrasonic baths, with exception for the PIPS detector. The cleaning procedure is discussed in appendix A.

The monitor was installed as illustrated in figure 8. The piping and instruments diagram of the XAMS gas system is shown in appendix B. The monitor is placed before the RIS and the getter, so that the (radon enriched) gas that passes through the monitor is pumped from TPC. In that case, the measurements in the radon monitor reflect the radon concentration in the TPC. In this setup, the radon monitor can be used in two ways: online and static. For an online measurement, the old circulation path is bypassed by closing valve V26 and opening V27 and V28. This way, the radon monitor has become part of the circulation route and is taking measurements as the xenon flows through it. For a static measurement, xenon gas can get trapped in the radon monitor by opening valves V27 and V28 for a short amount of time, after which they will be closed and V26 is opened to bypass the monitor. Valve V29 is a port to the ambient air and should be closed during measurements. The piping behind

(14)

valve V29 is added to keep the operation of the radon monitor flexible. For example, if new characterisation measurements are desired, a (calibrated) radon source could be installed here. It also adds the possibility to evacuate the radon monitor. Another valve, V30, is planned to be added in the near future. This enables the user to operate the radon monitor separately from the gas circulation of XAMS. Together with the valve V30 a pressure sensor will be installed. An overview of suggested measurements using this upgraded setup will be given in chapter 4.1.

Figure 8: A simplified diagram of the circulation gas system with the radon monitor implemented. Valves V11-V13 belong to the getter system, which is all together marked by the blue section. The radon insertion system and its valve V22 are indicated by the pink section. Valves V26-V30 are used for operating the radon monitor and are marked by the yellow section. The red components are not yet part of the system and will be added in the near future.

(15)

Figure 9: A picture of the radon monitor, implemented into the gas system of XAMS. The valves needed to operate the radon monitor are marked.

(16)

3 Characterisation of the radon monitor

Before the implementation of the radon monitor to the XAMS gas system, the monitor was characterised. For the characterisation, the efficiency of the detector was probed at differ-ent high voltages and pressures. With the detection efficiency the radon activity inside the monitor can be determined.

3.1 Overview of the characterisation measurements

For the characterisation of the radon monitor three sets of measurements were done. An overview of the three sets are given in table 1. For all these measurements the monitor was filled with xenon as a carrier gas. Next to this, the voltage bias on the detector was kept at 40 V and the monitor was kept at room temperature.

For the first two sets of measurements a single radon injection was done. The radon had been accumulating for 3.06 days in the226Ra source, where it reached an estimated activity of 54.1 Bq. The first set of measurements was taken on the 6th of February and studied the effect of the high voltage on the inner vessel, while the gas pressure inside was kept at 2.0 bar. These will be discussed in section 3.5.1. The second set was taken from the 7th until the 26th of February, where the voltage on the inner vessel was set to 2000 V and the pressure inside the monitor was 2.0 bar. During this period, the energy spectrum was recorded in 35 data sets. The first measurement of the set spanned over half an hour, while the rest of the measurements all took an hour of data, totalling to 34.5 hours of data.

The third and last set of measurements was taken from the 26th of February until the 1st of March, where the pressure inside the detector was varied. Radon was newly injected and had been accumulating for almost 20 days, which corresponds to an estimated activity of 123.5 Bq. The high voltage on the inner vessel was set to 2000 V. These measurements show the pressure dependency of the efficiency and are discussed in section 3.5.2.

Measurement set Date radon injection Injected activity [Bq] Measurements taken on Voltage on

inner vessel [V] Pressure [bar]

#1 06-02 54.1 06-02 500-2500 2.0

#2 06-02 54.1 07-02 till 26-02 2000 2.0 #3 26-02 123.5 26-02 till 01-03 2000 1.0-2.6

(17)

each radon injection the monitor was evacuated via valve C4. Five minutes of pumping shows to be sufficient to reach a vacuum of 10−2 millibar, determined with the Compact Pirani Gauge from Pfeiffer Vacuum [18]. Radon gas emanates from a 127 Bq 226Rn source and accumulates in the volume between valves C2 and C3, the so called sample line. The radon enriched carrier gas will flow into the evacuated monitor after C3 is opened. When the first part of the accumulated radon gas is injected, C3 is closed again. To get the remaining radon gas into the monitor, the sample line needed to be flushed again. This is done by opening valve C2, so the carrier gas flows into the sample line which is at a lower pressure. After the expansion, C2 is closed and in turn C3 is opened to have the radon enriched carrier gas flow into the detector, after which C3 is closed again. The flushing procedure is repeated three times for every radon injection. Additional carrier gas can be added through C1. When the monitor is at the desired pressure, it is ready for the measurements.

(a)

(b)

Figure 10: (a) The characterisation setup illustrated. The black lines represent the gas lines. Gas can directly be injected into the radon monitor through valve C1, or can be directed into the so called sample line, between C2 and C3. The radon monitor can be opened to the ambient air with valve C5 and can be evacuated through a vacuum flange after C4. The green components represent electronics. The cables for the high voltage (HV) input, the bias voltage (Vb) input and the signal output are shown, together with the pre-amplifier. (b) A photo of the radon monitor in the characterisation setup.

(18)

3.2 Quantifying the amount of radon

To calculate the amount of radon that gets injected, it is important to know how long it has been accumulating inside the sample line. When accumulating, radon builds up in the sample line until secular equilibrium is reached.

Secular equilibrium is the situation in which the production rate of a radioactive isotope is equal to its decay rate. In this case, the decay rate of226Ra determines the production rate of222Rn. When secular equilibrium is reached, the amount of radon will remain constant:

dNRn dt = λ|RaN{zRa(t)} production rate −λRnNRn(t) | {z } decay rate = 0 (3.1)

In equation 3.1, λ is the decay constant and N the quantity of the radioactive isotope. It takes several222Rn half-lives to establish equilibrium, as is shown in figure 11. To calculate the radon build up in this figure, a 226Ra source with an activity of 127 Bq is assumed and thus secular equilibrium is reached when the radon activity reaches the same value. Secular equilibrium can only occur for radioactive daughters that have a much shorter half-life than their parent nuclide. In that case, the production rate can be assumed constant due to the timescale that is being considered.

It is likely that some of the accumulated radon will be left behind in the radium source when injecting. To know how much radon gets emanated into the sample line a calibration needs to be done, which will be discussed in section 4.1. For the calculations in this work, it is assumed that all of the accumulated radon is injected.

Knowing the radon builds up as shown in figure 11 and knowing the accumulation time, the injected amount of radon and thus its activity can be calculated even though secular equilibrium has not yet been reached. The activity of radon inside the monitor after injection (at t = 0) is given by the function for radioactive decay:

ARn(t) = ARn(t = 0) · e−λRn·t. (3.2)

The volume of the detector is well known, which allows to calculate a corresponding activity concentration inside the radon monitor.

3.3 Energy calibration

(19)

Figure 11: A calculation of the accumulation of222Rn inside the sample line over time. A226Ra source with an activity of 127 Bq is assumed. Secular equilibrium is reached when the radon activity reaches 127 Bq.

alpha peaks, with their tail towards the lower energies, can be explained by the alpha particle not depositing all of its energy into the depletion zone. This increases the chances to detect a lower energy than the actual decay energy of the alpha particle. In the energy spectrum we can see that the 210Po peak overlaps with the 218Po peak. This needs to be accounted for when calculating the activity of218Po, since there is some delay in the 210Po events that can cause a background signal. The 210Po background is discussed in appendix C. Next to this, a difference between the shape of the218Po and214Po peak can be seen. For 218Po, the counts decrease faster at higher energies, whereas the 214Po peak is more broad. A possible explanation is the short half-life of214Po (164 µs), causing a BiPo event as explained earlier in section 1.2. At the lower channels (0-60) another type of event can be distinguished. The source of this signal is not investigated in this work.

The channel numbers belonging to the maximum of the peaks can be plotted against their corresponding decay energy, as it is done in figure 13. This is used to do a calibration, fitting a linear line in the shape of E = a · C + b, where E is the decay energy and C is the corresponding channel number. The fit yields a slope of a = (21.269 ± 0.024) keV/channel, with an intercept b = (50 ± 7) keV. This energy calibration is only reliable for alpha particles with a decay energy that is similar to that of the polonium isotopes. The full energy scale cannot be assumed linear, as indicated by the nonzero intercept with the y-axis.

(20)

Figure 12: The energy spectrum of the combined measurements of the second characterisation set. The measurements add up to a total of 34.5 hours of recording time. The peaks from the polonium isotopes are marked.

Figure 13: The three peaks for the polonium isotopes fitted with the linear function E = a · C + b, where E is the decay energy and C is the corresponding channel number. The fit parameters are given

(21)

3.4 Activity determination

To determine the activity of a peak, a region of interest (ROI) in the spectrum is defined. To have a measure for setting the limits of the ROI, a Gaussian curve is fitted to obtain a full width half maximum (FWHM) value. The proximity of the210Po and218Po decay energy is taken into account when defining the ROI limits, together with the the asymmetrical shape of the alpha peak. The limits have manually been defined as 8·FWHM on the left side of the maximum and 2·FWHM on the right side, and are the same for both the214_{Po and the}218_Po

ROI. An example of this selection criterion is shown in figure 14. With these limits, circa 96% of the counts of the peak are included in the ROI. The counts in the ROI are summed and divided by the time measured to obtain an activity.

Figure 14: Example of the definition of the region of interest. The limit on the left side of the maximum of the peak is defined to be 8·FWHM, with the limit on the right side at 2·FWHM.

After injection, the radon will decay as described in formula 3.2. The time behaviour of the radon daughters up until 210Pb is expected to follow the decay of the radon, once they are in equilibrium. This is because their half-lives are much shorter than that of 222Rn. Directly after radon injection, however, the activity of the radon daughters is zero and first needs to grow in, as shown in figure 15a. This ingrow period can also be seen in the measured data, of which an example is shown in figure 15b. For218Po the ingrow period takes less than half an hour after radon insertion, whereas it takes almost 4 hours for214Po. Once the polonium isotopes are in equilibrium, their activity reflects the radon activity inside the monitor.

(22)

(a) (b)

Figure 15: The period of ingrow after radon injection inside the monitor. (a) A calculation of the activity inside the radon monitor over time. This is shown for the radon daughters up until 210Pb. The different ingrow periods are visible. The214_{Po activity follows the same time evolution as} 214_Bi

since these two nuclides are already in equilibrium with each other. (b) The ingrow is also visible in the measured data. These are the first measurements performed after the second radon injection.

3.5 Detection efficiency

Not all of the activity inside the radon monitor gets measured. The detection efficiency of the radon monitor is defined as the measured activity of a radioactive isotope divided by the activity expected from the amount of radon injected:

= AX, detected(t) AX,expected(t)

(3.3) Where X refers to be the detectable 222Rn daughters: 218Po and 214Po. As described in section 2.2.1, the radon activity is determined by detecting alpha decays of polonium atoms that got collected on the PIPS detector.

Due to isotropic alpha emission only half of the collected atoms on the PIPS can be detected, which sets the theoretical maximum detector efficiency at max = 0.5. The geometry of the

monitor also plays a role for the efficiency. The total volume of the gas chamber is calculated to be 2.13 liters, of which 0.61 liters are encapsulated by the high voltage dome. This comes down to 29% of the total volume being detectable. Combining this with the factor of 0.5 we get max = 0.145. Next to this, not all polonium atoms are ionized upon creation, as

mentioned before in section 2.2.1. When we assume an ionization fraction of 87% [9], the maximum detection efficiency is reduced to max= 0.126. Lastly it is seen in figure 6 that the

(23)

The effective volume can be determined by dividing the measured efficiency by the maximum as described earlier: Veffective= VHV· measured max . (3.4)

The detection efficiency can also be effected by impurities in the gas, since they can neutralize the ions while being drifted towards the detector. The drift time tD is a way to describe how

these impurities play a role in the detection efficiency, which is given by the drift velocity νD

[19]:

νD = µ · E

tD =

d

µ · E. (3.5)

In this equation, the ion mobility is given by µ, the electric field by E and the distance to the PIPS detector by d. With a shorter drift time the effective volume increases, increasing the detection efficiency. Due to the shape of the electric field, the drift time and the efficiency are not linearly related.

Only the fraction of ionized 218Po atoms that get created within Veff drift onto the surface

of the PIPS detector. Subsequent radon daughters in the decay chain, however, also have a chance to be positively charged upon creation, which increases the chance of a 214Po atom to eventually be collected on the PIPS detector. The average detection efficiency of 214Po is therefore expected to be higher than that of 218Po. The data of the second set of characteri-sation measurements, as seen in figure 16, shows how the detection efficiency is constant over time and how it differs for the two polonium isotopes. Here, the average detection efficiencies are calculated to be Po-218 = 0.01994 ± 0.00018 and Po-214= 0.02283 ± 0.00019. The larger 214_{Po efficiency is most obvious for the first few days since the radon injection, when the}

activity was high and thus the error on the efficiency was relatively small. Using the 218Po efficiency and equation 3.4, the effective volume is determined to be approximately 0.10 liter.

(24)

Figure 16: The detection efficiency for 214_{Po and} 218_{Po during the second set of characterisation}

measurements. A voltage of 2000 V was applied on the inner vessel and the pressure inside the monitor was 2.0 bar. The average detection efficiency is seen to be higher for 214Po than for 218Po. The mean efficiencies are given in the figure.

Figure 17: The214_{Po and}218_{Po activity plotted over time. The function A}

measured(t) = A0· exp(−λ · t)

(25)

The second set of characterisation measurements spanned over a period long enough to see the exponentially decreasing activity as shown in figure 17. An exponential function in the form of Ameasured(t) = A0· exp(−λ · t) is fitted. The polonium isotopes have reached

semi-secular equilibrium for these measurements, and therefore they should decay with the radon decay constant. The difference between A0 for the two different isotopes can be explained by

the difference in detection efficiency. When the A0 values are divided by the corresponding

efficiency we get an injected activity of (54.1±0.7) Bq for the218Po value, which agrees with the injected amount of radon, and (55.1±0.9) Bq for the214Po value. The deviating injected activity for214Po can possibly be explained by the single data point taken in the earlier stages of the set. To get a more representative image of the activity around the first few hours after radon injection, the energy spectrum should have been probed more often. When the first measurement is not taken into account, the fit for 214Po yields a begin activity of (1.244 ±0.019) Bq. Dividing this by the detection efficiency gives an injected activity of (54.5 ±1.0) Bq, which agrees with the value that is based on the accumulation time.

Comparing the λ-values that have been determined through the fit with that of radon is a way to see if the polonium isotopes follow the decay rate of radon, as is anticipated. The fits yield decay constants of λP o−218= (2.100 ± 0.015) · 10−6 s−1 and λP o−214= (2.122 ± 0.022) · 10−6

s−1. The radon decay constant is λRn = 2.098 ± ·10−6 s−1 when it is rounded to match the

precision of the fit parameters, where the error on λRn goes to zero. The fit value for the 218_{Po data agrees with λ}

Rn but that for214Po is too high, implicating a faster decay rate. The

first data point can again be excluded from the fit, as discussed in the previous paragraph. This would return a decay constant of λP o−214 = (2.108 ± 0.023) · 10−6 s−1, which does agree

with the decay constant of radon. From this we can conclude that the measured activity of the polonium isotopes reflects that of the radon inside the monitor.

3.5.1 Influence of the high voltage on the efficiency

The electric field that mobilizes the positive ions is of influence on the detection efficiency through the drift time, as been described in equation 3.5. To quantify the dependency on the high voltage, measurements were done varying the voltage between 500 and 2500 V in steps of 250 V. For each voltage step two consecutive measurements of 30 minutes each have been done. The results in figure 18 show the 218Po activity against the different voltages. 218Po is most dependent on the electric field to reach the PIPS detector, since it is the direct daughter of 222Rn and has no chance to get collected on the detector after multiple subsequent decays, as is the case for 214_{Po. We can see that the two measurements are in agreement with each}

(26)

It is anticipated that minimizing the drift time of the ions will lead to a higher detection efficiency, which explains why we see an increasing detection efficiency with increasing voltage in figure 18. At a higher voltage, more ions get plated down onto the surface of the PIPS. From 2000 V and onward we can see the efficiency flattens to a value around 0.020. A possible explanation for this flattening is that all of the available ions are already mobilized at 2000 V, and so increasing the voltage further will not make a difference for the detection efficiency. A voltage of 2000 V was applied for all the other measurements mentioned in this thesis.

Figure 18: The efficiency calculated for218Po at different voltages on the inner dome. Two subsequent measurements where done at each voltage. For the second measurement, the the voltage value is offset slightly so the two data points do not overlap.

3.5.2 Pressure dependence of efficiency

The ion mobility µ in equation 3.5 is dependent on the ion charge q, the temperature T and the diffusion constant D, described by the Einstein relation [19]:

µ = q kB· T

(27)

where ν = q

3RT

M is the squared mean velocity of the gas. The type of gas determines the

molar mass M and the gas constant is given by R. In equation 3.7 the mean free path is described by λ = (σ · n)−1, where σ is the effective cross-sectional area for collision between particles and the pressure dependency is given by the particle density n. Operating at higher pressures cause the particle density to be larger, in turn causing the mean free path and diffusion constant to be smaller. With a smaller diffusion constant the ion mobility will also decrease, causing a longer drift time. Physically this means that at greater particle densities the amount of impurities capable of neutralizing the ions increases too, limiting the ion mobility and causing the detection efficiency to decrease.

For radon measurements in XAMS the monitor is implemented into the gas system, where only xenon will circulate through the monitor. Furthermore, the xenon gas is always kept at room temperature during circulation. This is why the dependency on the gas type and temperature are not quantified in this thesis, even though they do have an influence on the detector efficiency. To get a grip on these dependencies, it is helpful to write out ν in the ion mobility function: µ = q kb λ 2 r 3R M T. (3.8)

Minimizing the thermal motion of the carrier gas is predicted to increase the ion mobility and reduce the drift time. Operating at lower temperatures is therefore expected to result in a higher detection efficiency. Next to this, a smaller molar mass would also result in a larger ion mobility and therefore increase the detection efficiency.

To determine the pressure dependency of the efficiency, measurements were taken at pressures between 1.4 and 2.6 bar in steps of 0.2 bar. Additionally, measurements were done at 1.0 bar. At each pressure the energy spectrum was probed by taking at least 8 data sets, each with 15 minutes measuring time.

The first measurements at each pressure show a period of ingrow. Two examples are given in figure 19. The figures for the remaining pressures can be found in appendix D. In figure 19 the218Po activity is shown for each of the measurements taken at the same pressure. The activity is corrected for the radioactive decay by multiplying it with a factor of et·λRn_{, where}

t is the time since radon injection. A possible explanation for the observed activity ingrow could be that the newly introduced gas is not well distributed throughout the gas chamber. The inlet pipe adds new, radon-free gas into the inner vessel first, which pushes the radon enriched gas into the outer vessel of the monitor. The radon is not yet homogeneously mixed which causes a lower activity in the effective volume. After some amount of time the activity reaches an equilibrium, implicating a uniform distribution of the radon. To adjust for the

(28)

ingrow period, the first 4 measurements are not taken into account when determining the pressure dependence of the detector efficiency.

(a) (b)

Figure 19: The activity of the218_{Po peak for the measurements taken at a specific pressure. A time}

correction of a factor et·λRn _{is made for the activity. An ingrow period can be seen for the first few}

measurements. Afterwords, the efficiency stabilizes to a constant value. The measurements done at 2.0 bar are shown in (a) and the measurements done at 2.6 bar are shown in (b).

An average efficiency is calculated for the remaining data points at each pressure, which where all in agreement with each other. The average energy for the different pressures is shown in figure 20. The activity for the measurements up to 1.6 bar can be recognized as being constant. After this, the efficiency seems to show a decrease. This behavior could be described by a parabola, and so a fit in the form of = a · P2+ b · P + c is done, as can be see in figure 20a. The parameters for this fit are a = (−0.6 ± 0.5) · 10−3 bar−2, b = (0.3 ± 1.7) · 10−3 bar−1 and c = (21.4 ± 1.4) · 10−3. As a measure of testing the fit, the sample variance s is calculated for the fit. This is the difference between the measured efficiency compared to the fit [20]: s = v u u t 1 N − 1 N X i=1 d2 i, (3.9)

with i as the measurement number, N as the total number of data points and di as the

deviation for each data point:

di = 1 −

measured,i(Pi)

fit,i(Pi)

(3.10) The sample variance s is expressed in a percentage deviation from the fit, and is indicated in

(29)

The linear function = a·P +b is fitted in figure 20b, with fit parameters a = (23.0±0.4)·10−3 bar−1 and b = (−1.69 ± 0.22) · 10−3. For this linear fit, the sample variance holds a value of 1.78%. An explanation for the larger sample variance and larger reduced chi-squared can be the relatively large efficiency value measured at 1.6 bar. The parabolic fit gives room for a more constant behaviour of the efficiency at some range of pressures, while also allowing for a pressure dependency at a different pressure range.

(a) (b)

Figure 20: The detection efficiency for218_{Po plotted against the pressure inside the radon monitor. A}

parabolic and linear function are fitted by using the LMFIT package in Python. The fit parameters are given in the figures and the shaded region represent the sample variance of the fit. (a) The pressure dependency fitted with the parabolic function = a · P2_{+ b · P + c. This fit yields a sample variance}

of 1.57%. (b) The pressure dependency fitted with the linear function = a · P + b. This fit yields a sample variance of 1.78%.

(30)

4 Outlook and conclusion

With the radon monitor implemented into the XAMS gas system, insights can be gathered about the radon sources of the system.

4.1 Suggestions for future measurements

The efficiency is now determined on the assumption that all of the accumulated radon in the radon source gets injected into the monitor. The source that was used is a 226Ra source with a well known activity. The known activity does not guarantee that all of the222Rn that gets produced is also available for injection in the sample line; it could be that some radon is left behind in the source itself. In that case, the efficiencies determined in this thesis are underestimated. Because of this, I would recommend to calibrate the source. When the226Ra source is used to inject an amount of radon into a different radon monitor with a well defined detection efficiency, the emanated and injected amount of radon can be determined. When this is done, it could also be interesting to test how the source emanates radon under different pressure circumstances.

For all of the measurements in this thesis the gas in the radon monitor has been static. When circulating through the radon monitor, a gas flow is introduced. Depending on the geometry of the radon monitor this can have an influence on the effective volume, and therefore on the detection efficiency. In [9] a similar radon monitor was characterised, which has the shape of a semicircle. Here, an increasing flow reduced the detection efficiency. This could be due to the location of the gas inlet and outlet of the detector: the circulating gas can flow along the arch of the semicircle while leaving the gas in the middle static, creating ‘dead’ volume. In the XAMS radon monitor, the gas flows into the inner vessel, whereas the outlet is placed in the outer vessel. When introducing a flow in our radon monitor, the separation of the in-and outlet could reduce the possibility to create dead volume. The gas flow in the XAMS gas system can be varied and additional static measurements can be done, which offers the opportunity to study the influence of a flow on the detection efficiency.

4.2 Conclusion

In this thesis a radon monitor was characterised for implementation into XAMS. When it comes to the voltage on the inner vessel of the monitor, the detection efficiency is seen to be optimized at voltages of 2000 V or higher. When a voltage of 2000 V is applied and

(31)

is shown to be dependent on the pressure inside the radon monitor, which can be described by a parabolic function.

With the detected polonium efficiency and its corresponding efficiency the radon activity in the monitor can be determined.

(32)

Acknowledgements

This project would not have been possible without all of the amazing people that surrounded me during it. Alvaro, thank you for your daily support and your confidence in my abilities, you have been incredibly motivating throughout this project without ever losing you enthousiasm. Stefan, the radon master, thank you for being so patient and helpful during the past months, I am very glad you were there to make sure I have been doing this the right way. Auke Pieter, thank you for setting up and supervising this project, the ‘best supervisor’ award displayed in your office is shown to be very well deserved. In addition I would like to thank Peter, Joran, Leonora, Lucas, Gijs and Olivier for the extensive coffee breaks, table tennis matches and bergfests. I have had so much fun and learned a lot from your bright minds. Patick, Auke Pieter and the rest of the dark matter group, thank you for making this research group such an inspiring and inclusive learning space. I am grateful to have experienced this, and I am sure any physicist-in-the-making that encounters the dark matter group would agree with me! Finally, thank you for showing me the unfiltered business of building a XENON detector and trusting me with a shift at LNGS. I am glad I can stay for a little longer.

Papa, Camille, Tiba, Daan en mijn lieve vrienden aan de andere kant van de weg, dank jullie wel voor de schijnbaar moeiteloze ondersteuning; de adviesen, het lachen, al jullie liefde, en alle taart en kaas.

(33)

References

[1] E. Hogenbirk, M. Decowski, K. McEwan, and A. Colijn, “Field dependence of electronic recoil signals in a dual-phase liquid xenon time projection chamber,” Journal of Instru-mentation, vol. 13, no. 10, p. P10031, 2018.

[2] E. Aprile, J. Aalbers, F. Agostini, M. Alfonsi, F. Amaro, M. Anthony, B. Antunes, F. Arneodo, M. Balata, P. Barrow, et al., “The XENON1T dark matter experiment,” The European Physical Journal C, vol. 77, no. 12, p. 881, 2017.

[3] A. Loya Villalpando, “Characterization of silicon photomultipliers for event position reconstruction in a dual-phase xenon time projection chamber,” Master’s thesis, Univer-siteit van Amsterdam, 2019.

[4] E. Aprile, J. Angle, F. Arneodo, L. Baudis, A. Bernstein, A. Bolozdynya, P. Brusov, L. Coelho, C. Dahl, L. DeViveiros, et al., “Design and performance of the XENON10 dark matter experiment,” Astroparticle Physics, vol. 34, no. 9, pp. 679–698, 2011. [5] E. Aprile, K. Arisaka, F. Arneodo, A. Askin, L. Baudis, A. Behrens, E. Brown, J.

Car-doso, B. Choi, D. Cline, et al., “The XENON100 dark matter experiment,” Astroparticle Physics, vol. 35, no. 9, pp. 573–590, 2012.

[6] E. Aprile, J. Aalbers, F. Agostini, M. Alfonsi, F. Amaro, M. Anthony, L. Arazi, F. Ar-neodo, C. Balan, P. Barrow, et al., “Physics reach of the XENON1T dark matter ex-periment.,” Journal of Cosmology and Astroparticle Physics, vol. 2016, no. 04, p. 027, 2016.

[7] R. B. Firestone, The Table of Isotopes. 8th ed., 1996.

[8] E. Aprile, M. Alfonsi, K. Arisaka, F. Arneodo, C. Balan, L. Baudis, A. Behrens, P. Bel-trame, K. Bokeloh, E. Brown, et al., “Analysis of the XENON100 dark matter search data,” Astroparticle Physics, vol. 54, pp. 11–24, 2014.

[9] S. A. Br¨unner, Mitigation of 222Rn induced background in the XENON1T dark matter experiment. PhD thesis, 2017.

[10] E. Hogenbirk, “Development, commissioning and first results of XAMS,” Master’s thesis, Vrije Universiteit van Amsterdam, 2014.

[11] “Entegris website.” https://www.entegris.com/. Accessed: 23-03-2020.

[12] K. van Teutem, “PMT calibrations and radon measurements in a liquid xenon time projection chamber,” Master’s thesis, Universiteit van Amsterdam, 2017.

(34)

[13] “Mirion website.” https://www.mirion.com/. Accessed: 20-03-2020.

[14] J. Albert, D. Auty, P. Barbeau, D. Beck, V. Belov, M. Breidenbach, T. Brunner, A. Bu-renkov, G. Cao, C. Chambers, et al., “Measurements of the ion fraction and mobility of α- and β-decay products in liquid xenon using the EXO-200 detector,” Physical Review C, vol. 92, no. 4, p. 045504, 2015.

[15] “COMSOL website.” https://www.comsol.com/. Accessed: 18-03-2020.

[16] E. Abram, “Characterization of a high pressure radon detector,” bachelor’s thesis, Uni-versiteit van Amsterdam, 2017.

[17] “ORTEC website.” https://www.ortec-online.com/. Accessed: 22-03-2020.

[18] “Pfeiffer Vacuum website.” https://www.pfeiffer-vacuum.com/. Accessed: 27-03-2020.

[19] P. W. Atkins and J. De Paula, Atkins’ Physical Chemistry. Oxford university press, 2010.

(35)

Apendix

A

Cleaning procedure

Before implementing the radon monitor into the XAMS gas system it was disassembled and the compontents were cleaned, with exception of the PIPS detector. The procedure Nikhef recommends and provides in its cleanroom was followed:

1. The first ultrasonic bath was filled with a 5% Tickopur RT5 solution, which was heated to 50 °C. The monitor parts were emerged for a duration of 10 minutes. This bath cleans the components from minerals, rust and grease among other things.

2. After the first ultrasonic bath, the components were rinsed with demiwater.

3. A second ultrasonic bath was filled with a 5% Tickopur R36 solution heated to 50 °C and also took 10 minutes. This solution cleans the parts from oil, grease, organic and inorganic materials.

4. After the second ultrasonic bath the monitor components were rinsed again, after which they were dried with compressed air before getting reassembled.

(36)

(37)

C

Background measurements

To get a feel for the background that is observed with the radon monitor, measurements were done where no radon was injected. For the background measurements, the gas chamber of the monitor was evacuated to a pressure of approximately 10−2 mbar. After evacuating the gas chamber, it was filled with nitrogen gas at a pressure of 1.07 bar. Four subsequent measurements were done and are shown on page 37, where we focus on the energy range of the polonium peaks since this is our region of interest. Note here that these measurements are done with nitrogen instead of xenon as a carrier gas. However, we do not expect an influence from the carrier gas in the background measurements. This is because we assume to have pumped out all of the radon enriched gas, and therefore expect to only measure the radioactive atoms that are left behind on the surface of the PIPS detector.

In the first measurement, done immediately after evacuating the chamber, we still measure some 214Po activity at circa 3 mBq. After almost 2 hours the 214Po activity has already decreased to 0.2 mBq, as can be seen in the second measurement. In the second measurement we can also see a 218Po activity of 0.1 mBq. In the third and fourth measurement, the measured 218Po and 214Po activity stay stable at 0.1 mBq. The background measurements were executed on the 24rd on January. At that moment, the last radon injection was done on the 16th of January, where the injected radon had an activity of 93 Bq. Considering the last injection and assuming the detection efficiency is 2%, we would expect to measure a 218Po and214Po activity of 0.5 Bq when the monitor would not have been evacuated. Knowing this, we can conclude that the evacuation gets rid of approximately 99% of the radon daughters in the monitor.

The third polonium isotope in the222Rn decay chain,210Po, lasts the longest on the surface of the PIPS detector. This is due to the long210Pb lifetime of 22.3 years. For these measurements the210Po activity is 2 mBq at most.

(38)

(39)

Characterisation of a radon monitor and implementation into XAMS