Prototype of a Attenuation Length Monitor for JUNO

Transcription

1 Prototype of a Attenuation Length Monitor for JUNO Masterarbeit am Fachbereich Physik, Mathematik und Informatik der Johannes Gutenberg-Universität in Mainz Heike Enzmann geboren in Wuppertal Mainz, den

2 1. Supervisor: Univ.-Prof. Dr. Michael Wurm 2. Supervisor: Univ.-Prof. Dr. Lutz Köpke

3 A B S T R A C T The Jiangmen Underground Neutrino Observatory (JUNO) with its 20 kt liquid scintillator (LS) detector currently under construction in China. Its main goal is to determine the mass neutrino hierarchy via a precise survival probability measurement of reactor antineutrinos. Using electron antineutrinos from several 53 km distant nuclear reactors, it will be done by precisely measuring the substructure in the oscillated energy spectrum of the reactor neutrino events. A challenging energy resolution of 3% at 1 MeV is required for this measurement. To obtain this resolution in a detector of this scale, the properties of the liquid scintillator have to be optimized: The absorption length should be larger than 60 m and the attenuation length should be over 20 m. Therefore, it must be ensured that only liquid scintillator with high purity is filled into the central detector. In the context of this work the prototype of an On-line Attenuation Length Monitor for JUNO was designed, set up and tested. Its purpose is is to test the transparency of LS batches before they are filled into the central detector. The system determines the medium s attenuation length in a parallel measurement of the intensity loss of a split laser beam over two LS samples contained in tubes of different lengths equipped with two identical CCD-cameras. A test run with Toluol, which has a chemical structure similar to that of the LS for JUNO demonstrated that the On-line Attenuation Length Monitor works and revealed some possibilities for further improvement. iii

4 C O N T E N T S 1 introduction Neutrino Physics Standard Model Neutrino Oscillations Mass Hierarchy Scintillator Principle of Organic Scintillators Light Propagation 11 2 juno Location Central Detector Muon Veto Acrylic Sphere Stainless Steel Structure Buffer Photomultiplier Tubes Liquid Scintillator Composition Purity Requirements Filling Neutrino Detection 22 3 on-line attenuation length monitor Design Setup Light source Light detection Tubes Filling Mounting Components CCD Camera Tubes Laser Evaluation of Results Integration Testing 43 4 test measurements Empty Setup Attenuation length of Toluol Systematic uncertainties 48 5 summary and discussion 49 a appendix 51 iv

5 contents v a.1 Charge coupled device 51 a.2 Function of a Laser Diode 51 bibliography 53

6 I N T R O D U C T I O N 1 The Jiangmen Underground Neutrino Observatory (JUNO) is a new generation neutrino experiment currently under construction in Southern China. While Juno will explore various questions concerning neutrino physics, the primary objective is to determine the mass hierarchy of the neutrinos. For this purpose, the electron antineutrino flux of several nuclear reactors about 53 km away is to be measured with the aid of a 20 kt heavy detector based on a liquid scintillator. In addition to the mass hierarchy, the JUNO experiment will be able to investigate the properties of neutrino oscillation more precisely due to its expected enormous statistics (O(100k) events). By this, the solar mixing parameters θ 12 and m 2 21 shall be measured more precisely. In addition, the detection of supernova neutrinos, solar neutrinos, atmospheric neutrinos, geoneutrinos and possibly proton decay will also be possible in the proton-rich liquid scintillator. For the determination of the mass hierarchy, it is necessary to obtain an energy resolution of at least σ E E = 3%. In order to reach this E energy resolution, 1200 photoelectrons would have to be detected for a neutrino with 1 MeV visible energy. To achieve this resolution in a detector of this scale, the properties of the liquid scintillator have to be optimized. First of all the optical transparency is of utmost importance. The absorption length should be larger than 60 m and the attenuation length should be over 20 m. Also, high optical coverage ( 75%) with efficient photomultipliers (quantum efficiency 35%) is necessary to detect a sufficiently large fraction of the photons reaching the verge of the detection volume. The subject of this master thesis is the design of an On-line Attenuation Length Monitor to monitor the optical transparency of the liquid scintillator (LS) before it is filled into the main detector. The optical transparency of the liquid scintillator is monitored by measuring the attenuation length. Impurities in the liquid scintillator would lead to absorption and thus to a shorter attenuation length. In this chapter, an introduction to the relevant physics is given. Firstly it deals with neutrino physics. In this context, the main focus is on the neutrino oscillation and the mass hierarchy. Then liquid scintillators (LS ) as used in JUNO and their properties are presented. The following part about the JUNO detector takes a closer look at the design of the planned detector and its most important components. Chapter 3 discusses the design and individual components of the Online Attenuation Length Monitor. Finally, first measurements taken 1

7 1.1 neutrino physics 2 Figure 1: Elementary particles in the Standard Modell [2] with the setup of the On-line Attenuation Length Monitor and the results are presented. 1.1 neutrino physics Standard Model The Standard Model of particle physics describes the interaction of the point-like building blocks of all matter. These particles are called fermions [1]. Fermions have a spin of S = 1/2 and are divided into three families. The first family consists of the up-quark (u), the down-quark (d), the electron (e ), and the corresponding electron neutrino (ν e ). Together the up-quark (u) and the down-quark (d) make up protons and neutrons (the so-called hadrons). Together with the electrons, (e ) atoms are formed building all known stable matter. Particles interact via the four fundamental forces: the electromagnetic force, the weak force, the strong force, and gravity. The first three forces are described by the Standard Model, gravity has not been included so far. The forces are mediated by the gauge bosons: the photons (γ) for the electromagnetic force, the gluons (g) for the strong force, and the Z 0 and W ± bosons for the weak force. All the gauge bosons have a spin of S = 1 [1]. The three fermion families and the gauge bosons are shown in Figure 1 green. Coupling is only possible between certain particles. Quarks can couple to all three forces, leptons can only couple to the weak force unless they carry an electric charge in which case they also couple to

8 1.1 neutrino physics 3 the electromagnetic force. It is also possible for gluons to couple with other gluons and for W ± to couple with photons. The Z 0 and W ± have masses of 91 GeV/c and 80 GeV/c, respectively. However, the Standard Model is a renormalizable quantum field theory, where gauge bosons should be massless particles. The introduction of the Higgs field and the Higgs boson solves this problem. Fermions of the 2nd and 3rd family are created by inelastic interactions. These particles eventually decay into leptons and quarks (bound as protons) of the first family. This decay happens through chains via strong, electromagnetic, or weak force. Neutrinos are only subject to weak interaction and are massless in the Standard Model. The restriction to the weak interaction leads to low interaction rates with matter. In order to detect neutrinos, detectors with large active volumes are necessary, as in the Homestake experiment [3] and with the JUNO experiment that is currently under construction in China. Another method is the indirect detection of missing energy in the energy balance of a particle reaction. The observation of neutrino oscillations has shown that neutrinos have a finite mass. Figure 1 shows the currently known upper limits for the masses of the neutrinos. Within the framework of the Standard Model, many predictions have been made in the past, which have subsequently been experimentally confirmed. This applies to the Charm, Bottom and Top Quark. The Z 0 - and W ± - bosons were also correctly predicted. Despite the great success of the Standard Model of elementary particle physics, it will not be the definite proof. It does leave some phenomena unexplained. There are 25 free parameters that do not result from the theory thy can only be defined externally. For example, it does not explain why there are three generations of fermions or where dark matter comes from. In terms of neutrinos, the Standard Model does not incorporate neutrino oscillations or their non-zero masses. Therefore the classic Standard Model will have to be extended and/or modified Neutrino Oscillations Neutrino oscillations describe the change of neutrinos from one flavor eigenstate to another. In this process a neutrino of a particular flavor eigenstate ν α > (with α = e, µ,τ) changes in a different flavor eigenstate. The individual flavor lepton number is not conserved during the oscillation process, whereas the lepton number conservation for all three flavor states taken together is still valid.

9 1.1 neutrino physics 4 parameter value [ev 2 ] sin 2 (θ 12 ) ± sin 2 (θ 13 ) sin 2 (θ 23 ) sin 2 (θ 23 ) m 2 21 m 2 32 m 2 32 (2.19 ± 0.12) 10 2 (0.51 ± 0.05) (normal mass hierarchy) (0.50 ± 0.05) (inverted mass hierarchy) (7.53 ± 0.18) 10 5 (2.44 ± 0.06) 10 3 (normal mass hierarchy) (2.51 ± 0.06) 10 3 (inverted mass hierarchy) Table 1: Current world averages for the neutrino mixing angles and mass differences [4] If neutrinos are massive, the three flavor eigenstates can be described as a superposition of three mass eigenstates ν i > ( with i = 1, 2, 3) with the following relationships: ν α >= i U αi ν i > and ν i >= (U ) iα ν α >. (1.1) α In 1.1, U αi is the Pontecorvo-Maki-Nakagawa-Sakata-Matrix (U PMNS ) which assumes that neutrinos are Dirac particles (ν x = ν x ). The PMNS mixing matrix can be parametrized as follows: c 13 0 s 13 e iδ cp c 12 s c 23 s s 12 c 12 0 (1.2) 0 s 23 c 23 s 13 e iδ cp 0 c c 12 c 13 s 12 s 13 s 13 e iδ cp = s 12 c 23 c 12 s 23 s 13 e iδ cp c 12 c 23 s 12 s 23 s 13 e iδ cp s 23 c 13 (1.3) s 12 s 23 c 12 c 23 s 13 e iδ cp c 12 s 23 s 12 c 23 s 13 e iδ cp c 23 c 13 The parameter s ij and c ij stand for the sine and cosine of respective mixing angle θ ij, (Table 1). The phase factor δ is only different from zero if the neutrino oscillation violates CP symmetry. This is expected but has not been observed experimentally so far. Using the time development of mass eigenstates ν i (x, t) >= e ie it ν i (x, 0) > (1.4) and the PMNS matrix, the probability of a conversion between flavors α β can be calculated for a neutrino ν i (x, 0) >= e ipx ν i > (1.5)

10 1.1 neutrino physics 5 with momentum p emitted by a source x = 0 at a time t = 0: ν α (x, t) >= U αi Uβi eipx e ieit ν β >. (1.6) i,β A difference in the neutrino masses leads to different phase factor in equation 1.6. Therefore flavor content of the final state does not necessarily match that of the initial state. Even if the differences in the neutrino masses are small, this effect can be large for long distances. The time-dependent transition amplitude for the flavor conversion ν α ν β is in this case given by the equation: A(α β)(t) =< ν β ν i,β U βi U αie ipx e ie it. (1.7) Assuming relativistic neutrinos with p m and E p we can write E i = m 2 i + p 2 i p i + m2 i 2p i E + m2 i 2E and re-write Equation 1.7 to A(α β)(t) = U αi U βi exp ( i m2 i 2 with L = x the distance between detector and source. The transition probability P is (1.8) ) L. (1.9) E P(α β)(l) = A(α β) 2 (1.10) ( = U αi Uβi 2 + 2Re U αi Uαj U βi U βjexp i m ) ij L 2 E i j>i with m ij = m 2 i m 2 j. Thus, the probability that a neutrino will be found in its original state after traveling a distance L is given by: P(α α) = 1 P(α β). (1.11) α =β A more stringent derivation of the above presented equations can be found in [5]. Figure 2 shows the probability that an electron neutrino ν e will, for a given energy E and flavor, find itself in the eigenstate ν α > after traveling a particular distance L. The black curve represents the probability, that ν α continues to be found in the electronic flavor eigenstate, while the blue curve shows the probability for the muonic, and the red curve for the tauonic transition.

11 1.1 neutrino physics 6 Figure 2: Neutrino oscillation P(e µ) and P(e τ) Mass Hierarchy Through measurement of neutrino appearance or disappearance it is possible to obtain values for θ ij which are proportional to the amplitude and for m 2 ij which are proportional to the frequency of the oscillation. In most cases it is sufficient to describe the oscillation process with two neutrino flavors. Then, the transition probability can be written as [6] ( ) P(α β) = sin 2 (2θ) sin m2 (ev 2 ) L(m). (1.12) E(MeV) Without additional information, it is not possible to determine the sign of m that leads finally to the current mass hierarchy problem. The values for θ and m are extracted in experiments with neutrinos from different sources: Solar Neutrinos are the first neutrino source in which evidence for neutrino oscillations were found. They are electron neutrinos produced in the process of nuclear fusion in the sun. In total, neutrinos are emitted in five different energy spectra (Figure 3). The maximum energy reached is Eν max = MeV [1]. Early measurements by the Homestake experiment were only sensitive to ν e neutrinos via the process [8] v e + 37 Cl 37 Ar + e. (1.13) The measured flux was lower than expected. The SNO [9] and the Super-Kamiokande experiment [10] were able to demonstrate clear evidence for neutrino oscillation. In 2015, Takaaki Kajita (Super-Kamiokande Collaboration) and Arthur B. McDonald (SNO Collaboration) were awarded the Nobel Price for Physics for the discovery of neutrino oscillations, thus showing that neutrinos have mass. From measurements of solar neutrinos it is possible to determine θ 12 and m 12

12 1.1 neutrino physics 7 Figure 3: The predicted solar neutrino energy spectrum. The figure shows the energy spectrum of solar neutrinos predicted by the BP04 solar model [7]. as given in Table 1. The sign of m 12 can be determined due to propagation of the neutrinos in solar matter that introduces an additional phase (MSW effect [11][12][13]). Atmospheric Neutrinos are produced in the decays of pions π + µ + ν µ (1.14) µ + ν µ e + ν e. The pions are generated by protons from cosmic rays colliding with nuclei in the atmosphere. Atmospheric neutrinos were studied for example by Super-Kamiokande [14]. The Super-Kamiokande experiment consisted of a large water tank with tons of ultra-pure water. The neutrinos interacted with the nuclei of the water via chargedcurrent interactions ν l + N l + X (1.15) with l = e or µ. The created leptons l produce Cherenkov light which is measured by photomultiplier tubes. The Cherenkov light is emitted as ring whose opening angle depends on the β of the lepton. The flavor of the lepton is identified through the sharpness of the fringes of the rings. The heavier muons have a much more sharper ring due to less scattering. Additionally, the direction of the lepton and thus the

13 1.1 neutrino physics 8 Spectrum (ν/mev/fission) Cross Section 235 U 238 U 239 Pu 241 Pu 235 σ ( U) ν e (MeV) E ν cm 2 ) -42 σ (10 Figure 4: ν e energy spectra (four curves with negative slopes) for 235 U, 238 U, 239 Pu, and 241 Pu are shown. The curve with the positive slope represents the cross section of the inverse beta decay (IBD) process. The convoluted IBD spectrum, measured in experiments, is shown as the dotted line. [6]. direction of the incoming neutrino was extracted. The results showed that the muon neutrino flux from below the detector, i.e. the neutrinos were produced on the other side of the earth and traveled around 10 4 km, was lower than the flux from neutrinos produced above the detector. This effect was not observed for the electron neutrinos. The disappearance is explained with the change ν µ ν τ. From P(µ τ) studies are θ 23 and m 31 2 extracted as given in Table 1. Reactor Neutrinos are anti-electron neutrinos produced in nuclear power plant reactors. For each 1 GW of thermal power produced by fission, about ν e are emitted every second [6]. The energy spectra of different neutrino sources are given in Figure 4. The sensitivity to the mixing parameters depends on the baseline between source and detector. For short baselines of the order O(1 km), the measurements are sensitive to θ 13. Daya Bay experiment is such a short baseline experiment which measured θ 13 = 0[15]. The current values for θ 13 are given in Table 1. Long baseline (O(100 km)) experiments as KamLAND [16] are sensitive to the same neutrino oscillation parameters as solar neutrino experiments. As only the sign of m 12 is known, it is not known if the m 1 or m 3 is the lightest mass eigenstate this leads to the mass hierarchy problem (Figure 5).

14 1.1 neutrino physics 9 normal inverted Figure 5: The two possible mass hierarchies [1]. The major goal of JUNO is to discover the mass hierarchy of neutrinos. The survival probability for ν e can be expressed as [17] P ee = 3 i=i U ei e i m2 i 2E i U ei 2 =1 cos 4 θ 13 sin 2 2θ 12 sin 2 ( 21 ) sin 2 2θ 13 sin 2 ( 31 ) sin 2 θ 12 sin 2 2θ 13 sin 2 ( 21 ) cos(2 31 ) (1.16) ± sin2 θ 12 2 sin 2 2θ 13 sin(2 21 ) sin(2 31 ) The variables m i and E i are the mass and energy of the corresponding mass eigenstate, while θ ij is the corresponding neutrino mixing angle. U ei is the neutrino mixing-matrix element. It relates the electron neutrino to the mass eigenstate ϑ i. The oscillation phase ij scales with the distance mass squares between the mass eigenstates j and i. It corresponds to the following relationship: ij = m2 ij L 4E ν. (1.17) The distance L between neutrino emission and detection (also known as baseline) and the neutrino energy E ϑ also need to be considered. In equation 1.16, only the last term is sensitive to the mass hierarchy. The plus or minus sign depends on m 13 > 0 for the normal and < 0 for the inverted hierarchy. The precise measurements of neutrino flux with a resolution better than 3% is required to draw conclusions on the nature of the mass hierarchy.

15 1.2 scintillator scintillator The detection of particles and their accurate identification is very important for the JUNO experiment. Scintillators are widely used in this field. A scintillator is a substance whose molecules absorb kinetic energy and emit the excitation energy in the form of light (usually in the ultraviolet or visible range). This process is known as scintillation. The effect is mainly used in scintillation detectors to measure the energy and intensity of ionizing radiation. A scintillation detector is obtained when a scintillator is coupled to an electronic light sensor such as a photomultiplier tube (PMT), photodiode, or silicon photomultiplier. The deposited energy of each impact is calculated by measuring the amount of light. Usually, scintillators can be divided into two classes - organic scintillators and inorganic scintillators. While inorganic scintillators such as NaI or CsI have a solid crystal structure, organic scintillators occur as solid compounds as well as in liquid form. Most organic scintillator compounds include an aromatic core, which is also referred to as a benzene ring. In the following only organic scintillators will be discussed, such as used in JUNO [18] Principle of Organic Scintillators The principle of scintillation is based on the excitation of the organic molecules to higher energy levels, a so-called π-electron structure. A π-electron structure exists when the p orbitals of two atoms overlap [19]. Figure 6 shows a schematic of how these energy levels relate to each other. States with spin 0 (singlet) are marked with S i, states with spin 1 (triplet) with T i. The energy difference between the states S 0 and S 1 is about 3-4 ev in organic scintillators. For molecules, each of these levels split into more vibration levels marked as S ij and T ij. Here the energy difference is about 0.15 ev. If a particle transmits part of its kinetic energy to a molecule, the molecule is excited into a higher state. At room temperature, the molecules are in the ground state S 00, since the thermal energy of ev is not sufficient to reach the excited states. The scintillation light (fluorescence) is emitted during the transition from S 1 S 0. Molecules in levels higher than S 1 quickly fall to the S 1 - level through radiative transitions. The S 1 -level will empty following an exponential curve I = I 0 e t τ, where τ, the time constant, has the magnitude of nanoseconds. Besides the just described fluorescence, another effect takes place in the scintillator. This is known as phosphorescence. Instead of the S 1 -state returning to the S 0 -state, inter-system crossing occurs. The S 1 - state transitions into T 1 -state. For the T 1 -state less energy is required than for the S 1 -state, yet still more than the ground state. This triplet

16 1.2 scintillator 11 Figure 6: Schematic representation of the working of a scintillator [20] state has a considerably longer lifetime. Therefore the transition T 1 S 0 is delayed. Due to the energetically lower T 1 -states the emitted phosphorescence light has a longer wavelength than that fluorescence light [21] Light Propagation One important aspect of the JUNO detector is the light s ability to traverse through the solvent and reach the PMTs. All interferences have an impact on the energy and spatial reconstruction of the event. Two processes mainly occur as the produced photons propagate through the medium: absorption and scattering. These processes strongly depend on the emitted wavelength. Absorption describes processes in which photons interact with a component of the scintillator Subsequently their energy is released as thermal energy. This is most likely to occur when the scintillator has impurities. In this case all information from the photon is lost. Scattering describes interactions in which a photon interacts with another particle changing its direction and/or its wavelength as a result. This makes reconstruction difficult. There are two main contributions to scattering: the Rayleigh processes and Mie scattering[22].

17 1.2 scintillator 12 The Rayleigh processes describes elastic scattering of photons on particles whose diameter is small compared to their wavelength λ. The oscillating electric field of a light wave acts on the charges within a particle, causing them to oscillate according to the electric field vector of the incident photon. The particle becomes a small radiating dipole. Since the scintillator molecules are smaller than the wavelengths for the detector, Rayleigh scattering is the dominant process. It can not be reduced by purification processes or by the addition of substances. Rayleigh scattering sets a natural limit for the attenuation length L and thus for the transparency of a scintillator. An ideal scintillator without impurities and absorption effects would have a maximum attenuation length. [23] Mie scattering occurs when the diameter of the particles d is comparable to the wavelength λ of the scattered photon. Due to the size ratios, Mie scattering is primarily observed in the interaction with dust and dirt particles, which can be contained in the scintillator material. Therefore purification of the scintillator material is essential. If scintillator is as pure as expected, Mie scattering will play a negligible role in the absorption length of the JUNO detector. The intensity of the remaining light is related to the mean free path length of the photon before the interaction, the absorption length λ abs, and the scattering length λ sct. The light intensity I for a covered distance x is exponentially linked to these values: I(x) = I 0 e x λ abs e x λ sct = I 0 e x λ att (1.18) I 0 is the initial intensity. The combination of scattering and absorption is the attenuation effect, and thus I att is defined as the attenuation length: 1 λ att = 1 λ abs + 1 λ sct (1.19) For a large-volume particle detector, a long attenuation length is advantageous, since the light yield of a scintillator detector is very important. A better light yield leads to a better resolution.

18 J U N O 2 The Jiangmen Underground Neutrino Observatory (Juno) is a multipurpose neutrino experiment located in Kaiping, China. It is designed to determine neutrino mass hierarchy and precisely measure oscillation parameters (see Figure 1.1.2) by detecting reactor neutrinos from the Yangjiang and Taishan Nuclear Power Plants. It is also aimed at observing supernova neutrinos, studying the atmospheric and solar neutrinos and the geo-neutrinos, and performing exotic searches. This chapter will take a closer look at the setup of the Juno detector, looking at the different components of the Central detector as well as the used LS, its requirements and the purification. The following information is taken from the JUNO Conteptual Desing Roport [18]. 2.1 location The detector will be located in Kaiping, Jiangmen, in Southern China, 700 m underground with a distance of 53 km from both the Yangjiang and Taishan nuclear power plants. The Location is shown in Figure 7. Figure 7: Location of the JUNO detector and the Nuclear Power Plants Construction started in 2014 and is scheduled to be finished in The area will include a tunnel for trains, an underground experiment hall, a water pool, the central detector, a muon tracking detector, and some ancillary facilities. 13

19 2.2 central detector central detector Figure 8: Schematic of Juno detector [24] The central detector will be spherical and filled with 20 kt of LAB based liquid scintillator. It is surrounded by photomultiplier tubes (PMT). This inner detector will be placed in a water pool that protects the detector from natural radioactivity in the surrounding rocks and serves as a muon veto. There will also be a muon tracker above the central detector to better track cosmic muons. Figure 8 shows a schematic of the Juno detector, with the most important components Muon Veto The muon veto will be made up of two parts, the Top Tracker and the Water Cherenkov. The Top Tracker will be on top of the central JUNO detector as can be seen in Figure 8. The Top Tracker will be made up of three layers of plastic scintillator. That partially cover the central JUNO detector allowing to reject about 50% of all muons that reach the central JUNO detector. The Water Cherenkov will surround the detector. Muons entering the 30 kt of ultra pure water will create Cherenkov light. There are 2000 large PMTs installed to detect the Cherenkov light. The place-

20 2.2 central detector 15 ment of the PMTs as shown in Figure 8 maximizes the detection efficiency for cosmic muons arriving from above. The detection of muons is important in order to be able to exclude induced background caused be muons such as neutrons and radioactive isotope from spallation of 12 C. By placing the detector 700 meters underground the rock also shields the detector from cosmic radiation, reducing the number of muons that reach the central detector[25] Acrylic Sphere The acrylic sphere will have an inner radius of r=17,75 m and will be approximately 12 cm thick and made up of over 200 sheets of Acrylic. The used material is polymethylmethacrylate, also known as acrylic glass or plexus. It will hold 23,000 m 3 of the liquid scintillator. [26] At the top of the acrylic sphere, there will be a chimney with a diameter of 0.5 m, through which the central detector will be filled and where the calibration system will be integrated. The acrylic hollow sphere is fixed on a stainless steel structure (see Section 4.1.2), which also will carry about 16,000 inward facing photomultipliers (see 2.2.5) to detect the photons from neutrino events. [18]. For JUNO, the transmittance of the acrylic sphere is also of high importance. If the transmittance is too small, not enough photons would reach the PMTs and the energy resolution of 3% could not be achieved. Tests have shown that an acrylic layer of 4-5 cm has a transmittance of about T=96 %. It is shown that the refractive index has a similar wavelength dependence and a similar refraction index as the liquid scintillator LAB. This automatically reduces the probability of intensity losses at the optical transi- Acrylic sphere [26] Figure 9: 3D reconstruction of the tion, which would be given by two different values for the refractive indices. Acrylic has a high stability, is easy to process, and has a good resistance to many organic substances including the scintillator material as well as not being subject to aging. This makes acrylic a good choice. [18]

21 2.2 central detector Stainless Steel Structure The acrylic sphere will be held up by a stainless steel lattice shell. The steel lattice construction will have a diameter of 41.1 m and a total weight of 600 t. The acrylic sphere will be connected to the stainless steel lattice with over 500 stainless steel rods and submerged in the water of the Cherenkov detector. This structure will also carry the PMTs (see 2.2.5). Figure 10: 3D reconstruction of the Stainless steel lattice shell and acrylic sphere The variant of a steel lattice offers several advantages over an entire steel sphere. It is much lighter and consists of smaller individual components, making construction easier. The resistance against seismic activity is also increased. The photomultipliers used to measure the scintillation will also be attached to the steel frame. The inward pointing PMTs will be protected from outside light by a shield installed at the back of the PMTs protecting the entire inner detector. This is to ensure that the PMTs only register signals from inside the detector [18] Buffer The acrylic sphere will be surrounded by an approximately 1 m thick water buffer layer. The water molecules absorb part of the emitted energy and thus reduce their excitation capability. At a certain value, the radiation energy is no longer sufficient to generate light in the scintillator. This way the background caused by impurities can be

22 2.3 liquid scintillator 17 reduced. Possible sources for these types of impurities are located in the installed steel of the steel structure and the glass of PMTs Photomultiplier Tubes In JUNO two different photomultiplier tubes (PMTs) will be used. The main detection occurs using the large PMTs. These PMTs have a diameter of 508 mm (20 ) and cover about 75% of the detector surface. Large PMTs are beneficial because less are needed to cover most of the detector. Small PMTs will partially fill the gaps between the large PMTs. These small PMTs have a diameter of 76.2 mm (3 ) and cover another 3% of the Detector surface. These small PMTs are faster and have a better resolution of the large PMTs and of the small PMTs will be installed. The use of the small PMTs alongside the large PMTs helps to reduce systematic effects represented by non-stochastic terms in the energy resolution dependence and to extending the dynamical range i.e. for muon detection. High detection efficiency is required for the detection of scintillator photons because of the detector size and the light attenuation that occurs in the scintillator material [25]. 2.3 liquid scintillator The Juno detector will be filled with 20 kt liquid scintillator (LS). This liquid scintillator is the target medium for the detection of neutrinos and antineutrinos. To achieve the desired energy resolution of at least σ E E = 3% in E JUNO [27], a high detection efficiency is required. For this the scintillator needs a very high transparency. This allows light that is produced near the center of the detector to propagate to the PMTs. This is difficult because of the detector size, requiring unusually high LS purity. The main component of the LS is the liquid scintillator linear alkyl benzene (LAB) wich also is the solvent. To this 1.4-Bis(2-methylstyryl)- benzol (bis-msb) and 2,5-Diphenyloxazol (PPO) are added. The LAB in JUNO gets excited by ionizing particles and passes this excitation on to the two-component system of 2,5-Diphenyloxazol (PPO) and 1.4-Bis(2-methylstyryl)benzol (bis-msb). The subsequent Stokes shifts increase the wavelength of the emitted photons to about 430 nm. This wavelength shift is necessary to avoid self-absorption. At the new wavelength of 430 nm, the transparency of the liquid is largely governed by the Rayleigh scattering of photons on the solvent molecules. Other molecules in the liquid can reduce the transparency.

23 2.3 liquid scintillator 18 The LS serves as target medium for the detection of neutrinos and antineutrinos. The primary reactions for reactor electron antineutrinos ( ν e ) is the inverse beta decay on free protons, ν e + p + n + e +, resulting in a prompt positron and a delayed signal from the neutron capture on hydrogen (τ 200µs) Composition The liquid scintillator used in JUNO is composed of three different chemicals. The base of the liquid scintillator is LAB to which the solute PPO (c PPO 3g/l), and the wavelength shifter bis-msb (c MSB 15mg/l) are added. linear alkylbenzene (lab) In JUNO linear alkylbenzene also know as LAB is used as the liquid scintillator. The formula is C 6 H 5 C n H 2n+1. LAB encompasses all substances which have an aromatic ring as well as a hydrocarbon chain, which can vary in length. Figure 11 shows the structural formula of LAB. This scintillator is very suitable for JUNO due to its excellent Figure 11: Structural of LAB formula transparency and long attenuation length of 20 m at a wavelength of 430nm. The attenuation length is about twice as large as that of pseudocumene that is for example used in Borexino. The absorption maximum of LAB is 260 nm while the emission maximum is at 283 nm. 2,5-diphenyloxazole (ppo) 2.5-diphenyloxazole also known as PPO is one of two wavelength shifters used in the JUNO experiment. It has the chemical formula C 15 H 11 NO and is a solid under standard conditions. The absorption maximum of PPO is 303 nm while the emission maximum is at 365 nm. Figure 12 shows the structural formula of PPO with its 3 benzene rings. Figure 12: Structural of PPO formula In JUNO the concentration of PPO will be in the range of c PPO 3g/l, the exact final concentration is being determined with sensitivity and transparency studies [18].

24 2.3 liquid scintillator bis(2-methylstyryl)benzol (bis-msb) Figure 13: Structural formula of bis-msb The other wavelength shifter used in JUNO is 1.4-Bis(2-methylstyryl)benzol also known as bis-msb. It has the chemical formula C 24 H 22 and is a crystalline solid at room temperature. In JUNO the concentration of bis-msb will be in the range of c MSB 15mg/l [18]. Figure 15 shows the structural formula of bis- MSB. Bis-MSB is often used in combination with PPO because the emission maximum of PPO (365 nm) and the absorption maximum of bis-msb (345 nm) are very close to each other and therefore an optimum wavelength shift can be achieved. The emission maximum of bis-msb (420 nm) is far away from the absorption maxima of both PPO and LAB thus letting the shifted light propagate through the detector[18] Purity Requirements For the mass hierarchy measurement the energy resolution of JUNO is required to be least at 3% at 1 MeV. This corresponds to the detection of at least 1,100 photoelectrons for a deposition energy of 1 MeV. Both the initial light yield and the transparency of the liquid have to be optimized to achieve this result. [m] a 90 Light Yield LY(L,L a ) s L=30m 1500 Absorption Length L L=25m L=20m L=10m L=15m Scattering Length L [m] s 1100 Figure 14: Photo-electron yield and functions of λ Abs and λ scat [28] Figure 14 shows the dependence of the photo-electron yield and the attenuation length on the absorption and scattering length. The black lines show the position of specific attenuation lengths. It can be seen that the yield of photo-electrons is substantially more dependent on the absorption length than on the scattering length of the liquid

25 2.3 liquid scintillator 20 scintillator. The effect of the scattering length on the photo-electron yield is increased only from an attenuation of 20 m.[28] It is therefore more important to have a high absorption length. Figure 15: Emission spectrum of purified and raw bis-msb Especially the absorption of bis-msb at wavelengths between 420 nm and 460 nm increases, if impurities are present in the LS. This can be seen in Figure 15. Organic impurities have to be removed as far as possible since they reduce the transparency of the LS. Organic material has absorption bands in the optically relevant range. This leads to an intensity reduction. Besides this, any larger particles will also lead to Mie scattering (See Chapter 1.2.2). The radioactive impurities have to be reduced as far as possible in JUNO. Radioactive isotopes in the LS lead to higher background rates. For the measurement of reactor neutrinos the concentration has to be under g/g and for the detection of solar neutrinos under g/g for Uranium and Thorium[18] Filling The JUNO detector will be filled with 20 kt of liquid scintillator. Due to the large amount and the danger of recontamination during transport, the raw materials will be purified and mixed on site. The mixed LS is then filled into the central detector [18]. Figure 16 shows the currently planned procedure for preparation. The whole procedure has to take place in a nitrogen atmosphere because dissolved oxygen can act as a quenching agent and lead to reduced light output.

26 2.3 liquid scintillator 21 Figure 16: Flow chart for LS mixing and filling [24] After delivery the LAB, PPO, and bis-msb are filled into storage tanks. The PPO and bis-msb are then filtered and mixed to form the master solution. The LAB is purified in two steps. First, the Aluminum Column (Al 2 O 3 ) is used to remove organic and radioactive impurities. Organic impurities reduce the transparency and must, therefore, be removed. After this step, the quality of the LAB is checked for quality control (QC) [18]. Quality control is important because if on bad batch is filled into the central detector onto an otherwise good LS, the whole LS volume would be spoiled. Re-purification, after the LS was filled into the central detector, would be very time-consuming and difficult. It is, therefore, important to ensure good quality in every batch. At this point, only the optical properties of the LAB can be examined. To ascertain the quality of the LAB, the transparency is examined. In order to guarantee the necessary purity, the attenuation has to be measured with an accuracy of λ(20 ± 1) m. The effect of the Aluminum Column on the optical properties of LAB has been studied at IHEP. The results can be seen in Figure 17. The LAB purified by the Aluminum Column (green) has a far better transmittance in the interesting wavelength than any other LAB. It should also be noted that the high contamination of the used LAB (blue) does not affect the function of the Aluminum Column. If the quality is good, the LAB is then passed onto the Distillation Tower for distillation, if not the LAB is fed back into the Aluminium Column. In the Distillation Tower the LAB undergoes functional distillation to remove radioactive metal ions from the LAB. This is possible due to the vast difference in boiling points. LAB is far more volatile than metal. After distillation the quality of the LAB is again examined and if necessary the distillation process is repeated. Other-

27 2.4 neutrino detection 22 Figure 17: Absorption spectra of different LAB including LAB purified in Nanjing lab (green) [18] wise, the LAB is mixed with the master solution and filled into the central detector [24]. 2.4 neutrino detection In JUNO the neutrinos are detected via the weak interaction. Neutrinos can interact weakly over two different channels. ν e ν e ν e e Z 0 W e e e ν e Figure 18: Scattering process Neutrino scattering occurs when a neutrino interacts with an electron via a neutral current (Z 0 ) or a charged current (W ± ), hereby a part of the kinetic energy of the neutrino is transferred to the electron. This is shown in figure 18. The inverse beta decay (shown in Figure 19) is the other weak interaction between particles and neutrinos. An antineutrino interacts with a proton via a virtual W + boson. A neutron and an electron are created as described in the equation: ν e + p e + + n.

28 2.4 neutrino detection 23 ν e e + W + p + n Figure 19: Inverse beta decay Due to the prompt positron signal and the delayed neutron signal from the neutron capture on hydrogen, this interaction process can easily be separated from the background. Therefore, the inverse beta decay is used to detect the neutrino signals in JUNO. The cross sections for weak interactions are extremely low. The cross section for the inverse beta decay is σ IBD E e p e 1MeV cm 2 (2.1) The parameter E e is the energy and p e is the momentum of the emitted positron e + [1]. The kinetic energy of the proton and neutron after the collision can be neglected because of the relatively small energy of the neutrino (in the range of a few MeV) and the large masses of the proton and the neutron [29]. Figure 20: Spectra for the antineutrino signal and the five main backgrounds in JUNO [27]

29 2.4 neutrino detection 24 The energy resolution of at least 3% is not only influenced by the detection efficiency and scintillator purity but also by the background. The expected total background to signal ratio is 6.3%. Figure 20 shows the spectra for the reactor antineutrino signal and the five main backgrounds, geo-neutrinos, 13 C(α-n) 16 O, 9 Li 8 He, fast neutron, and accidental background [27]. It is important to reduce the background as far as possible.

30 O N - L I N E AT T E N U AT I O N L E N G T H M O N I T O R 3 Before filling the CD, the optical transparency of the LS has to be tested to assure that all purified batches fulfill the purity requirements. For this a measurement system monitoring the attenuation length has been developed in the course of this thesis. The measurement of the attenuation length allows conclusions about the purity of the liquid scintillator because the presence of organic impurities in the liquid scintillator would lead to absorption. This means that more photons are lost over the light path and a shorter attenuation length will be measured than for a liquid scintillator of higher purity. In the setup attenuation length is measured by determining the intensity loss of a laser over a tube filled with LS. The initial light beam is split and travels along two light paths, provided by two tubes of different lengths to allow a relative measurement. This way, many systematic uncertainties such as the initial intensity I 0, reflections, and other surface effects cancel out. The attenuation length L Att of a medium over the distance x is given by Ix I 0 = e x L Att. (3.1) The attenuation length can also be calculated comparing the drop of the same initial intensity over two different light paths x 1 and x 2 using L Att = x 1 x 2 ln(i x1 ) ln(i x2 ). (3.2) For this the final light intensies I x1 and I x2 have to be determined. This allows more precise measurements due to the fact that, except for the length of the medium, both light paths are identical. Accordingly any surface effects caused by the light entering or exiting the chamber can be ignored since they do not have an effect on the relative light intensity. Another positive effect of using the relative light intensity is that the time stability of the light output of the laser does not affect the measurement. 3.1 design When designing the attenuation length monitor there are three main components: the light source, the light detector, and the chamber for the medium (see Figure 21). In addition, the filling system and the mounting of the setup have to be considered. 25

31 3.1 design 26 Figure 21: Sketch of the setup of the purity monitor (not to scale) For the measurement, the light has to travel on a straight light path through the medium along the axes of the tubes. Diversions could occur either at the transitions between different materials or if temperature layering is present in the medium. The refractive index is influenced by the temperature of the medium. It decreases with higher temperatures. Both these effects can be minimized by perpendicular incline of the light beam. For the transition between different materials, this is achieved by aligning the components. In order to hit possible temperature layers present in the medium at a right angle, the tubes are mounted vertically so that the gradient in the reflective index is parallel to the light beam. This way no refraction is expected in accordance with the Snell s law. A horizontal setup was used to measure the attenuation length at the TU Munich, here the temperature effect was observed [30] Setup Figure 21 (not to scale) shows a sketch of the setup of the purity monitor. A laser is used to measure the attenuation length. A beam splitter and a mirror allow the light beam to travel through both tubes parallel. In this setup the laser is placed above the longer tube. This way the distance from the laser to the CCD Camera is the same for both arrangements. This ensures that the length of the light paths and thus

32 3.1 design 27 any potential widening of the beam diameters is equal for both light beams in the empty setup. The effects of different lengths the light travels through the medium can not be eliminated. The attenuation length of air is long enough, that it does not have an effect on the outcome of the measurement. The CCD cameras are placed below both tubes at an equal distance. The setup is fastened to the wall using optical rails. This allows easy adjustment and exchange of the components, if necessary, without having to disassemble the entire setup Light source A well-focused light source is preferred because it eliminates the need for a complex optical system to obtain a parallel light beam. This avoids many possible sources of error and makes alignment easier. If the light source has a fixed wavelength of 405 nm, no monochromator is necessary thus simplifying the setup. Though using a spectral light source with a monochromator would allow the measurement of the attenuation length at different wavelengths, this is not necessary for the monitoring of the purity via the attenuation length. Measuring at the wavelength of 405 nm gives enough information without the need for an expensive piece of equipment and, furthermore, is easier to set up. A wavelength of 405nm was chosen because at this wavelength organic impurities have a stronger effect on the attenuation length than at 430nm (see Figure 22) Figure 22: Absorption spectra of raw LAB and purified LAB Diode lasers have a fixed wavelength as well as a focused beam. Therefore a Violet Laser Diode Module with a wavelength of 405 nm is used. The intensity measured at the same time after passing through two different lengths of the medium is compared. This way

33 3.1 design 28 only the difference in intensity I is relevant, not the absolute intensity Light detection For the light detection it is important that small relative differences in the intensity can be detected. One option would be use a photodiode or a PMT. However, it would be necessary to always to maintain perfect alignment of the light beam in order to exactly hit the photodiode since it does not offer any spatial resolution. Spatial resolution is useful in this setup because it allows examination of the beam profile for disruptive effect, that could influence the intensity measurements. If the light beam hits the wall or blurs out too much, less intensity than expected will be measured. Another possibility is that the light beam is bent by refraction in the medium. In this case, the profile of the light beam might not or only partially be detected making measurements useless. Using a charge-coupled device (CCD) camera solves these problems since its has spatial resolution. However, the CCD chip has to be significantly larger than the diameter of the light beam (see 23). Figure 23: Laser spot on the CCD chip As already mentioned, the intensity is measured after the light beam has passed through two different lengths of the medium. This means that the two different light beams at different positions have to be detected. The ideal solution would be to lead both beams onto one camera but this would require a much more complicated setup using a larger CCD chip or a shutter that allows alternating measurements of each beam. For simplicity, this should be avoided. A more practical solution is therefore to use two separate cameras of the same type. This way all possible effects of the cameras should be identical so that the relative difference in intensity can be easily be measured.

34 3.1 design Tubes LS are based on organic solvents that attack and dissolve many surface materials Therefore the chambers containing the LS samples have to be made of nonreactive material. Two materials widely used for this purpose are Teflon and stainless steel. Large Teflon tubes are difficult to obtain. Stainless steel tubes, however, are easy to get and easy to work with. Therefore stainless steel is used in the setup. Another aspect that makes stainless steel preferable is that light reflection on the inside of the tubes should be avoided. Teflon tubes are white allowing more reflection and scattering then stainless steel tubes which can be blackened. The length as well as the diameter of the tubes are to be considered. The longer the tube the more precise measurements are possible. However, the length of the tube is constrained by the sample size and the height of the room. Since the other components have to be placed above and below the chamber, a realistic maximum length is 1.5 m. The other chamber has to be considerably shorter in order to obtain a large enough difference in intensity I. The second chamber has a length of 0.5 m. The diameter of the tubes should not be chosen too small. That could lead to the formation of standing waves in the medium. Standing waves would make measurements impossible due to the refraction of the light beam. Here a tube diameter of 7.5 cm is chosen. The tubes have to be closed at both ends while allowing the light to pass through and enabling the filling and draining. It is also preferable if the chamber can be opened. This is best achieved by attaching a flange to each end of the tube with the necessary opening and attachments Filling The filling system of the on-line attenuation length monitor has to be a closed system. This avoids contamination with ambient dust and will be mandatory for inserting the system into the filling line of JUNO. Moreover, an air tight system has the possibility of performing the measurements under a nitrogen atmosphere. This is of interest for the measurements with not only LAB but with the added wavelength shifters since oxygen binding reduces the transparency of the LS. Because chemical compatibility with the LS, the hoses of the filling system also have to be either Teflon or stainless steel. Since stainless steel is not flexible enough, the hoses are made of Teflon and the connector parts are made of stainless steel. It is necessary that the tubes can be completely filled and drained. This ensures that there is no air, gap, or bubbles in the light path during measurement. Otherwise, the additional optical transition could

35 3.2 components 30 cause refractive or reflective effects that would strongly impact the measurement. If the tubes are not completely drained, the residual of the first measurement contaminates the next measurement. This has to be considered for continuous operation during filling at JUNO. In this setup, to ensure complete filling and draining, the hoses have to be attached at the highest and lowest possible part. The hoses are therefore attached to the flanges closing the tubes Mounting The attenuation length monitor has to be mounted as stable as possible. It is important to minimize vibrations in the liquid that may influence the measurement results, which was observed at the TU Munich [30]. There are several sources for vibrations or other interferences. Firstly, there are many people working in the lab. This means vibrations of the floor are a common occurrence. Building vibrations also have to be considered. These are strongest in the floor and in non-load bearing walls. The weight of the setup constitutes another difficulty. The volume of 12 liters plus the stainless steel tubes give a total weight of about 50 kg. It is, therefore, necessary to consider the stability of any mounting option. The non-load bearing walls are plaster walls and would not hold the setup. Considering these factors, the attenuation length monitor is for the present time mounted on a load-bearing column. This keeps vibrations from activities in the lab as well as the effect of the building vibrations down to a minimum while definitely being able to hold the weight of the setup. A similar solution has to be found for the later setup at the JUNO site. The components of the setup have to be aligned along the light path and it is advantageous if the distance between the components can be adjusted. The setup is therefore fastened to the wall using commercial optical rails. This allows the use of standard optical components and holders, an easy adjustment, and the exchange of the components, if necessary, without having to disassemble the entire setup. 3.2 components This part will take a closer look at the components that are used in the on-line attenuation length monitor. The components are chosen based on the mentioned considerations.

36 3.2 components CCD Camera For light detection, two identical CCD cameras are used. It is crucial to use to identical cameras in order to measure the relative intensity. The Atik383L+ Camera (shown in Figure 24) has a Kodak KAF 8300 chip measuring 17.6 mm mm with 8.3 million pixels, each 5.4 µm. The Atik383L+ Camera has an active cooling system, in order to thermally stabilized the CCD and achieve the best result. Figure 24: Atik383L Data The data is recorded with the software Artemis Capture. Artemis Capture is a standalone program for Windows designed to control the Atik camera, to capture and preview images. The images captured by the camera are stored as TIFF-files (16-bit amplitude resolution). These TFF-files can not be manipulated by the user before saving and are considered the raw data. During operation of the camera, the pictures can be previewed using the Artemis Capture software. The brightness and contrast of this preview image can be adjusted by the user in order to better adjust the setup, i.e. to identify false images of reflections at the vessel surfaces. tif format The Tagged Image File Format (TIFF) is used to store image data in a raw format. It is very flexible and can store different types of images. Advantageous is the option to store images in a lossless format, i.e. every pixel information is available Characterization of the CCD Camera In order to characterize the camera, measurements were taken using an optical enclosure covered with black fleece. This ensures that no outside light disturbs the measurements. Hereby the camera units, the dark noise level, the linearity and distribution of hot pixels was studied. In order to adjust the light intensity for different measurements gray filters were used. The adjustment of the light intensity is necessary to avoid saturation of the CCD-Chip. Gray filters are neutral-density filters. They reduce the intensity of all wavelengths equally. This avoids overexposure of the CCD-Chip due to the high intensity of the laser. 31

37 3.2 components 32 camera amplitude units Due to its 16-bit amplitude resolution, the camera gives the intensity in arbitrary units with 0 meaning no light was detected and being the maximum. In order to compare the measured intensity, they are given in percent of the possible maximal intensity (% pmi). bit noise Since the CCD Camera outputs whole values between 0 and minimal uncertainty is at least ±1 equivalent to % pmi and about 0.2% of the measured intensity at the dark noise level. Besides the statistical uncertainties systematic uncertainties have to be considered. dark noise In order to determine the dark noise of the cameras, a picture at exposure lengths of 0.1 seconds, 0.3 seconds, 0.5 seconds, 0.6 seconds, 0.8 seconds, and 0.9 where taken. The histograms of the intensity distribution with the shortest exposure lengths (0.1 seconds) can be seen in Figure 25 and with the longest exposure lengths (0.9 seconds) can be seen in Figure 26. Entries Relative intensity [% pmi] Figure 25: Intensity distribution of dark noise over all pixels at exposure time of 0.1 sec For all these measurements, the intensity spectrum shows the expected Gaussian shape at low intensities. The dark noise level is 0.653% pmi with a σ = for all exposure lengths. The intensity distribution for all exposure lengths was fitted with a Gaussian curve. These are compared in Figure 27. It can be observed

38 3.2 components Entries Relative intensity [% pmi] Figure 26: Intensity distribution of dark noise over all pixels at exposure time of 0.9 sec Figure 27: Gaussian functions for the different exposure lengths that the Gaussian distributions agree quite well with each other. The exposure time has no effect on the mean or sigma (σ) of the Gaussian function the values are listed in Table 2. This is due to the fact that the mean dark noise of the pixels is steady over time and the spread in mean values largely exceeds the statistical uncertainties of individual pixels. The mean dark noise is not influenced by the exposure length. To determine the stability of the dark noise dark pictures were taken overnight (10 h), at exposure lengths of seconds without any light source. Pictures were taken every 10 seconds. The fluctuation of the dark noise average over all pixels is shown in Figure 28.

39 3.2 components 34 time [sec] mean [% pmi] σ Table 2: Mean and sigma values for different exposure lengths Figure 28: Fluctuation in dark noise average as percent of the possible maximum intensity The dark noise of the camera is mostly stable over long time periods. At the beginning of the measurement, the dark noise is at its lowest at % pmi. There is a clear rise in the intensity for the first 25 minutes where the dark noise rises up to % pmi. This is probably due to the chip heating up. After this, the dark noise is almost steady, yet rises by 0.12 % of the dark noise every hour. Dips can also be observed with the intensity of about -0.2 % of the dark noise. Hereby the intensity drops for 0.2 minutes and then rises for 3 minutes until it has returned to the original level. The fluctuations in the dark noise should not be problematic since all intensity measurements will be dark noise adjusted. The pictures are dark noise adjusted by measuring the dark noise level of each picture and subtracting it from the measured intensity. This leaves only the actual intensity allowing precise measurements of the attenuation length. In order to subtract the dark noise from each picture, the mean of all pixels with less than 5% pmi was calculated. This gives the maximum of the dark noise peak (Figure 29).

40 3.2 components 35 Entries Relative intensity [% pmi] Figure 29: Intensity distribution of a picture with laser spot hot cells CCD Cameras tend to have defective pixels. These pixels exhibit an unusual high dark noise and are referred to as hot pixels. In order to locate these hot cells, the intensity spectrum is examined. As shown in Figure 30, the main part of the dark noise has the expected Gaussian shape, with a majority of the pixels featuring a mean dark noise level of about 0.649% pmi. There are also pixels clearly brighter than expected. These pixels form a tail to the right of the Gaussian distribution. Pixels outside the six sigma threshold of the Gaussian function are classified as hot pixels and are removed from further analysis Relative intensity [%] Number of Pixels 106 Figure 30: Intensity distribution of dark picture with Gaussian fit (red) and six sigma cut (dashed red line) The spatial distribution of the hot pixels on the CCD chip is shown in Figure 31. The hot pixels are randomly strewn across the entire area of the CCD chip. There are no clusters. The hot pixels can, therefore, be excluded in the future analysis without needing to compensate for large uncovered areas.

41 3.2 components 36 Figure 31: Areal distribution of hotpixels definition of mean laser intensity Henceforth the mean intensity of the Laser will be defined as the mean intensity in a circle with a 400 pixel radius around the weighted center of all pixels with a intensity over 80% of the maximum of the measured intensity. This definition of the maximum is used because the whole laser spot is in the circle for all pictures and the number of considered Pixels is the same for all pictures, allowing a good comparison. The applied circular selection around the laser spot can be seen in Figure 32. The uncertainty on the mean is calculate using the bit noise for each pixel the uncertainty on the mean value is therefore ±0.004 %pmi. linearity To determine the linearity of the responses of both CCD-Camera, the laser was used as a light source. Pictures were taken at exposure lengths of 0.1 seconds to 1.0 seconds in steps of 0.1 seconds. At each exposure lengths, 10 pictures were taken to gain statistics. The mean intensity of each of the measurements was then determined and plotted against the exposure time. As can be seen in Figure 33, the CCD camera has very good linearity. It can be seen that the by extrapolation to 0 the expected intensity is still higher than the measured dark noise. This is due to scattered light. However, while both cameras have good linearity the gradient for both cameras is different. Therefore, in order to get a good relative

42 3.2 components y x Figure 32: Laser spot with circle for calculation of mean intensity 0.00 Relative intensity [% pmi] Integration time [s] Figure 33: Linearity of camera 1. The results for the linear fit f (t) = a t + b are a = (6.01 ± 0.07 )(% pmi)/s and b = (1.05 ± 0.04) % pmi. measurement, it is necessary to measure the difference between the two cameras. comparison of cameras Since the setup uses two cameras (see section 3.1.1), both cameras were compared.

43 3.2 components 38 Relative intensity [% pmi] Integration time [s] Figure 34: Linearity of camera 2. The results for the linear fit f (t) = a t + b are a = (4.61 ± 0.01 )(% pmi)/s and b = (1.10 ± 0.01) % pmi. First, the dark noise of the two cameras is compared by simultaneously taking dark pictures with both cameras for 8 hours. This was done for exposure times of 0.1 sec. This way the dark noise level as well as the dark noise stability can be compared. Figure 35: Dark noise of both cameras The different dark noise levels can be seen in Figure 35. It is clear that the dark noise of camera 2 is slightly higher than that of camera 1. The dark noise of camera 1 is steady at % pmi while the dark noise of camera 2 is steady at % pmi. In order to rule out the possibility that this is caused by the power supply or the socket, short time measurements were taken where the first the power supply and then the socket were interchanged. This did not have any effect on the dark noise levels, it can, therefore, be concluded that this difference is due to the manufacturing of the camera and constant under outside influences. The behavior of dark noise is mostly identical as can be seen in Figure 35. The difference of the dark noise to the corresponding mean of

44 3.2 components 39 the dark noise is identical for both cameras except for the dips in the dark noise. These dips are not correlated therefore it is assumed that they are the effect of the camera and not caused by the environment. Since the behavior of the dark noise is identical for both cameras, the intensity measured by the cameras can be compared after subtraction of the dark noise. Afterwards, the stability of both cameras was compared while the laser (see 3.2.3) was used. In order to measure the laser beam simultaneously with both cameras, the beam splitter was placed in the beam and both cameras were placed an equal distance from the beam splitter. Figure 36: Laser light detection of both cameras It can be observed in Figure 36 that again the intensity registered by camera 2 is higher than that registered by camera 1. If the intensity fluctuations of both cameras are compared as shown in Figure 36, it can be seen that both cameras measure the same intensity fluctuations. This means that the measurements of the two cameras can be compared allowing the measurement of the attenuation length Tubes As shown at the beginning of this chapter, two tubes of different lengths are used for a relative measurement. As explained in 3.1, both vessels are made of stainless steel and are identical except for their length. The length of tube 1 is (1.500± 0.001) m, the length of tube 2 is (0.5 ± 0.001) m with an inner diameter of (7.5 ± 0.15) cm and a wall thickness of (0.5 ± 0.1) cm. Both tubes were blackened by the company Inox-color to reduce reflection of scattered light inside the tubes. By adding a chromiumoxide layer to the surface, the processes enhances the inert properties of the stainless steel. The "Inox-spectral-Verfahren" adds the coating by an electrochemical process. At both ends of both tubes, flanges were attached in order to close the tubes. In the middle of each flange, a window of borosilicate glass is situated to allow the passage of the laser light. The glass has a diameter of (30 ± 0.1) mm, the hole a diameter of (20 ± 0.1) mm. This leaves an overlap of about 5 mm to

45 3.2 components 40 Figure 37: The short tube with flange and hose glue the window on. The windows were fixated using the two component epoxy EP42HT- 2LO. This epoxy is low outgassing and the compatibility with liquid scintillator has already demonstrated for parts of the Double Chooz experiment This makes it ideal for the use in this JUNO monitor. The windows in the flanges protrude out of the stainless steel by 3 mm. This is done to ensure a complete draining and filling of the light path. If the light path is not completely filled due to air bubbles being caught in the tube, reflection, deflection or interference may occur. This would strongly distort the measurement Laser The laser used for the purity monitor is a violet, compact size fix collimated laser diode, the CW from Roithner-Lasertechnik, with a wavelength of (405 ± 5)nm This laser has an optical output power of <1 mw and a beam diameter at aperture of 2 x 4 mm. The beam divergence is only 1.0 mrad Characterisation of the Laser Diode Using the CCD camera, the intensity profile of the laser beam was examined. As shown in Figure 38, the circular beam profile is cut in one spatial direction by the diodes aperture and moreover features a complex radial intensity pattern.

46 3.2 components 41 y x Figure 38: Profile of the laser beam at a distance of 50 cm 8.00 The picture was taken at a distance of 50 cm. The visible laser point is 2.03 x 4.05 mm. This is in line with the specified beam divergence of 1.0 mrad. Over the whole distance of 2 m, an expansion of 2 mm can be expected. This leads to a beam of 4 x 6 mm, which can be well contained on the area of the camera s CCD-chip of 17.6 x mm. stability Relative intensity [%] Time[min] Figure 39: Time stability of the laser diode The time stability of the laser was measured using one of the CCD cameras. After the camera had stabilized, the laser was switched on and the measurement series of 8 hours was started. Figure 39 shows the laser intensity recorded: In the first ten minutes after activation, the laser quickly loses about 2% of the maximum of intensity. After this period, the emitted intensity stabilizes and only fluctuates by 0.06% of the maximum for about 90 minutes. Later on, the fluctuations increase. If the fluctuations of the laser are to be minimized, the laser should only be used for measurements for 90 minutes at a time after a 10 minute heating period. The laser is suitably stable to be used for the measurement of the attenuation length with the CCD cameras. The fluctuations in the

47 3.3 evaluation of results 42 laser intensity are not problematic since the the two cameras perform a relative measurement. 3.3 evaluation of results The attenuation length is measured by comparing the intensity of the laser beam after traversing a different amount of material (i = 1, 2). Each camera takes k = 1... n images of the laser beam spot at the same time. In a first step, the intensities as measured by each pixel I ik,px,raw are corrected for dark noise: I ik,px,dark corr. = I ik,px,raw I i,dark noise. (3.3) After that, the intensity of each image I ik is computed as described in Now, the intensity for each camera I i is the mean intensity of the n pictures: I i = mean n k=1 (I ik). (3.4) The attenuation length L Att = x 1 x 2 ln(i 1 ) ln(i 2 ) = x 1 x 2 ln ( I1 I2 ) (3.5) is now accessible via length x i of the tubes and the measured intensities I i in the cameras. The expected ratio of the intensities I 1 /I 2 for LAB with L Att = 20 m at 405 nm and x 1 x 2 = 0.5 m is I(1.5 m) I(0.5 m) = e 1 m 20 m = (3.6) If the ratio is up to 5% or better, it can be used to test the purity of LAB. 3.4 integration The on-line attenuation length monitor will be integrated into the purification system (see Section 2.3.3) on site at JUNO. It will be filled directly from the pipeline and after measurement, the LS will be led back into the system. There are currently two options for the installation. In any case the monitor will be part of the quality control (QC). The possible options can be seen in Figure 40 marked with red arrows. The monitor could be installed directly after the Aluminium Column. The Aluminium Column removes the optical impurities. Measuring the attenuation length at this point could prove this process successful or, if not, lead the LS back to the column.

48 3.5 testing 43 Figure 40: Possible positions for the monitor in the filling system The monitor could also be installed after the Distillation Tower. Impurities could still be detected at this point, before the distilled LAB would be mixed with the PPO and bis-msb. However, the radioactive metal ions that are removed by the Distillation Tower can mostly not be detected by the on-line attenuation length monitor. 3.5 testing The setup had to be checked for leaks and the measurement parameters were tested. First the tubes were filled with water to check for any leaks in the construction using the same tubing system that would be used for the liquid scintillator. The water was drained from the setup, the tubes were dried and all leaks were eliminated. A second filling with water proved the tightness of the system. Both times the water was left in the system as short as possible to avoid altering the coating of the tubes. The fill level was checked as well, to make sure that there is no air in the light path, as this would cause unhelpful surface effects such as reflection, deflection or interference. After ensuring the function of the purity monitor, test measurements were taken with the empty setup. These measurements are presented in Section 4.1